Model Predictive Control for Enhancing Wind Farms Participation in Ancillary Services

The increasing penetration of Renewable Energy (RE) systems into the electric grid is creating new challenges into the power system. The unpredictable and variable nature of renewable power generation is increasing the imbalances between generation and demand. For this reason, wind farms, which are the main source of RE in Europe, are required nowadays to support the grid, providing services of voltage and frequency regulation. To be able to increase their power production during a frequency event, Wind Power Plants (WPPs) need to work below their maximum generation capacity, keeping an additional amount of power, called power reserve, that can be injected into the grid when required. The power reserve of a wind farm strongly depends on the interaction among the wind turbines. The wake effect produced by the upstreams turbines affects the wind condition that each turbine faces and reduces their maximum available power. This study aims to present the effects of different distribution of the Wind turbines (WTs) individual power contribution on the power reserve. Three control strategies, based on Model Predictive Control (MPC), are tested on a fifteen turbines wind farm under different wind conditions. Simulation results show that, in almost all cases, prioritizing the power contribution of the most downstream turbines and deloading the upstream ones, leads to a maximization of the wind farm power reserve. Furthermore, an additional MPC strategy aiming to combine active and reactive power control, for providing both frequency and voltage regulation at the Point of Common Coupling (PCC), is presented. The advantage of a combined active and reactive power control is the possibility of improve the voltage support capability of the WPPs, by controlling the active power set-points. The MPC is also tested on a fifteen turbines wind farm, in order to validate the performances of the controller while solving the multi-objective problem. The ability of the controller to handle simultaneously the different requirements is proven