Abstract
In order to facilitate building energy management, this study developed a method to disaggregate the hourly electricity demand of commercial buildings and classify buildings based on the extracted demand characteristics. The developed method was designed to be applicable to a variety of commercial buildings for which hourly electricity demand data is available. The method decomposes hourly electricity demand data into four components, considering the dependency on operation and meteorological conditions. The demand components representing weekly operating conditions extracted by principal component analysis are used for clustering buildings, along with other demand characteristics representing building’s peak demand and the composition of the four components. In a case study of 4947 commercial buildings, the proposed method showed high estimation accuracy with 66.9% of buildings having a CVRMSE of less than 20% between the sum of the four estimated components and the observed demand. It was also confirmed that the components well characterized the demand for end-uses. The clustering identified seven clusters different in temporal and quantitative characteristics of the four components. The demand characteristics of the clusters fitted well with empirical knowledge on subsectors’ demand characteristics. A building energy management method based on demand characteristics was also proposed.
