Applied Reliability Centered Maintenance

front matter i-xxiv.
qxd 3/3/00 2:28 PM Page i
Applied
Reliability-Centered
Maintenance
front matter i-xxiv.qxd 3/3/00 2:28 PM Page ii
front matter i-xxiv.qxd 3/3/00 2:29 PM Page iii
Applied
Reliability-Centered
Maintenance
by Jim August, PE
OME, Inc.
front matter i-xxiv.qxd 3/3/00 2:29 PM Page iv
Copyright 1999 by
PennWell
1421 S. Sheridan/P.O. Box 1260
Tulsa, Oklahoma 74101
Library of Congress Cataloging-in-Publication Data

August, Jim
Applied reliability centered maintenance / by Jim August.
p.cm.
ISBN 0-87814-746-2
1. Plant maintenance. 2. Reliability (Engineering) 3. Maintainability
(Engineering)
I.Title.
TS192 .A94 1999

658.202--dc21
99-050100
All rights reserved. No part of this book may be reproduced, stored in a

retrieval system, or transcribed in any form or by any means, electronic or
mechanical, including photocopying and recording, without the prior written
permission of the publisher.
Cover Design: Shanon Garvin and Brian Firth

Layout: Brian Firth
Printed in the United States of America
03 02 01 00 99 1 2 3 4 5
front matter i-xxiv.qxd 3/3/00 2:29 PM Page v
Table of Contents
Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .vii
Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .x
Acronyms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .xi
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .xv
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .xix
Chapter 1 Applied RCM: An Overview . . . . . . . . . . . . . . . . . . . . . . . .1
Chapter 2 Maintenance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .23
Chapter 3 RCM Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .71
Chapter 4 Plant Needs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .113
Chapter 5 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .161
Chapter 6 Lessons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .195
Chapter 7 Fast Track . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .255
Chapter 8 Maintenance Software . . . . . . . . . . . . . . . . . . . . . . . . . . .301
Chapter 9 Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .321
Chapter 10 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .341
Appendices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .351
Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .353
Further Readings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .391
RCM Software Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . .437
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .477
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .481
v
front matter i-xxiv.qxd 3/3/00 2:29 PM Page vi
front matter i-xxiv.qxd 3/3/00 2:29 PM Page vii
Figures
1-1 Idarado Bar and Rod Mill . . . . . . . . . . . . . . . . . . . . . . . . . . . .5

1-2 Failure Distributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7
1-3 Maintenance Terms Map . . . . . . . . . . . . . . . . . . . . . . . . . . . .12
2-1 Maintenance Continuum . . . . . . . . . . . . . . . . . . . . . . . . . . . .24
2-2 Practical PM Alignment . . . . . . . . . . . . . . . . . . . . . . . . . . . . .26
2-3 Maintenance Electromagnetic Spectrum Analogy . . . . . . . . . .28
2-4 Fossil Unit Forced Outage Rate . . . . . . . . . . . . . . . . . . . . . . .31
2-5 Boiling Water Reactor Air Operator Valve "Failures" . . . . . . . .34
2-6 Change! . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .38
2-7 Maintenance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .39
2-8 Specific Maintenance Processes . . . . . . . . . . . . . . . . . . . . . . .40
2-9 Condition-based Maintenance . . . . . . . . . . . . . . . . . . . . . . . .42
2-10 Engineering (Design)-Limited Failure Aging . . . . . . . . . . . . . .44
2-11 Time Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .47
2-12 PM Performance "Triggers" . . . . . . . . . . . . . . . . . . . . . . . . . .54
2-13 Coal Mill Fire Suppression . . . . . . . . . . . . . . . . . . . . . . . . . .69
3-1 Modern Blending Coal-fired Power Plant . . . . . . . . . . . . . . . .72
3-2 "Critical" Streamlined RCM/PMO Approach . . . . . . . . . . . . . .78
3-3 Equipment Maintenance Standard . . . . . . . . . . . . . . . . . . . . .80
3-4 System Component Hierarchy . . . . . . . . . . . . . . . . . . . . . . . .82
3-5 Blocking Tasks Reduces Outage Duration . . . . . . . . . . . . . .89
3-6 Coal Belt Assembly Functional Failures . . . . . . . . . . . . . . . . .93
3-7 Equipment Failure Hierarchy FMEA . . . . . . . . . . . . . . . . . . . .94
3-8 Ishikawa Fishbone Example . . . . . . . . . . . . . . . . . . . . . . . . . .97
3-9 Fault Tree: Loss of Cooling . . . . . . . . . . . . . . . . . . . . . . . . .106
3-10 Random Limit Floor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .107
vii
front matter i-xxiv.qxd 3/3/00 2:29 PM Page viii
3-11 Fault Tree: Loss of Cooling . . . . . . . . . . . . . . . . . . . . . . . . .108

3-12 "Run-in" . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .109
3-13 Turbine Failures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .110
3-14 Cumulative Turbine Failures . . . . . . . . . . . . . . . . . . . . . . . .111
4-1 Pareto Chart of System Loses . . . . . . . . . . . . . . . . . . . . . . . .117
4-2 PM Performance Process . . . . . . . . . . . . . . . . . . . . . . . . . . .126
4-3 IR Sootblower Failures . . . . . . . . . . . . . . . . . . . . . . . . . . . . .137
4-4a Nuclear 4160V Breaker Failures . . . . . . . . . . . . . . . . . . . . .139
4-4b Fossil 4160V Breaker Failures . . . . . . . . . . . . . . . . . . . . . . .139
4-5 System "Black Box" Model . . . . . . . . . . . . . . . . . . . . . . . . . .156
4-6 Two Sides of Failure Management . . . . . . . . . . . . . . . . . . . .157
4-7 Boiler Feedpump in a 50% Redundant Train . . . . . . . . . . . . .158
4-8 Failure Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .159
4-9 System Part Functional Relationships . . . . . . . . . . . . . . . . . .160
4-10 Failure Progression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .160
5-1 Failure Curves for 93% of Equipment Components . . . . . . . .179
6-1 Best Value? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .202
6-2 Breaker Failure Summaries . . . . . . . . . . . . . . . . . . . . . . . . .222
7-1 Maintenance Terms "Map" . . . . . . . . . . . . . . . . . . . . . . . . . .257
7-2 Failure Progression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .265
7-3 Traditional PM Development . . . . . . . . . . . . . . . . . . . . . . . .270
7-4 Backlog Maintenance Timing . . . . . . . . . . . . . . . . . . . . . . . .281
7-5 Plant Instrument Air/Service Air Equipment Group . . . . . . . .288
7-6 Modified Work Control Process Flow . . . . . . . . . . . . . . . . . .289
7-7 Residual Heat Removal Equipment Group Register . . . . . . . .291
7-8 Can It be Worked On-line? . . . . . . . . . . . . . . . . . . . . . . . . .293
7-9 Modified Maintenance Process . . . . . . . . . . . . . . . . . . . . . .297
8-1 CMMS Hierarchy of Equipment . . . . . . . . . . . . . . . . . . . . . .302
8-2a Sootblowing Air Compressors: Task Summary . . . . . . . . . . . .308
8-2b Compressors Loaded (Round) . . . . . . . . . . . . . . . . . . . . . . .309
viii
front matter i-xxiv.qxd 3/3/00 2:29 PM Page ix
9-1 Process Change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .321

9-2 Maintenance Cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .323
9-3 Maintenance Work Cost . . . . . . . . . . . . . . . . . . . . . . . . . . .336
10-1 Fort. St. Vrain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .342
10-2 Maintenance Strategy Map . . . . . . . . . . . . . . . . . . . . . . . . .345
ix
front matter i-xxiv.qxd 3/3/00 2:29 PM Page x
Tables
1-1 Common Outage Intervals . . . . . . . . . . . . . . . . . . . . . . . . . . . . .16
2-1 Example of Time Base Intervals . . . . . . . . . . . . . . . . . . . . . . . . . .49
3-1 Equipment for Standard Templates . . . . . . . . . . . . . . . . . . . . . .81
3-2 Strategy Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .85
4-1 Value . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .128
4-2 PM Basis Format . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .133
4-3 PM vs. CM and WO classes . . . . . . . . . . . . . . . . . . . . . . . . . . .154
4-4 Component, Function, Part Failure . . . . . . . . . . . . . . . . . . . . . . .156
6-1 Tick Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .221
6-2 Critical Instruments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .246
7-1 Areas Not Worked On-Line . . . . . . . . . . . . . . . . . . . . . . . . . . . .292
x
front matter i-xxiv.qxd 3/3/00 2:29 PM Page xi
Acronyms List
AB A or B (with respect to train, or piece of equipment)

A-E Architect-engineer
ALARA As low as reasonably achievable
ANI American Nuclear Insurers
ANSI American National Standards Institute
ARCM Applied reliability centered maintenance
ASCE American Society of Civil Engineers
ASME American Society of Mechanical Engineers
ASQC American Society for Quality Control
B/C Benefit/cost
BFP Boiler feed pump
BPV Boiler and pressure vessel
B&W Babcock and Wilcox
BWR Boiling water reactor
Cal. Calibration
CBM Condition-based maintenance
CC Channel checks
CCF Common cause failure
CDM Condition-directed maintenance
CDM (FF) Condition-directed maintenance (failure finding)
CE Combustion Engineering
CEM Continuous emissions monitor
CFR Code of federal regulations
CIC Component identification codes
CM Corrective maintenance
CNMM Condition-monitoring (based) maintenance
CMMS (Legacy) Computerized maintenance management systems
CNM Condition monitoring
CO Conditional overhaul
CRT Cathode ray tube
CT Combustion turbine
CWP Circulating water pump
CWT Circulating water tower
DC Design change
DCS Distributed-control system
DOE Department of Energy
xi
front matter i-xxiv.qxd 3/3/00 2:29 PM Page xii
DOT Department of Transportation

DP Differential pressure
E Emergency
EEI Edison Electric Institute
EFOR Equivalent forced outage rate
EG Equipment group
E MWR Emergency maintenance work request
EO Equipment operator
EPA Environmental Protection Agency
EPRI Electric Power Research Institute
EQ Environmentally qualified
FAA Federal Aviation Administration
FAI Failed as is
FC Fails closed
FD Forced draft
FERC Federal Energy Regulatory Commission
FF Functional failure
FMEA Failure modes and effects analysis
FMECA Failure modes and effects criticality analysis
FO Fails open
FOR Forced outage rate
FT Functional tests
FTA Fault tree analysis
FTM Fixed-time maintenance
GADS Generation availability data system
GE General Electric
GPA Grade point average
GUI Graphical user interface
HEU Hydraulic equipment units
HP Horse power
HRSG Heat recovery steam generator
HTGR High temperature gas reactor
I&C Instrumentation and control
IA Instrument air
ID Induction draft
IEEE Institute of Electrical and Electronic Engineers
INPO Institute of Nuclear Plant Operations
IPP Independent power producer
KISS Keep it simple, stupid!
LAN Local area network
LCM Life-cycle maintenance
xii
front matter i-xxiv.qxd 3/3/00 2:29 PM Page xiii
LCO Limiting conditions for operations

LTA Logic tree analysis
LWR Light water reactor
MIS Maintenance information system
MORT Management oversight risk tree
MOV Motor operated valve
MPFF Maintenance preventable functional failure
mREM One thousandth of a REM
MS Microsoft
MSG-3 Maintenance Steering Group-3
MTBF Mean time between failures
MTTR Mean time to repair
MW Megawatt
MWO Maintenance work order
MWR Maintenance work request
NCE New century energies
NDE Non-destructive examination
NEC National electric code
NERC North American Electric Reliability Council
NFPA National Fire Protection Association
NPM No planned maintenance
NPPD Nebraska Public Power District (utility)
NRC Nuclear Regulatory Commission
NSM No scheduled maintenance
O&M Operation and maintenance
OCM On-condition maintenance
OCMFF On-condition maintenance (failure-finding)
OEM Original equipment manufacturer
OOS Out-of-service
OM Operator manual
OSHA Occupational Safety and Health Administration
OTF Operate to failure
P&ID Process and instrumentation drawings
PA Primary air
PC Primary containment
PCRV Pre-stressed concrete reactor vessel
PdM Predictive maintenance
PG&E Pacific Gas & Electric
PM Preventive maintenance
PMO Preventive maintenance optimization
PRB Powder River Basin
xiii
front matter i-xxiv.qxd 3/3/00 2:29 PM Page xiv
PUC Public utility commissions

PV Present value
PWR Pressurized water reactors
R Reliability
RAM Reliability, availability and maintainability
RCA Root cause analysis
RCFA Root cause failure analysis
RCM Reliability centered maintenance
RD Response-driven
REM Roentgen equivalent, man
RL Random limit
RO Reverse osmosis
RPS Reactor protective system
RTF Run to failure
SAE Society of Automotive Engineers
SBAC Soot-blowing air compressor
SL Straight line
SNAFU Situation normal all fouled up
SOA Society of Actuaries
SP Surveillance programs
SPC Statistical process control
SRCM Streamlined reliability centered maintenance
SSC Structures, systems and components
SWOT Strength weakness opportunity threat (analysis)
TBM Time-based maintenance
TC Thermocouple
TMI Three Mile Island
TPM Total productive maintenance
TQM Total quality management
TRCM Traditional reliability centered maintenance
UAL United Airlines
VAR Volt amp reactive
VM Vibration monitoring
VOM Volt ohm meter
VWO Valves wide open
WO Work order
WSSC Western States Coordinating Council
Y2K Year 2000
xiv
front matter i-xxiv.qxd 3/14/00 5:07 PM Page xv
Acknowledgements
...a chaise breaks down, but doesnt wear out
-Oliver Wendell Holmes, The Deacons Masterpiece

(with credit to Stan Nolan & Howard Heap)
Elements of reliability centered maintenance (RCM) arent new;

time based maintenance, make-or-buy, re-work, performance testing
and corrective maintenance (CM) have been terms used traditionally
to describe aspects of scheduled maintenance programs. On another
level, RCM brought order to a confused and complex subject. Like
Inuit (Eskimo) languages many terms for snow, maintenance has many
descriptions. The beauty of RCM is the order that it brings to these
terms in the context of a strategy.
Strategy is an underlying theme of RCM. As in chess or war, strat-
egy requires supporting tactics. These are mastered as preliminaries.
Without appreciating the tactics of maintenance, the need and value
for strategy can be missed. Even with a strategy, tactical battles can be
lost. Yet with a strategy comes a comprehensive vision for managing
short, intermediate, and long-term maintenance. A strategy provides a
resource map that uniquely identifies the multiple roles that must be
played by various work groups.
Experts come into play in several ways: (1) developing the strate-
gy and supporting tactics for the existing plant, (2) identifying the
paths to future goals, and (3) managing the emerging maintenance
requirements of the plant. By whatever name its given, understanding
on-condition/condition-directed maintenance complementary tasks is
the general lesson of RCM. Too often in my experience, failures
werent recognized because limits werent defined. Redundancy or
sheer nerve were adequate support decision-making. At some point,
otherwise failed equipment can be operated. Perhaps not economical-
ly, but operated.
xv
front matter i-xxiv.qxd 3/3/00 2:29 PM Page xvi
To assist the reader with the terminology, I offer a rough term

equivalence guide. What we used to call preventative maintenance
(PM) is now recognized as mainly rework and restore time-based
maintenance (TBM). What we used to call corrective maintenance is a
combination of condition-directed, condition-based, and failure main-
tenance. True failures are infrequent in even reactive environments.
They virtually dont occur in some. Surveillance and operating tests
now have formal roles as special types of on-condition maintenance
tasks. These tasks trigger maintenance, of course. The contribution
of RCM is:
to put order into the maintenance lexicon and tool kit

identify how to reduce operating costs using condition monitoring
and on-condition maintenance (OCM) tasks while maintaining
operating goals
Id like to recognize all those who helped me with this work

friends, family, and peers. My friends patiently tolerated my probing
questions and helped review materials. My family tolerated my long
hours and absence. My reviewers tolerated my long-winded initial
drafts! (I recommend writing a book to everyone.) Perhaps foremost
though, Id like to recognize the giants who led the way, including the
commercial aviation pioneers, the Federal Aviation Administration
(FAA), and the Airline Transport Association. The risk of putting
together a list is leaving someone out. Published works capture the
ideas of many people. The Department of Defense deserves special
merit for noting a promising technology, and chartering its documen-
tation in published form (Reliability Centered Maintenance). Like
Newton, we stand on the shoulders of giants. But like Einstein, they
painstakingly put together pieces of a puzzle that were already avail-
able in various nebulous forms.
For my reviewers Id like to cite those who provided specific com-
ments and assistance. First, Id like to recognize Alan Bern, my col-
league at TU Electric. Alan painstakingly plowed through the first
roughs with exuberance and helpful comments. On the technical
level, Alans detailed review was invaluable. Second, my long-time pal
xvi
front matter i-xxiv.qxd 3/3/00 2:29 PM Page xvii
Joe Hunter, who shared an operators perspective. Next, my friend

and former boss who provided crucial support in times of need
Frank Novachek. (We havent always seen things the same, but our
friendship has endured.) Next, my colleague Jon Anderson, who
knows more practical maintenance than I could ever dream of. Lastly,
I want to thank the many fine crafts who have been so helpful sharing
their insights over the years. The fixers are a crusty crew, difficult in
the extreme to converse with, but a truly fine bunch of people whove
made this career worthwhile. Corporate supportersthose who pro-
vide work and a place to practice the craftare also most welcome.
These include Jim Love, manager at New Century Energies (NCE),
along with Dick Chuvarsky, Jim Stevens, Richard Roe, Mike Blossom,
at Arapahoe (NCE), and Mike Young, Chuck Fidler and Chuck
Gaines from Cooper (Nebraska Public Power District [NPPD]).
They supported the pragmatic work side to allow this to happen.
A former manager once exclaimed that he knew of no one else
who made so much effort to understand, and fundamentally grasp
things. I take that in a complimentary way. Make no mistake, low
cost maintenance ultimately boils down to technical expertise. If you
have technically competent staff youre lucky. Some companies treat it
superficially. But like chess, the overall strategy ultimately comes
down to understanding the roles and selections of the pieces.
Equipment and diagnostic technology must be understood, and this
understanding comes only with years of experience.
In many ways RCM is nothing but a framework. To achieve the
program requires finishing, so to speak. At this age I can finally admit
I know very little of the useful knowledge in this world, and learn
more every day. The beauty of well-designed equipment is that you
really need to know very little to operate it successfully. The corollary
is that you need to know useful knowledge very well to diagnose and
repair equipment effectively. While most people are reasonably suc-
cessful with modest levels of understanding, those who know equip-
ment well and have the best maintenance model turn in superior per-
formance. RCM provides a model to point out our weaknesses so we
all learn more quickly.
Finally, what can I say about applied RCM (ARCM)? Why the
xvii
front matter i-xxiv.qxd 3/3/00 2:29 PM Page xviii
new term? RCM has been given a black eye by analysis. Some would
make it a religion or take it into non-measurable, esoteric philosophy.
I believe that RCMs most appropriate application is fundamentally as
a technology. Theres little new except complexity, failure behavior,
numbers, exploration, and RCMs integrating perspective. Since no
one has offered a completely comprehensive text emphasizing mainte-
nance technology, I offer this work.
The most tedious and difficult aspects of RCM are selecting task
limits and intervals. Fortunately, RCM also provides techniques to
provide answerseven with incomplete informationwith powerful
alternatives that help us to manage risk. Technical competence is
taken for granted; it may or may not be available.
ARCM is fundamentally about using RCM for value, to under-
stand which paths to pursue, what ax to grind, and where to focus.
ARCM is finding those things that provide value in a specific setting,
and doing them. Just do it!
Know your equipment. Know how it ages. Know that it is aging,
and how to restore it. And when its broken, just fix it! This last mis-
sive is the hardest. A chaise breaks down but never wears out!
xviii
front matter i-xxiv.qxd 3/14/00 5:07 PM Page xix
Preface
Things that matter most must never be at the mercy of those that matter least.
-Goerthe
There hasnt been a maintenance best seller since Zen and the
Art of Motorcycle Maintenancea book on philosophy. The authors
theme was his love/hate relationship with technology, expressed as his
motorcycle.
Maintenance is a tough subject to write about. Its so dry! Yet its
a subject we all immediately relate to, both professionally and as people.
The last significant new book on technical maintenance intro-
duced us to RCM, in 1978. Since then, a few new maintenance terms
have been added, such as total productive maintenance (TPM) and
total quality management (TQM). Asset management is the latest twist
on the subject, as I write this. The original United Airlines (UAL)
work, Reliability Centered Maintenance, by Nolan and Heap, pub-
lished in 1978, has led to 20 years of implementation history in the
nuclear generation industry. Several re-interpretations have been pub-
lished. My purpose in this book is to provide new fundamental inter-
pretations of ARCM based upon the original theory, and to discuss
them in terms of real-world problems and experience. These prob-
lems at one time or another demanded hours of analytical thought,
planning and performancetheir learning was in some instances bit-
terly bought.
RCM offered a fresh perspective on maintenance, focusing on the
theory and methodology of traditional, non-aeronautics RCM.
However, the original RCM fieldaeronauticsdiffers radically from
power generation. RCM texts currently available could easily leave
the impression that applying RCM requires the skills of engineers and
mathematicians.
In fact, RCM summarizes practical experience, putting it on a firm
engineering and mathematical foundation. RCM provides a fresh per-
spective on maintenance to enable us to make better, more informed
xix
front matter i-xxiv.qxd 3/3/00 2:29 PM Page xx
maintenance decisions. Practically, RCM principles can be learned

and implemented in day-to-day work situations. Organizations that
want to apply simple RCM lessons can do so quickly. While engineers
and managers need RCM involvement, it is more important to achieve
shop-floor RCM recognition and application. This involves reducing
traditional RCM (TRCM) materials to simple catechisms and applying
them to daily, routine maintenance problems. RCM offers a powerful
tool to operationalize maintenance.
For maintenance managers, RCM offers a powerful tool to justify
effective programsespecially when working under distant, obscure,
or conservative vendor, government, or corporate guidelines. In
todays environmentwith its ever-more prescriptive regulation and
competitive pressuresRCM is a tool that organizations can use to
recapture the maintenance initiative.
But to have any value RCM must focus on implementationoper-
ationalizing many maintenance activities that the very best organiza-
tions have implicitly practiced all along. Such an emphasis on imple-
mentation is fundamentally different from that addressed by TRCM
approaches. Organizations like the Electric Power Research Institute
(EPRI) have developed simplified methods, termed streamlined RCM
(SRCM). A fossil version has been coinedpreventative maintenance
optimization (PMO). These differ in fundamental ways from the
approach presented herean approach I call ARCM.
Taken together, I believe that these methodsunlike those now
accepted as traditional generation RCM and documented in several
published textsrepresent a simpler route to RCM benefits and
return to the original RCM methods developed by Stan Nolan and
Howard Heap. The distinctions may appear arbitrary to those who
believe they are versed in the subject because they differ from utility
RCM in several subtle but important ways. But these minor distinc-
tions allow maintenance organizations to capture the value and effective-
ness discovered by the commercial air industry more than 20 years ago,
when RCM was first put to the testthe methods that still work best.
My ARCM approach is based on experience and based upon the
supposition that the primary barriers to RCM implementation are not
in the maintenance work force but in the prevailing state of mainte-
xx
front matter i-xxiv.qxd 3/3/00 2:29 PM Page xxi
nance. What is maintenance? Who administers it? How should it

be managed and performed?
Maintenance is democratic and relies upon autonomy in utility
environments. Few traditional engineering staffs are fluent in this
world. Most of the new tools and products were introduced by con-
tractors. Most non-military, non-aerospace maintenance departments
practice in the traditional maintenance environment. Utilities with
whom I have worked fit this model. For successful implementation, it
is essential that RCM materials and methods are suited to capital M
maintenance.
Yet, where formal RCM studies precede formal implementation,
results have been mixed. Management unfamiliarity with RCM princi-
ples and the absence of formal project management methods and
measures hurt. Shop-floor readiness for change has influenced success
but utilities traditionally have had low learning rates (supported by
regulation.) Sometimes changes (even necessary ones) overwhelm an
organization until they put the RCM effort on the shelves.
Implementation demands too much too fast.
RCM has been described as the right maintenance at the right
time, but who knows the right maintenance and the right time? How
do they find this out and inform the organization? When the front
office calls the shots, a planned maintenance effort succeeds to the
degree the shops follow its lead. Operations, maintenance, and techni-
cal support staffers often feel they intuitively know the right mainte-
nance and the right times, but cannot connect with those hardheaded
guys who schedule outages, select parts, or establish practical expendi-
tures. They feel swept aside by maintenance infrastructure.
If the perception is that RCM is theoryirrelevant to field mainte-
nanceARCM provides techniques and simple, applicable insights
that improve maintenance performance. RCM theoretically provides
complete maintenance solutions (and regulatory support bases) but
TRCM-based maintenance requires complete, closed-form system
studies. My experience as a practicing engineer is that RCM can be
implemented on a daily basis for most routine maintenance at indus-
trial facilities. Through time, properly documented and retained, RCM
measures support complete equipment and system maintenance pro-
xxi
front matter i-xxiv.qxd 3/3/00 2:29 PM Page xxii
grams. On another level, there are only a few unique and new dimen-
sions in RCM. One of these is inherent reliability. The single most
important maintenance lesson is that when you have a maintenance
problemfix it. No benefit results from deferring or ignoring mainte-
nance. The likely results are complicationscompounded problems,
secondary failures, and greater expenses.
This book encourages small, simple steps to help implement RCM
and to help guide big projects in organizations with formal mainte-
nance foundations. These methods counter prevailing wisdom and I
eagerly look forward to discussions I hope to generate.
Examples cited here all have basis in fact and are taken from an
historical perspective. Some were faced as long as 20 years ago! In
some instances the plant and even the companies no longer exist. All
are provided to stimulate thought and provoke reader reflection. I
believe that had we known of these methods at the time, and had
broader recognition of their validity, we would have been more effec-
tive and may have avoided expensive consequences. On the other
hand, where we decided to be effective at the time, we typically were.
So, what key points of ARCM are covered in this text?
The fundamental RCM strategy classifications

Simple, neat documentation
Measurement
Factual basis maintenance planning
Explicit failure resistance limits specified for on-condition
maintenance
Partial solutions, including:
outage reviews
PM reviews
safety modification reviews
Strategy
A fast-track approach to RCM described here can be applied at

most facilities. These differ across industries and facilities, but my
general experience is there are two high-value RCM areas:
xxii
front matter i-xxiv.qxd 3/3/00 2:29 PM Page xxiii
Instrumentation and Controls (I&C)

Implementation
I cannot over-emphasize that implementation is where the value is,

and that benefits come from ARCM discipline introduced at the shop-
floor level. ARCM can deliver the results for those who have the
stomach to go after them. Organizations operated with maintenance
backlogs provide a worker security blanket. To substantially change
these organizations will threaten equilibrium! ARCM can and will
change your maintenance processes. It may turn out that you discover
things you suspected all along, but could never prove!
Regulated utilities had the luxury of operating in the cost-plus
world for many years. The need for greater (and documented) pro-
ductivity is only dawning on these work forces. Maintenance workers
need tools that can help them make these changes, and make signifi-
cant contributions. ARCM provides just that.
This book is directed towards providing practical, useful material
to assist generation companies with process reliability. And while
many reliability aspects are inherently known, the subject has been
taught on the job with the sink-or-swim approach. This book offers
specific operations, maintenance, and engineering tools to improve
reliability and availability and reduce costs at the plant level. It shares
new ways of viewing maintenance so users can make more informed
decisions. Most readers will be able to follow this text knowing little
about TRCM theory. Although we bring it all together at the end, for
those wishing to firm up their theoretical RCM understanding, I rec-
ommend browsing any of the RCM texts cited in the references. Most
are in print, a few are excellent, and all are worthwhile.
xxiii
front matter i-xxiv.qxd 3/3/00 2:29 PM Page xxiv
chapter 1 1-22.qxd 3/14/00 5:08 PM Page 1
Chapter 1
Applied RCM: An Overview
The problem with twin-engine planes is that they double your chance of engine failure.
-(Aviation anecdote)
RCM is a maintenance perspective in an operational context

understanding plant goals, needs, and equipment (e.g., how equipment
serves, ages, and fails), and then developing a maintenance strategy to
optimize outcomes in the context of your goals. It opens an operations-
maintenance dialog. It recognizes the joint roles of operators in main-
tenance and maintenance in operations. When we understand and
quantify roles in the maintenance process and understand maintenance
limitationswhere and when to involve design engineeringwe get
better in all ways.
These words are obvious but, sadly, operations and maintenance
have been overlooked in recent industrial research in the U.S. It wasnt
always this way.
Precursors to RCM
Early in the 20th century, Frederick Taylor, William Shewhart, W.E.
1
Applied Reliability-Centered Maintenance
Deming, Joseph Juran, and other Americans developed models for

improving production, management, and organizational theory. Western
Electric, Ford, General Electric, United States Steel, and Westinghouse
led the world as manufacturing powerhouses using these models.
Americans prided themselves for producing the best products with the
best methods. Understanding how things worked, inside and out, was
taken for granted as American know-how. Profound process knowl-
edgethe foundation of American industrywas an accepted standard.
As skill levels and standards of living changed, the workforce
changed as well. Production prowess became less important as the post-
World War II era progressed. American goods were still in demand but
inflation of the 70s was paced by cultural and social change. Capital
availability dropped. The energy crisis developed. Global competitors
arose. American industries were put on the defensive, their industrial
hegemony ended. We still had what it takes, but as the century pro-
gressed, it became time to rethink some processes, to sharpen the com-
petitive edge. This was the backdrop for development of RCM.
Development of RCM
The term comes from the title of the work, Reliability Centered
Maintenance by Stanley Nowlan and Howard Heap. It was published
by United Airlines in conjunction with the U.S. Defense Department. It
remains available through the National Technical Information Agency
and the Department of Commerce.
RCM fills a void between reliability (R) engineeringfocusing on
the theory and mathematics of Rand the workplace, where maintain-
ing production is key. Applied conscientiously over time, RCM provides
production focus. While there are other tools (and no single one is per-
fect), and although tools and processes overlap in approachesand
adjunct tools include training, technology, and softwareRCM is par-
ticularly suited to American culture and needs.
Consultants sell versions of RCM. At least 10 different software
packages purport to allow users to perform RCM. Two-to-four page
magazine ads in maintenance periodicals promise to teach RCM in three
days. (I wish these guys had been around when I took integral calculus.
Perhaps I could have learned that in three days!) Some companies
2
practice more than one version. If for no other reason than to engage
small talk at industry conferences, its useful to know what RCM is
and what is it not.
RCM has other names. PMO is one. Common sense is another.
There are certainly competing versions of RCM, as well. An RCM
process standard has been drafted. Questions outnumber answers:
Are there fundamental RCM attributes? What are they?

What key factors characterize a program, person, or company as
RCM-based?
What key factors demonstrate the degrees to which different
programs achieve RCM?
The answers incorporate the best elements of the original aerospace

maintenance and R developments of the 50s and 60sfailure analysis
theory, work performance consistency, and quality. Learning these
answers requires an intense awareness of maintenance by participation
and benefits from experience implementing and improving mainte-
nance programs. How well RCM elements are implemented into work
practice determines the degree to which an organization embraces
RCM. What work practices indicate that implementation has been
achieved? They include:
dynamic maintenance programs

ongoing maintenance dialog among operations,
maintenance, engineering, and support staffs
awareness of operating and work strategies at the performer level
active, effective maintenance with frequent design
engineering interactions
cost-performance information at the system and equipment level
buttressed by failure statistics
focus on improvement and improvement ideas
continual cost reduction
improved availability
the ability to identify and eliminate low value work at all
organizational levels
3
personnel competence in areas of expertise; awareness of others

competencies (cross training)
obsession with continuous assessment and interpretation of
plant condition and health
the ability to take organizational action based on observed
performance trends
questioning attitudes
RCM-minded organizations dont just operate plantsthey

improve plant operations. Before you conclude this objective is obvi-
ous, ask yourself:
How many organizations truly focus on plant operations?

How many presume operations will follow?
How many support plant operations and production from
all perspectives?
High-performing maintenance organizations share attributes with

high performers identified in other fields.(Fig. 1-1)
Origin
Maintenance came into its own as a concept with the industrial
revolution. Before that time, machines were designed, built, and main-
tained by their users. Watt, Edison, Westinghouse, the Wrights, Sikorsky,
and a long list of other brilliant people conceptualized, developed, test-
ed, and debugged their own designs. They had few peers, for design is
the realm of sheer genius. Design-build-operate information exchange
wasnt necessary-they were integrated in one and the same person.
The industrial revolution differentiated processes. Product users
became separate from product makers. As production became depend-
ent upon machines, specialtiesoperators and maintainersemerged
and evolved into different jobs. Scientific work analysis (espoused
early in this century by Frederick Taylor) found that there were benefits
in specialization. The assembly linededicating low-skilled workers to
specific assembly taskstook this position to an extreme degree.
Operations diverged from maintenance. Managers didnt want opera-
4
Figure 1-1: Idarado Ball and Rod Mill, Pandora, Co. Informal on-the-job training,
remote owners, and lack of operating strategy lead to sporadic operations, high costs,
and eventual shut down. This plant employing 350 people kept the otherwise non-
descript town of Telluride from becoming yet another western ghost town in the 1960s.
tors to think about maintenancejust do the job. New technologyand

social developments such as unionizationenabled workforce differen-
tiation. Electricians evolved into crafts, as had boilermakers, mill-
wrights, and other trades. Engineering became a profession.
Instrumentation sprouted as a craft, evolved to encompass controls, and
then added software in modern distributed-control system (DCS) plants.
Specialization helped to create maintenance and other separate
work groups early in this century. This has been part of our problem.
RCM
When Stan Nolan and Rowland Heap coined the term R-centered
maintenance in their 1978 publication they summarized early jet engine
R development by the commercial airline industry and the FAA.
Ultimately, RCM was applied to jumbo airliners (beginning with the
Boeing 747) to capture practical R lessons in a highly visible field. This
work provides many of the concise RCM terms:
5
condition-monitoring
maintenance task
hard-time
logic tree analysis
on-condition
effectiveness
age exploration
failure-finding
time-based
R studies in the late 50s were driven by the large lead the Soviets
apparently held in missile technology. Spurred on by congressional
funding, R studies in defense and aerospace took many paths. Spin-off
benefits included development of:
failure modes and effects analysis (FMEA) theory

fault tree analysis (FTA)
general failure analysis
management system failure analysis and management oversight
risk tree (MORT)
along with other R techniques.

Failure study provided lessons that were counter-intuitive. Large
statistical populations were examined and preconceived notions were
put aside. Some surprising results developed. One, for example, was the
general thumb rule about wearout. (Wearout is the notion that every
part has a finite lifetime, after which it will deteriorate quickly to failure.
This in turn necessitates replacement.) Wearout couldnt be proven in
many real applications because most products never reached the
wearout stage in service. Yet, its one of the most fundamental assump-
tions of any maintenance program (Fig. 1-2).
Using systems analysis, it was recognized that in complex designs,
individual components and their functions could be replicated in ways
that reduced or eliminated the consequence of their failure. In this way,
individual component failures had little or no consequence by design.
6
Figure 1-2: Failure Distributions
Systems failednot individual components. Design could manage

component failures. This shift in perspective to a system approach was
profound. On the other hand, one couldnt assume anything about fail-
ures. Systematic study of equipment component failure modes and their
roles in overall functionality was necessary. Failures that couldnt be
controlled had to be addressed by redesign.
Cold War defense applications pushed design envelopes. There were
mistakesthe Thresher sinking, the Apollo 7 fire, and the B-58 release
problemeach representing significant design oversights. There were
also many successesthe Boeing B-52, the Navy Polaris program, the F-
4 Phantom, and the SR-71 Blackbird. From the confusion and lessons
of this period what we now call RCM evolved into a discernible strategy.
By the late 1950s, Americanslike people in other industrialized
nationshad a cultural concept of maintaining things. From this we
get the paradigm of preventive maintenancePM. Its based on the
7
notion that things shouldnt breakor wouldnt, if properly main-

tainedand when they did, someone was at fault. With proper PMs,
equipment would last indefinitely. PM went hand-in-hand with the
concept of a TBM routine and complemented the maturing technology
of the day.
Americans developed other significant concepts: preplanned obso-
lescence and operate to failure (OTF). Both play significant roles in the
development and applications of RCM. Its said that Adolph Hitler
once dismissed American industrial prowess, acknowledging only that
Americans build a good refrigerator. Intending to trivialize American
efforts, he missed his mark. Building a good refrigerator was a very wor-
thy endeavorbut the point is that the process can be transferred to
other products (some of whichtanks, fighters, bombers, destroyers,
and a host of other war productshelped to defeat him in World War
II). It also suggests that a complex appliancethe refrigeratorhad
evolved to be so reliable, it required virtually no maintenance whatso-
ever! That a complex product could operate over its entire economic
life with virtually no maintenance was an industrial milestone.
Preplanned obsolescencereplacement of a serviceable item by a
technically superior producthad arrived. Rapid advances in produc-
tion and technology and lowering of life-cycle costs led to products that
could be replaced before the end of their functionally useful lifea
uniquely American milestone. By the end of the 30son the eve of
World War IIthe stage was set for what would become RCM.
Precepts included:
PM-based maintenance strategies (TBM performance)

OTF: products that required virtually no service
for their entire useful lives
technological obsolescence: products which would be retired
prior to wearout failure
increasingly complex products in industrial and private use
a belief in technology and our collective ability to manage and
control it
8
Post-World War II
Soon after the war, television added a whole new dimension to
American life. Jet engines, rockets, and nuclear reactors were intro-
duced. Designs were refined and matured. The steam locomotive
benefactor of a hundred years of evolutionwas outflanked by subma-
rine diesel engines that had been modified for locomotive use. Post-war
production shifted to consumer products. By the 1950s, a new para-
digm presented American products and technology as the best in the
world, though new products provided new problems amid the techni-
cal advances. Technology growth led to a second preventive mainte-
nance paradigmpredictive maintenance (PdM).
PdM suited the rapid advances in diagnostics and equipment taking
place at the time. Using our insight into the mechanics of failures, we
would be able to predict when things were going awry, and then head
them off before they did. The Department of Defense applied PdM on
F-105s, fast-attack submarines, and M-100 Abram tanks. Maintenance
practitioners and managers embraced PdM applications such as vibra-
tion monitoring, oil sample analysis, multi-channel analyzers, and
remote telemetered data. Regulators also saw the appeal in these
philosophiesso much so, they sometimes mandated their use. Areas
of vital public interest, such as nuclear power and air transportation,
were early PdM proponents. Military procurement contracts specified
PdM use. Industrial safety and environmental protection followed.
Over time, however, requirements became more prescriptive.
Computerized maintenance management systems (CMMS) delivered
information with ease; suddenly, organizations were buried in mainte-
nance demands. More parties took interest in the maintenance process
and had resources to pursue their interests. The vast resources of the
federal government could be applied where the public interest was con-
cerned. The PdM experience bogged down and stalled.
PdM acknowledged time-based PM but emphasized that you
couldnt prevent all failures with TBM. You could do something near-
ly as good and possibly more useful, howeveryou could know when
things were starting to fail. All you needed was the right diagnostic tools
and the ability to interpret them. All it took was a little savvy and the
right technologyand Americans had both! The model held great
appeal. So much so, that thousands of predictive maintenance programs
9
were set up.

The development of the jet engine had shown shortcomings of the
PM model. PM did not always worktime-based overhauls had actu-
ally made things worse in case studies. Objective reviews uncovered the
not-so-obvious problem that maintenance did not always improve
equipment performance. The problem was not the maintenance per-
formers. Rather, it could be best explained in fundamental statistics
intrusive overhaul of previously satisfactory engines resulted in higher
failure rates.
Enter Traditional RCM

Commercial jet engines in the 1950s posed a dilemma. Under the
supervision of the FAA, and with competitive pressure plaguing air-
lines, jet engine R couldnt be guaranteed within the 1950s regulatory
rulebook. The technology was on the forefront of a lengthy product
learning cycle but low engine R had to be improved immediately. The
FAA applied the accepted maintenance standards of the day and pre-
scribed increased PM in the form of reduced hard-time overhauls. (It
halved mandatory time-based engine overhauls, in fact.) Yet, statisti-
cally, many jet engine failures reflected a kind of infant mortality-with
more frequent overhauls, engineers suspected total failures would
increase, not decrease, as desired. A better means had to be found.
In 1959, R engineers offered a plan to systematically collect and ana-
lyze actual jet engine failure data (targeting the Wright R-2800 CA-15 and
Pratt & Whitney JT-4 engines). R data they analyzedAir Force statis-
tical parts failure studies, as well as United Airlines parts usage records,
the best available data at that timesuggested re-examination of existing
failure data assumptions and interpretation. The idea of exploring aging
and wearout performance of equipment in-service was born.
The embryonic space program was also studied for failure process-
es in its man/machine systems. Failure modes, their effects, and
processes evolved to the recognition that processes as well as products
and systems failed. Root-cause failure analysis was developed and
thrived. Failure evaluation, categorization, and oversight methods such
as MORT were initiated. The data exhaustively collected by the United
10
Airlines and Air Force statistical failure studies supported new and fun-
damentally different interpretation of failures.
Advancing rapidly along several paths, failure pattern recognition
developed into the identification and study of failure modes and their
effects. Emphasis shifted from performing repairsthe historical focus
of maintenanceto understanding the causes of failure. The assump-
tion that maintenance was always effective was challenged. Systems the-
ory and evaluation of the Pratt & Whitney JT-4 engine maintenance
results laid the foundations of what has come to be called RCM.
Key aspects of the initial findings included:
systems focus
recognition of complexity as an important attribute in
modern equipment
failure classification by modes
assessment of failure mode effects on systems
numerical and statistical data evaluation of large
equipment populations
Theoretical assessment of what maintenance can (and cant) do, the

completeness of maintenance plans, and options for equipment assess-
ment identified this new approach. Spanning a period of 15 years, these
theories developed, through application, into RCM.
RCM brought together many loose maintenance ends under a com-
mon umbrella, covering the full maintenance spectrum to identify the
range of possible solutions. It brought closure to maintenance theory. If
we can identify the failure modes of interest to use.g., those whose
costs we wish to impactthen, we can identify options. These proce-
dures help us to generate a closed solution set around our options, or
indicate that the equipment cant be operated within our criteria.
RCM provides a standard, common methodology for assessing,
ranking and evaluating any maintenance environment. It encompasses
previous methods and then enlarges the spectrum to include testing.
RCM extends maintenance processes by providing a standard method
for the development and application of any maintenance program with
certain, objective results. RCM provides the structural glue that holds
together the three professions of operations, maintenance, and engi-
neering (Fig. 1-3).
11
CNM
Key:
PM -- Preventive Maintenance
CM -- Corrective Maintenance
TBM -- Time Based Maintenance
CDM -- Condition-Directed Maintenance
OCM -- On-Condition Maintenance
OCMFF -- (OCM) Failure Finding
NSM -- No Scheduled Maintenance
CNM -- Condition Monitoring
Figure 1-3: Maintenance Terms Map
Applied RCM
Because it evolved in a highly regulated environment, traditional
RCM (TRCM) includes a rigorous task selection methodology that fol-
lows detailed flow paths needed to document decision-making. This is
12
summarized in a logic tree analysis (LTA) process that works down a

systematic hierarchy to classify component decision logic.
Simplifying and summarizing the common results of this process is
reflected in ARCM.
RCM benefits come from applications. ARCM accelerates the
tedious and laborious traditional process to provide immediate applica-
tions. If the key RCM points include:
strategic mission-oriented thinking

systems equipment approach
function understanding
technology assessment
fact-based decision processes
failure understanding (especially root causes)
statistical failure analysis
profound process understanding
continuous improvement
completeness
functional failure focus
risk management orientation
benefit/cost (B/C) consideration
failure modes identification, classification, and study
then an ARCM perspective simplifies maintenance processes, supports

standardization, and identifies general strategies. It suggests where and
how to focus improvement efforts. It can also be as problematic as tra-
ditional RCM.
ARCM requires that information be applied to basic processes.
ARCM identifies the whatthe work required. Maintenance work
processes provide the how to accomplish the what. So, ARCM
requires a two-part process:
identifying the best solutions, using information, and simplified

RCM analysis
implementing the results
13
While RCM doesnt directly tell us how to improve actual mainte-

nance performance, it will tell us whether an improvement effort was
effective or not. High infant mortality rates point towards better main-
tenance practices for improvement efforts; high random failures, on the
other hand, point to operations processes. The numbers suggest where
to selectively concentrate.
ARCM helps develop a general philosophy of how operations and
maintenance best work together. It prevents functional losses by man-
aging failures. It leads to process improvement. The overall objective is
meeting mission goalsusually in the form of costs, safety, and riskso
the benefits of ARCM applications extend beyond the facility owners to
the operating and maintenance staffs, the community, and even to the
general public.
The R in RCM
Reliability defined
Mathematically defined, reliability (R) is a conditional probability
the ratio of acceptable outcomes to total trials. More exactly, R is the
probability that components, equipment, and systems will perform their
design functions without failure. Its based upon:
definition of successful outcome(s)

mission (including intended use and environment)
period of interest
conditions at onset of the mission period
While its not the purpose of this book to develop R theory, we need
to understand basic R concepts to appreciate the R in RCM. Intuitively,
we should have some benchmark R numbers in mind when we look at
any equipment. For example, is a feedpump R of 0.99995 satisfactory?
In what context? How about overall feedwater system R? Two 50%
pump combinations? Three? What are the benchmark comparison
standards? How can we relate these numbers to conditions that utility
managers more closely follow, such as equivalent availability, capacity,
and cost?
14
Basic R is noticeable mainly in its absence. Reliable products, equip-

ment, and plants establish benchmark comparisons that contrast
sharply with under-performing competitors. R has value, though devel-
oping supporting analysis is complex. For instance:
R = 1 - Unreliability
Often, unreliability is provided. In those cases, R is given by the

following:
0.99995 = 1 - 0.00005
The R0.99995can be viewed as a 99.995% probability of a suc-

cessful event outcome. Lets briefly look at R in more depth.
R engineering
R engineering applies R theory to solve engineering problems. This
is done by projecting a systems overall R and applying engineering
methods to assure those goals are achieved. When R is allocated among
constituent components, successful mission completion can be estab-
lished for new designs with relative confidence. For existing facilities,
sources of unreliability can be identified and traced back to causes
design, operation, maintenance, or a combination thereof.
Unlike military R applications that focus on individual mission
events, power plants look at operating periods. These could be:
periods between scheduled outages

calendar periods
budget periods
peak production periods
Scheduled outages vary greatly from application to applica-

tion with many possible issues of R engineering in play. Major sched-
uled outages occur on an interval of 12 months (or longer); a general
benchmark is 18 months. Most boilers and nuclear reactors adhere to
this interval for major inspections and rework. Special outages run on
15
longer intervalsturbines on 5-to-12 year intervals, for instance. In

todays economic climate, operators push design envelopes to extend
outage periods. Some common outage intervals are listed in Table 1-1.
Plant Outage Interval

Fossil Boiler Boiler Inspection 18 month
Combustion Turbine Combustor Inspection 12 month
Hydro Intake Inspection Spring
Turbine Stage Inspection 5 year
Nuclear Refueling 18 month
Table 1-1: Common Outage Intervals
Not surprisingly, a great deal of R assessment data comes to the

power industry from aerospace and military applications. In those cases,
R theory was used to assess small production runs, low volumes, single-
use components, and systems that usually involved specific, one-of-a-
kind missions. Analysis was mainly government funded. By contrast, R
studies funded in the private sector focus on high value, high volume
products (such as computers and peripherals) supporting hardware,
software, and telecommunications support. Devices may function in
hundreds of thousands of operations daily with a high cost for failures.
R assessment and benchmarking are less common in the power indus-
try today as deregulation focuses everyone on lower production costs,
and fewer companies willingly share their information.
Overall, R engineering is usually applied to high value product man-
ufacturing, and so most R work has supported weapons, space, com-
puters, and commercial air transport applications, where R in a product
can be apportioned with a R budget allocated among constituent
parts. Traditional heavy manufacturing is less frequently evaluated in R
terms, where products have typically been in productions for tens of
years. Generation is not traditionally thought of in manufacturing
termsand its not rocket science, even though R concepts apply.
Within power generation, R assessment has not penetrated non-
nuclear generation to a significant degree. With tight project budgets,
16
its easy to cut back on capital investments that have little or no short-
term payback, even if they could affect mean-time-to-repair for essen-
tial equipment. Utilities often establish a project budget to manage
costs without considering long-term operational consequences. This
provides an opportunity to apply R engineering.
That being said, in commercial generation designs, R engineering
uses general thumb rulesstandards and guidelinesto achieve
client contract goals. R is built upon incremental advances in produc-
tion methods and facilities, standardized redundancy, layout planning,
and common design packages. Designers use experience and similar
designs to project plant R. Probability risk assessments are reserved for
nuclear plants, where special requirements (such as the NRCs mainte-
nance rule) override simple economics.
There are two ways to evaluate Ra priori (before the facts) and a
posteriori (afterwards). Production R engineering looks at a facilitys a
posteriori performance, examining sources of unreliability and their
causes. By allocating unreliability downward to systems, equipment,
and components, engineers identify those areas with the greatest oppor-
tunity for improvement. They can then allocate resources to where they
will do the most good.
A priori calculations require the use of probability theory and
assumptions. This can be illustrated by tossing a coin 1,000 times. If you
get 493 heads, the a posteriori probability of a head is 0.493 or
49.3%. Probability theory tells us that for the toss of a fair coin, the a
priori probability of a head is 50%, exactly. (Strictly speaking, the mean
value probability approaches 0.50 after many tosses.)
The key assumption is that we have a fair coin. Overlooking or fail-
ing to appreciate such a simple, common assumption in a real-world
problem can be painful to the owner of a manufacturing plant stuck
with a costly retrofit, significantly different production costs, or both.
When they assess facility R projections, owners must carefully evaluate
how they were developedtheir basis. Numerical results providing R
whether casual R estimates or formal failure modes and effects critical-
ity assessment (FMECA)are rarely provided with designs. In their
absence, the owner must rely on:
17
the reputation of the architectural/engineering firm

design proposal experience
innovative context of the design
A reputable designer with a proven track record, an existing design,

and incremental enhancements supports low risk. Sometimes these fac-
tors cant all be met. Occasionally none of them can. In these cases, R
analysis can even help quantify and evaluate risk.
R theory doesnt specifically tie R outcomes to unreliability sources.
R engineering can measure system R, relate that to subsystems and com-
ponents, and reveal which individual reliabilities are needed to achieve
a given overall R. It can project R based upon supporting processes, sys-
tems, and components. It wont improve basic processes that determine
overall R and it cant assure that operations meet assumptions for avail-
ability of standby or backup systems. (This issue led, in part, to the
NRCs maintenance rule. If a site license assumes, for safety calcula-
tions, that a standby system is available 99.5 % of the time, then that
level of readiness should be maintained in the operating plan!)
Process R
TQM and statistical process control (SPC) address production
process R. Each process has different inherent design capabilities. This
concept of process capability has been thoroughly developed by manufac-
turing process engineers and statisticians. In addition, Deming, Stewhart,
Juran, Gryna and others provided many insights into the statistical basis
for production process improvement. While some companies are very
capable at improving production processes-most find it a struggle.
Yet, a goal of this book is to provide tools for generation engineers
seeking to improve plant process R to support higher unit, plant, and
system R goals. Like a body-builder developing muscle mass, however,
building intrinsic process R is a laboriously slow process.
Initially, theres lots of training and other investment with no imme-
diate payback. It takes time to generate results and earnings. Fast-track
methods can provide quicker paybacks but once advocates and sup-
porters of a process improvement project move on, its often back to
18
business as usual. Achieving fundamental change requires sustained

commitment.
A first step is the basic measurement process. R engineering builds
on the basic theory that identifies and quantifies the benefit of a two-
pump versus three-pump (one-redundant) feedwater supply system
basic configurations, series and parallel, mathematical models, and
where they are best suited. Operators should understand the relative
cost benefit of a two versus three pump configuration. Theory tells us
whether our strategies fit with the designers intent and how likely we
are to be successful. An out-of-service, functionally abandoned spare
pump fundamentally changes the designers intent. Plant staff may not
appreciate thiseven in an obvious case, such as this one.
A critical bearing-water system serviced large reactor cooling com-
pressors (circulators), in an advanced nuclear plant. A spare bearing-
water pump was intended to be always on standby, ready to run. In vir-
tually every bearing-water pump trip, the standby pump either kicked
in immediately or a plant shutdown quickly followed. If bearing-water
was injected into the circulated gas coolant, shutdowns were lengthy.
Yet, maintenance crews worked on these pumps while they were on-line
since operations could do without them. Every outage, it seemed, had
more high-priority work, more pressing problems than pump mainte-
nance. Practically, then, the pumps were out-of-service on-line. Many
times, a demand for the out-of-service standby pump failed and water
went into the gas coolant. Armed plant trips detected water in the
coolant gas and brought the unit down.
This is a classic example of what can happen when maintenance is
staffed initially from another plant. Personnels appreciation of the
importance of the bearing-water pumps was fundamentally out of sync
with the plants needs. They could have changed the prevailing main-
tenance and operations culture. The absence of a working dialog
between engineering and maintenance crippled the plant.
The lesson is clear: R theory should drive maintenancenot com-
pany practices nor plant maintenance preferences. In practice, compa-
ny and even industry culture are powerful change impediments.
19
Implementation
Implementing RCM process results is tough, and the reasons why
provide insight into generation production challenges. Some are simple,
others complex, but RCM analysis without implementation has no value.
An organization considering ARCM should first examine its basic
work management processes, to uncover implicit processes that may
be understood but not well defined. Managers and other parties to
existing processes may not know how work is actually performed; those
who perceive that potential gains would cause them to lose in any way
could block RCM applications.
Maintenance has traditionally been a craft process whose workers
usually have had great latitude to work flexibly, using their own meth-
ods, standards, and pace. For this reason, maintenance culture and
practices should be reviewed for RCM alignment. Some organizational
features align more naturally with RCM processes than others. They
must be discussed and emphasized to support the RCM effort and so
avoid later implementation pitfalls.
Many organizations find maintenance process commitments sub-
stantial, and (understandably) are reluctant to take them on. However,
once an RCM-based maintenance paradigm takes hold, RCM thinking
can provide compound returns. Simplified projects can achieve RCM
benefits quicklyeven within the budget year. The discipline can great-
ly focus efforts. This offers the added benefit of demonstrating change
success. If value can be demonstrated, most organizations have power-
ful incentives to improve.
As the pace of industry deregulation and reorganization continues
around competitive structures, companies will have to invest in mainte-
nance infrastructure to remain competitive. New emphasis will be
placed on R and process improvement offered through RCM.
Operations, maintenance, and engineering support will all benefit as
companies discover this fertile area of improvement.
Value added
If the megawatts arent available when the buyer demands them,
20
then no sale occurs. Availability creates the sales opportunity, since elec-
tricity storage is limited. Producing at a cost lower than a competitor
assures sales in a competitive market. This means that availability is a
significant value-adder and R is the practical indicator of availability
performance.
Cost is another factor. A plant with a high product costeven
though its product is availablewill not be as attractive in an eco-
nomic dispatch model. Because total generation costs include operat-
ing and maintenance production costs, plants that turn to RCM favor-
ably influence availability and cost to benefit their bottom lines.
Maintenance strategy
Nine times out of ten, operators initiate maintenance but their
operating success rests on the maintenance product delivered.
Maintenance cant correct fundamental design problems. In the past,
only combined group efforts identified design as the fundamental prob-
lem and eliminated maintenance as a solution. RCM enables us to iden-
tify misapplied maintenance quicklyfacilitating designer involvement
more effectively. This helps maintenance focus on things it can correct
and stick to a winning strategy.
Design change (DC) maintenancedesign changes initiated
where a maintenance solution is availableis expensive. Maintenance
organizations request and implement design changes either for non-
problems or for problems that have simple maintenance solutions,
because:
workers didnt understand the maintenance needs of the equipment

problems were patently maintenance issues but persistent root-
cause analysis didnt occur before initiating the DC support
request cost/benefit routine process
organizations fail to recognize the true cost of modifications
Great savings are realized when maintenance solutions are recog-

nized, design change requests canceled, and workers roll up their
sleeveswhen maintenance strategy is followed, in other words!
Daily maintenance strategies should be explicit. Reviewing implicit
21
ones often unveils substantial opportunitiesthe potential for more

online maintenance, the elimination of ineffective maintenance, the
opportunity to extend service intervals. Strategy development should
include maintenance-planning managers and the rank-and-file workers,
who have great insights on equipment performance in-service but who
dont always understand how to change and improve maintenance deci-
sion-making.
For fossil generators, an absence of plans can mean that work is per-
formed because equipment is availablenot because its needed. For
nuclear generators, complex, multiple path maintenance approaches,
extreme conservatism, and overbearing regulation cause productivity
losses. Each environment can learn from the other.
When companies find cost management so important, a strategy
can provide a clear road map of what is possible, and paths that must
be taken.
22
Chapter 2
Maintenance
How come dumb stuff seems so smart when youre doing it?
-Dennis the Menace
If it aint broke, dont fix it.

-Farmers adage
Maintenance practiced in North American power plants is crisis-

oriented. Crisisthe day-to-day emergence of random events and
directive managementis what the American industrial culture seems
to need to manage on a daily basis. Crises energize companies. When
business orientation is towards crisis, crisis is inevitable and structural.
Yet, lack of preparation for predictable events is what provides a
crisis orientation. While work will always be a dynamic environment,
proactive maintenanceas can be derived from a failure managent
based strategycan manage or remove a great deal of stress for every-
one. Isnt that a better way to operate?
The point is a simple one : O&M are harder to perform under
adverse circumstancesespecially when some plant managers wear cri-
23
Figure 2-1: Maintenance Continuum
sis events like purple hearts and some corporate cultures reward those
who promote and manage crisis, rather than stable productive work-
places. I advocate stable, predictable operations. We need to get the job
done, minimizing crisis responses! And everyone needs to go home at
the end of the day.
Maintenance Options
On a continuum, maintenance varies from purely reactivefailure
responseto purely preventivetime-based. (Fig 2-1) Looking at
maintenance across such a spectrum, theres less tendency to view any
particular maintenance approach as either good or bad. Theyre
just approaches.
I dont come to this discussion totally unbiasedI believe in
planned maintenance. Competence stems from knowing which method
is most effective, and when. Even response-based maintenance can be
planned! Different equipment with different design capabilities opti-
24
Maintenance
mizes costs at different points on the maintenance spectrum. By the

maintenance strategy we develop, we can choose where to place equip-
ment on the spectrum based on overall risk, cost, and operational objec-
tives. Equipment falls naturally into niches. To optimally place equip-
ment on the maintenance spectrum requires an understanding of the
equipment, its context, and the operating organizations culture, risk
tolerance, and goals. Is any one place better to be than another? That
depends on your operating goalsor perhaps your regulators.
With design evolution moving steadily towards complexity, redun-
dancies are incorporated into most designs. These provide opportuni-
ties to lower costs. Maintenance strategy(s) can use redundancy features
differently to lower costs while maintaining functionality.
Environment also influences component strategies. Lubrication
requirements in a dusty, dirty, or wet environment vary from those in a
clean, dry one, given the same level of equipment sealing. A dirty
environment requires frequent lubrication to purge contaminants. This
function is unnecessary or greatly diminished in a clean environment.
Constituent component capability and quality levels also influ-
ence how quickly a component ages. For example, a high-quality lubri-
cant with superior base stock and additives outlasts a simple mineral oil.
High-quality electrical insulating materials outlive simple, inexpensive
ones. Constituent material variations influence where a functional
itemoil, cable, or other materialfalls on the failure spectrum.
Understanding the capability of constituents, and how they influence
overall part capability, influence maintenance strategy. It may be more
cost-effective to use high-quality lubricating oils that possess pre-
dictable aging characteristics than to condition-monitor frequently for
degradation.
We usually associate low quality with product unpredictability.
High-quality components have longer mean time between failures
(MTBF), a more predictable lifetime, or (typically) both. Random
failure requires more efforts to:
mitigate the failure (e.g., introduce redundancy)

detect and correct the failures that ultimately will occur
25
suffer the operational loss-

es when uncontrollable
random failure occurs
The uncertainty of random-

ness in failure inherently raises
costs! This increases initial prod-
uct cost, operating cost, or both.
Ultimately, life cycle costs are
higher. Predictable failure mecha-
nisms add value. Quality manu-
facturers know this, and strive to
deliver predictable products
predictable operations, pre-
dictable failures.
Manufacturers engineering
staffs develop predictable Figure 2-2: Practical PM Alignment
products that command higher
returns. The more reliable the productthe lower its costthe greater
the demand. When quality is perceived in a reliable product, that prod-
uct commands higher prices that an informed purchaserthe guy
maintaining the equipmentis willing to pay. However, buyers indi-
rectly related to the work, who cant see or measure the costs of low
quality, may purchase on cost, not value. Most craft workers are painful-
ly aware of the cost of quality and since they have reworked many jobs
due to faulty parts.
The idea is to extend maintenance intervals through the use of
superior parts, engineering, and processes. (Fig. 2-2) The old saw that
all parts are created equal is all wet. Anyone in business eventually
learns that parts are not created equal. If analysis shows that lower
quality substitution is adequatebuy em. Until you have this profound
piece of knowledge, be careful! The unknown substitute probably isnt
a bargain. For a PM program:
use long-lived parts

align PMs
lasso the cowboys
26
Maintenance
Establish a process with rules and then ensure that everyone (even
cowboys) plays by the rules. Theres almost never a good reason to
shorten a PM interval just to get a price break on parts. The opposite
should occur: if you become aware of a premium part, analyze its
cost/benefit; if you find its cost-effective to use it, buy it. Only extend
the service-life interval based on the better part.
Doing part-lifetime analysis work is not trivial. Unfortunately, many
people think that it is, which is why theres such a large market for low-
quality parts. In a more rigorous, informed cost environment, many
cost-based part suppliers couldnt survive.
When youve analyzed, compared, and tested your components,
youre ready to build them into sub-assemblies, skids, and systems. The
overall integration determines the failures that ultimately cause overall
functional failure. Two things can happenequipment, with a life-lim-
iting part, can fail. It can also last indefinitely, with internal failures,
while preserving function. (This is the complexity principle.) If there is
a predominant age-based failure, it establishes an aging and failure pro-
file. The composite of all component failure modes over the expected
life of the equipment or assembly, and their redundancy in design,
establishes the overall composite failure characteristic and behavior.
This locates the equipment on a failure spectrum (Fig. 2-3).
Thus, the failure spectrum enables you to consider alternative
strategies and how the maintenance strategy must change when compo-
nents change. Ideally, only those changes that increase product lifetimes
would occur but, unfortunately, low-quality parts and/or services com-
promise lifetimes with the opposite effect. In some cases the systematic
downgrade of constituent parts leads to equipment capability lossthat
is, as equipment becomes less capable, useful, and maintainable, utility
decreases and aging accelerates.
Between major replacement programs, our ability to maintain
equipment drops as many small problems gradually sap the overall
equipment utility, its capability to be maintained, and its operating mar-
gins. When the operating margin is gone, failures occur. Taken togeth-
er, all across a plant, they raise overall costs. This is why planned main-
tenance programs can maintain nearly complete performance capacity.
But, how? A policy of conscious age-exploration and learninga
27
Figure 2-3: Maintenance Electromagnetic Spectrum Analogy
continuous-improvement environmentcan help assure that perform-

ance loss doesnt happen. Equipment condition deteriorates from igno-
rance, not conscious abuse. Some aging can be defined more accu-
rately as the systematic extraction of capital value from equipment by
compromising the original equipments specifications through lower
quality substitutions. A skilled, committed workforce can maintain
capital equipment capabilities to nearly original specifications and con-
trol the rate of deterioration. And because original specifications
include design margins, motivated plant personnel can maintain mar-
gins over part life.
The failure spectrum explains why composite operator diagnostic
skills are needed, and why operators are useful in maintenance. The nat-
ural differentiation between operators and maintenance workers comes
from the failures each addresses. If you have:
low MTBF
random failures
28
Maintenance
then only operator monitoring will be effective. The more the mainte-
nance strategy is oriented towards no scheduled maintenance (NSM),
the more dependent the strategy is on operators to identify failing
equipment. The best operators are literally integrated into the man-
machine process. Experienced, skilled operators can compensate for
most failures and identify developing problems through CNM. They
require little guidance. A facility with such operators who make well-
designed rounds in a well-designed plant with a responsive maintenance
process is an ideal plantfunctional failures are rare to non-existent!
Many plants meet this implemented ARCM definition today.
The failure spectrum suggests that to be effective we must really
manage risk. The effectiveness depends on plant design, combined R
factors and redundancies, and how equipment is operated. The key to
managing risk is education. We must master knowledge of:
design failure capacity

maintenance support
monitoring level
maintenance response timeliness
We can place ourselves anywhere on the risk management spec-

trum. Favorable outcomesincluding low costsdont result by acci-
dent. An uninformed acceptance of a maintenance strategy doesnt opti-
mize overall performance. Consistency does.
Consistency
Fossil generating-station maintenance processes and strategies are
implicit; nuclear plant processes are defined (though nuclear plant
processes are functionally similar to fossil). For effective RCM applica-
tions, information exchange must occur on several levels, no matter
what kind of plant is involved. These processes are unique to mainte-
nance optimization, continuous cost reduction, and performance
improvement but are not routine for many reasons.
Corporate cost information is often unavailable or inaccurate. Many
utilities are just learning cost management. Predictable costs are a chal-
29
lenge to achieve. Spontaneous, undocumented problem solving may be

commonplace, and standards absent. Predictability does not happen by
chance. Random approaches lead to higher costs. Developing consistent
information processes that support cost management requires practical
experience and time.
Consistent, integrated measurement processes provide the neces-
sary feedback information that enables staffs to tune the maintenance
plan. Available generation-oriented CMMS software and user-friendly
feedback mechanisms, accessible to any employee, provide effective
feedback. Twenty years ago, engineers had secretaries to type reports;
today they have PCs and word processors. They file their own reports
faster! The same transition can be projected for mainframe CMMS soft-
ware systems.
Consistency can be achieved! Several elements are involved.
Statistics
Informalitythe lack of a maintenance strategy that is universally
understood and appliedintroduces random factors into work per-
formance. This dilutes planned maintenance effectiveness and increas-
es the frequency and dispersion of failures. Maintenance plans must
address equipment to control failures, but this has been difficult to do,
except at the worker level.
Statistics tell us that around 85% of the tasks in a typical large gen-
erating facility are CNM and CDM, a large fraction of which should (or
needs to) be implemented by operators. Because the monitoring inter-
val for operator tasks is shorthours to dayswhat does this say for
plants that are characterized by:
informal (or no) operating procedures?

informal (or informally developed) rounds?
informal (or no) current system operating guides?
This traditional electric generation scenario helps explain why ran-

dom failure ratesin spite of mature plant designs with years of design
and operation developmentare high.
While each unique design requires a distinct interpretation, simple
30
Maintenance
Figures 2-4: Fossil Unit Forced Outage Rate (from NERC GADS)
procedures could help to assure consistent, timely performance of plant

operation tasks in fossil units. This statement is based on North
American Electric Reliability Council (NERC)-reported planned and
forced outage rates of 1.26 and 8.73, respectively, for all fossil units for
1996-1998. (Contrast these with figures of 1.23 and 2.96 for nuclear
units.) Both plant types plan for the same level of outages. Forced out-
ages at fossil units are just comparatively high (Fig. 2-4).
Maintenance assumptions, revisited

Applicability. Maintenance effectiveness based on periodic mainte-
nance cannot be assumed. Complex electronic systems developed dur-
ing and after World War II, the advent of complex mechanical systems
(such as jet turbines) and statistical life-failure studies first suggested
that PM activity effectiveness varied.
For example: The Weibull model mathematically characterized
31
infant mortalitybut applications to electron tube aging quantified it.

(Infant mortalityderived from mortality studies of human popula-
tionswas a general attribute of new unburned electronic equip-
ment. Burn-in significantly reduced early life failures.)
This lesson formally challenged the prevailing notion that PM was
automatically effective. Weibull generalized the mathematical model to
be useful to test proposed PM activity. Activity must be technically
effective to be considered valid PM. Since time-based jet engine over-
hauls actually increased in-service failures, technology should be care-
fully scrutinized with the applicability test.
New PM must be successful on two levels to be considered appro-
priate. First, is the activity technically appropriate? That is; does it real-
ly achieve its stated, intended failure-prevention purpose? We assume
here that we are evaluating state-of-the-art application of the technolo-
gy and that the analysis is performed by trained, qualified craft in a pro-
duction environment with production equipmentnot under lab con-
ditions. Does it actually work? is what were trying to answer.
Once this is assured, we can go on to the next, broader levelis
the intended work cost-effective in the production environment? In
practice, there are cut-off points where PM is no longer cost-effective
where equipment requires NSM. Formally tagging an equipment main-
tenance task or plan as NSM places it on a CNM basis and forces any
new PM task to pass the cost criterion to be applied.
Testing applicability and cost-effectivenessobjectively, statistical-
lyis demanding work. For this reason, few do it. They use other, less
rigorous techniquesor gut feelings! However, validating PM effec-
tiveness in the field is an ongoing chore that has high payback. The key
concept is to assure and validate that any PM task meets applicability
and effectiveness criteria.
Effectiveness. Technical applicability is necessary for any proposed
PM. But does it make money? Is it cost-effective to do? Cost-effective-
ness is a higher hurdle than applicability for new proposed PM tasks.
Before this test was generally applied, utilities routinely purchased
the latest test equipment as it came along and trained a person who then
became the applications advocate. The person promoted its use widely
often ignoring cost-effectiveness. As a result, programs built around
32
Maintenance
technology exploded. VM, non destructive examination (NDE), oil

analysis, acoustics, leak detection, and other test specialties were devel-
oped extensively for many components and problems. How can sam-
pling a 4 pint bearing sump be effective? The time involved to pull a
sample matches what it would take to replace the entire contents.
Whats the point in performing VM on a 10 HP motor with no (histor-
ical or conceivable) impact on plant availability? Unfortunately, these
techniques have been used in applications where there could never be a
significant returnor any return at all. Reviewed with an ARCM effec-
tiveness test, many of these tests simply cannot pass the cost-effectiveness
criteria.
Controlling PM program scope by ruthless application of the effec-
tiveness test is essential for program credibility.
Add a PM. Managers often respond to failures with the cry, do
a PM. This is especially true in response to regulatory pressure. PMs
provide a tool to keep an inquisitive inspector at bay. The assumption is
that its quick, simple, demonstrates action, and doesnt cost anything
right? Wrong. Time-based PMsany single, simple PM performed
through a typical maintenance organizationare expensive. In review-
ing bloated programs over the years, I can only conclude that many of
them had an insurance or regulatory origin, or originated with manage-
ment. In many management-derived PMs objectives are vague, task-
failure correspondence missing, crew input is absent, and planning or
understanding is missing. When these gaps are formally addressed,
theres concurrence that such PMs reflect no value addedand are
droppedor a much simpler solution is found (operator monitoring,
redesign, or NSM.)
Strict adherence to failure analysis and R engineering assures appli-
cable, effective PMs. As the primary source of operational R, what goes
into the routine maintenance system must be managed with great care
much like a checkbook. A program of controlled PMs is superior to
an informally managed program in which anyone can originate one.
The right PMs, performed consistently, add R and availability, and
they lower operations cost. A credible PM system can be a prime con-
tribution made by a conscientious plant engineering group.
Statistics and regulators. Regulators and insurers (in my experi-
33
Figures 2-5: Boiling Water Reactor Air Operator Valve Failures
ence) ignore statistics, concentrating instead on rare, improbable, yet

significant events. They track these with zeal. True risk-based regulation
considers the probability of occurrence before initiating expensive cor-
rective actions. This benefits the public interest when its applied in the
commercial aviation industry, but I dont expect to see it in electric
power generation. If anything, Environmental Protection Agency
(EPA)/environmental and Occupational Safety and Health
Administration (OSHA)/workplace focus on fossil-fuel generation
appears to be destined for more intervention.
RCM is an effective strategy to counter these trends. Truly manag-
ing failuresand having the statistics and programs to prove itwill
not only keep regulators more accountable, but will ultimately improve
regulation and enforcement. (Fig. 2-5) In many incidents in which reg-
ulatory action was threatened or carried out, in my estimation, regula-
tors were justified. In too many of these cases, as Ive said, PM was a
palliative substitute designed to keep inspectors at bay.
RCM studies suggest that regulators consider levels of redundancy
and backup before issuing citations. For example, in the nuclear indus-
try, violations can be issued for non-compliance on vague, ill-defined,
34
Maintenance
contentious, or indistinct secondary issues where no actual failures (of

people or physical hardware) can be identified. Violations are based on
what could have happened, or for breakdowns in management and sup-
port processes. Nuclear requirements are so exceedingly complex and
numerous, that 100% compliance is unrealistic. Many of the regulations
are subject to interpretation. Worse, an inspector makes his/her mark
based on citations. My opinion is that minor failures that leave the func-
tional redundancy of support systems intact, are acceptableunless
for cause or other compliance issues arise. Because nuclear power
plant designs today are mature with few new technical issues (plant
aging excepted), hypothetical failures should not be the issue. Such sec-
ondary guesswork represents an unproductive use of engineering
resources. The selective application of RCM-based statistics can clarify
cases.
In fossil plants, a three-barrier, defense-in-depth standard works
well. Design is an inherent barrier. Finding anything other than code-
specified materials and welds on high-pressure piping requires immedi-
ate correction including shutdown. The basis is that a fundamental
design assumption has been broken, putting a real hazard in place.
The second barrier is the redundancy for the primary failure cause
(usually instrumentation but perhaps other operating limits). Consider
the case of catastrophic blade separation failure in a high-speed fan due
to imbalance. Based on experience, it occurs with a warning periodif
you monitor vibration. If monitoring occurs, failure will be detected,
even if we break the first barrier. To assure that monitoring is in place
(with necessary trips) means we must have an instrument PM program.
If an operator is expected to initiate the trip (instead of an armed, pro-
grammed trip on excessive vibration), monitoring of operator perform-
ance is also needed.
The third barrier is general O&M equipment monitoring. Though
non-specific, most practical failures have predecessors, be they alarms,
warnings, limits, noises, vibrations, and even smells. An efficient moni-
toring program conducted by motivated and skilled operators effective-
ly identifies evidence of failure. Combine them and you have defense in
depth.
Application of these three barriers in fossil generation has been
35
effective in selecting appropriate CNM. By taking appropriate actions

while following the generally accepted environmental standards, we pla-
cate regulators and plant managers. Taken together, these three barriers
effectively control risk posed by 99% of failures. At plants that use all
three measures effectively, significant eventsaccidents, unit trips, and
major equipment lossesare very rare.
Instrumentation has been a sticky issue at some fossil units. Fossil
plants typically dont use armed trips when options to monitor high tur-
bine vibration levels, induction draft (ID)/forced draft (FD) fan vibra-
tions, and other faults, such as electrical faults are available.
Jumpering trips for startups is accepted practice. (Standards are
informal.) Fossil operators have great discretion to sustain operations
in the face of conditions outside normal limits. Expectations may be
unclear or unknown. Training emphasis is haphazardor absent. Yet,
when limits are exceeded, its important that operators act. From an
RCM perspective, knowing the essential functional instrumentation
limits that relate to basic safety and equipment performance is essential.
In almost all cases, the penalty is economic. Economic penalty is ever
more detrimental to competitive health though.
The philosophy often is, Let the operator run the plantwell
cross bridges when we come to them. My experience has been that in
many cases, an operator made a spontaneous callthe bridge was
crossedand someone then decided there could have been a better
response. An ARCM-based maintenance program would have assured
that the expected response is known before the bridge is crossed! With
no planned maintenance program for essential instrumentation, its easy
to see why an operator discounts an alarm. An essential alarm that is not
maintained is more than a nuisanceat best, its a trip waiting to hap-
pen in a fossil operating program with ambiguous instrumentation
guidelines.
I&C programs can greatly influence maintenance-controllable
station performance. Nuclear plant I&C is largely controlled by techni-
cal specifications, which are comparatively constant. Fossil plant opera-
tors have the discretion to establish I&C calibration intervals, alarm
check frequencies, and many other test intervals. However, because so
much I&C equipment is available, important instrument calibrations
36
Maintenance
(cals) can easily become buried beneath the trivial many. Screening
fossil I&C for unnecessary cals and other work is highly effective in
improving overall program results and assuring completion of those cals
that do make a difference.
Maintenance Process
Overview
Engineers appreciate highly complex chemical, mechanical, and
other engineering processes that can be analyzed objectively. This often
stands in contrast with organizational process awareness. Most man-
agers of production facilities are engineers, but there has been less
recognition of the soft processes as they apply to operating efficiency.
After years in the utility industryas both engineer and managerI
attribute this to a combination of cost-plus mentality and lack of pro-
found maintenance process awareness endemic to American industry.
Maintenance is not static. The constant introduction and improve-
ment of materials and processes has transformed the maintenance
process. Like other processes, the environment has influenced the pace
of change. Where 40 years ago, small simple-cycle plants and diesel
operations were replaced by huge vertically integrated utilities, today
the opposite occurs (Fig. 2-6).
Maintenance is one of many complex organizational processes that
benefit greatly from process improvement techniques. For example,
quality process theories found in manufacturing can be applied to main-
tenance performance. Maintenance can be viewed as a process that
delivers available equipment (products) in an operating facility on
a budget. (Fig. 2-7) Traditional maintenance organizations have done an
outstanding job delivering maintenance but rules are changing.
Maintenance organizations need to deliver operating equipment more
of the time at lower cost and take on more than the old maintenance
department has done. Some independent power producers (IPPs)
have replaced traditional maintenance staffs and annual-unit outages
with flexibility scheduled overhauls at lower cost. Workers literally wear
all hatsoperator, mechanic, and technicianto develop the jack-of-
all-trades utility worker.
37
Figure 2-6: Change! Although this engine captured 100 years of steam design learin-
ng, engineers could not overcome the inherent advantages of diesel locomotives.
Infrastucture requirements, high operating costs, and labor agreements made steam
no match for simpler, reliable diesels. Anachronisms lingered for 40 more years but
operating steam locomotives disappeared forever on Americas railroads between 1955
and 1960. No tear has turned back the inevitable march of technology progress.
With the modern emphasis on processand the competitive pres-

sure in the utility industrymaintenance processes are too ripe an
improvement opportunity to not change. This challenges the corporate
culture in most maintenance organizations, of course! Few floor-level
maintenance managers attend workshops and skills-improvement class-
es. Maintenance is a highly crafted product in many companies but too
many of them have historically tolerated a high level of rework.
Organizations with low training and high rework often convey an
implicit message that performing the job right the first time has little or
no organizational value. Emphasis on new, innovative work practices
means that more work performers need to see others methods.
Maintenance improvement throughout other industries includes
total quality maintenance (TQM) and TPM. While each represents a
good start at defining and advancing the maintenance profession, each
has intangible aspects that bring TQM to mind. TQM skills relate to the
38
Maintenance
Figure 2-7: Maintenance
work but arent recognized as critical organizational processes by tradi-

tional maintenance workgroups. While TQM has its place, tangible
techniques will have greater success in the North American power envi-
ronment. Successful maintenance hinges on bringing fundamental
39
Figures 2-8: Specific Maintenance Processes
organizational skills together with work performance.

W.E. Deming identified competence levels he called profound
business knowledge. These key processes, inherent to any business, are
exceedingly difficult to learn and perform and are usually developed
through experience inside the business. Employees learn these tech-
niques and processes in a work environment over many years, not nec-
essarily even understanding why they work. Such proprietary compe-
tencies present entrance barriers in competitive markets because of this.
Businesses wont openly divulge trade secrets or process details that
strongly influence cost or how a process works. (Fig. 2-8)
In this context, what are the profound elements that shape an orga-
nizations maintenance performance?
Key maintenance processes

Plan. Maintenance planning identifies necessary work and decides
how it should be performed. Repetitive work is standardized. Simple
methods increase work performance consistency. Step-by-step work
development supports the work itself. Competent crafts follow opti-
mized work plans to minimize work, trips, and parts usage, while con-
40
Maintenance
trolling key aspects of the work.

Weak maintenance organizations dont plan or adhere to planned
work. Maintenance unit managers and craft plan their own work. Work
plan standards are few or absent. Planners are not a specialized, skilled
group. Anyone can plan characterizes the approach. Yet unplanned
work is slower, has higher failure rates, and suffers from greater rework.
Statistically, most maintenance work is highly repetitive over a long-time
perspective. This supports planning.
Schedule. Weak maintenance organizations lack scheduling
processes. They maintain incomplete database work lists, work tracking
and control measures, or measurement capability in their CMMS. They
can expedite work when they must but the routine work horizon is
shortperhaps two or three days. Theyve learned to work mainte-
nance within a short horizon tuned for crisis management, but less
supportive of long term work plans.
Performance. RCM presumes maintenance performance is avail-
able. Plant and equipment failures due to inadequate performance show
up statistically as infant mortality failures with random causes.
(Consistent failure causes can be attributed to processes. Random fail-
ures must be attributable to lack of process control.) Maintenance per-
formance is a rich subject, but one that is not within the scope of this
book.
RCM (or other maintenance selection technology) helps identify the
maintenance repertoire an organization chooses to support. It can iden-
tify weak maintenance processes based on statistical and proximate fail-
ure analysis. It greatly increases the craft awareness of critical equip-
ment components, and increases their sensitivity in rework processes. It
cannot (at least, not by itself) establish or compensate for deficient craft
skills, processes, environments, materials, or other factors that are a
direct part of the maintenance process.
Many factors contribute to excellent maintenance performance.
Consistency comes from knowing the job, following standard practices,
working with close engineering support, and adequate training. These
elements are never perfect in any maintenance organization. They need
continuous focus to assure maintenance performance is consistent, and
constantly improving. Sadly, some organizations are only vaguely aware
41
Figure 2-9: Condition-based Maintenance
of factors contributing to maintenance excellence.

Training. Utility maintenance programs were traditionally centered
on the apprenticeship and journeyman steps that qualified an employee
to perform any work of that skill in the plant. Seniority and positions
were determined when the transition occurred. Apprenticeship pro-
grams provided a basic training but lacked engineering coordination
and involvement. Instead, union and managementthrough a joint
apprenticeship committeedetermined standards, periods, curricu-
lum, and other elements in managed apprentice programs that were
grandfathered through previous union contracts.
42
Maintenance
Smart utilities have initiated pre-selection testing to assure that

apprentices (and new operators, for that matter) possess the basic apti-
tude and knowledge to be successful in their work. If R and failure
study were added to apprenticeship training, it would help maintenance
restructuring towards age exploration and other strategic initiatives.
The administrative aspects of maintenancewhat information manage-
ment is, and what it tells us, from engineering and management per-
spectivesneeds to be conveyed to the craft in the field.
A mechanic who knows that a particular part gives poor service
must convey that on a work order (WO) at the time of premature
replacement, if R engineer is to know he found the part unsatisfactory,
and trigger an engineering assessment of part performance. Although
some assessments get quite detailed, many identify easily solved engi-
neering problems, once action is triggered.
Engineering. Maintenance and engineering are not traditionally
partners. Systematic support of maintenance is not the focus of engi-
neering. Where maintenance departments can justify an engineering
presence, engineers often focus on design. Few maintenance groups
having engineering support dedicated to R improvement, process
improvement, or cost reduction, because these responsibilities never
functionally flowed down to performance-level engineers. Traditional
utility engineering departments should redirect their efforts towards
operational and maintenance support to help achieve R improvement
and cost reduction. This redirection is fundamental, but difficult for
many engineering groups.
The maintenance process defined

Maintenance is initiated either by defectexception and deterio-
rated performanceor it derives from continuous monitoring, restora-
tion, and performance. The former is corrective, or responsive mainte-
nance; the latter, CDM. Organizational maintenance, therefore, can be
thought of as either fundamentally proactive or reactive. Either can be
effective.
The responsive maintenance processtraditionally used for CM
starts with an identified problem. An out-of-spec condition, a suspi-
cious noise, an unresponsive machine, a failed alarmall of these con-
43
Figures 2-10: Engineering (Design)-Limited Failure Aging
ditions trigger a response (Fig. 2-9).

