The safe operation of pressurised CO2 pipelines is key to the success of Carbon Capture and Storage as a viable means for tackling global warming. However, given their relatively small number currently in operation, the confident prediction of their puncture size failure probability and how it compares with hydrocarbon pipelines are fundamentally important questions that must be resolved. This paper presents the development and application of a robust statistical analytical technique for the confident prediction of the puncture failure size probability distribution for CO2 pipelines. The above involves fitting statistical distributions to the historical puncture size failure data using the Maximum Likelihood Estimator (MLE). The minimum acceptable sample size sufficient for acquiring a reliable MLE is determined by calculating the mean squared error of the MLE as a function of the data sample size. To improve the estimation confidence, given the scarcity of historical failure data, the bootstrapping method is employed to obtain the 95% confidence interval of the MLE. The application of the above technique to pressurised CO2 and hydrocarbon pipelines indicates that as compared to the latter, CO2 pipelines are more likely to experience smaller puncture size failures (e.g. 50 mm), thus resulting in smaller magnitude but more prolonged releases. This directly impacts the preventive and emergency response planning required especially in the case of buried CO2 pipelines where small leaks can remain undetected for long periods.
Authors: Jiahuan Yi, Sergey Martynov, Haroun Mahgerefteh
