You are on page 1of 9

Journal of Natural Gas Science and Engineering 26 (2015) 1050e1058

Contents lists available at ScienceDirect

Journal of Natural Gas Science and Engineering


journal homepage: www.elsevier.com/locate/jngse

A framework for making maintenance decisions for oil and gas drilling
and production equipment
Yang Tang a, *, Zhengwei Zou b, Jiajia Jing c, Zhidong Zhang c, Chong Xie a
a
School of Mechatronic Engineering, Southwest Petroleum University, Chengdu 610500, China
b
School of Computer Science, Southwest Petroleum University, Chengdu 610500, China
c
Safety, Environment, Quality Supervision & Testing Research Institute, CCDE, Guanghan 618000, China

a r t i c l e i n f o a b s t r a c t

Article history: There are few scientific maintenance decision-making methods in current maintenance and manage-
Received 30 January 2015 ment of drilling and production equipment (DPE). Conventional methods have some conspicuous de-
Received in revised form ficiencies and shortcomings, for example, unreasonable maintenance methods, surplus or insufficient
23 July 2015
maintenance, exorbitant maintenance costs and increasing failure frequency, which have caught a great
Accepted 24 July 2015
Available online 31 July 2015
influence to production safety and economic cost in the oil and gas exploitation process. In this study, a
framework for making maintenance decisions was presented in order to improve the maintenance and
management of the DPE. First, eight evaluation indexes and their scoring criteria were defined to
Keywords:
Importance level evaluation
quantify subjective evaluation of importance level of the DPE. Then, a linear weighted mathematical
Analytic hierarchy process model was presented to calculate importance level value and a weight computing method of each
Monte Carlo simulation evaluation index was put forward based on the Analytic Hierarchy Process (AHP). And the subjective
Maintenance decision-making effects were eliminated with Monte Carlo Simulation (MCS) in the scoring process. Next, maintenance
Drilling and production equipment decision-making trees (MDMTs) for the DPE were set up by reference to the logic decision tree of
reliability-centred maintenance (RCM). Finally, feasibility of the framework was verified by testing a well
control system in Tarim Oilfield. Therefore, the framework for making maintenance decisions can provide
reasonable maintenance methods and achieve scientific maintenance and management for the DPE.
© 2015 Elsevier B.V. All rights reserved.

1. Introduction out timely maintenance or replacement for the DPE. Moreover,


there were no effective decision-making methods or scientific
The management system of point inspection and regular repair theoretical models to guide the maintenance process for the DPE.
(PIRR) is regarded as the present core of maintenance and man- Negative outcomes, including surplus repair, insufficient repair,
agement for onshore and offshore oil field equipment, which unreasonable repair intervals, and higher maintenance costs, were
mainly adopt corrective maintenance (CM), time-based mainte- brought about during maintenance and management (Eti et al.,
nance (TBM), detection-based maintenance (DBM), etc. 2006; Kyriakidis and Dimitrakos, 2006). Therefore, it is necessary
(Doostparast et al., 2014; Perrons and Richards, 2013). According to to establish a new framework for making maintenance decisions
field investigation, we found that the overwhelming majority of for the DPE using a mathematical model and decision-making
major accidents and economic losses in the drilling and production theory in order to improve equipment reliability, simplify mainte-
process of oil and gas were caused by equipment failure and human nance decision processes and ensure production safety.
factors. Most equipment failures were attributed to the current There have been some widely used maintenance method and
maintenance and management method that was backward and theory in aerospace, military, rail transportation, electricity, ship-
unscientific on the basis of the experience of the maintenance and building and other industries (Dekker, 1996; Murthy et al., 2002).
repair personnel. The conventional maintenance strategies could Over the years, a lot of research works on maintenance strategies
not be applied by the maintenance and repair personnel to carry and decision-making models have been done as well. Bevilacqua
and Braglia (Bevilacqua et al., 2000) proposed a method to evaluate
importance level of the power plant equipment based on MCS and
* Corresponding author. modified FEMCA. Bertolini and Bevilacqua (Bertolini and
E-mail address: tangyanggreat@126.com (Y. Tang). Bevilacqua, 2006) presented a new technique to determine the

