Physics-Based Interval Data Models to Automate and Scale Home Energy Performance Evaluations

Fraunhofer USA is working with team members from Eversource Energy, Holyoke Gas & Electric, and National Grid to develop a tool for utility energy efficiency programs that remotely identifies homes with retrofit opportunities and quantifies the potential energy savings for each homes, using communicating thermostat data and interval electricity and gas data.

Home energy assessments are costly ($250–$400) and inconvenient, and only about one-third result in a major retrofit. This customer acquisition cost of ~$1,000 threatens program cost-effectiveness. The tool addresses the main challenges that impede the key steps of energy conservation measure deployment through utility energy efficiency programs using data analytics. This leverages data from the dramatic growth in the installed base of communicating thermostats to automate, simplify, and customize retrofit opportunity assessment, customer acquisition, and energy performance evaluation.

The team is developing, training, and validating this remote performance assessment tool by using home energy assessment information from homes with communicating thermostats in Massachusetts. The technical and market goals of the project are for the algorithm to: 

• Identify the ~20% of homes that would benefit most from at least one of the target energy conservation measures (ECMs) to reduce space heating energy consumption: insulation or air sealing
• Accurately predict the energy savings of the target energy conservation measures and evaluate the realized energy savings
• Increase the rate of on-site energy audits for households with at least one of the two target retrofit measures and provide household-specific energy savings estimates
• Increase the fraction of site audits resulting in implementation of the target energy conservation measures by focusing audits on homes with a target retrofit opportunity.

• Developed high-resolution PRISM model to predict runtime and future savings from retrofits, as well as perform EM&V
• Validated model for homes heated with boiler or furnace with 1 or 2 communicating thermostats using information from home energy assessments
• Successfully classified whole-home R-value with 89% accuracy as compared to ground-truth values
• Successfully classified ACH₅₀ with 96% accuracy as compared to blower-door test values

• Conduct randomized-controlled trial using customer-specific outreach, i.e., identifying specific ECMs and their expected energy savings based on developed tool
• Evaluate whether outreach with specific retrofit measures and predicted savings can increase uptake of:
  • On-site home energy assessments
  • Implementation of energy retrofit measures.

For more information, project presentations, and related publications, visit the Physics-Based Interval Data Models to Automate and Scale Home Energy Performance Evaluations web page.