A metamodel for long-term thermal comfort in non-air-conditioned buildings

Thermal comfort is a primary concern of building design. Accurate and rapid methods are needed to assess thermal comfort in non-air-conditioned buildings. In this work, a new flexible metamodel was developed for assessing long-term thermal comfort in non-air-conditioned buildings. Using five different indices, the metamodel was used to study the thermal comfort in an office during summer for two typical cold and hot European climates. The metamodel was fitted by multiple regression analysis with a low number of dynamic simulations using the design of experiments. A good fit was obtained for indices, such as maximum indoor temperature, that are consistently sensitive to the influencing factors. For the remaining indices, such as overheating degree-hours, the accuracy was dependent on the location and thermal mass of the building. An interpolation method was proposed that can be used when the metamodel fit is not good. This method gave accurate results with all the indices (R-2 > 0.98) when the values were obtained from the mean indoor temperature without additional computational efforts. The proposed metamodel could be used to rapidly assess long-term thermal comfort and to understand the relationships between building design and indoor thermal conditions.