Benchmarking of Three Girth Weld Flaw Assessment Models: A Validation Methodology

Abstract The application of quantitative risk assessment involves the utilization of pre-existing fitness for service calculators. This departure from the original intent of these calculators necessitates a comprehensive evaluation of their efficacy within an extended domain. This article elucidates the methodology employed to gauge the performance of three girth weld flaw assessment models: namely, the Kastner model, API 1104 Appendix A Option 2, and API 579. The proposed framework delineates a comprehensive analysis, originally devised for financial models, surpassing the conventional validation process, which typically entails a comparison of predicted values against selectively chosen experimental data, commonly referred to as the “least square error Olympics”. A sensitivity analysis of the models was employed to evaluate the impact of data sparseness and ascertain the optimal data collection strategy. Furthermore, we advocate for the measurement of the decrease in entropy as a more effective means of demonstrating the predictive information encapsulated within distinct models. The benchmarking exercise revealed that the venerable Kastner model exhibits a confined range of applicability but maintains a high level of accuracy. Additionally, it was observed that there are instances where a synergistic combination of models can be employed to enhance predictive capabilities.