A Review of the Energy Performance Gap between Predicted and Actual Use in Buildings

Abstract

It has become a scientific consensus that buildings are one of the most critical sectors to achieving zero emissions, efficiency, and resiliency, thus mitigating and adapting to the impacts of climate change. Building performance is the means to communicate energy efficiency and carbon emissions, as well as detect performance issues and improvement measures. The traditional approach to assessing building performance is modeling the building performance with physics-based computer simulation engines and using the simulated results as the indicator of compliance, labeling, and certification related to energy-efficiency codes and standards. A significant drawback of this design-only simulation approach is that designs of buildings and completed buildings differ, and with them also the design-phase simulation and the actual use. Studies have indicated that there is a substantial gap between the computer-predicted performance at the design stage and the actual performance of buildings once constructed. This gap suggests that the in-use building does not align with the expected performance targets. The performance gap could lead to skepticism in high-performance building concepts, undermining public confidence in the building energy efficiency and decarbonization movement. Growing research attention has been drawn to this problem in recent years. However, the presented understanding of the performance gap and hence its significance remain incomplete and inadequate.With the purpose of enlightening the latest research status and future research directions, this paper reviews existing literature on this performance gap problem and yields an in-depth discussion on the magnitude and reasons for the performance gap and ongoing strategies to close it. Furthermore, unlike other review articles on this problem, this paper discusses the application of state-of-the-art techniques, statistic regressions and machine learning, in improving the prediction accuracy of building performance in addition to those used in traditional physics-based simulation engines.

Topics & Concepts