http://dx.doi.org/10.1016/j.jngse.2015.07.038
1875-5100/© 2015 Elsevier B.V. All rights reserved.
Y. Tang et al. / Journal of Natural Gas Science and Engineering 26 (2015) 1050e1058 1051

best strategies for the maintenance of critical centrifugal pumps in should be different. So there is a relative importance level between
an oil refinery. Based on AHP, a maintenance strategy of a multi- them. The definition and the fundamental scale of the relative
criteria classification of equipment was proposed by Go  mez de importance level were made in Table 1. The relative importance levels
Leo n Hijes and Cartagena (de Leo n Hijes and Cartagena, 2006), and of eight evaluation indexes can be scored by reference to the Table 1.
oil pipeline projects were effectively evaluated by Dey (Dey, 2004) Then, we made scoring criteria of eight evaluation indexes based
with a multiple-attribute decision-making technique. Chang, on reviews from maintenance engineers and operating personnel
Chang and Zio (Chang et al., 2010) applied the MCS to estimate the in the field investigation. The influence levels of the evaluation
production availability in offshore installations. The recent research indexes were classified into three to five grades depending on their
on maintenance strategies includes work by Aida and Fathi (Azizi respective characteristics and scored with a 10-point system
and Fathi, 2014), who presented an empirical investigation to (Yuliang et al., 2003; Triantaphyllou et al., 1997). The definitions of
rank different factors of optimum maintenance strategies based on the influence levels and the scoring criteria for the evaluation in-
a fuzzy analytic hierarchy process, and by Arunraj and Maiti dexes of the DPE are as follows:
(Arunraj and Maiti, 2010), who exhibited a technique of mainte-
nance policy selection created based on the risks of equipment 2.1. Influence of failure on personnel and environment safety (S)
failure and costs of maintenance. Both AHP and MCS that are
feasible and reliable mathematical models are often used for Equipment failure may bring about some disasters (such as
making decisions in science management. However, very few ap- poison gas leakage and diffusion, fires, and explosions) in the field
plications of the AHP and the MCS exist for making maintenance during oil and gas exploration and development. These disasters
decisions for DPE through research and investigation. We have also will affect personnel safety and environmental pollution. However,
surveyed a few maintenance decision-making methods and models their possible influence levels should be considered in the case of
associated with DPE, especially ones that are based on different the DPE failure, and their scoring criteria are shown in Table 2.
importance levels.
However, the DPE is different from the equipment in other in- 2.2. Influence of failure on system functions (SF)
dustries, such as failures types and distributions, maintenance
methods and costs, and reliability and safety requirements, due to When equipment failures occur, the influence on functions of
their harsh construction environments, complicated working con- the overall system should be a major concern during the drilling
ditions and extremely high safety requirements in the drilling and and production process. The definition of the influence levels and
production process of oil and gas (Hmida et al., 2013; Du et al., the scoring criteria of SF are given in Table 3.
2013). Thus, these existing maintenance decision-making
methods and maintenance strategies that are applied to the 2.3. Average failure rate (FR) of equipment
equipment in other industries are not directly suitable for the DPE.
It is necessary to study a maintenance decision-making method for Mean time between failures (MTBF) of the DPE can be calculated
the DPE by focussing on the features of the drilling and production from their operational records corrected by the operators and re-
process. Therefore, we put forward a new framework for making pairers and the reliability databases for the relevant equipment. In
maintenance decisions for the DPE based on their different Table 4, the numeric range of the MTBF and the scoring criteria for
importance levels. Through the framework for making mainte- FR are shown.
nance decision in this paper, more reasonable and more effective
maintenance methods can be devised for the DPE to guarantee the 2.4. Maintenance costs (MC)
reliability and security of the drilling and production operation of
oil and gas. According to plenty of investigation for equipment maintenance
The remainder of this paper is organized as follows. In Section 2, and repair in the oil field, these factors, including structural
evaluation indexes and their scoring criteria for importance level of complexity, maintenance time, spare part costs, and maintenance
the DPE are defined. In Section 3, an evaluation model for impor- store location should be accounted for in the MC of the DPE. The
tance level of the DPE is established using AHP and MCS approach scoring criteria for MC are shown in Table 5.
which is applied to reduce the subjective influences in the scoring
process. Then, the DPE are divided into three categories according 2.5. Downtime loss (DL)
to their different importance levels and their three MDMTs are set
up based on RCM theory in Section 4. A case study for the well This evaluation index refers to the economic loss arising from
control system is carried out to verify the framework in Section 5. downtime. The downtime attributes economic loss to equipment
Finally, Section 6 provides some discussion and conclusions. maintenance or replacement in oil drilling operations. The scoring
criteria for DL are shown in Table 6.
2. Definition of the evaluation index and scoring criteria
2.6. Monitoring availability (MA) of equipment failure
According to extensive investigation and research, we found
that influence factors of importance level of the DPE could be Failure monitoring availability for DPE can be assessed based on
classified into four categories: reliability factors, economy factors, monitoring possibilities of equipment failure, monitoring
monitoring availability factors and maintainability factors. These complexity, monitoring cost, etc. The scoring criteria for MA are
four categories were further subdivided into eight influence factors, shown in Table 7.
as shown in Fig. 1 (Aven and Vinnem, 2005; Yuliang et al., 2003).
These eight influence factors were regarded as evaluation indexes 2.7. Downtime (DT)
of importance level of the DPE.
In order to determine the importance level of the DPE with In the maintenance and repair process of the DPE, the DT in-
mathematical methods, it is necessary to define scoring criteria to cludes the time (in man-hours) for idleness, maintenance and
quantify influence level of each evaluation index. For different restart. Combining with the actual situation on site, we presented
equipment or systems in the DPE, weights of the evaluation indexes the scoring criteria for DT in Table 8.
1052 Y. Tang et al. / Journal of Natural Gas Science and Engineering 26 (2015) 1050e1058

