Modelling and control of an N-removing activated sludge process

In this paper an approach to obtain a control strategy for optimal N-removal in alternatingly aerated, continuously mixed, continuously fed activated sludge processes (ASP's) using adaptive Receding Horizon Optimal Control (RHOC) is presented. This control strategy is successfully tested both in simulation and pilot plant experiments. The RHOC approach offers an excellent opportunity to link the higher level of plant economy and the lower level of plant control by expressing the plant economy in the RHOC's cost criterion, including constraints on in- and outputs. Essential for the performance of RHOC controllers is the availability of accurate predictions of near-future NH4 and NOx concentrations. A recursive estimator for the model parameters of a grey-box model identified from experimental data is designed to keep the model close to the real process behaviour