Maintenance Scheduling of Earth Moving

Equipment at the Klipspruit Colliery

Albert Greeff
25081439

Submitted in partial fulfilment of the requirements for the degree

Bachelors of Engineering (Industrial Engineering)

in the

Faculty of Engineering, Built Environment, and

Information Technology

University of Pretoria
Pretoria, South Africa

November 2009
 Why build a Model?

We build models to understand a particular phenomena or charac-

teristic of a system. We build models to understand the relationship
between cause and effect. We build models to improve our under-
standing because with a good model comes discovery, with discovery
come understanding and with understanding comes control.
DISCOVERY - UNDERSTANDING - CONTROL

- www.CrowdDynamicsLtd.com, April 2009

i
Abstract

An investigative study of BHP billitons Klipspriut colliery is done to deter-

mine the best design for a maintenance workshop that services track and tyre
based Earth Moving Equipment (EME). The workshop is under construction
and management requires a design that will optimise the number of entities and
minimize the throughput time of the workshop. There are different factors that
in infuence the number of entities ranging from warranty commitment to quality
of service.
By optimizing the throughput of the workshop the mine would have access to
more EMEs which will result in a more productive mine.
The simulation is done with Arena software and is modelled according to a
generic model of maintenance systems. The proposed system design through-
put is used to measure the theoretical designs productivity.
The simulation provides different improvements thus a optimal design can be
chosen depending on managements interest.
The number of shifts worked on the mine is increased with 33 percent if the bays
are divided 3:1 for tyre based services. A 78 percent increase in the productivity
is obtained by extending labour from five days to seven days. This will increase
labour cost but will result in a more productive facility.

ii
Contents

Abstract ii

1 Introduction 1
1.1 Background of BHP Billiton . . . . . . . . . . . . . . . . . . . . . 1
1.2 Introduction of the Problem . . . . . . . . . . . . . . . . . . . . . 2
1.2.1 Klipspruit Colliery . . . . . . . . . . . . . . . . . . . . . . 2
1.2.2 Warranty Service . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Project Aim . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.4 Simulation Modelling . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.5 Project Question . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.6 Project Methodology . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.7 Document Structure . . . . . . . . . . . . . . . . . . . . . . . . . 5

2 Literature Review 6
2.1 Maintenance Scheduling . . . . . . . . . . . . . . . . . . . . . . . 6
2.1.1 Breakdown of Maintenance . . . . . . . . . . . . . . . . . 6
2.2 Technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.3 Conceptual Simulation Model For Maintenance Systems . . . . . 8
2.3.1 Input and Maintenance Load . . . . . . . . . . . . . . . . 9
2.3.2 Planning and Scheduling . . . . . . . . . . . . . . . . . . . 9
2.3.3 Sub Models . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.4 Generic Klipspruit Colliery . . . . . . . . . . . . . . . . . . . . . 11

3 Data Preperation 14
3.1 Conceptual Model . . . . . . . . . . . . . . . . . . . . . . . . . . 14

iii
 3.1.1 Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.1.2 Normal Day-time Service . . . . . . . . . . . . . . . . . . 15
3.1.3 High-Priority Service . . . . . . . . . . . . . . . . . . . . . 16
3.1.4 Availability of Parts . . . . . . . . . . . . . . . . . . . . . 16
3.1.5 Quality Control . . . . . . . . . . . . . . . . . . . . . . . . 16
3.1.6 Breakdowns . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.1.7 Earth Moving Equipment . . . . . . . . . . . . . . . . . . 17
3.1.8 Theoretical Model . . . . . . . . . . . . . . . . . . . . . . 17
3.1.9 Earth Moving Equipment Process . . . . . . . . . . . . . 18
3.2 Computer Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.2.1 Creation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.2.2 Service Decisions . . . . . . . . . . . . . . . . . . . . . . . 20
3.2.3 The Mine . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.2.4 Workshop Queueing Conditions . . . . . . . . . . . . . . . 21
3.2.5 Workshops . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.2.6 Breakdowns . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.2.7 Availability of Parts . . . . . . . . . . . . . . . . . . . . . 25
3.2.8 Resource Scheduling . . . . . . . . . . . . . . . . . . . . . 26
3.3 Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.4 Influential Factors . . . . . . . . . . . . . . . . . . . . . . . . . . 27

4 Results and Discussion 29

4.1 Managenent Proposal . . . . . . . . . . . . . . . . . . . . . . . . 29
4.2 Distribution of Bays . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.3 Workshop Operations . . . . . . . . . . . . . . . . . . . . . . . . 30
4.3.1 Time Extension . . . . . . . . . . . . . . . . . . . . . . . . 30
4.3.2 Weekend Work . . . . . . . . . . . . . . . . . . . . . . . . 31

5 Conclusion 32

A Appedix A - Figures 33

B Appendix B - Arena 36

Bibliography 37

iv
List of Figures

2.1 Maintenance Process . . . . . . . . . . . . . . . . . . . . . . . . . 12

2.2 Planning and Scheduling . . . . . . . . . . . . . . . . . . . . . . . 13

3.1 Major Phases in a Simulation Study . . . . . . . . . . . . . . . . 15

3.2 Top Level of Maintenance Process . . . . . . . . . . . . . . . . . 19
3.3 Submodel of Workshop . . . . . . . . . . . . . . . . . . . . . . . . 22
3.4 Submodel of Breakdowns . . . . . . . . . . . . . . . . . . . . . . 26

