Strategic Energy Management Modeling: What’s good enough?
Strategic Energy Management (SEM) programs have expanded over the years in the United States and abroad (CEE 2016). SEM programs vary widely in implementation and scope, but typically involve a longer-term customer engagement to train and embed energy efficient practices. Energy Trust of Oregon (Energy Trust) has been implementing industrial Strategic Energy Management since 2009.
This analysis looks at the 59 SEM customer facilities that were in the sample selected for the Energy Trust’s Production Efficiency Program 2012-2013 impact evaluation and a separate Industrial SEM Process and Impact evaluation. The following questions are investigated:
- What are the characteristics of models developed for SEM?
- Are there ways to simplify and streamline model development?
- How well do the simple models perform compared to the models used to estimate savings?
Source: Energy Trust of Oregon
Published: August 2017
Download the White Paper
Abstract
Both commercial and industrial Strategic Energy Management (SEM) programs include the development of one or more models of energy consumption for each participating site. Protocols for model design have been put forward with many specifics on model requirements. Significant resources are often spent on developing models due to data and specification requirements. The question of what level of model development is good enough to estimate program level savings has not been addressed.
Energy Trust of Oregon has been implementing SEM for commercial and industrial customers for many years, and many hundreds of models have been developed. This study selects a sample of gas and electric models from the industrial SEM program, and reruns them using standardized simplified specifications. The overall savings estimated by the models specified by the program and those from the simplified models were compared, to determine if the difference in savings estimated by the two methods is statistically significant. The results prove useful in guiding programs on what level of analysis should be undertaken for SEM participants at the start of an SEM engagement. As simplified models rely on a reduced set of variables and data, the results of this study also provide guidance on what types of data the program should collect to develop the initial model.