Accepted Manuscript

A Cost-Effective Simulation Algorithm for Inspection Interval Optimization:

An Application to Mining Equipment

Onur Gölbaş ı, Nuray Demirel

PII: S0360-8352(17)30400-X
DOI: http://dx.doi.org/10.1016/j.cie.2017.09.002
Reference: CAIE 4888

To appear in: Computers & Industrial Engineering

Received Date: 17 April 2017

Revised Date: 30 August 2017
Accepted Date: 5 September 2017

Please cite this article as: Gölbaş ı, O., Demirel, N., A Cost-Effective Simulation Algorithm for Inspection Interval
Optimization: An Application to Mining Equipment, Computers & Industrial Engineering (2017), doi: http://
dx.doi.org/10.1016/j.cie.2017.09.002

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
 A Cost-Effective Simulation Algorithm for Inspection Interval Optimization: An Application

to Mining Equipment

Onur Gölbaşıa,* and Nuray Demirelb

aDr.; Department of Mining Engineering, Colorado School of Mines, CO 80401, Colorado, The USA; Phone:+1 720-327
3112; Fax: +1 303-273 3719; E-mail: golbasi@mines.edu

bAssoc. Prof. Dr.; Department of Mining Engineering, Middle East Technical University, 06800, Ankara, Turkey; Phone:
+90-312-210 5813; Fax: +90-312-210 5822; E-mail: ndemirel@metu.edu.tr

Abstract

In machinery maintenance policies, regular inspection intervals should be specified in such a way that the

cumulative of direct and indirect financial consequences of maintenance activities should be minimized while

supporting the functional health of system components. This study aims to develop a simulation algorithm,

called the time-counter, to optimize inspection intervals. In the algorithm, uptime and downtime behaviors of

the system components and production losses in the corrective repairs are considered random values. Delay

time concept is regarded when estimating failure detection periods and deciding on the required maintenance

type. In addition, the developed model is applied to two active draglines and their inspection intervals are
optimized for 232 and 184 hours for Dragline-1 and Dragline-2, respectively. The optimized values are

observed to decrease the total maintenance cost by 5.9 and 6.2 percent compared to the current interval of 160

hours. The main novelties of the study are that i) the proposed concept which allows for the simultaneous

assessment of system components in an incremental time span has not been proposed in the literature when

deciding on optimal inspection intervals, ii) it is the first initiative in inspection optimization of a mining

machinery system, and iii) it uses real datasets on lifetime, repair time, and financial values that are rarely

observed in the maintenance studies.

Keywords: Inspection interval; optimization; delay time; production systems; maintenance cost.

1 Introduction

Inspections are the integral parts of maintenance policies, especially in machine-based production cycles.

These activities are generally performed in regular intervals to identify, examine, and recover potential

abnormalities in working machinery systems. An inspection may entail various work packages such as: i)
visual inspection of system elements, ii) preventive or corrective rectification of defects via repair or
replacement activities, iii) preventive replacement of predefined components in deterioration period, iv)
lubrication of friction-intensive locations in systems, and v) overhauling complex components. Inspections

identify work packages covering the questions of whom, how, and how long for an effective implementation

schedule. These maintenance activities allow decision-makers to reduce overall maintenance costs and to

monitor functional efficiencies of system elements via taking precautions against failures that may take place

during operations.

Although inspections provide benefits for sustainability of operations and longevity of system

components, practicing them in short time intervals may lead to a frequent halt in operation. In this case,

production losses due to excessive downtimes overcome the economic benefits of inspections. On the other

hand, if time intervals between inspections are over-extended, this causes frequent failures of components
during operations and initiates wear-out problems for working components. Resultant economic

consequences of maintenance, once again, becomes a problematic issue. Therefore, inspection intervals should

be determined so that the overall maintenance cost is minimized by including both direct and indirect
considerations.

This paper proposes a simulation model for optimization of constant inspection intervals for a production

system with multiple components, each of which has a random lifetime and random repair time over a

predefined observation period. During the observation period, it is assumed that the following events can be
performed for individual components within the system:

i. Preventive Maintenance O Component is maintained preventively in regular inspections in case the period

of time between the defect arrival and expected failure points overlaps any inspection duration.

ii. Corrective Maintenance O Component is maintained correctively during operation in case the defect

observable period is between any two inspection points and that it cannot be recognized during any

inspection.

iii. Do Nothing O If a component is functionally healthy but any other operationally-dependent component

fails during operation hours, the healthy components stop deteriorating during the random repair time

of the failed one.

The proposed model is a simultaneous analysis of the events (i), (ii), and (iii) for each component in the

system in the given active time point which is incremented with small time intervals up to target observation

period. The model is mainly motivated by solving time-dependent failures and the operation behaviors of
large numbers of components with different operational dependency to obtain optimal inspection intervals.

This research integrates a maintenance decision mechanism with delay-time concept with a cost minimization

problem and employ a new time-incremental simulation algorithm to find optimal inspection intervals. The
second motivation of this paper is that the algorithm is applied to two massive draglines used in a coal mine

and actual datasets gathered from the operations are introduced to the algorithm. In this way, this study
offers an inspection interval optimization for a mining machinery for the first time where production is

significantly correlated with machinery performance, and maintenance cost alone may add up to 40-50% of

the total operating cost of machinery in a mine. In addition, a comprehensive reliability analysis is performed

for the dragline components so that the lifetime and the maintenance characterization of an earthmover

working in a demanding mining environment is revealed in detail for the first time.

This study covers five main sections. Following this introductory part, Section 2 gives a comprehensive

literature review on the delay-time based inspection optimization and mining machinery maintenance. The

originality of this paper is also discussed in this section. Section 3 presents the proposed simulation model. In

Section 4, the developed model is applied to the draglines by using actual datasets to find the optimal
inspection interval by minimizing the overall maintenance cost. Finally, the main conclusions derived from

the study are given in Section 5.

2 Literature Review

2.1 Inspection Interval Optimization with Delay-Time Concept

The delay-time concept introduced by Christer (1973) is extensively utilized in the literature to determine

optimal inspection intervals of the systems. This concept considers the two-stage failure process as a

summation of two successive periods: The first one is between the start-up time of the component and the
defect arrival time and the other one, called delay time, is from the defect arrival time to the expected failure

time if the defect is not rectified (Wang and Banjevic, 2012). In the literature, the main effort for the delay-time

concept has been given to building mathematical models under certain assumptions. These models have

heavily concentrated on the systems with i) single component, ii) a major failure mode or iii) a single failure
rate (Christer, 1991; Okumura et al., 1996; Jia and Christer, 2002; Wang, 2009; Wang, 2011; Berrade, 2013;

Wang, 2013; Wang et al., 2014; Flage, 2014; Van Oosterom et al., 2014; Berrade et al., 2015; Li et al., 2015; Wang

and Wang, 2015; Ramadan, 2016; Wang et al., 2017; Yang et al., 2017). There are comparatively fewer numbers

of models working on the systems with i) multi-component, ii) multi-failure mode or iii) multi-defect.

In more detail, Zhao et al. (2007) tried to maximize reliability of components with the multi-defect arrival

by regarding non-constant inspection intervals. The non-homogenous Poisson process was used to define the

defect arrival rate and minimal repair was assumed. Aven (2009) considered a monotone safety system with

multi-components to develop an inspection plan by using the delay time concept via minimizing long run
unit cost induced by inspections and component replacements in failures and inspections. Wang et al. (2010)

presented a delay time model for multi-component systems where components were regarded individually

but also in harmony to league together within a subsystem or system. This model allowed for the
identification of more than one failure mode for each component. Optimal intervals were obtained by
minimizing total cost in a renewal cycle. Wang (2012) used the delay-time concept in a joint model to find out
the optimal inspection intervals and to develop a spare part inventory plan for the systems holding large
amounts of identical component. Wang et al. (2012) proposed a dynamic model for the inspection policy of

multi-component systems where technological improvements in the systems may require non-constant

inspection intervals. The model objective function was developed in such a way that cost, downtime or

reliability could be selected as an optimization criterion. Taghipour and Banjevic (2012) proposed two models

to optimize the inspection interval of a complex system with components subjected to hard and soft failures.

In a research by Jodejko-Pietruczuk and Werbińska-Wojciechowska (2014), the effect of accuracy of delay-time

estimation on block inspection policy for a three-component k-out-of-n system was simulated by assuming
three different probability functions for the delay time values. Liu et al. (2015) proposed a model to achieve

optimal inspection intervals of a n-component parallel system by minimizing long-term cost.

Besides mathematical contributions, the delay-time approach has also been applied to various specific

systems such as automotive vehicles (Scarf and Majid, 2011), wind turbines (Andrawus et al., 2008),

metallurgical systems (Christer et al., 1995; Christer et al., 1998), marine systems (Christer and Lee, 1997; Wang
and Majid, 2000; Pillay et al., 2004; Arthur, 2005; Emovon et al., 2016), civil engineering structures (Christer,

1988; Redmond et al., 1997; Cavalcante et al., 2017), and other mechanical systems (Leung and Kit-Leung, 1996;

Martínez et al., 2015).

