Saep 349 PDF

Engineering Procedure
SAEP-349 3 December 2014

Advanced Process Control (APC) Systems
Document Responsibility: Process Optimization Solutions Standards Committee
Saudi Aramco DeskTop Standards

Table of Contents
1 Scope............................................................. 2
2 Conflicts and Deviations................................ 2
3 Applicable Documents.................................... 3
4 Definitions...................................................... 3
5 Responsibilities.............................................. 6
6 Instructions..................................................... 7
Appendix A – APC Feasibility and

Benefit Estimation Methodologies........ 26
Appendix B – APC Project Documentation......... 31
Appendix C – Post-Audits................................... 35
Previous Issue: 5 December 2012 Next Planned Update: 5 December 2017

Revised paragraphs are indicated in the right margin Page 1 of 41
Primary contact: Saif, Fouad Ali at 966-3-8800780
Copyright©Saudi Aramco 2014. All rights reserved.

Document Responsibility: Process Optimization Solutions Standards Committee SAEP-349
Issue Date: 3 December 2014
Next Planned Update: 5 December 2017 Advanced Process Control (APC) Systems
1 Scope
This Saudi Aramco Engineering Procedure (SAEP) with its appendices prescribes the
minimum mandatory requirements and guidelines governing the planning, engineering,
design, installation, testing and monitoring of Advanced Process Control Systems
(APC) projects in Saudi Aramco plants. SAEP-349 is applicable to all Saudi Aramco
projects either as stand alone or which include advanced process control systems.
This entire procedure may be attached to and made a part of purchase orders.
In the event that some or all of the steps are executed by contractors or vendors, the
project team shall ensure that the project is executed within the framework of this
procedure.
Selection of specific hardware or software platforms is not within the scope of these
standards and practices.
This document addresses the following list of steps for justifying, implementing, and
maintaining advanced control applications in Saudi Aramco refineries, gas plants and
other applicable facilities:
a) Feasibility Studies - Cost and benefit estimation
b) Implementation and execution
c) Inferential properties development
d) Documentation
e) Performance Monitoring
f) Post-audits
2 Conflicts and Deviations
2.1 Any conflicts between this procedure and other applicable Saudi Aramco
Engineering Procedures (SAEP), Engineering Standards (SAESs), Materials
System Specifications (SAMSSs), Standard Drawings (SASDs), or industry
standards, codes, and forms shall be resolved in writing through the Manager,
Process & Control Systems Department of Saudi Aramco, Dhahran.
2.2 Direct all requests to deviate from this procedure in writing to the Company or
Buyer Representative, who shall follow internal company procedure SAEP-302
and forward such requests to the Manager, Process & Control Systems
Department of Saudi Aramco, Dhahran.
Page 2 of 41
3 Applicable Documents
The engineering execution of APC projects may be considered with the latest edition of
the references listed below, unless otherwise noted.
The following reference contains additional information and industry guidelines

(Sections 3, 4, 6 & 7) for APC applications that may be referenced.
API RP557 Guide to Advanced Control Systems
4 Definitions
4.1 Abbreviations
APC Advanced Process Control System
APCU Advanced Process Control Unit
ARC Advanced Regulatory Control
CV Controlled Variable
DV Disturbance Variable
DCS Distributed Control System
ES Engineering Services
ESA Engineering Services Agreement
FPD Facilities Planning Department
KPI Key Performance Indicators
MIS Management Information System
MV Manipulated Variable
MVC Multivariable control
OO Operating Organization
PAS Process Automation System
P&CSD Process & Control Systems Department
P&ID Process & Instrumentation Drawing
PID Proportional, Integral, Derivative feedback controller
PT Project Team
SAEP Saudi Aramco Engineering Procedure
4.2 Definitions
Advanced Process Control System: Refers to any process control system

application that has functions beyond those commonly associated with
regulatory process control systems. Advanced Process Control Systems
Application may be characterized by any of the following:
Page 3 of 41
● A control system that controls or manipulates multiple variables in order to

maintain one or more operating objectives.
● A control system that performs calculations beyond those that could
normally be performed using standard algorithms available in DCS systems
or multi-loop controllers.
● A control system that may utilize a significant number of DCS standard
algorithms connected together in a complex manner.
APC is generally executed in a higher level computing resource such as a

process control computer or implemented in a programming environment at
lower control levels, irrespective of the complexity of the computations.
Multivariable (including blend property control), constraint and optimizing

controls will be labeled advanced controls. Controls that fall into this category
will be those that are supervisory in nature, i.e., they normally, but not always,
output to the set points of other control loops rather than to the final control
elements directly.
For the purposes of this document, a significant APC application project is

defined as:
● Implemented on a reasonably sized process unit, and
● Automates the setpoint changes to most of the unit's manipulated variables,
and
● Is performed either by a contractor and/or using in-house resources.
Constraints: limits in the process or equipment that should not be exceeded.

Constraints may take the form of physical limits such as a design temperature or
pressure or other pre-defined process limits such as a maximum feed rate,
composition or other value. Constraints may be either maximum values or
minimum values.
Controlled Variables (CV): process values that are maintained by the control
system by making appropriate adjustments to Manipulated Variables. These are
controller inputs to be kept at a set point. Examples are flow, pH, pressure,
temperature, level or concentration.
Distributed Control System (DCS): A process control system is composed of

distinct modules. These modules may be physically and functionally distributed
over the plant area. The distributed control system contains all the modules and
associated software required to accomplish the regulatory control and
monitoring of a process plant, excluding field instruments, remote terminal
units, auxiliary control systems and management information systems.
Page 4 of 41
Disturbance Variables (DV): process input values associated with an Advanced

Control System Application that are measured but are not controlled by the
application. An Advanced Control System Application often takes control
actions to maintain the control objectives when Disturbance Variables change.
Examples are ambient temperature, feed from another unit, etc. Disturbance
variables are also called Feedforward variables (FF).
Economic Variable (EV): an indication of process economics. This variable

can be directly measured using instrumentation sensors and transmitters, or
computed from other variables directly measured or values obtained from
laboratory testing or other techniques.
Engineering Service Agreement (ESA): are internal written commitments

between P&CSD and their customers to resolve most important issues. It is a
service guarantee that P&CSD will provide and help implement a solution in the
field. ESAs state the agreed upon objectives, schedules and success criteria for
individual projects. Customers are simultaneously committing their support and
resources.
Inferentials or Inferential Properties: are physical or chemical properties of a

process stream or a batch which are calculated (inferred, thus inferential) from
other readily measured physical properties such as temperature, pressure and
flow rates.
Manipulated Variables (MV): process values that are adjusted by the

Advanced Control System Application to meet operating targets and desired
values of Controlled Variables.
Multivariable Control (MVC): a form of an Advanced Control System

Application in which several control variables are maintained at desired values
through a complex relationship. Several Manipulated Variables may be adjusted
simultaneously in order to maintain an economic or other operating objective.
Multivariable Controllers typically execute at a frequency of one minute.
Operating Organization (OO): The operating facility or plant.
Process Variable (PV): an indication of process performance, which is directly,

measured using instrumentation sensors and transmitters, values that are
computed from these variables or values obtained from laboratory testing or
other techniques.
Project Team (PT): The team assigned the responsibility of implementing and
managing the project. The team may be led by a contractor (vendor), P&CSD,
operating organization, or any combination of the three.
Page 5 of 41
Regulatory Control (PID): a control application in which generally one

controlled variable is maintained at a desired value by manipulation of one
manipulated variable. Regulatory control may also include control applications
that utilize common calculations or predictions. Examples are steam drum level
controls, combustion controls or mass flow calculations.
Service Factor (SF): is a measure of the fraction of time that an APC application
is operating on-line over a specified period of time. It is expressed in percent.
Utilization Factor: is a measure of the effectiveness of an Advanced Control

