Enhancing thermal comfort in shared building spaces: A group thermal preference framework and optimization strategies

Due to intergroup differences in thermal preferences across shared spaces, advanced optimization strategies often fail to achieve consistent effectiveness in various scenarios. Early selection of an appropriate type of optimization based on group thermal preferences is therefore essential. However, the limited understanding of the underlying mechanisms of group thermal preferences makes this a significant challenge. To address this issue, this study introduces a three-indicator framework for defining group thermal preferences: optimal temperature, thermal sensitivity, and maximum satisfaction rate. This framework enables a clear characterization of intergroup differences. Using thermal preference data from ASHRAE Database II, we conducted stochastic simulations to analyze the correlations between fundamental group attributes and group thermal preferences. Primary findings reveal that group size and internal diversity significantly influence group thermal preferences. Larger groups (50–500 people) exhibit a 3 %-14 % lower potential maximum satisfaction rate than smaller groups (10-person), along with reduced thermal sensitivity and more fixed optimal temperatures. Additionally, every 1 °C increase in the standard deviation of preferred temperature among the group members can lead to an approximate 18 % decrease in the group's potential maximum satisfaction rate. These insights facilitate rapid assessments of group thermal preferences in early-stage planning. On this basis, the optimization priorities for groups with distinct characteristics are discussed, and a bottom-up decision-making framework for optimized thermal environment design is proposed. Additionally, a novel thermal environment evaluation metric, termed satisfaction achievement rate (SAR), is introduced, which helps to mitigate the existing unfairness and misinterpretation caused by intergroup differences in post-occupancy evaluation. © 2025 Elsevier Ltd. All rights are reserved, including those for text and data mining, AI training, and similar technologies.