CDM in the proactive organization is structured around monitoring,
rounds, and PM taskstraditionally, TBM. These tasks provide earlier
warning of failure and allow more planning time. CDM differs from CM
maintenance (CNMM) by explicit versus implicit failure definition
parameters, and limits that act as hard-triggers to initiate corrections.
Maintenance performance requirements are specific and clear.
The Achilles heel of electric power generation has been the inabili-
ty of operators to exercise discipline to establish operating limits. An
organization practicing RCM knows those limits (Fig. 2-10). PM is at
the heart of such a program.
Preventive Maintenance (PM)

PM is an add-on in traditional maintenance programs. Fix
things firstPM can follow. Managers and staff talk about PM impor-
tance, but many PM programs are low on the resource scale. This con-
dition reflects the fact that life-cycle maintenance (LCM) lacks the sim-
plicity and appeal of CNM. The LCM approach uses all maintenance
44
Maintenance
performance information to schedule inspections and replacements,

during which CDM is discovered and performed. The primary differ-
ence between CNM and LCM is the level of focus, organization, and
effectiveness. LCM requires a higher degree of equipment knowledge,
monitoring, and scheduling than a simple CNM program.
As companies stated objective is to continue to operate aging facil-
ities, useful equipment life will eventually be exceeded by operation.
Knowing how equipment wears out, and the techniques to restore it,
has become highly valuable. In many cases, projected facility lifetimes
have elapsed. Major componentsturbines and boilersrequire major
maintenance on shorter intervals than original facility life spans.
Maintain-design performance objectivesheat rate, cooling, and
generating capacity-deteriorate on a relatively short time scale-months
to years. Outage performancesimply PM on a larger, less frequent
scaleis a major aspect of any large generating units operations.
Traditional PM programs are based around three basic unit operat-
ing modeson-line, restricted load, and off-line. Many organizations
segregate scheduled outages from their overall PM program; they view
outage maintenance using a traditional CM paradigm. Practically,
though, only deferrable work can be scheduled for a planned outage.
Scheduled outage work, accumulated in reserve and worked to com-
pletion in priority during an outage, involves restoring fault-tolerant
equipment, instrumentation, and reserve capabilities. Some of it is time-
based restoration. Scheduled outages are essentially comprised of CDM
and TBM. In short, scheduled outages are all PM work!
Startup crews develop routine, online PMs during plant start-up
and rise-to-power testing phase. Skilled mechanics, planners, and
schedulers review vendor literature, combine this with their own expe-
rience, and develop PM activities to perform using a CMMS scheduler.
When complete, they have exhaustive lists that faithfully recreate ven-
dor recommendations. Yet, most of these vendor-based programs are
only fractionally worked. Why? Because the scope of the PM program
is so large, and credibility of the PMsespecially the performance
intervalsis questionable; once staff realizes they can extend intervals
with low risk, rigorous performance drops. Most vendor-identified,
TBM intervals are grossly conservative anyway, because vendors cover
45
every application with one recommendation, unaware of actual service

conditions for the equipment they supply. Service conditions largely
determine the appropriate intervals for performance. They include the
operating environmenttemperature, cleanliness, moisture, intervals,
loadingand compliance with operating guidance.
Vendors do their best to identify necessary equipment maintenance.
Some PMs can be stretched with little risk, some must be done reli-
giously. Maybe not on the vendors exact recommended interval, but on
some service interval.
This illustrates a great opportunity for many plants. By reviewing
and eliminating low value or no-value PMs (those that offer low or even
negative benefit-to-cost ratios), the value of average PMs can be substan-
tially raised. This adds to PM credibility and supports performance,
because as PM credibility is achieved, performance barriers drop.
PM models
A PM is any scheduled preventative task intended to reduce the
probability of failure. Key ideas are:
scheduled
intended failure prevention
effectiveness
A PM can be scheduled by a computer, a repetitive round, the

human memory, or another method, but it has a time stamp that triggers
PM performance. Whether the PM is effective or not is another matter.
Organizations have differing levels of PM feedback measurement.
Some systems have no PM performance measurements at all, in other
cases, when measured and presented, the results diverge from those
intended.
Some PMs are conservative to a great degree. They have no real
value, rarely get done and have no operations impact. Other PMs have
impact but dont get done consistently. Some PMs are actually detri-
mental to equipment conditioning but get done anyway! Not surpris-
ingly, programs with a high incidence of ineffective PM tasks have inef-
fective feedback. Since few companies have a R engineering specialty,
such deficiencies are not surprising.
46
Maintenance
PMs can be done on calendar, clock or many other bases. For

example, time limits for reactor refueling and boiler outages are regula-
tory and fuel depletion time-based. These outages occur on nominal 18
month intervals. They were extended from a 12 month interval that was
common more than 10 years ago. Equipment run-time provides anoth-
er common time measure. This is suitable for continuously run motors.
Processing facilities often use production age indicators. Tonnage (coal
belts, dumpers, crushers, feeders, mills), total air moved (compressors),
or integrated flow (pumps) can provide suitable age measures.
Demand equipment, like medium voltage breakers, see most of their
aging during operation, and so operation cycles are a suitable measure.
Manufacturers typically set the parameter(s) most suitable for equip-
ment age measurement, and they often provide suitable age-measuring
instrumentation.
Some measures require an integrator. Coal tonnage through
dumpers, feeders, or mills, resin regenerations, and breaker trips are
examples.
Aging parameters are so important, when known they should be
explicitly identified. Engineers may imply an aging parameter but the
operator, mechanic, and engineer in the field need explicit aging param-
eters identified for all equipment. For PM timing, aging parameters
must be explicit (Fig. 2-11).
Figure 2-11: Time Parameters
47
Sometimes regulations establish inspection standards with the force

of law. Such inspections must be performed as prescribed. For simplic-
ity and convenience, they usually follow a calendar interval. Fire inspec-
tions, when mandated by laws or insurance agreements (which may be
endorsed by law) are done annually. Many inspections have been
grouped with implicit or explicit activity (such as reactor refueling or
boiler outages) to assure the activity is performed.
In an ideal world, all time-based PMs would be specified to catch
equipment at significant aging marks. Regulated, time-based inspec-
tions can lead to exceptionally conservative PMs when viewed using the
equipments natural aging parameter. For example, many in-service
inspections for nuclear power plant valves prescribe a quarterly test to
assure function. For many of the valves in these programs, the test will be
the sole operation of the valve during the quarter. If the test is to detect in-
service aging, such a calendar PM is simply too frequent. Most of these
valves could be tested successfully on much longer intervals.
Time bases fall into several natural categories based on interval
(short, intermediate, and long) enabling us to select the appropriate
method to perform the PM activity (Table 2-1).
PM perspectives
Organizations view PM performance differently. A PM activity
issued may be considered as good as complete at some facilities. Others
treat work complete more formally, allowing equipment interpreta-
tion based upon their last performance.
Other organizations are PM intense, performing every vendor-rec-
ommended task. This approach initiates effective monitoring and time-
based PMs processes but can also break down if operators or the craft
discover they can skip task intervals with little or no failure conse-
quence.
When equipment doesnt fail, people tend to continue with an
existing program, even though it is over-conservative and performed
too often. The only way to find out what the equipment can support (in
terms of lifetime and PM replacement intervals), is to perform age
exploration.
And craft workers dont uniformly perform all PMs to completion.
48
Maintenance
Short (hours to weeks)

Performance checks
Area checks
Equipment checks
Alarm checks
Operator rounds
Medium (weeks to months): On-line, PdM

Lubrication level checks
Cleaning
Standby equipment tests
Performance tests
Alarm checks
Filter replacements
Long (years): Intrusive, PdM

Planned outages
Large equipment overhauls
Main control cals
Trip tests
Lubrication replacements
Relay cals
Table 2-1: Example of Time Base Intervals
If essential PMs slip and failures occur, then program credibility is cast
into doubt and everyones effectiveness is diminished. The systematic
clean-up of casual PM programs is a significant first step on the path
towards effective RCM implementation.
In reality, programs based on manufacturing recommendations can
49
typically extend task intervals greatly with little risk of failure conse-
quences because of conservative vendor recommendations.
So, whats best? Craft and operators need to critique and adjust rec-
ommended PM task intervals. Craft feedback on intervals is required
for the dual purpose of finding the best intervals and maintaining com-
mitment to actual task-monitoring performance. Craft worker feedback
is also essential to the CMMS (and to engineering) concerning how well
parts perform, in service, to continually manage and reduce parts costs.
(In ARCM, this is a formal, continuous process.) Fostering close opera-
tions-maintenance ties, whether by intent or accident, yields more effec-
tive PM programs.
Operators, in fact, provide first-level monitoring in plant PM sys-
tems. Like maintenance PMs, operators rounds (routinely scheduled
checks that monitor broad areas and systems) should be based on value.
Traditional rounds put operators into the plant on a non-specific, just-
in-case basis, but rounds can be based on the frequency and risk of
failures. Operators may extend certain rounds with no consequence but
they bear the responsibility to support their decision. For equipment
that requires no action until an alarm goes off, a monitoring and main-
tenance strategy must be based upon that. Actively recruiting operators
to develop, review, and turn rounds is a continuous, high value
process.
Support the craft workers doing what they know needs to be done
by means of a task list. Once the PM task list is developed, work
processes determine how much gets done. Some organizations have a
catch-as-catch-can approach. Others have a work-all approach. Some
leave work scope to the discretion of the workers. Others try to work
equipment that is available. Few systematically measure the degree to
which they adhere to and complete their plan.
In the absence of a measurable plan, theres reason to question
maintenance effectiveness. Good programs are carried forward by
knowledgeable and committed craft workers. Workers still lack infor-
mation that points in the direction of improvementunaware of the
degree to which they are dependent upon the collective memory of the
workforce to accomplish PM work. In an environment with turnover
their success is diminished.
50
Maintenance
Power plant outages are planned and executed to restore produc-

tion capability and to prepare for the next sustained operating period.
How a plant identifies and performs outage work tells us something
about their general work perspective:
Some plants will work all outage work to completion

Most have very aggressive targets but partial completion
Many traditional plants dont recognize how an outage is equiva-
lent to any other PM period
Most plants are much more aggressive with their outage
workscopes and management than they are with their ongoing
PM programs
Many fail to see opportunities as they occur. With proper work
planning and coordination, upwards of 40% of outage hours can
be worked online
When the plant is down, restoring production is important. But

preventing the plant from going down in the first place is achievable,
through highly effective levels of work, understanding, skills, coordina-
tion, and implementation. When PM is obscured by more visible activ-
ity, it can lead to a PM is not real work mentality. The craft often
avoid PMs because they arent organizationally focused, and prefer
overhauls that give them more opportunity to perform
disassemble/reassemble work management may even let them select.
PM delivery is taken for granted by many organizations even
though failure records indicate that existing PM programs are not being
followed. Documented performance, with periodic checking, is an
effective tool to assure PM performance is real. Peer self-checks and
periodic manager or operations checks are other ways to monitor PM
completions. An outside audit can help establish the delivery credibili-
ty of a PM program.
When failures occur in a well-designed program, investigation is in
order. Companies that review and assess their monitoring programs
historically experience few random, surprise failures. When PM is treat-
ed with discipline it will add confidence to the program that stands
behind the PM.
51
Identify the essential features of any PM program as a productive

first step. In fact, if everyone can agree on the essential elements, it will
save a lot of wasted effort. After years of consideration, Ive identified
some essential PM program features:
Activity list: a formally maintained list of PM activity the organi-

zation is committed to perform
Scheduling tool: a method that delivers routine PM performance
and assures priority for PM even when crises occur
Selection and issuing methods: ways to establish the scope and
routine issues that PMs must address for peak performance
Completion reporting: a feedback system to report completion
Measurement of completion rate: a system for management to
measure PM program health
Assessment: periodic review of the program for effectiveness
Standards: there are guidelines for performance, deferrals where
appropriate, grace periods, and so forth
I characterize these elements as memory, execution, and discipline.

Basic craft skill is necessary to perform the workthats assumed. But
to establish an efficient, basic program, you must have:
focus
group work organization and delivery capacity
effective tasks
Organizations that have only one or two of these elements suffer

fundamental flaws in their PM programs. Programs with major gaps are
not viable PM programs.
Before developing tasks, you should determine whether or not you
have a viable PM program. A little effort to create a viable program
before too much work is put on the list can help the organization grow
into PM more easily.
It is common to lack a PM delivery process. (i.e., High value
PMs are identified, but there are no means to assure that work is per-
formed consistently). This may be part of a larger problem, such as
52
Maintenance
overall work prioritization and scheduling. In any event, PMs are

deferred to the lowest priority level and get worked as fill-in, or catch-
as-catch-can. High-value PMs can truly be painful to track as premature
failures reoccur. These are avoidable with a credible, consistently per-
formed PM program.
How are typical PM programs operated? A PM sheet gets kicked
out of a computer CMMS system and goes to the shop, where a main-
tenance foreman personally prioritizes and schedules work based upon
his personal experience, workload, pre-assigned priorities, processes,
and organization. Program reviews indicate that many PMs get com-
pleted randomly under this systemtheyre performed with high vari-
ance.
Another weakness of traditional PM programs is found in typical
completion performance reporting and trending. Any program with
15% performanceand acceptable resultsis in need of some tuning!
Whether such numbers indicate low interest on the part of managers, or
a complete lack of understanding about the programs value, PM is a
complex subject that requires sophisticated understanding of equip-
ment, failure, statistics, people, and processes. Its complex scheduling
starts where project management ends.
PM triggers. PM, like any a repetitive activity, must have an initia-
tora time trigger, condition identifier, and condition-based action. It
could be as simple as a clock, or as smart as a specific condition. It could
be something in between. (Fig. 2-12) Seasons, weather, age on a com-
ponent, or a sixth sense that some mechanics seem to have all provide
clocks and initiating events. An individual often must work for years to
learn the subtle clues the operating plant offers about its condition and
needs. The more concern individuals have for reading these clues, and
the more plant knowledge and experience they have, the more success
they enjoy. Training improves anyones ability to read equipment.
There are seasonal, conditional, and equipment operating cycle trig-
gers. Many do not reside in the plant CMMS, either because they wont
fit or because time parameters preclude their use. Other times, triggers
are just habit. Understanding what causes actions to take place, is
important if you want to change a maintenance system. If, for example,
people work off personal scratch sheets rather than the organizations
53
CMMS printouts, the value of the

CMMS as a tool is diminished.
Current CMMS installations
provide the capability to schedule
PMs from flexible time parameters.
Most CMMSs handle common time
equivalentsprocess tonnage,
machine hours, or operations cycles.
Some can read plant DCS systems
automatically. The principlethe
time triggeris the same with each.
A known parameter, correlated by
experience with an aging character-
Figure 2-12: PM Performance Triggers istic, initiates a work trigger based
on elapsed time; monitoring with specific limits identifies an out-of-

spec condition and triggers work. The more work thats identified on
the CMMS, and the more time thats spent to assess it, the more effec-
tive a maintenance program will be.
Scheduled outages can pose a PM dilemma of sorts. Outages com-
prise large amounts of deferrable PM work on degraded, degrading, or
failing equipment. Performance loss can be tolerated to operate up to
the scheduled outage. However, at some point the benefit of deferral is
overruled by unscheduled outage risk. Also: Outages made up of major,
planned work (such as turbine overhauls, boiler cleanings, and reactor
refueling) that involve very large workscopes and personnel resources.
Not formulating an exact schedule raises the cost to carry these people
and resources. Finally, major work such as turbine overhauls provides
the tentan outage under which we can park a great deal of other
necessary work.
Organizations committed to optimum maintenance performance
can use outages to introduce PM techniques to the plant. Scheduled
outages are the most fundamental PM interval and activity. They repre-
sent opportunity.
The challenge of life-cycle vision

Theres a fine line between maintenance and CM, and it has to do
54
Maintenance
with how people are dispatched to perform their day-to-day work.

Scheduled workin which most of the time is scheduledminimizes
reactive responses. Such structured, planned work improves mainte-
nance effectiveness and reduces costs. LCM (in which the work is struc-
tured around available equipment windows) means that equipment is
more nearly in a state of continuous monitoring. This supports routine,
continuous operations. Organizations adopting LCM must understand
that a key aspect is the implementation of an effective maintenance
measurement plan.
Informality and chance. Planning, implementation, and effective-
ness measures that are vague characterize an informal maintenance pro-
gram. Informality in a large, complex facility will not support a cohesive
plan. An LCM approach to maintenance adds consistency and structure
to all maintenance performance indicators in a plant. Routine, habitual
work practices lead to consistent maintenance. An organization with
millions or billions invested in plant assets cannot afford random main-
tenanceparticularly when there are tools that can greatly increase effec-
tive task selection and maintenance performance.
Traditionally, plant PMs came from vendor manuals; others were
initiated in response to failures that werent adequately understood. Its
also common to see PMs initiated, suddenly, where equipment had been
operated for 20 years or more with no PM with reasonable success.
Such PMs cannot pass effectiveness actionable tests. Failure-prevent-
ed applicability reviews and expert cost/benefit assessments are miss-
ingassessments that need to be supplied by maintenance strategy spe-
cialists. Theres no obvious connection between work performed and
failure prevented in such cases. Workers and plant staff whose commit-
ment to PM programs is lukewarm, take risks when they defer or miss
PMs, betting that they can get away with no activity. And because fre-
quently they can, the program loses credibility.
Credible PMsand workforce PM program support and compli-
ancearise from explicit failure-prevented associations, the more
explicit the better. PM priority then naturally falls on an equal basis with
other work. The appropriate plant personnel must perform PMs. Non-
specific tasks and checks should be performed by operations. Intrusive,
technical, or direct work needs maintenance performers and their support.
55
Vague PM scopes are of low value. Because theyre indefinite,

theyre time intensive; workers fumble through what operators could
do better on their rounds. Non-specific scopes cannot be assembled
into credible work packages. One way to identify ineffective pro-
grams is their lack of distinct, actionable PM tasks assembled into
performance packages. Many PMs involve quick, easily performed
tasks repeated many times. When theyre developed, planned, and
blocked into work tasks, they optimize work time and increase orga-
nizational PM focus.
In some organizations, individuals select their work on a daily basis,
even when theres a work backlog. Effectiveness depends upon the pri-
ority that they assign to the work. When they pick work they like to do,
where does this leave organizational priorities? A significant part of PM
performance involves the discipline to select, issue, document, and
report. Scheduling discipline comes from aligning people and work.
Cleanliness, operations professionalism, and cost-management are
other indicators of discipline. Disciplined organizations have focus.
Their goals are clear and they work to them.
PMs functional elements

Time-based: clocks. CMMS issue WOs based on computer clock
time, or events such as an outage or even memory. Scheduling software
does no workpeople in an organization do work. Once a PM WO is
issued, how does work follow?
Intrusive work ranges from low skill to high skill. The more intru-
sive the work, typically, the more skill required, and the more special-
ized the personnel. Sometimes only an expert can interpret data. It may
take a mechanical engineer to evaluate a yielding failure or a civil engi-
neer to examine a railroad loop fill. For non-intrusive tasks, we assume
that people have skills to perform simple tasks, or they can learn them
quickly and effectively on their own. In each case, training introduces
consistency into task performance. Consistency in performance leads to
consistent O&M.
VM, oil analysis, and ultrasonic diagnostics all require training and
skill to enable workers to interpret equipment conditions. However,
most people in the plant are capable of learning about reading gauge
56
Maintenance
glasses, oil slingers, and oil contamination; recognizing smells (such

as burning coal); finding steam and water leaks; sensing vibration
through foundations, and picking up on normal or abnormal sounds.
Working in a production environment, one must call equipment
failure causes like an umpire calling an out at first baseand then get
back into production quickly. Learning should be a program attribute.
Change must be encouraged. Continuing on simple pathschanging
perfectly clean instrument air filters, rebuilding valves, replacing like-
new partsinstead of adjusting programs to reflect real-world equip-
ment and plant needs indicate static programs. When a production
environment accepts it, its not structured for change. The systematic
extension of replacement intervals to discover how long equipment and
materials last is a high value attribute of a living program that can be
missing in traditional facilities.
Operational based: surveillance. One reason that fossil units expe-
rience higher equipment failure rates than nuclear plants is the absence
of surveillance programs (SP) that are structured into nuclear.
Nuclear plants have extensive specifications imposed by their license
and SP demonstrate compliance. There are more preliminary technical
identifications of deterioration before the equipment proceeds to func-
tional failure. With few fossil equivalents, periodic alarms or protective
devices and standby systems run higher risks of failure. This in turn
increases equipment failures.
For any plant, a SP is defense plan against random failure that cap-
tures the operators value. The inherent design of fossil plants is more
nearly optimized, as well. Originated in the first quarter of the twenti-
eth century, their design has been evolving for nearly 100 years. The
more a design advances:
the more it supports NSM performance

the less its overall function is impacted by component failures
Fossil units are relatively fault-tolerantand more equipment is

taken to failure limits. They come down when real parts failnot
abstract specifications. On top of regulatory constraints, nuclear plants
are younger than fossilsboth of which mean that its difficult to
57
make significant design improvements. Nuclear units must operate

within specifications, and while failed specs are less interesting than
broken components, they cost less to correct.
Operate-to-failure (OTF), preplanned failure, and no scheduled
maintenance (NSM). OTF is widely misunderstood. NSM is a more
apt term. Nolan and Heap never use OTF in R-centered maintenance.
Rather, they use NSM.
Refered to as Operate-to-failure, OTF poses a barrier to operator
maintenance cooperation. Most workers dont want equipment to fail
but they like to work maintenance. They need to be backed off inher-
ently reliable equipment that can be trusted to self-identify failure dur-
ing monitoring with low risk. Where a failure has no functional impact
and maintenance can be readily scheduled after failure, OTF strategy is
cost-effective. This is what RCM is all about-taking advantage of the fact
that most equipment doesnt benefit from scheduled maintenance.
Where redundancies and risk management are absent, however,
OTF isnt effective. The organizations risk tolerancethe period dur-
ing which the redundant train, alarms, or other equipment can be
impaired prior to work performedinfluences OTF effectiveness. If
operations cant wait for a few days to schedule work on backup equip-
ment thats down, then savings will not result. OTF does not work for
nervous Nellies.
Design process evolution calls for equipment failure to become less
important. Advanced designs tend towards fail-safe by automatically
removing equipment from service or initiating self-shutdown rather
than creating severe events. Advanced designs demand abuse to force
failures. Most users give up well before the equipment becomes a haz-
ard. Advanced designs include equipment in production for more than
20 years that has been through three design iterations. After 50 years,
they approach their design limits. Few survive in their fundamental
form for more than 100.
Measurement. CMMSs lack the capacity to differentiate a func-
tional failure from other types of failure. Functional failures, as a prac-
tical matter, show up in operators logs as shutdown initiating events
fires, accidents, and other facility compromises. They influence long
term plant availability.
58
Maintenance
A nuclear plants failure interpretation by law is much more con-

servative than its fossil counterpart. As the definition of functional fail-
ures becomes more common, improved work-order measurement can
be expected and we can develop better ways to compare apples with
apples.
Availability is a telling measure. According to NERC submittals,
fossil units are actually more available than their nuclear counterparts.
(Shorter scheduled outage durations appear to be the cause.)
Overhaul. Overhauls are grouped activities in which intensive,
intrusive work must be performed, such as disassembly of a turbine to
clean blading. In RCM lingo, an overhaul as a PM doesnt exist. Rather,
its a group of small tasks, such as:
blade cleaning
root tip crack inspections
rotor inspections
blade erosion inspection
gasket replacement
rotor re-balancing
bore inspections
lube oil purification
cooler inspections
instrumentation bypass line inspections
casing bypass flow erosion inspection
generator winding examination
balancing
stop valves inspection
stem blush removable
weld repair
Many of these activities are either time-based or established failures

that require periodic checks and restoration. Risk is managed by a com-
bination of instrumented detection (like VM) supported by periodic
internal inspection.
Because internal examination work is so expensive, its usually cost-
effective to do it all once a machine is apart. Many PM tasks are
59
blocked around one intrusive activity, so that the overhaul-to-overhaul

life of large machines is prolonged. (Fig. 2-2)
In RCM, we restore components that are aging. The philosophy is
very useful for examining overhauls, since it may lead to eliminating
activities whose failure is determined to be low-risk or never possible.
Large activity PMs should be reduced to individual tasks so that each
added-value task can be separately assessed. A healthy, question every-
thing attitude must be encouraged from the top down.
What measurement level is appropriate? Workers often dont see
value in documenting work. They rarely see anyone use documented
results. Accountants are the only ones who routinely assess completed
work. Traditionally, maintenance organizations do maintenance
rather than avoid or reduce it, so they dont recognize measurement
value to reduce work and theyre insensitive to cost.
Traditional maintenance organizations with in-house horizons
havent benefited from out-of-company maintenance observation. Self-
training marks the traditional maintenance organization. They can even
view maintenance support staffplanners, schedulers, and engineers
as outside the maintenance process and may harbor skepticism of
new technologies and analysis. Theyre eager to work but planning,
preparing, and training suffer from the preference to do work.
However, doing work requires tools-measurement, documentation,
and analysis are also valuable tools.
Maintenance is a cost-plus proposition. Traditional maintenance
rewards are proportional to the volume of maintenance produced
not value. Financial incentives to increase maintenance productivity
dont make sense to those who suffer consequent income loss.
Maintenance measures have historically been cost-oriented, but
theyre measured predominantly at the high endthe unit, station, and
corporation. Major process areasprocurement, outages, stock, man-
hoursoften lack performance measures. System level measurements
can be lacking, as many older CMMS systems cannot support system
level cost allocations. Yet, detailed system measures would allow diag-
nostic trend reviews of the maintenance processes for high value
improvements.
Traditional plant managers focus on expenses. They never directly
60
Maintenance
received benefits of increased generation revenues in the regulated envi-

ronment. Completing additional maintenance work could bust budgets,
even though increased revenues could result. Even today, many man-
agers lack cost-management skills or CMMS systems to be able to track
such fine points to hold the line on costs.
Measures are valuable when useful information results and
improvement areas are identified. CMMS offers greater measurement
capabilities but information is only as good as data received. Achieving
consistent data inputhour reporting, costs, problem descriptions and
work doneis the essential first step. Several alternatives may be avail-
able to provide immediately accessible benefits.
E MWRs. Emergency maintenance WOs (E MWRs) correlate
to equipment failure rate. Since true functional failure often generates
an emergencythe Ework-order, tracking by E MWRs is a sim-
ple, functional failure rate measure.
Overtime. Like E MWRs, overtime results from functional failure
events. Though less definitive than E MWRs, overtime provides
another useful way to measure functional failure rates.
Failures. Most utilities that write a home-grown CMMS begin
the measurement process by assessing the types and quality of informa-
tion collected. Once measurement is established, organizations usually
find that:
there are many measurement capabilities they have that they

dont need (the trivial many)
there are a few measurements they need and dont have (the
critical few)
data quality or accuracy may be low (they need training or sim-
pler methods)
they need measures of both process simplicity or quality, or both
An organization that recognizes the need to measure has made a

giant step towards performance improvement. At the time it recog-
nizes that its measurement system needs improvement, its just a mat-
ter of time before broader measurement needs and improvements
are considered.
61
Maintenance rule. The NRCs maintenance rule [Title 10 Code of

Federal Regulations (CFR) 50.56] can be meaningfully interpreted in
RCM.
Three fundamental attributes of maintenancefixed time mainte-
nance, CNM, and operate-to-failuremust be followed for systems,
structures, and components that influence safety and are not inherent-
ly reliable. The rule requires monitoring for things that dont meet
organizational goals for availability and maintenance-preventable func-
tional failures (MPFFs).
The rule is workable, if ungainly. Areas of confusion involve defini-
tions of run-to-failure and have no PM program. If we dont have
a traditional PM maintenance program, are we running things to fail-
ure? And, if we do, must it be entirely and completely documented, and
with documented performance?
An example: A large, single-unit boiling water reactor (BWR) has
between 50,000 and 100,000 coded component identification codes
(CICs), for which there are between 5,000 and 10,000 PMs. Though
many PMs address more than one CIC tag number, do we have a PM
for everything? No. Do we come close? Hardly! Well, then do we have
a PM for every essential component? No. All environmentally qualified
(EQ) components? No. Many items lack a formal PM program; are we
running them to failure? Emphatically, no! First, remember that a fail-
ure must be a functional failure. With so much redundant equipment
installed, and with many minor and inconsequential failure modes, the
vast majority of reported failures are proximate, not functional. If there
are 100,000 CICs with 3 failure modes each, we should have up to
300,000 PM tasksbut there arent. Why not?
Operator monitoring covers about 80% of the work. Operator
rounds, surveillance testing, and other non-specific monitoring, such as
area checks, will identify the vast majority of failures. If we exceed tar-
get levels for unavailability or MPFFs, the maintenance rule says we
must initiate monitoring and corrective actionwhats known as set-
ting goals. If were already monitoring, we merely need to document
and in other ways demonstrate compliance. This means that corrective
action may entail the same monitoring program we had beforeif we
had an excellent programto ensure that we never experience either an
62
Maintenance
MPFF or availability below our target.

The system level challenge has been how to measure unavailability.
In the absence of simple unavailability measures, clearance tag-outs
or log entries are used to declare equipment to be inoperable and so
track unavailability. While these processes are simple and conservative,
they overstate unavailability, thereby artificially forcing equipment into an
a (1) category. Yet, probably the single greatest maintenance-rule ben-
efit has been the requirement to track system level performance. Prior to
the maintenance rule, few plants did this. System level availability and
failure analysis would greatly benefit many non-nuclear units.
The Hawthorne principal (based on manufacturing performance
studies done at Western Electrics Hawthorne plant in the 1930s) states
that managements visible interest in performance improves it. This is
exactly what happens after performance measures are instituted and
tracked at a production facility. Its a powerful case for measurement.
Costs
Operation and maintenance practices based on reactive (not proac-
tive) philosophies are very costly. Direct costs include rework, increased
scheduling, and increased risk. Risk ultimately translates into operat-
ing events that impact equipment and employees. High risk organiza-
tions mean higher cost operations, just as speeding drivers mean higher
costs for insurers.
Analyzing failures and costs confirms this intuitive knowledge
maintenance performance correlates with insurance claim losses.
Insurers periodically inspect client facilities to assess their insurance risk
and help clients better manage that risk.
There are wide variations in electricity cost, in part because some
producers are more expensive, based on their plant outage profiles,
while for others its routine maintenance practices. If, in fact, competi-
tion re-invigorates the generating industry, it will happen because com-
panies will be forced to re-evaluate and improve processes that have
been slow to change.
63
Case Examples
SBAC
A 7000-acfm soot-blowing air compressor (SBAC) experienced pre-
mature filter pluggage, filter element tear-out, and prolonged operation
with unfiltered intake air. The rotary compressor uses five high speed
compressor stages, each driven off a common bull gear. The compressor
required overhaul two years after the previous performanceabout two
years short of its early projected overhaul date but much shorter than the
previous stage overhauls. Installed as one out of three compressorstwo
in continuous servicethis unit achieved more than five years service
until its first overhaul. The nominal life for the compressors was placed at
four years, aside from the new operating period when all units ran
almost eight years. On these staged compressors, the high-velocity fifth
stage ordinarily wore out first, establishing the overhaul need.
The diminished compressor life (between overhauls) was two years.
At an overhaul cost of between $250,000 and $300,000, the shorter life
cost an additional $150,000 (around $75,000 annualized). The missed
PMs cost three hours every quarteror $1,000annualized, including
filters. Cost benefit is at least 75-to-1 based on maintenance costs. Such
a PM cannot be missed without adding substantial maintenance costs.
While down, boiler convection passage plugging increased. Because
operating staff had to be pulled aside for the compressor overhaula two-
man, one-month duration job (with contracted help), normal schedules
were interrupted. Because overtime was required in this union shop, the
whole plant was authorized overtime, further driving up costs.
During operations, big-ticket failures mean major unscheduled
events. Non-routine, non-turbine or boiler costs can be tracked by fre-
quency and cost category. Major unpredictable equipment failures can
cost up to hundreds of thousands of dollars. Such failures are an obvi-
ous target for reduction! They can be identified, counted, costed (annu-
alized), understood-by-cause(s), and corrected. Taking on unplanned
but statistically predictable big ticket events in a systematic manner
results in gradual improvement in equipment online cost performance.
Ultimately, all equipment wears between service intervals.
Achieving maximum predictable service intervals is a goal of a PM pro-
64
Maintenance
gram. Big-compressor soot-bacs are monitored over time.

Performance monitoring, combined with in-service PM between over-
hauls, should identify the projected overhaul period well in advance
of actual deterioration.
Compressor performance drops as the stage blading wears until a
combination of low flows at load with increased surging identifies the
need for overhaul. The overhaul consists primarily of removal of the
bull wheel (actually, the pinion), stage inspections for dimensional bear-
ing wear and visual balding wear, discharge plenum cleaning and
inspection, and reassembly. Normally, no bearings or gears are
replaced. Only the fifth, finishing-stage compressor blading is reworked
due to high wear from moisture and particulate erosion. Normal over-
haul focus is on the fifth stage and normal part reorders anticipate
replacement of these parts.
Premature turbine blade failure

A forced turbine outage due to blade deposits occurred in a unit for
which nominal turbine overhaul interval is five years. Overhaul at four
years was required due to derating from the lowered stage efficiency
and literal valve-wide-open generation loss due to plateout deposits. To
schedule an early overhaul, system dispatchers bought replacement
power, because the base loaded units outage was in the near-term unit
outage forecast and other unit outages couldnt be rescheduled to cover
the outage period.
The turbine overhaul took around two monthsannualized, 12
days on a 5-year interval, or 15 days at four. The cost of replacement
power was the difference between the marginal production cost at that
date and the scheduled replacement power cost (around $35), coming
in at (the ballpark) price of $15-20/MWhe; 350 MWe (rounded) for two
months at a price differential of $20 (the difference between $35
and$15), is around $10 million. This outage would have occurred any-
wayjust two years later, as a scheduled outage. At that time, the
replacement generation would have been provided internally or with
lower-cost, long term contract scheduled power. This wholesale power
rate averages lower, thereby reducing its cost. The differential cost of
replacement power, planned and scheduled, would have probably been
65
in the $5-10 range rather than $20/Mwhe. Factoring internal genera-

tion, it could have been as low as $2-3.
Considering present value (PV) calculations, from an accounting
perspective, the benefit of meeting scheduled outage periods includes
the PV of the deferrable expense. A $5 million expense, deferred two
years, costs $3.5 million (a finance charge of $1.5 million). In a finan-
cially competitive environment, this is real money. While this sort of cal-
culation is new to traditionalists, its a day-to-day calculation for IPPs
and co-generators. Its something the rest of us need to learn.
Generator retaining ring

Consider a generator retaining ring failure on a 350 MWe unit, that
cost around $10 million in direct repair expense and another $25 mil-
lion in revenue and replacement power expense. The event had pre-
cursors. Two separate vibration amplitude step increases had occurred
in an otherwise stable trend. The first preceded the event by about 10
days. The second occurred the morning of the final event. The second
offsetranging upwards of 10 milsso concerned operations that they
called in a VM crew to check the instrumentation that morning.
Checking the generator bearings-using hand held instrumentsthe ring
final failure developed.
The event was complex but had precursors and warnings that were
ambiguous and/or missed. The cost of crack detection and repair is so
high theres no real effective deterrent, except to run-to-failure. The key
to operations success in this area is never to initiate stress cracking. One
way to do that is to exclude moisturethe most common controllable
cause of stress cracking. This is the generator manufacturers strategy.
OOS, with the generator disassembled, moisture is difficult to control,
but preventive measures must be taken. Tented heaters and/or dehu-
midifiers can maintain non-condensing environments.
This particular event occurred in the West, and so humidity was an
improbable cause for the retaining ring failure. Hydrogen coolers, on
the other hand, are known to leak because during shutdowns, the
hydrogen cooling water in service with low hydrogen pressure created
potential for leakage into the hydrogen system. This, in turn, would pro-
vide the opportunity for internal condensation and stress-corrosion
66
Maintenance
cracking in the generator. By use of periodic hydrogen moisture sam-

pling, original equipment manufacturers (OEM) intend that generator
internalsincluding rotor and retaining ringsare kept above the dew
point. When hydrogen is monitored for moisture, leaks into the hydro-
gen system can be detected. Bleeding and feeding dry hydrogen can
avoid condensation even when cooler leaks are present.
In addition, liquid detectors identify and alarm for liquid that accu-
mulates in drain areas at the bottom of the generator frameareas with
isolation, vent, and drain capability. Any water or condensation accu-
mulation can be monitored, trended, blown down, and inspected. The
float-operated switches and alarms can be checked and replaced.
In this event, a third party performed root-cause investigations
because a large insurance settlement was at stake. It was rumored that
the generator in question (as well as its twin) had had cooler leaks.
Whether or not the plant had consistently performed recommended
monitoring was a matter of speculation, based on its overall PM com-
pletion rate. Taken as a whole, its PM program was weak, with overall
completion rates of scheduled PM work in the 5-10% range.
The presence of moisture in generators has a known, deleterious
effect on many components. Besides stress-corrosion cracking, insula-
tion windings can fail to ground. Generators have been in production
for more than 100 years and designs are highly refined. To remove great
amounts of heat internally generated, manufacturers have evolved
hydrogen gas cooling and water inter-coolers. Water coolers always
present the potential to introduce moisture to a gas, usually during tran-
sient or shutdown operations.
Vendor manuals warn operators to monitor moisture internal to the
generator at several levels:
H2 leakage monitors (offline)

drain alarm switches (alarms)
drain monitoring and blowdown
H2 moisture monitors online (optional)
This illustrates a typical, integrated equipment failure defense strat-

egydirect operator monitoring supported by condition-based mainte-
67
nance (CBM) upon failure detection. Essential monitoring equip-

mentinstrumentation and alarmsis maintained by an I&C group.
The failure-avoidance strategy relies upon monitoring and alarm equip-
ment availability, backed up by direct operator draining of blow-down
lines that require only periodic performance checks.
But moisture potential is everywhere. Twenty years of plant experi-
ence tells me to look for water where you dont want it and least expect
it because thats where its certain to turn up. This applies to instrument
air systems, air purge systems, or any kind of cooler with water on one
sidesensing lines, damper lines, Helium lines (gas reactors only), and
other gas lines. Every water cooler is certain to leak so make sure water
quality is high and then monitor appropriately for leakage. Fix any iden-
tified leaks as early as possible. Avoid direct dependence upon opera-
tor/human intervention to directly control failure events. Designs
should support operator activity by minimizing direct intervention
requirements.
In many cases, like this one, maintenance didnt happen and opera-
tors learned to tolerate the deteriorated condition. Over time, an expen-
sive repair would ultimately occur as secondary problems developed.
Addressing degradation before it becomes excessive is key to a solid
maintenance programand also the weak link in too many of them.
Operators typically do a credible job of identifying deteriorated and
unserviceable equipment but follow-up is where success or failure is
determined.
The problem is rarely that the installed instrumentation is inade-
quate. The problem is usually the absence of a maintenance strategy to
successfully operate and maintain essential equipment. Its a problem
when an operating philosophy includes reactive maintenancenot fix-
ing things until forced to do so. This is a problem that instruments cant
address and may even make worse.
This example illustrates the importance of critical instruments and
of CNM. I use the term essential instrumentsin essence, the instru-
ments that provide identification and protection from unacceptable
failures that must be avoided. Such instruments and their associated
monitoring programs must be maintained.
68
Maintenance
Figure 2-13: Coal Mill Fire Supression
Coal belt fire

A coal gallery fire occurred at a coal facilitys inclined yard belt.
Coal spillage, and a chronic station history of coal fires, combined with
a defective fire protection system causing a major loss. Fire alarms and
the deluge system failed. Insurer loss estimates (and rates) were based
on fire alarms and equipment availability. Inclined coal belts and the
right environmental conditions (fire doors left open, for example), gen-
erate an excellent draft. In this case, within an hour the belt gallery had
burned and collapsed.
Fire, casualty, and industrial insurers decided that coverage was the
thorny issue. Loss-risk assessments are based on actuarial assumptions,
including systems and alarms. Events in which devices are OOS fit
another loss categoryperhaps one that cant be insured.
For both companies and insurers, the best defense against cata-
strophic losses is predictable, consistent operations. A fundamentally
sound CNM program, supported by an effective CBM program, deliv-
69
ers such consistency. A great deal of fossil generation equipment has

fire protection alarms and instrumentation. These provide direct means
to manage fire risk, and should be treated with caution. (Fig. 2-13).
These fire detection alarm circuts need a planned maintenance pro-
gram. Coal mill CO monitors have great valueif they work!
70
Chapter 3
RCM Performance
Left to themselves, things will go from bad to worse.

-Corollary, Murphys Law
TRCM analysis is time-consuming, and companies must balance

time against value. Without implementation, there is no value but many
organizations have not benefited from RCM although they have done
complete, system-by-system RCM analysis. Perhaps theyre too interest-
ed in quick results. Perhaps its a lack of patienceultimately, everyone
suffers from that, too. However, there is no value in academic RCM
exercises or dismissing RCM altogether because you fail to take all the
steps.
As a practical matter, Ive rarely (probably never) had the luxury of
time to do a complete analysis. In fact, the complete RCM analysts
Ive met have never claimed that their complete analysis was final. (No
one would care [or dare] to do so, except when pressure to implement
RCM comes in major part from regulatory agencies.) My focus instead
has been implementation, as opposed to theoretical basis development.
Its most effective to pass along a maintenance optimization philosophy
71
Figure 3-1: Modern Blending Coal-fired Power Plant: Apparently simple, looks are
deceiving. This zero discharge plant ranks with the last nuclear units for complexity.
The plant is running at full load.
in the context of solving operational and maintenance problems.

Nuclear plants have extensive formal maintenance plans because
the NRC drives the need. Fossil plants have little to nothing. R rules are
the same. Nuclear plants simply have greater documentation, justifica-
tion, and traceability requirements. Each environment has needs, for
which the maintenance optimization answer is the same, as it should be.
Studies by EPRI on competing methods indicate PMO outcomes are
largely the same and depend less on specific process than upon skill and
knowledge of reviewers. Rules and regulations aside, the operating goals
of each plant type are generally the same.
Lower equipment failure rates at nuclear plants (based upon fail-
ure data at the plant level) are not due to inadequate documentation
there really are far fewer equipment functional failures. This makes failure
studies at nuclear units more difficult because the statistical population is
smaller. While fossil failure experiences at the component level dont trans-
fer directly to nuclear plants, they do offer insights. Industry-wide data-
bases can fill in for statistical data for any single unit (Fig. 3-1).
72
RCM Performance
Selection Based on Importance and Cost

Two factors determine whether a piece of equipment deserves
TBMimportance and risk.
Equipment that impacts production (due to failure) must be man-
aged. If failure has random components, there must be installed redun-
dancy to provide operational flexibility. This philosophy supports
OTFwait until it requires maintenance, then do it.
If it was worth buying and installing in the plant, every piece of
equipment deserves some maintenance. Some maintenance, in most
cases, means NSM, based upon function and CNM. An on
demand/OTF/no-scheduled-maintenance strategy is appropriate for
most installed plant equipment because one tenet of RCM is to do no
maintenance on equipment that doesnt inherently require it. A default
monitoring program captures most inherently reliable, randomly failing
equipment through non-specific operator monitoring, area rounds,
and periodic checks made in the course of doing other things.
Combined with alarms, redundancy, and trained operating staff, there is
a great reservoir of operating depth to draw from.
Every manager goes after high value opportunities, and this means
availability loss and the systems contributing to them. Generation
costsincluding lost production valuerepresent value lost when an
unplanned trip occurs. The RCM approach takes the unit apart, system
by system, according to how each system supports overall plant genera-
tion objectives. Systems affecting generation are high value systems,
particularly during forced outages, when steam supply and turbine sys-
tems determine their duration. Understanding how all work contributes
to outage duration and prevents failure is another availability benefit of
RCM analysis, however, because careful RCM analysis can trim outage
workscopes.
Consider the number of systems in any large generating planton
the low end, perhaps 40-at complex nuclear plants, it may be 100 or
more. Does this mean analyzing each and every system to get a useful
answer? This is the implied requirement with TRCM (especially at
nuclear plants). If we dont have to do everything, do we pick and
choose the systems and supporting parts we evaluate? The answer is
73
that we can select systems:
provided we retain the results in common retrievable format

commit to return regularly to continue analysis on an ongoing
basis
If you elect to go this route, you must consider integrating the

resulting programs. The death knell of many TRCM programs was the
failure to integrate the work. This requires performing LCM integra-
tion. Without this dimension, PM optimization is risky business.
Consider the mission of each unit in an owners portfolio. In the
past, we assumed generation was base loaded; today more varied mis-
sions are applied. Some plants conserve corporate assets while others,
provide peaking generation or support transmission services (voltage
support). Transmission support roles must be defined by facility owners
and be identified to plant operating staffs. Some units serve in dual
roles, others provide generation reserves. Just as phone industry dereg-
ulation has opened up new services and markets, the generating and
transmission industries have new markets, productsand missions. A
maintenance plan developed around an obsolete or wrong mission will
not effectively use resources.
Plant system units

There are several major classes of generating units. We can broadly
group these into five types:
gas-fired boilers
coal-fired boilers
CTs
hydro
nuclear
Size, vintage, fuel, cooling, architect-engineers (A-E), boiler, and

turbine OEMs differentiate the fossil units. Fossil coal plants with zero
discharge permits are the most complex. Many perform fuel receipt,
fuel processing, fuel storage, fuel movement, combustion, waste pro-
74
RCM Performance
cessing, emissions, and water treatment for plant equipment. Fossil

units can further be grouped by investment, production costs, capabil-
ity, and staffing. There are representative equipment and systems (often
from one to several major suppliers) in any category. Each unit is some-
how unique and can be treated as an individual exception.
Fossil CTs have gained market share spanning the last 20 years as
their missions have evolved-from providing fast-start backup generation
to base load combined-cycle facilities. They fill many niches today, from
base load to peaking. Complete facilities may support heat recovery
steam generators (HRSGs), which in turn supply a steam-driven tur-
bine. Such combined-cycle facilities are similar to large fossil generators
in their complexity, number of available models, and economic value.
Standardization has established common configurations. These units
often burn gas, have been brought into production quickly, and can be
modified over time to incrementally increase load capacity.
As a class, hydro spans a greater range of equipment and more ini-
tial service dates than any other category. Some of the earliest electric
generators ever sold are hydrosand many are still in production after
100 years! Age variation makes hydro unique. There are principally just
two major design applications-high and low head (or Pelton and Francis
wheel turbines, respectively). Many hydros are seasonal peakers,
although some provide pumped storage units. In the Northwest, many
see base load service. In the West, a lack of water limits hydro use to
spring runoff or peak load periods. Under these conditions, nominal
load ratings can be very misleading as many units run only a few weeks
per year.
There are two basic nuclear classes-BWRs and pressurized water
reactors (PWR). Major equipment suppliers further differentiate
PWRs. Each has unique demands and requirements and all share a
common regulatory environment. Although sizes vary, later-design base
loaded plants are industry standards.
Each plant category has many standard systems in common-design
features, aging characteristics, operational features, and other person-
ality traits. Some are suitable for a single type of service-others fill mul-
tiple roles. Benchmarked plants can be identified through NERC,
Institute of Nuclear Plant Operations (INPO), NRC, supplier, and
75
industry data-suppliers and can provide reference points for perform-

ance self-assessment.
Plant system functions

RCM maintenance analysis centers on retaining functionality, and
so focuses on system functions at a high level.
Systems are the building blocks of power plants. In a power plant
(or any large facility), they break down complex conduit, cable, and pip-
ing into discrete, understandable units. Classified by type, systems share
many similarities from application to application. (There are many types
of turbines, but turbines share many common subsystems and func-
tional requirements.) Experienced workers implicitly know all primary
functions of the major common plant systems. Its the unique, uncom-
mon, plant-specific functions that occasionally surprise even a seasoned
engineer. Engineers continually strive to enlarge the number of func-
tions a given operating design can perform to stretch economic benefits.
A-E design/owner documents identifying system functions include
process and instrumentation drawings (P&IDs), system descriptions,
training materials (where available), major modifications, procure-
ments, and engineering files. Additional documents include licenses
and virtually any maintenance or cost information that can provide
functional insight. Maintenance optimization requires that this infor-
mation be assembled into a current snapshot of system and equipment
status. For nuclear plants, training materials often provide a current,
complete overview of current systems status. For other plants, an RCM-
or PMO-type analysis may be the first effort to piece together a big-
picture systems-operating overview since startup.
RCM system analysis also documents operator insights that often
identify implicit functionality or hidden assumptionsparticularly in
vintage plants, where design documents are out of date. Capturing
operator insights into system operating goals, problems, equipment,
and integrity-items gleaned from years of experience-is a valuable exer-
cise with many maintenance plan benefits for companies seeking to
extend facility operations.
TRCM identifies system functions, boundaries, and interfaces in
great detail. In non-nuclear applications, the detail can usually wait until
76
RCM Performance
a need presents itself. Streamlined ARCM identifies systems and func-

tions implicitly. Nuclear facility functions are identified in detail due to
the maintenance rule and so receive little benefit from recreation of sys-
tem design documents. Identification of functional failures at the sys-
tem level, which in turn shapes performance monitoring and testing, is
key. Further definition can be added as needed when problems arise
during equipment grouping.
Functional equipment grouping requires selecting system bound-
aries and interfaces, which can be done implicitly. Traditional analysis
stops short of the kinds of grouping, scheduling, and other LCM fea-
tures. Deferring such analysis into the working elements of performing
effective maintenance, and identifying cost information and failure sta-
tistics, are yet other distinctions of ARCM.
Analysis often leads reviewers to discover simplified assumptions
and methods that reflect the designers intentions. Modifications made
to older plants sometimes change original A-E systems but this can be
captured by ARCMs basic, simplified documentation. This supports
additional detailed assessments that speed analysis.
To establish system importance, its important to explicitly identify
functions that support:
safety
environment
production
license technical specification or other formal commitments
agreed upon as conditions for
operations
practical support requirements
major equipment trains and redundancies
essential instrumentation
This review is best performed in conjunction with a full mainte-

nance strategy review and PM development plan that examines per-
formance, cost, component R and failures, operations equipment use,
and monitoring. Extant PM programs are evidence that once someone
viewed that equipment as important. This consideration should be car-
77
Figure 3-2: Critical Streamlined RCM/PMO Approach
ried forward, until more information identifies otherwise.

The analyst keeps these major functions in mind while reviewing
equipment. The plant equipment master register should serve as a
checklist for those reviewing systems and equipment. Consider equip-
ment importance, costs, maintenance demand (PM and CM), and inci-
78
RCM Performance
dental requirements. The list should include:
large equipment
equipment covered by existing PM, calibration, and test programs
equipment of regulatory, insurance or cost concern
major redundant equipment
As a practical matter, organizations with PM programs in place

should determine how it was developed and how much remains current
to simplify the review process. For new facilities, determining equip-
ment size and cost, production impact, and then benchmarking to stan-
dards, guides, and other facilities will identify important equipment. If
a similar facility has already been examined, use its equipment as a ref-
erence. Chances are the analysis will proceed along the same path.
Efficiently performing ARCM requires standards, standard meth-
ods, and processes. Large, visible, high-value equipment with mainte-
nance programs in place is often accessible only at outages and so
receives high organization maintenance focus. Such important pro-
grams cannot be altered until staff cut their teeth on more basic
equipment and simpler maintenance strategy evaluations. Developing
and applying standard templates at each facility, based upon site-unique
equipment and experience, greatly focuses attention on equipment that
is important in an overall sense, and requires routine resources to main-
tain (Fig. 3-3). Important equipment is that which warrants standard
maintenance plan development and includes any equipment that has
production impact, is used repetitively, and demands a significant
amount of maintenance resources.
Although the critical streamlining approach for RCM/PMO is used
and useful to many, its arbitrary splitting of equipment into critical
and non-critical classes has led me to search for other methods. By
reviewing and summarizing my actual method for performing PM opti-
mization applying RCM, I developed an inversion flow process that I feel
is more intuitive to follow (Fig. 3-2). The reverse logic, if you will, is that
we evaluate looking up the importance hierarchy. It is easier to see why
an item does qualify for scheduled maintenance, and add that than wres-
tle over why it does not, and develop a supporting justification.
79
Figure 3-3: Equipment Maintenance Standard
Organizing standards around major classes of equipment has the

added benefit of focusing plant staff around the spectrum of equipment
at their plant, its varying needs, and the options and strategies necessary
to maintain it (Table 3-1).
Standards focus us on what time-based and condition-based PM to
perform and steer us away from performing low value maintenance.
For example, its common to find PMs on manual isolation valves
though with rare exception, they have virtually no value as TBM. A
more appropriate approach is NSM. This doesnt imply the valve
should be allowed to fail or that once it has failed, to ignore it. It means
Do no planned scheduled maintenance and monitor valves until a
problem is noted.
Because we cannot predict when valve maintenance is needed, its
80
RCM Performance
Electrical Mechanical General

Motors Pumps I&C
Breakers Valves: Lubrication
Switchgear Air operated Predictive
Contacts Motor operated
Relays Solenoid operated
Batteries Hydraulic operated
Meters Manual
Pressure reducing
Safety
Relief
Filters
Compressors
Blowers
Traps
Dampers
Table 3-1: Equipment for Standard Templates
more effective to use operator non-specific monitoring to manage the

failure. For those rare valves with significant safety or other non-main-
tenance functions redundancy or TBM replacement may be effective.
Equipment
Hierarchy level
Equipment fits into the system hierarchy between systems and
components. Equipment is integrated into system support functions.
It is often redundant, or supports a redundancy feature.
Instrumentation also fills this role. Equipment is often identified in
trains (Fig. 3-4). Equipment can be alternately viewed as susystems.
That being said, the ways in which we identify equipment is often
arbitrary. It is convenient to view equipment as a combination of com-
81
Figures 3-4: System Component Hierarchy
ponents that come out of a box or that is made up of other subsystems.

The primary reason to differentiate equipment to the component level
is components fail and are repaired. In doing analyses, many items fit
as components while others work better as subsystems. The distinction
is perspective. Can you (or do you want to) further differentiate it as a
component? Do you consider the item functionally or discretely?
How you answer these questions influences whether you take a per-
formance monitoring and testing PM perspectiveappropriate in a
general senseor a proximate failure detection/correction perspective
that is specific to a given component and failure mechanism. Once
established, you cant easily restructure an equipment hierarchypar-
ticularly when using a software product to establish and maintain it.
The best structure-one that produces useful analysis-is one that pro-
vides the desired end productcommon or relevant failure mecha-
nisms (modes and causes), applicable and effective work tasks, and
related essential information (who, what, when, where, why, and how).
Functional classification structures can be very effective at monitor-
ing large groupings of equipment through the use of performance tests.
This cost effectively simplifies monitoring (though, ultimately, compo-
nents fail and must be maintained).
Components are further grouped through the use of codes. Plants
82
RCM Performance
code to different levels on a component register. Nuclear unit compo-

nent coding is controlled (as a design document) so that it is best used.
For fossil units, coding is largely the way equipment and components
are entered into the stations CMMS for work order retrieval. Hydros
(in my experience) code very little. Analysts must find an intermediate
ground if theres a choice. Since the purpose in coding is performing
maintenance, following the stations CMMS equipment coding methods
is usually an excellent choice. Analysts who structure their work to fol-
low CMMS equipment codes find benefits include:
recognizable structures (for station)

consistent equipment records
direct applicability to work order tasks (PM and maintenance
work order [MWO])
simplicity of PM coding
Where existing CMMS coding isnt unique or has no logical struc-

ture, simplification may be needed. Elimination of repetitive equip-
ment identifiers is a prerequisite to program consistency.
Equipment analysis can be developed by hand, spreadsheet, or a
variety of software products. Although software products have limita-
tions, they are a great asset to achieving consistency. Software must be
applied consistently with standardized products supported by quality
analysis. Software provides the ability to create and document on a pro-
gram basis, which can be carried forward over time to support a liv-
ing maintenance program. Software can provide essential standardiza-
tion and grouping tools and also be used as a training tool.
Failure descriptions
System failures are described functionally. These are general and
non-specific with regard to component and performance. System fail-
ures are ultimately caused by discrete component failure but can be
identified much more easily at the system level than at a discrete com-
ponent level.
For example, a system that provides hydraulic valve-position con-
trol might functionally fail to control. Any number of other things
83
can cause slow control response-fluid contamination, actuator plate-

out, leaky seals, worn hydraulic parts, low pressure. The beauty of the
performance focus is that it identifies the effect of the problem on per-
formance without requiring specific cause. When we know effects we
can project likely system failure causes.
We can take specific action when we know component failures. We
cannot when we find system functional failures. System problems
require us to diagnose a proximate failure at the component level to
resolve the system failure. This requires troubleshooting and other ana-
lytical skills. Component failures ultimately cause functional failures.
(Component failures are also known as engineering failures and some
engineering texts refer to them as root-cause failures.) These failures
are not necessarily root-caused however, but are grounded in the R
definition of failureand components are at the root of the physical
fault tree in an analysis. In a broader context, component failures can
in turn have root causes.
Failures and their root causes are addressed by FMECA. FMECA
is abstract, but grounds a complete, TRCM analysis. The inherent sim-
plicity of a PM program is realized when we take a statistical approach.
We want to know-statistically-what things fail, with what failure modes,
and with what statistical frequency. Key points to remember in ARCM
are that:
components fail
actionable tasks must address component failure
failure mechanism = failure mode and cause
typically, there are fewer than three common failure mechanisms
for a component type. Statistically, theres often one
ARCM perspective is statistical, not absolute; we worry about
the common modes overall
if a specific application has a known specific failure or failure
mode, we can address that
root cause does not have to be addressed for PM to be effective
the objective is to manage risk
84
RCM Performance
Normally, the focus of failure analysis is to get to root cause.

Knowing whether youve found a root cause is a prerequisite to pre-
venting recurrence. PMs dont absolutely have to be perfect to be sta-
tistically effective. For example, its common to find misapplied designs
in a failure scenario. A root-cause approach would involve a design
modification to correct the application. PMs sometimes provide an
effective, though not permanent, fix. Many in-plant problems have been
effectively addressed in this manner.
RCM provides three basic strategy options to address failuresfour,
including non-specific CNM for the non-specific maintenance. Any suc-
cessful maintenance strategy starts by applying tasks that are suitable and
effective, based on equipment function and dominant failure mechanisms.
The choices are outlined in Table 3-2 and further discussed below.
Table 3-2: Strategy Options
85
Time-based maintenance is statistically the least frequent task, but

when suitable, its often highly cost-effective. It applies to elements with
a fairly defined aging behavior. When looking for TBM PM candidates,
always consider aging characteristicswhat parts age, how you expect
them to age, what the aging parameter is. For low-cost replacements
filters, strainers, etc.TBM is often the most effective PM in the sense
that it doesnt expend time on CDM. Once a failure pattern is strongly
correlated to a time parameter, TBM is usually a good choice. It works
very well for inexpensive rework/restore tasks, less so for large-cost
tasks that could be extended even slightly by CDM.
Condition-directed maintenance is also known as on-condition
maintenance and CBM. Its a two-part task in its simplest form: A
time-based monitoring or inspection task is conditionally followed by
an on-condition rework or repair task. The item is reworked or
repaired on the condition that it fails inspection or test criteria. As
applied in the commercial airline industry, on-condition maintenance
is a highly controlled inspection procedure based upon a known,
proven onset of deterioration that is detectable prior to final failure. It
always has a specific go/no-go failure criterion. A failed item
receives the directed-maintenance rework/restore task associated with
the inspection. The rework/restore task is also direct, known, and
determinate.
Establishing engineering limits that determine on-condition
maintenance requires intrinsic knowledge of both the equipment and its
failure mechanisms. Its time-consuming at first and tough to doand
tough to sell. Until engineers and operators accept that limits are
definable, and management agrees to follow them, a CDM program has
no teeth. As a result, at many plants with informal PM programs,
equipment continues to run to failure. Establishing limits and the cul-
ture that can live by them is a fundamental paradigm shift for a tradi-
tional organization.
After CDM tasks are identified, implementation consists of build-
ing inspection tasks into appropriate vehicles for performance
rounds, tests, inspections, PMs-and assuring that rework/restore tasks
are defined in the form of procedures, checklists, or other formats for
quick, consistent performance. The full benefit of a planned mainte-
86
RCM Performance
nance program requires a plan that the organization knows it will per-
form repetitively. Herein lies another ARCM benefit-the ability to
reduce a substantial amount of maintenance to a production perform-
ance basis.
NSM. The choice to not perform scheduled maintenance is a pro-
found one. When there are no applicable, effective tasks, and no safe-
ty or environmental issues, the default is to select NSM. While this
selection seems the obvious choice, the role of the PM gatekeeper is
probably the toughest in maintenance. Ineffective or inapplicable PMs
arise from many sources. Regulators, managers, and engineers all feel
uniquely qualified to make maintenance decisions. Systems engineers at
nuclear plants have this in their formal job descriptions. The perception
is that PM is freewrite up a repetitive work order, and it just happens.
Properly developing PMs is as tough as laying out a facility design
tougher, where the concept of engineered PM programs hasnt been
sold. Organizations without gatekeepers perform lengthy lists of PM
activityonly a small percentage of which get done. Nuclear plants
spend inordinate sums on PMs that drive up their costs yet do not ben-
efit operations or safety.
Every organization needs a gatekeeper-type R engineer with author-
ity on the same level as the chief engineer at an architect-engineering
firm. Such an individual controls PM scopes and helps to achieve imple-
mentation on those PMs that matter.
Hidden failure. Hidden failures are those not evident to the oper-
ating crew under normal conditions. They usually result from instru-
mentation and/or control failures, where a component identifies a func-
tional failure not otherwise evident. Some relate to failure of redundant
and/or standby systems. For all critical functionsthose involved
with safety, that would not otherwise be evident to the operating crew
an instrument is typically provided to make the equipment failure evi-
dent. These can further be hard-wired to arm pre-set trips for critical
functions where the trip response time is essential for safety or eco-
nomics.
Nuclear units have many more hard critical trips than fossil
plants. In both cases, however, if the instrument, trip or alarm is the
operators sole line of defense against a critical safety failure, then the
87
component that provides that function is also critical. If it fails, an event

could occur without the operator aware of it or able to take action.
Consequently, instruments provided for safety require maintenance to
assure their hidden function is available.
Auto-start systems and standby trains in nuclear plants have the
same roles. Emergency diesel generators, standby core spray, coolant
injection, and other systems must be available in the event of an acci-
dent. Fire detection and control systems at all plants have similar roles,
and strategies are similar. Those fossil plants equipped with huge ID
fans have automatic high vibration alarms to protect against cata-
strophic failure due to imbalance.
Maintenance strategies for critical instruments, alarms, and trips
should include periodic function tests, especially when the potential for
random failure is highe.g., when alarm functions and subsystems are
complex. Checks of overspeed trips, periodic surveillance and fire
protection, and other routine, scheduled maintenance test these func-
tions.
The functional test is a general default strategy for any hidden
function failure prevention.
Whenever instruments and controls are involved, they should be
considered for hidden functions and functional test condition-directed
activity. Failure to alarm, trip, or otherwise perform the protective
action initiates CDM.
Rework/restore. For either time-based or condition directed main-
tenance, a rework/restore task may be specified. Rework means to
rebuild or otherwise bring a component back into spec. Restore
returns to an in-spec state by replacing parts with qualified spares or by
performing repairs. Repair has a specific maintenance context in
nuclear work. Reworking a cracked weld back to specification involves
a design change. In RCM, its only a restoration task.
From an ARCM-PM development perspective, the two tasks are the
same-from a performance perspective theyre extremely different.
Performing rework is often classified as light maintenance; perform-
ing restore work that involves repair or welding is classified as heavy.
A large, sophisticated, and capable facility will have more in-house
maintenance performance skills than smaller ones but in all organiza-
88
RCM Performance
tions, there are repetitive needs to assess, and the decision to be made,
whether to rework it or contract services to achieve in-specification con-
ditions. A combination of the two is needed at most facilities, and must
be factored into the RCM program.
Blocking tasks
After applicable and effective tasks have been selected, they must
be blocked for effective performance. (Fig. 2-2, page 26)
Blocking starts at the task level. For instance-achieving performance
effectiveness for a large turbine overhaul requires selecting and per-
forming between 20 and 50 major TBM and CDM rework/repair tasks.
Many of these in turn will be performed hundreds of times. We incor-
porate these into the disassemble/reassemble schedule, as a project to
assure task completion and coordination. This theory applies across the
board, even at the instrument calibration level. (We would never send a
technician out to calibrate just one instrument in a rack.) A good meas-
ure of the effectiveness of PM programs is the degree to which they
achieve blocking to conserve performance trip time. Blocking also
reduces equipment outage duration (Fig. 3-5)
Figure 3-5: Blocking tasks reduces equipment outages duration.
89
Assembling PM activity into natural groupings depends on:
skill requirement
the task
the interval
Once grouped, tasks indefinitely stay together for scheduled per-

formance. There is a temptation in traditional PM programs to attach
additional checks to scheduled maintenance packages that add no
value. The gatekeepers role is to keep these activities out. PM integrity
hinges on absolute credibility. In a discretionary environment, the
absence of credibility means such PMs would likely not be completed.
PM tasks and vendors

Maintenance addresses component failures and their causes.
Identifying common failure modes is an intermediate step to selecting
effective PM. TRCM formalizes this step and consciously evaluates
maintenance options and their effectiveness.
Vendors have the ability to provide adequate PM programs if they
write effective manuals. Many do not. PM is an arta specialty.
Vendors dont always know the best ways to present PM options for
their clients and their unique needs. They understand common failure
modes for their own equipment because product-development cycles
address these before products are brought into production. It is rela-
tively common to find the full spectrum of RCM options represented in
vendor recommendations:
do nothing until a problem manifests itself (NSM)

monitor for specific problems at some intervals, then fix them
(CM [CM]/CBM)
do rework/replace tasks on intervals (time-based maintenance
(TBM))
inspect to specific criteria (CDM)
Functional and performance tests are also prescribed. These

insights are excellent starting points but they dont guarantee an effec-
90
RCM Performance
tive program. Taken literally, and in full force, they often assure exces-
sive and redundant monitoring or overly intrusive maintenance with
excessive parts replacement. Furthermore, vendors cant anticipate a
users exact application. The application determines which of the com-
mon failure modes become dominant. Fortunately, almost every vendor
calls for adjusting program intervals and recommends performance
based on experiencethe universal out.
In most PM programs, vendor-prescribed tasks are adequate. From
an analytical perspectiveand when cost-effectiveness and managing
risk are involved (e.g., R engineering)we need to go further. Ideally,
our statistical frequency-of-failure information identifies dominant fail-
ures, their occurrence frequencies, and the risk they pose in each appli-
cation, so that our overall strategy can be tuned to manage risk. One
needs failures to do this from a R engineering perspective, as RCM is
essentially a R engineering derivative.
Basis history
The link among failure mechanisms and tasks, and our selection cri-
teria, is whats called the selected tasks basis. Using it, over time, we can
track and understand why a given maintenance program is in effect at
any point. This is important from a regulatory perspectivethe main-
tenance rule requires that a basis be carried forward for all PM tasks
at nuclear power plants. A basis is desirable to maintain a living pro-
gram in any plant, however.
Its much easier to assess changes made to a program when you can
trace its origins. The lions share of changes are based on an assessment
of the current state and attendant equipment needs but why a given
program was in place is almost never specifiedeven in the nuclear
plants. At best, change-histories justify an existing program and provide
the basis for it but in a regulated or LCM maintenance program, a doc-
umented basis has value. A basis is an important step to developing an
effective, living maintenance plan.
PM work packages
At the equipment level, we can package multiple PM tasks to facil-
itate efficient work performance. Work can be organized by the per-
91
former skill group-instrument techs, electricians, mechanics-by frequency

(weekly, monthly, quarterly), or by other convenient grouping attributes.
A PM program should be designed for each equipment group, based
on the degree to which PMs resemble and differ from one another.
Information sources
Many sources of information identify viable PM tasks and their
associated failures. Vendors provide more information about PM activ-
ities than failures, but its easy to infer associated failures from their rec-
ommended tasks by analogy, comparison, and experience. Manuals
from OEMs provide maintenance, diagnostic, PM, and calibration
guides. Performance interval information they provide is typically not
directly applicable, so the user needs experience and judgement to sup-
port intervalsor, better yet, a diagnostic capability with an age explo-
ration program. Users who presume continuous service, and lean
toward literal internal applications, design overly conservative intervals.
In addition to vendor operations and maintenance (O&M) guid-
ance, there are:
standards (industry, user groups, professional societies)

legal guidance (particularly in regulated fields)
insurance standards
shop practices
basic failure study and analysis
benchmark plant or equipment practices
Occasionally more sources are available. If we take boilers, for

example, there are:
OEM manual guidance

state laws
American Society of Mechanical Engineers (ASME) Boiler and
Pressure Vessel Code, V1
insurance agreements
EPRI guidance
plant practices
92
RCM Performance
Figure 3-6: Coal Belt Assembly Functional Failures

specific site-failure experience
specific code interpretations
failure experience (such as safety relief valves)
industry event experience
Peer programs, supplier processes and literature, and published

professional society papers provide a wealth of additional information.
In some instances, company or plant licenses may identify additional
specific requirements, particularly environmental or risk-management
requirements.
The point is that to develop a complete equipment failure perspec-
tive, we must review all information sources and establish a relevant
program based upon plant operating schedule, maintenance capability,
and policies. The plan, whatever it is, must meet optimization goals,
work simply, and be supported by workers. Worker commitment is cru-
93
cial to PM success. Successful plans are those developed by worker

teams, incorporating their ideas.
Effective, simple PM standards can address general classes of
equipment (supported by generic industry performance information),
site-specific failure experience, and unique worker insights. Developed
by teams, they not only capture different perspectives, they gain buy-in.
Even imperfect plans can rapidly optimize, under age-exploration
constraints, to quickly correct for any initial absence of data. Selecting
which classes of equipment to address can be based on valuedeciding
operational impact and cost. Once standards are developed, they sim-
plify life for workers, schedulers, and planners.
PM and maintenance performance consistencies support R in many
ways. Standards can reflect unit failure experience (Fig. 3-6).
Failure mode, failure mechanism, and cause

When components fail, the failure mode describes how they do so.
Mode and cause together define a failure mechanism. A failure mode
can be managed if you understand the failure mechanism. The goal of
FMEA analysis is to identifyconciselythe failure modes and mech-
anisms of interest (Fig. 3-7). Successful plant operations depend upon
achieving design failure modes and full component life.
The effects of failure types vary. Inconsequential modes and effects
can be ignored while major, intolerable instances must be understood
and managed carefully. Failure mode variations among manufacturers
and within a given manufacturers product lines can reveal radically dif-
ferent aging factors. For this reason, again, success with PM and ARCM
Figures 3-7: Equipment Failure Hierarchy FMEA
94
RCM Performance
requires that you know the manufacturers product lines and document
your experience. Manufacturer representatives usually do a wonderful
job helping to specify suitable products. These manufacturers cost
morebut they very often warrant the extra costs. They provide valu-
able selection criteria service.
Criticality
Risk is mathematically defined as:
Probability x Consequence
where failure effects determine consequence. Analytical data collected

from actual experience, industry data, failure libraries, vendor litera-
ture, similarity comparisons, or published failure studies quantify fail-
ure probabilities. Knowing failure effects completes a risk assessment.
However, risk itself has different measures. Overall equipment fail-
ure risk is meaningful, calculable, and measurable. FMECA (with C
for critcality) and fault free analysis provides systems-perspective risk
management tools. Overall failure risk for any major plant equip-
ment, component, or integrated system could be assigned a numerical
value at the design level. For example, the overall mission success
goal for a system might be set at 0.995. From this, a FMECA of all sys-
tem equipmentincluding supporting subassemblies and compo-
nentscould then be developed and overall R calculated. If the calcu-
lated R didnt achieve the mission-established goal, then criticality
analysis could identify R loss contributors to mission failure. These
contributors could be re-evaluated and the largest risk re-apportioned
as a design tool. Subassemblies or components in which failure-risk is
high or which offer opportunities for risk reduction, are improved until
the design-risk goal is achieved.
Risk-allocation looks at the desired final product and identifies an
overall target failure risk. It is broken down to the system and compo-
nent level, where FMECA identifies the main risk drivers. Design
changes (substitutions or redesigns) can address and improve overall
mission risksystematically and on a budget.
The term critical has origins in R engineering and FMECA.
95
Unfortunately, the association with R engineering has been lost in

common use. (FMECA ranks failure modes based on the contribution
to failure. In this context, criticality provides a relative numerical rank.)
Instead we have arbitrary delineation of equipment into critical and
non-critical categories. The results are not useful. Once arbitrarily
tagged non-critical it is generally assumed equipment can safely be
ignored-even after clear-cut fault evidence develops!
This is not the same context used by Nolan & Heap. Their criti-
cal definition was that a critical failure is any failure that could have a
direct effect on safety. The two differ on the basis of:
basic purpose
calculation ranking
failure focus
Thus, non-critical has wrongly come to mean low priority and

ignore. In truth, failure modes, mechanisms, probability and conse-
quences determine criticality risknot equipment. This is exactly
FMECA methodology. As an engineering FMECA points out, most fail-
ures arent major mission risks, though a disproportionate few usually
drive overall risk results. This is why risk allocation using FMECA
works! Focusing on the wrong failuresexactly as if you blindly adopt-
ed the critical equipment approachmisdirects resources. This is
the downside to a critical equipment mindset: Failure modes and
mechanisms are important!
The importance assigned to equipment determines criticality-how
high risk failure modes are converted into overall plant impact. Most of
the day-to-day failures on critical equipment (e.g., work orders on
important equipment) are not actually functional failures but are con-
dition-directed work (from CM lists) that must be prioritized in con-
text. An RCM mindset is an extremely powerful tool with which to pri-
oritize work and ensure that operations work control is effective.
Practical Difficulties and Fishbones

Formal FMECAs are an engineering tool that can bog down with
extraneous detail. There are literally thousands of ways that things can
96
RCM Performance
Figure 3-8: Ishikawa Fishbone Example
fail. What needs identificationat an appropriate levelare not the

many ways in which things can fail, but the one or two ways that things
do failpredictablyin plants achieving their design capability. This
means identifying the critical few failure modesusually three or
fewerand assuring these are managed.
This real data review brings authenticity to FMECAs. Practically,
equipment fails in a few, repetitive ways. To work smart, efficiently, and
cost-effectively, we must take advantage of this fact. One of my com-
plaints with nuclear licensing and regulatory bodies is their fundamen-
tal inabilities to come to grips with this statistical fact. Rather than
focus on the way things do fail in this world, they theorize ad infinitum.
Practically, this has no value in a mature industry.
Statistical Process Control (SPC) is a technique to focus work
groups on high risk failure contributors in manufacturing processes.
Ishikawa diagrams (named after the inventor)or fishbones, based on
their shapesallow systematic group assessment and tracking of
process failures (Fig. 3-8). Their effective display of statistically-based
failure cause and relative frequencies, allow continuous process
improvement to proceed systematically in a plant environment.
97
Fault tree analysis (FTA)

Once we understand an integrated system design, we can build a sys-
tem block diagram and develop a mathematical logic R calculation. This
is based upon the relationships among the assembled components and the
probability of individual component failure. Such an exercise is an FTA.
Primarily a design tool, FTA relates overall system performance risk
to the supporting component risk which quickly identifies where and
how design can be improved to lessen risk. Once FTAs are built, how-
ever, plants can also use them for troubleshooting and sensitivity study.
(Fig. 3-9, page 106)
Hierarchy and boundary

Plant structure proceeds in linear fashion from the site, with com-
mon support systems, to units, which have distinct and separate sys-
tems. Systems can be further expressed as subsystems, equipment skids,
and components. Many componentsassemblies replaced out of a
boxcan be further expressed as individually replaceable parts. This
forms a hierarchy.
Functional failures can occur at any of these levels. They flow up the
hierarchy depending on relational logic, redundancy, and design robust-
ness. As components and parts fail, these failure incidents eventually gen-
erate functional failures. Functional failures occur where you define func-
tions-the boundaries. Proximate failure occurs in components and parts.
Making corrections to components and parts to restore function and per-
formance is the ultimate focus of any maintenance strategy, whether it is
TBM, CDM, CM/CBM, or OTF. Failure analysis is developed from the
failure evidence. This will be found ultimately in failed components and
parts, and the hierarchy and boundaries in which they lie.
A common hierarchy includes:
unit
system
equipment/subsystem
sub-tier subsystem(s) (if any)
component
part
98
RCM Performance
failure(s)
causes(s)
The system/subsystem relationship can be replicated as needed. In

practice, there are rarely more than two system levels, but exceptions
can occur. Generally, analysts set the number of systems and levels of
detail. They occasionally desire more levels than software allows. We
must also remember that these are like an organizational chartan
abstraction. Real plant equipment can have dependencies and ties that
dont show up in design hierarchies. Take any model with an appropri-
ate grain of salt. Its utility is how well it provides useful insights.
TRCM establishes system boundaries, interfaces, and input/out-
puts; ARCM simplifies this process using existing, applied system defi-
nition that flows from plants, models, and CMMS structures. Not that
extensive system design layouts arent helpful. They are. But ACRM
doesnt strive to translate existing design information into immediate
forms. In addition to supporting the units needs more exactly, RCM
can be utilized to keep basic rules in mind:
allow no component duplication

seek closed systems, but work with open ones
treat fluids like equipmentprovide fluid failure modes
anticipate equipment grouping for PM performance and rounds
Equipment groups provide one reason to understand system bound-

aries. Equipment boundaries simplify PM programs when selected care-
fully. They naturally define closed fluid systems and groups. When
open-fluid processes are designated as systems, remember the open-sys-
tem rule: Fluids (including energy) that flow into a system influence the
system and can substantially influence system failure performance as they
touch all pressure boundary components. Fluid performance tracking is
simplified by treating fluids as another system component.
Functional Reviews
Engineers find new functions in familiar areas, and their design ele-
ments, major O&M systems, and equipment. These reveal functional
99
intentions for plant owners and operators. Reviews of systems and

equipment with which you are unfamiliar should be given extra consid-
erations while old standbys, like feedwater, should be skimmed.
Design-to-design variations warrant quick review. Most A-E
descriptions average 20 pages or fewer for a fossil system and are high-
ly repetitious from plant to plant. They provide high-level schematics
with important redundancies, which are useful for grouping. The
degree to which the original plant design basis has been maintained is
reflected in manuals, guidelines, and other available documents. Older
fossil plants are often significantly different from their as-built con-
figurations. The degree to which the existing plant deviates from docu-
mentation hints at the maintenance effort that will be required. TRCM
reviews exhaustively create the system design basis, while ARCM cap-
tures the essential highlights for PM use.
Some OEM documentation and manuals are not available.
Typically, in older plants, vendor manuals are missing or hard to locate.
Where unavailable, an RCM-type review can reconstruct systems based
upon similar systems already analyzed, station review, and previous
experience. Supplementary documentation developed by walkdown
may be needed. Personnel interviews provide valuable insights and the
major historical events that have influenced the plants overall produc-
tion, safety, maintenance, and cost records. Incorporating personnel
experience even when documentation exists, is a strategically valuable,
cost-effective RCM review aspect.
How equipment is coded in plant CMMS systems determines the
detail of failure review. It shapes failure, work, and cost-information
sorting. Equipment review can be facilitated by the ability to download
and sort CMMS information. Classifying equipment types for failure
analysis by MWR and PM reviews is a useful and effective way to speed
information processing.
History
Production plants develop a failure history quickly. Industry expe-
rience and history provide high value information for failure risk analy-
sis. Nuclear plants list functions subject to significant risk and support-
ing equipment under the NRCs maintenance rule. In fossil, person-
100
RCM Performance
nelespecially operatorsknow risk significance by experience.

Significant risk factors for major equipment can be extracted from plant
trip data submitted to the NERC at the unit level. Generic NERC data
can also be consulted by unit class. In a new plant, its essential to dis-
cuss projected risks with operators. For instance, fire risks differ radi-
cally with different fuels, and so an operator interview usually conveys
such information quickly.
The importance attached to personnel interviews and comparison
plant experience cannot be over-emphasized, both for gray beards who
remember that little valve up there that blew its packing, initiated
feedwater upset, flooded the drum, and tripped the unit.
In assessing equipment importance, obtain a representative history
of between 3 and 10 years worth of failure datausually as MWRs
obtained from the CMMS by system and sorted by equipment tag num-
ber. Typically, for any given class of equipment or component, several
hundred legitimate failures are necessary. This can be difficult to obtain
in nuclear studies, where many work orders are hypothetical audit ques-
tions, not failed equipment. Depending on the plants coding system,
assemble information according to a given equipment type. Large-sys-
tem reviews require methods to manage and display information that
are fast and standardized. Generally, you need to collect:
system equipment component lists with reference data

P&IDs
MWR lists for a statistically representative period
vendor PM recommendations (from vendor manuals)
existing PM plans and tasks (from CMMS)
standard component programs (if any exist. They could be informal)
management commitments
federal agency information, whether EPA, OSHA, NRC, etc.
state agencydepartments of health, state boiler inspector, etc.
insurance providers (fire, industrial, etc.)
standards (ASME codes, National Fire Protection Association
(NFPA), building codes, etc.)
other local or regional standards
operational reportsusually, summaries of operating logs and
101
production reports supplied to corporate managers, generation

accounting groups, Federal Energy Regulatory Commission
(FERC), and NERC
the station operating license
unit trip or other operating event reports
Operating reports provide insight into the stations major outage
events to detail a plant functional failure experience.
R engineers review failure descriptions, item-by-item, from hard
copy summaries, tallying up types of failures and operations impacts.
They summarize dominant equipment failures, frequencies, and costs to
provide raw statistical data for meaningful reviews of existing PM pro-
grams and whats needed to build PM templates. Once a failure history
has been reviewed (and with a list of important equipment in hand) they
review equipment importance and O&M recommendations. This pre-
pares them to talk with operating and maintenance crews about plant
strategies and experience.
Standards
In practice, there are two, possibly three broad categories of equip-
ment and components:
1. High-impact items without redundancy that operate singly or in

unspared-pairs and directly supports generation. Boiler feed pumps
(BFPs), turbines, ID air fans, circulating water pumps, boilers, and
reactors fall into this category. Large, expensive, and economically
important, they also have safety implications due to rotating inertia,
high temperatures and pressures of contained fluids, cooling functions,
and plant trip potential. These components get maintenance in-depth
analysis including equipment history reviewMWRs, overhauls, PM
program, emergency callouts, operating records, and historical practice
2. At the other extreme are replicated items found in significant

numbers throughout the unit. A single-unit 830 MWe BWR has several
hundred motor operated valves (MOVs) for instance. Smaller pumps,
motors, valves, and other components are present in even greater quan-
102
RCM Performance
tity. Some warrant maintenance but few warrant individual maintenance

analysis based upon similarity of design and failure modes (unless iden-
tified by failure reviews). A nominal component model can be con-
structed and a standard maintenance plan developed to be applied in
production as a plant standard
For example, an air-operated valve standard is developed, tailored

by plant, based upon:
operating air purity

general plant cleanliness
local environmentrural/industrial, etc.
availability and quality of ventilation
application
Standards must address substantial design differences only. When

similar equipment types have similar programs dominant failure modes
should be the same, or nearly so. Some of the equipment presented
above is unique and needs a unique standard. On the other hand, many
common motors, valves, and a host of other equipment are very simi-
lareven when provided by different suppliers. These beg to be
lumped together under a common standard.
3. Between the extremes, equipment maintenance plans can be

broadly based upon standards tailored to an application. For example,
multiple types of chain drive-style sootblowers have similar designs but
different parts and unique characteristics. A standard can be developed
for the general case and tailored to several special applications. Details
in the standard can vary. Because blowers can be associated in different
groups, a high-soot group could be maintained to one standard, a
low-soot group to another. Unimportant blowers could be given NSM
aside from routine lubrication
Development of standards is best done case-by-case, on a site-tai-

lored basis, working with the maintenance craft and engineering staff.
Special requirements from site to site will force standards to be tailored.
103
The environment, interfacing supporting system design, maintenance

standardsall play roles. One site may have excellent, dry instrument
air; another may have it comparatively wet. The former site may be in a
dry, dusty climate; the latter, a moist, damp one. The former site might
find that motor heater checks add no value at all but that air inlet filters
are essential. The latter site may require motor heater checks to avoid
damp and burned-out windings from motor startup.
Tailoring of vendor requirements to individual site conditions elim-
inates canned vendor-recommended PM maintenance and yields pro-
grams better suited to local conditions and practices. Skilled workers
who understand equipment needs implicitly can perform off-track PM
tasks and programs, year after year, without adjustment if their facilities
start with OEM-based programs and never develop a PM tuning
processes.
Standards also effectively address vendor requirements (especially
since vendors go overboard on suggested maintenance at times).
Either that, or vendors advertise that, Our equipment demands
virtually no attention beyond periodic checks. This poses some dilem-
ma for the system integrator who takes vendor promotional literature
literally at face value!
A standard identifies those few things that must be checked on rou-
tine operator rounds, TBM PMs, and special CNM tasks. It provides an
appropriate, site-specific interval to manage risk, as well as those things
that need to go into a PM schedule on a longer interval. We identify the
critical few by direct RCM analysis or comparison to a known analysis
and by using insights based upon other key information and lessons.
Comparison Analysis
TRCM always performs comparison analysis as a final key project
step. These before-and-after snapshots hold limited value when com-
pared to the value achieved with RCM-based PM reviews.
Comparison analysts require a high level of bookkeeping to main-
tain a spreadsheet documentation of project accomplishments.
Documentation is suspect if plant staff delays PM changes, misses
review meetings, forgets to prepare reviews for meetings and neglects
rework analysis.
104
RCM Performance
Comparison analysis is performed when plant owners want to know

how well a PM optimization performed. Comparison analysis will con-
tinue to be performed for this reason. But numbers should be careful-
ly considered to avoid a paper PM program that means little.
Summary
Detailed, TRCM analysis has great value primarily as an analytical
learning tool. Standardsapplied quickly and reapplied many times
speed PM assessment and implementation. People working to stan-
dards develop processes and production methods to perform repetitive
tasks with consistency, speed, and simplicity to make the standardized
application of formal RCM techniques effective.
However, dedicated equipment applications, failure consequences,
and the ways in which we decide equipment importance necessitate
adjustment that limits the depth of TRCM analysis. Fortunately, bench-
mark and composite references mean detailed RCM analysis isnt always
needed. When we identify important components, develop appropriate
programs and standards, then devise the appropriate PM task and
measure the results, we can further adjust individual component PM
programs to overall standards requirements, where needed. Simplify.
Standardize. Implement the best ARCM-PM program.
Maintenance Process
Traditionally, work orders are based on noted problems. This is the
CM maintenance model. A second modelscheduled maintenance
supplements and extends the fundamental model. But identification of
problems, a posteriori, is how traditional maintenance works.
Operators know the problems because they know the system, the equip-
ment, capabilities, and what they need it to do. Response-based main-
tenance was the first improvement over disposable equipment due to its
significant capacity to reduce cost.
Response-based maintenance is very cost-effective when compared to
the alternativenothing! Its the first basic step in any maintenance pro-
gram. The next step is an intuitively harder onescheduled maintenance.
105
Figures 3-9: Fault Tree, courtesy Item Software, Anaheim, CA.

