Predictions of summertime overheating: comparison of dynamic thermal models and measurements in synthetically occupied test houses
journal contributionposted on 20.05.2019, 13:32 authored by Ben M RobertsBen M Roberts, David AllinsonDavid Allinson, Susie Diamond, Ben Abel, Claire Das Bhaumik, Narguess Khatami, Kevin LomasKevin Lomas
Summertime overheating in UK dwellings is seen as a risk to occupants' health and well-being. Dynamic thermal simulation programs are widely used to assess the overheating risk in new homes, but how accurate are the predictions? Results from two different dynamic thermal simulation programs used by four different experienced modellers are compared with measurements from a pair of traditional, semi-detached test houses. The synthetic occupancy in the test houses replicated curtain operation and the CIBSE TM59 internal heat gain profiles and internal door opening profiles. In one house, the windows were always closed and in the other they operated following the TM59 protocol. Sensors monitored the internal temperatures in five rooms and the local weather during a 21-day period in the summer of 2017. Model evaluation took place in two phases: blind and open. In the blind phase, modellers received information about the houses, the occupancy profiles and the weather conditions. In the open phase, modellers received the test house temperature measurements and, with the other modellers, adjusted their models to try and improve predictions. The data provided to modellers is openly available as supplementary information to this paper. In both phases, during warm weather, the models consistently predicted higher peak temperatures and larger diurnal swings than were measured. The models' predicted hours of overheating were compared with the measured hours using the CIBSE static threshold of 26℃ for bedrooms and the BSEN15251 Category II threshold for living rooms. The models developed in each phase were also used to predict the annual hours of overheating using the CIBSE TM59 procedure. The inter-model variation was quantified as the Simulation Resolution. For these houses, the blind phase models produced Simulation Resolution values of approximately 3% ± 3 percentage points for TM59 Criterion A and 1% ± 1 percentage point for TM59 Criterion B. The Simulation Resolution concept offers a valuable aid to modellers when assessing the compliance of dwellings with the TM59 overheating criteria. Further work to produce Simulation Resolution values for different dwelling archetypes and weather conditions is recommended.
London-Loughborough Centre for Doctoral Research in Energy Demand (grant EP/L01517X/1).
- Architecture, Building and Civil Engineering