A Bayesian framework for quantifying uncertainty in the thermal history of curing composite structures

The development of thermal management approaches for composites manufacturing based on physics-based process simulation has become well-established in recent years. However, estimation of thermal boundary conditions, typically in the form of heat-transfer coefficients (HTCs) at the air-part and air-tool interfaces, during convective heat transfer-based curing processes (such as autoclaves and ovens) remains a challenge and a major source of uncertainty. Current deterministic process simulation methods are not suitable for capturing the effect of these HTC uncertainties and their consequential effects on the corresponding thermal histories of curing parts. This work develops and demonstrates the applicability of statistical inference-based models to estimate HTC distributions and the associated uncertainties using synthetic datasets generated from finite element simulations. An experimental case study with real data from the cooling of a heated tool is then presented on using the validated model for inferring, as well as quantifying the uncertainties in HTCs.