A two-step calibration framework for accurate building energy simulations: Integrating energy and indoor temperature data

Accurate building energy consumption simulation is essential for evaluating energy-saving strategies and enabling real-time control. Traditional calibration methods, which rely solely on energy consumption data, often fail to capture dynamic thermal behavior, particularly in systems with significant temperature lag, such as radiant floor heating. To address this limitation, this study proposes a two-step calibration framework using particle swarm optimization (PSO) to integrate both energy consumption and indoor temperature data. In the first step, static parameters such as building envelope performance are calibrated during unoccupied periods using conventional error metrics. In the second step, dynamic parameters such as occupant behavior and internal heat gains are adjusted during occupied periods to improve thermal accuracy. A case study of a childcare center with radiant floor heating demonstrated that the proposed method reduced the mean hourly deviation error (MBE) from 10.80 % to − 4.77 % and the coefficient of variation of the root mean square error (CV-RMSE) from 74.32 % to 22.79 % in terms of energy consumption. These results confirm that the method enhances both energy consumption and temperature prediction accuracy, offering greater reliability for real-time building operation and control.