System Application. It is more than a measure of whether the application is on
or off. This usually is a complex calculation that is based upon the numbers and
types of sub-functions within an application, the percentage of time that the
functions are operating and the relative economic weighting of each sub-
function.
5 Responsibilities
5.1 APC Project Team composition
The APC Project Team (PT) is assigned the responsibility of implementing and
managing the project. The team may be led by a contractor (vendor), P&CSD,
operating organization, or any combination of the three. A project may be
performed using one of the following three methods:
a) Contractor. The contractor leads, provides all necessary manpower and
has full responsibility for project completion and success under the
management of the appropriate Saudi Aramco representative.
b) In-House. Saudi Aramco employees (P&CSD/OO) lead the project and
are fully responsible for all manpower, hardware and software
requirements.
c) Combination of contractor and in-house. This type of project is led by
Saudi Aramco employees (P&CSD/OO) who will contract portions of the
work as needed.
5.2 Process & Control Systems Department (P&CSD)
Responsibilities include:
a) P&CSD shall be consulted before commencing any significant APC
project to effectively utilize corporate resources and ensure support for
long-term sustainability of the APC application. If appropriate, an
Engineering Service Agreement (ESA) will be developed between P&CSD
Page 6 of 41
and the Operating Organization to ensure a clear vision of project

objectives, responsibilities and deliverables.
b) Ensure that the procedures in this SAEP are used to perform APC project
planning, justification, implementation, testing, documentation,
maintenance, and monitoring.
5.3 Operating Organization
Responsibilities include:
a) Consulting P&CSD before commencing any significant APC project.
b) Ensuring that sufficient engineering, operations and maintenance support
are available and committed to implement and maintain the APC
application.
c) Following the procedures in this SAEP.
d) Correcting instrumentation deficiencies that hinder APC implementation
or operation.
e) Approving final acceptance of the APC application.
f) Review and approve any feasibility and post-audit studies/reports
associated with APC.
g) Revise operating procedures to include and reflect implementation of
APC.
6 Instructions
6.1 APC Feasibility Studies – Cost and Benefit Estimation
6.1.1 Methodologies of APC Benefit Estimation
Every significant APC project shall have an APC Feasibility Study

performed to estimate the potential benefits achievable with an APC
implementation. This requirement is not necessary if the proposed APC
application is:
a) On a process unit where an identical application and associated
post audit have already been completed and where similar
economics are applicable.
b) For an APC remediation or revamp project.
This section addresses the guidelines for estimating the benefits that
APC would bring to a particular process unit. APC benefits are
Page 7 of 41
estimated by a feasibility study (See Appendix A) and then proven by a

post-audit (See Section 6.5 and Appendix C).
There are four (4) major methodologies for estimating APC benefits.
These are:
a) Industry experience
b) Best Operator
c) Statistical estimations
d) Simulation techniques
6.1.2 Recommended Methodologies of APC Benefit Estimation
Saudi Aramco recommends using the statistical estimation method for

final justification of APC projects. Methods one (1) and two (2) may be
used for preliminary selection, budgetary and project prioritization.
See Appendix A for detailed methodologies and procedures of APC
benefit estimation.
Generally, the funding used for APC projects is expense and not capital
sourced. This is primarily due to software licensing and engineering work
that make up the primary component of such projects. Any large capital
funding greater than $2MM for APC projects should follow the Facilities
Planning Department (FPD) Procedures & Guidelines (See website:
http://sharek.aramco.com.sa/orgs/30003023/Pages/Default.aspx).
6.2 Implementation and Execution
6.2.1 Implementation Steps
An APC project will generally follow a series of discrete steps or phases

for an APC implementation. These steps usually include the following:
a) Feasibility Study – Cost and Benefits Estimation (See Section 6.1)
b) Procurement/Contractual Development (See Section 6.2.2)
c) Controller Scope/Preliminary Design
d) Kick-off Meeting
e) Plant Pretest
f) Formal Plant Test
g) Modeling and Simulation
h) Commissioning
Page 8 of 41
i) Training
j) Documentation
A project may reduce the number of steps by combining certain

functions. Also, functions in some steps such as training and
documentation may be completed at various times throughout a project.
Each project phase is described in more detail in the following sections.

The description includes recommendations and “best practices” that
should be followed. Tables 6.1 through 6.10 except for Table 6.3 list
specific deliverables for each project step and which organizations are
involved in producing that deliverable. Table 6.3 list which organizations
will take responsibility, assist or approve specific project tasks.
6.2.2 Procurement/Contractual Development
This task will likely need to be performed in conjunction with part of the
next step, Controller Scope/Preliminary Design, in order to have a good
handle on project scope to meet procurement requirements.
Advanced Process Control implementations generally use expense

funding due to required software licenses and the large amounts of
engineering time and effort required.
Software licensing requires a contract as per Saudi Aramco contracting

procedures. P&CSD has extensive experience in APC contracting issues
and is the sponsor of several GCS contracts for APC vendors. P&CSD
shall be consulted before commencing any APC projects to effectively
utilize corporate resources and ensure support for long-term
sustainability of the APC application. If appropriate, an Engineering
Service Agreement (ESA) will be developed between P&CSD and the
operating organization (OO) to ensure a clear vision of project
objectives, responsibilities and deliverables as shown in Table 6.1 below.
Table 6.1 – Contractual Agreements and Personnel Requirements
Personnel Requirements (Involvement indicated by 'X')

Contractor OO/APC
Deliverables P&CSD OO/Opns OO.Maint OO/Lab
(if used) Eng
Labor and software X X X
licensing Contracts
Engineering Service X X X
Agreement (ESA)
Page 9 of 41
6.2.3 Controller Scope/Preliminary Design
During this phase, P&ID's are reviewed and a tag list is developed along
with the initial controller design. The envelope or scope of the
controller(s) is established and a list is developed of the preliminary
manipulated variables (MV), disturbance variables (DV), and controlled
variables (CV). The control objectives and required inferentials are
reviewed as shown below: See Table 6.2.
Table 6.2 – Scope/Preliminary Design Deliverables and Personnel Requirements

Contractor OO/APC
P&CSD OO/Opns OO/Maint OO/Lab
(if used) Eng
o Tag list X X X
o Preliminary X X X X
controller design(s)
o Project Steps and
X X X
Schedule
o An Inferential
Development and X X X
Implementation
Schedule
6.2.4 Kick-off Meeting
A kick-off meeting shall be held at the plant site with all project
personnel involved in the project. This includes Vendor consultants,
P&CSD engineers, and from the operating organization (OO) including
representatives from operations, maintenance and engineering groups.
A responsibility table shall be developed to clearly present a summary of
personnel roles pertaining to specific project tasks. Personnel involved
may represent OO, central engineering or contractor. See Table 6.3
below for an example of an APC project led by a contractor or P&CSD.
A kick-off meeting is held to:
a) Review detailed description of process operation, control
objectives, and historical operating and control issues.
b) Confirm the project scope.
c) Review and confirm the project steps and schedule.
d) Review the initial controller design (initial list of matrix variables).
e) Review existing inferentials and establish an inferentials
development and implementation plan.
Page 10 of 41
f) Review step testing procedures.

g) Develop pre-test plan.
h) A formal Kick-Off Meeting minutes report delivered to
OO/P&CSD within one week after the kick-off meeting.
i) Form a team and define responsibilities for the members.
Table 6.3 lists deliverables and personnel requirements for this step.
Table 6.3 – Example of APC Project Responsibility Table

P&CSD/ P&CSD/
Task Description OO Task Description OO
Contractor Contractor
Pre-Testing A R Detailed Design Document X R
APC Calculation Design,
Detailed Design Review
Configuration and A R A/X R
Meeting
Commission
ARC Design and Commission A R Engineer Training R
Step Test Plan X R Operator Training A R
APC Controller
Step Testing A R A R
Commissioning
Model Identification A R As Built Documentation R
APC Controller
A R
Generation & Testing
Legend: R: Take Responsibility or Action, A: Assist, X: Approve
The deliverables are listed below in Table 6.4.
Table 6.4 – Kick-off Meeting Deliverables and Personnel Requirements

