Tuesday, April 20, 2010
10-31

Optimization and process dynamic modeling with neural networks of the enzymatic hydrolysis of sugarcane bagasse for the production of second generation ethanol

Edwin G. Boza Condorena, Laura L.G. Fuentes, Sarita C. Rabelo, Rubens Maciel Filho, and Aline Carvalho da Costa. Chemical Processes, School of Chemical Engineering, State University of Campinas, Av. Albert Einstein, 500. P.O. Box 6066, 13083-970. Campinas. SP. Brazil, Campinas, Brazil

This paper reports an application of neural networks for optimization and dynamic modeling of the enzymatic hydrolysis of sugarcane bagasse pre-treated with calcium hydroxide. The statistical model obtained by factorial design was not accurate in describing the process at the final time of the reaction; presenting a determination coefficient of 40.35%. The F test results for the model, one of them related with the regression and the other with the lack of fit, show that the model obtained can not be used for predictive purposes. Thus, neural networks were used to model the same experimental data set obtained in the factorial design. It was demonstrated that a neural network with 3 neurons in the first layer, 20 neurons in the hidden layer, and 1 neuron in the output layer, is able to describe the dynamic behavior of the glucose concentration in bagasse hydrolysis process, in the ranges of concentrations of cellulase, β-glucosidase and glucose considered by the experimental data set. According to this model, the concentrations of enzymes to achieve maximum glucose concentration are 482 FPU/L (16.1 FPU/g bagasse) of cellulase and 688 CBU/L (22.9 CBU/g bagasse) of β-glucosidase. This corresponds to a glucose concentration of 16.7 g/L and a yield of 83%. The results for the optimal point were obtained using the selected neural network model and optimization algorithms: Simulated Annealing, Threshold Acceptance, and Genetic Algorithm.


Web Page: www.bioetanol.org.br/hotsite//arquivo/editor/file/School%20of%20Chemical%20Engineering%20-%20University%20of%20Ca