2.2 Maintenance Researches on Mining Equipment

Mining is the extraction of economically-valuable minerals from the earth’s crust by using different surface

and underground application techniques. Thousands of large pieces of equipment with different capacities

and service types are employed at mine sites to satisfy the row material demand of various industries. More
particularly, surface mines dominate most of the mining equipment population all over the world. More than

70,000 pieces of equipment in 1,400 large surface mines with a total market value of USD 150 billion is

expected to be used in the surface mining operations worldwide. The given population can be examined in
detail in Table 1 (The data was retrieved from The Parker Bay Company, 2017).

Table 1. Population of Active Surface Mining Equipment in the World.

Equipment availability is the primary concern of sustainable operations in both surface and underground

mines since mining production rate and operating cost are significantly correlated with equipment

performance. Here, maintenance expenditure alone is expected to add up to 40-50 % of the total operating
cost of equipment. Inefficient maintenance policies may increase this amount as well as interrupt production

plans. Therefore, it is vital for mines to monitor operability and maintainability of their equipment fleet to

ensure sustainable operation and to lower ore extraction cost.

A considerable amount of research paper has been published on the assessment of mine equipment

operability which relies heavily on reliability modelling. Some of them have been motivated to obtain
reliability-based preventive maintenance intervals of equipment by conventional methods (Roy et al., 2001;

Samanta et al., 2002; Hall et al., 2003; Samanta et al., 2004; Barabady and Kumar, 2008; Elevli et al., 2008;

Vayenas and Wu, 2009; Uzgören et al., 2010; Vayenas and Wu, 2011; Hoseinie et al., 2012; Rahimdel et al., 2014;

Gustafson et al., 2015). The reliability assessment in these papers shows the similarity that the systems are

decomposed into subsystems or taken as a whole and reliability is evaluated by fitting the grouped lifetime

datasets into a distribution. Some innovative approaches such as genetic algorithm (Vagenas and Nuziale,
2001), fuzz logic (Tanasijević et al., 2013), neural networks (Dindarloo, 2016a), and support vector regression

(Dindarloo, 2016b) were also proposed in reliability evaluation of mining equipment. Moreover, some

researchers presented graphical methods that can be used in monitoring and management of mining
equipment failures (Hall et al., 2000; Rao and Prasad, 2001; Gupta et al., 2009; Pascual et al., 2009; Skotnicka-
Zasadzień and Bialy, 2011; Barberá et al., 2014). In addition, preventive replacement of working components

has been one of the research interests in mine equipment maintenance. Lhorente et al. (2004) incorporated

corrective and preventive maintenance with preventive replacement to determine an optimal age-based

maintenance policy and they applied the model to mining truck wheel motors. Knights (2009) optimized the
time interval for group replacement of shovel teeth by analyzing grouped failure data with multiple

suspensions and by finding rate of wear on the components. Wang et al. (2009) applied a joint model of

replacement policy and inventory policy to mining truck motors by using condition- based reliability in

characterization of unit failures. Demirel and Gölbaşı used a reliability analysis to find out the age-

based replacement intervals of worn-out dragline components and minimum repair vs. replacement cost

ratios for the economic validity of replacements. Moreover, there are some other mining-related studies which
focus on the effect of the working environment on maintainability (Barabadi et al., 2010), spare part inventory

problems (Louit et al., 2011), and risk-based failure assessment (Petrovic´ et al., 2 ; Gölbaşı and Demirel,

2017).

2.3 The Research Motivation

The recent researches discussed in Section 2.1 showed that multi-component systems haven’t been

discussed adequately in the delay-time literature. The available ones have generally considered certain
configurations of systems such as k-out-of-n or parallel. This paper focuses more on series-parallel systems

that may be configured with multi components in different functional dependencies. In this case, a simulation
algorithm which analyzes the operational conditions of individual components simultaneously in a real-time

horizon with small time increments, and which has not studied with this content previously in the literature is

proposed. The algorithm uses delay-time concept when calculating the probabilities of defects to be detected

in inspections prior to failures. Secondly, in the previous researches, repair time has generally been ignored
(Leung and Kit-Leung, 1996; Redmond et al., 1997; Arthur, 2005; Scarf and Majid, 2011; Wang, 2013; Yang et

al., 2015) or taken as constant (Christer et al., 1998; Wang and Majid, 2000; Jia and Christer, 2002; Pillay et al.,
2004; Wang, 2009). This study regards corrective repair times as a random variable for a more realistic

determination of maintenance downtime costs. Effect of corrective repair duration on deterioration of healthy

components is analyzed within the model and the biased conditions are eliminated so as not to interfere with

the random lifetimes of components.

The studies discussed in Section 2.2 could be utilized in a wide variety of applications in mining sector to

improve maintenance policies and equipment performability. However, none of them have concentrated on

optimizing intervals of inspections which is the most common preventive activity in maintenance policies in

mines. In addition, it is realized that there are very few works which discuss the financial outputs of

maintenance. In this basis, the current study works on draglines since the earthmover industry has the biggest
share among all the mining equipment with an expected future market volume of USD 30 billion in the

following decade. Inspection intervals of two draglines were optimized by using actual datasets of lifetime,

repair time, and maintenance cost gathered from the consultants and the specialists. As it stands, the paper
provides an insight in examining and analyzing inherent factors effective in a mining equipment inspection

for the first time. Therefore, it contributes to the literature by extending the frontiers of the existing
knowledge on mining equipment.

3 Notation and Assumptions

3.1 Notations

Target observation time in simulation

Active time in simulation where

Very small-time increment in comparison with component lifetimes and target period

Lifetime probability density function of component

Repair time probability density function of component

Randomly assigned time-between-failures for component during simulation

Randomly assigned time-to-repair for component during simulation

lifetime start point of component at the timeline during simulation

lifetime finish point of component at the timeline during simulation
lifetime period of component between related and points

total external halt of component during related due to planned system halts

and corrective maintenance of dependent system components

Expected percentile proportion of component’s lifetime just before defect arises

Total number of previous planned halts during inspections and shifts at ,

respectively
Total number of previous preventive maintenances applied in inspections for
 component at
Constant time between inspections, i.e. inspection interval

Constant durations of each inspection and other planned halts during shift,

respectively

Downtime cost per minute

Constant physical corrective and preventive costs of component, respectively

(excluding production loss due to maintenance downtime)

corrective and preventive costs of component at the end of related ,

respectively (including production due to maintenance downtime)

Cumulative downtime cost of planned halts during inspections and shifts at ,

respectively.

Cumulative cost of component at

Cumulative system cost at

3.2 Assumptions

1. A multi-component system with several failure modes is considered in the algorithm. Failure modes
can be functionally-dependent or -independent.

2. Effective service life of each components is assigned randomly by regarding the lifetime probability

density functions.

3. Maintenance recovery level is evaluated individually considering component lifetimes’ trend analysis
and resultant imperfect maintenance approach.

4. System is halted for regular inspections for a constant time interval. Inspection activity is assumed

perfect so that it can catch any defect if it is available. Operations can also be halted for other planned

breaks, e.g. breaks between shift changes.

5. Defects identified during inspections are rectified preventively. The total time required for preventive

maintenance cannot exceed the inspection duration.

6. If one component fails during operation, all functionally-dependent components are halted for a

random time-to-repair period of the failed component. All effort is given to recover the failed
component correctively.

7. Random time-to-repair values are determined with the probability functions of repair downtime

datasets. Each time-to-repair value covers the total duration to identify, repair, and inspect the failure.

8. Corrective maintenance cost is estimated by summation of direct and indirect economic consequences

of failures. Indirect cost is determined from the production loss of repair downtime. Since preventive
 maintenance only takes place in inspection duration, indirect cost of inspection is taken separately.

Therefore, preventive maintenance cost only includes physical cost.

9. Simulation ends at a pre-specified target time, i.e. finite period. In each run, for a given constant
inspection interval, the algorithm is simulated successively up to that the total maintenance cost comes

to a balance. The runs are repeated by incrementing the interval in each run and the interval which

minimizes the total cost gives the optimal solution.

3.3 Stages of the Simulation Algorithm

The proposed simulation algorithm is called the time-counter algorithm and it computes five dependent

sub-events in each loop: time counting, estimation of lifetime period length, maintenance decision-making,
assignment of a start-point for new lifetime, and cost estimation.

Time Counting: The algorithm starts at . Active time, , is updated with the unit time increments of

after each loop. In each up-to-date active time, a control point is created for the simultaneous and holistic
evaluation of system’s components to decide on their maintenance requirements. When reaches to any

production shift or inspection point in the timeline, it counts and updates the total number of shifts or

inspections from the simulation start-up point. In this way, the number of scheduled system downtimes is

updated and stored.

Estimation of Lengths of Initial Lifetime Periods: At , system elements start to operate simultaneously
with random survival times, i.e. , derived from . The algorithm locates the length of the functioning

times for the components, , in real timeline via assigning start ( ) and finish ( ) points for active ones.

