A Surrogate Model for Predicting Stress Intensity Factor: An Application to Oil and Gas Industry

Evaluation of the stress intensity factor (SIF) for a crack propagating in a structural component is the analytical basis of linear elastic fracture mechanics (LEFM) approach. Handbook solutions give accurate SIF results for simple crack geometries. For intricate crack geometries and complex loading conditions finite element method (FEM), is used to predict SIF. The main drawback of FEM techniques is that they are prohibitively expensive in terms of computing cost and also very time consuming. In this manuscript, authors have presented a Gaussian Process Regression Model (GPRM), which may be used as an alternative to FEM for predicting SIF of a propagating crack. The GPRM is firstly trained using 70 SIF values obtained by FEM, and then validated by comparing the values of SIF predicted by GPRM and FEM for 30 data points (i.e. combination of crack size and loading). On comparing the aforementioned values the average residual percentage between the two is 2.57%, indicating good agreement between GPRM and FEM model. Also, the time required to predict SIF of 30 data points is reduced from 30 mins (for FEM) to 10 seconds with the help of proposed GPRM.