Fig. 1. Evaluation indexes evaluating the importance level of the DPE.

Table 1 Table 5
The definition and fundamental scale of the relative importance level. Scoring criteria for MC.

Definition Score S/N Maintenance costs (US$) Score

Equally important 1 1 <1600 0e2


Moderately important 3 2 1600e16,000 2e4
Strongly important 5 3 16 000e80,000 5e6
Very strongly important 7 4 80 000e160,000 7e9
Extremely important 9 5 >160,000 10
Intermediately important 2, 4, 6, 8

Table 6
Table 2 Scoring criteria for DL.
Scoring criteria for S.
<S/N Economic loss Score
S/N Influence level Score
1 Very small 0e2
1 No influence 0e2 2 Small 3e6
2 Slight influence 3e4 3 Large 7e8
3 Greater influence 5e8 4 Huge 9e10
4 Significant influence 9e10

Table 7
Table 3 Scoring criteria for MA.
Scoring criteria for SF.
S/N Monitoring availability Score
S/N Influence level Score
1 High 0e4
1 No impact 0e2 2 Low 5e7
2 Slight impact 3e4 3 No monitoring 8e10
3 Obvious decline 5e8
4 Complete loss 9e10

3. Evaluation model for the importance level of the DPE

2.8. Complexity of maintenance (M)


3.1. Quantitative evaluation based on the AHP method

This evaluation index is subject to difficulty degree of equip-


According to the scoring criteria for the DPE in the Section 2, we
ment assembly and disassembly, the complexity degree of the
can obtain score values of the eight evaluation indexes based on
equipment structure, the difficulty degree of implementation (such
subjective evaluation of the equipment. But we cannot directly
as height, surrounding environment, etc.), and supply level of spare
determine importance levels of the equipment from these score
parts. The scoring criteria for M are shown in Table 9.

Table 4 Table 8
Scoring criteria for FR. Scoring criteria for DT.