During the active lifetimes, points stay stable where points change dynamically with the failures in

other dependent components and scheduled downtimes. Therefore, values for the related lifetimes are

extended dynamically by shifting points. The lengths of these extensions in active are referred to as

the component’s external halt, , and are determined by considering the constant downtime durations of

scheduled activities and random values of other failed components which are operationally dependent.

values do not change effective survival times of components, they only move points on the timeline.

Therefore, can be estimated as in Equations 1-3.

(1)

(2)

k TT if i k
ij (3)
TT ij ,k are dependent components
TT km if i k
 Decision of Maintenance Type: When catches any , the algorithm analyzes the probability of the

concerned defect to be detected during the previous inspection. The algorithm assumes that any component

defect can be detected during inspection if the detectability period starts before the inspection. In this case,

preventive maintenance takes place for the defective component within inspection hours. If there is a sudden

failure without any indication or the detectability period falls between any two inspections, the component is

allowed to fail during operations and immediately after, a corrective maintenance is performed in a
randomly-generated downtime derived from . The mathematical requirement for preventive

maintenance in inspections can be investigated in Equation 4. values in the equation are constant and the
subjectively evaluated values for components themselves, specify the expected percentile portion of

component lifetimes just before the defect identification. If Equation 4 is not satisfied, corrective maintenance
is carried out at the end of the related . Depending on the maintenance decision, a random TT ij value is

generated for the component as given in Equation 5.

(4)

TT ij hen rre tive maintan e is de ided (5)

TT ij
assi n hen reventive maintan e is de ided

Maintenance Cost Estimator: Corrective ( ) and preventive ( ) costs in each are accumulated

under component cost ( ) dynamically. only covers the direct cost of preventive costs ( ) since it

takes place within the inspection duration. In , random indirect cost is also included by considering

random which is derived from . Downtime costs of planned halts are also included separately in

the total cost estimation. Cumulative and up-to-date values are gathered under total system costs ( )

simultaneously. Based on these assumptions, the overall maintenance cost of each component at time can

be estimated as in Equations 6-7 where total system cost including costs due to components and planned halts

can be calculated as in Equations 8-9.

(6)

if
(7)
if

(8)
 (9)

Assignment of New after Component Maintenance: A new random is assigned to the related just

after maintenance of the concerning component. Starting point of new lifetime, , is determined by

considering the type of performed maintenance as: i) summation of the previous and after

corrective maintenance (Equation 10) or ii) finish time of the last inspection after preventive maintenance

(Equation 11).

(10)

(11)

The sub-events of the algorithm successively compute the mentioned variables until target observation

time has arrived, . The main output of the simulation is total system cost at time . The algorithm
simulation is repeated for target observation times until value comes to equilibrium. After achieving

value for a constant interval, , the algorithm run is repeated sequentially by incrementing the interval. .

The minimized point for gives the optimized value for the system as given in Equation 12.

(12)

One illustrative example is given in Figure 1. Here, a system with two functionally-dependent components

is considered. values are highlighted in green in the figure. The first failure attempt in the system is

observed for component-1 ( ). At that moment, the time-counter checks the probability of the failure to be

detected in the previous inspection. Since the delay time zone in does not capture any inspection,
corrective maintenance is applied within random . Therefore, the new lifetime of component-1, ,

starts after corrective maintenance. Simultaneously, deterioration of component-2 is also stopped and its

lifetime, , is extended with this generated value. Then, the ongoing lifetimes of both components
are also extended with inspection durations when the first inspection point is arrived. The second failure

attempt in Figure 1 is experienced for component-2. At the end of , the time-counter detects that the defect

in this component can be maintained in the previous inspection. Therefore, preventive maintenance is

performed for this component and its new lifetime, , is initiated after finish of this inspection. The mutual

interaction continues until that the active time exceeds target observation time, .

Figure 1. Time-Counter Algorithm for Sample Two-Component System Gölbaşı and Demirel,
 The general flowchart of the algorithm is given in Figure 2. Using this algorithm, the effect of inspection

interval changes on corrective and preventive maintenance costs and contribution of individual component

costs to overall system cost can be also evaluated as well as optimizing inspection intervals.

Figure 2. Flowchart of the Inspection Interval Algorithm for a Constant Inspection Interval

A data acquisition methodology is also offered for the developed algorithm (Figure 3). As seen in the

figure, not only quantitative data acquired via maintenance records, but also the subjective evaluation of

maintenance activities with expert opinions of maintenance crew, personal interviews with administrative

staff, and in-situ investigation of the production area are also vital to specify a boundary for both the system

and algorithm.

Figure 3. The Proposed Data Acquisition Methodology for the Algorithm

To clarify and validate the developed model, the algorithm was applied to two draglines currently

operating in an open-cast coal mine in Turkey, as given in Section 4. In this case study, financial worth of
maintenance events and downtime losses, estimations on delay times, and lifetime and repair datasets were

acquired from the maintenance crew at the production area and the senior dragline maintenance experts.

4 Application of the Algorithm: A Case Study for Walking Draglines

Draglines are massive earthmovers extensively utilized in overburden stripping operations of open-cast

coal mines. Draglines perform operations single-handedly without the need of a haulage system. They strip

the overburden material by dragging their buckets suspended from the boom tip towards the main frame.

Draglines have a structural weight of more than 4000 tonnes (Townson et al., 2003) and they may make an

annual production of 30-35 million m3 using buckets with varying capacities up to 125 m3 (Darling, 2011).
Downtimes due to dragline halts may lead to the financial loss of up to 1 million dollars per day (Townson et

al., 2003). Operational severity and the massive structure of these machines require the development and

practicing of comprehensive maintenance policies at mine sites. This study applied the developed inspection
interval algorithm to two draglines in an open-cast coal mine in Turkey by characterizing the

uptime/downtime behavior of dragline components and by estimating financial consequences of maintenance

and resultant downtime.

The algorithm initially requires estimation of lifetime, , and repair time, , probability density

functions of dragline components in order to determine their random uptime/downtime behaviors. In this

sense, Section 4.1 briefly describes datasets, system decomposition, and data trend and correlation analyses.

Section 4.2 states the uptime/downtime characterization of system components. Assumptions on delay times
and financial worth of maintenance activities are mentioned in Section 4.3. Simulation results for the draglines

are given in Section 4.4.

4.1 Preprocessing of Lifetime and Repair Time Datasets

The studied draglines perform their operations using buckets with a capacity of 20 yd3 (15.3 m3) and 40 yd3

(30.6 m3). In the analyses, they are referred to as Dragline-1 and Dragline-2, respectively. According to

maintenance records, operations of Dragline-1 and Dragline-2 are detected to be halted 938 and 903 times

between the period of 1998 and 2011, with a downtime of 13,954 and 16,471 hours, respectively.

Estimation of and functions for individual components entails the sensitive decomposition of

draglines into both subsystems and components considering functional and structural relevancies. In this
basis, a dragline can be decomposed into the following subsystems: i) Bucket unit which creates a contact

zone with ground material for an earthmoving operation, ii) boom unit which creates an operational radius

for bucket suspended from its tip, iii) dragging unit which strips ground material using the bucket which is
thrown away from the main frame, vi) hoisting unit which hoists the filled bucket and dumps into a spoil area

after a swing movement, v) rigging unit which creates a balance in axis of the bucket mouth when hoisting

and dumping it, vi) movement unit which helps rotation during hoisting and dumping activities and

reposition of dragline in production area with walking mechanism, and vii) machinery house unit which
controls all abilities of dragline and converts electrical power to mechanical power via motors and generators

for hoisting, dragging and movement actions. After the allocation of the maintenance records according to the

subsystems, percentile contribution of the subsystems to maintenance breakdowns and maintenance numbers

were revealed as shown in the Pareto charts in Figure 4. It was observed from the charts that the failure
frequency is by far the highest in the dragging unit with 29 percent in Dragline-1 where failure numbers are

more equally distributed among the machinery house, rigging, dragging, and bucket units in between 18-22

percent in Dragline-2. On the other hand, the machinery house is the main determinant of maintenance
downtimes in both draglines with a 56 percent and 47 percent contribution to Dragline-1 and Dragline-2

maintenance-based production loss, respectively.

Figure 4. Pareto Charts for Failure Numbers and Maintenance Durations According to the Subsystems

Uptime, , and downtime, , characterization in the study included failure-inducing components

alone which were detected with maintenance records, personal interviews with experts on dragline
maintenance, and machinery catalogues. Considering the definitions in the quantitative records, the system

was decomposed into bottom elements as far as possible. In this sense, dragline components to be analyzed in

the study, their observed failure conditions and their common maintenance types can be examined in Table 2.

In Table 2, Mode01 and Mode02 denotations were added to the descriptive names of rigging, dragging and

hoisting ropes and the rigging pulley to separate their non-reparability and reparability conditions,
respectively. Mode01 means that replacement is the only option if the component fails where Mode02 points

out that the component can be recovered without replacement if the component is dislocated from its major

mechanism without failure. Other components without any Mode01 and Mode02 expression are repairable.

