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

net/publication/321651592

Production performance analysis during operation phase: A case study

Article in Proceedings of the Institution of Mechanical Engineers Part O Journal of Risk and Reliability · December 2017
DOI: 10.1177/1748006X17744383

CITATIONS READS

12 2,751

3 authors:

Ali Nouri Abbas Barabadi

UiT The Arctic University of Norway UiT The Arctic University of Norway
56 PUBLICATIONS 406 CITATIONS 66 PUBLICATIONS 997 CITATIONS

SEE PROFILE SEE PROFILE

Yonas Zewdu Ayele

Oslo Metropolitan University
50 PUBLICATIONS 425 CITATIONS

SEE PROFILE

All content following this page was uploaded by Ali Nouri on 02 July 2021.

The user has requested enhancement of the downloaded file.

Original Article

Proc IMechE Part O:

J Risk and Reliability
1–17
Production performance analysis Ó IMechE 2017
Reprints and permissions:
during operation phase: A case study sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/1748006X17744383
journals.sagepub.com/home/pio

Ali N Qarahasanlou1, Abbas Barabadi2 and Yonas Z Ayele2,3

Abstract
Production performance analysis plays a significant role in supporting the decision-making process, for managers and
engineers dealing with the challenges of the optimization procedure of a system delivery capacity. Operational conditions
can influence the production performance of a system drastically, in a short or long term, by affecting the system config-
uration, reliability, maintainability, maintenance supportability, and functional capacity of its components. Moreover, these
impacts can lead to increased business risks and uncertainties. Such a situation demands the successful application of
tools and methodologies to minimize the total business risk through an accurate prediction of production performance,
addressing the effects of operational conditions. However, there is a lack of practical methodologies, for production per-
formance analysis of systems operating under the time-dependent operational conditions. Hence, the central thrust of
this article is to develop a systematic methodology, for the production performance analysis of a system considering the
time-dependent operational conditions. In the case study, the results show that the production performance of mining
equipment is affected significantly by the operational conditions. These changes occur across the working shifts and due
to any changes in the production plan of the mine. The result of the analyses shows that the critical items in the mine is
the loader. Hence, in order to increase the production performance of the system, the optimization of the reliability or
maintainability of the loader needs to be considered as a priority. Moreover, the wagon drill has a significant excess
capacity, which means that, by adding more trucks to the system, the production performance of the system can be
improved.

Keywords
Reliability, maintainability, maintenance supportability, production performance

Date received: 6 September 2017; accepted: 30 October 2017

Introduction and functional capacity (RMSF) of its items are main

terms of the PP. Figure 1 illustrates the terms of PP
Modern manufacturing and production systems have and their relationship. In general, reliability, maintain-
become more and more specialized, large, complex, ability, and maintenance supportability are intercon-
automated, and integrated. However, with the nected concepts; and, they can be significantly affected
increased mechanization and complexity in these sys- by the operational conditions, which the system is
tems, there is a rise in the number of unplanned com- expected to work.9–11
ponent stoppage (failure) scenarios.1 Failure of In real situations, there are many cases where these
components incurred downtime and unavailability of operational conditions may change during the opera-
the system, which can cause substantial production per- tion phase. For instance, in mine production industry,
formance (PP) losses and affect business perfor- the rock type can change rapidly in different sections of
mance.2,3 Furthermore, the other negative impacts of
system failure(s) include high maintenance costs and, in 1
Faculty of Mining, Petroleum & Geophysics, Shahrood University of
some cases, exposure to accidents.4,5 Technology, Shahrood, Iran
To ensure the production goals, a systematic analy- 2
UiT The Arctic University of Norway, Tromsø, Norway
3
sis of the PP of the system considering all influencing Faculty of Engineering, Østfold University College, Fredrikstad, Norway
factors are required. PP, in general, is defined as the
Corresponding author:
capacity of a system to meet a specified demand for Abbas Barabadi, UiT The Arctic University of Norway, Langnes, Tromsø
deliveries or performance.6 The system configuration, 9037, Norway.
reliability, maintainability, maintenance supportability, Email: abbas.b.abadi@uit.no
2 Proc IMechE Part O: J Risk and Reliability 00(0)

Figure 1. The relationship between the terms of production performance.

Source: Adapted from Barabady7 and Norsok Standard.8

the mine, causing different levels of stress on some min- and maintainability (RM) of a system; and, they did
ing machines, such as drill wagons and bulldozers. not consider the supportability of the system and its
Consequently, these different levels of stress would components in the analysis. Moreover, these studies
change the reliability of the equipment, and, subse- mainly deal only with the time-independent operational
quently the PP of the system. Hence, the rock type may conditions. That means that most of the available mod-
be considered as a time-dependent influencing factor in els and methodologies cannot be effectively employed
the PP analysis of these machines. The operational con- for analysing the PP of a given system during the opera-
ditions can be altered due to various factors, such as tion phase where the operational conditions may
changing on the operator and maintenance crew perfor- change according the production plan. These conse-
mance, organization and team management strategy, quently causes significant problems, to the engineers
customer demand, and ambient environment condi- and managers, when analysing PP of a system during
tions (e.g. temperature). operation phase.15,17,18
The effect of time-dependent operational conditions Based on the above discussion, it is an important
can lead to increased business risks and uncertainties. requirement to consider the impact of the time-
Hence, for minimizing the total business risks, it is dependent operational conditions when analysing the
essential to predict the performance of the system pre- PP of a system. In this article, a new methodology is
cisely, by employing various methodologies and by proposed, which considers the complex and fast-
considering the impact of time-dependent operational changing nature of the time-dependent operational
conditions. conditions. This article seeks to quantify the effect of
Over the years, several methodologies and standards time-dependent operational conditions on the system
have been proposed, for PP of a system.2,12–14 RMSF. The application of the proposed methodology
Barabady and colleagues,2,14 for instance, employed the is demonstrated by a case study of the PP of mining
proportional hazard model (PHM) for quantifying the equipment in Azarbayjan Molybdenum–Copper Mine
effect of the operational conditions on the PP of off- located in Iran.
shore production facilities. Naseri and Barabady14 sug- The rest of the article is organized as follows. The
gested an expert-based model for predicting the PP in proposed methodology is discussed in section ‘PP anal-
oil and gas facilities. Aven and Pedersen12 proposed a ysis in the operation phase’. Thereafter, the application
model for quantifying uncertainties in relation to the of the model is demonstrated by a real case study, in
PP analyses. Hjorteland et al.13 established a guideline section ‘Case study – PP analysis of mine production
based on the Bayesian approach for quantifying the systems’. Finally, section ‘Conclusion’ presents the con-
uncertainties in the PP analyses throughout the various cluding remarks.
phases of a project, prior to production. However, most
of available models or methodologies are primarily
PP analysis in the operation phase
developed for analysing the PP of a given system for the
oil industry2,10,15,16 which make it challenging to use in The production assurance concept was first developed
the other sectors such mining industry. Additionally, by the Norwegian oil industry (see e.g. Norsok Z-016
these models mainly focused on analysing the reliability Standard8). Thereafter, it became the basis for ISO/CD
Qarahasanlou et al. 3

Figure 2. Production performance analysis considering time-dependent operational conditions.

