Experimental Validation of the Adaptive Gaussian Process Regression Model Used for Prediction of Stress Intensity Factor as an Alternative to Finite Element Method

Abstract

Currently, in the oil and gas industry, finite element method (FEM)-based commercial software (such as ANSYS and abaqus) is commonly employed for determining the stress intensity factor (SIF). In their earlier work, the authors proposed an adaptive Gaussian process regression model (AGPRM) for the SIF prediction of a crack propagating in topside piping, as an inexpensive alternative to FEM. This paper is the continuation of the earlier work, as it focuses on the experimental validation of the proposed AGPRM. For validation purposes, the values of SIF obtained from experiments available in the literature are used. The experimental validation of AGPRM also consists of the comparison of the prediction accuracy of AGPRM and FEM relative to the experimentally derived SIF values. Five metrics, namely, root-mean-square error (RMSE), average absolute error (AAE), mean absolute percentage error (MAPE), maximum absolute error (MAE), and coefficient of determination (R2), are used to compare the accuracy. A case study illustrating the development and experimental validation of the AGPRM is presented. Results indicate that the prediction accuracy of AGPRM is comparable with and even higher than FEM, provided the training points of AGPRM are chosen aptly. Good prediction accuracy coupled with less time consumption favors AGPRM as an alternative to FEM for SIF prediction.