IDENTIFICATION OF WASTE-WATER TREATMENT PROCESSES FOR NUTRIENT REMOVAL ON A FULL-SCALE WWTP BY STATISTICAL-METHODS

The introduction of on-line sensors of nutrient salt concentrations on wastewater treatment plants opens a wide new area of modelling wastewater processes. Time series models of these processes are very useful for gaining insight in real time operation of wastewater treatment systems which deal with variable influent flows and pollution loads. In this paper nonlinear time series models describing the variations of the ammonia and nitrate concentrations in the aeration tanks of a biological nutrient removal WWTP are established. The models proposed herein are identified by combining well-known theory of the processes, i.e. including prior knowledge, with the significant effects found in data by using statistical identification methods. Rates of the biochemical and hydraulic processes are identified by statistical methods and the related constants for the biochemical processes are estimated assuming Monod kinetics. The models only include those hydraulic and kinetic parameters, which have shown to be significant in a statistical sense, and hence they can be quantified. The application potential of these models is on-line control, because the present state of the plant is given by the variables of the models which are continuously updated as new information from the on-line sensors becomes available.