1 s2.0 S235197892030696X Main

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Available online at www.sciencedirect.

com
Available online at www.sciencedirect.com
ScienceDirect
ScienceDirect
Available online at www.sciencedirect.com
Procedia Manufacturing 00 (2019) 000–000
www.elsevier.com/locate/procedia
ScienceDirect
Procedia Manufacturing 00 (2019) 000–000
www.elsevier.com/locate/procedia
Procedia Manufacturing 43 (2020) 696–703

17th Global Conference on Sustainable Manufacturing


17th Global Conference on Sustainable Manufacturing
Empirical and Distribution Approaches for Analysing Reliability
Empirical
and and Distribution
Maintainability of RadialApproaches forinAnalysing
Compressors Reliability
Oil and Gas Systems
and Maintainability of Radial Compressors in Oil and Gas Systems
Paul Amaechi Ozor
Paul Amaechi Ozor
Department of Quality and Operations Management, University of Johannesburg, Doornfontein Campus, Doornfontein 2094, South Africa
Department
Department of Quality of
andMechanical
OperationsEngineering,
Management, University
Universityof of
Nigeria, Nsukka Campus,
Johannesburg, Nsukka
Doornfontein 140001,Doornfontein
Campus, Enugu State,2094,
Nigeria
South Africa
Department of Mechanical Engineering, University of Nigeria, Nsukka Campus, Nsukka 140001, Enugu State, Nigeria

Abstract
Abstract
The pertinent aspect of an investigation into the reliability and maintainability of a radial Compressor equipment in the oil and gas
industry
The is presented.
pertinent aspect ofUse of empirical into
an investigation models and candidate
the reliability distribution to quantify
and maintainability maintenance
of a radial Compressorparameters
equipmentfromin thesystem
oil anddata
gas
was explored.
industry The approach
is presented. Use of was illustrated
empirical models with
anddata taken from
candidate a typicaltoOil
distribution and Gas
quantify company that
maintenance will be from
parameters referred to asdata
system IE
Company
was for privacy
explored. reasons.was
The approach Theillustrated
results show thatdata
with the taken
method canabetypical
from used toOil
simultaneously determine
and Gas company instantaneous
that will be referred reliability,
to as IE
maintainability,
Company level reasons.
for privacy of improvement etc.,show
The results on the system
that after any
the method can individual
be used to maintenance.
simultaneously The results indicate
determine that reliability
instantaneous of
reliability,
the system degraded
maintainability, level to
of approximately 0.703391
improvement etc., on theatsystem
the 12after
th
failure and repair maintenance.
any individual point. The probability thatindicate
The results 0.8998that
of the failing of
reliability or
already
the failed
system components
degraded can be repaired
to approximately in about
0.703391 at 108.75
the 12 thhours wasand
failure established. TheThe
repair point. degree of improvement
probability necessary
that 0.8998 of the to restore
failing or
the entire
already system
failed to the bestcan
components performance
be repairedstate at the108.75
in about periodhours
is 81.5625.
was established. The degree of improvement necessary to restore
the entire system to the best performance state at the period is 81.5625.

© 2020
2019 The Authors.
Authors, Published by Elsevier
Elsevier B.V.
B.V.
This
© is an
2019 open
The accessPublished
Authors, article under the CC BY-NC-ND
by Elsevier B.V. license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Peer reviewunder
Peer review underthetheresponsibility
responsibility
of of
thethe scientific
scientific committee
committee of Global
of the the Global Conference
Conference on Sustainable
on Sustainable Manufacturing
Manufacturing.
Peer review under the responsibility of the scientific committee of the Global Conference on Sustainable Manufacturing
Keywords: Reliability; maintainability; radial compressor; Empirical; distribution; oil and gas systems
Keywords: Reliability; maintainability; radial compressor; Empirical; distribution; oil and gas systems

