Multiscale stochastic fatigue analysis of CFRP laminate composites with Bayesian calibration-based characterization method

Abstract

This paper establishes a novel multiscale stochastic fatigue analysis framework to predict the uncertainty characteristics observed in the fatigue experiments of carbon fiber reinforced polymer (CFRP) laminate composites. A Bayesian calibration-based characterization method derives fatigue parameter distributions for constituent level (fiber, matrix, interface) from lamina fatigue experimental results. With multiscale fatigue analysis framework, a micromechanics theory-based constitutive model is defined to calculate the fatigue damage at the constituent level, and the degradation effects due to fatigue damage are reflected during the finite element (FE) analysis. Additionally, the uncertainty of material properties present in the specimens is captured using the Karhunen-Loe`ve (KL) expansion method, a spectral stochastic finite element method (SSFEM). As a result of multiscale stochastic fatigue analysis, a distribution of fatigue life and fatigue failure mechanisms can be predicted. Considering the stochastic properties observed in experimental results, it can be confirmed that the developed method accurately reflects the realistic fatigue behavior of CFRP laminate composites.