Contractor OO/APC
(if used) Eng
o Minutes of Meeting X X X X X X
o Responsibility Table X X X X X X
o Pre-Test Step Plan X X X X X X
6.2.5 Plant Pretest
The first step in the implementation of an advanced process control

project after the kick-off meeting is the pretest. A pretest procedure
document should be developed and endorsed by the OO before the start
of the pretest. The following should take place during the pretest:
Page 11 of 41
6.2.5.1 Verify the suitability of the existing preliminary design of the

controller strategies including the configuration of the
regulatory controllers under the proposed MVC controller.
If changes in configuration of the base regulatory system are
required, the plant should operate in this new mode for at least
one (1) day with an engineer present to determine that there are
no underlying problems with the new configuration.
6.2.5.2 Establish estimated gains, responses, noise, and linearity where

possible.
6.2.5.3 Identify control tuning issues and initiate immediate steps to

remedy them.
6.2.5.4 The filtering of all critical signals should be reviewed for noise.
Noisy signals should have the filter increased to remove the
noise as best as possible or use a heavily filtered copy of the
variable if it is not appropriate to filter – e.g., the variable is
also an input to a safety system function.
6.2.5.5 Identify potential process problems that may affect controller

implementation.
6.2.5.6 Verify functionality of critical instrumentation required for the

controller. Identify any shortcomings in the regulatory controls
and field instrumentation and determine remediation steps.
This includes valve hysteresis, slip/stiction and positioners.
6.2.5.7 Determine the necessary manipulated variable step sizes,

acceptable operating ranges and laboratory support
requirements.
6.2.5.8 Collect historical process and laboratory data as necessary to

support inferentials development and evaluation and
preliminary model identification. Ensure that all required tags
are being collected.
6.2.5.9 A control diagnostics product, if available, should be used to

help diagnose and identify regulatory control problems with the
selected manipulated variables on the process units.
6.2.5.10 Any calculated variables such as pressure compensated

temperatures required for the controller are developed at this
stage of the project, implemented in the DCS if possible, and
added to the data collection before the formal plant test.
Page 12 of 41
6.2.5.11 A logbook shall be created to note important observations that

occur during the pretest such as problematic control valves and
instrumentation, and poorly tuned controllers.
The following Table 6.5 below identifies “best practices” for

plant step testing.
Table 6.5 – Step Testing “Best Practices” and Guidelines
No. Practice Description

Do not move a second manipulated variable (MV) consistently at
1 Independent Movement the same time as another MV or at the same relative time to the
other MV.
Step sizes should be large enough to have an effect on the
process that is at least 6 times the noise level in the measured
2 Minimum Movement Size
variable – this is the control variable (CV). The moves should be
as large as operations will allow.
Hold times (time between moves) should be varied between
approximately ¼ of the settling time of the process and about
3 Hold Times 1 ¼ of the settling time. The hold times should be varied
randomly and uniformly over this range. The average hold time
should be approximately equal to ½ of the process settling time.
Start model identification during testing. Models should be
4 Model during Test checked each day to identify areas where additional testing may
be required.
Push towards known constraints if possible. This will mimic
5 Move Steady State
what the optimizer will do.
Monitoring all control valves, bypasses and analyzers to ensure
6 Monitoring Saturation
that none are at saturation. Saturation will invalidate step data.
Step test variables that may show non-linearities. These will
7 Test Non-Linearities
need to be identified to be linearized later.
Real-time communications with all project team members and
8 Good Communications operations is very important. Maintain a log of all abnormalities
such as sudden weather changes, saturations, bypasses, etc.
Following this pretest activity:
6.2.5.12 All deficiencies identified during the pretest are rectified.

Items are categorized as: High importance, Medium
importance and Low importance. It is imperative that the
formal plant test does not proceed unless all of the High
importance items are completed.
6.2.5.13 The new regulatory controls proposed in the preliminary design

shall be configured, implemented and tuned as required for the
formal plant test.
Page 13 of 41
6.2.5.14 Implement any inferential property calculations that are

completed.
6.2.5.15 Develop the Formal Plant Test Plan to provide the greatest
amount of information across a wide range of response
frequencies in the shortest amount of time. The plan will
specify the details of data collection including recommended
manner of collection, required software, etc.
See Table 6.6 below for deliverables and personnel

requirements for the Pretest step.
Table 6.6 – Pretest Deliverables and Personnel Requirements

Contractor OO/APC
(if used) Eng
o Regulatory controller tuning X X X X X
summary
o Instrument/equipment repair X X X X X
check list
o Formal Plant test plan X
X X X X X
o Preliminary plant test report. X
X X X X X
o Update the Preliminary
X X X X
Design.
6.2.6 Formal Plant Test
Plant testing is started once all necessary repairs to the instrumentation

and regulatory control layers are completed. The Plant Step Test is
conducted to obtain data on process dynamics, gains and loop
interaction. See Table 6.5 for testing guidelines and Table 6.7 for
deliverables. Similar to the pretest phase, a formal plant test procedure
document (See item 6 of Table B.1 in Appendix B) shall be developed
and concurred by plant operations before the start of the plant test.
6.2.6.1 During a plant test, each independent variable is stepped up

and down a sufficient number of times to obtain an acceptable
model. This number may be between 3 to12 times or more
depending on various factors. Data is collected generally on a
one minute frequency for the entire test. The collection
frequency will depend on a variable's time-to-steady-state.
The test is conducted on a 24 hour basis, with continuous
Page 14 of 41
engineering coverage, to collect as much continuous data as

possible.
6.2.6.2 Automatic step testing software such as SmartStep (AspenTech

Inc.) or ProfitStepper (Honeywell Inc.) should be considered, if
available. Automatic testing typically requires a good
knowledge of the plant gains to ensure safe operation and
should only be used if appropriate. It is strongly recommended
that safety critical processes (such as reactors, etc.) are
manually stepped.
6.2.6.3 Preliminary model identification is performed throughout the

test to ensure that the quality and quantity of collected data are
sufficient to facilitate good controller development.
6.2.6.4 The plant test is carried out in close cooperation with the console
operators. In case of manual plant test, the console operators are
consulted on all step moves and make the required set point
changes. If this is an automatic test, then the steps moves
maximum sizes are determined by discussions with the console
operators. The set points changes are made automatically if an
automated step testing tool is used. The plant test is a good time
to start discussing the APC application with the console
operators and performing some training.
6.2.6.5 Use an approved closed loop testing and modeling technology

to perform the plant tests especially in the case of revamping
MVC controllers. Products used by Saudi Aramco include
Smart Step from Aspen Tech, Profit Stepper from Honeywell
and Tai-Ji Identification from Tai-Ji Controls, which lately has
been acquired by Honeywell. These tools automatically step
test multiple variables simultaneously, which affords a more
concise model identification data set. The result is reduced test
times, analysis, model identification and ultimately lower costs
for implementing and maintaining MPC applications.
6.2.6.6 If the required inferentials have not been completed, it will be

necessary to collect lab samples during the plant test above and
beyond the normal samples collected. The additional sample
load could be manpower intensive therefore extra laboratory
personnel may be required for the task.
The step test deliverables are listed in Table 6.7.
Page 15 of 41
Table 6.7 – Plant Test Deliverables and Personnel Requirements

Contractor OO/APC
(if used) Eng
o Plant Test Response Data X X X X X X
o Plant Test Log X X X X X X
6.2.7 Modeling and Simulation
6.2.7.1 After completion of the plant test, the plant response data is
analyzed using the off-line data analysis tools. The relationships
between all of the independent variables (manipulated and
disturbance) and the dependent variables (controlled) are
identified, and placed into a model matrix for use by the
controller.
6.2.7.2 The resulting model matrix defines the dynamic relationships

between the manipulated variables in the controller and the
variables it controls.
6.2.7.3 Once the controller model is built, economics are developed.

The economic information and the off-line simulation software
shall be used to simulate the entire controller.
6.2.7.4 Tuning factors are adjusted to get the desired response action
of the controller. Initial controller parameters for
commissioning and field tuning shall be established from the
simulations.
6.2.7.5 The detailed design is finalized and a final design review

meeting shall be conducted.
6.2.7.6 A commissioning plan shall be developed and then reviewed

by the proponent.
Table 6.8 below lists the deliverables and personnel

requirements for this step.
Page 16 of 41
Table 6.8 – Modeling and Simulation Deliverables and Personnel Requirements

