Journal of Process Control, Vol.95, 55-66, 2020
Combination of cascade and feed-forward constrained control for stable partial nitritation with biomass retention
Ammonium removal is a key step in wastewater treatment which can be accomplished biologically. An interesting process option for this purpose is coupling partial nitritation with the Anammox process. The goal of the partial nitritation process is to convert half of the ammonium in the influent stream into nitrite, so both can be later converted into dinitrogen gas by the Anammox reaction. To obtain a stable partial nitritation, ammonium oxidizing bacteria (AOB) have to prevail over nitrite oxidizing bacteria (NOB) so as to avoid further conversion of nitrite into nitrate. The dissolved oxygen concentration is a key variable for the functional group selection. In this study, a constrained combination of cascade and feedforward control is proposed for reactors with biomass retention, aimed at suppressing unwanted NOB while keeping a nitrite:ammonium ratio suitable for coupling with Anammox. The master controller, aimed to regulate this effluent ratio, generates the set-point for the dissolved oxygen concentration slave controller. In addition to the cascade controller feedback loop, a feed-forward controller calculates the optimal dissolved oxygen concentration based on the current influent stream flow rate and concentrations. The resulting dissolved oxygen concentration set-point is compared to constraints that guarantee the suppression of NOB and survival of AOB. The proposed control strategy is simple to apply in common wastewater treatment plants with biomass retention. A sensitivity analysis is performed to assess the effect of model parameters uncertainty on the controller constraints and to determine which parameters need to be identified with more precision to avoid instability or poor results. The performance and the effect of the uncertainty of the most sensitive parameters on the proposed control algorithm are assessed through simulation using realistic streams as inputs of the process. (C) 2020 Elsevier Ltd. All rights reserved.