Kinetic modeling of batch fermentation for mixed-sugar to ethanol production

We analyze the growth dynamics of Saccharomyces diastaticus LORRE 316 utilizing mixed-sugar to produce ethanol and glycerol, using three alternative modeling frameworks, namely S-system, lin-log model and Monod's model. The S-system and lin-log model have identical steady-state solutions, but differ in their representation of transients. The global/local optimization method was applied to determine optimal parameter values for each model. Hybrid differential evolution with data collocation was first applied to determine a satisfied solution. Such a global search phase could avoid yielding a premature estimate and alleviate computational burden. The optimal estimate obtained from the global search phase was then served as the initial starting point for a local optimization method to obtain the refined solution. From experimental observation, the dynamical prediction using the S-system is more acceptable than those of the lin-log model and Monod's model.