Dynamic Multiobjective Global Optimization of a Waste Water Treatment Plant for Nitrogen Removal

This paper deals with the dynamic optimization of a wastewater treatment plant model for nitrogen removal. Two process variables (i.e., aeration factor in a tank and internal recycle flow rate) are selected as control variables and their time profiles are approximated using the control vector parameterization (CVP) technique. Two conflicting objectives have been considered to formulate a multiobjective optimization problem, namely the quality of the effluent and the plant economy. To solve the multiobjective problem and find the Pareto front the epsilon-constraint technique has been considered. In order to solve this complex problem and to prove its multimodality, a multistart procedure and the scatter search metaheuristic have been applied, showing that the problem is indeed multimodal. Results reveal that this approach is successful to find optimal operation policies in this type of plants.