OEE Calculator — Availability, Performance and Quality
Overall equipment effectiveness multiplies three rates that each answer a different question: was the machine running, was it running at speed, and was what it made good. Enter the planned time, the downtime, the ideal cycle time, the units produced and the rejects, and this calculator returns all three plus the OEE they produce — and the minutes each one cost you.
Overall equipment effectiveness
—
Results are estimates for planning purposes. Verify with your supplier, engineer, or a qualified professional before ordering material or performing work.
The six losses, and which factor each one lands in
NIST lists the six losses total productive maintenance sets out to eliminate. Two land in availability, two in performance and two in quality — which is the whole reason for splitting OEE into three factors instead of reporting one number.
| Loss | Shows up in | What it looks like on the floor |
|---|---|---|
| Equipment breakdown | Availability | The machine stopped and could not run. The loss everyone counts, and usually the smallest of the six. |
| Setup and adjustment slowdowns | Availability | Changeovers, first-piece approval, tooling changes, waiting for material or an operator. |
| Idling and short-term stoppages | Performance | Minor stops nobody logs: a jam cleared in twenty seconds, a sensor wiped, a chute knocked free. They are invisible one at a time and enormous over a shift. |
| Reduced capacity | Performance | The machine runs, but below its design cycle — dull tooling, a cautious operator, a feed turned down months ago and never turned back. |
| Quality-related losses | Quality | Scrap and rework produced during steady running. |
| Startup and restart losses | Quality | Parts made while the process comes up to temperature, pressure or dimension after every stop. |
The six losses are named in NIST Advanced Manufacturing Series 100-18. Where each one lands, and the descriptions above, follow from the definitions of the three factors.
Two definitions decide your score before any machine does
OEE is comparable only against itself, and only when the same two rules are applied every shift.
- What counts as planned production time. Count scheduled breaks as planned time and availability drops; leave them out and it rises. Neither is wrong, and neither can be changed halfway through a year of data without making the trend meaningless.
- What the ideal cycle time is. It has to be the fastest the machine makes one good unit — the nameplate rate or the best cycle ever demonstrated. Set it from a historical average and performance sits near 100 % permanently, hiding the largest loss on most machines.
- Where the numbers come from. Short stops are the loss that is chronically under-reported, because nobody writes down twenty seconds. A machine whose performance rate is poor while its downtime log looks clean is usually telling you about stops that never got logged.
- Rework counts against quality, not performance, even when the part eventually ships.
Reading the three factors
The product is the headline, but the three factors are the useful part — a single OEE figure hides which of them moved.
- Availability low, performance and quality fine — breakdowns and changeovers. Maintenance and setup reduction, in that order.
- Performance low, availability fine — short stops and slow running. Walk the machine for an hour and count the stops nobody logs.
- Quality low — scrap during running, or startup parts after every stop. Chasing OEE by running faster makes this one worse.
- All three around 90 % — an OEE near 73 %. Multiplying rates is unforgiving, which is what makes the metric useful in the first place.
- Measure the constraint. High OEE on a machine that is not the bottleneck often means it is building inventory nobody ordered.
The part an uptime target can quietly cost
Availability rewards getting the machine running again quickly, and that pressure lands on the people who go inside it. Two things are worth watching while the number improves:
- Lockout does not get skipped for a short job. 29 CFR 1910.147 applies to servicing and maintenance where unexpected energization or the release of stored energy could injure someone — clearing a jam included. A machine that stops often is a machine people reach into often.
- Guards do not get left off to save the next thirty seconds. Writing the energy control steps down for the machine is what the lockout/tagout procedure builder is for.
How it's calculated
- Availability
A = (planned time − downtime) / planned timeNIST writes it as (Required Availability − Downtime) / Required Availability × 100. The numerator is the run time, sometimes called operating time.- Performance
P = (ideal cycle time × total units) / run timeHow much of the time it did run was spent producing at the design rate. Above 100 % means the ideal cycle time is wrong, not that the machine beat physics.- Quality
Q = (total units − defects) / total unitsFirst-time-through. A unit that needed rework is a quality loss even if it ships.- OEE
OEE = A × P × QThree rates multiplied, so a mediocre score in each gives a poor one overall: 90 % × 90 % × 90 % is 72.9 %.- Fully productive time
ideal cycle time × good unitsThe same result read as minutes: OEE × planned time. It is the only part of the shift that produced a good part at full speed, and the cross-check that the three losses add up.- Where the time went
planned = downtime + speed loss + quality loss + fully productive timeSpeed loss = run time − (ideal cycle time × total units). Quality loss = ideal cycle time × defects.
Frequently asked questions
Multiply three rates: availability, performance and quality. Availability is run time divided by planned time; performance is the ideal cycle time times the units produced, divided by run time; quality is good units divided by total units. A shift with 480 planned minutes, 60 minutes down, a one-minute cycle, 350 units and 12 rejects gives 87.5 % × 83.3 % × 96.6 % = 70.4 %.
No number on this page will tell you, on purpose. The figures quoted as world-class come from a copyrighted TPM book and depend entirely on how a plant defines planned time and ideal cycle time — two plants running the same machine equally well can report scores twenty points apart. The score that matters is your own, measured the same way every shift, moving in the right direction.
Because the ideal cycle time you entered is slower than the machine actually runs. Performance compares real output against the design rate, so if the design rate is set from an average rather than from the fastest demonstrated good cycle, the machine will beat it and the arithmetic will pass 100 %. Fix the cycle time; it also flatters availability calculations everywhere else.
It depends on your plant's definition, and the definition is what has to stay constant. Counting scheduled breaks as planned production time makes availability look worse; excluding them makes it look better without a single machine improving. Write the rule down, apply it every shift, and compare like with like.
Quality. The quality rate is first-time-through: a unit that needed a second pass consumed the cycle and did not come off right, whether or not it was eventually sold. Counting rework as good is the most common way an OEE number quietly stops being useful.
NIST lists them as the losses total productive maintenance sets out to eliminate: equipment breakdown, setup and adjustment slowdowns, idling and short-term stoppages, reduced capacity, quality-related losses, and startup or restart losses. The first two land in availability, the next two in performance, the last two in quality — which is why splitting OEE into its three factors tells you where to look.
It earns its keep on the constraint. On a machine that is not the bottleneck, a high OEE can simply mean it is building inventory nobody needs, and chasing the number there costs more than it returns. Measure the constraint carefully, and measure the rest only well enough to know when one of them becomes the constraint.
It changes availability first, and it changes performance next, because a machine that is deteriorating tends to slow down and to stop briefly long before it fails outright. It also means more time spent inside the machine — which is exactly when 29 CFR 1910.147 energy control applies, and when a rushed changeover becomes a lockout that never got done.
Sources & references
- NIST Advanced Manufacturing Series 100-18 — The Costs and Benefits of Advanced Maintenance in Manufacturing: OEE = Availability × Performance Rate × Quality Rate, with the three definitions and the six losses TPM targets
- NISTIR 7911 — Assessment of Real-Time Factory Performance Through the Application of Standardized Data: the overall equipment effectiveness index among the factory performance indicators in use
- 29 CFR 1910.147 — The control of hazardous energy (lockout/tagout): servicing and maintenance that an OEE program will make more frequent still requires energy control
Content checked against these sources — last reviewed September 1, 2026.