Scheduled (or preventative) maintenance is much harder to implement
than failure-based maintenance, but many companies satisfy themselves
with response-based maintenance implementation. To see what scheduled
maintenance can do, consider first what it cannot (Fig. 3-10).
For any component, there is some residual failure rate tied to ran-
dom failures inherent in the component (based on design and produc-
tion processes). This minimum failure floor the best PM program can
only approach, with diminishing returns. Between the response-driven
(RD) and random limit (RL) floor is the area that scheduled mainte-
nance can influence. This RD-RL difference can only be improved by
fundamental design and manufacturing product process improvements.
The RD-RL difference varies with equipment. Factors establishing
the floor include:
design
materials
106
RCM Performance
Figure 3-10: Random Limit

construction
environment
operation
Failure results when stress exceeds capability. Design, materials,

and fabrication provide equipment with capability. Operating stresses
in a perfect, variation-free world would never exceed design limits but
in the real world, they do. Equipment designers must anticipate field
loads and conditions. Suitable materials, manufacturing, and dimen-
sions assure products perform adequately with a factor of safety.
Designers build systems from components and equipment. They
arent exact. System designers work off experience. They stretch
design envelopes with operating and environmental assumptions.
Some application stresses exceed design expectations. A residual fail-
ure rate is often present in efficient economic design. A perfect main-
tenance program would achieve the residual inherent capacity of the
design (Fig 3-11). Discovering this floor (with scheduled maintenance)
107
Figure 3-11: Fault Tree : Loss of Cooling

and extending it through design is the focus of ARCM. Ninety percent
of component failure modes do, in fact, realize this inherent capacity
with virtually no maintenance. This is the discovery of RCM and why
we must use tools with great care!
CDM lies somewhere between absolutely no maintenance and
inherent R limits. Response-driven maintenance works well, as a first
step-and this is where many organizations find themselves. Further
strategies move closer to design-limited R.
Scheduled maintenance effectiveness has been validated by long
term measurement of steam turbines. Here the failure rate curve looks
like Figure 3-11. Load capability determines overhaul times.
Peaks are limited by suitable PM tasks that reduce failure rate to
lower levels for a given period. Performing maintenance establishes
an intermediate failure level R curve. Scheduled maintenance plans
drive failure rates further towards the inherent R floor. To the degree
reduction is cost effective, scheduled maintenance is effective.
For some failures, operation changes or re-designs are necessary.
Failures caused by unanticipated external environmental factors require
108
RCM Performance
Figure 3-12: Run In
review of environmental control. Opportunities vary from one applica-

tion to the next-some are minor, others are large. Other factors influ-
encing task selection is the requirement that we add value. Infant mor-
tality or quality can influence the run-in period failure rate. After
some period a higher failure rate returns to an inherent baseline level
(Figures 3-12, 3-13 and 3-14).
109
Figure 3-13: Turbine Failures. Although turbine overhaul failures following overhauls
follow an infant mortality curve overall, the composition of individual failures is not so
clear. Limited extension intervals suggest that extended lifetimes between overhauls are
feasible for many turbines. Ultimate age-based turbine failures appear to be a compo-
sition of blade deposit and erosion failures for many machines. These cause stage effi-
ciency to fall. Overhauls are then a question of economic production tradeoffs.
Assessing numerically small failure numbers means interpolating between few fail-
ure events using judgement. This is at best a risky proposition. The comparison of
many machines with many failure modes at the other extreme is also fraught with risk.
In the final analysis, an engineering inference supported by detailed parts examinations
and performance tests is the most useful approach.
110
RCM Performance
Figure 3-14: Cumulative Turbine Failures. With a fleet, turbine failure periodically
approximates the overhaul interval. While this suggested wearout, closer examination
showed most failures occurred following start-ups and reflected infant mortality prob-
lems. In fact, that best explains the timing! Data like this suggests that turbine over-
haul intervals may be extended with minor risk. Until age exploration establishes a
wearout interval with more exact failure experience, the predominant risk is under-
utilizing the asset. OEMs complicate issues by providing traditional time-based over-
haul interval recommendations.
Defining efficiency and load loss failure further complicates the problem.
Some companies have vague standards for end-of-period performance that provide
the basis for overall intervals. Without and exact efficiency standard, failure to
achieve performance is a subjective call. Although the issues are complex, there are
simple measures and solutions.
Lastly, the data support the idea of random limit. For this fleet, some failures
persist throughout the overhaul cycle.
111
Chapter 4
Plant Needs
Operations is hours of boredom punctuated by moments of stark terror.

-Anon, Navy
Production and Delivery

Production processes have great varietysome are batch, some
continuous; some three-shifted, some one. Most power plants are
staffed 24 and 7but this is changing as operator functions are rede-
fined from shift engineers (who can start up, shutdown, and reconfig-
ure the plant) to utility workers performing minor maintenance and act-
ing as diagnostic technicians.
What will not change, is that plant processes require monitoring.
Even remote-operated site dispatchers assume monitoring status. When
monitoring is separate from maintenance and other production support
processes, there must be interfaces. Operations traditionally main-
tained plant configurations, production, monitoring, and control while
others provided support services. Supporting processes greatly influ-
ence maintenance effectiveness. Operators identify CDM needs and ini-
113
tiate most plant maintenance. Various PMO implementations show that

even with advanced diagnostic techniques, operators initiate at least
80% of non-routine maintenance MWRs. Operators form a key link in
maintenance performance.
The planning, scheduling and other maintenance groups can provide
effective support-to the degree operators initially monitor and identify
equipment discrepanciesclearly, accurately, and expeditiously.
Operator value is even more pronounced when they provide flexi-
ble, immediate maintenance responseon the spot packing take-up to
address a small leak, for example, or minor instrument adjustments. In
many instances, operators can be trained to regularly perform these
duties. IPPs have effectively trained operators for light maintenance
roles.
Like all other staff, operators are a resource that must be conserved.
Minimizing plant staff requires understanding exactly what needs to be
done, when and then performing this (directly or remotely) with mini-
mum wasted or redundant effort. This, of course, is in addition to the
operators fundamental job of managing and configuring the plant.
Flexibility ensures that operators will always have plant roles. These
roles may change or merge with other functions, but the basic opera-
tions role will remain until and unless plants become literally dispos-
able.
How will operator roles change? With new information manage-
ment systems, the control operator could easily be merged with a rov-
ing operator using virtual-reality technologywearing a hard-hat with
a heads-up display, like a fighter pilots helmet. The operator would
be free to control the plant, on the spot, in the course of making rounds.
Functions such as plant monitoring, control, and
startup/shutdown/reconfiguration capability are preserved in radically
innovative ways.
Perhaps the correct way to say this is that operators who add value
will always have roles. Companies must restructure processes and roles
to add value for all personnelor fade away. They need to adopt meth-
ods and processes that increase operator value. PMO and ARCM add
operator value because they improve the ways in which operators are
used for CNM. This improves R and maintenance performance by:
114
Plant Needs
identifying maintenance needs sooner

initiating condition-directed maintenance earlier
providing greater focus than a traditional corrective maintenance
program
Delivery
The value of any plant improvement process is limited by the abili-
ty to deliver the benefits to the customer. When RCM methods drive
PMO for equipment maintenance programs, LCM, and overall plant
work scheduling and coordination, its based upon an intrinsic belief
that they represent better ways.
The next question is: how can a plant get there? Two basic process-
es are required. If absent, they must be developed. They are:
a PM process
an LCM scheduling process
There is no such thing as a completely effective PM process. An
effective process must:
consistently deliver a high PM-completion performance

provide effective methods to set priorities and allow rescheduling
to deal with contingencies (like outages) that compete for
resources
provide personnel opportunities to learn and to change processes,
as needed
The LCM scheduling process must be able to deliver routine, con-

sistent, non-outage plant activity. The scheduling window must be large
enough that most work planning is feasible, and routines can be devel-
oped and learned. It needs to address management strategies for sys-
tems and conflicting equipment risk, divisionalization and equipment
grouping and how it supports the adoption of schedules and routines.
(Todays the first Monday of the month. We always do fire protection.
During the third week of the quarter, we always do 4160 Bus 1B fault
detection tests.) Such routines provide anchors. It allows groups to
plan exception work around a systems or an equipments known
115
work basis. Equipment comes up repetitively and can be worked

online. Risks are known, pre-approved, controlled, and accepted.
Routine work has preset rules, checklists, conflict control routines and
checks, and qualifications that avoid random performance, outcomes,
and failures.
Management must commit to support process development and
assure that maintenance orientation actually changes. Process changes
must support work structures, software, and processese.g., an LCM
schedule that requires operations and scheduling, working with per-
formance groups to lay out the systems they need to work on in some
overall strategy.
Systems Approach to RCM

Development
To begin to apply RCM select a part of a plant. A large plant is com-
plexan integrated coal-fired or nuclear plant has upwards of 100 sys-
tems. To implement PMO and reap tangible benefit requires focus. You
must identify, review, and evaluate high value systems in a project-ori-
ented, standardized, cost-effective way. (Fig. 4-1)
It takes years to develop a feel for the operability and functional
requirements of a specific plant. In-depth knowledge of 15 major sys-
temsand working knowledge of the restis required. Specialists
require more. Profound understanding of key generation-supporting
system requirements, including R and cost, helps to ensure optimal
plant operations.
Developing an in-depth understanding requires a cross-functional
team effort, a process, and a place to hold the collective conscious-
ness that results. An integrated LCM operating and maintenance plan
provides the process and place.
Training, design basis, and needs awareness

At a facility, the high-voltage power supply demanded very little
maintenance for many years. Then, in fairly rapid succession, a bus duct
from the main power transformer grounded and a 4160 breaker failed.
116
Plant Needs
Figure 4-1: Pareto Chart of System Losses

At another unit in the same system, a station reserve auxiliary trans-
former blew up, along with a 4160 switchgearall within several
months. Reviews indicated that plant operators were unaware of mod-
ifications and requirements that fundamentally changed the systems.
Had they not lost touch with their plants design basis, they would have
done things differently.
Plant personnel can lose touch with a design basis. In most cases,
only significant events re-focus engineering on recovering them. Senior
plant operating and support personnel often carry these requirements
around in their heads but only rarely does this get captured in a safe
environment. Nuclear plants fare better than fossil, because regulatory
requirements maintain design bases; and refresher training. Even with
that, there are problems in accessing, using, and modifying design basis
information.
One consequence of a thorough RCM-based system review is that
personnel relearn intended system functions, plant missions, equip-
ment criticality and those weaknesses in design, operations, and
maintenance inherent in an operating facility. The RCM process allows
the team to recapture this information. Once recaptured, the
RCM/PMO process can document it for easy retrieval and further use:
117
Training
Reassessment and updating of operator rounds for improved
monitoring
Identification and ranking of problems with redesign opportunities
Extension of service intervals on equipment through a systematic
application of age-exploration
RCM can be used to evaluate and rank design modification

requests, separating nice to do from must-do modifications by
objectively documenting risk. Even small modifications-like piping
rerouting, ladder installation, or structural support modifications-
demand substantial plant resources. Design engineering staff cost
these innocent changes and charge for them accordingly, but plants
often perform them under maintenance. This could be to lower
expenses, keep the mods off the capital budget, or to lower perceived
expenses, but such seemingly minor mods take a disproportionate
amount of the station resources. They often come in far above estimates
and budget. One lengthy study of plant design modifications performed
at a plant found that average design modification cost came in at more
than 10 times estimated costs! This trend was sustained over many
modifications and several years time. Modifications out of the shop are
usually underestimated. RCM can often add the disciplined control that
keeps the nice-to-dos on an even footing with the value-adders.
On a regular basis, RCM can screen two types of work that regularly
vex plant management: capital budget requests and safety-justified
work. Using RCM rigorously wont eliminate all ineffective capital
spending, but I believe with strict implementation and enforcement,
RCM encourages people to become more aware of options and possi-
bilities before they commit to any one plan. It will help some utility staff
to learn their jobs and will help control traditional committees plagued
by little accountability, who use vague processes to spend money for
projects that have no basis in actual need.
System definition
Operating and engineering personnel use the plant architect engi-
neers system structure to understand work and develop operating pro-
118
Plant Needs
cedures. Maintenance personnel often lack such perspective, which can

lead to missed teamwork opportunities. For example, when mainte-
nance is repetitively willing to take systems down to perform corrective
work, they add to operating costs. Operations inability to recognize
system-synergistic effects during troubleshooting leads to less-effective
root cause analysis (RCA).
Understanding systems is a trait necessary for any competent
operator. Operator training emphasizes systems and their requirements,
but some plant training pales by comparison to simple U.S. Army boot
camp gun assembly/disassembly drills! Usually its because trainers for-
get that operating requirements stem from plant design but that oper-
ating practice deviates from design intentions, to some degreesome-
times substantially. Requirements not reflected in design documents or
operating guides can be lost over time when designers original inten-
tions are forgotten, and increased costs are sometimes the result.
System performance measurement

The system metaphor is useful as a design and operating tool but
it carries over into cost accounting or performance measurement unit
operations areas. Only a few plants systematically follow system-based
costs. Few old CMMSs provide the capability to systematically track
costs below the unit level. Tracking capability is usually driven by FERC
reporting requirements and codes, based upon specific components
and FERC categories. FERC codes provide a place to start examining
production cost contributors but they developed from an
accounting/regulatory perspective. Better than nothingless than
ideal!
System performance measures break down losses into tangible,
graspable opportunities. Fossil generation losses attributable to coal
mill availability focus attention on mill operating strategies such as work
performance (on/off-line), overhaul intervals (quarterly inspection
schedule, tonnage based, etc.), CNM (motor amps trends, differential
pressure trends, fineness trends), and even risk management approach-
es for startup mill tripsa common problem. Volatile coal fires suggest
fire-risk measures are also important to select. System level performance
standards and measures often provide warnings of problems to come.
119
Intuitively, I think we recognize that maintenance deferral on an asset

like our house or car will result in higher costs later. This same level of
concern is needed on systems in plants.
The NRCs maintenance rule mandates that nuclear plants measure
the performance of risk-significant systems. When the rule was imple-
mented, it elevated system management awareness in nuclear plants.
Fossil units have equally complex systems-combustion, emissions, water
and disposal systems all operate subject to complex, regulated require-
ments. Even without a maintenance rule, the value of system perform-
ance measures shouldnt be overlooked. In my fossil experience, system
level measures identified many significant opportunities. Personnel at
some fossil plants may look at their nuclear brethrenwith the per-
ceived luxury of system engineersand claim its not feasible to imple-
ment such a degree of effort, given fossil staffing levels. However, some
IPPs and co-generatorssome of the best competitors-have proven
that they do have this capability, not just among staff engineers but on-
shift operating crews, as well. Developing maintenance capabilities
comes from an organizations attitudes and assumptions about mainte-
nance-not resources. I believe a lack of needs understanding and
training are as common among engineering support staffs as actual
operating costs and problems.
System-forced outage contribution rates (e.g., chargeable losses) need
tracking. Secondary effects from sootblowing, feedwater, circulating,
and makeup cause boiler outages, for instance. The availability, per-
formance, and R of important backup systems influences overall R.
Nuclear units declare limiting conditions for operations (LCO)
grace periods. Loss of fossil systems converts to a real outage! Nuclear
units suffer LCO-driven shutdowns, but the loss of a single sootblow-
ing air compressor at a Powder River Basin (PRB)-fired fossil boiler can
cause convection passage plugging which, once started, typically ends
with an outage, followed by blasting or lancing to free and remove slag
and ash deposits.
System monitoring
Operations monitoring combines an operators skills, knowl-
edge, experience and senses (sight, smell, sound, taste, feel). It also
120
Plant Needs
requires the instrumentation needed to extend the senses to fully mon-

itor all areas deemed important by equipment designers.
For operators to monitor the plant, the tools-whether tradition-
al, discrete, physical devices or completely integrated DCS systems-
must fit the maintenance strategy. A repetitive lesson of ARCM is the
need to adequately define strategies for instrumentation and to tie these
to the installed physical plant.
Existing work processes often:
devalue critical and essential instruments (the critical few

buried among the trivial many)
include extraneous setup and test equipment
inadequately relate critical and essential instrument functions to
operator actions
For operators to be effective with instruments-whether working

the boards, monitoring a cathode ray tube (CRT) touch screen, or
plying the traditional round-instrument roles need clear identification.
Expectations for instrument response need clear, unambiguous guid-
ance and instrumentation systems, again, must be part of an overall PM
strategy.
While these rules are enforced in the aerospace and nuclear indus-
tries, fossil generation has lacked an integrated strategic perspective
concerning the role and function of instrumentation. For many gener-
ators, the low hanging fruit in ARCM is the opportunity to organize
the maintenance I&C sector. But ARCM is a powerful tool to add focus
and integration to instrumentation maintenance.
Instrumentation is as complex as the plants it guides and guards.
Theres a lot of it and its essential to operations. I&C specialists are
among the highest paid and highest skilled personnel and the discipline
represents a major portion of operating costs and capital requirements.
Safety has strong ties to instrumentation as well.
Whats the state of your plants instrumentation? Do you have an
overall strategy? Do you know, at a moments notice, what any given
instruments function is? Its calibration status? Its annual costs? If you
cannot answer these questions confidently, then the ARCM will help
your operations staff integrate instrumentation into plant hierarchy.
121
System cost
Changes in system costs provide an early warning of items worth
further investigation. They also provide key measures for benchmarking
in competitive studies. How many generators know their air costs?
How, then, would a fossil generator evaluate a proposal by an air com-
pressor vendor to provide air at a unit volume price?
When system and service costs are measured, they provide serious
numbers for thought. In a competitive environment, cost oversights
raise unit cost. Loss of 1% generation in a year, and the associated gen-
eration R loss, is opportunity (and revenues) lost. Ask any IPP operator.
Assessing PM Programs
PM programs must pass the same muster as any other: They have to
contribute to the bottom line.
This means measures have to be in place to assess PM costs and
delivered benefits. Integrated effectiveness measuresthe statistical
and cost pictureare the key measures. PM activities must meet the
bottom-line acid tests of technical and cost adequacy. Failing either test
means the PM is probably an unnecessary expense. Like all expenses
time and otherwisegetting employees tuned to look for low value
or non value adding expense material is a key to long term financial suc-
cess.
Acid test #1: applicability

Applicability is a unique RCM attribute. ARCM uses applicability
with fewer rigors than called for with formal statistical tests. It is still a
rigorous test, however. A high level of assurance can also be attained by
benchmarking tasks to others by similarity, using expert opinion and
review, and performing statistical analysis of actual experience. In each
case, simple tests assure that proposed TBM tasks are appropriate and
incidences of ineffective PMs that weve examined are avoided.
However, applicability is not easy to gauge. It shouldnt be presumed.
Applicability requires that a reviewer be thoroughly versed in the
available technologies, their routine application, and results. Unskilled
performance of suitable tasks can result in failure. This is an imple-
122
Plant Needs
mentation issue. More basic is the application of correct technology in

proven formats as predictive tests on condition. The wrong test is not
applicable. Using ultrasonic tests to search for early bearing failure
could be fruitless if the test cant detect the fault in question.
Unfortunately, in the past some tasks have been based more on
hope than proven fact, and its not unusual to find inapplicable PM
tasks. Occasionally their lack of value becomes a political football,
rather than a learning exercise. Determining applicability calls for small
teams of knowledgeable expert reviewers.
Many time-based CNM tasks include the specific measurement and
assurance of chemistry limits, for instance. These activities fit the PM
definition but have traditionally been included as a part of a station
chemistry program. Its pointless to reclassify these as maintenance PMs
if theyre part of an effective chemistry program. On the other hand,
critical chemistry alarms and instruments generally require time-based
calibration or checks that are often not formally monitored under exist-
ing chemistry programs. Identifying, and finding a home for these
checks can be a valuable aspect of an ARCM general review.
Until the stations entire scheduled work task list is reviewed and
checked for applicability, there is an intrinsic barrier to implementation
of any of themand this general absence of value in all PM tasks
destroys the integrity of the entire program. Specifically, every PM WO
for an operationalized task must be reviewed and certified for effective-
ness to assure specific benefit that is not repetitive or redundant to
another PM check or activity. This detailed PM-by-PM credibility
check builds a stronger foundation for the overall program.
Its also hard work that plant staff frequently lacks the skill to per-
form. Oftentimes, it has never been performed! However difficult, the
benefits resulting from performing these reviews are substantial.
Reductions and consolidations in PM activity almost become a by-prod-
uct. Cleaning up the books has achieved up to a 60% reduction in
system tasks, providing time for proven, applicable ARCM-based PMs.
At the very least, a lot of extraneous work has been purged and the
focus applied to the relevant tasks.
The applicability test is straightforward for the assessment of 80%
of a typical program. In the absence of clear-cut evidence that predic-
123
tive methods are applicable, the opposite should be assumed.

Operational and economic-based failure prevention tasks should clear-
ly pass applicability tests. Inability to confidently support a monitoring
technical basis suggests it doesnt work!
Acid test #2: cost effectiveness

Effectiveness, when used in a discussion or application of RCM,
refers specifically to cost-effectiveness. Clearly, a non-applicable task
will not be cost-effective and separating the two effectiveness criteria is
unnecessary, for an obviously inappropriate task. Cost-effectiveness is,
by far, the tougher to handle. Therefore, each should be given separate
consideration to assure both tests are met and resources are applied
well.
Timing influences effectiveness. Frequently performed activities can
lose cost-effectiveness. Traditional airline RCM has rigorous applicabil-
ity requirements for on-condition task monitoring that havent been
carried over to electric generation use thus far.
For example, all scheduled, on-condition operator tasks involve
specific monitoring, incorporated as scheduled-rounds tasks. Rounds
tasks have historically not been rigorously based and have included
many non-applicable, non-effective tasks in addition to NSM-type,
non-specific area checks. Yet, rounds offer the typical plant a major
opportunity to reduce overall workload and scope while significantly
improving monitoring.
RCM strives for cost effectiveness by assuring that all OCM tasks
are performed at intervals that identify failures before the final failure
phase begins. Because some discrete failures lack this terminal phase,
often the best we can do is check for failed states. Since the design stage
includes these elements with redundant equipment and channels, our
tasks are channel, alarm, and status checks. (In some DCS systems,
these are automated.) While ARCM strives to assure that rounds inter-
vals are appropriate, it also recognizes that not all rounds will result in
detection of final phase wearoutnor do they need to. The essential
requirement for specific activity in operator rounds is to assure that
tasks specifically address failures and that intervals are appropriate
based upon failure rates. Traditional rounds failed to provide intervals
124
Plant Needs
for thorough monitoring. One had to be quickand superficialto

complete a large plants hourly round.
Sadly, the effectiveness test is shunned by engineering groups and
maintenance support. Preparing cost estimates to support longer inter-
vals and effectiveness hurdles is exciting only if youre an accountant.
However, for these cost checks to be done correctly, a technical engi-
neering support group must prepare them, develop effective, detailed
cost benchmark cases, and then evaluate other comparable tasks. As
much as I personally dislike this analysis, it sheds light on where the
value is and where to place resources. Very often, the plea heard in
plants, in defense of many PM tasksusually distant from perform-
ersis, The cost is minor. Just do it!
But once again, little things add up. This approach, carried forward
over time, builds ungainly programs that cant be managed or maintain
credibility. Typically these multitudes of things are only partially com-
pleted. PMs should be dropped if their cost benefit cant be determined
with some certainty. In most instances these cases are straightforward.
CNM efforts can often be trimmed based on cost. For big-ticket fail-
ures-when what if accountability could be an issue-responsible staff
sometimes initiates a PM to wash their hands of further involvement
despite the fact theyre best qualified to manage the problem. This cant
be allowed to happen. More than one PM program has failed based on
over-scoped work. These cost assessments are the staple of any plant
support engineers work. If they cant create them, they probably dont
have the basic skills to be effective in this gatekeeper role.
Process Improvement
Maintenance performance improvement must address two aspects:
the establishment of a basic PM process

high value, time-stamped tasks that really need to be performed
The two are parts of the same puzzle. Confusing the issue can be the
problem of establishing a basic maintenance PM process when another
system is already in place-even if its not performing well. A pilot proj-
125
Figure 4-2: PM Performance Process
ect can be an effective way to initiate a workable process on a limited

equipment scope when nothings in place. You can work up from there.
Because plants proceed on a business-as-usual basis, participant
selection is one key to pilot program success. Voluntary participation by
those with an interest in PMs and commitment to the long term well
being of the plant is required.
126
Plant Needs
Establishing a process
Since most large facilities have a corporate CMMS, its necessary to
build a CMMS-based PM system. Many companies treat their CMMS
separately from their basic maintenance processes, yet its the process by
which they determine, plan, and carry out their work and the process
they use to develop and maintain a maintenance strategy. Ideally, the
CMMS is designed around a working PM process. (Fig. 4-2) Many
legacy systems had PM added as an after thought. With so many facets
of efficient PM performance and so much equipment in a typical, mod-
ern plant, concurrent PM process development and CMMS implemen-
tation is not feasible. Getting a basic PM process instituted around a
small system or core group of equipment is a necessary first step to a
comprehensive site-wide program. Tying CMMS support processes into
the program follows. More PM process development follows addition-
al CMMS tuning.
CMMSs may lack someone to manage the program. In a crisis-ori-
ented plant, the PM portions of the CMMS are implemented incom-
pletely, so an effective PM process never develops. For these plants,
attaining a fully-implemented PM process is an especially high value
activity.
An effective PM process has several essential functionsperform
ongoing PM tasks; rank and prioritize CNM results for time based,
CDM work; incorporate improvements. These elements are based upon
identified failure mechanisms, costs, availability, and other improve-
ments to PM program processes. Plants that presume their PM program
process is adequate often find, after performing an RCM effort, that
essential elements are missing-PM elements, maintenance performance
elements, support elements.
Among case histories of failed TRCM efforts are those which failed
because of underlying assumptions. An ARCM focus creates the most
essential PM elements-quickly-where they are missing.
Processes. Developing a maintenance process model sounds silly
to those who have been doing it for years. On the other hand, why is
it that some organizations do maintenance creatively and uniquely-as
evidenced by their WOs, equipment, and other process aspects-and
some do not? Its precisely because maintenance processes are so often
127
taken for granted as a part of plant needs that occasionally we may need
to confirm our model. (Table 4-1)
Like the maintenance process model, the basic PM process has
many different interpretations. Maintenance outcomes are also influ-
enced by organizations different cultures and personalities. Some get
many miles out of equipment, some get less, but as long as the organi-
zation extracts what it considers to be fair value from its assets, and it
makes a profit, it makes no difference how quickly its used up. In evolv-
ing industries, a facilitys useful life may be five years. Typically, the
high-tech and information-technology industries are radically restruc-
tured that quickly. Competitors adapt quickly or die. In microprocessor
and memory electronics manufacturing facilities, plants are rebuilt or
product lines replaced far more frequently than the generation industry
is used to. In the electronics environment, extracting value from a facil-
ity in five years makes economic senseit may be obsolete at the end of
the period. Based on unit product cost, the least expensive alternative
may be to entirely replace the facility with new when that happens.
Utilities and petrochemicals lie at the other end of the useful-life
spectrum. Generators that cranked out MWs in 1910 are running
Table 4-1: Value
128
Plant Needs
todayand their product is little changed, right down to the cycles.

Electricity is a commodityor, rather, it has remained a commodity.
What has changed are the differentiators and premium services pack-
aged to make one supplier more attractive than another. Compare this
to phone service. True technical innovation in 1983 was limited. Today
(post deregulation) there are a host of exciting new phone services and
options. What changed was that entrepreneurs discovered they could
do a lot with phone service once it was opened to them. Expect the
same innovation in electricity-creative players differentiating their prod-
ucts with exciting new services.
Petrochemical products add high value and profits to the value
stream, driven by the demand for computers, consumer products, and
synthetic fabrics. These markets are expanding and enjoying high prof-
it margins. They are difficult to enter, however, requiring large capital
outlays, complex production processes, and many different skills for
effective production. Electricity, by contrast, is a single, highly refined
commodity that can be produced with off-the-shelf equipment.
Companies roles
More than ever, generation companies need units that operate with-
in predictable costs. Since total costs include payments to co-generators
and unplanned power purchases, in-house generation costs essentially
control costs. Factors effecting random, forced outages vary from com-
pany to company, but for net power purchasers, the unplanned loss of
a single unit means substantial costs. High electricity costswhether
generated by nuclear units or base load coal generationis the factor
driving large end-users to clamor for de-regulation.
Some companies are electing to get out of generation. Those choos-
ing to remain find a tough environment. State public utility companies
(PUCs) arent granting rate increases for those who remain regulated
and most are planning some form of deregulation. Companies whose
benchmarks prove that they arent competitive find it even more diffi-
cult to restructure for competition. In light of this new generating envi-
ronment, the traditional arms-length relationship between plants and
parent companies is likely to become more interested and concerned for
plant performanceif its not already here.
129
Despite stable fuel costs and new technologies, generating costs

remain both high and uneven and this cost disparity is also driving
deregulation. Some parts of the country are blessed with proximity to
resources or sweetheart hydro deals that keep prices low. Others are
on the wrong end of supply chains or committed to paying off late-built
nuclear generation that keeps prices high. Users dont understand, nor
do they care about price historythey see disparity in rates and want
better deals.
The market will determine the competitive clearing rate for genera-
tion. Older plants will either be made competitive or shut down.
Obsolete plants with high-heat rates or poor-performing facilities will
be candidates for re-powering. A third optionperformance enhance-
ment could be viable for marginal producers in which facility design
and operations contribute to uneconomic performance. Companies will
have several options for these plants:
capital investment to improve performance

people/process investments to improve performance
shut down the asset
re-power the asset
sell the asset
Many traditional utilities have been slow to adapt to the changing

environment. In a cash-bind situation, its probable that they will elect
to make a quick fixasset sale or shutdownto unload an unprofitable
facility. Such facilities may re-emerge in the hands of companies that
elect to become generators, provided that:
improvement in performance can be gained with minor-to-moder-

ate capital outlays
human performance issues can be resolved with new contracts,
incentive plans, training, and management
Companies with a fixer-upper will have to decide how best to

realize a gain from each asset.
130
Plant Needs
Nuclear generation
As an entity, the nuclear generation industry answers to regulatory
masters at NRC. Despite outstanding operating records achieved by
these plants, costs are high and hard to reduce. Nuclear processes-com-
plex and slow to change-place burdens on plants that need innovative
and cost-conscious improvements. Gradual attrition of high cost
nuclear units will continue as competitive-pressure increases.
Although overseeing a mature technology, the NRC generates
new regulations and findings, unabated. This maintains a regulatory
focusnot a productivity improvement one. Nuclear plants face chal-
lenges to simplify their processes just as fossil plants do, but fear of
NRC scrutiny gives rise to a conservatism that limits innovative jumps
and raises costs. Nuclear units need to be allowed to explore safe ways
to manage costs and risk in the public interest.
PM Bases
Justification is the concept of a basis. In fossil work, the basis is the
cost-benefit calculation. It should include safety, environment, codes,
insurance, and other compliance and general concerns. It can be explic-
it, but more often its impliedand never documented. In fossil gener-
ation the focus is to do things. Nuclear has no such luxury.
Documented justifications are expected to support changes.
Documenting a PM basis could be setting up changes to be blocked.
This is particularly true where its unclear why a task was even started
in the first place! In nuclear generation, a PM change history usually
provides such a basis. It merely needs to be collected and occasional
gaps completed before it is grandfathered to the original PM program.
Should something go awry, theres an opportunity to check the intent,
results, and see how things got off track. Developing and retaining a
basiswhy a PM is needed, selected, and at what intervalis valuable
information for history and review in either nuclear or fossil work
(Table 4-2).
Nuclear generation, with great many prescribed PM requirements,
requires the change-out of EQ components as specified in their aging
design basis documents and compliance with all vendor-directed main-
131
tenance on essential components (unless a justification supporting an

alternative is prepared and accepted). For a nuclear plant, a basis is a
useful PM change tool.
Vendors usually provide excellent guidance, but occasionally speci-
fy activities that dont make sense. Occasionally, their equipment
some of it manufactured more than 20 years agocan be maintained
more effectively with other methods. Plant staff is often reluctant to
question vendor recommendations, however, in any environment. In the
fossil environment, vendor recommendations are often difficult to
access. Even with overwhelming case histories suggesting certain pro-
grams, staff tenaciously adheres to vendor-based programs and recom-
mendations. An age-exploration program with documentation is an
asset, but most companies cant afford the expertise to develop a formal
parts-aging program. Engineers are sometimes asked to make judge-
ments on age exploration. As an occasional art, its difficult to learn to
practice with finesse. If you cant make an informed judgement on parts
performance in-service, its safer to stick with someone who has, like the
vendor.
Yet, when applied to non-essential parts, its relatively easy to make
informed decisions on intervals based on experience and judgement.
Life extensions based on service durations are also easy to justify.
Occasionally, a part performs far beyond the vendor-specified capacity
in serviceeven by accident. Once thats known, its a simple extend-
life decision. This is age exploration. Superior designs and equipment
are insensitive to part aging. So, one solution to the aging dilemma is to
specify high-quality equipment that includes ingenious methods of self-
identifying aging performance, in service.
Most vendors specify replacement requirements that include lati-
tude for experience-based adjustment. To do otherwise would be like a
manufacturer directing you to replace tires at so many miles, ignoring
your experience. (Nuclear EQ components are the notable exception to
this rule because of nuclear regulations and potential fines.)
Obviously-especially in nuclear or other high risk applications-PM
basis program development begins with vendor manual review. After so
many reviews, and after working with common vendors, most R engi-
neers memorize the dominant failure modes and applicable, effective
132
Plant Needs
Table 4-2: PM Basis Format
PM tasks! This also means that nuclear and fossil operators rarely need
to develop cost/benefit bases for doing many PMs, but can apply tem-
plates that implicitly include cost bases.
133
The bottom line is that any PM that supports generation is a must-

do. Countless calculations show that its cheaper to do PM than lose
generation. Thats the reason for maintaining PM bases. The trick is to
recognize the PMs that influence generation and those that dont.
Marginal PMs cause problems. Such work virtually never has produc-
tion impact in well-designed plants and shouldnt be done. Many stud-
ies have shown that PM is like playing lotto-given two work choices,
wed prefer the big payoff. Some PMs have big payoffs, more have small
ones, and many are losers. Plants need to find the winners and shun the
losers.
Conservatism
Nuclear and fossil plants are alike in that both are ultra-conserva-
tive in selecting PM task performance intervals. Both suffer limited
access to expert analysis and support, and so depend heavily on vendors
for analytical support. Part of whats driving this is conservatism.
Until one checks component performance, in service, first-hand,
theres a tendency to grossly underestimate their capabilities. Combined
with equipments inherent fault tolerance, there are many unrealized
opportunities to extend component life and service intervals. Use data
thats available to you, if only to avoid severely penalizing a maintenance
plan!
Conservatism offers traditionalist operators tools such as condi-
tional overhauls and age exploration, both of which force them far out-
side their comfort zones. But this is where the significant savings are
also.
Conditional overhaul is not more work. It is the directed rework of
a component focused to restore original performance. While not intu-
itive, conditional overhauls have been demonstrated to be statistically
effective for jet engine overhauls. Few generating companies have a
formal repair policy of conditional overhaul, however.
Age exploration and PM interval extensions are a second opportu-
nity. Virtually all companies that use age exploration extend intervals by
minimums of 10-30%. Benefits from such minor extensions take a while
to add up to real cost savings. A substantial lesson from ARCM-aggres-
sive use of age exploration, can significantly extend PM intervals.
134
Plant Needs
Where PMs support economic-based failures, extending intervals radi-

cally supports finding the appropriate time limits. Practically, most
PMs avoid economic-based failure.
Over-conservatism
Task interval conservatism is a requirement for any PM mainte-
nance program. That is: OCM intervals must be short enough to iden-
tify diminished failure resistance, but long enough to realize an items
useful life, so that on-condition maintenance tasks may be effective.
To identify appropriate intervals may require an actuarial analysis.
Those charged with adjusting intervals are not trained actuariesa skill
that requires advanced training in mathematics and many years of spe-
cial tests. A tendency in any programincluding maintenanceis to
utilize overly conservative requirements. Margin is hung on margins,
until many basic intervals reach their constraining limitthe annual
boiler or refueling outage. Of all PMs worked in power plants, the
scheduled outage interval often has the greatest PM frequency. This
represents the minimum interval that can be selected with no produc-
tion interruption. In the absence of hard R and actuarial engineering
analysis, these intervals have become accepted, implicit standards. They
also must be challenged.
An interval that represents half the appropriate (or capable) design
life of a piece of equipment puts severe restrictions on scheduled out-
age work and greatly increases expense. Statistically, annual or refuel-
ing-interval PM intervals disproportionately populate PM systems, and
reviewing annual outage work is a highly profitable task. You may find
that plant support staff selects outage replacement intervals in spite of
performance, vendor guidelines, and other recommendations that sup-
port longer intervals. CNM performed with inadequate specifications
identifies fault conditions early and overhauls prematurely. Overly con-
servative work can load up an outage with thousands of extra work
hours. Properly selecting intervals represent an immediate opportunity
for many plants.
There are other examples of over-conservatism. The asbestos
inspection requirement is one year in the absence of a monitoring pro-
gramthree years, otherwiseso a typical plant can inspect at three
135
years! Yet, almost none do so. Others use annual replacements for parts
faulted for a single failure. A nuclear plant automatically reduces by
25% its EQ service lifetime, just in case an EQ hard-time replacement
PM gets missed. Conservatism adds up. Costs for replaced parts are
high, as are infant mortality failure rates. Actuarial studies show that
overhaul activities are ineffective at improving R, yet they remain a
mainstay of the traditional generation industry.
Alternatives to regular and capricious applications of conservatism
will address any number of oversights. Competent R engineering help,
setting exact intervals, and age exploration standards represent excel-
lent opportunities to advance along the same maintenance cost-man-
agement learning curve in generation that occurred in the commercial
aviation industry.
If PM scheduling is a process problem go after the processdont
introduce common-cause conservatism. It cant correct the fundamen-
tal root-cause flaw an ineffective scheduling process presents. PM
restrictions defeat the purpose. Process errors that occur because of
complexity make it highly unlikely that more complexity in any
processjust like our failure process itselfwill reduce the error rate.
Based on documented parts performance, and provided the envi-
ronment is maintained, quality parts usually exceed expectations. When
environments must be maintained (e.g., protected from water, excessive
temperatures, caustic atmospheres, acid runoff, or excessive
wetting/drying cycles), then use the best materials available, perform
age explorationand condition-monitor.
When fossil environmental control equipment (ventilation and
cooling) is abandoned due to maintenance priorities or difficulty in
using it, its very much like abandoning chemistry specifications that are
too difficult to maintain. Theres no obvious immediate effect, but long
term consequences can be serious. In a number of cases, restoring
equipment to service was easily justifiable but hard to achieve.
Numbers can help define the objective-failure story-numbers that
most traditional generation RCM analyses lack. Some RCM analysts go
so far as to discount failure statistics and numbers. In my opinion, this
is a serious oversight. Implicitly or not, we live by frequency and conse-
quences. However, while numbers dont tell the whole story, those who
136
Plant Needs
Figure 4-3: IR Sootblower Failures
ignore them often chase the trivial few. This has given RCM a black
eye. While some analysts bog down in endless pursuit of rare or imagi-
nary eventsthings that dont happen in this world-my approach
reflects interest in measurement. But I also review large quantities of
data to identify failures, summarize statistics by failure categories, and
make estimates (Fig. 4-3).
The numbers I work with arent exact but they are in the right ball-
park. I view them like dose rate estimates: theyre order-of-magnitude
significant and they identify sensitivity to costs. Costs need to be under-
stood at a 10%, 100%, or 1000% payback during the period of inter-
est. A 10% payback on a turbine overhaul may be worth chasing but
probably not for a $20 filter replacement. A 500% savings on a $20 task
clearly outweighs the same for a $2 task, so we want to structure our
programs to capture that value. Practically, this means when it comes to
a trade-off (and it will), we must give up the $2 tasks to make room for
the $20s.
Ultimately, activities should reflect on-site statistical data and failure
experience. Environmentsincluding the work environmentare
unique to each plant and influence what fails (and when). The cultural
137
environmentlevels of skill, knowledge, and other intangiblescan be

inferred but is just as hard to measure, and also influences failures. Just
as two randomly selected individuals will experience different success
rates with the same auto (as measured in longevity and life cycle cost),
two similar plants experience distinctly unique operational outcomes.
These can only be explained in process and cultural terms (Fig. 4-4a
and 4-4b).
Avoiding rare failure eventsthe root cause of most heavy produc-
tion and financial lossesare the major controllable benefit from an
ARCM plan. These events are worth understanding.
Those in my experience show common traitsinaccurate, failed, or
unavailable instrumentationoften play a role. Second, general instru-
mentation status warns of structural process problems. Ineffective
maintenance of critical instrumentation and failure to incorporate that
into an overall operating plan, indicates a weak operating organization.
Just as there are strategies designed to reduce the risk of auto acci-
dentsand we accept that a driver with a perfect record must be a good
driverso plants with high performance records practice risk minimiza-
tion strategies. Conversely, plants with spotty records are those that fail to
follow operating and maintenance practices that help to manage risk. In
the absence of such strategies, they suffer more losses.
Before you conclude that the better plants must be higher-cost
operations, note that insurance industry statistics and risk presentations
indicate otherwise. Steady, consistent performers are not only low-cost
performers but are safe, low-risk performers, as well.
After a rare failure event, managers need analytical support and the
implementation of failure prevention strategies. The notion that a big-
ticket failurethe equivalent of a meteor strikeis a random event that
happens only once in a lifetime, is not true. In practice, they keep hap-
pening, over and over. Generator retaining ring failures, water chem-
istry upsets, bus-bar explosions, shorted buses, plant trips, firesman-
agers tire from probing questions or site trip reviews.
After many years spent examining major losses, I find that in most
cases, a chain of events presents a history. The progression towards ulti-
mate failure depends on systematic process weaknesses. Rare events
occur more frequently in the absence of process awareness and controls.
138
Plant Needs
Figures 4-4a: Nuclear 4160V Breaker Failures
Figures 4-4b: Fossil 4160V Breaker Failures
They reflect random, repetitive occurrences that together convert to an

operating event. The more and greater their frequency, the more prob-
able they will do so. Ultimately, statistics tell the story. Rare events can
be managed with conscientious, complete operation strategies. These
rules are well known in theory and practiced by professional operating
organizations. ACRM is another way to add clarity to an otherwise clut-
tered operating field.
139
Failure frequency
After leaving nuclear and working in fossil power plants for nearly
a decade, I developed several strong impressions. For those who havent
worked both environments, theres a great deal of information sharing
that is possible. Each has very focused strengths that are applicable to
the other.
Nuclear is focused on identifying technical failures that add both
clarity and certainly to help those working in unfamiliar terrain.
Embracing failure realityversus abstract considerations of imagi-
nary problemscan greatly improve nuclear competitiveness at virtual-
ly no risk to the general public.
Fossil focus means using inherent design availability that is
built into plants to perform work as needed. The advantage is the abil-
ity to perform real-time maintenance; the risk is potential functional
failure because margins are expended. Fossil units easier start-up and
load cycle, for the most part, minimizes production losses incurred from
a forced outage.
The ability to mobilize personnel and systems to get a job done
is another fossil capability. Paperwork and organizational systems are
compact, focused, and anchored in a vested and accountable individual
or group. This focus supports the performance of CBM. However,
because fossil maintenance is less formal, operating limits are occasion-
ally stretched or overlooked and reactive failures or forced outages can
result. Defining clear operating limits to trigger condition-directed
maintenance is a fossil generation need. The opportunity for fossil
(unlike nuclear) is the authority to make individual plant interpretations
of risk and benefit when engineered limits are reached. This can pro-
vide great operating flexibility. There is absolutely no benefit when lim-
its are blown over and failures result.
On many occasions, fossil plant staff clearly understands key
operating limits from a technical perspective but organizationally, they
fail to act in a timely manner. Expectations were not made clear, or man-
agers failed to support operator decision-making. Again, the point of
CNM is to identify and perform CDM prior to final failure. To do this
well, those who perform monitoring must be expected and empowered
to act.
140
Plant Needs
Coded components. Nuclear plants have more coded components

than a comparable fossil plant. By way of comparison: 500 to 1000 MW
nuclear plants have 40,000 to 100,000 coded components while a coal
plant of the same size will code as few as 500. Arguably, fossil plants
have more complex equipment and systems, and plant coding effective-
ly ends at the skid level. Beyond this, identification is by text descrip-
tion alone, which is adequate for equipment identification and failure
analysis.
Fossil units are solely coded for maintenance and their CMMS
equipment descriptions and codes are typically the only ones available.
Design basis equipment tagging is usually absent or abandoned once
CMMS equipment lists are prepared. Nuclear plants, in contrast, main-
tain regulatory design databases coded for configuration management,
regulatory oversight, and equipment control. Fossil plants tag too lit-
tle and inadequately. Tagging systems should uniquely identify coded
equipment to the skid level using consistent descriptors. Perhaps a tag-
ging standards committee is needed. The ultimate answer is that bal-
ance is required. Excessive detail introduces complexities and costs no
one needs and presents a burden to use and maintain. Too little detail
means that costs and failures cant be adequately traced. My experience
at the nuclear plant was that about 5% of the equipment tags gener-
ated 95% of the MWRs. These items truly need unique identification.
Based on this assessment, nuclear plants are over-tagged for practical
operations.
As a practical matter, tagging uniquely identifies equipment for
operations, maintenance, and modification. Detailed tagging is required
if there are modifications that must be controlled. For normal O&M,
skid-based tagging systems used by fossil plants are adequate.
Complexity. Technically, nuclear environments are little different
than that of fossil or hydro. The risks are greater, but the complexity of
the controls in a modern fossil plant is actually greater. The slowdown
in nuclear plant construction has dated its technology. Fossil and CT
plants, however, have had a full decade of DCS applications that nukes
dont use. Yet, new technology introduces more inherently reliable
designs.
Technical barriers to nuclear advances are countered by the (rela-
141
tively) generous nuclear budgets. In a utility environment under regula-

tion, cost pressure on nuclear plant management is slight. With large
capital assets at risk, utilities have historically spent generously on
nuclear projects-sometimes starving fossil cousins in the process. Its
this generosity that has encouraged nuclear complexity in the regula-
tory sense; the NRCs insensitivity to costs has aggravated it and further
damaged nuclear competitiveness. There can be no other way to inter-
pret the tremendous growth of nuclear support infrastructure during
the past 20 years, at a time when the technology has, if anything,
matured and gotten simpler!
Organizational complexity introduces organizational errorsor
organizational failures, if you will. These virtual errors have
become the focus of regulatory interest as much as anything real. The
net effect has been an even greater complexity and more unreliable
organizational systems. Management focuses more on covering its
exposed regulatory backside than cost management while the NRC con-
tinues to operate as if there is no numerical threshold or objective meas-
ure of performance that objectively establishes pass/fail.
The U.S. nuclear industry was conceived and built to emulate the
philosophy of Admiral Rickover and the Navy in commercial form but
has failed to learn several statistical and cost lessons. Statistical and
numerical measures in the commercial nuclear industry are virtually
absent and this has unfortunately led to inherently higher structural
costs.
Complex failures
Complex failures in this definition include interdependent and
logical-sequencing faults involving equipment and control interactions,
multiple failures, intermittent failures, secondary failures, loss of redun-
dancy, and drift. Theyre difficult to identify, troubleshoot, and correct.
Analytical difficulty arises because many variable facets present them-
selves in concert. Each emulates the problems of a plant startup, in
which defining and solving coincident problems takes a thorough test
plan, expert assistance, and persistence. Teams and specialists are need-
ed to ferret out complex failures.
Avoiding complex failures lowers costs and increases production. A
142
Plant Needs
thorough maintenance strategy helps identify failures before they gen-

erate secondary failures and propagate into complex failures. The con-
verse is also true: When a plant gets behind its optimum maintenance
curve, failures occur, failure complexity increases, and a plant loses pro-
duction and increases expenses.
Troubleshooting skills are of great value in a plant suffering a high
population of failed equipment and complex failures but such skills also
raise costs. Timely, CDM performance and discipline reduces the num-
ber and complexity of failures, and ultimately allows a plant to be mon-
itored at an overall average lower skill level.
Serious secondary failures with complex secondary failure effects
fires, systematic contamination, and deterioration of sophisticated plant
fluid system chemistry, makeup water, and waste water equipmentare
often identified but consistently downplayed. Over time, they blossom
into serious problems that shorten the useful lives of equipment and
ultimately, facilities.
Fires, flooding, and environmental changes can introduce perva-
sive, common cause failures-a failure that crosses assumed independ-
ence boundaries. Once they take hold, common cause consequences are
difficult to correct. Because they cross boundaries they invalidate
redundancy. Avoiding common failure modes has many paybacks, and
implementing strategies to avoid them arises from R engineering and
from understanding events that are influenced by O&M.
Operationalizing. A professional society can develop standards
but only an operating environment can operationalize activitycon-
vert a subjective goal into a measurable performance objective. Once a
goal is operationalized, subjectivity is gone. For example, you could
operationalize a goal of performing well during an academic career as
achieving an overall grade point average (GPA) of 3.4. Once the goal is
set, you either graduate with a 3.4 (or higher) or you dont. This goal is
very measurable. Operationalizing therefore supports goal setting.
Governments operationalize tasks, as do companies and the workforce
that labors for them. Developing operational guidance and implement-
ing standards is a necessary step.
Many agencies implement standards that call for an effective PM
program. But what is meant by an effective program? What are its
143
attributes? What objective performance level is effective? How does

this come about? To comply, companies must:
determine what constitutes an effective program

sell it to others in the organization
set standards
implement those standards
assure themselves that they achieved implementation
This means there must be measures. If a program doesnt measure

up (by all accepted standards) then:
correction is necessary to remain within an effective program
When governments impose new standards, by law or regulation,

utilities must determine how to operationalize them. For example, the
Americans with Disabilities Act passed in 1992 and caused an outcry
from industry, not so much over the breadth of the act, but over the
need to develop an operational interpretation of the requirements. Now
that operationalization has occurredreasonable standards have been
developed and implementedthere is less concern over the law.
Slow-changing environments cant operationalize as quickly as
dynamic ones. As the utility industry becomes more competitive, those
who respond quickly will be more viable than those who cant.
Implementation is how work standards are operationalized.
Operationalizing tools include:
goal-setting
documentation
procedures
training
measurement
feedback
Common mode failures

Common mode failures denote failures that are common among
144
Plant Needs
classes of equipment, equipment in common locations, or other com-

mon factors. Using an inferior grade of grease that separates and accel-
erates aging, failures increase. Perhaps we now survey a few of the failed
values and find that the problem is really incompatibility of greases. I
had assumed all the lubricated valves would fail individually; now, I
have all valves loaded with grease and a common problem-a common
cause failure (CCF).
CCFs violate the design assumption of redundancy. While they are
independent failures, they are also programmatic failures, for the most
part. It is the intention of all design and operating rules (and particu-
larly regarding nuclear plants) to avoid CCFs. Common cause failure
modes substantially change the overall probability of functional failure
by negating design redundancy. The NRC worries a great deal about
them-and rightfully so! Fossil environments also display CCFs where
expected conditions changed, were never met, or were lost.
CCFs can show up as environmental problems, or as defective sup-
port system services. The recent year two thousand (Y2K) controversy
reflected a common cause failure made for software. Another classic
example involves the instrument air that operates many plant solenoid-
operated pilot valves and air operators. Instrument air contamination
has been a relatively frequent occurrence in industry, when moisture
and dustespecially entrained rustclogs service instruments, valves,
and air-operated controls. At low temperatures, air line freezing has
caused complete failurethe loss of instrument air and related servic-
es. Prevention is provided by drying the air, and air-dryness monitoring,
using moisture monitors.
A second CCF involves exceeding design service temperatures-in a
boiler enclosure. Ambient temperatures around one fossil plant boiler
ran 30 higher than ambient in summer and colder in winter. In sum-
mer, excessive temperature failuressootblower overload trips from
misalignment (due to thermal expansion), ignitor and burner-flame
scanner logic failures, and other control failurespredominated. In
winter, ventilation damper instrument air lines froze and instrument
drift was a problem. The common mode failuredue to the loss of
environmental controlcould be tolerated, but maintenance and oper-
ational expenses rose.
145
Keep it simple, stupid! (KISS): Maintenance is a complex

process. Simplification is key. To borrow a phrase from the military,
When complexity beckons, think KISS. Maintenance reverses
entropythe progressive trend towards disorder over time. As thermo-
dynamics students learn, a systems entropy can only be reduced with
the injection of energy or control. An outside source creates order
where disorder would otherwise prevail. Order is not the natural state
of things and achieving it requires continuous, constant renewal. That
injection of energy or controlor bothis necessary to be able to
maintain. It doesnt just happen; its hard work. It contrasts greatly with
the traditional organizational maintenance view of the world!
Its not that we cannot achieve order. We can. But we cannot do so
without:
thought processes
applied effort
providing each in enough volume to offset the inherent disorder
of the system
The KISS dictum says that by keeping systems simple, we greatly

reduce the capacity for disorder and simplify the effort required to
maintain order.
The very best maintenance performers often have very simple and
effective implementation processes. ARCM says, in so many words,
There is value in our underlying maintenance processes.
The Japanese have a phrase, poke yoke, which roughly translates
to make fool proof or make it impossible to get it backwards. So-
called poke yoke devices have long been used by the Japanese to sim-
plify production processes. Viewing American maintenance as a pro-
duction process, I believe that it needs many more poke yoke devices.
Intelligent engineering focuses on making every process a poke yoke
process and the key to more poke yoke processing is engineering main-
tenance production teamwork. That requires communication.
Ambiguity. Plant information is often incomplete. Parts usage,
equipment histories, WO entries, aging documentationall lack com-
pleted CMMS fields or are never developed, leaving us to make deci-
146
Plant Needs
sions with the best information available.

Blame some incompletions on information systems-many main-
frame-based systems disallow user entry correction so errors or
incomplete information cannot be addressed. Worse, many information
providers have never been shown how information adds product value.
(Information must be used if its to provide value.) Information fields
that can be retained may lack standards for entries. Because system
installations lack direct worker input, use, or records guidance, users
stumble along, ignoring the CMMS unless a timecard is tagged to it. As
a maintenance expectation, CMMS use, system training, expectations,
and information retrieval have been tailored to engineers, schedulers,
and managers. Requests for user-friendly systems have gone largely
unheard and applications for common user problems missed.
PMO facility review projects can be intenseand rewardingand
if they actively involve workers in development of PM intervals they can
include CMMS use. Quality and use can be improved with a few simple
tools. One is basing PMs upon statistically representative, adequately
large samples. Sample sizes can be extended by using multiple unit data,
vendor data, and overall industry data. Consider industry surveys and
vendor recommendations for new equipment when data is unavailable.
In the absence of unambiguous experience, its wise to perform
similarity analysis and benchmarking. If you cant find history for the
specific model of pump in question use another, similar manufacturers
pump. Although its not perfect, its a fast way to grow an effective
program. Skilled workers can suggest similar models, manufacturers,
and environments with which theyre familiar. Materials, like compo-
nents, can be estimated for inservice life with benchmark comparison.
Getting a similar, proven-life component from the environment in ques-
tion provides an efficient way to identify service life.
Again, to limit ambiguities, there is a continuing need for standards
based on equipment populations and composite environments that are
as large as possible. Standards fill gaps in specific information. When
information is incomplete, we must augment it and develop standards
based on age exploration that can develop a wider, more complete
experience base.
147
Culture
Maintenance delivery
Its my belief that American maintenance performance is disorgan-
ized. We cope with high rework rates, ignoring statistical (and other)
tools to identify, measure and reduce, (or eliminate) rework. Substantial
maintenance coordination and improvement opportunities need to
include ongoing:
continuous improvement
innovative jumps
American culture supports innovative leaps and always has. Process

improvement technology is an area where an active approach to per-
formance improvement can pay off.
A single-unit nuclear generating facility with upwards of 10,000
scheduled PM maintenance activities sees perhaps 10,000 jobs
worked annually. The sheer volume of items to track and coordinate is
overwhelming, even before operational constraints and other compli-
cating factors are considered. Complexity, though attributed to nuclear
plants, is typical of any large production facility. A refinery, a fossil
power plant, or a chemical process facility has comparable levels of
complexity.
Plant complexity also leads to work complexity. An I&C technician
troubleshooting a tank level controller requires the tank be in service
to perform the task. A mechanic replacing a valve needs the tank
drained first. Operations cant have the tank back in service until the
work is closed and the paperwork is complete. Such a clearance tag
out takes hours to prepare on the front endand sometimes gets lost.
Right hand/left hand stories abound.
Though the resolution to such conflicts is usually obvious in hind-
sight, its difficult at the outset to remember that large, complex jobs
require many different skills, schedules, awareness, and work condi-
tions. Coordinating these difficult pieces requires the most skilled staffs
and capable systems available.
148
Plant Needs
The alternative to choreographed work is taking every day as it

comesembracing each work activity separately, distinctly, and singu-
larly. This approach results in scattergun maintenance performance and
repetitive equipment downtime. Its frustrating when workers find that
conditions dont support the work, or have changed. Trenching a newly
paved street to replace a sewer is urban-legend folklore in part because
its a common occurrence.
Yet, tools to improve worker focus and work coordination are more
available all the time. Computerized maintenance planning promises
better coordination. The widespread availability of PCs, Local Area
Networks (LANs), and other electronic tools greatly enhance our abili-
ty to coordinate work. ARCM also promises better work identifica-
tionseparating the wheat from the chaff, so to speak. There is so
much equipment in a large plant, that just doing work doesnt cut it.
Work must be structured and focused to high levels. Until now, work
practices havent changed in part because environments supported
business as usual. We must first understand work practices to be able
to invoke meaningful maintenance changes and better serve plant
needs.
Delivery of better services, once understood, is another issue. One
great frustration milestone of my early RCM career was recognizing
that, by itself, RCM offered limited benefits. Just knowing what to do,
has limited value. Delivery is an equal partner to knowledge.
Knowledge has to be packaged into delivery methods and systems that
integrate with the organization to provide lasting benefits. Many organ-
izations simply arent ready for this commitment to change without
extraordinary external forces working on them.
What are those forces? Competitive pressuresand a growing
awareness that maintenance both directly and indirectly offers tremen-
dous potential for cost improvementshave changed this perspective.
In the last couple of decades, an awareness of RCMapplied in aero-
space as a successful technologyhas enabled the potential for bene-
fits-transfer to other areas. Indeed, there have been considerable bene-
fits developed from both the nuclear and fossil generation areas, as sup-
ported by EPRI. At the same time, theres a nagging feeling that RCM,
like so many other programs, has under-performed thus far.
149
Maintenance performance
Maintenance normally occurs within an uncertain environment.
Maintenance organizations often share information verbally, which has
limitations. Workers cope with equipment problems with varying
degrees of engineering support. Theres little documented, easily
retrievable information concerning equipment failure and many ways to
approach it. PM programs are implicitly defined and rarely have a basis.
Available PM information only implies the failures it addresses while
prescribing monitoring or corrective tasks. Few vendors specify (or
perhaps even know) exactly how to perform organizational mainte-
nance or address appropriate maintenance intervals.
Times and characteristics of equipment failure mode attributes
actual or idealizedare very uncertain. Yet, its from them that we
obtain mean-life, conditional probability of failure curves, distributions
of failure type, and mean life variation. Conservatism, built into main-
tenance task performance to compensate, could come from institution-
alized monitoring frequencies that are too tight. Theres also a lack of
trust in supporting systems and processes, including the computer
maintenance management/information systems. Many facilities com-
pensate by over-performing maintenance.
Some conservatism arises from the very nature of large industrial
maintenance and the crafts inherent desire to do good work. Part of it
stems from the lack of effective CMMS PM systems. However, a huge
part of the problem arises from the uncertainty of equipment lifetimes
and use. Combined with a TBM modelthe traditional PM model
(assuming it was performed)means huge amounts of conservatism
have to be built in.
Maintenance doesnt need to be random, nor must there be so
much of it. In disorganized facilities, factors working together to help
control failures include conservatism, craft, design, and monitoring.
The tendency to maintain wide margins for error (on the assumption
that things will be missed) adds tremendous conservatism to part-life-
time calculations, randomness of failure assumptions, and other main-
tenance program features.
Craft workers in a stable working environment learn equipment
150
Plant Needs
needs and deal with equipment directly without organizational system

intervention or support. In some cases, they fight disorganized control
systems, succeeding with traditional techniques and skills that evoke the
meaning of craft. Insights shared among operators and maintenance
crews are intuitive, perceptiveand valuable. Supporting systems that
should benefit from this information, oftentimes do not. A PM program
may fail to capture many insights known at the worker level.
Great conservatism is intentionally incorporated into equipment
design, as well. To reduce the risk of equipment overload, use, or pre-
mature failures in uncontrollable environments, designers build in mar-
ginsexcess capacity. Users become aware of these margins (particu-
larly in fossil and hydro facilities) by pushing limits. Equipment toler-
ates these stresses or fails. Over time, practical operating limits are
established based on experience.
If we dont have (or require) huge conservatisms:
costs drops
production improves
waste drops
R improves
The case for improved R is therefore counter-intuitive: Less margin

forces us to operate closer to design limits. This places additional stress
on blading, tubes, casings, and other long-lived hardware. Design is
based on more carefully specified conditions. Increasing operating lim-
its without such insight is more likely to generate unacceptable per-
formance. These days, owners increasingly foot the bill for lost per-
formance.
Equipment groups
Work association (also known as work blocking) can speed and
streamline maintenance performance. Associations can occur at the
task, equipment, or systems levels among equipment, function, or
boundary groups. Such opportunities arise when its convenient or
mandatory to work on elements within an equipment group.
The basic objectivesimilar to the objective throughout industrial
151
manufacturingis to minimize work performance time and labor.

Maintenance time analysis shows that trip, part, and planning time rep-
resent significant amounts of total average maintenance work perform-
ance. A major maintenance performance cost factor in any large com-
plex facilitypower and chemical process plants, factories, or even
transmission systemsis the trip that is needed every time work is per-
formed.
Creating equipment groups (EGs) and work associations supports
the systematic reduction of trip time and allows coordinated corrective
and preventive work to be performed both within and between craft
groups by appropriately linking tasks. Associations take advantage of
whats been learned in previous work performances and support con-
tinuous learning. This can be lost in environments where the emphasis
is on doing work, not making money.
For repetitive, planned workespecially PMsits especially con-
venient to park activities on the plants schedule within a scheduled
EG. Once work is associated, you avoid the need to schedule it on a
detailed level within the group. This avoids PM realignment.
When performing a number of PMs, lining up intervals within and
across groups speeds performance. If you make mistakes-if work gets
misaligned and performance complicated-problems can be identified
and worked around. Problems occur, day in and day out, in any sched-
uled maintenance program. Rescheduling misaligned PM work
becomes quite a useful skill in a PM-oriented plant. Of course its bet-
ter to minimize the causes of misalignments so that few occur.
Grouping, scheduling, and man-loading around a 12 week LCM sched-
ule is crucial-and then working to the plan!
Routine work alignment can be based upon electrical divisions, for
instancethe rotating 12-week quarterly schedule and the plant sur-
veillance test schedule. A routine surveillance plan is required at
nuclear plants and at fossil plants with test requirements. However,
every plant has an ASME code and insurance test requirements and a
list of other required monitoring that is often specific and lengthy
even if its environment is considerably simpler. One spin-off benefit of
surveillance-level scheduling is the absolute need to develop and man-
age a plants 12-week schedule to meet intensely monitored nuclear
152
Plant Needs
license requirements. This schedule can offer benefits to all plants.

Where available, such a schedule provides a ready vehicle for a
monitoring test program such as developed by an ARCM-based failure
review. This benefits any large, complex facility operating and mainte-
nance process. A CMMS can provide the schedule software ticklers
that initiate the plan. The 12-week schedule offers a near-term window
that fits with other scheduled plant activity to become a tool that allows
the plant to perform moreand betterplanned work.
CNM
Most CNM-initiated maintenance originates in operations. CNM-
monitoring without specific failure criteria-can be hard to rank, priori-
tize, and perform due to its generality. In the absence of time-based and
on-condition WO categories, an organization can measure its CNM-
originated work, based on the work-fraction coming from operations. If
an operation originates 70% of the WOs unrelated to operational tests,
then about 70% of them are NSM. Scheduling and planning, and engi-
neering, initiate most of the balance of the outage, PM, and modifica-
tion WOs.
TBM comprises the planned maintenance that is traditional, and
time-based rework/replace task work. If a plant can identify condition-
based from time-based WOs, they can measure the RCM maintenance
workload as as shown on Table 4-3.
A small fraction of CNM identifies functional failures. Measuring
that fraction involves (1) reading WOs or (2) checking logs. Few
CMMSs have fields to record functional failures (FF) and few opera-
tors discriminate functional from other failures. Logs typically record
functional failures.
A quick way to re-align CMMSs to measure RCM-based work strat-
egy is to relate CDM to on-condition WOs. You can also perform all
condition-directed work as part of the original on-condition WO.
This establishes three basic WO classes:
This approach provides a quick way to measure existing processes.
153
Now, what do the numbers mean?

Benchmark profiles are only now being developed in the power
industry. Grouping numbers in this manner can show absolute WO
numbers any way desired, but hours worked is a common benchmark
comparison quantity. PM work hours are inherently low. Most non-out-
PM (time-based) (1) TBM
(2) OCM/CDM (including OCM FF/CDM)
CM (corrective) (3) NSM/OTF (Failed)
Table 4-3: PM vs. CM and WO classes
age PM jobs are simple tasks.

Organizations generally need to increase their on-condition work
fraction. This work involves an explicit failure resistance measure, which
initiates condition-directed work. This requires explicit failure limits,
work performance focus, and priority on work with a failure limit
exceeded.
This combination is often seen in traditional instrumentation pro-
grams, where a significant amount of out-of-calibration and failed
instrumentation work is identified. When restored, often at low cost,
immediate operational R improvement resultsa quick payback.
Two Perspectives on Failure

There are two failure perspective focusesfunction and compo-
nent. The functional perspective expresses what a component does, the
component perspective, and how it deteriorates. Part names have func-
tional rootsblower, pump, and breakeroften derived from a verb.
The name suggests the primary function. Because the perspectives dif-
fer, remembering the context whenever failure arises can help avoid
confusion.
When components, equipment, and systems fail, ultimately its
because a physical part deteriorated. This proximate part degeneration
eventually causes a functional failure. (A bearing wears until a pump
trips on high vibration, a motor winding resistance fails, the motor
154
Plant Needs
shorts, and trips the pump off, or an operator smells smoke and trans-
fers pumps, shutting down the offending pump.) Functions affect work
performance. Failures translate as lost functions. The operators com-
ponent is a black box that he or she may not understand. They only
need to see the functional outputs or note their absence and act appro-
priately (Fig. 4-5).
We require functions while operating plants. When functions
break (e.g., are lost) we diagnose failures, locate the source, and fix
parts. Operators perspective is inherently functional. But while identi-
fying a functional problem is one step, tracing that back to its physical
source is another matter. Success in managing failures depends on orga-
nizational diagnostic skills (Table 4-4).
Holding a functional perspective simplifies the operators required
equipment knowledge. Operators need only assure function availabili-
ty-which involves the senses-and interpreting instruments. Facility
instrumentation supports function monitoring, and the specific func-
tion, measurement requirements, and equipment redundancy deter-
mine the instrumentation needed.
The function-part failure dichotomy is important when we talk
about failure and operate to failure. There are few (if any) cases in
which plants intentionally operate to system functional failure. This
simply makes no sense. We provide robust equipment redundancy
specifically to avoid it. In the hierarchy of systems, subsystems, and
their functionality, however, redundant or incidental functionality is
provided at subsystem (or lower) levels that can tolerate failure, to some
degree. Risk accompanies function failures, but it can be managed.
NSM is meaningful on components where a failure will be evident,
can be managed, or has no functional impact. Redundant instruments
packages, inexpensive components, and even spare trains and equip-
ment support this approach. If redundant equipment can be run to
failure while maintaining system functions, the deciding factor is cost.
Sophisticated microprocessors and sensors can identify and shutdown
deteriorating equipment, limiting damage. The cost is loss of the
equipment until maintenance is completed. This strategy is viable for
wearout failures where there is installed redundancy (Fig. 4-6).
Consider a boiler feedpump in a 50% redundant train (three 50%
155
Figure 4-5: System Black Box Model
Component Functional Failure Part Failure

Pump Wont pump Seized bearings
Blower Low volume at speed Worn impeller blades
Breaker Cant extinguish arc Weak coils
Pump Wont pump Bad starter
Pump Wont pump Lost prime

Table 4-4: Component, Function, and Part Failure
pumps, any two of which provide 100% rated flow). (Fig. 4-7) This con-
figuration is a standard plant feedwater design, and meets boiler head
requirements for four to seven years of service. This approach is viable
and effective, provided the standby-train pump can start and load reli-
ably. Such assurance can be provided by periodic testing. When in-serv-
ice feedpump failure is identified, capacity is shifted to standby. The
worn-out, failing pump is removed from service and repaired. This could
be online or off-line, during a scheduled outage. Although equipment
must be restored, the systems functions are maintained (Fig. 4-8).
OTF as described here is a rational. We need to remember that the
failure considered here is an abstract engineering proximate function
156
Plant Needs
Figure 4-6: Two Sides of Failure Management: Risk and Cost
failure. (Fig 4-9) For many traditional engineers, this is not their per-
ception of failure. The function-equipment-failure seesaw makes OTF
confusing for some. For an operator, maintenance is of no consequence
so long as they always have necessary (or backup) equipment available.
OTF means little as long as black box system functions work.
This approach may not set well for the mechanic, however. OTF
must conserve equipment or economic consequences make it unreason-
able. Catastrophic failure fears explain why many mechanics object. In
fact, a great deal of equipment is designed to support an OTF strategy.
Internal sensing devices initiate shutdown on fault conditions causing
function loss. This limits equipment damage, but sacrifices functionality.
Cases can arise where sacrificing equipment for extended functionality is
preferred. Operators make the choice.
This function-to-physical failure mode relationship is summarized
with Figure 4-10. Functional failures observed using a system perspective
are the result of physical part deterioration. Functions can only be
157
Figure 4-7: Boiler Feedpump in a 50% Redundant Train
restored by addressing the failure mode of the physical part level.
158
Plant Needs
Figure 4-8: Failure Description
159
Figure 4-9: System Part Functional Relationships
Figure 4-10: Failure Progression
160
Chapter 5
Applications
Rule 1: Fly the airplane.

-Pilot saying
Overview
Operations
Generating plants employ upwards of hundreds, even thousands, of
direct employees. Plants are complex, with complex needs. Theyre
built with production capability and system-support roles in mind and
with the outright goal of making a profit for their owners while keeping
costs low.
This mission comes into play even before generating plants come
into being. Decisions on siting, project management, and other issues
depend on it. Once a certificate of necessities and benefits has been
issuedquasi-governmental authority allowing cost and earnings
recoveryand project construction proceeds, the utility traditionally
engages an A-E who develops plans and specifications based on needs.
With utility guidance, the A-E refines plant objectives and performs
initial scoping of required systems, their capabilities, and other plant
161
requirements. A design takes shape as goals and objectives are reduced

to paper specifications. Plant layout and supporting systems are based
upon years of experience and proven practice. The A-E uses previous
designs and experience for reference, but each new facility is a new
design and guidance from the utility takes precedence over the A-Es
previous work. Even when a plant is completed, new units are added,
one at a time, as loads grow. The focus is on initial cost, so standard-
ized plant designs are rare. Even common equipment such as soot-
blowers and boiler feedpumps differ on units adjacent to one another.
Yet, for operating organizations, unique designs increase the complexi-
ty of operations. Why is it done this way?
Unit design supports high-level operating roles and goals. Plants
dont just make electricitythey support company production goals,
filling multiple roles in a complex generating pattern that includes sea-
sonal load management, weather, system disturbances, long term pur-
chases, and other unknowns that have to be taken into consideration.
These roles are defined during the approval and design phase for any
proposed plant.
Over the life of a facility, roles evolve, dramatically change, or even
end. Some are identified that werent originally anticipated. Conditions
also changebusiness and political, as well as technical. Virtually all
fossil-fired boilers have been retrofitted with emissions monitoring and
control equipment to reflect environmental laws. After the accident at
Three Mile Island, the nuclear industry radically changed as the NRC
required major plant modifications.
When operating staffthe people who actually run the plants
understand current plant roles, they are better able to focus and man-
age competing needs. But operating staff see only part of the plant oper-
ations picture. Changes in mission, company production goals, regula-
tory intervention, load shifts, plant aging/obsolescence issues, fuel cost-
many factors that change over time are beyond operators direct influ-
ence. To maximize operating returns, operators must understand goals
and mission or their performance effectiveness is undermined.
Operations personnel operate the plantreconfiguring for the
mode of operations (full-load, regulation, part-load, or shutdown).
Reconfiguring the plant includes tagging out systems and equipment for
162
Applications
service. But operations is also responsible for plant CNMplanned

time-based equipment monitoring on rounds, as well as non-specific
general area monitoring performed during plant operations and rounds.
Operators initiate and prioritize much of condition-directed plant
maintenance. Most originates from a functional text schedule. Lastly,
operations is responsible for the material condition of the plantclean-
liness and general safety.
Overall, its a huge task list. In a general wayand only in a gener-
al way!plant training provides skills to accomplish these tasks. No
group does it perfectly, but those that do it well receive a wealth of oper-
ating benefitsincluding predictable cost and operations.
Balancing department goals

Station operating departments that deal with plant performance,
budgets, personnel, stock, services, and many other cost decisions on a
day-to-day basis can easily sub-optimize operations. For example, a pur-
chasing agent may want to minimize parts cost while a mechanic wants
a quality part from a specific vendor. These objectives potentially con-
flict. If the agent is unaware of the quality of different manufacturers
parts, he may second-guess the mechanic and substitute an equiva-
lent.
O&M goals may also conflict over the performance and timing of
maintenance, prioritization of work, work standards, and a host of other
issues.
Value added. Everyone on staff either has a direct plant support
role or an indirect service role. The former roles include operators and
mechanics who keep the facility running, as well as onsite engineers,
technicians, and others engaged in maintaining operations through
direct equipment support roles. The latter category includes clerical
staff, management, and off-site support staff. They enable those in pri-
mary roles to do their jobs. When an individuals contributions support
either role, they add value.
Engineering support role

Engineerings role is to enable primary workersoperators and
maintenance staffto perform their work more effectively. Plant engi-
163
neers support two major functions ongoing maintenance and opera-

tion of the plant, and plant re-design for improved performance or cost.
The latter role ties to original plant design and construction, but
includes services such as redesigning parts for life-extension. Most
engineers understand their traditional role as design/build. Few are
specifically trained for supporting plant O&M roles, which must be
understood to be effective. Traditional design-construct engineers
struggle with this issue.
What type of engineering support is needed? How much interven-
tion from plant operations is appropriate? Who plays Solomon when
operations and engineering goals differfor example, in plant operat-
ing envelope specifications? Competitive generating companies with
strong engineering cultures are able to establish operating and mainte-
nance standards that benefit both R and cost, reducing undesirable,
unexpected events.
Plant engineering fills the operations-design interface gap. Effective
system engineers require people skills, operating, and maintenance
experience, and general engineering competence, supplemented by cost
awareness and computer information management skills. Skilled system
or plant component engineers favorably influence plant operations by
reducing operating costs.
Instrumentation and control (I&C)

I&C groups calibrate, test, and maintain plant controls. Without
controls, the plant doesnt runparticularly those with newer DCS sys-
tems. I&C also provides the operators instrumentation window on
the plant, whether by traditional analog instruments or more current
digital or DCS display control screens.
I&C evolved out of maintenance, as a work specialization but I&C
remains a very critical maintenance role. I&C technicians in a sense are
super operators who know a plant functionally well enough to tweak
its controls without trips, yet fully understand the technical details of
their trade.
Malfunctions in I&C can cause plant trips and other undesirable
events but its not a black art. Controlling the risk associated with I&C
activity involves procedures, routine practices, and standardization of
164
Applications
equipment that are planned, developed, and utilized. The real potential
of I&C lies in improved availability. Cost reductions arent an especial-
ly promising or even desirable goal (except perhaps at nuclear units).
The tedious, time-slugging work of disassembly, rework, and reassem-
bly of major equipment-the traditional mechanical maintenance role
is absent, because I&C hours could greatly increase or decrease with
small impacts on overall costs. R comes from reliable instruments and
controls and for plant R, I&C holds great value. Direct I&C influence
on other areas is slight.
Understanding the factors that cause trips, and improving instru-
mentation until it plays no role is the major I&C goal. Instrumentation
R and availability is a significant concern for operations. Operations
and I&C must work closely.
Other players. Traditional mechanical, electrical, and I&C mainte-
nance is supplemented by welders, insulators, and specialists such as
non-destructive evaluation technicians, vibration analysts, direct-sup-
port engineers, and janitorial staff. As it fulfills its primary role of imple-
menting time-based and condition-directed programs, maintenance
must also coordinate with specialist and contract maintenance groups
brought in for special jobs and outages.
Maintenance holds the greatest influence over costs, through
planned and outage maintenance programs and budgets. Because of the
time-intensity of any major disassemble/reassemble work, maintenance
has tremendous leverage over operating O&M cost. In a forced outage,
or a delayed return-to-power situation, traditional maintenance costs can
increase with few questions because the value of lost generation is great.
Engineering
Operations: organizational relationships
Engineering, operations, and maintenance have historically been
distant cousins. Engineering performs design-build roles. O&M run
facilities. Their interactions were usually limited to day-to day operat-
ing issues. Engineering provides project management support for large
modification projects but it routinely works alongside operations in
plant support. Fossil plants may have two or three onsite engineers who
165
provide minor project support, controls, problem analysis, and special

plant needs but an organization in which engineering directly support-
ed plant operations is more exception than practice.
But the absence of a working relationship can cause coordination
and support problems. Special arrangements address outages, typically
two or three engineers supporting a large, multi-unit plant. The obvious
gap is that of ongoing, structured engineering support for maintenance
and operations. To fill it, O&M call upon their own resources.
Engineers from corporate engineering to plant engineering groups usu-
ally hold little operating background. Plant issuesCNM technologies,
controls, failure analysis, and maintenance supporthave to be learned
informally along the way. Companies rarely have specific job descrip-
tions, culture, or measures to identify plant support expected or meas-
ure how effective it is. Consulting engineers continue to fill out plant
support engineering ranks.
The exception is in nuclear plants, where adequate engineering
resources are mandated by law.
Yet, virtually every plants operating life begins with design prob-
lems that cry out for engineering involvement. Stations develop strate-
gies to cope with all sorts of high cost problemsnon-functional
equipment, failed instrumentation, analysis and introduction of new
methods, materials, or equipment to reduce costs. At some point the
size and scope of maintenance efforts increase to where facility re-pow-
ering is more attractive to manage long term costs. How do RCM, main-
tenance, and engineering combine to address these problems?
Traditionally, maintenance engages design engineering assistance
during new facility startup. After that, engineering supports corporate-
initiated changes such as new technology or replacement of large, exist-
ing equipment and facilities that are worn out, such as cooling water tow-
ers, circulating water tunnels, or large equipment foundations.
Maintenance problems given to design engineering staffs often lack
problem definition. Design engineering traditionally accepts project
requests regardless of projected payback. Cost justifications for many
design changes simply arent available before design engineering initi-
ates fixes. Management-initiated design fixes are often made in
response to regulatory, safety, or other cases without adequate research.
166
Applications
When engineers design fixes, or engage contractors, or incur other

expenses, its often without adequately understanding the problems,
their causes, options, value added benefits, or costs. The combination
of regulated environments and traditional engineering aversion to cost
awareness has combined to allow this kind of project evolution.
This is where RCM can help design engineering groups-enabling
them to add value by improving the design request prioritization
process. Analysis of failure-mode statistics identifies those plant prob-
lems that have design-only solutions, the kind that need to go to com-
petent plant engineers for resolution.
The flip side is that design-R features assist maintenance perform-
ance and support overall plant R.
As the industry continues to deregulatewith no additional capital
to spendthe single biggest opportunity for generating companies and
engineering staffs will be the preservation of assets. Unprofitable (or
marginal) assets in the competitive environment will need assessments
to identify their best options. Re-powering or topping cycles may
improve basic production costs. Unreliability will present opportunities
for improvement. Under-performing assets (based on competitive
benchmarks) will benefit from R engineering in concert with ARCM
O&M programs. They will be the quickest route to improved perform-
ance and lower costs.
Many facilities have been maintained with homegrown modifica-
tions over the years. Some of these plants will gain immediate improve-
ment by having basic hardware unreliability issues identified and
removed. In cases where new units were added with little consideration
to R or common services, complex and hard-to-run systems resulted.
Older, multi-unit plantssome hosting different models and vintages of
equipmentadded on equipment that compromised redundancy or
added tag-out complexity issues. Resolving these problems can lead to
quick improvements in performance at little capital and minimal oper-
ating expense.
Plant engineering support roles

When it first became clear that complex power plants need ongoing
plant engineering support, the nuclear plants established system engi-
167
neers to serve as system managers. They support operations with fail-

ure analysis, specific system design interpretations, modifications sup-
port, failure analysis, and many other useful functions. Ideally, fossil
units have someone with the same capacity and training.
In practice, nuclear systems engineers focus on regulatory compli-
ance but their ideal role is to improve system work, operating process-
es, system R, and lower system costs. An effective plant engineer, system
or otherwise, is someone who masters subjects not taught in engineer-
ing curricula-failure analysis, cost engineering, maintenance, controls,
R, and operations. Procedures, cals, test programs, and general industry
requirements take time to learn, but an operators role must be learned
over time. Learning to provide real-time support to O&M doesnt come
from sitting behind a desk pushing papers, but rather managing per-
sonnel at the plant level.
R engineering theory and RCM provide excellent guidance for plant
engineers in this endeavor. Developing the skills and capabilities neces-
sary to perform RCM, in streamlined format, can provide guidance for
support engineering groups. RCM requires engineering, technical, and
plant support competence. Plant support includes failure analysis, oper-
ating procedure analysis, poke yoke (human factors), engineering
simplifications, maintenance support, process improvement, I&C
understanding, measurement, cost and performance improvement
awareness. Its a tall order in anyones book. But its so importantand
those who do it well are so fewthat companies need to develop new
job descriptions and training programs to ensure it occurs. Capable
plant engineers are required at any facilityno matter what type. This
includes steel mills and food processors, not just generating stations,
and includes whatever title by which theyre currently known (plant
engineer, production specialist, services engineer, maintenance engi-
neer, project engineers, application engineer, and so forth).
This will not be easy, however. For utilities, the need is great, the
position is new, but organizational inertia is a barrier to anyone seeking
to fill this role. That inertia exists because utility generation has been
organizationally static for 40 years. The last great change came from the
nuclear units special regulatory, safety, and operations support needs-
the change that differentiated generation into fossil and nuclear. The
168
Applications
latest changesthe proliferation of gas-fired CTsmay again split the

industry. Deregulation and convergence with the gas industry will
fuel additional changes organizationally. Fossil generators may also
find battle lines drawn over issues of high costs created by re-regulation.
These battles can be best engaged through improved engineering
functions, for engineering traditionally improves the product.
Improving generation processes reduces costs and increases safety as it
increases generation. In the 1950s and 60s, the generation industry
thanks to more efficient processes and facilitiesenjoyed reduced
costs, improved product, better customer value, and ultimately further
industry growth. Todays CT technology is continuing along this path
the primary reason why new CT orders keep coming. Gas supply looks
adequate to support much more electric conversion to gas. In the mar-
ket-driven environment into which all analysts say were headed, mar-
kets will decide these and other issues. New opportunities for plants
and support personnel await those who can put engineering improve-
ments to work.
Plant modification
Meaningful improvements will not come cheaply because they
depend upon plant design modifications and such mods are expen-
sive. Those initiated within the plant tend to cost much more than esti-
mates suggest in my experience. Minor modifications managed on-site
often have the lowest level of control and stand as the worst offenders.
In concert they add up to a burden on operating budgets and available
staff. When the final numbers are in, such projects can cost more than
whats budgetedabout 10 times more, throughout my utility work
experience, based upon final-cost figures for many minor design
changes using a cost-accounting system that traced charge numbers to
jobs. Given that original cost benefit, justifications (where utilized) were
based on estimates that were a factor of 10 low, it stands to reason that
there must be a significant volume of design work of marginal value
or, more likely, of no tangible value when the goal is reducing unit oper-
ating costs or increasing generation.
Improving the design change screening process will thus have great
paybacks. ARCM can do just that. In RCM task selection logic, design
169
modifications are the last choice when there is no effective PM that can
be done, and failure cant be tolerated. In fact, these cases are rare.
Effective PM translates to technically effective, a case agreed to
by experts addressing a failure mode. It points to fundamental design
flaws uncommon in production components and equipment. More
commonly in these cases, maintenance fundamentally misses the mark.
i.e., the task performed has no applicablemuch less cost-effective
basis.
Until its proven that a design change is required, redesign is a cost-
ly proposition. If a maintenance solution is at hand, however, savings
and benefits will be substantial. To make this point, you must have done
your homework, and there must always be analysis on which to base
design changes and value. Formal RCM analysis provides the basis spec-
ifications for redesign.
Another common organizational weakness is the failure to pass
design-developed equipment assumptions (and support requirements)
to the facility operating and maintenance staffs in a manner useful to
them. After problems arise and designs are reviewed, it often becomes
apparent that plant management and engineering staffs never connect-
ed on procedures, training, drawings, or other key aspects of what was
supposed to be a joint effort. From my experience, in about half the
cases engineering did provide the product, but it got lost at the plant
level because the plant lacked the infrastructure to use the material pro-
vided.
Its hard to recall faulty designs, and so developing thorough failure-
based maintenance plans effectively identifies areas that can truly bene-
fit from design. Such reviews ensure that operating and maintenance
problems at the plant level get corrected at the plant levelwith little
or no engineering assistance-before going to the design engineer. In this
manner, step improvements occur in O&M. Operating groups
improve their understanding of plant design specificationsand limita-
tions. Plant operators better grasp design and operation factors
required for success. RCM considerations assure that design requests
are those that design personnel and processes can and should legiti-
mately address.
There is also value in having engineering staff work on product
170
Applications
improvement. O&M staff can identify costly failure problems, develop

alternatives (including the basic maintenance program strategies and
tasks), and support age exploration. Design engineering can focus on
mysteries that arent understood or require re-design. One inevitable
consequence is that engineering and maintenance draw closer-focusing
on facts, on quantifying costs (and benefits), and on supporting equip-
ment needs.
Engineering tools
A number of engineering tools provide R analysis for generating
units.
Hand-calculated until just a few years ago, R analysis was generally
not applied to complete designs. Instead, thumb rules, benchmarks,
and standard solutions were applied. Today, personal computers (PCs)
and specialty software offer greater capability to evaluate detailed
designs for availability, R, risks, and other life cycle R aspects. Many
analyses tie directly into plant operations. RCM evaluations support
implementation of the unit planned maintenance program. Other prod-
ucts provide similar services. For instance, Markoff Analysis can be
used to evaluate conditional probability of failure when important
equipment is OOS for maintenance.
In a deregulated environment, with many new plant and equipment
designs emerging, capital investments are put at greater and greater risk.
This increases the need for R tools for these design assessments. Here
are some of the best.
FMECA. A complete RCM analysis begins with a failure modes
and effects criticality analysis (FMECA). ARCM limits analysis to the
major hitters that can be identified and used, based upon experience.
ARCM/RCM for an existing facility is an a posteriori assessment
experience limits the scope of the review and focuses on value. New
facilities can be reviewed using a priori RCM, utilizing a variety of for-
mal R engineering tools, including FMECA. Projections of likely prob-
lems, availability, and maintenance costs can be generated based solidly
on analysis.
Analytical FMECAs have been used for years in aerospace applica-
tions to zero in on risk contributors and manage overall risk on a budg-
171
et. This analysis logically fits capital-intensive, single-mission design

applications that characterize the space program and many high-risk
military missions. Government and general specification MIL-STD-
1629A provides standards for FMECA preparation. Software to meet
these standards is available commercially.
Fault trees. Fault trees, like FMECAs, focus designers on weak-
nesses in a developmental design. They allow detailed assessments of
alternative configurations and any likely failures. Fault trees can be used
not only to assess designs, but as a corrective tool to assess existing
applications. As a side effect, they provide excellent risk-management
tools for operator training. (Fig. 3-9, page 106)
Availability simulation. Although applied primarily for design con-
sideration, availability simulation can project the impact of redundancy
loss during major maintenance. Many times availability simulation tells
a story better than mathematical analysis. If maintenance schedulers can
see risk impacts, they can more wisely schedule maintenance. Often the
design aspects of redundancy are not conveyed in ways useful to sched-
ulers. Simulation results can fill that gap. Again, software products can
perform this work.
Weibull analysis. Weibull analysis models failures into a Weibull
distributionthe most generally available failure distribution (in the
sense that both infant mortality and aging can be modeled). Processes,
programs, and equipment can be tested for infant mortality and for ran-
dom or lifetime aging failure behavior. Failure distribution can then
suggest strategies for corrective measures, particularly as they relate to
maintenance performance. More formal failure and design-out engi-
neering can be included.
Weibull analysis is of particular interest to organizations evaluating
suppliers and parts. Parts specifications can require a Weibull distribu-
tion evaluation and multiple parameters. Weibull analysis provides
excellent information about the performance of parts in-service. This
includes:
infant mortality failure rate and period

normal service period duration and residual random failure rate
expected life
dispersion of expected life
172
Applications
Weibull analysis can provide a competitive edge to parts suppliers.