1. Introduction
1. Introduction
There has been unprecedented call for increased maintenance planning and control strategies in the oil and gas
industry,
There hasconsidering high systemcall
been unprecedented availability and productivity
for increased maintenance demands placed
planning and on the sector.
control Analyzing
strategies in thethe
oilreliability
and gas
and maintainability
industry, consideringofhigh
the system
serviceavailability
systems usingand empirical
productivitymethods
demands andplaced
engineering scienceAnalyzing
on the sector. in the lighttheofreliability
relevant
parameters
and (time to failure
maintainability or time systems
of the service to repair)using
can provide
empirical a means
methods of identity and assurance
and engineering as in
science regards when
the light of and how
relevant
long maintenance
parameters (time toshould
failurebeordone. This
time to procedure
repair) requires
can provide that historical
a means performance
of identity dataasofregards
and assurance the particular
when and system
how
must maintenance
long have been established a-pri-ori,
should be done. Thisinprocedure
addition requires
to the wealth of experience
that historical of the analyst.
performance data of theTheparticular
essence system
of this
presentation
must is to established
have been explore a practical methodology
a-pri-ori, in additionforto deducing
the wealthreliability and maintainability
of experience indicators
of the analyst. appliedoftothis
The essence oil
and gas systems.
presentation The effect
is to explore of the said
a practical indicatorsfor
methodology candeducing
stand as reliability
a suitableand
support and maintenance
maintainability decision
indicators appliedtool
to for
oil
raising
and gassystem
systems.availability
The effectand productivity.
of the Reactive
said indicators and proactive
can stand maintenance
as a suitable aremaintenance
support and both defineddecision
for degradable
tool for
raising system availability and productivity. Reactive and proactive maintenance are both defined for degradable
2351-9789 © 2019 The Authors, Published by Elsevier B.V.
Peer review©under
2351-9789 2019the
Theresponsibility of the scientific
Authors, Published committee
by Elsevier B.V. of the Global Conference on Sustainable Manufacturing
Peer review under the responsibility of the scientific committee of the Global Conference on Sustainable Manufacturing

2351-9789 © 2020 The Authors. Published by Elsevier B.V.


This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Peer review under the responsibility of the scientific committee of the Global Conference on Sustainable Manufacturing.
10.1016/j.promfg.2020.02.119
Paul Amaechi Ozor / Procedia Manufacturing 43 (2020) 696–703 697
Author name / Procedia Manufacturing 00 (2019) 000–000

systems depending on whether it is aimed at restoring the failed system or forestalling predicted failure [1]. Elevated
reliability is very essential for achieving the objectives of any industrial system. Reliability is normally assessed and
allocated during the design stage of most systems [2]. Real world or on-site factors which can broadly be grouped into
age, usage and environment can alter reliability predictions and decisions reached at the design stage. This necessitates
the modeling of the practical reliability behavior of the system vis-à-vis the operating conditions. Many research
output have centered on reliability stability and improvement of different component, unit or system [3]-[5]. The
reliability of many systems have been assessed in literatures [6]-[8]. The consensus can be described by a simple
statement; reliability is diminishable. There must therefore be a minimum reliability threshold at which a system must
be maintained (repaired) for improved reliability. The repair process can be decomposed into further specific subtasks
and delay times as well as various measures of maintainability. Notable among others include the mean time to repair
(MTTR), median time to repair; the mode, most likely repair time as well as the number of maintenance hours per
operating hour. Maintainability is defined as “the probability that a failed component or system will be restored or
repaired to a specified condition within a period of time when maintenance is performed in accordance with
prescribed procedures” [1]. A school of thought [9] posits that quantification of the indicators results in addressing
the parameter as maintainability performance. The paper later described a procedure for obtaining the maintainability
assets in the industry during the life cycle. Past researchers have expressed discontent with the technique of assessing
maintainability indicators using traditional methods [10]. In the author’s opinion, better models, for example; the
proportional hazards model is more precise. Accordingly, a variant of the proportional hazards model called
proportional repair model was proposed and how it can be extended for evaluating maintainability amidst time-
dependent covariates explained. Attention was not paid to when repair process should be initiated, though the model
was validated on a real world case. Recalling the impossibility of knowing with a good degree of accuracy, where,
when and how a class of system should be maintained, Al Dallal [11] introduce a class maintainability model that
considers the internal qualities. This was however applied to object oriented software systems which are at best varied
in size, physic-mechanical structure and presentation to core hardware industrial systems. The task of modeling
reliability and maintainability can provide for just-in-time detection and renewal of a repairable engineering system.
Huge progress was made in the field when researchers [12] describe how reliability and maintainability could be
allocated to warship on the basis of mission reliability and inherent availability which were referred to as top-level-
parameters of the system. The work exposed the possibility and powerful effect of working out a synergistic allocation
of the parameters and indices of the two quantities (reliability and maintainability) to repairable systems, short of
handling each independently as earlier practiced.

