Enhancing thermal comfort in public spaces through agent-based and performance-based modeling

Human thermal comfort is a key indicator for assessing the quality of indoor environmental quality (IEQ), particularly in large public spaces of older buildings, such as stations and cafeterias, where the thermal environment often falls below standard. In recent years, many studies have integrated personal thermal sensation assessments into the control of air conditioning systems, striving to optimize the settings of heating, ventilation, and air conditioning (HVAC) systems to improve indoor thermal environments while reducing energy consumption. However, few studies have taken a human-centric approach, considering improvements beyond adjustments to HVAC equipment, such as through spatial restructuring and behavioral management to further enhance thermal comfort.

This paper introduces a user-driven modeling approach for guiding spatial division and function management to create more satisfactory indoor thermal environments. Our research aims to develop an agent-based model to simulate the use of a university cafeteria during the summer, demonstrating a complete process from data collection and occupancy modeling to performance assessment and design optimization. In the data collection phase, we employed fixed-point video recording to count and record the number of diners, their behavior trajectories, activity processes, and durations to obtain primary behavioral data, facilitating result calibration and comparative analysis.

In the occupancy modeling phase, leveraging the BSim simulation platform, we developed an agent-based behavioral simulation model for thermal environments. During the performance assessment phase, physical space and location data of individuals were input into COMSOL software for dynamic thermal comfort assessment. In the design optimization and evaluation phase, based on the results of the thermal comfort assessment, we proposed nine different scenario design options including spatial layout, window management, and equipment adjustments. Using a controlled single variable method in the BSim and COMSOL platforms—keeping diner attributes constant and only altering the spatial environment—we simulated different scenarios. Finally, using indices such as the thermal comfort performance index, activity efficiency of individuals, and space utilization rate, we evaluated each scenario to select the most optimal design.

The method proposed in this paper, combining agent behavior modeling with performance modeling to assess spatial designs, provides a realistic and dynamic basis for selecting advantageous design options, offering specific guidance for designers committed to improving the functionality and comfort of building environments.