Enhanced bioenergy production via operational optimization of anaerobic co-digestion of chicken litter and straw waste with magnetite nanoparticles

The optimal usage of magnetite (Fe3O4) nanoparticles in anaerobic co-digestion (Co-AD) of chicken litter (CL) and straw wastes (SW) for higher bioenergy production requires systematic investigation due to process complexity involving multiple factors. This study employed response surface methodology (RSM) based on a central composite design to mathematically predict and optimize methane yield (MY, NmL CH4/g VSadded) from batch Co-AD of CL and SW with Fe3O4 nanoparticles supplementation. Artificial neural network (ANN) modeling based on machine learning was further applied to identify alternative solutions for optimal conditions. Three factors were investigated: the co-substrates’ carbon-to-nitrogen ratio (C/N) and total solids level (TS, %), and Fe3O4 nanoparticles dosage (FNP, mg/L). The RSM generated a significant second-order quadratic equation predicting a maximum MY 318.4 NmL CH4/g VSadded achieved under C/N 34.65, TS 5.28%, and FNP 19.39 mg/L. ANN developed a trained multilayer perceptron network (3-14-1) with a superior predictive performance compared to the RSM model. The trained ANN coupled with a genetic algorithm generated a maximum MY of 318.3 NmL CH4/g VSadded under C/N 35.00, TS 4.24%, and FNP 17.42 mg/L. The optimization results were experimentally validated with low prediction error (<0.5%), achieving approximately a 15% increase in MY compared with the Fe3O4-free Co-AD process. Although derived under batch-scale conditions within the investigated experimental design space, this study provides methodological support for process simulation and optimization and informs the efficient use of Fe3O4 nanoparticles to enhance bioenergy production from agricultural wastes.