OEE Calculation: Formula, Examples and Where AI Helps
OEE Calculation: Formula, Examples and Where AI Helps
OEE calculation multiplies availability, performance and quality, so a line at 90%, 95% and 99% scores about 85% OEE, the usual world class mark.
Your plant reports an OEE number every week, but nobody on the floor trusts it. Downtime reasons get typed in at shift end, short stops vanish, and scrap counts arrive a day late.
This guide walks through the OEE formula with a worked line example. Then it shows where manual data bends each factor, and how machine data and AI-based loss tagging make the number good enough to act on.
Key takeaways:
OEE = Availability x Performance x Quality. A worked shift below scores 74.8%.
85% OEE is the long-standing world class benchmark. Many discrete plants run closer to 60%.
Manual logs tend to miss short stops, which makes availability look better and performance worse.
TEEP adds planned downtime and idle shifts, so it shows how much capacity you really have.
Automatic data capture and AI loss tagging fix the inputs, so the OEE number holds up in a meeting.
OEE calculation multiplies availability, performance and quality, so a line at 90%, 95% and 99% scores about 85% OEE, the usual world class mark.
Your plant reports an OEE number every week, but nobody on the floor trusts it. Downtime reasons get typed in at shift end, short stops vanish, and scrap counts arrive a day late.
This guide walks through the OEE formula with a worked line example. Then it shows where manual data bends each factor, and how machine data and AI-based loss tagging make the number good enough to act on.
Key takeaways:
OEE = Availability x Performance x Quality. A worked shift below scores 74.8%.
85% OEE is the long-standing world class benchmark. Many discrete plants run closer to 60%.
Manual logs tend to miss short stops, which makes availability look better and performance worse.
TEEP adds planned downtime and idle shifts, so it shows how much capacity you really have.
Automatic data capture and AI loss tagging fix the inputs, so the OEE number holds up in a meeting.
What Is OEE and Why Does It Matter?
What Is OEE and Why Does It Matter?
OEE stands for Overall Equipment Effectiveness, and it shows the share of planned production time that makes good parts at full speed.
Seiichi Nakajima made OEE popular as part of Total Productive Maintenance in the 1980s. It's still the most common way plants measure machine productivity.
The overall equipment efficiency formula has three factors:
Availability: run time divided by planned production time. It captures breakdowns and changeovers.
Performance: actual output speed against ideal cycle time. It captures short stops and slow running.
Quality: good parts divided by total parts. It captures scrap and rework.
The importance of OEE comes from what it hides less than other metrics. Output alone can look fine while a line loses a quarter of its time. For the basics in plain terms, see our guide to OEE meaning and formula for plant managers.
OEE stands for Overall Equipment Effectiveness, and it shows the share of planned production time that makes good parts at full speed.
Seiichi Nakajima made OEE popular as part of Total Productive Maintenance in the 1980s. It's still the most common way plants measure machine productivity.
The overall equipment efficiency formula has three factors:
Availability: run time divided by planned production time. It captures breakdowns and changeovers.
Performance: actual output speed against ideal cycle time. It captures short stops and slow running.
Quality: good parts divided by total parts. It captures scrap and rework.
The importance of OEE comes from what it hides less than other metrics. Output alone can look fine while a line loses a quarter of its time. For the basics in plain terms, see our guide to OEE meaning and formula for plant managers.
OEE Calculation: The Formula With a Worked Example
OEE Calculation: The Formula With a Worked Example
OEE Calculation: How One Shift Gets to 74.8%
Each loss shrinks the share of planned time that makes good parts at full speed
The OEE formula is Availability x Performance x Quality, or simply good count x ideal cycle time divided by planned production time.
Here's an OEE calculation example for one 8-hour shift, the same one the figure above breaks down:
Input | Value |
|---|---|
Shift length | 480 minutes |
Breaks | 60 minutes |
Planned production time | 420 minutes |
Downtime (breakdowns, changeover) | 47 minutes |
Run time | 373 minutes |
Ideal cycle time | 1.0 second per part |
Total count | 19,271 parts |
Good count | 18,848 parts |
Now work out how to calculate OEE step by step:
Availability: 373 / 420 = 88.8%.
