Failure prediction of buried pipeline by network-based geospatial-temporal solution

Abstract

Corrosion is a severe threat to the integrity of buried metal pipes due to the interaction of pipe materials with the surrounding soils. A review of the published literature shows that there are serious challenges for engineers to accurately predict failures in the buried pipeline system. In this paper, a network-based geospatial-temporal solution is developed to predict the risk of pipe failures, considering the spatial dependence and temporal variability of corrosion growth. An algorithm is developed integrating theories of reliability, corrosion science, random field, stochastic process, and copula within the framework of Monte Carlo simulation. The application of the developed algorithm is demonstrated using a real complex gas pipeline network. Also, the effect of discrete intervals, temporal variability and corrosion exposure area on probability of pipe failure is investigated. It is found that the failure probability of pipe segments varies spatially and the location of pipe segment with the largest probability of failure varies with time. It is also found that the largest probability of failure of pipe network is less sensitive to the discrete intervals although a fine discretization of the random field will increase the accuracy of simulation. It can be concluded that the proposed method can predict and visualize the location, time and magnitude of risk of a complex pipe network with reasonable accuracy.