As a plant engineer, I would have much greater faith in a product that
came with a Weibull specification than without.
Integrating The Big Three

Three basic functions are necessary to operate facilities over time-
operations, maintenance, and engineering.
Operations is the first supporting leg in every plants mission and
implementation. Unless theyre truly remote units, plants dont operate
without operators. Even remotes have dispatchers!
Maintenance includes all those organizations generally engaged in
maintaining the facility, including chemistry. Little-m maintenance
includes direct service roles such as scheduling and I&C. (Cost account-
ants call them direct labor.) These people turn the wrenches and per-
form tangible work. Their direct supportplanners, schedulers, and
dedicated clerical staffis also maintenance staff. Engineering owns
plant design, design improvement, and design-cost reduction over time.
These three groups influence, to a large degree, overall plant per-
formance and competitiveness within the cost constraints of supply and
demand and corporate structure. They determine the degree of pro-
duction success at any particular plant. This is where RCM improve-
ments live!
Operations roles
Failure identification. An operations staff is primarily responsible
for plant condition. However, operating staffs own the plants, to vary-
ing degrees, and so failure identification is a legitimate responsibility.
Recognizing failure requires knowledge, experience, skill, tools, and
failure standards-a perfect fit with operators plant-monitoring assign-
ments. Failure identification, as a rule, is sometimes assumed, over-
looked, or taken for granted. Again, operators have the abilities and the
obligation.
Operations spends more time than anyone else in the plantread-
ing instruments, operating equipment, feeling vibration levels, smelling
fluid leaks, hearing noises, and seeing how things do (or dont) perform.
173
They are naturally suited to recognize changes, identify faults, and initi-
ate correction.
Successfully identifying failures depends on experience and skill.
Some operators receive excellent training, either during career develop-
ment or prior to hire, while what others receive is very limited. Effective
operators in a competitive environment need higher-than-average skill
levels. Turnover increases training requirements. Nuclear plants have
excellent training programs because of license and industry standards
while fossil plant training is more on-the-job, hands-on, learn-as-you-go.
Both methods have their place. Training needs to be cost-effectivein
fact, measurement for cost-effectiveness is a training need in itself.
About 80% of all failure-identification tasks originate with opera-
tors, based upon RCM failure analysis. That is: fully 80% of all RCM-
based maintenance involves operator monitoring! In a CNM program,
then, maintenance starts with operations. Because operations monitor-
ing is so pervasive, failure recognitiona key feature of effective main-
tenancebegins with operator training.
Two primary operations tasks are CNM and functional testing.
CNM uses the senses and instrumentation to identify equipment
failure and failure trends. Functional testing for hidden failure func-
tionsalarms, trips, and other protective or standby devicesassure
function is preserved.
Nuclear plants wont discover large, available benefits from
increased functional (surveillance) testing because they already have
extensive surveillance plan requirements based upon their licenses, and
they generally have excellent availability. Fossil plants, however, may
find major gaps in their testing and equipment protection plans. Many
fossil surveillance plants are informally controlled and miss critical
and essential instruments and alarms. If implemented, these can assure
design conditions are met.
The second aspect of the operations monitoring program is the test-
ing program. Essential alarms and trips are typically tested on the
largest equipment in both nuclear and fossil plants. These include tur-
bine trips and vibration trips for large ID and FD fans. But other, less-
er alarms dont get testedchemistry out of spec condition alarms,
for example. Some critical alarms occur in remote locations and may not
174
Applications
go to the main control room. Calibration and testing programs for these
alarms implies value and perceived importance. Hard (unit trip) cal-
ibration limits are frequently neglected. Operations and engineering
personnel often interpret alarm values substantially differently, as if the
two arent reading the same set of guidelines.
Some utilities intentionally minimize hard trips when an equip-
ment supplier provides a hard-wired trip or status alarm (on high
vibration, perhaps) and the company uses a status alarm. The operator
then acts on the alarm appropriately. Such vibration status instru-
mentation is installed on virtually all the main turbines at one Midwest
utility I know of. This approach undermines effective instrument main-
tenance. Their position was that, We dont want any trips to occur due
to sporadic alarms. They expected their operators to interpret ambigu-
ous instrumentation from the same erratic instrumentation that no one
wanted hard-wired for trips. There was a R problem with the trip
instrumentation that the company was unwilling to address.
How an operating company addresses instrumentation indicates
much about its operations philosophy. In the case of critical instru-
mentation, critical has two connotationsRCM and common usage.
RCM is a direct safety consideration, common use is subjective, inexact
intuition. Ambiguous instrumentation guidelines indicate unclear man-
agement philosophy. An RCM-based instrumentation review can help
management select the instrumentation and limits for clear action.
OOS, uncalibrated, or otherwise unessential instrumentation
abounds in a typical plant. A vast majority of instrumentation provides
non-critical, non-essential status. Such instrumentation can readily have
non-scheduled maintenance (run to failure or self-identify) and be
maintained as operators recognize their need for, or dependence on, its
use. A few instruments provide early warnings of impending high cost
failurelarge-machine vibration monitors, for instance. These need
attention.
Although concern that hardwired instrument trips will lower unit R
is legitimate, there are more fundamental worries. Focus on essential
instrumentation improves unit R and safety. Clear instrumentation
maintenance standards supports safe operation. Usually, instrumenta-
tion and personnel protection for large equipment go hand-in-hand.
175
Concern that hard requirements get carried away (in the fossil world,
anyhow) is driven by fear and culture, not careful analysis. The oppor-
tunity to establish operations-administered guidance can improve safe-
ty and performance.
Once identified equipment failures are entered into a plants main-
tenance system, operators describe symptoms and provide other
insights. Getting the right maintenance starts with identifying prob-
lems correctly and that means clear WO problem descriptions. Even
someone with limited writing skills can quickly grasp WO specifica-
tions. The more defined the WO problem, the more diagnostics com-
pleted, the easier it is to troubleshoot, define scope, and perform work.
Operator monitoring. Operators monitor plant performance,
remotely and locally-in the control room and on rounds. Automated
DCS plants trend by CRTs or by automated-round logging devices.
Monitoring via DCS CRT or control room panel requires a big picture
perspective and the capacity to anticipate. DCS make monitoring the
plant easier, simplifies work, and improves alarm response. DCS sim-
plify round monitoring requirements because remotely monitored
points can be trended and need not be replicated in rounds. Invariably
there are instruments that arent monitored, or that need a physical
presence to visually review, or that cant be downloaded because appro-
priate drops arent available. In these cases, a round is still necessary.
DCSslike all other instrumentationneed oversight to control
information going into the system and the alarms safeguarding it.
Because DCS has the capacity to tie together large amounts of informa-
tion, scope-of-monitoring is even more important. RCM helps prioritize
and rank information value. Critical alarms can be emphasized and sta-
tus alarms de-emphasized. On a DCS upgrade, an RCM filter can eval-
uate alarms and instruments for monitoring, and limit the scope of mon-
itored equipment, hardware, and software. This substantially reduces
the amount of instrumentation required, and saves money. Savings con-
tinue over the life of the facility because the scope of monitoring and
maintenance has been limited.
Rounds optimization. Rounds consume the major portion of oper-
ators time. Ideally, time not spent reconfiguring the plant is devoted to
rounds and monitoring. Rounds are CNM tasks that incorporate fail-
176
Applications
ure-finding checks. Traditionally, rounds were a way to ensure operators

were in the plant. An hourly round was a typical service standard but in
systematically developing a CNM program, one finds that thorough
rounds through a large facility take several hours. The one hour round
is both not feasible and not desirable since it compromises monitoring
quality.
Because RCM-based CNM guidelines identify expected failure
modes for each item of equipment and equipment monitoring limits,
they can be incorporated into logsheets and hand-held logging devices
or remote DCS monitoring screens. RCM rounds data can be down-
loaded directly into equipment and CMMS files for detailed assessment,
history, and trending.
By comparison, traditional operators rounds were developed using
vendor guidance and general experience. They offer substantial
improvement opportunities including reduction in scope of numerical-
ly monitored points while increasing the completeness of failure mon-
itoring. Combined with interval extensions (that random failure data
suggests with benchmarked actuarial analysis), the simplification
opportunity is substantial. The results include:
reduction in rounds scope

elimination of redundant and inconsequential logging
addition of overlooked instrument and alarm checks
development of a rounds strategy (grouping)
consideration of regular and abbreviated rounds
During startups, it may be necessary to temporarily reduce rounds-

logging to avoid adding staff. (The increased risk from reduced moni-
toring is presumably acceptable.) Automated-rounds data-logging
requires that round tasks and routes be specifically determined.
Previously, many facilities used informally developed and managed
rounds. As a major strategy change, implementing rounds-logging is
the ideal time to perform a rounds review and integrate rounds into
an overall plant CM strategy. The assumptions and basis for rounds are
well worth examining before to re-institutionalizing rounds.
Implementing automated rounds-logging is a milestone to assure
177
rounds are firmly grounded by ARCM.
Parts
Age exploration. Actuarial failure statistics from commercial avia-
tion studies show most in-service parts (93%) never reach their design
end-of-life. Components are replaced on a hard-time basis though only
partly consumed (Fig. 5-1). This stems from conservatism and from
untested assumptions about wear-out and overhaul. Commercial avia-
tion experience and conclusions transfer directly to the generating
industry, supported by appropriate information control and manage-
ment. Age-explorationsservice and wear monitoring on parts in
service as they are replacedprovides information that can extend life-
times.
To improve part utilization, the involvement of craft performing
part in-service performance assessment is both necessary and logical:
They remove, service, and replace virtually all parts and so their assess-
ments of parts performance are essential for aging study. When skilled
workers ask the question, How much remaining serviceable life is
there? it orients them towards assessment of failure modes, criticality,
and part service performance. CMMSs offer the ability to track failures
and replaced-parts performance information with less effort.
However, no information trail begins without a skilled craft assessment
and data entry.
Evaluation of parts performance in-service is every plant persons
job. The savings potential is simply too large for such work to be
ignored, and many facets to parts service requires that all be involved.
These facets range from warehousing lifetimes to nuclear environmen-
tal and usability issues. Sometimes savings come where least expected.
While most CMMSs have the ability to develop age exploration
processes sensitizing the craft to age explorations as a routine practice
is more challenging. A simple assessment of a part as it is replaced is
more that adequate. Fancy material-failure analyses that are within the
capabilities of some companies are, for the most part, not needed. Parts
management subroutines in new CMMSs will enhance parts use and
tracking-but even good guesses are helpful!
Component monitoring and age exploration have been practiced as
178
Applications
Figure 5-1: Failure Curves for 93% of Equipment Components

an engineering discipline for a long time. Dissatisfaction with the cost
and availability of precipitators led to the development of fabric filter
dust collection systems, affectionately known as baghouses.
Improving economics and the performance of fixed-speed drives, with
the growth of power electronics, lead to development of variable-speed
drives. New materials, processes, and equipment all start with the real-
ization that equipment has inherent design limits that ultimately limit
the capability, which in turn starts the search for new alternatives.
Involving an entire organization at all levels with in-service performance
evaluation is a profitable first step towards the next breakthrough.
New software information-management capabilities make this more
feasible than ever before. More importantly, there are significance sav-
ings to be realized. Parts usage and improvement based upon service
requirements and performance are one of the prime features of RCM
applied in commercial aviation. In my own career, Ive regularly discov-
ered major savings opportunities by extending equipment service life-
times based upon age exploration and redesign. The latter are usually
179
not dramatic as engineering exercises, but in aggregate the cost savings

and performance enhancement have added up to some staggering
amounts.
In most plants, small parts (costing less than $10,000) and the
scheduling and planning process means that parts decisions are left to
the worker or planner. The planner-often in the dual role of scheduler-
makes decisions when ordering or purchasing the parts based upon
feedback from the shop floor. This is where part utilization improve-
ment begins.
Stocking levels. Parts usage and processes drive stocking. Some
parts need to be carried as critical (important) spares, most dont.
Given the spare everything strategy of the traditional utility industry,
there are ample opportunities to reduce stockjust as there are oppor-
tunities to make errors: No-parts-stocking strategies will always provide
a spare on demand.
In fact, the absence of any part problems indicates over-conserva-
tive stocking practices. Lacking sparesor being unable to locate a crit-
ical sparetranslates into lost production that makes for an expensive
spare. Spares carried in stock for years, with virtually no movement, is
no better. One expense is visible, the other is not.
To optimize stocking, operators and craft must understand and
accept the parts strategy and the larger operating strategy of which it is
a part. They manage failures, which requires an implicit appreciation for
the plants equipment R. Approaches differ, when using and extending
the life of used parts. Some would have you toss every consumable-oth-
ers, reuse everything. Theres a middle ground in most cases.
When workers know that parts arent readily available, do they
exercise greater care using them? You bet! Inexpensive and consum-
able parts get tossedexpensive and reusable gets reused. (But, which
is which?) An age exploration program can help identify them and fill
the gaps. If we dont know how a part failedfind out! Many parts can
be reused cost-effectively, e.g., not with a penny-pinching, refurbish-
ment-at-any-cost philosophy. Evaluating the reuse of serviceable parts
requires age exploration and training. Parts-life estimation is tied close-
ly to in-service parts-failure evaluation.
Many companies establish stocking levels with expert software, but
180
Applications
a conservative mentality and an awareness of parts aging and failure can

suffice. It requires training and instilling a questioning culture.
Theres a degree of risk, which may include not having key parts for
equipment that fails in service. Like owning a car or a home, risk-man-
agement exists in several formsfinancial, obsolescence, and down-
time.
The traditional strategy of carry an extra of everything was
advanced by vendors, who supplied lengthy lists of every spare imagi-
nable, requiring warehouses with huge inventories to control and man-
age. Taking more riskusing parts-sharing groups and vendor-main-
tained parts inventoryreduces part costs. Truly critical spares can
be managed with overnight delivery, parts sharing, and operational con-
tingencies. For overall success, reduced stocking must tie to knowledge
of part-failure risk, redundancy, risk profile, and equipment strategies.
Consistency. Reliable parts provide value. Seems simple enough.
However, study how parts are used in any given manufacturing setting,
and the strategies employed in selecting parts and vendors, and youll
come to some surprising conclusions. For instance:
manufacturers focus on reducing parts inventory as a cost-man-

agement strategy
general results of that strategy transfer to the generating plant
environment
parts R influences required stocking
unreliable parts carry hidden stocking costs
the more vendors and part sources a plant has, the greater its
overall part variability
part variability increases costs
Most engineers and planners have to deal with multiple suppliers

and mixed lots. By evaluating parts service, I conclude that manufac-
turers of quality parts understand the cost/benefit savings that superior
parts provide for usersand they price accordingly. Unfortunately,
planners, maintenance managers, and buyers-those who dont have the
information (or desire) to do life cycle parts-costing-make most of the
buying decisions. Many utilities require firm bids when a purchase
181
amount exceeds some nominal amount$5,000 is commonwhich

kicks decisions upstairs.
Special, custom-application parts increase life cycle management
costs. A coal-fired plant with six millsno two of which has the same
basic hot air ports, pyrites brushes, discharge valves, rolls, and tension-
ing plungersmeans that this individual crafting of equipment
requirements increases parts requirement. In the absence of formal
standardization policies, there are common parts-management thumb
rules. SPC (borrowed from manufacturing) provides the best guidelines
on how to monitor parts usage. Many parts events have demonstrat-
ed the folly of not following standardized part rules. Many corporate
buyers disagree with these guidelinesthey buy strictly on pricebut
they rarely face the consequences of using and managing parts that
dont fit or that break on installation (or soon after entry in service)
or that break randomly.
In general:
know critical functions of all secondary market parts selections

go with the OEM except for obvious substitutions
work with the suppliers engineering staff to understand parts in
service
low-quality appearing parts are usually what they arelow
quality
high-quality appearance doesnt guarantee performance; sup-
pliers do
workers have practical insights on parts performance
parts records will surprise you
Problems. When a part problem is identified, its best to check

records. However, its likely that records concerning parts and their use-
and costs-are incomplete. Usually, the craft identifies a part problem
based on service problemsor a feeling. Analysis confirms the concern
and quantifies an opportunity. More capable inventory management
systems and CMMSs can improve parts information and also age explo-
ration. Systems can help, but only if theres perceived benefit to their
use.
182
Applications
This is an area for poke yokethe use of many simple tools.

Simple retention-and-review storyboards are valuable for training
and evaluation of failed-parts experiences. Good or bad examples,
photographs, and failure descriptions all help improve understanding of
parts.
Troubleshooting. Some basic truths:
troubleshooting costs technical and maintenance personnel time

the more experienced personnel are more productive trouble-
shooters
troubleshooting new equipment is harder than diagnosing a
familiar machine
a learning curve progression must be followed before people
reach full diagnostic effectiveness in a plant environment
new and unforeseen failures demand more resources
FMECA and a fault tree assist troubleshooting by establishing rela-

tive probabilities of what can go wrong. High-probability events can be
checked first. They also provide failure symptoms that can validate
actual failure causes. An FMECA indicates sources of failures, benefit-
ing future troubleshooting.
Experienced personnel dont require these insights. Every organi-
zation has inexperienced staff and bringing new people up to speed
quicklyproviding all users with optional aidsis extremely benefi-
cial. Rare failures are hard to catch and diagnostic aids such as fault
trees, logic guides, and FMECAs are then very helpful.
To effectively diagnose equipment, however, the technician must
understand it. The more complex the equipment is, the harder it is to
diagnose. Technicians must understand the components that provide
the functions that make up the equipment as well as their interactions.
Focusing on functional descriptionsor even on component engineer-
ing failureslimits failure detail. (Complex equipment can also simply
fail.)
When equipment maintenance is permitted to slide and instrumen-
tation and redundant components reach failure, multiple failure modes
begin. Failure interactions make trouble-shooting much more difficult.
183
A time benefit of a fully implemented PM program is that failure-iden-

tification is as simple as it can be. Fewer failures get diagnosed with
complex interactions, so diagnostics are more straightforward.
One of the most difficult tasks in a complex plant is to restore aban-
doned equipment-just ask an engineer who has done plant restoration
after a fire! Its almost as hard as startup, since everything must be
checked out. Abandoned equipment also generally has multiple failed
states, each of which must be separately corrected to restore function
Secondary failures drive home this point. Every secondary damage
event results in many more problems. (Reconstruction of steam-dam-
aged cable can be extremely complex and time consuming.) Multiple
failures can result from flooding, fires, leaks, or a variety of other events
that fundamentally change or exceed the physical environment. Some of
the most severe damage (in terms of cost) comes from events leading to
steam and moisture attacks on components designed for dry environ-
ments.
I prefer to deal with primary-failure prevention and simple failure
modes. To do so, we should understand those primary failures that lead
to common-mode, general, and expensive secondary damage.
Failure numbers
Numbers are the best way to tell the objective failure story, yet
theyre missing from most traditional generation RCM analysis reports.
Some RCM books go so far as to discount failure statistics and numbers
altogether. In my opinion, this is a serious oversight. Implicitly or not,
we use frequency and consequences to draw conclusions, and numbers
tell that story. Those who dont understand this and live by the numbers
wind up chasing the trivial few. It gives RCM a black eye when ana-
lysts bog down in endless pursuit of rare or imaginary eventsthings
that dont happen in the real world.
My approach reflects my predisposition towards measurement
Im an engineer. In reviewing large quantities of failure data, identifying
failures, summarizing statistics by failure categories, and making esti-
mates, I work with numbers that arent exactbut theyre in the right
ballpark. I view them like health physic numbersorder-of-magnitude
significance. They identify sensitivity to costs that need to be under-
184
Applications
stood at 10%, 100%, or 1000% payback levels over the period of inter-
est. We need to structure our programs for value. A 10% payback on a
turbine overhaul may be worth chasing, but probably not for a $20 fil-
ter replacement. A 500% savings on a $20 task clearly outweighs the
same for a $2 task. Practically, this means when it comes to a tradeoff
(and it will), we must give up the $2 tasks to make room for the $20
tasks.
Activities should reflect on-site statistical data and failure experi-
ence. Work environments are unique and influence what fails and when.
The cultural environment influences what failures are recognized.
Available levels of skill, knowledge, and other intangibles can be
inferred, but are hard to measure. Just as two randomly-selected indi-
viduals will experience different success rates with the same make and
model of automobile (as measured in longevity and life cycle cost), two
similar plants experience distinctly unique operational outcomes. These
can only be explained in process terms.
So-called rare eventsthe second aspectpose an actuarial
problem. Rare events represent the highest-value RCM learning and
benefit opportunities. Most heavy production and financial losses arise
from them. They are certainly worth understanding.
After many years examining major losses, I find that in most cases,
a chain of events presents a history. The progression towards ultimate
failure depends on systematic process weaknesses-rare events occur
more frequently in the absence of process awareness and controls. They
reflect random, individual, repetitive occurrences. Individually, they
rarely convert to an operating event but if they happen frequently, that
event will most probably occurand, ultimately, statistics tell the story.
Rare events can be managed with conscientious, complete operations.
These rules, well known in theory, are well-practiced by professional
operating organizations.
Safety
Direct consequences
Generating plant safety presents two challengesmaximizing safe-
ty practices and minimizing costs.
185
Overall, we need to apply better safety practices in many plants.

Fossil generating unit accident rates are significant. High pressure
steam, high voltage, coal belts, and large rotating equipment perform
well under ordinary conditions, but are unforgiving. Safety awareness
begins with understanding equipment and how it fails. Many safety
issues develop in the course of returning failed equipment to service.
This includes diagnosis, tag out, physical work, test, and return-to-serv-
ice. Better understanding of equipment failure leads to better mainte-
nance practices, more specific work plans, and planned workand
planned work is safe work.
PM emphasis is on time-based and CDM, and occasionally, on
NSM. PM work is plannable. When planned, it contributes to safe
operations. Rework and repair tasks are routine and plannable. Some
plants develop detailed work steps, others leave details to the skill of the
craft. Whichever approach is taken, using standard PM tasks in stan-
dard blocked work formats, ensures that repetitive work is supported
and contributes to safety. Its easy to make a case for safety when a high-
er plant-conditional readiness state is achievedand an ARCM-based
maintenance strategy in fact leads to a higher state of equipment readi-
ness, both for critical instrumentation and redundant and lessor equip-
ment. When work is performed in logical rank, the need to work extra
hours is reduced. More efficient work practicelike equipment align-
ment and on-condition maintenance-also contributes to safety.
Operators and mechanics who learn the system review process also
acquire skills helpful in understanding equipment importance and pri-
oritization. Combined with the inherent improvement in R that occurs,
its easy to make a theoretical case that RCM favorably benefits safety.
However, actual measurements, to the best of my knowledge, have not
been made.
Serious injury and equipment hazard events can stem from per-
forming on-line maintenance. As units are put under more pressure to
generate, the trend to perform on-line maintenance will increase. The
need for routine, high value work plans will be even greater.
Potential consequences
A second, equally significant improvement in safety can be derived
186
Applications
from more expeditious use of a plants safety budget. Presenting a budg-

et case for safety modifications is a chronic problem. Conversely, many
convenience modifications are covered in safety terms to improve
their likelihood of approval.
The greatest practical safety issue in many non-nuclear plants is the
state of critical instrumentation. ARCM can and has significantly
improved plant awareness of instrumentation by focusing maintenance
efforts on high value instruments and alarms, while discontinuing
scheduled maintenance on the rest. Operations monitoring has benefit-
ed from this review. Some classes of plant equipment get taken for
granted, and this practice could benefit from an ARCM review.
Wherever a station spends a substantial amount, that warrants con-
sidering ARCM analysis. Chances are tidy savings will result. All-out
faith in contractors to identify plant material maintenance needs is gen-
erous, but financially risky. The chance is high that the plant will end up
with a Mercedes, rather than a Ford!
In its original development phase, RCMs criterion for critical
work was safety. Yet, safety is a routine issue at most generating stations-
companies will sacrifice at least an hour a quarter to have all hands
attend a safety meeting. ARCM can neatly operationalize many
safety issues. The airline criteria (RCM critical = direct safety conse-
quence) meant critical, mission terminated. i.e., For air transport, a
critical failurea safety issueis a show stopper. All critical failure
modes warranted scheduled maintenance. In the event that applicable
and effective maintenance tasks couldnt be specified, the default
action was to redesign. Utilitiesparticularly those operating fossil
plantscould greatly improve safety budget mileage by adopting sim-
ple, direct safety criteria of the airline industry. Hypothetical, possible,
and theoretical safety expenses that prevail at so many stations could be
a thing of the past.
Example and Case Histories

Circulating water tower
Circulating water towers (CWTs) offer a representative look at the
tradeoffs in large equipment maintenance. Many possible CWT main-
187
tenance approaches are available. Towers can be capitalized, and

replaced in 20 years. They can be continuously maintained, with ongo-
ing, cell-by-cell rebuilding. The choice depends on long term operating
goals and philosophy. Functions vary from cost-conserving mist-elim-
inators to functionally important debris screens.
Nobody loves a CWT. Theyre damp, dirty, infested with birds, and
they smell like chlorine and biological growth. In farming areas they fill
up with sediment and windblown soil. In industrial areas, they scrub
everyone elses emissions. (Maintaining a CWT adjacent to an oil refin-
ery was my personal headache.) Industrial area towers require chem-
istry different than those at other isolated tower sites.
Towers suffer several dominant failure effects:
biological growth and debris accumulation

aging of diffuser/distributor nozzles
random environmental damage (like ice)
cell hot water basin flow balancing adjustment
timbers and fill aging
structural deterioration, fastener relaxation shifting
Plant operators take towers for granted until performance is a prob-

lem. Theyre not complex, theyre easy to understand, yet theyre
sophisticated in design. In a common basin-distributor design, water
risers carry heated condenser cooling exhaust into cell basins through
shutoff/flow control valves. Water spreads and falls through diffuser
nozzles onto fill, forms droplets, and then evaporates and cools. Cool
water in the basin provides a suction reservoir for the circulating water
pumps (CWPs). CWPs draw cool, aerated water through large mesh
screens, knocking out debris and returning it to the condenser inlet
water boxes.
Cooling point depression is the measure of tower performance.
Margin means theres ample cooling capacity. Ample capacity assures
condenser vacuum is adequate.
Towers require seasonal maintenance. In summer theyre most
needed for cooling but in winter they need the most attention. Towers
can be icy, slippery, dirty, bitter cold (except inside), and demand con-
188
Applications
stant attention. Ice builds up, louvers tear off, fill comes down, and
screens plug. Work can be miserable. Fan electrical problems, lubri-
cating oil leaks, vibrations, and a host of other random problems make
towers tough maintenance areas.
During the summer, the units are often on the raw edge of load
reduction as the tower cell fans, spray patterns, distribution, fill condi-
tion, and other lesser problems make towers the key determinants of
load. Every last ounce of capacity may need to be coaxed from an old
and tired tower. Balancing cells on a shifting, deteriorated tower can be
almost impossible.
Towers have blown over, fallen down, burned down, and rotted
away. The last case is the most frequent. Deterioration of cell basin lev-
els, or leaky distributor shutoff valves, makes cell balancing difficult or
impossible. Structural sags can effect the hot water basins so that bal-
ance cannot be achieved. As towers age, their inability to balance,
maintain basin temperature, and maintain condenser vacuum in sum-
mertime (during load peaks) makes replacement inevitable.
The more dramatic tower episodes in my career involved lesser sub-
systems that werent appreciated until they became problems or failed
outright. Before fiberglass return lines and spargers became standard,
redwood staved-distribution piping was common. At one plant the
staving failed, sprinkling a waterfall out away from the tower basin.
Flooding resulted, and the basin went low. After the basin emptied, the
circulating water pumps tripped. The unit went down on combination
of low vacuum and no cooling water flow. A similar event involved a
tunnel access manhole cover bolt failure on the discharge side of the cir-
culating water pumps. The condenser tripped on low vacuum after the
basin emptied. This latter case destroyed a contractors onsite trailer.
Circulating water pump head is 30 to 60 feet at rated flow, nominally
enough for an impressive waterspout!
Tower fan problems are the stuff of legends. Fans throw blades
when ice damage occurs. Deicing practices aggravate this tendency.
Gearbox failures, due to water lube-oil contamination, are typical as age
increases. Corrective measures for throwing blades have involved cre-
ative modifications like enclosing the diffuser assemblies with heavy
wire mesh. (Consider the costs for this modification for a 16 cell tower
189
and you see the potential for RCM-based modification review! What
an opportunity for root-cause analysis, too!)
Motor failures are common as towers age in service. Most new
units use weather-enclosed motors. Theres no work to speak of and
they are essentially consumable. Sizes range from 5-75 horsepower
(HP).
Secondary failures from tower fill and structural debris have been
quite damaging in unique cases. We once rebuilt major tower sections
(with a contractor) and wood scrap debris was left in the basin. Startup
transported this debris to the water boxes and waterbox isolation valves
(seats), and ultimately into the condenser tubes. Screens removed a lot
of debris, but the volume and size of the debris, together with repetitive
screen cleaning during adverse winter weather conditions, allowed large
amounts into the condensers, where it accumulated. Silt accumulated
around the packed wood debris waterboxes and flow stagnated in the
partially blocked tubes. Local corrosion cells were established. The
resulting tube damage prematurely required condenser retubing due to
severe water conditions and the inability to control localized secondary
pitting corrosion. An admiralty brass condenser that had a design life
of 30 years was limping badly at 13. Production losses ran into the mil-
lions. Retubing ultimately cost around $5 million. Granted, the water
was aggressive, but the condenser had been performing well until the
wood debris episode.
This example reiterates the importance of such simple features as
screens, and the need to do simple PM tasks-screen cleaning-very well.
It must be timely, and conscientious, even in adverse weather. Our will-
ingness to start the unit in this state of unreadiness indirectly reflected
our standards. When standards are compromised at high levels, the
trickle-down effect can be significant. Ultimately, workers care when
they see that managers care. Standards must begin at the top.
Secondary damage of this nature is an expensive consequence of
low-quality work and otherwise inconsequential failures. It is very pre-
ventable-if youre aware of the risks. Of course, this event was an infant
mortality failure, but a very predictable one in light of the stations other
problem areas.
190
Applications
RCM-based maintenance options
1. NSM. Replace the tower at 20 years-suffer intermediate produc-

tion losses. Generally minor aging problems like lost fill, structural sag-
ging, and dry rot add up over time until the most cost-effective choice
is a complete replacement. For peaking units this may be an economic
option
2. OCM. Use performance test and inspection to identify when to
perform maintenance. Overall, balancing capability, structural integri-
ty, and wood condition can identify selective areas for rework. Rebuild
towers in part, cell by cell. (This requires skilled tower technicians.)
Generally, tower fill must be rebuilt after working into a damaged sec-
tion. This requires higher tower skills than a typical plant person
achieves over the course of routine work. The work is difficult and
requires specialists
3. TBM. Distributor nozzles cleaning and biological growth-con-

trol chlorination must be done repetitively on a schedule. Even with
chlorine treatment, biological growth occurs in hot water cell basins
where chlorine depletes. Basins need physical removal of growth with
squeegees, scrapers, or other tools. Distributor nozzles require visual
inspection and debris removal. Damaged nozzles require replacement.
Nozzles exemplify an area that plant staffs typically dont appreci-

ate. Made up of simple plastic fittings, nozzles must break up and dis-
tribute water into entrained droplets to achieve cooling. Fill helps this
by further dispersing the water as it splashes downward. Damaged noz-
zles merely dump water into the fill. With too much waterfall-type
flow, cooling is reduced. Not everyone appreciates that the splash zone
of a healthy tower effectively increases its height six to eight fold!!
We once assigned a new tender to clean tower nozzles. Although
briefed, he discovered an expeditious way to perform the jobrestor-
ing nozzles plugged with flaking deck paintby using a broomstick to
rod out the nozzles. Unfortunately, he broke the distributor fitting on
many nozzles that he cleaned! When we discovered this, months later,
during the summer peak load, we faced the double whammy of forced
191
load reductions and nozzle restoration. Unskilled people can damage

even simple CWTs!
4. NSM. Aside from lubrication and performance, additional work

can be identified, based on equipment condition. A typical multi-cell
tower has installed redundancy. Other equipment strategies include:
motors and motor starters (CM)

lube of motor bearings (TBM). (Many are now sealed bearings.)
balancing fans (OCM)
distribution valve seats restoration based on shutoff (OCM)
seasonally repair louvers (icing damage) (OCM)
replacing damaged fill (OCM)
correcting resonance from structural relaxation (OCM)
As perhaps you can appreciate, the distinction between mainte-

nance strategy types starts to overlap. At an excellent operating com-
pany it almost becomes academic!
There are many tower styles. All provide the same basic cooling
functions. Overall tower and individual cell performance provides
good opportunities for on-condition work.
The challenge is performing inspections and tests as scheduled and
then performing indicated, required, condition-directed work as
demanded. Because seasonal load periods are brief, its easy to miss the
summer peak season if work cant be responsive. Cell-by-cell perform-
ance assessment, specific inspections for resonance, dry rot, and other
structural deterioration are typical examples of effective OCM. Fan
motor, drive, and assembly maintenance are more conventional rotating
equipment OCM examples. CDM beyond the rotating equipment
often fits best as a part of a cell-by-cell rework program.
Towers are greatly affected by adjunct programs, such as water
chemistry. A good chemistry program helps a tower achieve its design-
capable life. Problems such as erratic chlorination will result in service
deterioration and premature aging.
Tower operation methods influence component life. Towers de-
iced by reversing fans suffer greater ice damage and require more fill
192
Applications
and louver rework. Fill restoration is specialist work. Replacement

requires removal by working-in, and restoration working-out.
Most towers have considerable as-new performance margins.
Many tower ratings can be adjusted as much as 25% on paper with lit-
tle performance impact based on the same hardware. A typical tower
will last 20 years with modest maintenance (focusing on mechanical fans
and drives). Alternatives to this involve spending around $10,000/cell
annually, continually servicing and rebuilding cells, or replacing the
tower entirely in 20-30 years at a cost of $200,000/cell. The solution
must fit with the companys long term unit operating profile.
Establishing and carrying out a tower maintenance strategy in a cost-
competitive environment is, in fact, a complex optimization problem.
Running a tower well on a limited budget means theres little room for
oversights or mistakes.
Aftermarket tower services and parts abound. Some suppliers are
outstanding, others marginal. Beware of services from anyone who is
not a tower specialist. The re-decking cited above was performed by a
local contractor, not a reputable tower service firm. At the time, every-
one saw it as a good deal, and an easy job. It turned out to be a good
deal more.
193
Chapter 6
Lessons
Wise men dont need advice. Fools dont take it.

Benjamin Franklin
Believe your instruments.

Pilot Adage
Calibrate your instruments.

Naval Reactors
Preventive and corrective maintenance models are built from basic

assumptionsgeneralizations not always born out by facts. The tradi-
tional PM model assumes:
PM is always less costly than failure

equipment can not be run effectively with failure-based main-
tenance
we understand and can recognize failure
the consequences of missing PM is failure
195
Can simple failures progress to functional failures? Can functional

failures impact equipment and systems? This is what we seek to under-
stand and, ultimately, remedy. These are the lessons.
Even the best, most optimized, experience-based designs are no
more than relatively insensitive to failures. The natural progression for
equipment over its production life is to increase failure resistance by
design as field experience increases; ultimately, however, how people
perform is what actually matters. Some organizations gun deck main-
tenance paperwork (like PMs); workers dont blindly buy into process-
es. Failure process ignorance, or a lack of participation in PM identifi-
cation, selection, and development limits worker process commitment.
Root cause analysis can be a factor. Until actual strategy and costs
are developed, its not often clear which maintenance approaches are
likely to yield the best overall task, or which combination of tasks best
address potential failures at the lowest cost. Until the plan is imple-
mented and statistically meaningful costs are collected, value cannot be
established. Backing off existing, low value PM is just as tough as estab-
lishing high value programs.
Compare CDM versus TBM alternatives for a filter change-out. The
accepted wisdom is that OCM is always less expensive and better than
hard time. Then consider plant pump VM-at some level, the benefit
derived from monitoring pump vibration vanishes. At a $50/hr (loaded)
labor rateand with no production impact no planned maintenance
balances against monthly monitoring with a $10,000 pump capital cost
and a 10 year lifetime. (i.e., if a pump lasts 10 years with no PM, costs
$10,000, and carries an 8% capital cost, 10 years of monitoring to
achieve a two year life extension has little or no payback.) Such num-
bers can be replayed with different assumptionstime value of money,
and so forthbut the general range doesnt change. Somewhere in the
comparison of capital cost range with projected life, PM becomes inef-
fective.
Until a benchmark case analysis is done, operating team assump-
tions about failures and costs may diverge greatly from realityto their
economic disadvantage. Note that as maintenance effectiveness increas-
es, the break-even point drops. Its more effective to perform PM in an
efficient organizationand less effective in an inefficient one.
196
Lessons
Task Intervals
PM task intervals are based on failure rate and mean life variability.
For random-failing components, special strategies may apply (within
the context of the design). For a predominantly random failure mode,
for instance, functional monitoring can effectively identify instrumenta-
tion failure. A check made at a fraction of the MBTF can identify
instrumentation failure while minimizing overall failure risk. An instru-
ment can be tolerated in a failed state for limited intervals, if effec-
tively redundant.
Operations is charged with identifying random equipment failures
in the plant. An effective round ensures that operators get through the
facility often enough to identify failuresparticularly random ones
without excessive monitoring. Usually, four to eight-hour round inter-
vals are effective. If they must be made more frequently, designs should
be evaluated; if they can be made less frequently, it raises questions of
risk, and staffing levels. Equipment in terminal failure may require addi-
tional monitoring. Requirements to monitor terminally failing equip-
ment can be quite substantial.
Selecting task intervals is part science, part art. Failure data alone
may be inadequate for infrequent failure modes. With limited failure
information, manufacturer recommendations, failure mode physics, or
mode type assessment, an expert opinion may be needed to establish
appropriate task intervals. In many instances, too, an exact interval is
not critical. With inexact information, intervals can be over-specified
(made too frequent), particularly by an unskilled analyst. An age explo-
ration-based monitoring program can adjust intervals based on failure
type and age exploration.
When craft participates in interval selection, its a significant orga-
nizational growth step. Craft develops ownership of monitoring and
parts-in-service assessment that in turn supports identification of
appropriate intervals.
For expensive, age-based failures, intervals need to be conservative.
Generator rotor cracking probably has an exact MTBF in excess of 30
years. Inspection on a disassemble-overhaul basis (typically, every 5-10
years) is appropriate since the equipment is at risk. Instruments moni-
197
toring for high cost equipment failures must be maintained and their
failure prevented. Brief outage periods are acceptable but entail risk.
Prolonged unavailability is not an option. VM on high-inertia rotating
equipment must be maintained constantly, with hard-wired trips, and
must have an operating limit. Anything above the limit is an automatic
trip.
The problem establishing intervals for instrumentation is ambigui-
ty. Manufacturers resist hard-wired trips to avoid spurious or undesired
events. They assume that an operator can discern inappropriate
demand trips on an instrument provided for status, and avoid bring-
ing down the unit. Its right in theory but wrong in practiceoperations
learns to ignore unreliable instruments. Regular, spurious instrument
trips and alarmsas there can be at status only instrumented
plantsmeans that instrument value plummets. Status only instru-
ments can lead to maintenance deferral when their importance is dimin-
ished. Well-maintained, high-quality critical instruments are essential.
Identifying critical instruments is helpful, of course! Chasing a
faulty alarm is frustrating and expensive. Sitting in the hot seat after
making the wrong call in an ambiguous situation is equally trying. You
dont get too many chances before the instruments are discounted and
ignored. In RCM (or ARCM) a spuriously alarming instrument is con-
sidered failed-one of the truly great contributions to instrument
maintenance programs and one that operators had been demanding for
years.
It all sounds overwhelming. But in reviewing thousands of compo-
nents and tens of thousands of WOs, one quickly becomes comfortable
estimating task intervals. It takes experience to develop a feel for fail-
ures, but, with some quick R training, most experienced people can
draw on their years of observations to make excellent judgements about
parts agingparticularly when wear, abrasion, or erosion processes are
at play. A failure model helps integrate a picture of the failure process
with plant culture and strategy.
In practice, MTBFs are often grossly underestimated. Plant staff
base their life estimates (and PM intervals) practically on a small frac-
tional sample of failed parts. While suitable for safe-life interval lim-
198
Lessons
its, it isnt obvious that this grossly underestimates the average life.
Predominantly, PMs are based on economics. What this says is that
informally developed economic PM intervals are almost always grossly
conservative.
Estimating organic failures is vexing. Aging is expected; however,
rubbers, cloths, elastomers, and similar materials deteriorate with time
and temperature. Even when visual aging evidence is missing, its risky
to assume theres been no aging. The Arhenius temperature character-
izes organic agingbelow it, little or no aging occurs; above it, aging
increases quickly. Visual evidence can be absent in the transition range.
Calculating an ageespecially for components in critical applications
helps avoid gross errors. When large organic expansion joints made of
reinforced cloth, rubber, and binders, reach their manufacturers spec-
ified life, life extension is risky. During installation, the absence (or pres-
ence) of offset, vibration pulsations, and other synergistic aging phe-
nomena complicates the picture. Only experience can determine actual
in-service aging.
Remaining aware of failure processes and how they workknowing
what to look forgreatly improves setting task intervals. Fortunately, a
few fundamental aging processes repeat over and over in most plant
applications. Learn these, and you have the basic tools to evaluate most
aging mechanisms. Developing failure mode data is a R engineering
exercise. Fortunately, experience, thumb rules, and training go far.
Age Exploration
Definition
Age exploration is the systematic examination of the lifetime a com-
ponent or part can support in an application in service. Its crucial to
setting task intervals. The term means literally to explore component
aging, and find out what service the component can provide.
It used to be assumed that all components had finite lifetimes
equipment wore outand needed replacement or overhaul. As first
examined in air transport, it was discovered that it doesnt hold true for
many components. Though powerful and intuitive, the assumption had
no basis. A statistically large number of components showed virtually
199
no deterioration during period of use, based upon early jet engine over-
hauls in the late 1950s and early 60s. These included actuarial analysis
of failure studies. On 90% of replaced components, life remained at
the end of their specified life.
Lifetimes have always been based on the best available informa-
tion. Aircraft turbine developmenttransitioning from prescribed
overhauls to age study-based monitoringsummarizes the experience.
When overhaul limits were eliminated and age exploration undertaken,
equipment lifetimes increased significantly resulting in quick economic
benefits and lower risk. Considerable actuarial analysis detailed mathe-
matical failure analysis to quantify lifetimes and conditional probability
distributions-and support this change. The concept of conditional
overhaul gradually emerged. The results can be applied to most other
industrial maintenance applications.
Conditional overhauls only address immediate failure causes and
correct other necessary parts to achieve specified performance. The
paradox is that conditional overhauls yield overhauled equipment
that statistically perform the same as traditionally overhauled ones. By
literally running fault-tolerant equipment with NSM until failures devel-
op, we can use the concept of age exploration, merged with design and
conditional overhauls, to give credence to the term NSM.
Early equipment manufacturers and maintenance experts did their
best to specify age-based replacements, recognizing the potential to do
better through age exploration. Extending useful equipment life
requires understanding how items age in service and how effective we
are at discovering itand then formulating how best to use this knowl-
edge. Profound understanding of statistics and actuarial lessonsles-
sons learned from those aircraft engine overhauls, failures, and actuari-
al analysisenables this conceptual leap. Evaluation of in-service per-
formance on an ongoing basis (particularly for new equipment and
components as they enter service) enables us to manage risks.
In the course of understanding jet engine aging and failures, the air-
craft industry discovered that even the very best maintenance and engi-
neering experts couldnt predict future engine performance based on
overhaul data. Experts predicted the imminent failure of apparently
worn-out equipment, only to have the equipment perform (statistically)
200
Lessons
identically to acceptable equipment. The inescapable conclusion was

to extend life literally until an age-based failure limit clearly emerged
and then develop lifetime limits around it. This is the lesson of the air-
craft industry. And while it takes a leap of faith to adopt, it has clearly
been successful for extending engine service life.
The inability of even the most qualified personnel to predict future
performance with statistical accuracy-the expert-life prediction para-
dox-flies in the face of all common expectations. It leads us to assume
that under normal circumstances, equipment has a finite lifea sto-
chastic interpretation. The statistically correct alternative would be to
assume (until proven otherwise) that all equipment has an indefinite
life. This is the approach followed with age exploration.
The maintenance implications are enormous. We can finally under-
stand why 90% of parts never fail in-service, and the powerful case to
be made for CNM to establish parts life expectation. It also explains the
legitimacy and validity of the case for NSM, and why the common
phrase, run to failure, is patently, statistically wrongnot only high-
ly-biased and charged, but statistically incorrect. This is the message
maintenance traditionalists need to learn.
We dont indefinitely extend life where environmental or safety
losses could occur. In these cases we benchmark against our best expe-
rience to establish appropriate monitoring intervals. In practice, experi-
ence enables us to predict likely failure mechanisms and their order of
magnitude. While acknowledging the paradox that says: use fault toler-
ant designs and then stretch, until facts point us to an age limit, we
select safe-life limits for critical safety-based failure modes.
Value
For major equipment overhauls (boilers and turbines), extending an
interval by even a few days can have value. Other small savings, in
aggregate, also add up. Age exploration achieves its greatest potential
value when a plant shutdown can be deferred.
For a nuclear BWR, replacing solenoid valves to meet an EQ often
falls in the 4-6 year range. Extending the qualified life for control rod
pilot solenoids (at 4 per control rod, or 137 control rods per unit) pro-
vides substantial savings. (This requires re-qualification testing, of
201
Figure 6-1: Best Value? Every nuclear plant spends time on the NRCs watch list or
so it seems. Impressive operating records lead regulators to suspect production is
emphasized over safety. Maintenance expectations differ in the highly regulated
industries despite the same equipment. The challenge of deregulation is to use
industry-wide best practices to achieve outstanding operations at low cost.
course.) For non-essential parts or in fossil applications, this means

keeping parts in continuous service until aging or in-service failure
demonstrates life limitations. Obviously, this must be done in a con-
trolled manner if failures have safety or environmental impact. Many
times, however, component failures have minor impact. Age exploration
may take an item out to failure to establish lifetime limitationor to
demonstrate there is none.
Solenoid valves statistically perform very well in service when con-
trol air quality is maintained. Failures are mainly sticky operation, at
which time the solenoid needs rework or replacement. However,
nuclear application failures are unacceptable. Its particularly important
to understand solenoid pilot actuator life in high-temperature environ-
ments (like pilot-operated relief valves). Plant shutdown can result if
its not managed. Those who work with such parts must be counted on
to observe in-service performance and lifetimes that can improve parts
usage. For cost-management purposes, systems used to pass this infor-
mation along must be adequate to conduct effective aging studies (Fig.
6-1).
202
Lessons
How well parts perform is information required to assess in-service

aging but its never been easy to get. New CMMS software can docu-
ment failure data and retain it for future use with relative ease; raising
craft awareness levels, and retaining history for future use, is a key next
step. Most maintenance personnel are capable of performing and doc-
umenting very useful summary observations about parts when its
expected. Simple one- or two-line descriptions identifying unusual in-
service aging performance are most useful. Normal aging warrants
periodic assessment and descriptions. Again, when entered directly by
users in their own terms, theyre most beneficial for future aging stud-
ies. Theres a balance between too much and too little detail that needs
to be maintained. Too much detail and too many reviews of large
amounts of data bog down personnel; too little, and the assessment is
inadequate. Workers need guidance on what level of detail to provide.
Some technicians and mechanics systematically perform replace-
ments using rebuild kits or available stock spares with little considera-
tion to part in-service performance. Others assume a new part is always
better than an old one. Age exploration teaches us to examine compo-
nents upon replacement, using aging knowledge and observation, to
assess remaining service life.
Infant mortality studies also dispel the notion that a new part is
always better. This is particularly true for electronic and electrical com-
ponents, and some mechanical parts. Living organisms demonstrate
infant mortality-in fact, the term originated here! For service in com-
plex equipment, proven, performance-aged and burned-in compo-
nents have a higher probability of service than newly rebuilt ones.
Everyone knows the period immediately after a startup carries the great-
est risk, and that in theory, when random failures are measured, a new
part doesnt always out-perform the old one.
Theory (based on experience) validates the use of age exploration
to extend service intervals for new designs as well as parts. Any new
component has some theoretical in-service life but until validated by
performance, it remains a theoretical projection. For new, complex
equipment, age-exploration is essential to establish potential life.
Regulators want assurance that new parts and components will perform
in service as expected. In some instances-safe-life limited equipment,
203
for examplethey may require disassembly/inspection as part of the

conditional certification of a new equipment subassembly or products.
Systematic application
Age exploration in new equipment begins by removing parts from
service for examination. New-failure mechanisms, premature aging, and
other unanticipated failure-mode evidence requires immediate atten-
tion. Over the long term age exploration provides the basis for predict-
ing how much service a given component can support. You can better
extend life when you realize the ultimate service life limit. Done sys-
tematically, it can provide the basis for improving many plant equip-
ment maintenance decisions.
Such age exploration principles have been known and used for
years, but havent found regular applications in electric power genera-
tion. Perhaps this simply reflects the traditional nature of power plant
maintenance. The need to improve part-cost performance hasnt been
a need in generating plants-until now. Legitimizing age exploration
neatly resolves this cultural problem. Utilities should develop formal
age exploration methods and hand the decision process back to those
who actually use the parts.
Effective age exploration requires:
awareness of part characteristics

documentation (usually via CMMS)
engineering assistance
a corporate environment that encourages learning
This last element is vital. Employees must expect that plants, equip-
ment, and systems will continually improve, and that performance will
increase and costs decrease through improved utilization. Without
learning, a part-aging program wont be effective.
The benefits-both obvious and subtleafforded by age exploration
include:
steady reduction in spare part expenses, both direct and indirect

increased awareness and understanding of the role that parts play
204
Lessons
in overall equipment considerationshow parts age; the service

they provide
increased awareness of suppliers and manufacturers and how their
parts perform in-service
the added value provided by premium parts
the R contribution from the best available part
How many mechanics understand these benefits well enough to

confidently perform parts life-extension today? Many organizations
buy on price. More simple case histories documenting how inferior
quality parts can impact a facility are needed to overcome this bargain
disposition.
Persons handling or installing parts of any sort need to assess
their usual suppliers for parts condition and further service suitability.
They need to do so on an ongoing basis. Once entered into corporate
maintenance databases this information can be used to evaluate future
supplier or vendor-supplied parts, procurements, and the suppliers
themselves. Decisions to upgrade existing parts need to be based upon
in-service performance, cost, and simple life-needs assessment. Its too
easy to see only the tip of the cost iceberginitial purchase price. The
total in-service cost can remain hidden beneath many other factors.
Overall costs include generation and service loss, as well as costs of
replacement and parts.
Part-use problems are compounded when multiple suppliers are
involved. No matter how carefully specified, manufacturers achieve dif-
ferent performance results. Mixed-part populations, in service, compli-
cate age exploration. Unless a company can design and run statistical
experimentsand few cansuppliers must be evaluated one at a time.
Find quality suppliers. Expect to pay more for their parts and serv-
ice. Develop long term relationships. This is the lesson on parts from
manufacturing.
Engineering Focus
When its included as part of an overall corporate strategy, age
exploration focuses corporate engineering on what matters. Issues of
205
redesign, statistics, cost analysis, and new CMMS tools can help put
engineering resources where they will add the greatest value.
Old and new facilities differ in their engineering design improve-
ment needs. In the past, rapid advances in design, lowered unit costs,
and load growth meant that engineering focused on construction. Plant
lifetimes were short. Disposable plants were expected to be techni-
cally obsolete after 40 years in service. Such was the design standard.
Today, plant replacement capital simply isnt available to old-line utili-
ties. Traditional, vertically-integrated utility generating units, whether
fossil or nuclear, are beset with high costs and complex processes, put-
ting their continued existence at risk. The challenge is to redesign and
re-deploy assets for competitive survival. When engineering groups lack
the experience needed to effectively improve plant operations, utility
engineering must turn to others-and this should not be the case.
Failure spectrums
A complete RCM review of a complex systemestablishing a main-
tenance spectrumquantifies optimum maintenance mix. At the
extreme are systems that support heavy monitoring: they have a high
number of random, low-consequence failures that dont (cost-effective-
ly) support fixed-time maintenance. Personal computersmany of
them controlling many complex subsystemsfit this profile. Overall,
they fail randomly, but a quality machines average age at failureits
MTBFis several years longer than prescribed useful life. Its highly
probably the device will be technically obsolete (and taken OOS)
before this point is reached. The MTBF is large20,000 hours or more.
Most failures are, in fact, randomly introduced software glitches or
random operational losses. Hardware failure incidence is low. There are
no effective tasks that will cost-effectively prevent failure so an effective
strategy is one that addresses failure identification and data preservation
instead.
The philosophy behind how equipment is designed also determines
its approach with respect to operator intervention, monitoring, and the
value placed on monitoring time. When system failure rates are low, it
demonstrates integrated man-machine design success. Time (in man-
hours) required to achieve failure rates may differ radically from one
206
Lessons
design to the next. If labor is valued low and capital requirements are
high, an overall optimum low cost solution is labor-intensive-a CNM-
intensive maintenance solution. If the cost of capital is low and man-
power high, the optimum mix is little manpower and more capital.
Here, the limiting case requires no man-hours for maintenance at all
the OTF case. Different cultures approach equipment maintenance dif-
ferently but often apply one of these two methods.
I think of the former model as German and the latter as
American because the way the design of a Porsche and a Chevrolet
reflect this different thinking. European maintenance strategies lean
towards more monitoring while Americans tend towards less. If equip-
ment is capable of extended life, we should seek that approach and
apply it for the optimum maintenance cost.
Operating costs can be reduced through reductions in capital
expenses, provided such reductions-or an increase in unreliability-dont
increase O&M. (One unplanned outage and all savings can be wiped
out) The purpose of many capital expenses are performance improve-
ment. Programs to extend life must assure against trade-offs-or, worse,
bottom line losses. Invariably, production losses carry high penalties but
are abstract and harder to quantify than PM. (How can we measure the
cost of opportunities lost when sales are missed?) Industry faces the
same opportunitiesand riskson a much larger scale. Every decision
in industry is a roll of the dice-and they roll hundreds or even thousands
of times a day. When we do, ineffective or incorrect strategies show up
on fairly short order. (Twelve to 24 months are needed to measure the
impact of a strategy change for average plant cases.)
Many technically advanced products carry specifications that assure
a specific design life at a specified level of performance. Boiler tubes will
last 40 years at design firing rates with specific water chemistry. (Firing
rate and chemistry specificationstechnical limitsassure design life.)
Exceeding these specifications causes immediate proximate failure.
Some books refer to this as engineering failure, or root cause failure. For
long-lived capital equipment, understanding this relationship reflects
business profound knowledge. Many companies do not (or cannot)
make this relational tie. Where equipment records indicate secondary
failures can be attributed to exceeding specifications, improved per-
formance monitoring can significantly improve economics. Where
207
expected plant life is 40 years, (with an institutionalized maintenance

strategy and plan) for instance, and an ultimate pre-obsolescent life of
80 years, extending equipment life improves future returns. This
reflects, for example, the future value of a sound water chemistry pro-
gram or firing a boiler within limits.
Monitoring identifies those specifications that are exceeded before
they convert into proximate failures. Minor adjustments and tweak-
ing are required constantly spanning 40 years, but long term benefits
are substantial if a unit that would otherwise be retired at 40 can sustain
80. Near-term, condensers and boiler sections re-tubed at 10-15 years
can achieve 30. Circulating water towers, ready to collapse at 18 years,
sustain 30.
Sophisticated maintenance practices include:
tailored performance monitoring plans based on vendor

recommendations
identifying problems early
addressing identified problems while minor, before the final
failure phase:
with redundancies in place or risk managed
while limits arent compromised
on a timely basis
planned replacements of known age-limited parts (including
lubrications)
quality parts and competent service
Plants and equipment become uneconomic when they become

unreliable. The context of R depends on the operating mission. As items
age, age-based modes failure probabilities gradually increase.
Eventually, failures increase. Managed carefully, the overall failure rate
for equipment can be controlled.
This final phase is the most challenging for operations within any
company: When do you pull out and re-capitalize or reinvest in an exist-
ing facility to restore R and performance? Effectively selecting alterna-
tives characterizes strong engineering programs; doing so poorly (or not
at all) ensures that failures will occur to make operating economics less
208
Lessons
favorable (faster), driving profits down more quickly. When costs

become unpredictable, operations become less economic, and finally-
uneconomic. Managing increases in aging unreliability allows a compa-
ny to determine facility end-of-life by technological and mission obso-
lescence means.
For an obsolete facility, increasing generating R and availability with
small capital investments provides attractive short-term earnings oppor-
tunities. For most aging plants, operating costs are highas are costs of
capital and risk. Such facilities remain competitive only by keeping their
total costs down. This points to improved O&M.
Maintenance can be abandoned in the final operational phase when
a company closes a facility. Such a decision precedes operational termi-
nation by some period, and is often irreversible, economically.
Railroads strapped for cash did this in the 50s and 60s. Once suitors
were found, or economics improved with deregulation in the 80s
many properties were exhausted to a point where abandonment was the
only economically viable decision. Extracting capital from a facility by
means of deferred maintenance should never be done lightly, or uncon-
sciously, as has happened in some companies strapped for cash.
By profiling typical plant systems to understand their technology
and basic maintenance process, RCM-based benchmark comparisons
help establish appropriate cost levels and identify effective methods to
manage production and control costs.
PM Implementation Models
Do your best
When some companies build or acquire facilities and maintain a
laissez-faire approach to facility maintenance, its because they discov-
ered they can run them much longer than vendor-specified intervals
with no apparent loss. Ultraconservative vendor intervals partly explain
ho-hum approaches to TBM.
Imagine, on the other hand, that missed PM intervals had (relative-
ly) immediate and severe consequences. Time-based monitoring pro-
grams and vendors would gain credibility! I believe this would happen
if manufacturers discarded the volumes of trivial, over-conservative
209
information they provide with much of their new equipment. Whats

needed is a Cliffs Notes of PM.
The truth is, few truly grasp vendor-recommended strategies for
maintenance. Vendor manuals (like company prospectus reports) are
under-rated and under-read. Their technical writing styles, required
skill levels, and even their basic information varies in quality and accu-
racyeven from the same vendor! Variation is even greater between
vendors. Some fail to disclose service life information. Others dont pro-
vide service manuals. Many dont offer in-service aging and perform-
ance information. Yet, in my opinion, the more product life cycle infor-
mation a vendor offers, the more competitive they become.
They may not offer product failure information because its
assumed that disclosure of this information is unwise competitively and
carries legal disadvantagesparticularly when the vendor recommen-
dations arent conservative and lawyers become involved. Vendors pro-
mote the virtues and longevity of their products in sales literature while
their service manuals are conservative. Development of an O&M strat-
egy rests heavily on the user, as a result. Vendors supply to sophisticat-
ed users but I have witnessed many disagreements between vendors and
users over product usage. Several notable ones were litigated; most were
resolved through user-vendor negotiations. Most were unnecessary.
Disagreements arise because maintenance requirements down-
played during the sale gain emphasis afterwards. To be economically
viable, equipment must deliver a period of reasonable, reliable service
with little or no maintenance. It needs to do this competitively. When
expectations are met, everyones happy. When not, accusations fly. Ive
personally evaluated failures as an owner but on some occasions,
problems developed based on our owner specifications, maintenance
performance, and other responsibilities. Vendors still worked with us-
they must have seen weaknesses in their own equipment or services or
they wouldnt have negotiated! What did everyone learn? General les-
sons included:
users can successfully exclude many performance-monitoring

recommendations and achieve reasonable performance
users who fail to act on known information and problems
incurred losses
210
Lessons
serious equipment shortcomings involved a bargain: An owner

cant economically pursue an incompetent, occasional, or in and
out supplier
quality companies meet their obligations; most go further than
owners can rightfully expect
conscientious, quality suppliers engage in frank discussions about
equipment, problems, and expectations. Often this is where large
operating organizations learn about operating limitations, despite
literature and training
Equipment vendors are expected to possess a high level of mainte-

nance awareness, but this exceeds reality in many cases. Between ven-
dor guidance and informal learning, many plant maintenance staff and
managers carry on with inexact, unspecified programs. Many are mere-
ly inferred from work practices. All maintenance departments strive to
be effective but only recently has effectiveness been defined.
Maintenance models-even those structured around CMMSs and work
practices-equate to defined maintenance processes. Some organizations
do work in regulated environments, and still dont have process maps.
ISO 9000 certification has driven maintenance process documentation
and definition. In recent memory, only the NRC regulations and the
maintenance rule has had greater impact.
I call the functional PM model the do your bestyou know what
that is model. Transferring performance responsibility from manage-
ment to performers, requires performance accountability, particularly
where there has been no objective performance measures. Doing
best can be highly subjective and support radically different out-
comes. Guidelines, expectations, and measurement may be vague or
completely absent. Craft worker guides can be non-existent or PM tasks
can lack specificity.
The old presumption, when specifying PM work, sounds like, PM
the pumppeople understand what to do. This has led to three
mechanics looking at three different things. RCM goes way beyond this
modeltoo far, in fact. ARCM appropriately stops at specifying what to
do.
211
Trust us (we know whats needed)

Indefinite PM programs are not written down yet adherents swear
they exist. Trust us says the staff. Undocumented plans lack field
performance consistency; yet bring this up and staff takes affront.
Informal audits reveal maintenance people who say they do PM
tasksthey know how, and they know why-but dont have cost, per-
formance measures, or other analysis. Since they have little experience
measuring performance, theyre concerned about measures. After all,
measurement has never been done before! Maintenance personnel with
only informal training in maintenance, PM theory, and methods gained
their knowledge on the job. That knowledge cant stand formal techni-
cal adequacy or value tests in many cases.
Of course, the acid test of a maintenance program is how it per-
forms. Continuous production increases and reasonable costs are satis-
factory program indicators. The reverse of thishigh failure rates and
steady or increasing costsare the warning signs of ineffective mainte-
nance.
Traditionally, increasing or unpredictable budgets were grandfa-
thered and the need for maintenance was universally recognizedwith-
in maintenance. General thumb rules and guidelines applied by the
manager were taken as gospel. These older, authoritative shops that
preceded the modern organization and CMMS produced some of the
best maintained plants around. Availability was good. But was money
spent effectively? Well never know. Worse, when the central authority
figure retired, it could be discovered that these leaders were the only
ones who knew the maintenance program strategy. In their absence, the
slate was wiped clean. Without a maintenance process, maintenance
cannot be developed and grown; the risk of a major redirection when
the central controlling figure changes is so high that the whole program
slides back to ground zero.
Such an intensely personal program cant be benchmarked,
improved, or even shared. Someone makes it work, on-site. The style
doesnt lend itself to standardization, large facilities and teams. Major
maintenance cost charges that were not controlled at the plant level and
budgets were largely swagged from previous historical performance,
with cost adjustments made for inflation and margin. Budgeting was
based on specific activity in principle-but not always practiced.
212
Lessons
With no competitive pressure and a cost-plus baseline budgeting

systems-this maintenance strategy suited regulated utilities for many
years. Many government and quasi-governmental agencies operated in
this same way. Todays maintenance strategy, technology, and theory are
changing expectations.
Today, the cry reflects former President Reagans arms limita-
tion motto: Trust...but verify!
When maintenance schedules are reduced to optional task lists,
they accumulate backlogs. These justify routine overtime hours, bud-
geted across the board. Maintenance can then work all the overtime
anyone wants. The only problem is that no one wants to work PMs. In
part this is because planned work can be deferredde-prioritized
and you cant justify overtime for it. Only emergencies warrant non-rou-
tine expenses, so planned work goes beggingand undone.
Discipline in managing the backlog is whats missing. It hasnt hap-
pened in part because of fear that regulatory authorities find backlog
trimming unacceptable. Yet, its convenient. Airlines, reactor operators,
chemical refiners, boiler operators, and other risky businesses must
have maintenance plans, by law. They must plan work, do TBM, and be
accountable for work performance on all plant equipment.
A better prioritization method would be activity-based, as it is in
accounting: All activities need to pass similar tests to be ranked for com-
mon resources. The automatic deferral of PM to crises is a fundamental
maintenance paradox. Priority work needs to be value-based. ARCM
provides value-based tests that can restore credibility to a maintenance
system.
Maintenance organizations occasionally downplay the value of
timed maintenance in a traditional hammer-and-wrench environment.
PM is not real maintenance, goes the jargon. If you can plan it
thats cheating! Combine this attitude with a utilitys tendency to let
workers self-direct (e.g., choose their work) and its no wonder that
PMs dont get done.
Typical PM implementation
In the unregulated maintenance arena, only a small number of PM
tasks are performed. The tasks themselves may not specify work to be
213
done. Review provides only vague notions about work to be done. From
years of plant experience, I can infer things worth doing, but auditing
work turns up many different results. The common trait is that the value
of many PM tasks cant be assessed or calculated. The plants entire
maintenance strategy may be suspect. At fossil generating stations Ive
audited 15% of the work on the PM list is performed-on the high end.
Typically its 7-10%. Low is 3-5%.
Nuclear plants have more aggressive lists and better measurement
because of regulatory requirements. Completion rates are regularly
between 80-95%. I watched a BWR achieve more than 90% of sched-
uled PMs worked to completion month in-and-out. Those not worked
were rescheduled. Failing to perform scheduled PMs had to be justified
in advance. Returning a PM to a backlog list was unacceptable. The
result? This unit did not suffer a plant trip in five years! Not that this
was solely due to PM completion-there were many other expectations
and practices that supported operations. But the culture was one of
commitment, competence, and maintenance across all groups
inspired by the regulatory environment.
Clearly, the contextual meaning of the PM program was radically
different in these two environments. To get this latter level of PM pro-
gram performance requires management commitment, and PM work
credibility.
Fossil plant R is a tribute to designthey run so well with so little.
But if most fossil units run well without a complete maintenance plan,
whats the upper performance limit? Would a more detailed plan bring
down performance and raise costs? What about other aspects of PM
performance, such as outages? My opinion is that more complete
strategies can raise production and lower costs.
Total PM performance
Some companies develop and execute maintenance strategies based
upon what I call total PM performance strategies. They aim to
enhance production and profitability goals by centering on facility uti-
lization. Ensuring high production, profitability, and performance facil-
ity utilization rates must be heavy and planned. The key to such projects
is that maintenance must support operating and facility use plansnot
214
Lessons
drive them. Exceptional operating organizations plan their mainte-

nance, using performance measures and production goals as feedback
tools. Some of the best (based on numbers) operate in areas not tradi-
tionally associated with heavy maintenance performance. At least one
prominent food service company is included.
Companies integrate these plans into their total operating strategies
in an effort to see maintenance support operations. They demonstrate
how maintenance can center on production-not provide a shop for peo-
ple who like to take apart and reassemble big, complex equipment at
their leisure!
To reinforce total PM performance, inspection, CBM, outage
scheduling, and TBM priorities should derive from production sched-
ules and goals. TBM (e.g., replace/restoration PM tasks and inspec-
tions) provides the structure for an operating facilitys scheduled work.
Developed week-by-week over a quarterly period, it can provide the
framework for a repetitive, long-term maintenance schedule. Combined
with outages and operations rounds, a comprehensive monitoring plan
takes shapeone that provides the overall foundation for implementing
a facility maintenance strategy. All the bases are covered.
To a traditionalist, this is backwards. Maintenance is delivered on
demand hed say. A PM can always be deferred. But to the R engineer,
applicable and effective TBM receives the top maintenance priority.
PM maintenance tasks avoid future losses and expenses. PM monitor-
ing tasks facilitate scheduling necessary maintenance. Applicability
means a task is technically appropriate and effective, as intended, when
performed by a qualified person. Effective means the value ratiocost
to benefitis favorable. When PMs are tied to economics, ranking
cost-effectiveness assures that high-benefit work precedes the rest.
PMs B/C ratios rank value. PMs with economic ratios less than 1.0
(e.g., the PMs benefit is less than its performance cost) are best per-
formed on a NSM basis. In those cases, equipment can self-disclose
maintenance requirements to operators on non-specific rounds or mon-
itoring. This is the most effective maintenance plan there is. The chal-
lenge is to get people thinking present valueone day at a time.
For companies that cant get the kids to play ball, there is the rail-
road approach: Take away the train set. (Railroads outsourced mainte-
215
nance unflinchingly in the 80s.) In this case, its called corrective main-
tenance. Contract all repair work to specialists so that all that remains
are PMs, performed in-house as the total responsibility of the remain-
ing workforce. This could be a multidisciplinary group-electrician, tech-
nician, and mechanic skills-supplemented with operations. This
approach clarifies, identifies, and prioritizes maintenance organization
work. The inherent conflict-of-interest between PMs and corrective
work for overtime, enjoyment, or other motivations is gone. PM is now
the only game in town. Selection of PM as core work is espoused by
some merchant generators.
Merchant co-generators tried this approach as an interim measure
because they lacked trained, skilled crafts. Onsite plant staff performed
all routine operations and light maintenance while outage and heavy
work was contracted out. Plant staff-clearly focused on the plant condi-
tion-used CNM as their primary tool to identify, diagnose, prioritize,
and plan outage and restorations. It was effective!
Vendor Perspective
The vendors dilemma is twofold. He must provide a good product
while generating sales. Ideally, he receives follow-up sale and service
calls for training, service, parts and so forthfor each customer. When
the client receives value and satisfaction from the equipment, the ven-
dors interest is best served when a product has a finite life. His best sit-
uation is technical, functional, or economic obsolescence before the end
of useful facility life occurs. The client retires the product in-service to
buy anotherunless the vendor can convince him to upgrade to some-
thing better.
Vendors are also repositories for product development knowledge.
In the course of their work, they must identify, understand, and remove
design, production, and operating impediments that cause failures.
They generally retain this information conveying it selectively to users.
Unfortunately, vendors cant provide complete failure data to equip-
ment owner/operators nor fully disclose product development and
applications, as they must protect competitive positions. They need to
exercise discretion in the event of legal action. In addition, plant tech-
216
Lessons
nical staff must understand and translate the operating data they receive
from the vendor. Generally, users dont need or require details about
the product. They might be intimidated by too much information
weak points and total costsor might comparison shop, or be steered
towards a competitor. Lastlyand most importantvendors dont
have complete information on actual in-service failures and aging per-
formance. They cannot possibly understand all conceivable environ-
mental and aging factors, applications, and uses imposed by the users
and their environments.
So, our dilemma is that vendor information, while good, is incom-
plete. It generally gives a fair assessment of common failures and
expected maintenance needs for anticipated service applications over
an intended period of use but doesnt provide environmental or appli-
cations information that summarizes a products stretch capacity
always the most exciting and challenging areas for users. However, ven-
dors are always a first source to identify both expectations and report-
ed experiences with new products.
If you can connect with a vendors engineering staff (assuming they
have one), you can resolve most questions with unpublished accounts
and experience for many product use applications and most failure his-
tory. Vendor engineers are more likely to offer critical failure informa-
tion over the phone than on paper.
Vendor recommendations
Vendor recommendations represent the best guide to maintenance
strategies that are appropriate for the equipment they offer. The quali-
ty of vendor recommended maintenance varies greatly. Some is truly
outstanding. Many dont provide any information at all. The vast major-
ity provides useful, but incomplete or sometimes inaccurate guidance.
At best, vendors provide a starting point, and so their O&M manuals,
sales literature, and drawings should be reviewed while developing any
maintenance strategy.
In a highly regulated environment, a vendors guidance may carry
the force of law. If the vendor specifies that a certain filter must be
changed every three months, you must change it. Rarely, however, are
vendors so direct-or consistent with plant time measures-in their guid-
217
ance. They may recommend lubrications based on service hours while

plants base it on calendar time. Its common for vendor tasks to be
inserted literally into CMMS scheduled PMs at the conservative recom-
mended interval. My review of gearbox lubes at a nuclear plant revealed
that most were changed on a one year interval, even though many would
not exceed their annual service hours during the entire licensed life of
the plant!
My experience is that most shop people trust vendors implicitly and
take their guidance without question. As a practical matter, however,
reviewing thousands of vendor guidelines has led me to the conclusion
that developing PMs is specialty work if you want to keep unnecessary
work to a minimum. It takes several years of plant, equipment, and fail-
ure experience to learn how to interpret and apply vendor recommen-
dations. You must understand a breadth of programsoverhauls, cals,
checksas well as traditional PM work tasks and you need to under-
stand how maintenance shops perform work. On top of this, you must
appreciate equipment functions, uses, and failures.
Failure Footprints
CMMS barriers
When performing ARCM, simple techniques can often significant-
ly improve analytical results. Typically, RCM analysis is highly abstract,
using esoteric, redundant, or even arcane statements. Many analysts
focus exclusively on expert interviews to flush out failure modes of
interest. While interviews are good, numbers tell the story as weve seen.
In my experience, the difficulty most people haveengineers
includedis penetrating corporate maintenance management systems.
CMMSs are difficult to learn and harder to interpret. In the role of
maintenance manager-at the mercy of others to develop and interpret
CMMS reports-I finally forced myself to learn how they work. Having
waded through the process several additional times at several compa-
nies and plants, I highly recommend anyone involved with maintenance
not yet fluent with these systems, reports, and numbers to learn at least
one. If you want to evangelize, you have to learn the language.
218
Lessons
Another CMMS barrier is report formats. Row upon row of unfor-

matted numbers and text represent little to uninformed readers.
Information is often coded and layouts must be compromised to stan-
dardize reports. Key record locations and formats must be learned.
Once you are able to skim CMMS reports, recognize key information,
and query interactively, you can use CMMSs effectively to understand
maintenance.
Most CMMS fields are not crucial. Important fields include the
component (equipment) identifier, component name, work type, and
description. For a time-based PM WO, a field should identify PM scope
and source. For corrective maintenance, the problem description origi-
nates and justifies the work. A basis (or analysis) that justifies doing the
work initiates a PM. CM is justified by a plant problem documented by
an observed symptom and problem description. Operators usually pre-
pare CM WOs because they first notice most problems while making
rounds, trying to operate equipment, reconfiguring, or just running the
plant.
The work done field complements the problem or scope
description. This field is the work performers assessment of the prob-
lem and its probable causes. Such information is only as good as the
report from the original work performer author/writer. So long as
descriptions are generally adequate, you can infer the scope of a prob-
lem and work completed based on two or three overview lines of work
done text. Practically, this is the most you can expect most mechanics
and technicians to write!
After participating in some CMMS redesign/rewrite efforts, Ive
become an advocate of commercial CMMS software products. Cost,
standardization, compatibility, and response-to-changes are a few of the
reasons why. Limitations imposed by in-house CMMS suppliers pose
severe restrictions on end users, yet, utilities have historically taken this
approach. Not surprising, few CMMSs are fully utilized in the field.
Many have extensive field variations. The basic process flows can be dif-
ferent, making software utility less.
If basic PM/CM work justification information is complete, and
the problem description/work done summary are completed faithful-
ly by work performers, then analytical failure study results can be ade-
219
quate. These two fields provide the most basic failure information nec-
essary for these studies. For CMMS users-particularly schedulers, plan-
ners, and workersthe maintenance process flow model and CMMS
must compliment one other. With the widespread application of file
transfer protocols (and products like Microsoft(c) [MS] Excel), its rel-
atively painless to export, sort, and reformat CMMS data for analysis
and presentation. (Day-to-day users are restricted by their unique sys-
tems and processes they support.)
Information displayed in a condensed report formatemphasizing
work initiation and completion fieldsenables a quick review of com-
ponents by type. An RCM failure-sample survey (by system and by com-
ponent type) provides both failures types and the work performed.
Several thousand WO reports are representative for an average sys-
tem. Typically, this encompasses one to three years of WOs for a fossil
plant and at least 30 for a nuclear facility.
I read the reports, placing tick marks for each failure group, and
then identify dominant failure modes. Based on the time period under
review, I make some rough estimates of MTBF and MTTR. In this way,
I can knock off a relatively large system in a day or so. My research into
system problems comes next, and takes more time, but the CMMSs can
be very effective when I download and sort vast quantities of informa-
tionexactly the uses CMMS promoters championed 20 years ago
(Table 6-1)!
These reports tell a story. They indicate aging-based failures, ran-
dom failures, and indeterminate areas that need more review. They
enable me to effectively structure questions for operators, maintenance
personnel, and engineering support personnel. Done well, these reports
can summarize failures in visual ways for storybook analysis and prob-
lem discussions. They provide a relevant basis for types of PM tasks and
their intervals (Fig. 6-2).
220
Lessons
Table 6-1: Tick Summary
OTF and No Planned Maintenance (NPM)

in the Real World
Failure
Unplanned failures occur, although functional failures are rare. Still,
signals can get missed, staff can be unaware of identified emerging fail-
ures, and instrumentation can be OOS. Failure patterns, conventional
wisdom, and misinterpreted conditions can lead us to miss the obvious.
221
Figure 6-2: Breaker Failure Summaries
Sometimes, organization cultures are a barrier to recognition.

Ignorance, fear, or inertia can combine with process barriers until a real
failure develops. Failure study is both morbidand fascinating. Any
unplanned event that compromises operating goals is a failure.
Unplanned events that dont compromise operating goals are the luck
of the draw. Everyone gets lucky to some degreealthough we usually
create our own luck.
Those companies adept at avoiding failures are those that work
hard at fundamentals. Those that understand failure modes and mech-
anismsdeveloping failure strategiesimprove their capacity to man-
age risk and so favorably influence their failure rate.
Imagine two utilities with opposite workplace cleanliness standards.
One staffs (and pays for) janitorsthe other doesnt. One has coal units
so clean you could easily mistake them for a nuclear plant. The other
guarantees your hands will turn black if you hold the handrails on stair-
222
Lessons
ways and ladderswhich you need to do for safety. One maintains cur-
rent plant drawingsno engineering job is complete until the drawings
are revised. The other never updates any post-construction drawings.
One plant finds time and money to maintain little extras, like ventilation
and air conditioning. The other cant seem to keep them upexcept for
the administrative offices.
Each plants policies offer powerful cues about management expec-
tations and equipment standards. The commitment to develop an O&M
planin contrast to a catch as catch can approach- delivers greater
confidence. Developing and maintaining standards takes fortitude. A line
is drawn in the sand. If you skip back and forth across the line too many
times, it becomes indistinct. Companies that set high standards out-per-
form those that do not, as reported in business books.
OTF in RCM
OTF simply means no scheduled maintenance tasks. The terms
run to failure and operate to failure are similar but generate nega-
tive interpretationsand thats the least of our problems.
Regulators have come to expect that everything has a planned
maintenance program and it must be understood that OTF is a planned
maintenance program. The plan is no scheduled maintenance like the
mathematical null solution. The work elements and failure modes can
still be virtually complete. Some nuclear plants must document their
null maintenance plan, literally. The inherent robustness of design cited
in the RCM classic by Nolan and Heap is, largely, beyond the grasp of
the general population, regulators, and particularly the media. Those
not versed in RCMand most people arentsimply dont understand
this distinction or its basis nor do they need to.
For the NSM option, there is no time-based WO to kick out, but
operating staff and those in the plant must identify and respond to
symptoms. NSM programs depend on their personal and informal diag-
nostic skill and knowledge though the tasks are non-specific or not
scheduled. This also illustrates why RCM-based, non-specific tasks
need to be ruthlessly purged from the CMMS task list. For operators,
knowing the plan depends on their initial condition assessment.
Removing redundant CMMS tasks encourages more thoroughness in
223
their condition-assessment learning and application.