20815. Production assurance is a term used to describe proposed and depicted in Figure 2. The proposed
how capable a system is, to meet the demand for PP. methodology has three key elements:
Hence, to have an effective production assurance man-
agement, an accurate estimation of PP is needed. PP  Estimating the baseline parameters of RMSF for all
can be quantified using various measures, such as pro- items;
duction availability, throughput capacity, deliverabil-  Updating the RMSF for all items as the covariates’
ity, or demand availability.6 Estimating the PP of a level changes;
specific system has two stages. First, the RMSF of all  Building up the system configuration, estimate the
items, which builds up the system, should be estimated. PP, and monitor the system.
Second, based on the configuration, which depicts the
pattern of how different items connected to each other In the first stage, of the proposed methodology, all
for building up the whole system, the PP of the system items that build up the system should be identified. The
can be estimated. Here, for quantifying the effect of the exact definitions of items and system boundaries are
time-dependent operational conditions on the RMSF necessary in order to estimate the RMSF parameters.
parameters, a systematic methodology has been In general, events that occur only inside the system
4 Proc IMechE Part O: J Risk and Reliability 00(0)

Figure 3. Data analysis using PHM and its extension.

boundary should be considered for PP analysis.19 After based on the nature of the collected data (e.g. homoge-
identification of the items and their boundaries, all fac- neous, heterogeneous, time dependency of the covari-
tors that may affect the failure mechanism, repair pro- ates, and trend type), an appropriate regression model
cess, and maintenance logistic should be identified and need to be employed, for estimating the RMSF baseline
considered as covariates. Covariates, in the context of parameters. For some guidelines regarding homoge-
this article, are defined as all those factors which may neous data analysis, see the literature,21–23 and for het-
have an influence on the RMSF of the items that build erogeneous data, see, for example, Garmabaki et al.24
up the system. Examples of the covariates are the sur- For instance, when the collected RM data are not
rounding environment (e.g. temperature, humidity, and homogeneous, the covariates-based model such as
dust), condition indication parameters (e.g. vibration PHM can be used for RM parameters analysis.17,20,24,25
and pressure), design modification, the skill of the oper- Figure 3 shows a systematic guideline for RM analysis
ator and maintenance crew, the repair history, and using PHM. In a nutshell, this figure demonstrates the
so on.20 specific steps that help the PP analyst to find the most
Furthermore, these covariates generally can be cate- suitable alternative model for analysing the heteroge-
gorized as time-independent and time-dependent. In neous data. For instance, one of the key steps is that
particular, the time-dependent covariates possess after formulation of the covariate, the omission of cov-
diverse levels; the possible levels of each time-dependent ariates and their dependency should be checked. If
should also be identified for precise estimation of the there is a dependency between covariates, a new covari-
RMSF parameters. For instance, on some mine pro- ate can be introduced, for instance, by combining of
duction sites, the dry summertime may increase the dust the covariates.20 Later, if the assumption of the propor-
level, and consequently, such conditions may affect the tionality (the ratio of any two hazard rates are constant
reliability of some equipment, while dust will not be an with respect to time) is justified, the basic PHM can be
issue during the wintertime. Hence, in this case, the dust used; otherwise, the extension PHM (e.g. stratified
is considered as a reliability covariate, with two levels – PHM) should be used.26 Graphical and numerical
summer level and winter level. A binary covariate can models can be used to check the proportionality (PH)
be used to represent these two levels, for example, 1 for assumption. Numerical tests such as Schoenfeld resi-
summer dust level and 0 for winter dust level. For an duals are used the goodness-of-fit (GOF) tests to check
effective PP analysis, an accurate prediction of the the PH assumption.20 For more information regarding
RMSF parameters of the items for various levels of the PHM and its application, see the literature.2,26,27
covariates should be carried out. Hence, as the covari- In addition, in some cases, there might not be covari-
ates’ level of a given item is changed, the RMSF perfor- ates associated with the collected failure or repair data.
mance of the item should then be updated. In such scenarios, a classical approach, such as a point
After identifying of the RMSF covariates, the base- process model, can be used for data analysis.24 The best
line operational conditions need to be identified. The model for a specific item of data can be selected based
baseline operational conditions, in the context of this on the trend behaviour and dependency of the collected
article, are referred as the operational conditions that data.28,29 Figure 4 shows a guideline regarding how
their effect of the covariates on RMSF parameter of the these tests can be applied. For instance, in the case of no
item can be estimated with the high degree of accuracy. trend, the renewal processes (RP) method (fitting best
Each item may have its own RMSF baseline condition; classic distributions such as 2P-Weibull or 3P-Weibull)
and, the baseline condition should be defined in such a is the suitable model for analysis, while, in the case of a
way to represent the most dominant operational condi- trend, the non-homogeneous Poisson process (NHPP)
tions that the items will experience during its mission model can be used for the analysis. For trend testing, a
time (planned usage time for the items). set of tests, including Mann Test, Laplace Test,
Thereafter, once the baseline conditions are identi- Likelihood-ratio test, and Military HB Test, need to be
fied, in the next stage, the baseline RMSF should be applied.22,23,30 Once the model is selected, then the para-
estimated for all items. For this purpose, all historical meters should be estimated.
PP data (e.g. time-to-failure (TTF), time-to-repair Furthermore, in some cases, the historical PP data
(TTR), and time-to-delivery (TTD)) and their associ- are inadequate or their quality is not acceptable. In
ated covariates should be collected. Subsequently, such cases, surrogate data, generic information,
Qarahasanlou et al. 5

Figure 4. Data analysis using the classical methods.

