Paper # 1 - Simulation As A Tool To Enable World's Best Mill Relining Practice
Paper # 1 - Simulation As A Tool To Enable World's Best Mill Relining Practice
Paper # 1 - Simulation As A Tool To Enable World's Best Mill Relining Practice
PRACTICE
Peter Rubie
Peter Rubie,
Chief Engineer, Russell Mineral Equipment Pty Ltd
149 Hursley Road (PO Box 100)
Toowoomba AUSTRALIA 4350
([email protected])
ABSTRACT
For many years, optimising the relining of hard rock grinding mills has been complicated by:
the large number of variables;
the low frequency of relines (slow feedback loop);
the potential economic risk associated with some changes
A technology that has its roots in the automotive manufacturing industry is being utilised to
construct a statistically accurate, configurable, and dynamic simulation of the mill relining process. The
effects of various mechanisation and process changes can be rapidly modelled and compared.
This technology will provide mill maintenance managers and contract relining companies with the
opportunity to benchmark current performance against worlds best practice. It will also provide an
opportunity for concentrator plant designers and mill manufacturers to plan for maximum plant efficiency,
by designing an infrastructure and equipment combination that maximises annual plant availability.
This simulation tool and its associated data set, has been developed using recent grinding mill
performance data from mills of varying dimensions, liner configurations, mill relining equipment, operator
and tool variables.
The simulation will provide an output that the various stakeholders can utilise to make optimal
decisions within the given financial, physical and technological constraints of the individual concentrator
plant.
KEYWORDS
mill relining, discrete event software, simulation, mill availability, process efficiency
INTRODUCTION
The duration of a mill reline is the sum of thousands of discrete and interdependent activities. The
choreographing of these activities occurs in an environment that is often noisy, full of other unrelated work
and where the visibility of, and communication with, dependent activities is often obscured by the mill
itself. Because of the complexity of the interplay between tasks and the difficulty of quantifying them the
estimation of reline times has, in most cases, been reduced to assigning an average time per piece, based on
historical data, to each type of liner and multiplying this by the number of pieces to be changed. This is
useful at a macro level but is a blunt instrument when it comes to gaining insight into the effects of
changing workflows or labour/machine/tool availability. Stephen Kittels Mill Metrics Survey (Kittel,
2004) which led to the development of RMEs O-ZONE liner lifting tool range, and significant reline time
reduction has demonstrated that a detailed understanding of individual tasks and their interaction can lead
to significant innovation and time saving.
This paper describes RMEs work in developing a simulation program that models the mill
relining process at individual operation level and mimics their complex interplay. The goal of this work is
to provide a tool that enables mill owners, plant and machine designers and reline planners to visualise and
quantify the effects of changing reline parameters and processes. It is envisaged that this program,
correctly applied, will lead to the optimal use of existing machinery at site, the design of more efficient
plant, the creation of new technologies and the implementation of new practices. All of these combined
will lead to a significant reduction in mill relining times.
BACKGROUND
Since 1985 Russell Mineral Equipment Pty Ltd (RME) has had the specific goal of mechanising
hard rock mineral grinding mill relining. It has been identified that mill liner life dictated the when of
mineral concentrator maintenance shutdowns. Furthermore, the exchange of mill liners dictated the
duration of each shutdown; the how long. In combination (the when and the how long) dictated
annual maintenance shutdown lost time and therefore the availability and the possible utilisation of the
mineral concentrator plant.
Since all mineral ore of value must pass through the grinding mill, liner life and liner exchange
rate can define the productive limits of the entire mine site.
There are four major contributors to controlling the timing of mill relining shutdowns and the
speed of liner exchange, these being:
the liners themselves (their size and quantity);
the reline planning process and relining crew proficiency;
relining equipment; and
plant design surrounding the grinding mills.
RME is often approached for advice on probable reline performance based on a particular plant,
mill and machine configuration. By necessity the advice given has been limited to the performance of the
machines, rather than the overall reline performance. Some typical question asked that are difficult to
answer follow.
