Multi-Fidelity Models for Failure Onset Analysis of Composites Under Uncertainties

Abstract Advancements in composite manufacturing have enabled innovative design strategies to enhance stress distribution, stiffness, and overall performance of composite structures. However, addressing uncertainties such as material variability and manufacturing defects remains a critical challenge. This study introduces a multi-fidelity Gaussian Process Regression (GPR) framework for the analysis of composite laminates under uncertainty. The approach integrates low- and high-fidelity models through the Carrera Unified Formulation (CUF), ensuring both efficiency and accuracy in structural analysis. Low-fidelity models rely on Equivalent Single-Layer (ESL) theories with through-the-thickness Taylor expansions, while high-fidelity models employ Layer-Wise (LW) theories with Lagrange expansions for detailed layer-level responses. The proposed framework is designed to investigate stress analysis, failure onset, and deflection of composite laminates in the presence of uncertainty. By accounting for material and load variability as well as manufacturing defects, it offers a practical tool for assessing structural behavior and improving composite design reliability.