The value of regression models in determining industrial energy savings

Determination of industrial facility energy savings is commonly undertaken to comply with legal requirements, meet sustainability goals, and quantify greenhouse gas emissions, cost savings, and as one way to demonstrate energy performance improvement per the requirements of an energy management system standard, such as ISO 50001. At their most basic level, energy savings are determined through a process of accounting for baseline and reporting period energy consumption followed by a calculation of energy savings as the difference in the levels of energy consumption between the two periods. The determination of industrial facility-wide energy savings can be undertaken in a wide variety of ways and in relation to energy performance improvement; energy savings can be expressed as any number of energy performance indicators (EnPIs). Energy savings values will vary depending upon if an absolute, intensity, or regression basis of calculation is selected for use. The use of one of the three basis of energy savings calculation will alter the relevance, meaning, and comparability of the resulting energy savings values To increase consistency in the determination of energy savings values, standards such as the International Performance Measurement and Verification Protocol (IPMVP), ISO 17747 - Determination of Energy Savings in Organization, and the U.S. Department of Energy (U.S. DOE) Superior Energy Performance (SEP) program Measurement and Verification Protocol (SEP M&V Protocol) have been published. Industrial facilities in the United States, Mexico, and Canada are certified to the U.S. DOE SEP program after becoming ISO 50001 certified and verification by a third party of calculated energy savings per the SEP M&V Protocol. This paper discusses some of the different meanings energy savings values will have when calculated by using the three different bases. Additionally, data from five SEP certified industrial facilities are used to calculate energy savings values using the three bases and compared as energy performance improvement percentage (EPI %) values. Discussion of why a large degree in variation in EPI % values are seen for any given set of data is provided. Ultimately, while each energy saving calculation basis has merit and can be used appropriately in its own context, the regression basis is shown to best translates energy savings values into contextualized energy performance improvement values by conveying comparable information regarding operational and behavioural improvements not otherwise observable with the absolute or intensity bases.