Thesis

See discussions, stats, and author profiles for this publication at: https://www.researchgate.
net/publication/280246123
Reliability and Life Cycle Cost Modelling of Mining Drilling Rigs
Thesis · December 2014

DOI: 10.13140/RG.2.1.1667.8243
CITATIONS READS
2 1,226
1 author:
Hussan AL-Chalabi
Luleå University of Technology
21 PUBLICATIONS 38 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
AtlasCopco View project
Ph.D Candidate View project
All content following this page was uploaded by Hussan AL-Chalabi on 22 July 2015.
The user has requested enhancement of the downloaded file.

DOC TOR A L T H E S I S
Hussan Saed Hamodi Al-Chalabi Reliability and Life Cycle Cost Modelling of Mining Drilling Rigs
Department of Civil, Environmental and Natural Resources Engineering
Division of Operation, Maintenance and Acoustics
Reliability and Life Cycle Cost Modelling
ISSN 1402-1544
of Mining Drilling Rigs
ISBN 978-91-7583-043-8 (print)
ISBN 978-91-7583-044-5 (pdf)
Luleå University of Technology 2014
Hussan Saed Hamodi Al-Chalabi

Reliability and Life Cycle Cost Modelling
of Mining Drilling Rigs
Division of Operation, Maintenance and Acoustics

Luleå University of Technology
SE -- 971 87
Sweden

Printed by Luleå University of Technology, Graphic Production 2014
ISSN 1402-1544
ISBN 978-91-7583-043-8 (print)
ISBN 978-91-7583-044-5 (pdf)
Luleå 2014
www.ltu.se
Dedicated
to my parents, Khalida and Saed,
to my family, Amani, Abdulmalek and Yaman,
for allowing me to be part of their life and love.

ACKNOWLEDGMENTS
The research work presented in this thesis has been carried out during the period 2011 to 2014
at the Division of Operation, Maintenance and Acoustics at Luleå University of Technology
(LTU). The research programme was financed by Atlas Copco, Boliden Mineral AB and LTU.
First of all, I would like to express my deepest gratitude to my main supervisor, Professor Jan
Lundberg, who has enriched my knowledge through stimulating discussions and fruitful
guidance. You have always believed in me and shown a positive attitude.
I would like to express my sincere gratitude to Alireza Ahmadi and Behzad Ghodrati, my co-
supervisors, for their invaluable guidance, suggestions and support.
My sincerest gratitude is extended to Professor Uday Kumar, Head of the Division of
Operation, Maintenance and Acoustics at LTU, for providing me with the opportunity to
pursue my research at the Division of Operation and Maintenance Engineering.
I would like to express my sincere gratitude to the Iraqi Ministry of Higher Education and
Scientific Research for providing a scholarship which made it possible for me to pursue
doctoral research at LTU.
I am grateful to all the group members in the drilling rig project for their valuable time in
meetings, sharing their experiences and suggestions for design improvement. Specific gratitude
is extended to Arne Vesterberg, Tommy Öhman, Curt Lindblad, Mats Johansson, Mikael
Anderson, Lars Karlsson, Andreas Nordbrandt, Ulf Sundberg and Jesus Retuerto for sharing
their expertise.
I would like sincerely and gratefully to acknowledge my colleagues (present and past) at the
Division of Operation, Maintenance and Acoustics for providing a friendly and open-minded
working environment. Special thanks are due to Malin Shooks, Rajiv Dandotiya, Mustafa
Aljumaili, Christer Stenström, Stephen Famurewa, Yasser Ahmed and Yamur Aldouri for their
discussions. Gratitude is also extended to my friend Hassan Ali. The administrative support
from Cecilia Glover and Marie Jakobsson is also gratefully acknowledged.
I wish to express my sincere gratitude to my dearest parents, Saed and Khalida, who have
always offered their full support throughout my life and who taught me the meaning of life.
Gratitude is extended to my siblings too. I am really thankful for all the support given to me.
Finally, I would like to express my deepest gratitude to my loving wife, Amani, and our
beloved sons, Abdulmalek and Yaman, for their enormous understanding and endless support
during my studies and research. It would not have been possible to complete this journey
without you by my side.

December, 2014
Luleå, Sweden

VI

ABSTRACT
In the context of mining, drilling is the process of making holes in the face and walls of
underground mine rooms, to prepare those rooms for the subsequent operation, which is the
charging process. Due to the nature of the task, drilling incurs a high number of drilling rig
failures. Through a combination of a harsh environment (characterised by dust, high humidity,
etc.), the operating context, and reliability and maintainability issues, drilling rigs are identified
as a major contributor to unplanned downtime.
The purpose of the research performed for this thesis has been to develop methods that can be
used to identify the problems affecting drilling rig downtime and to identify the economic
lifetime of drilling rigs. New models have been developed for calculating the optimum
replacement time of drilling rigs. These models can also be used for other machines which
have repairable or replaceable components. Based on an analysis performed in a case study, a
life cycle cost (LCC) optimization model has been developed, taking the most important
factors affecting the economic replacement time of drilling rigs into consideration. To this end,
research literature studies and case studies have been performed, interviews have been held,
observations have been made and data have been collected. For the data analysis, theories and
methodologies within reliability, maintainability, ergonomics and optimization have been
combined with the best practices from the related industries.
Firstly, this thesis analyses the downtime of the studied drilling rigs, with the precision and
uncertainty of the estimation at a given confidence level, along with the factors influencing the
failures. Secondly, the thesis identifies components that significantly contribute to the
downtime and the reason for that downtime (maintainability and/or reliability problems).
Based on the failure analysis, some minor suggestions have been made as to how to improve
the critical components of the drilling rig. Thirdly, a new method is proposed that can help
decision makers to identify the replacement time of reparable equipment from an economic
point of view. Finally, the thesis proposes a method using the artificial neural network (ANN)
for predicting the economic lifetime of drilling rigs through a series of basic weights and
response functions. This ANN-based method can be made available to engineers without the
use of complicated software.
Most of the results are related to specific industrial and scientific challenges, such as planning
for cost-effectiveness. The results of the case study are promising for the possibility of making a
significant reduction in the LCC by optimizing the lifetime. The results have been verified
through interaction with experienced practitioners from both the manufacturer and the mining
company using the drilling rig in question.
Keywords: Cost-optimization; Decision making; Drilling rig; Economic model; Life cycle
cost analysis; Mining industry; Optimal replacement time; Reliability analysis; Replacement
decision; Underground mining rig
VII

VIII

SAMMANFATTNING
Borrning är den process inom gruvdrift som åstadkommer hål i berg för att förbereda för
laddning och sprängning. Denna process innebär hård belastning på de borriggar som utför
borrningarna, vilket resulterar i oplanerade och kostsamma driftstopp. Dessa driftstopp beror
bland annat på damm och hög luftfuktighet, handhavande, samt borriggarnas inbyggda
funktionssäkerhet och underhållsmässighet.
Huvudsyftet med detta forskningsprojekt är att utveckla metoder att beräkna ekonomisk
livslängd för borriggar i gruvmiljö. Dessa metoder skall även lämpa sig för beräkning av
ekonomisk livslängd av andra typer av maskiner som består av utbytbara eller reparabla
komponenter. Därvid har en livscykelkostnadsmodell tagits fram baserat på en fallstudie, där de
viktigaste maskinpåverkande faktorerna tagits i beaktande. För att kunna utveckla dessa
metoder och denna modell har litteraturstudier, fallstudier, intervjuer, observationer,
datainsamlingar och modellering genomförts. Den så utvecklade modellen baseras på rådande
teorier och metoder inom driftsäkerhet, underhållsmässighet, ergonomi och optimering för
industriella applikationer.
De komponenter som väsentligt bidrar till driftstopp med avseende på funktionssäkerhet och
underhållsmässighet har kartlagts. Baserat på analys av tillgänglighet, ges också några förslag på
hur dessa kritiska komponenter kan förbättras. Slutligen föreslås en metod för att förutsäga den
ekonomiska livslängden hos borriggar, genom användning av så kallad viktning, och utan krav
på komplicerad programvara.
Resultaten är i första hand relaterade till industriella applikationer och förväntas hjälpa
beslutsfattare att ta kostnadseffektiva beslut. Resultaten från fallstudien gör det också möjligt att
uppnå betydande besparingar genom optimering av livslängden på borriggar. Resultaten har
verifierats genom kontinuerlig interaktion med både tillverkare av borriggar och gruvbolag. Ur
vetenskaplig synvinkel har också kunskapen ökats beträffande kritiska komponenters
tillförlitlighet.
Nyckelord: Optimering av kostnad; beslutsfattande; borrigg; borrmaskin; ekonomisk modell;
livscykelkostnadsanalys; gruvindustri; livslängd; tillförlitlighet; tillgänglighet; funktionssäkerhet;
underhållsmässighet; underjordsbrytning
IX

X

LIST OF APPENDED PAPERS
Paper I
Al-Chalabi, H., Lundberg, J., Wijaya, A. and Ghodrati, B. (2014), ‘‘Downtime analysis of
drilling machines and suggestions for improvements’’, published in the Journal of Quality in
Maintenance Engineering, 20(4), 306-332.
http://dx.doi.org/10.1108/JQME-11-2012-0038
Paper II
Al-Chalabi, H., Lundberg, J., Jonsson, A. and Ahmadi, A. (2014), ‘‘Case Study: Model for
economic lifetime of drilling machines in the Swedish mining industry’’, published in the
Engineering Economist.
http://dx.doi.org/10.1080/0013791X.2014.952466
Paper III
Al-Chalabi, H., Ahmadzadeh, F., Lundberg, J. and Ghodrati, B. (2014), ‘‘Economic lifetime
prediction of a mining drilling machine using artificial neural network’’, published in the
International Journal of Mining, Reclamation and Environment, 28(5), 311-322.
http://dx.doi.org/10.1080/17480930.2014.942519
Paper IV
Al-Chalabi, H., Lundberg, J., Al-Gburi, M., Ahmadi, A. and Ghodrati, B. (2014), ‘‘Model for
economic replacement time of mining production rigs including redundant rig costs’’,
submitted for publication in the Journal of Quality in Maintenance Engineering.
XI

XII

DISTRIBUTION OF WORK
The content of this section has been shared and accepted by all the authors who have
contributed to the papers. The contributions of each named author to the scientific papers
included in this thesis can be divided into the following main activities:
1. formulating the fundamental ideas of the study (initial idea and model development);
2. performing the study (data collection and analysis);
3. writing the paper and analysing the results;
4. revision of important intellectual content;
5. final approval for inclusion in the PhD thesis.
Table 1. Contributions of the main authors and co-authors of the appended papers
Paper IV
Paper III
Paper II
Paper I
Hussan Al-Chalabi 1, 2, 3 1, 2, 3 1, 2, 3 1, 2, 3
Jan Lundberg 1, 4, 5 1, 4, 5 1, 4, 5 1, 4, 5
Alireza Ahmadi 4 4 4
Behzad Ghodrati 4 4 4 4
Andi Wijaya 4
Adam Jonsson 1, 4
Majid Al-Gburi 2 2
Farzaneh Ahmadzadeh 3
XIII

XIV

ABBREVIATIONS
ANN Artificial neural network

BPP Branching Poisson process
CMMS Computerised maintenance management system
DFR Design for reliability
DMC Decreasing maintenance cost (%)
DOC Decreasing operating cost (%)
DRC Decreasing redundant rig cost (%)
ERT Economic replacement time (months)
GSI Geological strength index
GUI Graphical user interface
HME Heavy mobile equipment
IAC Increasing acquisition cost (%)
IFPP Increasing factor of the purchase price (%)
iid Independent and identically distributed
IPP Increasing purchase price (%)
K-S Kolmogorov-Smirnov test
LCC Life cycle cost
MC Maintenance cost (cu)
NHPP Non-homogenous Poisson process
OC Operating cost (cu)
ORT Optimal replacement time (months)
PP Purchase price (cu)
RFMC Reduction factor of the maintenance cost (%)
RFOC Reduction factor of the operating cost (%)
TBF Time between failures (h)
TOC Total ownership cost (cu)
TOCvalue Total ownership cost value (cu)
TTF Time to failures (h)
TTR Time to repairs (h)
XV

XVI

NOTATION
α Significant level
cu Currency unit
µx Mean time to repair (h)
µy Mean time to failure (h)
DTLL Lower limit of downtime (h)
DTM Mean downtime (h)
DTUL Upper limit of downtime (h)
TTRLL Time to repair lower limit (h)
MTTR Mean time to repair (h)
TTRUL Upper limit of time to repair (h)
mLL Lower limit of number of failures
mM Mean number of failures
mUL Upper limit of number of failures
cu/h Currency unit per hour
XVII

XVIII

TABLE OF CONTENTS

ACKNOWLEDGMENTS .................................................................................................... V
ABSTRACT ...................................................................................................................... VII
SAMMANFATTNING ...................................................................................................... IX
LIST OF APPENDED PAPERS ......................................................................................... XI
DISTRIBUTION OF WORK ......................................................................................... XIII
ABBREVIATIONS ........................................................................................................... XV
NOTATION .................................................................................................................. XVII
1 INTRODUCTION........................................................................................................1
1.1 Background ..............................................................................................................1
1.1.1 Mining drilling rig .............................................................................................2
1.1.2 Reliability literature review................................................................................3
1.1.3 Optimal replacement time literature review .......................................................4
1.2 Basic concepts and definitions ...................................................................................5
1.2.1 System reliability and performance .....................................................................5
1.2.2 System availability ..............................................................................................5
1.2.3 System maintainability .......................................................................................6
1.2.4 System life cycle costing .....................................................................................6
1.2.5 Life cycle costing fundamentals ..........................................................................7
1.3 Industrial motivation of the study ..............................................................................8
2 THE APPROACH OF THE THESIS .......................................................................... 11
2.1 Research problem description ................................................................................. 11
2.2 Overall research goal ............................................................................................... 11
2.3 Research questions.................................................................................................. 12
2.4 Scope and limitations of the study ........................................................................... 12
3 RESEARCH METHODS ............................................................................................ 15
3.1 Data collection and analysis ..................................................................................... 15
3.1.1 Research objects .............................................................................................. 18
3.1.2 Failure and cost data......................................................................................... 18
3.2 Method used in Paper I........................................................................................... 21
XIX

3.3 Method used in Papers II-IV................................................................................... 21
4 SUMMARY OF APPENDED PAPERS ...................................................................... 23
4.1 Paper I .................................................................................................................... 23
4.2 Paper II .................................................................................................................. 23
4.3 Paper III ................................................................................................................. 24
4.4 Paper IV ................................................................................................................. 24
5 RESULTS AND DISCUSSIONS ................................................................................. 25
5.1 Results and discussion related to research question 1 ............................................... 25
6 CONCLUSIONS ......................................................................................................... 41
7 SCIENTIFIC & INDUSTRIAL CONTRIBUTIONS ................................................. 43
8 SCOPE OF FURTHER RESEARCH ......................................................................... 45
REFERENCES .................................................................................................................... 47

XX

1 INTRODUCTION
1.1 Background
Mining is ranked as the second basic industry of early civilization after agriculture. Since
prehistoric times, mining has been an essential part of human existence, i.e. mining in the
broadest sense of the term, meaning the extraction of any naturally occurring mineral
substances from the earth or other heavenly bodies for utilitarian purposes (Hartman, 1987).
Humans began mining approximately 450,000 years ago (Hartman and Mutmansky, 2002).
Nowadays, mining is the foundation of the world’s industrial development.
Minerals have been mined in Sweden for over 1,000 years. Metals from these minerals and
other substances are important for society and are used in daily life. The mining industries
contribute nine percent of the Swedish gross domestic product and employ 0.5 percent of the
total industrial labour force (Lithander, 2004). In 2010, mining accounted for 13 percent of all
industrial investments and contributed SEK 26 billion to Sweden's gross domestic product. In
2012, the Swedish Trade Council stated that the Swedish mining industry had annual sales of
approximately SEK 70 billion and more than 30,000 employees, making the mining industry
an important engine for growth in Sweden (Swedish Trade Council, 2012). The year 2013
was a record year for the Swedish ore production, as it reached almost 80 million tons that
year, an increase of ten percent compared with 2012. Today, there are 17 mines in production
in Sweden, 15 metal mines and two clay mines.
In modern mining, there are five stages in a mine’s life in the utilization of a mineral deposit.
The first stage is the search for ores or other valuable minerals, which is called the prospecting
stage. The second stage in the life of a mine is determining as precisely as possible the value
and volume of the mineral deposit found. This stage is called the exploration stage.
Development is the third stage in the mine’s life and it involves the work of opening a mineral
deposit for exploitation. Exploitation, the fourth stage of mining, is associated with the actual
recovery of minerals from the earth. The final stage in the operation of most mines is
reclamation, the process of closing the mine (Hartman, 1987).
Mining can be classified into two main types based on the excavation technique. Type one is
surface mining, which is performed on the surface by removing layers of bedrock to reach
stone, coal, or ore deposits. Type two is underground mining, which concerns the exploitation
of geological materials or other minerals by extracting them beneath the surface of the earth.
Underground mining consists of making incisions into the earth in the form of underground
tunnels to reach ore deposits, and can be classified into three main types based on the access
type: slope mining, shaft mining and drift mining. Slope mining is a form of underground
mining used where the mineral deposits are located very deep. This type of mining is normally
carried out when there are problems drilling shafts vertically downward. Shaft mining is a type
of underground mining where shafts are driven vertically into the earth to access ore deposits
(Wijaya, 2012). Drift mining involves the use of a drift or horizontal access tunnel driven into
the earth for extracting or transporting ore or minerals; drilling rigs are used for face drilling in
this type of underground mining. This thesis focuses on drilling rigs used in underground
mines. A typical drift mining process cycle consists of six processes, namely drilling, charging,
blasting, scaling, loading and rock bolting. These processes form a cycle illustrated in Figure
1.1.
1

Figure 1.1. A typical drift mining process cycle
The process cycle starts with drilling, the process of making blasting holes in the mining room
face by crushing the rock. The next process, charging, is the process of filling the blasting holes
with an explosive, for example dynamite. The subsequent process is blasting, the action of
breaking rock using the charged explosives. After this the scaling process will start, in which
loose material is cleared from the roof, face and walls to make the mining room completely
safe. This is followed by the loading process, in which the broken rock is gathered and loaded
onto trucks for removal to a central loading area. After the loading process is finished, the
mining room has been cleared from broken rocks and prepared for the final process, which is
bolting. Bolting involves the insertion of steel rods into holes drilled into the mining room’s
roof or walls to provide support for the roof or sides of the mining room.
1.1.1 Mining drilling rig

A mining drilling rig is a machine used to create holes in the ground. Mobile drilling rigs can
be used to make tunnels and underground facilities, and small or medium-sized mobile drilling
rigs are utilized in mineral exploration, for example. Mining drilling rigs are used for two main
purposes, namely production drilling (for processes in the mining production cycle such as
bolting) and exploration drilling (for identification of the location of minerals).
As the description of the process cycle for drift mining shows, drilling is an extremely
important step in the workplace. From an economic viewpoint, drilling rigs make an
important contribution to the mine’s production rate, have a high acquisition, maintenance
and operating cost, and represent a possible critical bottleneck for production. The L2C
drilling rig is used as the object of a case study in this thesis and its components are presented in
Figure1.2. This rig is manufactured by Atlas Copco and is used by Boliden AB in Sweden.
In 1898 Atlas manufactured its first drill, driven by compressed air, and then in 1915 the
company produced drills equipped with four-cylinder single-acting piston motors. These drills
had a design that offered many advantages over those of Atlas’ competitors, including fewer
parts and good balance. The first mobile rig arrangement for underground drilling was
designed for sub-level caving in Kiruna in 1952. In 1973 Atlas Copco presented the first
heavy-duty impact hydraulic rock drill. The hydraulic rock drilling technique made it possible
to increase the drilling output. In 1998 the company launched a new generation of
underground drill rigs called boomers. Based on a modular design, this rig concept set new
standards for automation, computerization and performance.
2

1 Cabin 6 Front jacks 11 Cable reeling unit

2 Boom 7 Hydraulic pump 12 Diesel engine
3 Rock drill 8 Rear Jack 13 Hydraulic oil reservoir
4 Feeder 9 Electric cabinet 14 Operator panel
5 Service platform 10 Hose reeling unit
Figure 1.2. Mining drilling rig (source: Atlas Copco Rock Drills AB)
Economic competition has pressurized mining companies into achieving higher production
rates by enhancing the techniques of drilling and blasting and increasing mechanisation and
automation. Boliden’s historical data over a period of one year (2008) from one of their
underground mines in Sweden show that more than 15% of the unplanned downtime of
mobile rigs is related to drilling rigs, the greater part of whose downtime is attributed to the
poor reliability of their components. Thus, the drilling rig is a bottleneck in the production
cycle and is becoming an important object of research in underground mining. Due to a
combination of a harsh environment, the operation context, and reliability and maintainability
issues, the drilling rig is a major contributor to unplanned downtime.
In the following subsections, an account is given of literature reviews performed to identify the
current status of research on reliability and optimal replacement times, as documented in the
available literature.
1.1.2 Reliability literature review

A mine production system consists of many subsystems. To make a mining system profitable,
the optimization of each subsystem in relation to other subsystems should be considered
(Barabady and Kumar, 2008). To achieve this aim, reliability and maintainability analysis
should be performed for each subsystem in the mine production system. Since the mid-1980s,
reliability analysis techniques have been essential tools in automatic mining systems (Blischke
and Murthy, 2003).
Many researchers have studied the reliability and maintainability of mining equipment and its
failure behaviour. For example, Kumar et al. (1989) analysed the operational reliability of a
fleet of diesel-operated load-haul-dump (LHD) machines in Kiruna Mine in Sweden. They
divided the LHD machines into four main subsystems. They did separate failure analyses and
the reliability of the fleet was modeled. Kumar et al. (1992) performed reliability analysis on
the power transmission cables of electric mine loaders in Sweden. They used a proportional
hazards model to investigate the effects of two different designs on the reliability of a power
transmission cable of an electric mine loader. They have found that the proportional hazards
model can be used as an explanatory tool to seek explanations for undiscovered facts and on
that basis decisions for selecting the suitable design for a component can be taken. Kumar and
Klefsjö (1992) analysed the maintenance data of one subsystem (the hydraulic system) of a fleet
of six LHD machines divided into three independent groups at Kiruna Mine.
3

Reliability assessment of mining equipment was performed by Vagenas and Nuziale (2001);
using genetic algorithms, they developed and tested reliability assessment models for mobile
mining equipment. They found that the application of genetic algorithms in reliability
engineering may contribute to a better understanding of the reliability characteristics of
industrial systems. Vayenas and Xiangxi (2009) used reliability-based maintenance to study the
availability of 13 LHD machines in an underground mine. They were interested in the
influence of machine downtime on the productivity and operation costs of these machines.
They found that mechanical breakdowns and planned maintenance consume most of the repair
time of the fleet. Wijaya et al. (2011) developed a method for visualising downtime by using a
jack-knife diagram; they applied the method on a scaling machine at a mine in Sweden.
Gustafson et al. (2012) analysed the reliability of automatic and manual LHD machines and
performed a comparison between them. They found that the hydraulic and electric systems are
the biggest reasons for the downtime of both manual and automatic LHD machines.
In addition, many studies have considered reliability, maintainability and optimum replacement
decisions; readers are referred to Ahmadi and Kumar (2011) and Dandotiya and Lundberg
(2012), for example, for further studies on the recent literature in this field. The first study to
have been made regarding the downtime analysis of mining drilling rigs is presented in this
thesis.
1.1.3 Optimal replacement time literature review

Standard models for economic replacement time (ERT) decisions contain an estimation of the
discounted costs by minimising the total cost of the equipment. The assumption of these
models is that equipment will be replaced at the end of its economic lifetime by a continuous
sequence of identical equipment (Hartman and Tan, 2014). Bellman (1955) developed the first
optimal asset replacement model for a variable lifetime of assets. Wagner (1975) offered
dynamic programming formulation for the equipment replacement problem in which the state
of the system is the time period and the decision at each period is to keep the equipment for N
periods. His formulation has been extended by researchers to deal with the realities of
technological changes, for example see Oakford et al. (1984), Bean et al. (1985), Hartman and
Rogers (2006), and Hritonenko and Yatsenko (2008). These authors assumed a finite horizon
in their approaches to the problem of equipment replacement under non-stationary costs.
In 1976, Elton and Gruber (1976) showed that an equal life policy was optimal on an infinite
horizon under technological changes. In contrast, Hartman and Murphy (2006) studied an asset
replacement problem for a stationary finite horizon; they illustrated how a bound on the
number of times an asset is retained in its economic life can be obtained, thus suggesting that it
is optimal to replace the asset during its economic lifetime.
Dynamic programming models have been utilized in real cases of calculating equipment
replacement time because of the important uncertainties associated with life cycle costs
(Richardson et al., 2013). The net present value of all the life cycle costs associated with an
infinite sequence of equipment life cycles has also been used to make equipment replacement
decisions (Bethuyne, 1998; Scarf and Bouamra, 1999; Hartman, 2005; Yatsenko and
Hritonenko, 2005). Other researchers have used different equipment replacement models to
analyse a variety of equipment, such as forklifts, buses, and aircraft (Eilon et al., 1966; Keles
and Hartman, 2004; Bazargan and Hartman, 2012). Although Tanchoco and Leung (1987)
found that replacement decisions could be influenced by capacity considerations, others have
noted that technological changes can encourage decision makers to utilize equipment beyond
its economic lifetime (Cheevaprawatdomrong and Smith, 2003). The first study to have been
made regarding the optimal replacement time of mining drilling rigs is presented in this thesis.
4

1.2 Basic concepts and definitions
1.2.1 System reliability and performance
In this section we present the basic definitions of reliability and discuss the relationship
between reliability and performance. System or product reliability is defined as the ability of a
product or system to perform as intended (i.e. without failure and within specified
performance limits) for a specified period of time and in its life cycle conditions (Kapur, 2014).
Reliability is an important attribute of a system or product. In the case of the drilling rig, the
user expects the rig to operate properly for a specified period of time beyond its purchase date,
which usually depends on the purpose and cost of the rig. At a low cost, a throwaway rig may
be used as a redundant rig just for a limited period of operation. A reliable rig may be expected
to last for many years of operation when it is properly maintained. The behaviour or the future
performance of the system depends on its reliability. Thus, reliability has been considered as a
term that concerns quality (Kapur, 1986; O'Connor, 2000). In the past few decades, system
reliability has been defined in many ways, including the probability that a system, product, or
device will not fail for a given period of time under specified operating conditions (Shishko
and Aster, 1995). System reliability can be defined as the capability of a product to meet
customer expectations of product performance over period of time (Stracener, 1997).
The performance of equipment is typically associated with the functionality of that equipment,
what the equipment can do and how well it can do it. For example, the functionality of an
underground mining drilling rig involves drilling holes in the face and walls of mining rooms.
How it drills the holes and the quality of the holes involve functional performance parameters
such as rotational speed, feeding speed, ease of drilling hard rocks, and steering adjustments.
Improving the performance of equipment usually requires adding technology and complexity.
This can make the desired reliability difficult to achieve.
To summarize the relationship between reliability and performance, it can be stated that the
reliability of equipment refers to its ability to perform its function without failure for a certain
period of time in its life cycle application under specified working conditions and within
certain limits of performance. The performance parameters usually describe the functional
capabilities of the equipment. Finally, it can be stated that the performance of equipment is
related to its reliability.
1.2.2 System availability

Availability is defined as a percentage measure of the degree to which machinery and
equipment are in an operable and committable state at the point in time when they are
needed. This definition includes committability and operability factors which are contributed
to by the equipment itself, the process being performed, and the surrounding facilities and
operations (Katukoori, 1995). There are different classifications of availability and different
ways to calculate it. A common classification is based on the span of time to which the
availability refers and the types of downtime used in the computation. There are a number of
different classifications of availability, such as the following:
 point availability, defined as the probability that the system is in an operable state at a
specified time;
 interval availability, defined as the expected fraction of an interval of a specified length
of time during which the system or equipment is in an operable state;
 steady state availability, defined as the expected fractional amount of time in a
continuum of operating time during which the system or equipment is in an operable
state.
5

For further discussion about the classifications and ways to calculate the system availability, see
Katukoori (1995), Kumar and Akersten (2008), and Stapelberg (2009).
1.2.3 System maintainability

The maintainability of an item is defined as its ability, under stated conditions of use, to be
retained in, or restored to, a state in which it can perform its required function when
maintenance is performed under stated conditions and using prescribed procedures and
resources (Rausand, 2004). System maintainability is a main factor which can be used to
determine the availability of the system.
The maintainability of a system depends on design factors such as ease of reinstallation, ease of
dismantling, ease of access to the system, and so on. Maintainability is primarily determined by
the design of the system or the item and can be greatly enhanced if the repair procedure and
fault detection and isolation are worked out during the design stage of the system itself. The
objective of maintainability is to restore the function of the system in a minimum period of
time.
1.2.4 System life cycle costing

The life cycle cost of a system can be defined as the sum of all the acquisition and ownership
costs incurred during the useful life span of the system. These costs include direct, indirect,
recurring, nonrecurring, and other related costs incurred, or estimated to be incurred, in
design, research and development, investment, operations, maintenance and other support of
the system, and retirement of the system. The ownership costs represent the total of all costs
other than the acquisition or initial cost during the life span of a system. The term life cycle
costing was used for the first time in 1965 in a report entitled "Life Cycle Costing in
Equipment Procurement" (Dhillon, 2010). This report was prepared by the Logistics
Management Institute, Washington, D.C., for the Assistant Secretary of Defense for
Installations and Logistics, the U.S. Department of Defense, Washington, D.C. According to
many studies, the system ownership cost (i.e. operating and logistic cost) can vary considerably
and amount to 10-100 times the original acquisition cost (Ryan, 1968). Determining the
system’s life cycle cost is an important issue, because the acquisition is a small part in relation to
the total costs associated with owning and operating the system.
The life cycle cost of a system can be determined through one of three general methods (Farr,
2011). These methods are the engineering build-up, the analogy and the parametric method.
The engineering build-up method involves direct estimation at the component level leading to
a detailed engineering build-up cost estimate of the system. This can be achieved by using
machine element and mechanics theories for estimating the lifetime of equipment and practical
lifetime tests, see, for instance, Norton (2011). The engineering build-up methodology can be
used for individual estimates for each item, element, or component, and these estimates are
then combined into the overall cost estimate. Therefore, this methodology is sometimes
classified as a "bottom-up" estimating method. It involves computation of the cost of each
element by estimating costs at the lowest level of detail and computing quantities and levels of
effort to determine the total system cost.
In the analogy method an estimate is made using historical results from a similar system or
product. Analogy estimates are performed on the basis of comparison and extrapolation using
similar systems or items. In many instances this can be accomplished using simple relationships
or equations representative of detailed engineering build-up estimates of past projects. The
preferred means to conduct a cost estimate early in the system life cycle is to use data from
programmes that are technically representative of the programme to be estimated. The cost
6

data are then subjectively adjusted upward or downward, depending on whether the subject
system is felt to be more or less complex than the analogous programme (Farr, 2011).
The parametric method is based on mathematical models or equations. Simple mathematical
relationships such as nonlinear and linear regression are mostly utilized. Excel software, for
example, can be easily utilized to fit these relationships (Farr, 2011). In this thesis, the
parametric method is used to calculate the total ownership cost of the drilling rig and to
estimate its economic lifetime.
1.2.5 Life cycle costing fundamentals

Life cycle costing can be defined as a method for estimating the total life cycle cost of an
acquired item or acquired equipment. The total life cycle cost includes equipment
procurement and ownership costs. The equipment ownership cost could be quite significant
and in most cases exceeds the procurement cost. For example, various studies indicate that the
operating and maintenance costs over the life span of equipment could be many times greater
than its procurement cost (Ryan, 1968). There are many reasons why the industrial sector all
around the world is being compelled increasingly to use life cycle costing to make different
types of decisions that indirectly or directly concern engineering systems. Examples of these
reasons are increasing operation and maintenance costs, budget limitations, competition, and
the high cost of investment (Seldon, 1979). There are six primary areas where life cycle costing
can be used: comparing competing projects, controlling on-going projects, long-term planning
and budgeting, selecting a successful candidate among bidders competing for a project,
comparing logistics concepts, and deciding the optimal replacement of aging equipment
(Seldon, 1979). The focus of this thesis is directed on the economic replacement time of
mining drilling rigs. To perform life cycle costing studies, different types of information are
required, such as the costs connected with the acquisition, operation, maintenance, installation,
and transportation (delivery) of the equipment, the taxes to be paid, the disposal cost, the
second-hand value, and the useful operational life of the equipment or systems ( Brown, 1979).
There are many activities associated with life cycle costing, including the following (Earles,
1981):
o conducting appropriate sensitivity analysis;
o defining activities that generate a product’s or an item's ownership costs;
o establishing discounted life cycle costs;
o identifying all the cost drivers;
o defining a product's or an item's life cycle.
Over the years, many authors have proposed steps for performing life cycle cost analysis
(Wynholds and Skratt, 1977; Coe, 1981; Dhillon, 2013). The following ten steps have been
considered quite effective in performing life cycle cost analysis (Greene and Shaw, 1990):
1. determine the life cycle cost analysis objective,
2. define the scope of the system,
3. choose the effective models for estimating the life cycle cost,
4. obtain all the essential data and make the appropriate inputs into the selected model,
5. conduct sanity checks for outputs and inputs,
6. conduct essential sensitivity analysis,
7. formulate the life cycle cost analysis results,
8. document the life cycle cost analysis,
9. present the life cycle cost analysis as is appropriate,
10. update the life cycle cost analysis as is appropriate.
7

