Operations Deliver

Overall Equipment Effectiveness

Overall equipment effectiveness is availability times performance times quality: a measure of how much of a machine’s scheduled time becomes good output.

The questionOf the time this machine was supposed to run, how much produced good parts at full speed, and where did the rest go?

OEE. Related measures are TEEP, which counts against calendar time, and OOE.

Formula · Seiichi Nakajima, Early 1980s

OEE Good output share of planned time

Availability Run time ÷ planned production time

Performance (Ideal cycle time × total count) ÷ run time

Quality Good count ÷ total count

OEE is the product of three ratios, each with its formula.

When to use it

A plant says it needs more capacity, and you suspect the existing equipment is not producing what it could. Check this before approving capital spending.
Downtime is logged but speed losses and rejects are not, so the picture of the line is incomplete.
A TPM, lean, or maintenance program needs a baseline and one shared measure across shifts.
You need a ranked list of loss causes on a single bottleneck machine.

When to avoid it

The machine is not the constraint. Raising OEE on a machine that is not limiting output mostly builds inventory.
The work is paced by people or is knowledge work, so there is no ideal cycle time to measure against.
The aim is to rank plants, shifts, or operators. Products, changeover patterns, and definitions differ, and OEE says little about people.
The nameplate speed is not credible and nobody can agree an honest ideal cycle time. The performance figure will be fiction.

Overview

OEE reports how much of the time a machine was scheduled to run turned into good parts at the speed it was designed for. It multiplies three ratios, each of which isolates one kind of loss, and so supports the choice of which loss to attack on a given machine first.

The single percentage matters less than its parts. A machine can show high availability and still lose an hour a shift to slow running and small stops that nobody logged as downtime. Quote the three ratios together, and quote the definitions with them, because planned time and ideal cycle time are choices, and the number moves when they do.

The six big losses

Breakdowns

Equipment failures that stop production. This is an availability loss. It is the loss everyone already tracks, and often the smaller one.

Setup and changeover

Time between the last good part of one product and the first good part of the next, including adjustments. Also an availability loss. Some sources call it setup and adjustment, and some fold in waiting for material.

Minor stops

Short jams and pauses that the operator clears in seconds or minutes. A performance loss. Plants often skip logging stops under a set length, so these do not appear as downtime.

Reduced speed

Running below the ideal cycle time because of wear, worry about quality, or habit. A performance loss, and one of the least visible.

Startup rejects

Scrap and rework made while the process settles after a start or changeover. A quality loss. Plants differ on whether to charge it to the changeover or to quality, so decide once and keep to it.

Production rejects

Defective output during steady running. A quality loss. Newer definitions split it into scrap, rework, and parts that can be sold as lower grade. Sources label the six losses slightly differently, so agree the list with the plant.

What it is not

A fixed target. An OEE of 85 percent is often quoted as world class, but it is a rule of thumb that varies by industry and product mix. A line that runs many products with frequent changeovers cannot be compared with a single-product line.
A measure of operator performance. Equipment design, material quality, maintenance, and scheduling drive most of the losses, and operators control little of them.
A number that is comparable between plants by default. Planned stops, ideal cycle time, minimum logged stop length, and the point where a reject is counted all vary.

The steps

Fix the definitions first

Agree on planned production time, the ideal cycle time and how it was set, the shortest stop that will be logged, and what counts as a good part. Write them down. What comes out is one page that every shift uses.
Capture time and counts by shift

Record planned time, stop time, total count, and good count for each shift and run. Capture it from the machine where you can, because manual logs miss minor stops.
Calculate the three ratios

Work out availability, performance, and quality separately, then multiply them. Report all four numbers together, never OEE alone.
Convert the losses to minutes

Restate each ratio's shortfall as lost minutes of scheduled time, as in the example. Minutes make the losses comparable, and show which loss is biggest.
Attack the largest loss and remeasure

Code the causes behind the largest loss for a couple of weeks, pick the top cause, and run an improvement cycle on it. Remeasure with the same definitions, so a gain is not an artifact of a changed rule.

The formula

OEE = availability × performance × quality

The share of planned production time that produced good parts at ideal speed, found by multiplying three ratios, one each for stops, speed, and rejects.

Availability = run time ÷ planned production time. Run time is planned time less stops, such as breakdowns, changeovers, and waiting for material.
Performance = (ideal cycle time × total count) ÷ run time. It is below 100 percent when the machine runs slower than ideal or stops briefly. A figure above 100 percent usually means the ideal cycle time was set too slow.
Quality = good count ÷ total count. Parts that need rework are not counted as good when you use the first-pass definition.
Planned production time is the scheduled time less planned stops such as breaks, set by whichever definition the plant has agreed.

In practice

A packaging line reports 88 percent uptime, and the plant manager cannot see why output still trails the plan. The line is scheduled for 450 minutes in the shift, after breaks. A 30-minute breakdown and a 24-minute changeover take 54 minutes, so the line runs for 396 minutes. Availability is 396 divided by 450, or 88 percent. The ideal cycle time is 0.6 minutes per pack, and the line makes 560 packs. At ideal speed 560 packs would take 336 minutes, so performance is 336 divided by 396, or 84.8 percent. Of those 560 packs, 532 are good, so quality is 532 divided by 560, or 95 percent. OEE is 0.88 times 0.848 times 0.95, which is 70.9 percent.

Expressed in minutes, the 450 break down as 319.2 minutes of fully productive time, 54 minutes of downtime, 60 minutes of speed loss, and 16.8 minutes of rejects. The largest loss is speed, not downtime. Availability alone said nothing about it. The next step is to code the minor stops and slow running for two weeks, so the 60 minutes can be assigned to causes.

What you leave with

A one-page set of definitions for planned time, ideal cycle time, minimum logged stop, and a good part.
A baseline OEE by line and shift, with availability, performance, and quality shown separately.
A loss ledger in minutes, ranked from the largest loss to the smallest.
The machine and the single loss to work on first.
A view on whether the capacity problem is real or is locked inside existing equipment.

Failure points

Definitions that drift

Someone resets the ideal cycle time slower, or lengthens the minimum logged stop, when the number looks bad. OEE rises and nothing on the floor has changed. Fix the definitions in writing and log any change.

One headline number

Management sees 71 percent and sets a target for 80. Nobody can say whether to work on breakdowns, speed, or rejects, so the effort goes to whichever the loudest person prefers.

Planned time stretched

Hard periods such as long changeovers or material shortages are relabeled as planned stops, which removes them from the measure. OEE improves on paper and output does not.

Improving a machine that is not the constraint

The line is tuned and its output rises, but the machine behind it cannot take more. The extra output sits as work in progress and the plant's throughput is the same.

A loss category called other

Operators log stops by tapping a button or choosing from a list, and when the list is long or unclear, most stops land in other. The ledger cannot rank the causes, which defeats the purpose of the measure.

The failures above are usually Commercial Acumen: the work was not tied to money, for the client and/or for you.

Origin

Seiichi Nakajima developed total productive maintenance in Japan, and OEE is its main measure of equipment losses. The term appears in his Japanese writing on TPM in the early 1980s. Productivity Press published the English translation of his TPM nyumon as Introduction to TPM in 1988, and that is how the measure reached Western plants.

Nobody owns OEE and it carries no trademark. The calculation is standard but the inputs are not. Companies differ on what counts as planned time, how ideal cycle time is set, and how short a stop has to be before it is ignored. ISO 22400-2 defines a version for manufacturing operations, and the definition any given plant uses is often its own.