Re-examination of external temperature as a predictor of energy usage in buildings

• A unique relationship between external and internal temperature is proposed.
• Referred to as natural thermal lag (NTL) the relationship is explored in 3 buildings.
• NTL can be calculated by year-long measurement of the buildings ‘at rest’.
• In energy forecasting, NTL can guide the selection of the optimum predictor.
• NTL can also be derived by statistical regression analysis of energy usage alone.

The importance of reducing energy usage in buildings is difficult to overstate. A large part of the research interest in energy reduction focuses on the ability to accurately forecast consumption. The practise of using different forms of statistical models to achieve accurate forecasts is well established, both as an operational guide but also to accurately estimate savings following an efficiency programme. Within the range of these models, the most common form is that of regression and the simplest of these often establishes a relationship between the single predictor of external temperature and energy usage. Where an external temperature index is used in regression, the most common form is that of daily average temperature. Given the unique nature of commercial buildings, this paper seeks to examine the nature of a building's thermal response to local external temperature and examine how this parameter might influence a simple statistical model's energy prediction accuracy. Examination of three different large commercial buildings has shown that the application of daily average temperature may not provide the most accurate predictor. At a whole building level, a method has been devised to find the most influential range of external temperatures affecting energy usage.