Activity corrections are required for accurate anaerobic digestion modelling

The impact on the prediction of key process variables in anaerobic digestion (AD) when activity corrections are neglected (e.g. when ideal solution is assumed) is evaluated in this paper. The magnitude of deviations incurred in key variables was quantified using a generalised physicochemistry modelling framework that incorporates activity corrections. Deviations incurred on the intermediate and partial alkalinity ratio (a key control variable in AD) already reach values over 20% in typical AD scenarios at low ionic strengths. Deviations of moderate importance (∼5%) in free ammonia, hydrogen sulfide inhibition, as well as in the biogas composition, were observed. Those errors become very large for components involving multiple deprotonations, such as inorganic phosphorus, and their magnitude (∼40%) would impede proper precipitation modelling. A dynamic AD case simulation involving a series of overloads showed model underpredictions of the process acidification when activity corrections are neglected. This compromises control actions based on such models. Based on these results, a systematic incorporation of activity corrections in AD models is strongly recommended. This will prevent model overfitting to observations related to inaccurate physicochemistry modelling, at a marginal computational cost. Alternatives for these implementations are also discussed.