Proper Probabilistic Characterization of Uncertainties and its Impact to Reliability-Based Pipeline Integrity and Risk Management

Abstract Reliability-based integrity and risk management requires explicit consideration of various uncertainties in the modelling and decision-making process. In the case of likelihood of failure estimation of individual defects, the main uncertainties involved include uncertainties associated with the physical properties of the pipe, uncertainties associated with defect dimension and the model uncertainty associated with burst pressure calculation. In the case of likelihood of failure and risk analyses for segments with a large number of defects, additional consideration to aggregation of uncertainties from a joint or defect population is needed. The current industry practice for the reliability-based assessment often takes a simplified approach on the probabilistic characterization of various uncertainties and the aggregation of uncertainties, resulting in inaccurate and often conservative estimation of likelihood of failure and risk. The objective of this study is to through examples illustrate how proper probabilistic characterization of uncertainties can be constructed and its impact to reliability-based integrity and risk management. Specifically, this study investigates characterization and modelling of burst pressure calculation error and probabilistic modelling of additive measurement error. The influence of different assumptions and probabilistic representations of the uncertainties to the estimated likelihood of failure for individual defects and a defect population on the segment is examined. The results from the study demonstrated the importance of proper probabilistic characterization of uncertainties in reliability-based integrity management and risk assessment and hopefully open discussions on further improvements in this area in the industry.