S/N MTBF(hrs) Score S/N Man-hours (hrs) Score

1 >7000 0e2 1 <2 0e2


2 3000e7000 3e4 2 2e4 3e4
3 1000e3000 5e6 3 4e8 5e6
4 300e1000 7e8 4 8e24 7e8
5 <300 9e10 5 >24 9e10
Y. Tang et al. / Journal of Natural Gas Science and Engineering 26 (2015) 1050e1058 1053

Table 9 Step 3: Determine the eigenvector relative to the maximum


Scoring criteria for M. eigenvalue lmax, which is given by
S/N Man-hours (hrs) Score

1 <12 0e2
2 12e24 3e4 W ¼ ðu1 ; u2 ; /; un Þ (4)
3 24e72 5e6
4 >72 7e10
Step 4: Check the consistency of the judgement matrix D:

values of the evaluation indexes. In order to evaluate the impor- CR ¼ CI=RI (5)
tance level of the DPE, a linear weighted mathematical model was
presented to calculate the total score of eight evaluation indexes,
CI ¼ ðlmax  nÞ=ðn  1Þ (6)
namely, Index, as follows (Yuliang et al., 2003):
where
X
n
Index ¼ mi ai (1)
i¼1
CR e the random consistency ratio of the judgement matrix;
CI e the general consistency index of the judgement matrix;
where RI e the mean random consistency index of the judgement
matrix.
n- the total evaluation indexes;
mi- the scoring value of the ith evaluation index; and For the judgement matrix of steps 1e8, the standard value of the
ai- the weight coefficient of the ith evaluation index. mean random consistency index RI is given in Table 10 (Saaty,
2008).
We found that the weight coefficient of each evaluation index
for different equipment or systems is different from the analysis in Step 5: Adjust the consistency of the judgement matrix D and
the Section 2. As shown in the Eq. (1), the weight coefficient of obtain the weight coefficients of the evaluation indexes.
evaluation index ai in all of calculation parameters may have a
significant impact on the result of Index, which is regarded as the If CR < 0.01, the consistency of the judgement matrix D is
importance level of equipment. In order to calculate the Index, we satisfactory, which means that the weight apportionment of each
should determine weight coefficient of every evaluation index ai evaluation index is reasonable; if not, the judgement matrix D
first. The AHP is a theory of measurement through pairwise com- should be adjusted until the consistency meets the above
parisons that relies on the judgements of experts to derive priority requirement. Then, every value of the eigenvector W ¼ (u1,u2,/,un)
scales (Shen et al., 1998). Therefore, the AHP is consistent with this corresponds to the weight coefficient of each evaluation index.
case, and its calculating flow is given as follows (Yuliang et al.,
2003; Saaty, 2008): Step 6: Calculate the total score of the evaluation indexes Index
with Eq. (1) by plugging in the weight coefficient and score of
Step 1: Construct a judgement matrix D through pairwise each evaluation index. The total score represents the impor-
comparisons among the evaluation indexes: tance level of the equipment. Finally, we have fulfilled the
quantification of the weight coefficient of the eight evaluation
indexes for the DPE and obtained the importance level of the
equipment according to this method.
2 3
u11 u12 …… u1n
6 u21 u22 …… u2n 7
D¼6
4 ……
7 (2)
…… …… …… 5 3.2. Analysis of eliminating the subjective factors based on the MCS
un1 un2 …… unn
In this framework, the score values of the evaluation indexes
where uij is a relative importance level of the ith evaluation index to were directly given by the expert or the maintenance engineer in
the jth evaluation index, and uji is the relative importance level of the scoring process based on their scoring criteria. Therefore, the
the jth evaluation index to the ith evaluation index. Thus, the value scoring value and the importance level of the equipment might be
of uji is the reciprocal of uij, namely, uij$uji ¼ 1. Table 1 shows the subject to some influence from subjective factors and individual
definition and the fundamental scale of the relative importance differences among people, which would lead to some errors in the
level. end results of the importance level of the DPE. According to the
investigation and analysis of similar problems, the above problem
Step 2: Calculate the maximum eigenvalue lmax of the judge- can be solved using the MCS approach (Williams-Kovacs and
ment matrix D using the system of homogeneous linear equa- Clarkson, 2014). After the weight coefficients ai were solved using
tions as follows: the AHP, using many simulation calculations with the MCS
approach for the result of the previous step will effectively elimi-
nate the influence of subjective factors. With the application of the
8 MCS approach and the AHP, the robustness of the importance level
>
> ðu  lÞu1 þ u12 u2 þ / þ u1n un ¼ 0
< 11
u21 u1 þ ðu22  lÞu2 þ / þ u2n un ¼ 0 Table 10
(3)
>
> //// Standard value of the mean random consistency index RI.
:
un1 u1 þ un2 u2 þ / þ ðunn  lÞun ¼ 0 n 1 2 3 4 5 6 7 8

