CONTROL OF NUTRIENT REMOVING ACTIVATED SLUDGE SYSTEM

The objective of this paper is to evaluate several control algorithms for nitrogen removal using a simulation benchmark of a pre-denitrifying activated sludge process. Various PI and feedforward controllers are evaluated and compared with advanced multivariable and nonlinear model predictive control, which uses a perfect process model, perfect measurements and perfect knowledge of the disturbances. The simulation results indicate that PI nitrate and feedforward-PI ammonia control closely imitates the optimal operation strategy as the operating costs are only slightly higher compared to the case when model predictive control is applied. It appears that the improvement is thus more related to control structure, i.e. where the sensors and the actuators are located, than to the control algorithms.