Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

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

net/publication/236333520

Modeling Deterioration in Concrete Pipes as a Stochastic Gamma Process for


Time-Dependent Reliability Analysis

Article in Journal of Pipeline Systems Engineering and Practice · March 2013


DOI: 10.1061/(ASCE)PS.1949-1204.0000145

CITATIONS READS

42 974

2 authors:

Dr. Mojtaba Mahmoodian Morteza (Amir) Alani


RMIT University University of Greenwich
96 PUBLICATIONS 998 CITATIONS 109 PUBLICATIONS 1,682 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Dr. Mojtaba Mahmoodian on 01 February 2014.

The user has requested enhancement of the downloaded file.


Modeling Deterioration in Concrete Pipes as a Stochastic
Gamma Process for Time-Dependent Reliability Analysis
M. Mahmoodian 1 and A. Alani 2

Abstract: Concrete sewer pipes encounter corrosion as a predominant form of deterioration that reduces the service life of the pipeline
system and increases the costs of maintenance and rehabilitation. Reliability analysis can help pipeline owners and sewer-system managers to
Downloaded from ascelibrary.org by Greenwich University on 09/04/13. Copyright ASCE. For personal use only; all rights reserved.

predict the service life of the pipes and the time for repair or replacement. To include uncertainties in the reliability analysis, the corrosion
process can be considered a stochastic process. This paper presents a stochastic gamma process model to account for temporal variability and
uncertainties associated with the corrosion process in concrete sewer pipes that typically increase the probability of failure with the aging of
the pipe. The proposed methodology is then applied to a specific concrete sewer system. The key element of the proposed method is the use
of the gamma process concept. A gamma process can model the monotonic behavior of the ageing and deteriorating process. The application
of the gamma process presented in this paper can potentially contribute to widening the use of stochastic processes for the modeling of
structural degradation processes such as wear, fatigue, corrosion, crack, erosion, and creep. DOI: 10.1061/(ASCE)PS.1949-1204
.0000145. © 2013 American Society of Civil Engineers.
CE Database subject headings: Stochastic models; Time Dependence; Probability; Failures; Concrete pipes; Sulfides; Corrosion;
Reliability.
Author keywords: Stochastic modeling; Time-dependent reliability analysis; Probability of failure; Concrete pipes; Sulfide corrosion;
Gamma process.

