Application of the linearized ADM1 (LADM) to lab-scale anaerobic digestion system

Anaerobic digestion (AD) has been widely applied for extracting energy from organic wastes. Optimization of AD operation is difficult because of its complexity. Therefore, an effective control of the AD process has become a popular area of research. Anaerobic Digestion Model No. 1 (ADM1) is the most commonly applied model for predicting and controlling the performance of an AD system. However, it requires extensive model calibration/validation steps to improve its accuracy and sensitivity. In this study, ADM1 was modified by linearizing biochemical process equations to develop a linearized ADM1 (LADM) to be implemented in a model predictive control (MPC) for AD systems. Then, both ADM1 and linearized ADM1 (LADM) were firstly applied for a lab-scale AD system and their performances in predicting biogas production of the AD system were compared. The results showed that the LADM had a very similar performance to that of ADM1, as the mean relative error (MRE) of the former was 1.3%. However, the computational cost of the LADM was 18% and 15% lower than that of ADM1, as calculated by using Matlab's two integrators, i.e., ODE45 and ODE15, respectively. Then, the lab-scale AD system was operated by the MPC with LADM implemented. As a result, a very accurate prediction of the biogas production could be obtained. The MRE of the model prediction of biogas production for 80-days AD operation was only 6.9%, indicating that the LADM and MPC scheme developed in this study are promising for application in the process control of an AD system.