Abstract An autonomous sugarcane bioethanol plant was simulated in EMSO software, an equation oriented process simulator. Three types of fermentation units were simulated: a six parallel fed-batch reactor system, a set of four CSTR in steady state and one consisting of a single stoichiometric reactor. Stoichiometric models are less accurate than kinetic-based fermentation models used for fed-batch and continuous fermenter simulations, since they do not account for inhibition effects and depend on a known conversion rate of reactant to be specified instead. On the other hand, stoichiometric models are faster and simpler to converge. In this study it was found that the conversion rates of sugar for the fermentation systems analyzedwere predictable from information on the composition of the juice stream. Those rates were used in the stoichiometric model, which accurately reproduced the results from both the fed-batch and the continuous fermenter system.

At present, direct on-line measurements of key bioprocess variables as biomass, substrate and product concentrations is a difficult task. Many of the available hardware sensors are either expensive or lack reliability and robustness. To overcome this problem, indirect estimation techniques have been studied during the last decade. Inference algorithms rely either on phenomenological or on empirical models. Recently, hybrid models that combine these two approaches have received great attention. In this work, a hybrid neural network algorithm was applied to a fermentative process. Mass balance equations were coupled to a feedforward neural network (FNN). The FNN was used to estimate cellular growth and product formation rates, which are inserted into the mass balance equations. On-line data of cephalosporin C fed-batch fermentation were used. The measured variables employed by the inference algorithm were the contents of CO2 and O2 in the effluent gas. The fairly good results obtained encourage further studies to use this approach in the development of process control algorithms.

Cephalosporin C, a b -lactam antibiotic, is the starting molecule for industrial production of semi-synthetic cephalosporins. The bioprocess for its production is carried out in batch stirred and aerated tank reactors utilizing strains of the filamentous fungus Cephalosporium acremonium. In this work a comparison was made between the processes of production of cephalosporin C in a conventional batch bioreactor, with synthetic medium containing glucose and sucrose, and in a fed-batch reactor at several flowrates of supplementary medium containing invert sucrose. In general, the fed-batch process was shown to be more efficient than the conventional batch one, and the process in which the lowest supplementation flowrate was used presented an antibiotic production significantly higher than those obtained under the other conditions.

The production of penicillin G by Penicillium chrysogenum IFO 8644 was simulated employing a feedforward neural network with three layers. The neural network training procedure used an algorithm combining two procedures: random search and backpropagation. The results of this approach were very promising, and it was observed that the neural network was able to accurately describe the nonlinear behavior of the process. Besides, the results showed that this technique can be successfully applied to control process algorithms due to its long processing time and its flexibility in the incorporation of new data