Multivariate time series analysis for design and operation of a biological wastewater treatment plant

In the area of biological wastewater treatment, the more common approach to dynamical modeling has been the use of mechanistic models with differential equations. This study illustrates that multivariate time series analysis can be a valid alternative. A lab-scale activated sludge unit, treating brewery wastewater, was operated with a highly variable COD-load. For design and operation purposes, we wanted to model the behaviour of biomass and effluent-quality. Because of the dynamic nature of the system, it was not possible to use steady-state relations. With time series analysis, we succeeded in modeling the dynamic behaviour, using only daily routine measurements. Simple equations, describing biomass, effluent COD and effluent suspended solids are obtained and combined to form one system of difference equations. The use of this system is basically equivalent to the use of models consisting of differential equations. Thus by means of simulations, design and operation alternatives can be evaluated.