Trading energy yield for frequency regulation: Optimal control of kinetic energy in wind farms

Abstract

The burden on conventional units to regulate the system frequency increases if they are replaced due to wind farms. This paper explores up to which time scales the rotating kinetic energy in wind turbines can smooth frequency variations and assist with the regulation task. To this end, a comparison is made between a standard wind turbine controller and optimal control of wind turbines, respectively derived from causal time-domain simulations and an optimization algorithm that allows predicting. The latter algorithm is used to give a benchmark for the smoothing potential, shown by plotting the Pareto efficiency of the normalized standard deviation of frequency variability versus a normalized measure of the energy yield. Results indicate that smoothing comes with an energy loss that is determined by the energy content of power imbalances. It is shown that a wind share of 20%, within the instantaneous generation mix, can absorb frequency variations on timescales up to 100 sec while the energy loss is limited to only 2%. A higher share of wind power aggravates frequency variability. Nevertheless, in such circumstances the potential of rotating kinetic energy in wind farms increases.