Besides, it can be observed from Table 2 that the machinery house and movement units could not be

decomposed into the bottom elements sufficiently due to a lack of explanation in the maintenance records.

Therefore, they actually point to failure location rather than components themselves and their failure modes

were specified as general malfunction.

Table 2. Failure-Inducing Dragline Components and Their Failure and Repair Conditions

Lifetime and repair time datasets were allocated to the components. Then, these datasets were subjected to
data correlation and trend tests prior to the characterization process since the non-stability of time series in

the analyses changes the characterization method. In this sense, the Lag-1 scatter plot and Pearson correlation

test can be utilized to measure the correlation among sequential data values where Crow-AMSAA and
Laplace tests can be performed to analyze data trend Gölbaşı and Demirel, . In this basis, versus

values of individual lifetime and repair time values for each component were plotted using a Lag-1

graph. If the plots follow any specific pattern, it may point to possible data dependency. Qualitative outcomes

of the plots were also verified by the Pearson correlation test. An example in Figure 5 shows that the lifetime
data of the Dragline-1 bucket pin component is independently distributed. Other datasets were also tested

and no data dependency was observed.

Figure 5. Data Independency Test for Lifetime Data of Dragline-1 Bucket Pin Component

The second point to be considered in and estimation is the stationarity of the dataset. In case of
any continuous deterioration or upturn condition of components in time, their lifetime intervals can shorten

or extend, respectively. Although parameter estimations for a stationary dataset can be done by fitting data

into a distribution, a non-stationary dataset requires a stochastic process with the capability of measuring
increasing or decreasing trend in a time series. In this basis, Crow-AMSAA and Laplace tests were used in the

detection of data stationarity. Crow-AMSAA accepts the trend between sequential data within a dataset if
or where is the total number of failures, is the expected shape

parameter, is the score of chi-square distribution, and is a confidence interval. can be estimated

using Equation 13 (Wang and Coit, 2005; Demirel and Gölbaşı, . On the other hand, the Laplace test
points to a data trend if or where and are the number of failures and arrival time

(cumulative time between failures) of failure. can be calculated using Equation 14.

(13)
 (14)

Sample data stationarity results for drag chain assembly (DR1) and drag ringbolt (DR2) lifetime datasets
can be examined in Table 3. The tests show that the lifetime data of Dragline-2 DR1 follows a trend where the

others have stationary lifetime values. These tests also revealed that the components coded with HO3, HO5,

BU1, BU3, BU5, RI6, MH1, MH2, MO1, and MO3 for Dragline-1, DR1, HO1, RI1, and MO1 for Dragline-2 have

non-stable lifetime datasets. In addition, there is no data trend in repair time datasets of the components for
both draglines.

Table 3. Trend Tests of DR1 and DR2 Components

4.2 Uptime and Downtime Characterization of the System Components

In the study, parametric values of and functions were determined considering the assumptions

in Figure 6. In this sense, the parameters of the datasets with no correlation and no trend behavior, i.e.

identically and independently distributed (iid), were estimated using best-fit distributions where some other

methods were used for the remaining components. Here, non-stability in the datasets can be due to
deterioration or improvement in component lifetimes and resultant maintenance requirements.

Figure 6. Study Methodology in and Estimations

The obtained and parameters can be examined in Tables 4-5, respectively. In the tables, the p-

values higher than 0.05 for Anderson-Darling statistics in a 95 percent confidence interval show that the

functions provide good fits for the concerned datasets. It was also realized that the majority of functions
with stable datasets are well fitted with a Weibull distribution. The general renewal process (GRP) for

nonstable datasets also has the same descriptive parameters as in a Weibull distribution. Details of this

process can be investigated in Mettas and Zhao (2005). On the other hand, datasets were detected to be
stationary and also well fitted in Lognormal distribution with parameters log(mean), , and log(std), .

Table 4. Parametric Values of functions

Table 5. Parametric Values of Functions in Lognormal Distribution

4.3 Estimation of Delay Times and Maintenance Costs

In the algorithm, the lifetime threshold value determines the useful lifetime of components prior to

defect identification period in a percentage value. Once this identification period starts, the delay time
interval until the lifetime finish time gives an opportunity to maintain the defects which haven’t turned
to failure yet. In the study, preventive maintenance in these kinds of defects are only allowed during regular

inspections. In addition to delay times, financial consequences of maintenance activities are measured in the

study using unit downtime , direct corrective , and direct preventive costs. , , and

values specify constant production loss per minutes in the case of maintenance, physical costs of corrective

maintenance for component, and preventive maintenance for the component, respectively. To set values

for these parameters, one expert with more than 35 years’ ex erien e in dragline maintenance and one

authorized person working as representative of the target draglines in Turkey were requested to fill out a

questionnaire. In this basis, expected and , and values of the components, for which USD/Turkish

Lira currency rate is 2.2, can be viewed in Table 6.

Table 6. Expected Lifetime Threshold and Maintenance Cost Values of the Dragline Components

The cost values in Table 6 give the expectations on the physical costs of failures. In the questionnaires, the

experts stated that the maintenance of the ropes (DR3, HO2, and RI3) is performed out of regular inspection
hours. Besides, the mode02 failure condition in both ropes and pulley (DR4, HO3, RI4, and RI6) occur due to

the dislocation of the components from their major mechanisms and it generally takes place suddenly without

any signs. Therefore, the concerned values were stated as zero for these components where their

values were specified as 100 percent since there is no delay time. Moreover, the values of mode02 failure
conditions in the ropes (DR4, HO3, and RI4) were also given zero since they were assumed to cause

production loss alone, not any physical cost. The boom component is maintained out of the inspection hours

and no estimations were given to both the boom (BO1) and air conditioning unit (MH4). Therefore, the only

production loss due to maintenance is included for these components in the analyses even though they have
some certain maintenance costs.

On the other hand, unit downtime cost, , is the production loss per unit time in terms of USD/min. In the

case that the hourly rate policy is applied for a maintenance crew wage, this amount can also be included into
. However, a constant wage policy is applied in the studied mine and downtime durations do not have any

effect on the wages. Therefore, production loss alone was taken as . In this sense, financial loss in dragline

production can be measured by considering material removed per cycle, effective cycle time, and the financial

value of volumetric production as given in Equation 15. The value of should be estimated differently for

other systems according to their production factors.

(15)

The bank material removed per each cycle in Equation 15 can be expressed by using bucket
volume , swell factor , and the fill factor of the bucket . In a dragline operation area, in-situ
overburden material waiting to be stripped is called bank overburden. This overburden is swelled to an

extent during stripping operation due to stress relaxation. The volumetric ratio between swelled and in-situ

materials is called the swell ratio . Additionally, ratio between operational and actual volumes of the

bucket is the fill factor. In Equation 15, the actual cycle time of a dragline can be estimated using expected
cycle time and operator efficiency . The ratio between resultant material removal per cycle

and actual cycle time gives us the volumetric production rate. Therefore, the value can be estimated by

multiplying volumetric production per minute and financial value of unit volumetric production, as given in

Table 7.

Table 7. Estimation Results for the Target Draglines

Table 7 shows that system halt in Dragline-2 may cause a unit production loss about 73 percent higher

than for Dragline-1. Fill factor, swell factor, cycle time, and operator efficiency in the table are the expected

properties of the studied draglines and the mine site. They were acquired from 5zd ğan 1984). The obtained

values of , , , , , and , are utilized as input for the inspection algorithm.

4.4 Simulation Results

In the current condition, Dragline-1 and Dragline-2 are operated in 3 shifts with 8-hour working periods.
In each shift, a compulsory break of 30 minutes is given due to legal worker rights. Therefore, usage time of

each dragline is 22.5 hours a day. The dragline operations are also halted for 8 hours in regular inspections

with a time interval of 160 hours (one shift a week). Therefore, constant , , and values were

introduced to the algorithm as 8, 0.5, and 160 hours, respectively. The draglines were assumed to be operated
continuously except for these compulsory breaks and random repair times during operations. The target

observation time ( ) was set as 8,766 hours to reveal the annual maintenance cost profile of the systems. For

the given assumptions, the model was computed in 14 runs where inspection intervals in terms of hour were
incremented in each run by 24 between 16 and 326. For each run, the model was simulated 400 times. The

simulation results are given in Figure 7. Annual maintenance costs of the draglines covering direct and

indirect consequences of maintenance events were observed minimized at the inspection interval with 232

and 184 hours for Dragline-1 and Dragline-2, respectively. Compared to the current inspection interval of 160

hours, the optimized intervals can reduce the total cost by 5.9% and 6.2% for Dragline-1 and Dragline-2,
respectively (Table 8). This indicates that the extended inspection intervals for both draglines have a better

capability of minimizing the overall maintenance cost including the physical and nonphysical financial

factors.

Figure 7. Variation of Annual Maintenance Cost with respect to Inspection Intervals

 Table 8. Percentile Change of Cost with respect to Inspection Intervals

The algorithm also estimates the percentile contribution of corrective and preventive maintenance

activities to the overall maintenance cost in different inspection intervals (Figure 8). Production loss due to
regular machine stoppage in inspections and the related physical costs are included into preventive

maintenance cost. On the other hand, corrective maintenance costs cover physical costs of corrective repair

and production loss due to repair downtimes which are randomly derived from in each failure.

