Metrics
OEE in practice: the six big losses and where maintenance comes in
How to calculate OEE with a complete example, the six big losses of TPM, a quick cross-check of the math and the mistakes that hide the real loss.
An OEE of 100% means the machine produced for the entire scheduled time, at rated speed, without a single bad part. No machine gets there. What OEE shows is how far you are from it and, above all, where the loss is.
OEE (Overall Equipment Effectiveness) was born alongside TPM, Total Productive Maintenance, in the 1970s. It isn't a maintenance-only metric, but a good share of the losses it points to pass through the hands of the people who look after the machine.
The three factors
OEE = Availability × Performance × Quality
- Availability: of the time scheduled for production, how much the machine actually ran. It drops with stops.
- Performance: while running, how much it produced compared with what it would at rated speed. It drops with slow running and minor stops.
- Quality: of what it produced, how much came out good the first time. It drops with scrap and rework.
Because the three are multiplied, small losses in each one add up to a big loss overall. 90% × 90% × 90% is 73%, not 90%.
The six big losses
TPM splits what pulls OEE down into six types, two per factor:
| Factor | Loss | Shop-floor example |
|---|---|---|
| Availability | 1. Breakdowns and failures | Burned-out motor, seized bearing, hydraulic leak |
| 2. Setup and adjustments | Format changeover, tool change, adjustments after the change | |
| Performance | 3. Minor stops | Dirty sensor, part jammed on the conveyor, an alarm the operator resets |
| 4. Reduced speed | Machine "held back" below rated speed because of vibration, noise or fear of a breakdown | |
| Quality | 5. Process defects | Out-of-tolerance parts from play, a worn tool, unstable temperature |
| 6. Startup losses | Parts scrapped until the machine stabilizes after startup or a setup |
Worked example
A fictional filling machine, on an 8-hour shift:
- Shift: 480 min. Planned break (meal): 30 min. Scheduled time: 450 min.
- Stop due to failure: 35 min. Setup: 25 min. Run time: 390 min.
- Rated speed: 60 units per minute (ideal cycle time of 1 second).
- Produced: 20,300 units. Good: 19,700.
| Factor | Math | Result |
|---|---|---|
| Availability | 390 ÷ 450 | 86.7% |
| Performance | 20,300 ÷ (390 × 60) | 86.8% |
| Quality | 19,700 ÷ 20,300 | 97.0% |
| OEE | 0.867 × 0.868 × 0.970 | 73.0% |
To check the math: OEE also equals good parts divided by what the machine would make over the entire scheduled time at rated speed. Here, 19,700 ÷ (450 × 60) = 19,700 ÷ 27,000 = 73.0%. If the two paths don't match, some of the data is wrong.
On this shift, performance lost as much as availability. Almost 3,100 units never came out while the machine was on, because of slow running and minor stops nobody recorded as a "stop". You can try your own numbers in the OEE calculator, which shows which factor weighs the most.
Where maintenance comes in
Breakdowns are the obvious loss, but maintenance influences almost all the others:
- Breakdowns: well-sized preventive maintenance, predictive maintenance on critical assets and root cause analysis on repeat failures.
- Setup: quick-clamping fixtures, preset tools, marked settings. Maintenance is often production's partner in reducing setup time.
- Minor stops: these are the blind spot. They last seconds or minutes, the operator handles them alone and nobody opens a work order. Added up, they become hours. Cleaning and adjusting sensors, guides and conveyors solves many of them.
- Reduced speed: when the machine vibrates or makes noise, the operator slows it down "so it won't break". The mechanical problem stays hidden inside performance. Ask the operator why the machine runs below rated speed.
- Defects and startup losses: play, tool wear and unstable temperature control produce bad parts. Often the cause is mechanical.
Common mistakes
- Using the "usual" speed instead of rated speed. If the machine has always run at 50 and you use 50 as the reference, the speed loss disappears from the metric. Use design capacity or the best sustained speed ever demonstrated.
- Not recording short stops. Without that record, they only show up as low performance, with no cause.
- Comparing OEE across different machines. A press and a filling machine live in different realities. Compare a machine with itself over time.
- Taking a simple average of OEE across machines. A bottleneck machine with low OEE matters more than an auxiliary one with high OEE. Look at the bottleneck first.
- Looking only at the final number. An OEE of 73% on its own doesn't tell you what to do. The three factors and the six losses do.
A widely cited reference is 85% as world-class OEE. Treat it as a direction, not a target for every machine. Many lines run between 50% and 70%, and going from 60% to 70% is already a huge production gain without buying a new machine.
In taskfield, the downtime recorded on corrective work orders feeds each asset's availability, and the history shows which failures took the machine out of operation most often in the period.
Summary
- OEE = availability × performance × quality; small losses in each factor multiply.
- The six big losses: breakdowns, setup, minor stops, reduced speed, defects and startup losses.
- Check the math: good parts ÷ (scheduled time × rated speed) should give the same OEE.
- Minor stops and reduced speed hide mechanical problems that nobody records.
- Use rated speed as the reference and compare a machine with itself.
Calculate it now. Run the numbers with your own data in the free calculators: