How to Calculate OEE: Formula, Method, and Worked Example Shift by Shift

Calculating OEE — Overall Equipment Effectiveness — looks simple on paper and becomes systematically more delicate the moment you do it on a real line. The formula fits on one line. The calculation conventions, the time-frame perimeter, the stoppage classification, the reference cycle time — these fit on several pages and condition the reliability of the final number. This article does not rehearse the theory. It walks step by step through the calculation of OEE on a real 8-hour shift, with concrete numbers, explicit trade-offs, and common errors flagged along the way.

The method presented here is the one used by methods engineers and continuous-improvement leaders who want an OEE that is operationally usable — not a reporting figure. It takes about 90 minutes the first time on a shift, much less afterward. It requires nothing beyond Excel and a stoppage log. It produces an OEE that survives scrutiny.

The worked example covers a fictional but realistic CNC machining line, calibrated on the average data observed by TeepTrak across 450 client sites: 8-hour shift, run target of 480 pieces, manufacturer-spec cycle time of 60 seconds per piece. By the end of the article, you will know not only what OEE the line achieved, but also how to justify it to your management and how to defend it against your quality leadership.

The OEE formula in three lines

OEE is the product of three factors, each expressed between 0 and 1 (or as a percentage between 0 % and 100 %). The canonical formula adopted by virtually every industrial reference — the original Nakajima/JIPM framework, AFNOR NF E60-182 in France, SEMI E10 for semiconductors — is the following.

• Availability = Effective operating time ÷ Required operating time. Measures what the equipment actually ran versus what it was supposed to run.
• Performance = (Pieces produced × Theoretical cycle time) ÷ Effective operating time. Measures the ratio of actual cadence to theoretical cadence, over the time the machine was really running.
• Quality = Good pieces (first time right) ÷ Pieces produced. Measures the share of pieces that meet specification on the first pass, excluding rework and scrap.

OEE is obtained by multiplying the three: OEE = Availability × Performance × Quality. That is the whole formula. The trap is not in the formula — it is in the rigorous definition of each of the five temporal and quantitative parameters above. As of May 2026, the international conventions have been stable for over a decade, but the practical application remains the place where everyone gets tripped up at first.

Step 1 — Define the time frame honestly

The first methodological choice structures everything that follows: what do you count as required operating time? Three conventions coexist in practice and produce results that can vary by 15 to 25 OEE points on the same real line.

The total scheduled time convention takes as denominator the total time during which the line is staffed: 8 hours on a standard day shift. This convention is the strictest and produces the lowest OEE figures, but also the most actionable ones — it surfaces every loss, including the ones other conventions hide.

The planned production time convention excludes scheduled non-production stops: meal break, planned preventive maintenance, shift handover meeting. On an 8-hour shift, this typically removes 45 to 60 minutes. This is the most widely used convention in manufacturing across the US, UK, and EU industrial markets.

The required time convention further excludes planned product-mix-related times: planned changeovers, regulatory cleaning, scheduled calibrations. This convention produces the most flattering OEE figures but masks the real productivity of the capital equipment.

For our worked example, we use the planned production time convention: 8 hours of shift minus 45 minutes of breaks and shift briefing — leaving 7h15 of planned production time, i.e., 435 minutes or 26 100 seconds.

Practical recommendation: write your time-frame convention down in black and white before calculating anything. Have it validated by your manager and your quality director. Do not change it for at least 12 months. An OEE whose convention changes every quarter is an OEE that no longer serves any purpose.

Step 2 — Measure Availability in the worked example

On our 7h15 planned production time (26 100 seconds), we log machine stoppages over the day. Here is the real shift log of a typical day observed on a CNC line at an aerospace supplier — data from TeepTrak instrumentation then verified manually.

• 07:45 — Shift start: 8 min of warm-up and machine ramp.
• 09:12 — Chip evacuation jam: 4 min.
• 10:30 — Planned tool change: 18 min.
• 11:05 — Micro-stop (measurement probe misaligned, operator restart): 90 seconds.
• 12:00 — Meal break: already out of planned production time.
• 13:55 — Pallet feeder breakdown: 22 min until full recovery.
• 14:50 — Micro-stop (transient hydraulic pressure alarm): 2 min.
• 15:30 — Three cumulated micro-stops (cycle clearing, brushing, first-piece visual check on new lot): 6 min cumulated.