Figure 8. Percentile Contribution of Corrective and Preventive Maintenance Costs for Changing Inspection

Intervals

It was revealed from the figure that the percentile weight of the preventive maintenance cost is higher in

Dragline-2, compared to Dragline-1. These cost values converge to each other for the extended inspection

intervals. For the optimized intervals, the percentage of preventive maintenance cost decreases from 23.5 and

25.4 percent to 17.6 and 23.7 percent for Dragline-1 and Dragline-2, respectively. Therefore, it is understood
that these intervals provide a holistic benefit even though they increase the portion of corrective maintenance

cost.

5 Conclusions

Regular inspections are commonly applied in industries as an essential part of maintenance policies to
protect machine health by removing hidden or probable defects in the mechanisms. The time between

inspections should be determined carefully in a way where it should minimize physical and nonphysical cost

effects of inspections. This study proposed a simulation algorithm, referred to as the time-counter, to be

utilized in the cost-effective optimization of inspection intervals. The algorithm checks the operability of
system elements simultaneously in a real-time horizon with small time increments. The decision on

preventive/corrective maintenance for individual components is given according to their random lifetimes

and the related delay time estimations. The generated maintenance costs include direct costs and indirect
costs of downtimes. In the algorithm, the inspection interval is minimized with a sensitivity analysis between

the total maintenance cost per observation period and the different interval lengths. In the study, the

developed algorithm was applied to two draglines by using real lifetime, repair time, and financial datasets,

whose quality and validity were checked by both the maintenance experts and the sales representative of the

draglines. The simulation results showed that the economic consequences of maintenance can be decreased by
5.9 % and 6.2 % for Dragline-1 and Dragline-2 by extending the current interval of 160 hours to 232 and 184

hours, respectively.
 For future studies, the developed algorithm can be extended and improved by considering factors such as

spare part availability, maintenance crew availability, operational dependencies among equipment fleet, and

changes in production rates.

Acknowledgement

This research study was partially supported by The Scientific and Technological Research Council of Turkey
TUBITAK ith r je t number M . The auth rs uld like t thank Dr. Metin 5zd ğan and Enver
Şeker i for their support to this research study in acquiring data and sharing their expertise on dragline
maintenance. Three anonymous reviewers are also acknowledged for their comments in technical and
structural improvement of the paper.
References

1. Andrawus, J. A., Watson, J., Kishk, M., & Gordon, H. (2008). Optimisation of Wind Turbine Inspection
Intervals. Wind Engineering, 32(5), 477-490. doi:10.1260/030952408786411921

2. Arthur, N. (2005). Optimization of Vibration Analysis Inspection Intervals for an Offshore Oil and Gas
Water Injection Pumping System. Proceedings of the Institution of Mechanical Engineers, Part E: Journal of
Process Mechanical Engineering, 219(3), 251-259. doi:10.1243/095440805X8638

3. Aven, T. (2009). Optimal Test Interval for a Monotone Safety System. Journal of Applied Probability, 46(2),
330-341. doi:10.1239/jap/1245676090

4. Barabadi, A., Barabady, J., & Markeset, T. (2011). Maintainability Analysis Considering Time-Dependent
and Time-Independent Covariates. Reliability Engineering and System Safety, 96, 210-217.
doi:10.1016/j.ress.2010.08.007

5. 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 and System Safety, 93, 647-653.
doi:10.1016/j.ress.2007.10.006

6. Barberá, L., Crespo, A., Viveros, P., & Stegmaier, R. (2014). A Case Study of GAMM (Graphical Analysis
for Maintenance Management) in the Mining Industry. Reliability Engineering and System Safety, 121, 113-
120. doi:10.1016/j.ress.2013.07.017

7. Berrade, M. D., Scarf, P. A., & Cavalcante, C. A. (2015). Some Insights Into the Effect of Maintenance
Quality for a Protection System. IEEE Transactions on Reliability, 64, 661-672. doi:10.1109/TR.2015.2417431

8. Berrade, M. D., Scarf, P. A., Cavalcante, C. A., & Dwight, R. A. (2013). Imperfect Inspection and
Replacement of A System with A Defective State: A cost and Reliability Analysis. Reliability Engineering
and System Safety, 120, 80-87. doi:10.1016/j.ress.2013.02.024

9. Cavalcante, C. A., Alencar, M. H., & Lopes, R. S. (2017). Multicriteria Model to Support Maintenance
Planning in Residential Complexes under Warranty. Journal of Construction Engineering and Management,
143(4). doi:10.1061/(ASCE)CO.1943-7862.0001250

10. Christer, A. H. (1973). Innovatory Decision Making. In K. Bowen, & D. J. White (Ed.), The Role and
Effectiveness of Theories of decision in Practice (pp. 369-377). Hodder & Stoughton.

11. Christer, A. H. (1988). Condition-Based Inspection Models of Major Civil-Engineering Structures. Journal
of the Operational Research Society, 39(1), 71-82.

12. Christer, A. H. (1991). Prototype Modelling of Irregular Condition Monitoring of Production Plant. IMA
Journal of Management Mathematics, 3(3), 219-232. doi:10.1093/imaman/3.3.219

13. Christer, A. H., & Lee, S. K. (1997). Modelling Ship Operational Reliability over a Mission under Regular
Inspections. Journal of the Operational Research Society, 48(7), 688-699.

14. Christer, A. H., Wang, W., Baker, R. D., & Sharp, J. (1995). Modelling Maintenance Practice of Production
Plant using the Delay-Time Concept. IMA Journal of Management Mathematics, 6(1), 67-83.
doi:10.1093/imaman/6.1.67

15. Christer, A. H., Wang, W., Sharp, J., & Baker, R. (1998). A Case Study of Modelling Preventive
Maintenance of a Production Plant using Subjective Data. Journal of the Operational Research Society, 49(3),
210-219.
16. Darling, P. (2011). SME Mining Enginering Handbook (3 ed.). Society for Mining, Metallurgy, and
Exploration.

17. Demirel, N., & Gölbaşı, O. . Preventive e la ement De isions for Dragline Components Using
Reliability Analysis. Minerals, 6(2). doi:10.3390/min6020051

18. Dindarloo, S. (2016). Reliability Forecasting of a Load-Haul-Dump Machine: A Comparative Study of

ARIMA and Neural Networks. Quality and Reliability Engineering International, 32, 1545-1552.
doi:10.1002/qre.1844

19. Dindarloo, S. R. (2016). Support vector machine regression analysis of LHD failures. International Journal
of Mining, Reclamation and Environment, 30(1), 64-69. doi:10.1080/17480930.2014.973637

20. Emovon, I., Norman, R. A., & Murphy, A. J. (2016). An Integration of Multi-Criteria Decision Making
Techniques with A Delay Time Model for Determination of Inspection Intervals for Marine Machinery
Systems. Applied Ocean Research, 59, 65-82. doi:10.1016/j.apor.2016.05.008

21. Enders, C. K. (2010). Applied Missing Data Analysis (1st ed.). New York, The USA: Guilford Press.

22. Flage, R. (2014). A Delay Time Model with Imperfect and Failure-Inducing Inspections. Reliability
Engineering and System Safety, 124, 1-12. doi:10.1016/j.ress.2013.11.009

23. Gölbaşı, O., & Demirel, N. . evie f Trend Tests f r Dete ti n f Wear-Out Period for Mining
Machineries. Proceedings of International Mining Congress and Exhibition of Turkey (pp. 933-939). Antalya:
The Chamber of Mining Engineers of Turkey.

24. Gölbaşı, O., & Demirel, N. . O timisati n f Dra line Ins e ti n Intervals ith Time-Counter
Algorithm. International Journal of Mining, Reclamation and Environment , 1-14.
doi:10.1080/17480930.2017.1339168

25. Gölbaşı, O., & Demirel, N. 017). Risk-Based Reliability Allocation Methodology to Set A Maintenance
Priority Among System Components: A Case Study in Mining. Eksploatacja i Niezawodnosc – Maintenance
and Reliability, 19(2), 191-202. doi:10.17531/ein.2017.2.6

26. Gupta, S., Maiti, J., Kumar, R., & Kumar, U. (2009). A Control Chart Guided Maintenance Policy
Selection. International Journal of Mining, Reclamation and Environment, 23(3), 216-226.
doi:10.1080/17480930902916478

27. Gustafson, A., Schunnesson, H., & Kumar, U. (2015). Reliability Analysis and Comparison Between
Automatic and Manual Load Haul Dump Machines. Quality and Reliability Engineering International, 31,
523-531. doi:10.1002/qre.1610

28. Hall, R. A., & Daneshmend, L. K. (2003). Reliability Modelling of Surface Mining Equipment: Data
Gathering. International Journal of Surface Mining, Reclamation and Environment, 17(3), 139-155.
doi:10.1076/ijsm.17.3.139.14773

29. Hall, R. A., Knights, P. F., & Daneshmend, K. (200). Pareto Analysis and Condition-Based Maintenance
of Underground Mining Equipment. Mining Technology, 109(1), 14-22. doi:10.1179/mnt.2000.109.1.14

30. Hoseinie, S. H., Ataei, M., Khalokakaie, R., Ghodrati, B., & Kumar, U. (2012). Reliability Analysis of
Drum Shearer Machine at Mechanized Longwall Mines. Journal of Quality in Maintenance Engineering,
18(1), 98-119. doi:10.1108/13552511211226210

31. Jia, X., & Christer, A. H. (2002). A Periodic Testing Model for a Preparedness System with a Defective
State. IMA Journal of Management Mathematics, 13(1), 39-49.
32. Jodejko-Pietruczuk, A., & Werbińska-Wojciechowska, S. (2014). Analysis of Maintenance Models'
Parameters Estiamtion for Technical Systems with Delay Time. Eksploatacja i Niezawodnosc – Maintenance
and Reliability, 16(2), 288-294.

