Office representatives for cost-optimal energy retrofitting analysis: A novel approach using cluster analysis of energy performance certificate databases

• Published in: Energy and buildings 2020-01, Vol.206, p.109557, Article 109557
• Main Authors: Gangolells, Marta, Casals, Miquel, Ferré-Bigorra, Jaume, Forcada, Núria, Macarulla, Marcel, Gaspar, Kàtia, Tejedor, Blanca
• Format: Article
• Language: English
• Subjects: Algorithms; Buildings; Centroids; Certificates; Cluster analysis; Clustering; Consum d'energia; Cost analysis; Data bases; Decision making; Edificació; Edificis d'oficines; Energy; Energy conservation; Energy consumption; Energy performance certificates; Energy policy; Environmental aspects; Housing; Industrial buildings; k-means; Office buildings; Offices; Reference buildings; Refrigerators; Residential areas; Residential buildings; Retrofitting; Vector quantization; Àrees temàtiques de la UPC
• ISSN: 0378-7788; 1872-6178
• DOI: 10.1016/j.enbuild.2019.109557

•A replicable methodology for identifying reference buildings is presented.•The k-means clustering method is applied to the energy performance certificate database.•To illustrate the methodology, an EPC database including 13,701 offices is used.•Sixteen clusters and corresponding representative offices are identified.•Representative offices can be used for cost-optimal retrofitting analysis of the entire stock. A large number of buildings must be evaluated to formulate energy retrofitting policies for existing building stock. In this context, it is crucial to identify reference buildings that can effectively represent the entire stock, since such buildings can then be used to assess the individualized cost-effectiveness of retrofitting measures. This paper presents a novel approach for identifying and defining a set of reference buildings by applying the k-means clustering method to energy performance certificate databases. To this end, a four-step methodology has been envisaged. First, an energy performance certificate database is prepared and variables that have an impact on energy consumption are pre-selected. Selected data are then pre-processed. Next, the k-means clustering method is applied. Finally, the resulting cluster centroids are used to identify the closest energy performance certificates in the database, in other words, the representative buildings that will then be used for cost-optimal retrofitting analysis. The methodology is illustrated using the energy performance certificate database managed by the Catalan Institute of Energy (ICAEN), which includes a sample of 13,701 offices. Due to the large number of missing values in the database, the k-means clustering algorithm was finally performed over 6,083 energy performance certificates. Seven representative office blocks and offices in industrial buildings and nine representative offices in residential buildings were identified. The results establish the basis for supporting strategic decision-making for energy saving retrofit interventions in existing Spanish offices.