Experimental Investigation of Building HVAC Load-Shifting Efficiency

Commercial building Heating, Ventilation, and Air Conditioning (HVAC) systems offer a promising resource for demand response. Strategic sub-hourly adjustments of building temperature setpoints can change HVAC fan power usage to create a virtual battery for the grid without impacting occupant comfort. HVAC fans behave as "inefficient" virtual batteries as there is a discrepancy in energy consumed during a load-shifting event and the counterfactual baseline energy if no event had occurred. Previous results do not agree on the magnitude or direction of additional energy consumption, with differences between experimental and simulation results. In this paper, we leverage a large experimental dataset of load-shifting events to explore the extent of virtual-battery inefficiency. Unlike past experimental analyses, we analyze only energy-neutral events, i.e., events in which the virtual-battery "input" and "output" energy are similar, such that inefficiency is attributed to the building resuming normal operation after the event. Our results show that HVAC load-shifting efficiency is higher than previously reported in experimental analyses, suggesting that past results are influenced by the inclusion of non-energy-neutral events. Additionally, we examine the relationship between event efficiency and several metrics, finding that buildings which shift less power or are able to return to normal operation faster have higher HVAC load-shifting efficiency than buildings with larger power shifts or longer return times. Understanding these trade-offs is important for power systems operators; it enables them to fairly evaluate the cost and constraints of HVAC load shifting as compared to other resources.