Accelerating Composite Cure Cycle Optimization with Combined Probabilistic Machine Learning and Finite Element Process Simulation

Abstract

Advanced composites are widely used in industries such as aerospace. To certify aerospace structures, the manufacturing process of composites is tightly controlled to ensure compliance with process specifications (i.e., process specs). For example, in autoclave curing of thermoset composites, process specs often limit the maximum temperature in the composite part due to exothermic reaction, maximum thermal lag between the part and autoclave temperature cycle, as well as maximum part porosity and minimum degree of cure. Given all the requirements in process specs, optimizing the temperature cycle in autoclave curing to reduce cycle time and hence increase production throughput is often challenging. While Finite Element (FE) process simulation is used for coupled analysis of heat transfer and thermo-chemical curing reaction of composites during the curing process, due to the long simulation time of high-fidelity simulation models, optimization in the entire design spectrum is not feasible, and most often relies on trial-and-error and engineering insights. In this study, we present a novel framework based on combined probabilistic Machine Learning (ML) and FE to accelerate the process optimization of composites. The probabilistic ML is developed based on the underlying theory of curing and a mathematical foundation of Gaussian Process Regression (GPR). Instead of trial-and-error or time-consuming grid search, this approach relies on targeted FE evaluation of selected cure cycles by the ML model, such that in a step-by-step simulation approach, an optimized solution is obtained. While this approach minimizes the effort to optimize cure processing conditions, it also enables the evaluation of non-traditional cycles such as multi-ramps and holds to minimize the autoclave cycle time. For validation, the framework is applied to optimize the cure cycle of a thick HEXECL AS4/8552 composite laminate cured on a typical Invar tool. A validated FE model is used for process simulation along with the ML model for targeted evaluation. Results show that with only a few analysis steps, an optimized solution can be obtained to satisfy process specs while minimizing cycle time.