Big-data for building energy performance: Lessons from assembling a very large national database of building energy use. (15th February 2015)

• Record Type: Journal Article
• Authors: Mathew, Paul A.; Dunn, Laurel N.; Sohn, Michael D.; Mercado, Andrea; Custudio, Claudine; Walter, Travis
• Is Part Of: Applied energy. Volume 140(2015:Feb. 15)
• Page Start: 85
• Page End: 93
• Publication Date: 2015-02-15
• Subjects: CBECS Commercial Buildings Energy Consumption Survey -- RECS Residential Energy Consumption Survey -- CEUS Commercial End-Use Survey -- DOE Department of Energy -- BPD Buildings Performance Database -- USEIA United States Energy Information Administration -- EUI Energy Use Intensity -- BEDES Building Energy Data Exchange Specification -- ESCO Energy Service Company -- NBI New Buildings Institute -- USEERE United States Office of Energy Efficiency & Renewable Energy; Buildings Performance Database -- Building performance -- Big data -- Building data collection -- Data-driven decision support; Power (Mechanics) -- Periodicals; Energy conservation -- Periodicals; Energy conversion -- Periodicals; 621.042
• DOI: 10.1016/j.apenergy.2014.11.042
• Languages: English
• ISSNs: 0306-2619

Abstract

Highlights: The largest compilation of building energy data in the US, with over 750, 000 buildings. Most of the effort lies in data cleansing and mapping to a common data schema. Paper includes comparisons to data in CBECS and RECS – the US national statistical datasets. The database supports empirical comparison of energy use and data-driven savings analysis. Abstract: Building energy data has been used for decades to understand energy flows in buildings and plan for future energy demand. Recent market, technology and policy drivers have resulted in widespread data collection by stakeholders across the buildings industry. Consolidation of independently collected and maintained datasets presents a cost-effective opportunity to build a database of unprecedented size. Applications of the data include peer group analysis to evaluate building performance, and data-driven algorithms that use empirical data to estimate energy savings associated with building retrofits. This paper discusses technical considerations in compiling such a database using the DOE Buildings Performance Database (BPD) as a case study. We gathered data on over 750, 000 residential and commercial buildings. We describe the process and challenges of mapping and cleansing data from disparate sources. We analyze the distributions of buildings in the BPD relative to the Commercial Building Energy Consumption Survey (CBECS) and Residential Energy Consumption Survey (RECS), evaluating peer groups of buildings that are well or poorly represented, and discussing how differences in the distributions of the three datasets impact use-cases of the data. Finally, we discuss the usefulness and limitations of the current dataset and the outlook for increasing its size and applications.