Contractor OO/APC
(if used) Eng
o Final Process Model Matrix X X X
o Detailed Controller Design X X X X
Document X X
o Initial Controller Parameters X X X X
o Controller Simulation File X X X
o Commissioning Plan X X X
6.2.8 Commissioning
6.2.8.1 The off-line controller configuration file from the off-line

controller simulation shall be used in the online system.
6.2.8.2 Additional database points required for the controller are built,
any operator screen changes are made, and the controller is
started up to run on live plant data in open loop mode.
6.2.8.3 Verification shall be performed of all controller predictions

along with the constraints the controller is operating against.
This includes uncompensated predictions to check for model
accuracy.
6.2.8.4 Verification shall be performed of all process values, manual

entries and calculated values.
6.2.8.5 Check all graphics and operator interface procedures. Ensure

that all related engineering and operator displays are available
and functional including laboratory update screens. Once the
interface has been checked, the controller is ready to be
commissioned according to the developed plan.
6.2.8.6 Informal training of the operators shall be conducted prior to

and during the commissioning process. For most of the first
week, the commissioning is covered for 24 hours/day for new
controllers. Revamping existing controllers may require less
24 hours/day coverage since Operations are already familiar
with the technology and the concepts. After the first week of
Page 17 of 41
commissioning, the coverage is scaled back and there is a

continued focus on informal training.
6.2.8.7 Saudi Aramco shall verify successful startup and shutdown of

controllers including shedding to backup regulatory control
schemes on the PAS. Shedding verification for bumpless
transfer shall include loss of critical input to APC,
communication failures, or any other situation that may inhibit
the APC from performing safely.
6.2.8.8 Saudi Aramco shall verify the performance guarantees as

specified in the contract.
Table 6.9 below lists deliverables and personnel requirements

for this step.
Table 6.9 – Commissioning Deliverables and Personnel Requirements

Contractor OO/APC
(if used) Eng
o As-Built Engineering X X X X
detailed design document
o Minutes of final X X X X X X
Commissioning review
meeting.
6.2.9 Training and Technology Transfer
6.2.9.1 The operations personnel shall be provided with hands-on

training at the time the control system is commissioned.
The operator displays, functionality and the control objectives
shall be explained at this time. This training should provide the
operations personnel with an understanding of how to interface
with the control system.
6.2.9.2 Formal training of the operators shall be conducted during

appropriate project phases for new applications and first-time
installations of APC technology in a plant. The formal training
shall cover some background on the APC control technology,
the objectives of the controller, the specifics of using the
controller interface, and any other important points associated
with using the controller.
Page 18 of 41
6.2.9.3 Informal training of the operators shall be conducted as

appropriate for modified applications that impact controller
operability. These may include the addition/removal of
variables, tuning that affects controller behavior, and different
strategies or modes. Informal training is normally conducted at
the console with each board operator.
6.2.9.4 Saudi Aramco engineers assigned to the project shall also

become experienced with the control technology and
implementation steps through active involvement throughout
the project phases. The objective is to ensure proper
technology transfer so that they may carry out the long term
maintenance of the control system.
6.2.9.5 An operator training manual shall be developed.
Training deliverables are listed in Table 6.10.
Table 6.10 – Training Deliverables and Personnel Requirements

Contractor OO/APC
(if used) Eng
o Operator training class X X X X
o Operator Training Manual X X X X
6.3 Inferential Properties Development
6.3.1 Inferential Development
6.3.1.1 The use of an inferential calculation shall be considered for all

controlled or monitored properties associated with an APC
application whether or not an on-line analyzer exists or not and
assuming that the properties are capable of being calculated by
state properties such as flow, temperature and pressure, and lab
results are available.
6.3.1.2 An on-line, reliable analyzer is always preferred over an

inferential calculation only. The best engineering practice is to
have an inferential backed up by an analyzer (See Section 6.3.2).
Regardless of analyzer quality, most analyzers have slower cycle
times than that of the APC application. A properly designed
inferential will allow the APC to continue operating for a
reasonable period of time in case of analyzer failure.
Page 19 of 41
6.3.1.3 The development of inferentials shall be based on a

representative period of operational data. This includes
consideration of product specifications, seasonal variations in
the plant conditions and products, efficiency changes in
equipment such as fouling and catalyst deactivation, etc.
Special care should be taken in selecting the process data used
in developing the inferential models. Process data should be
properly aligned with the lab analysis results and/or analyzer
reading. Furthermore, the process data needs to be checked for
the effect of data compression employed by data historians;
highly compressed process data may not be useful for
inferential models developed due to the loss of key features in
process variables behavior in some cases.
6.3.1.4 The choice of inferential input variables shall be based upon

known engineering relationships as much as possible.
6.3.2 Inferential Feedback
6.3.2.1 Inferred properties shall be developed to augment on-line

analyzers to provide instantaneous feed back to the APC
controllers. The inferred properties shall allow either lab or
analyzer feedback, typically updating the bias.
6.3.2.2 The operator or APC engineer shall have the capability to

suspend the feedback. If feedback is automatically suspended
based on feedback data validation, it shall not be re-established
automatically. Operator or APC engineer intervention is
required to re-establish feedback if it has been suspended.
6.3.2.3 The APC engineer (or the operator) shall also have the
capability to select and change the lab (or analyzer) bias
fraction that is for updating the inferential. Suspending
feedback is also equivalent to setting the bias fraction to zero.
The on-line inferential shall have the capability to reject
outliers, frozen signal, and feedback values that exceed a pre-
specified rate of change.
● Lab Feedback - The operator shall have the facility to

evaluate the inferential prediction after lab feedback, and
have the option to reject the feedback.
● Analyzer Feedback - The inferential shall have the facility

to accept a signal from the analyzer indicating that a new
result is ready to be read. The inferential package must
Page 20 of 41
accommodate the dynamic differences (deadtime and lag)

between the analyzer and the predicted quality result when
using feedback from an analyzer.
These facilities for feedback shall be available to the operator

and/or APC engineer via a display in the DCS system.
6.3.3 Inferential Validation
6.3.3.1 Property inferentials shall have validation on the input signals,

feedback signal, and the final inferential output. As a
minimum validation shall include signal validity checking,
high and low limit checking, freeze checking, and spike
detection.
6.3.3.2 The inferential shall have the capability to declare some inputs
as “critical” such that the inferential can be declared a bad
value result or cease to operate if a critical input fails the
validation check. Failure of feedback validation shall not cause
the inferential to stop predicting. The operator shall be notified
(via message or alarm) if the feedback signal does not return to
a good signal within a specified period of time.
6.3.4 Documentation
The documentation for the inferential shall include the following

information:
a) Validation limits.
b) Model form and explanation for choice of form.
c) All inferential tuning parameters.
d) Description of original dataset, and record of data manipulation
prior to inferential development.
e) The correlation of each input variable to the predicted property.
6.3.5 Interface
6.3.5.1 Operator
The operator shall be notified (via message or alarm) if the

inferential ceases to update due to critical input variables
failing validation or ceases to update on feedback due to
feedback signal failing validation.
Page 21 of 41
The inferential monitoring screen shall show the active status

of feedback. A mechanism must be provided on the inferential
interface to deactivate and activate the feedback.
For lab feedback, the operator shall have a mechanism to enter

the lab data and timestamp. Preferably, the interface will allow
an operator to use a simple software switch to enter a
timestamp at the time of sampling.
6.3.5.2 Engineer Interface
The engineer's interface shall allow access to all aspects of the

inferential model. This includes tuning parameters, feedback
filtering, any feedback statistical calculations used within the
inferential, validation parameters, and the unbiased prediction.
6.3.6 Security Access
The operator shall have access to enter lab feedback information, and to
reject the result of a particular feedback calculation. The operator shall
also have the ability to suspend analyzer or lab feedback to the
inferential indefinitely.
The operator may be allowed view-only access to other aspects of the

inferential if desired. All other access to the inferential shall require
engineering access to modify the inferred properties.
6.4 APC Project Documentation
The complete engineering and operating documentation for an APC application

shall consist of the components listed and detailed in the appropriate tables in
Appendix B. As a minimum all documentation shall be electronically compiled
on a CD and provided to the Aramco operating organization site representative,
and a copy shall be provided or retained by ES/P&CSD/APCU.
6.5 Performance Monitoring
6.5.1 General Requirements
6.5.1.1 Each installed APC application with multivariable control shall

be included in a new or existing APC Performance Monitoring
package that has been approved by P&CSD. P&CSD
maintains a recommended list of such vendors. The
performance monitoring tool(s) for MVC and PID may or may
not be the same software or vendor package. Project team
Page 22 of 41
should contact the appropriate P&CSD personnel to obtain

current recommendations.
6.5.1.2 The performance monitoring application shall be able to

monitor the performance continuously without disrupting the
MVC and PID closed loop operations. The performance
analysis results shall be available at the supporting APC
engineers' desktops through the company's intranet using
web-enabled technology. It should use intelligent messaging
via DCS displays, workstation displays, or email for
performance exceptions or alerts.
6.5.2 MVC Performance Monitoring
The performance monitoring package shall adhere to the following

minimum requirements:
6.5.2.1 Provides MVC controller health status through key

performance indicators (KPI's) that are simple in nature and
portray concise information. Examples of these indices are:
a) MVC controller ON/OFF: To be used to indicate if MVC
is turned on and being used to control a process unit.
b) Service Factor: Is a historical trend of the ON/OFF status
for a controller.
c) Utilization factor: To measure how effectively the MVC
is being utilized during the time it is in-service.
d) Constraint analysis: Charts showing the number,
distribution and identification of actively constrained
Control Variables (CV's) and Manipulated Variables
(MV's) over a time period.
e) CV's/MV's statistics: This KPI will include standard
deviations for CV's/MV's, CV's/MV's target errors and
CV's/MV's limit errors.
f) MVC model quality: Verification of model quality by
measuring unbiased predictions and prediction errors.
g) CVs/MVs limit change frequency: Track number of
times that operator limits are changed over a one week
period. Excessive changes indicate trouble.
6.5.2.2 Provides immediate key statistics with the ability for the
engineer to solicit additional information on-demand through
Page 23 of 41
trends and “drill-down” processes if the higher level indices

indicate a problem.
6.5.2.3 Indicates more than just whether or not the application is on-
line. They shall show how effective the MVC controller is
accomplishing its objectives.
6.5.2.4 Monitors the performance of critical MVC-supporting

applications and systems such as inferential calculations and
analyzer performance.
6.5.2.5 Monitor potential symptoms created by any degraded

controller function that negatively impacts performance.
6.5.3 Regulatory Control (PID) Performance Monitoring
6.5.3.1 Regulatory control loops malfunctions can severely impact the

