Optimal Operation of a Wastewater Pumping Station by Model-based Predictive Control

Abstract

In this thesis, Model-based Predictive Controller for a midscale wastewater pumping station is developed. We used MATLAB Simscape blocks to build a full detailed model in a Simulink environment. The system consists of five pumps fed by a buffer tank and deliver sewage to a treatment plant through a pipeline. This model is run many times to simulate, check and validate the behavior of the system. A recursive auto-adaptive ARX model is developed for inflow rate prediction. The predictor is separated into two independent models, one for dry weather and the other for run-off weather. Stochastic data is generated to check the smartness of the predictor and an accepted result is achieved. To optimize the operation plan of the system, the detailed model is used to design a preliminary control strategy which is used later to set the optimization problem. North Gaza Terminal Pumping Station is taken as a case study. Fixed speed pump scheduling is found more efficient. A GA optimizer is used to decide the operation plan of the pumps taking into account power, maintenance, and sewage volume in the tank as an environmental factor with different weights. Upper and lower limits of the water level in the tank is constrained by the GA. The optimizer is tested and gave explainable results. Finally, the MPC is tested with 48 hour time horizon and an accepted operation plan is resulted. Detailed plots and charts are generated using MATLAB to explain what is going on through all levels of work in this thesis.