OEE

How OEE is worked out

Three components, multiplied together.

Availability is the time the equipment actually ran, divided by the time it was scheduled to run. Scheduled, not calendar: a sorter that was never meant to run on a Sunday has not lost availability on a Sunday, and measuring against the calendar turns your shift pattern into an equipment fault.

Performance is what it produced while running, divided by what it would have produced at its rated speed for that same running time.

Quality is the share of that output that was right first time, needing no re-sort, re-label or rework.

Multiply the three. Because it is a product, OEE behaves like the perfect order rate: three components at 90 per cent each give 73 per cent, which surprises people the first time they see it.

OEE comes from manufacturing, and it travels into a warehouse only where a machine sets the pace. On sorters, conveyors, palletisers, automated storage and automated vehicle fleets it is the right tool. On a manual pick face there is no rated speed, so there is nothing to measure against and picks per hour is the figure you want instead.

There is no good number

Published OEE figures are the least comparable numbers in this glossary, because the whole result depends on the rated speed somebody chose and on what counts as scheduled time. A generous rated speed and a narrow schedule produce a flattering figure from an indifferent machine.

So fix your definitions, write them next to the figure, and read the three components rather than the product. The headline tells you that potential was lost. Only the split tells you where.

In practice

A parcel sorter is scheduled for 480 minutes and loses 60 to a jam, a belt changeover and a late start. Availability is 420 divided by 480, or 87.5 per cent.

Its rated speed is 6,000 parcels an hour, so in 420 running minutes it could have handled 42,000. It handled 33,600. Performance is 80 per cent.

Of those 33,600 parcels, 336 went to the wrong chute and had to be recirculated. Quality is 99 per cent.

OEE is 0.875 times 0.80 times 0.99, which is 0.693, or 69.3 per cent.

The interesting number is the 80. The sorter was not broken and it was not slow; for a large part of those 420 minutes it was starved, waiting for parcels from packing benches that could not feed it at 6,000 an hour. The fix is upstream, and the OEE report is the thing that pointed at it. The figures here are an illustration.

What moves the figure

• Whether the machine is fed. In warehouses, starvation is the usual cause of a poor performance component, and no amount of maintenance addresses it.
• Jam causes. Most jams trace back to packaging and label quality rather than to the equipment: odd-shaped parcels, poor tape, labels that will not scan.
• Planned maintenance against breakdown. Planned downtime can be moved outside scheduled time. Unplanned downtime cannot.
• The rated speed you measure against. If it came from a sales brochure rather than from a commissioning test, the performance component is fiction.
• Changeovers. A machine reconfigured between order types loses availability, and the loss is a planning choice about batching.
• Mis-sort causes. The quality component shares most of its causes with first time right, and the two reports should agree.