Introduction corrosion (H2 S corrosion) is amongst the most challenging problems


regarding sewer operation and maintenance. Studies on sulfide
Concrete sewers suffer from sulfide corrosion. This may lead to formation, its corresponding modeling, and its effects on concrete
pipeline failure and potential flooding of neighborhoods and corrosion have been performed for many years (Pomeroy 1976).
residences, traffic disruptions, and pollution of underground water Recently, new findings of a long-term study of H2 S gas adsorption
resources. European cities spend in the order of 5 billion Euro and oxidation on concrete and plastic sewer pipe surfaces have been
per year on wastewater network rehabilitation. This amount will reported (Nielsen et al. 2009).
increase significantly over the coming years due to network aging The mechanism of H2 S in material degradation has been reason-
(Saegrov 2006). ably well understood. However the challenge for sewer system
In the UK, sewer failure has been identified as one of the most managers is the lack of monitoring data and uncertainty about
significant threats to infrastructure by the ICE State of The Nation the factors that affect and control the time-dependent process
Rep. Defending Critical Infrastructure (2009). There are approxi- of concrete corrosion (Pomeroy 1976; Sulfide 1989; Kienow and
mately 310,000 km of sewer pipes with a total asset value of ap- Kienow 2004; Nielsen et al. 2009).
proximately £110 billion in the UK (Hau et al. 2005). With more To deal with this data scarcity and associated uncertainties, the
than 40% of its network constructed before 1945, the UK sewerage use of stochastic models for the reliability analysis of concrete
system is known to be one of the oldest sewerage systems in the sewers may be considered. In a stochastic model, both sampling
world. The aging of sewers causes approximately 5,000 collapses and temporal variability associated with the degradation process
per year in the UK’s sewer pipeline network which costs more can be taken into account.
than £230 million per year, and this rate is increasing (Read and Various frameworks have been proposed to model the behavior
Vickridge 1997). of underground pipelines for different types of material, using a
The degradation of concrete sewer systems can be primarily reliability-based concept. Ahammed and Melchers (1994, 1995,
attributed to concrete corrosion induced by biogenous sulfuric acid 1997) conducted a comprehensive and continuous study on the
attacks, which cause severe structural deterioration, and, ultimately, reliability analysis of underground steel pipelines. To consider
structural collapse of the pipeline. Hydrogen-sulfide- related uncertainty associated with the rate of corrosion and the uncertain
location of its occurrence, they used a probabilistic approach
1
Ph.D. Candidate, School of Engineering, Univ. of Greenwich, Central (first-order second-moment reliability method, FOSM) for the
Ave., Chatham Maritime, Kent, UK (corresponding author). E-mail: analysis of pipeline reliability.
m.mahmoodian@gre.ac.uk Camarinopoulos et al. (1999) used a combination of the
2
Professor, School of Engineering, Univ. of Greenwich, Central approximate quadrature analytical and Monte Carlo methods to
Ave., Chatham Maritime, Kent, UK.
Note. This manuscript was submitted on August 8, 2012; approved on
evaluate the multiple integrals in their reliability analysis for cast-
March 7, 2013; published online on March 9, 2013. Discussion period open iron buried water pipes. They also used the model to assess the
until January 21, 2014; separate discussions must be submitted for indivi- sensitivity of structural reliability to the variation of some important
dual papers. This paper is part of the Journal of Pipeline Systems Engi- parameters such as wall thickness, unsupported length, and external
neering and Practice, © ASCE, ISSN 1949-1190/04013008(5)/$25.00. corrosion coefficient.

© ASCE 04013008-1 J. Pipeline Syst. Eng. Pract.

J. Pipeline Syst. Eng. Pract.


Benmansour and Mrabet (2002) studied the reliability of buried Let dðtÞ denote the deterioration at time t, t ≥ 0, and let
concrete pipes by using the first order second moment method. They the probability density function of dðtÞ, in accordance with the
studied two typical cases to assess the effect of laying on the circum- definition of the gamma process, be given by
ferential behavior of pipes and the effect of an existing continuity
within soil along the pipeline on longitudinal behavior. fdðtÞ ðdÞ ¼ Ga½djαðtÞ; λ ð2Þ
Amirat et al. (2006) used reliability analysis to assess the effect
of both the residual stresses generated during the manufacturing with mean and variance as follows:
process and the in-service corrosion of underground steel pipes. αðtÞ
A probabilistic-mechanical model was used to couple the residual E½dðtÞ ¼ ð3Þ
λ
stress model with the corrosion model to assess the aging effects
through the pipe’s service life.
αðtÞ
Li et al. (2009) presented a methodology for predicting the Var½dðtÞ ¼ ð4Þ
effect of corrosion on the remaining life of underground steel λ2
pipelines with a mechanically-based probabilistic model by taking
Downloaded from ascelibrary.org by Greenwich University on 09/04/13. Copyright ASCE. For personal use only; all rights reserved.

