Estimating Energy Savings Resulting from Strategic Energy Management Programs
A rapidly growing approach to energy efficiency, strategic energy management (SEM) programs realize savings by using capital in concert with operations and management improvements to facilities. By using various frameworks, evaluators of these programs can estimate reductions in energy consumption before and after program implementation.
In this study, the authors used a simulation approach to test different frameworks and model specifications.
Source: Cadmus Group
Published: August 2017
Methodology Comparison Paper
Abstract
Strategic energy management (SEM) programs seek to implement long-lasting, comprehensive changes in energy management and consumption at industrial facilities.1 Evaluators typically estimate energy savings associated with SEM programs using regression analysis to model facility energy consumption as a function of weather and other variables. They then use the regression models to compare energy consumption before and after program implementation. Evaluators can choose among several regression model frameworks, with the accuracy of results depending on the model selected. Therefore, selecting the right framework proves integral to completing an accurate, robust evaluation. Testing frameworks to determine those best suited to evaluate savings for a particular facility (and under a specific set of circumstances) presents a challenge as true energy savings value at a facility remain unknown. Using all available frameworks, evaluators can estimate savings and compare results, conjecturing about how well the frameworks estimate savings, but they cannot compare the estimates to a true energy savings value to determine, with certainty, which works best.2
By using simulated data, evaluators can begin answering questions about which regression models perform best and under what circumstances. If simulating facility energy consumption data using a function of weather and other known variables (and because they generate it), evaluators can determine “true” savings and savings drivers underlying the model. In this study, the authors use simulated energy consumption, based on models representing industrial facilities that participate in SEM, to answer two important questions:
-
Which regression analysis framework should evaluators use to calculate accurate, robust savings estimates?
-
How should the facility regression model be specified within that framework?
The simulation revealed several important findings.