Translating probabilistic climate predictions for use in building simulation

With the onset of global climate change it is necessary, when designingbuildings, to understand how a change in ambient conditions might affect theperformance of a building. This requires the integration of future climate predictionswith appropriate building simulation software. Previous approaches to this have useddeterministic predictions of climate, essentially design years that provide a singleprediction for what a future year might look like. The UK Climate Projections ’09(UKCP’09) have presented predictions in a probabilistic format – through multipleiterations of climate models, a large dataset of climate predictions can be constructedfor specific scenarios, where each climate has a certain probability of occurrence. TheLow Carbon Futures project described by this paper will suggest a methodology fortranslating this information into a building context. Not only will this provide a rangeof building performance predictions relating to a range of climate predictions, but alsothe effect of this predicted climate change on the “failure” of that building at sometime in the future can be discussed. Such a failure may involve overheating in anaturally ventilated building, or a cooling plant becoming insufficiently sized.Through the simulation of multiple climates for specific buildings (accounting foroccupancy, location, operation and construction), the project will demonstrate the linkbetween specific climate metrics and overheating criteria, and the probability of thisoccurring. The project will use this methodology to discuss the failure probability ofbuildings including naturally ventilated offices, dwellings without mechanical coolingand also schools, though this paper will focus on overheating in the domestic sector.Where simulation techniques identify a potential issue with overheating, adaptationscenarios will be suggested (and simulated) to look at the possibilities of “futureproofing”these buildings against climate change. The methodology relies on applyingestablished statistical methods to the climate data, to allow this information to beintegrated into ESP-r dynamic building software and to develop simplified statisticalmodels capable of estimating building performance, that closely approximate thepredictions of ESP-r, in relation to climate data.