In order to perform effective life cycle cost analysis, availability of reliable cost data is necessary.
This means that the existence of good cost data banks is very important. Thus, in developing a
new cost data bank, careful attention must be given to factors regarding the data such as
uniformity, volume, responsiveness, flexibility and comprehensiveness (Dhillon, 2010).
Although data for life cycle cost analysis can be obtained from many sources, their quality and
amount may vary quite considerably. Therefore, prior to starting a life cycle cost study, it is
vital to examine carefully factors such as data applicability, data availability and data bias
(Dhillon, 1999).
In 2010 Dhillon discussed the advantages and disadvantages of life cycle costing, stating that the
important benefits of life cycle costing are that they are useful for the following purposes:
o reducing the total cost,
o controlling programmes,
o comparing the cost of competing projects,
o making decisions associated with equipment replacement, planning, and budgeting.
In contrast, some of the main disadvantages of life cycle costing include the following:
o it is costly;
o it is time-consuming;
o the acquisition of data for analysis is a trying task,
o it has doubtful data accuracy.
For more details, see Dhillon (2010). The specific purposes of life cycle costing in product
acquisition management are to estimate the total ownership cost (TOC) of the product and to
decide when to buy a new one. Reducing the TOC by using LCC analysis is an important
issue in the development process of the product, and understanding TOC implications is
necessary to decide whether to continue to the next development phase, as well as to control
costs and assist the procurement decisions (Farr, 2011).
Life cycle cost analysis is an economic estimation technique that determines the total cost of
owning, operating, maintaining and disposing of a system over its life span. It must be used to
understand fully how to determine and interpret the total ownership cost of a system (Farr,
2011). A simple process for developing an LCC model is shown in Figure 1.3.
1.3 Industrial motivation of the study

Industrial companies, or more specifically mining companies, put huge funds, often millions of
dollars, into their annual budgets to purchase heavy mobile equipment (HME) such as drilling
rigs, scaling rigs, wheel dozers, wheel loaders, dump trucks, etc. Given the enormous costs of
acquiring, operating, and maintaining their HME, it is important for mining companies to
optimise their replacement and procurement strategies (Richardson et al., 2013). With
increased mine production, the HME operating hours will increase and, as a result, the
operating and maintenance cost will increase also. At some point in the equipment’s life span,
these costs will be too high and, since it will then no longer be economically viable to
continue using the old equipment, it should be replaced (Verheyen, 1979). An essential
economic consideration in industrial companies is that a model must be found for identifying
this point (i.e. the point at which the equipment replacement time is expected to yield a
minimal life cycle cost). Obviously, for mining companies, one of the most important decisions
concerns determining the ERT of capital equipment; this can be accomplished with the help
of life cycle cost analysis. The main reason for the increasing use of the life cycle costing
concept for HME is that at some point the operating and maintenance costs will exceed the
acquisition costs.
8

Understand the
manufacturer Define the
Collect data
and operator scope
requirements
Evaluate and
normalize the
Conduct
Sensitivity data
LCC
analysis analysis
Select the
variables
Regression Data analysis

and curve fit and correlation
Figure 1.3. Process for developing an LCC model (adapted from Farr, 2011)
Since the 1960s most of the research conducted in the field of mining equipment has
concerned the reliability and maintenance of such equipment. Nowadays, modern mining
machines have a higher technology level and a higher complexity in comparison with the old
ones (Hoseinie, 2012). Reliability analysis techniques have been gradually accepted as standard
tools for the planning and operation of complex and automatic mining equipment (Barabady
and Kumar, 2008). Generally, the downtime costs of mining equipment are too high due to
high production losses (Dandotiya, 2012). In order to enhance equipment reliability and to
reduce mining equipment downtime, a large amount of research has been conducted using
various methods to ensure the improvement of equipment design and maintenance procedures.
Reliability analysis can be used in identifying the sensitive subsystems and critical components
of mining machinery. Moreover, reliability analysis has been acknowledged as the basic
requirement for all maintenance planning, prediction of the remaining useful life, and life cycle
cost analysis of machinery. Therefore, we need to analyse the reliability of operating machinery
at the first stage of any comprehensive field studies.
A preliminary study performed by the author of this thesis in one underground mine in
Sweden revealed that more than 15 percent of the unplanned downtime of mobile equipment
is related to drilling rigs. This high amount of unplanned downtime leads to monetary losses,
which need to be minimized. Since drilling is the first process of a typical mining cycle and is a
key factor for production, it is also important to find solutions for drilling rig problems and
reduce the downtime. To minimize losses related to downtime, the reliability of the drilling rig
should be improved and the economic replacement time of the rig should be determined. The
replacement of rigs should be carried out in a cost-effective way.
The present study has focused on the development of a model that can be used by the
manufacturer and the operator companies for easy estimation of the optimal replacement time
of drilling rigs from an economic point of view. This model can in principle be used for other
machinery used in the mining industry or other industries. Furthermore, the study has also
aimed to calculate the downtime of drilling rigs to assist the manufacturer and operator
companies in identifying the problems affecting rig downtime and to provide solutions that
will reduce it. This collaborative research has been jointly conducted by Luleå University of
Technology, a mining company operating the drilling rig and the manufacturing company
which developed the rig.
9

10

2 THE APPROACH OF THE THESIS
2.1 Research problem description

Meeting an increasing demand for higher production, reducing the production equipment’s
downtime and increasing its productivity are the major concerns of the collaborating mining
company. Economic competition is pressurizing the mining company into achieving higher
production rates by enhancing its drilling and blasting techniques and increasing its
mechanisation and automation. This means that improving the availability of the company’s
production rigs should be investigated.
The production rigs are those rigs which contribute directly to the mine production rate and
create value for the company. One of these production rigs is the drilling rig. The maintenance
department at the mining company aims to reduce the number of failures and the downtime of
production rigs. The financial department at the company aims to reduce the costs associated
with the drilling rig’s acquisition, operation and maintenance by implementing a cost-cutting
strategy. There are many reasons why the mining company is being compelled increasingly to
use life cycle costing to make different types of decisions that indirectly or directly concern
production rigs. Examples of these reasons are increasing operation and maintenance costs,
budget limitations, competition, and the high cost of investment. The overall goal of the
mining company is to increase its mine production and the company profit.
To achieve the goal of the maintenance department, the engineers at this department asked the
following questions: which components of the drilling rig have mostly contributed to its
downtime and which problems have the greatest effect on the downtime of the rig? To
achieve the goal of the financial department, the financial experts at this department asked the
following question: what is the best time to replace the rig and buy a new one from an
economic point of view, with a view to reducing the total ownership cost of the rig? To
achieve the overall goal of the mining company, downtime analysis should be performed for
the drilling rig to identify the rig components that contribute most to its downtime, to identify
the problems affecting the downtime of those components, and, if possible, to make
suggestions for the solution of these problems. Another measure which would help the
company to achieve its overall goal would be to perform life cycle cost analysis for the drilling
rig in order to estimate the optimal replacement time of this rig, with a view to reducing the
ownership cost from an economic point of view and increasing the company profitability.
The main reason for an increased use of the life cycle costing concept for production
machinery is that at some point of its life span the operating and maintenance costs will exceed
the acquisition costs. To this end one should consider focusing one’s attention on optimization
models, techniques and tools which can help the decision makers in the mining company to
optimize their equipment lifetime.
2.2 Overall research goal

The overall research goal has been to develop methods for downtime analysis and replacement
time optimization. These methods should be useful for machinery in general and specifically
for drilling rigs in mining applications.
11

2.3 Research questions
To fulfil the overall research goal of the thesis, the following research questions (RQs) have
been formulated.
1. Which critical components make a large contribution to the downtime of the drilling
rig? What are the dominant problems influencing the drilling rig downtime, reliability
problems and/or maintainability problems?
2. How should a model be constructed that can calculate the optimum replacement time
of a drilling rig? Moreover, which cost factors have the largest impact on the
replacement time?
3. How should the above model be improved with respect to easy and accurate prediction
of the economic lifetime of this rig, with a high level of certainty and with a view to
enhancing decision making in this regard?
4. How should a model be constructed for estimation of the economic lifetime of mining
production rigs when also the costs from one redundant rig are considered?
These research questions are answered by the four appended papers, each of which makes its
own contributions toward the research questions (see Table 2.1).
Table 2.1. Relationship between the appended papers and research questions
Paper IV
Paper III
Paper II
Paper I
Research question 1 (RQ 1) X

Research question 2 (RQ 2) X X X
The research framework appears in Figure 2.1. Paper I analyses and compares the downtime of
four drilling rigs used in two underground mines in order to identify the critical components
of these rigs. Paper II develops a practical model to calculate the optimal replacement time of
the drilling rigs. Paper III develops a new model for easy and accurate prediction of the
economic replacement time of mining drilling rigs by using an artificial neural network
technique. Paper IV develops a model to determine the economic replacement time of
production rigs used in the mining industry when also the costs from one redundant rig are
considered.
2.4 Scope and limitations of the study

The scope of this thesis is limited to the drilling rig and, more specifically, to the study of the
replacement time of drilling rigs used in underground mines in Sweden, as well as the issues
related to the downtime of these rigs. The thesis is based on manually entered failure, repair
and cost data from the collaborating mines.
The limitations of this thesis can be described as follows. Firstly, the reliability of human beings
is an important area to consider, but is beyond the scope of this thesis. Secondly, the data are
related to specific mobile mining equipment, with specific environmental and operating
characteristics and working in specific underground mines. Thirdly, the results obtained from
this research are limited to the information that could be extracted from the collected data.
12

Fourthly, the results are limited to case studies of L2C drill rigs. Fifthly, the operating
conditions, for example the working environment, are not considered in the downtime
analysis. Sixthly, the influence of production performance of the drilling rig is not considered
in the model of the optimal replacement time. The model is only considering the costs.
Finally, in the present thesis, the replacement time analysis was performed based on the cost
data obtained from only one drilling rig operated in a single underground mine in Sweden.
Consequently, the operating conditions should be taken into consideration before applying the
study’s results, especially when the operating conditions are not similar to those in the present
study.
Paper I
Downtime analysis and
comparison between
different subsystems
Paper IV
Reliability and Life Cycle Paper II
Model for ERT Cost Modelling of Mining Optimal replacement time
considering different cost
factors
Drilling Rigs estimation
Paper III
Economic replacement
time prediction using
artificial neural network
Figure 2.1. Research framework
13

14

3 RESEARCH METHODS
3.1 Data collection and analysis

The data were collected from the database of the computerized maintenance management
system (CMMS) of the mining company participating in the present study. The literature
survey was carried out based on relevant peer-reviewed papers in conference proceedings and
journal articles from online databases. Some of the papers and articles were searched for on the
basis of references of other relevant articles, and some of the known papers and articles were
retrieved directly from the journal databases. Relevant books and reports were searched for in
Lucia (Luleå University Library’s online catalogue), and PhD and licentiate theses from various
universities were studied. Specific keywords were used to search for information in well-
known online databases, including Science Direct, Elsevier, Emerald, Springer and Google
Scholar. Moreover, discussions were held with reference groups.
The present research was performed on two underground mines in the north of Sweden.
Failure and repair data for two years (August 2009-August 2011) were collected to answer
research question one. These data were collected from the two mines for four drilling rigs.
Three rigs were working in the first mine and the fourth rig was working in the second mine.
Cost data for around four years (March 2009-January 2013) were collected to answer research
questions two, three and four. The cost data were collected for a single rig operating in the
second mine.
The analysis of failure data which can be used in reliability analysis should be based on the
assumption that the times between failures (TBFs) are independent and identically distributed
in the time domain. The magnitude of the recorded TBF data was valid for fitting the various
distributions for representing the population of the TBFs. However, before any reliability
analysis for Paper I was performed, trend and serial correlation tests had to be carried out to
check whether the usual assumption of independent and identically distributed TBF data in the
data sets would be contradicted or not (Kumar, 1989; Kumar and Klefsjö, 1992). The Laplace
trend test was used in this study to test the hypothesis that a trend did not exist within the TBF
data. We calculated the test statistic U for the TBFs of the drilling rig subsystems to be at a
significant level of 0.05. From the standard normal tables, with a significant level of 0.05, the
critical value is equal to 1.96. If -1.96<U<1.96, then we would accept the hypothesis of no
trend within the TBF data. After applying the trend test for the critical components of the
studied rigs, we found that there was no trend within the TBF data; for example, U was equal
to 0.55 in the feeder of the drilling rig used in the second mine.
The serial correlation test of the TBF data of the drilling rigs and their subsystems was carried
out to check the dependence of the TBF data. This test was performed by plotting the ith TBF
against the (i-1)th TBF. We tested the significance of the correlation by calculating the (r)
value and comparing it with the critical (r) value, which could be obtained from the
correlation tables. For example, the results of the serial correlation test of the above component
(i.e. the feeder) are given in Figure 3.1. Since the points are randomly scattered, the failure data
can be assumed to be independently distributed. The (r) value is equal to 0.05, while the
critical value of (r) from the correlation tables at the significance level alpha = 0.05 and a
degree of freedom of 30 for a two-tailed test is equal to 0.364. We can conclude that the
correlation between the TBFs of this subsystem is statistically not significant (i.e. no correlation
15

exists), since r < the critical r. Different types of statistical distributions were examined and
their parameters were estimated by using the Easy Fit and Minitab software.
Serial correlation test

400
300
ith TBF
200
100
0
0 50 100 150 200 250 300 350 400
(i-1)th TBF
Figure 3.1. Serial correlation test for the feeder of the drilling rig
As mentioned earlier, the cost data for the single machine used in one mine were collected for
a period of around four years. Figures 3.2 and 3.3 illustrate the maintenance and operating cost
for this period.
Maintenance cost Operating cost
150
55
125
Maintenance cost (cu)
Operating cost (cu)
100 45
75 35
50 25
25
15
0
70 80 90 100 110 120 70 80 90 100 110 120
Time (month) Time (month)
Figure 3.2. Maintenance cost Figure 3.3. Operating cost

With reference to Papers II-IV, since the user company planned to use the machine for 120
months, extrapolation was performed on the operating and maintenance cost data. Figures 3.4
and 3.5 illustrate the maintenance and operating costs determined by data extrapolation.
Maintenance cost Operating cost
400 150
300
Operating cost (cu)
100
200
100 50
0
0
-100
-200 -50
0 50 100 150 200 0 50 100 150 200
Time (month) Time (month)
Figure 3.4. MC after extrapolation Figure 3.5. OC after extrapolation

In Figures 3.4 and 3.5, the dots represent the real historical data for the maintenance and
operating costs. Curve fitting was performed using the Table Curve 2D software to show the
behaviour of these costs before and after the time when the data were collected. Note that the
fitting would have been better if more data had been available for a time period of more than
16

four years. This software uses the least squares method to find a robust (maximum likelihood)
optimisation for nonlinear fitting. It is worth mentioning that the drilling rig in this case study
had no multi-level preventive maintenance programme. The main reason why the
maintenance cost was quite low in the earlier months was that the rig was new at the start of its
utilization. The history shows that the user company began to keep track of cost data by using
CMMS when the maintenance costs started growing. This observation can be compared with
the three phases in the conventional bathtub curve (see Figure 3.6) used in reliability studies
(Blischke and Murthy, 2003).
Figure 3.6. Bathtub curve (adapted from Blischke and Murthy, 2003)
The first phase of this curve represents the period with a relatively large number of early
failures. This phase is not included in Figure 3.4 because the failures are considered to have
been identified and resolved before the delivery of the rig to the mining company. The rig is
assumed to have been delivered to the mining company while in the second phase. In this
phase the rig has random failures at a low and almost constant failure rate. This phase is visible
in the period between 0 and 76 months in Figure 3.4, where the cost is low and relatively
constant. The final phase represents a period with an increasing number of failures that are
related to the aging of the system. This corresponds to the period between 76 and 120 months
in Figure 3.4.
There was no information regarding the number of failures and the maintenance cost at the
beginning of the lifetime of this rig and up to the time when the company started to record
these data in its CMMS. Therefore, the ‘‘Lorentzian Cumulative’’ equation was used to
perform the extrapolation for the maintenance cost data.
The fitted curve for the maintenance and operating costs (in the second and third phases) using
the Lorentzian extrapolation is assumed to represent the behaviour of our data adequately,
2
since r (adj.) = 0.97 and 0.91 for the two costs, respectively. In Figures 3.4 and 3.5, the
inflection point after 120 months of operation is due to a change in the usage profile of the rig,
as the rig was to be used as a redundant rig. This was confirmed by the engineers at the
collaborating mine. The rig was to be used after 120 months of operation only when any of
the production drilling rigs should fail, and therefore the planned service and corrective
maintenance were assumed to be relatively constant.
In Paper IV, MATLAB code was used to generate data sets for different time scales for using a
redundant rig, after discussion with an expert group at the collaborating mining company (see
Figure 4 and Table 3 in Paper IV).
17

3.1.1 Research objects
The present research was performed on four drilling rigs working in two different
underground mines in Sweden. The first underground mine began operating in 1940 and is
classified as one of the oldest operating mines in Sweden. The predominant mining method
used in this mine is cut-and-fill mining with hydraulic backfill (Krauland et al., 2001). The
geology of the mine is irregular and complex. The complex ore is presently extracted at a
depth of 900-1,400 m below the surface. The ores extracted from the first mine are lead, gold,
copper, zinc, gold-copper and silver ores (Rådberg et al., 1992). In this mine, the uniaxial
compressive strength of the intact host rock varies from 65-150 MPa and the geological
strength index (GSI) varies between 50 and 80 (Edelbro, 2008). The second underground
mine began operating in 1952 and is located in northern Sweden. The extraction process in
this mine is the cut-and-fill method. The extracted ore is polymetallic and contains copper,
gold, silver, lead and zinc. The geological strength index of the rock varies between 70 and 80
at a depth level of •1115-1130 m and the rock mass is of high quality, while at a depth level of
•1130-1185 m, the rock mass quality is lower, with a geological strength index equal to 30-50
(Edelbro, 2009).
The drilling rig investigated in the present research is the L2C drilling rig, which is a model
typically used in underground mining in Sweden. The dimensions of this rig are as follows:
length 14.5-16.6 m; width 2.5 m; width of the rig with side platforms 2.9 m; height of the rig
with the cabin 3.15 m; weight 26-33 tons (see Figure 1.2). The rig has four retractable
stabilizer legs and an articulated four-wheel drive chassis. It can be operated by a water-cooled
turbocharged diesel engine with 120 kW at 2,300 rpm or by electric power with a capacity of
158 kW (Atlas Copco Rock Drills AB, 2010).
3.1.2 Failure and cost data

The failure data used in this research (see Paper I) were collected over a period of two years
(2009-2011). The source of the data is the database of the two underground mines in Sweden
participating in the study. This database belongs to the MAXIMO system, which is a
computerized maintenance management system (CMMS). In this research study, the time to
failure data (TTF data) and the time to repair data (TTR data) of four drilling rigs and their
subsystems were arranged in chronological order so that statistical analysis could be used to
determine if there was any trend in the failure and repair data. The framework for the basic
methodology used in this study for the analysis of the failure and repair data is presented step-
by-step in Figure 3.7.
The first step in analysing the data was calculation of the times between failures (TBFs) for the
system. In the CMMS, the failure data are recorded based on calendar time. Since drilling is
not a continuous process, the TBFs were estimated by considering the utilization of each rig.
Reliability and maintainability data analysis is usually based on the assumption that the TBF
and TTR data are independent and identically distributed (iid) in the time domain. Therefore,
it was critical to conduct a formal verification analysis of the assumption that the TBF and
TTR data were iid; otherwise completely wrong conclusions could be drawn (Ascher and
Feingold, 1984; Kumar and Klefsjö, 1992). The next step, after sorting and classifying the TBF
and TTR data based on the component level, was validation of the iid assumption. The failure
data were tested for trends using the Laplace trend test. This trend test is used to determine
whether the data set is identically distributed (Klefsjö and Kumar, 1992). If such a trend is
observed, then classical statistical techniques for reliability analysis may not be appropriate.
Therefore, a non-stationary model such as the non-homogenous Poisson process (NHPP) must
be fitted (Ascher and Feingold, 1984; Kumar and Klefsjö, 1992; Modarres, 2006; Birolini,
18

2007; Louit et al., 2009; Ghosh and Majumdar, 2011). Otherwise, the serial correlation test
can be used to test the dependence of the failure data. A dependence test determines whether
successive failures are dependent in data without a long-term trend (Klefsjö and Kumar, 1992).
If a dependence between successive failure data is observed, a branching Poisson process (BPP)
model can be used in this case (Ascher and Feingold, 1984). If a dependence is not observed,
then the iid assumption is valid. After testing the validity of the iid assumption, different types
of statistical distributions were examined and their parameters were estimated by using the Easy
Fit and Minitab software. The goodness of fit of the distribution was tested by using the
Kolmogorov-Smirnov (K-S) test with the Easy Fit software. In the present study, all the
statistical tests were conducted by using a significance level (α) equal to 0.05.

Figure 3.7. Flowchart for analysing failure data (adapted from Asher and Feingold (1984))
The cost data used in the present research (see Papers II-IV) were collected over a period of
around four years. The CMMS is the source of the cost data for the single rig operating in the
second collaborating mine in this study, as mentioned earlier. The cost data contain corrective
maintenance costs, preventive maintenance costs and repair times. The corrective and
preventive maintenance costs contain costs for spare parts and labour (repair personnel). It is
important here to mention that all the cost data used in this study concern real costs without
inflation. Due to the company regulations, all the cost data were encoded and expressed in
currency units (cu) for this research. A sample of the cost data can be seen in Table 3.1.
Table 3.1. Sample of cost data (Date format: ‘‘year-month-day’’)
Work description Working Materials Labour Service Total Date Component Work
time (h) cost (cu) (cu) cost (cu) cost (cu) Description type
Extension 2 bolts of 1 28.148 0.45 0 28.598 20xx-03-15 Feeder PM
V-feeder 13:23
FU1 Atlas L2C/2 5 9.836 0 4.182 14.018 20xx-03-15 PM
13:24
Mounting the sensor 6 0 2.7 0 2.7 20xx-03-15 Steering system CM
cables 22:41
Atlas Copco L2C 16 0 7.2 0 7.2 20xx-03-16 Electrical CM
13:17 system
Replacing the hose 0.5 0 0.225 0 0.225 20xx-03-19 Hoses CM
feeding 07:30
19

Figure 3.8 shows the flow chart for the data collection and analysis process for the LCC
optimization model in Papers II-IV.
Figure 3.8. Flow chart for the data collection and analysis process used for the LCC model
20

3.2 Method used in Paper I
Paper I analyses the failure and repair data of four drilling rigs working in two different mines
in order to estimate the downtime interval of each component for each rig. The aim of the
downtime interval estimation is to identify the problems affecting the rig availability, to
determine whether they are reliability problems and/or maintainability problems. Downtime
analysis usually involves different groups of people who have different knowledge and
backgrounds (Wikoff, 2008). For example, the present study involved three main groups,
personnel representing the manufacturing company, personnel representing the company using
the drilling rig, and representatives of academia. Given the groups’ diversity, it was vital to
make the communication between them more effective and clear. To do so, a downtime
visualization method developed by Wijaya in 2011 was used in this paper. This visualization
method is based on the jack-knife diagram and presents a point estimation (a single value) of
the downtime (Knights, 2001). To overcome the shortcomings of the point estimation of the
downtime, an interval estimation of the downtime was used in this paper. This downtime
interval involved three estimation points, namely the mean, the upper limit and the lower limit
of the downtime.
In making the interval estimation, the method assumed that the failure times followed the
Weibull distribution and the repair times followed the lognormal distribution. The failure and
the repair data for the iid assumption and the goodness of fit of probability distributions were
tested (see Figure 3.7). The confidence interval of the estimated downtime for a given uptime
was solved by finding a confidence interval for the ratio of the mean time to repair (µx) to the
mean time to failure (µy). In this study, the ‘‘exact method’’ (Masters et al., 1992) was
implemented to estimate the confidence interval of μx/μy. The interval estimation of the
downtime was then plotted on a log-log diagram as a function of the log number of failures
(the vertical axis) and the log repair time (the horizontal axis). The used method was then
applied to identify the critical components that made a greater contribution to the rig
downtime. These components could be ranked based on the three estimation points of the
downtime in order to highlight the component which should be prioritized and improved
first. Based on discussions with the reference group, interviews with operators and field
observations, some minor suggestions were made as to how to improve the critical
components of the drilling rigs used in the collaborating mines.
3.3 Method used in Papers II-IV

Papers II-IV propose an approach to life cycle cost optimization of a single drilling rig for one
mine in Sweden. The proposed approach is based on the calculations of the operating and
maintenance cost of the case study. Extrapolation for the maintenance and operating cost was
performed to generate data for the unavailable period during the life cycle of the rig. A
declining balance depreciation method was used to estimate the resale value of the case study
after each month of operation. In this method, a fixed percentage of the book value at the
beginning of the month represented the monthly depreciation of the rig (Luderer et al., 2010;
Eschenbach, 2010; Dhillon, 2010). The rig resale value is the rig’s value if/when the firm
wants to sell it at any time during its lifetime. A life cycle optimization model was developed
to estimate the optimal replacement time (ORT) of our case study. In Paper II, after the ORT
was calculated, a single-variable sensitivity analysis and a multivariable sensitivity analysis were
carried out to determine the effect of the input factors for the optimization model on the
ORT of the new model of the drilling rig. A linear regression analysis was also performed on
the results obtained from the MATLAB codes, and this analysis was used to carry out a
sensitivity analysis resulting in a mathematical model which could be used for the new model
21

of the ORT estimation. The graphical user interface method was used to visualize the results
of the optimization model and the sensitivity analysis.
In Paper III, the same approach as in Paper II was followed, but in Paper III artificial neural
network techniques were used for the following two main reasons.
 One aim was to help the engineers and decision makers in the user company to
estimate the ORT of new drilling rigs without needing to use complicated software.
 Another aim was to determine the relative importance of factors which were used in
the optimization model and which would affect the ORT of new drilling rigs. The
factor which has the highest impact on the ORT of new rigs should be prioritized in
the development process of new drilling rigs.
Paper IV presents a practical model for determining the economic replacement time of
production rigs. The objective is to minimise the total cost of this type of capital equipment,
and included in the total cost are acquisition, operating and maintenance costs, and costs
related to the rig’s downtime. The costs related to the rig’s downtime are represented by the
costs of using a redundant rig. The same approach as that used in Papers II and III was
followed in Paper IV, but in this paper our attention was focused on the loss of production due
to the rig downtime. Data generation was performed for different time scales used in the
calculations of the redundant rig cost for the lifespan of the rig.
22

4 SUMMARY OF APPENDED PAPERS
This chapter presents a summary of the four appended papers. Each paper makes its own
contribution towards the research questions and reports the findings of the case studies. The
relation between the papers and the research questions is illustrated in Table 2.1. For detailed
information, the reader is referred to the appended papers.
4.1 Paper I
Purpose: The purpose of Paper I is as follows: 1) to estimate the downtime of the drilling rigs’
components at a given confidence level; 2) to identify the critical components of drilling rigs;
3) to analyse the reliability and downtime of several drilling rigs to determine what kinds of
problems affect their downtime; 4) to specify which strategies, i.e. design for maintainability
and/or design for reliability, should be applied to reduce the drilling rigs’ downtime; and 5) to
suggest some improvements for the components that contribute most to the rigs’ downtime.
Findings: The visualization method used to analyse the downtime of the drilling rigs reveals
the following: 1) there are notable differences in the downtime of most of the studied
components for all the rigs; 2) the rigs' components are ranked based on three estimation
points for the downtime; 3) the hoses, feeder and rock drill have relatively the highest
downtime compared with the other components of the drilling rigs; 4) the downtime of the
studied components is due to reliability problems, and therefore design for reliability (DFR)
should be adopted to reduce their downtime; and 5) no maintainability problems have been
detected for the rigs’ significant components, and therefore a design for maintainability strategy
is not required.
4.2 Paper II
Purpose: Paper II presents a practical model that can be used to identify the replacement age
of drilling rigs from an economic point of view, using available data from the mining
company.
Findings: According to the results obtained from the optimisation curve, the absolute ORT
of the drilling rig at the case study’s mine is 115 months of operation. However, the ORT has
a range of 110 to 122 months during which the total cost remains almost constant. This means
that the company has the flexibility of being able to make replacements within the optimum
replacement age range, i.e. 12 months. Therefore, there is no fixed date or age at which the
total cost is at a minimum. In general, a range of months provides the minimum total cost. The
results of the sensitivity analysis performed to identify various factors’ effect on and correlation
with the optimal replacement time indicate that increasing the purchase price and decreasing
the operating and maintenance costs have a positive effect on increasing the ORT. The results
of the regression analysis using three factors, namely the increasing factor of the purchase price
(IFPP), the reduction factor of the operating cost (RFOC), and the reduction factor of the
maintenance cost (RFMC), show that the ORT of the new rig depends linearly on its IFPP,
RFOC and RFMC. These results confirm the computation and the results of the sensitivity
analysis. The results of the regression analysis show that the maintenance cost has the largest
impact on the ORT, followed by the purchase price and the operating cost. Hence, the
manufacturer must make a greater effort to improve the reliability and maintainability of the
23

drilling rig to reduce the costs associated with maintenance and to increase the ORT. The
personnel at the financial department of the user company can easily use the graphical user
interface (GUI) to estimate the ORT of a new rig and see the behaviour of its ORT for
different values of its IFPP, RFOC and RFMC. These factors will provide a clear view of the
ORT of the new rig. This knowledge will help the user company determine when to buy a
new rig and assist them in any negotiations with the manufacturer concerning the purchase
price.
4.3 Paper III

Purpose: This study presents a new model for easy prediction of the ORT of drilling rigs
using ANN techniques. The ANN was used to design a model which can be employed to
calculate the ORT of a drilling rig. The model is based on a series of basic weights and
response functions which make the ORT estimation easy for any engineer to use without
resorting to complicated software. An Excel spreadsheet can be used as a substitute for fast and
accurate calculation of the ORT.
Findings: The study finds that increasing the purchase price and decreasing the operating and
maintenance costs will increase a rig’s ORT linearly. Decreasing the maintenance cost has the
largest impact on the ORT, followed by increasing the purchase price and decreasing the
operating cost by the same percentages. The ANN gives a series of basic weights and response
functions which can be made available to any engineer without the use of complicated
software. In addition, the ANN also helps decision-makers determine the best time
economically to replace an old rig with a new one. The ANN model is very effective in
estimating and predicting the ORT of a mining drilling rig, since the correlation between the
input and output variables is very high and the accuracy is 99%. Moreover, the model’s
performance is very consistent with data used for training (seen) and testing (unseen).
4.4 Paper IV
Purpose: This paper presents a practical model to determine the economic replacement time
(ERT) of production rigs when also the costs from one redundant rig are considered. The
objective is to minimise the total cost of this type of capital equipment, and included in the
total cost are acquisition, operating and maintenance costs, and costs related to the rig’s
downtime. The costs related to the rig’s downtime are represented by the costs of using a
redundant rig.
Findings: The results show that the lowest possible total cost when the redundant rig cost is
equal to 1 cu/h can be achieved by replacing the rig at 104 months of its planned lifetime.
There is a range between 97 and 109 months when the minimum total cost can still be
achieved in practice. Finding the economic replacement range is an important result of our
study, as it can help decision makers in their planning. The results also show that the redundant
rig cost has the largest impact on the ERT, followed by the acquisition, maintenance and
operating costs. The study also finds that increasing redundant rig costs per hour have a
negative effect on the ERT, while decreases in other costs have a positive effect. Regression
analysis shows a linear relationship between the cost factors and the ERT.
24

5 RESULTS AND DISCUSSIONS
The findings of the conducted research are discussed and presented in this chapter according to
the stated research questions.
5.1 Results and discussion related to research question 1

RQ 1: Which critical components make a large contribution to the downtime of the drilling rig? What are
the dominant problems influencing the drilling rig downtime, reliability problems and/or maintainability
problems?
The first research question of the study is answered in Paper I, whose findings are briefly
discussed here.
The downtime visualisation approach proposed by Wijaya (2001) is used in this paper to
visualize the downtime of drilling rig components. In this approach, the failure times are
modelled with a Weibull distribution and the repair times are modelled with a lognormal
distribution, and these distributions are generally valid for this method (Abernethy, 2000;
Wiseman, 2001; Rausand and Høyland, 2004; Schroeder and Gibson, 2005; Bovaird and
Zagor, 2006).
Figure 5.1 represents the log-log graph used in this paper. The graph shows the log number of
failures (the vertical axis) and the log repair time (the horizontal axis). The curves of the
constant downtime appear as straight lines with a uniform negative and constant gradient. The
downtime confidence interval was used to analyse the downtime of the components of the
drilling rig. Three equations were used to establish the downtime confidence log-log plot. The
estimation points were the mean (DTM), the lower limit (DTLL) and the upper limit (DTUL) of
the downtime. To estimate the interval of the downtime, the three estimated points were
connected by a straight line. mM, mUL, mLL, MTTR, TTRUL and TTRLL represent the coordinates
of the three previous estimated points. The downtime was calculated for 5,300 hours, which
represents the theoretical number of production hours of one year.
In Paper I we made two main types of comparisons based on the downtime of the rigs’
components. The first type was a comparison between the three rigs working in the same
mine, which we call mine Y, and the second type was a comparison between two rigs working
in different mines (i.e. mine X and mine Y).
Figure 5.2 illustrates the application of the used method for the first type of comparison (for
the rest of the results, please consult Paper I). Figure 5.2 indicates that components C2 (the
hoses in rig 2), B3 (the feeder in rig 3) and A3 (the rock drill in rig 3) have a higher downtime
than all the equivalent components in any other rig used in the same mine. Components F
(cables), E (booms) and D (accumulators) have a lower downtime compared with components
C, B and A. Figure 5.2 also shows that components C and B have more than 100 failures and a
repair time of less than 3 hours. The results show that most of the drilling rig’s components
have reliability problems, and therefore design for reliability (DFR) should be adopted to
reduce the downtime of the drilling rig.
Figure 5.3 illustrates the application of the used method for the second type of comparison (for
the rest of the results, please consult Paper I).
25

