A Bayesian Network approach for risk analysis of composite subsea pipelines subjected to falling objects

Abstract

Thermoplastic composite pipes (TCPs) have potentials to be used as subsea pipelines. TCPs are superior in corrosion resistance, yet vulnerable to impact damages. In this study, a Bayesian Network (BN) model was developed taking the factors, such as water depth, weather condition, seabed stiffness, material aging, etc., as variables of the BN model to evaluate the probability of failure of TCP subsea pipelines when they were subjected to falling objects from passing vessels. Case study examples of an assumed TCP pipeline in shallow water area of Bohai sea of China were used to illustrate the topology structure of the BN model and the modeling procedures of the conditional probability table (CPT) of the BN model variables. The case study examples illustrated that the BN model can predict the probability of failure of a TCP pipeline and diagnose the likely causes of the failure. Additionally, the BN model can be updated by taking into account new data such as aging of materials for the entire service life of the pipeline. The predicted failure probability of the assumed TCP pipeline was not benchmarked due to lack of observation data; however, the proposed BN model is a useful tool for the integrity management of TCP pipelines when widely installed in the future.