Why does a plan of NSM work? Why, statistically, is it appropriate
for 80% of the equipment in complex plantsperhaps more? Three
reasons: people, design, and equipment.
People monitor. Design adds complexity and minimizes dominant
failure modes. Evolutionary equipment is inherently reliable. The line
between a condition-directed task performed by an operator on his
rounds and a non-specific operators CNM task gets blurry as operator
skill levels increase. Highly skilled operators with many years of experi-
ence apply their knowledge in every monitoring situation, often above
and beyond the call of duty. We may benefit but we havent planned on
it. And their absence has only minor impact on our programby
design.
Worker learning and knowledge buttress NSM. When the work
force has a high level of equipment understanding and job commitment,
informal maintenance programs like NSM work well. Operators sched-
ule inspection tasks while on their rounds. Between rounds, they play
free-form CNM, so to speak. What they find, through skill and atten-
tion, highlights problems that require further, specific follow-up
(though equipment that qualifies for NSM has minor or mitigated
plant-failure impact). Maintenance can be planned and scheduled with
virtually no safety risk.
NSM should never be used to ignore equipment. Once failure iden-
tifies equipment needs, work goes into a queue. Because the work it
identifies originates from non-specific rounds, NSM doesnt downplay
importanceits developing failure mode establishes it! It prioritizes
work with all other scheduled CM tasks, non-specific tasks, and sched-
uled maintenance. Although NSM-originated maintenance could be of
higher priority, thats highly improbable. Rather, NSM work is general-
ly lower priority because of the redundancy and mitigation capacity
inherent in the design. NSM effectiveness, (like CNM monitoring)
depends on the organizations ability to monitor to identify condition
and then to include and manage CBM maintenance work with priorities
in place for the planned maintenance process. Weekly reviews reshuf-
fle the hot and not planned work schedules. The schedule is dynam-
ic during the planning phase, but once a job is issued for work, resched-
224
Lessons
ules should occur only very infrequently. Routine, repetitive jobs need
standard corrective plans developed, planned, and shelved to work
on demand. Sophistication improves as people gain insights into plant
design depth-a depth developed with routine ARCM applications.
The term self-identifying maintenance also sheds light on both
NSM and OTF. The bottom line of that plan is that we dont schedule
formal tasks to perform maintenance - nothing more or less.
Equipment maintenance needs are self-identified by the equipment.
This is acceptable because of inherent R-the absence of known, effec-
tive age-based PM measures-and limited consequences of failure. No
one in their right mind would do any maintenance to a car headlamp
other than replace it upon failure. On failure, however, its very impor-
tant to replace promptly. Its the same concept.
Legitimate failure
Actual failures occur when we violate performance standards. If
standards are established, its unnecessary to discuss failure criteria. In
their absence we have nothing but a discussion about what constitutes
a failurea discussion beneficial only to the degree that it leads to
common failure definitions. The exercise is pointless-except to develop
failure standards. Without standards, people cannot agree on what is
important and what is failed. For better or worse, nuclear plants have
many guidance standards in place. Failures are well known. For fossil
plants, the concept is abstract. Failure tends to follow a free-form defi-
nition, literally linked to a primary function failure. At plants with spec-
ifications, failures are specification-based and incremental.
An RCM-based approach to failure definition forces people to
think about goals and limits, which in turn leads to earlier action. Goals
can be obscure to operating staffs. Obvious limits on measured vari-
ables such as opacity and emissions, material thickness and production,
can be missed. It could be argued that exceeded specifications and lim-
its are in themselves arbitrary failures, since in most cases violation does
not cause sudden, discrete events. Rather, an engineering limit has been
exceeded. Real failure comes some time later and with continued loss of
margin. Design specifications have margins thatif properly fol-
lowedsafeguard us from real proximate failure areas. Real proxi-
mate failure consequences include:
225
operating events that compromise company strategic goals

unplanned, unscheduled production losses
equipment losses that compromise design and operating margins
Risk correlates with real failure experience. A high risk plant

experiences more proximate failures. Causes are diverse, but result
from chronically pushing design margins. Plant practicesshorted time
for PM and monitoring, missed CDM cues, use of design redundancy,
or failing to operate within prescribed standardslead to higher failure
rates.
Examples of Group (1) failures include:
major accidents
excessive accident rates
excessive environmental releases
Group (2) includes:
unplanned unit outages

restrictions
unplanned plant system losses
unacceptable loss of production equipment margins
Group (3) includes:
failure to observe warning instruments

overriding interlocks
operating above tube temperature limits
overfiring
Convergence: OTF-CBM equivalence

As much as OTF and CDM appear to be extremes in strategy, they
are closely tied. The resulting condition-directed WO looks identical.
Equipment has no memory of whether condition deterioration was
caught by a formally scheduled task with a WO or by an inquisitive and
alert operator. Both cases are discovered in advance of any production,
226
Lessons
safety, or environmental impact; in both cases, the resulting mainte-

nance can be planned and scheduled (e.g., function failure hasnt
occurred). Both tasks become common work items in the condition-
directed (formerly corrective) maintenance backlog.
The difference is in how they originated. One originates with a
time-based, OCM task. The other begins when an operator notices
something that is not right. The former condition has a performance
limit; the latter may or may not. In maintenance performance, each type
of monitoring can provide the same outcome but the outcomes require
different levels of interpretation.
Since an OCM task that generates a CDM activity is tuned to a
specific failure mode, it should be easy to perform follow-up CDM. The
proximate failure should be clear and unambiguous. Functional fail-
urethough not clearly traceable as a functional lossshould be clear-
ly evident. On the other hand, non-specific failures are ambiguous.
Operator ability discriminates wheat from the chaff.
Indeed, non-specific monitoring depends on the operators discre-
tion, skill, training, and judgement. Establishing standards for monitor-
ing, and then training the operators, can help hone their monitoring
skills. Database structures show OTF monitoring response to be virtu-
ally the same as that for CNM. The distinction is in the initiating task
planned, scheduled, or none at all.
One could argue that identifying a task in the OTF bucket is arbi-
trary. Because the lions share of formally scheduled monitoring tasks
are operations-orientedfailure-finding tests and specs-performance
monitoringoperations performs most ARCM-derived CNM. In
CNM, this formal leadership distinction is, theoretically, reserved for
highly skilled and experienced personnel. Maintenance-performed
time-based monitoring is much more task-oriented. Maintenance per-
forms monitoring by WO. No WOno work.
This can become a legal exercisewhich ARCM is notbut the
key point to remember about CDM task WOs is that when a true-blue
OCM task appears:
its formally scheduled

its specific to a failure
227
its actionable (a discrete go/no-go outcome)

it has a specific limit, margin, measurement, or attribute that
unambiguously (to the trained analyst) is go or no-go failure
evidence
OTF equipment self-identifies failure. We dont have to extract fail-

ure elements with sophisticated tools like vibration analysis or ther-
mography because the benefit isnt there. Some operators possess a
sixth sense for identifying failing equipment (the power plant operators
version of a green thumb?). They anticipate failures with scant physical
evidence and often, their intuition is correct. They relish the diagnostic
roleits all they do. This demonstrates how the distinction between
CBM- and OTF-based maintenance can be rather arbitrary.
Mathematically, OTF is simply a case of OCM without the usual
time limits. The formal scheduled PM interval is taken to infinity. Can
the PM identify the failure if it happens? You betif its technically
effective and the failure occurs. Is it specifically worth doing? Generally
not. For a facility with a 40 year lifetime, maintenance performed on
even a 20 year interval is approaching plant-life limits. At such an inter-
val, such tasks look as if they are taken out towards infinity. They
qualify strongly for non-specific monitoring if the return is small. This
demonstrates how excessive amounts spent on PM can raise costs.
When all is said and done, OTF is rounds-based, non-specific oper-
ator monitoring. Everyone knows trained operators walking through
the plant can accomplish a lot on experience, whether specifically
directed or not. The lesson, pure and simple, is to keep it simple but
keep it up! ARCM offers a streamlined method to identify the moni-
toring tasks expected of operations, and establish appropriate intervals
to make them effective.
Complexity in failures
Failures have contextssimple or complex. Simple failures are eas-
ier to manage so we naturally prefer them to complex incidents.
Complex failures involve multiple failed items and interactionsinter-
actions that make them hard to diagnose. Multiple hidden failures are
harder to recognize, interpret, and correct. These include multiple coin-
228
Lessons
cident failures in complex equipment trains.

RCM analysis helps familiarize us with the spectrum of the antici-
pated failures. We are made aware of likely failures as well as prevention
strategies and counter-measures. Task performance can be made easier
by developing fault trees for complex system failures. They enable us to
step through the failure analysis quickly, and because fault trees can pre-
develop the probable failure modes, actual diagnostic performance take
less time. When failure events and their likelihood are established up
front, developing preventative measures also comes easier.
Bootstrapping
ARCM provides us with many ways to approach plant operations
improvement. Each can add value quickly. Once specific equipment
and subsystem analysis is complete, results can be transferred to similar
units with little effort. Applying previous learning to similar units and
systems without any formal detailed analysis is a form of bootstrap-
ping.
Contrary to TRCM analysis, power plant design is highly standard-
ized. All steam turbines use a Rankine cycle, for example. Suppliers are
the same in a given region of the world. Large generating facilities share
considerable standardization of equipment, systems, and layouts. A few
configurations have been developed and see many repetitions. Even
informal standards have proven their utility over time. Thus, once a
basic repertoire of RCM equipment and systems understanding is in
place, it forms the core for many common applications.
Systems. Common system configurations abound. In fossil plants,
feedwater, condensate, sootblowing air, and circulating water are very
similar from unit to unit. In nuclear plants, the GE BWR and the
Combustion Engineering (CE), Westinghouse, and Babcock & Wilcox
(B&W) PWRs share common design elements. This supports stan-
dardized RCM analysis.
Occasionally new systems are integrated into traditional ones, and
this requires system re-analysis.
Equipment. Most common equipment in power generating facili-
ties is supplied by two or three primary suppliers. Even where there are
many suppliersas in the cases of valves and motorstheres so much
229
functional commonality that many failure CNM tasks transfer directly.

This commonality supports the development of standard maintenance
strategies at the equipment level. While the relative frequency of the
major failure modes shifts from plant to plant and environment to envi-
ronment, the primary modes tend to be the same. Where they differ,
theres often a different operational strategy or use at work.
Commonality supports standards development. Using the craft to
identify and select commonly encountered failures and select strategies
is an effective way to focus general information and solicit buy-in. For
simple PM work formalized development and application of standards
eliminates large amounts of low value workthe annual replacement of
gearbox oil in a very clean plant environment, or monitoring and
replacement task intervals for hoists, cranes, and other infrequently
used equipment. It can lead to major labor savings.
Reduction in the frequency of parts replacements is another conse-
quence. Small stock items are carried because manufacturers recom-
mend it, and theyre replaced on intervals to support 24 hour, round-
the-clock operations. In standby, this equipment may not see the man-
ufacturers full year of operation over the life of the plant! To service it
on the manufacturers suggested interval is usually way too frequent.
Processes. Developing an RCM-based maintenance program
invariably forces an organization to review processes. Virtually all of
them must be tuned and new ones added, including cost and measure-
ment. This stressful exercise partly explains why many RCM-based
maintenance plans fail. Without a vision and commitment to change,
forces naturally align to resist.
Process redefinition, however, though painful, is also the most ben-
eficial long term aspect of an RCM-based review. Refocusing work
around a PMO philosophy is fundamental. New skills, perspective, and
commitment, and long term benefitssuch as more directly compara-
ble cost information between users, more useful information break-
down, and performance measurementmake the pain tolerable.
Potential maintenance savings of around 40% have been cited in vari-
ous electrical trade journals, studies and specific cases as the benefit
opportunity derived from improved maintenance.
In contrast with another popular process redefinition methodre-
230
Lessons
engineeringan RCM approach offers specific methods.

Organizations can use RCM approaches in discrete incremental
amounts, for example. Improved PM measurement can be initiated on
a CMMS with the introduction of failure criteria. New terminology can
be phased within a time. Decisions about software, basis maintenance,
training, and other issues can be implemented with only minor organi-
zation disruption.
The downside to incremental change is that organizations dilute les-
sons. More than one PMO effort has flagged and failed this way. You
may be successful at introducing new maintenance methods, but when
the old remains in place, an incremental approach is sure to fail.
Critical?
Traditional RCM definition
Traditionally, commercial air transport RCM reserves the term crit-
ical for those failures that have an immediate and direct safety impact.
A critical failure is any failure that could have a direct effect on safe-
ty. (Nolan & Heap) Note the word direct imposes specific qualifying
criteria not imposed elsewhere. This qualification excludes failures that
arent immediately evident, or werent single failures. For non-evident
failures, the absence of evidence means there is no direct impact on
safety. The immediate requirement screens out multiple, train-redun-
dant failures.
The first major permutation of RCMs definition of critical
occurred in the transition from aerospace to nuclear power. NRC defi-
nitions for what are now called essential componentsthose whose
failure could affect fission product environmental releasesgave crit-
ical a new dimension. Since many nuclear components occupy this cat-
egory, critical applications grew by default. The tendency to associate
the term with specific components, rather than failures, compounded
confusion.
The final application of the term critical to non-nuclear units,
produced a flow process that divided analysis into critical and non-crit-
ical. Dividing components in this way (a la nuclear) proved confusing as
fossil plants struggled to abandon their historical critical interpreta-
231
tion associated with production impact. By failing to rigorously apply

the original criteria, and indiscriminately applying it on the basis of pro-
duction impact, critical scope grew. The R engineering definition
from FMECA diverged, as well. Critical was based on a numerical
value that only had meaning as a relative ranking in the context of all
other equipment in the analysis.
Casual use
Many of us, occasionally use critical casually. Those SRCM meth-
ods that divide equipment into two broad categories-critical and
non-critical-depend upon the equipment involved, and determine
whether the review selected involved is a thorough failure review or a
quick and dirty (cost-based) sanity check. After evaluating thousands
of components, Ive found that the RCM process that identifies com-
ponents for PM is not that important. Most analysts can quickly deter-
mine a components suitability for PM, and the likelihood the PM is
effective. Whether a non-technical person can follow their analysis is
another matter! If your interest is solely the final product-an effective
PM program-you probably dont need the extra information. If you
must maintain it, you do!
One plant in which I worked developed an automated prioritization
CMMS that identified all equipment as critical (or non-critical) and elec-
tronically pre-assigned priority to the equipment WO. This system ulti-
mately produced a disproportionate number of critical, high-priority,
work-today WOs. Because no one had the inclination to override
default rank and rank any priority low, the predictable results were that
virtually everythingexcept PMswere critical. This was not only
not useful, but effectively eliminated meaningful priority.
The primary work prioritization/screening tool depended on two
attributes-is it emergency or deferrable? The tasks most likely to relieve
the workload, long termPMs-rarely made the cut.
The primary purpose of a priority system is to rank the importance
of workquickly. If there is no discernible priority attribute, or its
skewed, then the system-no matter how ingenuous at the software level-
has little value. This system had no value. When truly critical equip-
ment fails, it causes unsafe conditions. Economically critical equip-
232
Lessons
ment failures cause unit outages or other obvious design-intent and sys-
tem-function failures that relate to bottom lines, not safety.
Truly critical failure modes. People dont think in terms of failure
modes. Understanding failures does not come easily. Like any skill, it
takes years of experience and missed calls to get rightmost of the
time. Systems understanding comes first. Many people in responsible
positions havent had the time to hone this skill but are charged with
understanding plant risk and making safe, effective operate/shutdown
decisions. Such managers rarely make good candidates for failure engi-
neersthey like black and white cases, and clear-cut calls. (And,
indeed, someone has to make a call.)
Its extremely important to clarify distinctions among safety, pro-
duction, and economics, to best allocate scheduled maintenance
resources. After 15 years of discussion and analysis, I believe that pro-
ducers are inherently safer that non-producers. Plants that operate-just
like cars that put on miles with no breakdowns-have to be in good
hands to be able to do so. Its rare to find top performers that dont also
put operational safety at the top of the list.
Just as equipment groups are similar, so uniquedifferentcriti-
cal failure modes are relatively scarce. What they share in common is
the fact that most equipment potentially presents life-threatening failure
modes, eventually. When we buy it, get good service out of it, and expe-
rience its performance deterioration so legions of engineers have to
scratch their heads arguing over whether the time for overhaul (or
replacement, as the case may be) has comethats the way we like it!
Operations at this point wants a new onepump, compressor, valves,
belt, whateverbut understands the old beast well and so gets more
mileage from it.
Wearout. Wearout is the desired end-state for every component in
an operating organization. When equipment is worn out, evenly and
well, meeting a manufacturers promised life, it represents an ideal. Its
a matter of gradual, predictable performance loss, providing lots of lat-
itude to schedule replacementand manage risks by scheduled main-
tenance. Ideally, turbines age this way between overhauls, as full turbine
load gradually trends towards valves wide open (VWO) position.
Centrifugal pumps show gradual loss of head, as the rotating elements
deteriorate and seals wear.
233
Wearout is an important failure mode and the most desirable out-

come. Examples include:
loss of balance on high-speed rotating equipment

loss of steam (or other high pressure fluid) boundary
cracks
gaskets
ruptures
packing blowout
loss of lubrication on heavily loaded parts in relative motion
lubrication aging
contamination
aging breakdown
loading
Wearout relates to critical failure modes as one end of spectrum of

concern that can be lumped under two, maybe three groups. Consider
automobile tires.
Wearoutthe ideal and key failure that ultimately will end the life
of every tireis balanced at the other extreme by random failures. In
a car tire, it can be the unpredictable blowout or loss of air from punc-
ture, glass, debris, or even driving over a curb. Last, theres loss of air.
Its possible that a tire may not lose any air over its service life, but most
will.
Tire manuals offer many examples of rare event type failures.
Scalloping, uneven wear, ply separation, rubber breakdown from chem-
ical attack, stem failure...typical drivers see one of these failures once or
twice in a lifetime (assuming you maintain your car!). Secondary fail-
ures-from imbalance, improper adjustment, unintended service, sus-
pension damage, worn ball joints, and the like-can also be induced eas-
ily.
Confusion implications. The greatest downside to design redun-
dancy is the ambiguity it introduces into the definition of criticality.
After all-using our tire analogy-is a tire critical if you have a spare (and
the means to change it, and the free time to do so)? Power plants have
many redundant critical spares. Standby boiler feedpumps, redun-
234
Lessons
dant buses with auto-transfer breakers, spare turbine gland steam

exhausters...so long as the redundant feature is available and the correct
transfer sequence occurs, loss of the primary component has no impact.
Hence, it cant be critical.
Or can it? What if the backup fails or the auto start sequence fails,
or the backup is OOS or performance-impaired? In these cases, pro-
duction or other essential functions are lost the primary is critical.
This ambiguity is another reason why I dislike indiscriminate use of
the term critical. In some sense, every item in a plant is critical-or
mis-specified. i.e., There shouldnt be any equipment in a facility that
has no functional value. If this is the case, it should be abandoned with
no impact. What is critical is outcome. A critical failure is when some-
ones hurt-a non-critical when theyre not. In this context critical
reflects the airlines direct safety consequences interpretation.
Importance
Given that critical will probably endure as a named condition,
though in confusing permutations, lets consider one more attempt to
clarify and simplify its use.
Safety and cost drive all PM while safety and economics drive all
plant operations. Ignoring economics fails to adequately address our
sole operations purpose. For this reason, economically critical is an
acceptable concept-provided we restrict its application to failures. The
primary consequence of most safety equipment failures are operational;
we must terminate operations to address a key safety function. Using
our functional definition of failures, we could agree to use another term
for equipment classification based upon economics, then reserve the
term critical for safety functions and their failures.
We would then have two classes of equipmentimportant and non-
importantand a sole criteria for classification: whether we plan to
consider scheduled maintenance for the item or not. We could just as
well identify these as scheduled maintenance and non-scheduled
maintenance. Once past this barrier, we can review equipment for
applicable and effective PM tasks.
Practically, a reviewer looks at the following identifiers to discern
235
suitability of scheduled maintenance:
equipment size
failure reporting frequency
work frequency (including PM)
vendor recommendations
general industry practice
shop practice
equipment register
Equipment size. Large equipment is expensive. Safety considera-

tions are based upon enclosed energy and fluids. Large items are always
reviewed for PM activity. In addition, since the work scope involved in
opening large equipment is considerable, theyre expensive to maintain.
Documented PM or maintenance costs usually confirm this. Large
equipment vendor manuals and other recommendations usually suggest
appropriate safety and operations concerns, as well as cost-effective
tasks.
Failure reports. There are two sources of failure reports: operating
events and WOs. Until these are screened-ranked by criteria and evalu-
ated-the nature of the failure is merely supposition. Practically, any
equipment that carries operations impact needs careful review for
potential PM tasks; if operations impact is absent, equipment with sig-
nificant failure frequency warrants deeper review. Depending on the
plants equipment tag number coding detail, such WOs may be elec-
tronically hidden for some equipment. A comprehensive review will
always turn them up.
Operating event records are only as good as the level and detail of
the operating logs. In each case, log review conclusions should be
reviewed with operators and maintenance staff to confirm trends, as
well as flush out items not included in the formal records. Operating
groups with a shift rotation of more than five people, providing plant
coverage, should maintain a written operating log. Where logs are inad-
equate or lack detail, standardize them. In one event I know of, initiat-
ing new records to document the cost of operational failures was instru-
mental in gaining corporate support for capital investments to lower
236
Lessons
costs.
Work frequency. Systems that are costly to operate and maintain
demand that much extra work-and they require PM programs, WOs, or
both. Outage and contracted work not captured in plant WO systems
should be reviewed for scope that reflects maintenance tasks. Acid
tests for large work scope should be considered-is a basis established?
Does the plant know why the work is done, and what its performance
benefits are? Occasionally, capital modification work performed in off-
budget areas is charged as a maintenance expense, especially when engi-
neering groups perform maintenance support functions. This skews
costs.
Vendor recommendations. Vendor literature should be reviewed to
assure that vendor-identified work has been assessed. Vendors general-
ly understand their equipments economics better than anyone else.
They also appreciate safety considerations, although they may not
understand specific applications. Vendor recommendations often turn
up interesting insights (and oversights) that influence large equipment
costs. Small equipment that lacks obvious, integrated functions is usu-
ally NSM by definition, but reviewing this against the vendor recom-
mendations can identify tasks the vendor thought were cost effective to
perform, even for generic equipment.
Comparing recommendations from similar vendors is an effective
way to establish appropriate tasks and performance intervals. Many
times upgrades and enhancements influence maintenance frequency-
like a superior synthetic lubricant that extends a lubrication perform-
ance interval, for instance.
Industry practice. Benchmarking equipment for standard industry
practice is another effective way to establish appropriate levels of effort.
Comparing one industry to another, when both run the same equip-
ment, can provide insights. For example, how mines handle coal in dif-
ferent locations supports cost-effective improvements for utility coal
handling operations. As with all benchmarking, understanding the
methods and practices behind the numbers is essential to making
appropriate choices.
Shop practice. Every shop develops techniques to manage work
performance. Often some of them are unique and effective tasks that
237
can be adapted to other areas.

Once, working at a chemical plant, a worker pointed out a simple
method to determine bearing wear on a large fabric-making machine.
The applicability and efficiency of this simple quick test for bearing
tightness was a classic example of shop learning and creativity. The test
was faster and more effective at locating bad bearings on a 17 main
bearing drive than any other method Ive seen since.
Equipment registers. At initial plant startup, an equipment CMMS
register or hierarchy is created based upon design engineering or
accounting descriptions. These lists of equipment should be considered
for PM. High cost capital- and skid-mounted equipment-along with
equipment that has been erected on site-goes into the register.
Reviewing this list for PM candidates is an effective way to assure that
everything gets considered. Design modifications that support the reg-
ister also need review.
Area Checks
When operations personnel perform area checks, theyre engaged
in NSM. Area checks require operator rounds, to check the condition
of equipment installed in the plant. This was the original intent of the
hourly round.
In fossil generation, an operators complete round can take two
hours or more, even at a brisk pace. And brisk is not the point-you
must slow down to read instruments. Often, lighting and cleanliness can
make equipment monitoring additionally time consuming (especially
when gauges are dusted in coal or oil mist.) When I identify hourly
rounds sheets (except for the control room), Im immediately skeptical
of their effectiveness and applicability.
Yet, rounds are important. In complex plants, many failures are ran-
dom, and actual system functional failures are rare. Because of this, its
essential that the operator on a round identifies failing (and failed)
equipment. A log entry or CMMS trouble report are techniques to iden-
tify failures. After a complete ARCM review at a nuclear plant (review-
ing approximately 100,000 components), we found that the over-
whelming default PM activity, numerically, was NSM. This plant had
238
Lessons
exceptional operating rounds. Plant operators are uniquely positioned

to identify failed equipment.
This functional role was what was intended by the hourly rounds in
the U.S. Navy. Its the same for commercial operators: Monitoring is the
operators principle role except during startups, shutdowns, and prepa-
ration for maintenance.
Random failure identification is an operator value-adder. To be
effective, operators must perform an adequate, failure-based round
that is, a round that meets the effectiveness criteria discussed earlier.
Heres where ARCM criteria can help.
When developing monitoring requirements for systems and equip-
ment, ARCM methodology helps develop rounds in a shorter, less-arbi-
trary way. The net benefit is a better, more specific, actionable, and
focused rounds. Performing round reviews can achieve significant
improvements in operating plants. An RCM system review assists oper-
ators by identifying:
functionally important equipment (and their failure modes)

analyzed NSM equipment checks (important failure checks to be
done by operators)
formal, defined checks (for rounds) and observed process and
equipment parameters
optimum round intervals, based on MTBF
area checks
An area check is a general survey that integrates the senses and non-
specifically identifies failures. Like functional tests at the system level,
they integrate and pick up major failure indicators. Theyre like the
area check that driver education courses suggest you perform before
you get in your car and drive away. Airline personnel make them every
time a plane takes off, confirming the absence of functional problems
for various basic, yet critical components. They also enable the quick
discovery of problems that could have serious consequences over time.
Area checks in plants are also cost-effective ways to identify random
and general deterioration failures. Clean, well-maintained equipment
provides unambiguous results. If standards drop, however, and dirty
239
conditions develop, area check effectiveness drops. Cleanliness is essen-

tial in plant environments. Clean, organized areas enhance failure detec-
tion. A Japanese maintenance process called TQM places great empha-
sis on equipment cleaning as a tool to discover incipient failures.
This suggests other requirements for effective area checking.
Poorly illuminated areas cant be monitored effectively, so area check
value drops commensurately. Coal facilities can be dark and dirty.
Aging paint reflects less light. One mechanic friend refers to his fossil
plant as the cave. It consistently underperforms and suffers from
perennial operating and other performance shortcomings.
What level of cleanliness is appropriate? Its hard to say. I know of
one coal-fired facility operated as a low cost producer that maintained
a janitorial staff of eight exclusively for what was a (2) 550 MW unit
facility. With availability well above 90% and capacity factor close to
80%, it must be doing something right!
Area check strategies also work well for complex fluid system flows
where leakage can be identified by the trained eye (or the other senses),
or by sense-enhancing equipment (such as ultrasonic detectors or ther-
mograph imagers). They fit well with operator rounds. Examples of
specific area check failure targets include:
lube oil reservoir leaks

service water leaks
hydraulic leaks
instrument air (or other gas) system leaks
Obviously, as operators gain experience and skill, their area checks

become more effective. Innovative, as well! Operators of a large coal
facility, running volatile morpheline water chemistry treatments, point-
ed out to me one of the most effective area checks I ever encountered:
smell. Its extremely sensitive and the fastest way to identify a steam
leak: morpheline has a distinctive, pungent smell. In coal handling
areas, the pungent smell of burning coal is an effective tool to locate
smoldering coal piles in belt galleries and bunkers. Chlorine and sulfur
dioxide systems can quickly identify leakage by smell well in advance of
acute safety concerns.
240
Lessons
Instrumentation
The typical plant has a tremendous amount of instrumentation and
control equipment. Much of it was installed or packaged with equip-
ment skids, provided by the manufacturer/assembler. Many instru-
ments have as their primary function the setup and performance of pre-
operational or operational testing-lube oil skids, for example.
Equipment suppliers often provide remote panels for pre-operational
testing and operations. A typical plants equipment, augmented by skid
I&C, provides so much instrumentation that to tackle TBM of all of it
would be a Herculean task!
Major process control loops feed a plants DCS. This is done
through drops, in modern plants. The typical DCS runs two redundant
independent buses with self-checking diagnostics and the capability to
swap buses, should a problem occur. Each has a fully redundant back-
up with the same capability. These robust systems havent been the
focus of a detailed RCM assessment by many clients, and I&C techni-
cians and engineers, by and large, effectively maintain DCS controls.
This suggests low value added benefit here at the typical plant installa-
tion. Risk management, maintaining redundancy, and depth have
thus far been very effective. However, there are I&C opportunities.
The first is to select and identify candidates for NSM. I&C PMs
include cals, channel checks (CCs), and functional tests (FTs). Self-diag-
nostic equipment can reduce or eliminate the need to perform FT. Self-
calibration routines can eliminate the need to calibrate. Typically, a
trouble alarm sounds if a plant DCS loses a drop or channel. Periodic
checking of the channel alarm is all thats required.
Its tempting to feed every instrumentation point in a plant into a
new DCS during an upgrade project. However, large fossil units could
have 5,000 points fed into the DCS. For important equipment (like boil-
er feedpumps), this enables dropping more points than original plant
data logging could accommodate. More information is available on-line
than previously available, or available locally at the feedpump skid
(such as local hydraulic and control oil pressure and temperature) but
the downside is that we must maintain instruments-including the low
value instruments. Points fed into the DCS need conservative selection
with an operating monitoring strategy in mind. Selection should not
replicate an existing monitoring strategy. Extraneous nice to have
241
instrumentation can otherwise result.

Estimating installation costs (at one man-hour per point), a new
DCS installation can cost nearly 5,000 labor hours for point drops
alone. If 50% of the points get no functional use, then non value added
instrumentation costs $125,000 (2,500 x $50). ARCM in I&C programs
not only provides guidance to control the installation, but also the main-
tenance costs and application selection of low- or non-value instrumen-
tation.
Two key instrumentation applications are loop cals and functional
checks. Active control loops require calibration as determined by drift.
Functional checks of essential safety, production, or compliance alarms
is also necessary on an interval determined by alarm failure risk.
Examples include:
safety: steam-driven BFP vibration alerts

production: deaerator level alarms
compliance: continuous emission monitor (CEM) flow, opacity,
and species (and other alarms)
DCS systems allow taking alarms out-of-scan-for-nuisance status

and other low-level applications. While appropriate for temporary
problems, out-of-scan status can be forgotten. Routine checks for out-
of-scan alarms are a necessary scheduled task.
Spurious alarms
Consider the new car whose seat belt monitor tells you to buckle
up, over and over, as you cruise down the highway. Most people can
take about five minutes of that before they pull the plug-assuming they
can find it. While searching for it, theyre a safety hazard (unless they
pull over). Nuisance alarms are more than a nuisance-theyre a distrac-
tion and a potential safety problem.
In the context of RCM instrument functionproviding a clear,
unambiguous picturea nuisance alarm is a failed alarm! Critical
alarms should be corrected. If non-critical (e.g., you can safely tolerate
their absence for long periods), remove them.
Screening I&C calibration intervals for extension based on drift
242
Lessons
experience and importance can reduce and simplify I&C work. In fos-
sil plants, calibration frequency can be adjusted (extended) by installing
new controls, but it should be considered in all plants. Many times,
overly conservative intervals persist long after theyve been identified.
Excepting very old pneumatic control devices, there are many opportu-
nities to calibrate less often. Newer sensors and instruments can reduce
maintenance requirements by significantly reducing drift. Reducing
efforts expended on non-essential control loops calibration enables
more consistent focus on key control loops and essential alarms.
Critical instruments
I qualify my reservations about using the term critical with one
exceptioninstrumentation. The reason for this is very simple. Key,
essential safety and monitoring control instrumentation really has one
single function-to provide operators with an unambiguous window on
the plant world. When this is not the case, the instrumentation has
failed. For this instrumentation alone-because its sole function is unam-
biguous safety information-its failure alone is enough to justify a plant
trip.
Its like the driver with broken windshield wipersif you cant see,
its hard to proceed safely. Or a train with no clear signals. To proceed
with no window on the worlds critical features violates the basic pre-
cepts of safety. Its like flying blind. This is why I quantify essential mon-
itoring and control equipment as critical. Generally, if there is an
active control loop, its role is already captured as important. No con-
trol, no operations. This interpretation is largely limited to I&C with
safety status functions, and is consistent with Nolan and Heap.
Note that the vast majority of instrumentation doesnt meet these
criteriawell under 1%, and maybe 0.1%. And were not talking about
a little drift here or there in an operating event. Although drift also
has limits and boundaries, critical instrumentation that has drifted out
of range is failed.
The I&C equipment spectrum extends from non-essential to con-
venient controls to generation control loops to safety I&C. Equipment
not directly supporting generation control provides service, conven-
ience, time-savings, or another support function. It is non-critical.
243
Much of this instrumentation provides diagnostics capability, or comes

installed with manufacturer packages. Once operations begin, startup
instrumentation serves little or no further useful purpose. It may be use-
ful later for diagnostics, but this instrumentation usually provides no
further value.
The vast majority of such instrumentation supports NSM strategy.
When an operator identifies that its failed, it can be channel checked,
calibrated, or otherwise restored. Instrumentation used exclusively for
tests should be checked out just prior to test, and then revert to NSM.
As tests are scheduled, cals can be incorporated.
Spare, standby equipment such as spare boiler feedpumps and I&C
status equipment share a similar redundant roletheyre not called into
service unless a primary failure occurs in protected equipment. Some
SRCM methodologies refer to this spared equipment as non-critical.
This explains how SRCM methods can take three virtually identical
parallel components and determine that two are critical and the third is
non-critical. Theyve dedicated the third redundant item to spare
service (on paper, at least). This is particularly confusing when looking
for simple answers on whether or not to PM the spare. The design
symmetry is voided in operations.
Special equipment requires special consideration. That doesnt
mean no PM or NSM as a routine maintenance plan. Spare equip-
ment still requires periodic testing. Because its not in normal service
and cant self-identify failure, the periodic test provides a time-based
OCM task. Failure can only be detected by placing it in service, so func-
tional tests at some interval are both applicable and effective. Post-
maintenance tests after restorative work assure you have a functional
item available.
Instruments often provide condition alert or diagnostics. For con-
dition alerts, reliable, high-quality instruments increase effectiveness.
Ambiguous or faulty instruments can cause an inadvertent plant trip.
Lost production because a protective device failed is painful; this is why
plants avoid armed trips with such passion. If instruments that pro-
vide for status-only information show ambiguous results, thats not crit-
ical. There should be other, back-up items to resolve the ambiguity.
(We expect [and pay] operators to negotiate this sea of ambiguity.) For
244
Lessons
instruments that force immediate operating decisions, there can be no

ambiguity. Theres no time to diagnose whether the problem is in an
instrument or real. Such instruments are critical; they have high
impact and value.
Occasionally, operators arent aware which instruments are criti-
cal; more commonly, advice only instruments are identified as criti-
cal and vice versaand an avoidable plant trip occurs. This makes it
important to understand the criteria by which I&C are identified as crit-
ical to correctly specify the type, reduce ambiguityand avoid plant
trips. The RCM paradigm provides simple, clear guidance on how to do
this. Although the identification of critical instruments ties into the
review of a system and its failures, it can also be performed as separate
activity. A person familiar with a systems protected failure functions
can review instrumentation and identify critical failure mode protec-
tion. They can then tag critical instrument functions, as appropriate.
Catastrophic critical equipment failure modes must not occur
under any circumstance. Large fans and turbine generators are two
prime examples. Fans cant be credibly over-speeded, with fixed-
speed drives. Turbines can. Unless over-speeded, turbines cant credibly
shed large rotating mass parts-but fans can. Practically, avoiding these
disasters is achieved through armed trip devicesoverspeed and vibra-
tion trips for turbines, and imbalance trips for fans.
There are no explicit standards for armed vibration trips on large
rotating equipment but industrial safety and product liability has led to
informal standards that are in widespread use. With experience-partic-
ularly with large air handling fans (ID, FD, scrubber, and primary air
[PA])weve learned that high vibration amplitudes quickly lead to
catastrophic failure. While most manufacturers provide armed trip
device protection options, many companies choose monitoring with
alarm options. This places any ambiguitys interpretation squarely on
the operators shoulders. Boiler flame scanners, ignitor permissives,
main fuel trips, visual flame scanners, and other trip/alarm devices fall
into this category. Such devices provide essential safety and equipment
protection functions and, hence, are critical. O&M literature generally
identifies the critical nature of this instrumentation by emphasizing test
and maintenance requirements.
245
Unlike mechanical equipment, instrumentation has fewer functions.

Generally, theres an active control function or a passive (hidden) mon-
itoring function-sometimes both. Oftentimes status and monitoring are
tied in the same sensing or control loop. Decisions as to what constitutes
critical makes more intuitive sense when its applied to I&C, because
these extend the human senses and present condition information that
we otherwise would not have. Operators are critically dependent on the
extended sensory range these items provide! Without them, we are
potentially ignorant of otherwise unprotected conditions. Personnel
safety is always involved, as proven by general industry experience.
Some examples are shown in Table 6-2.
Hazard Instrumentation Experience

Imbalance VM Catastrophic rotating
machine missile
Combustion/explosion Methane detector Coal fires
Overload Breaker overload relay Electrical overload
mechanical faults
Bearing overheating Embedded thermocouple (TC) Prevent bearing babbit
damage and fire
Burner stability Flame scanners Explosions
Fires Carbon monoxide detectors Smoldering fires
Fire detection Rate-of-rise detectors Slow fire
Fire protection Deluge system Fast fire: transformer fire
Flame cameras Loss of flame/rich combustion Explosion
Table 6-2 Critical Instruments
Primary instrument functions include:
active process control

process measurement
operator shutdown action (e.g., vibration)
246
Lessons
abnormal condition (personal/equipment safety)

steam leak detection
fire
radiation
informational/diagnosticstatus
Critical instrumentation functions include:
active main power block control process

critical safety control processes
electrical shedding/isolation/overload protection
electrical emergency safety ties
automatic important equipment protective shutdown
protects important equipment
protects non-important equipment from safety failure modes
critical safety function
alert to main process control limit out-of-spec
explicit compliance role
status action taken on any of the above
non-critical instrumentation functions
status information
successful transfer
skid diagnosis/convenience
Insurers have a vested interest in stations maintaining these devices.

Insurance companies (or their representative) typically audit operating
conditions independent of maintenance monitoring programs. These
arrangements are generally effective. However, insurers lack the insights
of plant operators and can not investigate problem alarms that are
sources of ambiguous information. These instruments are sources of
risk when operators take them out-of-scan and a plant suffers loss of
an interlock, trip, or requires a jumpered condition. Most insurance
reps wont have the authority (or the knowledge) to open breakers or
control cabinets for inspection. Insurance representatives dont statisti-
cally analyze a stations MWRsan illustrative and telling indicator of
plant operations.
247
Vendors identify equipment failure modes that require special

instruments to detect. They generally provide these instruments or
identify their need. Good instrumentation comes with a quality prod-
uct, and vendor literature is the resource to ferret out this functional
information. Its essential to do an ARCM review on important
equipment as well as critical instruments-anything large, expensive, and
involving safety. Industry and insurance protection standards are a
good second resource.
Critical instrumentation is not hard to identify, and it can quickly
benefit from ARCM-based review to provide a quick R performance
return. This is especially true at units that have never scrutinized instru-
ment programs before. This guidance also provides insurers a focused
process to improve a clients risk profile. In operating companies with
functional OTF critical instrumentation programs this will mean learn-
ing new habits. They will find it difficult to change but they will also
derive the largest benefit from instrumentation focus.
Redundancy
Costs and layers
Where redundancy cost can be managed, redundancy adds value.
Large mechanical and electrical equipment costs are substantial. In the
absence of cost containment emphasis, costs quickly escalate. With
generation simply another competitive product, uneconomic costs are
an added burden.
Many American families have a spare car (some two). Its value is
mobility when the primary vehicle breaks down or goes into the shop.
There is a cost to maintain a spare. Typically its less than a primary vehi-
cle because operation is limited, but there are fixed costs. Consider the
pros and cons of a spare vehicle. Space, time-cost, and other less-obvi-
ous costs are incurred to have one. There are fixed and variable costs,
all of which are endured for the assurance and convenience of the spare.
Cost-savings are achievable if the R of one car is high enough to elimi-
nate the spare. Herein lies the problem of redundancy: How much,
before too much becomes a cost and organizational burden?
A Midwestern utility developed a system-wide blackout recov-
248
Lessons
ery plan-including a PM program-that caused resistance at several

plants and within the substation group. After progress stalled all par-
ties met for a problem discussion session. Applying ARCM to the
blackout recovery plan revealed that it exceeded five redundancy layers.
In some cases there were seven functional layers of transmission redun-
dancy! Workers in transmission and distribution were aware of this
redundancy, and had learned to use it, performing necessary work but
outside the recovery plan objectives. Budgets didnt account for the sheer
volume of needed work, and were constantly trimmed. Real redundan-
cyfunctional equipment in servicewas slipping, as a result.
Just as redundant equipment, components, and systems increases
complexity and demands resources, too much redundant equipment
work can raise costs. If workers learn maintenance complacency
because of redundancy, redundancy has backfired. Eventually system R
suffers. In a non-competitive world, carrying redundancy is easier.
Transmission and distribution R costs will be most impacted as the
market opens to wheeling and competitive power sales, but the value of
redundancy in generating systems will undergo review as well. If four
redundant layers have been required to generate functional redundan-
cy of two, R enhancement from maintenance improvement is in order.
Oh my God!
Rare failures occur with random predictability. When a plant expe-
riences a rare event, the loss is sobering. We hope that equipment dam-
age is the only consequence but occasionally, people are hurt. Major
equipment losses are rare, averaging less than one per year per unit
(based on NERC, my own experience, and other statistics) even at func-
tionally run-to-failure facilities. Individual, rare failure events occur
several times in the 40 year plus life of a unit. Scarcity is their problem:
statistically, theyre like auto speeding; many separate events are
required before an event registers. But plants with events account dis-
proportionately for overall losses. Precursor events are risk factors.
Control them, and you have made a substantial impact on practical risk.
For large equipment event protection, instrumentation extends
the human senses for failure modes (and events) that otherwise cant be
detected. Fossil units are at a significant disadvantage to their nuclear
249
brethren on this point. Not all objectively check critical instrumenta-

tion, or train personnel to follow operating guides based on what they
say. Nuclear plants, on the other hand, are held accountable by license
for their controls and trips, and have many of the trips hard-wired.
Hard-wired trips make it absolutely necessary to maintain instruments
and controls to operate reliably.
RCM recognizes that oh, my God! losses are often critical instru-
ment-related. The task is to identify vendor-supplied instrumentation
that provides major event loss protection, identify the essential mainte-
nance elements, and then make sure theyre maintained.
Major sources of significant event information include regulatory
authorities, industry groups, insurers, technical literature, and industry
conferences. Developing and maintaining a worst case oh, my God!
failure file may be a corporate insurance or safety group function, but it
requires operating and engineering input. Nuclear plants are required
to maintain industry operating event reviews. Most others do so on an
informal level. RCM analysis can take considerable opinion and guess-
work out of the risk estimation for these losses. A complete assessment-
based on industry event frequency and corporate plant risk manage-
ment programs-will identify risk levels, and help focus available
resources where they will do the most good.
R engineering tools that can help quantify and develop these risks
are principally FMECAs. FMECAs can quantify the risk of any partic-
ular catastrophic event. Once you quantify the risk (by extracting exist-
ing designs and using industry information), O&M can manage the
risks. This is where ARCM comes into play.
VM instrumentation is provided on large rotating equipment.
When available, it tells a story. A plants operating environment can use
this story to prevent or mitigate a high cost failure. However, VM instru-
mentation has no value if not used. Typically these systems are expensive,
costing about $200,000. Turbine ones cost up to $100,000. If you dont
use the information, or plan to use it-why have it? Save the bucks and put
them someplace else.
250
Lessons
Instruments In Utility Cultures

Nuclear plants operating requirements are less ambiguous than
fossil. More of their world is prescribed-whether by the nuclear steam
system supplier, NRC, INPO, American Nuclear Insurers (ANI), or
others, almost ad infinitum. Fossil plants until very recently were at the
other extreme-they could pick and choose their operating standards.
To date fossil lacks consistent processes and guidelines that assure
adherence to standards in practice.
Many instruments arent reflected on fossil plants design drawings.
Instruments run the gamut from irrelevant, unreliable to excellent test
and monitoring equipment, from the trivial to most essential in on-line
monitoring and controls. Yet, the operator is largely on his own to learn
and discriminate instrument importance and consistency. This gener-
ates ambiguity.
Many utility cultures distrust instrumentation. They tolerate and
support ambiguous operations. Instrumentation is considered inher-
ently unreliable. Suppose that it is-what does this say about the instru-
ment maintenance program, or the overall maintenance program?
Instruments that cannot be maintained to performance levels required
to leave them armed are frequently disarmed and used for status
checks alone. They are jumpered, ignored, allowed to fail, not calibrat-
ed-and the operators using them cannot be held accountable for their
actions because there are no firm guides or expectations on the instru-
ment standards, functional performance levels, or maintenance pro-
grams. For engineering and I&C folks with no clear guidance on instru-
ments importance-they work on hundreds, if not thousands of instru-
ments weekly-an instrument is solely an instrument. (Nuclear plants
have very clear standards by contrast. Mom, in the form of the NRC,
tells them exactly what to do.)
Do you think ARCM will help this picture? You bet it will!
Critical instrumentation is so essential to reliable, safe operation
that it behooves an operating company to understand the exact role of
all instrumentation, and to treat it accordingly. This means that every-
one with operating responsibility has to understand the processes, risks,
failures, and monitoring methods. This is awareness simply not in many
251
plants today. Outside pressure could force increased instrument aware-

ness but I believe the marketplace is more efficient. Unreliable plants
are uneconomic plants. Low cost producers deliver reliable operations
first, and then predictable, controlled generation on demand. They will
increase their market share.
Case Example
In 1998, a New Zealand transmission/distribution utility ran into
the worst utility nightmare: an inability to supply loads to a major load
centers customers. Most primary feeders to the downtown district in
Auckland were lost. Personnel were unable to restore service in a time-
ly manner. New overhead catenary had to be strung as an emergency
measure, requiring approximately five weeks but ending the crisis. The
event involved the predictable loss of two aging, deteriorated feeders
and the additional loss of two more in rapid succession. A few facts
bear scrutiny:
The utility knew it had a load and cable problem

The utility had attempted to obtain new underground cable
access-without success
Community support for the necessary construction and mainte-
nance was lukewarm
Management punted-failed to go public with the reality of the
untenable situation
The industry had recently deregulated
The utility in question was a vertically integrated distribution
utility in an unregulated generation environment (Institute of
Electrical and Electronic Engineers (IEEE) Spectrum, May
1998)
The utility had made many employees redundantwhat we
Americans call right-sizing
There were other issues. When the final failures developed, the util-
ity acted slowly to save their remaining good cables. The will within the
company to challenge its organizational path was absent. (What
252
Lessons
occurred we call group-think.)

In truth, its tough to roll out these issues before a typical companys
managementthe shoot the messenger syndrome is alive and well.
Those who can quantify and add substance to R issues go unrecognized
in many companies.
Yet, standard pat approaches result in lukewarm results when inno-
vation is necessary. At many plants, only the economic consequences of
a plant being pulled from the rate base gets action. While residual
transmission and distribution entities will be less affected and perhaps
entirely untouched by deregulation, they shouldnt be from a perform-
ance viewpoint. With no nationwide network to wheel over, its hard
for me to see how deregulation will really work. Perhaps a franchise
vote can award a system to a bidding operating entity that shows the
most R Skill. RCM principles apply to transmission and distribution
also.
Criticism aside, there has been little economic pressure on regulat-
ed entities because the alternatives to the public arent yet clear. I fore-
see a time in the future when the public is more aware of the nature and
costs of unreliability and demand better performance from residually
regulated entities and even the government! This may come solely
through the market because R has tremendous value while unreliability
has only cost.
253
Chapter 7
Fast Track
Conquest is easy. Control is not.

-James Kirk, Mirror, Mirror, Star Trek
Traditional maintenance programs-those structured on a combina-

tion of PMs, overhauls, cals, and repairs-have been revolutionized by
the advent of CMMSs. Maintenance planners have loaded activity iden-
tified by vendor O&M manuals and implemented them in the form of
time-based activities. Where done completely-such as in nuclear power
plants-the consequences were surprising.
Its always been difficult to perform all the work, work performance
was often inefficientsometimes very inefficient. Direct work request
input and WO generation amplified trip and coordination time.
Without a method to organize, consolidate, and perform work within
the utility process environment, a complex, stagnant maintenance pic-
ture emerged. Schedulers and planners were added to cope with this
workload-sometimes to little avail.
Companies with billions invested in facilities often formally budget
little to maintenance plan development-even though production losses
255
from load reductions, extended outages, and unit trips translate into
millions in lost sales. This can be tolerated in the regulated environ-
ment with capital investment guaranteed returns. In competitive
industries-refining, chemicals, fibers, process and manufacturing-main-
tenance losses cannot be tolerated.
Culturally, maintenance is an unglamorous backshop where status
quo has been accepted. It hasnt achieved recognition as a strategic
function supporting production. Traditional accounting treats mainte-
nance as a variable cost of production. It is required simply to keep a
facility in operating condition. But if investment in production is nec-
essary to support revenues, then maintenance is a viable production
investment. Industries squeezed by cash flow have tried to cut mainte-
nance but have discovered their competitive position erodes as produc-
tion capacity, services, and processes decline.
CMMS Strategy
To implement any maintenance plan, there must be a strategy devel-
oped on the plants CMMS. Most CMMSs use a PM/CM work model
even if they use an RCM maintenance philosophy. Work originates on
a CMMS as a timed event or work request. Both are internally gener-
ated and correspond to routine (pre-scheduled, timed event) and
response (requested, demand event) WOs. Pre-developed, pre-request-
ed, timed WOs are called PM. Everything else traditionally is CM.
This includes on-demand maintenance we prefer not to develop as rou-
tine, pre-scheduled work.
For instance, work is developed from scratch lists kept by engineers
and planners and put into CMMSs several months, weeks, or even days
prior to the scheduled outage as CM demand work requests. (Such
lists may have been maintained for years in hard copy despite CMMS
availability and capability.) This work looks, acts, and gets identified as
CM when in fact, most outage work is preventive in nature. Equipment
is operated into an outage; work is intentionally deferred into the
scheduled work period. When outage lists are prepared as WOs at the
last moment, potentially plannable/planned work becomes demand
work, and the benefits of planning-standardization, coordination, repet-
256
Fast Track
CNM
Key:
Figures 7-1: Maintenance Terms Map
itive performance, and preparation-are diminished. The CMMS strate-

gy is to create as much known, knowable, schedulable, and organizable
information as possible.
For fast-track RCM results, existing systems must be used to transi-
tion to a routine, planned work environment. To apply its concepts, we
must understand and map RCM terms and interpretations onto the
257
existing software and processes and then perform high value analysis
focused on implementation. (Fig. 7-1) As this occurs, the organization
develops a new maintenance paradigm focused on scheduled mainte-
nance.
When an organization acquires another CMMS they are forced to
adopt a new maintenance model. The selection of new CMMS software
facilitates the transition. This time is an opportunity to introduce
ARCM-based organization, planning, and scheduling methods.
Maintenance Infrastructure
Performing RCM requires a maintenance infrastructure, just as per-
forming planned work does. Someone needs the skills, time, and com-
mitment to do the work. The organization must have the confidence to
use the results. Processes and systems grow slowly, with nurturing care.
Even with focus, commitment, and expert help, learning is required.
The work force grasps most RCM concepts quickly, once they perceive
an organizational commitment to improve skills and manage costs. This
is infrastructure and it takes a dedicated period to develop.
In some instances, infrastructure development requires new capa-
bilities and measures. In others it requires processes-getting PM WO
change control processes, and creating PM owner responsibility.
Building infrastructure-awareness and sensitivity to a maintenance plan-
requires time and nurturing.
Traditional PM Programs
Consider VM-a traditional PM program. Immediate payback
comes from screening VM to limit and control scope. Only a few plant
areas benefit cost-effectively from VM. Although this might at first
seem like a complex task, its by no means that hard- particularly with
several benchmark VM cost/benefit studies. Developing and applying
benchmark cases can quickly establish where VM will be beneficial.
Using this template to quickly screen all equipment for VM can elimi-
nate large amounts of non-productive effort for better PM paybacks
elsewhere.
258
Fast Track
Other traditional PM programs can similarly be screened to get sim-

ilar results. I&C programs that typically perform excessive low value
equipment cals can use an ARCM filter to trim these cals to NSM.
Equipment can be functionally abandoned when no value added results
can be discerned to reduce daily work load. (Note that functionally
abandoned means no maintenance period. This is much different
than NSM.)
Substantial elimination of PM program activity can be achieved to
be consistent with PMO reviews cited elsewhere. ARCM provides the
value measure-every task must add value or get cut.
Managers may presume they have a PM program when in fact, they
dont. Before anything else, a manager needs to evaluate the stations
PM program state, acceptance, and commitment. This assessment is
best done independently, and should include PM program actual per-
formance measurement and capacity for measurement. Questions to
ask include:
Is a PM list maintained?
Whats the percentage completion rate of PMs on the list?
Who gets PM completion rate reports?
Who decides how to defer PMs?
Is there engineering responsibility for PM selection?
Is there an exception report for overdue PMs?
How are PM priorities ranked with regard to other work?
How are outage PMs maintained?
What is the process to add or remove PMs from the list?
Are the PM basic processes defined?
Who is responsible to maintain the stations PM program?
How does the PM program integrate (or fail to) with the CM pro-
gram?
Many plants run random PM programs. That is, they have laun-
dry lists of things to do as time becomes available. PM task selection,
performance, and reporting are hit-or-miss. Unfortunately, its virtually
certain that plants with this PM approach will suffer R and availability
losses. There is simply no credibility to this PM approach in a complex
259
plant. Further, unless operations can provide a supporting rounds pro-

gram, the culture simply doesnt give credence to PM. This program
delivers a better result than nothing at all, but fails to engage the organ-
ization in the very real, exciting task of operating a plant to reduce
unplanned events to virtually zero.
The adjunct to the plants routine PM program is the plants outage
PM program. For failure prevention (and RCM) purposes they are one
and the same. In plants operating with Legacy CMMS PM systems,
its common to find that outage work is maintained separately-even on
scratch paper! Partly, this may be a CMMS design fault; partly, its force
of habit and a failure to implement an outage management PM program
aspect on a software product. Outages are fleeting things-particularly
in companies that experience high unplanned unit outage rates. Its the
old shell game-unit X went down Monday, so defer unit Ys outage two
weeks while X recovers. In the mean time Z goes down unexpected-
ly...and so forth. All these unscheduled scheduled outage changes mean
plant schedulers cant plan for any outage window with any confidence.
The lack of commitment to schedule can reach all the way to top man-
agement, where VPs juggle units outage schedules rather than bite the
bullet and pay for replacement power. Nuclear units are spared because
of the expense of jumping an outage around and because of NRC over-
sight. Getting all system generating plants onto a planned schedule is
expensive-especially if unit R is low-but there is no other first step.
Corporate commitment to firm outage scheduling benefits plant
outage schedulers. Outage PMs can be rescheduled, even with outage
shifts, but, with a reasonable grace periodsay, 25 to 33%, and, assum-
ing application and implementation of CMMS methodsrescheduling
is efficient. Having seen CMMS outage scheduling systems that are
capable of this, my firm belief is that all PMs should be on the same
database in the same computer. If current software doesnt allow this,
find and buy some that does. Its available. Once this commitment is
made, theres no excuse for missed outage PMs.
Scheduling
Once an analysis for scheduled maintenance is complete, the real-
ities of making work happen takes over. Complex plants struggle to
260
Fast Track
schedule necessary work by common agreement. PM programs are at

greatest risk when successful scheduling methods cannot be imple-
mented.
Consider how traditional PM programs were built. Someone-prob-
ably a maintenance manager-searched his facilitys O&M manuals look-
ing for vendor PM recommendations. Extracting these into lists, they
built them into WOs-typically, one per WO-until the entire set of ven-
dor recommendations was incorporated into the plants CMMS. This
approach featured individually applied tasks, without regard to value,
service, risk, or organization. Is it the best approach?
Vendors recommendations represent a good stab at an initial pro-
gram, but of necessity are greatly conservative. Elements typically miss-
ing in this approach include:
group (operating team) assessment

work packaging
evaluation of the applicability and effectiveness for the recom-
mendations
absence of some normalizing routine with respect to the rest
of plant equipment and their PM
This approach creates an unmanageable program-paperwork

kicked out of the CMMS on a schedule that is:
inefficiently planned
hard to coordinate
indeterminate or questionable value
not ranked by some common value scale
With no planned approach to select and schedule PMs, is it any sur-

prise that organizations schedule PMs on a lower priority than correc-
tive maintenance, across the board? From an ARCM perspective, this
is an unequal playing field. If scheduled maintenance tasks can be iden-
tified that are applicable technically, effective individually, and cost-
effectively implementedwhich, by the way, are key attributes of
ARCM-based PMsthen PM performance should rank high on the
261
priority scale. The cost/benefit values of these PMs are much higher
than can be gained by fixing broken equipment. From a return-on-
investment perspective, returns on the most effective PM tasks range
from 50 to 200 times cost. Fixing broken equipment, on the other
hand, has no improvement return. It merely restores status quo. From
an investment perspective, then, which is more important - a 1/1, 5/1,
or 50/1 benefit? Common plant priority systems are structured as
appears below.
Priority Meaning
E Work immediately
1 Work next day
2 Work next week
3 Work when convenient
A value-based table would be inverted. Most PMs can be worked

when convenient, e.g., scheduled. They should also have the highest
priority to complete as scheduled.
In the scheme above, failures get top billing, corrective maintenance
comes next, and PMs occupy the tail end of the program. The irony is
that the highest value work receives the most meager resources! In the
absence of a developed system to level the playing field, priorities get
skewed.
When RCM tasks are developed with the involvement and commit-
ment of the owners, they realize value. Identified and selected this way,
the priority assigned such tasks is higher than by traditional PM priori-
ty schemes. With correct (normalizing) assessments, CDM can be
ranked on the same scale with PMs and all activity value can be ranked
on the same playing field as PMs. Uniform prioritization enables PM
comparison and ranking with emerging CM work.
The traditional, indiscriminate incorporation of vendor manual rec-
ommendations only compounds the priority crisis. It leads to the real-
ization that for some PMs, work had no value or was performed too fre-
quently. The net effect is a discredited PM program, diminished com-
mitment, and faint support. An ARCM-based PM task-selection
methodology selects credible PM tasks up front to facilitate scheduling
262
Fast Track
which permits less work and better commitment to selected work.

Traditional-environment PMs lack a clear tie between PMs and pre-
vented failures. PMs focus on task performance without presenting the
supporting failure prevented value-case. Thats often because ques-
tioning the basis for a task questions the competence of the entire main-
tenance organization. No one feels comfortable questioning the basis
for the work done, so the same work continues to be done, year after
year. This can lead to PMs performed on equipment abandoned in
place, PMs performed that contribute to failures, and minor-value PMs
performed with a work priority equal to major loss preventers. This is
not good business!
ARCM instead ties the best task to one prevented failure. With this
information-and the role of the failed equipment in the system-one can
evaluate the consequences of the failure and assess the value of the PM
activity. Each activity must stand on its own merit. This also rules out
piggybacking non-value work onto an approved PM task.
Tagging failure-prevented to a PM helps prioritize PM and pres-
ents a set of strategy options for a given piece of equipment. For exam-
ple, high-capacity rotary air compressors process sootblowing air for a
coal-fired unit. A key to achieving design life is filtering particulate mat-
ter out of the incoming air stream. Compressors are installed with a fil-
ter bank of 160 individually replaceable canisters in the intake ducts.
DP pressure instrumentation is installed on the blower intake head with
a high DP trip for the blower motor. Filters can be replaced on time
(TBM), provided filter pluggage has a time-aging characteristic. One
could simply schedule a hard-time replacement task on either runs
hours or for continuous loading (calendar time). Quarterly replace-
ment is required in a moderately dusty environment. A filter set costs
$500 (and time) so should be considered. This would check DP and
replace elements based on the filters plugging. This latter strategy pre-
sumes:
operator monitoring
calibrated instrumentation
maintenance work turnaround capacity
263
Item (3) requires that the organization process a OCM type WO

and perform it consistently before differential pressure (DP) plug-
gage becomes a blocked filter, and functional failure. If maintenance
cant deliver CDM with the necessary turnaround, and the compressor
filters plug, a high DP can partially tear the fabric elements. Then, the
torn filters appear the same as a new set-they pass air freely and the DP
is within specs. Unfiltered air rapidly erodes the high-speed, finishing-
stage blading and the compressors require premature overhaul. The
obvious lesson is the KISS principle. i.e., a simple time-based task can
be superior to the complex task-particularly if maintenance capacity to
deliver CBM is in any doubt.
As more programs embrace the philosophy of RCM, but without
the depth, the general push towards OCM will be interpreted as OCM
is always better. This could add unnecessary complexity to programs.
OCM/CDM presumes a sophisticated maintenance delivery system.
Unless this is demonstrably available, an across-the-board push towards
OCM should not be done.
Equipment priority systems are the first attempt to schedule day-to-
day work. All one needs is a relative importance code for each piece
of equipment in the plant CMMS. As WOs are generated, they can be
automatically or manually sorted based on this overall priority assign-
ment. As a first cut, this approach can be helpful, especially to a review-
er not familiar with the plant. It lacks dynamic capacity, however.
An effective program identifies degraded, not failed equipment, to
provide maintenance planning with a heads-up about equipment that
needs work. For outage work, it helps define the next window (Fig.
7-2). OCM identifies a work window in which we can maintain equip-
ment prior to final failure, but doesnt guarantee work input to the
CDM process on the right time schedule. To estimate a schedule win-
dow you need to understand the failure identified and the deterioration
window before final failure. Many seasoned O&M people know many
of these equipment windows for some equipment failures, but no one
knows them all. Theres also an element of chance that CM and CDM
processes account for.
Some failures are best understood as a continuum that ranges from
random to exact life at the extremes (Fig. 2-1). This model helps one
264
Fast Track
Figure 7-2: On-condition Maintenance Timing
realize that skid-mounted equipment has many sub-components, all of

which have multiple monitoring tasks. To manage it well, we must
understand and group components with a strategy. Experienced work-
ers can establish a components failure nature on this continuum. A sin-
gle individual alone rarely does this well. Typically, failures have pat-
terns and distributions. A broad, age-based failure distribution means
the window for ultimate failure will be less certain. This is why we must
estimate the window to prioritize each WO. Not all failures are suitable
for OCM/COM stratagies. Only some exhibit a specific resistance limit,
and failure window. RCM provides identification of the failure mode
and identification of the failure nature-characteristic and uncertainty,
associated with the mean. From these we can estimate the decision risk.
Failure modes and impacts set work priority. In a plant with two
50% boiler feedpumps, the loss of each pump causes 50% load loss.
On the surface, assigning a feedpump problem a high-priority seems
reasonable. However, 90% of boiler feedpump WO failures dont
involve loss of function. Many are for instrumentation. Some identify
operator confusion and misunderstandings. Some are for temporary
problems at startup. Relatively few problems involve loss or direct risk
265
to their function. Therefore, a CMMS approach that assigns WO pri-

ority solely based on equipment will not prioritize adequately. Few
WOs warrant top priority but you cant even identify those without
understanding the equipment failure mode, its specific risks to ultimate
machine failure, and overall unit operations. This is the role operators
fill.
People manage their cars intuitively-even non-maintenance people.
The challenge is to transfer simple intuitive sense to complex power
plants. Some prioritization systems dont support owner/operator
needs. (There is a place for maintenance theory!) The sooner organi-
zations recognize their need to learn maintenance, the better their
maintenance returns will be.
RCM supports maintenance theory by providing on-the-job equip-
ment failure training. One RCM-based equipment review consequence
is a much greater understanding of equipment failure modes and risks
based upon that facilitys actual experience. This enhances intuitive pri-
oritization-and better scheduling.
Scheduling Methods
Expedite
Traditional maintenance environments depend heavily on day-to-
day expedited work. With greater coordination required by the NRC,
nuclear generation has developed more routine scheduling methods.
They also have more detailed PM programs, and generally, a much high-
er degree of PM program implementation.
An operate to failure perspective simplifies scheduling. Priorities
are more obvious with failed equipment. Absence of daily and weekly
generation look-ahead scheduling reflects this acceptance, and the
inherent R of fossil designs. There are fewer imposed engineering
specification failures to contend with (in contrast to nuclear). There
is capacity in the traditional large generating station for operations with
compromised equipment, too. As a choice, OTF offers greater oppor-
tunity to perform real-time maintenance on demand, as needed. So
long as functional failures dont compromise production, and costs are
managed, the option to use OTF is a powerful one. It also requires
266
Fast Track
understanding failure context with maximum effectiveness.

Indiscriminately applied, or applied by default, it leads to higher cost.
Short term (weekly)

Planned work is efficient work. Planned, efficient maintenance
needs to be scheduled, simple, and standardized. CBM can virtually all
be preplanned to allow workers and managers to anticipate and pre-
pare. Weekly, look-ahead priorities should include:
PMs (in cost/benefit ranked order)

CBM (in priority order)
broken equipment
Working ranked PM and CBM ahead of failed equipment means

that overtime may be required for PM at reactive plants with many
equipment breakdowns, or that failed equipment with no functional
impacts may require operations tolerance. When ranked, the work
backlog provides a rolling stack of work that can be reshuffled dynam-
ically and scheduled to accommodate worker availability. This stack
approach can be adjusted to reflect current priorities and needs, but
once jobs are set, there should be an emphasis on avoiding schedule
changes. This reflects both human and cost sensitivity awareness.
Searching for corrective and CBM in the work backlog must be
understood in the value context. Operations plays a central role, but
the organization needs standards to rank equipment failures. Once a
system review is done-identifying key functions and operations impor-
tance-it becomes much easier to rank derived CBM and monitoring
tasks. A few carefully chosen benchmark cases provide an anchor for
comparing all work.
Ranking PM tasks can be reduced to identifying failure modes and rank.
Long term
Except for unit outages, most plants have short windowsand
memories. A 12-week schedule fills the middle ground between the
weekly look-ahead and the outage schedule.
The 12-week schedule was derived from surveillance tests at nuclear
267
plants. (A colleague, Jon Anderson, developed and implemented the

first 12-week schedules at Diablo Canyon with EPRI & PG&E.) It
coordinates routine on-line work performance-time-based equipment
change-outs, lubrications and monitoring inspections-that occur on a
shorter-than-outage schedule. The 12-week schedule places routine
work on a rotation (like an operations shift rotation) that comes around
every 12 weeks. The important concept is the simplicity of the result-
ing schedule. Weeks, months, quarters, semi-annually and yearly activ-
ity support simple coordination.
The 12-week schedule provides an intermediate window for plant
work unavailable in the past. This interim planning time frame suits
many moderate-scope, CBM tasks. It supports work coordination and
scheduling in EGs below the system level and so enables us to:
manage performance risk
improve work scheduling efficiency
facilitate and simplify tag outs
move outage work on-line
improve overall plant R
Now CMMSs have scheduling and alignment capabilities that
extend work alignment capacity substantially. Organizations with these
systems should use them to maximum advantage.
Scheduling equipment groups (EGs)

My first experience with a plant PM program was deciphering a
canned equipment maintenance program provided with a unit at
start-up. Company staff, supplemented by contractors, came to the site
for two years as startup proceeded. They read equipment vendor man-
uals, excerpted vendor PMs, and individually placed these into WOs on
the vendor-recommended performance interval. The identification and
performance of these start-up PMs wasnt comprehensive yet they gen-
erated some 600 PM tasks for a single 500 MW fossil unit. (These
excluded lubrications, which were performed as separate schedules)
(Fig. 7-3).
As one learns quickly with the literal download/implementation of
vendor PMs, many tasks are too frequent, usage assumptions are orders
of magnitude off (on the used side) and equipment in routine service
268
Fast Track
deteriorates much slower than youd ever expect, unless you had meas-
ured it. Vendors specify common lubricants, for instance, whereas pre-
mium ones increase lifetimes. Vendor information and ARCM can
jointly identify, select, and perform the right PMs-but we need both.
Work grouping should be intuitive to optimize performance.
Minimizing overall work-the number of times an area must be entered,
a tag out hung, an area cleaned-are examples of why project management
is effective. Power plants often suffer the frustration of work not coor-
dinated, equipment not cleared when crews are ready to work, or other
crews working when job site space is limited. Work conflict is a very real
and persistent problem. Re-entering areas for equipment rework is
another time-wasting frustration. The fewer surprises, the fewer over-
sights, the fewer items falling in cracks, the less rework there is.
Organizations that measure rework are often surprised at its level as
a percentage of the total. Studies Ive developed and seen have indi-
cated rework approaching 50% of all WO hours, at some plants.
Unless its measured, youll never exactly know the loss. If we accept
that rework is important to manage, then theres substantial opportuni-
ty to reduce this wasted effort.
EGs are very effective at doing this. An EGa logical assembly of
equipment identified and scheduled for work as a unitvary in their
group basis but commonly include:
single tag out and return
standard tag out boundaries
one clearance (including draining or otherwise prepping)
one post-maintenance test and calibration for all work done
multiple tasks performance while in the area
coordinated scheduling of all items in the group to reduce volume
of work items
enabling of establishing standard PM work plans and schedule
intervals for major trains
coordination of work within the group
coordination of LCO type license, safety, and insurance com-
pensatory measures
safety
269
Figures 7-3: Traditional PM Development
EGs can be established in many ways, but must share common

dynamics. First, they must exist for some logical purpose. While log-
ical often means making all the equipment available for work under a
270
Fast Track
common tag out boundary, this does not always have to be the case.
Another use of an EG could be to associate many of the same equip-
ment types into a round for common assessment under one trip.
Another EG could be for common lubrications. Another could be for
VM, or inspecting fire doors. There are as many group possibilities as
possibilities to associate work. Newer CMMSs provide the ability to
establish parent-child relationships, and therefore facility equipment
work grouping.
EGs work best when developed as a joint operating agreement
among all entities in the plant. They provide an agreed-upon standard
for work performancei.e., EGs have greater value if operations can
support them with a standard tag out boundary, and release the equip-
ment for work based on the group. Any plant considering work per-
formance at power (previously done only during scheduled outages),
needs to absolutely minimize the risk of trips. This makes groups a
powerful tool, especially as plants realize the value of doing more work
on-line instead of in traditional outages. (OCM and CNM must be per-
formed on-line.) Once a group is established, the risk of doing work
on-line can be assessed, managed, and controlled with greater focus and
certainty.
Groups fill the gap between systems and the componenti.e., at
the top we have hundreds, or even thousands of components per sys-
tem, based upon functions and identified in the form of system draw-
ings, descriptions, and component lists. Below the system level, there
are often equipment trains, logical subsystems, and the unique organi-
zational structures and work practices that require different groups.
Groups can change dynamically over time as well. For example, an
organization could move from hard time lubrications to OCM, and
back to hard time for a given class of equipment, such as a coal belt (or
eliminate them entirely with sealed bearings). Both types of monitoring
can run concurrently. Groups simplify and standardize this practice.
Why not associate tasks within a given functional work area in a
procedure and eliminate the EGs? This fixed grouping suffers from
its frozen nature and relative difficulty in modifying procedures. Tasks
are typically harder to internally reorganize. EGs retain work individu-
ally, and carry the flexibility to make new associations electronically.
271
They minimize the need to rework hard copy or text files, which makes
them easier to use. Grouping and ARCM complement each other, since
RCM identifies PM tasks to be done by functions, failures, and descrip-
tions-all potential grouping attributes. As PM scope grows, coordina-
tion requirements are likewise greater, and this motivates grouping.
Groups also facilitate work planning. All planned work PMs and
MWRs are identified and associated with a given EG by component
number and coordinated for most efficient performance. Once a group
window is established on the rolling 12-week schedule, time can be
allotted for work based upon experience, risk, and scheduled work
time. Groups can tag many CBM tasks onto one activity to improve the
capability to schedule and work CBM in a controlled way.
Outage
Outage planning and scheduling benefits from the rigor and sim-
plification introduced by ARCM methods. Outages affect unit avail-
ability and constitute the most expensive maintenance budget period.
RCM reviews simplify and standardize outage work to minimize them
and maximize benefits.
Getting outage workscopes, formally reviewing them for applicabil-
ity, effectiveness, and cost/benefit value, greatly benefits outage scope
and budget management. For units that maintain extensive outage
work scopes on a routine basis, the RCM screen is virtually identical to
that performed for existing PM programs. For those that develop
scopes just prior to coming down, there is opportunity to cut scope.
Given that most outages slightly to moderately run over scheduled
durations (based on my experience), theres great opportunity to
achieve substantial returns with unit outage scope reviews.
Outages are sometimes only partially planned. Consistency and
predictability of outage workloads comes with thorough, failure-based
work review for value.
Project Management Techniques

A project approach helps to implement and achieve quick RCM
benefits. RCM projects are soft programs. Support groups down-
272
Fast Track
play commitments and schedules at times. Engineering must provide

analytical parts life-extension and aging studies to support intervals,
EGs need development and coordinated implementation. CMMS
information must be downloaded-then uploaded. Without a project
focus its extremely difficult to identify problems, measure progress,
and achieve success.
RCM project management is harder than many realize. It takes
effort and special skills to create a core team, develop PM standards for
the common equipment, refine processes and methods, and plod
through the 20 to 40 systems that are typically reviewed in a major effort.
Where an RCM project approach has been used, its effective moving
analysis throughout the scope of the project. Where ignored, projects
become stalled, interest wanes, and results are imperiled. Its also possi-
ble to get cold feet. Groups tend to backslide on extended intervals, and
theres a pre-disposition to adjust all outage PM frequencies to the low-
est common denominator-the annual (or 18 month) outage.
Working to schedules
The great challenge for everyone is working to schedules. As nuclear
plants discovered, adding schedulers doesnt guarantee success without
somehow structurally simplifying the work. Operating groups re-sched-
ule easily to support the operating plan, but cause severe implications for
work managers. An eleventh hour operations outage housecleaning
designed to maintain a schedule led to the elimination of nearly 100 PMs
on one job. The limited availability of work windows, combined with the
long duration between plant outages, forced a significant amount of PM
work into grace periods and mandated schedule realignment. This short-
perspective schedule change hurts long-term operating goals.
Once PMs are set up and the schedule is aligned, seemingly super-
ficial changes can have long term consequences. Where PM programs
are mandatory, this can force unscheduled plant outages. Accepting
PM as a priority is a difficult lesson for any organization. Operations,
discovering that they incurred an unplanned plant outage by cut-and-
slash deferral of scheduled PM work, learns a sharp lesson. As
workscopes shift towards a maintenance strategy, schedules become
increasingly important.
273
Consider a plant that has 100% planned, scheduled work and

man-loads to that schedule. There is really no latitude for change.
Any rescheduled work will disrupt the schedule to some degree. For
this reason, an eleventh hour PM cleanup/defer effort is likely to lead to
future chaos. Clearly, the simpler the initial scheduled maintenance
effort the lessor the temptation to defer scheduled PMs.
Overhaul Intervals
Basis for overhauls
Overhauls associate individual, time-based rework tasks around a
common disassembly activity. Because much time can be invested in
disassembling and reassembling a large machine, as much work as pos-
sible is performed with any disassembly task. The consequence of any
single component wearing out before the next scheduled overhaul peri-
od is so great, its considered cost-effective to simply replace all compo-
nents, regardless of condition or cost. This is the basic overall strategy.
Take it apart, replace as much as possible, and button it up.
Many traditional mechanics-as well as instrument technicians and
others-follow an overhaul work philosophy. This has two disadvan-
tages. First, costs are higher than necessary when serviceable parts are
replaced. Second, by examining parts, workers learn to support a
plant- or company-wide age exploration program. Without examining
parts and asking the serviceability question, R engineers never get feed-
back on inservice performance unrelated to failures, and lose the oppor-
tunity to pursue systematic life extension based upon age exploration.
Overhaul intervals are typically based on accepted standards that
represent a composite wearout picture for many components. Once
established, these intervals have gone unchallenged for long periods.
Only the recent drive for cost-competitiveness, and the demonstration
by a few IPPs that overhaul envelopes can be stretched, has changed
this perspective. Unfortunately, executive committees are too often the
ones setting new turbine and boiler overhaul intervals in almost com-
plete absence of field engineering information on equipment capability.
With todays emphasis on CNM and life-extension, plants are
extending intervals, supplementing known aging problems with specif-
274
Fast Track
ic monitoring to assess the development of specific problems. Bearing

wear, for example, can be monitored by a combination of oil monitor-
ing for babbit and particulate products, visual bearing inspections
(through removable caps), VM and trending, and in situ dimensional-
wear measurement. The combination is as effective as a physical bear-
ing examination for diagnostics. We can infer about all there is to know
from these tests. Bearings can also be individually removed and inspect-
ed during light outages, of course.
Blade assessment can likewise be predicted from stage efficiency
and overall performance tests. Borescope examination (on newer
designs with removable ports) is also an option. But the assessment
strives to extend intervals by using aging experience and intelligence to
perform secondary CDM that is sensitive to real life problems and pos-
sibilities.
Optimizing strategy
Many secondary considerations go into the scheduling of a heavy
outage, such as a turbine. These include availability of other units, over-
all load, R, scheduling of replacement power and services, and value of
the deferral in present value terms. With the recognition that nominal
outage intervals may have been conservative, methods to extend outage
intervals (while managing risk) are considered. Methods using condi-
tional probability have been available for years.
From an RCM perspective, a single great potential savings comes
from the systematic examination of risk that comes from incrementally
extending an outage period from a known benchmark. The five-year
turbine standard was considered reasonably safe but companies are
shedding the known safety of this interval to take overhauls out to
seven, nine, and even longer nominal intervals. As they do this, they
seek to manage their risk with increased use of CNM. Extending large
machine outage intervals systematically is an obvious RCM capability.
Planning
Planned work. Efficient preplanning requires that work be antici-
pated-either because it gets performed over and over (like PM inspec-
tions) or because equipment failure modes follow statistical patterns.
275
For example, limit switches get loose and sticky in certain environ-
ments. Even in clean environments, the contact may oxidize. If your
plants primary experience with limit switches is that they come loose
and need to be reset, that job can be anticipated and preplanned. That
standard job should be planned and ready to go on a moments notice.
A job within the skill of the craft requires a job plan-just not
something on paper. (The shop practice and methods guide is the
generic standard plan.) The planner could use a preplanned job, the
standard, or none, at his discretion. But having a standard written job
plan facilitates training, establishes a standard, and enables learning and
revision as methods change. With electronic CMMSs, maintaining stan-
dard job plans is as simple as Windows cut and paste capability and
offers the opportunity to use standard electronic plans.
Preplanned corrective maintenance. Planners and others some-
times conclude that because a failure mode occurs randomly, the work
cant be preplanned. If the failure mode is predictable, and recurring,
the job can be preplanned. Using the 80/20 rule, all high-frequency CM
WOs can be preplanned and filed away (electronically) for immediate
recall. This makes the electrician who gets called in on the backshift for
an unpleasant job a little happier-he doesnt have to wait for the job plan
once he gets in or plan it himself-on the fly! He has something to work
from. Maintenance gets more consistent performance. When these
simple aids were made available workers found them useful.
Problems: costs and rank

PM priority. PM tasks that an organization performs, properly
selected, are its most important work. Based on the analysis of many
failures, its effective to establish a PM ranking system that uses three
criteria.
At the top of the priority list are those PM tasks required by law and
safety. Generally, these lawslike boiler codeswere put into effect in
response to past disasters. Performing these are a good business practice.
Very often, insurance endorsements require them. Insurance commit-
ments maintain basic facility safety and preserve equipment from major
loss. Very often this equipment work fills the same role as critical instru-
mentation. It reduces or eliminates the risk of unacceptable failures.
276
Fast Track
Environmental compliance requirements fit here. At face value,

they are in the public interest. There is simply too much at risk in terms
of perception and penalty to be less than completely in compliance.
While some regulations can be gamed, its not a good risk. PMs
addressing failure modes that would otherwise violate these standards
are the highest priorities to work. If they arent done, its guaranteed
that an emergency results upon discovery-for someone. Examples
include:
boiler code safeties: lift-off tests

stack lights: aeronautical safety
code boiler inspections: boiler safety
overspeed trip devices: turbine safety
CEM equipment: environmental compliance
Plant outages are the second level of PM. Any PMs that directly
prevent plant outage fit this criteria. Boiler tube inspections, condens-
er tube inspections, boiler chemistry monitoring, DCS two-channel
backup configurations checks, boiler camera checks, and main steam
safeties liftoff tests (where these are split, say, 3/35% for 105% total
relief capacity) are examples.
These tasks assure key redundancies, backup equipment, and/or
other capabilities are present. If we lose these devices or equipment,
and anything else happens, we go down. Plant DCS displays are anoth-
er example. By themselves, operator display consoles to the DCS con-
trol provide instrumentation. A plant can (and has) continued to oper-
ate with no active display monitors. (Its never supposed to happen, but
it hasat least once!) If anything else goes wrong, the plant probably
goes down because we cant respond. Redundancies for critical instru-
ments also fit this category. Important equipment goes here. Power
pops that protect code safeties fit here also.
At the third level is PM for purely economical reasons. This
includes PMs that avoid reactive maintenance or large equipment
replacement costs. The traditional work hours and materials B/C PM
slides in here. Theres no production impact at this level, but work tasks
at this level are not all equal.
277
For favorable PMs, there are safety, production or B/C ratios

greater than one. Benefit-to-cost ratios greater than one includes every-
thing from 1.1 (marginal PMs) to 1000/1-real benefits. Total perform-
ance costs range from a few to hundreds of thousands of dollars. We
want to go after high value, big ticket benefits first. We want to achieve
the combination of high B/C and total payback. Unless we rank the
PMs, we could be lubing a pump-well motor for a few bucks while a
sootblowing air compressor filter clogs up and tears out! The former
costs $20 and has a marginal benefit-the pump will still last 90% of
design life without its PM. The latter costs $200-500 but has B/C of
well more than 100, and close to 1000 for larger compressors. So, our
PM system must identify a B/C of 1000/1 and cost of $500 (total value
$500,000) over the B/C of 1.1 and cost of $20 (total value of $2). Most
traditional systems cant make this distinction, and a traditional sched-
uler sees two hours time for either task as equal.
Standards
Every facility has highly repetitive maintenance tasks, either because
of the number of identical components, or its repetitive maintenance
nature. Developing standardized methods to perform work improves
maintenance efficiency. Work standards should include planned
NSM/OTF and CBM jobs, in addition to traditional time-based PMs.
However, convincing craft people that theyll benefit from trouble
shooting procedures and diagnostic guides is a challenge. Once theyve
developed them, they support their use. Nuclear units have procedures
that provide a high degree of work conformity. Even fossil plant check-
lists and guides standardize work uniformity, consistency, and perform-
ance time.
Establishing maintenance standards that address common equip-
ment classes is a preliminary step to build maintenance programs. For
a fossil plant with 20 coal belts, the major componentsgearboxes,
take-ups, belts, drive motor, and soft start gyrolhave nearly the
same needs. Likewise, a nuclear unit with 200 Limitorque motor oper-
ated valves (MOVs), needs an MOV standard as the first step towards
overall work rationalization. A standard will need tuning for details,
such as environmental conditions, equipment class, importance and
278
Fast Track
usage, but the standard is a start. A failure review of each type ulti-
mately identifies specific performance issues. A maintenance plan stan-
dard establishes an efficient, relevant method to examine plant equip-
ment needs at the big picture level.
Standards take many conflicting issues into account. At the equip-
ment level, in a single plant failures and wear are similar. Environment
and usage factors will emphasize some failure modes while suppressing
others. Yet, usage and environment will be similar within a plant.
These influence failure modes.
The maintenance standard summarizes experience and establishes a
plant baseline-and they can be revised with lessons learned, at any time.
Standards provide a foundation for maintenance checkout lists or pro-
cedures, accommodating different requirements or classes of equip-
ment. Taken together, maintenance standards provide the basis for a
plants planned maintenance program.
PM Reviews
PM backlog review
All plants must carry backlogged work. Very often, this is a large
and mostly inactive file. WOs that have been on the list for more than
a year, for example, will never be worked unless a change occurs. A
quick way to establish work value is to review and screen backlogged
WOsboth PM and CMto identify the high value work. This
requires equipment familiarityknowing failure modes, the manufac-
turers guidance, and industry standards. When performed by an expe-
rienced analyst, the plant can eliminate low value WOs and retain valu-
able ones (Fig. 7-4).
Large backlogs may mask high-risk work that fell into a crackfor
example, feedwater heater tube inspections that were skipped, or
missed lubrications, filters, and inspections. A CMMS review can sim-
plify backlogs while pulling out high value work. A R engineer regularly
reviewing the lists can keep backlogs short.
Reviews divide corrective maintenance into CBM and failure main-
tenance. The difference depends on whether an in-service failure
occurs. Failures occur in all planseven in RCM-based plans. CBM
279
(originating mainly from operator monitoring) fails when operations

isnt a full partner. Operators who monitor and manage a work back-
log of CMs and PMs can recognize and extract buried important back-
log activities with immediate R and cost benefits.
The review and purging of a work backlog is, in every way, like the
ongoing review of problems for discrimination into CDM (FF) and
other. Most other is discretionary, cant have a defined scope, or
needs further assessment to define the equipment state. In the absence
of standards, a credible expert must perform the assessment and make
the go/no go call. Ironically, my experience has been that the tough
sells for the majority of the backlogged WOs is getting people to give
up WOs that arent ready for work! Of course, having an organization
tuned to this philosophy (via ARCM) makes it easier to set discrimina-
tion levels for work and work on failed equipment.
Completed RCM-based equipment reviews, based upon standards
for common, high value equipment, provide a ready reference to rank
equipment value. RCM failure modes document ways in which items
fail, at what relative frequency, with what equipment impact, and failure
identifying method(s). This provides a tool for operator, electrician,
and mechanic training and serves as the best way to get everyone on
the same page. Operators can document known failures in standard
terms that O&M groups understand. When operations participates in
maintenance, they improve prioritization of work resources.
Maintenance is more responsive to the plants needs.
PM list
Plants start up with OEM-based PM worklists based on specific
equipment preservation. Many vendor plans assume a continuous serv-
ice operating assessment. Most plant equipment is not in continuous
service and sees far fewer operating cycles than estimated in vendor
manuals. Adjusting vendor recommendations for these operating dif-
ferences generates the first large reduction in vendor-based PMs.
Other enhancements can extend vendor intervals for continuous or
difficult-to-service equipmentless frequent filter changes or lubrica-
tions, higher quality parts, lubricants, and filters, and minor modifica-
tions to improve service. These adjustments are fundamental for a
280
Fast Track
Figure 7-4: Backlog Maintenance
competent R or plant engineer. Vendors are great starting points for

maintenance plansbut only starting points.
Comparing RCM- and OEM-based PM programs illustrates how
RCM screens are effective at reducing PM hours. An immediate
RCM return comes from existing PM program review using experi-
enced maintenance engineers. Thereafter, direct vendor contact and
281
cooperative efforts improve the use and application of vendor products.

Problems should be reviewed with vendor engineers and regular con-
tact provides insights into how vendors rate the performance effective-
ness of various checks and inspections, as well as service intervals.
A surprising benefit is how control over maintenance improves
when the informal maintenance plans are put on a firm basis. This
establishes a baseline for age exploration and further improvements.
CMMS work backlogs

All plants suffer from WO backlogs and all WO systems operate
with in-process work. This is a fundamental rule of differential cal-
culus. The question is how much is right?
Absolute numbers for work backlogs (or work in process) dont
mean a lot. As we have seen, WOs can be associated, grouped, and
processed in many different ways that affect the numbers. Some plants
accomplish major tasks-such as entire turbine disassembly/reassembly
with a single WO, and do so effectively. Others might break this job out
into hundreds of tasks. There are many ways to kluge up a WO sys-
tem. Failing to screen, manage, or group work is one! Planned and PM
maintenance WOs introduce the greatest numbers problems. This is
why group strategies are important.
The default standard is usually one WO per identified problem.
The number of routine demanded WOs gives an idea of the amount
of emerging maintenance work. Here, the question of WOs no longer
seems academic, because one realizes a consequence of more is poten-
tially a lot more WOs. More WOs mean more planning review, more
backlog (numerically), and potentially, from a regulatory enforcement
perspective, more potential of identification as having an inadequate
maintenance program to meet the challenge.
Practically, work-in-process for degrading, but not yet inoperable
equipment is necessary and ideal. With effective prioritization and
lead-time to failure-such as provided by advanced monitoring tech-
nologies-theres more time to manage, control, and correct equipment
degradation. The CBM backlog can rightfully be viewed as a gold mine
of opportunity. The problem has been inadequate methods to screen
and prioritize work. Instead, oversight group standards have drawn
282
Fast Track
arbitrary absolute lines to suggest where the backlog is too much.

