(621j) Optimal Control of Fed-Batch Fermentation Process Using Iterative Dynamic Programming

This article proposes a feeding strategy based on a metabolic networks model to optimize ethanol production using Escherichia coli KO11.  In metabolic engineering, growth behavior and product secretion from microorganisms can be predicted using constraint-based analysis of metabolic networks. In this study, a new approach of using stoichiometric model combined with kinetic model allows for precise prediction of ethanol production and formation of other byproducts under specific fermentation environment.  An optimal feeding strategy and minimization of total fermentation time are accomplished by iterative Dynamic Programming approach, which maximizes ethanol production and also minimizes toxic byproduct formation. The switching time from batch to fed- batch is also taken as variable to be optimized.  All variables are computed by using iterative Dynamic Programming utilizing a stoichiometric model. Iterative dynamic programming is modified to increase computational efficiency. Simulation results are promising.