Free Trial

← Back to blog

Production managementReading time 12 min

A machine running 95% of the time can still have an OEE of 50%

OEE is the most quoted and most miscalculated figure in manufacturing, because the number can double depending on how you choose the denominator.

OEE

01Three factors multiplied

MetricCalculationWhat it deducts
AvailabilityActual running time ÷ loading timeBreakdowns, changeovers, material changes, waiting
Performance(Ideal cycle time × output) ÷ actual running timeIdling, minor stops, reduced speed
QualityGood pieces ÷ total piecesDefects, rework
OEEThe three multiplied
0.90 × 0.95 × 0.99 = 0.846. All three look respectable; together they are 84.6%.

02The six big losses

LossWhich factorTypical cause
Equipment failureAvailabilityUnplanned stops, waiting for repair
Set-up and adjustmentAvailabilityChangeover, first-piece adjustment
Idling and minor stopsPerformanceJams, sensor misfires, tens of seconds each time
Reduced speedPerformanceTool wear, or slowing down out of caution
Process defects and reworkQualityParameter drift, material variation
Start-up lossesQualityScrap between switching on and settling down
Minor stops are the hardest to see: a few tens of seconds each, never logged, and capable of eating ten per cent of the performance factor.

03From OEE to TEEP

  • OEE uses loading time as the denominator and asks whether the scheduled hours were used well.
  • OOE uses operating time, including planned downtime.
  • TEEP uses calendar time (24 × 365) and asks how much of the machine's capacity is being used at all. It is an investment metric.

The three answer different questions: OEE asks about shop-floor management, TEEP asks whether to buy the next machine. A high-mix, low-volume plant will naturally have a low OEE because it changes over often, and comparing it with a mass-production line is meaningless.

04How to improve it: look at the lowest factor first

  1. Low availability → changeover time is the first target. SMED separates changeover work into internal tasks, which need the machine stopped, and external tasks, which can be prepared while it still runs. Converting internal into external routinely halves changeover time.
  2. Low performance → measure the minor stops first. Most plants do not record stoppages under three minutes; making them visible comes before anything else.
  3. Low quality → go back to process capability (Cpk) and the measurement system (GR&R), and establish whether the process is genuinely unstable or the measurement is misjudging.

05Where the data comes from

The three factors need three kinds of data: machine running and stopped state, actual output against ideal cycle time, and good and defective counts.

  • State comes from machine I/O, current sensing or the CNC interface, far more accurate than a hand-written log.
  • Output comes from counters, sensors or labour reporting.
  • Good count comes from inspection records, and the basis for inspection comes from the drawing.

That last link is often broken: the inspection sheet was copied off the drawing by hand, a mistyped tolerance distorts the defect rate, and the quality factor in OEE is distorted with it.