How to calculate OEE: availability, performance and quality
Calculate overall equipment effectiveness from shift data: a worked example, how to combine shifts, SUMIFS formulas and the mistakes to avoid.
Overall equipment effectiveness (OEE) is availability × performance × quality. For a 480-minute shift with 30 minutes of breaks, 47 minutes of unplanned stops, a 1.0 second ideal cycle and 21,450 units made, of which 21,020 were good, OEE is 77.9%. The calculation needs four counts that most production lines already record: planned time, stop time, total units and good units. The input that needs the most care is the ideal cycle time.
This article defines each term, works one shift through to the result, explains why several shifts should be combined by adding minutes and counts instead of averaging percentages, and sets out a spreadsheet layout and the mistakes that most often distort the figure. The shift data is illustrative.
What are the definitions and formulas behind OEE?
- Planned production time is the time the line is scheduled to produce. Scheduled breaks and planned maintenance come out of the shift length first.
- Run time is planned production time minus unplanned stops, such as breakdowns, material shortages and unscheduled changeovers.
- Ideal cycle time is the fastest time to make one unit under ideal conditions, in seconds. It belongs to a product, not to a line.
- Total count is every unit that came off the line in the period, good or bad.
- Good count is the units that met specification the first time, without rework.
The three factors follow from those definitions:
- Availability:
run time / planned production time - Performance:
(ideal cycle time × total count) / (run time × 60) - Quality:
good count / total count
The 60 converts the ideal cycle from seconds to minutes so that it matches the run time. OEE is availability × performance × quality.
What does a worked OEE calculation look like?
The inputs for one shift:
| Input | Value |
|---|---|
| Shift length | 480 minutes |
| Breaks, excluded from planned time | 30 minutes |
| Planned production time | 450 minutes |
| Unplanned stops | 47 minutes |
| Run time | 403 minutes |
| Ideal cycle time | 1.0 second per unit |
| Total count | 21,450 units |
| Rejects | 430 units |
| Good count | 21,020 units |
Each factor, calculated from those inputs:
| Factor | Calculation | Result |
|---|---|---|
| Availability | 403 / 450 | 89.6% |
| Performance | (1.0 × 21,450) / (403 × 60) | 88.7% |
| Quality | 21,020 / 21,450 | 98.0% |
| OEE | 0.8956 × 0.8871 × 0.9800 | 77.9% |
Availability says the line ran for 89.6% of the time it was scheduled to run. Performance says that while it ran, it produced at 88.7% of the ideal rate. Quality says 98.0% of the units were good. The product of the three is 77.9%.
The same number comes from one formula, good count × ideal cycle time / (planned production time × 60), which gives 21,020 × 1.0 / (450 × 60) = 77.9%. The two forms agree because run time and total count cancel when the factors are multiplied.
Where the 450 planned minutes went shows the losses in time:
| Use of planned production time | Minutes |
|---|---|
| Good output at ideal speed (21,020 × 1.0 s / 60) | 350.3 |
| Unplanned stops | 47.0 |
| Running slower than the ideal cycle (403 − 357.5) | 45.5 |
| Making rejects (430 × 1.0 s / 60) | 7.2 |
| Total planned production time | 450.0 |
The average actual cycle was 1.13 seconds per unit (403 × 60 / 21,450), against an ideal of 1.00 second. The 45.5 minutes of slow running is what that gap costs in time.
Read the factors one at a time before deciding what to fix. Unplanned stops and slow running each take about a tenth of the planned time, 10.4% and 10.1%, while defects take 1.6%. A quality target alone would miss the two largest losses. Defects still matter, because the material and labor in a rejected unit do not appear in OEE and need their own measure.
How do you combine several shifts into one OEE figure?
A weekly or monthly OEE is not the average of the shift percentages. An average gives every shift the same weight, so a short run counts as much as a full production shift. Add the underlying minutes and counts across the shifts, then divide.
Take a second shift with 120 planned minutes, 60 minutes of run time, the same 1.0 second ideal cycle, 2,400 units made and 2,280 good. Its OEE is 31.7%.
| Shift 1 | Shift 2 | Combined | |
|---|---|---|---|
| Planned production time (minutes) | 450 | 120 | 570 |
| Run time (minutes) | 403 | 60 | 463 |
| Ideal run time (minutes) | 357.5 | 40.0 | 397.5 |
| Total count | 21,450 | 2,400 | 23,850 |
| Good count | 21,020 | 2,280 | 23,300 |
| OEE | 77.9% | 31.7% | 68.1% |
The average of the two shift values is (77.9% + 31.7%) / 2 = 54.8%. The combined figure is 68.1%, more than 13 percentage points higher. Those are percentage points, not a percentage change, as the percentage change question explains.