Interestingly, the method was adjudged capable of transcending its illustrative application to warship, in the sense that
its exploration in complex repairable industrial systems was proposed. Though reliability and maintainability
allocation operations can be defined for the system during the design stage, nevertheless; results established may not
apply very effectively when the system has been installed, operated and degraded. In particular, the insight obtainable
from the allocation problem can be extrapolated to prescribe effective on-site reliability and maintainability control
policies for already operational systems, which can account for the operating peculiarities and environmental effects
on the system. The options to either continue maintaining industrial systems or allow them to age to scrap had been
investigated [13]. The author noted that low quality designed system or systems whose state may not be readily
ascertained owing to low quality design can increase the spate of costly system issues. The presentation experimentally
validates already modified system metrics like coupling, instability and abstractness found in literature [14], through
establishing the relationship with reliability and testability. Prediction models for the two external quality attributes
were also constructed. To date, there is still paucity of research on the optimal modeling strategy for reliability and
maintainability (RM) quantities, especially for the special case of Oil and gas industry. This paper is intended to
present a practical RM decision support tool to maintenance practitioners in the field. The modelling technique
employed is clearly described and validated on data taken from an example Oil and Gas system. The author is not
privy to any publication directly of the sort in literature.
698 Paul Amaechi Ozor / Procedia Manufacturing 43 (2020) 696–703
Author name / Procedia Manufacturing 00 (2019) 000–000

2. Materials and Methods


The RM analysis method has been explored in this work as an assurance for radial compressors in Oil and Gas
Company, power plants and industries. The RM analysis can be an important deliverable not only at the product design
phase, but also during the entire life cycle for process industries like oil and gas and power plants. The materials used
in the study for the performance of RM analysis were based on qualitative and quantitative data. The study made use
of literature models (statistical, failure rate, reliability and maintainability), though from various sources as well as
specific software candidate distribution identification tool. Dataset on a typical radial compressor from the IE
Company was sourced from maintenance records. There were various interview sessions with the maintenance
personnel who have been working on the radial compressor for over fifteen years. The opinion of the personnel on the
values of some of the parameters that were not readily deductible from modelling and data analysis was sought and
fortunately obtained. There were two failure categories generally related to the equipment. For the case where repair
would be necessary within a limited period of time and without removing the equipment from the process line, it is
regarded as minor failure. If multiple components or a major part fails in the compressor and warrants bringing down
the entire system, amidst plenty of repair time, the failure is referred to as critical. Data on all the failure forms ensured
better accurate analysis of the failure-repair cycle. The full description of the repair time in case of reactive (corrective)
maintenance, the time for isolation, on-site mobilization, depressurization, actual repair time as well as reinstallation
time were all integrated into the repair time. Other downtime of interest was supply resources time, i.e. time for
procuring spare parts and administrative or logistics time. The generalized reliability model for multi-component
systems [15] was relied upon for analysing the relevant parameters. The model was basically chosen due to its capacity
to allow fractional operating time of the studied system to be quantified. The exact life time distribution that would
most accurately model the collected data was identified through descriptive statistics, Probability and least squares
plots as well as confirmatory tests. The parameters of the distribution were estimated. Specific fitness test was carried
out for the selected distribution for confirmatory purposes. The quantification of Maintainability indices was
conducted with the already established reliability threshold in retrospect.

3. Analysis

3.1 Reliability

As presented in previous works for the analysis of reliability of a multi-component system, immediately after an
individual maintenance, at any point in time, 𝑡𝑡 [15]:
𝛽𝛽𝑗𝑗
1
𝑅𝑅𝑗𝑗 (𝑡𝑡𝑖𝑖,2 ) = (1 − 𝜀𝜀) {𝑎𝑎1 exp(−𝑐𝑐𝑗𝑗 𝑘𝑘 ′ 𝑖𝑖,2 ) + 𝑎𝑎2 𝑒𝑒𝑒𝑒𝑒𝑒 [−
𝜃𝜃𝑗𝑗
(𝜔𝜔𝑗𝑗 𝑘𝑘 ′ 𝑖𝑖,2 )] } (1)
Where:
𝛽𝛽𝑗𝑗 = shape parameter of the jth unit, 𝜔𝜔𝑗𝑗 = fraction of time the 𝑗𝑗𝑗𝑗ℎ unit works, 𝜃𝜃𝑗𝑗 = scale parameter of the 𝑗𝑗𝑗𝑗ℎ
unit, 𝑎𝑎1 , 𝑎𝑎2 = coefficients of chance and wear out failures, 𝜀𝜀 = probability of failure before the next scheduled
maintenance, 𝑐𝑐𝑗𝑗 = constant factor for the jth unit that account for nonlinearity effect, 𝜂𝜂𝑗𝑗 = improvement factor in the
jth unit after an individual scheduled maintenance
𝑘𝑘 ′ 𝑖𝑖,2 = 𝑡𝑡𝑖𝑖 × 𝜂𝜂𝑗𝑗 (2)

