Practical Comparison of Aggregate Control Algorithms for Demand Response with Residential Thermostatically Controlled Loads

The fast deployment of renewable energy sources requires an increasing amount of frequency regulation reserves in the electric power system. Residential thermostatically controlled loads are ideal candidates to provide such services due to their thermal inertia, large energy buffers and high power ratings. However, the coordinated control of a large number of appliances represents a large-scale control problem with high computational requirements. To alleviate these needs, several aggregate modelling and control strategies have been developed in the recent literature. This paper compares two aggregate models for frequency regulation with electric water heaters used for domestic hot water: a generalized battery model and a set of representative tracer devices obtained with the cross-entropy method. Under realistic simulation assumptions, it is shown that, whereas the generalized battery model better describes the behavior of a group of electric water heaters, the tracers provide a better prediction of the available flexibility for reserves provision.