Biohydrogen synthesis from food waste hydrolysate: Optimization using statistical design of experiments (DoE) and artificial neural network (ANN)

Dark fermentation is an eco-friendly route for producing hydrogen, a clean next-generation fuel. The present study reports valorization of food waste to biohydrogen (bioH₂) through dark fermentation using Clostridium pasteurianum. The optimization of fermentation parameters using response surface methodology (RSM) with central composite design (CCD) resulted in bioH₂ yield = 1039 mL/L (1.58 mol/mol hexose sugars) for the conditions: pH = 6.5, temperature = 36 °C, TRS concentration = 10 g/L. An artificial neural network coupled with a genetic algorithm (ANN-GA) predicted the optimum parameter set as pH = 6.7, temperature = 36.8 °C, TRS concentration = 10.85 g/L. A bioH₂ yield of 1108 mL/L (1.73 mol/mol hexose sugar) was obtained for these conditions. The modified Gompertz model revealed a maximum bioH₂ production rate of 185.34 mL/L·h for ANN-GA conditions as compared to 153.74 mL/L·h for RSM-CCD predicted conditions. Fermentation at ANN-GA-predicted conditions revealed greater shift of metabolic intermediates towards acetic acid/butyric acid pathway, resulting in higher bioH₂ production. The ratio of acetic to butyric acid increased from 0.9 to 0.94, indicating metabolic shift favoring bioH₂ production. These results demonstrate superiority of ANN-GA technique for simulating behavior of a non-linear system like the metabolic pathway of C. pasteurianum.