OEE performance = (ideal cycle time × total units made) ÷ run time. It is the same as actual rate ÷ ideal rate. Count good and bad units, use run time only (stops are already in availability), and never let the result go above 100%.
What OEE performance measures
Performance is the speed part of OEE. It asks one question: while the machine was running, how close did it run to its ideal rate?
It does not look at stops. Stops belong to availability. It does not look at bad parts either. Those belong to quality. Performance looks only at run time and the units made in it.
In Overall Equipment Effectiveness (R. C. Hansen, 2001) this factor is called the operating speed rate, and it is built on Nakajima's TPM formulas. Hansen sets the net operating rate to 1.0 by definition, so speed rate is the whole performance factor.
Worked example, step by step
Example: a packing line is scheduled for 480 minutes. It stops for 60 minutes in total. Its ideal cycle time is 1.5 seconds per unit, which is 40 units per minute. It makes 14,700 units.
| Step | What | Calculation | Result |
|---|---|---|---|
| 1 | Run time | 480 min scheduled − 60 min stops | 420 min |
| 2 | Time the units needed at the ideal rate | 14,700 units ÷ 40 units per min | 367.5 min |
| 3 | Performance | 367.5 ÷ 420 | 87.5% |
| 4 | Check: actual rate ÷ ideal rate | (14,700 ÷ 420 = 35 per min) ÷ 40 | 87.5% |
| 5 | Check: ideal cycle time ÷ actual cycle time | 1.5 s ÷ (420 × 60 ÷ 14,700 = 1.714 s) | 87.5% |
Three routes, one answer. That is a good habit: calculate it two ways and check they match.
The 12.5% is real time. It is 52.5 minutes that the line was "running" but not making units at full speed. That is the loss to hunt for.
Use all units, not only good units
Performance uses the total count: good plus bad. In the example, 14,112 of the 14,700 units were good, which is a quality of 96%.
If you divide only the good units by the ideal output, you get 14,112 ÷ (420 × 40) = 84%. But quality has already taken its 4%. You would count the same bad units twice.
The right chain is 87.5% availability × 87.5% performance × 96% quality = 73.5% OEE. Check it with Hansen's simpler method: good units ÷ ideal rate ÷ scheduled time = (14,112 ÷ 40) ÷ 480 = 352.8 ÷ 480 = 73.5%.
What counts as run time
Run time is scheduled time minus every stop you logged. Changeovers, breakdowns, waiting for material and so on are already in availability. Do not take them out a second time.
The catch is the stops you did not log. A jam cleared in ten seconds leaves no record, so the line looks like it ran. The lost time stays inside run time and shows up as a lower rate. That is why performance is really "slow running plus small stops".
This is the link to the hidden factory. If your performance is low and you cannot see why, small stops are the first suspect. How to reduce micro-stops shows what to do about them.
Several products on one machine
Do not average the rates. Add up the ideal time each product needed, then divide by run time.
Example: in 400 minutes of run time, a machine makes 6,000 units of product A (ideal rate 50 per minute, so 0.02 min each) and 3,000 units of product B (ideal rate 20 per minute, so 0.05 min each).
Use one ideal rate for every product of the same size and format. Hansen says to keep that family rate even when one product is harder to run, and to note the reason. That makes a difficult product visible instead of hiding it.
Why the result is capped at 100%
If a line makes 20,160 units in 480 minutes of run time at an ideal rate of 40 per minute, the sum gives 20,160 ÷ 19,200 = 105%. That cannot be right, because the ideal rate is supposed to be the best the line can do.
Either the ideal rate is too low, or the counter is wrong. Cap performance at 100%, then fix the cause. On a machine that is not the bottleneck, the ideal rate is the rate needed to feed the bottleneck, so a faster machine simply caps at 100%. More on that in ideal cycle time.
How to measure it on the floor
You need three numbers per shift or per run:
- Total units from a counter at the machine output, good and bad together.
- Run time from an hour meter or a stop log. Scheduled time minus stops.
- Ideal rate for that product family, written down and agreed.
Hansen notes that cycle counters, run-hour clocks and simple chart recorders are enough to start. Then check the counter against the run clock: if the two disagree, one of them is wrong.
Our own tip: plot the rate per hour. A flat line below the ideal rate means slow running. A line that jumps up and down, with dips to zero, means many short stops. The fix is different for each.
Common mistakes
- Using the nameplate speed that the line never reached. Performance looks poor and nobody trusts it.
- Using last month's average rate as ideal. Performance then reads close to 100% and hides the loss.
- Dividing good units by run time and counting quality twice.
- Removing logged stops again. They are already in availability.
- Averaging rates of different products instead of adding ideal time.
- Leaving it above 100% without checking the ideal rate.
The full list of calculation errors is in OEE calculation mistakes. To try your own numbers, use the OEE calculator. To see where performance sits among the other losses, read the six big losses.
Performance is only as good as the count and the run time behind it. Fabrico takes stops, speed and counts live from PLCs or IoT sensors (external sensors on older machines), so the rate is not an end-of-shift guess. See how Fabrico connects machines.
What people ask next
Can OEE performance be over 100%?
Not in a correct calculation. If you get more than 100%, the ideal rate is set too low or the count is wrong. In Hansen's method the speed factor is capped at 100%. Re-check the ideal cycle time and note the change.
Is performance the same as speed loss?
Speed loss is the part of performance you lose by running slower than the ideal rate. Performance loss is 100% minus performance, so it also holds short stops that nobody logged. Read more in speed losses and micro-stops.
Do I use good units or all units?
Use all units made, good and bad. Bad units are counted in quality, and counting them twice makes OEE look worse than it is. See how to calculate OEE quality.
What if I do not know the ideal rate?
Hansen's fallback is the rate from the best 4 hours of the last 400 hours of running. Use it until you can accredit a proper rate. The full method is in ideal cycle time.
Where does performance fit in the full OEE formula?
OEE is availability × performance × quality. Performance is the middle factor. The full formula and a worked example are in how to calculate OEE, and the three factors are explained in availability, performance and quality.
Related guides: Production Efficiency vs OEE: How They Differ · Fully Productive Time: Definition and Waterfall · Net Operating Time: Definition vs Run Time · OEE vs Machine Efficiency: Why They Disagree · Speed Loss Calculation in OEE: Step by Step