Computation and implementation issues for biodiversity simulations with extended anaerobic digestion model No 1 (ADM1_N) in Matlab/Simulink

Abstract

The anaerobic digestion process comprises a whole network of sequential and parallel reactions, of both biochemical and physicochemical nature. Mathematical models, aiming at understanding and optimization of the anaerobic digestion process, describe these reactions in a structured way, the IWA Anaerobic Digestion Model No. 1 (ADM1) being the most well established example. While these models distinguish between different microorganisms involved in different reactions, to our knowledge they all neglect species diversity between organisms with the same function, i.e. performing the same reaction. Nevertheless, available experimental evidence suggests that the structure and properties of a microbial community may be influenced by process operation and on their turn also determine the reactor functioning. In order to adequately describe these phenomena, mathematical models need to consider the underlying microbial diversity. This was demonstrated in one of our previous work by extending the ADM1 to describe microbial diversity between organisms of the same functional group. The resulting model was called ADM1_N. Due to its complexity and stiffness, the implementation of the model is not a simple task and several computational aspects need to be considered. In this paper, the experiences gained from a Matlab/Simulink implementation of ADM1 into the extended ADM1 Model (ADM1_N) are presented. Aspects related to ODE vs. DAE implementations, system stiffness and varying time constants, algebraic solvers for pH and other troublesome state variables, numerical solvers and simulation time are discussed. The resulting model has been compared with the traditional ADM1 in describing experimental data of a pilot-scale hybrid Upflow Anaerobic Sludge Filter Bed (UASFB) reactor. The obtained results show that the extended model improves the fit of experimental data.