Res–Intel goes beyond predictive analytics by providing customized, specific recommendations that prescribe cost-effective Demand Side Management for each single family residence!
Data always speaks and is always predictive. Seigel (2013) notes that this “data effect” is a leap of faith for organizations to take prior to engaging in predictive analytics.
They often don’t know what they will find prior to exercising their data muscles.
Fortunately, the developers of Res-Intel have validated real-world billing and real estate data and the results show that massive benefits can accrue from systematically combining re-purposed real estate data with building energy regression modeling.
Benchmarking LA County’s Energy & Water Use
Regression Modeling
Our understanding of building energy consumption can be enhanced through numerical modeling efforts. Res–Intel utilizes proven building energy regression modeling techniques that have a successful track record in identifying energy efficiency opportunities.
Most building energy modeling is performed with DOE-2 simulation models that typically overestimate the energy efficiency savings available from retrofits, often by 100% or more. These energy models require an incredible amount of detailed building data. LBNL’s Home Energy Scoring Tool, a standardized energy rating for home buyers and sellers, requires users to input attic and foundation insulation levels, window areas, window types, window glazings, frames, window fill, U-Factor and a solar heat gain coefficient.
Acquiring this information on widespread basis to rate all residential single family buildings’ energy ratings is impossible.
Res–Intel’s regression modeling approach uses parsimonious building data to predict energy and water consumption and savings opportunities.
Variables Accounting for Energy Use Predictions
Targeted Marketing: Energy and Water Conservation Uplift
Res–Intel is designed to first benchmark energy and water consumption at every residence in a jurisdiction and estimate their energy and water conservation conservation opportunities. Next, customer participation propensity is estimated from utility program data. Finally, a list of high savings, high propensity customers is generated for marketing outreach by the utility or 3rd parties.
A/C Mainenance
Using econometric methods, we have estimated the predicted probability of different households to perform A/C maintenance at different levels of key household variables.
- Baseline Uptake: Statewide, about 38% of single family households that pay for their own electricity performed some maintenance on their A/C system over the previous 12 month period.
- High kWh Usage Uptake: These customers have about a 5% higher probability of maintenancing their central A/C system than baseline.
- Res–Intel Uptake: By fully utilizing our software, utilities could target households that have more than a 9% higher probability of maintenancing their central A/C system than baseline. The uplift from the Res–Intel approach is nearly double the current targeted approach used by utilities.
By using Res–Intel prescriptive analytics over a 5 year program cycle, kWh savings from a utility A/C maintenance program could nearly double from targeting marketing uplift.
Water Savings
The following analysis highlights how the modeling results can be used to increase water savings by utility programs. We analyzed the penetration of front loading clothes washers by households in California.
- Baseline Uptake: Statewide, about 25% of single family households that pay for their own electricity in California have purchased an energy efficient front loading clothes washer that use only 1/3 of the water of a top loading washer.
- High kWh Usage Uptake: High kWh customers show about a 5.6% higher probability of purchasing a front loading clothes washer than baseline households. Electric utilities could target these customers with front loader purchase rebates.
- Res–Intel Uptake: By fully utilizing the our software, utilities could target households that have a 29% higher probability of purchasing a front loader clothes washer than baseline households. For this indoor water and energy conservation measure, the uplift from the Res–Intel approach is over 5 times the targeted marketing approach currently used by utilities.
Water conservation program cost-effectiveness can be increased through the use of Res–Intel’s predictive analytics. Targeting high use households adds only 5% marketing uplift each year, while targeting customers using the full suite of Res–Intel’s predictive analytics can potentially increase water savings by nearly 150% over a 5–year period.