OEE calculator: availability x performance x quality, with a worked example

Every machine is scheduled to run for some planned production time. OEE asks one question about that time: how much of it was spent making good parts at the fastest speed the machine can sustain? If the answer is 100 percent, the machine never stopped, never ran slow and never made a bad part. In real plants the figure is far lower, and the useful part of OEE is that it splits the shortfall into three separate causes that different people can fix.

Those causes are stops (availability), slow running (performance) and bad parts (quality). A packaging line that loses most of its time to jams has an availability problem and needs maintenance and changeover work. The same line running at ninety percent of its rated speed all shift has a performance problem, which is a different conversation involving settings, material and operators.

OEE is the product of three ratios. Each one is a fraction between zero and one, and each one answers a different question about the planned time.

There is a handy shortcut and cross-check. Because the run time and total count cancel out, OEE also equals (ideal cycle time x good count) / planned production time. In other words it is the time it would take to make the good parts at ideal speed, divided by the time you planned. If your three-factor answer and this shortcut disagree, one of your inputs is wrong.

The calculator starts with these numbers, so you can follow the arithmetic against the screen. A filler is scheduled for 480 minutes. It stops for 60 minutes in total (a jam, a label reel change and a short wait for empty containers). Its ideal cycle time is 20 seconds per part. It makes 1,000 parts, of which 950 pass inspection.

The ledger in the calculator shows what the 66.0 percent hides. Of the 480 planned minutes, 60 were stops, about 86.7 were lost to running below ideal speed, about 16.7 were spent making parts that were not good, and about 316.7 minutes were fully productive. The biggest single loss is slow running, not the stops that everyone remembers, which is a common finding when a plant first measures OEE.

OEE is only as good as its inputs, and most arguments about OEE are really arguments about inputs. Planned production time comes from the shift schedule. Stop time is best taken from the machine itself: a PLC can run a timer while the machine state is stopped or faulted, and log each stop with a reason code, which is more reliable than operators writing times on a clipboard at the end of the shift.

Total count normally comes from a counter on a sensor at the end of the machine, and good count is total count minus the rejects counted at an inspection or reject station. These are ordinary counter and timer jobs in a PLC. If you want to practise building them, the counters page and the timers page walk through the instructions, and the data logging page covers how the totals reach a database or a report.

Ideal cycle time deserves care. Use the fastest rate the machine can sustain, from the design rate or a best demonstrated run, not the average of last month. If you use the average, performance will look perfect and hide the speed losses that OEE is meant to expose.

Total productive maintenance practice groups the causes of lost time into six categories, often called the six big losses. They map neatly onto the three factors, which tells you which factor to look at when a particular problem dominates.

Small stops are the hardest to see and often the most valuable to find. A one-minute stop usually gets written down. A five-second pause that happens two hundred times does not, yet it can cost more. This is why automatic logging from the PLC beats manual records.

• Counting planned breaks as downtime. The usual convention takes planned breaks out of planned production time. Some companies instead measure against all calendar time, a different figure called TEEP, so state which one you use.
• Using the average rate as the ideal cycle time, which hides speed loss.
• Mixing products on one line without changing the ideal cycle time. If two products run at different speeds, calculate each run separately or use a weighted ideal time.
• Counting reworked parts as good. Quality is first-pass yield, so a part that needed rework is not a good part for OEE.
• Comparing OEE across different machines or plants without checking that they use the same definitions. The number is for tracking one machine over time and for finding its biggest loss, not for ranking plants.
• Reading a performance figure above 100 percent as success. It means the ideal cycle time or the count is wrong.

You will often see 85 percent quoted as a world-class figure for discrete manufacturing, built from commonly cited component targets for availability, performance and quality. Treat it as a reference point and not a target. A batch process with long cleaning cycles and a high-speed bottling line are not comparable, and a machine that is not the bottleneck may not justify the effort needed to raise its number.

A better use of OEE is as a trend on one machine. Measure it with a fixed definition for a few weeks, find the largest of the three losses in the ledger, fix that first, and watch whether the number moves. Improving the biggest loss is worth more than polishing the total.