scientific knowledge about the design, manufacture, PP analysis can be performed by analytical or simula-
and operation of the equipment can be used directly or tion methods. Simulations are costly and time-consum-
in combination with (limited) system-specific opera- ing; however, they have great capacity for capturing the
tional data for PP analysis.15,31,32 Moreover, direct time-dependent behaviour of the systems.
operational experience and test data from identical Finally, during the operation phase, operational
items under identical/different environmental condi- conditions should be monitored carefully, and all
tions can be used for this aim. The Offshore and related PP data and information should be collected.
Onshore Reliability Data (OREDA) database, Non- Thereafter, whenever it is needed, the RMSF should be
electric Parts Reliability Data,33 and MIL-HDBK- updated. PP data should be reported in such a way that
217F34 are some example of the available databases, enables systematic PP analysis to be carried out.6
which can be used, in particular, for reliability analysis. Hence, the data should be collected in the right format
In the next stage, after quantifying the baseline to facilitate their application. For this aim, the operator
RMSF of all items, the time dependency of the covari- and maintenance crew should be trained regarding the
ates should be checked. For this aim, identified covari- data collection process. Moreover, all data should be
ates should be mapped, for checking any changes in the available for all involved parties in PP analysis. For
covariates’ level during the mission time of a particular details of how to effectively collect a PP data, see ISO/
item. By collecting this information, we can find DIS 20815:20086 and ISO 14224:2006.41
whether a significant change in the RMSF of items
should be expected during its mission time. Later, this
Case study – PP analysis of mine
collected data and information will be used to update
the RMSF parameters of those items. Mathematical production systems
methods and expert opinions can be used to update the Mining is an important industry that provides raw
RMSF of items.14,35 For instance, the PHM or acceler- materials, which mainly are essential inputs for other
ated failure time model (AFTM) can be used as a industries. Nowadays, mining equipment is becoming
mathematical tool to update RMSF parameters in the large in size and complex. In addition, any unplanned
different levels of the covariates.9,20,26,36 stoppages (failures) of equipment will significantly
Once updating the RMSF parameters of all items, the reduce the PP of the mining process. Furthermore,
next step is to build up the configuration of a system these failures would cause significant financial losses.
using an appropriate tool. The most widely used tools for Azarbayjan Molybdenum–Copper Mine is the
this purpose are reliability block diagrams (RBDs), fault second-largest copper mine in Iran with an estimated
trees (FTs), and reliability graphs (RGs).37,38 Although deposit of 828 million tons with the average copper
RBD, RG, and FT are commonly used, they are limited grade of 0.62%. The Azarbayjan Molybdenum–Copper
in their capacity to model systems that have no sequential Mine (Sungun Copper Mine) is located 75 km north-
relationships among their subsystem failures. Moreover, west of the provincial town of Ahar, East Azarbaijan,
they do not provide the capabilities to model RMSF Varzaqan County, Iran. The region where the mine is
interactions among components or subsystems, or to rep- located is regarded as one of the coldest regions in Iran
resent changes in system performance configuration due with an average temperature of 0°C. Furthermore,
to the time-dependent operational conditions. Dynamic heavy precipitation and heavy fog are common atmo-
fault trees (DFTs),39 reliability phase diagrams (RPDs),19 spheric phenomenon all year round.
and dynamic reliability block diagrams (DRBDs)40 are The production line of the mine is illustrated in
the three most popular models that can be used to model Figure 5. The line comprises one drill wagon, one bull-
the time-dependent behaviour of the system. dozer, one loader, and six trucks. Furthermore, the set
In the next step, after estimating the RMSF of all of a mining fleet and the corresponding codes, as well
items at the different levels of the covariates and system as their mean functional capacity, are depicted in
configuration, the PP of the system can be estimated. Table 1.
6 Proc IMechE Part O: J Risk and Reliability 00(0)

Figure 5. Block diagram for a line of the production at the Sungun Copper Mine.

Table 1. The mining fleet characteristics.

Row Mine equipment Model Code Mean functional

capacity (m3/h)

1 Wagon drill Hasher Wa. 384

2 Bulldozer Caterpillar-d11n Bl. 258
3 Loader Caterpillar-988b Lo. 300
4 Dump truck Komatsu-785-5 DT.1 24
5 Dump truck Komatsu-785-5 DT.2 24
6 Dump truck Komatsu-785-5 DT.3 24
7 Dump truck Komatsu-785-5 DT.4 24
8 Dump truck Komatsu-785-5 DT.5 24
9 Dump truck Komatsu-785-5 DT.6 24

The mine has five different working zones including Estimating the RM and maintenance supportability
Monzonite – South, Trachyte – North, Monzonite – of items
Middle, and Ore-Sulphur-Oxide – Dump.
Referring the proposed methodology (Figure 2), once
To analyse the PP of the line of the production, the
items are identified, then the baseline conditions and
proposed methodology (Figure 2) is employed. The fol-
covariates should be defined. In this case study, the
lowing basic assumptions and limitations govern this
baseline condition and covariates were identified based
PP analysis:
on discussions with the experts working at the mine site.
 For this aim, first, the PP analysis procedures and its
The mining process consists of 10 identical line of
concept were described for the mine experts to ensure
the production. One of the production lines is illu-
that all involved experts had a common understanding.
strated in Figure 5.
 Thereafter, with the teamwork of the experts, using
Each mining machine (wagon drill, bulldozer, loa-
archival records, documentation, and direct observa-
der, and dump truck) is considered as an item.
 tions, the covariates and their potential levels were iden-
The defined items have two states: working state
tified for all defined items.
(there is production) or failure state (there is no
production).
 Maintenance supportability data of items have no RM analysis using PHM. The identified RM covariates
covariates. for various items, as well as their potential levels, are
 The functional capacity of items is considered con- depicted in Table 2. As Table 2 shows, some of the
stant where the nominal value is the mean of the reliability covariates are common between all items,
functional capacity of each machine between 2013 including working shift, rock type, site visibility, preci-
and 2015 (Table 1). pitation, temperature, and road condition. However,
some items have their own reliability covariates. For
Here, the deliverability is used to quantify the PP of example, as the drilling depth increases, the stress on
mining equipment. Deliverability, in general, is defined the wagon drill will increase, which can increase the
as the ratio of deliveries to the planned deliveries over a failure rate of this item. Hence, the drilling depth is
specific period of time, when the effect of compensating defined as a reliability covariate for drilling wagons. In
elements, such as substitution from other producers addition, after blasting, rock fragmentation may have a
and downstream buffer storage is included.8 significant effect on the reliability of the loader; hence,
Table 2. Identified reliability and maintainability covariates.

Reliability covariates Maintainability covariates

Loader Dump truck Drilling Bulldozer All items

Covariate Covariate level Covariate Covariate level Covariate Covariate level Covariate Covariate level Covariate Covariate level
Qarahasanlou et al.