A concrete sewer pipe is said to fail when its corrosion depth,


the effect of randomness into account in pipeline corrosion. Monte
denoted by dðtÞ, becomes more than its concrete cover (Gravity
Carlo simulation was used to calculate the remaining life and its
2007). That is to say, the concrete sewer is failed when all concrete
cumulative distribution function.
cover on the reinforcement is corroded. Assuming that concrete
Mahmoodian and Li (2011) developed an analytical stochastic-
cover a0 is deterministic, the time at which failure occurs is denoted
reliability method to predict the probability of failure for concrete
by the lifetime T. By considering probability density function
sewer pipes. The analytical results obtained from first passage
[Eq. (2)], the probability of failure can then be written
theory were then verified by the Monte Carlo-simulation method.
Z a
In this study a comprehensive approach to the model corrosion 0 Γ½ðαðtÞ; ao λÞ
process in concrete sewers is presented. To model the monotonic PrðT ≤ tÞ ¼ Pr½dðtÞ ≥ a0  ¼ fdðtÞ ðdÞdd ¼
0 Γ½ðαðtÞÞ
progression of a deterioration process, this paper proposes a
gamma-process model for the reduction of pipe-wall thickness ð5Þ
due to corrosion. where Γðν; xÞ ¼ ∫ ∞ ν−1 e−t dt is the incomplete gamma function
t¼x t
A gamma process is a stochastic process with independent, non- for x ≥ 0, ν > 0, and a0 is concrete cover.
negative increments having a gamma distribution with an identical To model concrete corrosion due to a hydrogen sulfide attack,
scale parameter and a time-dependent shape parameter. in terms of a gamma process, the question which remains to be
A stochastic-process model, such as gamma process, incorpo- answered is how the expected deterioration increases over time
rates the temporal uncertainty associated with the evolution of (i.e., the shape function for the gamma-process model). Empirical
deterioration (e.g., Bogdanoff and Kozin 1985; Nicolai et al. 2004; studies show that the expected corrosion depth at time t is often
van Noortwijk and Frangopol 2004). proportional to a power law: [Ellingwood and Mori (1993);
The gamma process allows modeling of gradual damage mono- van Noortwijk and Klatter (1999)]:
tonically accumulating over time, such as wear, fatigue, corrosion,
crack growth, erosion, consumption, creep, swell, and a degrading αðtÞ ¼ ctb ð6Þ
health index. The mathematical aspects of gamma processes can
be found in Dufresne et al. (1991), Ferguson and Klass (1972), where c and b are physical constants. Most of the time, there is
Singpurwalla (1997), and van der Weide (1997). engineering knowledge available about the shape of the expected
From the available literature it appears that Abdel-Hameed deterioration and therefore parameter b in Eq. (6) can be assumed
(1975) was the first to propose the gamma process as a proper constant (van Noortwijk et al. 2007). Table 1 presents the typical
model for deterioration occurring progressively in time. In his values for b from some examples of expected deterioration.
paper he called this stochastic process the “gamma-wear process”. To model the corrosion as a gamma process with shape function
An advantage of modeling deterioration processes through gamma αðtÞ ¼ ctb and scale parameter λ, the parameters c and λ should be
processes is that the required mathematical calculations are rela- estimated. For this purpose, statistical methods are suggested. The
tively straightforward. The stochastic-differential equation from two most common methods that can be used for parameter estima-
which the gamma process follows is presented by Wenocur 1989. tion are maximum likelihood and method of moments. Both meth-
ods for deriving the estimators of c and λ were initially presented
by Cinlar et al. (1977) and were developed by van Noortwijk and
Problem Formulation Pandey (2003).

The mathematical definition of the gamma process is given in Table 1. Typical Values for Exponential Parameter, b, in Different
Eq. (1). Recall that a random quantity d has a gamma distribution Deterioration Types
with shape parameter α > 0 and scale parameter λ > 0 if its prob-
Exponential
ability density function is given by
Deterioration type parameter, b Reference
λα α−1 −λd Degradation of concrete 1 Ellingwood and Mori (1993)
Gaðdjα; λÞ ¼ d e ð1Þ
ΓðαÞ due to reinforcement
corrosion
where αðtÞ = a nondecreasing, right continuous, real-valued func- Sulfate attack 2 Ellingwood and Mori (1993)
tion for t ≥ 0, and αð0Þ ¼ 0. The gamma process is a continuous- Diffusion-controlled aging 0.5 Ellingwood and Mori (1993)
time stochastic process ½dðtÞ; t ≥ 0 with the following properties: Creep 1=8 Cinlar et al. (1977)
• dð0Þ ¼ 0 with probability of one; Expected scour-hole depth 0.4 Hoffmans and Pilarczyk (1995),
• dðτ Þ − dðtÞ ∼ Ga½αðτ Þ − αðtÞ; λ for all τ > t ≥ 0; and and van Noortwijk and Klatter
(1999)
• dðtÞ has independent increments.

© ASCE 04013008-2 J. Pipeline Syst. Eng. Pract.

