Model Selection for Predicting the Return Time from Night Setback

This paper reports a comparison of models for estimating the return time from a night setback condition. Fifty-seven models are compared using simulation data that include the influence of climate, building mass, controller tuning, room orientation, and the unoccupied control strategy on return time. Two-parameter models are recommended for estimating the return time for both heating and cooling. The models use the room air temperature and an EWMA of the normalized heating demand (for the heating model) or cooling demand (for the cooling model) as predictor variables. The outdoor air temperature, a common input for predicting return time, is not used in the recommended models, but influences the return time through its effect on the heating and cooling demands.