With basic RCM theory and streamlined ARCM application, oper-
ating groups acquire a powerful tool to manage backlog. The RCM
screen can quickly reject irrelevant and exploratory WOs and WOs
where someone thinks there may be a problem. Theres no basis
for any WO until a problem is established. The I think, maybe,
please check it out... directives represent someone dumping work
onto someone else that theyre unwilling or incapable of doing.
OCM gives the plant more heads up about work coming down
the pike. More emphasis on CNM will place more work in process-log-
ically. Mathematically, by processing more work (all else being the
same) a new equilibrium backlog level will be established-higher than
before. This will be one outcome of implementing RCM. This is almost
certain to be viewed negatively.
Why? For maintenance, backlog means continuous maintenance
production. (This presumes that the work can be fed into the main-
tenance process systematically.) The traditional maintenance problem
is too much good work to do and too little time and too few resources
available to do it. Its also much easier to create WOs than to close
them. Uncontrolled, unscreened, and unprioritized work requests can
flood a system with requests that divert valuable resources to less-
important work. A WO system with effective screening and prioritiza-
tion, reduces absolute numbers of WOs going into the system, allowing
more good work.
One spin-off benefit of the formal RCM review of failure modes and
equipment impacts is the potential to simplify and standardize plant
work prioritization. Most priority assignments are superficially based
upon the equipment primary function and not on specific failure mode
importance. When frustrated operators fight degrading equipment
they over-emphasize their problems. Pre-assigned priorities by failures
classes, based upon importance, put equipment failure MWRs on com-
mon ground. This comes through a systematic review of the system
equipment using a FMEA-like assessment, which is built into the formal
RCM model.
Backlog review and prioritization is an excellent way to start apply-
ing RCM concepts in the very environments that will most benefit from
283
RCM. Significant amounts of backlogged work may be on hold, await-

ing failure, and nothings more frustrating than to see a WO languish in
a backlog while the item fails in the interim.
Equipment that fails with an outstanding MWR is a maintenance
process effectiveness measure. Equipment, with a wearout period that
fails with no MWR, likewise measures the health of the process and pro-
vides an effective measure of PM program performance!
Outage Work Review

Outages are major plant work. Their impact on overall plant avail-
ability and cost means that outage performance is critical.
Outages need to be planned around specific periods and staffs need
to hold firm to schedules. With costs of replacement power running at
two or more times the generated cost, and the cash flow penalties, there
are great consequences in outages overrunning schedules. Predictable
factors identify whether or not an outage schedule is realistica firm
schedule, project management, and a realistic assessment of the work
scope. Another is how the plant manages emerging work and the level
of contingency (back-up) resources.
Of all these problems facing outages, the biggest is managing the
scope of the work, before and during the project. Whenever you go
into equipment, you discover thingsfeedwater heaters with
tubesheet cracks, a deaerator with severe corrosionthings you dont
expect but are predictable. Many are only interpreted as problems dur-
ing inspection. Much outage work need not be done, but gets caught
up in the zest for doing work. Despite the excitement, schedule driv-
ing, and momentum that accompanies an outage, however, costs are
high. Plants tend to discount technical advice in outages in their zest to
do work. If plants chronically disregard their technical advice, the solu-
tion is simple-let the staff engineers go. They add no value.
RCM-based outage work screens can reduce that non-value work.
Reported results are favorable-up to 40% reductions in outage scope,
based upon work hours. RCM outage work screenswhich pass
work based upon identifying an explicit failure preventedcontrol
pre-outage work selection and work-in-progress scope. They also put
284
Fast Track
staff engineering to work on real problems to derive real value.

Finally, consider the growth of emerging work outage workscope.
The outage manager must manage it, using benchmarks, based on types
of plants, operating service, and other criteria. As outage performance
falls outside the benchmark, the effectiveness of emerging work control
needs to be reviewed. A substantial amount of unpredicted, unplanned
work means either the plants program and previous outage work done
summaries are incomplete, or the plants outage performance is ineffec-
tive. In either case, theres cause for concern and further investigation.
Event analysis
The best time to perform RCM failure mode identification and
analysis in an operating plant is concurrent with any major failure. Staff
is analyzing the failure; everyones energized and focused. Now is the
time to capture the lessons for the future. Ive found that RCM analy-
sis, done concurrently in computer database format, can help focus the
failure event analysis on facts, as well as document other potential hypo-
thetical and real failure modes discovered during investigation. Its
learning that will carry into the future.
One of the frustrations in determining root cause failure is the pre-
sumption of a single failure mode. Root cause failure analysis (RCFA)
doesnt work well for truly complex, synergistic failures-failures with
complex physical, and perhaps even organizational interactions. These
types can be decomposed to be represented as independent failures
on an Ishikawa drawing. Interactions, for which you lack adequate
information to determine the cause that developed into failure, are irrel-
evant in RCM-type process improvement analysis. The focus is learn-
ing all the mechanisms that could have lead to failure (and if you find it,
so much the better!). You need to separately address each independent
failure in your prevention strategy. In this regard, an RCM review of an
event can be more proactive and less fault-finding in nature than tradi-
tional failure analysis. This is exactly why Ishikawa diagrams help to
understand failure patterns-they seek not only the exclusive cause of a
particular event, but demonstrate the interrelationships that can lead to
failure. With this, and with process understanding, frequencies of
occurrence can be measured and action can be adjusted based on risk,
285
probability, and consequence.

Daily, a concurrent significant failure review contributes another
benefit. The plants engineer can evaluate and assess the most trouble-
some failures in the plants daily meeting. In this manner, all impor-
tant equipment in the plant will get at least a cursory SRCM review,
and the plant engineer will become thoroughly versed in RCM thinking.
This sort of review is enhanced by simple user-friendly software to doc-
ument failures. Software, using the plants CMMS database, can quick-
ly create a comprehensive list of the failures that matter.
Parts and Outages

During the years, through many outages, Ive observed many dif-
ferent parts applications in many different contexts. The ones that
stick are the problems. In many cases, we couldnt get the desired
parts on short notice and we took substitutessometimes exact after-
market sales equivalents. Other times, we reworked parts. In still other
cases, different parts were substituted. Results were not consistent!
Outages and other parts-driving events are often the result in crisis pro-
curements and so we took whatever we could getincluding substan-
dard parts.
A high cost item in any generation environment is engineering
redesignfor any reason. My part substitution experience is that many
engineering costs are carried off the plants direct budget, and therefore
unseenexcept as a general corporate overhead. Trust me, this is
expensive! It takes time to assess and reassign part functionality and
specification. Even when the work is relatively straightforward, the time
spent redesigning is considerable, and requires familiarity with the
equipment in question to assess the true equivalence of the part.
Plants cannot always get equivalent parts. Even when they do, what
staff remembers about their service lifetimes, they carry around in their
heads. However, every part has a characteristic failure curve when used
in nominal service. If we statistically group and evaluate them, we
derive mean life and failure standard deviation (assuming a bell curve).
What is nominal service? How legitimate is a normal distribution?
(Short lifetime parts turn out to be very unpredictable, even if they fit
the error in the curve.) In practice, with such uncertainty (and with
286
Fast Track
generation riding on parts decisions) we must justify the cost of carry-

ing excessive stock, simply because they will probably be used. Slow
moving stock can easily be evaluated and parts usage improved with a
joint failure and parts risk control strategy.
Focused strategy development

When a maintenance emergency develops, ARCM is an effective
tool to evaluate and put the threat into a workable context. New regu-
lations, regulatory violations, fines, or an operations upset pose chal-
lenges to continued plant operations. Likewise, near-misses due to
equipment failures can be evaluated by ARCM. The potential for recur-
rence, need for new maintenance strategies or perhaps even redesign
can be established.
RCM supports FMECA and often resolves fuzzy issues that other-
wise could go beyond bounds. PM program additions can also often be
evaluated in context with ARCM statistics and analysis.
Equipment Groups
EGs provide a method to make PM task performance efficient. The
concept derives from work blocking by Nolan & Heap, and is espe-
cially critical when trip time may be a significant portion of the total job
time, or a significant effort must be made to make equipment available.
For a plant, it could mean the time required to tag out and return
equipment to service. When equipment is available, all potential
required work ideally will be performed. In nuclear applications, the
NRC maintenance rule requires plants to monitor the unavailability of
risk-significant systems and minimize unnecessary work. This impetus
has always had value (Fig. 7-5).
The practical utility of developing and implementing EGs with a
CMMS is that when a group is scheduled down, you do all planned
work in the group. This requires a process of aligning all PMs in the
group in such a way that they occur in the scheduled downtime slots.
The other CMMS benefit is that when a group must come down, all the
backlogged work for that group is immediately retrievable (by sort) for
quick assessment. This greatly simplifies the job of the workscope
287
Figure 7-5: Plant Instrument Air/Service Air Equipment Group
planner and outage scheduler. EGs make it far less likely that an item
will fall into a crack and be lost. The greatest benefit is work perform-
ance consistency.
An operators job includes large amounts of travel between relative-
ly short periods of equipment monitoring. As much as 60% of an oper-
ators time may involve travel. Likewise, when maintenance crews must
spend a large fraction of their work time getting to and returning from the
job site, they need to coordinate trips for maximum time utilization. (In
practice, when work isnt effectively blocked, feedback from crews is usu-
ally quick and critical. The tragedy is that this doesnt always become
incorporated into work as improved job planning.) PM tasks, like proj-
ect activity blocks, are most effective when thought out and coordinated.
The concept is like taking your car to the garage: ideally, you identify all
the work and get everything taken care of with one trip. Obviously,
power plants are much more complex than cars but the principles the
same (Fig. 7-6).
288
Fast Track
Figure 7-6 Modified Work Control Process Flow
EGs therefore, allow:
efficient use of resources

coordination of work
control of work risk (especially for on-line work)
combination of work when equipment is available
What organizations can fail to realize is that in production, most

work is repetitive. i.e., after any plant has been in service several years,
289
very few work tasks are new. The bread and butter work of the organ-
ization is repetitive. Because of this, its highly effective to spend a great
deal of effort to plan and coordinate the repetitive tasks. This means
EGs and blocking of some sort.
The more planned maintenance an organization works, the more
important EGs become. In a purely reactive maintenance environment,
things breakas they break, they come down. All thats minimally nec-
essary is to quickly get them back into service. As a transition to
planned maintenance occurs, more work originates as:
TBM WOs and tasks
OCM derived work tasks
CNM/OTF operator-identified degradation work tasks
With a fully developed and effective PM program, fewer things

break. With close operations-maintenance coordination (to get access
to equipment to perform work), and by blocking equipment trains and
components into EGs, plants greatly simplify the performance and
scheduling of work. They conserve and simplify operation tag outs and
standardize work.
If task blocking is so effective, why do so few organizations use it?
I believe this comes back to the unanalyzed state of maintenance at too
many plants. Few managers have backgrounds necessary to develop
complete maintenance strategies. They lack the time to understand the
detailed operating and maintenance requirements that suggest the best
maintenance (and operating) strategies. Few have the maintenance
support engineers to assist them. They dont always grasp the essen-
tially repetitive nature of the work, and the need to standardize and
streamline performance. Maintenance is widely regarded (with great
justification) as a highly crafted skill, but that doesnt mean that every
job has to be developed new, from scratch. The essential lack of main-
tenance processes perspective leads to this acceptance. In addition,
some plants work under an outage mentality. They save all sorts of
major work for outage periods when resources are more available, fewer
questions are asked, and then just work like crazy for the entire outage.
Where applied comprehensively, EGs can strip away up to 40% of the
hours that are added onto outages and remove confusion about what
outage maintenance work really is.
290
Fast Track
Figure 7-7: Residual Heat Removal-Equipment Group Registar
New CMMSs promise to help classify and coordinate maintenance

performance. With respect to work grouping and scheduling, new
CMMS hierarchies have the functional tools to attain this promise. As
simple as the concept of blocking is, the actual development and imple-
mentation of processes that block and work based upon EGs is rather
291
Fossil Nuclear-BWR
Baghouses Minor work

Reactor core isolation cooling
High pressure coolant injection
Coal mills Residual heat removal
Coal belting Radwaste
Dust Suppression Off gas and augmented off gas
Feedwater heaters/boiler feed Service water
and condensate pumps
Circulating water tower cells Instrument air
Condenser waterboxes Turbine light maintenance [blocked by
As Low As Reasonably Achievable (ALARA)]
Fire control
Table 7-1: Areas Not Worked Online
sophisticated. Groups must be developed. PM tasks for equipment

must be identified and blocked into convenient work activities. These
activities must then be tagged with and scheduled onto EGs. Without
a computerized CMMS and the availability of an implemented, func-
tioning PM program, there is no structure to support working EGs.
On the other hand, where these elements are presentas in com-
mercial aviation and large power plantstheres substantial opportuni-
ty to use computer-scheduling tools to simplify work. This opportuni-
ty has been overlooked in some older plants. Several large integrated
coal-fired sites that Ive worked with have made easy work gains by EG
development. (Fig. 7-7)
Available plant design redundancy is used ineffectively because
most plants have never crossed the threshold of performing all the con-
trollable work on-line thats available. Some examples of things not
worked on-line (with the potential to be) are higlighted in Table 7-1.
One common reason things arent worked on-line is that isolation
valves wont allow it by leaking through. Another is that plants usual-
ly wont isolate tie buses to facilitate work. This means that one com-
292
Fast Track
Figure 7-8: Can It Be Worked On-Line? Can 40% coal-burning efficiency hold the
line? What level of reliability must back this up? These plants require staffs of 100.
Forty maintenance workers stretch to get all the work done. Yet plants in Australia
use two-shift operations with idle shutdown periods on automatic startup sequencing
unheard of in the US. Although unit outages will always be required, systems and
equipment can support more online work performance -- provided maintenance is
carefully coordinated with operations. In fact, as more maintenance work is initiated
by on-condition maintenance/condition monitoring, online work fraction increases.
This increases revenue.
mitment to performing on-line work involves raising the perceived

value of maintenance so that valves, dampers, operators, and switchgear
are available and clean tag out isolations are possible.
To test for on-line work availability, ask, if it fails, do you always
take the unit down? If the answer is, take the unit down, on-line
work may truly be too risky. But in those cases in which plants work
failures routinely on-line, something may be missing. Generally, my
experience is that planned work is much less risky, and more thought
out than the emergencies. They really entail little risk when planned
and supported by sound tag out and maintenance processes.
Obviously, every plant environment and feasibility differs, but with
more CNM, more on-line inspection is a necessity (Fig. 7-8).
293
EG development steps
EGs can be developed in two fundamental ways. One is based on
designP&IDs, trains, and equipment layout. The second is based on
existing operation tag out boundaries. Each has its advantages. The
designers intended EGs are often described in A-E system operating
procedures, plant operating modes, and system descriptions. These
materials can occasionally provide surpriseslike the fact that design-
ers anticipated isolating equipment for maintenance on-line that plant
managers never envisioned possible.
In essence, every designer had EGs in mind as they laid out their
plant. This is often the reason, for example, for the check and isolation
valves in standby lines. Its obvious that any redundant standby train
must incorporate the means to both bring it on-line as well as isolate it
for work. Designers have learned standard layouts and methods for
incorporating redundancy into designs and have used them for years.
The problems occur once the design leaves the designers control.
Although many design intentions get faithfully reproduced in the real
plant, as-built systems occasionally dont function as expected, either
due to design oversight or quality problems. If contractors use substan-
dard materials or undersized components to manage costs, equipment
doesnt perform in service as expected. Thats not the designers fault.
The greater problem occurs, however, when plant operators and
maintenance staff do not fully own a design at the time they become
responsible for it. Designers usually provide training, but theres no
assurance operators will comply with the designers intentions. The
design is too often compromised as its incorporated into operating
rounds, PMs, monitoring strategies, and allocations for maintenance.
Existing operating cultures have powerful influence on planned opera-
tion regardless of equipment or system capabilities. Given these facts,
and the haphazard methods by which we bring new systems on-line, its
inevitable that some design compromises occur.
Because the primary reason for forming EGs is to perform mainte-
nance, EGs dont need to be based upon a physical boundary. An EG
should provide a unique identifier in the plant CMMS for sorting.
294
Fast Track
Other group possibilities include:
operator rounds
VM round
calibration round
loop calibration
thermography round
lubrication round
fire door inspections
fire alarm inspections
In fact, any logical grouping of equipment based on inspection or

monitoring criteria, locationor bothcan be effective criteria for an
EG. While the first criteria for developing roundsperforming on-line
workis more powerful from a work-leveling perspective, the concept
of EGs has great potential to make many more work associations possi-
ble. In fact, an operator round is essentially an on-line EG structured
around a fixed amount of time and a route. The on-line work group is
hardly different except that the work belongs to maintenance and has
an accompanying WO.
An obvious concern among many managers today is workload lev-
eling. To control the tremendous augmentation and overtime worked
for regular outages, more plants today are performing on-line work. Yet
the predisposition to lump all heavy, major work into outages continues.
Coal handling is an example. How many coal plants take coal systems
down to avoid performing heavy coal handling work in outages? Few,
I imagine.
Certainly, doing on-line work requires coordination. If you fool
around the units could go down. But controlling risk by equipment
grouping was an intended use in the first place.
An EG strategy is essential in nuclear plants, where more armed
trips and license-LCOs come into play. Nuclear units have larger
amounts of on-line surveillance testing, failure-finding instrumentation,
alarm checks, and standby equipment tests. Nuclear plantswith their
large staffs for work coordinationoffer the opportunity to align PM
work into groups once and then keep it there indefinitely.
295
EG types
One method to develop EGs is using plant A-E P&IDs to identify
multiple trains, logical sub-groupings, and other sub-units that so natu-
rally form work organization skeletons. These EGs are a physical type,
based upon the tangible layout for installed equipment. Such groupings
are most often based upon designer intenti.e., the check valves in a
feedwater pump loop were installed to facilitate on-line work. A second
common grouping is the maintenance round. Thermography, VM, even
fire door inspections, logically fit into this kind of special EGa do it
all at once group. The key to this group is the equipment availability
on-line.
The first EG type includes a physical, energy, or pressure bound-
aryareas where steam, water, voltage, air, hydraulic, or special gases
(CO2, H2, He) are common. Usually, these can be isolated on-line by
train, using block and/or root valves, breakers, and other isolation
devices. The boundary points for EGs on a P&ID are also typically tag
out points. The second group type is a convenience group. These facil-
itate the use of a checklist to perform a PM round on a routine basis.
A combination of the boundary and convenience groups results,
when a system has separate trains that can be conveniently grouped for
simple work, even though trains can be crossed. Reactor water cleanup
at a BWR involves two pump trains on the front end and two filter dem-
ineralizer trains at the back end, separated by common regenerative and
non-regenerative heat exchangers. These can be grouped for simplicity
and convenience, even though the A/B pumps and A/B demineralizers
could be cross-connected. We can thus construct an arbitrary group,
but a simplifying and convenient one, within the pressure boundary.
Similarly, the BWR augmented off gas system has two similar yet inde-
pendent front- and rear-end trains, which grouping simplifies.
Implicit grouping results when routine work within a system is
aligned on a 12 week schedule. Once equipment PMs are placed on
the schedule, and worked as scheduled, the work groupings naturally
stay together as time goes on because theyre aligned to the same work
window. This grouping, developed as a natural outcome of alignment,
minimizes work periods, system outage time, and the other negative
work impacts. On a system with non-intrusive PMs (such as lubrica-
296
Fast Track
Figure 7-9: Modified Maintenance Process
tion, calibration, and electrical switchgear checks), alignment helps

associate, and keep associated, all work naturally occurring in a PM
group. The alignment group and boundary group may be the same, but
this is not required. External valve operator PMs, for example, line up
with other non-intrusive work that may be scheduled separately from
the boundary group.
Grouping also occurs at the skid and equipment level. Its natural
to group electrical, mechanical, and operator stroke tests of motor
operated valves together at nuclear power plants. Here groups are
297
based on skills required. Valves and their operatorsmotor operated,

air operated with solenoid pilots, and hydraulically operatedform
equipment level groups. Groups can also be developed around similar
technology, such as VM, thermography, or air-operated valve testing.
Groups can then be associated with fixed work-order scopes for per-
formance in convenient work hour blocks.
The primary goal in developing groups is improved work perform-
ance. Consequently, as WO lists for equipment are displayed, they may
suggest other unique groupings that offer convenience, simplicity, and
time savings (Fig. 7-9).
Operations
Operations copes day in and out with residual, random equipment
problems. They run complex facilities, and a general trend is for com-
plex equipment to fail randomly. Outstanding operations reduce ran-
domness. Mediocre operations introduce it. What factors separate out-
standing from mediocre to control randomness of operations?
Factors that have been identified by risk analysis and good operat-
ing practices for years include poke yoke methods and devices. Some
are:
simplicity
procedures
standards
marking
lighting
cleanliness
training
These are obvious. But consider some not-so-obvious methods.

Using the knowledge of the bodys natural clock to schedule shift rota-
tions makes for more alert operations. This improves response to
events. (We used to have a joke-All bad things happen on graveyards.
In fact, they frequently did as often as not because the response of an
un-alert operator to an event is just about random.)
298
Fast Track
RCM applications can be extended to the operator interface. An

alarm may correctly notify an equipment problem or exceeded limit,
but if the operator doesnt recognize the alarm, it might as well have
failed. Any factors that increase the eligibility and predictability of
operator response to mitigate random events have great value.
Modification reviews
Ranking unit modification capital allocation requests is an annual
budget exercise that is becoming more complex. Safety and environ-
mental improvements often are handled as separate line items, leaving a
limited amount of money to be divvied up among many needed
improvements. Ideally, improvements have high paybackscertainly
high enough to pay their way. Instrumentation upgradessuch as fos-
sils transition to distributed controlsmay be needed to continue eco-
nomic operations. Every modification should have a projected benefit.
RCM thinking has identified modifications and upgrades that made
no sense, and, once identified and cast in a R perspective, could be
deferred or eliminated entirely. Several of these discoveries and ARCM
has paid its way.
A single unit PRB coal unit was originally sited for two units, with
coal handling service sized and built to allow a two-unit operation. All
major belts except the transfer and tripper belt, along with the crusher,
and dust collection system had completely redundant spares. When
one of the long, inclined-yard belts became an aging concern, the coal
handling people requested to replace the belt as a part of the annual
capital budget request. The belt replacement (with a cost more than
$100,000) made the cut, even though there was no generation improve-
ment to be derived from the upgrade.
This improvement could easily be deferred by simply spacing out
the aged belt-using it as a backup for the otherwise redundant paired
belt. This was possible at virtually no production risk. Of course, long
term, the operating strategy in this case requires that everyone under-
stand the modified approach and accept the marginal increase in risk.
In another case, a $1 million dollar capital rail loop construction
project was replaced by $50,000 of capital improvements, and an over-
all greater operating monitoring program. This capacity to focus capital
improvements is an obvious ARCM benefit.
299
Chapter 8
Maintenance Software
Theres a right way, a wrong way, and a Navy way. We do it the Navy way.
-Master Chief, USN
Goals
Why do we need software to help us perform maintenance? People
and organizations have performed maintenance for years without it. On
the other hand, the primary purpose of software is to improve mainte-
nance productivity and work performance. With so many prescribed
statutory rules, its hard to imagine organizations doing work without
software and many have now used it for 20 years. Some do without,
however, often very effectively.
Its instructive to remember the promise of maintenance informa-
tion system (MIS) maintenance software as we review the changes nec-
essary to successfully implement RCM. The softwares original objec-
tive was to vastly improve the use of maintenance information. Did this
promised benefit occur? In many situations, it didnt. Access to com-
puters was functionally limited to the front offices. Software never
301
Figure 8-1: CMMS Hierarchy of Equipment

accurately modeled the work-control process and in some cases, the
process did not lend itself to modeling, since it was indeterminate.
What was involved with goal setting, and where did it fail to miss
the mark?
Hierarchy
Software was developed to facilitate the whole of maintenance plan-
ningincluding the entire equipment hierarchy.
Systems are the highest unit levelparts that can be removed from
service and replaced singly are at the lowest. In between we have equip-
ment trains, logical equipment groupings, major equipment assemblies,
and components, all of which can be tagged, isolated, and worked at
one time. (I think of equipment as component assemblies providing
302
major functions within systems.) Componentsitems that are built

into equipment and replaceable as units out of a boxare the dis-
crete building blocks. Logical groupings are available between the
component and system levels to better coordinate, schedule, and facili-
tate work (Fig. 8-1).
EGs simplify tag out work. Groups once established advance
planned work performance by tag-out processes to a consistent, pre-
dictably scheduled effort. EGs with standardized tag-out routines have
less potential for error and facilitate the coordinated scheduling of large
work blocks in single packages. Once a groups work is aligned, blocks
of work stay together until they are realigned.
At the lowest work coordination level equipment can remain in
service while its monitored and maintained. VM routines can be placed
in miscellaneous groups and be scheduled to work anytime. They
have no impact on operations. Shops may schedule this work for their
own convenience.
At the other extreme is equipment that requires unit outage to per-
form work. They go into outage groups. Boundary groups have these
two general default categories-on-line or outage work. The balance of
the EGs is made up of trains and associations that facilitate tag out, on-
line work, and standardization. Physical boundary groups depend on
design, but they form equipment associations that can be worked
together. A plant coal-mill or baghouse fly ash train could constitute a
group.
Convenience groups facilitate working activities such as fire door
inspections. Rather than bog down the CMMS with 50 different WOs,
a checklist can support one. The PM now involves one MWR and the
checklist.
Grouping success depends on organizing for simplicity and impact.
Maintenance success depends on grouping.
Coding levels
Equipment can be coded to any detail level. The key requirement
is to be able to uniquely identify equipment with no ambiguity for main-
tenance. Inconsistently coded equipment systems pose problems. The
primary reason to code and identify equipment in a CMMS is to facili-
303
tate maintenance and so the coding must support this.

Informal coding doesnt work well at large plants with many main-
tenance workers. Coding consistency leads to work coordination, fewer
mix-ups, and a usable maintenance history. Accurate logical coding
supports not only monitoring and maintenance, but clear tag out
boundaries.
Standardize
Equipment coding structure reflects how you do maintenance.
Coding must fit the CMMS and tie into any equipment grouping
scheme the plant intends to use. Equipment grouping is a powerful tool
that enables the free association of equipment for the primary purpose
of accomplishing maintenance. A CMMS should facilitate the develop-
ment and use of arbitrary EGs.
Applications
The value of EGs increases where equipment coding systems
are detailed. (At fossil plants, a boiler feedpump could be the smallest
coded unit in a group. A nuclear plant might have 100 coded identifiers
for the same equipment.) This concept supports natural grouping. The
trick is to uniquely locate the skid-identified components. For a plant
coded to the component level, group associations coordinate work.
Newer CMMS systems that support hierarchies automatically provide
grouping logic.
CMMS Computer Software

The maintenance process
The maintenance process used at a plant determines how a CMMS
needs to be structured. There are many different flow paths for main-
tenance performance that lead to similar outcomes. But if measurement
and trending are the goals, the path needs to be consistent with data and
paper flows.
For example, a unit with a CMMS but an operator who cant make
computer entries means hell have to delegate MWO entry to a clerk.
Quality and accuracy of problem descriptions will drop. Its important
304
to use plant staff for credibility of work as well as to validate MWO

requests. Any CMMS with inaccurate MWO entries that arent quick-
ly purged will suffer credibility problems.
Unique CMMS
Companies used to maintain large information services groups.
These groups are vanishing, victims of outsourcing and cost/benefit
assessment. Their legacy is custom software productsthe most impor-
tant aspect of the CMMS. These large legacy mainframe codes must be
mastered to extract data, generate reports, and otherwise interact with
and manage the day to day work of the organization. Few engineers and
managers have learned these systems. CMMS systems are largely the
software tool of planners, schedulers, and maintenance staff. However,
all the functions of maintenance at most stationsfor better or worse
are computerized, with most permanent records maintained this way.
CMMS systems offer operators great power to access and interrogate
information. Its kept maintenance managers either in control (or in the
dark, as the case may be) the past 20 years. Those who could access and
generate their own reports had an inherent advantage.
The advent of second- or even third-generation CMMS products
offers greater flexibility-though at the expense of tailored applications.
These products, with their Graphical User Interface (GUI), Windows-
based environments are truly exciting. From a standardization perspec-
tive, you will likely see the kinds of evolutionary paths that accompanied
the adoption of Word and WordPerfect as document software standards
in the PC worldi.e., much greater exchange of documents and other
information developed in the same Windows-based applications.
As this transition occurs, maintenance organizations must learn to
adapt systems they grew up with. Just as industry-wide application of
word processors has developed some incredibly powerful and common
routines, CMMS capabilities depend on having and learning the soft-
ware. Since a CMMS installation for a modest-sized utility runs well
into millions of dollars, some wont be able to afford the transition until
market pressures force it. These organizations will continue to struggle
with their specific software application. In addition, newer systems are
capable of more efficient import and export routines, which will facili-
305
tate strategic steps such as standardized work plans and procedures.

Improved sorting and information-management capabilities, as well as
user friendliness, should put CMMS applications much closer to the
worker and prepare for a transition to a paperless work environment.
This is an attractive advance. Real-time maintenance information is a
reality for some plants.
From an ARCM perspective, one of the greatest benefits of new
CMMS technology and software will be the access and availability of
failure and performance data at the system, group, or component level.
This enables system level management of common equipment. In the
nuclear industry, system engineers perform this role. When system
level data (particularly cost) can be reviewed, and theres ownership for
system management, I anticipate real performance advances. For the
first time, operating groups will be able to see how their maintenance
approaches roll up to the system performance and cost levels. With this
information, in integrated formats, it will be easier for professional
maintenance personnel to manage cost and production.
System level measurement

In ARCM, the functional level is most natural for performance eval-
uation. Integrated system performance includes functional failures,
costs, and unit availability/R impact. Other measures identify the main-
tenance strategy, system health, and results. For example, whats the rel-
ative breakdown between time-based and CBM? Or the proportion of
hours and work orders attributable to emergency work? Or percent-
age of all work that is planned? What is the relationship between esti-
mated and actual worked hours across the entire system? These are
some of the questions and measures that weve used in the past to
uncover major areas of opportunity. The key to achieving these results
was the ability to access performance numbers in the CMMS at the sys-
tem level.
In the absence of this capability, some or all of this information
could require manual manipulation and presentation. With the wide
availability of systems such as MS Access, a database can often be
attached and data manipulated to extract information, even though the
primary database is on an older mainframe computer.
306
Age exploration
To perform age exploration, there must be a convenient way to cap-
ture the performance date of in-service parts. For instrumentation and
controls, age exploration is as simple as looking at calibration data after
some period of service, and estimating the allowable drift before the
instrument goes out of range. Obviously, many assumptions and a lot
of skill are needed-not the least of which is familiarity with the equip-
ment. The principle is the same with mechanical parts, but there may
be multi-dimensional requirements to consider. Opinion and judgment
are still a key factor.
The tendency with age exploration is to underestimate mean life.
An interpreter sees the first few instruments drift or datapoints out-of-
range and they estimate the mean life at this age. In this manner,
extremely short mean life estimates result.
RCM Software Development

Process standardization
Software speeds and standardizes RCM analysis. If you need a
basis for every plant component (the case at nuclear facilities), then soft-
ware is indispensable. Software packages are available that cover the
full spectrum of capabilities, including at least one that faithfully repro-
duces TRCM, including the classic LTA approach. Most perform some
sort of SRCM/PMO simplification necessary for simplicity and speed.
Each supports different raw data sources, ranging from user input to
electronic download data at the two extremes.
Having developed bases by hand, in word processors and on
spreadsheets, my opinion is that software documentation packages add
value (Figs. 8-2a and b).
Identifying common PM tasks for standard components, with easi-
ly reproducible tasks and adjustable inputs, is a desirable software fea-
ture. If your organization is also considering an RCM-type basis for
your PM program, youll need to task group features. Before you con-
sider software, ask whether you need a documented basis at all. Some
companies dont document their PM work tasks, much less their task
basis. Some are very effective in this way, but its because their people
307
Figure 8-2a: Sootblowing Air Compressors Task Summary
308
Figure 8-2b: Compressors Loaded Round
309
are highly skilled and motivated. Even when there are lapses, the sim-
plicity is unmistakable.
A basis-to-PM relationship reflects the analogy of an architect to pro-
vide the basis for the builders construction. Most builders would not con-
sider building without a plan, but this wasnt always the case. Is it possi-
ble to build a house without a plan? It sure ismy granddad built at least
two that way. Is it effective? Perhaps marginally. Is it competitive? No.
Existing buildings constructed on an as you go basis cost more.
Maintenance programs built as you go can likewise be expensive
anywhere from 20% to 40% more, and they dont develop as much
inherent equipment R based on post-implementation PMO/RCM
reviews. Existing facilities must have some plan. Making adjustments
to the plan as you go is often the most cost-effective approach. On the
other hand for a new facility, LCM costs can be a major opportunity to
achieve long term production benefits and lower cost. Is there value to
increase production from the same capital asset? Ask any banker.
Most current RCM software are user friendly and save efforts.
Value doesnt come from documentation of every plant components
PM case, however, although such software is available for nuclear
plants. Rather, value comes from building effective maintenance strate-
gies. Building and maintaining standards that reproduce failure mech-
anisms and identify effective preventive tasks is most cost effective in
the long run. The organization that can retain strategies over time to
support a living maintenance program will maintain lower production
costs.
Ultimately, theres a tie between basis software and a plants CMMS.
Basis software provides the justification for the tasks performed. Since
PM tasks (as a part of a work performance package) address individual
failures, PMs get scatter-gunned, as well. Software that facilitates
grouping is helpful. Groups statistically increase the odds of PM on-
time performance. More advanced RCM packages provide task group-
ing and sort capabilities. In a seamless way, the things you do and how
you do them can be imported into CMMSs to provide the PM tasks and
frequencies. No single software has this functionality at this time.
Ideally RCM software should be simple and intuitive to use. Some
general goals include:
310
facilitate ground-up basis construction

facilitate existing PM program reviews and basis creation
facilitate PM task upload to the mainframe or CMMS upon com-
pletion
provide real-time inquiry support to users
support growth of new applications and technology
support gradual trimming of legacy PM systems with RCM prin-
ciples over time
Basis
The basis for any PM task is the reason why the PM task makes
sense to do (or not do, for the NSM case). In nuclear applications, there
is great emphasis on developing and maintaining a basis. In non-
nuclear applications, the justification for a basis is underlying task value.
Implicit basis requirements can be lost during time. Plants have aban-
doned very good PM tasks because they forgot why they did them. The
tasks very success may have been its downfalltasks effective at pre-
venting failures werent recognized as adding value, and were dropped
and a painful learning process begins anew.
This occurred once at a two-unit coal unit in a dramatic way. The
plant manager eliminated virtually the entire sootblower PM program,
based on low failure rate. For the better part of a year his costs dropped
with no apparent consequences. Then sidewall boiler tube blower tube
cutting developed in a major way. Within three months theyd experi-
enced two cut boiler tube leak outages. In a year theyd had six. This
was eventually traced back to corrective maintenance by untrained
mechanics. After two years it was clear that the net gain had actually
become a significant lossmore than five times the promised operating
savings, at the plants wholesale production rate. One outage saved
could have paid for a whole year of sootblower PMs.
When you dont know why youre performing an activity, theres
temptation to change. For PMs, you must know why youre doing them
and what the underlying value is. From the perspective of a living pro-
gram, thats when a task basis is most valuable. Changes can be evalu-
ated with more focus, clarity, and with less regulatory or safety risk.
Developing a basis for an activity is like keeping a log. Every time you
311
change or note something, you log it, and file the result where its
retrievable for later use.
For large plants and companies, this means reliance on computers
and their networks. The value of a basis is underscored in this age of
cost-trimming, when many young Masters of Business Administration
(MBAs) with virtually no practical experience are tempted to make
career-serving cost reductions. These only reveal their true impact later.
A documented basis gives staff more armor to defeat such moves.
Analysis
Software enables us to perform detailed failure and cost analysis
quickly. Several software products have this capability. Where genera-
tion could be lost, and expensive corrective action is a necessity or
option, cost-calculating software can establish B/C ranges and total
costs to enable us to focus on high value activity. Exact costs arent nec-
essary, but we need ranges to know if were talking B/C ratios of 1/1,
100/1 or 1000/1. Ideally, we would like bound our approximate upper
and lower PM and failure cost estimates.
In non-regulated environments, cost is the primary focus of all PM
activity and naturally forms the primary basis for any activity.
Practically, even with regulation cost is the basis for most PM hours.
You dont have to cost out every PM case in detail. Rather, you need
enough benchmark cases-on the order of 5 to 10, so that you can quick-
ly assess any new case by comparison-for cost analysis and priority rank
purposes. Standard development has an obvious cost tie.
The development of benchmark cases aids in regulated PM activi-
ties, as well. Regulated PMs are usually easy to identify and because a
law or license specifically requires PM tasks, these tasks often record
their legal source in the basis. Rules for reports concerning continuous
emissions-monitoring performance are a straightforward example.
Under the older discharge permits or licenses, PM requirements were
non-prescriptive except at the brush-stroke level. They once went no
further than general mandates for appropriate PM programs.
(Whats an appropriate program? An outcome-based answer is obvi-
ousone thats applicable and effective!) Nonetheless, theres consid-
312
erable latitude for interpretation. However, a rule for annual lift-off

tests of main steam safety valves has no such latitude. Both rules require
compliance. The latter is more explicit! The challenge is (always) to
operationalize maintenance strategies. Laws just make this cursory.
(CEM requirements are now explicit and detailed.)
The RCM process was originally developed to resolve problems in
a regulatory environment, so its a naturally effective fit. Regulators
often have trouble with maintenance engineerings finer points that are
not so obvious. (One is OTF.) Many regulatory bodies work with a
TBM mentality because its concrete and simple. Its obvious when the
job is done. (The irony is that the FAA nurtured the development of
RCM in the 60s and 70s) Until regulators see more direct applications
of RCM-based inspections and age exploration by competent organiza-
tions, they wont understand it. Another challenge is streamlining exist-
ing programs. There are many regulatory proponents of TRCM main-
tenance program development methods, but few SRCM advocates. Yet
SRCM is just as valid in regulated environments as any other. Seeking
a closed-form solution is simply forgone in full keeping with the spirit
of original RCM.
If an empirical approach is effective then use it. ARCM takes an
empirical approach. TRCM advocates will say that TRCM provides a
complete solution, but no one can guarantee that there wont be improve-
ments. No one can guarantee that all failure modes are enumerated,
though reams of paper analysis give some confidence. New technologies
are always developing, and the better you understand any failure, the
more options and ideas you gain to manage it. Working with plants over
time into implementation, we always learned and refined our RCM results
as we gained more experience and learned more about costs, techniques,
and methods. We often compromised exact RCM results to achieve a
middle ground that supported plant personnel. Software must accom-
modate such learning; it must be a part of a living system.
PMO streamlines PM programs. While PMO (the fossil equivalent
of SRCM) isnt as thorough as ARCM from a basis perspective, it offers
attractive maintenance returns by eliminating ineffective PM activity
and extending otherwise conservative task intervals. Theres a bias in
any maintenance program to increase work in the form of ineffective
313
PMs to fix a host of partially examined problems.

Addressing this problemwhether by PMO, RCM, or ARCM tech-
niquesrequires discipline. Its stressful. Those charged with making
judgments, such as extending intervals, often have little formal training
or background to do so. (They do so incrementally.) Management
rarely can spare the time to train. This is specialist work that often calls
for a contractor. Its difficult to achieve appropriate intervals based on
actuarial data alone. Until a person gains confidence with equipment
and the process of routine PM interval adjustment, hes reluctant to
make any substantial changes in tasks or intervalseven those that
would greatly benefit the existing program. Maintaining schedules and
managements expectation is likewise difficult.
One solution is a tiger team task force on PM task selection and
intervals, headed by a central R maintenance engineer with the author-
ity to make calls. Nuclear plants have a R engineer role for the inter-
pretation of maintenance rule compliance and measures. But theoreti-
cal compliance and practical support of PM activities are at extremes.
My experience with maintenance rule engineers is that they dont offer
useful hands-on experience and support developing appropriate main-
tenance tasks and intervals. They are immersed in the arcane world of
nuclear licensing.
Even with something as innovative as a tiger team, however, tra-
ditional CMMS processes start at a problem report from operations.
Other work tasks arent the primary focus of the CMMS. WOs origi-
nated from OCM need a follow-on work tag. PM follow-on condition-
directed tasks head off equipment failures. RCM associates secondary
condition-directed work to primary time-based monitoring. The result-
ing process is particularly simple to model. A RCM-designed CMMS
would be different from current failure-based models that presume
maintenance starts with operator-identified problems.
Tracking and measuring time-based work could be handled more
easily in a CMMS that was designed in an RCM format. Any new PM
task (as well as any proposed changes) would need justification. Basis
maintenance is a maintenance engineering function either way. When
PM originators, assignees, or deferrers sign for a change decision,
there are fewer problems with deferred PMs (such as safety/relief valve
314
tests) that somehow just disappear! This discipline diffuses through the
PM program so that it gains credibility.
Air transport, hazardous material handling, and nuclear generation
are areas of public concern that will need to continue to document and
justify maintenance programs from a regulatory perspective. This, too,
is an opportunity for software heavyweights who develop an interest
in maintenance performance models. A CMMS ultimately manages
workany workthat must be performed. A CMMS that provides a
seamless tie to an RCM-based work development system will always
have an inherent advantage. This transition to RCM software will fol-
low the implementation of standard CMMS programs.
Backing-off an existing program (one with too high frequency or
where regulation is relaxed) requires a basis document. PMO or RCM
software that provides basis maintenance as a feature has a one up on
other methods.
A non-regulated market environment really doesnt care why a
change occurs, in contrast. It is more than sufficient to justify where
you are at a point in time, based on cost. In this case, a well-prepared
basis document must present a case for a PM activity for economic rea-
sons. RCM development software and its related CMMS database
must have work efficiencies as a goal whether the environment is regu-
latory, economicor whatever.
Many PM programs developed informally and were subsequently
grandfathered by regulators. A basis is implicit. We presume that at
one time there was a good reason to do all the PM work specified.
Justifying a specific change is referred to as a partial basis. It documents
the purpose of one change. A complete PM program justification, on
the other hand, is comprehensive and includes all relevant program
documentation. These are known as full bases. Many programs sur-
vive on partial basis PM changes.
Documentation
Basis development in a non-regulated environment only needs to
suit the company and economic conditions. At a very basic level, all PM
is based on cost. Regulation-mandated PMs means that the potential
315
cost is the risk of being shut down for not doing legally required PM
tasks. Or worsethe costs associated with injury to the public or
employees from an event.
Typically these involve rules, agreements and understandings with
state health departments, boiler inspectors, federal agenciesFAA (for
stacks), EPA (emissions), OSHA (industrial safety), Department of
Transportation (DOT) (gas transport) and occasionally others. Its
common to have a regulatory body endorse a professional groups stan-
dard, like the NRC and state endorsements of the ASMEs Boiler and
Pressure Vessel (BPV) Code. Occasionally, rules or standards conflict,
like the mixed-waste jurisdiction issue only recently settled between
the NRC and EPA. Overall, these standards are the first levels of com-
pliance that need to be assured in a PM program. Their source docu-
ments are voluminous.
Many companies also separately endorse building codes, the
National Electric Code (NEC), and many of the supporting ANSI,
American Society for Quality Control (ASQC), ASME, IEEE,
American Society of Civil Engineers (ASCE), SAE, ASCLE, and other
technical body codes. These codes are impressive in size. Codes repre-
sent the best effort of a group of knowledgeable and interested people
to provide guidance on how to do something. Theyre often general,
vague, confusing, and subject to interpretation and change, but theyre
also the best source of information on any subject for which you arent
already an expert. Occasionally theyre dated, or organized based on
changes.
Insurers develop and maintain inspection standards in addition to
codes. Boiler and fire protection requirements are two examples, but
there are many others. Fossil boiler insurers and their representatives
often want to know specific ways that an owner implements a code
requirement. Occasionally an authority designates an implementing
agency for a code requirement. (I worked in a plant that had this
arrangement with the state boiler inspector. The state recognized that
industrial insurance agent and his engineers recommendations had the
authority of the state boiler inspector behind them.)
Industrial insurers may identify risk areas they would like
addressed. Sometimes this has to do with a facility. More often its for
316
new monitoring or other risk management equipment. It could be fire

detectors, wet pipe deluge, or an upgrade in physical plant. These
agreements carry the force of contract behind them, and should be
tracked like regulatory commitments, when made. Being able to track
and assure that insurance commitments are in effect has a favorable
impression on insurers. Obviously, they dont carry the force of law and
can be more easily negotiated. But if management agrees that some-
thing is really good to do, we can only assume we should do it until
directed otherwise. We should provide the means to performance
within our processes.
At that point, procedures and checklists should be based on
ARCM-derived work activity.
Products
The products of RCM basis documentation are the tasks done to
avoid failures. Organizations that have traditionally performed them,
as well as the products themselves, can summarize what these tasks
are. Most nuclear organizations would recognize the on condition-fail-
ure finding RCM task as the rough equivalent of their SP. (The SP is
based on the literal technical specifications of the nuclear plant, so the
correspondence isnt exact.)
This illustrates why the RCM paradigm is so useful. Its helpful in
cutting through the organizational muck, and getting to the basics.
Sound engineering, sound maintenance, appropriate to the situation.
Round checklists
Checklists for rounds are the staple of the roving operator. They
provide guidance on what to check, how often, limits, and other inci-
dental information. For the control operator, they are summarized in
software as screen pop-ups that require entry from a DCS.
Rounds are being modernized with hand-held wands and monitors.
Portable-monitoring devices can provide a seamless tie from the rover,
reading nonDCS data, into the CMMS or even DCS through a down-
load. This in turn supports trending.
Obviously out-of-limit equipment is a candidate for immediate
317
remediationclassic RCM-based CDM. The largest fraction of

rounds is non-specific checks, where the operators judgment deter-
mines whether an item is serviceable or not.
PM tasks
Scheduled work activity that includes tasks implemented by WOs
using personnel assigned to maintenance is the traditional scheduled
PM program. Two primary task groupings comprise the TBM plan-the
traditional PM and on-condition-maintenance. While OCM has no
exact equivalent in a traditional program, it does reflect the intentions
of the traditional program PdM. The distintion is the level of diagnos-
tic task scheduling (OCM) and follow-up task performance (CDM).
Organization
Organization of PM activity into useful sub-categories is a prime
benefit of RCM. The distinction between a non-specific CNM task and
an on-condition one is the simplification of routine scheduling and
work priority this provides. In so many words, the mature program
schedules failing-condition equipment for maintenance before indeter-
minate-condition equipment.
CMMS integration
The best RCM systems integrate cleanly with the CMMS. On the
front end, they use similar coding and system definitions to organize
strategies. They tie in instrument plans. They easily upload completed
grouped plans into the CMMS.
They should not require repetitive entry of CMMS WO plans for
the maintenance work plans. They should allow the later addition of
work plan detail from reference documents. They should support stan-
dard work plans for repetitive work-the case for virtually all PM.
RCM/CMMS idealization
What would the ideal RCM-based CMMS/RCM maintenance strat-
egy development and implementation system look like?
It would probably have a very different emphasis than traditional
CMMS systems that are based on the concept of broken equipment.
318
(Someone sees broken equipment and makes a trouble report/work

request.)
Practically, this latter model is not only highly reactive, its not a
good model of how world-class maintenance works. The very best
maintenance organizations maintain a continual process of monitoring
equipment and anticipating its failure. They dont wait to notice prob-
lems. In this regard, the traditional model lags workers and is more hin-
drance than help.
An ideal system would provide ongoing monitoring guidance, as
well as authorization to correct degraded equipment on an ongoing
basis. This requires very highly skilled, proactive maintenance, and
operating staff.
Configurations Simulation
What maintenance work is best in a given situation? With a simu-
lation model, we can take the maintenance plan, enter all the factors,
perform a Monte Carlo simulation, and see what R answers result.
Simulation is available today on PC to help facilities test strategies for
redundant trains and instruments to see which, in fact, is best. This can
help avoid learning lessons the hard way. Very often minor design
changes can have significant R paybacks. Maintenance routines are also
likely to benefit.
Software can simulate the availability impact of proposed modifica-
tions, before theyre made. Abandoned modifications resulted in some
cases only after the plant saw the impact of the modification in a plant
trip. Plant trips at even a modest-sized plant cost well upwards of tens
of thousands of dollars. For large base-loaded facilities, they may
approach six figures quickly. The age of trial-and-error design changes
in the utility industry is rapidly coming to a close. Its simply too expen-
sive!
Simplicity
One risk of simplified, computerized, RCM analysis capability is
analysis-paralysisdocumenting every potential failure possible.
319
Having been down this path myself, it is necessary for me to exercise

great discipline to avoid it on any given project. As an expert, you can
easily go into hypothetical mind dump. Clients may encourage it. In
reality theres rarely more than several important failures, and these
occur only on larger subsystems or components. We need to remind
ourselves of the potential benefits of effective maintenance as well as
what maintenance cannot do. Most components have one to two dom-
inant failures that occur commonly. Some might prefer exhaustive
analysis done as a matter of course. But its not necessary to paint a
clear R picture. For those who disagree with this position, consider how
you maintain your car or house. Chances are you plan around the one
or two things that do happen.
Policy
Some corporations may find it useful to establish company-wide
RCM standards. After an analysis has been completed, it represents a
significant amount of learning, and its logical to apply this at similar
facilities.
In the final analysis, the system that makes the work elegantly sim-
ple is the best one. Where learning is transferable with existing
processes, it should be transferred.
320
Chapter 9
Measures
We dont see things as they are, we see things as we are.

-Anais Nin
Measurement
Global
Accepting the challenge of an RCM/ARCM program is an example
of a process shift. When a process shift occurs, what precisely happens?
At the plant or company level, it doesnt mean that we instantaneously
have a new process, with new results. Any major organizational change
requires time, effort, and resource dedication to implement. But what
if we could change a system or its inputs instantaneously? We check a
control system response by feeding in a new signal. (Fig. 9-1) Treating
a system in the same way, we should see a new response (with some
dynamic delay). If we could model in this way, what would the response
level be?
Theoretically, output would start to respond once the change
321
Figure 9-1: Process Change
occurred. When control system input takes a step change, the process
output instantaneously has a new equilibrium value. It just takes time
for the process to get to the new value. Taking the system to be the
maintenance process, the input as maintenance selection, what are
some suitable output measures? Based on theory and our projections,
what do we expect to change? Our outputs are maintenance costs, unit
production cost, and R. To see change, we must measure their
response. Ideally, we achieve an appropriate level (Fig. 9-2).
We change a maintenance or operating plan to either increase pro-
duction, reduce costs, or both. R is a byproductits difficult to meas-
ure until we drop to the more sensitive system level. At the system level,
responses are easier to seeif we have system level measures. We seek
322
Measures
Figure 9-2: Maintenance Cost
measures that tell us we whether our change influenced performance as

expected. The questions are, what do we measure? How can we meas-
ure projected benefits?
The process and inputs are:
maintenance planning (input)

maintenance performance (the process)
measures (output)
No change can occur until the maintenance plan changes. Thus, the
initial effort after an ARCM effort must be implementation of the plan.
This takes time, because ARCM is implemented at the system level. To
speed the results and identification process, those systems with the
largest potential for improvement need to be selected. Problem sys-
tem selection usually isnt difficult, but depending on the level of plant
measurement awareness and sophistication, it can require some time to
identify the potential value-adders. Sometimes secondary failures arent
accurately reflected by their root causes.
The kinds of system problems reported in NERC statistics are gen-
erally the same for a given class of units. For example, coal-fired boil-
323
ers tend to have a high rate of boiler tube failures induced by fly ash-
erosion. Boiler and turbine losses typically represent the top two loss
contributors. To understand the spread of losses at a particular unit
requires understanding root cause losses in depth. This starts with
loss reports, but doesnt end until the loss drivers are understood.
This identifies the low hanging fruit. Reviews require an up-front
assessment-an intermediate step to assure the effort focuses on the best
improvement targets.
When a maintenance management system provides accurate num-
bers up front, theyre most helpful. Many CMMSs can identify these
statistics, but only to the degree data allows. Sometimes a surrogate sta-
tistic must be sought when a key field or other indicator isnt available,
or available but unreliable. For example, CMMSs that record hours
independent from time cards are suspect for time accounting data accu-
racy. CMMS reports of emergency WOs may be suspect based on the
uniformity and control of the category emergency.
Systems with performance problems often have multiple problems.
Systems with low availability are also typically high cost systems. More
hours are worked on these systems, much of it overtime, on short notice.
By focusing on the half-dozen highest cost systems, the effort has
much greater probability of success. There are often masked secondary
failures in the measurement, so analytical review of system costs is
required. Three factors that trend together are availability, R (evidenced
by forced outage rate), and costs/work hours. By simple thumb rules of
estimating, non-labor cost and work hours tend to roughly approximate
each other in total cost terms. i.e., a staff with an annual payroll of
approximately $5 million spends $5 million on parts and services.
Focused
Every system has inherent cost profiles, based upon designs show-
ing their inherent R with regard to cost, availability, and man-hours
needed to support a given level of production. Benchmark comparison
figures for similar plants are very helpful to understand where nominal
levels should be. After selecting improvement areas for focused effort,
several change iterations may be needed to achieve the desired result.
A detailed performance measurement system is necessary to meas-
324
Measures
ure the effectiveness of the effort. Competing measurement factors can

pose difficult alternatives. The three major measures are:
production
R
cost
Workers today are more aware of plant level production costs.

Value added varies by day, season, and conditions, and is highly vari-
able. Plants cant plan to market projections. They must focus on well-
known production factors. Market projections are at best an estimation
of expectations.
Benchmarking processes have gained credibility as a viable method
to compare costs. Widespread distribution of operating and mainte-
nance generation methods makes process benchmarking a useful tool to
understand competitive position. However, access to useful benchmark
information is also difficult to attain! Potential competitive threats
make benchmarking partners less likely to provide useful inside infor-
mation that may point towards fundamental weaknesses. Third-party
vendors, contractors, and consultants provide another way to obtain
competitive information. Benchmarks outside the industry are also use-
ful. Comparison with manufacturing and service providers often points
to innovative opportunities.
There are, for given classes of generation, intrinsic costs. Nuclear,
coal and gas all have intrinsic factors that fundamentally shape their cost
profiles. These profiles change slowly during time, but can sometimes
be influenced fundamentally. Three Mile Island upped the nuclear ante,
just as the energy crises of the 1970s impacted fossil generation.
Greenhouse gas remediation may eventually increase the cost and utili-
ty of high-carbon fuel combustion. While the burden of regulation has
been born most heavily by nuclear, indicators point towards steady
increases in the fossil generation areas.
Regulatory costs vary dramatically site to site and may be the most
significant hidden cost factor. Certainly, few nuclear planners dreamed
that the impact of licenses and construction delays would so strongly
influence their competitive position. Some plants racked up licensing
325
delays of nearly a decade in high-interest finance costs added to their

fundamental capitalization. Surely, in hindsight, those plants would
have never been built. This explains a hurdle to new base load generat-
ing units today-too many risks in one site.
My favorite ARCM effectiveness measures are based upon financial
performance supported by subjective interview assessment. Financial
costs include, by rank:
unit cost of production

system production cost
major equipment production cost
major department costs
work type costs
services and parts costs
Effective CMMS data sorts are indispensable to account for costs.

Where these CMMSs are not available, cost collection becomes
dependent upon FERC reports and in-plant techniques.
Performance measures include:
unit forced outage rate

unit availability
market-weighted availability
system forced outage rate
system availability
major equipment forced outage rate
emergency work
overtime work
Employee interviews help establish perceptions as well as problems.

Where there are discrepancies between perceptions and facts, percep-
tions establish the organizational perspective, beliefs, and focus.
Perception at one plant was that coal handling had no impact on
production despite a severely degraded coal handling system. Aside
from high cost, coal handling functions had been severely compro-
mised. One obvious deficiency was that the plant had no means to
326
Measures
remove iron or debris from the coal feed stream. All such material was
fed into the crushers, bunkers, feeders, and mills, where it was either
pulverized or caused random trips that required isolated, unplanned
entry to remove it from the equipment. All the while generation was
lost-up to 1/2% availability due to these events alone. At the wholesale
generation value added rate, this added up to a cool $486,000 annually
for that base loaded unit. Year after year the capital budget request for
tramp iron removal equipment and metal detection upgrades (budget-
ed between $100-200,000) was edged out by more glamorous projects.
Focus can be redirected once interrelationships are understood.
But it shows-from a RCM perspective-how design basis system func-
tions have gradually eroded and even vanished over a units in-service
life-to the units cost-competitive detriment. Nuclear units do not suf-
fer design memory loss, but they pay dearly to maintain that memory.
Training and documentation expenses are correspondingly higher.
All measures start with operating goals. Awareness of the competi-
tive profile, as well as industry standards and capabilities, are helpful in
establishing meaningful goals. The pursuit of meaningful goals is
exhaustively covered in literature on TQM process methods.
The key is to find a parameter that provides improvement focus.
Even in those rare cases where systems or equipment dont show obvi-
ous stratification, there are ample opportunities to focus on perform-
ance.
Changes and Measures

As deregulation progresses, generation R, and cost per net MW
hour, take on greater importance. These integrating measures give you
health overview, but they dont focus on improvement opportunity at
the plant level. For individual units, opportunities must be evaluated in
terms of specific mission goals. Measurement capacity depends on the
company (and stations) CMMS and other management system capabil-
ities. My experience has been that within the same company, different
units, and different departments within units, will use CMMS software
differently so that comparisons are both hard to make and take lots of
manual data manipulation. Ideally this would not be the case.
327
Measures
Consider traditional system level unit cost measures. Units can also
be measured at the systems level with an FERC-like cost reporting sys-
tem. FERC categories are:
boiler steam supply

turbine generator
auxiliary
FERC measures dont roll up as a system-oriented hierarchy and

FERC categories dont match A-E system boundaries and descriptions.
Information is ideally accessible at both the systems and equipment
level so that a cost hierarchy can be rolled up for coded equipment to
any appropriate level. Many legacy CMMSs today cant meet this need
because measurement hasnt been essential.
Measures need to focus on risk and economic importance. Safety
(accidents) or lost generation measures are most difficult. Maintenance
costs are known at the generating unit levels, but their allocation down-
ward to equipment is typically unavailable. To quantify risks, consider
such things as:
unit forced outage data

equipment emergency work orders
overtime
special part usage
insurance claims and audits
regulatory information or audit findings
industry experience
Extracting data requires reading computer and summary reports.

These are accessible from the right sourcesthe insurance department,
for example. There are also simple default measures and subjective staff
interviews. Operators and their direct maintenance support, have
excellent risk perception. Interviews can confirm other operating data.
Industry experience is also an excellent tool to quantify risks.
Experience around plants for many years, knowing how they are run,
328
Measures
and what fails, helps interpret risk patterns. Operating risks can be
managed as long as theyre known.
At an operating level, economics factors (corporate wage rates, his-
torical costs, and trends) are known. Total systems/equipment hours
worked, maintenance strategy mixes, equipment CM/PM (by hours),
emergencies, and total costs are also known. To interpret these num-
bers, you must also develop system/equipment risk profiles. Theres
nothing inherently good or bad about working 60% corrective/40%
preventive work on a system, unless you compare it to a known bench-
mark. Knowing that a systems ratio of reactive maintenance and its
competitive operating costs are high suggests finding out how competi-
tors operate similar systems or performing general benchmark studies.
One of the most exciting aspects of ARCM implementation is the
opportunity to view performance data from an entirely new perspective.
Historically, plants followed costs and work hours. Some further broke
these down into CM and PM. But if maintenance is better off per-
formed on demand, PM/CM categories have no meaning. New RCM
maintenance categories enable consideration of measurement-and what
can be measured. Most of these categories cant be measured directly
with existing CMMSs. But, new CMMSs can.
System measures
System level cost measurement is the minimum level to ensure that
unit performance expectations are measurable. Usually a system must
meet minimum safety standards and support pre-set production levels.
There are two broad systems categories. The first directly supports pro-
duction, the latter provides production support service. A loss of serv-
ice system functionality has a delay before production halts. Only a few
production systems directly support production.
Examples of production systems include:
fuel system
primary coolant system
main steam
reheat steam
feedwater
329
circulating water
flue gas (boiler)
electric conversion
Examples of service systems include:
waste-water
ash handling
service air
coal handling
domestic water
Theres always a level of risk with production. A coal-fired unit

operating without flame-proofing boiler protection is taking a grave
risk. Companies (and their insurers) apply grace periods to restore a
failed system of this sort to service. Instrument air (IA) systems could
fall into either the service or production category depending on equip-
ment. IA usually supports feedwater-regulating control valves, and tur-
bine extraction valves-important equipment at startup.
Minimum availability performance for production systems is nearly
100% at the unit system level when the plant is scheduled to be avail-
able. For support systems, a predefined service level is based upon his-
torical performance. A coal unit with a coal handling system availabili-
ty of 85% (defined as the capability to run coal to the bunkers) can run
indefinitely at full load. Another unit, with fewer redundant belts,
found that 95% availability is necessary to assure continuous produc-
tion with the same maintenance approach.
When scheduled maintenance periods are considered, system avail-
ability will shift. But pre-planned, on-line maintenance periods can
usually be planned and managed so theres less risk of unit outage.
Systems that impact productionwhether its sootblower-induced
boiler tube erosion or feedwater upset that trips the unit on high deaer-
ator leveldemonstrate explicit system functional failures. These
should be displayed in Pareto fashion by availability loss contribution.
Doing so accurately requires careful record reviews and RCA. If soot-
blower-induced boiler tube failure is never identified as a secondary
330
Measures
failure, then the boiler is charged for the tube failure. This bias misdi-
rects our effort on the wrong systems. Inconsistent failure reporting
makes this doubly difficult. Theres just no way to avoid becoming
familiar with the actual performance numbers.
One refrain heard many times over from station managers has been,
That failure will never happen here again. It wontat their unit
for five to ten years. After that the lessons learned are forgotten and the
potential for failure reoccurs. Monitoring one companys fleet of 20
large generating units, I found that fleet problems such as winding fail-
ures do recur. Unless they change some aspect of basic operation, over-
all risk levels remain the same.
Statistical failures can be compared to traffic tickets. Speeding tick-
ets, accidents, and insurance losses correlate. Eliminating risks from an
insurers portfolio begins with elimination of speederscharging that
risk category a higher premium to cover the higher risk. At the plant
level, few stations keep speeding tickets. Major loss precursors often
go unnoticed, but at the system level, its much easier to identify and
track precursor near misses and use them to predict future risk and
system level performance.
In the absence of a near-miss program, how can you identify sys-
tem level risk performance? One method is to track two measures that
correlate system risksystem equipment emergency WOs and over-
time. These indicate the degree to which unplanned events influence
system performance. These indicators can serve as red flags. Of course,
the absence of a system management plan, a system owner, and opera-
tional awareness are the big warning signs. With one or more of these
factors present, loss factors decrease.
Cost measures including total man-hours worked, total costs, and
how these are allocated between and among various work categories
need to be followed at the system level. Remember that a PM hour is
an effective work hour-the work is planned, predictable, and the value
added has been pre-identified-while an emergency hour is most ineffec-
tive. Given alternatives the PM hour is preferred.
After system outage work is controlled, emergency work should be
addressed. A system profile of planned CNM, TBM, and OCM offers
the lowest cost. Superficially similar systems can have very different
331
cost characteristics. A system maintenance plan compares and contrasts

cost factors for future needs and reference.
Failures
Another exciting aspect of RCM is the new perspective it provides
on failures. This is true both in regulated and unregulated environ-
ments. In place of the easy and conservative (and 100% bulletproof)
position of, When in doubt, or questioned, call it inoperable! we have
proactive, pre-planned assessment engineering and failure descriptions.
A confident pre-assessment can be provided to those who must make
shutdown decisions.
Failures influence economics, and are therefore worth measuring.
Some failures have license or emissions impacts, as well as production
loss value. There needs to be a consistent basis for measuring failures
at the system level. The NRC imposed this by regulation with the
Maintenance Rule at nuclear facilities. Nuclear plants must track sys-
tem availability and MPFFs at the system level for risk-significant sys-
tems. Risk-significant functional failures can be identified from operat-
ing logs, and E WOs.
Two difficult failure types involve redundant equipment. The first
is redundant trains, like a standby feedwater pump train. The second is
protective devices. Their function is redundant in the sense that they
serve to alert or prevent another primary function failure. Protective
devices whose function isnt required (nor typically desired) until an
event occurs spend the majority of their lifetime in standby waiting for
an event. Like redundant equipment, no backup need occurs until we
lose the primary. An unintended transfer constitutes a control transfer
functional failure. In the case of the redundant train, inadvertent trans-
fer doesnt constitute functional failure-it merely swaps trains. But for
a protective device, the component functional failure as an inadvertent
activation very often creates an unplanned event and system level func-
tional failure results. So, an unplanned and incorrect feedwater level
trip is an unintended event-and failure. A spurious high-vibration alert
trip on a steam-driven boiler feedpump is a failure.
Sometimes a functional test, a near miss, or a demand-event
uncovers a protective device failure. Because safety devices have multi-
332
Measures
ple backups, demands rarely utilize all devices and trains. Events typi-
cally are near misses, and most of them reveal loss of protective device
redundancy and greater risk of functional failure. Nuclear power plants
are effective at testing and identifying device functionality under
nuclear technical specifications, safety programs, and general public
safety considerations. Fossil unit rules arent as structured.
System functional failures provide an objective measure to track
failure performance. Though subject to interpretationmore so where
the functional requirements havent been formally identifiedthey are
still very useful. Total WO numbers are arbitrary as a system perform-
ance measure, but system work hours and costs are not. For nuclear
plants, MPFFs remain a convenient measure. For fossil, large
unplanned expenses may provide the failure measure. Because fail-
ures themselves are hard to track, I developed two measures that corre-
late with failures and are simpler to use: system emergency work orders
as a total number and percentage system overtime. These are easy to
extract from CMMSs at most plants. Both are indicators of functional
failures.
Costs
System operating costs are the obvious summary performance
measure. Some systems are more cost intensive than others. A soot-
blowing air system is an inherently high cost system in a PRB-fired boil-
er. Cost performance per standard cubic foot per minute of air pro-
duced is one measure of this systems performance. Unit efficiency and
boiler pluggage events are others.
Because CMMS systems dont always allow cost monitoring below
the plant or unit level, system costscombined labor and material
provide meaningful cost data. As important as total system costs are,
other numbers can tell more. The cost of overtime hours worked per
system, or costs of irregular part expense, are examples.
At one plant, we arbitrarily selected unplanned failures costing
above $25,000 to measure for overall performance. This selective analy-
sis required manually tracking CMMS entries (subject to interpretation)
but the results told a subtle story. Quantified in this way, eight major
failures a year dropped to five because of our efforts. Stratifying meas-
ures must be performed with great care, and the advice of a statistician.
333
PM Hours
Real hours
Although in an ideal world all planned PMs get worked, in the real
world this is never the case. Auditing PM jobslike other CMMS-
reported jobsprovides insights into how completely a program is
implemented.
Many Legacy CMMS systems ran redundant time accounting sys-
tems separate from the time cards submitted for pay. WO completers
could report any amount of time on WOsthe CMMS couldnt check
or enforce simple time accounting rules. One aspect of this was that
workers could work (according to CMMS time reports) any amount of
time. In my experience, workers are biased by work planning time esti-
mates and management expectations. Without an independent time
accounting system, they arent grounded by real world limitations.
Measurement helps to provide this!
For accuracy, a system should charge time concurrently from time
cards to WOs (or jobs, as appropriate). Fractional time charges down
to the decimal hour are needed for PM time measurement. Many PMs
are brief jobsshort enough that several may be worked in a morning
or an afternoon. Accurate cost accounting is necessary to understand
where time costs are allocated. In a typical plant, only 40-45% of work
time gets charged against work jobs. The challenge for maintenance is
to put in wrench time. Other things (like training) are important but all
compete for limited available time. When wrench time drops below
40% the plant needs to worry about time usage competitors. These
include safety meetings, extracurricular duties, and other supernumer-
ary tasks. Mechanics who dont turn wrenches have little value.
WO time accounting must be controlled like checks. Paid time
must be managed. Major time charges should be organized by Pareto
chart, be tracked, and audited. Routine time consumersrework, tool
and parts shagging delays, tag-out delays, engineering support delays,
and job planning (by the workman)should be given charge numbers,
so detailed time charges can be allocated. Bottlenecks, delays, and
other losses can be identified for organizational review. The goal must
be to continually increase work time charged to jobs in spite of the
334
Measures
steady stream of worthy programs and other distractions calling for

worker time and support. Without a wrench time productivity increase,
corporate programs wont justify their costs.
Cost of a maintenance hour

A maintenance hour costs much more than a maintenance line item
budget indicates. The range is several times more. (Fig. 9-3) This is why
maintenance is more expensive than most people assume. Estimators
peg job costs at two times the labor hour rate. When the time charged
to hours and actual time on the job are factored in, one quickly con-
cludes its hard to get hours worked in plants. The cards are stacked
against the poor maintenance worker. He has to do many thingsespe-
cially if he plans and coordinates his own work. This makes expensive
maintenance.
Why perform PM? A PM hour is a leveraged hour. It provides a
payback (when properly developed) that is many times the invested
time. Corrective maintenance isnt so leveraged. Unfortunately, many
personnel fail to grasp the value of these different kinds of work hours.
Emergency. Emergency work hours are underestimated, based on
detailed correlation with actual time worked. An emergency work hour
is between five and ten times underestimated, by estimating bias. This
estimate is based on numerical studies of work categories. (Many
organizations fail to evaluate and learn their estimating biases.) If the
estimate goes unmeasured and uncorrected, and it changes the capital
hurdle rate costs over a factor of five-to make a modification financial-
ly successful, you must get at least five times more than the estimated
return!
PM/CM Mix: Maintenance Process Measures

Effectiveness
No discussion of PM is complete without considering the measures
of the entire maintenance process. While there are many measures
requiredFERC, NERC, INPO, Edison Electric Institute (EEI),
EPRI, and othersutilities vary in their measurement performance.
Regulatory-based need measures fall short when used for other purpos-
335

Figure 9-3: Maintenance Work Cost
336
Measures
es, such as cost accounting. Measures based on first principles have

more direct use.
Responsiveness
As maintenance programs evolve towards CNM, measurement bal-
ance needs to be sought. Appropriate levels of CNM depend on system
type and strategy. Many strategies can achieve the same operating
objectives though at different cost and complexity levels. A strategy
must fit an organization.
Failure measurement is not possible without accepted failure stan-
dards. A traditional program lacks explicit definitions. Even nuclear
maintenance programs lack function-based failure criteria, as relatively
minor eventsthe charging motor run-on in a 4160 breakerget
described as failures. Major failures can go entirely unrecognized. The
secondary failure that results is often the focus of investigation. A
breaker fails to trip on overload, causing a fire, or an alarm fails to
annunciate an unsafe condition, like methane or carbon monoxide gas.
Events that should have been excluded by operation in fact become the
focus of investigation.
Total hours/system
Total work hours per systembroken down by PM (TBM +
OCM), CM (CBM + OTF) and functional failuresare a meaningful
RCM-based measure. The key ratios are the percentage of each ARCM
category. These profile the system. The continuous improvement goal
is to reduce required hours and costs. Where systems lack total cost
measurement capacity, tracking total work hours is a second useful
measure.
Trends
System downtime and functional failures are important perform-
ance measures. Identifying functional failures is a challenge when func-
tional-failure definition and perspective is absent. Having these meas-
ures requires that a company has engaged in goal setting for the unit.
This establishes relevant failures. Many havent.
337
Aging studies
WO populations should show continuous progression to completion.
If we examine progressively aged WO group snapshots over time, we
should see incomplete WO numbers decline. Mathematically, WOs are
worked proportional to number and age. When this doesnt happen, the
systems in trouble. Regular aging reports are useful for telling manage-
ment whether their maintenance system fundamentally works.
Costs
Change should generate improved performance, lower unit costs,
more flexible operations, or all of the above. Historically, plants have
never had income benefits allocated at the unit level and so higher
income generating units have requests buried in with peer units. Utility
cost and income aggregation are to blame.
Merchant-independents stand alone from a financial perspective.
Even then, unit costs allocate downward to systems and equipment by
tedious manual methods. The before/after snapshot of any significant
process change can be obscured. Utilities, as vertically integrated struc-
tures, suffer incomplete cost accounting at the unit level.
New CMMS systems greatly improve cost tracking. Theyre
dependent on data entry, but use hierarchies that are interact with and
extract data easier. They promise better information capture and pres-
entation.
Its difficult to tackle more than one plant system improvement at a
time. System cost trendstotal, routine, outage, emergency, and mod-
ification costsare major interest categories. Some cost expenditures
are most important. PM time and expense are among them. These
need tracking categories. For the cost-driving systems, these cost trends
will be important.
Ratios
Maintenance ratios tell a story. The emergency to routine mainte-
nance ratio-by hours-reveals how a systems work is managed. One
CMMS coded work priority on a continuum range from E to 3s (E-
1-2-3). Es were unscheduled, unplanned WOs; 3 were planned and
scheduled. High E/3s reflected reactive maintenance. (High and low
338
Measures
are relative.) Compared to other systems, units, plants, and companies-

these tell a story.
A high E/3 ratio may be suitable and effective for some systems.
Generally, for high intensity mechanical equipment, its not.
Companies pay maintenance workers overtime for emergency (E)
and unplanned work. It requires additional support, lacks parts, and
when all is said and done, is the least effective work an organization
does. A plant should avoid E work. Routine, planned, repetitive work
is at the other extreme. Its efficient and low cost. More routine plan-
ning should be sought. The ratio emphasizes this relationship.
Rework
Rework is a significant cost-contributor because maintenance is
expensive. If you can track rework causes, you can reduce them with
significant benefits. Manufacturers follow rework and scrap cost in
depth. Manufacturers neither want to make scrap nor send it out,
incurring warranty or other cost charge-backs. Maintenance is a
processbut tracking rework is like tracking scrap. WOs need to iden-
tify reworked equipment and jobs for trending and root-cause assess-
ment.
Workers, generally readily identify rework on jobs. With their par-
ticipation, rework maintenance can be measured. Things you can meas-
ure can be improved. Sources of rework should be identified for
process improvement.
Screening for effectiveness

Effective maintenance screens new WO requests as theyre input.
Different systems generate different work values. Many utility systems
focus on capturing all WOslegitimate, undefined, and speculative.
Nuclear plants in particular allow the documentation of incompletely
specified work. This is a great burden to a WO system, particularly
when it could be screened.
Maintenance screens quickly identify and reject inappropriate
work. WO screening is improved with a developed RCM process
because operator non-specific work requests are identified as CNM
type work orders. These cant be worked directly until someoneusu-
ally a work planner or engineerhas defined their scope.
339
Contrast this with a developed TBM or OCM (or OCMFF) type

WO. These are exactly defined and have exact follow-up work plans.
They have well-defined workscopes and failure criteria. The adoption
of an RCM model allows schedulers to quickly separate the known,
defined work for immediate work, from the vague, conjectural, ill-
defined and speculative work that constitutes OTF. These latter work
requests need specification, or return to originators until a failure can
be defined. A failure you cant specify, you cant fix!
340
Chapter 10
Conclusions
Learn from the mistakes of others. You wont get to make them all yourself.
-Eleanor Roosevelt
Just because somethings old doesnt mean you throw it away.
- Scotty, Relics, Star Trek
God integrates empirically.
-Albert Einstein
Organizational Entropy
Entropy explains why our natural order trends with time towards
disorder. Entropy explains why thermodynamic cycles have limits, heat
flows in a single direction, and why temperature and time have mean-
ing. Entropy is a powerful concept and one of the three laws of ther-
modynamics, (paraphrased):
You cant win (conservation of energy)

You cant break even (entropy)
You cant quit the game (thermodynamics provides the rule book)
341
Figure 10-1: The Fort Saint Vrain power plant in Platteville, CO., was operated as a
conventional nuclear plant until its sporadic operations and high cost caused its shut-
down and decomissioning. Today, its operated as a combined-cycle gas generator.
Recent organizational models apply thermodynamic principles to
organizations and processes. Entropy helps to explain the apparent
confusion and disorder among some large organizations as they do so.
Entropy can help us understand operating environments. We might
view a situation normal all fouled up (SNAFU) as an individual
faulta person messing up the jobbut an entropy model suggests it
is the nature of the system. Things will malfunction without continuous
addition of energy and intelligence to the process (Fig. 10-1).
This explains accepted aspects of operations that have never been
theoretically considered before. Operations demand intelligence and
energy. Every conscientious worker in an operating environment knows
this. Outstanding operations demand more! The assumption that
order is the normal state of affairs is simply founded on idealism.
Complex operations need information and control to produce value.
This only happens if intelligence offsets the systems natural tenden-
cy to unwind.
Management provides the framework that provides order.
342
Conclusions
Traditional management techniquesstrong-arming, intimidation,

shooting from the hip, power plays, scapegoatingare inherently weak-
er than techniques founded on facts, processes, and scientific methods.
TQM provides a general management model for process improvement.
Two distinct characteristics are organizational systems and processes
with feedbackprecepts similar to ARCM.
What are some intelligence inputs?
operating goals
operating plans
training
staff selection
work processes
standards
How can these be used to achieve consistency and predictability of

operations?
Organizational process intelligence is, in part, the shared knowl-
edge of the organizations members. Randomness results when workers
dont understand that individual effort matters. Successful organiza-
tions increasingly tap workers to help create value, control entropy, and
to retain and increase market share. Making a productany product
is a difficult task, even in a simple shop. Doing this profitably demands
creativity in a competitive environment. Electricity is a product and
generating it demands these same traits.
Maintenance is a complex organizational process with different lev-
els and perspectives. Through the years, organizations have developed
methods of performing maintenance based on experience, and applied
these with success. Thumb rules reflect fundamental rules and laws.
We dont need to understand theory to apply them successfully. On the
other hand, understanding the theoretical basis may provide insights to
enable us to apply the rules more broadly, and ultimately provide a com-
petitive edge.
Natural laws also govern business. Top performing organiza-
tions are those that embrace fundamental processes and principles,
those that actively search out new theory and technology to further
define their processes, and so maintain a cutting edge.
343
What is ARCM?
The general focus of RCM is identifying and preventing functional
failures. ARCM goes one step further. It throws out the dogmatic styles
in favor of pragmatics. In effect, If it works, use it. ARCM retains the
unique, fundamental principles introduced by Nolan and Heap,
Matteson, and all the other RCM pioneersthe factual basis, applied
statistics, applied engineering methods, benchmarks, and applications
based upon basic logic principles. Basics that validate NSM in a com-
plex equipment strategy. Basics that can effectively control high-
impact, large equipment outages with on-condition/CDM, and use
CNM as a general operations strategy. Basics that faithfully apply and
operationalize on-condition limits that so uniquely delineate Nolan
and Heaps published work from others. Methods to schedule time-
based and OCM equitably with the balance of non-specific work, with
assurances that CDM is worked. Methods that accept and apply risk
management to substantially improve performance, and lower risks and
cost (Fig. 10-2).
The bottom line is improved R (reducing functional failures),
reduced costs, improved quality, and well-supported corporate missions.
ARCM shares basic similarities with TQM. Both are founded on
statistics. TQM summarizes generalized lessons from early SPC appli-
cations that became tools for present day managers. Some conclusions
still apply. Others provide insights, but must be taken in context.
Embracing ARCMdeveloping strategies to reduce costsis what
facility operations are all about. A facility or company can practice
RCM and still not know about LTA or other detailed TRCM methods.
Does an understanding of RCM methodology help? Absolutely!
Years ago, corporate cost-cuttersaccountantstrimmed plant
budgets and cleaned out the shops. Experienced people left, planned
maintenance and training programs were cut. The opportunity to
improve processes was lost. Workers obviously disliked the top-down
cutting, but no one understood the value of the losses, cuts, or future
costs. Several years later, R was down, production down, unit costs up,
and maintenance was more reactive than ever.
Maintenance is an inherent cost of production, a fundamental con-
344
Conclusions
Figure 10-2: Maintenance Strategy Map

sequence of the second law of thermodynamics. You can trim mainte-
nance costs, but you cant eliminate the randomness or time elements of
failure. Improving maintenance requires process improvement. You
can play accounting tricks to flavor costs, but ultimately equipment and
system failures tell the bigger story. Creative numbers cant subdue
entropy or reality.
For those people who enjoy maintenance (just as others love design
and still others like to operate), who understand that the three roles
interconnect, and who pursue maintenance theory and technology in
large facilities hell-bent on the being the best, ARCM can help bring
order to a crowded, complex field.
Japanese authors, Masaji Tajiri and Fumio Gotoh describe a main-
tenance strategy called TPM. TPM shares some aspects of ARCM in
the final result. The end producttasks scheduled and workedare
similar. Different equipment and cost optimization schemes can lead to
similar ends. However, based on the published descriptions, TPM and
RCM have paths that are radically different. TPM presumes a large
amount of time available for group learning and fails to explain how the
345
infrastructure that supports maintenance performance gets built. (Yet,

infrastructure is required.)
TQM has supporting techniques such as SPC that share a great deal
in common with RCM, as originally defined. Statistical analysis, factu-
al basis, analytical methods, processes, and ways of deducing causes and
focusing on the critical few are common threads. TPM (the Japanese
approach) and TQM depend on an intense pursuit of work that, in my
experience, isnt a part of todays American industrial culture. Not that
people in American industry arent very proud of their work and work
effortsthey are. They just generally dont work 60-or-more-hour
weeks without extra compensation as other cultures do!
In addition, TPM and TQM are iterative approaches. In the absence
of another specific strategy, either (or both) appear to be good ways to
iterate general process improvements. ARCM, on the other hand, has the
capacity to generate the end-runthe substantial leap that companies
strive to find in benchmarking outside their own industries.
As a maintenance process, ARCM provides an improvement path.
ARCM is an objective, rational, statistically-based engineering strategy
that can be applied theoretically to new facilities to help initiate pro-
grams. Different means can achieve the same end but one approach
may support the end better based on culture, cost, and commitment.
There can be multiple paths but ARCM best integrates two functions
that have for years been culturally disjointed in plant maintenance in
North America-engineering and maintenance.
Consider baseball. You can commit an error here and there and still
win the game. You cant commit more than a few, though. Given the
tools available, developing a near error-free maintenance plan has never
been easier than it is today. Planned maintenance performance is very
controllable. To suffer losses because the maintenance plan isnt devel-
oped, implemented, or followed represents inadequate fielding. In
baseball, players need to field consistently. A similar standard is being
raised for facility managers today. Those who can manage maintenance
will step forward, others may need to step aside. Competitive markets
will force the changes.
346
Conclusions
Statistical maintenance
Maintenance itself is inexact, with many strategies, no single one of
which is right. Many work. Statistically, we must:
understand basic equipment failure types and frequencies

put controlled processes into place to manage failures
use statistical tools to improve actions taken to address an
identified equipment deficiency
learn and improve
A simple feedback loop goes a long way toward improving per-

formance. Consider the metaphor of the basic operational amplifier.
Feedback allows this relatively insensitive but high gain electronic
device to be transformed into a powerfully accurate, simple signal
amplifier. Sure, you sacrifice a lot of the theoretical gain, but you now
have a smart device locked on the input signal with plenty of practical
gain left. Feedback into the maintenance process can likewise greatly
help to reduce output variability and improve results.
In hindsight, most high impact events in my experience had obvi-
ous monitoring precursors. Some were so pathetically obvious we had
to tell ourselves (at the time) that nothing could have been done to over-
come glaring equipment deficiencies. These organizational blinders
were maintained even as outsiders repeatedly pointed them out. A
common thread was our failure as operators to consciously appreciate
our decisions and their risks.
There is a cultural weakness in the engineering transfer of intelligent
designs to operating groups. In many cases, the designers anticipate even-
tual failures and provide the means to monitor the high-impact ones.
Operators, on the other hand, systematically ignore essential instrumen-
tation in the vast sea of available equipment and fail to act based upon
instrumentation and senses when major failures are imminent.
Our ability to monitor for failure is too often compromised. ARCM
provides a substantial opportunity for designers and engineers to devel-
op better operator guidance for anticipated events over the facilitys
economic life. I envision a day when, along with the vendor manual,
operators receive a detailed ARCM-based optimized vendor mainte-
347
nance program. Operators, through their A-Es, will have the founda-
tion for a cost-effective maintenance strategy from startup. This in turn
will support better staffing, life cycle decision-making, and, ultimately,
lower overall facility maintenance costs. Performance levels will be
known for operators to benchmark. In short, a step-advance in facility
O&M performance is dawning.
Maintenance is to some degree an art. There are no panaceas that
suddenly make all maintenance decisions simple and clear. Even with
the very best RCM plans in hand, hard diagnostics, interpretations, and
choices are and will be required. But, armed with a maintenance plan,
operators will have better tools to interpret equipment, plan for main-
tenance, and perform work in a reduced cost fashion. This has been the
lesson of the commercial aviation industry. The challenge is to intro-
duce the appropriate degree of rigor into an ambiguous environment to
improve managing risk and cost.
A general, repetitive RCM lesson has been identifying a basic essen-
tial instrumentation strategy to help manage operating risks. The
opportunity suggested here is obvious: if the reader grasps an apprecia-
tion for the need to quantify, understand, maintain, and manage essen-
tial instrumentation from reading this book, they will gain great value.
The flipside of the coinlearning to manage non-essential instru-
mentsis a corollary. In this world of expanding hardware capacity, it
is particularly important to control the vital few versus the trivial
many. Operators must learn to discriminate and act on essential
instruments. This is the low hanging fruit that many maintenance
managers should grasp. Unknown or inadequate instrument mainte-
nance plans cost dearly in production, cost, and (rarely) in employee
and public safety.
The general lesson of RCM is CNM. Organizations with a CNM
philosophy are reliable. Its possible to go overboard but generally the
other case occurs. Little or no CNM, and absence of follow-through on
the insights provided by monitoring, are the trademarks of unreliable
and unsafe operators. Like the person who feels ill but is afraid to visit
the doctor for fear of having a worst fear confirmed, failure to act on
CNM adds risk. Understanding problems and alternatives enables us to
select options. Rarely are we saddled by Hobsons choice: take whats
348
Conclusions
offered or take nothing at all. Which organization is more successful in

the long runone that avoids reality and flees from information? Or
the one that embraces the future and its risks, and makes the best pos-
sible decisions with the available information? From a plant perspec-
tive, a heads up of a few days (or even hours) can often significantly
reduce operating impacts of imminent failures. The very best plant
operators demonstrate that unplanned production outages and costs
can be safety put to rest with a plan.
Applied to existing facilities, the ARCM approach can zero in on
areas of maximum value. Facilities are often inhibited from changing
existing practices. Change is expensive and hard to sell. There is a cul-
tural foundation at every plant thats hard to move, even in light of com-
pelling cases for change. New operators are more receptive to new
processes, and an owners desiresbut its still difficult.
Are there alternatives?

