An energy performance indicator is a measure of energy performance that can be compared over time: energy per unit made, or the ratio of actual energy to what a baseline model expected. ISO 50001 (clauses 6.4 and 6.5, as the US DOE 50001 Ready Navigator describes them) has the plant set one or more EnPIs and an energy baseline to compare them with.

Why not just track energy per unit made?

It is a fair EnPI when nearly all energy varies with output. Where part of it is a fixed load, energy per unit rises when output falls. In the worked example the year used 2.2% less energy than the model expected while energy per piece rose 1.2%. The regression separates the fixed load, the energy per unit made and the effect of weather.

How many months of data do I need?

12 months for the baseline. The DOE 50001 Ready protocol requires at least 12 months of data to build a model and compares 12 consecutive baseline months with 12 reporting months so that seasons are covered. The Results sheet calculates the model only when all 12 baseline months are entered.

What do the model tests check?

The four statistical tests from the DOE protocol: the F-test p value of the whole model below 0.10, every driver below 0.20, at least one driver below 0.10 and R squared of at least 0.50. Then three more: coefficients above zero, reporting months inside the range the baseline covered, and your own confirmation that the drivers make sense and the meters and product mix did not change.

How is this different from the environment pillar 90-day plan?

The plan is a 13-week method for cutting waste and energy in one area. This template is the measuring sheet for the energy side: the baseline, the monthly check against expected and the saving. The plan sets a baseline energy per unit; this sheet shows whether energy fell once output and weather are accounted for.

Free template

Updated October 14, 2026

Energy use and EnPI tracker

An energy performance indicator (EnPI) is a measure of energy performance that can be compared over time, and the energy baseline is the reference period or model it is compared with. Energy per unit made is the simplest EnPI, but it moves with output even when nothing in the plant has changed, because part of the energy is a fixed load (lighting, compressors, heating, idle equipment) that does not fall when output does. This template keeps a plant's monthly electricity, gas and other fuel in one workbook and converts them to one unit. It fits a regression line on a 12-month baseline (energy against output and, where weather matters, degree days) and shows for every month what the plant should have used. The gap is the result: actual minus expected, flagged when it is bigger than the normal scatter of the baseline months, added up into an EnPI ratio and an improvement percentage. The method and the model tests follow the US Department of Energy's 50001 Ready measurement and verification protocol. The PDF has the blank monthly log, a worked example for a made-up plant and a how-it-works page. It is a tracking tool: not a verified savings claim and not a statement of what ISO 50001 requires.

Free to use: print it, copy it and edit it for your team. Enter your name and work email once to download.

Page 1 of the energy use and EnPI tracker — Electricity, gas and other fuelOne common unitOutput and degree days12-month baseline modelExpected against actual each monthEnPI ratio and improvementFixed load shareRunning saving chartModel validity testsWorked example

When to use it

When to use an energy use and EnPI tracker

When you need to know whether the plant uses less energy for the work it does, not just less energy in a slow month.
When a utility bill or an energy project needs a saving figure that still holds up after output and weather changed.
When you start an energy management system, such as one based on ISO 50001, and need an energy baseline and an EnPI.
When energy per unit went up and someone asks whether the energy work failed.
Every month in the plant review, to catch the month that came in above expected while the year still looks fine.

How to fill it in

1
Set the common unit and the factors
MMBtu is the default: 0.003412 MMBtu per kWh and 0.1 MMBtu per therm (3,412 and 100,000 BTU, the factors in the DOE protocol). Change them to your own unit if you prefer kWh or GJ. Use the same factors in every month.
2
Enter 12 baseline months
The first 12 rows: electricity in kWh, gas in your gas unit, other fuel if any, output in units made and, if weather moves your energy, degree days from one source and one base temperature. Costs are optional.
3
Pick the drivers for each model
Electricity often depends on output only. Gas for heating usually needs degree days too. The Results sheet fits electricity, gas and total energy and shows the fixed load, the energy per unit made and the model tests.
4
Add one row a month
Expected energy comes from the baseline model at that month's output and weather. Actual minus expected, the flag and the running saving fill in. A month further from expected than 2 standard errors of the model is flagged.
5
Read the EnPI ratio
Actual divided by expected over 12 months. Below 1.000 is better; improvement is 1 minus the ratio. Compare it with the simple energy per unit made, which sits beside it. If they disagree and the model passes its tests, trust the model.
6
Start a new baseline when the plant changes
A new line, a new product family or a new meter changes the fixed load. Comparing against a year that no longer describes the plant gives a wrong answer.

A filled-in example

Illustrative, not a benchmark

An example: a made-up plant with 24 months of data, 12 baseline and 12 reporting, output counted in pieces (illustrative numbers).

The baseline electricity model: 53,216 kWh a month of fixed load plus 1.84 kWh for each piece made. That fixed load is 37.0% of baseline electricity. Gas depends on output and on cold: 3.14 therms for each heating degree day.
The total energy model explains 98.7% of the monthly swing (R squared 0.987), passes all the DOE tests, and a typical gap between a month and the model is 19.6 MMBtu.
Reporting year: 12,703.0 MMBtu used against 12,983.6 expected. Saved 280.7 MMBtu, EnPI ratio 0.978, improvement 2.2%, worth $4,472 at the example prices.
Output fell 5.8% (591,000 to 557,000 pieces). Energy per piece rose 1.2%, from 0.02254 to 0.02281 MMBtu.
Month 22: output 8.9% below the same month a year earlier, energy only 0.3% below. Expected 1,110.8 MMBtu, used 1,157.0: 46.2 above (4.2%), flagged. Months 17, 20 and 24 were more than 2 standard errors below expected.

The year is a 2.2% improvement even though energy per piece went up 1.2%: the fixed load does not fall when output does. Month 22 is the one to investigate.

Common mistakes

Judging the plant by energy per unit alone
When output falls, the fixed load is spread over fewer units and energy per unit rises. A plant that cut its fixed load can still show a worse number. Look at actual against expected.
A baseline from an unusual year
A long shutdown, a start-up year or a year before a new line gives a model of a plant that no longer exists. Pick 12 months that describe how the plant runs now.
Adding a variable because it is available
A driver that does not move the energy fails the p value tests and makes the model worse to use. Degree days on an electric load that does not depend on weather is the usual one.
Changing the factors or the degree day source in the middle
A different multiplier, a different heat content or a different base temperature after month 12 makes the reporting months incomparable with the baseline. Keep them fixed or restart the baseline.
Reading 5 months as a year
The DOE protocol compares 12 months with 12 months so that every season is covered. Fewer months can show a saving that is only the weather.

Download the template

Free to use: print it, copy it and edit it for your team. Enter your name and work email once to download.

Run this template in LeanSuite

LeanSuite's production dashboard tracks OEE and its three factors next to downtime by reason and output against plan, and KPI values can be pulled on a schedule from Excel on SharePoint or OneDrive or from Google Sheets.

See it in a 45-minute demo

In the lean glossary

Free tools

Guides

FAQ