Working shift ðzl1 Þ Morning[0] Working shift ðzt1 Þ Morning[0] Working shift ðzd1 Þ Morning[0] Working shift ðzb1 Þ Morning[0] Shift ðwx1 Þ Morning[0]
Afternoon[1] Afternoon[1] Afternoon[1] Afternoon[1] Afternoon[1]
Night[2] Night[2] Night[2] Night[2] Night[2]
Rock type ðz10 Þ Ore[2] Rock type ðzt10 Þ Ore[0] Rock type ðzd2 Þ Ore[0] Rock type ðzb6 Þ Ore[0] Weather Sunny & Clear[0]
Monzonite[1] Monzonite[1] Monzonite[1] Monzonite[1] Condition Semi cloudy[1]
Trakhite[2] Trakhite[2] Trakhite[2] Trakhite[2] ðwx3 Þ Overcast[2]
Site visibility ðzl4 Þ Sunny[0] Site visibility ðzt6 Þ Sunny[0] Site visibility ðzd3 Þ Sunny[0] Site visibility ðzb2 Þ Sunny[0] Dense fog[3]
Semi cloudy[1] Semi cloudy[1] Semi cloudy[1] Semi cloudy[1] Precipitation Continuous covariate
ðwx4 Þ
Overcast[2] Overcast[2] Overcast[2] Overcast[2] Temperature Continuous covariate
ðwx5 Þ
Dense fog[3] Dense fog[3] Dense fog[3] Dense fog[3] Involved one[0]
Precipitation (zl4 ) Continuous covariate Precipitation ðzt7 Þ Continuous covariate Precipitation (zd4 ) Continuous Precipitation ðzb3 Þ Continuous maintenance More than one[1]
covariate covariate crew ðwx2 Þ
Temperature (zl5 ) Continuous covariate Temperature ðzt8 Þ Continuous covariate Temperature (zd5 ) Continuous Temperature ðzb4 Þ Continuous
covariate covariate
Road condition ðzl7 Þ Normal[0] Road condition ðzt9 Þ Normal[0] Road condition ðzd6 Þ Normal[0] Road Normal[0]
Abnormal[1] Abnormal[1] Abnormal[1] condition ðzb5 Þ Abnormal[1]
Working place ðzl2 Þ Dump[0] Rock fragmentation Ore, Oxide, Sulphur, Bit size ðzd7 Þ Suitable (6.5$)[0] Bulldozer Most Suitable (same
ðzt4 Þ Dump[0] displacement level and fix in N-S)
ðzb7 Þ [0]
Mining face[1] Middle-Monzonite[1] Unsuitable Suitable (same level and
( . or \ 6.5$)[1] variable in N-S) [1]
Rock Ore, Oxside, Sulphur, North-Monzonite[2] Drilling depth (zd8 ) Very Suitable[0] Unsuitable (variable level
fragmentation Dump[0] and variable in N-S) [2]
ðzl9 Þ Middle-monzonite[1] Trakhite[3] Suitable [1] Working Dump[0]
North-monzonite[2] South-Monzonite[4] Unsuitable [2] place ðzb8 Þ Mining face[1]
Trakhite[3] Slope of road ðzt5 Þ Same level[0]
South-monzonite[34] Down[1]
Allocated truck Suitable (0–12) [0] Up[2]
to loader ðzl3 Þ Partly suitable (12–20) [1] Hauling distance Short distance
ðzt11 Þ (0–1350)m[0]
Unsuitable (20–100) [2] Normal distance (1350–
2300) [1]
Allocated truck Suitable (Truck 30t) [0] Long distance (2300–
capacity ðzl8 Þ 8000)m[2]
Suitable (Truck 85) [1] Allocated truck to Suitable (0–13) [0]
Unsuitable (Truck 100t) [2] loader ðzl3 Þ Partly suitable (13–43) [1]
Unsuitable (43–176) [2]
Allocated loader Suitable (Truck 30t) [0]
capacity ðzl2 Þ Suitable (Truck 80) [1]
Unsuitable (Truck 100t) [2]
Dump truck code DT.1
(dummy variable) DT.2
DT.3
DT.4
DT.5
DT.6
7
8 Proc IMechE Part O: J Risk and Reliability 00(0)

Supportability
rock fragmentation is considered as a potential covari-
ate for the reliability of the loader.

(min)
TTD
data

43
43

64
The maintainability covariates include the shift, site

258
172

107

129
visibility, precipitation, and the number of involved
maintenance crews. The shift generally represents the

Temperature
diverse maintenance crew, whose different skills and
expertise may have an effect on the maintainability per-

(°C)

11
11
11
1
5

7
5
formance of the mining equipment. In addition, some
maintenance tasks can take long time to be completed.

Precipitation
In such scenarios, several maintenance crews will be
working to repair the failed items during the course of

(mm)

0.2

0.3
0.4

0.6
a number of shifts. That means that under such condi-

0
0

0
tions, repair crews need effective communication.

condition
Weather
Ineffective communication will reduce the maintainabil-
ity significantly, as some jobs need to be repeated or

1
1
1
1
1
1
2
must be double-checked. Hence, the number of mainte-

maintenance
nance crews working on a failed item is considered as a

Involved
maintainability covariate. Moreover, as most of the

crew
maintenance is performed outdoors, the visibility at the

1
1
0
0
1
0
0
site and precipitation are considered as covariates.
Thereafter, the covariate levels were identified based

Working
on the production plan and different operational condi-

shift

1
1
1
1
1
2
1
tions that items will experience during their mission Maintainability data
time. The mine production is carried out in three differ-
ent shifts: (1) Shift A, which is the morning shift (7 h); Status

0
1
1
1
1
1
1
(2) Shift B, which is the afternoon shift (8.5 h); and (3)
Shift C, which is a night shift (6 h). Each shift has its
TTRs
(min)

5
5

3
64
94

70
13
own operator and maintenance crew with different
backgrounds (e.g. skills, experience, attitude, and cul-
Drilling
depth

ture). These differences may have different effects on

the RM of the defined items. Hence, three levels of cov-
0
0
0
1
0
1
ariates need to be associated with the shifts. The other 0
size
Bit

reliability covariate with various levels is rock fragmen-

0
0
0
0
0
0
0

tation. In general, different rock fragmentations are

condition

expected at the mine site, which is due to the current

Road

blasting practices and geological characteristics of the

0
0
0
0
0
0
0

rock. Here, based on the site observation and the opin-

ion of the mining experts, rock fragmentation is defined
Temperature

as a categorical covariate with four possible levels.

After identifying the potential covariates and their
(°C)

9
7
5
5
10
11
Table 3. A sample of the collected data of wagon drill (Wa.).

associated level, the baseline conditions for each item

were identified and formulated. In Table 2, the covari-
Precipitation

ate levels, which are highlighted in italics and bold font,

are the baseline conditions. For instance, morning shift
(mm)

0.3
1.2

0.1
0.6

(Shift A) is considered as baseline conditions for work-

0
0

ing shift. The numbers in brackets represent the covari-

visibility

ate coding. Having the baseline conditions, baseline

Site

TTR: time to repair; TTD: time to delivery.

RM should be estimated.
1
1
1
2
1
2
2

Here, the RM data are heterogeneous. Hence,

Rock
type

Figure 3 can be used as a guideline to find a suitable

2
2
2
2
2
2
2

model for RM analysis. For carrying out the RM anal-

ysis, the historical operational records of daily opera-
Working

tion and maintenance reports in the period from 2006

shift

1
0
0
0
1
0
1

to 2015 are selected. In this period, for each recorded

failure, the time-between-failures (TBF) and TTR were
Status

extracted. Moreover, the covariates associated with

Reliability data

0
1
1
1
1
1
1

each failure and repair process were identified and for-

mulated. A sample of the collected data for wagon drill
TBFs
(h)

(Wa.) is shown in Table 3. A cell with a value of zero in

62
24
15
11

31
1

1
Qarahasanlou et al. 9

Figure 6. Baseline reliability and maintainability of items.

Table 4. The result of Schoenfeld residuals for Bl. TBFs.