Twenty-five years of maintenance have led me to look for new meth-
ods that could fundamentally improve maintenance performance. Ive
reviewed TQM and TPM materials, both of which offer valuable
insights. Yet, I dont believe they are technology as much as philoso-
phy. Tajiri and Gotoh (TPM) outline methods that, in their final form
offer some of the same general insights as ARCM. If a solution is
unique, then differing methods should point towards the same solution.
This is a comforting validation.
What some maintenance optimization approaches ignore are spe-
cific methods of technical approach and process. This book has avoid-
ed detailed RCM process theory tedium or general strategy but readers
can readily find suitable materials in the references that will guide them
through RCM subjects. Weve focused instead upon the lessons and
practical elements of maintenance performance as they relate to opera-
tions.
Maintenance is a sophisticated process mostly taken for granted
especially by corporate financial staffs, managers, and executives, who
presume they understand the finer details when in fact that takes years
in the trenches. The good news for them is that there are many oppor-
tunities to improve maintenance while holding the line or even reduc-
349
ing costs. Many organizations can and have positioned themselves to

take advantage of this. Many more will join them with existing staffs
and processes largely in place. External forces will redirect some oth-
ers. For traditional utilities, competition is a powerful outside force,
and most will need to make a conscious decision to find new mainte-
nance approaches.
For innovators, these are exciting times. A tremendous amount of
theory, technology, and process understanding has led to practical prob-
lem approaches. OEMs are more aware of user needs. Software sup-
pliers are finally grasping software friendliness and accessibility needs.
New monitoring technologies have been perfected and become accept-
ed. Corporate utility staffs can now visualize profit and loss differences
in a competitive environment in plant availability and R terms. Some
players are getting out of generation altogether, creating more opportu-
nity for others staying in. Competitive forces are forcing serious players
to wake up. For operating staffs, theres a better chance that manage-
ment will fully engage them to develop methods that deliver perform-
ance. They will be supported with tools and training. For the tradi-
tional mom and pop utility, the writing is on the wall. Their family
approaches cant be competitive with best performers. In the absence
of a regulated economic protection, they too will pass away.
To understand maintenance, you must understand operations, engi-
neering, and statistics. Outage management, computer scheduling, and
data management are also needed. But finally, in this world of change,
a profound knowledge of equipment design, its functionality and weak-
nesses are necessary to support safe, cost-effective decisions. The
excitement is that a relatively new maintenance theory is still available
to reshape and simplify fundamental plant operations thinking. And it
can still transform your way of viewing and performing maintenance!
350
appendices final 351-352.qxd 3/3/00 2:59 PM Page 351
Appendices
1. Glossary of Terms
2. Further Readings
3. RCM Software Applications
4. References
351
appendices final 351-352.qxd 3/3/00 2:59 PM Page 352
glossary 353-390.qxd 3/3/00 3:00 PM Page 353
Glossary
80/20 Rule
A rule attributed to the Italian statistician Pareto based upon his
study of 18th century economic wealth distribution in Italy. For our
purposes, it attributes 80% of the problems to 20% of the equipment.
Generally the Pareto rule can be found in many skewed distributions.
Abandoned-in-place
Equipment removed from service and left in its location in a plant
because the cost of removal exceeds the scrap value. Equipment that
adds marginal value to a process compared to cost, may be left un-
maintained in place with no cost or production impact.
Acceptance criteria
Specific limits for acceptance. A term with general meaning, but
nuclear origin. Also used for operationalized failure criteria, generally
describing a test. Time-based PM will sometimes have acceptance cri-
teria. Large turbine bearings will be reused as is, i.e., unless wear
exceeds so many thousandths.
Actuarial
Mathematical statistical failure analysis using practices accepted by the
insurance industry and Society of Actuaries (SOA) or other profession-
al groups (as appropriate) to measure aging, risk, and mortality (based
on study for human populations).
A-E
Architect-engineer. The facility designer usually hired as a contractor to
provide a facility design. Occasionally also the constructor (design-
construct)
After-market
The parts and services supply market for equipment other than through
the OEM. After-market parts can be superior to OEM parts, but theres
much greater risk in the after market for parts reliability and quality.
This risk is generally borne by the buyer.
353
Age
Time correlation factor. The accumulated time since the equipment was
placed in service. In general, time equivalent for the aging process in
question. This could be tonnage for wearing parts, for example.
Age exploration
A systematic process of using conditional overhauls and opportunity
samples, with formal cost analysis, to evaluate and improve designs.
Age parameter
The measure of aging, which correlates with resistance to failure for
parts that age (dependent on the part, application, and use). Used as
the general time basis for maintenance.
Aging
A process that can reduce a part or components resistance to failure
over time. To grow old or show signs of growing old. Synonyms: dete-
riorate, fatigue, waste, tire, exhaust, flag, droop, ply-out, drained, spent,
depleted, obsolete, erode, consume, fret, rub, fray, erode, weather, cor-
rode, oxidized, rust, disintegrate, spoil, decay, decompose, break.
ALARA
As low as reasonably achievable: a program of minimizing radiation
exposure required for all nuclear license holders. The basis is to avoid
unnecessary radiation exposure and its long-term damaging somatic
and genetic effects.
Align
Block: put multiple PM tasks together for performance at one time
when an equipment train, unit, or even plant is available. Aligning
intrusive PMs into a unit outage window is an obvious example. For
scheduled work, once work is aligned, it facilitates the performance of
scheduled maintenance.
ANI
American Nuclear Insurers: a consortium of insurers providing nuclear
insurance backed by law.
ANSI
American National Standards Institute: a standards organization that cer-
tifies and maintains standards. These include many industrial and power
354
Glossary
generation standards. Common ones for nuclear plants are ANSI N45.2
and N18.7, for procedure use and maintenance management.
Applicability
In traditional RCM use, the requirement that assures a PM activity is
technically and statistically effectivei.e., it actually prevents failures.
Applicable
Prevents failure.
ARCM
Applied RCM: one abbreviated version of RCM that retains the funda-
mental salient elements of RCM (as described in the document by
Nolan and Heap). Includes maintenance strategies TBM, OCM, CDM,
OCMFF, and NSM, but simplifies analysis.
Area inspection
A general walk-around inspection that checks for random or other fail-
ures. For an airplane or car, pre-service visual check for obvious
problems.
ASCE
American Society of Civil Engineers.
ASME
American Society of Mechanical Engineers.
ASQC
American Society for Quality Control.
ASTM
American Society for Testing and Materials
Availability
Defined exactly by NERC. The period of time a unit is available to be
dispatched for generation, whether it is or not. Expressed as a fraction
of calendar time.
Base load
Loading a unit to full rating between scheduled down periods. Typical for
nuclear and low-cost generators. The opposite load term is peak load.
355
Basic interval
Prime intervalthe most fundamental interval when aligning a PM
task that fits the frequency, and provides reasonable multiples for the
overall maintenance program. A car requiring PMs at 12, 24, 30, 48 and
60 months has a basic interval of 12. Often the interval that also carries
a fundamental condition-directed maintenance program inspection
task. Missing a basic interval PM in a completed program carries a seri-
ous consequence.
Basis
The justificationthe reason why. Usually at least partially implicit.
B/C
Benefit/cost: benefit-to-cost ratio, commonly represented backwards as
cost/benefit ratio. For instance: replacing the oil has a cost/benefit ratio
of 10/1. It is really the other way but most people understand and state
it this way.
Benchmark
Comparing costs for similar processes, facilities, or equipment. Often
performed within an industry to validate competitive position, and out-
side to identify world-class performers.
Bit map
An image in a computerized format literally imported as a map of image
bits.
Block
Aggregate, group, or align for performance.
Blower
A low head fan that supplies gas (usually air).
Blue blush
Deep blue-hued carbon deposits on high-pressure, high-temperature
steam valves. Through time, these build up to where the valve may bind
during stroke. These require periodic removal.
Bootstrap
In computer jargon, a preliminary short software routine that allows
loading the main operating system. A system that gets the machine up
to minimum smarts to run.
356
Glossary
Breakdown
Fail suddenly, with little or no warning. A failure that impacts opera-
tions and production schedules.
Breaker
Circuit breaker
Burn-up
Slang: in nuclear work, reaching the radiation exposure administrative
limit. It pulls a person off the available worker list; they are referred to
as burned up based on reaching their maximum weekly or monthly
radiation exposure limit.
Bus
Electrical bus
Bus bar
The output bus that connects the generator (stepped-up) output to the
transmission grid. Often used in the context of bus bar costthe
cost to generate at the grid connection.
B&W
Babcock and Wilcox: a large industry supplier of boiler and nuclear
steam supply systems.
BWR
Boiling water reactor
Calibrate
To adjust an instrument for zero and span due to drift. A basic PM
activity that is often time-based.
Call out
Calling out a person for work after normal work hours end, usually in
response to a plant need.
CBM
Condition-based maintenance: the same as condition-directed mainte-
nance. Sometimes used to maintain a distinction between condition-
monitoring program derived maintenance tasks and formal on-condi-
tion-derived condition-directed maintenance.
357
CDM
Condition-directed maintenance: maintenance directed by condition.
Obligatory maintenance based upon defined failure limits exceeded.
CDM is the fundamental differentiator of a firmly RCM-based PM pro-
gram. To use effectively, it must be reserved for those on-condition
tasks with explicitly defined failure limits.
CDM (FF)
Failure finding condition-directed maintenance. A special type of con-
dition directed maintenance in which the acceptance criteria constitutes
satisfactory test performance. For instance, a diesel generator in stand-
by mode could be required to demonstrate that it can meet its design
specifications by starting and loading to 650 KWe within 10 seconds as
the test criteria.
CE
Combustion Engineering: a supplier of power generating plants and
equipment. Now ABB CE.
CFR
Code of federal regulations.
Chargeable loss
A loss that can be charged to a specific cause. Used in particular for
forced outage measurements. A restriction due to water chemistry dur-
ing startup is a chargeable loss to water chemistry.
CIC
Component identification code: a unique equipment identification
code.
Clearance
Tag out: a method to control equipment for work. A tag out or clear-
ance is required to isolate equipment from energy sources for personnel
safety.
CNM
Condition monitoring: operations-implemented general equipment
monitoring for failure. Tasks for which no exact on-condition limit
can be determined. When theres agreement that a benefit exists these
358
Glossary
tasks are implemented as non-specific condition monitoring, i.e., CNM

tasks for a coal belt walk-around check includes lighting, leaks, oil and
water, and coal accumulation. All have distinct benefits, but its hard to
create an on-condition task for each.
Caution: Some writers use the term synonymously with on-condition

used in this book.
CMMS
Computerized maintenance management system (from 1990s onward).
Called maintenance information system (MIS) in the 1980s. A comput-
erized WO initiation, tracking, planning, scheduling, approval, and
archive system. A sophisticated computer software system thats at the
core of maintenance management in complex operating environments.
CO
Conditional overhaul: an overhaul that corrects the proximate cause of
failure, secondary failures, and restores equipment to performance speci-
fication. CO does not completely disassemble nor replace all replaceable
parts, although it does replace any aging components prior to the next
scheduled overhaul period. CO zero-times the equipment. (See also
Control Operator).
Cogen
Cogeneration: a type of generation authorized by law to allow non-util-
ity participation in the generating market. Being phased out in favor of
independent power producers and separation of generating and trans-
mission and distribution assets.
Common mode failure

A failure mode that compromises redundant train, equipment, or com-
ponent independence. Because of this, it changes assumptions of inde-
pendent failures in FTA. Common mode failures significantly change
the overall probability of accident events and are of great concern to
regulators. Practically, maintenance practice has the potential to intro-
duce common mode failures systematically, so performers need to be
sensitive to this. For instance, one person using inappropriate grease
lubrication could inadvertently set up a common mode failure on all the
equipment greased incorrectly with that lubricant. This is what hap-
pened with motor operated valve operators and the use of incompatible
grease in the 1980s at nuclear plants.
359
Complex item
An item that fails to exhibit dominant failure modes and thus fails ran-
domly in service with no known age. Practically, an item that fails to
show aging.
Conditional probability of failure

The probability of failure in the next operating time interval, given sur-
vival into the current one. Also, one of several mathematical failure
probability curves.
Conditional Overhaul
An overhaul that conditionally addresses on the observed discrepencies,
returning the cycle-dependent time measure to zero. An overhaul that
addresses all proximate failure causes, as well as any aging to restore the
units post-overhaul aging parameter to zero. A maintenance activity
that corrects and zero-times a piece of equipment.
Condition-based
Condition-directed, including the non-specific results of condition
monitoring, which are separate from condition-directed in that no com-
monly agreed failure resistance has necessarily been exceeded.
Conservatism
The tendency to prefer an existing situation to change; safe. In engi-
neering, the provision of design margins to accomodate uncertainty.
Control Operator
The operator who manages the control room boards or DCS CRT.
Practically, the operator who is running the plant.
Constructor
The facility builder. Occasionally the same as the A-E.
Corrective maintenance
In former days, work on demand to correct failed equipment. A term
gradually falling out of favor due to its limitations and bias. Most cor-
rective maintenance is a combination of condition-directed, condition-
based no-scheduled maintenance but has never been differentiated. Old
MIS systems categorized work two wayspreventive, and corrective.
360
Glossary
Cost
Maintenance cost: combination of hourly cost, material cost, services
cost, and overhead. Typically five to six times the hourly cost. Excludes
opportunity cost of lost generation.
Cost effective
Worthwhile based upon cost/benefit perspective in a general sense.
Since PM implicitly considers the time difference from performance
cost incurred to benefit received, PM cost effectiveness implies using
the time value of money.
Critical
Immediate and direct safety consequences, usually unacceptable.
Critical is used in common language to identify any important failure.
Readers of the literature and particularly in RCM should skim to dis-
cern the authors context for the use of critical.
Critical failure
A failure with an immediate and direct safety consequence. A failure
whose risk is unacceptable based upon accepted safety standards.
Usually an immediate and direct threat to personnel, public, or (in rare
cases) the environment.
Critical few
The statistical few that predominantly drives the totals presented in
Poreto format by category. This comes from statistical analysis of data
in TQM and Process Improvement Technology. As an example: one
finds that the combination of bearing failures and insulation resistance
failure account for most motor failuresstatistically.
Critical instruments
Instruments that identify critical failures, such as excessive vibration for
a large rotating machine.
Criticality
Criticality analysis is analysis done in some streamlined RCM approach-
es to identify how important a failure mode really is. It is subjective,
based on interview and conjecture, and therefore of limited use for
assessment and management of experiential risk.
361
CT
Combustion turbine
DCS
Distributed control system: a plant-wide control system common in
many non-nuclear applications. The state of the art in control system
technology at this writing.
Design
Involving the specification, selection, and layout of equipment and
materials, in contrast with maintenance. Maintenance and design often
overlap in indistinct ways. Maintenance in fossil environments often
performs light design rolessometimes without being aware of the
design role.
Direct cost
Cost at the point of applicationin contrast with indirect cost or over-
head. Direct labor, material, and services costs show up as plant
expenses in traditional utilities. Indirect costs (in my experience)
dontalthough they ultimately effect the bus bar generation cost to
the consumer.
Discard
A type of PM task where a part is replaced based on time.
Dissimilar metal welds

Welds in boiler tubing used where the tube makes a transition from steel
to stainless or Inconel, and vice versa. Typically, the design objective is
to use expensive alloy only where necessaryin the high-temperature
area. This introduces a DSM transitionwhich is a weak point.
DOE
Department of Energy
Dominant failure mode

A predominant failure mode. Often used in reference to complex
equipment and equipment with multiple failure modes. Dominant fail-
ure modes are based on statistical experience. Their utility is that they
help focus failure management effort.
362
Glossary
DOT
Department of Transportation
Economic dispatch
Dispatch of the next unit of available generation based upon marginal
cost as the system load increases. Economic dispatch calculates the cost
of the next available MWe of generation and dispatches that unit that
provides it. Most PUCs require public utilities operating distribution
systems to follow such an economic dispatch model. The automatic
generation control (AGC) will identify which units are to be loaded
next (or removed first) as the load varies over the course of the shift and
day. In reverse, as the load drops, the most expensive generation is shut
down first.
EEI
Edison Electric Institute: a generating utility trade group.
Effective
Used casually in two contexts: technically effective and cost-effective.
RCM reserves the term to address cost-effectiveness and uses applica-
ble for technical effectiveness.
EFOR
Equivalent forced outage rate: forced outage rate adjusted to reflect the
equivalent effect of forced load restrictions. Defined exactly by NERC.
Empirical
Based on trial and errorexperiential.
Engineering cause
The local cause responsible for failure. A bearing experiencing a wipe
with no lubrication has fretting, spalling, or wipe as the engineering
cause. A plugged filter, mixed-up round, or broken supply line is not
the engineering causeeven though any one of them could have been
a root cause. Engineers often refer to engineering failure cause as root
causes. I call them proximate causes. The confusion is due to the mul-
titude of different definitions in various standards.
EO
Equipment operator: one level up from a tender. The roving operator who
starts and stops most heavy equipment requiring local monitoring.
Typically a senior experienced operator with 10 or more years of experience.
363
EPA
Environmental Protection Agency
EPRI
(The) Electric Research Power Institute: a voluntary, industry spon-
sored research organization that performs most research for the gener-
ating industry.
EQ
Environmentally qualified: a special class of (essential) nuclear equip-
ment required to perform shutdown and monitoring activity following
a hypothetical design basis accident. Typically, equipment with organic
and elastomeric materials that is susceptible to temperature aging.
These require special scheduled maintenance programs by law.
Event tree
A logic tree that shows the pathways from a primary event upward to a
final outcome. Used to identify contributors to overall outcome risk.
Evident
Evident failure: a failure in which the failed item should be evident to a
qualified operator performing their normal duties. Contrast with hid-
den failure: one that no one would be aware of except by monitoring or
performing special checks and tests.
FAA
Federal Aviation Administration
Fail safe
Fails in the safe direction or position, e.g., a fail-safe air-operated con-
trol valve would fail shut if that were the safe direction. This is how
the feedwater-regulating valve in some fossil plants fails since it avoids
flooding the steam drum. Fail-safe valve positions include FAIfails as
is, FCfails closed, and FOfails open, for example.
Failure Mode
Repetitive manner of failure intrinsic to design and application.
Failure finding
Testing to identify a hidden failure. Startup test of a redundant train,
364
Glossary
channel checks, and alarm checks are examples of standard (on-condi-

tion) failure-finding tasks.
Failure
(Broad) An unsatisfactory condition that fails to meet expectation.

(Specific) A condition that fails to meet a performance standard.
Failure maintenance
Failure-based maintenance: maintenance based on a failed condition. A
type of corrective maintenance.
Failure mechanism
Failure mode and a cause.
Failure substitution
Substitution of a major failure with a minor one. Redesign to lower the
consequences of failure.
Fault tree
Logic tree of outcomes traced to primary events through system logic
modeling that enables the calculation of failure probabilities for certain
events.
Feeder
Equipment that feeds a continuous, controllable stream of product into
a process. A coal feeder, for example.
FERC
Federal Energy Regulatory Commission: a federal commission charged
with the regulation of interstate power sales.
Fishbone
Fishbone diagram: Ishikawa diagram.
Flash
For computer applications, memory set permanently in EPROM. Also
short for flashover.
365
Flashover
Electric plasma arc due to failed equipment or protective device. A
high-energy arc that can cause injury or fire. Commonly occurs in
switchyards, on switchgear, or in large motors and controllers, which
use medium voltage (2000-8000 V) electric equipment. Requires high
voltage due to the electric resistance of air. Once initiated, however,
requires another circuit interruption to terminate.
FMEA
Failure modes and effects analysis: systematic review of the ways that
equipment is expected to fail, and consequences of the failure. A qual-
itative enumeration of failure modes. Developed as a discipline in the
early 1960s as a technique to improve reliability. Made into a Mil spec
standard, and required as a part of some defense design proposals
FMECA
Failure modes and effects criticality analysis: FMEA with criticality cal-
culation based on standard published component reliability figures.
FMECA extends FMEA to a numerical basis. This in turn allows tech-
niques such as reliability allocation to be used as a design tool.
FOR
Forced outage rate: forced outage hours(forced outage and service
hours) expressed as a percentage. A NERC reliability measure.
Fossil
Fossil-fired boiler or combustion turbine fueled with fossil fuel.
Function
Output(s) provided by a system.
Functional failure
Loss of one or more system outputs. Loss of a system purpose.
GADS
Generation availability data system: a statistical reporting system oper-
ated by NERC that summarizes broad categories of electric generating
plant performance.
Gaitronix
Plant public address and personal communication system.(A trade name.)
366
Glossary
GE
General Electric. Supplier of power plants, turbines, and electrical
equipment.
Graybeard
Experienced personnel who know the ropes. Usually with 20 or more
years of experience.
Gun Deck
Perform superficially but completing the paperwork to perfection.
GUI
Graphical user interface: a screen interface that uses a mouse to isolate
and execute commands
Gyrol
(slang) A soft-start single speed transmission between a motor and a
load. Used on large coal belts. Named for a manufacturer.
Hard time
Time-basednot condition-based. Sometimes used for emphasis to
indicate activity that could be worked as OCM, but is left at hard time
due to program maturity, or intent.
Hard wired
Unchangeable, unmodifiable. (Slang) Impossible to mess up. Contrast
with jumpers.
Heat rate
Heat required to generate a MWhr of load. Literally, the efficiency of
the plant to convert fuel to generation. The inverse of efficiency.
Typically fossil range is 6000-12000 BTU/MWhr
HEU
Hydraulic equipment units: a hydraulic equipment skid.
Hidden
Not evident to operators performing their normal routine duties.
367
Hidden failure
Opposite of evident. Describes a failure not normally evident to the
operator without special instrument and test.
House
Main plant.
HRSG
Heat recovery steam generator: steam generator used for combustion
turbine heat recovery.
HTGR
High temperature gas reactor: a nuclear reactor cooled by helium. The
retired Peach Bottom 1 and Fort St. Vrain plants were these types of
plants.
Hydro
Hydroelectric
I&C
Instrument & control
IEEE
Institute of Electrical and Electronic Engineers
Ignitor
Device used to ignite a fuel stream in a boiler, usually oil or pulverized
coal. An electric ignition source (spark plug) and fuelusually gas or
oil that assures a flame for combustion.
Important
Worthy for consideration of scheduled maintenance. A classification
used by EPRI for streamlined RCM.
In service
A device during its useful life. Typically used to describe equipment,
train, or plant that is operating, or capable of being operated.
Infant mortality
Failures shortly after entry into service due to quality, defect, and other
latent causes that drop as service life increases. For electronics, this used
to be the basis for burn-in of equipment.
368
Glossary
Inherent capability
The equipments intrinsic service capability related to design.
Inherent reliability
The reliability level supported by the intrinsic design. An upper limit
to the performance reliability of a plant or equipment.
In-op
Inoperable
Inoperable
(Nuclear terminology) Not capable of performing design-basis func-
tions. A system can be operating, yet be inoperable because certain acci-
dent or other scenario design assumptions cant be met. Based upon
technical specifications, inoperable status can force a plant to shut down
until operability is reviewed and assured. (Secondary) Unevaluated for
operability; standing by until a qualified person (usually an engineer)
evaluates and declares the equipment capable of performing its design
function. A real or virtual equipment status category.
INPO
Institute of Nuclear Plant Operations: a self-regulated nuclear industry
oversight body originated by the ANI after TMI. A quasi-regulatory
nuclear body performing many self-oversight functions.
Instruments
Devices that make a transducer conversion of condition to human-read-
able format. Commonly visual, but occasionally acoustic and other for-
mats.
Interlock
A device that prevents one device from operating until other require-
ments are met. These are usually based on personnel or, machine pro-
tection, or both. Sometimes the protection of the general public is a fac-
tor. For example, mid-1970s vintage cars had seat belt ignition inter-
locks. The engine wouldnt crank until the seat belt was fastened.
These were eventually eliminated based upon public outcry.
Interlocks
Devices that prevent undesirable equipment operations. For instance,
369
automobile hood releases are interlocked with a manual release latch

requiring the vehicle be stopped to open the hood.
IPP
Independent power producer. A producer outside the generating and
transmission companys jurisdiction but who has the right to sell power
to the generating companys grid on an economic dispatch model.
ISO 9000
A European common market standard requiring process and mapping
certification. A certification standard that assures basic process con-
trols are in place for production manufacturing.
Jumpers
Temporary power or control cables that defeat the purpose of a control
device, including interlock. The purpose of a jumper is to temporarily
defeat a control or interlock to facilitate maintenance, or bypass a fail-
ure.
KISS
Keep it simple stupid: a military term.
LAN
Local area network: a shared computer or microprocessor network.
LCM
Life-cycle maintenance
LCO
Limiting condition for operation: the technical specification limit that
requires shutdown or entry into a grace period when exceeded. A grace
period, when expired, must be followed by appropriate action-often
shutdown, if the condition cant (or hasnt) been corrected. A risk man-
agement tool for nuclear power plants.
Learning
A process that reduces times and cost to perform an activity. Used in
manufacturing to represent the general improvement in design and cost
as products enter and proceed through a production life cycle.
370
Glossary
Legacy systems
Existing, company-developed systems.
Life-cycle
The progression of a product from introduction through production
and into obsolescence. It occurs over many years.
Life-cycle maintenance
A maintenance plan with the overall product life-cycle strategy in mind.
Life-cycle cost
Total cost throughout the product life cycle including disposal costs.
Often the initial cost is the driving factor in a purchase decision. Like
the owner of a new European sports car, the owner may find that the
total operating costs far outweigh the initial cost.
Like-for-like
Like-for-like replacement: exact replacement. In contrast with replace-
ment by an improved or superior part. A replacement that minimally
maintains performance.
Living
Ongoing, changing, and evolving.
Loaded cost
Costs including overhead charges that may not be applied at the plant
level. Overhead charges pay for staff and corporate services not ordi-
narily charged directly at the plant level.
LTA
Logic tree analysis: an RCM decision analysis for classifying failures by
type. This in turn influences the maintenance strategy selected. One of
the more confusing aspects of traditional RCM for new users.
Lubrication
A process of replenishment of aging lubricants that are a part of the
equipment.
LWR
Light water reactor: the types of commercial nuclear plants licensed in
the United States. They in contrast with a heavy water reactor, such as
the Canadian Candu reactors.
371
Maintain
To preserve in an orderly state.
Maintainability
The capacity to maintain equipment. The design consideration of main-
tenance to provide access, turnaround, tools, and other support
requirements to facilitate maintenance.
Maintenance
(as defined by 10CFR50.65) The aggregate of those functions required
to preserve or restore safety, reliability, and availability of plant struc-
tures, systems, and components.
Maintenance rule
(10CFR50.65) An NRC (federal) rule that requires nuclear plants to
perform maintenance monitoring for in-scope structures, systems, and
components, and their safety functions and take corrective action
appropriately. Informally called the maintenance rule.
Maintenance strategy
A plan for the maintenance of a component or equipment in an RCM
format using a combination of CNM, OCM, TBM, OCMFF, and the
resulting CDM. NSM is the null strategy.
Markov analysis
A type of conditional probability analysis used for the prediction of suc-
cessful events with preconditions. One use is the likelihood of starting
emergency diesel generators after faults.
Mechanism
See: Failure mechanism.
Metal clad
Metallic cladding applied to some nuclear fuel types for protection. For
many fuels, the primary fission product barrier that restrains fissionable
gases from release to the environment.
MIL spec
Military specification. A military standard derived from U.S. DOD
equipment procurement specifications. MIL-STD-2173 (AS), i.e.,
addresses provision for FMECA reliability analysis for procurement.
372
Glossary
Mill
Coal mill. A machine that pulverizes coal to a fine combustible dust,
mixing it with air in the process.
MIS
Maintenance information system (See: CMMS).
Mixed waste
Waste that includes both radioactive material under the jurisdiction of
the NRC and hazardous material under the jurisdiction of the EPA.
Mod
Design modification: a change to a plants fundamental design.
Sometimes as simple as a part upgrade, or as complex as replacing a
precipitator with a bag house. Most fall somewhere between these
extremes, and many times wont be recognized as a design change (in
non-nuclear applications).
Morpheline
A volatile organic chemical used for treating feedwater where inorgan-
ic chemicals arent acceptable.
MORT
Management oversight risk tree
MPFF
Maintenance preventable functional failure
mREM
One thousandth of a REM: the common practical measure of exposure.
Typical radiation jobs incur several mREM of exposure. Big jobstens
of mREM. Bigger jobs correspondingly more. Typical limits are 40
mREM/week.
MSG-3
Maintenance steering group standard-3: commercial aviation mainte-
nance standard for RCM-based maintenance programs maintained by
the Airline Transport Association (ATA).
MTBF
Mean time between failure: the average period between failures for a
failure mode.
373
MTTR
Mean time to repair: the average time to restore a failed component to
service for a given failure mode.
Multiple failure
More than one concurrent failure. In contrast with single failure.
MW
Megawatt: 1,000 Kilowatts. A city of one million people has an electric
demand of about 1,000 megawatts, or 1,000 watts per capita. This is the
output of a relatively large two-unit generating station of 500 MW each.
This is a common standard.
NDE
Non-destructive examination: evaluation of a material condition such as
welds without destructive examination. Uses methods such as radi-
ograph, ultrasonic inspection, and replication to assess condition.
Near miss
An event which breaches several levels of protectionusually leaving
one remaining fault barrier.
NEC
National Electric Code
NERC
North American Electric Reliability Council. A voluntary organization,
that sets standards and rules for interconnected transmission system
generating station requirements to assure reliability of the transmission
system. The country is divided into 10 contiguous interconnected
regions. NERC regional committee members take responsibility for
meeting requirements that assure transmission system reliability, such as
establishing and maintaining rotating reserve and standby reserve
requirements. These are generating units immediately available or
available on short notice to come online to meet contingencies. NERC
also measures the overall reliability of member units. Whenever a unit
is brought online or removed from service, the plant records the nature
of the status change and cause.
374
Glossary
NFPA
National Fire Protection Association
NRC
Nuclear Regulatory Commission
NSM
No scheduled maintenance. A plan of using condition monitoring to
wait for a maintenance requirement to become evident. Legitimacy is
based on theoretical and actuarial studies that form the basis for relia-
bility centered maintenance.
O&M
Operations and maintenance
OCM
On-condition maintenance. The first check/inspect part of an on-con-
dition/condition-directed maintenance pair. A combination of task,
limit, and performance interval.
OEM
Original equipment manufacturer: the original supplier. Contrasts with
the secondary market or after-market supplier.
Old hat
Graybeard. A very experienced person.
On-condition
A scheduled maintenance activity with a specific monitoring method
and failure resistance limit identified that experts have agreed detects
resistance to failure. Upon exceeding this limit, resistance to failure has
declined so that failure will occur. Equipment is removed from service
for maintenance, at this point. It can be as simple as measuring the
thickness of remaining tread on a tire or as complex as modal analysis
for vibrations. The key concepts are resistance to failure and explicit,
repeatable limits that trigger condition-directed maintenance.
On-condition/condition-directed pair
A two-part maintenance activity that is unique to RCM.
375
OOS
Out of service.
Operate
To run, maintain, and dispatch production from a facility. To exercise
disgression in the asset to generate income.
Operationalize
Make useable in an operating environment. For example, a standard
must always be operationalized. This process develops the infrastruc-
ture and insures that the utility can make an activity work in a produc-
tion environment.
Operator
The owner-operator entity. More commonly, a person charged to mon-
itor, configure, and report plant conditions, working on shift.
OSHA
Occupational Safety and Health Administration
OTF
Operate to failure. A characteristic of RCM over-emphasized in after-
market books. No scheduled maintenance or maintenance required
on an interval exceeding the assets useful life is a more appropriate
term in the authors opinion. See also NSM.
Outage
A scheduled production down period to facilitate maintenance. Outage
maintenance is comprised of on-condition, time-based, and condition-
based maintenance. For nuclear units, this also provides the window
to refuel the reactor for American BWRs and PWRs.
Overhaul
To rebuild by teardown, reassembling with new consumable parts, and
reworking all parts and components to a like new condition.
Terminology applied to any large complex piece of equipment, from
diesel engine to turbine disassemble/reassemble work.
Pareto
Vinceto Pareto, Italian mathematician and statistician. Pareto demon-
strated the statistical presentation of data in block chart format by fre-
376
Glossary
quency and expressed an early version of the now trite 80/20 rule that
summarizes skewness often present in statistical data.
Pareto chart
Data presentation in block chart format ordered by frequencymost
frequent to least.
PC
Primary containment
PCRV
Pre-stressed concrete reactor vessel
PdM
Predictive maintenance
Peak load
Load added only as demand requires. Since demand varies by hour,
day, week, and season, some units will be started and stopped with
demand according to economic dispatch rules. These units are loose-
ly termed peakers and largely comprise hydro, combustion turbine,
gas-fired, and a few coal-fired boilers. Nuclear units are not peakers.
Peaker
A plant used to supply system peak load periods. Gas turbines, gas-
fired boilers (sometimes re-powered coal-fired units), pumped storage,
and diesel may comprise units in peaking service. Usually not econom-
ically dispatched until all base load is available due to high fuel cost.
Permissive
Logic permissive: a control scheme that must be completed for an
action to be permitted.
P&ID
Process and instrumentation drawings: design drawings supplied with
the plant (along with vendor O&M manuals) by the A-E to aid in per-
formance plant maintenance and modification.
Pillow block
A bearing housing shaped literally like a pillow. Often installed as a sep-
arate assembly for large equipment like coal belts.
377
Planned maintenance
Prepared maintenance plans for equipment that requires repetitive
maintenance. May include standard clearance points, parts, tools , and
resources such as labor and contractors. Planned maintenance is made
up of scheduled, on-condition, condition-directed, and some condition-
based maintenance.
PM
Preventive maintenance: planned scheduled maintenance activity. Also,
slang for PM work orders. Also, a scheduled maintenance program.
Sometimes used to refer to the discretionary part of the scheduled main-
tenance program in a regulatory enviroment.
PMO
Preventive maintenance optimization: a maintenance optimization
process that streamlines RCM. Primary advantage is simplification of
paperwork and coding based on the LTA of TRCM.
Poke yoke
Make user-friendly and simple: a Japanese term summarizing a tech-
nique that stresses task simplification to remove or diminish the possi-
bility of error. Poke yoke devices are devices which serve the same pur-
pose.
Population
Statistical population
Pot
Potentiometer: a variable resistor often installed in instrument loops to
facilitate calibration.
Potential failure
A failure that is imminent based upon exceedance of a failure resistance
standard. Examples: High-pressure boiler tube wall thickness less than
0.025 inches, machine vibration amplitude in excess of that machines
specified limit (say 5 mils at 1800 RPM).
PRB
Powder River Basin (coal): a distinctive Western low-sulfur coal char-
acterized by low heat rate, high volatility, and dust. Widely used based
on low sulfur content and price.
378
Glossary
Precursor
Precursor event. An event that predicts susceptibility for (precurses)
a more severe event. An event that predicts future events of the same
nature, but more severe in consequence. A leading risk indicator.
Predictive maintenance
Maintenance to diagnose conditions and predict future maintenance
requirements. Gradually being supplanted by condition-directed and
on-condition terminology.
Premature failure
Failure prior to planned end-of-life.
Premature removal
Removal from service ahead of schedule due to unsatisfactory service,
or selection as part of an age-exploration sample for new equipment.
Present value
Present value of money: total cost adjusted by discount factor and time.
Preventive maintenance
Traditional term for scheduled maintenance (10CFR50.65). Predictive,
periodic, and planned maintenance actions taken prior to failure to
maintain SSC within design operating conditions by controlling degra-
dation or failure.
Preventive maintenance program

PM program: a process that maintains equipment in a state of readiness
to support production requirements. All the supporting elements
required to support the PM process, including work identification, part
support, training, scheduling, and work planning.
Primary failure
The immediate failure. The first failure. A tire blowout causing an acci-
dent is the primary failure. See also secondary failure.
Process
A defined way in which something is transformed. A process must have
specified inputs/outputs and processing technology.
379
Profound knowledge
Intrinsic, hard-to-replicate knowledge of a process. Almost always pro-
prietary, whether by formal intent or functional cost to extract. Often
the basis for competitive advantage. Term coined by W. E. Deming.
Proximate cause
The immediate, local cause. Not necessarily a root cause (based on
RCFA), but known in engineering circles as root cause. The apparent
cause of failure. The cause evident at the failure location.
PUC
Public Utility Commission: the government entity that regulates the tra-
ditional utility environment in the public interest. Also known as PRC,
RC, and other acronyms.
Puff
Low-pressure explosion, large enough to damage large low-pressure
boiler walls, ducts, and mills. Overpressure on the order of several
inches of water. Because of the limited extent of over pressurization,
commonly called a puff. to designate minor nature
Pulverizer
Coal mill
PWR
Pressurized water reactor
RAM
Reliability availability and maintainability. A type of analysis that looks
at the total reliability of a system and factors in maintenance turnaround
time. Developed for aerospace and other high-cost applications such as
weapons programs to establish theoretical baselines for performance
expectations.
RCFA
Root cause failure analysis. Analysis to uncover root causes of prob-
lems. There are approximately 10 different root cause techniques.
They can be further delineated into stochastic and statistical groups.
Its important to understand the RCFA context. Nuclear units, for
example, dont use statistical RCFA.
380
Glossary
Redundant
(Websters) More than enough. Excess. Important safety, operational,
instrumentation, and other features are provided with redundancy in
engineering designs to assure their availability. Anyone whos ever
gazed into the cockpit of a commercial airliner has seen the four-fold
redundancy of critical instruments such as altimeter, direction, and roll.
Redundant equipment is provided in duplicated and triplicate precise-
ly because the function is critically required.
Regulate
Broadly speaking in a process control perspectivecontrol. For
transmission and distribution control, regulate describes the remote
operation of a generator to provide instantaneous load following. Some
power plants are reserved for base loadingprincipally nuclear units
and very large fossil units which are hard to start up and shut down, or
which dont follow load well. Typically, hydro, small fossil, and com-
bustion turbine peakers are used for regulation. They follow the load
through the course of the day, adjusting for instantaneous load changes.
Reliability
The ratio of successful missions to total trials. The degree to which an
operating unit meets the expectations of the operating entity (usually
owner) between scheduled down periods. The expectation that the
SSC will perform its function upon demand at any future instant
(10CFR50.65).
REM
Roentgen equivalent man: the primary measure of radiation dose expo-
sure used in the nuclear industry in the 1980s. Superseded by dose
equivalents in Sievelts (Sv), where 10 0 REM = 1 Sv.
Repair
Restore to specifications using welding or other processes. More than
rework/replace. Typically involves certified personnel and testing to
assure specification.
Replace
A rework task where a like for like part replacement is performed.
Commonly applied to filters, lubricating fluids, greases, and other con-
sumables that require repetitious service during equipment operation.
381
Technically, exact replacement. An upgrade to a superior partlike an

improved filter designinvolves a technical specification change. This
distinction is important in controlled aerospace and nuclear applica-
tions.
Restore
Return to exact specification. Most PM work is technically of a restora-
tion nature.
Rework
Unneccesarily perform work again. A job may require rework due to
infant mortality failure (relatively common in power applications),
unserviceable parts, failed performance tests, or (occasionally) lack of
appropriate documents and certifications (nuclear and aerospace
work). For fossil, bad welds are typical of a rework requirement.
Risk
What can happen (scenario), its likelihood (probability) and its level of
damage (consequences).
RO
Reverse osmosis: a common type of water makeup train purifier.
Root cause
The basic cause. In traditional RCFA, a cause that, once removed, pre-
vents recurrence (a stochastic perspective). This context is different
from a root cause on an Ishikawa diagram, which takes a probabilistic
perspective.
Round
A scheduled activity that checks a large part of a facility. Generally
comprised of a series of area checks with intermittent on-condition
checks. Also, a scheduled review of screens on DCS systems.
Round sheet
A round logsheet, updated in real time by round logging devices. A
sequence of readings currently being superceded by round logging devices.
382
Glossary
Run to failure
A misnomer, a term intended to summarize the no scheduled mainte-
nance aspect of many planned maintenance tasks. Misleading because
virtually none of these tasks result in functional failure. See also NSM.
Safe life limit

A part lifetime based on a known aging characteristic for a part with
direct safety impact. The part lifetime is limited based on manufactur-
er tests, conservative design factors, and standards or codes in the
absence of service aging experience.
Schedule
Enter an activity into a scheduling system.
Scheduled maintenance program

The series of scheduled planned maintenance activities. The program
that schedules the planned maintenance program Commonly, PM pro-
gram..
Secondary failure
Indirect failure. Failure that results from a primary failure. A boiler
tube leak caused from steam cutting from another tube leak is a sec-
ondary failure.
Service
Maintain.
Shifter
Shift supervisor, called operating engineer at some plants.
Significant
Equipment that has either a safety or economic impact, thereby war-
ranting review for potential PM benefits.
Simple item
An item characterized by very few failure modes. A relative term.
Review of the failure history of a simple item results in very few repeti-
tive failure modes recurring with great frequency. A filter and a journal
bearing are two example of simple items. The building block for com-
plex items. Contrast with complex item.
383
Single Failure
One concurrent failure. A simple failure to diagnose and correct, in
contrast with a multiple failure.
Six Sigma
A quality goal based on the reduction of failure frequency to less than
one in two million events.
Smoke
To destroy something. Occasionally from overload, continued use in
failure, or abuse. Smoking a motor, breaker, or starter are examples.
SNAFU
Situation normal all fouled up: a fiasco on a large scale. An organiza-
tional mix-up.
SP
Surveillance program: a planned functional test program made up
largely of on-condition and on-condition failure finding tasks used at
nuclear power plans to verify the functional capability of standby, pro-
tective, and alarm equipment
SPC
Statistical Process Control. A statistical study of processes that provides
measures of process capability and control. Widely used in manufac-
turing of high-quality products. Advocated for floor-up quality control.
SRCM
Streamlined RCM: an abbreviated version of RCM that simplifies RCM
using a two-path critical/non-critical approach. Developed by the
EPRI.
SSC
(10CFR50.65) Structures, systems, and components.
Standby system or train

One that is not operating and only performs its intended function when
initiated by automatic or manual demand signal.
Startup
Plant startup. A special period lasting from several minutes (for fast-
384
Glossary
start combustion turbines) to several days (for large baseload coal and
nuclear plants) that requires manual intervention, reconfiguration, and
direct support to place the plant in an operating phase.
Stroke
Operate, or test operation of, as stroke a valve.
Substitution
Replacement of an OEM part with an aftermarket one.
Super session
A super-session resembles MS Windows in which multiple applications
can be kept running so the user can jump between applications without
need to formally shutdown and restart applications. Early applications
could only run one per terminallike some DOS-based PCs even
today. This is a significant productivity tool.
SWOT analysis
Strength-weakness opportunity threat: a type of subjective risk analysis.
Synthetic
Synthetic oil: chemically constructed lubricants, in contrast with dis-
tilled column fractionated oil common in traditional lubricants. Such
lubricants generally possesss superior qualities, but at a price.
Synthetics cost four or more times more than specified traditional lubri-
cants.
System
A defined equipment group that performs a specific set of functions.
Usually, the A-Es plant documents provide a list of all plant systems,
their major equipment, functions, and expected operating conditions.
Very often all the related CIC lists and vendor O&M manuals are pro-
vided in binders (1970-1980s vintage units) that organize all informa-
tion about a plant. They are retained along with A-E design drawings
in document centers or shops for reference doing maintenance.
Tagno
Tag number; same as CIC.
385
Tag outs
An equipment control technique that facilitates work on equipment in
an operating plant. Also known as clearance. A controlled technique
to seperate energy from equipment to perform work.
Task
A single activity with a failure prevention aspect. The basic building
block of a PMWO. Usually, a PM consists of enough tasks to make
effective use of operator trip time to and from the work location. Many
simple tasks require 10 to 15 minutes to perform and are listed in ven-
dor manuals. Invariably, tasks must be organized into larger work activ-
ities or rounds to be done cost effectively.
TBM
Time-based maintenance. Roughly equivalent to hard time mainte-
nance with one slight distinction: The on-condition part of a two-
part on-condition/condition-directed maintenance pair can be consid-
ered as time-based. Its scheduled off the same software system as the
TBM task activities and looks virtually the same from a scheduler per-
spective.
Tech spec
Technical specifications. All equipment has technical specifications
used for reference in performance testing for deterioration. Nuclear
plants also have technical specifications that provide a basis for operat-
ing licenses. They must operate within these specifications or shut
down.
Tender
Job title for the lower seniority operators who roam the plants service
and outside areas, monitoring, servicing, and configuring equipment.
Time card
A charge for time, typically made against an activity or account. In some
CMMS systems, a time card and a work order are combined for mainte-
nance workers.
TMI
Three Mile Island. The Pennsylvania nuclear plant whose trip and shut-
down in March 1979 set the nuclear industry spinning from the adverse
386
Glossary
publicity and cost. The most serious commercial nuclear power plant
event in North America.
Total cost
Total life cycle cost. The total cost of operations, as distinct from oper-
ating and maintenance (O&M) cost, or startup or installation cost.
TPM
Total productive maintenance
TQM
Total quality management: a field of quality management that received
a great deal of promotion in the 1980s as traditional manufacturing
faced competitive pressure form overseas suppliers.
TRCM
Traditional RCM
Trip
Automatic or manual shut down of a piece of equipment. An operator
can manually trip a turbine or a breaker could trip on a ground fault
protection relay.
Tripper
Tripper belt. The coal unloading belt, the last of a series of belts that
moves coal from a railroad unloading point (often a rotary dumper) to
the housethe plant.
UAL
United Airlines
Unit
Generating unit. One increment of generating capacity at a plant
Useful life
Economically useful life. The period of time when an item can be
expected to operate with predictable cost and performance.
VAR
Volt amp reactive: in power flows, this portion of power provides volt-
age, but does no work. Its necessary to support the voltage in trans-
mission and distribution systems
387
Violation
A citation for violation of an article of law. Common jargon used in the
nuclear industry for citations under 10CFR50 and related parts of the
federal registrar. Very much subject to interpretation and established
precedent. Becoming common usage at fossil plants as EPA, OSHA,
DOT, and other agencies spread their wings.
VOM
Volt-ohm meter
VWO
Valves wide open: for a turbine, the maximum practical load that can
be placed on a machine.
Walk around
Area check: A tour looking for general failure evidence or environmen-
tal factors. Sometimes performed by management or non-routine per-
formers at plants.
Wear
To impair, consume or diminish by constant use, handling or friction; to
tire or exhaust.
Wearout
Fail gradually, with degrading performance allowing a long period to
evaluate alternatives and options. Similar to non-failures except that
performance specifications and expectations arent met. Gradual per-
formance deterioration until further service is no longer cost-effective.
Weibull
A specific mathematical distribution named after Lauritz Weibull, who
first used it extensively to model failure with age, infant mortality, and
randomness characteristics.
Weibull analysis
An analysis of failure data to fit the observed measurements to a Weibull
distribution. This can be done with specific Weibull analysis paper or
using software.
WO
Work order: a work authorization that conveys not only information
388
Glossary
about a job, but job accounting and performance information. Very

similar across all industries from auto repair to power plants.
WSCC
Western States Coordinating Council: the region of NERC covering the
Western states. One of eight NERC regions.
Xerox-style benchmarking
Process benchmarking. More complex than traditional benchmarking
because the process is also examined and compared.
Zerks
Grease fittings. After the manufacturer trade name.
Zero time
To reset the component aging clock to zero after an overhaul. To make
the item statistically indistinguishable from new based on mission
goals, performance, and failure criteria.
Zonal inspection
Inspection of an area or zone. Typically includes environmental con-
ditions, leakage, and other non-specific conditions that an experienced
person is expected to know. A pre-flight walk around aircraft check is
a zonal inspection.
389
further reading 391-436.qxd 3/3/00 3:01 PM Page 391
Further Readings
No Silver Bullets
RCM is not a silver bullet. Ultimately, improved performance comes
from better maintenance selection, timing, and performance. RCM helps
with selection timing and provides tools to raise awareness. Both timing
and performance benefit from heightened equipment awareness. Timing
improves first, maintenance performance, later.
As maintenance programs improve, two things become evident.
Crises decrease, but maintenance costs run higher. As crises decrease,
overtime, low productivity work, material parts expense, and service
expenses fall. After a yearlong enough to capture secondary cost fac-
torsproduction unit costs start to fall. More mega-wiggles are pro-
duced, so unit costs drop. This decrease in unit production cost due
to increased availability is a major benefit.
Long term effects are an increase in worker productivity and a
decrease in maintenance costs. These changes take approximately 2-5
years. This time period allows the benefits of reduced overtime, better
quality work, and improved productivity to build up. The requirement
for measurement to roll-up, as well as the time to allow fundamental
changes to influence machine performance causes the delay. Expect 1-
2 years to see an improvement in maintenance unit costs. For highly
reactive environments with high maintenance costssuch as those sole-
ly focused on in-service failuresimprovement can take place more
quicklyas short as six months! Improvement is seen as a decline in
unbudgeted maintenance expense and a decrease in both total hours
and cost by system. In order to see the decrease, you must first have sys-
tem-level performance measurement.
Aggressive RCM implementation can increase short-term costs.
Implementation makes the staff aware of the weaknesses in measure-
ment and administrative systems. A measurement infrastructure must
then be developed. The need for increased infrastructure begins to
grow in other areas. Requests for productivity tools and cost-effective
modifications rise. Training needs are recognized and their requests
391
increase. It takes time to recover these up-front investments. To see

RCM benefits accumulate, measurement capacity is required. Some
facilities are only just gaining the capability to conduct performance and
cost measurements on systems and equipment by drilling down. New
CMMS systems are creating these exciting measurement possibilities.
Tough Love Maintenance

Work Screening and Prioritization
Condition monitoring programs generate work orders when no
failure has occurred. This work request typically goes into a planning
backlog, where it will be reviewed, discussed, worked, and eventually
closed. Low-priority work arises because there is no obvious failure.
Very often this work resides in backlogs where it languishes until it is
canceled. These backlogs can be large and visible. Regulated nuclear
plants have agencies that base effectiveness on backlogged counts
(approximately 500 triggers escalated maintenance program action).
Few people are willing to cancel work requests, at any plant. Thus
a backlog becomes a Catch-22 barrier. Some environments backlog
WOs to work, where they become lost. Operators do not like can-
cellations. They perceive them as a slight unless there is personal follow-
up and contact. Cancellation is distinct and cancelers are distinctly
accountable. I reviewed the backlog of the service water system work
orders at a decommissioned, re-powered nuclear plant in preparation
for its restart as a fossil combined-cycle plant. In more than 1,000 of the
MWRs still active at the time of nuclear shutdown, fewer than 10%
could be classified as representing failures in the most general context.
In fact the mechanics had already identified these failures and had
completed (or were completing) them. Most failures were hypothet-
ical conjectures written up by anyone (and everyone) in the plants final
operational phase documenting what they felt were problems.
I was asked to help clean mess this up. With the regulatory veil lift-
ed, we quickly canceled most of them. Many were written by QA audi-
tors, outside inspectors, and contractors who had no basis for identify-
ing anything as a failure. At that time, this plant lacked an engineer-
ing group capable of WO screening. Screening was beyond the author-
392
Further Readings
ity of the scheduler/planners. Few had the gumption to do what was

needed during plant operationsdisapprove the groundless requests.
We did not know of a theory called RCM at that time and if we did, let
me assure you, we would have implemented it!
The lesson is garbage in, garbage resides. An important thing any
plant can do is to reduce this cholestric CMMS traffic. Setting up spe-
cific on-condition failure criteria helps. Any work order request that
fails to meet these criteria should go onto a clock suspense file to have
work specified or time out. If no resolution can be found within a
reasonable period, such as ten days, it should be canceled. Obviously
you cannot fix a problem that cannot be specified. Maintenance, as with
medicine, is patently expensive exploratory surgery. If, after you
request it, a WO does not exceed system or equipment performance
specifications, then it is a design change request. Maintenance is very
ineffective at performing design changes, yet many WOs strive to do
just that. The work initiation process should ask originators to specify
the exact nature of the failure. When it does, the justification monkey
is on their backs.
Plants that take this approach see a dramatic reduction in hypo-
thetical and non-failure WO tasks. This clears the air to focus on things
that have truly failed. The on-condition/condition-directed mainte-
nance pair is also a powerful focusing tool. For failed CDM tags
there should be little, if any, waiting. Prioritization is unnecessary. The
equipment basically needs to be fixed.
This is tough-love maintenance. It is hard at first, but a world of
benefit follows.
Missed PM
A nuclear plant declared a reactor core isolation cooling (RCIC)
pump inoperable based on a low-lube oil level. On his rounds, an
operator dipped the sump and the level came up just below the low-
level mark. The obvious and simple thing to do was to add oil.
However, the plant was operating, and HP (Health Physics) was leery
about exposure risk for a simple PM. They held up the work order, put-
ting Operations in the awkward position of having to perform an oper-
ability assessment. Asked for my opinion, my initial reaction was, You
393
have got to be kidding. I compared the exercise to driving my car

about 600 miles to the station in a winter storm, low on oil. In that sit-
uation a running engine was certainly a personal safety issue. Their
equipment was only in standby. This answer was all too simple. Several
hours later, the system engineer (armed with my assurances) put pen to
paper and with a few documents, laid the issue to bed. Everyone
breathed a sigh of relief, the clock ended, and the crisis was over. This
was a bread and butter system engineer routine. About 10 similar events
happened per week. Some points are worth discussion. If operability is
really a concern, why not simply add oil? The dose was several millirem
and oil addition only takes two-hours (with all the paperwork) to per-
form. This would have required assessing the PM program value, vis--
vis the ALARA (As Low As Reasonably Achievable) method.
I have made enough decisions to burn up my share of some equip-
ment, including bearings, on various operational problems. We never
lost a piece of equipment with low oil that we were monitoring. The
RCIC level was being monitored, in addition to being in standby service.
It is likely that correct oil level was never initially established since the
equipment was in mint condition and had no leaks. Although oil sumps
vary, wetted roller bearing levels can get well under nominal range and
still perform nicely. This equipment had synthetic oil. The supply oil
function failure level falls well below the low-level limit on typical lubri-
cated equipment. It is a simple matter to confirm this from the equip-
ments oil sump drawing. Failure modes show up as pumping air (from
a low oil level) and foaming (from a high oil level), but you need equip-
ment in service to see this.
Clearly, we prefer to keep the oil level in range, but my experience
has been that this equipment is insensitive to small oil level variations.
There are many occasions when performing oil addition at inopportune
times can be prevented, keeping this allowable variation in mind. This
was what I had alluded to with the car metaphor. I will operate with
low oil on a 4.5-quart system (e.g., down to three quarts), and never
have problems, even driving hundreds of miles in winter weather.
Equipment margins are available to be used, and should be used to sup-
port ALARA and HP, but not to the degree that Operations must esca-
late to a plant shutdown level.
394
Further Readings
About 10% of the checks I do on the road indicate I am low on oil.

I drive high mileage cars, odometered at 140,000 miles or more, and
these use a bit of oil. The car I previously mentioned used about a quart
per thousand miles. I also add oil at favorable times and weather (day-
light, wind less that 30 mph, no precipitation), except in emergencies. I
have driven upwards of a thousand miles with oil below the dipstick on a
couple of select occasions. I have yet to burn up an engine. Likewise, the
RCIC pump could have used some oil, but consumption was well known
and trendable. Oil could have quickly been added in an emergency.
A broader question routinely asked in a nuclear plant is Is a missed
PM cause to declare equipment inoperable? Based on statistical
reviews of missed PM operability assessments the answer would gen-
erally be no. I cannot recall one that declares equipment inoperable! In
an RCM context, the answer could be yes. This is particularly true if the
task is the condition-directed maintenance follow-up to an on-condition
measurement. In this case, the clock is running. Limits are based on fail-
ure resistance. The beauty of engineering (specification) failure is its slow,
progressive nature. There is a period of time before the failure progress-
es to rendering the equipment inoperable.
This is the dilemma of PM programs. Intervals are created for a rea-
son, but invariably some are missed in even the best of programs. An
effective program performs a high percentage of PMs on schedule; an
ineffective program does not. Intervals are set with great conservatism
and are widely spaced in most programs. When completion percentages
of PMs on the books are in the 5-10% range, it is hard to claim that
there is an actual PM program. Often, these hopelessly over-conserva-
tive paper programs are set up in the hope that a few PMs will be
accomplished. My observation is that many facilities fall into this latter
category. This is partly due to over-conservative intervals and absence
of life cycle maintenance program design. Few people run their cars
this way while most are generally successful. Few are engineers or
mechanics. The real tribute is the robust nature of the equipment, and
the capacity of operators to perform condition monitoringthey keep
these facilities viable. How much cleaner, efficient, and simple is the
specified program that is largely completed! Where these program
types are in place, costs and failure rates are substantially lower!
395
ALARA
An HTGR nuclear plant had intermittent control rod problems. As
the reactor engineer, I proposed PM to restore several of the nine inop-
erable spares to service level. We did not have any spare control rod
assemblies. Unfortunately, control rods are not only contaminated, but
activated, and potentially cause substantial exposure. Working on the
lower activated areas resulted in exposures of up to 10 millirem per
hour. The neutron absorbers were far too hot to work with directly.
Fortunately, assembly work was many feet away. For three years, Health
Physics (HP) held the PM work orders based upon ALARA (as low as
reasonably achievable radiation exposure). We could not perform
control rod assembly maintenance. Then an event occurred. The plant
scrammed during startup, and six control rod drives failed to insert.
The alert went up the corporate ladder to the president, since the plant
was under shutdown order. Suddenly, it was an all-out sprint to devel-
op and implement control rod-drive PM plans. Now HP was receptive.
We knew what needed to be done, but were woefully short of spare
parts. We puttered around developing work plans and failure informa-
tion. It was months before we could start work. When we did, the
exposures from control rod overhauls were between 20 and 500 mil-
lirem per drive. The total overhaul project, as I recall, required a grand
total of around a 100 man-REM over the course of one and a half years.
HP learned to tolerate control rod PM-related exposures. But HP was
still a maintenance work barrier. All work orders in this plant went
through HPeven work orders in the switchyard! On a good day,
walking a WO around for sign-offs took four hours. The fact that the
plant was the radioactively cleanest in the country and less than 1% of
work hours involved contamination or radiation spaces could not influ-
ence this turnaround.
At one critical point, while rebuilding control rods, we reattached
the highly-radioactive neutron absorbers to the drive assembly. We
would figuratively burn up mechanics on their weekly administrative
radiation exposure limits. A few subtle points help explain why the
plant was eventually forced to shut down. First, HP and ALARA fun-
damentally did not recognize PM in their plan. They discounted any
396
Further Readings
work that was not a crisis. HP was much more receptive to broken
equipment maintenance directly supported by the plant manager. This
reflected the prevailing culture at the plant.
Second, most of the control rod drive failures were secondary fail-
ures. The absence of a startup maintenance strategy on control rod
drives (and a host of other equipment) necessitated earlier overhaul per-
formance. For radioactive equipment, the longer an overhaul interval
can be stretched the lower the man-REM exposure. HP ALARA, as
practiced, ultimately increased the life-cycle exposure for the plant.
Theoretically, PM warrants ALARA recognition. Most HP administra-
tors and technicians know little about maintenance. They do not trust
the maintenance supervisors and workersafter all, they cause 95% of
the HP workload, including contamination events! Work groups
reflect prevailing culture. This HP department approved work based
on a call from the plant manager. This was ultimately not competitive
for a commercial nuclear plant.
Optimizing radiation exposure and maintenance costs remains a
challenge at nuclear units today. Recently, HP concerns were again a
barrier to full-scope PM plan implementation. The nuclear world has
not improved in 20 years. One solution is to put the life-cycle mainte-
nance strategy on an RCM-based foundation and pre-approve planned
work. Condition-monitoring programs, on-condition maintenance,
fixed time maintenance, and condition-directed maintenance activities
can be reviewed and pre-approved by HP. ALARA should not be a bar-
rier to PM. ALARA is, in so many ways, simply another cost of doing
nuclear maintenance that has to be optimized in an overall plant con-
text. As with safety, ALARA concerns must be placed on a common
playing field. An activity avoided today that will incur a future expo-
sure ten times as great does not implement ALARA.
One Use of OTF

Some plants are reluctant to drop activities that have no PM value.
Some tasks are applicable in identifying a failure, but are not cost
effective. OTF can document an uneconomic plant activity. When
there is no identifiable failure that can be related to the task, the activi-
ty is recommended to drop. Many drops are scheduled outage tasks.
397
For example, at one plant the weld liners on a continuous blowdown

tank were identified as low value. The liners protected the tank from
flushing blowdown water. In another case, welding of fatigue cracks
from thermal expansion was a drop. The functionality required was
compressive, not tensile strength. Yet welding cracks is a major outage
work activity.
Some other non-beneficial PM examples include rebuilding valves
that were like new upon disassembly; time-based solenoid valve part
replacements for low-importance equipment; and overhauling redun-
dant boiler feedpump lube oil system exhausters. Inexpensive equip-
ment with a redundant installation is always a candidate for condition
monitoring. A host of literature provides guidelines as to when equip-
ment can be effectively run to failure (remembering that this case still
benefits from operator monitoring). Standby and spare equipment run
times are often only a fraction of manufacturers calendar-time rework
recommendations. All too often these recommendations become the
basic interval for equipment PMs. This partly reflects traditional
CMMS scheduling limitations. Reducing these activities is a substantial
saving for HVAC and other support systems.
Typically, work performers know over-performed PMs well. They
visually examine filters, read the differential pressures during replace-
ment, or notice as found calibrations consistently dead on specified
calibration values. Operators can also identify ineffective PM candi-
dates. They know partially loaded equipment, well-maintained envi-
ronments, and other favorable factors that lead to low stresses and
longer-than-average service intervals. At some plants the PM list itself
is an ideal starting point to screen ineffective PM. PMs which have not
been done as scheduled or have only been sporadically performed are
candidates for interval extension or outright elimination.
Standards can document the activities that the plant selects to
perform as time-based or on-condition PMs and those they reserve for
condition monitoring. They can also control factors such as intended
parts, lubricants, and services that have a substantial impact on service
life. A premium lubricant, contrasted with a discount one, can improve
lubrication time intervals by a factor of five. A superior part supports
longer service intervals than an inferior one. It is common to see lubri-
398
Further Readings
cants and other consumables vendors replaced without consideration

for product performance or service interval appropriateness. Standards
can help avoid discount purchases that backfire. For your vehicle, you
might purchase a superior lubricant on the basis that it would reduce
service intervals. Would you also increase service frequency when you
dropped back to a low-quality substitute? If you did not, you would
greatly accelerate deterioration and reduce the overall life of your vehicle.
Maintenance mythology exists everywhere. Some is founded on
facts, but much of it is culturally based. OTF can help manage the
things people want to do, but which have no value. As with revving an
auto engine before shutting it downwhich is supposed to make it eas-
ier to startthere is no factual basis for many activities. To trim inef-
fective tasks, someone has to document that they see no value in a task
and put that in the organizations history. Leave it to the work advo-
cates to explain how it adds value. Many times a credible story simply
cannot be found.
For an organization moving toward condition monitoring (from
conservative time-based maintenance or fixed time overhauls) an
important caution is to assure that the organization builds an appropriate
maintenance performance response. A condition-directed maintenance
task has no benefit when the maintenance organization cannot turn it
around in time to avoid final failure. Indeterminate work that takes pref-
erence over the time-based and condition-directed tasks will sidetrack the
program. Maintenance credibility is the key to overall success. A credi-
ble program works simply, understandably, and with commitment.
Risk Management: FEAR

Why do plants perform low-value maintenance work? In a nutshell,
the answer is fear. When you do not understand what things do and
how they interrelate, you fall back to doing what has always been done.
When you are uncertain about failure causes, a defensible action is to
fix something. I have never heard a performer, manager, or utility cited
for performing unnecessary maintenance, even where a long history of
infant mortality problems was evident. Much exploratory surgery
maintenance has gone awry, particularly in outages. Medium-sized
399
equipment such as compressors, pumps, and motors, which are not

highly visible equipment, result in a great deal of spontaneous work.
Many engineers and managers have college degrees, but limited
hands-on experience for how things actually work, age, and fail. Many
maintenance workers have wonderful insight, skill, and intuition, but
are at a loss to challenge official work scopes and descriptions for-
malized in plant CMMS systems. Maintenance workers ignore written
instructions, or take them as general guidance based on this knowledge.
Maintenance is tradition-oriented. Change happens slowly. Blame is
common. Any change creates a target for blame. In one case, a rotary-
dumper motor-brake setup was extended from two-month intervals
(inconsistently performed) to four months, assuming maintenance
could perform it to the schedule. The total work scope declined by
50% in the maintenance streamlining effort. The interval was missed,
however, putting the scheduling and the PM team on the firing line.
Assessment revealed:
The plant missed the two previous scheduled quarterly PM per-
formances (on the same equipment).
The scheduling system lost this PM hardcopy for two months.
The nominal quarterly PM was used as fill-in by this shop.
The electrician performer had physical limitations.
The electrician performer had never done the work under training.
The dumper used a complex master-slave control that was sensi-
tive to control adjustment.
The work done had not been documented.
Peers questioned the assigned work performers selection for the

job. The physical location required light build and this man was big.
The unresolved questions pointed towards plant maintenance adminis-
tration, not RCM intervals. Unfortunately, the plant had no process to
perform a root-cause assessment, so few things changed. With little
documentary evidence, and only speculation as to what actually hap-
pened, it is hardly more than hearsay to guess at the work performed or
its problems. When companies work without documentation and work
plans, tracing mix-ups is difficult. This basic administered scheduled
maintenance program could not deliver TDM or OCM/CDM. We had
400
Further Readings
pushed the performance interval to the limit without considering the

plants capacity to perform and deliver PM! We thought that simplify-
ing the work significantly would be adequate to assure task perform-
ance; it didnt.
What is the solution? When more people have access to informa-
tion that supports performing maintenance (or not), better decision-
making will result. The development of maintenance strategy is helpful.
Consistent compliance with approved maintenance plans, expected
from licensed airline mechanics and maintenance, reduces the seat of
the pants decisions. These decisions may lack insight into the task per-
formance basis, but remain essential to keep informal programs run-
ning. More dialogue and development of trust between and among
workers, maintenance schedulers, and analysts is necessary. The adop-
tion of an RCM-based PM format takes the guessing out of all
TBM/CDM tasks. Just do it becomes the mantra. Maintenance can
be objectively measured based upon the completion degree of the result-
ing plan. Uncertainty and judgment will still be present, but in the devel-
opment and diagnostics of CM-type work request where problems must
be defined. As Franklin Delano Roosevelt once said, The only thing we
have to fear is fear itself. The same applies today in maintenance.
A Case for Overhauls

Nolan and Heap make a strong case against traditional overhauls.
They address the commercial aircraft industry, focusing on jet turbine
engines. This creates the case for conditional overhauls. Conditional
overhauls correct primary and secondary failures but do not exhaus-
tively replace non-aging parts and components. They should be used
more in traditional utility maintenance to manage costs. For turbines
and other large rotating machines, however, there is still a case for
overhaul, based on grouping many multiple independent failure
mode PM tasks and the extensive disassembly required for large
machines. For a turbine, many inspection tasks need to be performed
based on time and risk. Some include:
401
Instruments
Penetration weldment inspection

Failed thermocouple replacement
Failed pressure sensing line replacement
Calibrations
Control connector inspections
Stages
Blade deposits removal

LP stage tie wrap inspections
Blade root tip crack inspection
Bearing dimensional checks
Steam cut checks
Across gaskets
Along the split casings
At penetrations
Bolt crack inspection
Rotor crack inspection
Overhaul activity requires turbine stage disassembly. Disassembly/

reassembly for a large 500 MW machine alone is a 4,000-work-hour
task. Even with a three-shift coverage working six days per week, it
takes a 20-man/shift crew several weeks to complete! Most hours are
required in performing disassembly/reassembly. For this basic time
investment, it is absolutely critical to achieve maximum leverage and
minimize risk. Large grouped activities, driven by disassembly time,
create overhauls. I believe the term does have utility.
Effective overhauls require using both time-based and on-condition
maintenance risk management. For example, consider the rotor bore
crack inspection. Rotor bore cracks are low probability events, but ones
with great potential consequences. What is the utilitys risk profile for
such events? Are cracks minor cost factors, or major generation risk
factors? Manufacturers recommend performing inspections on every
overhaul. Our experience has been that doubling those intervals was
402
Further Readings
suitable. This adjustment provides risk-management, but also substan-

tially reduces overhaul costs. A decision such as this can only be made
unit by unit based on specific inspection results. This requires main-
taining a history. Unit loading influences crack propagation. Double-
shifted units might not support this decision. Changing a units service
would require re-evaluating this on-condition task. Overhaul tasks can
be time-based or on-condition. For example, performance efficiency,
loading behavior, and main bearing vibration-level trends are on-condi-
tion indicators. Time based age mechanisms include blade root tip
cracks, tie wrap cracks, and control valve deposits. Instrumentation can
convert time-based tasks to on-condition tasks. Blade deposits can be
monitored by careful stage efficiency tests. That necessitates instru-
mentation maintenance such as calibration.
Feedwater chemistry influences the rate of metal transport as well
as blading deposits. Feedwater heater leakage or boiler operation can
influence oxygen levels in transient periods, as well as the need to per-
form an overhaul. Condenser leaks influence feedwater chemistry
especially at facilities that lack full-flow demineralizers. These second-
ary factors can be primary failure causes. Overhaul timing can be
improved using a combination of:
(1) known aging performance
(2) history since last overhaul
(3) condition-monitoring as an ongoing risk control practice
What are the on-condition (condition-directed) tasks? Typically,

they include:
bearing temperatures (thermocouple replacement)

bearing vibration (bearing inspection and rework)
performance (specific problem identification and correction
like blade deposits and erosion)
especially stage performance
load capability
response
control valve position trends (valve stem and seat rework)
valve stroke tests (valve packing and operators)
turbine protection tests (protective devices rework)
403
A turbine in baseload service needs to operate smoothly at steady

load. A load-following machine, on the other hand, must ramp
(move to a new load level) smoothly. Failure to perform either well
could indicate control valve seat erosion or stem bending.
Owners are extending turbine overhaul intervals. Five-year nomi-
nal overhaul intervals are being pushed upwards of 12 years, based on
condition monitoring. The nominal overhaul period is less important
than what the unit performance and experience support. One unit
could not achieve five-year overhaul intervals due to copper transport
from feedwater heaters. Copper turbine-stage blading plate-out severe-
ly limited loads. Overhauls should be extended using combined infor-
mation. As generation gets more competitive, achieving maximum pro-
duction with no schedule interruptions is more important that ever.
Conditional Overhauls
Conditional overhauls specifically correct equipment failure, its cause,
any secondary failures, and nothing more. A conditional overhaul is an
opportunity for traditional workshops. The basic idea of only fixing the
obvious primary and secondary damage from a failure is widely used in the
commercial world. Aircraft turbines receive conditional overhauls. A
conditional overhaul extends to automobiles, diesels, and other large
equipment such as power turbines. We conditionally overhaul equipment
at home. When an automobile engine fails due to a main bearing, we eval-
uate the remaining life on the engine. Then we either perform a selective
complete bearing replacement (on a relatively new engine), or bearings
and cylinders on an older engine. If there is any ring or valve damage we
fix that too, based on the equipments age and our inspection. To apply
conditional overhauls, one must understand the time-based needs for the
equipment. Then, when a failure occurs, they must evaluate the failures
occurring in the context of the equipments age. Effort shifts to fixing the
failed equipment. This contrasts with a traditional shop practice of tear-
ing failed equipment down to be rebuilt up from the frame.
We lost a large compressor at a plant. We knew we had finishing stage
problems, but we lacked the staff adequate to follow up the job. Although
the nominal overhaul interval had been four years, only 18 months had
elapsed. The assigned mechanic did a complete compressor teardown.
404
Further Readings
When it was complete, the only problem noted had been the premature
failure of the fifth stage compressor wheel. It was an expensive over-
haul at $250,000; on the other hand, fifth stage replacement ran around
$30,000.
In another example, a gearbox was lost due to a failed retainer. A
grinding noise had resulted in the early gearset shutdown. The mechanic
tore into the gearbox with no specific instructions other than to correct the
problems. We found a missing retainer immediately after opening the
gearbox, 10 minutes into a six-hour job. We proceeded through the entire
disassembly from one end to the other, although we suspected the missing
retainer was the sole problem. On completion, we confirmed this. Guide
bearing misalignment from the missing retainer was the sole problem. It
could have been replaced, the cover installed, and the entire assembly run
successfully even before we had finished our exploratory surgery. RCM
states that this partial rebuild strategy is sound and that we should use it.
This approach should be built into a facility work maintenance process. Jet
engine actuarial studies showed that there is no statistically different per-
formance between a conditionally overhauled machine and the complete-
ly overhauled one.
This lesson is counter-intuitive to most shop thinking. The feeling is
that it cannot be good until you have looked the entire machine over.
Given that many shops implicitly direct mechanics to perform work as
they see necessary, conditional overhauls are a prime opportunity to
reduce low-value work. To practically implement conditional overhauls,
however, mechanics must recognize the difference between a conditional
and a full overhaul, committing to perform conditional overhauls when
appropriate. Many mechanics enjoy equipment work, especially tear-
down maintenance. This is why they are mechanics; they excel in their
jobs. Everyone in maintenance has to remember that performance of the
maintenance, in itself, is not the purpose. It is to keep the equipment func-
tional with as few resources as necessary.
Part Aging Dispersion

In maintenance, we often lack specific information on when com-
ponents and parts entered service. We may also lack failure mode sta-
tisticswhich modes are dominant, their likely causes, age dispersion at
405
failure, and related factors. We need statistically representative sam-

ples, but these are hard to develop other than at plants experiencing
very high failure rates. Even here, samples vary due to different classes
of equipment, vendors, operating modes, and maintenance plans. We
rarely have more than a few detailed samples upon which to base con-
clusions. Multiple parts suppliers, alternate mechanics and technicians,
and rework all increase failure modes complexity and variability.
Process standardization reduces dispersion and is common in man-
ufacturing. Standardization can simplify maintenance failure analysis.
We can considerably improve equipment by using maintenance per-
formance information to identify dominant failure modes and bench-
mark these to manufacturer standards and industry data. With specific
failure resistance criteria, we can sample and project comparable equip-
ment failure rates for similar environments. Specific failure resistance
criteria are one implementation key. Applying similarity analysis to
identify similar applications for specific failure modes of concern is
another. These should reflect operating and environmental factors such
as the plant, external operating factors, service, and use. This approach
is similar to the design of experiments. We make inferences when there
are many factors at play. The analysts skill enables us to sift through
history, find meaningful data for the failures in question, and develop
the results into simple conservative programs. This is as much an art as
it is a science. It must be based firmly on what facts are known.
Techniques to firm up and manage risk improve the result and lead to
risk management strategies that are insensitive to exact timing. Actual
experience and performance in service matters most. Teams perform
better than individuals. Experienced equipment reviewers help. Exact
analysis, documentation, and peer reviews can base an assessment on
objective opinion. This can also backfire when analysts are too conser-
vative. Conservatism results when analysts arent trained in statistics.
Another way to manage risk is to assure an effective condition mon-
itoring program is in place. With this, analysts can be comfortable
selecting meaningful intervals. This is particularly true for economical-
ly based failures.
Economic-based failure prevention constitutes the lions share of a
PM program by task numbers and work-hours. For economic-based
406
Further Readings
failure prevention tasks, the lesson is that intervals are extended too
conservatively. The use of condition monitoring, age exploration, and
other hedges can reduce the tendency to incrementally extend equip-
ment inspection intervals. A database of equipment components and
their failure modes is also helpful, as are benchmark intervals. A char-
acteristic of modern equipment is the combination of one or several
dominant age-based failure modes and an underlying complexity. The
composite exhibits mixed characteristics. A strategy of managing the
known aging failures with on-condition or time-based maintenance, as
appropriate based on certainty of aging and organizational capability,
combined with condition monitoring, maintains this equipment very
well. The challenge for operating organizations is to develop simple
standard applications of this strategy.
People lacking confidence and experience are uncertain. Sunday
analysts are squeamish as opposed to hands-on staff or experienced ana-
lysts. Databases and experience provide confidence. For economic fail-
ures, rapidly getting intervals to the age-based failure range is the only
way to learn what aging equipment failure modes are present, how long
they take to develop, and what applications they can support.
Maintenance Budgeting
To focus on managing maintenance costs a plant or company must
have a motivating driver. The traditional utility environment has not
provided this focus. Some companies do not invest in productivity
growth. Operating budgets can influence how productivity grows.
In my experience, changes are made in a previous years budget
adjusted for corporate goals (minus 5-10%, whatever the corporate
accounting office desires). Take existing staff salaries and add overtime
percentage (about 5%), services, and historical material costs. Budget
for non-routine events such as major scheduled outages. Adjust for his-
torical trends such as inflation. The resulting budget is last years plus a
percentage. Then hope for the best!
In my years as manager, we wound up chronically over our mainte-
nance budget. A catastrophic year, such as one involving a turbine fail-
ure, could double budget expenses. How could budget overruns be
407
sustained year in and out? How could they be tolerated with no cor-
porate response? Companies have bought the farm on maintenance.
They accept budgets and expenditures as they occur since no one ever
figured out an alternative. Corporate offices presume that historical
performance adjusted for predictable events is the most reliable budget
performance predicator. Corporate staffs manage catastrophic plant
outage risk by spreading them over all of the plants in the system, effec-
tively self-insuring. If a company owns 20 major generating units and
one major unbudgeted failure occurs per year (such as a $10-million
generator rewinding), it is buried in the $150 million of budgeted main-
tenance expenses for generation. At the fleet level, costs, including
catastrophes, are relatively predictable.
Corporate accounting departments can project likely expenditures
with certainty beyond formal budget submittals. This approach pro-
vides a simple cost-plus expense budgeting plan. It has no bias to
reduce costs. The approach presents barriers to promising new main-
tenance processes. It is averse to changes and to risks. It does not pro-
vide a return to plants for better production or maintenance perform-
ance. Emphasis is on existing staff and contract expenses, not innova-
tion. It fails to allocate money for long term improvements. In an envi-
ronment like this one, struggles over money for RCM, or any non-rou-
tine activity, will always exist. What are the ways to overcome these
mindset barriers?
innovation
standard measures
value focus
contingency budget reallocation
Maintenance and plant managers have been rewarded for being

conservative. This incentive is powerful. To achieve change, incentives
must change. Incentive should encourage innovation. To see what
innovative approaches work requires measurement. Simple measure-
ment and feedback has shown its value repeatedly. Executive manage-
ment should lead the way in promoting risk acceptance. An outstand-
ing approach would be for executives to literally raid the pantryallo-
408
Further Readings
cate part of the plant loss contingency budgets for developing cost-
effective, innovating technology. They should also fairly address any
work loss from productivity changes with affected workers. Success
with RCM means this problem will be encountered. The competitive
generation market will stimulate change. How much and how quickly is
a matter of conjecture. The beauty of RCM is the profound maintenance
process knowledge it provides. For those willing to master the subject,
maintenance need not be a mysterious, budget-busting free agent.
Bigger Opportunities
Most plants adequately manage turbines and boilers. RCM benefits
come from in-between systems that are high in cost and potential pro-
duction impact, but without chronic outage causes. These systems often
include:
sootblowers
sootblowing air
flue gas and overfire air
ash Handling
coal handling
coal Milling
Savings are found at the knee of the Pareto system cost curve.(See
fig. A-1 and A-2. These are discussed earlier in the book.)
These system losses provide high-value opportunities. They influ-

ence the generation of secondary failures (sootblowers for example) or
redundancy loss (coal handling). Their elimination pays double divi-
dends providing cost reductions with production increase!
Success in one area eventually spills over into others. Leadership,
persistence, and commitment are necessary. Pilot programs that were
not nurtured withered on the vine. As RCM is implemented there are
some specifics to look for, including:
condition monitoring
PMs defined as tasks
total existing PM count
409
Figure A-1 Pareto System Cost Curve : Costs
Figure A-2 Pareto System Cost Curve : Losses
410
Further Readings
formal on-condition/condition-directed P maintenance pairs

failure limits explicitly identified
must-do CDM task identified
These increase numerically.
Prioritization rules
Change
Prioritization simplified
These are more defined.
For high-cost, failure-prone systems, materials, services, and work-

hour costs are high. Costs decrease, first as work-hours, then in serv-
ices and material costs. For example:
E MWRs
Overtime
are leading indicators, and decline. Systems that lack redundancy

will see reliability and availability increase. For redundant systems
high cost activities drop.
For existing PM programs:
hours
CNM
work hours
shift to planned work
percentage of work tasks completed as preplanned OCM/CDM pairs
preplanned CDM work tasks
increase.
411
Checklist: State of Maintenance Health
Review List
Purpose: This checklist helps identify the availability of an effective
time-based maintenance scheduling system. This scheduling system sup-
ports the basic PM program foundation.
PM Health
scheduled maintenance work list

active PM Scheduler (usually residing on the CMMS system)
PM completion reporting
operationalized PM Tasks
PM grouping
system level performance measurement
Cost
Hours
Reliability/availability
RCM Health Sub-list
unit operations goals

systems ranking for importance
system goals
PM bases and identification as TBM/OCM(CDM)/CM
OCM/CDM limits
OCM/CDM pairing
CDM performance measures
system functional failures measured
failure analysis
Maintenance Process
Traditional CMMS system request work is based on noted prob-
lems. This is the corrective maintenance model. Maintenance begins
412
Further Readings
with a problem. A second model, scheduled maintenance, supplements

and extends the fundamental model. Combined, these two processes
comprise the basic CMMS software areas. Identification of problems, a
posteriori, is how traditional maintenance works. Proactive maintenance
requires shifting to an a priori model. This is what RCM provides.
Operators understand equipment problems. You cannot under-
stand the problems until you become familiar with the system, equip-
ment, its capabilities, and what you need to do with it. Response-based
maintenance is the first improvement over disposable equipment. It has
a significant capacity to reduce cost. This was the motivation for devel-
oping the space shuttle. A $500 million satellite that can be recovered
and reused at the expense of a single $100 million shuttle flight repre-
sents substantial savings.
Response-based maintenance is very cost-effective compared to the
alternative, which is nothing. It is the first basic step in any mainte-
nance program. The next is an intuitively harder onescheduled main-
tenance. Scheduled, or preventive, maintenance represents the second
maintenance program improvement level. Scheduled maintenance is
much harder to implement than failure-based maintenance. Many com-
panies satisfy themselves with response-based maintenance implemen-
tation. To see what scheduled maintenance can do, consider first what
it cannot.
For any component, there is some residual failure rate tied to ran-
dom failures that are inherent in the component based on design and
production processes. This represents a minimum failure floor below
which we cannot drop. No minimum limit represents perfection (an
impossibility), but through highly evolved, mature designs are not far
removed from perfection. Equipment is often economically retired
before aging is evident. Factors establishing the floor include:
design
materials
construction
environment
operation
413
Figure A-3: Cumulative Turbine Failures.
Failure occurs when stress exceeds capability. Design, materials,

and fabrication provide equipment with capability. Operating stresses
in a perfect, variation-free world would never exceed design. In the real
world they do. Designers must anticipate field loads and conditions.
Suitable materials, manufacturing, and dimensions assure products per-
form adequately, with a factor of safety. Systems designers build sys-
tems from components and equipment. They are not exact. They
stretch design envelopes with operating and environmental assump-
tions. Some application stresses exceed design expectations. Designers
work based upon experience. A residual failure rate is always present
in efficient design. A perfect maintenance program would achieve the
residual inherent capacity of the design, which is the minimum floor in
our example above. Discovering this floor with scheduled maintenance
and extending it through design is the focus of applied RCM. That
90% of component failure modes do in fact realize this inherent capac-
ity with virtually no maintenance is the discovery of RCM. For this rea-
son we must use PM tools with great care!
414
Further Readings
So, where does PM fit in? Condition-directed maintenance lies

somewhere in between absolutely no maintenance and inherent reli-
ability limit. Response-driven maintenance works well as a first step
and this is where many organizations find themselves. Further strate-
gies move closer to design-limited reliability. Scheduled maintenance
effectiveness has been validated by long-term measurement turbine
overhauls. (Fig. A-3)
Peaks are limited by suitable PM tasks that lower failure rates for
some period. Performing maintenance establishes an intermediate
failure level reliability curve. The scheduled maintenance plan drives
the failure rate towards the inherent reliability floor.
For some failures, operational changes or re-design are necessary.
Failures caused by external environmental factors require review of
environmental controls. The characteristic opportunity of each failure
varies from one application to another. Some are minor, others huge.
Many other factors require a consideration of their added value. Infant
mortality or quality control issues can influence the run-in period
with a high failure rate. After some period of time the failure rate drops
to an inherent baseline level.
Condition Monitoring
Most condition monitoring (CNM) maintenance is initiated
through operations. Condition monitoring is monitoring without
specific failure criteria. This is a double-edged sword. It can be hard
to rank, prioritize, and perform condition monitoring due to its gener-
ality. In the absence of time-based and on-condition work order cate-
gories an organization can measure its CNM-originated work. This is
based on the work percentage coming from Operations. If Operations
originates 70% of the work orders, then about 70% are no scheduled
maintenance. Scheduling, planning, and engineering initiate most of
the balance of outage, PM, and modification work.
Time-based maintenance comprises the traditional on-condition,
failure finding, and time-based rework/replace planned maintenance.
If a plant can tag originated work from time-based work orders, then
they can measure the RCM maintenance workload as follows:
415
Summarizing,
PM (time-based): rework/replace TBM
check/inspect/test OCM (including OCMFF)
CM (corrective): operations NSM
engineering/S&P & design changes
A fraction of the CM reflects functional failures. Measuring the

fraction involves reading WOs or checking logs. Few CMMS systems
have fields to record functional failures. Few operators discriminate
between functional and other failures. Logs typically provide function-
al failures, where they are maintained.
A quick way to re-align CMMS systems to measure RCM-based
work strategy is to relate condition-directed maintenance (CDM) to
OCM work orders. Or simply perform all on-condition directed
work as part of the original OCM work order. There are now three
basic work order classes:
PM (time-based) (1) TBM

(2) OCM/CDM (including OCMFF/CDM)
CM (corrective) (3) NSM/OTF
(Failed)
This approach provides a quick way to measure existing processes.

