Statistician William Edwards Deming once said, *“If I had to reduce my message to management to just a few words, I’d say it all has to do with reducing variation”*. In fact, manufacturing has always sought to limit variability by establishing standards or adopting best practices. And yet when we take the necessary steps to monitor Energy Performance Indicators (EPIs) in operational increments – hourly or every 10 minutes – we realise that there are still variations: a wealth of information that manufacturers would be misguided to overlook.

**1- Definition**

**2- An infinite number of variables for a wealth of information**

**3- Four steps to continuous improvement**

**4- Tips for analysing variability**

## 1- Definition

Instability, fluctuation or *“state of that which is variable”*: according to the dictionary, variability is simply a *“tendency to vary”*. In statistical terms, variability is the *“dispersion of the values of a frequency distribution, often measured as the standard deviation”*, while the standard deviation is the mean deviation from the median. In other words, variability, calculated as a percentage, is measured by dividing the standard deviation by the mean.

## 2- An infinite number of variables for a wealth of information

**An alternative to Excel spreadsheets**

Any complex system is subject to variability. And the innumerable sources of variability in manufacturing make it a good place to start for manufacturers in search of energy savings. There is nothing new in that. What is new, however, is the advent of new technologies and in particular Big Data as an alternative to Excel spreadsheets for operational staff. In a matter of clicks, the latter can finally pinpoint the operating parameters behind the variability of the EPIs and take the appropriate action without having to call in a data scientist.

**Measurement**

The only stipulation is that, in order to visualise the variability, they need to use the smallest time segment possible for each performance indicator. This means that the variability deviation detected between a monthly measurement and a measurement every minute or every hour can vary in the ratio of one to three, if not more.

**Visualize**

Not forgetting the process data… Shown below in orange, the boiler room data can be used to track gas consumption per tonne of steam produced. Little or no variability can be seen. The steam used by the process is shown in blue. Here we can see that, to produce the same tonne of finished product, the manufacturer might use up to two or three times as much steam! The energy performance gains are therefore to be found in the efficiency with which the steam is used, not in the efficiency of its production. And by reducing variability, the manufacturer stands to gain not just a few per cent but tens of per cent.

## 3- Four steps to continuous improvement

In practice, it takes four steps to analyse the variability of your indicators and continuously improve your energy performance:

**1. Measure the potential gain** between the average performance and the best stable, realistic performance;

**2. Identify the control parameters and settings associated** with the best stable performance;

**3. Apply best practices for operation**/factory or workshop settings;

**4. Rapidly measure the effective application of the set values** and their impact on energy performance.

## 4- Tips for analysing variability

Before you can start analysing the variability of your industrial facility’s energy performance, the first step is to set up and structure an operational database. Our free guide “Industry 4.0: how to optimise your energy consumption” will show you how to set it up.

**Sources**

Blu.e guide: “Industry 4.0: how to optimise your energy consumption”