3.2. Analysis of Failure and Repair Data

It is normally required that the failure/repair time (FRT) data be rank-ordered and values of the cumulative failure
function, 𝐹𝐹(𝑡𝑡), at time, 𝑡𝑡 calculated empirically, using the following relation [15]:
(𝑖𝑖−0.3)
𝐹𝐹(𝑡𝑡) = (𝑛𝑛+0.4) (3)
Where:
𝑖𝑖 = 1, 2, … 𝑛𝑛, 𝑛𝑛 = total number of observed failures
Paul Amaechi Ozor / Procedia Manufacturing 43 (2020) 696–703 699
Author name / Procedia Manufacturing 00 (2019) 000–000

3.3. Histogram

Fig. 1 presents the histogram for the time to failure data. The horizontal axis (𝑥𝑥), and vertical axis, 𝑓𝑓(𝑥𝑥) are the time
to failure ranges and corresponding frequencies respectively. No particular life time distribution can be inferred from
the histogram, and this necessitated further statistical description of the data. The computational results are displayed
in Table 1. The disparity in the sample mean and median times to failure is obvious. This indicates that the data may
not have come from a symmetrical or nearly symmetrical distribution, for instance; normal or a Weibull whose shape
parameter lies between 3 and 4. The mean being significantly larger than the median can imply that the data is skewed
to the right. Therefore the exponential distribution with hazard rate between 1 and 5, Weibull with shape parameter of
approximately 0.5 or lognormal distribution with shape parameter of 1 can all be tested. On the basis of te above
premise, the component cannot be immediately repaired upon failure. Use of Poisson distribution might not suffice as
well.

Fig. 1: Histogram of time to failure (TTF or x) for the radial compressor in hours

Table 1: Descriptive Statistics of TTF(hours)


Statistic Value Percentile Value
Sample Size 32 Min 0.1
Range 984.48 5% 0.1195
Mean 217.92 10% 0.471
Variance 58068.0 25% (Q1) 73.625
Std. Devtn 240.97 50% (Median) 153.54
Coef. of Var 1.1058 75% (Q3) 264.48
iation
Std. Error 42.598 90% 541.57
Skewness 1.957 95% 946.94
Excess Kurt 4.0738 Max 984.58
osis

3.4. Probability Density Function (pdf)

The pdf was determined for the distributions suspected on the basis of the descriptive statistics. Though specific
models for computing the pdf of the distributions are found elsewhere [1], a special software tool (EASYFIT) was
employed to obtain Fig. 2 which shows the pdf for the suspected distributions. It can be observed that the Exponential
distribution provides the best fit to the data while the Normal counterpart gave the worst fit. The data was subjected
to further analysis to confirm the visual inference. A probability plot can provide approximate values of the distribution
parameters and was explored as a means of ensuring that the best candidate distribution is selected. Both Weibull and
Lognormal distributions were ruled out because no better than 0.88 index of fit value was obtained for each. The
exponential distribution fitted the data throughout the whole range. Further analysis based on least square fitting was
700 Paul Amaechi Ozor / Procedia Manufacturing 43 (2020) 696–703
Author name / Procedia Manufacturing 00 (2019) 000–000

implemented to deduce the life time distribution of the system’s data with higher precision. The underlying statistical
index of fit was the coefficient of regression (𝑅𝑅2 ) which ranges from zero (0) to one (1). The closer the index is to
one, the better the fitness of the data to the particular distribution. From the least squares plots, the initial parameters
of the selected distribution was estimated. In the case of the radial compressor, least square plots were constructed for
each of the suspected distributions, after rigorous analysis of the time to failure (TTF). Some important and appropriate
parametric transformations was required to enable good least square plotting for some distributions. The values of 𝑅𝑅 2
for some of the tested distributions are given as: Lognormal distribution 𝑅𝑅2= 0.7544; Weibull distribution; 𝑅𝑅2 =
0.8834, Normal distribution, 𝑅𝑅 2 = 0.6759 and Exponential distribution, 𝑅𝑅 2 = 0.953. The least square plot for the
Exponential distribution is displayed in Fig 4 as the best candidate and subjected to advance confirmatory testing.
Relevant models guiding the plotting positions for the models are found in literature [1].

