A Stochastic Framework for Multiaxial Fatigue Analysis of Composite Structures Under Variable Amplitude Loading

This study presents a comprehensive probabilistic framework for predicting the fatigue life of laminated composite materials subjected to multiaxial and variable amplitude loading under uncertainties, with specific application to hot structure skin panels. Traditional deterministic fatigue models often fail to capture the complex interplay of stress components, material anisotropy, and environmental influences, leading to inaccurate life predictions. A stochastic modeling approach is developed to address this, incorporating uncertainties in material properties, loading conditions, and thermal effects. An adaptive L_1-minimization-based polynomial chaos expansion approach (PCE) is employed to quantify the uncertainties' effects, enabling efficient estimation of fatigue response variability and global sensitivity analysis. The methodology includes constant life diagrams (CLDs), a ply-level damage accumulation model using the Tsai-Hill criterion, and peak–valley detection for decomposing stress histories. This integrated framework captures the synergistic effects of longitudinal, transverse, and shear stresses while accounting for load sequence and interaction effects. Finite element analysis of woven carbon-carbon composite panels is used to evaluate fatigue under realistic dynamic loadings. The results demonstrate that the proposed probabilistic design framework enhances fatigue damage prediction in a computationally efficient manner and provides a reliable foundation for the structural integrity assessment and optimization of next-generation aerospace systems.