OEE explained: formula, worked example, common errors

Overall equipment effectiveness, or OEE, is one number that says how much of your planned production time turned into good parts at full speed. It is built from three factors that each answer a separate question, and its main use is to show you where the lost time went.

The formula

OEE = availability x performance x quality

• Availability = run time / planned production time. Did the machine run when it was supposed to?
• Performance = (ideal cycle time x total count) / run time. While it ran, did it run at full speed?
• Quality = good count / total count. Of what it made, how much was good?

Planned production time is the time you meant to run, such as a shift less its scheduled breaks. Run time is planned production time minus downtime. Ideal cycle time is the fastest time per part the machine can sustain while making good parts. Total count is everything made, good or bad.

A worked example

These are made-up numbers for one machine on one shift.

• Planned production time: 450 minutes, an 8-hour shift less 30 minutes of breaks.
• Downtime: 45 minutes, a changeover and a jam.
• Ideal cycle time: 30 seconds per part.
• Total count: 740 parts.
• Good count: 715 parts.

Run time is 450 minus 45, which is 405 minutes. Availability is 405 / 450 = 0.900, or 90.0 percent.

At the ideal rate, 740 parts should take 740 x 30 seconds = 22,200 seconds, or 370 minutes. Performance is 370 / 405 = 0.914, or 91.4 percent.

Quality is 715 / 740 = 0.966, or 96.6 percent.

OEE is 0.900 x 0.914 x 0.966 = 0.794, about 79.4 percent.

There is a shortcut that makes a good check. Good count x ideal cycle time / planned time = 715 x 30 seconds / 450 minutes = 357.5 / 450 = 0.794. The same answer comes out because run time cancels. You can try your own numbers in the OEE calculator.

Read the result as minutes. Of 450 planned minutes, only 357.5 minutes’ worth of good parts were made. The other 92.5 minutes went like this: 45 to downtime, 35 to running slower than ideal (405 minus 370), and 12.5 to the 25 bad parts (370 minus 357.5).

The six big losses

Lean and TPM practice sorts losses into six groups, and each group belongs to one factor.

What each factor points to

A low factor tells you which kind of loss to look for first.

• Low availability points to breakdowns, changeovers and waiting. Look at the stop log.
• Low performance points to small stops and slow running. Look at how often the machine halts for a minute or two, and at whether operators run it below its rated speed.
• Low quality points to scrap and rework. Look at when the rejects happen, at startup or throughout the run.

Two machines can show nearly the same OEE for different reasons. Machine A runs at 70.0 percent availability, 95.0 percent performance and 98.0 percent quality, which gives 65.2 percent. Machine B runs at 95.0, 75.0 and 93.0 percent, which gives 66.3 percent. The headline numbers are close, but A needs a maintenance and changeover plan, while B needs a look at its speed settings, its small stops and its rejects.

Setting up the data

• Start with the bottleneck. An hour gained there is an hour gained for the plant, while an hour gained elsewhere may only add inventory.
• Count parts at the machine, ideally with a counter, not from an estimate at the end of the shift.
• Record every stop with a start time, an end time and a reason from a short list the operators agree on.
• Decide what counts as planned time and write it down.

Before you trust the number

A few checks catch most data problems.

• Good count x ideal cycle time can never be larger than planned production time. If it is, the ideal cycle time or the count is wrong.
• Compare the machine’s counter with a physical count of one pallet or one bin over a shift. Counters that read high or low skew every factor.
• Add up the stop log. The minutes should match planned time minus run time, or you are missing stops or counting some twice.
• Make sure reworked parts are counted as good only once they have been fixed and passed inspection, and that a part is not counted as good at one station and scrap at the next.
• Run the same shift through the numbers by hand once, as in the example above, so you know the system’s answer is the one you meant to ask for.

Common mistakes

• Guessing the ideal cycle time. Use the rated speed or the best sustained run, not the average. If performance comes out above 100 percent, the ideal is set too slow.
• Mixing definitions. If one line treats changeovers as planned time and another treats them as downtime, their OEE numbers cannot be compared.
• Averaging percentages. To combine machines, add up the underlying times and counts and work out the factors from the totals. Averaging each machine’s percentage gives a short-running machine the same weight as a long-running one.
• Holding different machines to one target. A press with long changeovers and a packaging line with short ones have different ceilings.
• Chasing the number. OEE tells you which factor lost the time. The cause is in the stop reasons.
• Quoting a benchmark you cannot source. Compare against your own baseline.

Using OEE well

Collect a baseline for a few weeks before changing anything. Then add up the stop minutes by reason, sort them from largest to smallest, work on the top one, and measure again. Many of the top reasons turn out to be maintenance items, which is where the preventive maintenance checklist helps, and others turn out to be things the next shift never heard about, which is what the shift handover reference is for.