90
70
60
mUL DTUL
mM DTM
mLL DTLL
No. of failures
10
5
4
3
2
1
1 TTRLL TTRUL 7 10 20 30 50 80 130
MTTR Time to repair
Figure 5.1. Log-log plot of downtime confidence interval
135 300
C2
A4 A3 200
100 C4
B3
A2
100 C3 B2
57.
E2 B4
No. of failures
No. of failures
E4
D2
E3
F3 F2
14.2
D3 F4
10 10
D4
6 5 Time to repair (hours)
0.7 1.1 Time to repair (hours)10 16.6 24 1 2 40 70
Figure 5.2. Confidence log-log plot for the first type of comparison
Figure 5.3 illustrates that components A1 and B1, used in rig 1 in mine X, have less downtime
than the corresponding components of drilling rigs used in mine Y. This may be due to the
differences in the rock properties between the two mines. The geological strength index (GSI)
of mine Y varies between 50 and 80 (Edelbro, 2008), while the GSI of mine X varies between
30 and 50 (Sjöberg, 2003, as cited in Edelbro, 2008).
26

190 160
B3
100 B2
80 B4
A4 A3
100
No. of failures
No. of failures
A2
A1
B1
20
40
30 10
1 1.3 1.6 Time to repair (hours) 7.5 1 2 Time to repair (hours) 10 20
Figure 5.3. Confidence log-log plot for the second type of comparison
The order of the significant components has been prioritised based on three estimation points
of the downtime (see Table 5.1). These estimation points are the mean, the upper limit and
the lower limit estimation point of the downtime. This ranking is important to the
maintenance decision makers for certain maintenance activities in the mine, for instance
planning new operations, making new service schedules and budgeting maintenance. For
example, if the maintenance management department determines the highest acceptable
amount of rig downtime at the component level, it is essential for the maintenance staff to
know which components are likely to exceed the acceptable limit. If is decided that the
highest acceptable amount of downtime is 350 hours per year, based on the upper limit of the
estimation point of the downtime, they can observe from Figure 5.2 that components C2, B3
and B2 will exceed 350 hours of downtime. In this case, a good strategy may be to convince
the manufacturing company to improve the reliability of these components (i.e. the hoses and
feeder) in order to reduce their number of failures and increase their lifetime. Another
possibility is to increase the preventive maintenance on these particular components.
Paper I also presents some minor suggestions as to how to improve the critical components in
order to reduce their downtime. Discussions with maintenance personnel have revealed that
most of the failures of the feeder hoses are due to the mine’s environment. For example,
during drilling, the feeder hits the wall at different angles, especially when the feeder
movement is restricted because of spatial limitations. To overcome this problem, the feeder
could be equipped with steel plates on both sides; the plates should be large enough to prevent
the hoses from being scraped and scratched and to prevent the nipples at the necks from being
broken, see Figures 5.4, 5.5 and 5.6.
Another problem in the feeder is breakage of the pull rope, which can happen for two reasons.
Firstly, the pull rope can become slack due to usage and then hang over the edge of the cradle
plate when the plate moves back and forward on the slide bar. Secondly, the operator or repair
technician may put extreme tension on the pull rope when making an adjustment, see Figures
5.7 and 5.8.
27

Table 5.1. The order of the drilling rigs’ components based on their downtime
Type 1 Type 2 Type 3 Type 4
Mine Y three machines Mine X&Y four machines Mine X&Y three machines Mine X&Y two machines
Scenario Scenario Scenario Scenario
Order 1 2 3 Order 1 2 3 Order 1 2 3 Order 1 2 3
1 C2 C2 C2 1 C2 C2 C2 1 G1 G1 G1 1 I2 I2 I2
2 B3 B3 B3 2 B3 B3 B3 2 G3 G3 G3 2 I1 I1 I1
3 B2 B2 C4 3 B2 B2 C4 3 H3 H3 H3
4 C4 A3 B2 4 C4 A3 B2 4 G4 G4 G4
5 A3 B4 B4 5 A3 B4 B4 5 H1 H4 H4
6 B4 C4 A3 6 B4 C4 A3 6 H4 H1 H1
7 C3 E2 C3 7 C1 C1 C1
8 A4 C3 A4 8 C3 C3 C3
9 E2 A4 A2 9 A4 A4 A4
10 A2 A2 E2 10 A2 A2 A2
11 E4 E4 E4 11 A1 A1 A1
12 E3 E3 E3 12 B1 B1 B1
13 D2 F2 F2
14 F2 D4 D2
15 F3 F4 F3
16 D4 D3 D3
17 D3 D2 D4
18 F4 F3 F4
Figure 5.4. Nipples and necks Figure 5.5. Scraped and scratched hoses
Figure 5.6. Suggested plate of steel

28

Figure 5.7. Hanging pull rope Figure 5.8. Broken pull rope
To solve this problem, an electric motor with a control circuit should be designed to make
automatic adjustments of the pull rope, or a coil spring should be installed, in order to keep the
rope at a constant desired tension, as shown in Figure 5.9. Stronger roller bearings are another
possibility.
Electric Motor
Gear
Figure 5.9. Suggestion to prevent pull rope failures

Discussions with maintenance personnel have also revealed that the frequent failures of the
rock drill damage the third and fourth cup seals located inside the front head (nose) of this
component, as shown in Figure 5.10.
Figure 5.10. Damaged cup seal

A possible cause is the high water pressure inside the nose. Water is used to cool the front head
and flush it during the drilling process. However, damaged cup seals will cause water and oil to
mix, leading to abnormal wear of the valves used in the hydraulic system. To solve this
problem, the water pressure inside the front head should be reduced by increasing the number
of holes, especially in the area between the third and the fourth cup seals. Further research is
needed to confirm the above explanation for cup seal damage.
29

RQ 2: How should a model be constructed that can calculate the optimum replacement time of a drilling
rig? Moreover, which cost factors have the largest impact on the replacement time?
The second research question of the study is answered in Papers II, III and IV. The drilling
rig’s optimal replacement time (ORT) and the cost factors affecting the rig’s ORT are briefly
discussed in Paper II. In this paper, a data-driven optimisation model was developed for the
operation and maintenance costs, the purchase price and the rig resale value. The paper also
presents the values of these costs by using the discount rate.
Figure 5.11 shows that the absolute lowest possible total ownership cost value (TOC value) can
be achieved by replacing the rig with an identical new one every 115 months. Figure 5.11 also
shows that there is a range of 110-122 months when the minimum TOC value can still be
achieved in practice. In this study, we call it the optimum replacement range. Finding the
optimum replacement range is an important result of our study, as it can help users in their
planning.
A decision to replace the equipment before or after this optimum replacement range will incur
greater cost for the company. The use of a lower replacement age (i.e. less than 110 months)
will incur higher costs due to the high investment cost. Moreover, if the lifetime of the
machine exceeds the upper limit of this range (i.e. more than 122 months), the losses will
increase for the following two reasons.
1. The cost of operation and maintenance will increase due to rig degradation when the
operating time increases.
2. The rig’s resale value will decrease for each month of operation until the rig reaches its
scrap value at the end of its planned lifetime.
5 Optimal replacement time
x 10
2
Total ownership cost value (cu)
1.8
1.5
1.2
0.9
0.6
0.3
115
0
0 25 50 75 100 125 150 175 200 225 250
Replacement time (month)
Figure 5.11. Optimal replacement time of existing drilling rig
In order to demonstrate the influence of possible options of bathtub curves on the ORT of
mining drilling rig, it is interesting to see the effect of the first stage (i.e. 0-25 months) of the
curve in Figure 5.12 on the ORT of our case study.
30

Figure 5.12. Maintenance cost (first model)
However, since the curvature in Figure 5.12 is difficult to present as a continuous function,
which is needed for calculations, the function in Figure 5.13 is used instead as an approximate
representation of a possible first stage of a bathtub curve in our case study.
Maintenance cost
400
Expected maintenance cost (cu)
300
200
100
-100
-200
0 50 100 150 200
Time (month)
Figure 5.13. Maintenance cost (second model)
The function of extrapolation for the expected maintenance cost in Figure 5.13, which was
obtained using the Table Curve 2D software, is expressed as
Y  a  b  x 2 x  c  x3 (5.1)
where Y represents the expected maintenance cost, a=65.45, b=-0.006, c=0.0006, r2

(adj.)=0.97, and x represents the time (1, 2, 3, 4,…, n months).
Since the maintenance cost (MC) is increasing too much in Figure 5.13, this curve is not
realistic in the third stage (i.e. 76-145 months). In this case, the values of the MC for a period
of 0 to 120 months are used in ORT calculations. The result for the ORT obtained by using
the function in Figure 5.13 (i.e. 0-120 months) is 113 months. This result is close to the result
31

for the ORT of 115 months which was obtained by using the function in Figure 5.14 for the
same interval.
The MC in the first stage of Figure 5.13 is higher than the MC in the same stage of Figure
5.14. This is the reason why the calculated ORT is less than 115 months. Assuming that the
first stage in Figure 5.12 can, regarding the influence on the ORT, be approximated by the
first stage in Figure 5.13, since the two ORT results (i.e. 113 and 115) obtained using the two
functions in Figures 5.13 and 5.14 show only a small difference in their values, it can be
concluded that the difference between the ORT calculations using the two functions in
Figures 5.12 and 5.14 will be even smaller. In conclusion, the function used in Figure 5.14 is
reasonable for calculating the ORT of the drilling rig. This conclusion is also preferred by the
Boliden experts, who consider the curve shown in Figure 5.14 to be the most reasonable curve
(no recorded data exist for the period before 76 months of operation).
Maintenance cost
400
300
200
100
-100
-200
0 50 100 150 200
Time (month)
Figure 5.14. Maintenance cost (third model)

The function of extrapolation for the maintenance cost in Figure 5.14, which was obtained by
the Table Curve 2D software, is expressed as
a  x b   
Y  arctan     (5.2)
  c  2
2
where Y represents the expected maintenance cost, a=217.42, b=112.37, c=13.63, r (adj.) =
0.97, and x represents the time (1, 2, 3, 4,…, n months).
In order to ascertain which cost factor has the largest impact on the ORT of the drilling rig,
sensitivity and regression analyses were performed for the operating and maintenance costs, as
well as the rig purchase price. Figure 5.15 illustrates the effect of an increasing purchase price
(PP) on the ORT of the drilling rig.
Figure 5.15 shows that the ORT is an increasing step function of the PP (based on percentage
increases in the purchase price); the ORT remains constant for a specific range of PP
increments, and then increases stepwise. The reason for this stepwise behaviour is that, due to
the monthly calculations for the operating cost, the maintenance cost and the rig resale value,
the resultant ORT from the optimization model is then computed on a monthly basis.
32

121
ORT (month)
119
117
115
113
0 10 20 30 40 50
PP increase (%)
Figure 5.15. Effect of increasing purchase price

Figure 5.16 illustrates the effect of decreasing the rig operating cost (OC) (based on percentage
decreases in the operating cost) on the ORT. It is obvious that when the rig’s operating cost
decreases, the ORT will increase. This means that the ORT is not sensitive to a specific range
of operating cost reductions and will increase stepwise at specific OC rates of reduction, i.e. 15
and 34%.
117
ORT (month)
116
115
114
0 10 20 30 40 50
OC decrease (%)
Figure 5.16. Effect of decreasing operating cost

Figure 5.17 illustrates the effect of decreasing the rig maintenance costs on the ORT of the
drilling rig. When the maintenance cost (MC) decreases, the ORT will increase. In addition,
note that the ORT increases at certain reduction percentages of the MC, i.e. 7, 15, 23, 30, 36,
42 and 48%.
122
ORT (month)
120
118
116
114
0 10 20 30 40 50
MC decrease (%)
Figure 5.17. Effect of decreasing maintenance cost

33

Figures 5.15, 5.16 and 5.17 show that, with an increasing purchase price and decreasing
operating and maintenance costs, the ORT of a new model of this rig will increase stepwise at
specific percentages of these factors. The results also show that these factors exert a significant
effect on the total ownership cost at these specific percentages of the IFPP, RFOC and
RFMC. The explanation is that a new model of the drilling rig is assumed to be more reliable
than the replaced one. This will lead to a decrease in the number of failures in a new model of
this rig, which, in turn, will reduce the maintenance cost. A new model of this rig will also be
more productive than an old one and thus finish the same job in less time. This will decrease
the energy consumption of a new model of the rig, which will lead to a reduction in the
operating cost.
Our regression analysis of the ORT results obtained from the previous three cases used the
Minitab software and the least squares method. The ORT of a new model of the drilling rig
was modelled as a linear function of the IFPP, RFOC and RFMC. The regression analysis
results in the following mathematical model:
ORT  114  0.133  IFPP  0.0682  RFOC  0.164  RFMC (5.3)
2
The high R-squared adjusted value obtained from the regression analysis, R (adj) = 98.6%,
indicates that the ORT of a new model of this rig depends linearly on the IFPP, RFOC and
RFMC. Following the results of the sensitivity and regression analyses, the rank of the factors
affecting the ORT of a new model of a drilling rig is as follows:
1. reduction in the maintenance cost,
2. increase in the purchase price,
3. reduction in the operating cost.

RQ 3: How should the above model be improved regarding easy and accurate prediction of the economic
lifetime of this rig, with a high level of certainty and with a view to enhancing decision making in this
regard?
The third research question is answered in Paper III. This paper proposes a new model for easy
prediction of the economic replacement time of a drilling rig by using the artificial neural
network technique. In this study, three cases, represented by three MATLAB codes for the
optimization model, were used. The ORTs resulting from these codes were fed as inputs into
an ANN and the results translated into a relatively simple model based on weights and response
functions. Ninety percent of the data determined by the MATLAB codes were used in the
training of the neural network, see Figure 5.18. The model shown in this figure has very high
values for R2 = 0.99. However, the neural network model yields outputs very close to the
desired targets with a high level of accuracy.
After the training was completed, the network was tested for its generalization capabilities by
investigating its ability to respond to the input sets (unseen data) not included in the training
process. Therefore, ten percent of the data determined by the MATLAB codes were used for
tests with the obtained network (see Figure 5.19). As is evident in Figure 5.19, the model has
2
very high values for R = 0.99 for the ANN. However, the neural network model yields
outputs very close to the desired targets with high levels of certainty.
Also studied in Paper III is the relative importance of each factor affecting the rig replacement
time. As is evident in Figure 5.20, the most important factor influencing the ORT of the
drilling rig is a decreasing maintenance cost (DMC), followed by an increasing purchase price
(IPP). A decreasing operating cost (DOC) has the least influence. This result confirms the
34

results presented in Paper II regarding the factors exerting the greatest influence on the rig
replacement time.
Outputs vs. Targets, R=0.9962 Outputs vs. Targets, R=0.99653
Outputs Y, Linear Fit: Y=(1)T+(-2)

115
Outputs Y, Linear Fit: Y=(1)T+(-0.055)
115
Data Points Data Points
Best Linear Fit
Best Linear Fit
Y=T
Y=T
110 110
105 105
100 100
95 95
95 100 105 110 115 95 100 105 110 115
Targets T Targets T
Figure 5.18. Learning capability Figure 5.19. Generalization capability
50 49.5
40
Importance (%)
35.0
30
20 15.4
10
0
IPP DOC DMC
Figure 5.20. Relative importance of the factors affecting the ORT of the drilling rig
The main advantage of the proposed neural network model is its ability to produce acceptable
results; the correlation between the input and output variables is very high and the accuracy is
more than 99%. Moreover, the model’s performance is very consistent for data used for
training (seen) and testing (unseen). Therefore, the model is very effective in estimating and
predicting the ORT of a mining drilling rig with higher levels of certainty. Further, the ANN
gives a series of basic weights and response functions which can be made available to any
engineer without the use of complicated software; for more details see Paper III among the
appended papers.

RQ 4: How should a model be constructed for estimation of the economic lifetime of mining production
rigs when also the costs from one redundant rig are considered?
The fourth research question of the study is answered in Paper IV. This paper aims to present a
model to determine the economic replacement time (ERT) of production rigs. The objective
35

is to minimise the total cost of this type of capital equipment, and included in the total cost are
acquisition, operating and maintenance costs, and costs related to the rig’s downtime. In the
study, the costs related to the rig’s downtime are represented by the costs of using a redundant
rig. We tested the model on a drilling rig used in one underground mine in Sweden. The
artificial neural network technique was used to identify the relative importance of factors
influencing the drilling rig’s ERT.
In Paper IV we focus on the redundant rig cost as one of the critical factors affecting the rig’s
ERT. Mining companies lose a large amount of money each year because of lost production,
which, in turn, is due to the production equipment’s downtime. In fact, this may be the most
important factor affecting the ERT of production rigs.
In this study, a discussion held with the reference group at the mining company revealed that,
after a drilling rig has failed, it is sent to the workshop for maintenance. To continue
production without stoppages, the company uses a redundant rig which has the same
performance as the existing faulty production rig. Since in the mining industry the downtime
in production is almost zero, the compensation cost in this case represents the cost of using a
redundant rig. In this study, the experts at the collaborating mine classified the rig’s failures
into the following three categories:
1. failures fixed by the maintenance team at the workshop,
2. failures fixed by the maintenance team at the production point (the mining room),
3. failures fixed by the operators at the production point (the mining room).
For more details about the method for calculating the redundant rig cost based on these three
failure categories, see Paper IV among the appended papers.
Figure 5.21 shows the optimization curve and the ERT of our case study when the redundant
rig is used for a cost equal to 1 cu/h. The results show that the lowest possible total cost can be
achieved by replacing the rig at 104 months of its planned lifetime. As Figure 5.21 also shows,
there is a range of 97-109 months when the minimum total cost can still be achieved in
practice.
4 Economic replacement time of the drilling rig
x 10
18
Redundant rig cost = 1 (cu/h)
15
Total cost (cu)
10
5
ERT=104 months
0
0 20 40 60 80 100 120 140 160 180 200 220 240
Figure 5.21. Economic replacement time of the drilling rig
To show the effect of the redundant rig cost per hour on the ERT of our case study, we
changed the values of the redundant rig cost per hour from 1 to 6 cu/h. Figure 5.22 shows the
result.
36

Effect of redundant rig cost per hour
CRi=1 (cu/h). ERT=104 month CRi=2 (cu/h). ERT 94 month
180000 CRi=3 (cu/h). ERT 87 month CRi=4 (cu/h). ERT 82 month
CRi=5 (cu/h). ERT 79 month CRi=6 (cu/h). ERT 76 month
150000 CRi = Redundant rig cost per
Total cost (cu)
120000
90000
60000
30000
0
0 20 40 60 80 100 120 140 160 180 200 220 240
Figure 5.22. Effect of the redundant rig cost per hour on the ERT of the drilling rig
It is clear from Figure 5.22 that increasing the redundant rig cost per hour has a negative effect
on the ERT of the drilling rig.
To determine the effect of other factors on the ERT, we also performed a sensitivity analysis
on the rig acquisition, operating and maintenance costs, and the redundant rig costs (cu) using
the ANN technique. Four MATLAB codes for six cases of the redundant rig cost (1-6 cu/h)
were used to identify the effect of an increased acquisition cost (IAC), a decreased operating
cost (DOC), a decreased maintenance cost (DMC) and a decreased redundant rig cost (DRC).
The resulting ERT from these four codes was fed as input into the ANN and the results
translated into a relatively simple equation for estimating the ERT of the drilling rig. The
method of partitioning weights proposed by Garson (1991) and adopted by Goh (1995) was
used to determine the relative importance of the various input factors, see Figure 5.23.
60
50
Relative 1mportance (%)
40
30
20
10
0
1 2 3 4 5 6
IAC (cu) 33,1 33,1 31,7 34,6 30,5 32,8
DOC (cu) 14,1 11,9 8,0 9,4 10,3 3,2
DMC (cu) 20,4 15,0 17,7 10,1 12,3 6,2
DRC (cu) 32,2 39,8 42,3 45,7 46,6 57,5
Figure 5.23. Relative importance of input factors on the ERT of the drilling rig
It is evident in Figure 5.23 that the most important factor is the redundant rig cost, followed
by the acquisition, maintenance and operating costs. As mentioned earlier, four MATLAB
37

codes were used to identify the effect of IAC, DOC, DMC and DRC on the ERT of the
drilling rig. We chose the case where the redundant rig cost is equal to 1 cu/h to demonstrate
the effect of these factors on the ERT. Figure 5.24 shows the correlation of DRC and IAC for
a given 25% DOC and DMC. Figure 5.25 shows the correlation of DRC and DMC for a
given 25% IAC and DOC. Figure 5.26 shows the correlation of DRC and DOC for a given
25% IAC and DMC.
126 IAC=10% DOC=25%
IAC=20% DMC=25%
124 IAC=30%
IAC=40%
122
IAC=50%
ERT (month)
120
118
116
114
112
110
0 5 10 15 20 25 30 35 40 45 50
Decreasing redundant rig cost (%)
Figure 5.24. Correlation of DRC and IAC for a given 25% DOC and DMC
126 DMC=10% IAC=25%

DMC=20% DOC=25%
124 DMC=30%
DMC=40%
122 DMC=50%
ERT (month)
120
118
116
114
112
0 5 10 15 20 25 30 35 40 45 50
Figure 5.25. Correlation of DRC and DMC for a given 25% IAC and DOC
124
DOC=10% IAC=25%
DOC=20% DMC=25%
122 DOC=30%
DOC=40%
DOC=50%
120
ERT (month)
118
116
114
0 5 10 15 20 25 30 35 40 45 50
Figure 5.26. Correlation of DRC and DOC for a given 25% IAC and DMC
38

As Figures 5.24, 5.25 and 5.26 show, DRC, IAC, DMC and DOC have a positive effect on
the ERT of the drilling rig, but it is also evident that DRC has a more positive effect, followed
by IAC, DMC and ROC.
The ANN technique is used to construct a formula to calculate the ERT of our case study.
The formula is transformed to an Excel spreadsheet to make ERT estimation quick and easy
for any engineer to apply. The structure of the optimal ANN model is shown in figure 5.27;
its connection weights and threshold levels are summarised in Table 5.2.
IAC (%)
Input factors
DOC (%)
Output
DMC (%) (ERT)
DRC (%)
Figure 5.27. Optimal structure of artificial neural network (ANN) model

Table 5.2. Weights and threshold levels of proposed ANN
Hidden wij Hidden
layer weights from node i th input threshold
nodes layer to node j th hidden layer (θj)
i=1 i=2 i=3 i=4
j=5 0.085 0.089 -0.03 0.108 -8.776
j=6 3.006 0.379 1.441 1.784 -1.120
j=7 1.416 0.452 1.509 1.876 -5.001
Output wij Output
layer weights from node i th hidden Threshol
nodes layer to node j th output layer d (θj)
i=5 i=6 i=7
j=8 4.679 3.410 5.376 -3.581
The formula length depends on the number of nodes in the hidden layer. Adopting three
nodes gives an accuracy of 99%. The small number of connection weights of the neural
network enables the ANN model to be translated into a relatively simple formula, in which the
predicted ERT can be expressed as follows:
ERTs  1 (5.4)
   



1  w 1 1 
8  w5:8
   w
x x 
  7:81e x3  
 1e 1   6:81e 2
1 exp     
where ERTs represents the scaled ERT derived from the ANN model, θj represents the
output threshold and wij represents the weight from node i in the hidden layer to node j in the
output layer. Hence,
x1  5  w5:1  IAC  w5:2  DOC  w5:3  DMC  w5:4  DRC (5.5)
x2   6  w6:1  IAC  w6:2  DOC  w6:3  DMC  w6:4  DRC (5.6)
x3   7  w7:1  IAC  w7:2  DOC  w7:3  DMC  w7:4  DRC (5.7)

39

To obtain the actual value of ERT, the predicted ERTs must be re-un-scaled using the
following formula:
 ERTmax  ERTmin   ERTmax  ERTmin 
ERT  ERTs    0.9    ERTmax (5.8)
 0.8   0.8 
where ERTmax and ERTmin represent the maximum and minimum values of ERT respectively
derived from the optimisation model.
Since the accuracy of ANN model is 99% for tested data set, the connection weights and
threshold levels are valid for this particular case study. The ANN model can be used for ERT
prediction of another production rigs by using new connection weights and new threshold
levels which can be extracted from the ANN based on new cost data for these production rigs.
A discussion held with the reference group at the collaborating mining company revealed that
the overhaul service cost for the redundant rig during its life is 300 cu. The group also stated
that the cost of using the redundant rig is about equal to the maintenance cost of the existing
rig during its lifetime. Based on that, the cost of using the redundant rig per hour could be
calculated. Dividing the sum of the overhaul service cost and the maintenance cost of the
production rig by the number of usage hours of the redundant rig gives 1.5 cu/h. Based on
this new redundant rig cost (i.e. 1.5 cu/h), the ERT of the drilling rig was calculated and
found to be 96 months (i.e. 8 years).
40

6 CONCLUSIONS
The following conclusions can be made on the basis of this thesis.

 There are notable differences in the downtime of most of the investigated components
of all the drilling rigs used in two different mines. For the rigs used in the same mine, 3
out of 6 components show significant differences. For rigs used in different mines, 5
out of 6 components exhibit significant differences.
 The rigs’ components have been ranked based on three estimation points of the
downtime (i.e. the mean, the upper limit and the lower limit). This ranking will help
the maintenance decision makers in redesigning the rigs’ preventive maintenance
schedule and in prioritizing the components that have the highest downtime.
 The downtime (DTUL • 200 hours) of components C (the hoses), B (the feeder), A (the
rock drill), E (the boom) and G (the steering system) stem from reliability problems. It
is concluded that DFR should be implemented to decrease their downtime. Overall,
no maintainability problems were detected for the rigs’ significant components, and
therefore, a design for maintainability strategy is not required.
 The harsh working environment of mining drilling rigs has a strong effect on the
occurrence of failures of the feeder hoses. In this case, the installation of steel plates on
both sides of the feeder can reasonably be assumed to reduce the number of failures.
 The breakage of the feeder pull rope can be caused by the rope becoming slack due to
usage or being subjected to excessive tension; this can possibly be remedied and
reduced by installing a coil spring or installing an electric motor with a control circuit
to keep the pull rope at a constant desired tension.
 Increasing the number of holes between the third and the fourth cup seals inside the
nose of the rock drill could possibly solve the problem of cup seal damage.
 There is an absolute ORT of the drilling rig. However, the ORT also has a range of
months during which the total ownership cost remains almost constant. Therefore, the
company has the flexibility of being able to make replacements within the optimum
replacement age range.
 The results of the regression analysis show that a reduction in the maintenance cost has
the largest impact on the ORT, followed by an increase in the purchase price and a
reduction in the operating cost. Therefore, the manufacturer must make a greater effort
to improve the reliability and maintainability of the drilling rig to reduce the costs
associated with maintenance and to increase the ORT.
 The decision makers at the user company can easily use the optimization model to
estimate the ORT of a new rig and see the behaviour of its ORT at different IFPPs,
RFOCs and RFMCs. These factors will provide a clear view of the ORT of the new
rig. Therefore, this will help the user company determine when to buy a new rig and
assist them in any negotiations with the manufacturer concerning the purchase price.
 The main advantage of the proposed neural network model is its ability to produce
acceptable results; the correlation between the input and output variables is very high
and the accuracy is more than 99%. Moreover, the model’s performance is very
consistent for data used for training (seen) and testing (unseen). Therefore, the model is
very effective in estimating and predicting the ORT of a mining drilling rig. Further,
41

because the ANN model uses a series of basic weights and response functions, it helps
engineers in estimating the ORT of drilling rigs without using complicated software.
 Although the problem of the equipment replacement time has been solved previously
by other researchers using different models, our ERT model examines the relationship
between the factors affecting the ERT of production rigs, especially the cost of using a
redundant rig. The model can also be extended to other capital production assets in
other industries.
42

7 SCIENTIFIC & INDUSTRIAL CONTRIBUTIONS
The scientific contribution of this thesis is that it expands the existing body of knowledge
about the reliability and maintainability analysis of drilling rigs and the problems affecting their
downtime, as well as methods for modelling the optimal replacement time of production
equipment. The knowledge created will have useful applications in the future.
The scientific contribution of this thesis can be summarized as follows:
 deeper insight into how the drilling rig behaves regarding its downtime and related
problems (Paper I),
 models that can be used to estimate the economic replacement time of a drilling rig
(Papers II-IV),
 identification and explanation of the factors that have the largest impact on the rig’s
ERT (Papers II-IV).
The industrial contribution of this thesis is the provision of a method which can be used in
mining companies for estimation of the economic replacement time of production mining rigs.
This estimation will support decision makers in their planning and in their improvement of
components and machinery.
43

44

8 SCOPE OF FURTHER RESEARCH
Based on the conducted research, the following areas are recommended for future research.
 This thesis has dealt with the reliability and maintainability analysis of drilling rigs and
future research should focus on greater use of RAMS analysis (availability due to
maintenance supportability) for such rigs.
 More research is needed on calculation of the cost of improving the reliability and
maintainability of drilling rig components and on estimation of the level of
improvement which has to be attained in order to obtain the optimum reliability of
drilling rigs in a cost-effective way.
 An extension of the ORT study is needed which will include profit maximization by
also using the rig revenue.
 Due to competition, the manufacturing company may need to reduce the price of its
drilling rigs. As a consequence, the reliability of its subsystems will eventually be
downgraded. More research is needed to determine which subsystems can be selected
for reliability reduction and to what extent this should be done to achieve a maximum
price reduction and a maximum system ORT.
 More research is needed to estimate the optimal preventive maintenance replacement
intervals of the drilling rig components.
 The proposed ORT model can be improved so that it can be used to determine the
optimum replacement interval for other systems by modifying the model parameters
based on industry specifications.
45

46

REFERENCES
Abernethy, RB (2000), The New Weibull Handbook, Gulf Publishing: Houston.