Probability Density Function

0.56
0.52
0.48
0.44
0.4
0.36
0.32
f(x)

0.28
0.24
0.2
0.16
0.12
0.08
0.04
0
0 200 400 600 800
x

His togram Norm al Exponential Lognorm al Weibull

Fig. 2: Probability density function for tested distributions

P-P Plot
1

0.9

0.8

0.7

0.6
P (Model)

0.5

0.4

0.3

0.2

0.1

0
0.2 0.4 0.6 0.8 1
P (Empirical)

Norm al Exponential Lognorm al Weibull

Fig. 3: Probability plot of time to failure (TTF) in hours

4 y = 0.0035x + 0.192
y=ln[1/1-H(t)

3
R² = 0.953
2
Series1
1
0 Linear (Series1)
0 500 1000 1500
TTF

Fig. 4: Exponential least-square plot in hours


Author name / Procedia Manufacturing 00 (2019) 000–000
Paul Amaechi Ozor / Procedia Manufacturing 43 (2020) 696–703 701

3.5 Parameter Estimation

Exponential distribution is the choice. The initial and maximum likelihood values of its parameters can be estimated.
As reported in literature for the calculation of the cumulative distribution function (4) and least squares parameters
(5) for the exponential distribution respectively [1]:
̂
𝐹𝐹(𝑡𝑡) = 1 − 𝑒𝑒 −𝜆𝜆𝑡𝑡 (4)
∑𝑛𝑛
𝑖𝑖=1 𝑥𝑥𝑖𝑖𝑦𝑦𝑖𝑖
𝜆𝜆̂ = 𝑏𝑏 = ∑𝑛𝑛 2 (5)
𝑖𝑖=1 𝑥𝑥𝑖𝑖
The initial estimate of the parameter 𝜆𝜆̂ can be obtained directly from the least square plot for the exponential
distribution model by comparison of (5) with data of Fig. 4. Thus; 𝜆𝜆̂ = 0.0035. The maximum likelihood estimator
(MLE) for the parameter 𝜆𝜆̂ is given by:
𝑟𝑟
𝜆𝜆̂ = 𝑇𝑇 (6)
Therefore the MLE for the selected distribution is 0.004589. Bartlett’s (𝐵𝐵) goodness-of-fit test specified for the
exponential distribution can guarantee that the data is tailored down to the distribution or rejected completely, short
of the general Chi-Square test. Equation (7) can be used to compute the test statistic, 𝐵𝐵, which has the following null
hypotheses:
𝐻𝐻0 : Repair times are exponential with, 𝜆𝜆 = 0.004589, 𝐻𝐻1 : Repair times are not exponential with 𝜆𝜆 = 0.004589
1 1
2𝑟𝑟[ln(( ) ∑𝑟𝑟𝑖𝑖=1 𝑡𝑡𝑖𝑖)−( ) ∑𝑟𝑟𝑖𝑖=1 𝑙𝑙𝑙𝑙𝑡𝑡𝑖𝑖]
𝐵𝐵 = 𝑟𝑟
(𝑟𝑟+1)
𝑟𝑟
(7)
1+ (6𝑟𝑟)

The test statistic, 𝐵𝐵, has a Chi-square distribution with 𝑟𝑟 − 1 degrees of freedom. A level of significance of 0.1 was
chosen for the data and the value of the test statistic obtained.
𝛼𝛼 𝛼𝛼
For 𝛼𝛼 = 0.1, 2 = 0.05, 1 − 2 = 0.95, with degree of freedom, 𝑟𝑟 − 1 = 31,
𝑋𝑋 21−𝛼𝛼⁄2,𝑟𝑟−1 = 0.95 < 𝐵𝐵 = 36.5218 < 𝑋𝑋 2 𝛼𝛼⁄2,𝑟𝑟−1 = 44.98534
𝐻𝐻0 is accepted. The exponential distribution is best suited for modelling the maintainability of the radial compressor.

