In Section 7.1.2.2., the principle and applications of fed-batch fermentation tech­nologies for ethanol production were discussed. The operation of fermenters under fed-batch conditions is very difficult to model due to the fact that micro­bial cells grow under permanently changing conditions including volume changes explained by feeding of fresh medium into the bioreactor (Cardona and Sanchez, 2007). To tackle this problem, da Silva Henriques et al. (1999) developed a hybrid neural model for alcoholic fermentation by Z. mobilis in a fed-batch regime. The model uses all the information available about the process to deal with the dif­ficulties in its development and could be the basis for formulation of the optimal feed policy of the reactor.

Precisely, the optimization of feeding policy plays a crucial role for increas­ing both productivity and ethanol yield of fed-batch fermentations. Kapadi and Gudi (2004) developed a methodology for determination of optimal feed rates of fresh culture medium during fed-batch fermentation using differential evolu­tion that resulted in a predicted augment of ethanol concentration at the end of each cultivation cycle. Wang and Jing (1998) developed a fuzzy-decision-mak — ing procedure to find the optimal feed policy of a fed-batch alcoholic fermenta­tion using recombinant yeasts able to assimilate glucose and xylose. The kinetic model involved expressions that take into account the loss of plasmids. To solve this problem, a hybrid differential evolution (HDE) method was utilized. The application of HDE has also been carried out in order to simultaneously deter­mine the optimal feed rate, fed glucose concentration, and fermentation time for the case of S. diastaticus during ethanol production. The optimal trade-off solution was found using a fuzzy goal attainment method that allowed obtain­ing a good agreement between experimental and computed results (Chen and Wang, 2003). HDE also has been used for estimation of kinetic parameters dur­ing batch culture of the mentioned yeast (Wang et al., 2001). Other strategies of optimization have demonstrated their usefulness for evaluation of hybrid con­figurations involving reaction-separation integration (Cardona and Sanchez, 2007). For instance, vacuum fermentation technology has been modeled, simu­lated, and optimized by means of factorial design and response surface analysis (da Silva et al., 1999).