Uncertainty analysis of a simplified anaerobic digestion model applied to dynamic agricultural experimental data

This study provides an uncertainty analysis for a simplified anaerobic digestion model in agricultural biogas production, with implications for model output prioritization during parameter identification. Uncertainty in feed measurements is identified as a primary source of parameter uncertainty, magnifying and cascading into model output uncertainty. Experimental co-digestion of maize silage and cattle slurry is used for model validation. The model results in normalized Nash-Sutcliffe efficiencies of 0.87 for the methane flow rate, 0.8 for the acetic acid concentration, 0.74 for total ammoniacal nitrogen, and 0.7 for effluent volatile solids. Effluent total solids and pH value are modeled at 0.2 and 0.37, respectively. The model gives precise estimates of pH and gas concentrations, with coefficients of variation of 1 and 3-4%, respectively. Reasonably precise output states include effluent total and volatile solids, the methane flow rate and total ammoniacal nitrogen with coefficients of variation of 16, 7, 10, and 18%, respectively. Acetic acid concentration shows the highest coefficients of variation, as high as 81%. These findings provide implications for the development of more reliable and trustworthy forecasts in anaerobic digestion, contributing to more efficient and reliable process control.