33. Knights, P. F. (2009). Optimal Replacement Intervals for Shovel Dipper Teeth. International Journal of
Mining, Reclamation and Environment, 23(3), 157-167. doi:10.1080/17480930902916221

34. Leung, F., & Kit-Leung, M. (1996). Using Delay-Time Analysis to Study the Maintenance Problem of
Gearboxes. International Journal of Operations and Production Management, 16(12), 98-105.
doi:10.1108/01443579610151779

35. Lhorente, B., Lugtigheid, D., Knights, P., & Santana, A. (2004). A Model for Optimal Armature
Maintenance in Electric Haul Truck Wheel Motors: A Case Study. Reliability Engineering and System
Safety, 84, 209-218. doi:10.1016/j.ress.2003.10.016

36. Li, X., He, R., Yan, Z., Hu, H., & Cheng, G. (2015). A New Imperfect Maintenance Model Based on Delay-
Time Concepts for Single Components with Multiple Failure Modes. International Journal of Systems
Assurance Engineering and Management, 6(4), 479-486. doi:10.1007/s13198-014-0306-6

37. Louit, D., Pascual, R., Banjevic, D., & Jardine, A. K. (2011). Optimization Models for Critical Spare Parts
InventoriesPA Reliability Approach. Journal of the Operational Research Society, 62, 992-1004.
doi:10.1057/jors.2010.49

38. Martinez, A., Lara, G., Pascual, R., & Droguett, E. L. (2015). Optimal Failure-Finding Intervals for Heat
Shields in a Gas Turbine Combustion Chamber Using a Multicriteria Approach. Journal of Engineering for
Gas Turbines and Power, 137. doi:10.1115/1.4029202

39. Mettas, A., & Zhao, W. (2005). Modeling and Analysis of Repairable Systems with General Repair.
Proceedings of IEEE Reliability and Maintainability Symposium, (pp. 176-182). Alexandria.

40. Okumura, S., Jardine, A. K., & Yamashina, H. (1996). An Inspection Policy for a Deteriorating Single-Unit
System Characterized by a Delay-Time Model. International Journal of Production Research, 34(9), 2441-
2460.

41. 5zd ğan, M. . Çekmeke çe Dra line 5rt(kazı Yöntemleri ve Tunçbilek Uy ulaması. Madencilik,
23(2), 25-42.

42. Pascual, R., Castillo, G. D., Louit, D., & Knights, P. (2009). Business-oriented prioritization: A novel
graphical technique. Reliability Engineering and System Safety, 94, 1308-1313. doi:10.1016/j.ress.2009.01.013

43. Petrovic, D. V., Tanasijevic´ , M., Milic, V., Lilic, N., Stojadinovic, S., & Svrkota, I. (2014). Risk Assessment
Model of Mining Equipment Failure Based on Fuzzy Logic. Expert Systems with Applications, 41, 8157O
8164. doi:10.1016/j.eswa.2014.06.042

44. Pillay, A., Wang, J., Wall, A., Ruxton, T., & Loughran, C. G. (2004). Formal Safety Assessment of Fishing
Vessels: Risk and Maintenance Modelling. Proceedings of the Institute of Marine Engineering, Science and
Technology Part A: Journal of Marine Engineering and Technology(4), 29-42.

45. Rahimdel, M. J., Ataei, M., Khalokakaei, R., & Hoseinie, S. H. (2014). Maintenance Plan for A Fleet of
Rotary Drill Rigs. Archives of Mining Sciences , 59(2), 441-453. doi:10.2478/amsc-2014-0031

46. Ramadan, S. (2016). A Bi-Objective Inspection Policy Optimization Model for Finite-Life Repairable
Systems Using A Genetic Algorithm. Advances in Production Engineering & Management, 11, 38-48.
doi:10.14743/apem2016.1.208
47. Rao, K. R., & Prasad, P. V. (2001). Graphical Methods for Reliability of Repairable Equipment and
Maintenance Planning. Proceedings of IEEE Reliability and Maintainability Symposium, (pp. 123-128).
doi:10.1109/RAMS.2001.902453

48. Redmond, D. F., Christer, A. H., Rigden, S. R., Burley, E., Tajelli, A., & Abu-Tair, A. (1997). O.R.
Modelling of the Deterioration and Maintenance of Concrete Structures. European Journal of Operational
Research, 99(3), 619-631.

49. Roy, S. K., Bhattacharyya, M. M., & Naikan, V. N. (2001). Maintainability and Reliability Analysis of A
Fleet of Shovels. Mining Technology, 110(3), 163-171. doi:10.1179/mnt.2001.110.3.163

50. Samanta, B., Sarkar, B., & Mukherjee, S. K. (2002). Reliability Assessment of Hydraulic Shovel System
Using Fault Trees. Mining Technology, 111(2), 129-135. doi:10.1179/mnt.2002.111.2.129

51. Samanta, B., Sarkar, B., & Mukherjee, S. K. (2004). Reliability Modelling and Performance Analyses of An
LHD System in Mining. The Journal of The South African Institute of Mining and Metallurgy, 104(1), 1-8.

52. Scarf, P., & Majid, H. A. (2011). Modelling Warranty Extensions: A case Study in the Automotive
Industry. Proceedings of the Institution of Mechanical Engineers, Part O: Journal of Risk and Reliability, 225(2),
251-265. doi:10.1177/1748006X11401801

53. Skotnicka-Zasadzień, B., & Bialy, W. . An Analysis f P ssibilities t Use A Paret Chart f r
Evaluatin Minin Ma hines’ Failure Frequen y. Eksploatacja i Niezawodnosc-Maintenance and Reliability,
3, 51-55.

54. Taghipour, S., & Banjevic, D. (2012). Optimal Inspection of a Complex System Subject to Periodic and
Opportunistic Inspections and Preventive Replacements. European Journal of Operational Research, 220,
649-660. doi:10.1016/j.ejor.2012.02.002

55. Tanasijevic, M., Ivezic, D., Jovancic, P., Ignjatovic, D., & Bugaric, U. (2013). Dependability Assessment Of
Open-Pit Mines Equipment O Study On The Bases Of Fuzzy Algebra Rules. Eksploatacja i Niezawodnosc –
Maintenance and Reliability, 15(1), 66-74.

56. The Parker Bay Company. (2017, January 31). Web Site of The Parker Bay Company. Retrieved June 28,
2017, from Mining Equipment Population: http://parkerbaymining.com/mining-
equipment/earthmoving-equipment.htm

57. Townson, P. G., Murthy, D. N., & Gurgenci, H. (2003). Optimization of Dragline Load. In E. W. Blischke,
& D. N. Murthy, Case Studies in Reliability and Maintenance (pp. 517-544). John Wiley & Sons Inc.

58. Uzgören, N., Elevli, S., Elevli, B., & Uysal, Ö. (2010). Reliability Analysis of Draglines' Mechanical
Failures. Eksploatacja i Niezawodnosc-Maintenance and Reliability, 48(4), 23-28.

59. Vagenas, N., & Nuziale, T. (2001). Genetic Algorithm for Reliability Assessment of Mining Equipment.
Journal of Quality in Maintenance Engineering, 302-311. doi:10.1108/13552510110407087

60. Van Oosterom, C. D., Elwany, A. H., Çelebi, D., & Van Houtum, G. J. (2014). Optimal Policies for a Delay
Time Model with Postponed Replacement. European Journal of Operational Research, 232(1), 186-197.
doi:10.1016/j.ejor.2013.06.038

61. Vayenas, N., & Wu, X. (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. doi:10.1080/17480930902916494

62. Vayenas, N., & Wu, X. (2011). Maintenance Study of A Skip Hoist. International Journal of Mining,
Reclamation and Environment, 25(2), 177-186. doi:10.1080/17480930.2010.540369
63. Wang, H., Wang, W., & Peng, R. (2017). A Two-Phase Inspection Model for A Single Component System
with Three- Stage Degradation. Reliability Engineering and System Safety, 158, 31-40.
doi:10.1016/j.ress.2016.10.005

64. Wang, L., Chu, J., & Mao, W. (2009). A Condition-Based Replacement and Spare Provisioning Policy for
Deteriorating Systems with Uncertain Deterioration to Failure. European Journal of Operational Research,
194, 184-205. doi:10.1016/j.ejor.2007.12.012

65. Wang, P., & Coit, D. W. (2005). Repairable Systems Reliability Trend Tests and Evaluation. Annual
liability and Maintainability Symposium, (pp. 416-421).

