Effects of building energy optimisation on the predictive accuracy of external temperature in forecasting models

• Predictive relationship between external temperature and energy usage is explored.
• Energy efficiency programmes are developed for two buildings.
• Results of programmes show substantial reductions in energy usage in both buildings.
• Predictive power of external temperature is seen to substantially improve.
• Developing forecasting models should take account of building operational efficiency.

The practice of using external temperature as a predictor of energy consumption in commercial buildings is well understood and it has been shown that this predictor is the most important when compared to other weather-based factors. In the literature, where a study has shown the external temperature not to be sufficiently accurate in forecasting energy usage by statistical methods, more predictors can be added to derive a more accurate forecasting model. The influence of the operating efficiency of the building on this accuracy has not been properly explained. This may be because of a lack of internal performance data for the building under review.This study has been developed to examine the effects of building operating efficiency on the forecasting accuracy of a statistical model by isolating external temperature as the sole predictor. The mechanism used to demonstrate the effects of the building's operating efficiency on the merit of using solely external temperature as a predictor is linear regression between external temperature and energy usage data.The study was carried out on two buildings of differing scale, occupational use and construction techniques. Using revised operational schemes involving more carefully controlled energy consumption and by monitoring local external temperatures, the relative accuracy of the various regression models is seen to be enhanced significantly following the efficiency programmes. The study has also facilitated an examination of the relative contributions to forecasting model accuracy resulting from energy efficiency actions taken in both buildings compared to the use of various external temperature indices.