What will our reline times be if I install 8,000kg liners instead of 5,000kg liners? It is known that
fewer larger liners can lead to reduced reline times. RMEs liner placement Mill Relining Machines have
allowed the successful application of liners up to 7,250 kg. With mills continuing to increase in size and
liner suppliers capable of producing liners in excess of 10,000 kg this trend is likely to continue. The
question is now being asked, is there a point at which the benefits of this approach are reduced due to the
logistics of handling ever larger liners before, during and after the relining process?
How many people do I need on my reline crew? Reline process planning and execution varies
significantly from site to site and between reline teams. These factors have been recognised as some of the
largest contributors to variations in mill reline performance, yielding the question what constitutes a well
organised and executed reline?
What is the optimum floor height and lay down area size and location? Plant design is of vital
importance for new, yet to be built projects. As recognised by the Institute of Asset Management (AIM,
2008) and the SALVO project (Thomson, 2011), the overall life of asset cost, capital and operating, can be
greatly reduced by proficient plant design.
In an effort to answer questions such as these and to continue its drive towards removing mill
relining from the critical path of plant maintenance, RME has been seeking a tool that can combine all of
the contributing factors and provide a clear understanding of the impact of change. This has led to the mill
reline modelling project described in this paper.
DISCUSSION
Simulation Platform
Software Selection
Discrete event software was chosen as the platform for this project. This type of software is used
to model complex process flow where the interplay of operations or part/labour flows cannot be predicted
using analytic methods and where the only other alternative to modelling is trying out the changes on a live
system. Common uses of discrete event software include the modelling of passenger and bag flows at
airport passenger check in lounges to determine maximum throughput and queue times, the modelling of
hospital emergency wards to estimate waiting times and the modelling of car production lines. In addition
to providing the flexibility to program the characteristics of equipment and the logic of their interaction
within a system, the discrete event software used animates the movement of parts, personnel and
machinery. Visualisation of the mill reline will provide greater understanding of the effects of changes and
will provide the analyst with confidence that the results are realistic and representative of real world
outcomes.
A three dimensional (3D) user friendly interface was developed for the simulation, which is used
to set up the mill parameters, assign variables and define scenarios for testing.
Identifying Variables, Flexibilities and Logic
In 2010, RME received permission from four sites, each owners of either RUSSELL 8 or
RUSSELL Twin 8 Mill Relining Machines, to video record the relining of their mill and to interview reline
crew and staff. This resulted in the compilation of a comprehensive list of the variables and flexibilities
that must be considered in the modelling process, and an understanding of the logic of the interactions
between them. Table 1 lists the major variables defined.
A fundamental requirement of the program was flexibility. An analyst should be able to easily
change and analyse the effects on reline time of something as small as:
the speed of the liner cart;
the size of, or distance to, the liner lay down area;
the size, type, priority or number of any tool in any particular work zone;
whether the liner cart is loaded from the rear or the side.
Furthermore, analysis of significant elements should be easy to account for, such as:
using a Single or Twin Mill Relining Machine;
changing the number of liners in the mill; or,
installing an elevating platform;
the number of personnel on dayshift or nightshift.
All of these flexibilities plus many others have been included in the program in a way that allows
the logic to adapt with the change. Some of the flexibilities and logic will be discussed later in this paper.
Tool
Nut Resistance
Rattle gun
Washer removal
Bolt resistance
THUNDERBOLT Hammer
THUNDERBOLT Hammer
THUNDERBOLT Hammer
Fastener Removal
Liner Removal
THUNDERBOLT relative effectiveness x liner
resistance x liner mass scaling factor
Applies to liners not removed using THUNDERBOLT
hammers and is unique to each liner
Applies to any liner needing machine intervention prior
to slinging a user specified ratio
Liner resistance
THUNDERBOLT Hammer
Liner turning
Slinging time
Sling/O-ZONE Tool
Liner Release
Liner Cart
Liner pick up
Grapple Tool
Liner positioning
Operator skill
Grapple Tool
Tool type
Fork Lift
Operator skill
Liner Placement
All variables are treated as discrete time/frequency distributions; five discrete intervals are used
for each variable. The time measurement of the operations associated with each variable are grouped or
averaged together into five logical time intervals and a count of the number of occurrences at that time
entered in the frequency column.
