Real-time optimisation of a wastewater treatment network

The aim of this master thesis is to present a control strategy to optimise the partitionning of the sewage between the wastewater treatment plants of the SIAAP (Syndicat Interdépartemental pour l'Assainissement de l'Agglomération Parisienne) network. The problem has been addressed by the use of Model Predective Control (MPC) law. A simple model of the west part of the Seine bassin has been developped for this purpose. Evaluation on real data demonstrated the model was effective for reproducing the dynamic of total suspended solids, biological oxygen demand, ammonia nitrogen, nitrites and nitrates with root mean square error below 0.14. A systematic approach has been proposed for the parameters tuning of the MPC and evaluated using fuzzy methodology. The control law was tested against real measurements and showed an increase of 15% of performance for the ammonia nitrogen and the nitrites in dry conditions. In wet conditions, the control law showed slightly lower performance compared to real measurement but still address water in good state with a margin of 18%. The developed control law showed that the actual partitionning was not optimal and that MPC law seems a promising solution to optimise the partionning and improve the water quality.