How OEE is really calculated

Walk into any plant that has been chasing Overall Equipment Effectiveness — OEE — for more than a few months and you will find people who can recite the formula and people who do not fully trust the number. Both groups are right. OEE is a three-factor score, and each factor has its own failure mode. If you do not know where the inputs come from, the number in the weekly report is a shadow of a shadow.

This post walks through how OEE is actually calculated — the math, the inputs, and the traps — so the next time a plant manager asks "is that 82% real?", you can answer with confidence.

The formula

OEE is the product of three factors:

OEE = Availability × Performance × Quality

Each factor is a percentage between 0 and 100, so OEE is also a percentage. World-class discrete manufacturing is around 85% — a plant hitting 85% has very little room to improve on any of the three. Most plants start below 60% and spend years getting to 70%.

Factor 1: Availability — was the machine running when it was supposed to be?

Availability answers one question: of the time you planned to produce, how much did the machine actually spend producing?

Availability = Run Time ÷ Planned Production Time

Planned Production Time is the scheduled window minus planned stops (lunch, planned maintenance). Run Time is Planned Production Time minus unplanned stops (a breakdown, a material shortage, a changeover that ran long).

The trap

The trap is that downtime that does not get logged does not count against availability. If an operator does not mark the line as stopped when it stops — because it is faster to catch up later or because the paper log is in the supervisor's office — the time disappears into Run Time and the availability number lies upward. This is the single most common way OEE is inflated.

Factor 2: Performance — when it was running, did it run at full speed?

Performance compares how fast the machine actually ran while it was running to how fast it is rated to run.

Performance = (Ideal Cycle Time × Total Units Produced) ÷ Run Time

Ideal Cycle Time is the time the machine takes to produce one unit under perfect conditions — the nameplate speed. Multiplied by total units produced, you get the theoretical minimum time to produce that output. Divided by actual run time, you get the percentage of ideal speed.

The trap

Performance losses are quiet. A machine running at 80% of rated speed does not stop. Nothing flashes red. Nobody walks over. But a 20% performance loss eats a fifth of your throughput, every shift, without anyone noticing. This is the factor continuous monitoring surfaces that paper logs almost never do.

Factor 3: Quality — of what came out, how much was good?

Quality is the simplest of the three.

Quality = Good Units ÷ Total Units Produced

The trap

What counts as a good unit. If rework is counted as good because the customer eventually accepted it, Quality is inflated. If scrap is measured at the end of the line instead of at the inspection station, the feedback loop to the machine that produced the scrap is too slow to be useful. Quality that comes from a live, station-level capture is a very different number than Quality that comes from the finished-goods inventory report.

A worked example

A line is scheduled for 480 minutes (an eight-hour shift). 30 minutes are planned (lunch and planned cleanup). That is 450 minutes of Planned Production Time. The line suffers 60 minutes of unplanned downtime during the shift, leaving 390 minutes of Run Time.

• Availability = 390 ÷ 450 = 86.7%
• The line has an Ideal Cycle Time of 30 seconds per part and produces 650 parts in the shift.
• Theoretical minimum time = 650 × 30s = 19,500s = 325 minutes.
• Performance = 325 ÷ 390 = 83.3%
• Of the 650 parts, 26 are rejected.
• Quality = 624 ÷ 650 = 96.0%
• OEE = 86.7% × 83.3% × 96.0% = 69.3%

That is a realistic plant — availability and performance losses of roughly 13–17%, with quality in the high-90s. Moving any one of the three by two points moves OEE meaningfully. Moving all three by two points moves OEE by more than six points.

How DOVA computes OEE

DOVA captures the inputs for all three factors at the point they actually happen — at the station, in real time.

• Availability: operators register the line as running or stopped at the station. Stops are logged with a reason code the moment they happen, not at the end of the shift.
• Performance: parts completed are entered or counted as the shift runs. Performance is computed continuously against the Ideal Cycle Time configured for that machine and part.
• Quality: rejections are entered at inspection with a reason code. Quality is computed from the real time stream, not from a nightly reconciliation.

The result is an OEE number that the plant manager can actually trust — because every input has a timestamp and a source, and the traps described above are closed off.

What to do Monday morning

If you already have an OEE number you are not sure about, three questions are worth asking:

1. Where does the downtime number come from? If it is typed at end of shift from memory, availability is probably high.
2. How is ideal cycle time set for each part? If it is not set — or set too generously — performance is hiding losses.
3. Is rejection captured at inspection or at finished-goods? If the latter, quality is lagging by hours or days.

Fix those three and the OEE number stops being a shadow and starts being a tool.