Covariate Pearson correlation p-value Result

Working shift 20.008 0.946 PH justified

Road condition 0.034 0.772 PH justified
Bulldozer displacement 20.056 0.626 PH justified

the column ‘Status’ indicates that the Wa. was stopped where i indexes failure times, and j indexes failure and
due to the failure of another item, while 1 shows the suspension times.
Wa. was stopped due to its own failure. In general, the likelihood function (equation (2)) can
In RM analysis, it is assumed that there is no omis- be used to estimate the PHM parameters. Tables 5 and
sion or interaction between the covariates. Thus, the 6 depicted the covariates that have a significant effect
Schoenfeld residuals are used for checking the PH on the RM performance of each item. In both tables,
assumption.20 Table 4 illustrates the Schoenfeld resi- the third column shows the estimates of regression
duals test for failure data of the bulldozer (Bl.). Here, coefficients corresponding to each covariate. The
the null hypothesis is that the PH is not valid. Based on fourth column shows the p-values for testing the signifi-
the p-value in Table 4, it can be concluded that the null cance of each coefficient. The fifth column depicts the
hypothesis is rejected and then the PH is justified. Weibull distribution parameters. The last column
Hence, the PHM can be used for RM analysis of the shows the reliability (Table 5) and maintainability
defined items. (Table 6) of items, considering the effect of all effective
By considering the Weibull distribution as the base- covariates. In addition, Figure 6 depicts the baseline
line repair rate, the repair rate of an item can be written RM of items.
as30

b t  g b1 Maintenance supportability analysis using point process

mðt, uÞ = expðruÞ ð1Þ model. As mentioned above, the collected maintenance
l l
supportability data do not possess any covariates.
where b, l, and g are the shape, scale, and location Thus, a classical approach, such as a point process
parameters of the Weibull distribution, respectively. model, can be used for maintenance supportability
Based on the same principle, the likelihood function for analysis.24 Table 3 illustrates the extracted maintenance
equation (1) is expresses as30 supportability data of wagon drill. Here, the main vari-
" # able for the maintenance supportability analysis is the
Yb ti b1 TTD of request resources that are required for the
Lðb, l, rÞ = exp(rui ðti Þ
l l repair such as the spare parts, repair tools, and repair
i
2 0 tj 13 ð2Þ crew. As it is shown in Figure 4, the trend test should
ð
Y   t be carried out to find an appropriate model for the
4 @ exp ruj ðtÞ d bA 5
l data analysis. Table 7 shows the result of a trend test of
j
0 the maintenance supportability data of defined items.
 10
Table 5. Estimated parameters of the reliability performance of items.

Sub. Covariates a p-value Weibull parameters Reliability in the present of all covariates
b h g
!(0:646zd1 1:121zd7 0:586zd8 )
Wa. Working shift ðzd1 Þ 20.646 0.003 1.062 3.824 0.375  
t  0:375 1:062
RWa: ðt, zÞ = exp 
3:824
Bit size ðzd7 Þ 21.121 0.000
Drilling depth (zd8 ) 20.586 0.000
!(0:409zb1 + 1:429zb5 + 0:395zb7 )
Bl. Shift (zb1) 20.409 0.060 1.065 26.845 0.375  
t  0:375 1:065
RBl: ðt, zÞ = exp 
26:845
Road condition (zb5) 1.429 0.000
Bulldozer displacement (zb7) 0.395 0.049
!(0:505zl1 0:489zl2 + 0:260zl4 0:086zl5 )
Lo. Shift (zl1) 20.505 0.007 1.165 10.527 0.592  
t  0:592 1:165
RLo: ðt, zÞ = exp 
10:527
Working place (zl2) 20.489 0.004
Site visibility (zl4) 0.260 0.007
Precipitation (zl5) 20.086 0.020
DT. Working shift ðzt1 Þ 20.312 0.000 1.109 19.577 0   (0:312zt1 0:457zt3 + 0:071zt4
t 1:109 + 0:430zt5 + 0:083zt6 + 0:224zt9 0:624zt10 + 0:640zt11 + aztdummy )
RD:T: ðt, zÞ = exp 
19:577
Allocated truck to loader ðzl3 Þ 20.457 0.000
Rock fragmentation ðzt4 Þ 0.071 0.052
Slope of road ðzt5 Þ 0.430 0.000
Site visibility ðzt6 Þ 0.083 0.006
Road condition ðzt9 Þ 0.224 0.001
Rock type ðzt10 Þ –0.624 0.000
Hauling distance ðzt11 Þ  0.640 0.000
Dump truck code ztdummy 0.075
Proc IMechE Part O: J Risk and Reliability 00(0)
Qarahasanlou et al. 11

1:510 (3:176wt2 0:099wt3 bztdummy )

Table 7. The results of trend and autocorrelation tests of the
system’s items.

1:315 (2:366wd1 3:391wd2 )

1:047 (0:551wb1 2:250wb2 )

1:230 (2:06wl2 0:261zl3 )

Item Trend tests MIL-HDBK-189 Laplace’s Mann–
Kendall
Maintainability in the presence of all covariates

Wa. Statistic 381.43 0.58 –

p-value 0.617 0.561 –
Result No Trend
Bl. Statistic 38.88 21.58 –
p-value 0.056 0.113 –
MWa: ðt, w Þ = 1  exp  t0:781
0:436

MLo: ðt, w Þ = 1  exp  t0:375

MBl: ðt, w Þ = 1  exp  t0:375

MD:T: ðt, w Þ = 1  exp  2:604

3:114
1:763

Result No Trend

t
Lo. Statistic 42.11 20.65 –
 
 

 
 

p-value 0.894 0.518 –

Result No Trend
DT. Statistic 215.33 127.01 52.96
p-value 0 0 0
Result No Trend
0.781

0.375

0.375
g

0
0.436

1.763

3.545

2.604
Weibull parameters
h

1.315

1.047

1.166

1.510
b
p-value

0.000

0.000
0.005

0.000
0.000

0.001
0.001

0.001
0.007

Figure 7. The maintenance supportability of selected items.

Table 6. Estimated parameters of the maintainability performance of items.

From the result, it can be deduced that the TTD

data of all items are trend free and there is no depen-
22.366

23.391

22.250

20.261
23.176

20.099
2.551

22.06

dency between them. Hence, to find the best-fit classi-

cal distribution, Weibull, lognormal, logistic, normal,
b

and exponential distributions were nominated. The

analysis of the best-fit distribution for the collected
TTD and the estimated parameters are depicted in
Involved maintenance crew ðwd1 Þ

Involved maintenance crew ðwb2 Þ

Involved maintenance crew ðwt2 Þ

Involved maintenance crew ðwl2 Þ

Table 8. Furthermore, the maintenance supportability

of some of the selected items is depicted in Figure 7.

Weather condition ðwl3 Þ

ðwt3 Þ
Truck type wtdummy

Update the baseline parameters of RMSF based on

Working shift ðwd1 Þ

Working shift ðwb1 Þ

Weather condition

covariates


After estimation of the baseline parameters, the ques-

Covariates

tion regarding whether any significant changes in the

RMSF of the items are expected to occur during the
mission time of the system needs to be answered.
The production plan of the mine company stated that
in the coming year, around 7,005,510 t of ore rock
should be delivered to the production process. In addi-
Sub.

