The Impact of Uncertain Physical Parameters on HVAC Demand Response

Heating, ventilation and air conditioning (HVAC) units are one of the major resources providing demand response (DR) in residential buildings. A DR program requires a large population of units to make a significant impact on power grid services like peak shaving and balancing. This paper investigates the importance of various HVAC physical parameters and their distributions that affect the aggregate response of a population of units to DR signals. This is a key step to the construction of HVAC models with DR functionality, given insufficient data, to predict the DR capacity available for dispatch. The HVAC model parameters include the size of floors, insulation efficiency, the amount of solid mass in the house, and efficiency. These parameters are usually assumed to follow Gaussian or Uniform distributions over the population. The impact of uncertainty in parameter distributions are quantified through the following steps: 1) Simulate the response of an HVAC population during the transient phase and during steady state for a given DR signal; 2) Use a quasi-Monte Carlo sampling method with linear regression and Prony analysis to evaluate the sensitivity of the DR output to the uncertainty in the parameter distributions; and 3) Identify important parameters based on their impact to the aggregate HVAC response. Utilities or DR providers can use this analysis as guidance in the collection of data to derive an effective DR model.