OEE = availability x performance rate x quality rate

where

availability =	(planned run time - downtime) x 100
	------------------------------
	planned run time

note: planned run time is the total time when needed

performance =	(units produced x design cycle time) x 100
	------------------------------
	actual run time

or	(cycle time x quantity produced) x 1000
	-----------------------------
	actual run time

(note that this element often is capped at a maximum of 100 %)

quality rate =	(total output - defects) x 100
	--------------------
	total output

or (parts right first time) / (completed cycles) x 100

Notice:

• OEE is a very severe test since it is the product of three factors. If each factor were a high 90%, then the OEE would be only 72%. (Actually, 85% is good performance in world terms).
• OEE is directly related to the 6 big losses:
  
  Breakdown and Changeover form the availability measure
  Minor Stoppages and Reduced Speed form the performance measure
  Non conforming items and changeover scrap form the quality measure

The components of OEE should be tracked separately. It is a good idea to display 4 graphs; one for overall OEE, and one each for the 3 elements. Below each graph, the reasons may be given in the form of a fishbone diagram.

Where there are several similar machines, say presses a very useful exercise is to identify the highest performing machine on each of the three dimensions, and then to calculate the current "best demonstrated practice". This would be the current best OEE in the shop if the best feature of each were used. This gives an immediate, attainable target. Find out why or how the best figures are obtained for the best machine in each category.

OEE figures often come as a shock to management. They should certainly be calculated before any capital acquisitions are made, and as an indicator for continual improvement activities.

Some plants try to calculate an overall OEE for the entire plant. This questionable practice involves weighting the various OEE machines on some criteria say by total output. The problem is so much time is spent arguing about the best weightings, you may never get around to actually improving OEE.