To achieve carbon neutrality in a city or country, it is required to evaluate energy performance and energy conservation measures for large buildings. The energy performance of existing buildings are widely evaluated by annual EUI (energy use intensity, kWh/㎡yr). However, this annual value have limitations on analyzing seasonal effect and establishing energy conservation strategies. In this paper, we analyze monthly energy use patterns of large buildings and proposed general energy conservation strategies and measures according to the patterns. To classify the energy use patterns, we investigated clustering techniques on monthly energy use of office buildings in Korea. A k-means algorithm was implemented, and two different methods were compared: feature based k-means and time series k-means. The methods were performed with Euclidean distance metric and we tested our methods on energy use data from national database. The results show that feature based k-means method is significant in energy use pattern analysis. The energy use patterns of office buildings were divided into five clusters. We analyzed the characteristics of clusters by building size, annual and seasonal energy use.