APC application performance. In addition to monitoring the
MVC performance, the performance of the PID loops
associated with a MVC application shall be monitored.
6.5.3.2 The PID monitoring software application should be able to

calculate KPIs such as, but not limited to:
a) Service factor and effective service factor which highlight
controllers in their normal mode and percent of time
controllers are not at a limit.
b) Performance index: The performance of the PID loop can
be compared to a benchmark loop performance. Several
criteria can be used such as minimal variance, closed loop
time constant, rise time, etc.
c) Oscillation index: This index will detect loops that show
oscillating behavior and measure oscillation frequency.
d) Valve Stiction/Hysteresis index: When possible, valve
Stiction/Hysteresis level should be estimated.
e) Standard deviation: The variance of the process variable
(PV), the Setpoint (SP) and the Controller Output (OP)
should be monitored.
6.5.3.3 The PID monitoring package should include capabilities to sort

PID loops by level of performance and pinpoint the ones that
need attention. Whenever possible, the package should include
capabilities of detecting the origin of performance degradation
Page 24 of 41
(Tuning problem, Valve/transmitter problems, interaction,

external disturbances) and give indication of possible remedies.
6.6 Post-Audits
6.6.1 All significant APC applications should have a post-audit performed.
6.6.2 The post-audit is an essential tool for accountability of spent funds.

This serves to inform management of actual benefits obtained in order to
receive management support for maintenance and further control
opportunities.
6.6.3 The purpose of this section is to provide a set of guidelines to

characterize how well an advanced process control (APC) project
satisfied its objectives. In addition, this document will present different
approaches to benefits estimation as well as several guidelines to ensure
consistent basis of comparison between the before and after APC process
information. See Appendix C for detailed instructions.
Revision Summary
5 December 2012 Revised the “Next Planned Update”. Reaffirmed the contents of the document, and
reissued with minor changes.
3 December 2014 Editorial revision to transfer responsibility from Process Control to Process Optimization
Solutions Stds. Committee.
Page 25 of 41
Appendix A – APC Feasibility and Benefit Estimation Methodologies
A.1 Estimating Methodologies
There are four (4) major methodologies for estimating APC benefits. These are:
a) Industry experience
b) Best Operator
c) Statistical estimations
d) Simulation techniques
Saudi Aramco recommends the statistical estimation method for final justification of
APC projects. The first two methods may be used for preliminary selection, budgetary
and project prioritization. The simulation technique may be the most accurate and
comprehensive but depends heavily on accurate process steady-state and dynamic
models. The development of these models can be very resource demanding and
prohibitive in cost. This method should be considered only if such models already
exist. The first two methods are discussed below in more detail.
Industry experience benefits are usually published or presented by implementers of

APC technology or vendors of APC offerings. These benefits are based on an
aggregated average across similar processes based on twenty years of industry
experience and well publicized. Minimal effort is required to use experience numbers.
One needs only to identify the process and the throughput and look up the generic
industry benefit with some interpolation for the throughput. This “rule of thumb”
method has an accuracy of only ± 30% at best since a proponent's process may be
different in design, operation and economics. This method is not recommended for
formal justification use within Saudi Aramco and is generally used by proponent's who
have not implemented APC previously and suitable for budgeting purposes only.
The 'Best Operator' method is the next easiest to perform. It only requires historical
data for some key process controlled variables (CV). The average value of a variable is
then compared to the best achievable by an operator. The difference between the
average and the best operator represents the potential improvement with APC installed.
The assumption is that the APC will consistently perform like the best operator.
Accuracy with this method is around ± 20%. This method is not recommended for final
justification use with Saudi Aramco since it assumes that APC can do no better than the
best operator and optimum operating conditions may be outside the historic operating
region of the plant.
The recommended statistical estimation method is presented in detail in the following

sections. See Table A.1 for a summary of the estimation techniques:
Page 26 of 41
Table A.1 – APC Benefit Estimation Methodologies
Method Effort Required Accuracy Comment

Use for budgeting and preliminary
Industry Experience Easy ± 30%
selection only
Best Operator Easy ± 20% Use for preliminary selection only
Statistical Moderate ± 15% Recommended
Simulation Difficult ± 10% Use only if models exists
A.2 Statistical Methodology of APC Benefit Estimation (Recommended Method)
The statistical method is the most widely used methodology for estimating benefits prior
to the implementation of APC. It takes the form of a formal “Feasibility Study.” This
method requires more effort than the best operator approach, because of the need for more
extensive historical data and statistical calculations. The mean and standard deviation of
the data are calculated. A reduction in the standard deviation is assumed based on
experience with similar applications. This reduction is the greater stability that APC can
provide. Statistical formulas based on confidence regions are then used to estimate the
possible changes to the data mean value enacted by APC, which results in the benefits.
Although statistical analysis of plant data is usually the best way to accurately estimate
APC benefits, there may be cases when this approach is not valid. Some refinery
processes have characteristics that tend to invalidate the statistical approach.
For example, an objective function that is based on maximum product rates may
invalidate estimated benefits that rely on reduced variability in the process. In these
cases, another method, such as the simulation method should be explored.
A.2.1 General Requirements for Identifying APC Benefits

● Understand the economics and business objectives of the processes.
● Understand how the processes work and interact.
● Understand the process' constraints.
Process plant data should be collected and analyzed. Opportunities for

improving the process economics should be discussed with the site personnel to:
a) Compare actual plant performance with production/operating targets,
b) Compare plant performance with other benchmarks.
c) Consider any anticipated process plant modifications or expansions.
d) Identify improvements that have been realized by existing Advanced
Control System applications.
Page 27 of 41
A.2.2 Plant Data Collection
Proper collection of meaningful historical data is essential to the credibility of a

feasibility report.
The following types of data should be collected in sufficient quantity to be

statistically significant and consistent:
a) Product specifications.
b) Operating targets - such as product qualities, rates, yields.
c) Actual values achieved - feed and product rates, yields; laboratory and on-
line analyzer results.
d) Process operating constraints and limits, reasons and values should be
collected under as wide a variety of operations as possible. This could
include throughputs; pressure/flow/temperature limits, valve position and
equipment capacity limits, limitations imposed by other processes or
planning restrictions, environmental restrictions or safety restrictions.
e) Availability of process (stream days) and reasons for outages or restricted
operation.
f) Measurement availability - analyzers, flows etc., including accuracy,
repeatability and reliability.
g) Discrete events such as coke drum sequencing, dryer switches, etc.
h) Control system performance indicators such as poorly performing control
loops, control valves, etc.
i) Future plans and timing for process modifications and their potential
impact.
j) Production operating target changes, which may be planned.
k) Future operational flexibility requirements.
l) Economic driver data and long term plan.
Assumptions will have to be made where necessary data and information are not
available. Assumptions relative to missing data and the impact upon the
confidence of economic predictions should be discussed with and accepted by
site personnel.
A.2.3 Statistical Analysis of Benefits
An important method in analyzing potential control improvement benefits is

statistical analysis. Control improvement generally results in reducing the
standard deviation of controlled variables (targets, setpoints) and moving
Page 28 of 41
operation closer to constraints. Often a 50% reduction is assumed, but the current
value of the standard deviation should be considered and evaluated before
assuming further reduction. Often a 50% reduction can be somewhat conservative
as shown by actual audits of control applications. The improvement in standard
deviation will translate to tighter control to specifications and targets, and can be
represented in financial terms.
The statistical calculations can easily be performed in a spreadsheet such as

