Thermal comfort evaluation for mechanically conditioned buildings using response surfaces in an uncertainty analysis framework

An uncertainty analysis methodology is proposed to aid in quantifying the risks of thermal comfort under-performance posed by changes to variations in physical and operational characteristics of a building and its environment. This includes those implemented for building energy savings, peak electricity load reductions, or those due to climatic changes. Using building performance data as input, a response surface methodology is used to develop a model to predict building thermal performance for ranges of user-defined design variables. This model is verified for accuracy using in- and out-of-sample data. Uncertainly analysis is then used to estimate the probability of achieving an acceptable threshold of thermal comfort performance. A case study is presented to demonstrate the implementation and interpretation of the results of this methodology, which evaluates the effects of a 1-h demand response event on thermal comfort of a residential mechanically-conditioned building. The case study finds that a second-order response surface provides a reasonably accurate model of thermal comfort. For the studied single family home, compared to varying the air exchange rate, the indoor set-point temperature has a greater influence on achieving an acceptable level of thermal comfort.