Artificial neural network modeling of biochar enhanced anaerobic sewage sludge digestion

Abstract

Despite the considerable amount of data generated with respect to biochar application in the anaerobic digestion of sewage sludge, there is a research gap for correlations linking biochar physico-chemical parameters to anaerobic digestion performance. In the current study, cumulative methane production (CMP) from anaerobic digestion of sewage sludge amended with biochar is modeled by an Artificial Neural Network (ANN) (2 hidden layers, 12 neurons each) based on data compiled from 51 published biomethane potential tests (BMP). The model reflects the effects of 13 operational parameters covering physico-chemical properties of biochar, sludge characterization and operating conditions. Various types of sewage sludge and biochar under both mesophilic and thermophilic conditions have been successfully modeled with an R2 of 0.9924. An importance analysis is conducted to evaluate the significance of the model input parameters to CMP. Results indicate that operating conditions are more significant to CMP and that CMP is strongly correlated with biochar physical properties and chemical composition where chemical composition has the dominant effect. Overall, this study proves that biochar physico-chemical parameters are correlated to CMP and enables its forecasting under unexamined conditions thus assisting in process optimization, scale up and techno-economic analyses without resorting to BMP tests.