Availability, Performance and Quality

The three OEE factors are computed from different signals. Knowing which ones, before starting a project, tells you in advance what you will be able to measure and with what precision: a machine that does not count the Units it produces cannot give a Performance figure, and no calculation can derive it after the fact.

Availability

Availability = Run time / Planned time

Planned time is the time the machine was expected to be in production, and it is defined by the shifts. Run time is planned time minus unplanned stops: failures, waiting, unexpected blocks.

Required signal: the machine state, that is, a variable that takes distinct values at least for "in production" and "unplanned stop".

The distinction that matters is between planned and unplanned stops. A scheduled setup or a planned maintenance job is not an Availability loss: if it lands in the same category as a failure, the indicator penalises the company for being organised. How the categories are tracked is covered in machine states.

Performance

Performance = (Units produced x Nominal cycle time) / Run time

The nominal cycle time is the theoretical duration to produce one Unit at the maximum speed declared by the manufacturer, or the one defined by process engineering. It is the most delicate parameter of the calculation: if the nominal cycle is optimistic, Performance stays low forever; if it is cautious, the number sits near 100% and signals nothing.

Required signal: a count of Units produced. Either a counter that resets every period or a cumulative counter that only grows is fine, as long as you know which of the two it is.

Quality

Quality = Conforming Units / Total Units produced

Conforming means right first time, with no rework. Rejects and rework must be subtracted, otherwise the indicator measures volume, not quality.

Required signal, in order of preference:

1. a count of conforming Units, separate from the total produced;
2. a count of rejects, from which conforming Units are derived by difference;
3. neither: Quality stays at 100% by assumption and OEE comes out overestimated.

The third option is common at the start of a project, when rejects are still counted on paper. It is worth knowing: an OEE with no quality data is not comparable with that of a line which has it.

Recap

When one of these signals does not exist on the machine, it can be added in the field: a cycle sensor, a counter, a signal taken from the panel. It is a project choice, and it pays to make it knowing which OEE factor you are buying.