EXCEL.
It should also be noted that the improvement in stability of the controlled variables
is often achieved by more frequent adjustments of the manipulated variables.
However, the net effect is a reduction in variability of the controlled variables.
Another valuable analysis is to compare operations variability among operating

crews. Often significant benefits can be quantified by performing sensitivity
analysis and comparing “best” and “worst” operations versus average process
unit operation. This figure can then be used to estimate a realistic reduction in
variability.
Skill is required in selecting which prediction method to employ. Limitations in

information or current capability may indicate one method when in fact another
is more appropriate. It is also necessary to consider the time weighted average
of control targets when those targets are changing.
The prediction of reduced variability may then affect the overall process yield,
feed rate or energy usage. Economic factors for these effects should be readily
available from the site. The same data can be obtained from a rigorous steady
state process model and appropriate pricing information. Such a model can also
be used to assess the impact of the predicted change(s) on other process
parameters to confirm that the change is realistic and viable with respect to
known process operating limits/constraints.
It is likely that several operational and economical cases must be examined to

present a realistic view of potential APC benefits achievable. It is essential to be
able to predict if any increased yields of products due to APC do not negatively
impact downstream units, storage capacity or ability to sell the extra production.
The following steps shall be generally followed to calculate benefits:

a) Determine current feeds and product production from averaged historical
data that includes both process and lab. An accurate material and energy
balance is required at this step.
b) Compare the current production with maximum possible (based on
previous capacity runs if available) and system design capabilities.
Page 29 of 41
c) Calculate the average variability of CV's that impact economics based on

historical data. Compare these averages with maximum and/or minimum
specifications to determine the opportunity increase or delta in profits.
These may be calculated based on true boiling point curves (TBP) to
determine volume increases or decreases. The amount of shift in CV's
typically can be 50%, but a smaller or larger number may be calculated
depending upon process gains and degrees of freedom available
d) Ensure that material and energy balances are maintained after calculating
potential increases/decreases in product yields.
e) Using valid projected corporate market prices for feed and products to
calculate estimated benefits over at least the future 5 years.
f) A review of the current and future market to accommodate the predicted
increases in product shall be conducted with the relevant Saudi Aramco
marketing organization.
g) Where throughput increases are predicted with APC, an engineering
review of the existing downstream systems and storage capacity shall be
made. Any required changes to downstream systems or storage capacity
for APC shall be documented and associated costs estimated. Downstream
changes may not be required for successful APC implementation but may
represent additional opportunity benefits to be considered.
Experience has demonstrated that post-implementation audits are only as good as

the quality of the pre-implementation (base case) information. The base case must
be fully developed before implementation to ensure validity of a post-audit.
The information and data produced for a benefits prediction study provides a
very sound basis (Base Case) for carrying out a subsequent post-application
audit. The application implementation will always be in the future. Hence, it is
important that the impact of any intervening changes; both non-control (process
modifications, catalyst/target changes, etc.) and control are evaluated to check
the validity of the base case.
Page 30 of 41
Appendix B – APC Project Documentation
B.1 Documentation Requirements
Table B.1 lists the minimum documentation requirements for an APC project.
Table B.1 – Minimum Documentation Requirements for APC Projects

Item
Document Title Description
No.
This study details the potential benefits that APC will
provide to the process unit(s) under consideration for
1 Pre-Audit/Feasibility/Benefits Study
APC implementation. See Section 5 for further details
and content.
Complete minutes of key meetings shall be issued and
2 Minutes of Meetings distributed to the project team to ensure proper
communication and follow-up.
This document outlines a very basic and preliminary
controller design. It shall provide details of which
manipulated variables are to be stepped, how big the
moves will be and a time schedule for making the moves.
3 Pre-Test Step Plan
It shall detail which controlled variables will be monitored
along with any special or additional lab requirements.
In addition, data collection requirements shall be
addressed.
The project implementation team shall decide on the
frequency of status reports. If a Contractor is involved,
he should provide a weekly status report to Saudi Aramco
during the times whenever the Contractor is performing
work on-site. The Contractor shall provide a monthly
status report to Saudi Aramco during the times whenever
the Contractor is performing work off-site in the
Contractor's offices. The report should be in Microsoft
4 Status Reports WORD format and can be emailed to Saudi Aramco.
The report shall contain the following minimum items:
 Project issues
 Accomplishments
 Existing outstanding items
 New outstanding items
 Present schedule and any revisions
 Any other pertinent comments concerning the project.
The preliminary design documentation shall contain as a
minimum process description, operating targets and
5 Preliminary Design Document constraints, control objectives and strategies, tuning, point
architecture, supporting calculations including inferentials,
instrumentation issues, and simplified P&IDs.
Page 31 of 41
Item
Document Title Description
No.
Details the procedure and schedule for the rigorous plant
tests and follows the required content as described in
6 Formal Plant Step Test Plan
item 3 of this table – “Pre-Test Step Plan'. It may also
contain expected CV changes based on MV moves.
Once the Plant Test, model identification and controller
development are completed, the Detailed Design
Documentation shall be prepared and delivered for
review by Saudi Aramco prior to the controllers
7 Final Detailed Design Document
commissioning phase. In addition to an update of what
the Preliminary Design contains, this document will
include the matrixes of the final controllers, models,
configurations and calculations.
This document describes the plans for commissioning
activities for each MVC. It should contain controller
design features and constraints that operations should be
8 Controller Commissioning Plan aware of during commissioning. It should have a
commissioning schedule for each controller and MV.
It should detail steps for PAS setup, operations review,
shedding mode testing, and any impact to operations.
This document is intended to introduce the plant
operations staff to the applications which are installed,
and to provide the essential information required to
9 Operator Manual operate these applications. This document is intended to
be a quick reference, not an in-depth reference manual.
The reader is referred to the application specific manuals
for more detailed or background information.
Once commissioning of the controller has been
completed as part of the field implementation phase, the
10 As-Built Design Document Detailed Design Document will be updated and re-issued
to an “as-built” status. See Table B.2 for content of the
Design Document.
Plant P&IDs shall be updated to reflect the APC loops,
11 Updated P&IDs inferential calculations, and any medications or additions
to the regulatory controls.
B.2 As-built Engineering Design Document Content
Table B.2 lists the minimum required content for the as-built engineering design
documentation. The sections are suggested for the document, but some can be renamed
or combined if appropriate. The engineering manual shall be provided as an indexed
PDF document or Microsoft Word document.
Page 32 of 41
Table B.2 – Content of As-Built Engineering Design Document

