Reliability Analysis of Underground Pipelines Using Subset Simulation

An advanced Monte Carlo simulation method, called Subset Simulation (SS) for the time-dependent reliability prediction for underground pipelines has been presented in this paper. The SS can provide better resolution for low failure probability level with efficient investigating of rare failure events which are commonly encountered in pipeline engineering applications. In SS method, random samples leading to progressive failure are generated efficiently and used for computing probabilistic performance by statistical variables. SS gains its efficiency as small probability event as a product of a sequence of intermediate events with larger conditional probabilities. The efficiency of SS has been demonstrated by numerical studies and attention in this work is devoted to scrutinise the robustness of the SS application in pipe reliability assessment. It is hoped that the development work can promote the use of SS tools for uncertainty propagation in the decision-making process of underground pipelines network reliability prediction. pipeline can be quantified in terms of 'performance margin' with respect to specified design objectives. In reliability engineering 'performance margin' is denoted as reliability index, probability of failure and safety margin etc. The failure events in the pipe failure analysis can be formulated as the exceedance of a critical response variable over some specified threshold level with an acceptable factor of safety. By predicting the pipeline reliability, the safe service life can be estimated with a view to prevent unexpected failure of underground pipelines by prioritising maintenance based on failure severity and system reliability (3), (4). There is no general algorithm available to estimate the reliability for a buried pipeline system. The pipeline reliability is usually given by an integral over a high dimensional uncertain parameter space. Methods of reliability analysis such as first order reliability method (FORM), second-order reliability method (SORM), point estimate method (PEM) and Monte Carlo simulation (MCS), etc. are available in literature (5)-(7). In this context, a robust uncertainty propagation method whose applicability is insensitive to complexity nature of the problem is most desirable. Many of the methods are inefficient when there is much number of random variables and failure probabilities are small. Furthermore, some need a large number of samples which is a time consuming procedure. Advanced Monte Carlo methods, often called 'variance reduction techniques' have been developed over the years. In this respect, a promising and robust approach is SS which is originally developed to solve the multidimensional problems of structural reliability analysis (8). The structural