Model Predictive Control for water management of a two-reservoir hydroelectric power plant

In the present work, model predictive control techniques are applied to the water management of a two-reservoir hydropower plant. The availability of forecasts on water requests by the hydroelectric plant was assumed and this motivated the adoption of a model predictive control approach. Ad hoc model mismatch compensation strategies has been developed which takes into account for unknown and unmeasured water flows and model uncertainties. The MPC strategy has been designed based on first principles linear model equations and on assigned control specifications. MPC parameter tuning is suitably adapted to the occurrence of specific state conditions, thus enabling optimal management of conditions of possible reservoir overflow, limiting water wastage while ensuring prompt reaction to possible variation in water demand related to the hydropower production plan. The reliability of the proposed approach has been tested through tailored simulation experiments, taking into account different significant scenarios.