4. Maintainability
One of the most important decisions for maintenance practitioners regarding a proficiently designed and successfully
installed repairable system is the maintenance schedule frequency or mean time to repair (𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀). Other parameters
are the median time to repair as well as the standard deviation. The three quantities can be obtained by using (8). The
failure rate or hazard rate can be deduced by (9) whereas the repair rate is given by (10).
1
𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 = 𝑡𝑡𝑚𝑚𝑚𝑚𝑚𝑚 = 𝜎𝜎 = 𝜆𝜆 (8)
𝜆𝜆(𝑡𝑡) = 𝜆𝜆 (9)
𝑡𝑡
𝐻𝐻(𝑡𝑡) = 1 − 𝑒𝑒 −𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 (10)

The mean time to repair can be calculated from the values of the repair times. Since the data came from exponential
distribution; maintainability can directly be quantified as a function of time in line with the selected repair distribution.

𝑀𝑀(𝑡𝑡) = 1 − 𝑒𝑒 −𝜇𝜇𝜇𝜇 (11)


Where:
1
𝑀𝑀(𝑡𝑡) = maintainability at time, t, 𝜇𝜇 = repair rate, 𝜇𝜇 = 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀
, 𝑡𝑡 = time in hours.
702 Paul Amaechi Ozor / Procedia Manufacturing 43 (2020) 696–703
Author name / Procedia Manufacturing 00 (2019) 000–000

5. Result and Discussion

The modelling procedure was applied to ascertain the reliability and maintainability of the example repairable system,
that is; radial compressor whose TTF data was obtained from IE Company. The mean time to repair obtained from
1
the methodology (𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 = ) was in general agreement with the values originally estimated through standard
𝜆𝜆
statistical procedures (Table 1). The implication of this result is that the maintainability modelling approach and the
life time distribution eventually selected were quite appreciable. It further shows that the method modelled the
operation (failure and repair) of the case study system closely, especially within the extent of the data used. The
approach was further shown to be able to give instantaneous reliability and maintainability status of the component as
displayed in Table 3. Values of parameters of (1) are presented in Table 2. Equation (11) was used to determine the
probability of completing 90% of the repair actions within the necessary time, in each repair period. The abridged
version of various maintainability values are displayed in Table 3. A lot of other maintenance planning and control
decisions can be made on the basis of the results in Table 3. One of such courses of action is determination of the
acceptable level of operational reliability of the studied system and estimation of the component maintainability at the
time. Consider the 12th failure period for instance, the maintenance engineer can ascertain that the reliability of the
component have degraded to somewhere around 0.703391. The practitioner can as well estimate the reasonable
probability that 0.8998 of the failing or failed components can be repaired in about 108.75 hours. The degree of
improvement necessary to restore the entire system to the best performance state at this period is 81.5625. Having this
dataset, it will be easy to guide the maintenance practitioners as well as the administrative resources on the best period
to maintain the component for maximum benefits.

Table 2: Values of parameters of model 1


𝜺𝜺 𝜼𝜼 𝝎𝝎 𝜷𝜷 𝜽𝜽 𝒌𝒌′ 𝒊𝒊,𝟐𝟐 =𝒕𝒕𝒊𝒊 ×
𝜼𝜼𝒋𝒋
0.2 0.75 0.843285 1.29376 884.5206 Table 3

Table 3: RM values for the case study system


S/N TTF 𝑹𝑹(𝒕𝒕𝟐𝟐 ) TTR 𝑴𝑴(𝒕𝒕𝟐𝟐 ) 𝒌𝒌′ 𝒊𝒊,𝟐𝟐
(hr) (hr)
1 0 0.94448 0.1 0.899709 0.075
2 66.5 0.939202 0.13 0.899652 0.0975
3 70.25 0.93825 0.33 0.899642 0.2475
4 70.92 0.900489 0.8 0.899932 0.6
5 96.6 0.847993 4 0.899719 3
6 131.67 0.844433 14.08 0.899702 10.56
7 134.08 0.829528 25.25 0.899625 18.9375
8 144.25 0.808692 73.25 0.899425 54.9375
9 158.73 0.745128 74.75 0.899903 56.0625
10 205.5 0.706095 78 0.899812 58.5
11 236.75 0.703391 81.75 0.899805 61.3125
12 239 0.703391 108.75 0.899805 81.5625
13 239 0.701597 112.33 0.8998 84.2475
14 240.5 0.684374 114.5 0.899749 85.875
15 255.17 0.57177 120.58 0.899952 90.435
⋮ ⋮ ⋮ ⋮ ⋮ ⋮
30 2606.0 0.037446 577.42 0.899786 433.065
31 3914.3 0.026069 926.68 0.899593 695.01
Paul Amaechi Ozor / Procedia Manufacturing 43 (2020) 696–703 703
Author name / Procedia Manufacturing 00 (2019) 000–000

Conclusion
The paper explores the empirical analysis of reliability and maintainability of degradable but repairable radial
compressor in an oil and gas industry. The methodological procedure used to achieve the thrust of the study is simple,
well explained and graphically displayed. An application to a typical example of the case study repairable system
further strengthened the validity of the procedure as fruitful results were obtained. Though the method was originally
applied to a specific case, it promises to be useful in other spheres where service producing repairable systems are put
in use.

