발표자  :  So-I Seok

논문 제목  :  A comparative study of building energy performance evaluation in existing buildings using change-point model and k-means cluster

Abstract  :

   In the case of large-scale energy remodeling for existing buildings, it is important to identify and prioritize buildings with the lowest energy performance. The current criteria for classifying aging buildings with low energy performance, based on the completion year, is inadequate to determine the target of energy remodeling. Inverse model is currently widely used to evaluate the energy performance of existing buildings. Among various approaches of inverse model, the change point model assesses the energy performance of a building by analyzing the relationship between the actual energy usage of the building and the outdoor weather conditions. However, one of this method’s shortcomings is its lack of flexibility as it is only applicable to buildings with a linear relationship between outdoor temperature and building energy consumption. Therefore, another method is needed for energy performance evaluation the existing buildings. The purpose of this study is to classify aging buildings through change-point model, the commonly applied approach, and the k-means cluster in order to explore the applicability of cluster analysis for building energy performance evaluation. The target building is selected as apartment, a building type that accounts for a large proportion of the existing building located in Ulsan Metropolitan City, South Korea. The selected apartments were, constructed between 1988 and 2008. In total, 244 apartments are selected and their energy performance are evaluated based on monthly gas consumption of the year 2014. According to the results, 72 apartments are identified as aging buildings by the change-point model and k-means cluster. Among the identified aging apartment, 57 (65.5%) apartments are identified by both methods. In conclusion, it confirmed that possibility of introducing a new methodology applying k-means cluster analysis, that can improve the shortcomings of change-point model.