Ahmadi, A, Kumar, U (2011), Cost based risk analysis to identify inspection and
restoration intervals of hidden failures subject to aging, IEEE Transactions on
Reliability, 60(1), 197-209.
Altman, DG, Machin, D, Bryant, TN, Gardner, MJ (2000), Statistics with Confidence, BMJ
Books: Bristol.
Ascher, HE, Feingold, H (1984), Repairable Systems Reliability: Modeling, Inference,
Misconceptions and Their Causes, Marcell Dekker: New York.
Atlas Copco Rock Drills AB (2010), Atlas Copco Boomer L1C, L2C Mk 7B Operator’s
Instructions, Manual ed., Atlas Copco Rock Drills AB, Sweden.
Barabady, J, Kumar, U (2008), Reliability analysis of mining equipment: a case study of a
crushing plant at Jajarm Bauxite Mine in Iran, Reliability Engineering & System
Safety, 93(4), 647-653.
Bazargan, M, Hartman, J (2012), Aircraft replacement strategy: model and analysis, Journal
of Air Transport Management, 25, 26---29.
Bean, JC, Lohmann, JR, Smith, RL (1985), A dynamic infinite horizon replacement
economy decision model, Engineering Economist, 30(2), 99-120.
Bellman, R (1955), Equipment replacement policy, Journal of the Society for Industrial and
Applied Mathematics, 3(3), 133-136.
Bethuyne, G (1998), Optimal replacement under variable intensity of utilization and
technological progress, Engineering Economist, 43(2), 85-105.
Birolini, A (2007), Reliability Engineering: Theory and Practice, 5th ed., Springer-Verlag:
Heidelberg, New York.
Blischke, W, Murthy, D (2003), Case Studies in Reliability and Maintenance, John Wiley &
Sons: Hoboken, N.J.
Bovaird, RL, Zagor, HL (2006), Lognormal distribution and maintainability in support
systems research, Naval Research Logistics Quarterly, 8(4), 343--
-356.
Brown, RJ (1979), A new marketing tool: life-cycle costing, Industrial Marketing
Management, 8(2), 109-113.
Cheevaprawatdomrong, T, Smith, RL (2003), A paradox in equipment replacement under
technological improvement, Operations Research Letters, 31(1), 77-82.
Christer, AH, Scarf, PA (1994), A robust replacement model with applications to medical
equipment, Journal of the Operational Research Society, 45, 261--
-275.
Coe, CK (1981), Life cycle costing by state governments, Public Administration Review, 564-
569.
Condra, L (2001), Reliability Improvement with Design of Experiment, Marcel Dekker:
New York.
Cooper, DR, Schindler, PS (2006), Business Research Methods, 9th ed., McGraw-Hill Irwin:
New York.
Council of Science Editors (2006), Scientific Style and Format: the CSE Manual for Authors,
Editors, and Publishers, Rockefeller Univ. Press: Virginia.
Curran-Everett, D, Benos, DJ (2004), Guidelines for reporting statistics in journals published
by the American Physiological Society, American Journal of Physiology ---
Endocrinology and Metabolism, 287(2), 189---191.
47

Dandotiya, R (2012), Decision Support Models for the Maintenance and Design of Mill
Liners, PhD thesis, Lulea University of Technology: Luleå, Sweden.
Dandotiya, R, Lundberg, J (2012), Economic model for maintenance decision: a case
study for mill liners, Journal of Quality in Maintenance Engineering, 18(1), 79-97.
Dane, FC (1990), Research Methods, Brooks/Cole Publishing Company: Pacific Grove, Calif.
Das, P (2005), Improving the maintenance system of tippers: a case study, Economic Quality
Control, 20, 51-- -67.
Dhillon, BS (1999), Design Reliability: Fundamentals and Applications, CRC Press: New
York.
Dhillon, BS (2010), Life Cycle Costing for Engineers, CRC Press: New York.
Dhillon, BS (2013), Life Cycle Costing: Techniques, Models and Applications, Routledge.
Earles, ME (1981), Factors, Formulas and Structures for Life Cycle Costing, Eddins-Earles,
Concord, Mass.
Edelbro, C (2008), Strength, Fallouts and Numerical Modelling of Hard Rock Masses,
Doctoral thesis, Luleå University of Technology: Luleå, Sweden.
Edelbro, C (2009), Numerical modelling of observed fallouts in hard rock masses using an
instantaneous cohesion-softening friction-hardening model, Tunnelling and
Underground Space Technology, 24(4), 398-409.
Eilon, S, King, JR, Hutchinson, DE (1966), A study in equipment replacement, Journal of the
Operational Research Society, 17(1), 59-- -71.
Elton, EJ, Gruber, MJ (1976), On the optimality of an equal life policy for equipment
subject to technological improvement, Operational Research Quarterly, 27(1), 93-99.
Eschenbach, T (2010), Engineering Economy: Applying Theory to Practice, 3rd ed., Oxford
University Press: New York.
Farr, JV (2011), Systems Life Cycle Costing: Economic Analysis, Estimation, and
Management, CRC Press: New York.
Gardner, MJ, Altman, DG (2000), Estimating with confidence. In: Altman, DG, Machin, D,
Bryant, TN, Gardner, MJ (eds.), Statistics with Confidence, BMJ Books: Bristol.
Garson, GD (1991), Interpreting neural-network connection weights, AI Expert, 6(4), 47-51.
Ghosh, S, Majumdar, SK (2011), Reliability modeling and prediction using classical and
Bayesian approach: a case study, International Journal of Quality Reliability
Management, 28(5), 556-586.
Goh, ATC (1995), Back-propagation neural networks for modeling complex systems,
Artificial Intelligence in Engineering, 9(3), 143-151.
Greene, LE, Shaw, BL (1990), The steps for successful life cycle cost analysis.
In: Proceedings of the IEEE 1990 National Aerospace and Electronics Conference,
NAECON 1990, IEEE: New York, pp. 1209-1216.
Gustafson, A, Schunnesson, H, Galar, D, Kumar, U (2012), The influence of the operating
environment on manual and automated load-haul-dump machine: a fault tree analysis,
International Journal of Mining, Reclamation and Environment,
DOI:10.1080/1755182X.2011.651371.
Hartman, HM (1987), Introductory Mining Engineering, 1st ed., John Wiley and Sons: New
York.
Hartman, HM, Mutmansky, JM (2002), Introductory Mining Engineering, 2nd ed., John
Wiley and Sons: Hoboken, N.J.
Hartman, JC (2005), A note on a strategy for optimal equipment replacement, Production
Planning & Control, 16(7), 733-739.
Hartman, JC, Murphy, A (2006), Finite-horizon equipment replacement analysis, IIE
Transactions, 38(5), 409-419.
Hartman, JC, Rogers, JL (2006), Dynamic programming approaches for equipment
48

replacement problems with continuous and discontinuous technological change, IMA
Journal of Management Mathematics, 17(2), 143-158.
Hartman, JC, Tan, CH (2014), Equipment replacement analysis: a literature review and
directions for future research, Engineering Economist, 59, 136-- -153.
Hennessy, C, Freerks, F, Campbell, JE, Thompson, BM (2000), Automated analysis of failure
event data. In: Proceedings of the Modeling and Analysis of Semiconductor
Manufacturing Conference, March 27, Tempe, Arizona, USA.
Hoseinie, SH, Ataei, M, Khalokakaie, R, Ghodrati, B, Kumar, U (2012), Reliability analysis
of drum shearer machine at mechanized longwall mines, Journal of Quality in
Hoseinie, SH, Khalokakaie, R, Ataie, M, Ghodrati, B, Kumar, U (2013), Monte Carlo
reliability simulation of coal shearer machine, International Journal of Performability
Engineering, 9(5), 487- 494.
Hritonenko, N, Yatsenko, Y (2008), The dynamics of asset lifetime under technological
change, Operations Research Letters, 36(5), 565-568.
Kamat, SJ, Riley, MW (1975), Determination of reliability using event-based Monte Carlo
simulation, IEEE Transactions on Reliability, 24(1), 73-75.
Kapur, KC (1986), Quality evaluation systems for reliability, Reliability Review, 6(2), 19-20.
Kapur, KC, Pecht, M (2014), Reliability Engineering, John Wiley & Sons: Hoboken,
N.J.
Katukoori, VK (1995), Standardizing Availability Definition. University of New Orleans:
New Orleans, La. Available at:
http://www.plant-maintenance.com/articles/Availability_Definition.pdf.
Keles, P, Hartman, JC (2004), Case study: bus fleet replacement, Engineering
Economist, 49(3), 253-- -278.
Klefsjö, B, Kumar, U (1992), Goodness-of-fit tests for the power-law process based on the
TTT-plot, IEEE Transactions on Reliability, 41(4), 593-- -598.
Knights, PF (2001), Rethinking Pareto analysis: maintenance applications of logarithmic
scatterplots, Journal of Quality in Maintenance Engineering, 7(4), 252-263.
Kortelainen, H, Kupila, K, Silenius, S, Päivike, A (2003), Data for better maintenance plans
and investments policy, TAPPI Journal, 2, 8-- -12.
Krauland, N, Marklund, P, Board, M (2001), Rock support in cut-and-fill mining at the
Kristineberg Mine. In: Hustrulid, WA, Bullock, RL (eds.), Underground Mining
Methods: Engineering Fundamentals and International Case Studies, Society for
Mining, Metallurgy and Exploration: Littleton, Colo.
Kumar, D, Klefsjö, B, Kumar, U (1992), Reliability analysis of power transmission cables
of electric mine loaders using the proportional hazards model, Reliability Engineering
& System Safety, 37(3), 217-222.
Kumar, U (1989), Maintenance Strategy for Mechanized Mining Systems: a Reliability
Analysis Based Approach, Licentiate Thesis, Luleå University of Technology: Luleå,
Sweden.
Kumar, U, Akersten, PA (2008), Availability and maintainability. In: Melnik, EL and Everitt,
BS (eds.), Encyclopedia of Quantitative Risk Analysis and Assessment, John Wiley &
Sons: Chichester, pp. 77-84.
Kumar, U, Klefsjö, B (1992), Reliability analysis of hydraulic systems of LHD machines
using the power law process model, Reliability Engineering & System Safety, 35, 217-- -
224.
Kumar, U, Klefsjö, B, Granholm, S (1989), Reliability investigation for a fleet of load haul
dump machines in a Swedish mine, Reliability Engineering & System Safety, 26(4),
341-361.
49

Lin, T, Titmuss, D (1995), Critical component reliability and preventive maintenance
improvement to reduce machine downtime, Computers & Industrial
Engineering, 29(1), 21-23.
Lithander, E (2004), Sweden. In: World of Information, Europe Review 2003-4: the
Economic and Business Report, 5th ed., Wolden Publishing: Saffron Walden.
Louit, DM, Pascual, R, Jardine, AKS (2009), A practical procedure for the selection of time-
to-failure models based on the assessment of trends in maintenance data, Reliability
Engineering & System Safety, 94(10), 1618-1628.
Luderer, B, Nollau, V, Vetters, K (2010), Mathematical Formulas for Economists, 4th ed.,
Springer: Heidelberg, Dordrecht, New York.
Masters, B, Lewis, T, Kolarik, W (1992), A confidence interval for the availability ratio for
systems with Weibull operating time and lognormal repair time, Microelectronics and
Reliability, 32(1-2), 89-99.
Modarres, M (2006), Risk Analysis in Engineering: Techniques, Tools, and Trends, Taylor &
Francis Group: Boca Raton, Fla.
Neuman, WL (2003), Social Research Methods: Qualitative and Quantitative Approaches,
5th ed., Pearson Education, Allyn and Bacon: New York.
Norton, RL (2011), Machine Design: an Integrated Approach, 4th ed., Pearson: Upper Saddle
River, N.J.
Oakford, RV, Lohmann, JR, Salazar, A (1984), A dynamic replacement economy decision
model, IIE Transactions, 16 (1), 65---72.
O'Connor, PD (2000), Commentary: reliability -- - past, present, and future, IEEE
Transactions on Reliability, 49(4), 335-341.
Rådberg, G, Nordlund, E, Nyström, A (1992), Case study on the determination of input data
for numerical modeling at the Kristineberg mine. In: Hudson, JA (ed.), Rock
Characterization, British Geotechnical Society: London.
Rausand, M, Høyland, A (2004), System Reliability Theory: Models, Statistical Methods, and
Applications, John Wiley and Sons: Hoboken, N.J.
Richardson, S, Kefford, A, Hodkiewicz, M (2013), Optimised asset replacement strategy in
the presence of lead time uncertainty, International Journal of Production
Economics, 141, 659-667.
Ryan, W (1968), Procurement views of life cycle costing. In: Proceedings of the 1968 Annual
Symposium on Reliability, Boston, Massachusetts, January 16-18, pp. 164-168.
Scarf, PA, Bouamra, O (1999), Capital equipment replacement model for a fleet with
variable size, Journal of Quality in Maintenance Engineering, 5(1), 40-49.
Scarf, P, Dwight, R, McCusker, A, Chan, A (2006), Asset replacement for an urban railway
using a modified two-cycle replacement model, Journal of the Operational Research
Society, 58(9), 1123---1137.
Schroeder, B, Gibson, G (2005), A Large-scale Study of Failures in High-performance-
computing Systems, Technical Report (CMU-PDL-05-112), Carnegie Mellon
University: Pittsburgh, Pa.
Seldon, MR (1979), Life Cycle Costing: a Better Method of Government Procurement,
Westview Press: Boulder, Colo.
Shishko, R, Aster, R (1995), NASA Systems Engineering Handbook, NASA Special
Publication, 6105, NASA: Washington, D.C.
Stapelberg, RF (2009), Handbook of Reliability, Availability, Maintainability and Safety in
Engineering Design, Springer: London.
Stracener, JT (1997), Mathematical basis of reliability. In: Cruse, TA (ed.), Reliability-Based
Mechanical Design, Marcel Dekker: New York, pp. 91-122.
Sullivan, TJ (2001), Methods of Social Research, Harcourt College Publisher: Fort Worth, Tex.
50

Sumser, JR (2000), A Guide to Empirical Research in Communication, Sage Publications,
Thousand Oaks, Calif.
Swedish Trade Council (2012), Internationalisering och Export som Viktigt Inslag i
Regeringens Mineralstrategi. Näringsdepartementet: Stockholm. Available at:
http://www.regeringen.se/content/1/c6/19/82/82/89102eae.pdf.
Tanchoco, JMA, Leung, LC (1987), An input-output model for equipment replacement
decisions, Engineering Costs and Production Economics, 11(2), 69-78.
Vagenas, N, Nuziale, T (2001), Genetic algorithms for reliability assessment of mining
equipment, Journal of Quality in Maintenance Engineering, 7(4), 302-311.
Vayenas, N, Xiangxi, W (2009), Maintenance and reliability analysis of a fleet of
load-haul-dump vehicles in an underground hard rock mine, International Journal of
Mining, Reclamation and Environment, 23(3), 227-238.
Verheyen, PA (1979), Economic interpretation of models for the replacement of
machines, European Journal of Operational Research, 3, 150-156.
Waddell, R (1983), A model for equipment replacement decisions and policies, Interfaces,
13(4), 1---7.
Wagner, HM (1975), Principles of Operations Research, 2nd ed., Prentice-Hall:
Englewood Cliffs, N.J.
Wijaya, AR (2012), Methods for Availability Improvements of a Scaling Machine System,
Doctoral thesis, Luleå University of Technology: Luleå, Sweden.
Wijaya, AR, Lundberg, J, Kumar, U (2011), Downtime analysis of a scaling machine,
International Journal of Mining, Reclamation and Environment,
DOI:10.1080/17480930.2011.603515.
Wikoff, D (2008), Root cause analysis. In: Mobley, RK, Higgins, LR, Wikoff, D (eds.),
Maintenance Engineering Handbook, 7th ed., McGraw-Hill: New York.
Wiseman, M (2001), Reliability management and maintenance optimization: basic statistics
and economics. In: Campbell, JD, Jardine, AKS (eds.), Maintenance Excellence:
Optimizing Equipment Life-Cycle Decisions, Marcel Dekker: New York.
Wynholds, HW, Skratt, JP (1977), Weapon system parametric life cycle cost analysis.
In: Annual Reliability and Maintainability Symposium, Philadelphia, Pa, pp. 303-309.
Yatsenko, Y, Hritonenko, N (2005), Optimization of the lifetime of capital equipment using
integral models, Journal of Industrial and Management Optimization, 1(4), 415-432.
Yin, RK (1994), Case Study Research: Design and Methods, 2nd ed., Sage Publications:
Thousand Oaks, Calif.
51

APPENDED PAPERS

PAPER I
Downtime analysis of drilling machines and suggestions for

improvements
Al-Chalabi, H., Lundberg, J., Wijaya, A. and Ghodrati, B., 2014. Downtime analysis of
drilling machines and suggestions for improvements. Published in the Journal of Quality in
http://dx.doi.org/10.1108/JQME-11-2012-0038

The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1355-2511.htm
JQME
20,4
Downtime analysis of drilling
machines and suggestions for
improvements
306 Hussan S. Al-Chalabi
Division of Operation, Maintenance and Acoustics, Luleå University of Technology,
Received 20 November 2012
Revised 23 October 2013 Luleå, Sweden and Mechanical Engineering Department, College of Engineering,
Accepted 11 January 2014 Mosul University, Mosul, Iraq, and
Jan Lundberg, Andi Wijaya and Behzad Ghodrati
Division of Operation, Maintenance and Acoustics,
Luleå University of Technology, Luleå, Sweden
Abstract
Purpose – The purpose of this paper is to analyse and compare the downtime of four drilling
machines used in two underground mines in Sweden. The downtime of these machines was compared
to show what problems affect downtime and which strategies should be applied to reduce it.
Design/methodology/approach – The study collects failure data from a two-year period for four
drilling machines and performs reliability analysis. It also performs downtime analysis utilising a log-
log diagram with a confidence interval.
Findings – There are notable differences in the downtime of most of the studied components for all
machines. The hoses and feeder have relatively high downtime. Depending on their downtime, the
significant components can be ranked in three groups. The downtime of the studied components is due
to reliability problems. The study suggests the need to improve the reliability of critical components to
reduce the downtime of drilling machines.
Originality/value – The method of analysing the downtime, identifying dominant factors and the
interval estimation for the downtime, has never been studied on drilling machines. The research
proposed in this paper provides a general method to link downtime analysis with potential component
improvement. To increase the statistical accuracy; four case studies was performed in two different
mines with completely different working environment and ore properties. Using the above method
showed which components need to be improved and suggestions for improvement was proposed and
will be implemented accordingly.
Keywords Reliability analysis, Confidence interval, Downtime analysis, Drilling machine
Paper type Research paper
1. Introduction
Underground mines are a main source of minerals. Growing demand for metals as a
result of modern lifestyles and the industrial development of recent decades has
r Hussan S. Al-Chalabi, Jan Lundberg, Andi Wijaya and Behzad Ghodrati. Boliden AB is
gratefully acknowledged for their financial support. Published by Emerald Group Publishing
Limited. This paper is published under the Creative Commons Attribution (CC BY 3.0) licence.
Anyone may reproduce, distribute, translate and create derivative works of this paper (for both
commercial and non-commercial purposes), subject to full attribution to the original publication
Journal of Quality in Maintenance and authors. The full terms of this licence may be seen at http://creativecommons.org/licences/
Engineering
Vol. 20 No. 4, 2014 by/3.0/legalcode
pp. 306-332 Boliden AB is gratefully acknowledged for their financial support. The maintenance
Emerald Group Publishing Limited
1355-2511 personnel and maintenance management system employees at Boliden AB who helped the
DOI 10.1108/JQME-11-2012-0038 authors in this research are also thanked for their support.
focused our attention on the factors affecting the extraction of minerals. One of Downtime
the most important factors is unscheduled downtime of the machines used in the analysis of
extraction of ore. Lost production due to downtime will obviously increase the
production costs (Roman and Daneshmend, 2000). A mine production system consists of drilling machines
many subsystems. To make the system both profitable and practical, the optimisation of
each subsystem in relation with other subsystems should be considered (Barabady and
Kumar, 2008). To achieve this aim, reliability and maintainability analysis for each 307
subsystem in mine production system should be performed. Since the mid-1980s,
reliability analysis techniques have been essential tools in automatic mining systems
(Blischke and Murthy, 2003; Barabady and Kumar, 2008).
A drilling machine is very important to the extraction process. Drilling is the
process of making holes in the mining room face. From an economic viewpoint, the
drilling machine dominates a mine’s production rate, since drilling is the first process
of a typical mining cycle. Economic competition has pushed mining companies to
achieve higher production rates by enhancing techniques of drilling and blasting and
increasing mechanisation and automation. A significant cost issue is the maintenance
of underground mobile machines: 30-65 per cent of a mine’s total operation costs
typically come from maintenance (Cutifani et al., 1996; Gustafson et al., 2013). Maintenance
costs include the cost of planned and unplanned maintenance.
Historical data over the period of one year from an underground mine in Sweden
show that more than 15 per cent of unplanned downtime of mobile machines is related
to the drilling machine. Since the drilling machine is key to production, it is important
to find solutions for machine problems and reduce downtime.
This study performs downtime analysis of drilling machine to identify which
components and what type of problems (maintainability problems and/or reliability
problems) contribute to downtime, and to determine which strategies, designs for
maintainability and/or designs for reliability should be applied to reduce it. To better
understand the downtime of the drilling machine, we conducted an analysis of the
historical data for three machines of the same model used in one Swedish mine and for
one machine used in another mine, using jack-knife diagrams with confidence
intervals. The results of a study of downtime with or without confidence intervals for
only one machine used in only one mine may not give a clear picture of the overall
behaviour of the machine’s components. Thus, it will prevent the designers to give
good suggestions to improve the reliability and/or maintainability of the machine. To
overcome this shortcoming, a comparison study and an analysis of the downtime by
using jack-knife diagrams with confidence intervals for one machine used in mine X
and three machines same model used in mine Y was conducted during this study. The
reason of using data for one machine, which used in mine X, is that this mine has only
one machine of the same model of the machines that used in mine Y. This diagram was
used in order to overcome the shortcomings of using Pareto diagrams in maintenance
engineering applications (Knights, 2001).
A jack-knife diagram has one shortcoming; it presents downtime as a single value
(point estimated), and because of involved uncertainties it has been considered insufficient
(Altman et al., 2000; Curran-Everett and Benos, 2004; Wijaya et al., 2012). System
designers and users have a tendency to be risk-averse regarding downtime. They prefer a
design with a slightly higher estimated downtime (lower reliability) if the estimated value
is known to be accurate (as characterised by the upper limit of a confidence interval of the
downtime) rather than a design with possible inaccuracies in point estimation; therefore,
it is important to consider a confidence interval for system downtime (Colt, 1997).
JQME 1.1 Literature review
20,4 Many researchers studied the reliability and maintainability of mining equipment
and its failure behaviour. For example, Kumar et al. (1989) analysed the operational
reliability of a fleet of diesel operated load-haul-dump (LHD) machines in Kiruna
mine in Sweden. Kumar et al. (1992) performed reliability analysis on the power
transmission cables of electric mine loaders in Sweden. Kumar and Klefsjö (1992)
308 analysed the maintenance data of one subsystem (hydraulic system) of a fleet of six
LHD machines divided into three independent groups at Kiruna mine.
Reliability assessment of mining equipment was performed by Vagenas and Nuziale
(2001); using genetic algorithms, they developed and tested mobile mining equipment
reliability assessment models. Vayenas and Xiangxi (2009) studied the availability
of 13 LHD machines in an underground mine. They were interested in the influence of
machine downtime on productivity and operation costs and used a reliability-based
approach and a basic maintenance approach to determine the machine’s availability.
Wijaya et al. (2012) developed a method for visualising downtime by using a jack-knife
diagram; they applied the method on a scaling machine at a mine in Sweden. Gustafson
et al. (2013) used a fault tree analysis to analyse the idle times of automated LHD
machines at a Swedish underground mine. Hoseinie et al. (2012) performed reliability
modelling of the drum Shearer machine used at Taba’s coal mine in the central desert of
Iran. They analysed the failure rate of the machine’s subsystems.
As the literature review shows, there are many reliability and maintainability
studies of underground mining equipment but none of these has looked at drilling
machines. Given the importance of underground mining mobile equipment for production,
as well as the complexity of the equipment and the harsh mining environment, reliability
analysis of the drilling machine must meet rigorous requirements. This study is based
on data from several drilling machines working in different mines. In these mines the
working environment, ore properties and operators are different.
The aims of this study are as follows:
(1) to analyse the reliability and downtime of several drilling machines to
determine what kind of problems affect their downtime;
(2) to specify which strategies, design for maintainability and/or design for reliability
(DFR) should be applied to reduce the drilling machines downtime; and
(3) to suggest improvement for the components that most contribute to the
machines downtime.
2. Drilling machine and data collection

All drilling machines for mining applications are composed of similar operational
design units, such as cabin, boom, rock drill, feeder, service platform, front jacks,
hydraulic pump, rear jack, electric cabinet, hose reeling unit, cable reeling unit, diesel
engine, hydraulic oil reservoir, operator panel and water tank. A typical example of
a drilling machine and its components are presented in Figure 1.
Drilling machines manufactured by different companies have different technical
characteristics, e.g. capacity and power. Based on the operation manuals, field
observations and maintenance reports from the collaborating mines, in this study the
drilling machine is considered a system divided into several components and
connected in series configuration; if any component fails, the operator will stop the
machine to maintain it. Thus, all machine components work simultaneously to achieve
14 3 Downtime
2
analysis of
1
drilling machines
4
12 13
10 11
309
5
9 6
8 7
1 Cabin 6 Front jacks 11 Cable reeling unit
2 Boom 7 Hydraulic pump 12 Diesel engine Figure 1.
3 Rock drill 8 Rear Jack 13 Hydraulic oil reservoir A typical example of
4 Feeder 9 Electric cabinet 14 Operator panel a drilling machine
5 Service platform 10 Hose reeling unit
the desired function. Table I shows the critical components for each machine mentioned
in this study.
The dimensions of the drilling machine are: length 14.5-16.6 m; width 2.5 m; width of
rig with side platforms 2.9 m; height of rig with cabin 3.15 m; weight 26-33 tonnes.
It has four retractable stabilizer legs and an articulated four-wheel drive chassis. It can
be operated by a water cooled turbocharged diesel engine with 120 kW at 2,300 rpm or
by electric power with a capacity of 158 kW.
3. Reliability and downtime analysis

The failure data used in this study were collected over two years for four drilling
machines operated in two different underground mines in Sweden. The Maximo
computerised maintenance management system (CMMS) is the main source of the failure
data. In CMMS, the failure data are recorded based on calendar time. Since drilling is
not a continuous process, the time between failures is estimated by considering the
utilisation of each machine. In this study, we test and validate the failure and repair data
after collection. We test for trends using the Laplace trend test; we also test for serial
correlation (Ansell and Phillips, 1994). When these tests are used, depending on the
results, classical statistical techniques for reliability modelling may be appropriate
(Ascher and Feingold, 1984; Kumar and Klefsjö, 1992; Modarres, 2006; Birolini, 2007;
Louit et al., 2009; Ghosh and Majumdar, 2011). The Kolmogorov-Smirnov (K-S) test is
Mine X Mine Y
Machine 1 Machine 2 Machine 3 Machine 4
Component Sym. Component Sym. Component Sym. Component Sym.
Rock drill A1 Rock drill A2 Rock drill A3 Rock drill A4

Feeder B1 Feeder B2 Feeder B3 Feeder B4
Hoses C1 Hoses C2 Hoses C3 Hoses C4
Accumulators D2 Accumulators D3 Accumulators D4
Boom E2 Boom E3 Boom E4
Cables F2 Cables F3 Cables F4 Table I.
Steering system G1 Steering system G3 Steering system G4 Critical components
Cylinders H1 Cylinders H3 Cylinders H4 for each machine
JQME classically used for the selection and validation of the probability distribution models (for
20,4 further information, refer to Louit et al., 2009). In this study, we conduct all component
analysis based on the black box approach.
To estimate the interval of downtimes, we assume that the failure times have
a Weibull distribution and the repair times have a lognormal distribution. In addition
to its flexibility, the Weibull distribution gives a reasonably accurate failure analysis
310 even with a small sample size (Masters et al., 1992; Abernethy, 2000). The shape and
scale parameters of the Weibull distribution are determined by using the maximum
likelihood estimation method. Lognormal distribution is generally used to model repair
times (Rausand and Hoyland, 2004; Schroeder and Gibson, 2010). The long tail to the
right of lognormal distribution provides the best fitting representation of the repair
situation. Most repairs are accomplished in a small period of time, but in certain
cases, repairs can take a much longer time (Wijaya et al., 2012). The Weibull
distribution has a probability density function given by:
" #
b y b1 y b
g ð yÞ ¼ exp ; y40 ð1Þ
Z Z Z
where b and Z are the shape and the scale parameters of the Weibull distribution,
respectively. The expected mean time between failures (MTBF) is given as:

1
MTBF ¼ my ¼ ZG þ1 ð2Þ
b
Lognormal distribution has a probability density function given by:

" #
1 1 lnðxÞ m 2
f ðxÞ ¼ pffiffiffiffiffi exp ; x40 ð3Þ
xs 2p 2 s
where s and m are the standard deviation and the mean of the variable’s natural
logarithm, respectively. The expected mean time to repair (MTTR) is given as:

s2
MTTR ¼ mx ¼ exp m þ ð4Þ
2
The confidence interval of the downtime is determined according to the following

equations:
DT
MTTR ¼ ð5Þ
m
UT
MTBF ¼ ð6Þ
m
DT MTTR
¼ ð7Þ
UT MTBF
where DT is the downtime, m is the number of failures and UT is the uptime. Thus, the Downtime
downtime can be formulated as (Wijaya et al., 2012): analysis of
mx drilling machines
DT ¼ UT ð8Þ
my
The confidence interval of the estimated downtime can be solved by finding

311
a confidence interval for mx/m y. In this study, the exact method (Masters et al., 1992) is
implemented to estimate the confidence interval of mx/m y (for more clarification of the
method to estimate the interval of the downtime, see Wijaya et al., 2012). All tests are
conducted using the Minitab, Matlab and Easy Fit software; the significance level
a used in all tests is 0.05.
4. Methodology
A simple yet important graphical method to visualise downtime is the jack-knife
diagram (Knights, 2001). In this diagram, the failure data are presented as a log-log
graph. The graph shows log number of failures (vertical axis) and log repair time
(horizontal axis). The curves of constant downtime appear as straight lines with
a uniform and constant gradient (Knights, 2001). This study uses the jack-knife
diagram with the downtime confidence interval to analyse the downtime of the
components of the drilling machine. Three equations are used to establish the
confidence log-log plot. The estimation points are the mean, lower limit and upper
limit of the downtime. The value of these points can be estimated from the following
equations (Wijaya et al., 2012):
2 3
exp m þ s2
2
DTM ¼4 5 UT ð9Þ
ZG b1 þ 1
2 2 3
exp s2 G
DTLL ¼ 4a P
5 UT ð10Þ
b
G b1 þ 1 2Z1b m y
i¼1 i
2 2 3
exp s2 G
DTUL ¼ 4b P
5 UT ð11Þ
b
G b1 þ 1 2Z1b m y
i¼1 i
where DTM is mean downtime, DTLL is a lower limit of the downtime, a is a constant
for the lower limit, G is a geometric mean of the lognormal distribution, y is time
between failures, DTUL is an upper limit of the downtime and b is a constant for the
upper limit.
JQME The next step is the determination of coordinates of the three previous estimated
20,4 points. For the DTM, the abscissa is the MTTR, and the ordinate is determined as:
UT
mM ¼ ð12Þ
MTBF
312 For the lower and upper limit of the downtime, the abscissa and the ordinate are
determined as:
pffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffi
log DTL log DTM þlogðMTTRÞ
TTRL ¼ 10 ð13Þ
DTL
mL ¼ ð14Þ
TTRL
where TTRL is time to repair of the limit (upper or lower); DTL is the downtime of the
limit (upper or lower); MTTR is mean time to repair; mL is the number of failures of the
limit (upper or lower). For more information on derivation of the abscissa, the lower
and upper limits of downtime refer to Wijaya et al. (2012).
To estimate the interval of the downtime, the three estimated points are connected
by a straight line (Figure 2).
5. Results and discussion

