Green Energy & Expo

Biography

Biography: Yeshona Sewsynker

Abstract

Renew interest in biohydrogen production as a potential alternative to the depleting fossil fuels is driving the development of this bioprocess. Its scale up requires the development of process models that relate the key operational parameters with the hydrogen yields, models that are accurate and reliable at various scales of the process development. In this paper, Response Surface Methodology (RSM) and Artificial Neural Networks (ANN) were used to model and optimize biohydrogen production at two different process scales. The input variables consisted of inoculum size (10-50%), molasses concentration (100-300 g/L) and Hydraulic Retention Time (10-48 hours) and the output was the hydrogen yield. The considered process scales were the culture volumes of 80 and 800 ml. Seventeen experimental data were generated at each scale and used for model development and process optimization, thus a total of two models at each scale. RSM models gave coefficients of determination (R2) values of 0.97 and 0.89 for 80 and 800 mL respectively. Process optimization with these models predicted a yield of 1.09 and 0.72 mol H2/mol sucrose for 80 and 800 ml scales respectively. Experimental validation gave yields of 0.99 and 0.70 moles H2/mol sucrose for 80 and 800 ml respectively. Thus, deviations from predicted values of 0.1 and 0.02 were obtained from 80 and 800 ml scales. These models showed relatively negligible deviations from their predicted values. These findings suggest that miniaturization of experiments for biohydrogen model development does not significantly impact on the model accuracy. This minimizes the process development cost.