Wa.

DT.
Lo.
Bl.

tion, when extracting the required amount of ore, there

12 Proc IMechE Part O: J Risk and Reliability 00(0)

Table 8. The best-fit distribution and related parameters for selected items.

Item Best-fit distribution Parameters Mean time to Baseline supportability

delivery (min)
   
Wa. Loglogistic-2P m = 21.139; s= 0.711 54 lnðtÞ + 1:139 lnðtÞ + 1:139
sWa: ðtÞ = exp =1 + exp
0:711 ! 0:711
Bl. Weibull-3P b = 0.643; h= 0.392; g = 0.024 34  
t  0:024 0:643
sBl: ðtÞ = 1  exp 
0:392
  !
Lo. Weibull-3P b = 0.881; h= 0.405; g = 0.023 27 t  0:023 0:881
sLo: ðtÞ = 1  exp 
0:405
  
DT. Weibull-2P b = 0.696; h= 0.017 20 t 0:696
sBl: ðtÞ = 1  exp 
0:017

Figure 8. Baseline and updated failure and repair rates of Bl. in Phase 2 during different shifts.

is also a production of waste rock, which is about shift is smaller than the day shift, by a factor equal to
44,474,522 t. The production plan also stated that in the the exp (ad1 ) = exp (  0:646) = 0:629. Hence, the RM
coming year, the mining equipment will work at three of the items should be updated during each shift as well
different locations including: (1) monzonite rock for as during each phase. The expected RM covariates and
5 months (from March to August), (2) on ore rock for their associated levels in each phase are depicted in
2 months (from August to October), and finally, (3) on Table 9. Having the expected covariates level in each
monzonite rock for 5 months (from October to March). phase (Table 9) and the regression coefficient of covari-
These locations named as Phase 1, Phase 2, and Phase ates (Tables 5 and 6), the RM of items in each location
3, respectively. This means that, during the mission time can be updated.
of the system, the items will experience different levels Figure 8 shows a comparison between the baseline
of RM covariates. For instance, the depth of the drill and update failure and repair rate of Bl. in Phase 2 dur-
blast holes will change during different phases. As ing the different shifts.
Table 5 demonstrates, the depth of blast hole has a sig-
nificant effect on the failure rate of the wagon drill.
Building up the system configuration, estimating the
Hence, as the depth of blast hole is changed, the relia-
bility of the wagon drill should be updated. Moreover, PP, and monitoring the system
each item will operate under three different shifts (Shift The items exhibit dynamic behaviour due to the effect
A, Shift B, and Shift C) a day, 365 days of the year. of the time-dependent operational conditions during
Hence, the items will also experience different levels of the day and across the defined phases. Hence, the
covariates in a single day. For example, as Table 5 selected model for system configuration should be able
shows, the failure rate of the wagon drill on the night to address this dynamicity. RPDs are a practical and
Table 9. The covariates and their expected levels in the defined phases.

Equipment Phase 1 (from March to August) Phase 2 (from August to October) Phase 3 (from October to March)
Reliability covariates Maintainability covariates Reliability covariates Maintainability covariates Reliability covariates Maintainability covariates

Wa. ðzd1 = 0, 1, 2Þ (wd1 = 0, 1, 2) ðzd1 = 0, 1, 2Þ (wd1 = 0, 1, 2) ðzd1 = 0, 1, 2Þ (wd1 = 0, 1, 2)

Qarahasanlou et al.

ðzd7 = 1, 1, 1Þ (wd2 = 0, 0, 0) ðzd7 = 0, 0, 0Þ (wd2 = 1, 1, 1) ðzd7 = 1, 1, 1Þ (wd2 = 1, 1, 1)

ðzd8 = 1, 1, 1Þ ðzd8 = 1, 1, 1Þ ðzd8 = 1, 1, 1Þ
Bl. ðzb1 = 0, 1, 2Þ (wb1 = 0, 1, 2) ðzb1 = 0, 1, 2Þ (wb1 = 0, 1, 2) ðzb1 = 0, 1, 2Þ (wb1 = 0, 1, 2)
ðzb5 = 0, 0, 0Þ (wb2 = 0, 0, 0) ðzb5 = 0, 0, 0Þ (wb2 = 1, 1, 1) ðzb5 = 1, 1, 1Þ (wb2 = 1, 1, 1)
ðzb7 = 0, 0, 0Þ ðzb7 = 1, 1, 1Þ ðzb7 = 1, 1, 1Þ
Lo. ðzl1 = 0, 1, 2Þ (wl2 = 0, 0, 0) ðzl1 = 0, 1, 2Þ (wl2 = 0, 0, 0) ðzl1 = 0, 1, 2Þ (wl2 = 0, 0, 0)
ðzl2 = 0, 0, 0Þ (wl3 = 0, 0, 0) ðzl2 = 1, 1, 1Þ (wl3 = 2, 2, 2) ðzl2 = 2, 2, 2Þ (wl3 = 3, 3, 3)
ðzl14 = 0, 0, 0Þ ðzl14 = 2, 2, 2Þ ðzl14 = 3, 3, 3Þ
ðzl5 = 1:4, 1:4, 1:4Þ ðzl5 = 0:1, 0:1, 0:1Þ ðzl5 = 1, 1, 1Þ
DT.1 ðzt1 = 0, 1, 2Þðzt3 = 0, 0, 0Þ (wt2 = 0, 0, 0) ðzt1 = 0, 1, 2Þðzt3 = 1, 1, 1Þ (wt2 = 0, 0, 0) ðzt1 = 0, 1, 2Þðzt3 = 0, 0, 0Þ (wt2 = 0, 0, 0)
ðzt4 = 0, 0, 0Þðzt5 = 0, 0, 0Þ (wt3 = 0, 0, 0) ðzt4 = 1, 1, 1Þðzt5 = 0, 0, 0Þ (wt3 = 2, 2, 2) ðzt4 = 2, 2, 2Þðzt5 = 0, 0, 0Þ (wt3 = 3, 3, 3)
ðzt6 = 0, 0, 0Þðzt9 = 0, 0, 0Þ (wtdummy = 1, 1, 1) ðzt6 = 2, 2, 2Þðzt9 = 0, 0, 0Þ (wtdummy = 1, 1, 1) ðzt6 = 3, 3, 3Þðzt9 = 0, 0, 0Þ (wtdummy = 1, 1, 1)
ðzt10 = 0, 0, 0Þðzt11 = 0, 0, 0Þ ðzt10 = 1, 1, 1Þðzt11 = 0, 0, 0Þ ðzt10 = 0, 0, 0Þðzt11 = 0, 0, 0Þ
     
ztdummy = 1, 1, 1 ztdummy = 1, 1, 1 ztdummy = 1, 1, 1
13
 14
Table 10. Phase deliverability analysis of the selected system.

