Uncertainty in the evaluation of the Predicted Mean Vote index using Monte Carlo analysis

Today, evaluation of thermohygrometric indoor conditions is one of the most useful tools for building design and re-design and can be used to determine energy consumption in conditioned buildings. Since the beginning of the Predicted Mean Vote index (PMV), researchers have thoroughly investigated its issues in order to reach more accurate results; however, several shortcomings have yet to be solved. Among them is the uncertainty of environmental and subjective parameters linked to the standard PMV approach of ISO 7730 that classifies the thermal environment. To this end, this paper discusses the known thermal comfort models and the measurement approaches, paying particular attention to measurement uncertainties and their influence on PMV determination. Monte Carlo analysis has been applied on a data series in a “black-box” environment, and each involved parameter has been analysed in the PMV range from −0.9 to 0.9 under different Relative Humidity conditions. Furthermore, a sensitivity analysis has been performed in order to define the role of each variable. The results showed that an uncertainty propagation method could improve PMV model application, especially where it should be very accurate (−0.2 < PMV<0.2 range; winter season with Relative Humidity of 30%).