Adaptive Demand Limiting Control Using Load Shedding

This paper presents a new algorithm for controlling the peak electrical demand in buildings below a target level through load shedding. The algorithm uses statistical methods to determine the load shedding requirements, without unnecessarily shedding loads. Specifically, a random walk model is used to forecast the uncontrolled electrical demand, and a statistical parameter is used to characterize the forecast errors. The statistical parameter is adjusted in an adaptive manner. Thus, the algorithm is easy to use because it automatically adjusts to a specific building's characteristics. Electrical demand data from a fast food restaurant were used to develop and test the algorithm. The algorithm has successfully been used to limit peak electric demand in a number of buildings.