Model Predictive Control of BSM1 benchmark of wastewater treatment process: A tuning procedure

The BSM1 (Benchmark Simulation Model N° 1) has become the standard simulation tool for performance assessment of control techniques applied to wastewater treatment plants (WWTP). Control of WWTP is not trivial because of large disturbances in the influent, nonlinearities, delays, and interaction between variables. In this work a multivariable Model Predictive Controller (MPC) is implemented and optimally tuned. The paper presents results for different tuning situations without the use of long time consuming simulations and considering uncertainty by means of multiple linearized models. Control of the dissolved oxygen in the last aerated reactor and nitrate level in the last anoxic compartment has been tested in simulations using the MPC control strategy. The obtained results show the MPC benefits when it is properly tuned.