Biohydrogen production using anaerobic mixed bacteria: Process parameters optimization studies

Abstract

Process parameters optimization using response surface methodology (RSM) and artificial neural network (ANN) coupled with Genetic Algorithm (GA) were carried out to analyze the influence of process parameters on maximum hydrogen production and H2 yield using sucrose as a sole carbon source. Alkali pretreated mixed culture obtained from the food waste was used for fermentative hydrogen production. Box-Behnken design was applied to examine the interactive effect of the significant variables (sucrose concentration, initial pH, inoculum size, and peptone concentration). Characterization of culture indicated that the culture was gram-negative facultative anaerobe. Maximum experimental hydrogen yield of 2.36 mol H2/mol sucrose was achieved at the optimal points predicted by the RSM. Modified Gompertz model and logistic model adequately fitted well and described the fermentative hydrogen production and bacterial growth, respectively. Process modeling abilities of ANN and RSM were compared on the basis of parameters such as estimated values of root mean square error (RMSE), multiple correlation coefficients (R2), and standard error of prediction (SEP). GA couple with ANN was used to find the global optimum point and the maximum H2 yield of 2.39 mol H2/mol sucrose was found at sucrose concentration 15.5 g/l, pH 8, inoculum size (% v/v) 8.4, and peptone concentration 4.9 g/l. The estimated values of RMSE, R2, and SEP for ANN model and RSM model confirm that fitness and prediction accuracy of ANN model was higher when compared to RSM model. From this study, we confirmed that genetic algorithm coupled with ANN technique can be a powerful tool to obtain global optimization in biochemical systems.