The use of distributions rather than overall average times is essential to the accuracy of the
program. For example if the time taken for a mill relining machine to retrieve a worn liner is represented
by the distribution in Table 2 and the time taken for the liner cart to return to pick up the liner is
represented in Table 3, if overall average times (75 sec for the machine and 74 sec for the liner cart) were
used, it appears that the machine never waits for the liner cart. However it is obvious from the tables that
(using this fictitious data) the machine may wait at times up to 250 30 = 220 seconds for the liner cart.
The total time the machine waits depends on the order in which the variables are combined. A random
number stream is used to select the time assigned to an event based on its relative frequency in the
appropriate variable. This type of relationship is repeated throughout the reline thousands of times and
illustrates why a simulation such as this is necessary to provide sensible real world results. It should be
noted that the distributions in Tables 2 and 3 are for illustration purposes only.In the program each of these
actual activities is the result of the combination of many variables.
Table 2 - Liner Retrieval Distribution
Time
(Seconds)
Relative
Frequency
Time
(Seconds)
Relative
Frequency
30
20
45
60
120
300
15
30
10
3
30
50
100
250
6
25
10
4
Average
time
Average
Time
75 Sec
74 Sec
The data populating each variable is mined from video recordings of actual relines. The initial
recordings taken at the four relines mentioned earlier were not complete and were only used to test the
practicality of obtaining data in this way and for some preliminary functional testing. The method proved
successful, however a larger data set will be required for simulations to be statistically accurate.
The relative effectiveness of various tools, for example the time difference between a
THUNDERBOLT 1500 and a THUNDERBOLT 750 Hammer removing the same liner bolt, needed to be
considered. To allow one variable to be used by different sized hammers a theoretical relative
effectiveness index was calculated. This will be validated in the future via experiment. Similar
experiments will be required for comparing rattle guns and suspension methods.
Modelling Overview
Mill Set Up
The modelling process begins by specifying the geometry of the mill and its surroundings and
defining the shape and location of each liner and fastener. A 3D model is generated progressively as data
is entered; this provides visual validation of the progress and accuracy of construction. The area around
the mill is divided into six independent work zones two each on the feed end, discharge end and shell. The
floor levels in each zone are entered and also the height that is accessible at each zone. The charge level
and access height inside the mill are also specified. These inputs result in transparent surfaces at the
heights specified and are provided as visual prompts to the operator when planning a reline scenario.
Figure 1 shows a fully populated model of the outside of a mill.
The next phase of the reline, liner placement, commences once all liners specified for the current
mill position are removed. Because the RUSSELL Twin machines are completely independent, there is an
option in the program to allow either of the twin machines to commence placing liners before the other
machine has completed removal. Liner placement is largely dictated by the type of liner being placed and
the placement variables specified during mill set up.
Running the Scenario
The program, as it executes the scenario, creates an animation of the reline. This is a very useful
tool as it provides visual verification that everything is interacting as expected. The speed of the animation
can be varied however it does significantly slow the program down. Once satisfied that the scenario is
running correctly, the animation can be turned off allowing fast computation. As the reline runs data
relating to the timing of every function is collated and stored in the program. Various reports are generated
that reveal the performance of the reline. Bottlenecks and areas that require focus are also highlighted.
Analysing Results
At the time of writing this paper the program is fully functional however the animation screen is
still in its raw form and the graphical representation of the output data for easy analysis is not complete.
Figure 2 shows the animation screen for the simulation at the time of this paper.
10