A.1 Quality Module . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

B.1 Modules of Arena software . . . . . . . . . . . . . . . . . . . . . . 37

B.2 Top Level of Arena Simulation . . . . . . . . . . . . . . . . . . . 38

v
List of Tables

1.1 Service Details . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

3.1 Parameters to Trigger Service . . . . . . . . . . . . . . . . . . . . 20

3.2 Conditions to Enter Normal Day Workshop . . . . . . . . . . . . 23
3.3 Conditions to enter high priority workshop . . . . . . . . . . . . 24
3.4 Conditions to Bypass Workshops . . . . . . . . . . . . . . . . . . 25

A.1 Earth Moving Equipment to be Delivered . . . . . . . . . . . . . 34

vi
Chapter 1

Introduction

1.1 Background of BHP Billiton

BHP Billiton is the worlds largest diversified resources company, employing
more than 38 000 people in 25 countries with interests in minerals ranging from
such as copper, coal, manganese, iron ore, uranium, nickel, silver to liquefied
natural gas, oil and diamonds(DeepSA, 2008).
BHP Billiton Energy Coal South Africa Limited is one of the largest energy coal
exporters in the world, operating in 5 different collieries (Khulata, Klipspruit,
Optimum, Douglas and Middelburg) in Mpumalanga, South Africa of which
Klipspriut is the youngest and also the main focus of this project. The Klipspruit
Colliery in Ogies is the first new mine to be established by BHP Billiton Energy
Coal South Africa Limited in more that a decade. The new mining activities at
Klipspruit, established in 2003, utilise the latest technologies in order to ensure
that the colliery becomes a leader in the latest production methods, labour
practises, environment-, health- and safety issues(DeepSA, 2008).
Klipspruit currently produces over 4 million tonnes of coal with a workforce of
111 full-time employees and 500 contractors.(DeepSA, 2008)

1
1.2 Introduction of the Problem

1.2.1 Klipspruit Colliery

As mentioned in the previous section, the Klipspruit colliery is a young devel-

opment by Ingwe Coal, a division of BHP Billiton. Ingwes vision is to develop
a mine that is the leader in the latest production methods and technologies in
order to acquire optimal productivity. In essence, if all systems of the mine,
including support systems, deliver optimally the productivity of the mine will
grow at a greater rate than anticipated.
One of the facilities that is in the design phase is a four-bay support system that
services the earth-moving equipment. This includes all the tyre- and track-based
vehicles that are active on-site. Aspects that have to be taken into consideration
when designing the workshop are:

Amount of each type of machinery that has to be serviced

Warranty aspect of each service

Time deviations of services

Breakdowns that cannot be attended to in the field

Amount of bays assigned to each type of vehicle base

The maintenance support system is currently outsourced to an external mainte-

nance company due to the unavailability of the workshop. By the time construc-
tion of the workshop is completed and the workshop is utilised (anticipated to
be completed during June-July 2009) the delivery of all the earth-moving equip-
ment will be finished. From here onwards minimal outsourcing will be required.

1.2.2 Warranty Service

One of the most important parts of buying an EME is acquiring an after sale
agreement that includes a warranty for the equipment, which is only applicable
to an EME if a service is conducted every 250 worked hours. The supplier offers
a lenient gap of 50 hours for the service to be conducted before the warranty
becomes obsolete.

2
 Service Scheduled Hours Duration of
Numbers of Machine Service(Hours)
1 250 5
2 500 5
3 750 4
4 1000 8
5 1250 4
6 1500 5
7 1750 4
8 2000 48

Table 1.1: Service Details

The warranty is of fundamental importance to the mine, thus the Earth Moving
Machine (EMM) will be relieved of working duty if it approaches the 50 hour
gap deadline. The service time differs depending on the amount of hours that
an EME has worked. The different times for services are shown in table 1.1.

1.3 Project Aim

The current design of the workshop assigns two bays to track-type EMEs, and
two are assigned to tyre-based EMEs, even though the number of track- and
tyre-based vechiles differs. There are 10 track-based vehicles and 30 tyre-based
vehicles active on the mine (the description and amount of EMEs that will be
delivered is shown in appendix A).
An investigation into the effect that the number of bays assigned to track- or
tyre-based EMEs have on the productivity of the mine, will be done by means
of simulation. The influence of shift time deviation will also be investigated.
Other workshop distributions will be investigated to determine the optimal value
of assigned bays to track and tyre based vehicles. The simulation will generate
mathematical data regarding the queue lengths, waiting time and resource uti-
lization from which conclusions will be drawn.
It should be noted that the current situation at other maintenance workshops is
not of concern, but the opportunity exists to make these workshops more pro-

3
ductive and thus increase the productivity of the mines individual output. The
optimisation can be done by integrating the data of the Middelburg mine with
the Arena simulation program to find a optimal solution. A more productive
maintenance program results in less vehicles to queue and this will in turn result
in a bigger workforce on site, as well as equipment warranties that last longer
due to better scheduling of work.
If simulation is optimal and the workshop does not deliver sufficient results,
mathematical optimization should be used to determine the amount of services
to be outsourced in order to create a profitable workshop that contributes to
the efficiency of the mine.

1.4 Simulation Modelling

According to LTS Consulting a simulation model is an effort to replicate a
real life situation to investigate the implications of the system and how it will
be affected by varying circumstances. By changing variables, predictions can
be made about the performance of the system. Discrete event simulation is
often used to model various systems in industries ranging from health care
to manufacturing. Arena simulation software is used in the Klipspuit colliery
project(LTSConsulting, 2008).

1.5 Project Question

With the information regarding the Klipspruit colliery and simulation modelling
the objective of the project is to develop a simulation model that will investigate
the maintenance scheduling at the workshop of the mine. Thus, a research
question is compiled as follows:

Given the amount of Earth Moving Equipment, maintenance equip-

ment, shift times and workbays, what is the optimal assignment of
bays and shift durations to be used in order to maximize productiv-
ity of Klipsruit collierys maintenance workshop and mine?

