Stochastic design of thick autoclave moulded composites

Material and geometric variability in prepreg systems strongly influence defect formation and dimensional non-conformance in thick autoclave-moulded composite laminates, yet laminate design is typically deterministic. This study presents a stochastic, simulation-driven design methodology in which variability in ply thickness and length is incorporated into three-dimensional consolidation simulations using a fully automated mesoscale framework. An industrially representative tapered laminate comprising 218 plies (340 × 320 mm) is used as a case study. Automated ply-book generation, model creation, execution and post-processing enable repeated simulations across defined variability distributions, allowing ∼200 full 3D compaction analyses in three days. The resulting dataset quantifies how cumulative per-ply deviations propagate through the laminate, driving wrinkle formation and thickness non-uniformity. Based on these insights, the laminate is iteratively redesigned to improve robustness under identical variability assumptions. Re-analysis shows reduced sensitivity to variability and improved dimensional compliance. Manufacture and inspection confirm predicted thickness distributions and wrinkle morphology, including accurate reconstruction of wrinkle topography. The results demonstrate that stochastic process simulation can proactively redesign thick composite laminates for improved manufacturability, shifting defect mitigation from reactive process adjustment to variability-informed laminate design.