Overall Equipment Effectiveness (OEE)
Overall Equipment Effectiveness (OEE) is a single percentage that multiplies availability, performance, and quality to show how much of the planned production time a machine spent making good parts at its rated speed. A score of 100 percent means no stops, no slow running, and no defects.
Overall Equipment Effectiveness (OEE) is the product of three factors, each measuring a different way a machine loses output. Availability is run time divided by planned production time, so it captures breakdowns, material starvation, and changeovers. Performance is ideal cycle time multiplied by total count, divided by run time, which captures slow running and the short stops too brief for anyone to log. Quality is good count divided by total count, counted at first pass, so a part that needed rework is not a good part. Planned production time is the denominator that decides everything, because it excludes time the plant deliberately chose not to schedule the machine.
Multiplying rather than averaging is the whole point. Three respectable factors produce an unremarkable product: 90 percent availability, 95 percent performance, and 99 percent quality give roughly 85 percent, and the structure forces each loss to stay visible instead of being smoothed away. Each factor owns two of the six big losses, with breakdowns and setup or adjustment under availability, small stops and reduced speed under performance, and startup rejects and process defects under quality. Related measures change the denominator rather than the idea: utilization asks how much of the available time was scheduled at all, and total effective equipment performance stretches the denominator to all calendar time.
Because planned production time is a choice, OEE is the easiest number in the plant to improve on paper. Moving changeover into planned downtime, declaring an idle hour as unscheduled, loosening ideal cycle time to the rate currently achieved rather than the design rate, and counting reworked parts as good will all lift the score without shipping one more part. An OEE that rises while output does not is a definition change, and it is worth auditing the definition before celebrating the trend. The other common misuse is measuring a machine that is not the constraint, where running harder converts raw material into work in process rather than revenue.
In a real shop the useful artifact is not the percentage but the downtime reason codes underneath it. Start on one machine, write down what planned production time means and what the ideal cycle time is per part number, then log every stop with a duration and a reason, plus total count and good count by shift. Paper works. Automated capture from the control or a machine monitor removes the operator's rounding once the reason codes have proven they describe the losses the floor actually has. Benchmarks between plants are close to meaningless given how differently the denominator gets defined; a stable baseline on one definition is what makes the trend worth anything.
OEE = Availability x Performance x Quality. Availability = Run Time / Planned Production Time. Performance = (Ideal Cycle Time x Total Count) / Run Time. Quality = Good Count / Total Count.
A CNC lathe is scheduled for two 480-minute shifts, so planned production time is 960 minutes. It loses 60 minutes to a spindle fault and 90 minutes to two changeovers, leaving 810 minutes of run time, so availability is 84.4 percent. Ideal cycle time is 1.5 minutes and the lathe produced 450 pieces, giving 675 minutes of value-adding time and performance of 83.3 percent. Eighteen pieces failed first-pass inspection, so quality is 96 percent. OEE is 0.844 x 0.833 x 0.96, or 67.5 percent.
How is OEE different from capacity utilization?
OEE measures what happened inside planned production time: whether the machine was running, running at its rated speed, and making good parts. Capacity utilization measures how much of the time available the machine was scheduled at all. A plant can report excellent OEE on an asset it only schedules one shift a week.
Which machines should I track OEE on?
The constraint first. An hour recovered on a bottleneck converts directly into plant output, so the measurement pays for itself. The same improvement on a non-bottleneck builds work in process and inventory without shipping anything more. Expand to other assets only once the constraint is stable, instrumented, and understood.
How do I start tracking OEE on a machine?
Write down the definition of planned production time and the ideal cycle time per part number before collecting anything, because both decide the answer. Then log every stop with a duration and a reason code, plus total count and good count by shift, on paper if necessary. Automate capture once the reason codes describe your real losses.
What does OEE fail to show?
Anything outside the machine. OEE says nothing about whether the part was worth making, whether the schedule was right, or whether material arrived on time. It also hides its own denominator, so two plants reporting the same figure can be running very differently. Treat it as a loss-finding tool against your own baseline, not a benchmark.
What data has to be recorded to calculate OEE?
Planned production time by shift, every stop with its duration and a reason code, the ideal cycle time for the part that ran, total pieces produced, and good pieces counted at first pass. Keep the raw stop records and the reason codes rather than only the percentage, because the codes are what tell you where to act.