66. Wang, W. (2009). An Inspection Model for a Process with Two Types of Inspections and Repairs.
Reliability Engineering and System Safety, 94(2), 526-533. doi:10.1016/j.ress.2008.06.010

67. Wang, W. (2011). An Inspection Model Based on a Three-Stage Failure Process. Reliability Engineering and
System Safety, 96(7), 838-848. doi:10.1016/j.ress.2011.03.003

68. Wang, W. (2012). A Stochastic Model for Joint Spare Parts Inventory and Planned Maintenance
Optimisation. European Journal of Operational Research, 216(1), 127-139. doi:10.1016/j.ejor.2011.07.031

69. Wang, W. (2013). Models of Inspection, Routine Service, and Replacement for a Serviceable One-
Component System. Reliability Engineering and System Safety, 116, 57-63. doi:10.1016/j.ress.2013.03.006

70. Wang, W., & Banjevic, D. (2012). Ergodicity of Forward Times of the Renewal Process in a Block-Based
Inspection Model using the Delay Time Concept. Reliability Engineering and System Safety, 100, 1-7.

71. Wang, W., & Majid, H. B. (2000). Reliability Data Analysis and Modelling of Offshore Oil Platform Plant.
6(4), 287-295. doi:10.1108/13552510010346824

72. Wang, W., & Wang, H. (2015). Preventive Replacement for Systems with Condition Monitoring and
Additional Manual Inspections. European Journal of Operational Research, 247(2), 459-471.
doi:10.1016/j.ejor.2015.06.035

73. Wang, W., Banjevic, D., & Pecht, M. (2010). A Multi-Component and Multi-Failure Mode Inspection
Model Based on the Delay Time Concept. Reliability Engineering and System Safety, 95(8), 912-920.
doi:10.1016/j.ress.2010.04.004

74. Wang, W., Carr, M. J., & Chow, T. W. (2012). A Two-Level Inspection Model with Technological
Insertions. IEEE Transactions on Reliability, 61(2), 479-490. doi:10.1109/TR.2012.2183911

75. Wang, W., Zhao, F., & Peng, R. (2014). A Preventive Maintenance Model with a Two-Level Inspection
Policy Based on a Three-Stage Failure Process. Reliability Engineering and System Safety, 121, 207-220.
doi:10.1016/j.ress.2013.08.007

76. Yang, L., Ma, X., & Zhao, Y. (2017). A Condition-Based Maintenance Model for a Three-State System
Subject to Degradation and Environmental Shocks. Computers & Industrial Engineering, 105, 210-222.

77. Yang, R., Zhaofei, F. Z., Kang, J., Li, H., & Teng, H. (2015). Inspection Optimization Model with
Imperfect Maintenance Based on a Three-Stage Failure Process. Eksploatacja i Niezawodnosc, 17(2), 165-
173. doi:10.17531/ein.2015.2.1

78. Zhao, J., Chan, A. H., Roberts, C., & Madelin, K. B. (2007). Reliability Evaluation and Optimisation of
Imperfect Inspections for a Component with Multi-Defects. Reliability Engineering and System Safety, 92,
65-73. doi:10.1016/j.ress.2005.11.003
 TABLES WITH CAPTIONS

Table 1. Population of Active Surface Mining Equipment in the World.

Primary Manufacturers Min. Active Population Population Range

Caterpillar, Komatsu, BelAZ,

Mining Truck 42000 90-360 mton
Hitachi, Liebherr, BEML and XEMC

Electric Shovel Bucyrus, Joy Global/P&H and OMZ 1850 5-64 m³

BEML, Caterpillar, Hitachi, Joy

Hydraulic Excavator 4400 9-44 m³
Global/P&H, Komatsu, Liebherr
Caterpillar, Komatsu and Joy
Wheel Loader 3700 8-50 m³
Global/P&H
Bucyrus, Joy Global/Harnischfeger
Dragline 500 9-126 m³
and OMZ

Mining Dozer Caterpillar and Komatsu 12500 Net Power Rating > 350kW (500 hp)

Mining Grader Caterpillar and Komatsu 2700 Net Power Rating > 160kW (275 hp)

Atlas Copco, Caterpillar, Sandvik,

Drill 2300 Net Pulldown Rating > 18,000 kg (40,000 lbs)
Joy Global, OMZ
 Table 2. Failure-Inducing Dragline Components and Their Failure and Repair Conditions

Unit Code Component Failure Condition Repair Type

DR1 Chain assembly Breakage Replacing and welding of individual chain
DR2 Ringbolt Breakage Welding
Dragging

DR3 Rope-mode01 Rupture Replacement

DR4 Rope-mode02 Dislocation from pulley Recovering the mechanism
DR5 Control General Malfunction General Repair
DR6 Socket Breakage Welding
HO1 Brake Fail to brake Mechanical Repair
HO2 Rope-mode01 Rupture Replacement
Hoisting

HO3 Rope-mode02 Dislocation from pulley Recovering the mechanism

HO4 Sockets Breakage Welding
HO5 Control General Malfunction General Repair
BU1 Bucket Body Wear and tear Welding
BU2 Chain Assembly Breakage Replacing and welding of individual chain
Bucket

BU3 Digging Teeth Dropping, breakage Replacing and welding of individual tooth
BU4 Pins Breakage Replacement of individual pins
BU5 Ringbolt Breakage Welding
RI1 Socket Breakage Welding
RI2 Ringbolt Breakage Welding
Rigging

RI3 Rope-Mode01 Rupture Replacement

RI4 Rope-Mode02 Dislocation from pulley Recovering the mechanism
RI5 Pulley-Mode01 Irrecoverable malfunction Replacement
RI6 Pulley-Mode02 Mechanical disintegration Recovering the mechanism
Removal of brush dust, fixing armatures, bearings or
MH1 Generators General Malfunction
couplings
Removal of brush dust, fixing armatures, bearings or
Machinery

MH2 Motors General Malfunction

House

couplings
Fixing injectors, valves, pumps, air compressors or timing
MH3 Lubrication General Malfunction
mechanism
MH4 Air Conditioning General Malfunction General repair

Fixing transmission box, bearings, felts, pinion gears, turret

MO1 Rotation General Malfunction
traversing mechanism, rails or flanges
Movement

Fixing transmission box, bearings, felts, walking axle, journal

MO2 Walking General Malfunction
bearing, pins or steel construction of walking feet

MO3 Warning General Malfunction Fixing connection couplings or warning brushes

Boom

BO1 Boom Chords Fracture Preventive Welding

 Table 3. Trend Tests of DR1 and DR2 Components

Dragline-1 Dragline-2
Test Name Test Statistics
DR1 DR2 DR1 DR2
134.8 54.0 108.2 37.3

Crow 108.3 43.2 60.4 33.1

AMSAA 162.0 79.1 101.9 65.2
Decision No Trend No Trend Trend No Trend
-0.59 0.43 -2.76 0.14
Laplace 1.64 1.64 1.64 1.64
Decision No Trend No Trend Trend No Trend
 Table 4. Parametric Values of functions
Dragline-1 Dragline-2
Code
Model Parameter p-value Model Parameter p-value
Dragging Unit
DR1 Weibull-3P 0,258 GRP Not iid*
DR2 Weibull-2P >0,250 Weibull-3P 0.354
DR3 Loglogistic-2P 0,168 Weibull-3P >0.500
DR4 Weibull-3P 0,233 Weibull-3P >0.500
DR5 Weibull-2P >0,250 Weibull-3P >0.500
DR6 Weibull-2P >0,250 Lognormal-2P 0.364
Hoisting Unit
HO1 Lognormal-2P 0,284 GRP Not iid*
HO2 Loglogistic-2P 0,205 Normal-2P 0.93
HO3 GRP Not iid* Lognormal-2P 0.519
HO4 Weibull-2P >0,250 No Failure Data
HO5 GRP Not iid* Weibull-2P 0.16
Bucket Unit
BU1 GRP Not iid* Weibull-3P 0.492
BU2 Weibull-2P >0,250 Exponential-2P >0.250
BU3 GRP Not iid* Weibull-2P 0.191
BU4 Weibull-3P >0,500 Weibull-3P >0.500
BU5 GRP Not iid* Weibull-3P >0.500
Rigging Unit
RI1 Weibull-2P >0,250 GRP Not iid*
RI2 Weibull-2P 0,224 Weibull-2P >0.250
RI3 Weibull-3P >0,500 Loglogistic-2P 0.178
RI4 No Failure Data Weibull-2P >0.250
RI5 Lognormal-2P 0,836 Normal-2P 0.882
RI6 GRP Not iid* Weibull-3P >0.500
Machinery House Unit
MH1 GRP Not iid* Weibull-3P 0.475
MH2 GRP Not iid* Exponential-2P >0.250
MH3 Exponential-2P >0,250 Lognormal-2P 0.339
MH4 No Failure Data Lognormal-2P 0.212
Movement Unit
MO1 GRP Not iid* GRP Not iid*
MO2 Weibull-2P 0,156 Weibull-3P >0.500
MO3 GRP Not iid* Exponential-2P >0.250
Boom Unit
BO1 Weibull-3P >0,250 Exponential-1P 0.348
*Not identically and independently distributed
Table 5. Parametric Values of Functions in Lognormal Distribution

