Prediction of fatigue crack propagation behavior of AA2524 after laser shot peening

Abstract

Laser shot peening is a relatively novel surface treatment technology generating deep compressive residual stress and further retarding fatigue crack growth rate (FCGR). Meanwhile, the load ratio, R, plays an important role in fatigue crack growth behavior. Evaluating the remaining life of components accurately is essential to damage tolerance design, many efforts have been applied to correlate the R-effect. In this paper, the effect of load ratio on the gain of AA2524′s fatigue life after laser shot peening have been thoroughly studied. Initially, the fatigue crack growth rate of AA2524 after laser shot peening was measured under different load ratios. Subsequently, a comparative analysis between the FCGR for AA2524 before and after laser shot peening was thoroughly conducted. Then, the crack closure model has been applied to correlate the R-effect of peened specimens, but the results are unsatisfied due to the ignore of compressive residual stress. To address this issue, a novel method of predicting crack growth life considering compressive residual stress was proposed. The artificial neural network model and support vector regression model based on particle swarm optimization algorithm were trained by the FCGR data, and the fatigue crack growth life is predicted by a cycle-by-cycle method. The result shows the artificial neural network model is more accurate than the support vector regression model, and this intelligent algorithm can be a potentially effective way to predict fatigue crack growth behavior.