Pipeline Risk Assessment by Bayesian Belief Network

Abstract

This study focuses on identification of risk factors in pipeline system and also, concentrates on identification of relationship between parameters. In order to achieve this purpose, Bayesian Belief Network with historical data was used to provide a framework for assessing risk relative to the company’s petroleum pipeline system. Each of the variables in the Bayesian Belief Network is described by nodes and each node has a state. Relationships between parameters are presented by arrows. Probability of any node being in state was shown in conditional probability tables. Historical data were helpful to build conditional probability tables. Variables were defined as corrosion, third party damage, mechanical and operational failure.