What OEE measures.
Overall Equipment Effectiveness compares what a machine produced with what it could have produced in the time it was required to run. A machine that ran for every minute of required time, at its ideal rate, making only good parts, would score one hundred per cent. Every shortfall from that falls into one of three factors, and OEE is their product.
Two plants quoting the same figure may be treating different stops as planned, working from different reference cycle times, and disagreeing about whether a reworked part counts as a good one. None of them is lying. The figure is only comparable when the definitions behind it are, which is why the definitions come first here.
The three factors, and what each one carries.
Each factor is a ratio between zero and one, and each carries one family of losses. Stops reduce availability, speed losses and micro-stops reduce performance, and rejects reduce quality. Multiplying the three gives OEE.
OEE = Availability × Performance × Quality
Availability = Operating time ÷ Required time
- Required time
- The time the machine was expected to produce: opening time less planned stops.
- Operating time
- Required time less every unplanned stop, whatever its cause.
Performance = (Units produced × Ideal cycle time) ÷ Operating time
- Ideal cycle time
- The shortest time to make one unit of this product on this machine.
Quality = Good units ÷ Units produced
Availability
Carries breakdowns, changeovers and every wait. What counts as planned should come from the shift calendar rather than from a judgement made afterwards, and short stops below the usual reporting threshold should be counted rather than absorbed into running time.
Performance
Carries speed losses and the stops too short to record. Counts should come from the machine’s own counters, and the ideal cycle time should be a property of the product on the machine, not a figure adjusted shift by shift to make a number work.
Quality
Carries scrap and rework. Rejects should be declared at the line against a reason, so that the quality factor and the scrap Pareto are built from the same declarations rather than reconciled against each other at month end.
The time categories underneath.
AFNOR NF E60-182 names each layer of time, so that a loss always lands in the same place. Read the table from top to bottom: each line is the one above it less one family of losses.
| Time | What it is |
|---|---|
| Total time | Every hour of the period, open or not. |
| Opening time | Total time less the hours the plant is closed. |
| Required time | Opening time less planned stops: planned maintenance, breaks, no orders. |
| Operating time | Required time less unplanned stops: breakdowns, changeovers, waits. |
| Net time | Operating time less speed losses and micro-stops. |
| Useful time | Net time less the time spent making rejects. |
A worked example.
One machine over one eight-hour shift. The counts below are the kind a controller reports; the factors are computed from them exactly as above.
| Item | Value |
|---|---|
| Required time | 480 min |
| Unplanned stops | 62 min |
| Ideal rate | 120 units per hour |
| Good units | 760 units |
| Rejected units | 24 units |
| Operating time | 418 min |
| Availability | 87.1% |
| Performance | 93.8% |
| Quality | 96.9% |
| OEE | 79.2% |
Each line can be checked by hand from the ones above it. A team that gets a different OEE from the same counts is using a different definition somewhere, and that is the conversation worth having.
Illustrative counts for one machine, not a figure from any plant.
Where two OEE figures stop being comparable.
Each of these is a defensible choice. The figures diverge when different sites make different choices without saying so.
Planned time defined differently
Moving changeovers, cleaning or maintenance out of required time raises availability without changing anything on the line. State what is planned, and keep it the same everywhere the figure is compared.
A reference rate adjusted to fit
Lowering the ideal rate to what a line usually achieves makes performance look healthy and hides exactly the loss the factor exists to show. Set the rate once, from a validated source, and record where it came from.
Rework counted as good
Counting a reworked unit as good raises quality and hides the time spent correcting it. Either answer can be right; what matters is that it is the same answer on every line.
Machine figures averaged into a line figure
An average of percentages weights a machine that ran one hour like one that ran eight. Combine machines from their times, useful over required, and on a coupled line report the constraint’s own figure beside the line’s.
Where the inputs should come from.
The formula is the same everywhere; the inputs are what decide whether the figure can be trusted. States and counts are best read from the machine controller, because a person writing them down at the end of a shift rounds durations and leaves short stops out.
What a machine cannot know is the reason for a stop or a reject. That is the part to ask of the operator, at the moment it happens, and only for the stops long or frequent enough to warrant it.
Terms this guide uses.
Each one is defined in full in the glossary, with the terms around it.
- OEE
Overall Equipment Effectiveness is the share of planned production time that produced good parts at full speed.
- Availability
Availability is the time a machine actually ran as a share of the time it was required to run.
- Performance
Performance is what a machine produced while running, as a share of what it would have produced at its ideal rate in the same time.
- Quality rate
The quality rate is the number of good units as a share of all units produced.
- AFNOR NF E60-182
AFNOR NF E60-182 is the French standard that defines the time categories of a production resource and the rates computed from them, including the TRS, the French equivalent of OEE.
- Ideal cycle time
The ideal cycle time is the shortest time in which a machine can produce one unit of a given product under good conditions.
- Micro-stop
A micro-stop is a stop short enough that it is cleared on the spot and rarely recorded by hand: a jam, a misfeed, a sensor that needs a reset.
- TEEP
TEEP measures good output against all calendar time, not just planned production time.