Modeling and Control of Aggregated Heterogeneous Thermostatically Controlled Loads for Ancillary Services

This paper presents a novel modeling and con- trol approach for the aggregation of large numbers of hetero- geneous thermostatically controlled loads, such as refrigera- tors, electric water heaters, and air conditioners, and their usage for Demand Response. Unlike traditional Demand Re- sponse methods that act on time scales of hours, this ap- proach is able to provide short-term (e.g., second-to-second) ancillary services, such as balancing and frequency control. A statistical modeling approach based on Markov Chains is used to describe the evolution of probability mass in a tem- perature state space. The Markov state transition matrix is identified using state information from the population of thermostatically controlled loads. A predictive controller is used to control the aggregate population of loads such that it tracks a signal. A simulation example shows the applicabil- ity of the approach to realistic systems, and includes a com- parison of control performance depending on available state information and controller parameterization.