Multi-criteria selection of optimum WWTP control setpoints based on microbiology-related failures, effluent quality and operating costs

This work aims at optimising the performance of WWTP with simultaneous biological C, N and P removal by improving its automatic control system. The success of control strategies is highly dependent on the selected operational setpoints of the controlled variables. These setpoints should be optimised to achieve the best effluent quality with the lower operating costs and, at the same time, ensuring an operation with low probability to develop biomass settling problems. Two different objective functions were used to optimise the setpoints of the tested control strategies: a cost function based on the operational costs by converting the effluent quality into monetary units and a multi-criteria function based on the effluent quality, the operational costs and settling problems of microbiological origin. Three different control strategies were evaluated: (i) open loop; (ii) optimised fixed setpoints for the ammonium and nitrate control loops and (iii) daily optimised setpoints. The multi-criteria optimisation resulted in a set of optimal setpoints with a Pareto distribution. This multi-criteria function was developed so that any of the criteria (effluent quality operating costs or microbiological risks) was conditional to another and thus, optimal scenarios with a high risk of failure due to microbiological issues could be rejected. Moreover, it was concluded that the optimisation process could be enhanced by using both objective functions in a complementary way. While the multi-criteria function enabled a more extensive evaluation of the different alternatives, once the weights are selected the operational costs function optimisation could be used to define an optimum set of setpoints to adapt the system to influent variations.

► We included microbial-related failures and EBPR when optimising WWTP control strategies.
► The multi-criteria function enabled a better evaluation of WWTP control alternatives.
► The setpoints from multi-criteria optimisation were approximated to a Pareto surface.