The situation

Management says the press has a quality problem. Scrap is high on the board. But when Jan, the shift supervisor, walks the line, the press is idle more than it is making bad parts. Everyone is arguing about the wrong loss because nobody has put the losses on one scale.

Overall Equipment Effectiveness, or OEE, puts them on one scale.

What OEE is for

OEE tells you what share of your planned production time was truly productive. “Truly productive” means the machine was running, running at full speed, and making good parts first time.

It multiplies three ratios:

  • Availability: how much of the planned time the machine actually ran.
  • Performance: how fast it ran compared with its ideal speed.
  • Quality: how many parts were good first time.

Here is the formula, shown once:

OEE = Availability × Performance × Quality

Because the three are multiplied, three decent-looking numbers give a weak result. 90% × 90% × 90% is only 72.9%.

OEE is a diagnostic. Its value is not the single number but the three parts. The lowest part tells you where to look first.

A short history

OEE grew out of Total Productive Maintenance (TPM). Seiichi Nakajima developed it in Japan, working with the Japan Institute of Plant Maintenance, from the late 1960s into the 1970s. He set out the idea and the six big losses in a Japanese book in 1982, which appeared in English as Introduction to TPM: Total Productive Maintenance (Productivity Press, 1988). Car makers took it up widely, and it spread to other discrete and batch manufacturing. Details of exactly who first wrote the formula down are not clear to us, so we only credit Nakajima’s published work.

Where it applies, and where it does not

It fits:

  • automated or semi-automated machines such as presses, CNC centres, molding cells and SMT lines;
  • the bottleneck machine, where lost time is lost output for the whole plant;
  • before-and-after tracking of a TPM or changeover project.

It fits badly:

  • Non-bottleneck machines. Pushing a feeder machine to a high OEE just builds a pile of work in progress.
  • Manual bench assembly. There the limit is people, layout and balance, not machine speed. Use takt time and line balancing.
  • Continuous process plants (chemicals, for example). They are governed by yield and reaction, not part counts.

A note on targets. You will hear “world class OEE” quoted as a fixed number. Targets vary by plant, machine and process, and a number that suits a bottling line may be silly for a job shop. We do not quote a universal benchmark. Compare your machine with its own history first.

How to do it

Our example is a fictional 400-ton transfer press, PR-08, at Silesia Precision Stamping. It makes door hinge brackets. Jan runs the shift. Agata leads maintenance. Check the arithmetic by hand.

  1. Planned production time. The shift is 8 hours, 480 minutes. Planned breaks are 30 minutes. Planned time = 480 − 30 = 450 minutes.
  2. Availability. The press stopped for 40 minutes for a coil change and setup, and 30 minutes for a feeder jam. That is 70 minutes. Run time = 450 − 70 = 380 minutes. Availability = 380 ÷ 450 = 84.4%.
  3. Performance. The press’s ideal cycle time is 10 seconds per stroke. It made 1,900 parts. Ideal time for 1,900 parts = 1,900 × 10 = 19,000 seconds. Run time is 380 minutes = 22,800 seconds. Performance = 19,000 ÷ 22,800 = 83.3%.
  4. Quality. 76 parts were scrapped for burrs and oil pits. Good parts = 1,900 − 76 = 1,824. Quality = 1,824 ÷ 1,900 = 96.0%.
  5. Multiply. 0.8444 × 0.8333 × 0.9600 = 67.6%.

Here is a shortcut to check. OEE is also the ideal time to make the good parts, divided by planned time: (1,824 × 10 s) ÷ (450 × 60 s) = 18,240 ÷ 27,000 = 67.6%. Same answer. If your two routes disagree, you made an arithmetic slip.

So about 32.4% of planned time was lost. Look at the parts: availability is 84.4%, performance 83.3%, quality 96.0%. Quality is the best of the three, not the worst. (The three gaps do not add up to 32.4 points, because the ratios multiply.) Management was looking at the smallest piece. The big pieces are setup, the feeder jam and slow strokes.

The six big losses, in plain words

Nakajima’s six losses sit under the three ratios.

Ratio Loss In plain words
Availability 1. Breakdowns The machine fails and stops.
Availability 2. Setup and adjustments Changing the job, then tuning it.
Performance 3. Small stops Short stops: a jam, a sensor to clear.
Performance 4. Reduced speed It runs, but slower than it should.
Quality 5. Startup rejects Bad parts while the machine warms up.
Quality 6. Production rejects Bad parts during the normal run.

Three ways it goes wrong

  1. Fudging the ideal cycle time. Someone sets the ideal time slower than the machine’s rating, so Performance looks great. The number is nice and the loss is hidden. Use the rated or proven best speed.
  2. Chasing OEE everywhere. A demand for the same OEE on every machine makes upstream machines overproduce. The aisles fill up. Focus on the bottleneck.
  3. Not logging small stops. A ten-second sensor clear feels too small to write down. Fifty of them a shift are not small. Without them, slow performance gets blamed on operators.

Our recommendations

  • Define “planned time” and “ideal cycle time” in writing, once, and keep them fixed.
  • Log stop reasons, even short ones. A tally sheet is enough.
  • Be suspicious of 100% availability or performance above 100%. They usually mean missing data or a wrong ideal time.
  • Work on the smallest of the three ratios first.
  • Treat the numbers as a tool for the team, not a way to rank operators.

Try it

The OEE tool will calculate the three ratios, show the six losses and give you a shift handover page. It runs in your browser. It is coming soon: OEE calculator.

From the book

OEE is part of Chapter 20 (Lean, OEE, KPIs). See the book page.