The present study only considers corrective maintenance. It assumes that the
failure times follow a Weibull distribution and the repair times follow a lognormal
distribution; therefore, the iid assumption is validated before analysis. The maximum
likelihood estimation method is used to estimate the corresponding parameters
using Minitab software. K-S test is used to validate the distributions using Easy
Fit software.
90
70
60
mUL DTUL
mM DTM
0
00
mLL DTLL
1,
No. of failures
10
0
20
e
0
im
10
nt
5
ow
40
4
3
20
10
Figure 2. 1
Log-log plot of downtime 1 TTRLL TTRUL 7 10 20 30 50 80 130
confidence interval
MTTR Time to repair
Parameters of the distributions are determined the lower bound, mean and upper Downtime
bound for a confidence interval of 95 per cent. For the failure data, blow and bupp are analysis of
the estimates of the lower and upper limits, respectively, of the maximum likelihood
estimate of shape parameter b, denoted by best. Zlow and Zupp are the estimates of the drilling machines
lower and upper limits, respectively, of the maximum likelihood estimate of scale
parameter Z, denoted by Zest. MTBFlow and MTBFupp are the estimates of the lower
and upper limits, respectively, of the maximum likelihood estimate of the MTBF, 313
denoted by MTBFest. For the repair data, mlow and mupp are the estimates of the
lower and upper limits, respectively, of the maximum likelihood estimate of the mean
parameter m of the lognormal distribution, denoted by mest. slow and supp are the
estimates of the lower and upper limits, respectively, of the maximum likelihood
estimate of the standard deviation parameter s of the lognormal distribution, denoted
by sest. MTTRlow and MTTRupp are the estimates of the lower and upper limits,
respectively, of the maximum likelihood estimate of the MTTR, denoted by MTTRest.
Tables II-V show the data analysis of the critical components of four drilling
machines working in two different mines in Sweden.
We compare the machines’ downtime using a jack-knife diagram with a confidence
interval. Using Equations (9-11), we determine three downtime estimation points,
DTM, DTLL and DTUL. The study uses theoretical production hours for one year as
uptime. When the data were collected for this study, each mine averaged about 16.5
working hours per day, or approximately 115 hours per week. By fitting lognormal
distribution for preventive maintenance (PM) data (service data) to all machines, we
find that mean time to service averages 6.7 hours. Since PM is scheduled for every
week, the theoretical service time for 49 working weeks for one year is calculated as
approximately 330 hours. Consequently, the stoppage for PM is three weeks per year;
hence, each mine is worked for 46 weeks per year. We can conclude that each mine has
approximately 5,300 production hours per year. The coordinates of the three downtime
estimation points are determined by using Equations (4), (12-14). Tables VI-IX shows
the values of the three estimation points and their coordinates.
We make four types of comparisons of the downtime of the machines’ components.
Type 1 is a comparison of six components on three machines used only in mine Y. Type
2 is a comparison of three components on four machines (1, 2, 3 and 4) used in both
mines. Type 3 is a comparison of two components on three machines (1, 3 and 4) used
in both mines. Type 4 compares one component on two machines (1 and 2) used in both
mines. Figures 3-9 are visualisations of these four types of comparisons.
Since the number of failures and repair time are noted manually, we expect these
data have some errors. Based on interviews with several maintenance persons
who work in the mines collaborating in this study, we conclude that 80 per cent of the
data does not include documentation errors. With this in mind, we define a ratio R,
whereby R ¼ maximum mean downtime/minimum mean downtime, and determine
the following limits:
(1) no significant differences; 1pRp1.2;
(2) indicated differences; 1.2oRp1.4; and
(3) significant differences; R41.4.
Figure 3 shows for a given uptime of 5,300 hours, there is a significant difference
between E2, E3 and E4 (boom), as R ¼ 3.5. Also there is a significant difference
between A2, A3 and A4 (rock drill), as R ¼ 1.7. But there is no significant
20,4
314
JQME
Table II.
Summary for
machine 1, mine X
Time to failure (95% normal CI)
Weibull parameters
Component m blow best bupp Zlow Zest Zupp MTBFlow MTBFest MTBFupp
(a) Failure dataa
A1 Rock drill 21 0.68 0.94 1.3 62.8 102.6 167.6 66.2 105.1 166.8
B1 Feeder 7 0.76 1.47 2.85 186.5 330.8 586.7 172.4 299.3 519.6
C1 Hoses 61 0.68 0.82 0.99 23.9 33.3 46.4 27.1 37.0 50.6
G1 Steering system 26 0.71 0.95 1.29 54.5 84.0 129.5 56.9 85.6 128.8
H1 Cylinders 5 1.48 3.78 9.62 262.8 343.3 448.4 233.4 310.2 412.2
I1 Hydraulics 5 0.74 1.55 3.23 186.0 363.5 710.5 170.9 326.8 624.8
J1 Fuel system 5 0.61 1.29 2.71 138.4 309.7 692.8 132.9 286.3 616.5
Time to repair (95% normal CI)
Lognormal parameters
Component m slow sest supp mlow mest mupp MTTRlow MTTRest MTTRupp
(b) Repair datab
A1 Rock drill 21 0.42 0.57 0.77 0.09 0.33 0.58 1.2 1.6 2.1
B1 Feeder 7 0.44 0.74 1.2 0.17 0.72 1.27 1.4 2.7 5.0
C1 Hoses 61 0.41 0.49 0.59 0.13 0.25 0.38 1.2 1.4 1.6
H1 Cylinders 5 0.19 0.36 0.67 0.20 0.52 0.83 1.2 1.7 2.4
I1 Hydraulics 5 0.38 0.72 1.34 0.55 0.08 0.71 0.6 1.4 2.8
J1 Fuel system 5 0.14 0.27 0.51 0.10 0.13 0.38 0.9 1.1 1.5
Notes: m is the number of failures. aZ and MTBF are given in units of hours; bMTTR is given in units of hours
Weibull parameters
(a) Failure dataa
A2 Rock drill 39 0.77 0.97 1.24 47.7 69.0 99.9 49.1 69.7 98.7
B2 Feeder 37 0.65 0.83 1.07 36.0 55.4 85.4 40.5 60.9 91.6
C2 Hoses 124 0.80 0.92 1.06 16.7 20.7 25.6 17.6 21.5 26.3
D2 Accumulators 10 0.70 1.22 2.13 156.1 272.7 476.3 149.6 254.9 434.4
E2 Boom 19 0.70 1.04 1.54 87.5 146.2 244.5 88.5 143.8 233.8
F2 Cables 7 0.79 1.50 2.83 298.1 555.1 1033.8 275.0 501.0 912.7
I2 Hydraulics 5 0.73 1.51 3.12 143.8 286.6 571.1 132.9 258.4 502.6
J2 Control panel 19 0.86 1.30 1.95 87.5 131.8 198.7 82.3 121.7 180.1
K2 Water cooler 8 0.45 0.84 1.54 121.6 332.2 907.6 140.4 363.1 938.6
L2 Valvesa 23 0.67 0.93 1.3 67.9 109.9 178.0 71.7 113.2 178.5
M2 Manual valves 16 0.78 1.18 1.79 113.5 180.9 288.1 109.5 170.5 265.4
N2 Movement device 6 0.61 1.28 2.65 211.2 434.6 893.9 202.3 402.6 801.2
(b) Repair datab
A2 Rock drill 39 0.41 0.52 0.65 0.27 0.44 0.60 1.5 1.7 2.1
B2 Feeder 37 0.46 0.58 0.73 0.72 0.91 1.10 2.4 2.9 3.6
C2 Hoses 124 0.56 0.64 0.72 0.48 0.59 0.70 1.9 2.2 2.5
D2 Accumulators 10 0.29 0.45 0.71 0.13 0.15 0.43 0.9 1.2 1.7
E2 Boom 19 0.51 0.71 0.97 0.75 1.07 1.39 2.6 3.7 5.4
F2 Cables 7 0.36 0.62 1.05 0.24 0.70 1.17 1.4 2.4 4.0
I2 Hydraulics 5 0.23 0.43 0.81 0.16 0.21 0.60 0.9 1.3 2.0
J2 Control panel 19 0.37 0.51 0.71 0.10 0.34 0.57 1.2 1.6 2.0
K2 Water cooler 8 0.52 0.85 1.39 0.007 0.58 1.17 1.2 2.5 5.1
L2 Valves 23 0.51 0.68 0.91 0.17 0.45 0.73 1.4 1.9 2.7
M2 Manual valves 16 0.42 0.60 0.85 0.07 0.36 0.66 1.2 1.7 2.4
N2 Movement device 6 0.37 0.66 1.17 0.36 0.89 1.42 1.6 3.0 5.5
drilling machines
Downtime
machine 2, mine Y
315
Table III.
analysis of
Summary for
20,4
316
mine Y
JQME
Table IV.
Summary for machine 3,
Weibull parameters
(a) Failure dataa
A3 Rock drill 38 0.62 0.78 0.99 32.1 50.0 77.7 37.8 57.3 87.0
B3 Feeder 54 0.66 0.81 1.01 29.7 42.4 60.7 33.6 47.3 66.4
C3 Hoses 45 0.83 1.05 1.33 43.2 58.1 78.0 42.9 56.8 75.2
D3 Accumulators 6 0.66 1.39 2.94 169.4 325.5 625.5 157.9 296.7 557.5
E3 Boom 8 1.14 1.93 3.28 266.7 400.9 602.5 237.7 355.5 531.9
F3 Cables 7 0.57 1.08 2.04 182.9 399.6 873.0 146.2 387.6 811.9
H3 Cylinders 14 0.56 0.86 1.33 90.2 180.0 359.0 100.7 193.2 370.6
(b) Repair datab
A3 Rock drill 38 0.63 0.79 0.99 0.36 0.62 0.87 1.9 2.5 3.4
B3 Feeder 54 0.56 0.68 0.82 0.61 0.79 0.97 2.2 2.7 3.4
C3 Hoses 45 0.48 0.60 0.73 0.31 0.49 0.66 1.6 1.9 2.3
D3 Accumulators 6 0.27 0.47 0.83 0.26 0.11 0.49 0.8 1.2 1.8
E3 Boom 8 0.38 0.62 1.02 0.33 0.77 1.20 1.6 2.6 4.2
F3 Cables 7 0.24 0.41 0.70 0.14 0.45 0.76 1.2 1.7 2.3
H3 Cylinders 14 0.28 0.42 0.62 0.41 0.64 0.87 1.6 2.0 2.6
Weibull parameters
(a) Failure dataa
A4 Rock drill 46 0.83 1.05 1.33 45.0 60.6 81.7 44.7 59.4 78.9
B4 Feeder 49 0.88 1.10 1.38 43.9 57.9 76.3 42.9 55.8 72.5
C4 Hoses 78 0.81 0.96 1.15 28.0 35.9 46.0 28.8 36.4 46.0
D4 Accumulators 6 0.68 1.48 3.19 270.1 502.0 933.1 249.6 453.8 825.0
E4 Boom 15 0.62 0.93 1.41 100.6 181.7 328.1 107.0 187.0 326.7
F4 Cables 5 0.58 1.27 2.79 305.7 685.8 1,538.4 293.8 635.5 1,374.6
H4 Cylinders 6 0.84 1.81 3.88 255.5 452.8 802.4 229.5 402.5 705.9
J4 Generator 6 0.41 0.99 2.40 108.0 299.8 831.8 112.0 299.9 802.8
K4 Pumps 5 0.69 1.54 3.41 259.2 507.3 992.9 238.6 456.6 873.7
(b) Repair datab
A4 Rock drill 46 0.49 0.60 0.74 0.28 0.45 0.63 1.5 1.8 2.2
B4 Feeder 49 0.60 0.73 0.89 0.42 0.62 0.83 1.9 2.4 3.0
C4 Hoses 78 0.45 0.53 0.62 0.29 0.41 0.52 1.5 1.7 1.9
D4 Accumulators 6 0.40 0.71 1.25 0.15 0.41 0.98 1.0 1.9 3.7
E4 Boom 15 0.51 0.74 1.06 0.28 0.66 1.04 0.5 2.5 3.9
F4 Cables 5 0.33 0.62 1.16 0.13 0.68 1.22 1.3 2.4 4.3
H4 Cylinders 6 0.27 0.47 0.83 0.19 0.57 0.95 1.3 1.9 2.9
J4 Generator 6 0.44 0.78 1.38 0.08 0.71 1.34 1.3 2.7 5.6
K4 Pumps 5 0.35 0.66 1.23 0.36 0.94 1.52 1.6 3.2 6.1
drilling machines
Downtime
Table V.
mine Y
Summary for machine 4,
317
analysis of
JQME difference between D2, D3 and D4 (accumulators), as R ¼ 1.2 in this particular case.
20,4 Moreover, component A3 in machine 3 has more downtime than the similar component
used in machines 2 and 4. Similarly, component E2 in machine 2 has more downtime
than components E3 and E4. Figure 3 shows there are notable differences in the
downtime of most of the studied components for all machines used in mine Y.
318
Downtime Repair time/failure No. of failures
Component DTLL DTM DTUL TTRLL MTTR TTRUL mLL mM mUL
A1 Rock drill 55.5 83.0 127.8 1.3 1.6 2.0 41.2 50.4 62.5
B1 Feeder 23.0 48.2 111.4 1.8 2.7 4.1 12.2 17.7 26.9
C1 Hoses 171.7 209.2 274.8 1.3 1.4 1.6 129.5 142.9 163.8
G1 Steering system 120.3 179.2 273.0 2.3 2.8 3.5 50.7 61.9 76.4
Table VI. H1 Cylinders 14.5 30.7 74.3 1.2 1.7 2.7 11.7 17.0 26.5
Three estimation points I1 Hydraulics 9.7 22.8 62.1 0.9 1.4 2.3 10.5 16.2 26.7
of the downtime and J1 Fuel system 10.6 22.0 52.1 0.8 1.1 1.8 12.8 18.5 28.4
their coordinates for
machine 1, mine X Notes: DTM, DTLL, DTUL, MTTR, TTRLL and TTRUL are given in units of hours

A2 Rock drill 117.0 136.0 213.4 1.6 1.7 2.2 70.5 76.0 95.2
B2 Feeder 208.3 259.4 390.5 2.6 2.9 3.6 77.9 86.9 106.6
C2 Hoses 510.6 547.9 726.3 2.1 2.2 2.5 237.6 246.2 283.5
D2 Accumulators 15.4 26.8 49.5 0.9 1.2 1.7 15.7 20.7 28.2
E2 Boom 106.8 139.5 271.8 3.3 3.7 5.2 32.2 36.8 51.4
F2 Cables 15.6 26.1 70.3 1.9 2.4 4.0 8.1 10.5 17.3
I2 Hydraulics 13.1 28.1 69.6 0.9 1.3 2.1 14.0 20.5 32.2
J2 Control panal 55.6 69.9 131.6 1.4 1.6 2.2 38.7 43.5 59.6
K2 Water cooler 21.2 37.7 101.7 1.9 2.5 4.2 10.9 14.5 23.9
Table VII. L2 Valves 65.4 93.6 151.5 1.6 1.9 2.5 39.1 46.8 59.5
Three estimation points M2 Manual valves 36.3 53.9 95.9 1.4 1.7 2.3 25.4 31.0 41.4
of the downtime and N2 Movement device 18.7 40.2 98.0 2.0 3.0 4.7 8.9 13.1 20.5
machine 2, mine Y Notes: DTM, DTLL, DTUL, MTTR, TTRLL and TTRUL are given in units of hours

A3 Rock drill 174.1 236.5 345.5 2.1 2.5 3.0 79.2 92.3 111.6
B3 Feeder 254.2 312.8 437.5 2.5 2.7 3.3 100.9 111.9 132.4
C3 Hoses 137.9 182.5 245.0 1.7 1.9 2.2 80.9 93.1 107.9
D3 Accumulators 11.0 22.4 51.2 0.8 1.2 1.8 12.5 17.8 26.9
E3 Boom 20.3 39.2 81.6 1.8 2.6 3.7 10.7 14.9 21.5
Table VIII. F3 Cables 12.3 23.4 48.9 1.2 1.7 2.4 9.8 13.6 19.7
Three estimation points G3 Steering system 78.5 105.0 172.9 1.9 2.2 2.8 40.9 47.3 60.7
of the downtime and H3 Cylinders 44.7 67.3 123.8 2.0 2.4 3.3 22.3 27.4 37.1
machine 3, mine Y Notes: DTM, DTLL, DTUL, MTTR, TTRLL and TTRUL are given in units of hours
Downtime
Component DTLL DTM DTUL TTRLL MTTR TTRUL mLL mM mUL analysis of
drilling machines
A4 Rock drill 134.0 169.1 237.2 1.6 1.8 2.2 79.3 89.1 105.5
B4 Feeder 183.8 234.4 330.1 2.1 2.4 2.9 84.1 94.9 112.7
C4 Hoses 213.9 253.1 326.9 1.5 1.7 1.9 133.7 145.4 165.3
D4 Accumulators 10.4 22.8 56.8 1.3 1.9 3.0 7.8 11.6 18.4 319
E4 Boom 43.5 72.5 126.5 1.9 2.5 3.3 21.9 28.3 37.4
F4 Cables 8.8 20.0 53.3 1.5 2.4 3.9 5.5 8.3 13.6
G4 Steering system 30.4 47.8 77.9 1.3 1.6 2.1 22.6 28.3 36.1
H4 Cylinders 16.1 26.2 74.9 1.5 1.9 3.3 10.3 13.1 22.2 Table IX.
J4 Generator 27.3 49.0 157.3 2.0 2.7 4.9 13.1 17.6 31.6 Three estimation points
K4 Pumps 16.1 37.2 101.0 2.1 3.2 5.2 7.6 11.6 19.1 of the downtime and
Notes: DTM, DTLL, DTUL, MTTR, TTRLL and TTRUL are given in units of hours machine 4, mine Y
135
A4 A3
100
0
A2
00
1,
57
E2
No. of failures
rs e
E4
ou tim
)
(H wn
D2
o
D
E3
0
10
14.2
D3
10
D4 Figure 3.
Confidence log-log plot
10
6 comparison between three

0.7 1.1 10 16.6 24 machines used in mine Y
Time to repair (hours)
When we interpret Figures 4-9 in the same way, we conclude the results of the ratio R
for all machine components, as shown in Table X.
Table X it shows notable differences in the downtime of most investigated
components of all machines used in both mines. For the machines used in the same
mine, three out of six components have significant differences. For the machines used
in different mines, five out of six components have significant differences.
Figure 4 indicates that the components C2 and B3 have higher downtime than the
equivalent components in another machine used in the same mine.
Figure 5 and 6 illustrate that the components A1 and B1 used in machine 1 in mine
X have less downtime than the same components used in the machines in mine Y.
This may be due to the differences of the rock properties between the two mines.
The geological strength index (GSI) of mine Y varies between 50 and 80
JQME 300
C2
20,4 200
C4
B3
100 C3 B2
0
00
1,
320 B4
No. of failures
rs e
ou tim
)
(H wn
o
D
0
10
F3
F2
F4
10
10
Figure 4.
Confidence log-log plot
comparison between three 5
machines used in mine Y 1 2 40 70
190
A3
A4
100
0
No. of failures
50
A2
rs e
A1
ou tim
)
(H wn
o
D
0
10
40
60
30
1 1.3 1.6 7.5
Figure 5.
Confidence log-log Notes: A1 (rock drill) and B1 (feeder) are components used
plot comparison in machine 1 in mine X, while A2, A3 and A4 (rock drill),
between four machines
used in both mines B2, B3, and B4 (feeder) are components used in machines 2,
3 and 4 in mine Y
160 Downtime
B3 analysis of
0
drilling machines
00
100 B2
1,
80 B4
321
rs e
ou tim
No. of failures
)
(H wn
o
D
0
10 B1
20
20
10
1 2 10 20
Figure 6.
Note: A1 (rock drill) and B1 (feeder) are components used in Confidence log-log plot
machine 1 in mine X, while A2, A3 and A4 (rock drill), B2, comparison between
four machines used
B3, and B4 (feeder) are components used in machines 2, 3 in both mines
and 4 in mine Y
(Edelbro, 2008) while the GSI of mine X varies between 30 and 50 (Sjöberg, 2003 as
cited in Edelbro, 2008).
Figure 7 compares the downtime of component hoses C in four machines in different
mines. Component C2 (hoses) has more downtime than the same components, C1, C3
and C4, used in the rest of the machines. A possible explanation is the difference in
how the various machines were handling. However, further research is needed to
confirm this explanation. Figures 5-7 clearly show that components B2 and C4 have
approximately the same downtime (259 h and 253 h, respectively). Similarly,
components A3 and B4 have approximately the same downtime (236 h and 234 h,
respectively). In addition, at the upper limit of the downtime interval, components
B4 and C4 have approximately the same downtime. The figures also show that at the
lower limit of the downtime interval, components C4 and B2 have approximately
the same downtime, as do components A3 and C1 and components C3 and A4.
Figure 8 compares two components used in machines 1, 3 and 4 in two different
mines. The components G4 (steering system), H4 and H1 (cylinders) have
approximately the same downtime at the upper limit of the downtime interval.
For comparison type four, Figure 9 shows no significant difference in the downtime of
the component I (hydraulics) used in machines 1 and 2, as R ¼ 1.2 in this particular case.
The order of the significant components has been prioritised by Wijaya et al. (2012),
based on three scenarios: the mean estimation point of the downtime (the high-likelihood
scenario), the upper limit estimation point of the downtime (the worst-case scenario)
and the lower limit estimation point of the downtime (the best-case scenario).
JQME 400
20,4
300
C2
0
00
322
1,
No. of failures
C4
rs e
ou tim
)
(H wn
o
D
C1
0
30
C3
100
80
0
10
60
1 1.1 1.5 5 6.6 8
Notes: C1 (hoses) is a component used in machine 1 in mine X,
Figure 7. while C2, C3 and C4 (hoses), are components used in
Confidence log-log plot machines 2, 3, and 4 in mine Y. G1 (steering system) and
comparison between H1 (cylinders) are components used in machine 1 in mine X;
four machines used
in both mines G3 and G4 (steering system), H3 and H4 (cylinders) are
components used in machines 3 and 4 in mine Y
These three scenarios are important to decisions about certain activities in the mine,
for instance, planning new operations and budgeting maintenance. Table XI illustrates
the order of the significant components from the prioritisation of maintenance action
point of view. During smooth operations, the component B4 (feeder) has a downtime
of about 330 h more than component C4 (hoses) which has a downtime about 327 h.
In the worst-case scenario of comparison type 1, component C4 should be given
priority because it has a high number of breakdowns, combined with breakdowns of
short duration; in comparison, component B4 has fewer breakdowns, combined with
breakdowns of long duration; refer to Figure 4 and Table IX. This is important because
a high frequency of breakdowns leads to a lower production rate as the machine needs
more time to reach normal performance after each breakdown.
To cite another example, if the maintenance management department determines
the highest acceptable amount of machine downtime at the component level, it is
essential for the maintenance staff to know which components are likely to go beyond
the acceptable limit. For example, if they decide that the highest acceptable amount of
downtime is 350 hours per year for one component, in the worst-case scenario, we can
observe in Figure 4 that components C2 (hoses on machine 2), B3 (feeder on machine 3)
and B2 (feeder on machine 2) will exceed 350 hours of downtime. In this case, a good
strategy may be to convince the manufacturing company to improve the lifetime of the
components; another possibility is to increase the PM on these particular components.
However, more research is needed on the topic.
80
G1
Downtime
70
analysis of
60
drilling machines
50
G3
G4 H3 323
0
No. of failures
30
0
H1
20
H4
0
10
rs e
ou tim
)
(H wn
o
10 D
30
10
8
1 1.25 7.5 10 12
Note: C1 (hoses) is a component used in machine 1 in mine X,
while C2, C3 and C4 (hoses), are components used in Figure 8.
machines 2, 3 and 4 in mine Y. G1 (steering system) and Confidence log-log plot
H1 (cylinders) are components used in machine 1 in mine X; comparison between
three machines used
G3 and G4 (steering system), H3 and H4 (cylinders) are in both mines
components used in machines 3 and 4 in mine Y
5.1 Link between analysis and improvements

The following general method is suggested regarding the link between analysis and
critical components improvement:
(1) collecting of maintenance and repair data;
(2) using confidence log-log diagram with reliability vs maintainability;
(3) identifying the most critical components (largest downtime);
(4) identifying the contribution of influencing factors (reliability and\or
maintainability) for each of the critical components; and
(5) redesigning of the critical components with respect to the finding.
Using confidence log-log plots, we pinpoint reliability/maintainability problems in
Figures 3-8. The plots show that at the upper limit of downtime (DTULX200 hours)
per year, components C (hoses), B (feeder), A (rock drill), E (boom) and G (steering
system) have reliability problems, with a high number of failures and low levels of
repair time. Thus, DFR strategy should be adopted to reduce their downtime.
5.2 Suggestions for critical components improvement

It has been concluded that the most common problem for all critical components
is reliability problems. Therefore, this section will focus on suggestions on how to
redesign the critical components to improve the reliability and reduce the machine’s
JQME 44
20,4 37.5
0
10
I2
I1
324
No. of failures
70
rs e
ou tim
)
(H wn
o
D
30
12.5
20
10
11.2
Figure 9. 10
Confidence log-log plot 0.8 0.9 1 3 4
comparison between Time to repair (hours)
two machines used
in both mines Notes: I1 (hydraulics) is a component of machine 1 used in mine
X; I2 (hydraulics) is a component in machine 2 in mine Y
Component Symbol Mine Machine Figure R Status
Rock drill A2, A3, A4 Y 2, 3, 4 3 1.7 Significant differences

Boom E2, E3, E4 Y 2, 3, 4 3 3.5 Significant differences
Accumulators D2, D3, D4 Y 2, 3, 4 3 1.2 No significant differences
Hoses C2, C3, C4 Y 2, 3, 4 4 3 Significant differences
Feeder B2, B3, B4 Y 2, 3, 4 4 1.3 Indicated differences
Cables F2, F3, F4 Y 2, 3, 4 4 1.3 Indicated differences
Rock drill A1, A2, A3, A4 X, Y 1, 2, 3, 4 5 2.8 Significant differences
Feeder B1, B2, B3, B4 X, Y 1, 2, 3, 4 6 6.5 Significant differences
Hoses C1, C2, C3, C4 X, Y 1, 2, 3, 4 7 3 Significant differences
Table X. Steering system G1, G3, G4 X, Y 1, 3, 4 8 3.7 Significant differences
R ratio for machines Cylinders H1, H3, H4 X, Y 1, 3, 4 8 2.5 Significant differences
components Hydraulics I1, I2 X, Y 1, 2 9 1.2 No significant differences
downtime. Discussions with maintenance personal reveal that the most of the failures
in the feeder hoses are due to the mine’s environment. For example, during drilling, the
feeder hits the wall at different angles, especially when the feeder movement is
restricted because of spatial limitations. To overcome this problem, the feeder could be
equipped with an iron plate on both sides; the plate should be large enough to prevent
hoses from being scratched and to prevent nipples at the necks from being broken
(Plates 1-2 and Figure 10).
Another problem in the feeder is the pull rope breaking. This happens for two
reasons. First, the pull rope relaxes with usage; it then hangs over the edge of the cradle
plate when the plate moves forward and back on the slide bar (see Plates 3 and 4).
Type 1, mine Type 2, mine X and Type 3, mine X and Type 4, mine X and
Downtime
Y three machines Y four machines Y three machines Y two machines analysis of
Scenario Scenario Scenario Scenario drilling machines
Order 1 2 3 Order 1 2 3 Order 1 2 3 Order 1 2 3
1 C2 C2 C2 1 C2 C2 C2 1 G1 G1 G1 1 I2 I2 I2
2 B3 B3 B3 2 B3 B3 B3 2 G3 G3 G3 2 I1 I1 I1 325
3 B2 B2 C4 3 B2 B2 C4 3 H3 H3 H3
4 C4 A3 B2 4 C4 A3 B2 4 G4 G4 G4
5 A3 B4 B4 5 A3 B4 B4 5 H1 H4 H4
6 B4 C4 A3 6 B4 C4 A3 6 H4 H1 H1
7 C3 E2 C3 7 C1 C1 C1
8 A4 C3 A4 8 C3 C3 C3
9 E2 A4 A2 9 A4 A4 A4
10 A2 A2 E2 10 A2 A2 A2
11 E4 E4 E4 11 A1 A1 A1
12 E3 E3 E3 12 B1 B1 B1
13 D2 F2 F2
14 F2 D4 D2
15 F3 F4 F3
16 D4 D3 D3 Table XI.
17 D3 D2 D4 The order of the
18 F4 F3 F4 significant component
Plate 1.
Scratched hoses
Second, the operator or repair person may put extreme tension on the pull rope
when making an adjustment. This excessive tension leads to two undesirable
occurrences: first, a reduction in the lifetime of the pull rope; second, a high load on the
pulley wheel leading to a reduction of the lifetime of the roller bearing inside it (see
Plate 5).
To solve this problem, an electrical motor with control circuit should be designed to
make automatic adjustments for the pull rope, keeping it at a constant desired tension,
as shown in Figure 11. Stronger roller bearings are another possibility.
JQME
20,4
326
Plate 2.
Nipples and necks
Iron
Plate
Figure 10.
Suggested plate of iron
Discussions with maintenance personal also reveal that the frequent failures in rock
drill are damaging the third and fourth cup seals located inside front head (nose) of this
component, as shown in Plates 6-7 and Figure 12.
A possible cause is the high water pressure inside the nose. Water is used to cool the
front head and flush it during the drilling process. However, damaging the cup seals
will cause water and oil to mix, leading to the adhesion of the valves used in the
hydraulic system. To solve this problem, the water pressure inside the front head
should be reduced by increasing the number of holes, especially in the area between the
third and the forth cups seals, as shown in Plate 8. Further research is needed to
confirm this explanation.
It is worth to mention that all suggestions for improvement were discussed and
agreed up on together with the maintenance experts and product development team of
the manufacturing company. The product development team also decided to take the
suggestions on and implement them in the future.
Downtime
analysis of
drilling machines
327
Plate 3.
Broken pull rope
Plate 4.
Hung pull rope
6. Conclusions
The downtime analysis of drilling machines shows a significant difference between
the three machines used in same mine (mine Y) in the downtime of components A
(rock drill), C (hoses) and E (boom). There is no significant difference between these
machines in the downtime of component D (accumulators). The analysis also finds
differences in the downtime of components B (feeder) and F (cables). Components A
and B used in mine X have less downtime than the same components used in the
machines of mine Y, most probably as a result of the differences in the rock properties
between the two mines. Further research is required to explain the differences in the
downtime between the same models of the drilling machine. There is a significant
difference in the downtime of component G (steering system) found in machines used
within a single mine and across mines. In contrast, there is no significant difference in
the downtime of component I (hydraulics) found in machines 1 and 2 used in different
mines. In general, there are notable differences in the downtime of most investigated
JQME
20,4
328
Plate 5.
Roller bearing with pulley
Figure 11.
Suggested electrical motor Gear Electrical Motor
Plate 6.
Damaged cup seal
components of all machines used in both mines. For the machines used in the
same mine, three out of six components have significant differences. For machines used
in both mines, five out of six components have significant differences.
The downtime analysis of drilling machines also shows that the machines’
components can be ranked on their downtime, using three different scenarios (the high
likelihood, the worst-case, and the best-case scenarios) based on three estimation
Downtime
analysis of
drilling machines
329
Plate 7.
Cup seals
Cup seals
Figure 12.
Cup seals
Plate 8.
Front head (nose)
of rock drill
JQME points of downtime. The components C2 (hoses on machine 2), B3 (feeder on machine 3)
20,4 and B2 (feeder on machine 2) have more downtime (DTULX350) hours per year.
The downtime (DTULX200) hours per year of components C (hoses), B (feeder), A (rock
drill), E (boom) and G (steering system) stem from reliability problems. Because they
have a high number of failures and short repair times, as shown in Figures 3-8, a DFR
should be created to decrease their downtime. Overall, no maintainability problems
330 were detected for the machines’ significant components; therefore, a design for a
maintainability strategy is not required. The failures of the feeder hoses are likely due
to the harsh work environment during drilling. In this case, putting iron plates on both
sides of the feeder may reduce the number of failures. The breakage of the feeder pull
rope is due to usage and excessive tension; this can be treated and reduced by
installing an electrical motor with a control circuit to keep the pull rope at a constant
desired tension. Finally, increasing the number of holes between the third and the
fourth cup seals inside the nose of the rock drill could solve the problem of cup
seal damage.
In summary, the suggested “DFR” solution is found to be applicable. In order to
judge for cost effectiveness, one should perform life cycle cost analysis to identify, if the
solution is also economically viable.
References
Abernethy, R. (2000), The New Weibull Handbook, Gulf Publishing Co., Houston, TX.
Altman, D.G., Machin, D., Bryant, T.N., Gardner, M.J., Gardner, S., Bird, S., Campbell, M.,
Daly, L.E. and Morris, J. (2000), Statistics with Confidence: Confidence Intervals and
Statistical Guidelines, BMJ Books, London.
Ansell, J. and Phillips, M.J. (1994), Practical Methods for Reliability Data Analysis, Oxford
University Press Inc., New York, NY.
Ascher, H. and Feingold, H. (1984), Repairable System Reliability, 7th ed., Dekker, New York, NY.
Barabady, J. and Kumar, U. (2008), “Reliability analysis of mining equipment: a case study of a
crushing plant at Jajarm Bauxite mine in Iran”, Reliability Engineering System Safety,
Vol. 93 No. 4, pp. 647-653.
Birolini, A. (2007), Reliability Engineering Theory and Practice, 5th ed., Springer-Verlag, Berlin,
Heidelberg, New York, NY.
Blischke, W. and Murthy, D. (2003), Case Studies in Reliability and Maintenance, Wiley-
Interscience, New Jersy.
Colt, D.W. (1997), “System-reliability confidence-intervals for complex-systems with estimated
component-reliability”, IEEE Transactions on Reliability, Vol. 46 No. 4, pp. 487-493.
Curran-Everett, D. and Benos, D.J. (2004), “Guidelines for reporting statistics in journals
published by the American physiological society”, American Journal of Physiology-Heart
and Circulatory Physiology, Vol. 287 No. 2, pp. H447-H449.
Cutifani, M., Quinn, B. and Gurgenci, H. (1996), “Increased equipment reliability, safety and
availability without necessarily increasing the cost of maintenance”, Proceedings, Mining
Technology Conference, Cooperative Research Center for Mining Technology, 11-12
September, Fremantle, WA.
Edelbro, C. (2008), Strength, Fallouts and Numerical Modelling of Hard Rock Masses, Luleå
University of Technology, Luleå.
Ghosh, S. and Majumdar, S.K. (2011), “Reliability modeling and prediction using classical and
Bayesian approach: a case study”, The International Journal of Quality Reliability
Management, Vol. 28 No. 5, pp. 556-586.
Gustafson, A., Schunnesson, H., Galar, D. and Kumar, U. (2013), “The influence of the Downtime
operating environment on manual and automated load-haul-dump machine: a fault
tree analysis”, International Journal of Mining, Reclamation and Environment, Vol. 27 analysis of
No. 2, pp. 75-87. drilling machines
Hoseinie, S.H., Ataei, M., Khalokakaie, R., Ghodrati, B. and Kumar, U. (2012), “Reliability analysis
of drum Shearer machine at mechanized longwall mines”, Journal of Quality in
Maintenance Engineering, Vol. 18 No. 1, pp. 98-119.
331
Knights, P.F. (2001), “Rethinking Pareto analysis: maintenance applications of logarithmic
scatterplots”, Journal of Quality in Maintenance Engineering, Vol. 7 No. 4,
pp. 252-263.
Kumar, D., Klefsjo, B. and Kumar, U. (1992), “Reliability analysis of power transmission cables of
electric mine loaders using the proportional hazards model”, Reliability Engineering
System Safety, Vol. 37 No. 3, pp. 217-222.
Kumar, U. and Klefsjö, B. (1992), “Reliability analysis of hydraulic systems of LHD machines
using the power law process model”, Reliability Engineering System Safety, Vol. 35 No. 3,
pp. 217-224.
Kumar, U., Klefsjö, B. and Granholm, S. (1989), “Reliability investigation for a fleet of load haul
dump machines in a Swedish mine”, Reliability Engineering System Safety, Vol. 26 No. 4,
pp. 341-361.
Louit, D.M., Pascual, R. and Jardine, A.K.S. (2009), “A practical procedure for the selection of
time-to-failure models based on the assessment of trends in maintenance data”, Reliability
Engineering System Safety, Vol. 94 No. 10, pp. 1618-1628.
Masters, B., Lewis, T. and Kolarik, W. (1992), “A confidence interval for the availability ratio for
systems with weibull operating time and lognormal repair time”, Microelectronics and
Reliability, Vol. 32 Nos 1/2, pp. 89-99.
Modarres, M. (2006), Risk Analysis in Engineering: Techniques, Tools, and Trends, Taylor &
Francis Group, Boca Raton, FL.
Rausand, M. and Hoyland, A. (2004), System Reliability Theory: Models, Statistical Methods, and
Applications, 2nd ed., John Wiley and Sons Inc., New Jersy.
Roman, P.A. and Daneshmend, L. (2000), “Economies of scale in mining – assessing upper
bounds with simulation”, The Engineering Economist, Vol. 45 No. 4, pp. 326-388.
Schroeder, B. and Gibson, G.A. (2010), “A large-scale study of failures in high-performance
computing systems”, IEEE Transactions on Dependable and Secure Computing, Vol. 7
No. 4, pp. 337-351.
Vagenas, N. and Nuziale, T. (2001), “Genetic algorithms for reliability assessment of mining
equipment”, Journal of Quality in Maintenance Engineering, Vol. 7 No. 4,
pp. 302-311.
Vayenas, N. and Xiangxi, W. (2009), “Maintenance and reliability analysis of a fleet of load-haul-
dump vehicles in an underground hard rock mine”, International Journal of Mining,
Reclamation and Environment, Vol. 23 No. 3, pp. 227-238.
Wijaya, A.R., Lundberg, J. and Kumar, U. (2012), “Downtime analysis of a scaling
machine”, International Journal of Mining, Reclamation and Environment, Vol. 26 No. 3,
pp. 244-260.
About the authors

Hussan S. Al-Chalabi received BEng Degree in Mechanical Engineering from the Mosul
University, Iraq in 1994 and MSc Degree in Mechanical Engineering in Thermal Power from the
Mosul University, Iraq in 2008. Then he joined the Department of Mechanical Engineering at the
Mosul University as a Lecturer. In 2011 he joined the Division of Operation, Maintenance and
JQME Acoustics at the LTU as a Doctoral Student. Hussan S. Al-Chalabi is the corresponding author
and can be contacted at: hussan.hamodi@ltu.se
20,4 Jan Lundberg is a Professor of Machine Elements at the Luleå University of Technology
and also in Operation & Maintenance with focus on product development. During the years 1983-
2000, his research concerned mainly about engineering design in the field of machine elements in
industrial environments. During the years 2000-2006, his research concerned mainly about
332 industrial design, ergonomic and related problems as cultural aspects of design and modern tools
for effective industrial design in industrial environments. From 2006 and forward, his research is
completely focused on maintenance issues like methods for measuring failure sources, how to do
design out maintenance and how to design for easy maintenance.
Dr Andi Wijaya received BEng Degree in Mechanical Engineering from the Gadjah Mada
University (GMU), Indonesia in 1998 and MSc Degree in Industrial Ergonomics and Licentiate
Degree in Sound and Vibration from the Lulea University of Technology (LTU), Sweden in 2001
and 2003, respectively. Then he joined to the Department of Mechanical and Industrial
Engineering at the GMU as a Lecturer. Since 2009, he received PhD Degree from the Division of
Operation, Maintenance and Acoustics, Luleå University of Technology, Sweden in 2012.
Behzad Ghodrati is an Associate Professor of Maintenance and Reliability Engineering
at the Luleå University of Technology. He obtained his PhD Degree on “Reliability based spare
parts planning” from the Luleå University of Technology and he was awarded the Postdoctoral
Research Fellowship from the University of Toronto in 2008.
To purchase reprints of this article please e-mail: reprints@emeraldinsight.com

Or visit our web site for further details: www.emeraldinsight.com/reprints
PAPER II
Case Study: Model for Economic Lifetime of Drilling

