The summertime overheating signature of UK homes

The summertime overheating of homes across the UK is a growing concern. Although there is no absolute definition of what constitutes overheating, it generally refers to raised temperature within a home that will impact a person’s thermal comfort or health. Previous research has revealed that one-fifth of English homes are subject to overheating.

Knowing the relationship between indoor and outdoor temperatures in different dwellings would help classify whether a home presents a lower or higher risk of weather-driven overheating for its occupants than the population average. This would help with better-targeted heat-protection interventions.

Building on existing work on data-driven modelling of internal temperatures, this thesis asked to what extent models derived from monitored data can predict the overheating risk of homes across the UK?

The research used three stages of investigation to develop empirical models from temperature data. In study one of the thesis, synthetic data from simulations of simple ‘shoe-box’ buildings in free-running mode were used to determine the relationship between outdoor and indoor thermal conditions. Further work involved data from two large-scale field trials in which the indoor temperatures in homes in the English Midlands and London were measured. In both these studies, a linear mixed effects model with random slope and intercept was used to identify the relationship between indoor and outdoor temperature whilst accounting for the differences between individual dwellings.

The use of a simple ‘shoe-box’ model in study one enabled the concept of ‘overheating signatures’ to be tested without the noise and uncertainty that emanates from ‘real world’ occupied homes. The results revealed that different building parameters or operating regimes produce a distinctive temperature sensitivity plot that relates indoor temperature to outdoor temperature.

Study two showed that the slope of the mixed effects model indicated the impact of a one-degree rise in outdoor temperature on indoor temperature across the sample of homes. On average, living rooms in the sample experienced an increase in indoor temperature of 0.39°C for each 1°C rise in outdoor temperature. However, the analysis revealed that homes had different intercepts, indicating whether they were warmer or cooler than the sample average. The slope also differed between the homes, indicating a greater or lesser sensitivity to changes in outdoor temperature. Further analysis using the indoor 2 day mean maximum temperature (2DMMT) as an overheating metric showed that the model could be used to identify homes that had a greater sensitivity to increases in the outdoor 2DMMT. The model was then used to predict the indoor 2DMMT for each home at an outdoor 2DMMT threshold appropriate for the region. Study three validated the findings from study two as the analysis was conducted on a sample of 27 London flats-a housing typology prone to overheating.

Overall, the results indicate the potential for simple data-driven models to map the overheating risk of free-running homes in a temperate climate. The method produces an overheating signature that is specific to the room and the way occupants behave regarding heat amelioration activities. A notable advantage of the modeling approach lies in its independence from the need for household or dwelling-specific information to predict overheating risk. The method relies solely on hourly internal air temperature measurements, alongside readily available external air temperature obtainable from local weather stations.

The mixed effects linear regression modelling method allows data from field studies conducted at different times to be combined without introducing bias to the analysis. The signature model can be used alongside probabilistic weather data to estimate the likelihood of exceeding indoor temperature thresholds. Identifying the homes at a higher risk of overheating is important information for developing retrofit interventions and conducting robust heat-health risk mapping exercises.

The limitations of the modelling technique are that it is unique to the specific room and the way it is operated by the occupants. Furthermore, care should be taken not to extrapolate beyond the range of temperatures for which the regression was devised. However, provided a warm summer is monitored, then indoor temperatures above the thresholds that represent thermal discomfort and an increased relative risk to health can be included, and so the frequency of exceedance of this in future weather can be estimated.

Further studies are needed to investigate the applicability of the modelling technique across the wider housing stock and to develop the technique to evaluate the effects of overheating mitigation strategies.