The Building Data Genome Project 2, energy meter data from the ASHRAE Great Energy Predictor III competition

Area of Science:

• Building energy analysis
• Sustainable energy systems
• Data science for built environments

Background:

• The need for comprehensive, real-world building energy data is critical for advancing energy efficiency.
• Existing datasets often lack the granularity, duration, or diversity required for robust model development and validation.
• The Great Energy Predictor III (GEPIII) competition highlighted the demand for advanced machine learning applications in building performance.

Purpose of the Study:

• To introduce a novel, open-access dataset comprising hourly energy and water consumption from diverse non-residential buildings.
• To provide a standardized resource for benchmarking prediction models, developing anomaly detection algorithms, and classifying building types.
• To facilitate research in measurement and verification (M&V) and long-term energy forecasting.

Main Methods:

• Collection and curation of time-series data from 3,053 energy and water meters across 1,636 buildings in North America and Europe.
• Integration of building metadata and complementary hourly weather data for enhanced analytical capabilities.
• Data cleaning and convergence processes to ensure data quality and time-series integrity.

Main Results:

• A dataset containing approximately 53.6 million hourly measurements spanning two full years (2016-2017).
• Inclusion of whole building electrical, heating/cooling water, steam, solar energy, water, and irrigation data.
• Data sourced from 19 distinct sites, offering geographical and building-type diversity.

Conclusions:

• The presented dataset serves as a valuable, open resource for the building science and data science communities.
• It enables advanced research in energy prediction, anomaly detection, and building performance analysis.
• The dataset's scope and quality support the development and validation of next-generation energy management strategies.