Machines in the Swedish Mining Industry
Al-Chalabi, H., Lundberg, J., Jonsson, A., Ahmadi, A., 2014. Case Study: Model for
Economic Lifetime of Drilling Machines in the Swedish Mining Industry. Published in the
Engineering Economist.
http://dx.doi.org/10.1080/0013791X.2014.952466
The Engineering Economist, 00:1–17, 2014
Published with license by Taylor & Francis Group, LLC
ISSN: 0013-791X print/1547-2701 online
DOI: 10.1080/0013791X.2014.952466
Case Study: Model for Economic Lifetime of Drilling

Machines in the Swedish Mining Industry
Downloaded by [Lulea University of Technology], [Mr Hussan Hamodi] at 23:06 02 October 2014
HUSSAN AL-CHALABI,1 JAN LUNDBERG,2

ALIREZA AHMADI,2 AND ADAM JONSSON3
1
Division of Operation, Maintenance and Acoustics, Luleå University
of Technology, Luleå, Sweden and Mechanical Engineering Department,
College of Engineering, University of Mosul, Mosul Iraq
2
Division of Operation, Maintenance and Acoustics, Luleå University
of Technology, Luleå, Sweden
3
Division of Mathematic Science, Luleå University of Technology, Luleå,
Sweden
The purpose of this article is to develop a practical economic replacement decision

model to identify the economic lifetime of a mining drilling machine. A data-driven
optimization model was developed for operating and maintenance costs, purchase price,
and machine resale value. Equivalent present value of these costs by using discount rate
was considered. The proposed model shows that the absolute optimal replacement time
(ORT) of a drilling machine used in one underground mine in Sweden is 115 months.
Sensitivity and regression analysis show that the maintenance cost has the largest impact
on the ORT of this machine. The proposed decision-making model is applicable and
useful and can be implemented within the mining industry.
Introduction
Economic globalization increases competition among mining companies, pushing them to
achieve higher production rates by increasing automation and mechanization and using new
and more effective equipment. This forces companies to use more reliable capital equipment
with higher performance capabilities; naturally, these are more expensive. The equipment
used in underground mining industries is subject to degradation throughout its operating
life. This increases the operating and maintenance costs and reduces production rates,
causing a negative economic effect. In addition, the equipment used in underground mining
is subject to a harsh working environment, and this accelerates degradation. Given all of
these factors, key questions for the mining industry include the following. When should
the company replace the equipment to minimize cost? How can the maintenance manager

C Hussan Al-Chalabi, Jan Lundberg, Adam Jonsson, and Alireza Ahmadi
This is an Open Access article. Non-commercial re-use, distribution, and reproduction in any
medium, provided the original work is properly attributed, cited, and is not altered, transformed, or
built upon in any way, is permitted. The moral rights of the named author(s) have been asserted.
Address correspondence to Hussan Al-Chalabi, Division of Operation, Maintenance and Acous-
tics, Luleå University of Technology, Luleå 97187, Sweden; E-mail: hussan.hamodi@ltu.se
1
2 H. Al-Chalabi et al.
convince finance and production managers to replace capital equipment at a specified time
in its life cycle? To answer these questions, life cycle cost analysis should be done in
advance of an equipment replacement decision.
The optimum replacement age of equipment is defined as the time at which the total cost
is at its minimum value (Jardine and Tsang 2006). In the mining industry, the costs associated
with owning equipment can be grouped into categories: initial purchase, installation, direct
downtime, maintenance and operating, financing, and cost recovery on disposal. The sum
of these costs represents the total cost required to own the mining equipment (Hall 2007).
Life cycle cost analysis helps decision makers justify equipment replacement on the basis
of the total costs over the equipment’s useful life. It allows the maintenance manager to
specify the optimal replacement time at the time of the equipment’s purchase.
Cost function models can be allocated to the various categories to allow easy estimation
of the total cost. Such models can be generally classified as detailed models, analogous
models, and parametric models. A detailed model uses estimates of material quantities
and prices, labor time, and rates to estimate the direct costs of equipment. Analogous
models identify similar equipment and adjust costs to account for differences between it
and the target equipment. Cost estimation with a parametric model is based on predicting
the equipment’s total cost by using regression analysis based on technical information
and historical cost (Asiedu 1998). Life cycle cost (LCC) analysis should not be seen as
a method for defining the total cost of the equipment but as a help in decision making;
thus, LCC analysis should be restricted to costs that can be controlled. In general, LCC
is determined by summing up all of the potential costs associated with equipment over its
lifetime. It is well known that the value of expenditure today costs more than the same value
of expenditure next year because of the “time value of money.” A discount rate is used
to take into consideration the time value of money. To compare costs incurred at different
times we must shift expenditure to a reference point in time. Thus, in this article, we are
interested in estimating the equivalent present value of earlier or future costs.
Literature Survey
Standard models for economic replacement time decision contain an estimation of the dis-
counted costs by minimizing the LCC of the equipment. The assumption of these models
is that equipment will be replaced at the end of its economic lifetime by a continuous se-
quence of identical equipment (Hartman and Tan 2014). Recently, a number of researchers
have studied the economic lifetime of capital equipment. Some consider the optimal life-
time of capital equipment using economic theories and vintage capital models, represented
mathematically by nonlinear Volterra integral equations with unknown limits of integration
(Boucekkine et al. 1997; Cooley et al. 1997; Hritonenko 2005; Hritonenko and Yatsenko
2003; Yatsenko 2005). Others use the theory of dynamic programming considering tech-
nological changes under finite and infinite horizons (Bellman 1955; Bethuyne 1998; Elton
and Gruber 1976; Hartman 2005; Hritonenko and Yatsenko 2008; Mardin and Arai 2012).
Yatsenko and Hritonenko (2005) studied the lifetime optimization of capital equipment
using integral models. The study designs a general investigation framework for optimal
control of the models. Hritonenko and Yatsenko (2007) studied optimal equipment replace-
ment without paradoxes. Using an integral model to calculate the economic lifetime of
equipment and considering technological changes (TCs), they showed that the economic
lifetime of equipment is shorter when the embodied TC is more intense. Hartman and
Murphy (2006) offered a dynamic programming approach to the finite-horizon equipment
Model for Economic Lifetime of Drilling Machines 3
replacement problem with stationary cost. Their model studies the relationship between the
infinite-horizon solution (continuous replacement of equipment at the end of its economic
lifetime) and the finite-horizon solution. Hritonenko and Yatsenko (2009) constructed a
computational algorithm to solve a nonlinear integral equation. The solution is important
for finding the optimal policy of equipment replacement under technological advances.
Kärri (2007) considered the optimal replacement time of an old machine, using an opti-
mization model that minimizes the machine cost. The model is built to handle capacity
expansion and replacement situations. Using real costs without inflation, Kärri (2007) mod-
eled the costs of the old machine with simple linear functions. He also used an optimization
model that maximizes profit. Scarf and Bouamra (1999) addressed the capital replacement
problem using a discounted cost criterion over a finite time horizon. They presented a robust
approach to solving the fleet replacement problem in which the fleet size is allowed to vary
at replacement. A survey of multiple and single asset solution techniques under a variety of
settings, including tax, variable utilization, various uncertainties, and technological change,
was addressed by Hartman and Tan (2014). They also illustrated a number of open problems
that are worthy of future research. Generally speaking, these studies focus on estimating
the economic lifetime of equipment, considering technological changes and using integral
models, theories of dynamic programming, vintage capital models, and algorithms to solve
nonlinear integral equations.
Despite the available information, it can be difficult for users to implement complex
models to calculate the optimal replacement time of equipment. Moreover, these models
sometimes require specific types of data that, as in our case study, are not available. These
can include data on production output, technological labor/output coefficient, revenue,
profit, etc. Thus, the aim of this study is to identify the replacement age of a mining drilling
machine from an economic point of view, using available data from a mining company,
specifically, the operating and maintenance costs, purchase price, and machine resale value.
In this study, equivalent present value of these costs was considered by using a discount rate.
Description of Drilling Machine

The drilling machines typically used in mines are manufactured by different companies
and have different technical characteristics; for example, capacity and power. An example
of a drilling machine and its components is presented in Figure 1.
The drilling machine is divided into several subsystems connected in series configu-
ration (see Atlas Copco Rock Drills AB 2010). If any subsystem fails, the operator will
stop the machine to fix it. Thus, all machine subsystems work simultaneously to achieve
the desired function.
Data Collection
The cost data used in this study were collected over 4 years in the Maximo computerized
maintenance management system (CMMS). The cost data contain corrective maintenance
costs, preventive maintenance costs, and repair time. The corrective and preventive main-
tenance costs contain spare parts and labor (repair person) costs. In CMMS, the cost data
are recorded based on calendar time. Because drilling is not a continuous process, the
operating cost is estimated by considering the utilization of the machines. It is important
here to mention that all cost data used in this study are real costs without inflation. Due to
the company regulations, all cost data are encoded and expressed as currency unit (cu) for
this study. Samples of cost data can be seen in Table 1.
Figure 1. Drilling machine. Source: Andreas Nordbrandt, Vice President Service Operations, Atlas
Copco Rock Drills AB (2010).
© Atlas Copco Rock Drills AB. Reproduced by permission of Atlas Copco Rock Drills AB.
Permission to reuse must be obtained from the rightsholder.
Methodology and Model Development

In this study the following assumptions are made for the optimization model:
• The cost of capital is given by the mining company involved in this study.
• Acquisition cost of the machine remains constant at each replacement.
• There is no installation cost for the machine.
• The optimization model is used for a finite time horizon.
• Production losses due to lead time during machine replacement are not considered.
• The machine will be used as a redundant machine after it reaches to its scrap value.
The taxes are included in the purchase price and operating and maintenance costs; for
that reason, taxes are not included as an independent parameter in the optimization model.
The study develops a practical optimization model based on the total cost. Associated
operating and maintenance costs, as well as purchase price and machine resale value, are
considered. The maintenance costs (MC) for each month of operation consist of corrective
maintenance (CM) and preventive maintenance (PM) costs:
MC = CM + PM. (1)
The corrective and preventive maintenance costs are given by
CM = SPc + LCc (2)

PM = SPp + LCp . (3)
Table 1
Sample of Cost Data
Actual
Actual materi- Total Actual Actual Actual Inventory
Work working als cost real cost labor service start descrip- Work
description time (h) (cu) (cu) (cu) cost (cu) date tion type
Extension 1 28.148 28.598 0.45 0 20xx- Feeder PM
Extender 03-15
2 bolts 13:23
of
V-feeder
FU1 Atlas 5 9.836 14.018 0 4.182 20xx- PM
L2C/2 03-15
13:24
Mount the 6 0 2.7 2.7 0 20xx- Steering CM
sensor 03-15 system
cables 22:41
Atlas 16 0 7.2 7.2 0 20xx- Electrical CM
Copco 03-16 system
L2C 13:17
Replacing 0.5 0 0.225 0.225 0 20xx- Hoses CM
the hose 03-19
feeding 07:30
shift
Because drilling is not a continuous process in the collaborating mine, operating cost
(energy cost and steel rod cost) is calculated for each month based on the utilization of
the drilling machine. The company plans to use the machine for 120 months. Therefore,
extrapolation for the operating and maintenance cost data was done. Figures 2 and 3
illustrate the maintenance and operating costs determined by the data extrapolation.
In Figures 2 and 3, the dots represent the real historical data for maintenance and
operating costs. Curve fitting was done using Table Curve 2D (Alfasoft AB, Göteborg,
Sweden) software to show the behavior of these costs before and after the time when data
were collected. Note that the fitting would be better if more data were available for a time
period of more than 4 years. This software uses the least squares method to find a robust
(maximum likelihood) optimization for nonlinear fitting. It is worth mentioning that the
drilling machine in this case study has no multilevel preventive maintenance programme. In
addition, it was new at the start of utilization. This is the main reason why the maintenance
cost is quite low in earlier months. The history shows that when the maintenance costs
started growing, the user company began to keep track of cost data by using CMMS.
The Lorentzian cumulative equation of extrapolation for expected maintenance cost
obtained by the software is expressed as

a x−b π
Y = arctan + , (4)
π c 2
Figure 2. Maintenance cost.
where Y represents the expected maintenance cost, a = 217.42, b = 112.37, c = 13.63,

r2 (adj.) = 0.97, and X represents the time (1, 2, 3, 4, . . . , n months). Similarly, the
Lorentzian cumulative equation of extrapolation for expected operating cost is expressed
as

a x−b π
Y = arctan + , (5)
π c 2
Figure 3. Operating cost.

where Y represents the expected operating cost, a = 79.89, b = 109.2, c = 13.85, r2 (adj.)
= 0.91, and X represents the time (1, 2, 3, 4, . . . , n months).
As the figures show, the operating and maintenance costs increase over time. In fact,
the number of failures increases with time and/or the machine consumes more energy due
to machine degradation.
A declining balance depreciation model is used to estimate the resale value of
the machine after each month of operation. The machine’s resale value is its value if
the company wants to sell it at any time during its planned lifetime. The resale value of the
machine, denoted S(i), is assumed to be given by the following formula (Eschenbach 2010;
Luderer et al. 2010):
S (i) = BV1 × (1 − Dr)i , (6)
where i represents time (month), i = 1, 2, 3, . . . , 120, and BV1 is the machine’s value at
the first day of operation. In addition,
BV1 = PP × a, (7)
where a represents the percentage that multiplied by the machine purchase price to represent
the machine value at the first day of use. During discussions with us, company experts agreed
that the machine’s purchase price decreases by 10% on the first day of use (i.e., a = 0.9).
In this study, the machine purchase price is 6,000 cu. Hence, the machine’s value on the
first day of use is 5,400 cu.
The depreciation rate that allows for full depreciation by the end of the planned lifetime
of the machine is modeled by the following formula (Luderer et al. 2010):
T1
SV
Dr = 1 − , (8)
BV1
where T represents the planned lifetime of the machine, 120 months in the case study. The
machine was assumed to reach scrap value after 10 years. The machine’s resale value is
given by
S (i) = (PP × a) × (1 − Dr)i . (9)
The declining balance depreciation model is suitable in this case because it assumes
that more depreciation occurs at the beginning of the equipment’s planned lifetime and less
at the end. It also considers that the equipment is more productive when it is new and its
productivity declines continuously due to equipment degradation. Therefore, in the early
years of its planned lifetime, a machine will generate more revenue than in later years.
In accountancy, depreciation refers to two aspects of the same concept. The first is the
decrease in the equipment’s value. The second is the allocation of the cost of the equipment
to periods in which it is used. The scrap value is an estimate of the value of the equipment
at the time it is disposed of. In this case study, 50 cu is assumed to be the scrap value of
the machine at end of its planned lifetime, a figure given to us by experts at the company.
Figure 4 shows the drilling machine’s resale values using the declining balance depreciation
model.
It is clear from Figure 4 that the machine’s resale values decrease with time until it
reaches scrap value at the end of its planned lifetime.
The next step in the calculations is to calculate the total ownership cost over each
operating month. In this study, the economic lifetime of the drilling machine is defined as
Figure 4. Machine resale value.
the machine age that minimizes the machine total ownership cost. The total ownership cost
over period i is denoted by TOCi , i = 1, 2, 3, . . . , n, where n is the number of operating
months. By definition,
RT

TOCi = PP + (MCi + OCi ) − S (i) , (10)
i=1
where MCi and OCi is the maintenance and operating costs for the ith month.
The reason for using total ownership cost is that the machine’s PP, OC, and MC
represent costs, whereas the resale value represents income for the company when it is
willing to sell the machine.
The objective is to determine the optimal replacement time that minimizes the total
ownership cost over the machine’s planned horizon. We assume that the replacement
machines (i.e., the new machines) have the same performance and cost as the existing
machine (i.e., identical machines). The number of replacement cycles during the planned
horizon is modeled as

Planned lifetime T
M= = . (11)
Replacement time RT
Figure 5 illustrates the expected total ownership cost of the machine over the planned
horizon.
As Figure 5 shows, the total ownership cost increases with time for two reasons: first,
operating and maintenance costs increase over time; second, the machine’s resale value
decreases over time until reaches its scrap value.
The optimal replacement time is the value of RT that minimizes the total ownership
cost value, as shown in Eq. (12). A discount rate of 10% was used to consider the time
value of money as mentioned by the collaborating mining company.
⎡⎧⎛ ⎡ ⎤ ⎞ ⎫ ⎤
⎨ RT
1 ⎬
TOCvalue = ⎣ ⎝PP + ⎣ MCi + OCi ⎦ − S (i)⎠ × × M⎦ . (12)
⎩ (1 + r) 12 ⎭
i
i=1
Figure 5. Expected total ownership cost.
Results and Discussion

Figure 6 shows the results when MATLAB (MathWorks, Natick, MA) software is used to
enable a variation of the parameter RT of Eq. (12) for a planned horizon of 120 months.
This is done to identify the optimal replacement time (ORT) of a drilling machine that
minimizes TOCvalue . The figure shows the TOCvalue versus a different replacement time RT.
To show the behavior of the optimization curve for a period more than the planned
horizon, we assume that the optimization is used for a new finite time horizon of 240
months; see Figure 7. The total ownership cost for each operating month of the new
planned horizon (i.e., i = 1, 2, 3, . . . , 240) is computed by using the total ownership cost
Figure 6. Total ownership cost versus replacement time of existing drilling machine.
Figure 7. Optimal replacement time of existing drilling machine.
function obtained from Figure 5. This function is the fit of the calculated total ownership
cost over the machine’s old planned horizon (i.e., 120 months).
As is evident, the absolute lowest possible TOCvalue can be achieved by replacing the
machine with an identical new one every 115 months. However, it must be noted that RT
= 115 months generates the absolute minimum cost. As Figures 6 and 7 also show, within
that, there is a range (e.g., 110–122 months) when the minimum TOCvalue can still be
achieved in practice. In this study, we call it the optimum replacement range. Finding the
optimum replacement range is an important result of our study because it can help users in
their planning. A decision to replace equipment before or after this optimum replacement
range incurs greater cost for the company. The use of a lower replacement age (i.e., less
than 110 months) incurs higher costs due to the high investment cost. Meanwhile, if the
lifetime of the machine exceeds the upper limit of this range (i.e., more than 122 months),
losses will increase for two reasons:
1. The cost of operation and maintenance increases when the operating time increases
due to machine degradation.
2. The machine’s resale value will decrease each month of operation until it reaches
its scrap value at the end of its planned lifetime.
Sensitivity Analysis
We next perform a sensitivity analysis to identify the effect of purchase price and operating
and maintenance costs on the ORT of the drilling machine. However, because most of
the factors may be interrelated, we use a multisensitivity analysis to identify the effect of
multiple changes of cost factors.
Figure 8. Effect of increasing purchase price.
Single-Variable Sensitivity Analysis

A single-variable sensitivity analysis varies one factor and keeps the others constant. The
factors considered in our sensitivity analysis include machine purchase price, as well as
operating and maintenance costs. Figure 8 illustrates the effect of an increasing purchase
price on the ORT of the drilling machine.
Figure 8 shows that the ORT is an increasing step function of PP (based on the per-
centage of purchase price); the ORT remains constant for a specific range of PP increments
and then increases stepwise. As an example, if the purchase price increases from 1 to 4%,
the ORT is constant. This means that the ORT increases stepwise at specific PP percentage
increments; that is, 5, 12, 19, 26, 34, and 42%.
Figure 9 illustrates the effect of decreasing machine operating cost (based on the
percentage of operating cost) on the ORT. It is obvious that when the machine’s operating
cost decreases, the ORT will increase stepwise, although it remains constant within a
specific range of decreasing OC. This means that the ORT is not sensitive to a specific
range of operating cost reductions and will increase stepwise at a specific OC rate of
reduction; that is, 15 and 34%.
Figure 10 illustrates the effect of decreasing machine maintenance costs on the ORT
of the drilling machine. When the maintenance cost decreases, the ORT will increase as a
step function of MC reduction. In addition, note that the ORT increases at reduction steps
of MC—that is, 7, 15, 23, 30, 36, 42, and 48%—and remains constant within these steps.
Figures 8, 9, and 10 show that with increasing purchase price and decreasing operating
and maintenance costs, the ORT of a new model of this machine will increase stepwise at a
Figure 9. Effect of decreasing operating cost.

Figure 10. Effect of decreasing maintenance cost.
specific percentage of these factors. This may occur because there is a significant effect of
these factors on the total ownership cost at these specific percentages of increasing factor
of purchase price (IFPP), reduction factor of operating cost (RFOC), and reduction factor
of maintenance cost (RFMC).
Multivariable Sensitivity Analysis

To increase our understanding of the correlation of input and output variables in the op-
timization model, a multisensitivity analysis was performed considering three different
cases. MATLAB software was used to enable a variation of the three factors, IFPP, RFMC
and RFOC, to show their effects on the ORT of the drilling machine. In all three cases, the
purchase price increases while the operating and maintenance costs decrease. Case 1 repre-
sents the effect of decreasing machine maintenance costs while increasing purchase price
and decreasing operating costs at different percentages at the same time. Figure 11 shows
the correlation between decreasing machine maintenance cost and increasing purchase
price for a given 15% reduction in the cost of operation. As the figure shows, decreasing
maintenance cost while increasing purchase price has a positive effect on increasing the
machine’s optimal replacement time.
Case 2 studies the effect of increasing machine purchase price while simultaneously
decreasing the maintenance cost at a given percentage of operating cost reduction. Case 3
considers the effect of decreasing the machine’s operating cost while decreasing mainte-
nance cost at a given percentage of increasing purchase price.
Figure 11. Effect of RFMC and IFPP for a given 15% RFOC.
From the results of the three cases (Figure 11), it is clear that the ORT of a new
model of this machine will increase as a result of increasing its purchase price, decreasing
the maintenance cost, and decreasing the cost of operation at different percentages. The
explanation is that a new model of this machine is assumed to be more reliable than the
old ones. This will lead to a decreased failure rate in a new model of this machine, which,
in turn, reduces the maintenance cost. In addition, a new model of this machine is more
productive than an old one; thus, it will finish the same job in less time. This will decrease
the energy consumption of a new model of this machine, which leads to a reduction in the
operating cost of it.
Regression Analysis
Our regression analysis of the ORT results obtained from the previous three cases uses
Minitab (Minitlab Inc., State College, PA) software and the least squares method. The
ORT of a new model of drilling machine is modeled as a linear function of IFPP, RFOC,
and RFMC. IFPP is defined as the percentage increment on the machine’s purchase price.
RFOC is the percentage reduction in the machine’s operating cost, and RFMC is the
percentage reduction in the machine’s maintenance cost. The regression analysis results in
the following mathematical model:
ORT = 114 + 0.133 × IFPP + 0.0682 × RFOC + 0.164 × RFMC, (13)
where IFPP = 5%, RFOC = 6%, and RFMC = 12%. The ORT resulting from the regression
model is calculated as follows:
ORT = 114 + 0.133 × 5 + 0.0682 × 6 + 0.164 × 12 = 117 (month).
The ORT obtained from the regression model is compatible with the values shown in
Figure 11. The other values of IFPP, RFOC, and RFMC can be calculated and checked as
well.
The high R2 adjusted value obtained from regression analysis, R2 (adj.) = 98.6%,
indicates that the ORT of a new model of this machine depends linearly on the IFPP, RFOC,
and RFMC. Following the results of the sensitivity and regression analyses, the rank of the
factors affecting the ORT of a new vintage model of a drilling machine is as follows:
1. The reduction in maintenance cost.
2. The increase in purchase price.
3. The reduction in operating cost.
Many studies have considered reliability, maintainability, and optimum replacement
decisions; readers are referred to, for example, Ahmadi and Kumar (2011), Wijaya et al.
(2012) and Dandotiya and Lundberg (2012) for further studies in the recent literature.
Graphical User Interface

During the study, we noticed that the user company is not always able to go through
the process introduced here. Therefore, to facilitate the decision-making process and to
enhance the company’s ability to make the right decision at the right time, we developed
a graphical user interface (GUI) to compute the ORT. The proposed GUI is designed to
enable checking of the effect of changing any of the factors; that is, IFPP, RFOC, or RFMC.
Figure 12 represents the GUI for case 1.
Figure 12. Graphical user interface.
The selected input factors appear on the left side of Figure 12; the program calculates
the ORT of the machine according to the selected input. The generated fields shown on
the right of the figure represent the ORT values, calculated after applying the proposed
optimization model. A plot representing the ORT trend appears in the figure’s central
column. From this, decision makers can determine the best time economically to buy a
new machine. They can choose one of three factors: purchase price, operating cost, or
maintenance cost. They can determine its effect on the ORT by observing the plot on the
interface. This method also provides decision makers with useful information if they are
negotiating with manufacturers over the purchase price of a new model of this machine.
Concluding Remarks
This article presents a comprehensive and practical approach that can be used to provide
the optimal replacement time of an underground mining drilling machine. The following
conclusions can be derived from this study:
1. Although many other models require reliability and failure data to identify the
optimum replacement age, the approach presented herein is based on financial data
on the purchase price, operating and maintenance costs, and the machine’s resale
value. This makes it very practical for industries.
2. According to the results obtained from the optimization curve, the absolute ORT of
the drilling machine at the case study’s mine is 115 months of operation. However,
the ORT has a practical range of 110 to 122 months, during which the total ownership
cost remains almost constant. This means that the company has the flexibility to
make replacements within the optimum replacement age range; that is, 12 months.
Therefore, there is no fixed date or age at which the TOCvalue is minimum. In
general, a range of months provides the minimum TOCvalue .
3. The results of the sensitivity analysis indicate that increasing the purchase price and
decreasing the operating and maintenance costs have a positive effect on increasing
the ORT.
4. The results of the regression analysis show that the ORT of the new machine depends
linearly on its IFPP, RFOC, and RFMC. These results confirm the computation and
the results of the sensitivity analysis.
5. The results of regression analysis show that the reduction in maintenance cost has
the largest impact on the ORT, followed by the increase of purchase price and
reduction of operating cost. Hence, the manufacturer must make a greater effort
to improve the reliability and maintainability of the drilling machine to reduce the
costs associated with maintenance and to increase the ORT. However, a detailed
RAMS analysis is required to identify the weakest points of the machine from
reliability, maintainability and supportability points of views.
6. Economists at the user company can easily use the GUI to estimate the ORT of a
new machine and see the behavior of its ORT at IFPP, RFOC, and RFMC. These
factors will provide a clear view of the ORT of the new machine. Knowing this will
help the user company determine when to buy a new machine and assist them in
any negotiations with the manufacturer over the purchase price.
Acknowledgments
The people at Boliden AB and Atlas Copco who helped in this research are gratefully
acknowledged. The authors also thank Behzad Ghodrati for his help. My sincerest gratitude
is extended to the reviewers and the editor of this journal for the valuable comments that I
received from them, which helped to improve this article.
Funding
The authors thank Atlas Copco for their financial support.
Nomenclature
a Purchase price percentage at first day of operation (%)
BV1 Machine’s value at first day of operation (cu)
LCc Labor cost for corrective maintenance (cu)
LCp Labor cost for preventive maintenance (cu)
M Number of replacement cycles
S(i) Resale value (cu)
SPc Spare part cost for corrective maintenance (cu)
SPp Spear part cost for preventive maintenance (cu)
T Planned lifetime (month)
TOCvalue Total ownership cost multiplied by number of replacement cycles (cu)
References
Ahmadi, A. and Kumar, U. (2011) Cost based risk analysis to identify inspection and restoration
intervals of hidden failures subject to aging. IEEE Transactions on Reliability, 60(1), 197–209.
Asiedu, Y. (1998) Product life cycle cost analysis: state of the art review. International Journal of
Production Research, 36(4), 883–908.
Atlas Copco Rock Drills AB. (2010) Atlas Copco Boomer L1C, L2C Mk 7B operator’s instructions.
Atlas Copco Rock Drills AB, Örebro, Sweden.
Bellman, R. (1955) Equipment replacement policy. Journal of Society for Industrial and Applied
Mathematics, 3(3), 133–136.
Bethuyne, G. (1998) Optimal replacement under variable intensity of utilization and technological
progress. The Engineering Economist, 43(2), 85–105.
Boucekkine, R., Germain, M. and Licandro, O. (1997) Replacement echoes in the vintage capital
growth model. Journal of Economic Theory, 74(2), 333–348.
Cooley, T., Greenwood, J. and Yorukoglu, M. (1997) The replacement problem. Journal of Monetary
Economics, 40(3), 457–499.
Dandotiya, R. and Lundberg, J. (2012) Economic model for maintenance decision: a case study for
mill liners. Journal of Quality in Maintenance Engineering, 18(1), 79–97.
Elton, E.J. and Gruber, M. J. (1976) On the optimality of an equal life policy for equipment subject
to technological improvement. Operational Research Quarterly, 27(1), 93–99.
Eschenbach, T. (2010) Engineering economy: applying theory to practice, 3rd ed. Oxford University
Press, New York.
Hall, B. (2007) A lifetime of equipment optimisation. Australian Mining, 99(5), 58–59.
Hartman, J.C. (2005) A note on “A Strategy for Optimal Equipment Replacement.” Production
Planning & Control, 16(7), 733–739.
Hartman, J. C. and Murphy, A. (2006) Finite-horizon equipment replacement analysis. IIE Transac-
tions, 38(5), 409–419.
Hartman, J. C. and Tan, C. H. (2014) Equipment replacement analysis: a literature review and
directions for future research. The Engineering Economist, 59(2), 136–153.
Hritonenko, N. (2005) Optimization analysis of a nonlinear integral model with applications to
economics. Nonlinear Studies, 12(1), 59–72.
Hritonenko, N. and Yatsenko, Y. (2003) Applied mathematical modeling of engineering problems.
Kluwer Academic Publishers, New York.
Hritonenko, N. and Yatsenko, Y. (2007) Optimal equipment replacement without paradoxes: a con-
tinuous analysis. Journal of Operations Research Letters, 35(2), 245–250.
Hritonenko, N. and Yatsenko, Y. (2008) The dynamics of asset lifetime under technological change.
Journal of Operations Research Letters, 36(5), 565–568.
Hritonenko, N. and Yatsenko, Y. (2009) Integral equation of optimal replacement: analysis and
algorithms. Journal of Applied Mathematical Modelling, 33(6), 2737–2747.
Jardine, A. and Tsang, A. (2006) Maintenance, replacement, and reliability theory and application.
Taylor & Francis Group, New York.
Kärri, T. (2007) Timing of capacity change: models for capital intensive industry. Lappeenranta
University of Technology, Lappeenranta.
Luderer, B., Nollau, V. and Vetters, K. (2010) Mathematical formulas for economists. Springer Verlag,
Berlin.
Mardin, F. and Arai, T. (2012). Capital equipment replacement under technological change. The
Engineering Economist, 57(2), 119–129.
Scarf, P. A. and Bouamra, O. (1999) Capital equipment replacement model for a fleet with variable
size. Journal of Quality in Maintenance Engineering, 5(1), 40–49.
Wijaya, A. R., Lundberg, J. and Kumar, U. (2012) Robust-optimum multi-attribute age-based re-
placement policy. Journal of Quality in Maintenance Engineering, 18(3), 325–343.
Yatsenko, Y. (2005) Discrete and continuous-time modeling of optimal equipment replacement, Paper
read at the 2005 International Conference on Scientific Computing, Las Vegas, Nevada.
Yatsenko, Y. and Hritonenko, N. (2005) Optimization of the lifetime of capital equipment using
integral models. Journal of Industrial and Management Optimization, 1(4), 415–432.
Biographical Sketches
Hussan Al-Chalabi received a B.Sc.Eng. degree in mechanical engineering from Mosul University,
Iraq, in 1994 and an M.Sc degree in mechanical engineering in thermal power from Mosul University,
Iraq, in 2008. Then he joined the Department of Mechanical Engineering at Mosul University as a
lecturer. In 2011, he joined the Division of Operation, Maintenance and Acoustics at Luleå University
of Technology as a doctoral student.
Jan Lundberg is Professor of Machine Elements at Luleå University of Technology and also
Professor in Operation & Maintenance with a focus on product development. During 1983–2000, his
research concerned engineering design in the field of machine elements in industrial environments.
During 2000–2006, his research concerned industrial design, ergonomics, and related problems as
cultural aspects of design and modern tools for effective industrial design in industrial environments.
Since 2006 his research is completely focused on maintenance issues including methods for measuring
failure sources, how to design out maintenance, and how to design for easy maintenance.
Alireza Ahmadi is an assistant professor in the Division of Operation and Maintenance Engineering,
Luleå University of Technology (LTU), Sweden. He received his Ph.D. degree in operation and
maintenance engineering in 2010. Alireza has more than 10 years of experience in civil aviation
maintenance as a licensed engineer and production planning manager. His research topic is related to
the application of RAMS and maintenance optimization.
Adam Jonsson is a senior lecturer in the Department of Engineering Sciences and Mathematics,
Luleå University of Technology, Sweden. He received his Ph.D. in statistics in 2008. Jonsson’s
research is in applied probability and social welfare economics.
PAPER III
Economic lifetime prediction of a mining drilling machine

using an artificial neural network
Al-Chalabi, H., Ahmadzadeh, F., Lundberg, J., Ghodrati, B., 2014. Economic lifetime
prediction of a mining drilling machine using an artificial neural network. Published in the
International Journal of Mining, Reclamation and Environment, 28(5), 311-322.
http://dx.doi.org/10.1080/17480930.2014.942519
International Journal of Mining, Reclamation and Environment, 2014
Vol. 28, No. 5, 311–322, http://dx.doi.org/10.1080/17480930.2014.942519
Economic lifetime prediction of a mining drilling machine using an

artificial neural network
Hussan Al-Chalabia,b*, Farzaneh Ahmadzadehc, Jan Lundberga and Behzad Ghodratia
a
Division of Operation, Maintenance and Acoustics, Luleå University of Technology, Luleå,
Sweden; bMechanical Engineering Department, College of Engineering, University of Mosul,
Mosul, Iraq; cDivision of Product Realization, Mälardalen University, Eskilstuna, Sweden
(Received 15 June 2014; accepted 4 July 2014)
This study develops models for predicting the economic lifetime of drilling
machines used in mining. It uses three cases, each represented by a MATLAB code,
to develop an optimisation model. The resulting ORT is fed as input to an artificial
neural network (ANN) and the results translated into a relatively simple equation.
The study finds that increasing the purchase price and decreasing the operating and
maintenance costs will increase a machine’s ORT linearly. Decreased maintenance
cost has the largest impact on ORT, followed by increased purchase price and
decreased operating cost. The ANN method gives a series of basic weight and
response functions which can be made available to any engineer without the use of
complicated software. It also helps decision-makers determine the best time econom-
ically to replace an old machine with a new one; thus, it can be extended to more
general applications in the mining industry.
Keywords: artificial neural network; decision support system; mining drilling
machine; optimisation model; replacement time
1. Introduction
Mining is a large, integrated, automated and complex industry; having safe, reliable
and cost-effective production is a strategic necessity if a company is to meet customer
requirements and gain a competitive advantage globally. Mining industries are under
constant pressure to improve their production performance at minimum cost. On the
one hand, they need more reliable capital equipment with higher performance capabili-
ties; on the other hand, these are more expensive.
Drilling machines are a key element of mining production, especially in the mineral
extracting process (see Figure 1). These machines are subject to degradation throughout
their operating life; this reduces production rates, with a resulting negative economic
effect. In addition, because drilling machines are used in underground mines, they are
subjected to a harsh working environment, thereby accelerating degradation. A key
question for mining companies is when to replace the drilling machines to minimise
cost and maximise profit.
As operating and maintenance costs are the main contributing factors to the total
cost, researchers concerned with cost optimisation are especially interested in the opti-
mum replacement time of production and repairable equipment, defined as the time at
*Corresponding author. Email: hussan.hamodi@ltu.se
© 2014 The Author(s). Published by Taylor & Francis.

