The Blog of operational excellence and energy management for industry

Browse our articles, expert advices and clients testimonials: our experience save your time!

Blu.e Wiki: measuring improvements in energy performance in industry with the IPMVP

The International Performance Measurement and Verification Protocol (IPMVP) is the methodological framework recommended by ADEME (the French Environment and Energy Management Agency) for measuring the energy efficiency of buildings. It is also very widely used in industry, but requires a number of adjustments for use in an industrial setting. Here are a few guidelines for adapting the protocol to the specific features of a factory whose production plant is being changed to cater for new products or organisational changes.


The principles of the IPMVP protocol

“Energy savings cannot be directly measured, because savings represent the absence of energy consumption.” Accordingly, the method consists in using the energy consumed during a baseline period to model the adjusted baseline consumption, i.e. the energy that would have been consumed during the period of analysis if nothing had been done.

However, Energy Conservation Measures (ECMs) have been implemented in the factory: work may have been done or line-of-business initiatives taken, either with or without CAPEX, such as changing a boiler, advanced system regulation, insulating a heating or cooling network, setting up a predictive tool to assist with management, making the associated changes, etc.

The method factors in all of the variables affecting energy consumption at an industrial facility, in order to measure and verify the financial gains achieved through each of the ECMs.

Setting up an Energy Information System is a vital prerequisite for collecting logs and the site’s exogenous and endogenous data for use in constructing models.


IPMVP Protocol Industry - Measure improvements energy performance


5 tips for adjusting the protocol to industry

The IPMVP protocol is usually carried out in 13 steps. Here is a close-up on the five key steps for its deployment in an industrial setting.

1. Choose Option D

The IMPVP proposes a choice of four methods (A, B, C and D) for determining savings. Option D is the one best suited to industry. It uses a simulation tool, like those offered by Blu.e, to determine the savings, and can factor in structural changes to the industrial tool. However it requires the model to be accurately modelled on an adequate data history.

2. Define the baseline period

The baseline period is used to construct the model. It must be representative of the site’s standard operation, so must cover all of its states (seasonality, production cycles, etc.). The data history to be recovered will vary from one case to another. You may need to exclude certain specific periods, such as a month in which the plant operated in downgraded mode because some of the equipment had broken down.

3.  Identify the impact variables

To construct the model, you will need to make an exhaustive list of all of the endogenous and exogenous impact variables that might affect energy consumption (volume, throughput and type of production, calendar, weather conditions, area of the factory used, work in progress, quality of the raw materials, energy prices, etc.). Once this selection has been made, an impact estimation algorithm can be used to assign a weighting to these variables in the model (or even remove them from the model).

4. Divided into phases and compose the models

The models will have to be constructed separately for each phase of the factory or workshop’s operation, depending on the level of detail required (production, non-production, transitional phases, annual idle time, etc.), The detail of the phases and the rules for the division into phases should be worked out with the site’s teams. After that, make linear combinations of the factors chosen for each phase.

5. Work out the required degree of precision

It is difficult to obtain models that are accurate to within 1%, and hence to measure ECMs that represent a few per cent of savings. There may not be any point in looking for the highest possible R² (the correlation coefficient), even if it means using variables that have a smaller impact. It is important to choose the right balance between representative variables in the models, and an acceptable R². Certain statistical tools, such as the p-value, can help with this choice.