An Evaluation of Probabilistic Integrity Assessment Codes

Abstract

In recent years, a large number of failure assessment models (FAMs), like SINTAP, BS7910 and API 579 have been developed to evaluate the structural integrity of pipelines. Based on them, various software, such as PRO-LOCA, PROST, xLPR, etc, have been established to evaluate the safety and reliability of pipelines. However, which of these codes fit best the user’s requirements is a difficult problem because it depends on scientific as well as on personal criteria. Therefore, in this paper, we propose several multiple evaluation criteria to discuss the integrity assessment codes. Furthermore, we apply mathematical methods to evaluate the prediction performance (PP) of the codes PRO-LOCA and PROST. The PP covers seven criteria: correlation, multimodality, dispersion, risk, conservativeness, robustness and accuracy. The correlation, multimodality and dispersion reflect the stability of the predicted results, while risk, conservativeness, robustness and accuracy illustrate distributional location characteristics (DLC) of the prediction accuracy (PA).