4
1.6 Project Methodology
By executing the projects aim, the main deliverable is a working simulation
model of the EME workshops which is applicable to the Klipspruit colliery. The
document will show different scenarios of the workshop, of which changes will
be made to assigned number of bays and working hours.
The EMEs will only be affected by the service time, the amount of bays (assigned
to different types of EMEs) and unavailability of parts. Areas which are not
considered but may have an influence on the real-life situation is man-power
availability, working enthusiasm and travel time from the mine to workshop.
It should also be noted that the simulation study only focuses on the tyre based
vehicles as this is the type which has the greatest possibility to create bottle
necks. When the simulation is finished track based vehicles will be integrated
into the simulation to verify the decisions regarding the variable aspects.
Assuming that each week has five working days, of which all are worked every
week, and two days for a weekend (where only the mine is operational and not
the workshop) the simulation follows a fairly trivial flow. Entities are created
and move towards the mine. This is a seven day schedule and entities are only
directed towards the workshop when their service is due. After the workshop is
completed the entities return to the normal schedule of working on the mine.
The simulation will use mathematical distribution to determine the time values
for each process. These distributions are computer generated with the time-
study values as inputs.

1.7 Document Structure

In Chapter 2 maintenance scheduling and simulation is reviewed in literature.
The most important areas of a generic maintenance system are discussed and
compared to the simulation that is planned at Klipspruit. Chapter 3 will dis-
cuss the conceptual model and assumptions regarding the model, as well as the
computer model.
Chapter 4 will analyse the results of the simulation and provide improvements,
while Chapter 5 concludes findings of the project and offers suggestions for
future research relating to the development of maintenance modelling.

5
Chapter 2

Literature Review

2.1 Maintenance Scheduling

Mr S.A Oke Oke (2004) reasons that maintenance scheduling spans across a
wide variety of research fields, mostly dominated by mathematics, computing
and engineering. Maintenance scheduling is applicable in the aviation industry,
vechile fleet management, process industry, road and railway maintenance and
refinery- and production facility(Oke, 2004).
Maintenance scheduling was first introduced in 1972. The development of the
field was slow due to the complexity of the mathematics involved. Ground
breaking work was done on maintenance scheduling using integer- and linear
programming, simulation modelling and a probabilistic approach(Oke, 2004).

2.1.1 Breakdown of Maintenance

The complexity of the procedures that machines undertake is directly related

to the level of skill required to maintain the machine. Production work time
is maintained by developing maintenance groups to ensure that workers do not
experience down-time.(Paz and Leigh, 1994)
Control measures are also implemented to control the inventory of special parts.
Mr. Noemi PazPaz and Leigh (1994) reasons that as reduction of repair cost
became important, periodic inspection and repair became part of the mainte-
nance function. It is essential to use scheduling effectively, as it is the main

6
factor that determines productivity. Studies have shown that without mainte-
nance scheduling productivity of workers reached an estimated average of 40
percent. Due to this fact certain maintenance policies exist, which are covered
in the next section(Paz and Leigh, 1994).

Maintenance Policies

Maintenance is implemented by adopting certain policies depending on entity

type, process type and duration of the process. Maintenance policies can be
divided into two categories(Paz and Leigh, 1994):

1. Policies for reducing frequency of failures are further divided into:

Preventative Maintenance: Maintenance to keep on item in specific

condition. Based on earliest expected failure time.

Early Equipment Replacement: Replaced before first failures can

occur.

Predictive Maintenance: Maintenance is done in advance of break-

down.

Rebuilding of equipment (Paz and Leigh, 1994)

2. Policies for reducing the severity of failures are further devided into:

Speeding the repair task by modular equipment design

Alternate output during repair by alternate job routing (Paz and

Leigh, 1994)

Historical data has to be analysed to schedule preventative maintenance, whereas

predictive maintenance is performed by measuring a specific range of data con-
cerning the output of machine that has to conform to certain limit values.
If the facility has a machine that is easily repairable and it is not critical to part
productivity, the run-to-break policy can be implemented. This policy implies
that the machine runs until it breaks and it is then repaired or replaced(Paz
and Leigh, 1994) .

7
2.2 Technique
Due to the complexity that the Klipspruit model leads to when using mathe-
matical programming, simulation modelling will be used. Simulation modelling
will provide the opportunity to reconstruct the model towards a more complete
and optimized answer. Mathematical scheduling will require a program that
uses complicated mathematics to obtain an answer that requires editing.
The simulation model can also be repeated with different inputs and will require
that the optimal situation be determined by inspection. The amount of differ-
ent factors that constrains a mathematical programming problem can cause
this amount of optional answers to exceed a million. Thus, with the use of a
generic model for maintenance systems that is reviewed in the following section,
the Klipspriut Collieries maintenance problem will be solved using simulation
modelling.

2.3 Conceptual Simulation Model For Mainte-

nance Systems
The model, that is designed by Duffuaa, Ben-Daya, Al-Sultan and AndijaniDuf-
fuaa et al. (2001) from King Fahd University of Petroleum and Minerals, is bro-
ken down into different sub models to indicate the importance and the effect
that some of the areas have on the model(Duffuaa et al., 2001).
A model requires planning, scheduling control and deployment to be developed
for planned and unplanned maintenance. The key areas that in uence a main-
tenance system are:

maintenance load

maintenance resources

maintenance planning, scheduling and execution;

maintenance support functions

reports and performance measures(Duffuaa et al., 2001).