Cumulative stoppages across the 7h15 planned production time: 8 + 4 + 18 + 1.5 + 22 + 2 + 6 = 61.5 minutes, i.e., 3 690 seconds.

Effective operating time = 26 100 − 3 690 = 22 410 seconds (i.e., 6h13’30”).

Availability = 22 410 ÷ 26 100 = 0.8586 i.e., 85.86 %.

First lesson: if you compute Availability using only the “real breakdowns” (the 22 minutes of feeder breakdown), you get 96 % and you are off by 10 points. Most sites that compute OEE manually make this exact error — they forget micro-stops, omit warm-up time, and exclude planned events like tool changes that should logically be inside the Availability perimeter (a tool being changed is a machine not producing).

Step 3 — Measure Performance in the worked example

During the 22 410 seconds of effective operating time, how many pieces did the machine produce? End-of-shift physical count: 312 pieces.

The manufacturer-spec cycle time states 60 seconds per piece. But the sustainable real cycle time on this product, in today’s conditions (material, ambient temperature, tool wear state), is actually 65 seconds per piece — figure obtained by stopwatch-timing 30 consecutive incident-free cycles.

Methodological question: which cycle time should we use? The international rule and industrial common sense both impose using the manufacturer-spec cycle time, not the sustainable cycle time. Why: the spec cycle time represents the equipment’s theoretical potential, and the Performance factor must reflect the gap between this potential and reality. If you adjust the reference cycle time to the sustainable rate, you mechanically mask speed losses — and you fabricate an artificially flattering OEE.

Performance = (Pieces produced × Theoretical cycle time) ÷ Effective operating time

Performance = (312 × 60) ÷ 22 410 = 18 720 ÷ 22 410 = 0.8354 i.e., 83.54 %.

Interpretation: the machine ran 22 410 seconds, but at perfect cadence it should have produced 22 410 ÷ 60 = 373.5 pieces. It produced only 312. The difference — 61.5 pieces not produced — represents the cumulated speed loss: micro-slowdowns not logged as stoppages, gradual cycle-time drift toward end of shift, operator adjustments on long cycles.

Second lesson: without a sensor, this calculation is defensible but imprecise. The 61.5 missing pieces typically hide between 5 and 15 micro-stops under 30 seconds that appear in no manual log. The gap between declared Performance and real Performance is typically 3 to 8 OEE points on most sites measured by TeepTrak.

Step 4 — Measure Quality in the worked example

Out of 312 pieces produced, end-of-shift quality control records:

• 298 first-time-right conformant pieces, releasable without rework.
• 9 minor non-conformant pieces requiring rework (deburring, dimension adjustment). Rework takes on average 2 minutes per piece.
• 5 scrap pieces (out-of-tolerance dimension, critical surface defect).

The standard methodological trap is to count the 9 reworkable pieces as “conformant” because they will eventually ship to the customer. The international OEE rule is the opposite: the Quality factor must reflect first-time-right performance, so reworked pieces count as non-conformant for OEE calculation, even if they are not physically scrapped.

Quality = Good first-time pieces ÷ Pieces produced = 298 ÷ 312 = 0.9551 i.e., 95.51 %.

If you wrongly include rework in the “good” count, you get 307 ÷ 312 = 98.4 %, a 3-point overstatement. On a line where scrap is low but rework exists, this methodological error overstates Quality by 2 to 5 points.

Step 5 — Compute the final OEE and decompose

With our three factors: OEE = 0.8586 × 0.8354 × 0.9551 = 0.6852 i.e., 68.52 %.

Sixty-eight point five two percent. That is the real OEE of a typical mid-market aerospace CNC line in May 2026, measured honestly, on a representative shift.

Loss decomposition versus the maximum potential:

• Availability losses: 100 % − 85.86 % = 14.14 points. Dominant cause: the pallet feeder breakdown (22 min, i.e., 8.4 points alone).
• Performance losses: 100 % − 83.54 % = 16.46 points. Dominant cause: unlogged micro-stops and cumulative cycle-time drift over 7 hours.
• Quality losses: 100 % − 95.51 % = 4.49 points. Dominant cause: surface-defect rework.

