Assessing the risk of climate change for buildings: A comparison between multi-year and probabilistic reference year simulations

Given a changing climate, there is a need to provide data for future years so that practicing engineers can investigate the impact of climate change on particular designs and examine any risk the client might be exposed to. In addition, such files are of use to building scientists in developing generic solutions to problems such as elevated internal temperatures and poor thermal comfort. With the release of the UK Climate Projections (UKCP09) [1], and the publication of a methodology for the creation of probabilistic future reference years using the UKCP09 weather generator [2], it is possible to model future building performance. However, the collapse of the distribution of possibilities inherent in the UKCP09 method into a single reference year or a small number of reference years, potentially means the loss of most of the information about the potential range of the response of the building and of the risk occupants might be subject to. In this paper we model for the first time the internal conditions and energy use of a building with all 3000 example years produced by the UKCP09 weather generator in an attempt to study the full range of response and risk. The resultant histograms and cumulative distribution functions are then used to examine whether single reference years can be used to answer questions about response and risk under a changing climate, or whether a more probabilistic approach is unavoidable.