Modeling thermal comfort holistically: Bayesian estimation of thermal sensation, acceptability, and preference distributions for office building occupants

Abstract

The three concepts of thermal sensation, acceptability, and preference each contribute to a holistic understanding of a building occupant's thermal comfort and how it can be effectively predicted. Nevertheless, there is currently no integrated framework for evaluating sensation, acceptability, and preference together as part of thermal comfort assessment in the built environment. Indeed, the only relation given between these variables in existing comfort guidelines - the Predicted Mean Vote – Predicted Percentage Dissatisfied (PMV–PPD) curve – rests on the tenuous assumption that occupants only find sensations at or near “Neutral” to be acceptable.This paper uses occupant response data from both the laboratory and field settings to develop an integrated approach for assessing office occupant thermal comfort through the multiple lenses of thermal sensation, acceptability, and preference. Specifically, probability distributions are developed for each of these comfort variables using Bayesian probit analysis. Given these distributions, we present revised PMV–PPD curves for field offices, and construct a new set of curves that represent the relationship between PMV and direct thermal acceptability and preference ratings. The probit analysis reveals that PMV is a significant predictor of thermal sensation distribution in the field; suggests that thermal acceptability and preference responses are subject to seasonal influences; and shows differences in thermal sensation, acceptability, and preference distributions for occupants in Air-Conditioned and Naturally Ventilated buildings. The usefulness of the developed distributions to practical thermal comfort assessments is discussed, as is the potential for these distributions to be updated in the future as more data are collected.