Evaluation of Reaction Models in CFD Simulations of P. putida Pilot-Scale Bioreactors

Scale-up of bioreactors is a challenging task that can be assisted by CFD simulations provided an accurate model of the microbial reaction behavior. Two common approaches to modeling the bioreactor kinetics are Monod and Michaelis-Menten type kinetics due to their relative simplicity and computational efficiency. More complex models such as metabolism and expression (ME) models are significantly more computationally intensive and as such are not typically used in large-scale CFD simulations. In this work, we introduce a machine learning (ML) reduced order model trained on the underlying ME model dynamics which is then directly integrated into the CFD simulation. The resulting ME/ML model provides similar accuracy of the full ME model but near negligible cost. In this talk, we will examine the differences in results between these various reaction kinetic models and their implications for large-scale bioreactors.