J. Pipeline Syst. Eng. Pract.


X  
Maximum-Likelihood Estimation n
c^ tbn
ðtbi − tbi−1 Þfψ½^cðtbi − tbi−1 Þ − log δi g ¼ tbn log ð9Þ
In statistics, maximum-likelihood estimation (MLE) is a method of i¼1
dn
estimating the parameters of a statistical model. When applied to a
data set and given a statistical model, MLE provides estimates for where the function ψðxÞ is the derivative of the logarithm of the
the model’s parameters. gamma function
In general, for a fixed set of data and underlying statistical
Γ́ ðxÞ ∂ log ΓðxÞ
model, the method of maximum likelihood selects values of the ψðxÞ ¼ ¼ ð10Þ
model parameters that produce a distribution that gives the ob- ΓðxÞ ∂x
served data the greatest probability (i.e., parameters that maximize
the likelihood function). Let us assume n observations are denoted
by x1 ; x2 ; : : : ; xn. The principle of maximum likelihood assumes Method of Moments
that the sample data set is representative of the population with In statistics, the method of moments is a method of estimation of
a probability density function of fx ðx1 ; x2 ; : : : ; xn ; θÞ, and chooses population parameters such as mean and variance by equating
that value for θ (unknown parameter) that most likely caused the
Downloaded from ascelibrary.org by Greenwich University on 09/04/13. Copyright ASCE. For personal use only; all rights reserved.

sample moments with unobservable population moments and then


observed data to occur, i.e., once observations x1 ; x2 ; : : : ; xn are solving those equations for the quantities to be estimated.
given, f x ðx1 ; x2 ; : : : ; xn ; θÞ is a function of θ alone, and the value Assuming transformed times between inspections wi ¼ tbi − tbi−1 ,
of θ that maximizes the preceding probability density function is i ¼ 1; : : : ; n, the method-of-moments estimates c and λ can be
the most likely value for θ. found from [van Noortwijk and Pandey (2003)]
In the current study a typical data set consists of inspection times Pn
ti , i ¼ 1; : : : ; n where 0 ¼ t0 < t1 < t2 < · · · < tn , and correspond- c^ δ d
¼ Pni¼1 i ¼ bn ¼ δ̄ ð11Þ
ing observations of the cumulative amounts of deterioration di , i ¼ λ^ i¼1 w i tn
1; : : : ; n are assumed to be given as inputs of the model. Fig. 1
schematically shows a time-dependent degradation model in the  Pn  X
dn w2 n
case of two inspections with a deterministic path. 1 − Pn i¼1Pin ¼ ðδ i − δ̄wi Þ2 ð12Þ
λ^
2
The maximum-likelihood estimators of c and λ can be deter- ð i¼1 wi Þ i¼1 wi i¼1
mined by maximising the logarithm of the likelihood function
of the increments. The likelihood function of the observed deterio- The first equation from both methods [i.e., Eqs. (8) and (11)] are
ration increments δ i ¼ di − di−1 , i ¼ 1; : : : ; n is a product of in- the same and the second equation in the method of moments is
dependent gamma densities: simpler because it does not necessarily require iterations to find
the unknown parameter (^c).
Y
n
lðδ 1 ; : : : ; δ n jc; λÞ ¼ fdðti Þ−dðti−1 Þ ðδi Þ
i¼1
Application of the Methodology
Y
n
λcðti −ti−1 Þ
b b
cðtb −tb Þ
¼ δ i i i−1 e−λδi ð7Þ The methodology mentioned in the preceding section is applied to a
i¼1
Γ½cðti − ti−1 Þ
b b
worked example for a concrete sewer pipe subjected to sulfide
corrosion. It is assumed that inspection has been carried out
To maximize the logarithm of the likelihood function, its
every 10 years from the installation time of the pipe. The results
derivatives are set to zero. It follows that the maximum likelihood
indicate that the depth of concrete corrosion for t1 ¼ 10 years
estimator of λ is [van Noortwijk and Pandey (2003)]
and t2 ¼ 20 years, are d1 ¼ 3 mm and d2 ¼ 8 mm. Assuming
c^ t b 35 mm of internal concrete cover on reinforcement and a linear
λ^ ¼ n ð8Þ deterioration (i.e., b ¼ 1), Table 2 presents the two parameter
dn
estimation methods that will give the shape and scale parameters
where c^ must be computed iteratively from the following for the Gamma deteriorating process.
equation: Fig. 2 shows the results obtained for the probability of the pipe
failure by using Eq. (5). It can be concluded from the figure that the
Corrosion Depth, d results from both methods of parameter estimation are in good
agreement.
For repair and rehabilitation planning analysis of the pipe, the
time for the pipe to be unserviceable, i.e., T c , can be determined for
a0 Failure Level a given acceptable risk Pa . For example, using the graph for the
Life time, Tc maximum likelihood method in Fig. 2, a time to failure of T c ¼
Inspection
63 years is obtained for Pa ¼ 0.1. This assumes that there is no
intervention during the service period of 0 to 63 years on the
pipe, such as maintenance and repairs. The ability to develop a
defensible estimate of T c (i.e., time for interventions) is of practical

