04.1 - Aerospace Grade Materials, Structural Analysis, Fatigue and Damage Tolerance
EFFICIENT SIMULATION AND UNCERTAINTY QUANTIFICATION OF PROGRESSIVE CRUSHING IN COMPOSITE LAMINATES
J. Reiner, Deakin University, Australia; R. Vaziri, The University of British Columbia, Canada
This study introduces a computational, data-driven framework for simulating the variability observed in progressive crush tests of composite laminates. A large virtual dataset is generated using efficient finite element analyses (FEA) of flat coupon carbon fibre–reinforced polymer (CFRP) laminates subjected to axial crushing. This dataset enables a systematic global sensitivity analysis to identify the FEA input parameters that most strongly influence the structural crush response. These insights guide the development of machine-learning surrogate models capable of rapidly predicting the crush behaviour of composite specimens with minimal computational cost. Leveraging these surrogate models, uncertainty quantification is performed using a Markov Chain Monte Carlo (MCMC) scheme to estimate parameter distributions that reproduce experimentally observed response variations. The proposed framework provides a scalable pathway for probabilistic assessment of progressive crushing in composite structures. By enabling fast surrogate-based predictions and statistically grounded parameter calibration, the methodology substantially reduces computational cost and accelerates the certification workflow for CFRP materials and components.
DOI: 10.71945/icas2026_0160