8
Maintenance interacts with many different areas of a business whether support-
or main-function. Figure 2.1 (page 10) indicates how maintenance interacts
with its own support functions as well as how maintenance interacts with the
rest of the organisation. The maintenance function has a continuous feedback
and improvement function in maintenance planning as well as in its interaction
with the rest of the organisation. It is also important to take note of the priority
listing of entities which serves as a constraint.(Duffuaa et al., 2001)
Maintenance scheduling is divided into different sections, of which the critical
areas are analysed.

Figure 2.1: Maintenance Process

(Duffuaa et al., 2001)]

9
2.3.1 Input and Maintenance Load

All simulations need an input or a create module. The function is to generate

different values for certain entities and areas, for example priority, time values
and quality values.
This will create the maintenance load that divides the generated load into a
critical load and a super load of which both consist of planned and unplanned
maintenance.
The load which relates to planned maintenance is known in advance and this
determines the crew mix and job sequence. The unplanned maintenance is trig-
gered by a probability distribution that is based on time studies. All planned
and unplanned maintenance is combined to form the maintenance load.

2.3.2 Planning and Scheduling

This is regarded as the most important module of the whole simulation as

it forms the base of the simulation and all areas interact with the simulation
through this process. A detailed diagram of the process ow is provided in Figure
2.2. The purpose of this module is to plan, schedule, monitor and execute all
the jobs that move through the maintenance department. The following steps
form the planning and scheduling module:

1. The maintenance load is the starting point of this process. The first
step differs, depending on whether the type of maintenance is planned
or unplanned. The crew size, mix and priority are known for planned
maintenance as well as the entitys priority. The same data is generated
for the unplanned maintenance with a probability distribution.

2. Maintenance spares and material availability is checked by using a sub

model.

3. Equipment and tool availability is checked by using the appropriate mod-

ule which is described in Section 2.3.3

4. The job ordered is prepared and scheduled if the resources are available.

5. The job is executed and data pertaining to the job is recorded.

10
Planned Maintenance Unplanned Maintenance
Maintenance Load

Crew mix, size and trade Generate crew mix, size and

sequence given as attribute trade sequence

Standard time given as Generate time standards

attribute

Priority given as attribute Generate priority

Check maintenance spares and materials availability

using materials and spares supply module.

Check tools and equipment availability

using equipment availability module.

Prepare work order

Schedule job

Execute job

Check Quality using quality module

Collection of statistics for performance measure

Figure 2.2: Planning and Scheduling

(Duffuaa et al., 2001)]

11
 6. The quality of the job is checked by using the quality module. If the
quality of the job does not meet the required standard the entity is sent
back for the job to be repeated. The data is recorded for performance
measures(Duffuaa et al., 2001).

2.3.3 Sub Models

The tools and equipment as well as materials and spares are sub models of the
base of the simulation. These modules check the availability of the specifc area.
If this item or area is not available, the execution of the model is delayed until
it becomes available.
The quality module is also a sub model but if the quality is not up to standard
the entity is redirected to be serviced again. This process is displayed in Ap-
pendix A.
Performance measures are put in place to analyse the system and to ensure that
continuous improvement occurs. The performance measure can be divided into
three parts, namely:

Maintenance effectiveness

Maintenance administration

Maintenance cost (Duffuaa et al., 2001)

2.4 Generic Klipspruit Colliery

The model provides a good framework for the simulation of the Klipspuit colliery
to be based upon. The Klipspruit colliery is also a maintenance system that
has interaction of different components such as quality, equipment and different
priority levels. The most important focus of the simulation is to optimise the
through put and the model provides a broken-down view of each area which will
indicate problematic opportunities.
Areas that are not the same when comparing the model with the colliery are
manpower availably, as well as the super load of the colliery which will not
be taken into account. The workshop bays are scheduled as equipment, thus
equipment availability will not be taken into account either.

12
Chapter 3

Data Preperation

3.1 Conceptual Model

It is imperative to understand the flow of entities between processes before
conducting the computer simulation. A generic approach to the major phases
of simulation modelling is shown in the Figure 3.1 (page 14). The conceptual
model furthers the understanding of the compilation and is the genesis of nearly
all simulation models.

3.1.1 Method

The first step in creating the conceptual model is to obtain enough information
from the management of the colliery to get an overview of the current oper-
ations. The current practises were understood by conducting site visits and
conducting interviews with the designers and the engineers working on the col-
liery.
A baseline for the conceptual model was constructed and approved by the man-
agement. The baseline was refined and improved through prolonged research
and brainstorming(Snyman,2008).
The output of the simulation will identify congestion of entities at key areas.
Some of these areas are the normal day time- and high-priority services, quality
control, breakdowns and parts availability.

13
 Start

Study the objectives of the system variables that effect the

Define the problem
achievement of objectives

Construct the simulation model Specification of variables, parameters, decision rules and
probability distributions.

Specify values of variables and Determination of starting conditions and run length
parameters

Run the simulation

Determination statistic tests and compare with other

Evaluate results
information

Verification and validation

Propose changes

Stop

Figure 3.1: Major Phases in a Simulation Study

(Snyman, 2008)]

3.1.2 Normal Day-time Service

This area is the collierys main workshop and all earth moving equipment is
scheduled to move through this workshop. The workshop runs on ten-hour
shifts where operations commence daily at 5 am. The 10 hour shifts will be
extended until the last service of the day is completed.
EMEs (Earth Moving Equipment) are flagged every 250 hours to advance to
the workshop in order to undergo a mandatory warranty service. These services
vary in length from 4-, 5-, 8- to 16 hours.