Performance: (1.0 s x 19,271) / (373 x 60 s) = 86.1%.
Quality: 18,848 / 19,271 = 97.8%.
OEE: 88.8% x 86.1% x 97.8% = 74.8%.
Check it the short way. 18,848 good parts x 1 second = 18,848 seconds, divided by 25,200 seconds of planned time, also gives 74.8%.
The OEE formula is Availability x Performance x Quality, or simply good count x ideal cycle time divided by planned production time.
Here's an OEE calculation example for one 8-hour shift, the same one the figure above breaks down:
Input | Value |
|---|---|
Shift length | 480 minutes |
Breaks | 60 minutes |
Planned production time | 420 minutes |
Downtime (breakdowns, changeover) | 47 minutes |
Run time | 373 minutes |
Ideal cycle time | 1.0 second per part |
Total count | 19,271 parts |
Good count | 18,848 parts |
Now work out how to calculate OEE step by step:
Availability: 373 / 420 = 88.8%.
Performance: (1.0 s x 19,271) / (373 x 60 s) = 86.1%.
Quality: 18,848 / 19,271 = 97.8%.
OEE: 88.8% x 86.1% x 97.8% = 74.8%.
Check it the short way. 18,848 good parts x 1 second = 18,848 seconds, divided by 25,200 seconds of planned time, also gives 74.8%.
Stop Guessing. Start Measuring.
See what downtime, energy waste, and manual processes are really costing your plant.
Stop Guessing. Start Measuring.
See what downtime, energy waste, and manual processes are really costing your plant.
Stop Guessing. Start Measuring.
See what downtime, energy waste, and manual processes are really costing your plant.
What Is a Good OEE Score?
What Is a Good OEE Score?
OEE benchmarks put world class at 85%, typical discrete plants near 60%, and lines under 40% in clear need of attention.
The 85% figure comes from Nakajima's targets: 90% availability, 95% performance and 99.9% quality. Few lines hit all three for long.
Don't chase a benchmark from another industry. A bottling line and a job shop can't share a target. Compare a line with itself over time, and compare lines that make similar products.
A drop from 75% to 70% on one line is worth a meeting. A gap between your plant and a blog's "world class" number isn't, until your data is clean.
OEE benchmarks put world class at 85%, typical discrete plants near 60%, and lines under 40% in clear need of attention.
The 85% figure comes from Nakajima's targets: 90% availability, 95% performance and 99.9% quality. Few lines hit all three for long.
Don't chase a benchmark from another industry. A bottling line and a job shop can't share a target. Compare a line with itself over time, and compare lines that make similar products.
A drop from 75% to 70% on one line is worth a meeting. A gap between your plant and a blog's "world class" number isn't, until your data is clean.
OEE Benchmarks at a Glance
Compare a line with itself over time before comparing it with these marks
Where Manual Data Distorts Your OEE Calculation
Where Manual Data Distorts Your OEE Calculation
Manual OEE data hides short stops, guesses reason codes and delays scrap counts, which can swing the reported number by 10 points or more.
Here's what goes wrong with each factor when people log it by hand:
Factor | Manual logging problem | Effect on OEE |
|---|---|---|
Availability | Stops under a few minutes go unrecorded | Availability looks too high |
Performance | Ideal cycle time set too slow, or never updated | Performance looks too high |
Quality | Scrap counted at shift end or next day | Quality lags reality |
Reason codes | "Other" or "mechanical" chosen for speed | No clear fix to work on |
The ideal cycle time is the quiet one. Many plants use the speed the line usually runs, not the speed it was designed for. That makes performance look close to 100% while the line loses real output.
Lost time often lands in unplanned downtime, and that's where the money goes.
Manual OEE data hides short stops, guesses reason codes and delays scrap counts, which can swing the reported number by 10 points or more.