This is an Open Access article. Non-commercial re-use, distribution, and reproduction in any medium, provided the original
work is properly attributed, cited, and is not altered, transformed, or built upon in any way, is permitted. The moral rights of
the named author(s) have been asserted.
Figure 1. Drilling machine.
which the total cost is minimised [1]. For example, Ahmadi and Kumar [2] developed
a cost rate function to identify the optimum interval and frequency of inspection and
restoration of an aircraft’s repairable but ageing components whose failures are hidden.
Wijaya et al. [3] proposed a robust-optimum multi-attribute age-based replacement pol-
icy. A model for estimating the lifetime of mill liners was developed by Dandotiya and
Lundberg [4] based on the properties of the ore used. Their proposed lifetime model is
combined with a replacement interval model to determine the optimum replacement
interval for the mill liners; it considers process parameters of multiple ore types.
Some researchers have considered the optimal lifetime of capital equipment using
economic theories and vintage capital models, represented mathematically by non-linear
Volterra integral equations with unknown limits of integration [5–8]. Others have used
the theory of dynamic programming to consider technological changes under finite and
infinite time horizons [9–12]. For instance, Yatsenko and Hritonenko [13] studied the
lifetime optimisation of capital equipment using integral models; they designed a gen-
eral investigation framework for optimal control of the models. Hartman and Murphy
[14] suggested a dynamic programming approach to the finite time-horizon equipment
replacement problem with stationary cost. Their model considers the relationship
between the infinite time-horizon solution (continuous replacement of equipment at the
end of its economic lifetime) and the finite time-horizon solution. Hritonenko and
Yatsenko [15] constructed a computational algorithm to solve a nonlinear integral equa-
tion; the solution helps determine the optimal policy of equipment replacement given
the technological advances. Finally, Kärri [16] considered the optimal replacement time
of an old machine by modelling the costs of the old machine with simple linear func-
tions. The resulting optimisation model minimises the machine cost and maximises the
profit. It can handle capacity expansion and replacement situations.
The artificial neural network (ANN) has been widely touted as solving many fore-
casting and maintenance decision modelling problems. For example, Gebraeel et al.
[17] proposed a set of neural network models using exponential approximations to esti-
mate the remaining life of thrust ball bearings in real time. Later, Wu et al. [18] devel-
oped an integrated neural network-based decision support system for predictive
maintenance of rotational equipment. The integrated system is platform-independent
International Journal of Mining, Reclamation and Environment 313
and is aimed at minimising the expected cost per unit operational time. Jafar et al. [19]
used an application of ANN to model the failure rate and estimate the optimal replace-
ment time for individual pipes in an urban water distribution system.
Other researchers have studied the remaining life predictions of various components
using ANN [20–22]. For example, Ahmadzadeh and Lundberg [23] developed a method
which predicts the remaining useful life of grinding mill liners. Using this method,
there is no need to stop the mill, enter it and measure the liners’ wear to make appro-
priate maintenance decisions, leading to enormous monetary savings.
Although many studies have estimated the optimal lifetime of equipment consider-
ing the possibility of technological change, more practical methods of ORT prediction
are required. It is well known that ANNs are able to model easily any type of paramet-
ric or non-parametric process and automatically transform the input data into optimal
and accurate outputs, in this case, the ORT of a drilling machine. In addition, using a
neural network gives a clear picture of the relative importance of the input variables
and suggests how variations in inputs will affect the ORT. Finally, because the neural
network can be represented by a series of basic weight and response functions, the
results can be made available to any engineer without the need for complicated
software.
The present study develops an ANN-based optimisation model using available cost
data to estimate the ORT of a drilling machine. The available cost data are the machine
purchase price, maintenance costs and operating costs. The production loss cost of
machine downtime is not considered due to the availability of a spare drilling machine
in the case study. The overall goal of model development is to minimise costs in the
mining industry.
2. Data collection
Four years of cost data were collected in a Maximo computerised maintenance manage-
ment system. Since drilling is not a continuous process, the operating cost can be esti-
mated by considering the utilisation of the machine. Data include corrective
maintenance costs (CM), preventive maintenance costs (PM) and repair time. The cor-
rective and preventive maintenance costs comprise spare parts costs (SP) and labour
costs (LC), calculated as real costs without inflation.
3. Optimal replacement time prediction using ANN

ANNs can perform nonlinear modelling without prior information and are able to learn
complex relationships between inputs and outputs; the process is also fast. One of the
first successful applications of ANNs in forecasting was reported by Zhang et al. [24];
these researchers designed a feedforward ANN to accurately simulate a chaotic series.
In general, feedforward ANNs trained with the back propagation algorithm have been
found to perform better than classical autoregressive models for trend prediction in a
nonlinear time series [25,26].
In this study, ANN is applied to find nonlinear or linear relationships between inputs
(costs) and outputs (ORT) for a drilling machine. The proposed ANN model combines
multi-layer perceptions with the back-propagation Levenberg-Marquardt algorithm. In
function approximation problems, this algorithm is considered to have the fastest
convergence and is one of the most powerful learning algorithms [25,27].
The analyses are based on three cases, increasing purchase price (IPP), decreasing
operating cost (DOC) and decreasing maintenance cost (DMC), each represented by a
MATLAB code. The resulting ORT is fed as input to ANN and the results translated
into a relatively simple equation. The equation is transformed to an Excel spreadsheet
to make ORT estimation quick and easy for any engineer to apply.
As noted above, the proposed model has three inputs: the increased purchase price
(IPP (%)), decreased operating cost (DOC (%)) and decreased maintenance cost (DMC
(%)). A hidden layer with three neurons and a nonlinear transfer function allows the
network to learn nonlinear and linear relationships between input and output variables.
The number of neurons in the output layer is constrained to one, as the output only
requires one parameter, in this case, the ORT of the drilling machine.
The number and size of layers between network inputs and the output layer are
determined by testing several combinations of numbers of layers and various numbers
of neurons in each layer. Each of the selected combinations is tested with several differ-
ent initial conditions to guarantee the proposed model is the best solution. The structure
of the optimal ANN model is shown in Figure 2.
4. Calculation of the ANN inputs

The optimal value for costs is required to obtain the ORT using ANN, so the following
optimisation model is developed to find the replacement time (RT) that minimises the
total cost value (TCvalue ), as shown in Equation (6). The term ‘total cost value’ is
defined as the summation of the machine purchase price (PP), operating cost (OC),
maintenance cost (MC) and resale value S (t) over a long period, with replacements
occurring at intervals of n periods. Therefore,
MC ¼ CM þ PM (1)
CM ¼ SPc þ LCc (2)
PM ¼ SPp þ LCp (3)
S ðt Þ ¼ PP a ð1 DrÞt t ¼ 1; 2; 3; . . .; 120 ðmonthÞ (4)
Figure 2. Optimal structure of ANN model.

where a represents the percentage that is multiplied by the machine purchase price to
represent the machine’s value on the first day of use. During discussions, a group of
experts agreed the machine’s purchase price decreases by 10% at this time. The depre-
ciation rate that allows for full depreciation by the end of the planned lifetime of the
machine is modelled by Equation (5) [28]:
1
SV L
Dr ¼ 1 (5)
BV1
"( " # ) #
X
RT
TCvalue ¼ PP þ MCk þ OCk S ðt Þ N (6)
k¼1
where
T
N¼ (7)
RT
MATLABTMTM software is used to enable a variation of the parameter RT of Equation
(6) for the optimisation time horizon in order to identify the ORT of a drilling machine
that minimisesTCvalue . The results show that the lowest possible TCvalue can be achieved
by replacing the machine every 96 months.
When MATLAB codes are used to identify the effect of IPP, DOC and DMC on
the optimal age of replacement, in all cases, the purchase price increases while the
operating and maintenance costs decrease (e.g. Figure 3). Case one represents the effect
of IPP from (1–50)% with DOC and DMC at different percentages. Case two repre-
sents the effect of DOC from (1–50)% with DMC and IPP at different percentages.
Case three represents the effect of DMC from (1–50)%, with IPP and DOC at different
percentages. For greater clarity, part of case three is selected (Figure 3) to show the
simultaneous production of (1–50)% DMC and 7% DOC and IPP at different percent-
ages. For this case alone, 400 different input data for ANN are considered. The proce-
dure is repeated for all cases, producing 6150 data-sets.
The produced data-sets (6150 data) from these codes are treated as inputs for the
proposed ANN model to predict the ORT of the drilling machine at any percentage of
IPP or DMC and DOC.
Figure 3. Producing 400 input data for ANN as part of case three.
5. Training and testing the proposed ANN model

Pre-processing data by scaling improves the training of the neural network. To avoid a
slow rate of learning, specifically near the end points of the output range (due to the
property of the sigmoid function, which is asymptotic to values 0 and 1), the input and
output data are scaled between the interval 0.1 and 0.9 [29]. It should be noted that
any new input data should be scaled before being presented to the network and the cor-
responding predicted values should be un-scaled before use [30]. The linear scaling
equation is expressed by:

0:8 0:8Xmax
Xs ¼ X þ 0:9 (8)
D D
where Xs represents the scaled value of input variables and X represents the un-scaled
value of input variables. The input variables are IPP, DOC and DMC. Equation (8) is
used here for an IPP, DOC and DMC between minimum increasing or decreasing percent-
age 1% (Xmin) and maximum increasing or decreasing percentage 50% (Xmax), such that
D ¼ Xmax Xmin (9)

From the 6150 scaled data, 90% are used in the training of the neural network; see
Figure 4. The model shown in Figure 4 has very high values of R2 = 0.9962 for ANN.
However, as also shown in the figure, the neural network model yields outputs very
close to the desired targets with high accuracy.
After the training is completed, the network is tested for its learning and generalisation
capabilities. Its learning ability is tested by examining its ability to produce outputs for
the set of inputs (seen data) used in the training phase. Its generalisation ability is tested
by investigating its ability to respond to the input sets (unseen data) not included in the
training process; thus, 10% are used for tests with the obtained network (Figure 5). As
evident in Figure 5, the model has very high values of R2 = 0.99653 for ANN. However,
the neural network model also yields outputs very close to the desired targets with high
levels of certainty.
Figure 4. Learning capability: Output of the proposed ANN for seen data.

The proposed ANN model is used to design a formula to calculate the ORT of a drilling
machine. The structure of the optimal ANN model is shown in Figure 2; its connection
weights and threshold levels are summarised in Table 1.
The equation length depends on the number of nodes in the hidden layer. Adopting
three nodes gives an accuracy of 99%. The small number of connection weights of the
neural network enables the ANN model to be translated into a relatively simple formula,
in which the predicted ORT can be expressed as follows:
1
ORTs ¼ n o (10)
h7 þð
w4:7 1þe1x1 Þþðw5:7 1þe1x2 Þþðw6:7 1þe1x3 Þ
1þe
where ORTs represents the scaled ORT derived from the ANN model, θj represents the
output threshold and wij represents the weight from node i in the hidden layer to node j
in the output layer. Hence,
x1 ¼ h4 þ w4:1 IPP þ w4:2 DOC þ w4:3 DMC (11)

Note that before using Equations (10–13), all input variables need to be scaled between
0.1 and 0.9 using Equation (8). It should also be noted that the predicted ORTs
obtained from Equation (10) is scaled between 0.1 and 0.9; however, in order to obtain
the actual value of ORT, it must be re-unscaled using the following formula:
Figure 5. Generalisation capability: Output of the proposed ANN for unseen data.
Table 1. Weights and threshold levels for the proposed ANN.
wii weights from node ith input layer to

node jth hidden layer
Hidden layer nodes i=1 i=2 i=3 Hidden threshold (θj )
j=4 −0.0648 −0.03030 −0.1297 10.4780
j=5 −2.6814 −0.78166 −2.1337 0.18142
j=6 −1.2496 −0.8258 −2.3112 5.023854
Output layer nodes wij weights from node ith hidden layer to Output threshold (θj )
node jth output layer
i=4 i=5 i=6
j=7 7.795 −5.3591 −10.801 2.919081

D D
X ¼ Xs 0:9 þ Xmax (14)
0:8 0:8
Equation (14) is obtained by solving Equation (8), considering the input variable (X) as
unknown and the output variable (Xs) as known. An Excel spreadsheet can be used as
a substitute for fast and accurate calculation of the ORT. Equation 15 is applied to esti-
mate the actual value of the ORT of the drilling machine as follows:

ORTmax ORTmin ORTmax ORTmin
ORT ¼ ORTs 0:9 þ ORTmax (15)
0:8 0:8
where ORTmax and ORTmin represent the maximum and minimum values of ORT
6.1. Study of relative importance of input parameters

The method of partitioning weights, proposed by Garson [31] and adopted by Goh
[32], is now used to determine the relative importance of the various input parameters;
see Figure 6.
As evident in Figure 6, the most important parameter influencing the ORT of the
drilling machine is the DMC, followed by the IPP. The DOC has the least influence.
The design for reliability and maintainability should be adapted to reduce the mainte-
nance cost, thus increasing the ORT of the drilling machine.
Figure 6. Relative importance of the variables affecting the ORT of the drilling machine.
6.2. Study of variables influencing the ORT of the drilling machine

With applying Equation 15, the ORT of drilling machine is calculated for different
variables, including IPP, DOC and DMC, to see the influence of these variables. The
DMC is the most important factor, as shown in Figure 6. Generally, the ORT increases
with a decrease in maintenance cost, compatible with the results shown in Figures 7 and 8.
6.3. Effect of increasing machine purchase price

As Figure 6 shows, the second most influential variable on the ORT is the machine’s
IPP. The correlation of IPP and DMC when the cost of operating is reduced by 25% is
shown in Figure 7. As is evident, both IPP and DMC have a positive effect on increas-
ing the ORT.
Figure 7. Effect of IPP and DMC for 25% OC decrease.
Figure 8. Effect of DOC and DMC for 25% PP increase.

6.4. Effect of decreasing machine operating cost

Figure 8 shows the correlation of DOC and DMC for a given 25% increase in PP.
Clearly, both DOC and DMC have a positive effect on increasing the ORT.
7. Conclusions
The main advantage of the proposed neural network model is its ability to produce
acceptable results: the correlation between input and output variables is very high and
the accuracy is more than 99%. Moreover, the model’s performance is very consistent
for data used for training (seen) and testing (unseen). Therefore, it is very effective in
estimating and predicting the ORT of a mining drilling machine. Further, because the
ANN model uses a series of basic weight and response functions, it does not require
expensive and sophisticated equipment for data recording and analysis.
The predicted ORT from the ANN model indicates that increasing the purchase price
and decreasing the operating and maintenance costs will increase the ORT of the dril-
ling machine. Results also show that the maintenance cost has the largest impact on the
ORT, followed by the purchase price and operating cost. Hence, the manufacturer must
make a greater effort to improve the reliability and maintainability of the drilling
machine to reduce the costs associated with maintenance and increase the ORT.
This study presents a comprehensive and very practical approach using ANN which
can provide the economic replacement time of a drilling machine with higher levels of
certainty. It helps engineers and decision-makers determine when it is best to economi-
cally replace an old machine with a new one without using complicated software.
Because of its effectiveness and simplicity, the model can be extended to more general
applications in the mining industry. The dynamic nature of the proposed methodology
opens up future studies of economic lifetime prediction for different categories of
mining equipment.
Notations
ORT optimal replacement time (months)
ANN artificial neural network
CM corrective maintenance cost (CU)
CU currency unit
PM preventive maintenance cost (CU)
IPP increase purchase price (%)
DOC decrease operating cost (%)
DMC decrease maintenance cost (%)
RT replacement time (months)
TCvalue total cost value (CU)
PP purchase price (CU)
OC operating cost (CU)
MC maintenance cost (CU)
S(t) resale value (CU)
SPc spare part cost for corrective maintenance (CU)
LCc labour cost for corrective maintenance (CU)
SPp spare part cost for preventive maintenance (CU)
LCp labour cost for preventive maintenance (CU)
Dr depreciation rate
t machine lifetime (months)
SV scrap value (CU)
BV1 booking value on first day of operation (CU)
L planned lifetime (month)
N number of machine replacements
T optimisation time horizon (120 months)
Xs scaled value of input variables
Δ difference between maximum and minimum values of input variables
X unscaled value of input variables
Xmax maximum value of input variables
Xmin minimum value of input variables
θj output threshold
wij weight from node i in the hidden layer to node j in the output layer
Acknowledgement
The authors would like to thank Boliden AB, Atlas Copco for supporting this research. Special
appreciation is extended to the operating and maintenance engineers at Boliden AB for sharing
their valuable knowledge, experience and data to improve this study. The authors would also like
to thank Majid Al-Gburi, Alireza Ahmadi for their help.
References
[1] A.K.S. Jardine and A.H.C. Tsang, Maintenance, Replacement, and Reliability Theory and
Applications, Taylor & Francis Group, New York, 2006.
[2] A. Ahmadi and U. Kumar, Cost based risk analysis to identify inspection and restoration
intervals of hidden failures subject to aging, IEEE Trans. Reliab. 60 (2011), pp. 197–209.
[3] A.R. Wijaya, J. Lundberg, and U. Kumar, Robust-optimum multi-attribute age-based
replacement policy, J. Qual. Maint. Eng. 18 (2012), pp. 325–343.
[4] R. Dandotiya and J. Lundberg, Economic model for maintenance decision: a case study for
mill liners, J. Qual. Maint. Eng. 18 (2012), pp. 79–97.
[5] R. Boucekkine, M. Germain, and O. Licandro, Replacement echoes in the vintage capital
growth model, J. Econ. Theory. 74 (1997), pp. 333–348.
[6] T.F. Cooley, J. Greenwood, and M. Yorukoglu, The replacement problem, J. Monetary Econ.
40 (1997), pp. 457–499.
[7] N. Hritonenko and Y. Yatsenko, Applied Mathematical Modeling of Engineering Problems,
Massachusetts: Kluwer Academic Publishers, 2003.
[8] N. Hritonenko, Optimization analysis of a nonlinear integral model with applications to eco-
nomics, Nonlinear Stud. 12 (2005), pp. 59–72.
[9] R. Bellman, Equipment replacement policy, J. Soc. Ind. Appl. Math. 3 (1955), pp. 133–136.
[10] E.J. Elton and M.J. Gruber, On the optimality of an equal life policy for equipment subject
to technological improvement, J. Oper. Res. Q. 27 (1976), pp. 93–99.
[11] G. Bethuyne, Optimal replacement under variable intensity of utilization and technological
progress, Eng. Econ. 43 (1998), pp. 85–105.
[12] N. Hritonenko and Y. Yatsenko, The dynamics of asset lifetime under technological change,
Oper. Res. Lett. 36 (2008), pp. 565–568.
[13] Y. Yatsenko and N. Hritonenko, Optimization of the lifetime of capital equipment using inte-
gral models, J. ind. manage. optim. 1 (2005), pp. 415–432.
[14] J.C. Hartman and A. Murphy, Finite-horizon equipment replacement analysis, IIE Trans. 38
(2006), pp. 409–419.
[15] N. Hritonenko and Y. Yatsenko, Integral equation of optimal replacement: Analysis and
algorithms, Appl. Math. Model. 33 (2009), pp. 2737–2747.
[16] T. Kärri, Timing of Capacity Change: Models for Capital Intensive Industry, Lappeenranta:
Lappeenrannan teknillinen yliopisto/Lappeenranta University of Technology, 2007.
[17] N. Gebraeel, M. Lawley, and R. Liu, Vibration-based condition monitoring of thrust
bearings for maintenance management, Intell. Eng. Syst. Artif. Neural Networks 12 (2002),
pp. 543–551.
[18] S. Wu, N. Gebraeel, M.A. Lawley, and Y. Yih, A neural network integrated decision
support system for condition-based optimal predictive maintenance policy, IEEE Trans. Syst.
Man Cybern. Part A Syst. Humans 37 (2007), pp. 226–236.
[19] R. Jafar, I. Shahrour, and I. Juran, Application of Artificial Neural Networks (ANN) to model
the failure of urban water mains, Math. Comput. Model. 51 (2010), pp. 1170–1180.
[20] C.S. Byington, M. Watson, and D. Edwards, Data-driven neural network methodology to
remaining life predictions for aircraft actuator components, IEEE Aerosp. Conf. Proc. 6
(2004), pp. 3581–3589.
[21] M. Mazhar, S. Kara, and H. Kaebernick, Remaining life estimation of used components in
consumer products: Life cycle data analysis by Weibull and artificial neural networks,
J. Oper. Manage. 25 (2007), pp. 1184–1193.
[22] F. Ahmadzadeh and J. Lundberg, Application of multi regressive linear model and neural
network for wear prediction of grinding mill liners, Int. J. Adv. Comput. Sci. Appl. 4
(2013), pp. 53–58.
[23] F. Ahmadzadeh and J. Lundberg, Remaining useful life prediction of grinding mill liners
using an artificial neural network, Minerals Eng. 53 (2013), pp. 1–8.
[24] G. Zhang, B. Eddy Patuwo and M.Y. Hu, Forecasting with artificial neural networks, Int.
J. Forecast. 14 (1998), pp. 35–62.
[25] S. Hykin, Neural Networks: A Comprehensive Foundation, 2nd ed., New Jersey: Printice-
Hall, 1999.
[26] R. Yam, P. Tse, L. Li, and P. Tu, Intelligent predictive decision support system for condi-
tion-based maintenance, Int. J. Adv. Manuf. Technol. 17 (2001), pp. 383–391.
[27] G.P. Zhang and M. Qi, Neural network forecasting for seasonal and trend time series, Eur.
J. Oper. Res. 160 (2005), pp. 501–514.
[28] B. Luderer, V. Nollau, and K. Vetters, Mathematical formulas for economists, Berlin:
Springer Heidelberg, 2010.
[29] A.W.C. Oreta, Simulating size effect on shear strength of RC beams without stirrups using
neural networks, Eng. Struct. 26 (2004), pp. 681–691.
[30] S.T. Yousif, Artificial neural network modeling of elasto-plastic plates, Ph.D. thesis, College
of Engineering, Mosul University, 2007.
[31] G.D. Garson, Interpreting neural-network connection weights, Artif. Intell. 6 (1991),
pp. 46–51.
[32] A. Goh, Back-propagation neural networks for modeling complex systems, Artif. Intell. Eng.
9 (1995), pp. 143–151.
PAPER IV
Model for economic replacement time of mining production

rigs including redundant rig costs
Al-Chalabi, H., Lundberg, J., Al-Gburi, M., Ahmadi, A., Ghodrati, B., 2014. Model
for economic replacement time of mining production rigs including redundant rig costs. Submitted for
publication in the Journal of Quality in Maintenance Engineering.
Model for economic replacement time of mining production rigs including
redundant rig costs
Hussan Al-Chalabia,b, Jan Lundberga, Majid Al-Gburic, Alireza Ahmadia and Behzad Ghodratia
a
Division of Operation, Maintenance and Acoustics. Luleå University of Technology. Luleå, Sweden.
b
Mechanical Engineering Department, College of Engineering, University of Mosul, Mosul, Iraq.
c
Division of Structural Engineering and Production. Luleå University of Technology. Luleå, Sweden.
Abstract
Purpose - This paper presents a practical model to determine the economic replacement time
(ERT) of production machines. The objective is to minimise the total cost of capital
equipment, where total cost includes acquisition, operating, maintenance costs and costs related
to the machine’s downtime. The costs related to the machine’s downtime are represented by
the costs of using a redundant machine.
Design/methodology/approach - Four years of cost data are collected. Data is analysed,
practical optimisation model is developed and regression analysis is done to estimate the drilling
rigs ERT. The artificial neural network (ANN) technique is used to identify the effect of
factors influencing the ERT of the drilling rigs.
Findings - The results show that the redundant rig cost has the largest impact on ERT,
followed by acquisition, maintenance and operating costs. The study also finds that increasing
redundant costs per hour have a negative effect on ERT, while decreases in other costs have a
positive effect. Regression analysis shows a linear relationship between the cost factors and
ERT.
Practical implications - The proposed approach can be used by the decision maker in
determining the economic replacement time of production machines which used in mining
industry.
Originality/value - The research proposed in this paper provides and develops an
optimisation model for economic replacement time of mining machines. This research also
identifies and explains the factors that have the largest impact on the production machine’s
ERT. This model for estimating the ERT has never been studied on mining drilling rigs.
Keywords Decision support model, Life cycle cost, Optimisation, Replacement time
Paper type Research paper
Abbreviations
ERT Economical replacement LCCi Labour cost for corrective maintenance (cu)
time (month)
ANN Artificial neural network SPVi Spare part value (cu)
IAC Increasing acquisition cost SPLi Spare part logistic cost (cu)
(%)
DOC Decreasing operating cost rt Repair time (h)
(%)
DMC Decreasing maintenance cost nl Number of labours
(%)
DRC Decreasing redundant rig cl Man hour cost (cu /h)
cost (%)
RT Replacement time (month) SPPi Spare part cost for preventive maintenance
(cu)
ERTs Scaled economical LCPi Labour cost for preventive maintenance (cu)
replacement time
TC Total cost (cu) RCi Redundant rig cost (cu)
1
cu Currency unit PTi Using time of redundant rig (h)
AC Acquisition cost (cu) CRi Redundant rig cost per hour (cu/h)
i Time period (month) TRi Logistic time for redundant rig (h)
MCi Maintenance cost (cu) TFi Restoring time of faulty rig to operation (h)
OCi Operating cost (cu) T1i Moving time of redundant rig from its
location to production point (h)
COi Compensation cost (cu) T2i Moving time of redundant rig from
production point to its original location (h)
Si Resale value (cu) TMi Moving time of faulty rig from production
point to workshop (h)
r Discount rate (%) TWi Time in workshop of faulty rig (h)
N Number of replacement TLi Moving time of repaired rig from workshop
cycles to production point (h)
CM i Corrective maintenance cost tdi Delay time in workshop of faulty rig before
(cu) repair (h)
PM i Preventive maintenance cost tri Actual repair time of faulty rig (h)
(cu)
SPCi Spare part cost for corrective t Ii Idle time in workshop of faulty rig after
maintenance (cu) repair (h)
1. Introduction
Industrial companies, or more speci¿cally, mining companies put huge funds, often millions of
dollars into their annual budgets to purchase heavy mobile equipment (HME) such as drilling
rigs, scaling rigs, wheel dozers, wheel loaders, dump trucks, etc. Given the enormous costs of
acquiring, operating, and maintaining their HME, it is important for companies to optimise
their replacement and procurement strategies (Richardson et al., 2013). As the HME operating
hour’s rise, so too do the maintenance and operating costs. At some point in the equipment’s
life span, these costs will be too high; it will no longer be economically viable to continue
using the old equipment, so it should be replaced (Verheyen, 1979). An essential economic
consideration in industrial companies is to find a model that can discriminate this point (i.e. the
point at which the equipment replacement time is expected to yield minimal life-cycle cost).
Obviously, for mining companies, one of the most important decisions is determining the
ERT of capital equipment; this can be done with the help of life-cycle cost (LCC) analysis.
The main reason for the increasing use of the life cycle costing concept for HME is that at
some point the operating and maintenance costs will exceed their acquisition costs.
In general, LCC is determined by summing up all potential costs associated with equipment
over its life time (i.e. the total of ownership and acquisition costs). It is well known that the
value of expenditure today costs more than the same expenditure next year because of the
decreasing “time value of money”. In this study we use a discount rate to account for the time
value of money. To compare costs incurred at different times, we must shift expenditure to a
reference point in time. Thus, we calculate the present equivalent value of the costs by
considering the discount rate factor.
1.1 Literature review

Standard models for ERT decisions contain an estimation of the discounted costs by
minimising the total cost of the equipment. The assumption of these models is that equipment
will be replaced at the end of its economic lifetime by a continuous sequence of identical
equipment (Hartman and Tan, 2014). Bellman (1955) developed the first optimal asset
replacement model for variable lifetime of assets. Wagner (1975) offered dynamic
2
programming formulation for the equipment replacement problem in which the state of the
system is the time period and the decision at each period is to keep the equipment for N
periods. His formulation has been extended by researchers to deal with realities of
technological changes, for example, see (Oakford et al., 1984; Bean et al., 1985; Hartman and
Rogers, 2006; Hritonenko and Yatsenko, 2008). These authors assumed a finite horizon in
their approaches to the problem of equipment replacement under non-stationary costs. In
1976, Elton and Gruber showed that an equal life policy was optimal on an infinite horizon
under technological changes. In contrast, Hartman and Murphy (2006) studied an asset
replacement problem for a stationary finite horizon; they illustrated how a bound on the
number of times an asset is retained at its economic life can be obtained, thus suggesting it is
optimal to replace the asset at its economic lifetime.
Dynamic programming models have been utilised in real cases of calculating equipment
replacement time because of the important uncertainties associated with life cycle costs
(Richardson et al., 2013). The net present value of all life cycle costs associated with an in¿nite
sequence of equipment life cycles has also been used to make equipment replacement decisions
(Bethuyne, 1998; Scarf and Bouamra, 1999; Hartman, 2005; Yatsenko and Hritonenko, 2005).
Other researchers have used different equipment replacement models to analyse a variety of
equipment, such as forklifts, buses, and aircraft (Eilon et al., 1966; Keles and Hartman, 2004;
Bazargan and Hartman, 2012). Although Tanchoco and Leung (1987) found replacement
decisions could be influenced by capacity considerations, others have noted that technological
changes can encourage decision makers to utilise equipment beyond its economic life
(Cheevaprawatdomrong and Smith, 2003). Still other researchers have considered reliability,
maintainability and optimum replacement decisions; readers are referred to, e.g., Wijaya et al.
(2012); Dandotiya and Lundberg, (2012), Golmakani and Pouresmaeeli, (2014) and Al-Chalabi
et al. (2014) for further discussion of the recent literature.
1.2 Aim of the study

Blanchard et al., 1995 mentioned that the costs associated with equipment support, operation
and maintenance can account for more than 75% of the equipment LCC. Thus, careful
consideration must be given to estimating the capital equipment ownership costs. Given the
importance of operating and maintenance (O&M) costs and in order to assure a specific level
of availability, industries must consider using redundant equipment to overcome production
loss when a failure occurs. Another way to ensure production performance is to make a
pooling agreement with other companies, renting commonly owned equipment to ensure that
failed equipment will be replaced by serviceable machines. But any of these compensation
strategies cost money for the operator. In the assessment of equipment replacement time, the
compensation cost (i.e. redundant rig cost in this case study) associated with a machine should
be taken into account. Thus, the aim of this paper is to develop a model to determine the
economic replacement time of production equipment, in this case, a drilling rig, considering
redundant rig cost. In this paper, we also consider the relative importance of other cost factors
on the rig’s ERT; these factors are the equipment acquisition, operating, maintenance and
redundant rig costs. Finally, in the model, we consider the time value of money by using a
discount rate.
2. Case study
In this study, our model of ERT for production machines is implemented in a case study of a
drilling rig used in mining industry. Our case study was selected from the mining sector, as
maintenance costs in this sector account for 20-30% of the total cost of production (Kumar,
1994). Kumar (1994) also maintained that to achieve optimal performance from the capital
intensive mining equipment and systems, mine operators must ensure world-class maintenance
in line with other advanced industries. The drilling rig is selected as a case study for several
3
reasons: drilling is the first step in a typical mining cycle and thus is extremely important (see
figure 1); the drilling rigs are heavily loaded; the rigs’ acquisition and maintenance costs are
high; finally, drilling represents a critical bottleneck for production.
1. Drilling
6. Bolting 2. Charging
5. Scaling 3. Blasting
4. Loading
Figure 1. Typical underground mining cycle
3. Model formulation
The ERT of capital equipment is the age that minimises its total cost. In this study, the total
cost is represented by acquisition (initial or investment) cost and ownership cost. The
ownership cost includes operating and maintenance costs and compensation cost. All repairable
systems wear over time; consequently, the ownership cost increases and the resale value
decreases.
In this study, the ERT is defined as the value of the replacement time (RT) which minimises
the total discounted cost, calculated on a monthly basis as follows:
ª ½ º
« °§ ª RT º · 1 ° »
Min TC Min « ®¨¨ AC « ¦ MCi OCi COi » Si ¸¸ u i ¾ u N »
(1)
«° © ¬i 1 ¼ ¹ 1 r 12 ° »
¬¯ ¿ ¼
The objective of the proposed model (i.e. Eq. 1) is to determine the ERT which minimises the
total discounted cost over the rig’s planned lifetime. We assume the replacement rig (i.e. the
new rig) has the same performance and cost as the existing rig (i.e. identical rigs). The number
of replacement cycles during the planned lifetime is represented as:
T
N (2)
RT
where T and RT represent the planned lifetime and the replacement time (in months)
respectively.
3.1 Maintenance cost

