A new algorithm for predicting overheating in predominantly naturally ventilated houses using dynamic thermal models

Dynamic thermal models are used to predict indoor temperatures for overheating building regulation compliance assessments. Models must reliably predict overheating to avoid the negative health implications that arise from exposure to high temperatures. Previous work, however, identified differences between measured and modelled overheating – particularly the speed at which the modelled temperatures changed relative to the measured. These differences were apparent and almost identical irrespective of the dynamic modelling simulation (DSM) software package used. Even extreme changes to the model set-up, such as unreasonably high levels of thermal mass, could not align the model predictions with the measured data, suggesting that the modelling algorithms required review and revision. This paper presents the new formulation of the modelling algorithms with a focus on the computational fluid dynamics (CFD) turbulence models used to simulate air movement and therefore heat transfer. These findings may improve the reliability of Part O building regulations compliance assessments and could require changes to the processes for analysing overheating in the wider building simulation community.