Zhao, Boxuan2022-04-292022-04-292022-04-292022-04-14http://hdl.handle.net/10012/18203Ultra-high strength steel (UHSS) such as Al-Si coated 22MnB5 are commonly used in the hot forming die quenching (HFDQ) process to produce light-weight automotive parts while maintaining good crashworthiness. The steel blank is typically austenitized in a roller hearth furnace, according to independently set heating zones and other parameters such as roller speed and part spacing. Most often these parameters are chosen heuristically based on experience, resulting in sub-optimal efficiency and part quality. To improve process efficiency and ensure complete austenitization before forming, a complete thermal-metallurgical furnace model that predicts the blank heating profile and the austenitization progress inside a roller hearth furnace is needed. This work introduces a framework for the furnace model, then evaluates three candidate austenitization submodels of different levels of complexity, including: a first order (F1) kinetics model, an Internal State Variable (ISV) model, and a phenomenological model. To address the drawbacks of conventional goodness-of-fit model derivation and evaluation method, the Bayesian model selection technique is introduced and used to evaluate the three candidates. This technique considers the uncertainties in the data, and the trade-off between model complexity and accuracy. Dilatometry data is used to calibrate the models and validate them. The selected austenitization submodel, ISV model, is integrated into the overall furnace model and its performance is validated using roller hearth furnace trials collected with instrumented blanks. The resultant coupled thermo-metallurgical furnace model provides a useful tool for researchers and industrial engineers to maximize production rate and ensure consistent part quality.enhot stampingroller hearth furnace22MnB5 steelMaster Thesis