Maintenance can be defined as any work used to keep something in an appropriate condition.
Maintenance can corrective or preventive. Corrective maintenance refers to actions which take
place after an unscheduled breakdown to return an item to a specified condition. Preventive
maintenance refers to regularly scheduled actions planned to keep an item in the desired
condition. The maintenance cost can be labelled as a summation of the materials and labour
expense required to keep an item in suitable working condition. In this study, due to the
company’s regulations, all costs data are encoded and expressed as currency unit (cu). The
maintenance cost is represented as follows:
MCi CM i PM i (3)
where CM i and PM i represent corrective and preventive maintenance cost (cu) respectively.
4
CM i SPCi LCCi (4)
where SPCi and LCCi represent spare part and labour costs for corrective maintenance (cu)
respectively.
SPCi SPVi SPLi (5)
where SPVi and SPLi are spare part value and spare part logistic costs (cu) respectively.
LCCi rt u nl u cl (6)
where rt represents repair time (h), nl is number of repairs and cl is man hour cost (cu /h).
PM i SPPi LCPi (7)
where SPPi and LCPi represent spare part and labour costs for preventive maintenance (cu)
respectively.
SPPi SPVi SPLi (8)
LCPi rt u nl u cl (9)
3.2 Operating cost

Data on O&M costs were collected over four years and stored in the MAXIMO computerised
maintenance management system (CMMS). Operating cost can be defined as recurring costs
for efficiently operating the equipment, in our case study, a drilling rig. The operating costs
include administration, energy, fuel, indirect overhead costs, consumables like steel rods,
operator’s salary, a figure given to us by experts at the collaborating company. In CMMS, the
cost data are recorded based on calendar time. Since drilling is not a continuous process, the
operating cost is estimated by considering the utilisation of the drilling rig. The company plans
to use the machine for 120 months. Therefore, extrapolation for the operating and
maintenance cost data was done. Figures 2 and 3 illustrate the maintenance and operating costs
determined by the data extrapolation.
Figure 2. Maintenance cost Figure 3. Operating cost

In Figures 2 and 3, the dots represent the real historical data for maintenance and operating
costs. Curve fitting is done by using Table curve 2D software to show the behaviour of these
costs before and after the time when data were collected. Note that the fitting would be better
if more data were available for a time period of more than four years. This software uses the
least squares method to find a robust (maximum likelihood) optimisation for nonlinear fitting.
It is worth mentioning that the drilling machine in this case study has no multi-level
preventive maintenance programme. In addition, it was new at the start of utilisation. This is
the main reason why the maintenance cost is quite low in earlier months. The history shows
that when the maintenance costs started growing, the user company began to keep track of
cost data by using CMMS. The equation “Lorentzian Cumulative” of extrapolation for
expected maintenance cost obtained by the software is expressed as
5
aª § x b · S º
Y arctan ¨ ¸ » (10)
S «¬ © c ¹ 2¼
where Y represents the expected maintenance cost, a=217.42, b=112.37, c=13.63,
r2 (adj.) = 0.97 and X represents the time (1, 2, 3, 4,…, n months). Similarly, the equation
“Lorentzian Cumulative” of extrapolation for expected operating cost is expressed as
aª § x b · S º
Y « arctan ¨ ¸ » (11)
S¬ © c ¹ 2¼
where Y represents the expected operating cost, a=79.89, b=109.2, c=13.85, r2 (adj.) = 0.91
and X represents the time (1, 2, 3, 4,…, n months).
As the figures show, the operating and maintenance costs increase over time. In fact, the
number of failures increases with time and/or the machine consumes more energy due to
machine degradation.
3.3 Compensation cost

In the present study, we focus on the compensation cost by using a redundant rig cost as one
of the critical factors affecting the ERT. Mining companies, for example, lose a large amount
of money each year from lost production which, in turn, is due to the production equipment’s
downtime. In fact, this may be the most important factor affecting the ERT of production
machines. In this study, we assume when a drilling rig fails and is sent to the workshop for
maintenance, to continue production without stops, the company uses a redundant rig which
has the same performance as the existing faulty rig. Since in the mining industry, downtime in
production is almost zero, the compensation cost in this case represents the cost of using a
redundant rig. In this study, the experts at the collaborating mine classified the rig’s failures in
three categories as follows:
1. Failures fixed by maintenance team at the workshop.
2. Failures fixed by maintenance team at the production point (mining room).
3. Failures fixed by operators at the production point (mining room).
Note: we obtained information on the drilling process and maintenance of drilling rigs by
talking with experts at the user company (U) and manufacturing company (M). Detailed
information, such as experience in years and work position of the experts, appears in table 1.
Table 1. Description of expertise of the experts used in the present study
Current position at companies (U) and (M) Expert field and experience (# years)
Maintenance Engineer for open pit and underground Maintenance of mobile and fixed equipment’s
mines (U) (23)
Mine production Foreman (U) Underground drill machines (30)
Mine production Manager (U) Mine drilling and production (15)
Mine production Planner (U) Mine production planning (22)
Maintenance Supervisor (U) Maintenance of mobile equipment’s (30)
Maintenance Manager (U) Maintenance of mobile equipment’s (26)
Mine production Manager (U) Mine drilling and production (32)
Maintenance Foreman (U) Maintenance of mobile equipment’s (25)
Maintenance Engineer for fixed equipment (U) Maintenance of fixed equipment’s (10)
Global Service Operations Manager (M) Maintenance of equipment (20)
Design Engineer–Underground Drill Rigs (M) Designing underground equipment (10)
Global Fleet Manager Marketing and business management (8)
Vice President Service Operations (M) Parts and Service Business management and
Maintenance of mobile equipment (18)
Regional business-Europe and product line manager- Project management and business management
Rental (M) (10)
6
The compensation cost based on a category one failure of the rig is modelled as follows:
COi RCi (12)
where RCi represents redundant rig cost (cu).
RCi PTi u CRi (13)
where PTi and CRi represent the used time of the redundant rig (h) and redundant rig cost per
hour (cu/h) respectively.
PTi TRi TFi (14)
where TRi and TFi represent the logistic time of the redundant rig (h) and the time to restore
the faulty rig to operation (h) respectively.
TRi T1i T2i (15)
where T1i and T2i represent the time to move the redundant rig from its location to the
production point and the return time from the production point to its original location (h)
respectively.
TFi TMi TWi TLi (16)
where TMi , TWi and TLi represent for the time to move the faulty rig from the production
point to the workshop (h), time in workshop (h) and return time after repair (from workshop
to production point) (h) respectively.
TWi tdi tri tIi (17)
where tdi , tri and tIi represent delay time in workshop before repair (h), actual repair time
(h) and idle time in workshop after repair (h) respectively.
Figure 4 illustrates the time the redundant rig is used due to a category one failure in the existing
rig:
G
A B C D E
T TMi tdi tri tIi TLi T

1i 2i
Time
TWi
TFi
PTi
Figure 4. Time the redundant rig is used due to a category one rig failure
Table 2 represents the clarifications of symbols A, B, C, D, E and G of Figure 2.
Table 2. Clarifications of symbols A, B, C, D, E and G of Figure 2
Symbol Clarification
A Production stops and a redundant rig starts moving from its location
B Production starts with a redundant rig and a faulty rig starts moving to the workshop
C Faulty rig enters the workshop
D Faulty rig exits the workshop
E Faulty rig starts work after repair and redundant rig starts moving to its original location
G Redundant rig arrives at the original location
7
Since the moving speed inside the underground mine is limited to low speed, we assume the
moving time of the maintenance team from the workshop to the production point is almost
equal to the moving time of the faulty rig from a production point to the same workshop.
Thus, the using time a redundant rig is used after a category two rig failure is modelled as
follows:
PTi TRi TMi tri (18)
Note: as the failures fixed by operators are classified as small failures and take only a short time,
the mining company does not use a redundant rig in the third category of failures. Table 3
illustrates the minimum and maximum time values given by the maintenance expert in the
collaborating mine; these are used in the model.
Table 3. Minimum and maximum time values (minute) used in the model
Time Minimum Maximum
Moving time of faulty rig from production point to workshop ( TMi ) 30 60
Delay time in workshop before repair ( tdi ) 30 90
Idle time in workshop after repair ( t Ii ) 30 60
We assume the moving time of the redundant rig T1i is equal to the moving time of the faulty
rig TMi . It is worth mentioning that the time values tdi , tIi , TMi , TLi , T1i and T2i are
randomly generated by using MATLAB code, since this type of data is not available from the
collaborating mine. We use a discount rate of 10% to consider the “time value of money”
following the suggestions of the collaborating mining company.
3.4 Resale value

A Matheson formula (declining balance depreciation model) was used to estimate the resale
value of the rig after each month of operation. In this method, a fixed percentage of the book
value at the beginning of the month represents the monthly depreciation of the rig. The rig
resale value is its value if/when the firm wants to sell it at any time during its lifetime. The
resale value denoted by Si is calculated by using the following model (Luderer et al., 2010;
Eschenbach, 2010; Dhillon, 2010):
Si BV1 u (1 Dr )i (19)
where “i” represents a time (number of months), i=1, 2, 3, …120 planed lifetime, BV1 and
Dr represent the rig value on the first day of operation and depreciation rate respectively. In
addition,
BV1 AC u A (20)
where “A” represents the percentage of decrease multiplied by the rig acquisition cost to
represent the rig value on the first day of use. During discussions with us, company experts
agreed that the rig acquisition cost decreases by 10% on the first day of use (i.e. A=0.9). In this
study, the rig acquisition cost is 6000 (cu). Hence, the rig value on the first day of use is 5400
(cu).
The depreciation rate that allows for full depreciation by the end of the planned lifetime of the
rig is modelled by the following formula (Luderer et al., 2010, Dhillon, 2010):
1
§ SV ·T
Dr 1 ¨¨ ¸¸ (21)
© BV1 ¹
8
where T and SV represent the planned lifetime of the rig, 120 months, and rig scrap value
respectively. The rig is assumed to reach scrap value after 10 years. The rig resale value is
calculated by:
Si AC u A u (1 Dr )i (22)
The declining balance depreciation model is suitable in our case study because this model
writes off the cost of the rig early in its lifespan at an accelerated rate and at correspondingly
lower monthly charges close to the end of its lifespan. It also considers the rig to be more
productive when it is new, and its productivity declines continuously due to rig aging.
Therefore, in the early years of its lifespan, a rig will generate more revenue than in later years.
In accountancy, depreciation refers to two aspects of the same concept. The first is the decrease
in the rig value. The second is the systematic allocation of the capital cost of the rig over its
lifespan. The scrap value is an estimate of the value of the equipment at the time it is sold or
disposed of. In our case study, 50 (cu) is assumed to be the scrap value of the rig at the end of
its planned lifetime, a figure given to us by company experts.

We tested the model for ERT on a case study of a drilling rig. This rig is manufactured by
Atlas Copco Company and used by Boliden mineral AB Company in Sweden. MATLABTM
software is used to enable a variation of the replacement time (RT) of Eq. (1) which minimises
the total cost.
Figure 5 shows the optimisation curve and the ERT of our case study at a redundant rig cost
per hour equal to 1 (cu/h).
4 Economic replacement time of a drilling rig
18 x 10
Redundant rig cost = 1 (cu/h)
15
Total cost (cu)
10
5
ERT=104 months
0
0 20 40 60 80 100 120 140 160 180 200 220 240
Replacement time RT (month)
Figure 5. Economic replacement time of the drilling rig

The results show the lowest possible total cost can be achieved by replacing the rig at 104
months of its planned lifetime. A decision to replace the rig before or after its ERT incurs
greater costs for the user company. The use of a lower replacement age (i.e. less than 104
months) incurs higher costs due to the high investment cost. Meanwhile, if the lifetime of the
rig exceeds its ERT (i.e. more than 104 months), losses will increase for two reasons:
1. The O&M and redundant rig costs increase when the operating hours increase due to
rig degradation.
2. The rig resale value will decrease for each month of operation until it reaches its scrap
value by the end of its planned lifetime.
As Figure 5 also shows, there is a range 97-109 (months) when the minimum total cost can be
still achieved in practice. In this study, we call it the economic replacement range. Finding the
9
economic replacement range is an important result of our study, as it can help decision makers
in their planning. To show the effect of the redundant rig cost per hour ( CRi ) in the ERT of
our case study, we change the values of the redundant rig cost per hour from 1 to 6 (cu/h).
Figure 6 shows the result.
Effect of redundant cost per hour
CRi=1 (cu/h). ERT=104 month CRi=2 (cu/h). ERT 94 month
180000 CRi=3 (cu/h). ERT 87 month CRi=4 (cu/h). ERT 82 month
CRi=5 (cu/h). ERT 79 month CRi=6 (cu/h). ERT 76 month
150000 CRi = Redundant rig cost per hour
Total cost (cu)
120000
90000
60000
30000
0
0 20 40 60 80 100 120 140 160 180 200 220 240
Figure 6. Effect of the redundant rig cost per hour on the ERT of the drilling rig
It is clear from figure 6 that increasing the CRi (cu/h) has a negative effect on the ERT of the
drilling rig. To determine the effect of other factors on the ERT, we perform a sensitivity
analysis on rig acquisition, operating, maintenance and redundant rig costs (cu) using the ANN
technique. Four MATLAB codes for six cases of CRi (1-6 cu/h) are used to identify the effect
of increased acquisition cost (IAC), decreased operating cost (DOC), decreased maintenance
cost (DMC) and decreased redundant rig cost (DRC). The resulting ERT from these four
codes is fed as input to the ANN and the results translated into a relatively simple equation to
estimate the ERT of the drilling rig. The method of partitioning weights, proposed by
(Garson, 1991) and adopted by (Goh, 1995), is used to determine the relative importance of
the various input factors; see figure 7.
60
Relative 1mportance (%)
50
40
30
20
10
0 1 2 3 4 5 6
IAC (cu) 33,1 33,1 31,7 34,6 30,5 32,8
DOC (cu) 14,1 11,9 8,0 9,4 10,3 3,2
DMC (cu) 20,4 15,0 17,7 10,1 12,3 6,2
DRC (cu) 32,2 39,8 42,3 45,7 46,6 57,5
Figure 7. Relative importance of input factors on ERT of drilling rig

10
As evident in figure 7, the most important factor is the redundant rig cost, followed by the
acquisition, maintenance and operating costs. Therefore, a design for reliability and
maintainability should be adapted to reduce the downtime and maintenance costs of the
drilling rig. As mentioned earlier, four MATLAB codes are used to identify the effect of IAC,
DOC, DMC and DRC on the ERT of drilling rig. We choose the case where CRi = 1 (cu/h)
to demonstrate the effect of these factors on the ERT. Figure 8 shows the correlation of DRC
and IAC for a given 25% DOC and DMC. Figure 9 shows the correlation of DRC and DMC
for a given 25% IAC and DOC. Figure 10 shows the correlation of DRC and DOC for a
given 25% IAC and DMC.
126 IAC=10% DOC=25%

IAC=20% DMC=25%
124 IAC=30%
IAC=40%
122
IAC=50%
ERT (month)
120
118
116
114
112
110
0 5 10 15 20 25 30 35 40 45 50
Figure 8. Correlation of DRC and IAC for a given 25% DOC and DMC
126 DMC=10% IAC=25%

DMC=20% DOC=25%
124 DMC=30%
DMC=40%
122 DMC=50%
ERT (month)
120
118
116
114
112
0 5 10 15 20 25 30 35 40 45 50
Figure 9. Correlation of DRC and DMC for a given 25% IAC and DOC
124
DOC=10% IAC=25%
DOC=20% DMC=25%
122 DOC=30%
DOC=40%
DOC=50%
120
ERT (month)
118
116
114
0 5 10 15 20 25 30 35 40 45 50
Figure 10. Correlation of DRC and DOC for a given 25% IAC and DMC
As figures 8-10 show, DRC, IAC, DMC and DOC have a positive effect on the ERT of the
drilling rig, but it is also evident that DRC has a more positive effect, followed by IAC, DMC
and ROC.
11
4.1 Training and testing the proposed ANN model
Artificial neural networks can perform nonlinear modelling without prior information and are
able to learn complex relationships between inputs and outputs; the process is also fast
(Ahmadzadeh and Lundberg, 2013). Our ANN analyses are based on the results obtained from
the four cases represented by four MATLAB codes, as explained above. The resulting ERTs
from these codes are fed as inputs to ANN and the results translated into a relatively simple
equation which can be used to estimate the overall ERT of the drilling rig. The equation is
transformed to an Excel spread-sheet to make ERT estimation quick and easy for any engineer
to apply. As mentioned earlier, the proposed model has four inputs: IAC, DOC, DMC and
DRC. A hidden layer with three neurons and a nonlinear transfer function allows the network
to learn nonlinear and linear relationships between input and output variables. The number of
neurons in the output layer is constrained to one, as the output only requires one parameter, in
this case, the ERT of the drilling rig. 90 % of the data are used in training and 10 % in testing
the neural network; see figures 11 and 12. The model shown in figures 11 and 12 have very
high values of R = 99 % for ANN. However, as also shown in the figures, the neural network
model yields outputs very close to the desired targets with a high level of accuracy.
Outputs vs. Targets, R=0.99819 Outputs vs. Targets, R=0.99829

135 135
Outputs Y, Linear Fit: Y=(1)T+(-0.44)
Data Points Outputs Y, Linear Fit: Y=(1)T+(-0.81) Data Points

Best Linear Fit Best Linear Fit
130 130 Y=T
Y=T
125 125
120 120
115 115
110 110
105 105
105 110 115 120 125 130 135 105 110 115 120 125 130 135
Targets T Targets T
Figure 11. Training capability Figure 12. Testing capability
The proposed ANN model is used to construct a formula to calculate the ERT of our case
study. The formula is transformed to an Excel spreadsheet to make ERT estimation quick and
easy for any engineer to apply. The structure of the optimal ANN model is shown in figure 13;
its connection weights and threshold levels are summarised in Table 4.
IAC (%)
Input factors
DOC (%)
Output
DMC (%) (ERT)
DRC (%)
Figure 13. Optimal structure of artificial neural network (ANN) model

12
Table 4. Weights and threshold levels of proposed ANN
Hidden wij Hidden
layer weights from node i th input layer threshold
nodes to node j th hidden layer (șj)
i=1 i=2 i=3 i=4
j=5 0.085 0.089 -0.03 0.108 -8.776
j=6 3.006 0.379 1.441 1.784 -1.120
j=7 1.416 0.452 1.509 1.876 -5.001
Output wij Output
layer weights from node i th hidden layer Threshold
nodes to node j th output layer (șj)
i=5 i=6 i=7
j=8 4.679 3.410 5.376 -3.581
Pre-processing data by scaling improves the training of the neural network. To avoid a slow
rate of learning, specifically near the end points of the output range (due to the property of the
sigmoid function, which is asymptotic to values 0 and 1), the input and output data are scaled
in the interval between 0.1 and 0.9 (Oreta, 2004). It should be noted that any new input data
should be scaled before being presented to the network and the corresponding predicted values
should be un-scaled before use (Yousif, 2007). The linear scaling equation is expressed by:
§ 0.8 · § 0.8 X max ·
Xs ¨¨ ¸¸ X ¨¨ 0.9 ¸ (23)
© ' ¹ © ' ¸¹
where Xs represents the scaled value of input factors and X represents the un-scaled value of
input factors. As it used in MATLAB code for neural networks, Eq. (23) is used here for an
IAC, DOC, DMC and DRC between a minimum increasing or decreasing percentage of 1%
(Xmin) and a maximum increasing or decreasing percentage of 50% (Xmax). This results in:
' X max X min (24)
The equation length depends on the number of nodes in the hidden layer. Adopting three
nodes gives an accuracy of 99%. The small number of connection weights of the neural
network enables the ANN model to be translated into a relatively simple formula, in which the
predicted ERT can be expressed as follows:
ERTs 1 (25)
· § ½
°®
T
§
1 ·¸§¨ w 1 1 ·¸°
8 ¨¨ w5:8
¨ ¸ ¨ w
x1 ¸¸ ¨¨ 6:8 x ¸¸ ¨¨ 7:8 x3 ¸¸ ¾°
° 1 e 1e 2 1e ¹¿
1 exp ¯ © ¹ © ¹ ©
where ERTs represents the scaled ERT derived from the ANN model, șj represents the
output threshold and wij represents the weight from node i in the hidden layer to node j in the
output layer. Hence,
x1 T5 w5:1 u IAC w5:2 u DOC w5:3 u DMC w5:4 u DRC (26)
To obtain the actual value of ERT, the predicted ERTs must be re-un-scaled using the
following formula:
13
§ ' · § ' ·
X Xs ¨ ¸ 0.9 ¨ ¸ X max (29)
© 0.8 ¹ © 0.8 ¹
Equation (29) is obtained by solving Equation (23), considering the input variable (X) as
unknown and the output variable (Xs) as known. An Excel spread-sheet can be used as a
substitute for fast and accurate calculation of the ERT of the drilling rig. Eq. 30 is applied to
estimate the actual value of the ERT of the drilling rig as follows:
§ ERTmax ERTmin · § ERTmax ERTmin ·
ERT ERTs ¨ ¸ 0.9 ¨ ¸ ERTmax (30)
© 0.8 ¹ © 0.8 ¹
where ERTmax and ERTmin represent the maximum and minimum values of ERT respectively
In this paper, it is worth to mention that the artificial neural network techniques were used for
the following two main reasons.
x One aim was to help the engineers and decision makers in the user company to
estimate the ERT of new drilling rigs without needing to use complicated software.
x Another aim was to determine the relative importance of factors which were used in
the optimization model and which would affect the ERT of new drilling rigs. The
factors which have the highest impact on the ERT of new rigs should be prioritized in
the development process of new drilling rigs.
4.2 Regression analysis

The regression analysis of the results obtained from the above four MATLAB codes uses
Minitab software and the least squares method. ERT is modelled as a linear function of IAC,
DOC, DMC and DRC. The regression analysis results in the following mathematical model:
ERT 104 0.19 u IAC 0.04 u DOC 0.14 u DMC 0.17 u DRC (31)
It is evident from the regression analysis for this particular case (i.e. CRi = 1 cu/h) that the
IAC has the greatest effect on the ERT of the drilling rig, followed by DRC, DMC and DOC,
but at the same time, CRi (cu/h) increases from 2 to 6 for other cases. Ultimately, the DRC has
the largest effect on ERT in our case study; see figure 7. The R-squared value obtained from
the regression analysis, R2 (adj) = 99 %, indicates that the ERT of the drilling rig depends
linearly on the factors of IAC, DOC, DMC and DRC, supporting the results obtained in the
sensitivity analysis.
5. Conclusions
This paper presents a model for the economical replacement time of production machines.
Although the problem has been solved previously by other researchers using different models,
our model can more readily examine the relationship between the factors affecting the ERT of
production machines, especially the cost of using a redundant rig. The model is found to be a
good choice for estimating the ERT in a case study of the drilling rig used in underground
mines in Sweden, and it can be extended to other production capital assets in other industries.
In our case study, the results of the sensitivity analysis show that the redundant equipment cost
has the highest impact on the ERT followed by equipment acquisition, maintenance and
operating costs. The results of the sensitivity analysis also indicate that decreasing the operating,
maintenance and redundant rig costs have a positive effect on increasing the ERT. The results
obtained from the optimisation curves show that increasing the redundant rig cost per hour has
a negative effect on the ERT. Therefore, improving the reliability and maintainability of
production equipment is essential to reduce their downtime and maintenance costs.The
14
absolute ERT of the drilling rig when CRi = 1 cu/h is 104 months. However, the ERT has a
range of 97 to 109 months, during which period the total cost remains almost constant. This
means the user company has the flexibility of making replacements within the optimum
replacement age range (12 months). The results of the regression analysis show that the ERT of
the new equipment depends linearly on its acquisition, operating, maintenance and redundant
rig costs. These results confirm the results of the sensitivity analysis. In summation, this study
presents a comprehensive and very practical approach which can determine the ERT of any
mobile equipment with higher levels of certainty by using ANN analysis.
Acknowledgment
The authors would like to thank Atlas Copco and Boliden mineral AB, for supporting this
research. Special appreciation is extended to the experts at Boliden mineral AB and Atlas
Copco for sharing their valuable knowledge and experience. The authors would like also to
thank Arne Vesterberg at Boliden mineral AB and Andreas Nordbrandt at Atlas Copco for
them supports. The authors would like also to thank for the support of CAMM (Centre of
Advanced Mining & Metallurgy) project in this research work. My sincerest gratitude is
extended to the reviewers and the editor of this journal for the valuable comments that we
received from them, which helped to improve this article.
References
Ahmadzadeh, F. and Lundberg, J. (2013), “Remaining useful life prediction of grinding
millliners using an artificial neural network”, Minerals Engineering, Vol. 53, pp. 1-8.
Al-Chalabi, H., Lundberg, J., Ahmadi, A., and Jonsson, A. (2014). Case Study: Model for
economic lifetime of drilling machines in the Swedish mining industry.The Engineering
Economist, (ahead-of-print), 1-17. Available at:
http://dx.doi.org/10.1080/0013791X.2014.952466
Al-Chalabi, H., Ahmadzadeh, F., Lundberg, J., and Ghodrati, B. (2014), “Economic lifetime
prediction of a mining drilling machine using artificial neural network”, International
Journal of Mining, Reclamation and Environment, Vol. 28 No. 5, pp. 311-322.
Bazargan, M. and Hartman, J. (2012), “Aircraft replacement strategy: Model and
Analysis”, Journal of Air Transport Management, Vol. 25, pp. 26-29.
Bean, J.C., Lohmann, J.R. and Smith, R.L. (1985), “A dynamic infinite horizon replacement
economy decision model”, The Engineering Economist, Vol. 30 No. 2, pp. 99-120.
Bellman, R. (1955), “Equipment replacement policy”, Journal of Society for Industrial and
Applied Mathematics, Vol. 3 No. 3, pp. 133-136.
Bethuyne, G. (1998), “Optimal Replacement Under Variable Intensity of Utilization and
Technological Progress”, The Engineering Economist, Vol. 43, No. 2, pp. 85-105.
Blanchard, B. S., Verma, D., and Peterson, E. L. (1995), Maintainability: a key to effective
serviceability and maintenance management, John Wiley & Sons, New York.
Cheevaprawatdomrong, T. and Smith, R. L. (2003), “A paradox in equipment replacement
under technological improvement”, Operations Research Letters, Vol. 31 No. 1, pp.
77-82.
Dandotiya, R. (2012), Decision support models for the maintenance and design of mill liners,
PhD thesis, Lulea University of Technology, Sweden.
Dhillon, B. S. (2010), Life cycle costing for engineers, Taylor & Francis Group, New York.
Eilon, S., King, J. R. and Hutchinson, D. E. (1966), “A study in equipment
Replacement”, Journal of the Operational Research Society, Vol. 17 No. 1, pp. 59-71.
Elton, E.J. and Gruber, M.J. (1976), “On the Optimality of an Equal Life Policy for
Equipment Subject to Technological Improvement”, Operational Research Quarterly,
Vol. 27 No. 1, pp. 93- 99.
Eschenbach, T. (2010), Engineering economy: applying theory to practice, 3rd ed.,
15
Oxford University Press, New York.
Garson, G. D. (1991), “Interpreting neural network connection weights”, Artificial
Intelligence, Vol. 6 No. 4, pp. 46-51.
Goh, A. T. C. (1995), “Back-Propagation neural networks for modeling complex systems”
Artificial Intelligence in Engineering, Vol. 9 No. 3, pp.143-151.
Golmakani, H. R., and Pouresmaeeli, M. (2014), “Optimal replacement policy for condition-
based maintenance with non-decreasing failure cost and costly inspection”, Journal of
Quality in Maintenance Engineering, Vol. 20 No. 1, pp. 51-64.
Hartman, J. C. and Murphy, A. (2006), “Finite-horizon equipment replacement analysis”, IIE
Transacations, Vol. 38 No. 5, pp. 409-419.
Hartman, J. C. and Rogers, J. L. (2006), “Dynamic programming approaches for equipment
replacement problems with continuous and discontinuous technological change”, IMA
Journal of Management Mathematics, Vol. 17 No. 2, pp. 143-158.
Hartman, J. C., and Tan, C. H. (2014), “Equipment Replacement Analysis: A Literature
Review and Directions for Future Research”, The Engineering Economist, Vol. 59, pp.
136-153.
Hartman, J.C. (2005), “A note on a strategy for optimal equipment replacement”, Production
planning & control, Vol. 16 No. 7, pp. 733-739.
Hritonenko, N. and Yatsenko, Y. (2008), “The dynamics of asset lifetime under technological
Change”, Journal of the Operations Research Letters, Vol. 36 No. 5, pp. 565-568.
Keles, P. and Hartman, J. C. (2004), “Case study: Bus fleet replacement”, Engineering
Economist, Vol. 49 No. 3, pp. 253-278.
Kumar, U. (1994), “Recent Trends in Mine Maintenance Technology and
Management”, Corrosion Reviews, Vol. 12 No. 3-4, pp. 191-200.
Luderer, B., Nollau, V., and Vetters, K. (2010), Mathematical formulas for economists. 4th
ed., Springer Heidelberg Dordrecht, London New York.
Oakford, R.V., Lohmann, J.R. and Salazar,A. (1984), “Adynamic replacement economy
decision model”, IIE Transactions, Vol. 16 No. 1, pp. 65–72.
Oreta A. W. C. (2004), “Simulating size effect on shear strength of RC beams without stirrups
using neural networks”, Engineering structures, Vol.26 No. 5, pp. 681-691.
Richardson, S., Kefford, A. and Hodkiewicz, M. (2013), “Optimised asset replacement
strategy in the presence of lead time uncertainty”, International Journal of Production
Economics, Vol. 141 No. 2, pp. 659-667.
Scarf, P.A. and Bouamra, O. (1999), “Capital equipment replacement model for a fleet with
variable size”, Journal of quality in maintenance engineering, Vol. 5 No. 1, pp. 40-49.
Tanchoco, J. M. A. and Leung, L. C. (1987), “An input-output model for equipment
replacement decisions”, Engineering costs and production economics, Vol. 11 No. 2,
pp. 69-78.
Verheyen, P. A. (1979), “Economic interpretation of models for the replacement of
Machines” European Journal of Operational Research, Vol. 3 No. 2, pp. 150-156.
Wagner, H.M. (1975), Principles of Operations Research, 2nd ed., Prentice-
Hall, Inc., Englewood Cliffs, New Jersey.
Wijaya, A. R., Lundberg, J., and Kumar, U. (2012), ”Robust-optimum multi-attribute age-
based replacement policy”, Journal of Quality in Maintenance Engineering, Vol. 18
No. 3, pp. 325-343.
Yatsenko, Y. and Hritonenko, N. (2005), “Optimization of the lifetime of capital equipment
using integral models”, Journal of industrial and management optimization, Vol. 1,
No. 4, pp. 415-432.
Yousif, S. T. (2007), Artificial neural network modeling of elasto-plastic plates. Ph.D. thesis,
College of Engineering, Mosul University, Iraq.
16
View publication stats

Thesis

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Thesis

Uploaded by

Copyright:

Available Formats

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

Reliability and Life Cycle Cost Modelling of Mining Drilling Rigs

Thesis · December 2014

AtlasCopco View project

Ph.D Candidate View project

The user has requested enhancement of the downloaded file.

Luleå University of Technology 2014

Hussan Saed Hamodi Al-Chalabi

Hussan Saed Hamodi Al-Chalabi

Division of Operation, Maintenance and Acoustics

Hussan Saed Hamodi Al-Chalabi

ANN Artificial neural network

1.1.1 Mining drilling rig

1 Cabin 6 Front jacks 11 Cable reeling unit

1.1.2 Reliability literature review

1.1.3 Optimal replacement time literature review

1.2.2 System availability

1.2.3 System maintainability

1.2.4 System life cycle costing

1.2.5 Life cycle costing fundamentals

1.3 Industrial motivation of the study

Regression Data analysis

2.1 Research problem description

2.2 Overall research goal

Research question 1 (RQ 1) X

2.4 Scope and limitations of the study

Figure 2.1. Research framework

3.1 Data collection and analysis

Serial correlation test

Operating cost (cu)

Figure 3.2. Maintenance cost Figure 3.3. Operating cost

Operating cost (cu)

Figure 3.4. MC after extrapolation Figure 3.5. OC after extrapolation

3.1.2 Failure and cost data

3.3 Method used in Papers II-IV

4.3 Paper III

5.1 Results and discussion related to research question 1

Figure 5.6. Suggested plate of steel

Figure 5.9. Suggestion to prevent pull rope failures

Figure 5.10. Damaged cup seal

where Y represents the expected maintenance cost, a=65.45, b=-0.006, c=0.0006, r2

Figure 5.14. Maintenance cost (third model)

Figure 5.15. Effect of increasing purchase price

Figure 5.16. Effect of decreasing operating cost

Figure 5.17. Effect of decreasing maintenance cost

5.3 Results and discussion related to research question 3

Outputs Y, Linear Fit: Y=(1)T+(-2)

5.4 Results and discussion related to research question 4

126 DMC=10% IAC=25%

Figure 5.27. Optimal structure of artificial neural network (ANN) model

x2   6  w6:1  IAC  w6:2  DOC  w6:3  DMC  w6:4  DRC (5.6)

x3   7  w7:1  IAC  w7:2  DOC  w7:3  DMC  w7:4  DRC (5.7)

The following conclusions can be made on the basis of this thesis.

Abernethy, RB (2000), The New Weibull Handbook, Gulf Publishing: Houston.

Downtime analysis of drilling machines and suggestions for

2. Drilling machine and data collection

3. Reliability and downtime analysis

Rock drill A1 Rock drill A2 Rock drill A3 Rock drill A4

Lognormal distribution has a probability density function given by:

The confidence interval of the downtime is determined according to the following

The confidence interval of the estimated downtime can be solved by finding

5. Results and discussion

Downtime Repair time/failure No. of failures

Downtime Repair time/failure No. of failures