Failure prediction model of oil and gas pipelines

Several failures of oil and gas pipelines have occurred during the operational phase of these infrastructures which have caused monetary and environmental effects. A literature review proves the shortage of studies in the failure assessment of oil and gas pipelines. Moreover, the operators need a prediction tool to estimate the failure probability of the pipelines to decide on their repair and inspection priority before the failure happens. The research, reported in this paper, introduces a new method to forecast the failure probability of oil and gas pipelines. Time dependent failures have been subject of the analysis and the authors benefit from a non-linear regression technique to predict the failure age of the pipelines. Historical data on oil and gas pipeline failures are obtained to find the relationship between the age of failure and four of the most important attributes of the pipelines. They include the operating pressure of pipeline, the external pipe diameter, the pipe wall thickness and the specified minimum yield strength. Stepwise regression analysis is done to find the best subset of the primary variables and their combinations with the second and squared root terms. Validation proves the acceptability of the model. Mont Carlo simulation is applied on the regression model in the next step to consider the uncertainty existed on the operating pressure of the pipelines. Results confirm improvements in the accuracy of the model especially in the beginning and middle ages of the pipelines.