RI 0.00 0.00 0.58 0.9 1.12 1.24 1.32 1.41


1054 Y. Tang et al. / Journal of Natural Gas Science and Engineering 26 (2015) 1050e1058

of the DPE is further increased. The logic block diagram of the MCS Class B equipment with an area percentage of 25%e65% and Class C
is shown in Fig. 2 (Yuliang et al., 2003; Marseguerra and Zio, 2000). equipment with an area percentage of 65%e100%.
As shown in the Fig. 2, eight random numbers in [0, 1] were From field surveys and the related literature, the existing
generated in the MCS process. The random numbers were regarded maintenance methods for DPE include lubrication (LUB), service
as the weight coefficient of the eight evaluation indexes and were (SVC), corrective maintenance (CM), time-based maintenance
assigned with the priority order obtained with the AHP in the (TBM), hidden failure detection (HFD) and condition-based main-
previous step. In other words, for any group of random numbers, tenance (CBM). To effectively implement these maintenance
the largest random number is assigned to the top priority, the methods, the primary failure mode, the failure effect and the cause
smallest one to the lowest priority, and the rest of the random of the three categories of equipment are analysed before estab-
numbers to the other evaluation indexes. Then, in an MCS lishing the MDMTs of the DPE. Then, the MDMTs of the DPE are
computation, the total score of all of the evaluation indexes of the established by referencing the logic decision diagram of reliability
equipment can be calculated using Eq. (1), and the importance level centred maintenance (RCM). The MDMTs of the DPE are described
of the equipment can be obtained and ranked according to the in detail as follows:
Index. Using N simulation calculations, several ranking values for
the same equipment are obtained based on their different impor- (1) The failure of Class A equipment has little or no influence on
tance levels. Then, the importance level of a single piece of equip- the function of the entire drilling and production system or
ment can be determined based on its cumulative frequency causes lower maintenance costs. Increasing the spare parts
sequence reaching 1; more specifically, the faster the cumulative inventory or decreasing the fault frequency for Class A
frequency of one piece of equipment reaches “1”, the higher the equipment cannot affect the drilling and production process
importance level of that equipment (Yuliang et al., 2003). of oil and gas. Four maintenance strategies, including LUB,
SVC, CM and TBM, are suitable for Class A equipment. A
maintenance decision tree for Class A equipment is shown in
4. Basing maintenance decisions on the importance level of Fig. 3 (Rausand, 1998).
the equipment (2) When Class B equipment fails, it might result in more severe
consequences, but it usually does not influence personnel
According to the statistical data indicating the priority order of safety or the environment. The failure frequency of Class B
the equipment, their cumulative frequencies can be plotted. Based equipment could be reduced through reasonable mainte-
on the principle of establishing a cumulative frequency curve chart, nance strategies that might cause higher maintenance costs.
the percentage of area on the right side of the curve can be taken as An MDMT for Class B equipment with the appropriate
another representation of the importance level of the equipment. A maintenance strategies, including LUB, SVC, TBM, CM and
larger percentage indicates a higher importance level. Based on the HFD, is shown in Fig. 4 (Rausand, 1998).
different area percentages of the equipment, namely, the different (3) The failure of Class C equipment might endanger personnel
importance levels among them, the DPE can be divided into three safety, pollute the environment or cause significant eco-
categories: Class A equipment with an area percentage of 0e25%, nomic consequences. To ensure the operation reliability and
maintenance economy of Class C equipment, the failure
frequency should be reduced by increasing the maintenance
costs with advanced maintenance methods. From the anal-
ysis above, a maintenance decision tree for Class C equip-
ment is shown in Fig. 5, adopting LUB, SVC, TBM, HD and
CBM (Rausand, 1998).

5. Case study

The usage and maintenance records for a well control system


used in drilling operations were collected from an oil company's
drilling crew in the Tarim Oilfield. We performed analyses of their
importance levels and maintenance methods with the framework
of making maintenance decisions for the DPE. First, a hierarchy tree
of the well control system was established by analysing its
matching equipment, as shown in Fig. 6.

Fig. 2. Logic block diagram of the MCS. Fig. 3. MDMT for Class A equipment.
Y. Tang et al. / Journal of Natural Gas Science and Engineering 26 (2015) 1050e1058 1055

Fig. 4. MDMT for Class B equipment.

In order to obtain the relative importance level of the eight coefficients and priorities of eight evaluation indexes were deter-
evaluation indexes, the operation and maintenance records of the mined for the equipment in the well control system, as shown in
equipment in the well control system were collected and analysed Table 11.
from the related resource information database. Then, the pairwise Based on the logic block diagram of the MCS shown in Fig. 2, a
comparison among the eight evaluation indexes was evaluated by programming calculation of the MSC was carried out by using the
the maintenance engineer and expert. The score value of each Matlab software. In the calculation process with the MSC, the
pairwise comparison was obtained by reference to the Table 1. After simulation times was taken as N ¼ 2000, and a dataset that includes
determining the score values, a judgement matrix D was con- 2000 sequences for a single piece of equipment can be obtained.
structed with Eq. (2) as follows: After a series of calculations on all of the well control equipment,
the statistical approach was applied to process these data about the
2 3 sequence of the equipment in the MSC. Finally, the cumulative
1 6 7 7 8 8 9 9
6 1=6 1 7 5 6 6 7 77 frequency diagram about the sequence of the importance level of
6 7
6 1=7 1=7 1 1=4 1=3 1=3 1=2 27 the well control equipment, including Hydraulic Accumulator,
6 7
6 1=7 1=5 4 1 2 3 4 47 Hydraulic Pump, Oil Suction Filter and Oil Return Filter, is plotted in
D¼6
6 1=8
7
6 1=6 3 1=2 1 2 3 277 Fig. 7.
6 1=8 1=6 3 1=3 1=2 1 1=2 27 In the Fig. 7, the cumulative frequency of both Hydraulic Accu-
6 7
4 1=9 1=7 2 1=4 1=3 2 1 45 mulator and Hydraulic Pump reach “1” when the sequence of the
1=9 1=7 1=2 1=4 1=2 1=2 1=4 1 equipment is at “8”. However, the cumulative frequency of the
Hydraulic Accumulator increases faster than that of the Hydraulic
Entering the judgement matrix D into Eq. (3) in Matlab software,
Pump before reaching “0.8”, which indicates that the importance
its maximum characteristic value lmax was calculated:
level of the Hydraulic Accumulator is higher than that of the Hy-
lmax ¼ 8:9204 draulic Pump. The cumulative frequency of the Hydraulic Pump is
faster than that of the Oil Suction Filter and the Oil Return Filter.
The eigenvector related to the maximum eigenvalue lmax was Therefore, the sequence of the importance level is as follows:
obtained with Eq. (4) as follows:

W ¼ ð 0:4696 0:2345 0:0289 0:0950 0:0627 0:0401 0:0458 0:0236 Þ

Hydraulic Accumulator, Hydraulic Pump, Oil Suction Filter and Oil


Return Filter. Additionally, the percentage of area on the right side
And the consistency of the judgement matrix D was checked of each cumulative frequency curve can be determined based on
using Eqs. (5) and (6): the Fig. 7. Then their area ratios are given in the form of histograms
in Fig. 8.
CI ¼ 0:1315 The area ratio of the Hydraulic Accumulator is between 65% and
100% in the Fig. 8, so it is sorted as Class C equipment. The Hydraulic
Accumulator should adopt HD and CBM based on the MDMT for
CR ¼ 0:0923 < 0:1 Class C equipment in the Fig. 5. The Hydraulic Pump is also Class C
equipment from its area ratio, which is higher than 65% in the Fig. 8.
The results above show that the judgement matrix D that were
According to the MDMT in the Fig. 5, HD and TBM were applied to
constructed through pairwise comparisons for eight evaluation
the Hydraulic Pump. The Oil Suction Filter belongs to Class B
indexes abided by the consistency demand. Therefore, the weight
1056 Y. Tang et al. / Journal of Natural Gas Science and Engineering 26 (2015) 1050e1058

Fig. 5. MDMT for Class C equipment.

