Development of an accurate material response model of PICA for an oxy-acetylene test bed system

Material response (MR) modeling is critical to understanding and predicting the behavior of thermal protection system (TPS) materials. Developing an accurate MR model can be a complex, expensive, and time-consuming process. This thesis details the development of a surrogate MR model of an Oxy-Acetylene Test Bed (OTB) system utilizing machine learning methods. 1dFIAT (Fully Implicit Ablation and Thermal response program) and Insulation Thermal Response and Ablation Code (ITRAC) were used to develop the MR model. Two different surface thermochemistry programs were evaluated to create B’ tables for the models. PICA (Phenolic Impregnated Carbon Ablator) was used as a model ablative material and was evaluated at 5 different heat fluxes for 60 seconds on the OTB. A sensitivity analysis of 1dFIAT was then performed to investigate the parameters that were contributing most to error between the predicted and experimental values. Using this method, the environmental inputs were highlighted. A machine learning (ML) model was used to generate and predict boundary conditions. The accuracy of the ML-assisted MR model was quantified against the OTB experimental results. The ML-assisted 1dFIAT model is compared to the ITRAC model. Learnings and challenges associated with the creation of the model using this method and future applications of this approach are also discussed.