Table 2. Shape and Scale Parameters Estimated from the Two Presented
Methods
0 Maximum likelihood
t1 t2 Time
Parameter estimation method Method of moments
Fig. 1. Time-dependent degradation model in case of two inspections αðtÞ 0.17t 0.094t
(deterministic path) λ 0.425 0.235

© ASCE 04013008-3 J. Pipeline Syst. Eng. Pract.

J. Pipeline Syst. Eng. Pract.


1 1
0.9 0.9
0.8 0.8
Probability of failure

Probability of failure
0.7 0.7 2 Inspections
0.6 0.6
0.5 0.5 4 Inspections
0.4 0.4
0.3 0.3
Method of Maximum Likelihood
0.2 0.2
Method of Moments
0.1 0.1
0 0
0 25 50 75 100 125 150 175 200 0 25 50 75 100 125 150 175 200
Time (year) Time (year)
Downloaded from ascelibrary.org by Greenwich University on 09/04/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 2. Probability of failure for different methods of estimation Fig. 4. Probability of failure for different number of inspections

1 with the case of two inspections. It can be concluded that having


more inspection data can give a better estimate for c and λ leading
0.9
to a more accurate service life prediction. For instance, although
0.8 service life for acceptable probability of failure of Pa ¼ 0.1 for
Probability of failure

