Dynamical modelling and state estimation of anaerobic wastewater treatment plants

A mass balance based model has been derived to represent the dynamical behaviour of two anaerobic digester (one pilot-scale, one industrial-scale). Its parameters have been calibrated for both reactors on the basis of experimental data and following a predefined design of experiments. This mass balance model constitutes the basis for software sensor design that takes advantage of the partial information on the process dynamics and uses on-line gas and alkalinity measurement. It computes the concentration of inorganic carbon and volatile fatty acids (VFA) in the digester. Simulations show the theoretical efficiency of this software sensor for realistic variations of the organic loading rate. The predictions of this software sensor when it is applied to experimental data are very close to the actual off-line measurements. The sensor predicts the accumulation of VFA, and thus very early detects a destabilization of the digester due to an overloading. The presented methodology demonstrates the usefulness of advanced monitoring techniques for a better understanding of the internal working of the process.