Using principal component and cluster analysis in the heating evaluation of the school building sector

In the field of energy savings in buildings, the interest towards the school sector is deeply motivated: schools have standard energy demands and high levels of environmental comforts should be guaranteed. The University of Athens in collaboration with the School Authority of Greece undertook a complete program on energy classification and environmental quality of school buildings. Data on energy consumptions were gathered and analysed with the participation of 1100 schools from all the prefectures of Greece. The data have been provided by the school authority of the country (OSK), in collaboration with the management of each school building. With regards to the size of the building and the external climate variability (HDD-method) energy normalization techniques have been applied in order to homogenize the data set. An energy classification tool has been created through clustering techniques, using the collected data regarding the heating energy consumption and as a result five energy classes have been defined. To evaluate the potential energy conservation for each class, the typical characteristics of school buildings belonging to an energy class have to be identified. A new methodology based on the use of the principal components analysis (PCA) has been developed. The method allows to define in an accurate way the typical building of each energy class and thus to perform analysis on the potential energy savings for the specific group of school buildings. By reducing the dimensionality of the problem, a bi-dimensional graphic in the first two PCs coordinate system promotes the understanding of the correlation between the examined variables, as well as the determination of sub-groups of school buildings with similar characteristics. The typical school of seven variables sample is defined as the closest to the medians in the principal components' coordinate system. (C) 2009 Elsevier Ltd. All rights reserved.