本研究對象係針對台灣地區各政府機關單位面積耗電量,進行影響耗能因子分析及建立耗能指標。藉由財團法人台灣綠色生產力基金會所提供的資料,分析影響耗能之各種因素探討及建立機關單位面積耗電量與影響因子間的關係。 本研究主要目的在探討九十八年各地方政府機關能之戶政事務所能源使用效率的情況,透過迴歸分析法得到具有顯著關係的因子DH(冷房度時)及員工數/m2,並利用集群分析法將具有顯著關係的因子DH(冷房度時)及員工數/m2進行分群,得到低、中、高耗能密度群,第一群位於不炎熱的氣候,且人口密集度較低的地區為低耗能密度群約佔13%;第二群位於較炎熱的氣候,且人口密集度較低的地區為中耗能密度群約佔65%;第三群位於較炎熱的氣候,而且人口密集度較高的地區為高耗能密度群約佔22%。 後續研究方面,本研究採用多元迴歸分析法以及集群分析法,分析的結果可提供政府機關績效評比的參考,而本文分析的結果有待實際測量來驗證,並找出造成能源績效使用不良的原因。
The purpose of the study is to analyze the factors affecting energy consumption and establish energy criteria based on power consumption per unit area of government agencies in Taiwan. According to the information provided by Taiwan Green Productivity Foundation, this study analyzes the factors affecting energy consumption and established the relationship between power consumption per unit area of institutions and impact factors. This study aims to investigate the energy efficiency of the census administration office in local government authorities in 2009. Through Multiple Regression analysis, significant factor has been found. That is the relationship between DH (Cooling Degree Day) and Number of employees per unit area. And using Cluster analysis of factors will have a significant relationship between DH (Cooling Degree Day) and Number of employees per unit area. They can be divided into the low, medium and high energy density group. The first group located in a mild climate and areas with low population density for the low energy density is about 13%. The second group located in the hot climate and areas with low population density for the medium energy density is about 65%. The third group located in the hot weather and areas of high population density for high energy density group is about 22%. For the following research, the results of the analysis can provide government agencies for reference by multiple regression analysis and cluster analysis. The results of this analysis need to be verified by the actual measurement. And the following research can find out the main cause of poor energy performance.