Dragline-1 Dragline-2
Code
p-value MTTR (hr.) p-value MTTR (hr.)
Dragging Unit
DR1 1.16 0.71 0.380 4.11 0.96 0.53 0.888 3.01
DR2 0.75 0.75 0.239 2.80 0.48 0.49 0.139 1.82
DR3 1.22 0.77 0.061 4.56 1.64 0.80 0.442 7.11
DR4 0.60 0.80 0.306 2.50 0.35 0.59 0.090 1.69
DR5 0.76 1.14 0.092 4.08 1.16 1.17 0.071 6.38
DR6 0.52 0.80 0.593 2.31 0.16 0.37 0.100 1.25
Hoisting Unit
HO1 0.43 0.78 0.139 2.08 0.59 1.07 0.110 3.18
HO2 2.05 0.78 0.072 10.54 2.40 0.69 0.476 13.99
HO3 0.37 0.70 0.871 1.84 0.49 0.35 0.192 1.74
HO4 1.24 1.36 0.593 8.77 No Repair Data
HO5 0.77 1.56 0.287 7.25 0.85 1.32 0.086 5.60
Bucket Unit
BU1 0.48 0.83 0.679 2.28 1.00 1.21 0.094 5.68
BU2 0.81 0.52 0.230 2.59 1.22 1.03 0.737 5.74
BU3 0.82 0.83 0.103 3.20 -0.02 0.64 0.068 1.21
BU4 0.03 0.57 0.087 1.21 0.08 0.61 0.091 1.30
BU5 0.70 0.77 0.075 2.70 0.43 0.63 0.191 1.88
Rigging Unit
RI1 0.33 0.52 0.912 1.59 0.16 0.70 0.512 1.49
RI2 0.02 0.46 0.268 1.13 0.51 0.64 0.066 2.05
RI3 0.11 0.53 0.086 1.28 0.44 0.58 0.072 1.83
RI4 No Repair Data 0.48 0.59 0.271 1.91
RI5 0.83 0.69 0.651 2.92 0.72 0.69 0.252 2.59
RI6 0.36 0.61 0.484 1.73 0.31 0.78 0.083 1.84
Machinery House Unit
MH1 3.96 1.40 0.547 139.38 2.63 1.95 0.091 92.7
MH2 2.98 1.70 0.091 83.16 2.76 1.73 0.065 70.07
MH3 0.20 0.68 0.207 1.53 0.76 1.04 0.139 3.65
MH4 No Repair Data 0.34 0.74 0.118 1.85
Movement Unit
MO1 0.59 0.96 0.353 2.89 0.55 1.09 0.085 3.14
MO2 0.84 1.39 0.604 6.07 1.46 1.56 0.071 14.63
MO3 1.46 1.58 0.083 14.86 1.23 1.27 0.929 7.70
Boom Unit
BO1 1.17 1.38 0.983 8.31 5.52 0.53 0.237 287.21
Table 6. Expected Lifetime Threshold and Maintenance Cost Values of the Dragline Components

Dragline-1 Dragline-2
Code
(%) (USD/failure) (USD/failure) (%) (USD/failure) (USD/failure)
DR1 90 568 65 95 1,114 98
DR2 85 80 16 90 56 16
DR3 100 644 0 100 1,132 0
DR4 100 0 0 100 0 0
DR5 85 386 196 90 500 295
DR6 95 80 65 95 95 65
HO1 80 45 33 85 45 65
HO2 100 705 0 100 1,216 0
HO3 100 0 0 100 0 0
HO4 95 80 65 No failure data
HO5 90 523 123 95 591 164
BU1 95 227 262 95 309 327
BU2 90 186 196 90 295 229
BU3 95 84 65 95 109 65
BU4 90 386 98 90 659 98
BU5 90 245 245 90 614 245
RI1 95 25 16 95 34 16
RI2 90 105 109 90 164 131
RI3 100 107 0 100 98 0
RI4 100 0 0 100 0 0
RI5 98 447 22 95 843 33
RI6 100 459 0 100 655 0
MH1 95 300 164 90 364 295
MH2 98 186 44 95 159 87
MH3 75 273 49 80 341 49
MH4 No estimation No estimation
MO1 98 2,955 327 95 3,977 589
MO2 95 1,795 491 90 2,205 785
MO3 90 227 196 90 291 393
BO1 No estimation No estimation
 Table 7. Estimation Results for the Target Draglines

Factors Dragline-1 Dragline-2

Bucket Capacity, Vbucket (m3) 15.29 30.58
Fill Factor, F 0.85 0.85
Swell Factor, S 1.45 1.45
Expected Cycle Time, Tcycle (min) 0.75 0.87
Operation/Operator Efficiency, (%) 73 73
Unit Production Loss, ($/bank m ) 3
0.60 0.60
Production Loss ($/min) 5.23 9.03

Table 8. Percentile Change of Cost with respect to Inspection Intervals

Inspection Interval Dragline-1 Dragline-2
(hours) Total Cost ($) Change (%) Total Cost ($) Change (%)
16 953,511 +75.2 1,599,547 +83.0
40 672,219 +23.5 1,033,099 +18.2
64 604,655 +11.1 893,757 +2.3
88 577,143 +6.0 892,107 +2.1
112 547,009 +0.5 966,352 +10.6
136 550,133 +1.1 832,068 -4.8
160 544,219 0.0 873,946 0.0
184 548,292 +0.7 819,428 -6.2
208 532,885 -2.1 886,724 +1.5
232 512,151 -5.9 871,997 -0.2
256 564,688 +3.8 921,108 +5.4
280 558,785 +2.7 924,187 +5.7
304 572,038 +5.1 1,036,597 +18.6
 START

Set Initial Lifetime Start

Active time, Yes Points, TS Component
t ? for i=1,…,n where n is the lifetime data
number of components
No

t t ∆t
here t ,..,∞
a d ∆t ≪ t

Yes
t >t ? Display C END

No

Yes Update and

A y i spectio  
Num Num Store as
ti e is arri ed?
Num

No

Yes Update and

A y shift cha ge  Num Num Store as
ti e is arri ed? Num
No

Input data of
components

Calculate system halts

between TS and t for
each component

Update and
Store as D

Update Lifetime Finish Points,

TF TS TBF D

No Yes Component wear−out Yes Perform

TF t ? can be catched in the preventive
recent inspection? maintenance

No
Perform
corrective Calculate
maintenance Cpm

Store Generate
TTR random TTR
values in 95% C.I.

Calculate
Ccmij

SUM

Update C for i=1,…,n

where n is the number
of components

Recalculate Lifetime Start

Point,TS , for i=1,…,n
where n is the number of
components

Store previous
system halts
before TS for
each component

Generate new
random TBF
in 95% C.I.

Start Estimation of
f ( x ) i and g ( x ) i
Request maintenance
failure/repair time
records and
machinery catalogues
Detect failure-intensive
components in the system
Decompose data into the
components and perform
data debugging
Perform data independency
and data trend tests and
reliability analysis. Estimate
function parameters
START
Start Estimation of Start Estimation of Start Estimation of
Cci ,Cpi ,and Thsi TBI, Tsh ,and Tins Cu
Prepare a questionnaire to be asked to Request information Request information about
maintenance experts with the following about inpection intervals the following proposed
proposed content: and all planned halts and content:

their constant durations
 Work packages of regular Economic
measurement unit of

inspections?
Components prevetively the operation ($/m3, $/

ton, $/min or $/unit) ?

maintained in inspections?
Components only correctively Factors to convert the
measurement unit to $/

maintained out of inspections?
min ?

Inspection groups of components?
Failure modes and repair types of

components?
Preventive/corrective maintenance

costs of components?
Subjective estimation on lifetime
threshold values (proportion of
defect detectable period to whole
lifetime for each components)?
END
 Branching-Poisson General renewal
process process

No No

Data Is dataset independently Is dataset identically

decomposition distributed? Yes distributed without trend? Yes
Best-fit distribution
into system - Lag-1 Scatter plots - Crow-AMSAA test
components - Pearson correlations - Laplace test
 HIGHLIGHTS OF THE MANUSCRIPT

 A simulation algorithm, called the time-counter, was developed for cost-effective optimization of

inspection intervals of production systems.

 A data acquisition strategy was presented to be utilized in the algorithm.

 The offered model was applied to two active draglines by using actual financial, lifetime and repair time
datasets.

 Sensitivities of annual corrective and preventive costs to inspection intervals were evaluated.