Overall equipment effectiveness (OEE) is a single percentage that shows how much of a machine’s planned time turns into good parts made at full speed. This guide shows plant staff how to calculate it from shift data, how to split the result into its three factors and where the number can mislead.

What OEE measures

The National Institute of Standards and Technology (NIST) lists OEE among the metrics manufacturers use to measure and improve performance. It defines OEE as the product of availability (or uptime), performance (the speed at which equipment runs as a percentage of design) and quality (first pass yield rate). The same July 2020 NIST report names OEE and utilization among typical uses of MTConnect, an open data-exchange standard for manufacturing equipment.

The calculation below follows guidance from Vorne Industries, which publishes OEE explainers and sells monitoring software.

The inputs

  • Planned production time: shift length minus time when production is not intended, such as breaks.
  • Stop time: time production was meant to run but did not, from unplanned stops such as breakdowns or planned stops such as changeovers.
  • Run time: planned production time minus stop time.
  • Ideal cycle time: the fastest cycle time the process can achieve under optimal conditions.
  • Total count and good count: all parts produced, and parts produced minus rejects.

Calculating OEE step by step

Vorne’s worked example covers one 480-minute shift with two 15-minute breaks and one 30-minute break, 47 minutes of downtime, an ideal cycle time of 1.0 second, 19,271 parts made and 423 rejected.

  1. Planned production time: 480 minus 60 minutes of breaks gives 420 minutes.
  2. Run time: 420 minus 47 minutes of downtime gives 373 minutes.
  3. Availability is run time divided by planned production time: 373 / 420 = 88.81%.
  4. Performance is ideal cycle time multiplied by total count, divided by run time: (1.0 second × 19,271) / (373 × 60 seconds) = 86.11%.
  5. Quality is good count divided by total count: 18,848 / 19,271 = 97.80%.
  6. OEE is the product of the three: 0.8881 × 0.8611 × 0.9780 = 74.79%.

A shortcut gives the same answer: good count multiplied by ideal cycle time, divided by planned production time, or (18,848 × 1.0 second) / (420 × 60 seconds) = 74.79%. Vorne prefers the three-factor method because the shortcut does not show which kind of loss is responsible.

Where the losses sit

Vorne’s Six Big Losses page maps typical losses to the three factors:

  • Availability: equipment failure, meaning unplanned stops such as breakdowns, tooling failure or lack of materials; and setup and adjustments, meaning planned stops such as changeovers, cleaning, warmup and planned maintenance.
  • Performance: idling and minor stops such as jams, misfeeds and misaligned sensors; and reduced speed, meaning running slower than the ideal cycle time because of worn equipment, poor lubrication or substandard materials.
  • Quality: process defects during steady-state production, and reduced yield from startup until steady state, both including scrap and reworkable parts.

What counts as a good score

Vorne’s page on world-class OEE says a world-class score is often thought to be 85%. It traces OEE to Seiichi Nakajima, who introduced it with total productive maintenance and the Six Big Losses in the early 1970s at the Japanese Institute of Plant Maintenance, and whose 1984 book included the world-class figures. Vorne says most manufacturers are closer to 60%, that three factors at 90% each give an OEE of only 73%, and that these benchmarks apply to discrete manufacturing rather than process industries.

Common problems

  • Performance above 100%. Vorne says performance should never exceed 100%; if it does, the ideal cycle time is probably set too high.
  • A headline number that hides trade-offs. In Vorne’s two-week example, OEE rose from 85.1% to 85.7% while availability rose five points and quality fell 4.5 points, a trade the company says most companies would not want.
  • Untracked small stops. Vorne says most companies do not accurately track idling and minor stops, which typically last a minute or two and are often chronic.
  • Unclear stop rules. Whether a stop counts as a failure or a minor stop depends on a site’s tracking policy; Vorne’s example is requiring a reason for any downtime longer than two minutes.
  • Benchmarking unlike lines. Vorne advises against comparing dissimilar processes and notes that products with many changeovers can score lower than simpler lines, so one plant-wide target may be counterproductive.
  • Treating changeovers as fixed. Vorne counts them as stop time and improvement opportunities, and says they are typically the largest source of setup and adjustment losses.
  • Weak data. NIST notes that many manufacturers believe they understand their performance until confronted with real data, and recommends open, standard data formats over proprietary ones when connecting equipment.