Section
Title Description
No.
This section shall describe the scope of the control
1 Introduction application(s) and an overview of the project schedule to
develop the controller(s).
This section shall provide an overview description of the
process that is being controlled by the advanced
2 Process Descriptions
controller(s). A simplified process diagram shall be
included.
This section is to describe the “big picture” control
Overall Control Objectives, purposes and issues. This section may be optional and
3
Challenges and Constraints included in section 4 below if the application involves only
a single controller.
This section shall provide an overview of how the
controller application(s) are arranged when multiple
applications are configured, and when a controller has
subsections/subcontrollers.
The key objectives underlying the design and tuning shall
4 Control Objectives be described in written format here.
The rationale for each controller's variable limits, as
commissioned, shall be provided. This shall be in the form
of a short 1-2 sentences description for each variable.
The rationale for essential/critical variables shall be
documented in this section.
Any identified instrumentation problems and issues should
5 Instrumentation Issues be fully described here along with resolutions and
recommendations.
The section shall list all of the variables used in the
controller, including the tag name, tag description, DCS
Control and Manipulated
6 source parameter, APC variable type (MV, CV, DV, EV,
Variables description
etc.). Essential/critical variables shall be identified in this
section.
This section covers the optimization strategies and
Optimization and Control economics of the APC controllers. This shall include
7
Economics various operating modes and their impact on controller
objectives.
The dynamic control and optimization (dynamic and/or
steady-state) tuning parameters for all variables shall be
8 Controller Tuning
listed. A short comment indicating the rationale for the
choice of tuning parameters shall be provided.
If there are operating procedures, instrument maintenance
procedures or considerations, or variable limit
9 Special Controller Procedures considerations specific to the control application(s), these
shall be described in this section. Procedures shall also
be included for controller startup, shutdown, and shedding.
Page 33 of 41
Section
Title Description
No.
This section shall list all special online calculations done
during pre-processing or post-processing of the control
10 Custom Controller Calculations application(s). This includes variable transformations,
custom calculations, etc. A print out of the calculations
shall be included.
Includes hardware and system architecture of DCS with
APC system. Should include configuration
details/procedure of APC system within the DCS and
address any site specific system issues.
All new DCS tags, calculations, code, filtering, etc.,
created or generated to support the MVC control
application(s) shall be described in this section.
Process Automation System
11 Any custom code written in the DCS shall be listed.
Configuration and Calculations
All DCS configuration changes made to existing DCS tags
shall also be listed.
All tuning parameters of regulatory loops that are modified
shall be listed.
Any project related changes to the data historization
should be noted and explained.
All property inferential calculations developed as part of
12 Inferential Property Calculations the control application(s) project shall be documented.
The formulae shall be listed.
Describes variables, interfaces and procedures for a
13 Performance Monitoring
performance monitoring component if applicable.
This section shall describe typical procedures for
managing the control application(s) online. This includes
starting and stopping the online MVC software, updating
and changing control application tuning, or other
14 Controller Management modifications to the application(s).
Special accounts, passwords, and server names required
to access the control applications on the control servers
shall be documented.
Controller model and gain matrices should be included.
15 Controller Model A short comment indicating the rationale for the choice of
models/gains shall be provided.
Controller configuration file should be included with any
customizations. All text and binary files required for the
off-line and online portions of the APC or MVC application
shall be provided in the as-built revision. The specific files
are not enumerated since each MVC vendor implements a
16 Configuration File
different structure. However, all files required to duplicate
the off-line development environment data set (not
included in the application software itself) and to
completely rebuild the online application (not included in
the application software itself) shall be provided.
Page 34 of 41
Appendix C – Post-Audits
C.1 General
The purpose of this section is to provide a set of guidelines to characterize how well an
advanced process control (APC) project satisfied its objectives. In addition, this
document will present different approaches to benefits estimation as well as several
guidelines to ensure consistent basis of comparison between the before and after APC
process information.
To reap the most benefits from an APC application, it is necessary to monitor and
analyze the APC application performance. In evaluating an APC project, it is necessary
to evaluate the application's control, process, and economic performance measures.
Benefits from APC projects are typically realized in the form of:
● Increased conversion and yield of higher value products
● Decreased utilities and energy cost
● Improved product quality control
● Qualitative benefits that improve overall operations and productivity which are
difficult to quantify.
The proposed performance measures shall help identify such benefits and possibly
highlight areas of opportunity for further improvements.
C.2 Post Audit Study Basis and Timing
One of the challenges in developing post audit reports is maintaining consistent basis of
comparison between the before and after APC implementation process conditions and
information. This consistency lends credibility that benefits reported are actual
(realistic) benefits attained. The following are some guidelines that could help in
establishing the study base case and timing considerations:
a) Controller should be stable (service factor greater than 90%) and running for at
least three (3) months before conducting a post audit study.
b) Benefits calculations shall consider controller operating for 300 to 350 days per
year, with a projected service factor (controller up-time) of 85%. If actual service
factor is known then it shall be used.
c) Compare the current operating condition of the plant with the conditions
documented in the base case. Conditions like feed rate, feed composition,
ambient temperature, season, down stream unit capacities, product specifications,
etc., have to be as close as possible to the base-case data to ensure a meaningful
comparison.
Page 35 of 41
d) If the above requirement is not possible, then as a minimum, base case data should
represent normal plant operation, and data that is one (1) year apart is conducive
to matching seasonal conditions.
e) Ensure that current operating conditions are normal plant or process unit operating
conditions (e.g., feed rate, feed composition, quality targets, etc.)
f) Base case and post-APC implementation data quality must be ensured. Process
upsets, excursions, and abnormal operating modes shall be excluded from the data
sets to maintain a consistent basis for comparison.
g) Post audit report shall highlight if the base case and post-APC data are corrected
for energy/material balance and/or normalized.
h) If for some reason, base case pre-APC data is limited or not representative of
current normal operating conditions, regression based models or off-line
simulation could be used to generate the base case data.
i) Only when all other means to obtain base case data are not possible, the controller
can be turned-off for a sufficient period of time to collect the pre-APC base case
information. This is not a good method for establishing a comparison as operators
can learn how the multivariable controller works and try to match, and they often
rise to the challenge.
j) Confirm quality and feed rate targets with the Production and Planning
organizations at the facility. If the feed rate and feed composition are significantly
different from the base case conditions, the performance run will not be
representative of typical operating conditions and a meaningful comparison with
the base case data will not be possible.
k) Confirm that important plant equipment and field instrumentation (pumps, valves,
transmitters, analyzers, etc.) are working properly. Any serious instrumentation or
analyzer problems have to be corrected before the performance test run can
commence.
l) Confirm that all the important valves have sufficiently small hysterisis and
stiction. Any cycles in PID controllers must be small enough not to adversely
affect the controller. Any significant cycles in the process when the controller is
switched off, must be corrected.
m) Confirm that the quality of inferential models predictions is acceptable as
compared to analyzer readings and/or lab analysis results.
n) Investigate any process or equipment modifications made since the completion of
commissioning. Major changes may change the dynamics of the process, and the
controller models may not adequately represent process behavior anymore.
(Additional step tests may have to be performed to improve the controller
performance after a major process modification. If this is the case, include it in
the recommendations.)
Page 36 of 41
o) Product and energy prices information need to be included in the post audit report.
Use of current product and energy prices is encouraged, however using pricing
information from the pre-APC project benefit analysis phase (if any) may give an
insight on actual vs. estimated benefits of the APC project. Whenever possible,
the report shall calculate the benefits using both current and benefit analysis phase
pricing information.
p) Compare PID controller tuning present at the time of the step test with current
tuning, and confirm that any changes made since commissioning does not
adversely affect the controller's performance. Inappropriate tuning changes have
to be corrected.
q) Ensure that the MV and CV limits of the controller are set up in such a way as to
maximize the economic benefits while maintaining safe plant operation.
r) Observe the controller performance during the post audit evaluation period to
ensure that its performance is still comparable with the results achieved at the end
of commissioning. Any major issues related to the controller performance have to
be identified and resolved before the post audit performance run is started.
s) Ensure that the data collection system is functioning properly, and that all the
important tags are collected at the same frequency as the base case data. Any data
compression or filtering that was not active when the base-case data was
collected, has to be removed for the test period. A data back-up procedure has to
be in place as well.
t) Review the data on a daily basis. Any abnormal incidents unrelated to the
controller (e.g., loss of feed pumps, reflux pumps, fan trips, loss of electricity
supply, or abnormal operating conditions) have to be removed from the data set.
The performance run has to be extended until an adequate period of normal
operating data has been collected.
u) If controller performance unexpectedly degrades (e.g., due to a major change in
operating condition or equipment changes), the post audit test run has to be stopped.
If required, the controller has to be modified before the test run is restarted.
v) Present a short training session to operations, engineering, and planning personnel
explaining the post audit methodology, and the performance run conditions.
w) Finally, it is recommended that a review process be conducted with Operations,
Planning, and Engineering staff to validate and agree upon the results before the
post audit report is issued.
C.3 Post Audit Methodology
APC benefits estimation methodologies typically include the use of statistical data
analysis (mean, standard deviation, and variance) of process variables, or products yield
and quality information represented in financial terms. Sometimes statistical methods
Page 37 of 41
alone cannot quantify the benefits attained from an APC application, thus a combination
of these methods may be required. Other methods may include performance analysis
from a material/energy balance perspective, or from a pure controller variables behavior
point of view (e.g., constraints handling and cost variables information.) How these
methods are applied at each site may be unique, therefore, the post audit report needs to
describe the methodology followed.
Although the main objective of APC post audit studies is to highlight the tangible benefits
reaped from the application, almost all APC implementations result in numerous
intangible benefits that cannot be represented in financial terms, yet result in noticeable
improvement in aspects related to the process unit or plant operation. As an example,
APC applications can dramatically improve the stability of the process, reduce operators'
intervention, and improve flexibility of the process especially when switching to different
modes of operation. The post audit report must highlight such intangible benefits and
possibly present some before and after APC information as a measure of improvement.
C.3.1 Controller Performance
This is a measure of how well the controller satisfies its objectives. Statistical
analysis measures (standard deviation, variance, and mean) as well as regulatory
control measures (set point tracking, speed of response, etc.) can be used to
describe this measure.
Controller and process information relevant to this measure may include items
listed in Table C.1.
Table C.1 – APC Measures of Performance
Measure Controller/Process Information

