Minimising Overheating in Passive and Low Energy Buildings Using Kriging-based inverse modelling techniques

Preventing summertime overheating within passive buildings is important for the comfort of the occupants. The likelihood that a building will overheat depends on several factors, including the form of the building, the percentage glazing and the building’s thermal mass and insulation. Furthermore, the amount of overheating depends on the criteria we use to measure it. We investigate the CIBSE TM52 overheating criteria and look at how they are affected by changes in the design of a PassivHaus style building. We calculate the
percentage of possible buildings that pass each of the three CIBSE criteria using the Gaussian process regression-based efficient global inversion (EGI) technique. Our work is divided into two stages. First, we look at the
sensitivity of the overheating criteria to the design (i.e. examining the building parameters that have the greatest effect on the overheating criteria). Second, we calculate the percentage of all possible building designs that meet these criteria using the EGI technique. This method provides an estimation of whether a building design will meet a criterion. This surrogate modelling method can be very accurate because the EGI technique ‘tunes’ the Gaussian process regression model to determine whether variables exceed a threshold. We explore the overheating criteria for 60,000 potential building designs. Our findings show that the relative glazed area has the greatest influence on the overheating criteria, whereas properties such as thermal mass and insulation have less effect than expected. Further work is needed to explore the effects on different building types in other climates.