Dual boundary element method for comparative studies on fatigue crack growth models

Abstract

Fatigue crack growth studies require models that accurately predict component life with low uncertainty. Despite the large number of proposed models, there is no clarity on their applicability, which justifies a comparative analysis between some of them. The dual boundary element method (DBEM) was applied for cracked bodies, whereby the stress intensity factors (SIF), the growth rate, and the number of cycles were computed. Three crack increment models were studied under constant amplitude fatigue loads: the Paris, the Klesnil-Lucas, and the Forman models. Results were validated with experimental literature and through the finite element method, indicating that each model represents a specific zone of the crack growth curve. Klesnil-Lucas model reproduces the region near the fracture threshold, Paris fits the controlled crack growth zone, whereas Forman’s model recreates the unstable fracture zone, i.e., when the stress intensity factor approaches the material’s fracture toughness. The J-integral with stress field decomposition gave errors below 0.8% for mode I. Results were similar for the propagation path and the number of cycles to those obtained with the finite element method, with errors of about 3% considering different K-effective approaches. Klesnil-Lucas accurately predicts the number of cycles with an error margin below 3%, considering the curved region in the growth rate at the propagation onset, while the Paris model becomes very conservative, predicting values up to 50% lower than experimental data. The Klesnil-Lukas model is advised for simulating the entire crack propagation.