Analysis of a simplified calibration procedure for 18 design-phase building energy models

Abstract

This paper evaluates the accuracy of 18 design-phase building energy models, built according to LEED Canada protocol, and investigates the effectiveness of model calibration steps to improve simulation predictions with respect to measured energy data. These calibration steps, applied in professional practice, included inputting actual weather data, adding unregulated loads, revising plug loads (often with submetered data), and other simple updates. In sum, the design-phase energy models underpredicted the total measured energy consumption by 36%. Following the calibration steps, this error was reduced to a net 7% underprediction. For the monthly energy use intensity (EUI), the coefficient of variation of the root mean square error improved from 45% to 24%. Revising plug loads made the largest impact in these cases. This step increased the EUI by 15% median (32% mean) in the models. This impact far exceeded that of calibrating the weather data, even in a sensitivity test using extreme weather years.

Keywords:

• energy simulation
• plug loads
• process loads
• unregulated loads
• weather
• design-phase models
• compliance models

Acknowledgements

The authors sincerely thank the following individuals for sharing their work and knowledge, in particular Steve Kemp, Victor Halder, Antoni Paleshi, and Eric Rubli. The authors also thank Dru Crawley for providing the extreme-year weather files, as well as Adam Scherba, Cathy Turner, Mark Frankel, and Guy Newsham, for providing the data set of LEED projects for comparison. This work was supported by the Harvard Graduate School of Design, the Harvard Real Estate Academic Initiative, and the Massachusetts Institute of Technology.

Notes

1. This includes the 92 buildings for which measured and predicted EUI data were available.

2. The authors have made available the custom Python scripts created for this research to batch process DOE (.sim) results files and extract key monthly and annual results. These scripts may be helpful to EE4, eQuest, and other .sim file users. The public can download them here: https://gist.github.com/hollywas/4757770 and here: https://gist.github.com/alexstorer/4219834.

3. The modellers followed the EE4 modelling protocol (Natural Resources Canada Office of Energy Efficiency & CANMET Energy Technology Centre 2008) building zoning principles, thereby creating separate model zones per HVAC system, space operation/function, heating/cooling loads, and solar orientation.

4. Compliance-type model: energy model used to demonstrate compliance with a building code or standard.

5. For example, in Building 8, the added retail displays not only consumed electricity directly, which the modeller tallied in the spreadsheet, but also added heat to the space, which would be ignored with this method.