0.7 the case of two inspections is T c ¼ 63 years, in the case of four


0.6 ao = 25mm inspections it is T c ¼ 70 years.
0.5 ao = 35mm
0.4
ao = 45mm Conclusion
0.3
0.2 In this study, corrosion in concrete sewers was modeled using the
gamma-process concept. The proposed method showed that a
0.1
gamma process can properly model the monotonic behavior of
0 the ageing and deterioration process of pipes. The service life of
0 25 50 75 100 125 150 175 200
a concrete pipeline can be predicted with high accuracy for differ-
Time (year)
ent levels of probability of failure by using the model presented in
Fig. 3. Probability of failure for different concrete cover this study.
Two common methods (maximum likelihood method and
method of moments) were suggested and applied for parameter
estimation in the stochastic analysis. A good agreement between
significance to structural engineers and asset managers with regard
the results from both methods was observed.
to planning for repairs and/or rehabilitation of the pipeline. An op-
In addition to the assessment of in-service pipes, the proposed
timum funding allocation for the pipeline system can be calculated
reliability analysis method can also show how change in the design
by doing cost analysis for the repair and replacement of corroded
parameters of pipes (such as concrete cover) can affect the service
pipes with higher risk of failure.
life of the system.
To evaluate how the thickness of concrete cover may affect the
The effect of frequency of inspections on the prediction of the
service life of the pipe, Fig. 3 shows the results for different thresh-
probability of failure was also estimated. It can be concluded that
olds (thicknesses of concrete cover, ao ). Although it is obvious that
having more inspection data can give a better estimate for statistical
a pipe with greater concrete cover will last longer, the differences
parameters leading to a more accurate service life prediction.
among the graphs in Fig. 3 can assist design engineers in assessing
the effectiveness of the thickness of concrete cover on reliability of
the pipeline system.
For instance, although for an acceptable risk of 10% (probability References
of failure Pa ¼ 0.1), the service life for the pipe with 25 mm Abdel-Hameed, M. (1975). “A gamma wear process.” IEEE Trans. Reliab.,
concrete cover is 40 years, increasing the concrete cover to R-24(2), 152–153.
45 mm will increase the service life to 85 years. This remarkable Ahammed, M., and Melchers, R. E. (1994). “Reliability of underground
increment in service life can give rational guidance for designing pipelines subject to corrosion.” J. Transp. Eng., 120(6), 989–1002.
new pipes with longer service life considering more capital invest- Ahammed, M., and Melchers, R. E. (1995). “Probabilistic analysis of pipe-
ment for production and using pipes with thicker concrete cover. lines subjected to pitting corrosion leaks.” Eng. Struct., 17(2), 74–80.
To illustrate how the frequency of monitoring data can affect the Ahammed, M., and Melchers, R. E. (1997). “Probabilistic analysis of
results, the problem is solved with more frequent inspection data. underground pipelines subject to combined stress and corrosion.”
Let us assume there are four inspections (rather than two) every Eng. Struct., 19(12), 988–994.
Amirat, A., Mohamed-Chateauneuf, A., and Chaoui, K. (2006). “Reliabil-
five years over a period of 20 years. Therefore in Eqs. (8) and (9),
ity assessment of underground pipelines under the combined effect of
n ¼ 4, and the depth of inspected concrete corrosion for every active corrosion and residual stress.” Int. J. Pressure Vessels Piping,
five years (ti ¼ 5, 10, 15, and 20 years) is di ¼ 1.5, 3, 5.5, and 83(2), 107–117.
8 mm for i ¼ 1, 2, 3, and 4, respectively. ASCE. (1989). ASCE manuals, and reports on engineering practice No. 69,
The illustrated results for four inspections in Fig. 4 show less Sulfide in wastewater collection, and treatment systems, ASCE,
probability of failure for the area of interest (Pa ≤ 0.1) compared New York.

© ASCE 04013008-4 J. Pipeline Syst. Eng. Pract.

J. Pipeline Syst. Eng. Pract.


ASCE. (2007). ASCE manuals and reports of engineering practice No. 60, Nicolai, R. P., Budai, G., Dekker, R., and Vreijling, M. (2004). “Modeling
Gravity sanitary sewers, 2nd Ed., ASCE, New York. the deterioration of the coating on steel structures: a comparison of
Benmansour, A., and Mrabet, Z. (2002). “Reliability of buried pipes, methods.” IEEE Int. Conf. on Systems, Man and Cybernetics, IEEE,
asranet (integrating advanced structural analysis with structural Danvers.
reliability analysis).” international colloquium 8–10 July, Glasgow, Nielsen, A. H., Jensen, H. S., Hvitved-Jacobsen, T., and Vollertsen, J.
Scotland. (2009). “New findings in hydrogen sulfide related corrosion of concrete
Bogdanoff, J. L., and Kozin, F. (1985). Probabilistic models of cumulative sewers.” Proc. Int. Conf. on Pipeline: Infrastructure’s Hidden Assets,
damage, Wiley, New York. Vol. 1, ASCE, San Diego, 344–353.
Camarinopoulos, L., Chatzoulis, A., Frontistou-Yannas, S., and Pomeroy, R. D. (1976). “The problem of hydrogen sulphide in sewers.”
Kallidromitis, V. (1999). “Assessment of the time-dependent structural Clay Pipe Development Association.
reliability of buried water mains.” Reliab. Eng. Syst. Saf., 65(1), 41–53. Read, G., and Vickridge, I. (1997). Sewerage, Rehabilitation and New
Cinlar, E., Bazant, Z. P., and Osman, E. (1977). “Stochastic process Construction, Repair and Renovation, Arnold, London.
for extrapolating concrete creep.” J. Engrg. Mech. Div., 103(6), Saegrov, S. (2006). Computer aided rehabilitation of sewer and
1069–1088. storm water networks, CARE-S, European commission, Community
Dufresne, F., Gerber, H. U., and Shiu, E. S. W. (1991). “Risk theory with research, The Netherlands.
Downloaded from ascelibrary.org by Greenwich University on 09/04/13. Copyright ASCE. For personal use only; all rights reserved.