Tightness of control, controller stability, and
Set point & Range tracking performance
tuning information.
Controller stability, smoothness, tuning, and
CV/MV movement
model accuracy.
Controller speed of response, tightness of
Closed loop Vs. open loop CV response
control, and disturbance rejection capabilities.
Performance improvements relative to pre-APC
CV/MV mean, variance, and std. dev.
data.
Controller robustness, and plant operators
Controller availability (service factor)
confidence (intangible benefits).
Controller prediction error plot. However, it is
necessary to distinguish between models inaccuracy
and unmeasured disturbances when looking at
Quality of controller models. prediction errors.
Note: This information could also provide a
benchmark for future post audit studies.
Page 38 of 41
The post audit report shall highlight, whenever possible, the observed control
performance improvements supported by process data plots and statistical data
analysis for both the before and after APC scenarios. Sometimes, conducting a
test-run for the controller is necessary to determine the performance
improvement information and to verify whether the controller is able to satisfy
its objective(s) or not.
C.3.2 Process Performance
This measure focuses on the controller's influence on the process behavior under
APC. Products yield and quality information may give an indication to the
tangible benefits of the APC controller. Statistical analysis tools can be used to
identify the process performance improvement; however, the intangible benefits
attained from the APC application also need to be highlighted in this measure.
Process stability, flexibility, and operators’ intervention information are
particularly important to highlight.
Other indications to the benefits from a process performance point of view are
the optimization and constraints handling abilities of the controller. Collected
process data need to show that the controller is driving the process to the
optimum operating point as per the controller objective, and is riding the right
constraints for the current mode of operation. For controllers with multiple
modes of operation, a test run shall be conducted to test their performance in all
modes unless the expense to the unit operation is too excessive.
Controller and process information relevant to this measure may include items
listed in Table C.2.
Table C.2 – Process Performance Measures
Process Information Indication
Active constraints Control objective(s) fulfillment

Process operating point (CV/MV targets) Controller optimization ability
Products yield information APC tangible benefits
Quality control performance APC tangible benefits
Energy and/or fuel consumption APC tangible benefits
Process stability and ease of operation Intangible process improvement
Level of operators intervention Intangible process improvement
One may chose to combine both the control performance and process
performance measures together when conducting the post audit study.
Page 39 of 41
The combined measure then shall present all the tangible and intangible benefits
attained by the APC application.
C.3.3 Economic Performance
This measure represents both control and process tangible performance

improvements in financial terms. It is often the strongest indication of the
profitability of an APC project. Deriving the financial benefits can sometimes
be straightforward. For example when control and process performance
measures indicate increased yield of valuable products, increased unit
conversion, reduced quality (specification) giveaway, or reduced energy
consumption; then in this case converting such data into profit or savings
information in financial terms is relatively easy.
However, sometimes it is necessary to analyze the controller's cost variables, as

well as CV/MV movements and represent them in financial terms. The post
audit report must explain the methodology followed in developing the economic
performance measures and the basis of deriving the financial information.
Pricing information used in the report shall be developed as per the directions
and guidelines followed by Planning and Economics organizations at the
facility. Controller and process information relevant to this measure may
include items listed in Table C.3.
Table C.3 – Economic Performance Measures
Process Information Indication

Increased yield of valuable products Profit
Increased conversion Profit
Increased throughput Profit
Reduced quality giveaway Savings
Reduced energy consumption Savings
C.4 Inferential models and regulatory control performance
Since the inferential properties and the regulatory loops are essential to the APC
application, the post audit report shall provide an indication about the performance of the
inferential models (if any) and the regulatory control loops during the evaluation period.
The performance of the inferential models could be analyzed through a comparative

study between lab analysis values and/or analyzer readings, and the inferential models
output. Lab or analyzer bias update need to be examined during the post audit period
by observing bias update trends to identify any model errors or drifts.
Page 40 of 41
The performance of the regulatory control layer needs to be addressed briefly in the
report. Poor performing loops need to be highlighted, tuning and remediation efforts
need to be included in the report for action and future reference.
C.5 Post Audit Report Outline
Table C.4 contains a list of sections that a post audit report should contain.
Table C.4 – Post-Audit Report Sections

Section
Title Description
No.
This section contains a brief synopsis of the study results
1 Executive summary
for Plant and P&CSD management.
This includes an overview of the process, project scope,
2 Overview cost and schedule, as well as the controller scope
summary.
A discussion of the methodology used for the study
3 Methodology including rationale for data selection and selected economic
basis.
This should include both control and process performance
measures as a result of APC implementation. Regulatory
4 APC performance measures
control and inferential models performance could be
included in this section.
This section shall include the economic performance
information and includes all data, assumptions and
5 Benefits calculations calculations used to support the study conclusions. Benefit
calculations should be provided in physical units and in
monetary terms when economic data is available.
This section should explain both the positive and negative
aspects of the project. Specifically, it should address how
the project is capturing benefits. Areas where benefits are
unrealized need to be presented with possible causes and
6 Observations
recommended solutions. Intangible benefits could be
highlighted in this section. Difference between study
findings and project expectations can also be reported and
explained in this section.
This section identifies areas where benefits can be
7 Future opportunities increased. Additional instrumentation, analyzers, process
modifications, etc., may be included in this section.
This section should be a brief summary of the results and
8 Conclusions
recommendations, as well as future areas of improvement.
Page 41 of 41

Saep 349 PDF

Uploaded by

Document Informationclick to expand document information

Document Informationclick to expand document information

Copyright:

Available Formats

Saep 349 PDF

Uploaded by

Document Information

Original Title

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

Saep 349 PDF

Uploaded by

Copyright:

Available Formats

Engineering Procedure

SAEP-349 3 December 2014

Saudi Aramco DeskTop Standards

Appendix A – APC Feasibility and

Previous Issue: 5 December 2012 Next Planned Update: 5 December 2017

Copyright©Saudi Aramco 2014. All rights reserved.

2 Conflicts and Deviations

The following reference contains additional information and industry guidelines

Advanced Process Control System: Refers to any process control system

● A control system that controls or manipulates multiple variables in order to

APC is generally executed in a higher level computing resource such as a

Multivariable (including blend property control), constraint and optimizing

For the purposes of this document, a significant APC application project is

Constraints: limits in the process or equipment that should not be exceeded.

Distributed Control System (DCS): A process control system is composed of

Disturbance Variables (DV): process input values associated with an Advanced

Economic Variable (EV): an indication of process economics. This variable

Engineering Service Agreement (ESA): are internal written commitments

Inferentials or Inferential Properties: are physical or chemical properties of a

Manipulated Variables (MV): process values that are adjusted by the

Multivariable Control (MVC): a form of an Advanced Control System

Operating Organization (OO): The operating facility or plant.

Process Variable (PV): an indication of process performance, which is directly,

Regulatory Control (PID): a control application in which generally one

Utilization Factor: is a measure of the effectiveness of an Advanced Control

5.1 APC Project Team composition

5.2 Process & Control Systems Department (P&CSD)

and the Operating Organization to ensure a clear vision of project

5.3 Operating Organization

6.1 APC Feasibility Studies – Cost and Benefit Estimation

6.1.1 Methodologies of APC Benefit Estimation

Every significant APC project shall have an APC Feasibility Study

estimated by a feasibility study (See Appendix A) and then proven by a

6.1.2 Recommended Methodologies of APC Benefit Estimation

Saudi Aramco recommends using the statistical estimation method for

6.2 Implementation and Execution

6.2.1 Implementation Steps

An APC project will generally follow a series of discrete steps or phases

A project may reduce the number of steps by combining certain

Each project phase is described in more detail in the following sections.

6.2.2 Procurement/Contractual Development

Advanced Process Control implementations generally use expense

Software licensing requires a contract as per Saudi Aramco contracting

Table 6.1 – Contractual Agreements and Personnel Requirements

Personnel Requirements (Involvement indicated by 'X')

6.2.3 Controller Scope/Preliminary Design

Table 6.2 – Scope/Preliminary Design Deliverables and Personnel Requirements

Personnel Requirements (Involvement indicated by 'X')

6.2.4 Kick-off Meeting

f) Review step testing procedures.

Table 6.3 – Example of APC Project Responsibility Table

Legend: R: Take Responsibility or Action, A: Assist, X: Approve

The deliverables are listed below in Table 6.4.

Table 6.4 – Kick-off Meeting Deliverables and Personnel Requirements

Personnel Requirements (Involvement indicated by 'X')

6.2.5 Plant Pretest

The first step in the implementation of an advanced process control