Why is the reliability of building simulation limited as a tool for evaluating energy conservation measures?

Buildings account for approximately 32% of the total energy consumption globally and up to 40% in the developed countries, which makes buildings a prime target for energy conservation. Various energy conservation measures (ECMs) have been proposed to improve the energy efficiency in buildings, and these ECMs are usually designed and assessed using calibrated building energy models. However, there is empirical evidence that reveals noticeable discrepancies between simulated performances of ECMs reported in building energy models and their actual performances measured in buildings. This paper examines two possible causes of such discrepancies. Specifically, this paper tests the following two hypotheses: (1) using assumed occupancy data as opposed to actual occupancy data in building energy simulation reduces the reliability of estimated performance of demand-driven ECMs; and (2) using an energy model built for one ECM to cross estimate energy consumption of another ECM is statistically inaccurate. An educational building was used as a test bed. The results proved both hypotheses true, showing that estimations were more accurate and consistent for models calibrated using actual occupancy compared with those using assumed occupancy, and that cross-ECM estimation resulted in statistical inaccuracy. The findings indicated that current building energy modeling methods have limited reliability in ECM performance assessment, and need to be improved to better support the design and implementation of ECMs in buildings.