Phase Items Shift A Shift B Shift C

Downtime Uptime Delivery Excess Downtime Uptime Delivery Excess Downtime Uptime Delivery Excess
(h) (h) (m3) capacity (m3) (h) (h) (m3) capacity (m3) (h) (h) (m3) capacity (m3)

Phase 1 Wa. 88 997 278,910 103,761 165 1153 359,873 82,869 366 564 186,657 29,782
Bl. 77 1008 25,332 12,986 96 1222 32,595 13,824 57 873 16,738 16,435
Lo. 270 815 217,898 26,539 216 1102 281,151 49,358 78 852 145,826 109,907
DT.1 127 958 22,723 262 118 1200 28,274 514 68 862 18,983 1695
DT.2 108 977 23,195 264 100 1217 28,697 521 58 872 19,236 1699
DT.3 151 934 22,167 254 140 1178 27,756 505 83 847 18,660 1674
DT.4 161 924 21,920 254 150 1167 27,511 500 88 842 18,553 1655
DT.5 133 952 22,594 261 123 1194 28,144 512 71 859 18,919 1690
DT.6 130 955 22,672 259 120 1198 28,227 513 68 862 18,985 1696
Phase 2 Wa. 223 57 13,860 7919 189 151 13,380 44,658 124 110 4446 37,642
Bl. 26 254 5325 82,562 29 311 1080 10,724 10 224 389 8105
Lo. 21 259 10,828 66,985 116 224 10,453 56,658 98 136 3473 37,319
DT.1 16 264 1779 4568 14 326 1722 6098 8 226 576 4850
DT.2 13 267 1790 4614 12 328 1728 6144 7 227 577 4880
DT.3 19 261 1749 4521 17 323 1707 6034 9 225 570 4821
DT.4 20 260 1738 4492 19 321 1689 6013 11 223 570 4790
DT.5 16 264 1778 4558 15 325 1716 6087 8 226 577 4839
DT.6 16 264 1771 4569 14 326 1717 6099 8 226 572 4855
Phase 3 Wa. 264 786 189,270 112,451 529 746 86,242 200,201 411 489 17,724 170,093
Bl. 599 451 6088 11,053 376 899 3330 30,816 105 795 753 29,439
Lo. 293 757 147,868 79,283 521 754 67,377 158,716 443 457 13,847 123,268
DT.1 156 894 20,499 959 146 1129 11,137 15,961 85 815 2303 17,252
DT.2 134 916 21,010 963 126 1149 11,415 16,167 72 828 2352 17,513
DT.3 184 866 19,840 948 173 1102 10,836 15,601 102 798 2254 16,905
DT.4 197 853 19,554 925 187 1088 10,705 15,414 110 790 2234 16,738
DT.5 163 887 20,324 959 154 1121 11,090 15,822 88 812 2286 17,196
DT.6 160 890 20,386 983 150 1125 11,107 15,882 86 814 2289 17,253
Proc IMechE Part O: J Risk and Reliability 00(0)
Qarahasanlou et al. 15

Figure 9. Reliability phase diagram of production line.

item, with a FCI value of 29%; in Phase 2, Wa. is the

most critical item, with 31% of system failure being
caused by the failure of Wa.; and, finally, in Phase 3,
the Lo., with FCI value of 43%, is the most critical
item.

Conclusion
In the real world, engineers and managers may face
many problems in the process of analysing the PP of a
system considering the effect of time-dependent opera-
tional conditions. Operational conditions may change
the RMSF parameters of items, as well as the config-
uration of the system. Such effects will change the
availability of the system in complex ways and conse-
quently the PP of the system. In this study, a systematic
approach has been developed to quantify the effect of
Figure 10. Mean availability of the system in Shift A for three time-dependent operational conditions on the PP of a
phases. system. In this model, in the first stage, the baseline
operational conditions for all items are defined; there-
suitable model for this type of system. Figure 9 illus- after, the RMSF parameters of items will be quantified
trates the RPD of the defined three phases. As men- under the baseline operational conditions. As the
tioned, the mine works 21.5 h daily in three separate operational conditions change, the RMSF parameters
shifts with the same configuration, namely, Shift A of the items will be updated and modified. The updat-
(7 h), Shift B (8.5 h), and Shift C (6 h). ing of the RMSF parameters is a challenging task that
Here, the PP of the system is predicted using a simu- needs accurate data and information on the failure
lation model for 1 year of operation (12 months). The mechanism and the repair and logistic process, as well
results of the analysis are presented in Table 10. In as on the operational conditions of the system.
Table 10, the item PP (or delivery) is the total amount In the case study, the results of the PP analysis show
of rock processed/produced by the item over the analy- that the RM of mining equipment in Sungun Copper
Mine is affected significantly by the operational condi-
sis time (a year). In Table 10, excess capacity is the
tions. These changes occur across the working shifts, as
additional amount of rock that an item could have pro-
well as based on any changes in the production plan of
cessed/produced while up and running. The excess
the mine. RPD can be used to model such a system
capacity is not processed by the item, because the pro-
with time-dependent RMSF characteristics. In the case
duction line is stopped due to the failure of other items.
study, the application of a PHM for parameters’ esti-
For example, the excess capacity of the loader can be
mation and updating is illustrated. The result of the
due to a failure on trucks or the wagon drill.
analyses shows that the critical item in the mine is the
Moreover, the mean availability of production line
loader. Hence, in order to increase the PP of the sys-
during the different phases is illustrated in Figure 10.
tem, the optimization of the reliability or maintainabil-
The mean availability is the proportion of time during
ity of the loader needs to be considered as a priority.
a mission that the production line is available for use.
Moreover, the wagon drill has a significant excess
From the result (Figure 10), it can be deduced that the
capacity, which means that, by adding more trucks to
mean availability of the production line during Phase 3,
the system, the PP of the system can be improved.
that is, during the harsh wintertime, is much lower than
in Phase 1 and Phase 2. Finally, to find the critical
items, the failure criticality index (FCI) was estimated. Declaration of Conflicting Interests
The FCI is defined as the percentage of times that a The author(s) declared no potential conflicts of interest
failure of specific items caused a production line failure. with respect to the research, authorship, and/or publi-
The result shows that, in Phase 1, the Lo. is the critical cation of this article.
16 Proc IMechE Part O: J Risk and Reliability 00(0)

