Our Technology

BOLT – Building Optimization and Learning Technology

The BOLT (Building Optimization and Learning Technology) is a proprietary energy analysis platform. Our engineers perfected the system in both the academic and business sectors. It is built on a foundation of internationally-accepted statistical utility analysis processes.

BOLT’s powerful algorithm models whole building regression analysis of utility data. This allows the removal of weather variance to capture the change in energy performance for a building (or home) over time, whether to capture energy conservation measures (ECMs) impact, verify operational changes, identify system issues, or to ensure continued energy savings achievement.

Empower’s proprietary algorithms and architecture provides a powerful tool for the residential and small commercial sectors, where program engagement has been limited to broad offerings or broad-based marketing. Every building is unique – whether a home or a small business. Our BOLT system recognizes these differences, and provides predictive analysis that helps energy consumers, efficiency providers and utilities develop accurate, relevant strategies for energy efficiency engagement.

The engine quickly runs large numbers of regressions. These performance profiles allow energy intensities to be generated for all the homes or buildings in a region. Metrics can be compared to similar structures in the area, allowing identification of those building owners who would benefit most from energy efficiency measures.

Rather than spend time and money in broad-based marketing and outreach, BOLT allows marketers to focus resources on those whose needs are highest, where return-on-investment is shortest, and where the measured success of a project will be highest. BOLT can also integrate with our partners’ data sets to deliver even more accuracy.

The Predictive Analysis Software Engine Process

The Customer Journey

DATA MERGE

At the onset of any customer engagement, a means of triangulating the obtained electric utility, gas utility, and county auditor data sets is required. These large files, vary in format and content between utilities and auditors. The software and team has built methods for processing the raw data from disparate sources, and not only extracts the necessary utility and house characteristics, but also through comparison of the files determines additional information about each building that can be used for precise targeting.

For example, in selling energy efficiency products to homeowners, our data includes categorical determinations such as whether the building is a landlord-tenant arrangement, single family dwellings or even if a household is currently receiving government-sponsored subsidies.

REGRESSION ANALYSIS

The software is able to process large numbers of accounts (residences) in a very short time. The capabilities have been tested to process 1 million data sets in under 7 minutes. The results generated from these computations are statistical regression models. These “performance models” are piecewise models where the “best fit, least squares” regression is generated for each building based upon its own energy consumption history.

The theoretical model our algorithms define are recognized and broadly used for thermodynamic models which relates energy consumption to outdoor air temperature. The model coefficients generated relate to both building performance and characteristics.

TARGET THE HIGHEST IMPACT

Using the performance models generated, along with building square footage and other information from the county auditor data, each structure is normalized and then compared to similar structures in the same geographic area. This allows the identification of worst performers quickly. Currently, our process identifies the most energy intense houses using region based comparison, identifying three target groups. Through inverse modeling, we don’t just assess annual consumption but compare heating energy, cooling energy, and average daily usage (lights, TV, etc.). Through this process marketing efforts can be applied to the most likely population of the homes within the targeted home population

MEASURE and VERIFY

International Performance, Measurement, and Verification Protocol (IPMVP) allows whole building models to be used when quantifying energy impacts of ECMs. Consequently, the performance model generated for targeting can be compared to a new model generated after improvements are made to verify the impact of ECMs, and re-inform the overall strategy and process.