Prediction of the Long-Term Effect of Iron on Methane Yield in an Anaerobic Membrane Bioreactor Using Bayesian Network Meta-Analysis

A method for predicting the long-term effects of ferric on methane production was developed in an anaerobic membrane bioreactor treating food processing wastewater to provide management tools for maximizing methane recovery using ferric based on a batch test. The results demonstrated the accuracy of the predictions for both batch and long-term continuous operations using a Bayesian network meta-analysis based on the Gompertz model. The prediction bias of methane production for batch and continuous operations was minimized, from 11~19% to less than 0.5%. A biochemical methane potential-based Bayesian network meta-analysis suggested a maximum 2.55% ± 0.42% enhancement for Fe2.25. An anaerobic membrane bioreactor improved the methane yield by 2.27% and loading rate by 4.57% for Fe2.25, operating in the sequenced batch mode. The method allowed for a predictable methane yield enhancement based on the biochemical methane potential. Ferric enhanced the biochemical methane potential in batch tests and the methane yield in a continuously operated reactor by a maximum of 8.20% and 7.61% for Fe2.25, respectively. Copper demonstrated a higher methane (18.91%) and sludge yield (17.22%) in batch but faded in the continuous operation (0.32% of methane yield). The enhancement was primarily due to changing the kinetic patterns for the last period, i.e., increasing the second methane production peak (k₇₁), bringing forward the second peak (λ₇, λ₈), and prolonging the second period (k₆₂). The dual exponential function demonstrated a better fit in the last three stages (after the first peak), which implied that syntrophic methanogenesis with a ferric shuttle played a primary role in the last three methane production periods, in which long-term effects were sustained, as the Bayesian network meta-analysis predicted.