H2S mitigation for biogas upgrading in a full-scale anaerobic digestion process by using artificial neural network modeling

Abstract

Biogas production by anaerobic digestion (AD) has emerged as a prominent bio-renewable energy source in recent years. However, the process also produces undesirable by-products, including H₂S, thereby negatively impacting the biogas quality. This study focused on mitigating H₂S in a full-scale AD located at a wastewater treatment plant (WWTP) by controlling the internal operational parameters by using an artificial neural network (ANN) model. Data from 54 days of AD operation were used to train and validate a structured ANN with a 5-3-1 topology. To minimize the H₂S content, optimum values for dry solid (DS), volatile solid (VS), pH, temperature, and primary sludge fraction (PS) were determined to be 6.2 %, 63 %, 7.7, 35.6 °C, and 67.6 %, respectively, using the particle swarm optimization (PSO) algorithm. This optimization indicated a 49 % reduction in the average H₂S concentration, from 6117 ppm to 3107 ppm. The analysis of relative importance (RI) showed that the pH (RI = −29.5) and PS (RI = −28.7) were the most critical factors affecting biogas quality. Additionally, several solutions derived from the optimization results were practically implemented in the Qom WWTP to achieve optimal conditions, and the outcomes were discussed.