Bayesian thermal comfort model

Abstract

Thermal comfort assessment is a prime measure in indoor environment design to evaluate occupant satisfaction. Fanger's thermal comfort model using heat balance theory conducted by chamber test has been widely adopted for thermal environment design criteria. However, rising numbers of thermal comfort field studies show that Fanger's model is not a good predictor of actual thermal sensation and many field measurements were statistically insignificant. This study proposes a Bayesian approach to update our current beliefs about thermal comfort and shows that the maximum likelihood of posterior estimates is close to the actual percentage dissatisfied (APD) obtained from large sample field surveys. For small sample sizes, the Bayesian estimation is close to Fanger's prediction and gives a solution for the discrepancy of Fanger's model. Congruence between Fanger's model prediction and contemporary field survey data is quantified. This quantitative assessment on the belief in newly yielded thermal comfort data can be a solution to the choice of thermal comfort criteria in future thermal environment designs.