What Is OEE and How to Track Equipment Efficiency Automatically
Practice4 minInnovialabs editorial team
In short
OEE (Overall Equipment Effectiveness) multiplies availability, performance and quality to show what share of planned time a machine actually produced good parts; automatic collection from machines plus clear downtime reasons turns it into a daily improvement tool.
OEE in one formula
OEE (Overall Equipment Effectiveness) shows how much of the planned production time was truly productive. It is the product of three factors:
OEE = Availability × Performance × Quality
- Availability — did the machine run when it was supposed to? Run time divided by planned production time. Losses: breakdowns, changeovers, waiting for material or an operator.
- Performance — when it ran, did it run at full speed? (Ideal cycle time × total count) divided by run time. Losses: slow cycles, small stops, idling.
- Quality — were the parts good? Good count divided by total count. Losses: scrap, rework, startup rejects.
A quick example with round numbers: planned time 480 minutes, 60 minutes of stops, so run time is 420 minutes and availability is 420/480 = 0.875. Ideal cycle time 1 minute, 380 parts made: performance is 380/420 ≈ 0.905. 361 of them good: quality is 361/380 = 0.95. OEE ≈ 0.875 × 0.905 × 0.95 ≈ 0.75.
The value of OEE is not the single number but the breakdown of losses: it tells you whether to fight breakdowns, slow speed or scrap.
Defining the terms before measuring
Most OEE disputes come from definitions, not data. Agree on:
- Planned production time — is lunch excluded? Planned maintenance? Shifts with no orders?
- Ideal cycle time — the machine's real best rate for this product, not an optimistic catalog figure.
- Small stops — below what duration a stop counts as a performance loss rather than downtime.
- What counts as good — first-pass good only, or after rework too.
Write these down and apply them the same way on every line, otherwise comparisons are meaningless.
Collecting data from machines
Manual paper forms are a starting point but usually understate small stops and slow cycles. Automatic options:
| Source | What you get | Notes |
|---|---|---|
| PLC / CNC controller | Run state, cycle counts, alarms, program | Most accurate; via OPC UA, MTConnect or vendor protocols |
| Digital I/O module | Running/stopped signal, part counter pulse | Works on old machines; simple and cheap |
| Current sensor | Running vs idle by power draw | Non-invasive; needs threshold calibration |
| Scanner or scale at the output | Good part count | Useful where counting at the machine is hard |
| Operator terminal | Downtime reasons, scrap reasons, product changes | Always needed — machines don't know why they stopped |
A practical combination: automatic state and counts from the machine + operator input for reasons.
Downtime reason codes
An automatic signal says the machine stopped; only a person knows why. Make it fast to answer:
- Keep the list short — roughly ten to fifteen top-level reasons, with optional detail.
- Group by category: breakdown, changeover, no material, no operator, quality check, planned stop.
- Show the prompt on a terminal at the machine right when the stop happens, with large buttons.
- Allow assigning one reason to a long stop and ignore very short ones (they count as performance loss).
- Review the "other" category regularly — if it grows, the list needs a new reason.
Shift dashboards
Different people need different views:
- Operator at the machine: current state, count vs target for the hour, unanswered stops.
- Shift supervisor: all machines on one screen, colored by state, top downtime reasons this shift.
- Production manager: OEE trend by line and week, Pareto of losses, comparison of shifts and products.
- Maintenance: breakdown frequency and duration per machine, mean time between failures.
Show hour-by-hour targets on the shop floor: a team that sees it is behind at 10:00 can still react, while a daily report only explains what was lost.
Rolling out step by step
- Pick one line or a bottleneck machine.
- Agree on definitions and the reason list with operators and supervisors.
- Connect state and counts automatically; add an operator terminal.
- Run for several weeks, compare with manual records, fix gaps.
- Start a short daily meeting around the top losses.
- Extend to other machines and integrate with planning and maintenance systems.
Common mistakes
- Chasing a "world-class" benchmark instead of improving your own trend.
- Inflating ideal cycle time so OEE looks better.
- Using OEE to blame operators — they will stop entering honest reasons.
- Collecting data but never discussing it at shift meetings.