The combined factors follow the same rule. Availability is total run time over total planned time: 463 / 570 = 81.2%. Performance is total ideal run time over total run time: 397.5 / 463 = 85.9%. Quality is total good count over total count: 23,300 / 23,850 = 97.7%. Their product is 68.1%, the same as the direct calculation. The ideal run time needs its own column, because it is the only way to add performance correctly across shifts.
What are the six big losses that reduce OEE?
The six big losses are a classification from total productive maintenance (TPM). Each one reduces one of the three factors:
| Big loss | Factor it reduces | Example on the line |
|---|---|---|
| Equipment failure | Availability | Breakdowns that stop the line |
| Setup and adjustment | Availability | Changeovers and the warm-up after them |
| Idling and minor stops | Performance | Short stops cleared without a breakdown record |
| Reduced speed | Performance | Running below the ideal cycle |
| Process defects | Quality | Scrap and rework |
| Reduced yield | Quality | Units lost during startup until the process settles |
A maintenance project that targets one loss should move its matching factor. If the factor does not move, check whether the loss is recorded in the right place.
Is 85% OEE a standard benchmark?
You will often see 85% quoted as "world-class" OEE. The figure is usually attributed to Seiichi Nakajima's work on total productive maintenance. It is a commonly cited benchmark, not a standard. A single number cannot account for product mix, changeover frequency or process type, so compare a line with its own history before comparing it with a benchmark.
How should an OEE spreadsheet be laid out?
Keep inputs and calculations in separate columns, as described in spreadsheet model design. Use one row per shift and line on a sheet named Log. Columns A to G hold the inputs: shift, line, planned production time (minutes), unplanned stops (minutes), ideal cycle time (seconds), total count and good count. Columns H to M hold the calculations. In row 2:
- Run time, H2:
=C2-D2 - Ideal run time, I2:
=F2*E2/60 - Availability, J2:
=H2/C2 - Performance, K2:
=I2/H2 - Quality, L2:
=G2/F2 - OEE, M2:
=J2*K2*L2
Fill the formulas down. The ideal run time column is what makes combined performance correct.
A Summary sheet gives one row per line. With the line name in A2, these formulas sum the log with SUMIFS. The ranges run to row 500 so that new shifts are included:
- Planned time, B2:
=SUMIFS(Log!$C$2:$C$500, Log!$B$2:$B$500, $A2) - Run time, C2:
=SUMIFS(Log!$H$2:$H$500, Log!$B$2:$B$500, $A2) - Ideal run time, D2:
=SUMIFS(Log!$I$2:$I$500, Log!$B$2:$B$500, $A2) - Total count, E2:
=SUMIFS(Log!$F$2:$F$500, Log!$B$2:$B$500, $A2) - Good count, F2:
=SUMIFS(Log!$G$2:$G$500, Log!$B$2:$B$500, $A2) - Availability, G2:
=C2/B2; performance, H2:=D2/C2; quality, I2:=F2/E2 - OEE, J2:
=G2*H2*I2
With the two shifts above logged against Line 1, the summary returns 570 planned minutes, 463 run minutes and 68.1% OEE.
The OEE calculator is the Sheet Reserve template for this calculation. Set the ideal cycle time from the machine's rated speed or from a careful time study, then check it against the best sustained run you have. To record the stopwatch readings, use the time study template.
What mistakes distort an OEE calculation?
- Counting planned stops as downtime. Breaks and planned maintenance come out of planned production time. If the 30 minutes of breaks stay in planned time and are logged as unplanned stops, availability falls to 84.0% (403 / 480) and OEE falls from 77.9% to 73.0%, although the line made exactly the same units.
- Using the ideal cycle of the wrong product. Performance should use the ideal rate of the product that ran. Taking the 0.8-second ideal of the fastest product on the line for this run, which had a 1.0-second ideal, drops performance from 88.7% to 71.0%. Using an ideal that is too slow inflates it instead.
- Performance above 100%. This means the ideal cycle time is set too long for what the line actually made, or that the total count and run time cover different periods. Find and fix the error. Capping the value at 100% hides it.
- Leaving out the 60. The ideal cycle is in seconds and the run time is in minutes. Without the division by 60, performance comes out at 5,323%.
- Averaging percentages. The shift example above shows the cost of this error: 54.8% instead of 68.1%.
For the short version of the formula, see how to calculate OEE.