3.1.3 High-Priority Service

The warranty of the EME allocates a 50 hour gap for the services to be com-
pleted. If EMEs approach the workshop and the queue for that current day is
full they are transferred back to the mining pit to continue work and will then
be called in for service the following day. If the EMEs are not serviced before

14
the gap is reached they are removed from active service and are placed on a
high-priority waiting list to be serviced.
These high priority EMEs will be serviced before any of the normal day time
services are performed. The high priority service is conducted within the same
workshop as the normal day time service.

3.1.4 Availability of Parts

Due to human error and deviations from planned situations, parts for services
are not always available. In this case the servicing of the EME that is queued for
service is placed on halt while the unavailable parts are delivered. The EME for
servicing which is delayed, does not occupy a bay, thus service of other EMEs
continues as scheduled.

3.1.5 Quality Control

To keep the mine working at an efficient rate the quality of a sample of the ser-
vices should be inspected. If the quality of the sample does not meet acceptable
standards the services have to be repeated. A vehicle that fails the quality test
is not marked as high priority, due to its ability to work if there is still time
remaining on the warranty gap. Therefore, the destination of a vehicle which
fails the quality test depends on the amount of hours that the vehicle has been
in use.

3.1.6 Breakdowns

The occurrence of a breakdowns is an unplanned event and the level of repair

required should be investigated and identified. An EME can be repaired in the
field if the damage is minor or if the repair process is trivial and does not require
workshop equipment.
If it is a bigger problem, the EME requires a workshop and a high priority repair
is to be conducted, which implies that the resources used to service the normal
day time service and the scheduled high priority services will instead be used to
repair the EME that has broken down.
Bottlenecks in the breakdown department are parts availability, quality control

15
and resource allocation in the workshops. Breakdowns do not have their own
allocated workshops, thus these repairs are also done in the normal day-time
service workshops. The field service is not done within a workshop so the delay
only impacts the relevant EME and not the rest of the mine.

3.1.7 Earth Moving Equipment

Only certain amount of the EMEs that are delivered for servicing are critical
to the working environment and the efficiency of the mine. These EME will be
referred to as the critical load, while the rest of the EMEs will be referred to as
the mass load.
The simulation will mostly deal with the critical load as its services and down-
time impacts on the efficiency of the mine. The model simulates the tyre based
EMEs as these are the largest in number.

3.1.8 Theoretical Model

As the efficiency of the mine is determined by the amount of coal that the
colliery can produce, the efficiency of the EMEs will be the determined by the
amount of coal that can be transferred in a certain time frame. To ensure that
the workshop contributes to the efficiency of the mine it has to service vehicles
in the shortest possible time and highest quality manner possible in order to
avoid congestion in the workshop.
To ensure that our model mirrors reality as closely, possible time studies will
have to be done on the following aspects:

working time per day of different EMEs

actual duration of a scheduled 4-,5- and 8-hour service

waiting time for delivery of unavailable parts

duration of a breakdown (both services performed in the field and the

workshop)

Data-gathering will be done on:

16
 unavailability of parts

quality-test failure

breakdowns that occur

The simulation software used to model the system provides an application to

determine the mathematical distribution for the data gathered. This makes it
possible for computer-generated distribution functions to be used in the com-
puter model. This is also the required input for Arena.

3.1.9 Earth Moving Equipment Process

For the purpose of this study, a tallied hour rate will be used for the tyre-,
lubrication- and mechanical services. This rate will be the assigned rate of 4-,
5-, 8- or 16 hours.
The time it takes one EME to travel between the mine and the workshop is not
taken into account as the mine and the workshop is situated very close to one
another.

3.2 Computer Model

The theoretical model that is described in the previous section of this document
is altered into a computer simulation. The simulation will be developed with
the use of Arena software. At the end of the simulation a report is generated
to obtain results from which conclusions regarding the system can be drawn.
Some of the statistics that are included in the reports are queue-length and
queue-time values, entity statistics, process details, time related attributes and
user-specified statistics.
The simulation model will be constructed around the generic model Duffuaa et.
al. Figure 3.2 (page 18) provides a graphical representation of the top level of
the simulation model. The layout of the model in Arena and a short description
on Arena is provided in Appendix B.

17
 Creation Assign start up Due for Assign service Maintenance
values service? time Process

Daily work Breakdown Maintenance

process occur? process

Figure 3.2: Top Level of Maintenance Process

3.2.1 Creation

All the entities are created in 12 different create nodes which are combined into
one module on Figure 3.2. Different quantities of tyre-type vehicles are created,
of which the Caterpillar 777 has the most entities. The quantity of each entity
can be seen in Table A.1 in Appendix A.
The entity will move towards an assign node. There are two attributes assigned
to each entity namely:

Worked hours: All the EMEs will be assigned a value from a mathematical
distribution in order to determine the amount of hours that the EME has
already spent on the mine. All attributes are assigned in values between
0 and 250, representing the number hours prior to the first mandatory
service.
The assumption that EMEs have different work levels has to be made in
order to exclude poor resource utilisation and to bring the model closer
to reality.

Level of service: This attribute has to be assigned to the EME to indicate

whether the entity has any time left on the allocated warranty gap. This
attribute, in correlation with the work hours-attribute, triggers if an entity
should be removed from the mine and queued for high level priority service.

18
3.2.2 Service Decisions

All the entities that have been created are sent through a decide node. The de-
cide node has two directions, according to the parameters that were assigned
by the previous module, to which the entities are sent. The EME can be sent
to the mine or to the workshop in order to be serviced. Table 3.1 on page 18
depicts the parameters for which an entity is sent for service.
If the worked hours for the EME fall between the stated values and the service

Parameter Worked Hours Gap Level of Service

1 250 and 300 0
2 500 and 550 1
3 750 and 800 2
4 1000 and 1050 3
5 1250 and 1300 4
6 1500 and 1550 5
7 1750 and 1800 6
8 2000 and 2050 7

Table 3.1: Parameters to Trigger Service

level is as indicated then the entity will be directed to the workshop, otherwise
it will be directed to the mine.