The priority improvement lever is therefore clearly Performance (invisible losses). The secondary lever is Availability with preventive maintenance on the pallet feeder. Quality is already strong and does not warrant structural investment in the short term.

This is exactly what a rigorous OEE calculation produces: a clear prioritization of improvement levers, grounded in numbers rather than intuition.

Step 6 — Check coherence with the site’s official OEE

If your line’s official OEE is declared at 82 % or higher, you have a 13-to-15-point gap with what honest measurement produces. That is the typical gap observed between manual self-reported OEE and rigorously measured OEE — a pattern TeepTrak has observed systematically on sites where sensor instrumentation is deployed in parallel with existing manual reporting.

The three structural causes of this gap are the same everywhere. First, under-reporting of micro-stops in manual logs (the operator resolves the issue in 30 seconds and moves on, the stop is never recorded). Second, silent absorption of speed losses into the reference cycle time (the manufacturer’s spec cycle time is adjusted downward “to avoid penalizing OEE”). Third, accommodating classification of unproductive time (a series changeover that drifts 20 minutes is counted as “planned changeover” rather than “changeover + overrun”).

A rigorous manual OEE calculation by an operator or methods engineer takes 90 minutes per shift the first time, 30 minutes thereafter. An automated OEE calculation via sensor instrumentation takes zero minutes — measurement is continuous, granular, timestamped, irrefutable. When the transition to sensor instrumentation reveals a 15-point gap with the official OEE, the question is no longer “is the sensor right” but “how long have you been letting the self-reporting error run”.

Methodological summary: the six-rule rigorous OEE calculation checklist

To calculate OEE correctly on any line in May 2026, six rules hold.

• Write down the time-frame convention first. Choose between scheduled time, planned production time, or required time, and freeze it for 12 months.
• Log every stoppage, including those under 30 seconds. Otherwise, Availability is overstated by 5 to 10 points.
• Keep the manufacturer-spec cycle time as the reference. Any other cycle time amounts to lying to yourself about real performance.
• Count rework as non-conformant for the Quality factor. OEE measures first-time-right, not what eventually ships.
• Decompose losses before setting an improvement plan. Without decomposition, the plan optimizes in the wrong place.
• Check coherence with the official OEE. If the gap exceeds 10 points, the official OEE is probably self-reported and instrumentation becomes the priority.

From manual calculation to continuous measurement

The manual method described above is the foundation: it teaches the conventions, it surfaces the trade-offs, it produces a defensible baseline. Most sites that follow it carefully on one shift quickly recognize the limits of doing it on every shift, every line, every day.

The transition to continuous automatic measurement is the natural next step. Wireless external sensors capture every stoppage at one-to-two-second resolution, operators qualify stoppages from a touch terminal in under five seconds per event, and the platform aggregates the data into shift-level, day-level, and month-level OEE with full loss decomposition. The manual 90-minute exercise becomes a continuous data stream available in real time.

The cost-effectiveness threshold for this transition is reached on most sites once three or more lines are being tracked seriously, or once methods engineering time spent on OEE data entry exceeds two hours per week. Below those thresholds, manual calculation remains viable. Above them, the operational overhead of manual measurement starts to dominate the cost equation, and the precision penalty (15-25 OEE points of structural understatement of losses) further widens the gap.

For the practical deep dive on running OEE in Excel and the six pitfalls to avoid, see the dedicated article Calculating OEE in Excel: Ready-to-Use Template and 6 Pitfalls to Avoid. For the comparison between OEE and adjacent indicators (TEEP, OOE), which clarifies why OEE alone may not be the right metric in every context, see OEE vs TEEP vs OOE: Three Indicators, Three Use Cases.

Move from manual OEE calculation to real-time automatic measurement in 48 hoursWireless external sensors installed in under 30 minutes per machine. No PLC modification, no IT integration, first real OEE data by day three. Free 48-hour POC on site, 450+ factories deployed across 30 countries.Request an OEE demo

External references

Overall Equipment Effectiveness — Wikipedia · AFNOR (NF E60-182) · JIPM — Japan Institute of Plant Maintenance · SEMI Standards (E10)

Related TeepTrak reading: Calculating OEE in Excel: Template and 6 Pitfalls · OEE vs TEEP vs OOE: Three Indicators, Three Use Cases · How to Calculate OEE: Complete Guide · OEE Calculation Mistakes That Inflate Your Numbers