Investigation and development of methods for optimal control of the activated sludge process

Abstract

This project was started as a result of strict environmental and health regulations together with a demand tor cost effective operation of wastewater treatment plants (VVWTPs). The main aim of this project is how to keep effluent concentration below a prescribed limit at the lowest possible cost. Due to large fluctuations in the quality and quantity of the influent concentrations, traditional control methods are not adequate to achieve this aim The major drawback with these methods is that the disturbances affect the process before the controller has time to correct the error (Olsson and Newell, 1999: 454). This problem is addressed through the use of modern control systems.

Modern control systems are model based predictive algorithms arranged as feed-forward controllers (Olsson and Newell. 1999: 454). Normally a controller is equipped with a constant set point; the goal In this project is to calculate an optimal DO trajectory that may be sampled to provide a varying optimal set-point for the Activated Sludge Process, In this project an optimal control problem Is formulated using DO concentration as a control variable. This requires a model of the process to be controlled a mathematical expressions of the limitations on the process input and output variables and finally the objective functional. which consists of the objectives of the control.

The structures of the Benchmark plant (developed within the COST 682 working group) and the Athlone WWTPs are used to implement this opt.mat control strategy in MATLAB. The plant's full models are developed based on the mass balance principle incorporating the activated sludge biological models: ,ASM1, ASM2, ASM2d and ASM3 (developed by the IWA working groups). To be able to develop a method that may later on be used for online control, the full models are reduced based on the technique In Lukasse (1996). To ensure that the reduced models keep the same prediction capabilities as the full models, parameters of the reduced models are calculated based on the Least Squares principle, The formulated optimal control problem is solved based on the decompostion-coorcdination method that involves time decomposition in a two layer structure.

MATLAB software [5 developed to solve the problems for parameter estimation. fun and reduced mode! simulation. and optimal control calculation for the considered different cases of plant structures and biological models. The obtained optimal 00 trajectories produced the effluent state trajectories within prescribed requirements. These DO trajectories may be implemented in different SCADA systems to be tracked as set points or desired trajectories by different types of controllers.