Acknowledgments
The material and financial assistance of the NRF-TWAS fellowship: award number: PD-TWAS160531166951; UID:
105554, towards this research is hereby acknowledged. However, opinions expressed and conclusions arrived at, are
those of the author and are not necessarily to be attributed to the NRF-TWAS. The Company staffers and participants
that contributed to make this work a success are hereby acknowledged.

References

[1] C. E. Ebeling, “An introduction to reliability and Maintainability Engineering”, Waveland Press, INC, Long
Grove, Illinois, Second Edition, Chapter 11, pp. 283- 297, 2010
[2] D. Lee, and R. Pan, “A Nonparametric Bayesian Network Approach to Assessing System Reliability at Early
Design Stages”, Reliability Engineering and System Safety, doi: 10.1016/j.ress.2017.11.009, 2017
[3] J-H. Yoon, I-H. Yang, J-E. Jeong, S-G Park and J-E Oh, “ Reliability improvement of a sound quality index
for a vehicle HVAC system using a regression and neural network model”, Applied Acoustics, Vol. 73, pp.
1099–1103, 2012
[4] S. Dalkilic, “Improving aircraft safety and reliability by aircraft maintenance technician training”,
Engineering Failure Analysis, Vol. 82, pp. 687–694, 2017
[5] F. Corvaro, G. Giacchetta, B. Marchetti and M. Recanati, “Reliability, Availability, Maintainability (RAM)
study, on reciprocating compressors API 618”, Petroleum, Vol. 3, pp. 266-272, 2017
[6] M. Catelani, L. Ciani, and M. Venzi, “Component Reliability Importance assessment on complex systems
using Credible Improvement Potential”, Microelectronics Reliability, Vol. 64, pp. 113–119, 2016
[7] J. Wu, S. Yana, J. Li, and Y. Gu, “Mechanism reliability of bistable compliant mechanisms considering
degradation and uncertainties: Modeling and evaluation method”, Applied Mathematical Modelling, Vol. 40,
pp. 10377–10388, 2016
[8] J. Mi, Y-F. Li, Y-J. Yang, W. Peng, and H-Z. Huang, “Reliability assessment of complex electromechanical
systems under epistemic uncertainty”, Reliability Engineering and System Safety, Vol. 152, pp. 1–15, 2016
[9] P. M. De Leon, D. G. Lez-Prida, L. B. Martinez, and A. C. Marquez, “A practical method for the
maintainability assessment in industrial devices using indicators and specific attributes”, Reliability
Engineering and System Safety, Vol. 100, pp. 84–92, 2012
[10] A. Barabadi, J. Barabady, and T. Markeset, “Maintainability analysis considering time-dependent and time-
independent covariates, Reliability Engineering and System Safety, Vol. 96, pp. 210–217, 2011
[11] J. Al Dallal, “Object-oriented class maintainability prediction using internal quality attributes”, Information
and Software Technology, Vol. 55, pp. 2028–2048, 2013
[12] X. F. Liang, L. Y. Chen, H. Yi and D. Li, “Integrated allocation of warship reliability and maintainability
based on top-level parameters”, Ocean Engineering, Vol. 110, pp. 195–204, 2015
[13] S. Almugrin, W. Albattah and A. Melton, “Using indirect coupling metrics to predict package maintainability
and testability”, The Journal of Systems and Software, Vol. 121 pp. 298–310, 2016
[14] R. C. Martin, “Agile Software Development Principles, Patterns & Practice”, Prentice Hall, Upper Saddle
River, New Jersey, 2003
[15] P. A. Ozor, and S. O. Onyegegbu, “Design of Preventive Maintenance Scheduling Model for Deteriorating
Systems”, Nigerian Journal of Technology, Vol. 30, Issue 3, pp. 90-96, 2011

You might also like