Professional Documents
Culture Documents
Structural Reliability of Corroded Pipeline Using The So-Called Separable Monte Carlo Method
Structural Reliability of Corroded Pipeline Using The So-Called Separable Monte Carlo Method
J Strain Analysis
1–8
Ó IMechE 2018
Structural reliability of corroded Reprints and permissions:
sagepub.co.uk/journalsPermissions.nav
pipeline using the so-called Separable DOI: 10.1177/0309324718782632
journals.sagepub.com/home/sdj
Abstract
The evaluation of the failure probability of corroded pipelines is an important calculation to quantify the risk assessment
and integrity of pipelines. Traditional Monte Carlo simulation method has been widely used to solve this type of prob-
lems, where it generates a very large number of simulations and takes longer time in computing. In this study, enhanced
computational method called Separable Monte Carlo is employed to evaluate the time-dependent reliability of pipeline
segments containing active corrosion defects, where a practical example was used. The results show that the Separable
Monte Carlo simulation method not only minimizes the computational cost strongly but also improves the calculation
precision.
Keywords
Pipeline, probability of failure, corrosion defects, Crude Monte Carlo, Separable Monte Carlo
Figure 1. (a) Failure of gas pipeline and (b) environmental damage caused by pipeline failure.2
lengths of defects), and errors related to the manufac- taken into account in this analysis. The results obtained
ture of the pipes. by the SMC are compared with those obtained with the
The probabilistic approach is based on a limit state traditional method (MCS).
function (LSF) represented by the difference between
the burst pressure of corroded pipes and operating
pressure, where the LSF represents the safe region of Reliability of a corroded pipeline
applications if it has positive value (LSF . 0) and the The residual life of a pipeline with corrosion defects can
failure region for negative values (LSF 4 0).12 This be evaluated using an LSF. This later defines the safe
approach takes into consideration the uncertainties of region (LSF . 0) and the failure region (LSF 4 0).
the inspection tools, human errors, and design manu- Two types of failure modes can result from an active
facturing. Therefore, corroded pipeline data are taken corrosion defect: the leak or the rupture failure
in a probabilistic way as random variables that follow modes.14,19 In this work, the studied failure mode is the
specific distributions. Two main types of methods can burst (rupture). These failures modes can be expressed
be cited in this category. First analytical methods, the using equations (1a) for the leak and equation (1b) for
first-order reliability method (FORM) and second- the rupture
order reliability method (SORM) are used several times
to solve reliability problems of corroded pipelines.10,13 LSF = d k t ð1aÞ
These methods are based on the calculation of the relia- LSF = PBurst POp ð1bÞ
bility index, where the LSF is used. Given the difficul-
ties of modeling of the latter, the presence of non- where d is the corrosion depth, t is the pipe wall-thick-
Gaussian variables pose a great problem that makes ness, k is multiplying factor to indicate the safety limits
this type of methods less used. Second, the simulation and can take the value of 0.8 (meaning 80% of pipe
methods represented by the Monte Carlo simulation wall-thickness) in equation (1b), PBurst is the failure
(MCS; classical) method,12,14–16 which is based on the pressure calculated by one of the known international
generation of a large number of random values accord- standards (e.g. ASME B31G, modified B31G, DNV
ing to different types of distributions (normal, log-nor- RP F101, and Shell-92), and POp is the operating pres-
mal, Gumbel, etc.) for estimating the failure probability sure. Therefore, the LSF depends on the same para-
of corroded pipelines. The MCS method is character- meters as the failure pressure model, the operating
ized by its flexibility, widely used for the calculation of pressure, POp , and the evolution of corrosion defect
the Pf of corroded pipelines.17 The main disadvantage dimensions. Thus, the LSF is function of the following
of this technique is that it is very expensive in time, parameters
especially when failure probability is greater than 1026.
