The OEE formula is simple. Deciding the inputs is not.
One shift worked end to end — ISO 22400-2 first, then the same data on the internal variant, then the loss converted to rupees at a real machine-hour rate.
This is the arithmetic, worked end to end with numbers from an Indian machining floor. The formula is simple; the difficulty is entirely in deciding what goes into it, so each input is defined before it is used.
The ISO 22400-2 calculation comes first. The MachineWise variant follows, with the same shift data, so the difference between the two is visible rather than asserted.
| Quantity | This example | How to determine yours |
|---|---|---|
| Shift length | 480 min | Actual scheduled shift, not nominal |
| Scheduled breaks | 45 min | Tea, lunch and any planned stoppage you do not intend to produce through |
| Planned production time | 435 min | Shift length less scheduled breaks. This is the ISO denominator for availability. |
| Unplanned stops | 94 min | Everything that stopped production and was not planned |
| Actual production time | 341 min | Planned production time less unplanned stops |
Be careful here: whether planned maintenance sits inside or outside planned production time is the single largest source of cross-plant OEE disagreement. Decide, write it down, and apply it consistently.
Availability = actual production time ÷ planned production time = 341 ÷ 435 = 78.4%
Effectiveness = (produced quantity × ideal cycle time) ÷ actual production time. With 212 parts at an 88-second ideal cycle: (212 × 88) ÷ (341 × 60) = 18,656 ÷ 20,460 = 91.2%
Quality = good quantity ÷ produced quantity = 206 ÷ 212 = 97.2%
OEE = 0.784 × 0.912 × 0.972 = 69.5%
Availability = machine on-time ÷ reporting time = 372 ÷ 435 = 85.5%
Performance = productive time ÷ machine on-time = 268 ÷ 372 = 72.0%
Quality = 206 ÷ 212 = 97.2%, identical to ISO
OEE = 0.855 × 0.720 × 0.972 = 59.9%
Availability is higher on the variant because the machine was powered for part of the time it was not producing — setup, for instance. Performance is much lower because a substantial share of powered time involved no cutting. The two figures disagree by nearly ten points and both are correct; they answer different questions.
Losses only get fixed when they have a rupee figure attached. Take the machine-hour rate your costing team already uses for quoting — not the electricity cost. At ₹850 per hour, the 94 minutes of unplanned stops in this shift cost roughly ₹1,330.
That is unremarkable until it is annualised. The same loss across 26 shifts a month is about ₹4.2 lakh a year on one machine. On a thirty-machine floor with a comparable loss profile, the figure approaches ₹1.2 crore annually — and the point of the exercise is not the total but that it is now large enough to justify a supervisor spending twenty minutes a week on a Pareto.
The downtime cost calculator does this arithmetic at your rates, and the OEE calculator runs both definitions above with your own shift numbers.
Counting cycles as parts
A program producing four components per cycle makes output and performance wrong by exactly four, and it looks entirely plausible.
Treating setup as downtime
A machine in setup is not broken. Conflating them produces a figure the shop floor will dispute, correctly.
Using a stale ideal cycle time
The single most common cause of a meaningless Effectiveness figure. See ideal cycle time explained.
Counting produced parts as good parts
The control counts cycles, not acceptable components. Quality then reads 100% and Performance is flattered.