A system identification of radiant ceiling panels using simplified RC network model for cooling performance prediction under tropical climate in Thailand

Accurate thermal modeling is essential for energy-efficient operation of building cooling systems, particularly in tropical climates where buildings are predominantly cooling-dominated. This study develops a gray-box thermal model for radiant cooling using a Resistance–Capacitance (RC) network with Dynamic Mode Decomposition (DMD) for automated parameter identification. Experiments were conducted in a 27 m² test chamber at King Mongkut’s University of Technology Thonburi (KMUTT) in Bangkok, Thailand, during the peak hot season, with outdoor temperatures reaching 35°C and solar radiation up to 1,390 W/m². A four-state RC model representing ceiling panel, indoor air, interior surface, and operative temperatures was developed. Using one week of data at 1-minute intervals, the model achieved 0.34°C RMSE in operative temperature prediction. Identified parameters showed physical consistency, with air time constants of 15–30min and surface time constants of 4–6h. Parameter identification required less than 5s, and forward prediction required less than 0.001s per step, supporting real-time applications. The method combines physical interpretability with computational efficiency, providing a practical tool for sustainable building energy management in tropical climates.