Availability times performance times quality. 90 percent on each gives 73 percent overall, which sounds like a line with a modest problem. It is a line with three separate problems, each of which is losing you a tenth of the plant, and the single number tells you nothing about which one to go and look at.Availability times performance times quality. 90 percent on each gives 73 percent overall, which sounds like a line with a modest problem. It is a line with three separate problems, each of which is losing you a tenth of the plant, and the single number tells you nothing about which one to go and look at.

Not between plants, and often not between lines in the same plant, because the definitions move. Is planned maintenance in availability? Is a changeover? Is the ideal cycle time the machine builder's figure or the fastest you have ever actually run? Every one of those is a defensible choice and every one of them moves OEE by several points. Benchmarking against another site's number is comparing two different measurements that share a name.Not between plants, and often not between lines in the same plant, because the definitions move. Is planned maintenance in availability? Is a changeover? Is the ideal cycle time the machine builder's figure or the fastest you have ever actually run? Every one of those is a defensible choice and every one of them moves OEE by several points. Benchmarking against another site's number is comparing two different measurements that share a name.

Small stops. A line that stops for thirty seconds every four minutes and a line that stops for two hours once can report the same availability. They are not the same line, they do not have the same cause, and they do not have the same fix. If your system only logs stops over five minutes, it is not measuring the failure mode that is most likely to be your biggest loss.Small stops. A line that stops for thirty seconds every four minutes and a line that stops for two hours once can report the same availability. They are not the same line, they do not have the same cause, and they do not have the same fix. If your system only logs stops over five minutes, it is not measuring the failure mode that is most likely to be your biggest loss.

Performance loss is the quietest of the three because the machine is running the whole time. A line held at 92 percent of rate by an upstream restriction shows green on every screen in the building. It is only visible if the ideal cycle time in the system is honest, and there is always pressure to set it to something the line can comfortably hit.Performance loss is the quietest of the three because the machine is running the whole time. A line held at 92 percent of rate by an upstream restriction shows green on every screen in the building. It is only visible if the ideal cycle time in the system is honest, and there is always pressure to set it to something the line can comfortably hit.

Report the three components separately and let people multiply if they want a headline. Fix the definitions once, write them down, and stop changing them. Then use OEE for what it is genuinely good at, which is trending one line against itself over time. It is a bad number for judging a plant and a very good one for telling you whether last month's fix worked.Report the three components separately and let people multiply if they want a headline. Fix the definitions once, write them down, and stop changing them. Then use OEE for what it is genuinely good at, which is trending one line against itself over time. It is a bad number for judging a plant and a very good one for telling you whether last month's fix worked.