the gamma process.” ASTIN Bulletin, 21(2), 177–192. Singpurwalla, N. (1997). “Gamma processes and their generalizations:
Ellingwood, B. R., and Mori, Y. (1993). “Probabilistic methods for condi- An overview.” Engineering probabilistic design and maintenance for
tion assessment and life prediction of concrete structures in nuclear flood protection, R. Cooke, M. Mendel, and H. Vrijling, eds., Kluwer
power plants.” Nucl. Eng. Des., 142, 155–166. Academic Publishers, Dordrecht, 67–75.
Ferguson, T. S., and Klass, M. J. (1972). “A representation of independent van der Weide, H. (1997). “Gamma processes.” Engineering probabilistic
increment processes without Gaussian components.” Ann. Math. Stat., design and maintenance for flood protection, R. Cooke, M. Mendel,
43(5), 1634–1643. and H. Vrijling, eds., Kluwer Academic Publishers, Dordrecht,
Hau, Y., Clarke, B., Howes, C., Cunningham, R., and Mathews, M. (2005). The Netherlands, 77–83.
“Defects in sewer pipe joints and water tests.” Proc. Institution of Civil van Noortwijk, J. M., and Frangopol, D. M. (2004). “Two probabilistic
Engineers, Water Management 158, Institution of Civil Engineers, U.K. life-cycle maintenance models for deteriorating civil infrastructures.”
Vol. WM3, 119–125. Probab. Eng. Mech., 19(4), 345–359, 77–83.
Hoffmans, G. J. C. M., and Pilarczyk, K. W. (1995). “Local scour down- van Noortwijk, J. M., and Klatter, H. E. (1999). “Optimal inspection
stream of hydraulic structures.” J. Hydraul. Eng., 121(4), 326–340. decisions for the block mats of the Eastern-Scheldt barrier.” Reliab.
Institution of Civil Engineers (ICE). (2009). ICE State of The Nation Rep. Eng. Syst. Saf., 65(3), 203–211.
Defending Critical Infrastructure, State of the Nation reports, Institu- van Noortwijk, J. M., and Pandey, M. D. (2003). “A stochastic deterioration
tion of Civil Engineers, London. process for time-dependent reliability analysis” Proc. of the Eleventh
Kienow, K. E., and Kienow, K. K. (2004). “Risk management, predicting Int. Federation of Information Processing Working Group 7.5 Working
your next concrete pipe sewer failure before it happens.” ASCE Int. Conf. on Reliability and Optimization of Structural Systems, Banff,
Conf. on Pipelines, ASCE, San Diego. Canada, Taylor & Francis Group, London, 259–265.
Li, S. X., Yu, S. R., Zeng, H. L., Li, J. H., and Liang, R. (2009). “Predicting van Noortwijk, J. M., van der Weide, J. A. M., Kallen, M. J., and Pandey,
corrosion remaining life of underground pipelines with a mechanically- M. D. (2007). “Gamma processes and peaks-over-threshold distribu-
based probabilistic model.” J. Pet. Sci. Eng., 65(3–4), 162–166. tions for time-dependent reliability.” J. Reliab. Eng. Syst. Saf., 92(12),
Mahmoodian, M., and Li, C. Q. (2011). “Service life prediction of under- 1651–1658.
ground concrete pipes subjected to corrosion.” 4th Int. Conf. on Concrete Wenocur, M. L. (1989). “A reliability model based on the gamma process
Repair, Concrete Solutions, Taylor and Francis Group, London. and its analytic theory.” Adv. Appl. Probab., 21(4), 899–918.

© ASCE 04013008-5 J. Pipeline Syst. Eng. Pract.

View publication stats J. Pipeline Syst. Eng. Pract.

You might also like