Impact of occupant behavior–driven metabolic rate variations on PMV prediction uncertainty through global sensitivity analysis

초록

This study investigates the impact of metabolic rate variations on Predicted Mean Vote (PMV) calculations in air-conditioned university office environments. We conducted a three-month experiment monitoring 34 occupants using infrared thermal imaging sensors, translating observed activity patterns into time-weighted average metabolic rates for probabilistic PMV calculations that incorporate measured distributions of environmental parameters. Results demonstrate that actual metabolic rates frequently exceed conventional sedentary estimates (1.0–1.2 met). When incorporating the measured metabolic rate distributions, PMV probability distributions shift notably toward warmer sensations and exhibit greater variance compared to scenarios using fixed metabolic assumptions. Global sensitivity analysis revealed variations in metabolic rates as the dominant contributor to PMV uncertainty (approximately 60 %), surpassing the influence of radiant temperature and clothing insulation that dominate under fixed metabolic assumptions. This pronounced influence of metabolic variability significantly undermines the reliability of conventional thermal comfort predictions, systematically shifting PMV distributions towards warmer sensations and increasing prediction variance. The comfort zone probability increased from 39.3 % to 62.4 %, but more importantly, this shift demonstrates how traditional sedentary assumptions may lead to a systematic cool-side bias in PMV predictions. These findings highlight the importance of explicitly incorporating realistic occupant activity patterns in thermal comfort modeling, suggesting potential improvements for PMV-based control strategies to improve both energy efficiency and occupant satisfaction in well-conditioned research environments. © 2025 Elsevier Ltd

키워드