LSFðTÞ = f D, t, sy , su , dðTÞ, LðTÞ, Pop ð2Þ
The purpose of this article is to apply new method
so-called Separable Monte Carlo (SMC) for the estima- where D is the pipeline outer diameter, sy and su are
tion of failure probability of corroded pipelines, which the yield and ultimate tensile strength, respectively, and
minimize both the number of simulations and the com- d(T) and L(T) are the depth and the length of the cor-
putational cost, and maximize the calculation precision. rosion defect at time T, respectively. Based on this
In this study, inspection results carried out in a pipeline assumption, the initial value of the length and the depth
built with API 5L X52 steel were used.12,15,18 Data col- can be indicated by L0 and d0 , respectively. Therefore,
lected as the properties of the material, pipeline design the steady growth prediction equation of corroded
parameters, and the defects geometries found were pipeline at T moment is given by
Ben Seghier et al. 3
dðTÞ = d0 + Vd ðT T0 Þ ð3Þ
LðTÞ = L0 + VL ðT T0 Þ ð4Þ
SMC simulation c 1 1 XN X M
Pf = I G PBurst 4POp ð12Þ
N M i=1 i=1
Traditionally, Monte Carlo method has been one of
the most successful methods proposed to calculate the Figure 3 shows a comparison between the MCS and
failure probability represented by equation (8). A major SMC principles, while Figure 4 shows the algorithm
advantage of using MCS method is that it is the sim- steps of the SMC approach.
plest and the straightforward calculation method, the According to Smarslok and colleagues,20,21 the CoV
most commonly used, and it proceeds in three steps: (1) is calculated using the following expressions
randomly sampling the sets of the input parameters of
h i 1 1 ð N 1Þ 1
the LSF according to underlying distributions; (2) Var Pd
SMC = g +u + g
repeated step (1) N times for all samples; and (3) post- N M R, R N M R1 , R2
calculating the failure probability. Then, equation (8) ð13Þ
can be approximated by
where
1 X
N h i
c
Pf = I G PBurst 4POp ð10Þ u = E FPOp ðPBurst Þ2 P2f ð14Þ
N i=1
4 Journal of Strain Analysis 00(0)
Figure 3. Illustration of the simulation procedure mechanism of the (a) Monte Carlo simulation and (b) Separable Monte Carlo
simulation.
h i
gR, R = Pf E Fc ðPBurst Þ2 ð15Þ using the above methodology, a set of pipeline para-
h i meters were applied. These later were gathered from
gR1 , R2 = E FPOp (minðPBurst1 , PBurst2 Þ2 P2f ð16Þ Caleyo et al.,12 Qian et al.,15 and Zhou.18 Moreover,
these parameters were detected based on inspection
operation. The LSF is represented by equation (5) and
used in the calculation determined by material property
Illustrative example (yield and ultimate tensile strength), pipe design para-
In order to implement the SMC simulation for the esti- meters (diameter, D; wall thickness t), defects geome-
mation of the failure probability of corroded pipeline tries (depth and length of defects), operation pressure,
Ben Seghier et al. 5
Table 2. The results of failure probabilities using CMC and SMC methods for various number of simulations.
CMC: Crude Monte Carlo; SMC: Separable Monte Carlo; CoV: coefficient of variation.
Note: The bold numbers represent the minimum simulation number to satisfy the condition of COV \ 0.05.
22. Zhou W. System reliability of corroding pipelines. Int J 25. Hasan S, Khan F and Kenny S. Probability assessment of
Press Vessel Pip 2010; 87: 587–595. burst limit state due to internal corrosion. Int J Press Ves-
23. Vinod G, Shrivastava OP, Saraf RK, et al. Reliability sel Pip 2012; 89: 48–58.
analysis of pipelines carrying H 2 S for risk based inspec- 26. Zhou W, Hong HP and Zhang S. Impact of dependent
tion of heavy water plants. Reliab Eng Syst Saf 2006; 91: stochastic defect growth on system reliability of corrod-
163–170. ing pipelines. Int J Press Vessel Pip 2012; 96–97: 68–77.
24. De Leon D and Macı́as OF. Effect of spatial correlation 27. Keshtegara B and Miri M. Reliability analysis of cor-
on the failure probability of pipelines under corrosion. Int roded pipes using conjugate HL-RF algorithm based on
J Press Vessel Pip 2005; 82: 123–128. average shear stress yield criterion. Eng Fail Anal 2014;
46: 104–117.