3.2.3 The Mine

The simulation for the mine is very trivial and forms part of the model as a sub
model. As the entities enter the sub model, labelled Daily Work Process (Fig-
ure 3.2 page 18), they arrive at a process node, which determines the amount of
work that the EME does that particular day. This is done by using a random
distribution. There is no queue at this process.
Following the process node is an assign node to update the worked hours at-
tribute. The amount of hours that the EME spent on the mine is added to this
attribute. The EMEs exit the sub model and approach the breakdown decide
node.
This node decides if a breakdown occurred. This choice is made by probability

19
- in other words, it is achieved by analysing historical data. If the EME suffers
a breakdown it is moved to the breakdown sub model. Alternatively it is sent
back to the Service Decide-node.

3.2.4 Workshop Queueing Conditions

If the entitys attributes fall within one of the parameters specified in Table 3.1
(page 19), the EME will move towards the workshop to be serviced.
The assign nodes, labelled Assign Service time, introduce a new attribute
towards the entity, mainly the service duration. The value assigned to this
property depends on the level of service (from level 1 to level 8) that the entity
requires. The service duration varies between four hours and two days. Table
1.1 (page 3) indicates the different service durations.
The assign nodes connect to a sub model labelled Maintenance Process. Figure
3.3 (page 21) provides a view of the layout of this sub model. There are mainly
three different results that can occur as entities move through this sub model.

Normal Day Workshop

There is only one set of requirements to enter an entity into this queue. The
first node is a decide node which determines the total value of all the service
times in the queue of all the workshops. If this value is more than 10 the entity
will bypass the first workshop-option, as the queue for that day is already full
and the entity will not be service on that particular day. To enter the Normal
Day workshop queue the sum of the value must be less than ten.
Another decide node follows, which stipulates that the time of arrival must be
between 5am and 3pm to enter the Normal Day workshop queue, thus the entity
arrives in the given window.
The final decide node that the entity must pass in order to enter the specific
workshop determines whether the current day falls on a weekday as the workshop
is not active on weekends. If the entity complies with this parameter it will be
included in the Normal Day workshop queue. Table 3.2 (page 22) summarises
the conditions for entry into the Normal Day queue.

20
 Is Is Queue Is it a Required Normal priority
M.P
workshop length < weekday? parts Maintenance is
open? 10? available? done

Wait for parts

M.P

Is warranty Remove entity Required High priority

going from service parts Maintenance is
expire? done
available?

21
Wait for parts

Figure 3.3: Submodel of Workshop

 Condition Yes or No
Queued entities service time less than 10 hours? Yes
Arrival time between 5 a.m. and 3 p.m? Yes
Arrival on Saturday or Sunday No

Table 3.2: Conditions to Enter Normal Day Workshop

High-Priority Queue

There are three ways of entering this queue. As the entity arrives at the first
decide node the sum of the service times of the queue length can be less than
10 to send it to the next decide node, which depends on the time at which the
entity arrives at the workshop - this can also be in the time window. If the
entity enters the final decide node either on a Saturday or on a Sunday the final
condition will be true (if this was a weekday it would be identical to the first
case and it would qualify for the normal week day queue).
Meeting the above conditions will send the entity to the decide node named
Warranty expire node. This node examines the amount of time that is left on
the warranty gap before it becomes obsolete. If the value is less than 20 the EME
is taken out of service and placed in the high priority queue. This workshop
will service its load before servicing the normal work day load. Therefore the
entity should have a gap value of less than 20 hours to access this queue.
The second and third sequences of conditions to enter the high priority queue
are listed in Table 3.3 (page 23). An entity must be directed away from the first
workshop and should have a gap value of less than 20 hours.

Bypass All Workshops

The three conditions that were stipulated for the high priority queue are the
same conditions that the entity has to meet in order to bypass all the queues
and continue work on the mine. The only difference is that the warranty gap
should be in excess of 20 hours. The warranty expire decide node will send the
entity back to the mine. Table 3.4 (page 24) summarises these conditions.

22
 Condition Yes or No
First Condition
Queued entities service time less than 10 hours? Yes
Arrival time between 5 a.m. and 3 p.m? Yes
Arrival on Saturday or Sunday Yes
Worked hours within 20 hours of the expire value? Yes
Second Condition
Queued entities service time less than 10 hours? Yes
Arrival time between 5 a.m. and 3 p.m? No
Worked hours inside 20 of the expire value? Yes
Third Condition
Queued entities service time less than 10 hours? No
Worked hours within 20 hours of the expire value? Yes

Table 3.3: Conditions to enter high priority workshop

3.2.5 Workshops

The entities will enter the workshop and will be serviced according to the service
duration attribute that is assigned to them. As there are only two workshops
(starting model) the high-priority queues load will be serviced first and then
the normal day-load. If the entity receives a two day service the process will
be reset and the scheduled services will be repeated from one to eight at the
appropriate time.

3.2.6 Breakdowns

When an EME leaves the mine and the breakdown decide node indicates that a
breakdown has occurred the entity would move to a sub model labelled Break-
down workshop.
Figure 3.4 (page 25) displays a flowchart of the sub model. The first module is
a decide module that uses a random distribution to indicate if the repair can be
done in the field or if the entity should be sent to the workshop.
If field service can be performed a hold entity is put in place to delay the entity
under review. For this service no resource that is of importance is used.

