What is OEE?
OEE stands for Overall Equipment Effectiveness. It tells you what share of a machine's planned time was spent making good parts at full speed. It is the product of three numbers: availability, performance and quality. A machine that is busy all day can still have a low OEE.
Every plant knows the feeling. The machine ran all day. Everyone was working. And still the quantity is short at dispatch.
The log shows one breakdown, yet the output is down by a fifth. The machine is "fine", but nobody knows how fine. And the owner is thinking about buying a new machine, without knowing whether the old one is really full.
Busy is not the same as productive. OEE shows the difference.
What are the three parts?
OEE answers three questions about one machine.
| Part | Question | Formula |
|---|---|---|
| Availability | Did it run? | Running time ÷ planned time |
| Performance | Did it run at speed? | (Parts made × ideal cycle time) ÷ running time |
| Quality | Were the parts good? | Good parts ÷ parts made |
OEE = Availability × Performance × Quality
100% would mean only good parts, at the ideal speed, with no stops, for the whole planned time. Every real machine is below that. The gap is the loss, and OEE shows where it is.
A worked shift
This is an example with made-up figures. A CNC lathe has a planned time of 450 minutes in a shift. The ideal cycle time is 8 minutes a part.
- The machine was stopped for 68 minutes (a breakdown, a tool change and a wait for the crane). So it ran for 382 minutes.
- Availability = 382 ÷ 450 = 85%.
- In the running time, the lathe made 41 parts. At 8 minutes each, that is 328 minutes of work. Performance = 328 ÷ 382 = 86%.
- One of the 41 parts was rejected. So 40 were good. Quality = 40 ÷ 41 = 98%.
- OEE = 0.85 × 0.86 × 0.98 = about 71%.
A quick check: good parts × ideal time ÷ planned time = 40 × 8 ÷ 450 = 71%. If the two answers do not match, there is a mistake somewhere. But the three parts are what tell you which loss to attack.
What are the six big losses?
There are two losses in each part:
- Availability: breakdowns, and setup and adjustment.
- Performance: small stops, and reduced speed.
- Quality: start-up rejects, and running rejects.
In the example, the log showed 68 minutes of stops. That is easy to see. But the slow running and the small stops were invisible in the log. They only showed up when the team counted parts against the ideal time.
When the team watched the machine for half an hour, they saw the chip conveyor jam three times, and the feed set low. Asked why, the operator explained that the tool chipped at full feed. A better tool grade fixed both problems.
Nobody was hiding anything. The operator turned the feed down for a good reason. The log simply could not see it. Fix the cause, never blame the operator.
Which time do we count as planned?
Keep your planned time honest. Count the working time of the shift, after fixed breaks. Take out only the time when there is no order for the machine, and report that separately.
Do not take setup time out of the planned time. A shop that removes setups from the plan shows a fine OEE and never works on its setups.
For a week or a month, work from totals. Add all good parts × ideal time, and divide by all planned minutes. Do not average the daily percentages.
What does a point of OEE give you?
If the machine decides the output of the whole plant, every extra good part is an extra part sold, if you have orders. If the machine is not the bottleneck, the gain is still real, but it shows up differently: less overtime, faster delivery, and a later need for a new machine.
Before you buy another machine, ask three things. How much does the old machine really produce? Which of its losses can be removed? What would the same money do if spent on the losses?
Busy is not the same as productive. OEE shows the difference.
How to start in your plant
- Choose one machine, the one that limits your output.
- Fix the planned time and the ideal cycle time for the main part.
- Every shift, write down the stops with a reason code, even the short ones.
- Count the parts made and the parts rejected.
- Work out availability, performance, quality and OEE for the shift.
- Once a week, make a Pareto chart of the loss reasons. Work on the biggest one first.
- Tell the team clearly that nobody will be blamed for what the data shows.
The question at the review is always "what stopped us?", never "who stopped us?" Use the record for training, never for blame.
Key points
- OEE = availability × performance × quality, for one machine.
- A busy machine can still lose a third of its planned time.
- The log shows breakdowns; only counting parts shows slow running and small stops.
- Do not take setup time out of the planned time.
- Measure the bottleneck first, and use OEE to find losses, not to blame people.