Fig. 6. Hierarchy tree of the well control system.

equipment due to its area ratio falling within the range of 25%e65%. without using mathematical model and decision-making theory,
From the MDMT for Class B equipment in the Fig. 4, the Oil Suction there were some undesirable phenomena and problems during the
Filter should be implemented with the HD and CM maintenance maintenance and management for oil and gas DPE. In order to
methods. The area ratio of the Oil Return Filter is less than 25%, so it improve the situation of the maintenance and management, we
is Class A equipment. Its maintenance method should be SVC and presented a framework for making maintenance decisions for DPE
CM, referring to the MDMT in the Fig. 3. Finally, the maintenance in this study to restrain safety accidents and economic losses in the
strategy and method for the Hydraulic Accumulator, the Hydraulic oil and gas exploitation process. We summarized four categories
Pump, the Oil Suction Filter and the Oil Return Filter were formu- influence factors, including eight influence factors to evaluate the
lated and performed. Through the reviews of the relevant experts importance level of the DPE. Eight influence factors were regarded
and maintenance engineers for the maintenance methods of the as evaluation indexes and their scoring criteria were defined to
above equipment, the framework of the maintenance decision quantify the result of subjective evaluation. The evaluation model
making of the DPE is considered reasonable and effective. calculating importance level of the DPE was established based on
the AHP and the MCS. By applying the model, not only the impor-
tance levels of the DPE were obtained, but also the human sub-
6. Conclusion jective effects of the scoring process were eliminated. Then, we
divided the DPE into three categories, including Class A, B and C
Due to the previous maintenance decision-making process based on their importance level values and established their
MDMTs by reference to the logic decision tree of RCM theory.
Table 11
Weight coefficients and priorities of the eight evaluation indexes.
Finally, a well control system in the Tarim Oilfield was made
maintenance decisions according to the framework. Their mainte-
S/N Factor Weight Priority
nance methods from the MDMTs were considered reasonable and
1 S 0.4696 1 effective by the relevant experts and maintenance engineers. So the
2 SF 0.2345 2 feasibility of maintenance decision-making framework is verified
3 FR 0.0289 7
effectively.
4 MC 0.0950 3
5 DL 0.0627 4 In this study, the mathematical model and decision-making
6 MA 0.0401 6 theory applying in the framework for making maintenance de-
7 DT 0.0458 5 cisions improve scientificity and reasonability of the mainte-
8 M 0.0236 8
nance and management for the DPE. On the premise that the
Y. Tang et al. / Journal of Natural Gas Science and Engineering 26 (2015) 1050e1058 1057

Fig. 7. Cumulative frequency sequences of the equipment.

Fig. 8. Area percentage of the equipment.

evaluation indexes and scoring criteria are revised according to Conflict of interests
their industry standards and maintenance features, the frame-
work can be popularized not only in the petroleum and petro- The authors declare that there are no conflicts of interest
chemical equipment but also in other industries equipment. regarding the publication of this article.
There are three evaluation methods of equipment importance
level that were presented in this paper. They can be applied to Acknowledgements
more complicated equipment or systems in other industries to
identify key components and parts and simplify the analysis The work was supported by the Natural Science Foundation of
tasks in next step. Moreover, these research methods and results China (Grant no.51274171) and the Graduate Student Innovation
in this study will probably provide some references for me- Fund of the School of Mechatronic Engineering, Southwest Petro-
chanical integrity management in the petroleum industry. But leum University (CX2014BZ04).
some relevant topics have not yet been completely worked out in
our study, for example, optimizing inspection and maintenance
intervals and reducing maintenance costs. In our next research References
work, we will focus on these topics to improve the framework
Arunraj, N.S., Maiti, J., 2010. Risk-based maintenance policy selection using AHP and
further. On account of the framework possessing fixed pro- goal programming. Saf. Sci. 48 (2), 238e247.
cedures and contents in application process, we are also planning Aven, T., Vinnem, J.E., 2005. On the use of risk acceptance criteria in the offshore oil
to develop application software so that its usability and effec- and gas industry. Reliab. Eng. Syst. Saf. 90 (1), 15e24.
Azizi, A., Fathi, K., 2014. Selection of optimum maintenance strategies based on a
tiveness will be improved in engineering applications. fuzzy analytic hierarchy process. Manag. Sci. Lett. 4 (5), 893e898.
1058 Y. Tang et al. / Journal of Natural Gas Science and Engineering 26 (2015) 1050e1058

Bertolini, M., Bevilacqua, M., 2006. A combined goal programmingdAHP approach Kyriakidis, E.G., Dimitrakos, T.D., 2006. Optimal preventive maintenance of a pro-
to maintenance selection problem. Reliab. Eng. Syst. Saf. 91 (7), 839e848. duction system with an intermediate buffer. Eur. J. Oper. Res. 168 (1), 86e99.
Bevilacqua, M., Braglia, M., Gabbrielli, R., 2000. Monte Carlo simulation approach for Marseguerra, M., Zio, E., 2000. Optimizing maintenance and repair policies via a
a modified FMECA in a power plant. Qual. Reliab. Eng. Int. 16 (4), 313e324. combination of genetic algorithms and Monte Carlo simulation. Reliab. Eng.
Chang, K.P., Chang, D., Zio, E., 2010. Application of Monte Carlo Simulation for the Syst. Saf. 68 (1), 69e83.
Estimation of Production Availability in Offshore Installations. In: Simulation Murthy, D.N.P., Atrens, A., Eccleston, J.A., 2002. Strategic maintenance management.
Methods for Reliability and Availability of Complex Systems. Springer London, J. Qual. Maint. Eng. 8 (4), 287e305.
pp. 233e252. Perrons, R.K., Richards, M.G., 2013. Applying maintenance strategies from the space
de Leo  n Hijes, F.C.G., Cartagena, J.J.R., 2006. Maintenance strategy based on a and satellite sector to the upstream oil and gas industry: a research agenda.
multicriterion classification of equipments. Reliab. Eng. Syst. Saf. 91 (4), Energy Policy 61, 60e64.
444e451. Rausand, M., 1998. Reliability centered maintenance. Reliab. Eng. Syst. Saf. 60 (2),
Dekker, R., 1996. Applications of maintenance optimization models: a review and 121e132.
analysis. Reliab. Eng. Syst. Saf. 51 (3), 229e240. Saaty, T.L., 2008. Decision making with the analytic hierarchy process. Int. J. Serv.
Dey, P.K., 2004. Analytic hierarchy process helps evaluate project in Indian oil Sci. 1 (1), 83e98.
pipelines industry. Int. J. Oper. Prod. Manag. 24 (6), 588e604. Shen, Q., Lo, K.K., Wang, Q., 1998. Priority setting in maintenance management: a
Doostparast, M., Kolahan, F., Doostparast, M., 2014. A reliability-based approach to modified multi-attribute approach using analytic hierarchy process. Constr.
optimize preventive maintenance scheduling for coherent systems. Reliab. Eng. Manag. Econ. 16 (6), 693e702.
Syst. Saf. 126, 98e106. Triantaphyllou, E., Kovalerchuk, B., Mann, L., Knapp, G.M., 1997. Determining the
Du, G., Ren, M.P., Yu, Y.F., Li, X.F., Zhang, Y.Q., Chen, X.L., et al., 2013. Risk ranking and most important criteria in maintenance decision making. J. Qual. Maint. Eng. 3
well control measures of drilling blowout in Xinjiang oilfield. Adv. Mater. Res. 6 (1), 16e28.
(16), 844e849. Williams-Kovacs, J.D., Clarkson, C.R., 2014. A new tool for prospect evaluation in
Eti, M.C., Ogaji, S.O.T., Probert, S.D., 2006. Reducing the cost of preventive mainte- shale gas reservoirs. J. Nat. Gas Sci. Eng. 18, 90e103.
nance (PM) through adopting a proactive reliability-focused culture. Appl. En- Yuliang, D., Yujiong, G., Kun, Y., 2003. Criticality analysis on equipment in power
ergy 83 (11), 1235e1248. plant based on Monte Carlo simulation. In: Proceedings-Chinese Society of
Hmida, J.B., Gaspard, A.J., Lee, J., 2013. Tqm-based equipment maintenance in oil- Electrical Engineering, 23(8), 201e205.
field service industries. Glob. Perspect. Eng. Manag. 2 (2), 60e66.

You might also like