Now, what do the numbers mean?
Different equipment and systems have different profiles.
Redundancy shifts the profile towards OCM/CDM, and ultimately to
condition-monitoring (NSM-OTF). Defined aging with high-produc-
tion impact biases towards TBM. This profile can be helpful to bench-
mark existing maintenance plans quickly. An irregular profile suggests
a more in-depth review.
Benchmark profiles are only being developed in the power industry.
Grouping can show absolute work order numbers any way desired, but
hours worked is a common benchmark comparison quantity. PM
work hours are inherently low. Most non-outage PM jobs are simple
416
Further Readings
tasks. Organizations generally need to increase the OCM/CDM work

fraction. This work involves an explicit failure resistance measure and
when exceeded that measure initiates work. This requires:
explicit failure limits

work performance focus and priority on failure limit exceeded WOS
This combination is often seen in traditional instrumentation pro-

grams. A significant amount of out-of-calibration and failed instru-
mentation work is identified. When restored, immediate operational
reliability improvement resultsa quick payback.
Redundancy and Functional Failure Measurement

Redundant equipment presents two important failure modes. The
first is redundant train failure, such as with two-out-of-three in a stand-
by feedwater pump train. The second is protective device failure.
Protective device functions are redundant in one sense; they alert oper-
ators to a primary function failure. In the absence of this failure mode
they serve no useful purpose. Protective devices should not act until an
event occurs. Their lifetimes are spent in standby. Like redundant
backup equipment, no backup is needed until a primary fails. For pro-
tective devices, inadvertent component failure can create an unplanned
event. In this case, a functional failure results. An unintended, incor-
rect feedwater level trip constitutes a protective action failure. A spuri-
ous, high-vibration alarm trip on a steam-driven boiler feedpump, with
no actual elevated vibration, is another alarm control failure.
Functional tests, near misses, or demand events often reveal pro-
tective device failure. Because safety devices have multiple redundant
backup devices, demands rarely utilize all backup devices and their
trains and infrequently fail to actuate protection. Precursor events are
often near misses revealing lost redundancy and risk exposure. Nuclear
power plants monitor hidden functionality under required technical
specifications and general public safety considerations. Fossil units are
not as structured. System functional failures (FFs) provide a perform-
ance measure. Identifying FFs is subjective, particularly where func-
tional requirements have not been formally identified. Total work
417
orders as a system performance measure is arbitrary; system work

hours and costs are not. Work hours relate directly to labor costs, and
indirectly to total cost. A measure is needed to track failure perform-
ance. Maintenance preventable functional failures (MPFFs) are meas-
ured for nuclear plants. Some fossil plants track large unbudgeted
maintenance expenses, but few track system level availability. Vague,
subjective measures are hard to track and have less value for improve-
ment measures.
Two measures generally correlate with failures and are simple to
use. First, system emergency work orders are useful as an absolute
number and percentage of the units total WOs. The second is system
overtime. Reactive maintenance strategy for a system reflects a high over-
time rate and emergency work order percentage. Both measures are wide-
ly available. For plants that measure hours and work orders down to the
system level, these provide a ready indicator of performance.
Condition Monitoring or
Condition-Directed Maintenance?
Morpheline has a pungent smell. Once you smell morpheline, as
with smoldering coal, you remember it. It is used as a volatile feedwa-
ter treatment at some steam plants. At a fossil plant, on a Main Steam
maintenance optimization project, we were working on steam leak
detection tasks. Valve packing, turbine steam seals, and pipe cracks
cause steam leaks. The question is What is an appropriate mainte-
nance task to identify steam leaks?
Large leaks, noise, steam release, and increased makeup signal that
there is a problem needing to be identified. Noise usually accompanies
steam leaks. Saturated steam leaks exhibit vapor. Inability to maintain
makeup is a sign that a system is not secure. Changes in make-up trends
are one clue to the presence of a small leak. Leaks in inaccessible areas
of the boiler must be inferred. For accessible leaks local inspection,
vapor tests, and ultrasonic tests are the best identifiers. Valve packing
is the most common source of steam leaks. Checking a valves lantern
ring compression is a good time-based packing measure. Most opera-
tors and mechanics learn this on the job. Steam piping is all lagged so
418
Further Readings
minute hairline cracks are not evident. Some condensation dripping

out from covered lagging is all that is seen. Visual inspection for con-
densation is a highly effective leak check.
The operators at a morpheline-treated plant shared one detection
method they use, which is the sense of smell. Your sense of smell is
acutely sensitive. They identified steam leaks by morpheline odor.
When you have a sensitive diagnostic like this, there is no point in con-
sidering test equipment. Our sense of smell is adequate to monitor a
whole boiler unit in an enclosed building. It can be effective for detec-
tion of other serious conditions, such as coal fires, too. Of course,
knowing that there is a leak and finding it are two different things.
Finding leaks can require sensitive equipment.
What is an appropriate strategy for steam leaks? Unless you have
piping fatigue, piping leaks are random. Welds and stress risers experi-
ence aging and NDE exams for cracks are appropriate in aging-stress
areas. Where erosion is a concern, wall thickness measurement is valu-
able. Inspection with radiography can be effective. The assumption
that cracks will be detectable before catastrophic failure supports on-
condition inspections and condition monitoring. Plant lighting, clean-
liness, and accessibility are equally important. Thus for steam leaks we
have:
Source Action Strategy Type

Packing Check for leaks CNM
Take-up lantern ring CDM (Take-up)
Seals Check for leaks CNM
Performance test CDM
Cracks Check for leaks CNM
What is the difference between CDM and CNM? CDM has explic-
it thresholds and is scheduled. CNM is informal, although the two are
very close in performance. For operating tasks, it becomes somewhat
arbitrary as to the category in which an activity fits. CNM generally
requires more experience and skill to apply. Interpretation is subject to
opinion. Some operators note everything, while others see very little.
Experienced, skilled operating staffs use CNM with high degrees of
419
success. If CNM is effective, and the organization can support on-

demand CDM effectively, then CNM should be used (if only) because
scheduling is simplified. CNM depends on environmental factors.
When suitable environments are absent, monitoring sensitivity drops.
As with finding oil leaks on an oil-soaked floor, CNM evidence can be
missed when:
background noise is high
randomness prevails
lighting is poor
housekeeping is poor
Operator Training and Life Cycle Maintenance Cost

After WWII, my dad was stationed in Japan as a part of the occu-
pation army. My mother followed him over. Lacking much to do and
wealthy (compared to the devastated Japanese), she looked for ways to
increase her mobility by traveling, shopping, and entertaining herself.
Unfortunately, she had never learned to drive, and Pop was not inter-
ested in teaching her how. He cited her innate physical inability, cer-
tain loss of the car in a wreck, insurance costs ... obviously groundless,
irrelevant reasons why she should not drive. She had a friend, Cookie,
of a similar independent persuasion. Cookie, having a car, offered,
encouraged, and even insisted to teach her how to drive. The point of
this story is to point out the value of training.
Cookie taught my Mom to drivein a manner of speaking. She
taught my mom to drive with two feetone on the gas pedal, the other
on the brake, at all times. My mother never was a good driver in the
sense of being consistent on the accelerator or brake. (On the other
hand, she never had a serious accident.) And Cookie was the reason.
And she was hard on brakes. If she was lucky, she would get 10,000
miles on brake linings. Typically it was around 5,000. In high school my
brother and I kidded about how, when mom drove away, her brake
lights remained lit until she dissapeared over the hill. That was about a
half-mile up a long grade from our house!
Average drivers get around 40,000 miles to a set of brake pads or
linings, and many get much more. I get around 80,000 to 100,000
reguarly. ( Of course, this depends on where you drive, how you drive,
quality of linings you purchase, and other factors.)
420
Further Readings
Think a moment about the life-cycle costs. She put at least 300,000
miles on various cars over the years. (She worked and commuted 50-100
miles each workday over most of her life.) At around $200 per brake
joba competitive 1990s rate ($300 is probably more like it) we have
20 (100,000/5,000) jobs per 100,000 miles, or around 60 total jobs. In
todays dollars these added up to 60 ($200) or $12,000, conservatively.
Probably more like $18,000 considering secondary damage when the
brake lining work got missed. Then throw in the present value costs
over the years and you are up around $20,000. Then consider we
havent started to value anyones time! What would the training cost
have beena few hours? At perhaps, $50/hour for a skilled driver
(using todays rates). Cost benefit (benefit to cost) is 10,000/$100
(round terms), or well over 100/1 conservatively.
Missing this kind of opportunity on a personal level is expensive; in
business it is uneconomic. Unfortunately, for all the same reasons, busi-
nesses regularly miss the opportunity to train employees, especially
operators, in the optimum use of equipment to manage costs. I hesitate
to use correct because that presumes there is a correct way, and
there is not one. There are only costs.
Strategically, one distinction between excellent (low-cost, high-reli-
ability) companies and also-rans is the ability to train people cost-
effectively. Why are there so many also-ran companies in business? One
reason is protected markets, such as the traditional utility industry.
Another is market inefficiency. Many American companies see their
product benchmark costs as unfavorable overall, while their training
costs are negligable, and they cannot make the connection. They lack
the profound business knowledge to relate training costs to final prod-
uct costs. A former boss of mine, R. O. Williams, used to jokingly ask
Whos the most expensive person on the payroll? At the time he was
the highest-compensated executive in the company. His answer, The
worker who is not trained.
Failure Complexity
Failures can be classified as simple or complex. Simple failures
involve single faults and modes without interactions or secondary fail-
ures. Aging failuresin which a specification is exceeded, such as
421
pump weardemonstrate simple failure. Complex failures involve hid-

den failures and secondary failures. Multiple failures may be complex.
We might have known that we had a steam-packing leak, but had been
unaware of secondary instrumentation damage that developed after-
wards. Hidden secondary failures mask the extent of the failure.
Instrumentation failures are often hidden. This suggests a strategy to
diagnosis complex failures. Complex failures can result from common
causes such as environments. For example, a complex economic failure
in a prototype nuclear plant resulted from a failure to act on the helium
interspace gas moisture monitor alarms. These monitors forewarned of
moisture in purge helium used in keeping instrument penetrations dry
and clean. The moisture-related condensation increased the instrument
failure rate. By the time the scope of moisture contamination was iden-
tified, we had a major instrumentation rework effort on our hands.
This type of failure is protected by instrumentation. These failures are
categorized as common mode failures when equipment independ-
ence assumptions are invalidated. This is of particular concern in the
nuclear and aerospace industries.
Overlooking environment maintenance is a common mistake.
Environments or process inputs can fluctuate from normal specifica-
tions causing complex failures. Examples include:
Building HVAC for a PRB coal-fired unit.
The buildings louvers and dampers were an integral part of the

HVAC. They froze due to moisture in the wintertime. They were hard
to get to, so gradually they went out-of-service. With about 50% of the
total cooling capacity out, the building was extremely hot in the sum-
meraround 120F in areas of many ignitor logic circuits. Mechanics,
electricians, and instrument technicians had to cope with higher rates of
ignitor failures under such adverse circumstances. They understandably
did not want to work in the hot areas during these high-failure periods.
In winter, with many louvers stuck open, the cold air blasts and mois-
ture coming in froze more equipment. Although not as cold as it was out-
side of the building, the cold blasts from the prevailing Arctic winds
caused temperature excursion failures.
422
Further Readings
High flame scanner, ignitor failures and control drift costs were the
secondary failures; they had a common root cause in the absence of design
environmental conditions.
Instrument cables in a nuclear power plant.
A steam leak occurred on a complex, high-pressure, steam pipe-articu-

lated thermal expansion joint. Although the immediate area of leakage
was shielded with a heat-resistant blanket, the increased temperature and
moisture resulted in condensation along many of the connector pins for
the plants primary coolant flow instrumentation. About 20 large control
cables terminated with 100 pins each. When these started to fail by inter-
mittent grounds, the plant went into technical specification grace periods
while the connectors were removed and cleaned. Not all connectors could
be reworked at load. The scope of the problem was not evident until an
unscheduled shutdown. The joint was repaired at this time, but the cable
pin problems persisted for months despite extensive rework.
Water chemistry for a coal unit.
The unit suffered a condenser tube leak. The contamination of the

condensate quickly dropped the feedwater pH to the under 7.0 range
with no subsequent boiler trips. Dispatchers were pleading with the units
crew to not trip the boiler. The unit went down four hours later on a boil-
er tube leak, but not before extensive scaling occurred in the boilers
waterwalls. Subsequent tube leaks and heat transfer imbalance required
special chemical cleaning. Cost of the chemical cleaning alone amounted
to nearly one million dollars. The cleaning added over a week to the out-
age scope of a 350 MWe unit. Lost generation cost amounted to $500,000
at the companys generation value added rate (difference between pur-
chase cost and generation cost).
Fire monitoring for a coal unit.
The units methane detection and fire protection deteriorated due to

wet environmental conditions causing a high rate of control failures. Acid
423
runoff from coal-dust water-suppression sprays caused a high incidence of

instrumentation faults. Water spray was required because the plants orig-
inal dry dust suppression was inadequate for the dusty coal supplied.
Water sprays were added when the dry dust system proved incapable of
managing the dust and being maintained at the same time.
The dry dust suppression system was rebuilt after it was pointed out
that the stations license certificate explicitly demanded it. The replace-
ment system was installed under adverse winter conditions at over three
times the budgeted cost under the pressure of a regulatory compliance con-
sent agreement. The replacement system, in spite of installation troubles,
was highly successful and returned the unit to license compliance Local
papers groused afterwards that the company was inappropriately excused
from fines for voluntary compliance.
Instrumentation and controls for a coal handling area.
Coal handling equipment that monitored the tramp iron, belts, and
alarms went out of service for a variety of reasons. Coal handling did not
warrant resources beyond the emergency level. At the unit age of 15
years coal handling system costs had taken a number three position behind
the boiler and the turbine. It appeared a matter of time before direct coal
handling outages impacted production. Coal handling equipment:
failed to crush coal to specified size

failed to remove iron or metal
required heavy washing to manage fire risk
suffered repetitive spills that required manual cleanup and heavy
washing
Hydraulic fire at nuclear unit.
A nuclear unit had hydraulically actuated bypass valves. The hydraulic

fluid, a commercial, stabilized synthetic, was thought to be immune to fire,
based upon supplier promotional literature. Hydraulic leaks from the
valves occurred chronically. Pans were installed to catch and funnel leak-
ing fluid into catch cans of approximately ten gallons capacity.
424
Further Readings
One of the drains plugged up, the tray overflowed, and the leaky fluid
dripped down onto exposed reheat steam safety valve hardware two levels
below. These started smoldering. Because of the non-conventional plant
design, the exposed parts of the safety valves were slightly above the flash
point of the fireproof fluid. The smoldering fluid ignited, and the little
fire eventually triggered the fire detection system. An operator responded
and extinguished the fire. Subsequent flashover after the flame was extin-
guished extensively damaged an area of intense cable, instrumentation,
and control equipment adjacent to a cable spreading room. Damage repair
took a focused effort of nearly 90 days and a special release from the NRC
to restart the unit.
The area of the fire was congested, dirty, poorly lit, and the facility had
historical problems of the hydraulic valves, especially oil leakage, that
were a root cause of the blaze. Direct costs of repairs were between $10
million and $20 million dollars.
Common to each of these events was the failure to maintain a spec-

ified environment or conditions followed by the inability to control sub-
sequent multiple equipment failures. In some cases, they led to a major
event. At the time, plant operators thought they had no alternatives;
their focus was on generation. After the event, or when the cost and
reliability impact of the secondary failures was evident, the primary fail-
ure problems were corrected at great expense.
Some events are humorous in hindsight but for those that suffered
through the crises, or participated in the front-end decisions that later
lead to problems, they were disheartening. The residual problem cor-
rections required extensive efforts that detracted from the strategic
needs of the plants. The lessons here include the importance of main-
taining environments and the relative ease of diagnosing and correcting
simple failures in contrast with complex ones. This lesson, in fact, is so
important it needs emphasis.
A primary benefit of a comprehensive (implemented) RCM-based PM

program is the focus on identification and correction of failures while they
are simple and less expensive to correct from either a production or cost
perspective.
425
In conclusion, after events such as these, the availability of a com-

prehensive hidden-failure test program is very helpful to assure all con-
trols and instruments of consequence are restored. Nuclear plants have
these programs implemented as surveillance plans. Fossil units do not
have an equivalent.
Interval Extension with Age Exploration

Age explorationa fundamental design toolis formalized by RCM.
Design engineers have always seen product improvement as their design
role and goal, but widespread TQM application has shown wrench turn-
ers can improve designs, too, including quickly extending maintenance
intervals.
My experience has been that PM interval extension is done with great
fear and caution. Whether it is the oil in your car or the turbine overhaul
for a power plant, everyone gets queasy extending intervals, particularly if
his name is on the extension request! Consequently, we extend intervals
incrementally. Five to 10% increases are typical, and we do not see the
accelerated effect of any major faux pas. Increases in minor intervals dilute
the benefits of major technical, parts, or materials upgrades. This is true
when a valve diaphragm polymer is improved for temperature resistance
or the lube oil in equipment is upgraded to synthetic. These superior
products age much better in service, and we need that knowledge. To see
the difference in performance, we must either perform a painfully detailed
condition-monitoring assessment, or select a few candidates for accelerat-
ed aging and see how their materials perform. The safe way is to do a
material aging characteristic study. For example, fatigue aging relates to
the stress energy levels of the metal, and there are a number of ways to test
this in the lab. A lubricants replacement interval is based upon successful
machine performance up to a given level of oil deterioration. We can
measure the end-of-life characteristics of the oil (or other material), and
then compare the superior products performance.
At end of life a traditional distillate-based lubricant shows physical
and chemical property changes. Viscosity, dissolved metals, particulate
count, total acid number (TAN), and other indicators demonstrate aging.
If we can correlate any one of these to the aging of the equipment, such as
total acid number, then we can estimate lubricant aging for other lubricant
426
Further Readings
comparison, such as a synthetic. We may find that the other lasts twice as
long in service. This approach is exact, engineering-based, and controlled.
Complex equipment provides a different challenge. It may not exhib-
it a dominant failure mode. We may not have enough experience to see
how it performs in service. However, we need a maintenance plan. If we
use the OEMs recommended interval and observe no failures over the
service period, how should we go about extending the interval? Here is
where RCM provides a useful tool. First, we need to quantify the failure
mode in question. It must not have a safe-life limit. We must be assured
that the failure will not create a personnel, public, or other hazard. If it
does, we should have, through the supplier and other agencies, a great deal
of information on which to fall back. If it does notas is the case in 90%
of the PM activity in a typical plantour next task is to reasonably extend
the interval. The RCM approach says that actuarially we have a solid basis
to extend the interval a substantial amountabout 50%! This is usually
a shock. For me, this is still like leaping off a cliff. Based upon studies per-
formed for no predominant failure mode complex equipment, we will
do very well with large service interval extensions by fitting a no experi-
ence template. These large extensions are exactly what we need to iden-
tify dominant failure mode characteristics in complex equipment.
This type of extension either very quickly extends parts out to where
a lifetime can be identified, or very quickly achieves substantial reductions
in PM hours performed and associated cost. In the context of an ARCM-
based approach, it can be done with very little economic risk. In this way,
we greatly accelerate the rate at which we learn the dominant failure
modes and their appropriate PM intervals.
Before RCM, few would perform substantial part life extensions.
Now we can extend intervals with some comfort. Not only are large exten-
sions possible, but they are statistically justified. In fact, there is very little
statistical justification for initial intervals for most equipment. Typically,
the first few failures are assumed to approximate mean life. We wind up
with greatly conservative service intervals from the onset. A corollary con-
cerns cases where PMs have been missed with no adverse failures devel-
oping. This experience justifies extending intervals to the discovery limit.
These add legitimacy to interval extension. With work performers includ-
ed in age exploration of parts performance, we can advance quickly to
more accurate realizations of potential equipment lifetimes.
427
Operate-to-Failure Is Not What It Seems

Operate to failure (OTF): Definition: A planned-maintenance
choice to perform no scheduled maintenance activities prior to equipment
failure. Equipment failure identifies a service requirement; e.g., main-
tenance is performed when the equipment self-identifies a maintenance
need. OTF selection can be based upon (1) absence of a functional failure
impact on a system, its safety, or environmental performance; or (2) the
absence of an applicable, effective maintenance task.
Operators monitor OTF equipment for condition and perform-

ance. OTF is more accurately described as no scheduled mainte-
nance or NSM. NSM better conveys the meaning, for the plan does
not allow OTF equipment to proceed to functional failure. NSM is a
legitimate strategy because failure actuarial studies show that 90% of
component types will not fail during service in a preventable way. They
are inherently reliable and will not benefit from PM over their serv-
ice lifetimes. OTF is a misleading term because it implies that compo-
nents will eventually fail in service, when in fact, they will not. This is
not appreciated practically by many operating and maintenance staff.
Key Points
Part of a system: An item selected for OTF is part of a system.

Item failure should have no impact on the system functionality to be an
OTF candidate. This could be due to:
redundancy
low failure impact
acceptable risk (for random failures, for example)
inherent reliability
No Direct System Impact: The item must not impact any essential
system functions.
functional failures absence
428
Further Readings
production impact absence

safety impact absence
environmental impact absence
Engineering-Specified Failure
Engineering specification-defined failure is based on a gradual

deterioration towards out-of-specified condition. When a spec-
ification has been developed it provides a measure of failure
resistance. In many instances failure measures are implicit; an
exact failure limit has never been identified. Frequently, this is
because the component is inherently reliable and very little failure
experience is available. (Fig. A-4)
Failure deterioration is based upon a specification; the item
retains residual performance capability. The failure in this
case is a continuous process and is tolerable for a brief period
beyond the specified limit as performance deteriorates.
Spec-based failures are proactive; they are based on design
limits, not catastrophic events. Design limits incorporate margins.
There are no known or specified applicable/effective tasks or cor-
responding failure resistance limits that have been identified or
agreed upon.
Random Failure Nature
Random failure characteristics often cannot be eliminated; design

has reduced the failure risk to an acceptable level by redundancy
or inherent reliability characteristic
Cost
The cost of replacing the failed equipment is lower than that of

preventing the failure.
429
CNM
Key:
Figure A-4: Maintenance Terms Map
Identification
The failures are typically identified by operators on area checks,

walk-arounds, or in use.
Failures may be identified through outage work.
430
Further Readings
Typical Candidates
any function where the failure resistance is ambiguous, not

explicitly identified, or not yet worthwhile
instrumentation (non-critical)
small items with local failure effects and no secondary damage
failures
engineered equipment with wear-out limits for gradual deteri-
oration
small tools
minor/hidden function items
* not in service
* event driven
* risk-acceptable
* joint probability of failure vanishingly small; risk acceptable
Largest Typical Plant Application
I/C program instrumentation
Very often, substantial amounts of non-critical instrumentation can

be effectively run-to-failure for calibrations & other maintenance.
These savings can be large.
Examples:
Home
1. light bulbs
2. small TVs, other consumer electronics
3. small appliances
4. watches
Plant
431
inherently reliable components

passive components
service components
1. small motors, pumps, valves

2. corrosive service pipe
3. tank liners for acids, caustics
4. flooring
5. structurally redundant steel
6. small piping
7. out buildings (pump houses, etc.)
8. cable
9. conduit
10. pipe
Maintenance Discipline
When maintenance programs struggle with PM, it may reflect a
problem with discipline. Developing and following a work plan reflects
maintenance discipline. Discipline means the ability to comply with
standards, no matter what their source. Correctly initiating work
orders, working to procedures, working to schedules, meeting dead-
lines, writing work summaries on completed work orders, signing com-
pleted workall can be reduced to basic work habits that demonstrate
commitment to standards. Work habits are hard to learn and easy to
compromise.
The Navy relaxed standards in the early 1970s. Candy, food, and
beverages increased the food residue in sleep areas. In short order, on
some ships, shipboard spaces looked like dumps. Cockroaches became
shipmates.
Discipline requires standards, training, and reinforcement.
Unfortunately, reinforcing behaviors is not the strength of traditional
maintenance. Unaccountability can prevail. Lawsuits have been filed
against companies stemming from the most trivial attempts to exercise
standards and authority. Submitting signed, accurate time cards, keep-
ing tools stored, cleaning work areas, even wearing shoes to work were
432
Further Readings
all cause for complaint. Companies that lack discipline coincidentally

struggle with PM. No amount of paperwork performs PM. It takes
someone who knows and cares about equipment and abouty his facility.
If discipline is absent, companies may need to add it to their strategic
goals.
I am not an advocate of authoritarianism. I do, however, believe there
are fundamental standards and processes. Everyone working at his own
pace will not cut it competitively. Companies must identify and adopt
their own basic standards. Without standards,there is no maintenance
processno foundation to build upon. Focus on political and social
goals at some utilities left them fundamentally without discipline.
Companies without standards will not be competitive.
Pre-stressing Tendon Buttonheads

Fifteen years ago, a nuclear plant had a corrosion problem with its
containment concrete pre-stressing tendons. Water droplets had accu-
mulated in some of the tendons, and around 40 buttonheads (144 per
tendon) on the anchor hardware had popped, up to five on a single ten-
don. These indicated wire failures inside the tendon conduit tube.
Without going into too much detail, accumulated moisture had initiat-
ed a corrosion cell on about eight of the 500 tendons and the result was
an operability concern of the vessel containment pre-stressing system.
As it was an essential system, nuclear safety was involved. The episode
became a plant startup issue. Analysis and review showed that the ten-
dons that experienced failures were a small population of the longitudi-
nal and bottom-circumferential tendons. The common factor was a ten-
dency for the tendon tube grease to drain away from the tendon head
and button hardware where it had been applied, after exposure to the
heat. Later, moisture in the tubehypothetically from original con-
struction and with the temperature differentialformed mass transport
cells and condensed as water onto susceptible bare anchor wire, but-
tonheads, and buttonhead wire extensions.
The few popped heads had given a warning. The heads were nom-
inally inspected every five years, and several had been found popped on
the first tendon inspection. Functional performance could be demon-
strated by measuring prestressing tendon lift-off forcea lift-off test.
433
This constituted on-condition maintenance. After the second event, we

embarked upon a mad rush to develop diagnostic techniques for the
button head corrosion monitoring that did not require liftoff of the
prestressing tendon. Liftoff of a single tendon required the location and
installation of a short stroke pancake jack, shims, and techniques to
unload the tendon and hardware. Once unloaded, they could be inspect-
ed. Performing one liftoff occupied a three-man crew an average time of
two shifts. We were performing many lift-offs.
Corrosion developed in the area of wire within 10 inches of the
anchor hardware. Most of it was located just under a heavy anchor
plate, which acted as a cold trap. One diagnostic technique was to use
NDE ultrasonic exams to monitor the acoustic sound reflection from
the buttonhead. It could detect whether or not there were button heads
in good condition. Tuned to look only for those heads that exhibited
corrosion within five inches of the button, the instrument would detect
buttons with complete corrosion nearly 100 % of the time. The corro-
sion process was incomplete, so many wires had partial or slowly devel-
oping corrosion. They would give weak reflectionssometimes none
at allwith a wire that exhibited surface corrosion. Ergo, the test was
not perfect, but it would detect complete failure with certainty if that
failure occurred within five inches of the end hardware. As one who did
much of the testing, I was confident that we had a method that would
detect failed tendons with high accuracy. I felt we had a test that would
save us thousands of dollars in liftoff tests. Liftoff testing of the acces-
sible tendons had been our only recourse. Unfortunately, this was a
nuclear plant. Our Quality Control and Quality Management voided
our test on the basis that it was not perfect. It could not prove that but-
tons were in good condition; it could only detect failed buttons.
Therefore, we continued to perform liftoffs at a rate of two to three per
week. This kept a team of three mechanics busy doing liftoffs for a year.
Did this test meet applicability criteria? Effectiveness was not an
issue, if it could be used! The time to remove and inspect one tendon
head was about two man-hours; one to remove the cover, another to
perform ultrasonic inspection of 144 buttonheads. In contrast, it took
48 hours to lift-off and visually inspect. Therefore, we had an imperfect
test in a nuclear environment and a charter that sought perfection. We
434
Further Readings
continued doing liftoffs until the plant was permanently shut down for
high cost. Developing effective on-condition/condition directed main-
tenance pairs is challenging work. It can depend greatly on the plant
regulatory and cultural environment for its success. Static, regulated
environments will not be conducive to any new techniques or methods.
They are far too demanding!
Statistical Process Control (SPC) Systems

Process control is a sensitive indicator of overall system health. A
process that is in control can be verified by performing a series of
quick, statistically-based, running-average checks. Statistical process
control (SPC) provides a wide range of literature addressing tech-
niques to identify and measure process control. Most plant people can
look at controller output signals and make momentary judgments as to
whether or not a process is controlling. This is an intuitive check.
Knowledge of SPC can make monitoring more intelligent and useful.
Other processes become candidates for control assessment. We may
not realize that different processes exhibit controls, and that a problem
originates from a secondary process or system input.
Failure of a process to be in control is statistically easy to deter-
mine. A process that is out of control has other problems that need
identification and correction. Determining when a process is out of
control requires control limits. For example, it is common to find that
operators adjust controller setpoints for personnel preference. This
results in processes that are statistically out-of-control. While this
subject is of paramount importance in manufacturing, it also applies in
power generation.
For example, different consequences resulted when operators pref-
erentially loaded along the generator excitation-loading D curve.
Adjustments eventually percolated throughout the unit, establishing a
new equilibrium. In another instance, we found that different operators
had different control schemes for boiler sootblowing. Some were more
effective than others based on blowing times, boiler differential, and
temperatures at a given loading and firing rate. Operator preference
was an expensive choice, for this boiler became plugged when slag and
ash buildups ran away. The stations philosophy though was that the
435
twenty blowing schemes assured a clean boiler. Therefore, this operat-

ing practice was preferred, despite the evidence of SPC.
Typically operators are unaware of the broader ramifications of
their operating styles and the impacts this can have on equipment relia-
bility and aging. Some may be beneficial, while most are not. SPC pro-
vides a technique to view specific controlled system outputs and evalu-
ate their condition. SPC identifies systems that are becoming unstable
well before they fluctuate out of specification, asserting its usefulness as
a powerful predictive tool. As with all predictive tools, it is only cost-
effective when applied judiciously. Broad-based SPC application with-
out cost-effectiveness consideration is inappropriate.
436
RCM 437-476.qxd 3/3/00 3:04 PM Page 437
RCM Software Applications
Courtesy: Item Software, Inc., Anaheim, CA
Fault Tree Analysis: FaultTree+ provides an example of a

Windows-based engineering reliability software tool. In the past, engi-
neers avoided exact reliability analysis partly due to the tedious nature
of developing manual fault trees. Software like FaultTree+ simplifies
analysis so that relatively large fault trees 6000 elements and upcan
be developed and analyzed with relative ease. Lookup tables of stan-
dard component failure rates are combined with analytical simplifying
techniques. These make fault tree analysis a real option for the shop-
floor engineer. The fault trees themselves focus efforts on fault areas of
concern from the minimum cut sets of interest (and non-interest),
to the overall contributions that various faults have on overall reliabili-
ty. (A cut set is one way the top event can occur. The top event is typ-
ically an undesired outcome a failure.) The locomotive example here
is typical; the fault tree illustrates design areas not previously considered
in the overall unit reliability. Sensitivity analysis can test the numerical
values used for the event probabilities, as well as the model logic itself.
437
RCM 437-476.qxd 3/3/00 3:04 PM Page 438
Often overall top event reliability is known, but the individual relia-
bilities are not. Individual reliabilities may be taken from generic tables.
In any event, the Fault Tree identifies the fault paths of interest, and their
logic, providing the opportunity to focus on the critical few that matter.
In the example, we build the fault tree on FaultTree+, assigning fail-
ure data as we go. We have several selections of failure models and
information to select from. Once complete, we can run the analysis and
see whether the frequency of occurrence of the top event here,
engine failure to loadfits our experience. Very often it doesnt, but
we now have specific guidance on where to look for additional data.
The fault tree thus supports the continuous evolution of a maintenance
and operating plan based on facts. For users who have never used fault
trees, they help understand the complexity of multiple failure data and
help focus efforts in selective areas for maximum results. Its common
for a fault tree model of a problem to draw out risk areas not previous-
ly appreciated.
Courtesy : Item Software, Inc., Anaheim, CA
Failure Modes and Effects Criticality Analysis: FMECA provides a

failure mode and criticality analysis tool. The difference between
Failure Modes and Effects Analysis (FMEA) and FMECA is the calcu-
lation of criticality a numerical calculation of the combination of
probability and consequences (criticality) that allows the ranking and
438
RCM 437-476.qxd 3/3/00 3:04 PM Page 439
Courtesy : Item Software, Inc., Anaheim, CA

focus on failure categories of interest. The traditional weakness of
FMEAs is the inclusion of the trivial few, often over-ranked, to the
point where analysis isnt cost-effective. The addition of criticality can
avoid this problem. Practically, the engineer must learn by rote memo-
rization what typical probabilities are and apply these with speed and
impact. Contrary to what some say, numbers should always be based on
data, or validated. Those available or derived from seat-of-the pants
estimates are almost always exceptionally conservative in plant applica-
tions. A failure modes and effects analysis documents the modes of
interest; the criticality hangs a number on the mode that allows relative
comparison. FailMode is an easy to use FMECA tool that also supports
later analysis in related ways. The equipment hierarchy, in particular,
can be reused many times for different analysis.
RCM Analysis: RCM can serve as a product design optimization
tool. Product optimization (as used in manufacturing) can view prod-
ucts from a life-cycle cost perspective, optimizing the combination of
initial, operating and maintenance costs. RCMCost provides a design
tool for the evaluation of products and development of suitable main-
tenance programs. Starting with an hierarchical model, FMEA is devel-
439
RCM 437-476.qxd 3/3/00 3:04 PM Page 440
oped from which costs (like criticality) can be developed. The common
file structure in Item Softwares RCM Cost and related products (Fail
Mode, Fault Tree+...) means that designers can develop a FMECA, per-
form fault tree analysis, and review critical failure modes together
during design. Alternative maintenance strategies can then be devel-
oped, explored, simulated, and optimized. Any combination can help
to develop a product manufacturers initial installation and recom-
mended scheduled maintenance program. Clearly, the same software
can be used by the end users (operating organizations) to review and
evaluate their maintenance practices, costs, and risks, and optimize their
maintenance programs. There are at least seven RCM software prod-
ucts available. Some support one or more applications more easily than
others. Software users must understand their own needs and then
explore the market alternatives.
440
RCM 437-476.qxd 3/3/00 3:04 PM Page 441
CMMS Example: Power FM

Courtesy: Asset Works, San Antonio, TX
and New Century Energies, Denver, CO
Splash Menu
Splash Menu. The startup or splash menu shows the major func-
tions offered by the CMMS and graphically suggests use of a mouse
the trademark of a GUI interface. Since many non-routine CMMS users
are not typists the GUI interface is essential for speed and convenience.
Note that most CMMS systems even today use traditional terminology
since this is what users know. (We could equally call PM Management
Scheduled Maintenance.) Users can view and update different areas
by controlled authorization. Most systems offer generous view only
data privileges but restrict updates to specific work areas.
Equipment Hierarchy. The hierarchy provides a convenient way for
plant workers and staff to quickly locate any equipment of interest for
the purpose of identifying, selecting, or reviewing work and related fail-
ures, resources, and costs. The hierarchy (in a GUI environment) dou-
441
RCM 437-476.qxd 3/3/00 3:04 PM Page 442
Plant Registar (System Level)

bles as what was once called the plant register. It relationally lists all
equipment that the plant anticipates may require work over the facilitys
life. The hierarchy should uniquely identify components down to the
level just above that where you expect parts replacement. For example,
if the tubes in a compressor aftercooler heat exchanger tube bundle are
the last items replaced, then the aftercooler heat exchanger should be
identified as a unique part in the compressor assembly.
Developing an appropriate level of detail is a fine art form that only
comes with experience. Too little and work cant be traced to failed
components; too much and the hierarchy becomes complex and diffi-
cult to query. A good compromise is that a moderate-sized single fos-
sil-fired generating unit or process facility should have between 500 -
2500 coded components. Over 10,000 should raise a concern for
excessive detail; under 250 raises concern for two little. The purpose
for coding any equipment (of course) is to identify and perform appli-
cable and effective maintenance. This hierarchy was selectively opened
442
RCM 437-476.qxd 3/3/00 3:05 PM Page 443
by drilling down with

the mouse to the area of
interest. Folders with a
+ on the folder icon
contain embedded chil-
dren, so the user knows
where to find the details.
Trouble Reporting.
Condition-Monitoring
begins when trouble is
identified. Initiating
Trouble Reports (TRs)
has traditionally been a
primary means of initiat-
ing maintenance. In an
RCM context, the vast
majority of failures will
fall under the No
Scheduled Maintenance
Equipment Hierarchy (to find WOs) category and will be initi-
Trouble Reporting: TR-WO relational list
443
RCM 437-476.qxd 3/3/00 3:05 PM Page 444
ated as TRs. Success with TRs hinges on ranking the nature and
importance of the failures. Documented failures and planned actions
for important equipment equipment in the plants CMMS register
facilitate sorting through TRs quickly to (1) extract high-impact failures
that warrant high-priority, and (2) allow the option to pre-plan work,
and work pre-planned work on many NSM-type failures. A TR should
clearly identify a problem in the title not just a piece of equipment.
Since the TR title converts into supplemental documents like a work
order, the TRs title, initiator, plant impact and priority, reported date,
and work start target date need identification. Obviously, a planner and
scheduler will have to review and adjust the initiators request with over-
all plant schedule and resources.
Work Orders: Shedulers Active WO List
Work Orders. After screening, planning, and approval, TRs

become Work Orders (WOs). PM masters issue as Work Orders.
(They are prescreened and preplanned.) The CMMS scheduler an
internal clockchecks PM scheduling information against time and,
when scheduled, converts a PM master into real WOs. A WO on a
CMMS is an electronic document. By itself, a WO can do nothing. But
444
RCM 437-476.qxd 3/3/00 3:05 PM Page 445
Work Orders: CM Work Order

it is an electronic authorization to use resources and perform work.
This may be printed as hardcopy or left as an electronic authorization
to allow work to be performed and reported entirely through PC inter-
faces. A WO number generally authorizes time charges, parts usage,
and contractor support. Thus, a CMMS generally ties electronically
445
RCM 437-476.qxd 3/3/00 3:05 PM Page 446
Page 2 CM Work Order
into time reporting and payroll, purchasing, and inventory systems. On

our splash menu we see these as Time Management, Purchasing, and
Inventory. (Human Resources provides a register of employees that
supports Time Management.) WOs originate as (1) TRs, (2) PM mas-
ters, and (3) work originating from the Planning and Scheduling
Department. The last category includes design changes and in some
cases, informally scheduled outage PM work and condition-directed
work that isnt formally tied to a TR or PM. From a measurement per-
spective, the goal is to drive all WOs into their appropriate categories
for measurement & accounting purposes.
446
RCM 437-476.qxd 3/3/00 3:05 PM Page 447
WO Lists PM: WO Masters
447
RCM 437-476.qxd 3/3/00 3:05 PM Page 448
PM: Periodic Maintenenance Crew WO List
Scheduling: WO Scheduler
448
RCM 437-476.qxd 3/3/00 3:06 PM Page 449
PM WO - Coal Mills WO
PM Periodic Maintenance. A PM master (or standard) must be

available on file as the source for all scheduled maintenance work
orders. This may also provide a convenient repository of preplanned
work that is initiated on-demand in response to TRs or Condition-
Directed Maintenance. The PM attributes includes the work descrip-
tion, plan, resources, parts, and scheduling information. A PM master
is extracted and converted electronically into a work order by the
CMMSs subroutine scheduler or an on-demand issue now action,
much as a person used to make copies of a master form for data entry
using a copier in former times.
Scheduling. Work Scheduling reconciles the reality of resources and
time (to perform work) against the desires of the work requesters (to
have problems corrected). Scheduling is the hallmark of maintenance
effectiveness, so a convenient and simple scheduling system is critical to
overall work scheduling success. Power FM offers here a modified
Gantt Chart format that visually displays key scheduling information.
WO numbers and title can be displayed by GUI mouse features,
449
RCM 437-476.qxd 3/3/00 3:06 PM Page 450
PM: Periodic Maintenance PW WO with Route
WO status is displayed visually by color (Past Due, Delayed, Assigned,

In Progress, Completed, and Locked). In Progress and degree of com-
pletion are visually confirmed by the white completion line that displays
hours charged as a fraction of estimated workhours. Little features like
450
RCM 437-476.qxd 3/3/00 3:06 PM Page 451
PM: Periodic Maintenance WO
this allow anyone to quickly confirm (1) that work is in fact in progress,
and (2) that its had so many hours of time charged and presumably
worked. Work that is stalled or parked is also visually clear. Key infor-
mation for selected WOs is displayed on the same page without jump-
ing around.
WO/PM Lists. Lists can be generated by many sort orders to sup-
port any of a variety of standard plant work review activity: daily work,
outage work, scheduled work, skill category or department work.
These lists must provide the key information WO number, title, pri-
ority importance, crew ID, and scheduled completion datein any sort
order. For PM masters the priority is supplemented by WO type that
should roughly translate into RCM scheduling options Hard Time:
TBM, Scheduled Tests & Checks: OCM, On Demand and No
Scheduled Maintenance (preplanned): CDM. Other categories such as
Overhaul allow convenient grouping into scheduled work categories for
451
RCM 437-476.qxd 3/3/00 3:06 PM Page 452
Overhauls: Boiler 18-month Outage WO List
outage. Double clicking the top field of any column resorts the list by
that column, and a second time will reverse the sort order (top-to-bot-
tom goes to bottom-to-top).
Overhaul. Overhauls are special groups of work activity that are
issued as large groups of activity at the same time. Overhauls such as
six-year turbine tear-down and inspections can be developed as many
separate individual WOs for the many activities that must separately be
performed. These can then be tagged as a specific outage such as an
18-month boiler inspectionand issued as one single group (or be
issued based on selected items from the group). The overhaul group
then gets issued as a single clump of work. While the benefits to this
are not immediately apparent, let me personally attest that many hours
were spent in former days issuing many of the individual WOs that
made up large power plant outages as many as 1,000! The time sav-
ings and simplicity of this feature are tremendous. Typically, these activ-
ities are the same ones the plant wants to download for their Project
Management Software, to be able to schedule the outage in rote detail. The
452
RCM 437-476.qxd 3/3/00 3:06 PM Page 453
PM Periodic Maintenance: CEM PM Work Orders (by Crew)
PM Periodic Mainrenance: Crew, CEM, PM, Wos (All Selected)
453
RCM 437-476.qxd 3/3/00 3:06 PM Page 454
Routes: PM WO Route
software facilitates download for detailed outage planning.

Work Tasks. Work Descriptions can be added in text format to pro-
vide master preplanned jobs that can be pasted into a Scheduled PM
master or appended to a WO. Work Tasks are unique based upon work
crews. An IC technician work task would have little value for a
454
RCM 437-476.qxd 3/3/00 3:06 PM Page 455
Routes: PM WO Route (Continued)

mechanic. A task could be viewed as a simple procedure that can be
pasted onto a WO to provide guidance, rather than typing in the plan
on that specific WO. The tag is electronic, however, so the transfer only
occurs when the work is issued. When you statistically analyze the work
455
RCM 437-476.qxd 3/3/00 3:07 PM Page 456
Query: WO Query Selection
Query: Completed WO Query List
456
RCM 437-476.qxd 3/3/00 3:07 PM Page 457
Route: PM WO Route (for MOVs)
done in a large facility, you find it is highly repetitious (e.g., a few fail-
ure modes dominate), and the advantages of this feature to pre-plan
even on demand CDM and NSM-type failure work are tremendous!
The beauty of this feature is the capacity to standardize planned work
plans and revise the work plan for tens or even hundreds of equipment
PMs and WOs with one standard change.
457
RCM 437-476.qxd 3/3/00 3:07 PM Page 458
Hardcopy: Documents Site Vehicle PM WO
Routes. Routes are groups of repetitive PM activity performed

together on one WO. A lubrication route, for example, lists the lubri-
cation locations for one type of grease or lube oil in a given plant area.
It conveniently packages activity for streamlined performance. A cali-
bration route does likewise. A group of pressure transmitters with the
same calibration interval is calibrated with the same standard job plan
in repetitive sequence. The key to successful route performance is the
ability to group and perform many brief PMs together in sequence as
a group reducing trip time.
Query. Many query features are available to organize and sort infor-
mation. Pre-programmed queries (built into the basic software) pro-
vide several options for standard information displays. Special queries
can be created real time, used, and saved into libraries (where
desired). Information selection and display
Hardcopy Documents. When printed, hardcopy should clearly
illustrate the work scope and performance information easily, to facili-
tate work. Once complete, the required work performance summary
and time charges may be returned to a plant clerk for data entry, or (as
ever more common) be entered directly into the CMMS by the worker
as a job completion step.
458
RCM 437-476.qxd 3/3/00 3:07 PM Page 459
Routes: PM Route Developemnet
WO/PM Lists: Prioritized List of Combined CM/PM WOs
459
RCM 437-476.qxd 3/3/00 3:07 PM Page 460
RCM Software Examples: RCMtrim

Courtesy: ERE, Inc., Arvada, CO
Startup Menu. The startup menu defines the major RCM software
functions. Since users are often not typists, a GUI interface improves
speed and convenience. Different users can view or update different
areas. Since use is restricted to a small group of engineers and analysts,
control requirements are simpler than for a CMMS. View only use
includes interrogating the database for known hardware failures and
failure data, the failure bases (e.g., failure basis, plural), and strategies
addressing known failures. Systems should be at the highest level in the
database and be identifiable by general category such as fossil, nuclear,
or chemical process, and/or other general classification. Broad systems
classes such as control, service and power conversion should be avail-
able for later analysis and sorting.
Startup Menu
460
RCM 437-476.qxd 3/3/00 3:08 PM Page 461
Pull-down Menus and Finders: Pull Down Selection
461
RCM 437-476.qxd 3/3/00 3:08 PM Page 462
Pull-down Menus & Finders. Pull-down menus should provide

pre-defined selection values and a category based on lookup tables that
standardize and speed user data entry. All basic system types serv-
ice, production, safety, etc.should be included for easy reference with
the capability to quickly add more. Pull-down data selection and char-
acteristics for many data field attributes such as manufacturers, compo-
nent types, failure modes and symptoms, basis data source, and sec-
ondary failures should be provided. Users should be able to quickly
scan options and extract and enter data. System, equipment, and com-
ponent finders speed locating relevant equipment, information, and
location for modeled equipment in the database. Finders should also
facilitate jumping to other analysis data source areas and back to the
home information for detailed review and selectionquickly.
Standards: Air Operated Valves and Suppliers

System Functions & Instruments Hierarchy. Functions should
be provided beginning at the system level. The user should be able to
select a system to model or review, drill down to the systems functions,
identify their importance, the instruments and controls used to monitor
462
RCM 437-476.qxd 3/3/00 3:08 PM Page 463
System Functions and Instruments Hierarchy
System Functions and Instruments Hierarchy: Key Instruments and Limits
463
RCM 437-476.qxd 3/3/00 3:08 PM Page 464
Pull-down Menus and Finders: Browse Lists and Finders System Equipment
Computers Hierarchy
these functions, and identify instrument range limits. Function impor-

tance rank should be standardized with pull-down menus and scaled
based on the relative rank of each category. Functions should further
identify type. All systems in the database are readily available for quick
development reference, and the functions can be categorized based
upon RCM-based attributes such as safety, environment, production,
and cost.
Functions Zoom. Instruments that provide key monitoring and
safety information should allow further development for operations
impact and requirements. These important instruments require cali-
brations, channel, or alarm checks. Lists of similar instruments and
requirements based on functions should be available to be developed as
instrument calibration programs based upon importance.
System-Equipment-Components Hierarchy. Systems are broken
down by major pieces of equipment such as a sootblowing air compres-
sor assembly or a startup boiler feedpump. Equipment can vary from
464
RCM 437-476.qxd 3/3/00 3:09 PM Page 465
Components Parts Failure Modes

simple skids and associated components on a skid to complex trains and
subsystems. Abstract associations of equipment based on proximity,
interfaces, and joint functionality should also be allowed to develop as
the users needs require. A feedwater pump train, for example, could be
treated as a skid. User-defined fields allow special requirements to be
added for general purposes, such as nuclear environmentally qualified
equipment. Hot links embedded in peripheral Browse Lists speed the
identification and selection of useful development material from similar
systems. The model should emphasize similarity of equipment and sys-
tems, standardizing plans implicitly as the plant model is built.
Components-Parts-Failure Modes. For each component, parts can
be identified and described in terms of failure information. Failure
modes, causes, system impact (based on redundancy and standby/run-
ning applications), location, notes, and user-defined fields can uniquely
describe the parts and their affect on the component and systems based
on each failure. Fundamental failure characteristics can be classed sta-
465
RCM 437-476.qxd 3/3/00 3:09 PM Page 466
Groups and Routes
Data Copying: Components
466
RCM 437-476.qxd 3/3/00 3:09 PM Page 467
Pull-down Menus and Finders: Lookup Tables
tistically (random, age-based,) and by symptom, occurrence frequency,

age estimate, and description. Failures can be ordered by importance
rank to emphasize high-risk failures. Basic part failure information can
be quickly scanned and validated by workers, or reviewed and used by
operators or engineering. Secondary failures and failure causes are also
identified. Models are available at every level Systems, Equipment,
Component, and Partsand their failures are available for immediate
review using browse and hot link features.
Applicable models and their subordinate information can be
extracted from any level, and be incorporated or recreated as new sys-
tems and subordinate equipment conveniently. Users can capitalize on
what they already know, rather than bog down in endless creation of
new documents and files. Menus allow users to summarize all-impor-
tant monitoring equipment in the database. Operators can quickly
locate an instrument and determine its importance to plant functions.
Loss of function consequences are clearly identified, allowing the
operator to evaluate impact on unit operations. Support instrumenta-
467
RCM 437-476.qxd 3/3/00 3:09 PM Page 468
Detailed Information Part Failure Preventative Maintenance Tasks-Reccomended

PM Tasks.
tion maintenance managers and crews can quickly review instruments,

ascertain intended operational role, and review failures and mainte-
nance plans. Failure risks can quickly be determined based on operator
instrument usage, design functionality, and maintenance strategy. The
requirements of operator training programs can be identified to support
cost-effective, failure-based development of formal and informal on-
the-job training.
Part Failure-Preventive Maintenance Tasks-Recommended PM
Tasks. From the failure information (especially causes), applicable and
effective PM tasks can be identified. Tasks can be uniquely identified
by number, labeled for RCM-type (time-based, on-condition, condi-
tion-directed, condition-monitored), be ranked by cost-effectiveness
and risk management value, and provide performance information (fre-
quency interval and units). Where the primary maintenance activity
requires condition-directed maintenance on failure, the failure defini-
tion limits and CDM tasks can be clearly identified. Where condition-
468
RCM 437-476.qxd 3/3/00 3:09 PM Page 469
Data Copying: Systems
System Equipment Components Hierarchy
monitoring is the strategy selected, any condition-directed maintenance

requirements may also be specified with or without limits. When no
469
RCM 437-476.qxd 3/3/00 3:09 PM Page 470
Operators Rounds
scheduled maintenance is selected, the factual review of a failure mode

and its impact can be documented with no scheduled maintenance pro-
gram as the planned strategy. The review is available for engineering,
diagnostics or reassessment at a later time.
Recommended Preventive Maintenance Tasks and Frequencies may
also be identified based on origin, task(s), frequency, authority, and ref-
erences. Users that require detailed bases for all work done such as
nuclear power plantswill be able to provide this information. All
users will be able to explicitly document requirements such as ASME
470
RCM 437-476.qxd 3/3/00 3:10 PM Page 471
Reports
codes, EPA Title 5 emissions, and other requirements should they so
desire. When documented, users can see those aspects of their sched-
uled maintenance program that are based on the force of law. Every
PM task has task resources necessary to perform the task identified.
This includes work classification, department, work hours, travel and
slack time. PM tasks can be grouped at the component, and part fail-
ure level to arrange packaged activity that can be performed as conven-
ient work packages.
Data Copying. Data copying subroutines allow the user to select,
extract, and apply failure data at many levels and copy large chunks of
pre-existing systems up to and including the entire system itself
into new or existing models. A process that avoids recreating basic
engineering and failure data available elsewhere for the multitude of
replicated components, equipment, and even systems present in a large
industrial facility can be used to develop plans quickly. Data copying
allows rapid similarity modeling at multiple levels.
471
RCM 437-476.qxd 3/3/00 3:10 PM Page 472
Part FailurePreventative Maintenance TasksReccomended PM Tasks
Reports
472
RCM 437-476.qxd 3/3/00 3:10 PM Page 473
Standards. A special set of equipment, component, or even system

standards is set up for one special purpose providing a library of
existing components and their failure modes, applicable PM tasks, sup-
pliers, and supplier or other recommended programs. The standards
should be able to jump-start the creation of a new system model from
scratch, providing a ready source of component failure data. Further,
the standards should provide the option for a facility to tailor their
maintenance strategy to reflect their predominant failure modes and
experience in their facility, based on their particular methods of opera-
tion and environment.
Detailed Information. The database ideally offers the opportunity
to capture and reuse detailed equipment information such as manufac-
turer-specific failure and maintenance strategies without starting over
for each new component type. Characteristics of different equipment
types should be readily available to easy application. Design similarities
with fleets or product evolutionary cycles can be captured.
Operators Rounds. Operator rounds can be developed as hard-
copy printouts or exportable files for the performance of rounds.
Operator rounds provide information about the monitoring strategy
that helps operators perform effective rounds and understand rational
behind monitoring intervals.
Reports. A variety of useful management, maintenance, operations,
and training reports should be available based upon systems, their func-
tions, work classifications, and risk. Cost management reports should
support maintenance strategy adjustments to reflect experience and
costs. Management should easily be able to document, maintain and
use the basis rational for the operating maintenance strategy. The infor-
mation should provide a useful tool to perform comparison analysis for
parts, manufacturers, and applications to derive future cost reductions,
performance improvements, and other benefits. The plan should inte-
grate operations monitoring, maintenance rework and replace tasks,
and engineering technical assessment. Useful derivative products
such as the calibration schedule, time-based maintenance plan, and
lubrication schedulesshould be available from the basic plant model.
Export/Import Routines. Predefined routines should allow export
of completed maintenance and monitoring strategies to CMMS and
473
RCM 437-476.qxd 3/3/00 3:10 PM Page 474
Groups and Routes
operating rounds management systems. Hardcopy results of the same

information should also be available to support training and review.
Existing facilities should be able to download existing equipment files
and PM plans as delimited files, have these loaded into the PMO data-
base to provide a starting point, and have finished plan results up-load-
able to the database as complete, organized tasks in file format. Little
or no data re-entry should be required to accomplish this transfer.
Groups & Routes. Work should be easily organized into groups
such as overhauls, surveillance tests, or special work activity that can be
issued at any time. Those planning work should have the option to de-
construct and re-plan maintenance activity in many different ways
based on work that mitigates failure. The software should be a flexible
work-planning tool. The activities the plant would like to download for
their Project Management Software, such as outage work, should be
available in exportable file format. The software should facilitate out-
age vs. online work and risk analysis. Routes of repetitive light PM
activity that can be performed together should be available.
474
RCM 437-476.qxd 3/3/00 3:10 PM Page 475
Lubrication, calibrations, and even operating inspections should conve-

niently package activity for streamlined performance. Since a key to
successful route performance is grouping and performing many brief
PMs together in sequence reducing trip time, the software should facil-
itate this use.
475
RCM 437-476.qxd 3/3/00 3:10 PM Page 476
references 477-480.qxd 3/3/00 3:11 PM Page 477
References
1. F.S. Nolan, H.L. Heap, et al, Reliability-Centered Maintenance,
United Airlines, San Francisco, CA, Dec. 1978 NTIS AD/A066579
2. Reliability Centered Maintenance (edited summary of Nolan &
Heap), R. Keith Young, MQS, PdMA , Millersville, MD, 1996
3. Smith, A.M., Reliability-Centered Maintenance, McGraw-Hill, New
York, NY, 1993
4. Scherkenbach, W.W., The Deming Route, CEE Press Books,
George Washington University, Washington, DC, 1990
5. Ishikawa, Ki., What is Total Quality Control? The Japanese Way
(Translated by D.L. Lu), Prentice-Hall, Inc., Englewood, NJ, 1985
6. Bloch, H.P., Geitner, F.K. Machine Reliability Assessment, Van
Nostrand Reinhard, New York, NY, 1990
7. Tajiri, M., & Gotoh, Total Productive Maintenance
ImplementationA Japanese Approach, McGraw-Hill, Inc., New
York, NY, 1992
8. Rao, S., Reliability-Based Design, McGraw-Hill, New York, NY,
1992
9. Kececioglu, D, Reliability Engineering Handbook Vol. 1-2, Prentice
Hall, Englewood Cliffs, NJ, 1991
10. Ireson, W. Grant, et al, Handbook of Reliability Engineering and
Management, McGraw-Hill, 1988
11. Equipment Maintenance Optimization Group Meeting Minutes,
ERPI Boston, MA, 1995
12. RCM Handbook, EPRI, 1994
13. Reliability Centered Maintenance Implementation, EPRI NDE
Center, Charlotte, NC, RCM Maintenance Training, Nov. 1993,
(NUS)
14. RCM for Substations Technical Reference, EPRI NUS
Gaithersburg, MD, June 1996
15. RCM Proceedings RCM for Substations Conference, EPRI Cambias
& Associates, August 1996
16. Predictive Maintenance Primer, NMAC EPRI (NUS), Palo Alto,
477
CA, April 1991

17. RCM Generic Applications Guide, EPRI, Feb. 1991
18. Demonstration of RCM, Project Description & Results, VI, Schwan,
et al, EPRI NUS, April 1991
19. Demonstration of RCM, First Annual Progress Report, V3.,
Anderson, J, et al, EPRI Southern California Edison/ERIN
Engineering, April 1991
20. Demonstration of RCM, Final Report of San Onofre NGS, V3.,
Betros, et al, EPRI Southern California Edison/ERIN Engineering,
April 1991
21. Proactive Operations and Maintenance Workshops
22. Root Cause Analysis Workshops, PSCo Improvement Technology
Group, 1996, Denver, CO
23. Process Improvement Technology, Brunetti, Wayne, 1994
24. Reliability Centered Maintenance, MQS, R. Keith Young, 1996,
PdMA, Millersville, MD
25. Code of Federal Regulations 10CFR50.65, Part 10 Section 50.65
The Maintenance Rule
26. Industry Guideline for Monitoring the Effectiveness of Maintenance
at Nuclear Power Plants, Nuclear Management and Resource
Council, Inc. (NUMARC), May 1993
27. Inspection Procedure 62002, 50.65 Maintenance Rule Inspection,
NRC Inspection Manual. US Nuclear Regulatory Commission
28. Inspection of Maintenance Rule Implementation at Nuclear Power
Plants Inspection Procedure 62706, Implementation, NRC Inspection
Manual, US Nuclear Regulatory Commission
29. Inspection of Structures, Passive Components, and Civil Engineering
Features at Nuclear Power Plants Inspection Procedure 62007,
Maintenance Observation, NRC Inspection Manual, US Nuclear
Regulatory Commission
30. NRC Information Notice 97-18: Problems Identified During
Maintenance Rule Baseline Inspections, US Nuclear Regulatory
Commission, April 14, 1997
31. The Maintenance Rule Presentation and Interpretation, American
Nuclear Society Winter Annual Meeting, Albuquerque, NM, Nov
16, 1997
478
References
32. Understanding Reliability Centered Maintenance: A Practical

Guide to Maintenance, Second Edition 1998, Jack Nicholas, R.
Keith Young, MQS, Millersville, MD 21108
33. Moubray, J. Reliability Centred Maintenance, Industrial Press,
New York, NY, 1992
34. In Search of Excellence, Peters, T.J., Waterman, R.H., Harper and
Row, Cambridge, MA, 1982
35. Faults & Failures: The Auckland Outage, Sweet, W., IEEE
Spectrum, April 1998, p.72
36. Auckland Unplugged: The Story of a Blackout, Ackermann, T. &
Muller, D., Electric Light and Power, Nov 1998, p. 20-23
37. Coal Handling Maintenance Optimization at Pawnee, Rohde, S. &
August, J., P/PM Technology, October 1996
38. Damn the Torpedoes! Hit or Miss (American Submarines entered
WWII armed with dangerously unreliable torpedoes that took
almost two years to fix), Murphy, D., American Heritage of
Invention and Technology, Spring 1998, Vol. 13, Num. 4, p. 56-63
39. Auckland: City in Crisis, Internet, www.nzwires/crisis, 2/09/98-
4/28/98
40. Proceedings, Predictive Maintenance Technology National
Conferences, 1996, 1998
41. Notes, Poke Yoke, Fury Enterprises, Dallas, TX
42. Metteson, Gene, The Air Transportation Industry-Birthplace of
RCM, (presentation & paper), RCM for Substations Conference,
Dec 4, 1995, Newport Beach, CA
43. Netherton, Dana, Standard to Define RCM, Maintenance
Technology, pp. 17-24, June 1999
44. The Deacons Masterpiece, Oliver Wendell Holmes, McGraw-Hill,
1965
479
Index 481-500.qxd 3/3/00 3:12 PM Page 481
Index
A
Accuracy (instrumentation), 243-248
Acronyms, xi-xiv
Age exploration, 6, 178-180, 199-205, 307, 338, 426-427:
definition, 199-201;
value, 201-204;
systematic application, 204-205
Aging analysis, 47, 270, 338, 405-407
Alarms, 242-243
Alternative solutions, 349-350
Ambiguity, 146-147
Analysis software, 312-315
Applicability criterion, 31-32, 122-124
Applications software, 304
Applications (RCM), 12-14, 161-193, 344-350:
overview, 161-165;
engineering, 165-173;
integration of functions, 173-185;
safety, 185-187;
case histories, 187-193;
statistical maintenance, 346-349;
alternatives, 349-350
Area checks, 238-240
Areas not worked online, 292-293
As low as reasonably achievable, 396-397
Assessing programs, 122-125:
applicability, 122-124;
cost effectiveness, 124-125
Assumptions, 31-37:
applicability, 31-32;
effectiveness, 32-33;
481
Index 481-500.qxd 3/3/00 3:12 PM Page 482
PM, 33;
statistics and regulators, 33-37
Availability simulation, 172
B
Backlogs, 279-280, 282-284:
work order, 282-284
Basis history (equipment), 91
Black-box model, 156
Blocking tasks, 89-90
Bootstrapping, 229-231
Budgeting, 407-409
C
Case histories, 63-70, 187-193, 252-253:
maintenance practices, 63-70;
soot-blowing air compressor, 64-65;
turbine blade, 65-66;
generator retaining ring, 66-68;
coal belt fire, 69-70;
circulating water tower, 187-190;
maintenance options, 191-193
Casual use, 232-235:
critical failure modes, 233;
wearout, 233-234;
confusion implications, 234-235
Changes and measures (output/response), 327-333:
measure types, 328-329;
system measures, 329-332;
failure, 332-333; costs, 333
Checklist (maintenance health), 412
Checklist (round), 317-318
Circulating water tower, 187-190
Clock-based PM, 56-57
482
Index 481-500.qxd 3/3/00 3:12 PM Page 483
References
CMMS, 218-221, 256-258, 304-307, 318, 441-459:

barriers, 218-220;
software, 304-307;
integration, 318; example, 441-459
CMMS example (Power FM), 441-459
CMMS software, 304-307:
maintenance process, 304-305;
custom products, 305-306;
system level measurement, 306;
age exploration, 307
Coal belt fire, 69-70
Coded components, 141
Coding levels (software), 303-304
Common mode failures, 144-147:
keep it simple stupid, 146;
ambiguity, 146-147
Company roles, 129-130
Comparison analysis, 104-105
Complex failures, 142-144
Complexity (failure), 141-144, 228-229
Component failure, 156
Components hierarchy, 464-465
Condition monitoring, 6, 153-154, 415-420
Condition-based maintenance, 6, 42, 153-154, 226-228, 392-393,
401-405, 415-420
Condition-directed maintenance, 226-228, 418-420
Configurations simulation, 319
Confusion implications, 234-235
Consequences, 185-187
Conservatism, 134-139:
over-conservatism, 135-139
Consistency (maintenance), 29-37:
statistics, 30-37;
assumptions, 31-37
Consistency (parts), 181-182
Corporate maintenance management system. SEE CMMS.
483
Index 481-500.qxd 3/3/00 3:12 PM Page 484
Corrective maintenance, xv, 276

Cost effectiveness, 32-33, 124-125
Cost measurement, 338
Cost analysis, 73-81, 117, 323-327, 333, 335-338, 409-411
Cost criteria, 73-81
Costs and layers (redundancy), 248-249
Costs and rank, 276-278
Costs, 32-33, 63, 124-125, 248-249, 323-327, 333, 335-336, 420-421:
maintenance, 63, 323-327, 335-336;
life-cycle, 420-421
Critical failure, 231-235:
RCM definition, 231-232;
casual use, 232-235
Critical instruments, 243-248
Criticality, 19, 78, 95-96, 171-172, 231-235, 243-248, 438-439
Culture, 7-8, 148-153, 256:
maintenance delivery, 148-149;
maintenance performance, 150-151;
equipment groups, 151-153
Custom products (software), 305-306
D
Delivery (maintenance), 148-149
Delivery (operations), 115-116
Department goal balancing, 163:
value added, 163
Design basis, 116-118
Design-change maintenance, 21
Development steps, 294-295
Development (RCM), 2-4
Development (systems approach), 116
Do your best (strategy), 209-211
Documentation (software), 315-317
484
Index 481-500.qxd 3/3/00 3:12 PM Page 485
References
E
Effectiveness measures, 6, 335, 337
Emergency maintenance, 61, 335
Engineering applications, 165-173:
operations-organizational
relationship, 165-167;
plant support roles, 167-169;
plant modification, 169-171;
tools, 171-173
Engineering focus, 205-209:
failure spectrum, 206-209
Engineering maintenance, 43
Engineering reliability, 15-18
Engineering support role, 163-164
Engineering tools, 171-173:
failure modes and effects criticality analysis, 171-172;
fault trees, 172; availability simulation, 172;
Weibull analysis, 172-173
Engineering-specified failure, 429
Entropy (organizational), 341-343
Environment maintenance, 422-426
Equipment groups, 151-153, 268-272, 287-299:
development steps, 294-295;
types, 296-298;
operations, 298-299;
modification reviews, 299
Equipment hierarchy, 81-99:
level, 81-83;
failure descriptions, 83-88;
blocking tasks, 89-90;
PM tasks/vendors, 90-91;
basis history, 91;
PM work packages, 91-92;
information sources, 92-94;
485
Index 481-500.qxd 3/3/00 3:12 PM Page 486
failure mode/mechanism/cause, 94-95;

criticality, 95-96;
practical difficulties, 96-97;
fault tree analysis, 98;
hierarchy and boundary, 98-99
Equipment register, 238, 291, 441-443
Equipment selection, 73-81:
plant system units, 74-76;
plant system functions, 76-81
Equipment size, 236
Establishing process, 127-129
Event analysis, 285-286
Examples (practice), 63-70, 187-193, 252-253
Examples (software), 441-475
Expedite, 266-267
F
Failure analysis, 6, 10-11, 85, 312-315, 332-333
Failure complexity, 421-426
Failure description, 83-88, 159
Failure footprints, 218-221: CMMS, 218-221
Failure frequency, 140-142:
coded components, 141;
complexity, 141-142
Failure identification, 6, 173-176, 430-432
Failure management, 157-158
Failure modes, 6, 94-97, 171-172, 233, 438-439, 465, 467-468
Failure modes and effects analysis, 6, 94
Failure modes and effects criticality analysis, 94, 96-97, 171-172, 438-
439
Failure numbers, 184-185
Failure perspectives, 154-160
Failure reports, 236-237
Failure spectrum, 28-29, 206-209
Failures, 34, 61, 83-84, 218-226, 249-250, 332-333. SEE ALSO Case
486
Index 481-500.qxd 3/3/00 3:12 PM Page 487
References
histories.
Fast track maintenance, 255-299:
CMMS, 256-258;
maintenance infrastructure, 258;
traditional programs, 258-260;
scheduling, 260-266;
scheduling methods, 266-272;
project management techniques, 272-274;
overhaul intervals, 274-279;
PM reviews, 279-284;
outage work review, 284-287;
equipment groups, 287-299
Fault tree analysis, 6, 98, 172, 437-438
Focused measurement, 324-327
Function integration, 173-185
Functional elements (PM), 56-63:
time-based (clocks), 56-57;
operational-based (surveillance), 57-58;
operate to failure, 58;
preplanned failure, 58;
no scheduled maintenance, 58;
measurement, 58-59;
overhaul, 59-61;
emergency maintenance, 61;
overtime, 61;
failures, 61;
maintenance rule, 62-63
Functional failure, 156, 417-418:
measurement, 417-418
Functional reviews (RCM), 99-102:
history, 100-102
G
Generating units, 74-76
Generator retaining ring, 66-68
487
Index 481-500.qxd 3/3/00 3:12 PM Page 488
Global measurement, 321-324

Glossary, 353-389
Goal balancing, 163
Goals (software), 301-302
Group types (equipment), 296-298
H
Hard time, 6
Hardcopy documents, 458
Hierarchies, 81-99:
level, 81-83
Hierarchy (software), 302-304:
coding levels, 303-304;
standardize, 304;
applications, 304
Hierarchy and boundary, 98-99
I
Implementation, 20-22, 209-216:
value added, 20-21;
maintenance strategy, 21-22;
models, 209-216
Implementation models, 209-216:
do your best, 209-211;
trust us, 212-213;
typical implementation, 213-214;
total performance, 214-216
Importance (criteria), 235-238:
equipment size, 236; failure reports, 236-237;
work frequency, 237;
vendor recommendations, 237;
industry practice, 237;
shop practice, 237-238;
equipment registers, 238
488
Index 481-500.qxd 3/3/00 3:12 PM Page 489
References
Industry practice, 237

Informality, 55-56
Information sources, 92-94
Infrastructure, 258
Instrumentation, 36, 163-165, 240-248, 251-252:
and control, 163-165;
spurious alarms, 242-243;
critical instruments, 243-248;
accuracy, 243-248;
utility cultures, 251-252
Instruments hierarchy, 462-464
Integration of functions, 173-185:
operations roles, 173-178;
parts, 178-184;
failure numbers, 184-185
Intelligence role, 341-343
Interval extension, 426-427
K
Keep it simple stupid, 146
L
Legitimate failure, 225-226
Lessons learned, 195-253:
task intervals, 197-199;
age exploration, 199-205;
engineering focus, 205-209;
implementation models, 209-216;
vendor perspective, 216-218;
failure footprints, 218-221;
operate to failure, 221-231;
no planned maintenance, 221-231;
critical failure, 231-235;
importance criteria, 235-238;
489
Index 481-500.qxd 3/3/00 3:12 PM Page 490
area checks, 238-240;

instrumentation, 240-248;
redundancy, 248-250;
instruments in utility cultures, 251-252;
case history, 252-253
Life-cycle cost, 420-421
Life-cycle maintenance, 44-45
Life-cycle vision, 54-56:
informality/chance, 55-56
Logic tree analysis, 6
Long term schedule, 267-268
M
Maintenance budgeting, 407-409
Maintenance cost, 63, 323-327, 335-336:
maintenance hour, 335-336
Maintenance delivery, 148-149
Maintenance discipline, 432-433
Maintenance infrastructure, 258
Maintenance options (RCM based), 191-193, 349-350:
no scheduled maintenance, 191-193;
on-condition maintenance, 191;
time-based maintenance, 191-192
Maintenance performance, 41-42, 126, 150-151, 214-216
Maintenance perspective (RCM), 1-22:
precursors, 1-2;
development, 2-4;
origin, 4-5;
reliability perspective, 507, 14-19;
post World War II, 8-10;
traditional RCM, 10-12;
applied RCM,12-14;
implementation, 20-22
Maintenance practices, 23-70:
options, 24-29;
490
Index 481-500.qxd 3/3/00 3:12 PM Page 491
References
consistency, 29-37;
maintenance process, 37-44;
PM, 44-63;
costs, 63;
case examples, 63-70. SEE ALSO Case histories.
Maintenance process, 6, 37-44, 105-111, 127-129, 297, 304-305,
335, 337-340, 412-415:
plan, 40-41;
schedule, 41;
performance, 41-42;
training, 42-43;
engineering, 43;
definition, 43-44;
model, 127-129;
software, 304-305
Maintenance process measures, 335, 337-340:
effectiveness, 335, 337;
responsiveness, 337;
total hours/system, 337;
trends, 337;
aging studies, 338;
costs, 338;
ratios, 338-339;
rework, 339;
screening for effectiveness, 339-340
Maintenance process model, 127-129
Maintenance process software, 301-320:
goals, 301-302;
hierarchy, 302-304;
CMMS software, 304-307;
RCM software development, 307-312;
analysis, 312-315;
documentation, 315-317;
products, 317-319;
configurations simulation, 319;
simplicity, 319-320;
491
Index 481-500.qxd 3/3/00 3:12 PM Page 492
policy, 320
Maintenance ratios, 338-339
Maintenance rule, 62-63
Maintenance strategy, 21-22
Management oversight risk tree, 6
Mean time between failures, 25
Measure types, 328-329
Measurement, 58-59, 321-327:
PM functions, 58-59;
output/response, 321-327;
global, 321-324;
focused, 324-327
Measures (output/response), 321-340:
measurement, 321-327;
changes and, 327-333;
PM hours, 334-335;
maintenance process, 335, 337-340
Mechanisms of failure, 94-95
Missed PM, 394-395
Models (PM), 46-48
Modification reviews, 299
Monitoring, 120-121, 176-178, 208, 238-240, 392-393, 418-420
N
Needs awareness, 116-118
No planned maintenance, 221-231:
failure, 221-223;
RCM environment, 223-225;
legitimate failure, 225-226;
condition-based maintenance, 226-228;
complexity in failures, 228-229;
bootstrapping, 229-231
No scheduled maintenance, 58, 191-193
Nuclear energy generation, 131
492
Index 481-500.qxd 3/3/00 3:12 PM Page 493
References
O
Obsolescence, 8
On-condition maintenance, 6, 191, 265
Operate to failure, 8, 58, 221-231, 397-399, 428-432:
failure, 221-223;
RCM environment, 223-225;
legitimate failure, 225-226;
condition-based maintenance, 226-228;
complexity in failures, 228-229;
bootstrapping, 229-231
Operational-based PM, 57-58
Operations (equipment groups), 298-299
Operations overview, 161-163
Operations roles, 173-178:
failure identification, 173-176;
operator monitoring, 176;
rounds optimization, 176-178
Operations-organizational relationships, 165-167
Operator monitoring, 176
Operator training, 420-421
Options (maintenance), 24-29
Organization of activity, 318
Organizational relationships, 165-167
Origin (RCM), 4-5
Outage, 16, 31, 51, 120, 272, 284-287:
intervals, 16;
work review, 284-287;
parts and, 286-287
Outage work review, 284-287:
event analysis, 285-286;
parts and outages, 286-287;
strategy development, 287
Over-conservatism, 135-139
Overhaul, 59-61, 110, 200, 274-279, 401-405, 452, 454:
493
Index 481-500.qxd 3/3/00 3:12 PM Page 494
basis, 274-275;
intervals, 274-279;
conditional, 404-405
Overhaul basis, 274-275
Overhaul intervals, 274-279
Overhaul schedules, 274-279:
basis, 274-275;
optimizing strategy, 275;
planning, 275-276;
costs and rank, 276-278;
standards, 278-279
Overtime, 61
P
Pareto system cost analysis, 117, 409-411
Part aging dispersion, 405-407
Part failure, 156, 286-287
Parts, 156, 178-184, 286-287, 405-407:
failure, 156, 286-287; age exploration, 178-180;
integration, 178-184;
stocking levels, 180-181;
consistency, 181-182;
problems, 182-184;
troubleshooting, 183-184;
and outages, 286-287;
aging dispersion, 405-407
Parts and outages, 286-287
Parts integration, 178-184
Performance (RCM), 41-42, 71-111, 126, 150-151, 214-216:
equipment selection, 73-81;
equipment hierarchy, 81-99;
functional reviews, 99-102;
standards, 102-104;
comparison analysis, 104-107;
maintenance process, 105-111
494
Index 481-500.qxd 3/3/00 3:12 PM Page 495
References
Periodic maintenance, 449-451

Perspectives (PM), 48-53
Planning, 40-41, 275-276
Plant modification, 169-170
Plant needs, 113-160:
production and delivery, 113-116;
systems approach, 116-122;
assessing programs, 122-125;
process improvement, 125-131;
PM bases, 131-147;
culture, 148-153;
condition monitoring, 153-154;
failure perspectives, 154-160
Plant support roles, 167-169
Plant system functions, 76-81
Plant system units, 74-76
PM, 33, 44-63, 90-92, 131-147, 279-284, 311-312, 318, 334-335,
468-471:
models, 46-48;
perspectives, 48-53;
triggers, 53-54;
life-cycle vision, 54-56;
functional elements, 56-63
PM bases, 131-147, 311-312:
conservatism, 134-135;
over-conservatism, 135-139;
failure frequency, 140-142;
complex failures, 142-144;
common mode failures, 144-147
PM hours, 334-336:
real hours, 334-335;
maintenance hour cost, 335-336
PM reviews, 279-284:
backlog, 279-280;
worklists, 280-281;
work order backlog, 282-284
495
Index 481-500.qxd 3/3/00 3:12 PM Page 496
PM tasks, 90-91, 318, 468-471:

vendors, 90-91
PM work packages, 91-92
Policy (software), 320
Practical problems, 96-97
Precursors (to RCM), 1-2
Predictive maintenance, 9-10
Preplanned failure, 58
Prestressing (tendon buttonheads), 433-435
Preventive maintenance, SEE PM.
Prioritization, 262, 276-278, 392-393
Probability, 17-18, 95
Problems (parts), 182-184
Process change, 321-322
Process control, 435-436
Process improvement, 125-131:
establishing process, 127-129;
company roles, 129-130;
nuclear generation, 131
Process reliability, 18-19
Process standardization, 307-311
Production and delivery, 113-116
Products (software), 317-319:
round checklists, 317-318;
PM tasks, 318;
organization of activity, 318;
CMMS integration, 318;
RCM/CMMS idealization, 318-319
Project management techniques, 272-274:
working to schedules, 273-274
Q
Query features, 458
496
Index 481-500.qxd 3/3/00 3:12 PM Page 497
References
R
Random failure, 25-26, 429
Rare failures, 249-250
Ratio measurement, 338-339
RCM analysis, 307-312, 439-440:
software development, 307-312;
software, 439-440
RCM definition, 231-231
RCM environment, 223-225
RCM software development, 307-312, 439-440:
process standardization, 307-311;
task basis, 311-312;
example, 439-440
RCMtrim (tm), 460-475
RCM/CMMS idealization, 318-319
Readings in RCM, 391-436
Real hours, 334-335
Redundancy, 35, 248-250, 417-418:
costs and layers, 248-249;
rare failures, 249-250
Reference materials, 477-479
Regulatory agencies, 33-37, 315-317
Reliability concepts, 5-8, 14-19:
definition, 14-15;
engineering, 15-18;
process, 18-19
Reliability perspective, 5-8, 14-19
Responsiveness measures, 337
Rework measures, 339
Risk management, 95, 399-401
Root cause analysis, 119
Round checklist, 317-318, 470, 473
Rounds optimization, 176-178
Routes (groups), 458, 474-475
Run-in period, 109
497
Index 481-500.qxd 3/3/00 3:12 PM Page 498
S
Safety, 186-188: direct consequences, 185-186;
potential consequences, 186-187
Scheduling, 41, 260-274, 281, 289, 449-451:
methods, 266-272
Scheduling methods, 266-272:
expedite, 266-267;
short term (weekly), 267;
long term, 267-268;
equipment groups, 268-272;
outage, 272
Screening for effectiveness, 339-340
Shop practice, 237-238
Short term schedule, 267
Simplicity criterion (software), 319-320
Software, 2, 179, 203, 304-312, 317-319, 437-475:
products, 317-319
Software applications, 437-475:
fault tree analysis, 437-438;
failure modes and effects criticality analysis, 438-439;
RCM analysis, 439-440;
CMMS example (Power FM), 441-459;
RCMtrim, 460-475
Soot-blowing air compressor, 64-65
Spurious alarms, 242-243
Standardization, 13, 80-81, 102-104, 278-219, 304, 307-311, 320:
software, 304
Standards, 102-104, 278-279
Statistical analysis, 331
Statistical maintenance, 347-349
Statistical process control, 18, 97, 435-436
Statistics and regulators, 33-37
Stocking levels, 180-181
Strategy, 275, 287, 344-346:
498
Index 481-500.qxd 3/3/00 3:12 PM Page 499
References
development, 287
Surveillance-based PM, 57-58
System cost, 122
System definition, 118-119
System failure, 83-84
System hierarchy, 81-83
System level measurement, 306
System measures, 329-332
System monitoring, 120-121
System performance measurement, 119-120
Systematic application, 204-205
Systems approach, 6-7, 116-122:
development, 116;
training, 116-118;
design basis, 116-118;
needs awareness, 116-118;
system definition, 118-119;
system performance measurement, 119-120;
system monitoring, 120-121;
system cost, 122
Systems approach, 116-122
T
Task basis, 311-312
Task intervals, 197-199
Terminology, xi-xiv, 12, 257, 353-389, 430
Time accounting, 334-335
Time-based maintenance, xvi, 6, 49, 56-57, 191-192
Tools (engineering), 171-173
Total hours/system measures, 337
Total PM performance, 214-216
Total quality maintenance, 38-39
Traditional maintenance programs, 10-13, 258-260
Training, 42-43, 116-118
Trend analysis, 337
499
Index 481-500.qxd 3/3/00 3:12 PM Page 500
Trends measurement, 337

Triggers (PM), 53-54
Trouble reports, 443-444
Troubleshooting, 183-184
Trust us (strategy), 212-213
Turbine blade, 65-66
Turbine failure, 110-111
Typical implementation (PM), 213-214
V
Value, 20-21, 128, 163, 201-204
Vendor perspective, 216-218:
recommendations, 217-218
W
Wearout, 233-234
Weibull analysis, 172-173
Work descriptions (tasks), 454-455, 457
Work frequency, 237
Work grouping, 269
Work orders, 282-284, 444-459:
backlog, 282-284
Work review, 284-287
Work screening/prioritization, 392-393
Working to schedules, 273-274
Worklists, 280-291
World War II, 9-10
500

Applied Reliability Centered Maintenance

Uploaded by

Copyright:

Available Formats

Applied Reliability Centered Maintenance

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Applied Reliability Centered Maintenance

Uploaded by

Copyright:

Available Formats

What are some of the key concepts discussed in the book?

What are some of the key concepts discussed in the book?

What are some examples of applications of RCM discussed in the book?

What are some examples of applications of RCM discussed in the book?

front matter i-xxiv.

qxd 3/3/00 2:28 PM Page i

Library of Congress Cataloging-in-Publication Data

TS192 .A94 1999

All rights reserved. No part of this book may be reproduced, stored in a

Cover Design: Shanon Garvin and Brian Firth

Printed in the United States of America

1-1 Idarado Bar and Rod Mill . . . . . . . . . . . . . . . . . . . . . . . . . . . .5

3-11 Fault Tree: Loss of Cooling . . . . . . . . . . . . . . . . . . . . . . . . .108

9-1 Process Change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .321

AB A or B (with respect to train, or piece of equipment)

DOT Department of Transportation

LCO Limiting conditions for operations

PUC Public utility commissions

-Oliver Wendell Holmes, The Deacons Masterpiece

Elements of reliability centered maintenance (RCM) arent new;

To assist the reader with the terminology, I offer a rough term

to put order into the maintenance lexicon and tool kit

Id like to recognize all those who helped me with this work

Joe Hunter, who shared an operators perspective. Next, my friend

maintenance decisions. Practically, RCM principles can be learned

nance. What is maintenance? Who administers it? How should it

The fundamental RCM strategy classifications

A fast-track approach to RCM described here can be applied at

Instrumentation and Controls (I&C)

I cannot over-emphasize that implementation is where the value is,

RCM is a maintenance perspective in an operational context

Applied Reliability-Centered Maintenance

Deming, Joseph Juran, and other Americans developed models for

Applied RCM: An Overview

Are there fundamental RCM attributes? What are they?

The answers incorporate the best elements of the original aerospace

dynamic maintenance programs

Applied Reliability-Centered Maintenance

personnel competence in areas of expertise; awareness of others

RCM-minded organizations dont just operate plantsthey

How many organizations truly focus on plant operations?

High-performing maintenance organizations share attributes with

Applied RCM: An Overview

tors to think about maintenancejust do the job. New technologyand

Applied Reliability-Centered Maintenance

failure modes and effects analysis (FMEA) theory

along with other R techniques.

Applied RCM: An Overview

Figure 1-2: Failure Distributions

Systems failednot individual components. Design could manage

Applied Reliability-Centered Maintenance

notion that things shouldnt breakor wouldnt, if properly main-

PM-based maintenance strategies (TBM performance)

Applied RCM: An Overview

Applied Reliability-Centered Maintenance

were set up.

Enter Traditional RCM

Applied RCM: An Overview

Theoretical assessment of what maintenance can (and cant) do, the

Applied Reliability-Centered Maintenance