발표자  :  Irakoze Amina

논문 제목  :  Classifiying buildings of various thermal performance through cluster analysis and energy consumption of different time unit

Abstract  :

  For an effective energy upgradation on a large scale of existing building stock, it is important to identify and prioritize building group for energy retrofit. Research has proven that the application of energy use intensity (EUI) approach to benchmark and identify energy inefficient buildings tend to oversimplify the complexity of building energy consumption; and hence give fuzzy building classifications. The purpose of this study is to classify buildings with similar envelop thermal characteristics by applying cluster analysis on their energy consumptions. Considering the difficulty in obtaining fine-grained building energy data, the applicability of the cluster analysis classification is assessed for three different building energy granularities: namely hourly, daily, and monthly heating energy consumption. For the purpose of this study, heating energy consumption of 51 buildings of different thermal characteristics is predicted through EnergyPlus/DesignBuilder energy simulation tools. The analyzed buildings have varying wall, window, and roof thermal transmittance, window solar heat gain coefficient (SHGC), and infiltration rate (ac/h). The results indicate that although k-means cluster analysis can be used to classify buildings of similar characteristics, the obtained building categories are more precise when cluster analysis is carried out based on daily and monthly heating energy consumption compared to hourly energy use. With monthly heating energy consumption, clustering algorithm is able to identify energy inefficient buildings that fall short of ASHRAE’s recommendations for thermal performance of building envelop.