Heat-related morbidity and mortality is anticipated to increase as climatic change induced overheating become increasingly common. The development of building-specific predictive models has the potential to alert occupants and emergency services to the severity of impending risks. This research aims to evaluate the implementation of a newly developed time series model for overheating prediction. Since risk forecasting is contingent upon the accuracy of the model at different future time steps, the sensitivity of model outputs to the uncertainty in the data inputs needs to be understood. Internal and external climatic variables were monitored in an unoccupied domestic dwelling in order to evaluate the empirical model's predictive accuracy. The uncertainty related to the proximity of external weather stations was evaluated using data taken from four nearby weather stations and further bespoke data sets derived by interpolation. The results confirmed the overall accuracy of the newly developed time series predictive model, whilst highlighting the benefits of climatic data interpolation in reducing predictive uncertainties. The empirically derived modelling approach showed a low variance to the actual temperature evolution over a seven-day predictive period, pointing to its validity as a robust model for the prediction of future overheating risks.
|Title of host publication||Proceedings of the PLEA 2017 in Edinburgh, Scotland, United Kingdom, 3-5 July 2017|
|Publication status||Published - 5 Jul 2017|
|Event||33rd International on Passive and Low Energy Architecture Conference: Design to Thrive: PLEA 2017 - Edinburgh, United Kingdom|
Duration: 3 Jul 2017 → 5 Jul 2017
|Conference||33rd International on Passive and Low Energy Architecture Conference: Design to Thrive|
|Period||3/07/17 → 5/07/17|