Evaluating multiple parameters dependency of base temperature for heating degree-days in building energy prediction

To improve the prediction accuracy of heating demand, an appropriate base temperature should be estimated before using the heating degree-days (HDD) approach. This study collected the measured data for gas consumption at half-hourly resolution and the building physical characteristics from 89 educational buildings over four years. To determine the base temperature, in addition to the ambient temperature, more detailed independent variables, i.e. solar insolation, relative humidity, wind speed, and one-day ahead residual temperature, were incorporated into a three-parameter change-point multi-variable regression (3PH-MVR) for heating. The mean base temperature using the 3PH-MVR approach was about 0.4 degrees C lower than the results from the 3PH method only. The relationships between base temperature and annual HDD (based on 15.5 degrees C), building location, and mean daily solar insolation were evaluated. It is found that the annual HDD and the daily insolation had clear impacts on base temperature, while there was a plausible relationship between base temperature and building location. Compared with traditional approach, the proposed 3PH-MVR method considers multiple weather parameters and determines a more robust base temperature, thus improving the prediction accuracy of HDD with higher average R-2 value at 0.86 than that of univariate regression (0.82).

Automated summary

By incorporating more detailed independent variables into the model, the proposed 3PH-MVR method in estimating base temperature for heating demand prediction shows a higher accuracy compared to traditional approaches. The relationships between base temperature and annual HDD, building location, and solar insolation were evaluated in this study.