Measurement and indicators

Overall Equipment Effectiveness (OEE)

In French: TRS (taux de rendement synthétique)

OEE measures the share of required time actually turned into good parts at nominal rate, as the product of three rates: availability, performance and quality.

What it is

Overall equipment effectiveness condenses three families of losses into a single number: stoppages, slowdowns and non-conformities. Its teaching value lies less in the result than in its breakdown — an OEE of 65 % says nothing, but 95 % × 80 % × 85 % immediately points to where to act.

In France, the standard distinguishes three indicators that are commonly confused: the TRS (OEE) is related to required time, the TRG to opening time, the TRE to total time. Comparing the OEE of one workshop with the TRG of another makes no sense, and yet it is frequent as soon as two sites are compared.

The six losses, and the rate that carries them

Breaking OEE down is not enough; one still has to know what each rate contains. The classic grid splits the losses into six families, two per rate.

Availability — breakdowns and unplanned stoppages, and changeovers and adjustments. The first family belongs to maintenance, the second to SMED: they are two different projects, and merging them into one action plan rarely leads anywhere.

Performance — minor stoppages and idling, and speed losses. These are the least well measured losses, because none lasts long enough to be reported: nobody logs a forty-second stoppage, and it happens thirty times per shift.

Quality — scrap and rework at start-up, and in steady state. Separating them has a practical interest: the former are dealt with through set-up and start-up validation, the latter through process stability.

It is this grid that makes OEE actionable: an OEE that falls without anyone knowing which of the six families moved leads only to a meeting.

What it is for, on the shop floor

OEE answers a management committee question: of all the time paid for to produce, what fraction came out as sellable parts? Without that figure, the choice between buying a machine and making better use of the one you have is settled by intuition — and intuition always says buy.

Three uses hold. Sizing: before a capacity investment, look at what the existing line would yield with fifteen more points. Prioritising: the weakest rate designates the project, and the family of losses that carries it designates the team that must lead it. Tracking, provided the three rates are tracked separately — a stable OEE can cover an availability going up and a quality going down.

One packaging station, figure by figure

Eight-hour shift, i.e. 480 minutes of opening time. Take off 30 minutes of planned breaks and 10 minutes of team briefing: the required time is 440 minutes. Those 40 minutes are not losses but decided stoppages — counting them as losses would amount to setting a target one has forbidden oneself to reach.

Then a 22-minute breakdown and a 48-minute format changeover are recorded. The operating time falls to 370 minutes: availability = 370 / 440 = 84 %.

The machine is rated at 60 parts per minute, i.e. a nominal cycle time of one second. It turned out 18,500 parts where 370 minutes at nominal rate allowed 22,200: performance = 83 %. Of those 18,500 parts, 370 are rejected at inspection: quality = 18,130 / 18,500 = 98 %.

OEE = 0.84 × 0.83 × 0.98 = 69 %. A check to do once and for all: the 18,130 good parts related to the 26,400 that the required time allowed give the same 69 %. The product of the three rates is not an accounting convention, it is the same quantity written differently.

What the single figure keeps quiet, the breakdown says. The heaviest loss is not the breakdown: the 17 % missing from performance are worth 62 minutes evaporated in minor stoppages, against 22 minutes for the breakdown that nonetheless kept three people busy. The workshop lives through the breakdown and suffers the minor stoppages without seeing them go by.

Last calculation, the one that triggers arguments between sites: related to the 480 minutes of opening time rather than the 440 required, the same shift shows an availability of 77 % and an indicator of 63 %. That is no longer an OEE (TRS), it is a TRG. Six points of difference, for the sole reason that nobody said which time it was related to.

Calculation

OEE = Availability × Performance × Quality

Availability = operating time / required time

Performance = (quantity produced × nominal cycle time) / operating time

Quality = conforming quantity / quantity produced

The nominal cycle time must be the reference rate of the machine, not the average rate observed: otherwise the performance loss disappears from the calculation by hiding in its own denominator.

The common mistakes

Managing OEE as a goal in itself. A line can show an excellent OEE by producing at full rate parts nobody needs — the machine has then been optimised by manufacturing overproduction, the first of the seven wastes. OEE is read on the stations that constrain the flow, not everywhere.

Taking as nominal rate the average observed over the last few months, on the grounds that the manufacturer’s value "is not realistic". A line that slows down lastingly then sees its reference come down with it, and its OEE go up: it is the only construction flaw that pushes the indicator up while the workshop is doing worse.

Publishing a monthly average. An average OEE of 70 % can cover twenty shifts at 78 % and two shifts at 20 %, and it is those two that carry the information. OEE is read shift by shift, with its spread.

On the shop floor — Food processing

On a bottling line, the measured OEE was good and the workshop was still running late. The loss was not on the machines but between the machines: OEE, measured station by station, did not see the line’s waiting times.

The other tools illustrated by this trade: food and beverage.

Frequently asked questions

What is the difference between TRS (OEE) and TRG?

Both are calculated the same way, but not on the same reference time. The TRS — OEE — relates production to required time, the time during which it had been decided to produce; the TRG relates it to opening time, breaks and planned stoppages included. The TRG is therefore always lower than or equal to the TRS, and the gap measures what was planned as non-production. The TRE, third of the family, relates it to total time.

What is a good OEE?

The value of 85 % has circulated as a benchmark of excellence since Seiichi Nakajima’s work on TPM, and has no normative standing. Above all it is not comparable from one workshop to another: a continuous process and a small-batch workshop with ten changeovers a day are not playing the same game. The only benchmark worth anything is your own history, measured with written definitions that have not moved.

How do you calculate the OEE of a line made up of several machines?

You do not average the OEE of each machine; the result would have no physical meaning. You calculate the OEE of the line on the machine that constrains the flow, taking as conforming quantity what comes out at the end of the line. An upstream machine running fast while downstream is blocked does not produce performance, it produces work in progress.

Do the maths

To enter your own times and see the breakdown into availability, performance and quality, open the OEE calculator, on the Kernel Ops website, beltup’s publisher.

Standardised term

NF E60-182 — Production equipment: overall equipment effectiveness (TRS), overall effectiveness (TRG), economic effectiveness (TRE)

The authoritative definition is in that standard. This entry is an explanation written by us: it does not reproduce the text and does not replace it.