23
 Conditions Yes or No
First Condition
Queued entities service time less than 10 hours? Yes
Arrival time between 5 a.m. and 3 p.m? Yes
Arrival on Saturday or Sunday Yes
Worked hours inside 20 of the expire value? No
Second condition
Queued entities service time less than 10 hours? Yes
Arrival time between 5 a.m. and 3 p.m? No
Worked hours inside 20 of the expire value? No
Third condition
Queued entities service time less than 10 hours? No
Worked hours within 20 hours of the expire value? No

Table 3.4: Conditions to Bypass Workshops

When the entity has to go to the workshop, the highest priority is assigned to
the entity and it is placed in the relevant queue. When the entity is repaired no
value changes are made to any of the attributes as it was not a warranty based
service.

3.2.7 Availability of Parts

Before an entity moves into a queue the necessary parts for repair should be
available. With the use of a decide node, which uses a random distribution,
parts availability is checked. If the parts are not available the servicing of the
entity is delayed until the parts arrive. This is applicable to the Normal Day
queue, the High Priority queue and the Breakdown queue, as well as to field
service.

3.2.8 Resource Scheduling

The workshop entities run on two schedules, namely Normal Day schedule and
Graveyard schedule. On a Normal Day schedule two workshops are available
from 5am to 3pm or until the work load for the day is completed. This schedule

24
 Should EME Required High priority
B.P be brought to B.P
parts Maintenance is
workshop for
service? available? done

Wait for parts

Required Field service is

parts done
available?

Wait for parts

Figure 3.4: Submodel of Breakdowns

applies to all three the types of workshops mentioned above.

The Graveyard schedule only services breakdowns and only has one workshop
available. This resource is available from 3pm until 11pm or until the load for
the day is completed.

3.3 Output
The simulation model for the Klipsruit Colliery runs for a year. When the
simulation is complete, statistical data is generated, at which point entity-,
queue- and resource data is available. The most important data gathered is:

waiting time and number of entities in Breakdowns queue

waiting time and number of entities in Normal Day queue

waiting time and number of entities in High Priority queue

the total time for entity to move through a scheduled service

25
 the amount of entities taken out of service due to warranty expiry

amount of EMEs bypassing the workshops

amount of times the EMEs have had to wait for parts

The desired output of the model should minimise all of the above mentioned
values. The EMEs should move through the workshop while minimising the
length of the queue. The High Priority queue should not take any entities out
of service due to warranty issues and parts availability should improve.

3.4 Influential Factors

The following factors in influence the model output and take the form of variable
data that determine the efficiency of the workshops:

random distribution of working hours

probability of breakdown occurrence

random distribution of available parts

probability of workshop breakdowns

amount of bays assigned to tyre based EMEs.

The factors that in influence the model that take the form non-variable data
are:

the minimum amount of vehicles in the system

the warranty conditions

service times

total amount of bays

resource availability.

26
Chapter 4

Results and Discussion

The Klipspruit Colliery will first be simulated in the way that management
proposed the implementation of the maintenance workshop. This simulations
results will be compared to other simulations of the collierys working with
different variable data, which will provide management with a few different
options, as they should decide what is feasible. Arena simulates the colliery
for a year and this simulation is repeated ten times in order to generate an
acceptable average.

4.1 Managenent Proposal

The data that is generated from this particular simulation will be compared to
all the other simulations as this is the current design. The bays are distributed
evenly between the tyre and track based EMEs.
The maintenance workshop has three different areas namely the Normal- , High
Priority-, and Breakdown workshops, of which the High Priority queue has the
largest waiting time (almost two hours). Of the 268 services that occurred dur-
ing the year 100 were of high priority, where the entity was removed from service
to meet its warranty obligations.
Some entities approached the workshop just to be sent back to the mine due
to different reasons which included day of the week, time of day or the queue
length. This event happened 312 times.

27
4.2 Distribution of Bays
Of the forty EMEs on the Klipspriut Colliery twelve are track-based and the
rest are tyre based - thus, a different distribution of work bays is the next sim-
ulation that will be investigated. If the simulation has a distribution of 3:1 for
tyre based EMEs the average waiting time in the queue is less. The workshop
has a through put of 392 entities, of which 157 are high priority and will be re-
moved from the mine. This is more than the management proposal but overall
it proves an improvement of productivity for the workshop.
Due to this change the mine increased the amount of shifts that workers worked
with 33 percent which will lead to more coal being mined, and to a more prof-
itable organisation. The increase in shifts that the mine operates on is the
reason why the number of High Priority entities increased.
The one bay of track EMEs will cause a small amount of drag but the system
will be more efficient with the bays divided in the same ratio as that of the
EMEs.

4.3 Workshop Operations

Currently the mine has two shifts, one where two bays operate from 5am to 3
pm and another where one bay operates from 3pm to 11pm. These shifts are
applicable to every day of the week except Saturday and Sunday. A simulation
will be done to extend the workshop hours for two bays until 11pm as well as
to investigate the outcome should the workshops be made operational during
weekends.

4.3.1 Time Extension

With the workshop running 16 hour shifts it will have to be divided into two
shifts. This will result in bigger salary expenses but the average waiting time for
a service will be lower than the current design. A massive 393 services occurred
during the year of which only 57 were high priority services. The number of
shifts that were worked during the year were 6960, which is less than the three
bay simulation.
This option results in increased expenses but proves a feasible productivity.

28
4.3.2 Weekend Work

The mine is currently working on a 24-hour shift, seven days a week, while the
maintenance workshop is only working 10 hours a day five days a week. If the
workshops are extended to seven days a week the improvement to throughput
is conciderable. A total of 476 entities are serviced of which only 19 is removed
from service and 136 are sent back to the mine without service. This equals
a productivity increase of 78%. The mine will have a 12% increase in shifts
compared to the management proposal.
By extending the work through weekend it is important to remember that ad-
ditional cost such as over time payment for maintenance workers will have to
be included in the budget.
If the weekend work is extended to 16 hour shifts, the model will be more
productive but the expenses increase more than the productivity does.

