Classification and energy analysis of bank building stock: A case study in Curitiba, Brazil

An engineering bottom-up technique has been selected for classification and energy analysis of 72 bank branches in the Brazilian city of Curitiba. A multi-stage approach has been adopted in this study to collect and analyse the dataset of these bank branches, virtually survey and map geometrical and design parameters of the branches and classify them according to shapes, sizes, geographical layout, and construction age. Four types (Type A to D) of bank branches have been identified. Type A branches have the greatest mean conditioned floor area (over 100 m(2)) with largest deviation of energy use intensity (over 47 kWh/m(2)year). Type B branches are most commonly found in densely populated downtown area. However, over 61% of the branches belong to Type C and D, which are mostly located in less congested sub-urban areas. The study also shows that post-2010s branches are among the largest energy consumers. An in-depth parametric assessment on specific design details from architectural drawings of selected branches was further conducted to identify the effects of building parameters on energy consumption of the bank branches. Despite the minute correlation obtained between energy consumption and design variables (R-2 of 0.253), a Linear Discriminant Analysis suggests that percentage of openings and occupant density are among the defining parameters of a low and medium-high consumer as regards bank branches in Brazil. This study provides insights into a better understanding of the design characteristics of bank branches in Curitiba and parameters potentially affecting their energy consumption. The archetype bottom-up technique presented in this study that involved studies at different stages can also be used to define typologies for other types of buildings. This study shows how complex building stock, such as, bank buildings could be classified and their energy performance collectively analysed using multi-stage approach. The approach combined a range of methods, which had been conducted in sequence, such as, virtual survey, building classification, further in-depth parametric and energy assessments of selected bank branches based on architectural drawings using Multiple Regression Analysis and Linear Discriminant Analysis. Such approach enables critical design parameters to be determined in the absence of complete set of building data, which could be adopted for other similar types of complex building stock.