A Data-Driven Approach for Providing Frequency Regulation with Aggregated Residential HVAC Units

The growing level of integration of renewable energy sources into power grids around the world has increased the need for cheap, reliable and effective fast frequency regulation resources. On that note, most grid operators are already considering the usage of demand-side resources for frequency regulation purposes. A major candidate for such initiatives is the heating, ventilation and air conditioning (HVAC) unit in both residential and commercial buildings. In this paper, we present a simplified data-driven approach for providing frequency regulation using aggregated residential HVAC units. The approach involves the use of a residential load model to generate data for frequency regulation model identification. Afterward, a relationship between instantaneous aggregated HVAC power consumption, HVAC power changes and thermostat setpoint offsets is established using a simple multiple linear regression model. Actual regulation qualification signals from the PJM market and the regression model are then used to evaluate the ability of the units to satisfactorily respond to frequency regulation signals. The obtained results show that while the results are satisfactory using PJM's performance metrics, improvements can still be made by accounting for model errors.