This paper presents the use of the IWA ADM1 to predict and interpret results from two full-scale anaerobic digesters fed with thermal hyrolysate (waste activated sludge with a long upstream sludge age) from a Cambi hydrolysis process operating at 165°C and 6 bar-g. The first digester was fed conventionally—though intermittently, while the second was heavily diluted through a substantial component of the evaluation period (110 days). There were a number of important outcomes—related to both model application, and model predictions. Input and inert COD: mass ratio was very important, and was considerably higher than the 1.42 g g−1 used for biomass throughout the IWA activated sludge and anaerobic digestion models. Input COD: VS ratio was 1.6 g g−1, and inert COD: VS ratio was 1.7 g g−1. The model succeeded on a number of levels, including effective prediction of important outputs (degradability, gas flow and composition, and final solids), clarification of the substantial data scatter, prediction of recovery times during operationally poor periods, and cross-validation of the results between digester 1 and digester 2. Key failures in model performance were related to an early incorrect assumption of the COD: VS ratio of 1.42 g g−1, and intermittent high acetate levels, most likely caused by inhibition, and rapid acclimatisation to ammonia. The acute free ammonia limit was found to be 0.008 M NH3–N, while the chronic inhibition constant (KI,NH3,ac) was 0.007±0.001 M NH3–N. Overall, this is a complex system, and application of the model added significant confidence to the initial operational decisions during an aggressive startup on an atypical feed.
Keywords: activated sludge, ADM1, ammonia inhibition, degradability, model, thermal hydrolysis