Model-Based Optimization of Cyclic Operation of Acetone-Butanol-Ethanol (ABE) Fermentation Process with ex Situ Butanol Recovery (ESBR) for Continuous Biobutanol Production

Abstract

This paper proposes a model-based optimization strategy for a fermentation process coupled with an ex situ butanol recovery-by-adsorption (termed ""ESBR-by-adsorption"" hereafter) process used for continuous biobutanol production. The ESBR-by-adsorption system exhibits cyclic dynamic behavior caused by the periodic switching of the adsorption column for its renewal. Since performance of such a system is largely determined by its dynamic behavior seen after converging to Cyclic Steady State (CSS), the optimization strategy should search for the optimal operating condition leading to the most profitable CSS. For the CSS optimization, we select key optimization variables and define the objective function and constraints. The resulting CSS optimization problem is strongly nonconvex, largely due to the various nonlinearities in the objective function and constraints, e.g., those in the kinetics of the ABE fermentation and adsorption. To alleviate the numerical convergence problem associated with nonconvex optimization problems, we adopt an initialization strategy of identifying a feasible solution region and a ""good"" initial guess through a coarse grid search. With the initialization strategy, two CSS optimization approaches, ""sequential"" and ""simultaneous,"" are examined for the system. With the model and simulation, performances of the two approaches are compared with respect to varying qualities of the initial guess to propose an effective practical CSS optimization strategy for the ESBR-by-adsorption system. The optimized continuous production'by the ESBR-byadsorption system showed significantly improved volumetric productivity of butanol, 5.5- and 3.7-fold increases respectively over the batch fermentation or semibatch fermentation with in situ product recovery.