29
Chapter 5

Conclusion

The simulations results provide opportunities for improvement in different ar-

eas. The productivity of the mine is enhanced by 33 percent if the layout of the
workshop is changed to 3:1 in the favour of tyre-based vehicles. This is a logical
approach, as the EMEs of the mine is divided in the same ratio. This decreases
the percentage of EMEs that are removed but does not solve the problem, as
38 percent of the services are still high priority.
In comparison to this, if the shift durations are prolonged to encompass week-
ends the productivity of the mine increases to 18 percent but the workshop
shows a increase in throughput of 78 percent.
If the mine needs to be optimized the different bay distribution is the best option
as this will financially be the optimal solution. The change in shift times would
be the best option if the workshop needs optimization. Maintenance workshop
labour costs will increase by a minimum of 40 percent.

Another flaw in the human character is that everybody wants to

build and nobody wants to do maintenance.

Kurt Vonnegut (1922 - 2007)

30
Appendix A

Appedix A - Figures

31
Quantity Mode Description
2 D11T Track Type Tractor
1 D10T Track Type Tractor
3 PV275 Overburden Drills
2 EMC720 Atlas Copco EMC 720 Drill Rig Coal Drill
1 834H Wheel Dozer
4 993k Wheel Loader(Front end Loader)
12 777F Off-Highway Truck(Haul Truck)
2 16M motor Grader
1 422 Backhoe Loader
1 988H Tyre Handler
1 988H Cable Realer
2 246B Bobcat
1 CAR740 Water Tanker
2 CAT740 Diesel Tanker
2 Massey Ferguson Tractor
1 UE 150 Tonne Lobed
1 CAT IT 14 G Integrated Tool Carrier
1 CAT TH 414 Telescopic Handler

Table A.1: Earth Moving Equipment to be Delivered

32
 Q

Generate a random number

R form U (0;1)

Job is considered complete Check job quality

r < Pq

Job is returned back from

rework

Return to Planning and

Scheduling Module

End

Figure A.1: Quality Module

33
Appendix B

Appendix B - Arena

Arena uses generic flow chart modules to illustrate the process that is simulated.
The EMEs in the simulation are entities and they move between modules or
nodes. These entities are created in a Create-module. This module can edit the
distribution in which entities are created, the time between entity arrival and
the names of the entities.
The entity would then move from the creation node to the module connected
to it. This can be any module except a Dispose-module. The dispose module
is used to take the entity out of the simulation - this is normally the final part
of the simulation. The current simulation doesnt use a dispose node as all the
vehicles stay on the mine and it is important to update a vehicles properties
regarding the model as its time spent on the mine increases.
A decide node is used to execute decisions. This can be done either by chance
or by condition. The decide node consists of the ability to provide the user with
two or more options.
In the simulation of the Klipspruit colliery an assign node follows the create
node. An assign node has the ability to assigns entity pictures, attributes, vari-
ables and the type of entity.
A process model is one of the most important modules in the simulation model
as this is the module that uses resources to process the entities. The dura-
tion, resource and delay type can be assign. A schedule to determine the re-
sources working hours can be defined and the amount of resources utilised can

34
be changedcite.
The hold entity is nearly the same as the process entity due to the fact that it
occupies the entity for a specified period of time. This time delay value can be
changed. A Record node captures the data, at a specified point in the model.
All the above mentioned modules are shown in Figure3.

Figure B.1: Modules of Arena software

35
 Cre a te tri p l e 7 s ta rt u p v a l u e s 1

0
1 2 6 s e rv i c e s
Cre a te fro n t e n d
loader s ta rt u p v a l u e s 2
s erv ic e or no s erv ic e
0
3 5 7 s e rv i c e s
Cre a te m o to r g ra d e r Else
s ta rt u p v a l u e s 3 wo rk s h o p fo r n o n
b re a k d o wn s
0

4 s e rv i c e 0
Cre a te wh e e l d o z e r s ta rt u p v a l u e s 4

0
pit
c re a te b a c k h o e
loader s ta rt u p v a l u e s 5 8 s e rv i c e

0 0

36
Cre a te ty re h a n d l e r s ta rt u p v a l u e s 6

0
0
Tr ue wo rk s h o p fo r
Break down b re a k d o wn s
Cre a te wa te r ta n k e r s ta rt u p v a l u e s 7

0 0
0 Fals e

Cre a te 1 d i e s e l
ta n k e r s ta rt u p v a l u e s 8

Figure B.2: Top Level of Arena Simulation

Bibliography

DeepSA (2008). A celebration of the south african mining industry. Deep SA,
pages 124125.

Duffuaa, S. O., Ben-Daya, M., Al-Sultan, K. S., and A.Andijani, A. (2001).

A generic conceptual simulation model for maintenance systems. Journal of
Quality in Maintenance Engineering, 7(3):207219.

LTSConsulting (2008). LTS consulting industrial engineers.

http://www.ltsconsulting.co.za/index.php?id=62.

Oke, S. A. (2004). Maintenance scheduling: Description,status and future di-

rections. SA Journal of Industrial Engineering, 15(1):101117.

Paz, N. M. and Leigh, W. (1994). Maintenance scheduling: issues, results and

research needs. International Journal of Operations and Production Manage-
ment, 14:4747.

Snyman, V. (2008). Sasol Crude Oil Pipeline study. PhD thesis, University of
Pretoria.

37