Funding 16. Moreno-Trejo J, Kumar R and Markeset T. System

assurance of subsea petroleum production systems: a case
The author(s) received no financial support for the
study mapping factors influencing the sourcing strategy.
research, authorship, and/or publication of this article. Int J Syst Assur Eng Manag 2012; 3: 255–265.
17. Barabadi A, Gudmestad OT and Barabady J. RAMS
References data collection under Arctic conditions. Reliab Eng Syst
Safe 2015; 135: 92–99.
1. Hassan J, Khan F and Hasan M. A risk-based approach
18. Barabady J, Markeset T and Kumar U. Review and dis-
to manage non-repairable spare parts inventory. J Qual
cussion of production assurance program. Int J Qual
Mainten Eng 2012; 18: 344–362.
Reliab Manag 2010; 27: 702–720.
2. Gao X, Barabady J and Markeset T. An approach for
19. Barabadi A, Barabady J and Markeset T. A methodology
prediction of petroleum production facility performance
for throughput capacity analysis of a production facility
considering Arctic influence factors. Reliab Eng Syst Safe
considering environment condition. Reliab Eng Syst Safe
2010; 95: 837–846.
2011; 96: 1637–1646.
3. Furuly S, Barabadi A and Barabady J. Reliability anal-
20. Kumar D and Klefsjö B. Proportional hazards model: a
ysis of mining equipment considering operational
review. Reliab Eng Syst Safe 1994; 44: 177–188.
environments: a case study. Int J Perform Eng 2013; 9: 21. Barabady J and Kumar U. Reliability analysis of mining
287–294. equipment: a case study of a crushing plant at Jajarm Baux-
4. Barabady J and Kumar U. Availability allocation ite Mine in Iran. Reliab Eng Syst Safe 2008; 93: 647–653.
through importance measures. Int J Qual Reliab Manag 22. Garmabaki A, Ahmadi A, Block J, et al. A reliability
2007; 24: 643–657. decision framework for multiple repairable units. Reliab
5. Pandey MD, Cheng T and Van der Weide J. Higher Eng Syst Safe 2016; 150: 78–88.
moments and probability distribution of maintenance 23. Louit DM, Pascual R and Jardine AK. A practical proce-
cost in the delay time model. Proc IMechE, Part O: J dure for the selection of time-to-failure models based on
Risk and Reliability 2016; 230: 354–363. the assessment of trends in maintenance data. Reliab Eng
6. ISO/DIS 20815:2008. Petroleum, petrochemical and nat- Syst Safe 2009; 94: 1618–1628.
ural gas industries – production assurance and reliability 24. Garmabaki AHS, Ahmadi A, Mahmood YA, et al. Relia-
management, https://www.iso.org/standard/69983.html bility modelling of multiple repairable units. Qual Reliab
(accessed 30 August 2017). Eng Int 2016; 32: 2329–2343.
7. Barabady J. Production assurance (concept, implementa- 25. Deloux E, Dijoux Y and Fouladirad M. Generalization
tion and improvement). Division of Operation and Main- of the proportional hazards model for maintenance mod-
tenance Engineering, Luleå University of Technology, elling and optimization. Proc IMechE, Part O: J Risk and
Luleå, 2007. Reliability 2012; 226: 439–447.
8. Norsok Standard. Z-016: regularity management & 26. Barabadi A, Barabady J and Markeset T. Maintainabil-
reliability technology. Oslo: Norwegian Technology ity analysis considering time-dependent and time-
Standards Institution, 1998, https://www.standard.no/ independent covariates. Reliab Eng Syst Safe 2011; 96:
no/Nettbutikk/produktkatalogen/Produktpresentasjon/ 210–217.
?ProductID=132422 27. Carr M and Wang W. A case comparison of a propor-
9. Ayele YZ, Barabadi A and Barabady J. Dynamic spare tional hazards model and a stochastic filter for condition-
parts transportation model for Arctic production facility. based maintenance applications using oil-based condition
Int J Syst Assur Eng Manag 2016; 7: 84–98. monitoring information. Proc IMechE, Part O: J Risk
10. Barabadi A. Reliability and spare parts provision consid- and Reliability 2008; 222: 47–55.
ering operational environment: a case study. Int J Per- 28. Eryilmaz S. Review of recent advances in reliability of
form Eng 2012; 8: 497–506. consecutive k-out-of-n and related systems. Proc IMechE,
11. Hu J, Jiang Z and Wang H. Preventive maintenance for a Part O: J Risk and Reliability 2010; 224: 225–237.
single-machine system under variable operational condi- 29. Medjoudj R, Aissani D, Boubakeur A, et al. Interruption
tions. Proc IMechE, Part O: J Risk and Reliability 2016; modelling in electrical power distribution systems using
230: 391–404. the Weibull–Markov model. Proc IMechE, Part O: J Risk
12. Aven T and Pedersen LM. On how to understand and and Reliability 2009; 223: 145–157.
present the uncertainties in production assurance analy- 30. Jardine A, Anderson P and Mann D. Application of the
ses, with a case study related to a subsea production sys- Weibull proportional hazards model to aircraft and
tem. Reliab Eng Syst Safe 2014; 124: 165–170. marine engine failure data. Qual Reliab Eng Int 1987; 3:
13. Hjorteland A, Aven T and Østebø R. Uncertainty treat- 77–82.
ment in production assurance analyses throughout the 31. Naseri M, Baraldi P, Compare M, et al. Availability
various phases of a project. Reliab Eng Syst Safe 2007; assessment of oil and gas processing plants operating
92: 1315–1320. under dynamic Arctic weather conditions. Reliab Eng
14. Naseri M and Barabady J. An expert-based approach to Syst Safe 2016; 152: 66–82.
production performance analysis of oil and gas facilities 32. Stamatelatos M, Dezfuli H, Apostolakis G, et al. Prob-
considering time-independent Arctic operating condi- abilistic risk assessment procedures guide for NASA man-
tions. Int J Syst Assur Eng Manag 2016; 7: 99–113. agers and practitioners, https://ntrs.nasa.gov/archive/
15. Barabady J and Aven T. A methodology for the imple- nasa/casi.ntrs.nasa.gov/20120001369.pdf
mentation of production assurance programmes in pro- 33. Denson W, Chandler G, Crowell W, et al. Nonelectronic
duction plants. Proc IMechE, Part O: J Risk and parts reliability data 1995. Rome: Reliability Analysis
Reliability 2008; 222: 283–290. Center, Griffiss Air Force Base, 1994.
 Qarahasanlou et al. 17

34. MIL-HDBK-217F. Military handbook: Reliability predic- 38. Zacks S. Introduction to reliability analysis: probability
tion of electronic equipment. USA: Department of models and statistical methods. New York: Springer Sci-
Defense, 1991. ence & Business Media, 2012.
35. Barabadi A. Reliability analysis of offshore production 39. Distefano S and Puliafito A. Dynamic reliability block
facilities under Arctic conditions using reliability data diagrams vs dynamic fault trees. In: Proceedings of annual
from other areas. J Offshore Mech Arct 2014; 136: 021601. reliability and maintainability symposium RAMS ’07,
36. Barabadi A, Barabady J and Markeset T. Application of Orlando, FL, 22–25 January 2007, pp.71–76. New York:
accelerated failure model for the oil and gas industry in IEEE.
Arctic region. In: 2010 IEEE international conference on 40. Kim MC. Reliability block diagram with general gates
industrial engineering and engineering management and its application to system reliability analysis. Ann Nucl
(IEEM), Macao, China, 7–10 December 2010, pp.2244– Energy 2011; 38: 2456–2461.
2248. New York: IEEE. 41. ISO 14224:2006. Petroleum, petrochemical and natural
37. Crowe D and Feinberg A. Design for reliability. New gas industries – collection and exchange of reliability and
York: CRC Press, 2001. maintenance data for equipment.

View publication stats