Here's what goes wrong with each factor when people log it by hand:
Factor | Manual logging problem | Effect on OEE |
|---|---|---|
Availability | Stops under a few minutes go unrecorded | Availability looks too high |
Performance | Ideal cycle time set too slow, or never updated | Performance looks too high |
Quality | Scrap counted at shift end or next day | Quality lags reality |
Reason codes | "Other" or "mechanical" chosen for speed | No clear fix to work on |
The ideal cycle time is the quiet one. Many plants use the speed the line usually runs, not the speed it was designed for. That makes performance look close to 100% while the line loses real output.
Lost time often lands in unplanned downtime, and that's where the money goes.
Unplanned Downtime in Manufacturing: The Real Cost
Unplanned Downtime in Manufacturing: The Real Cost
Where AI Helps With OEE Monitoring
Where AI Helps With OEE Monitoring
AI improves OEE monitoring by reading machine signals directly, tagging each stop with a likely cause and catching quality drift early.
AI doesn't change the formula. It fixes the inputs.
Automatic capture: PLC signals, sensors and counters feed run time and part counts every second. Short stops show up.
Loss classification: a model groups stops by pattern and suggests a reason code. Operators confirm with one tap.
Vision for quality: cameras count rejects as they happen, so scrap shows up in the shift it happens.
Early warnings: drift in cycle time or vibration flags a likely breakdown, which ties into predictive maintenance.
The trap is buying AI before the basics work. If your PLCs don't expose a running signal, start there. A clean signal from 10 machines beats a smart model guessing on 100.
AI improves OEE monitoring by reading machine signals directly, tagging each stop with a likely cause and catching quality drift early.
AI doesn't change the formula. It fixes the inputs.
Automatic capture: PLC signals, sensors and counters feed run time and part counts every second. Short stops show up.
Loss classification: a model groups stops by pattern and suggests a reason code. Operators confirm with one tap.
Vision for quality: cameras count rejects as they happen, so scrap shows up in the shift it happens.
Early warnings: drift in cycle time or vibration flags a likely breakdown, which ties into predictive maintenance.
The trap is buying AI before the basics work. If your PLCs don't expose a running signal, start there. A clean signal from 10 machines beats a smart model guessing on 100.
How to Improve OEE on One Line First
How to Improve OEE on One Line First
Improving OEE starts with the biggest loss on one constraint line, fixed with clean data, and then the next loss, not every metric at once.
Use the six big losses to sort where time goes: breakdowns, setup and changeover, short stops, slow running, startup rejects and production rejects.
A practical order:
Make data capture automatic on the constraint line.
Rank losses by minutes lost, not by how often they occur.
Fix the top loss with maintenance, setup work or process changes.
Re-measure for four weeks before moving on.
Many plants find their first 5-10 points of OEE in changeovers and short stops. Those are cheap fixes once you can see them.
Improving OEE starts with the biggest loss on one constraint line, fixed with clean data, and then the next loss, not every metric at once.
Use the six big losses to sort where time goes: breakdowns, setup and changeover, short stops, slow running, startup rejects and production rejects.
A practical order:
Make data capture automatic on the constraint line.
Rank losses by minutes lost, not by how often they occur.
Fix the top loss with maintenance, setup work or process changes.
Re-measure for four weeks before moving on.
Many plants find their first 5-10 points of OEE in changeovers and short stops. Those are cheap fixes once you can see them.
TEEP vs OEE: When Each Number Matters
TEEP vs OEE: When Each Number Matters
TEEP measures OEE against all calendar time, while OEE measures only planned production time, so TEEP shows hidden capacity.
TEEP = OEE x Utilization, where utilization is planned production time divided by all time. A line at 75% OEE running one shift a day has a TEEP near 25%.
That gap matters when someone proposes a new line. Before buying capacity, check if a second shift or fewer planned stops would cover the demand.
TEEP vs OEE isn't a contest. Use OEE to run the line well, and TEEP to decide how much line you need.
