OEE Explained: What the Number Tells You and What It Hides
What is OEE?
Overall equipment effectiveness, or OEE, measures how much of a machine's or a line's planned production time turns into good output. It combines three kinds of loss into one number:
- Availability: the share of planned production time the equipment was actually running
- Performance: how fast it ran while it was running, compared with its ideal rate (parts produced × ideal cycle time, divided by run time)
- Quality: the share of its output that was good first time
OEE is the three multiplied together. A line that runs most of the time, close to its ideal speed, with very little scrap, has a high OEE. A weakness in any one of the three pulls the whole number down.
How is OEE calculated?
The example below shows the arithmetic for one shift. The figures are illustrative, chosen to show the calculation, and do not describe any real plant.
| Measure | Illustrative shift | Result |
|---|---|---|
| Planned production time | 480 minutes | — |
| Stops (breakdowns and changeover) | 60 minutes | — |
| Run time | 420 minutes | Availability = 420 / 480 = 87.5% |
| Ideal cycle time | 1 minute per part | — |
| Parts produced | 380 | Performance = (380 × 1) / 420 = 90.5% |
| Good parts | 361 | Quality = 361 / 380 = 95.0% |
| OEE | — | 87.5% × 90.5% × 95.0% = 75.2% |
There is a quick way to check the result. Multiply the good parts by the ideal cycle time and divide by the planned production time. Here, 361 good parts at one minute each is 361 minutes of good output in a 480-minute plan, which is also 75.2%.
What does a single OEE number hide?
One OEE figure is useful for tracking the same line over time. On its own, it can mislead.
- Different losses can give the same number. A line with frequent breakdowns and a line that runs slowly can show the same OEE, and the fixes are completely different. Always look at the three components, not only the total.
- Definitions change the result. Whether planned maintenance, changeovers or waiting for material count as lost time changes the figure. If two sites define planned production time differently, comparing their OEE says more about their definitions than their performance.
- The ideal cycle time can be set wrong. If it is set to the speed the line usually runs rather than the speed the equipment is designed for, performance looks better than it is and speed losses never show up.
- Manual records miss short stops. Stops of a minute or two are easy to forget when downtime is written on paper at the end of a shift, yet they add up. They are hard to capture reliably without collecting data directly from the machine.
- Averages smooth over problems. A weekly average can hide one shift or one product that performs very differently from the rest.
Should you compare your OEE with other plants?
Be careful. Benchmark figures for "world-class" OEE are widely quoted, but they rarely come with a stated definition or data set, and they assume your products, changeovers and equipment look like someone else's. Your own line's trend, measured the same way every week, is a far better guide. Improving the machine that limits the whole line usually matters more than raising OEE everywhere.
How do you get OEE data you can trust?
- Agree the definitions first. Write down what counts as planned production time, which stops are excluded, and the ideal cycle time for each product.
- Take run, stop and count signals from the machine. The PLC often has them already. The work is getting them off the floor cleanly and naming them consistently.
- Record a reason for each stop. Machine signals tell you that it stopped. An operator's reason code tells you why. Keep the list of reasons short enough that operators will actually use it.
- Store the data once and build every report from the same model. When maintenance, production and management each pull from their own source, the numbers never agree.
- Check the data against the floor. Before anyone makes a decision from a dashboard, compare it with a shift someone actually watched.
Our IIoT and plant data layer service is built on the same principles: machine data taken off the floor once into a single model, and missing or drifting instrumentation fixed before anything is plotted. For how machine data is read in a consistent way, see OPC UA explained for plant managers.
Where does OEE fit in an automation project?
OEE is often one of the numbers an automation project is judged on, so it should be measured before the project starts, in the same way it will be measured afterwards. That baseline belongs in the assessment, and the signals and definitions behind it belong in the functional design specification. Our guide to what a functional design specification should contain covers that document.
Related reading: capturing machine downtime without operator guesswork, collecting data from an existing PLC and when a plant needs a data historian.
Not sure your OEE figures can be trusted? Our process starts with a fixed-fee plant assessment that measures the baseline before anything is proposed. Questions about network security, ownership and support are answered in our FAQ.