Predictive control of urban wastewater systems
Within recent years, technological advances and stricter regulatory requirements have seen the increased use of automation and instrumentation within the wastewater treatment industry. As a result, advanced control strategies are required, to fully exploit the potential of these complex systems in addressing water quality concerns. Model based control strategies can be appropriate within the multivariable constrained wastewater system. In particular, the inherent model based nature of this approach can be valuable in the prediction of the treatment plant effluent quality required over a considered time period, to meet water quality standards. Multivariable linear predictive control is implemented for a benchmark treatment plant model, demonstrating the constraint handling ability of the predictive control structure. The limitations of an effluent-based control strategy in the maintenance of river quality is discussed. A more global approach to wastewater control must be considered in order to compensate against disturbances within the system. Tackling this concern, the incorporation of receiving water quality objectives within the control strategy is proposed. To this end, the application of linear MPC to the control of dissolved oxygen concentrations in the receiving waters under storm conditions is demonstrated. The drawbacks involved in a linear model based approach within a nonlinear urban wastewater system are considered. Several nonlinearities are present: the bioprocesses involved are by definition nonlinear, and are affected by varying wastewater load and characteristics. These can be the result of varying stormwater effects upon the treatment plant or emergency overflows to receiving waters. This therefore motivates the development of nonlinear strategies in the control of the wastewater processes. The control of SISO nonlinear processes within the urban wastewater system, such as dissolved oxygen, is demonstrated via the use of fuzzy gain-scheduled and Wiener model based predictive control. Additionally, the use of existing nonlinear process models in the control of wastewater processes is shown in the application of state dependent model predictive control.