TEEP measures OEE against all calendar time, while OEE measures only planned production time, so TEEP shows hidden capacity.
TEEP = OEE x Utilization, where utilization is planned production time divided by all time. A line at 75% OEE running one shift a day has a TEEP near 25%.
That gap matters when someone proposes a new line. Before buying capacity, check if a second shift or fewer planned stops would cover the demand.
TEEP vs OEE isn't a contest. Use OEE to run the line well, and TEEP to decide how much line you need.
What to Look For in the Best OEE Software
What to Look For in the Best OEE Software
The best OEE software captures machine data automatically, shows losses by reason in real time and connects to your MES at a fair price.
Look for these features when you compare OEE software:
Direct links to PLCs and sensors, not just manual entry screens.
Reason codes operators can confirm in seconds on the line.
Dashboards by line, shift and product, with drill-down to each stop.
Links to your MES and maintenance system.
OEE software price per machine or per line that stays fair as you add lines.
Some plants get OEE from their MES. Others add a light OEE layer on top. Our guide to MES software for manufacturing plants helps you decide which fits.
The best OEE software captures machine data automatically, shows losses by reason in real time and connects to your MES at a fair price.
Look for these features when you compare OEE software:
Direct links to PLCs and sensors, not just manual entry screens.
Reason codes operators can confirm in seconds on the line.
Dashboards by line, shift and product, with drill-down to each stop.
Links to your MES and maintenance system.
OEE software price per machine or per line that stays fair as you add lines.
Some plants get OEE from their MES. Others add a light OEE layer on top. Our guide to MES software for manufacturing plants helps you decide which fits.
Frequently Asked Questions
How do you calculate OEE?
OEE is calculated as Availability x Performance x Quality. You can also divide good count x ideal cycle time by planned production time, which gives the same result.
What is a good OEE score?
An OEE of 85% is the common world class benchmark, and many discrete plants run near 60%. Compare each line with its own past results before using outside benchmarks.
What is the difference between OEE and TEEP?
OEE measures performance during planned production time, while TEEP measures it against all calendar time. TEEP shows how much extra capacity unused shifts and planned stops hide.
Why is my OEE number wrong?
OEE numbers often go wrong because short stops are not logged, the ideal cycle time is set too slow, or scrap is counted late. Automatic data capture from PLCs and sensors fixes most of these errors.
How does AI improve OEE?
AI improves OEE by capturing machine data automatically, suggesting reason codes for each stop and spotting quality drift early. It improves the inputs to the formula rather than changing it.
What are the six big losses in OEE?
The six big losses are breakdowns, setup and changeovers, short stops, slow running, startup rejects and production rejects. Each maps to one OEE factor, which helps you decide where to act first.
Conclusion
OEE calculation is easy, and the hard part is trusting the inputs, so fix data capture on one line before chasing a benchmark.
Pick your constraint line this week and check three things: the ideal cycle time, how short stops are logged and when scrap is counted. If you want help connecting PLC data to a live OEE view, the KGT team can scope a one-line pilot with you.
Conclusion
OEE calculation is easy, and the hard part is trusting the inputs, so fix data capture on one line before chasing a benchmark.
Pick your constraint line this week and check three things: the ideal cycle time, how short stops are logged and when scrap is counted. If you want help connecting PLC data to a live OEE view, the KGT team can scope a one-line pilot with you.
Sources:
Seiichi Nakajima - Introduction to TPM: Total Productive Maintenance
Vorne Industries - OEE Calculation and World Class OEE benchmarks
Society of Manufacturing Engineers - Total Productive Maintenance and the six big losses
Sources:
Seiichi Nakajima - Introduction to TPM: Total Productive Maintenance
Vorne Industries - OEE Calculation and World Class OEE benchmarks
Society of Manufacturing Engineers - Total Productive Maintenance and the six big losses
Industrial Autonomous Floor
Newsletter
Actionable insights on industrial AI, automation, and smart operations built for safe, secure, and compliant real-world environments.
