A Statistical Approach to Plant-Level Energy Benchmarks and Baselines: A Manufacturing-Plant Energy Performance Indicator

摘要

Energy efficiency is a metric of relative performance, to either a benchmark or plant specific baseline. As such, these metrics are an integral component of energy and carbon management practices. This paper presents an approach used by Energy Star to implement manufacturing plant energy benchmarking, or Energy Performance Indicators (EPI), for a variety of industries. To date, EPI have been developed for industries ranging from “light” manufacturing such as auto assembly and food processing to “heavy”, energy intensive sectors like cement, glass, and paper. Since “all plants are different,” the EPI control statistically for differences between plants including product mix, climate, utilization, and vertical integration. After adjusting for differences between plants the EPI statistically “scores” plants from 1 to 100 in terms of their percentile ranking. When characteristics change within a plant over time, these EPI can also be used to construct adjusted energy baselines. This paper describes the approach used for developing the EPI. It compares the results for 10 industries, in terms of the types of variables that are included and the range of performance, measured by the inter-quartile range. The paper gives an example from pharmaceuticals of how the EPI can be applied to create adjusted baselines, in this case normalizing for year to year differences in weather. The paper provides examples of how the distribution of performance has changed over time for auto assembly and cement manufacturing. The range of performance for both sectors has narrowed, contributing to an industry wide reduction in energy and carbon. Since Energy Star for Industry was launched, the number of EPI in use or under development has grown to 19 sectors within 11 industries. Until now, no industry-wide, plant-level, energy benchmark previously existed for these industries. We find that every industry has plant specific factors that influence energy consumption, so that a measure of energy efficiency must account (normalize) for those differences to be a useful management tool.