Paper Conference

Proceedings of uSim Conference 2022: 3rd uSim Conference of IBPSA-Scotland



Giulio Tonellato, Michaël Kummert

Abstract: Building stock simulation has evolved from attempting to reproduce yearly energy use to accurately representing the dynamic behaviour of the stock and its impact on peak electricity demand. As all building performance analyses, simulating the performance of a stock requires using weather data files, and “typical meteorological years” (TMY, or their variants TRY for typical reference year) are often used to obtain a synthetic data file representing an “average” year. The different methods used to obtain representative data files (EN ISO 15927-4:2005, TMY2 and TMY3 data files in the USA, or CWEC files in Canada) have been developed with individual weather stations in mind. Representative months are individually selected for each weather station in order to represent a typical month (Morris, 2016; Pernigotto et al., 2014; Wilcox and Marion, 2008). Using these typical weather files to simulate the performance of a building stock across a large region (e.g. country, state or province) would not allow to investigate the overall dynamic behaviour of the stock, such as its impact on the peak electricity demand, because the selected months would not necessarily be consistent across the different weather stations. This paper presents a methodology inspired by Huang (2020) to obtain population-weighted typical weather files which results in selecting the same reference months for a large set of weather stations, therefore allowing to model an entire building stock across a large geographical area with consistent weather patterns. The methodology is applied to Canada using the CWEEDS dataset provided by Environment Canada (ECC Canada, 2020).