Lesson 4 of 10
Performance, and the losses nobody logs
Lesson three ended with a machine that stopped two hundred times and still reported ninety-four per cent availability. This is where those minutes went. Performance is the only fraction that catches losses no one wrote down — and it is also the easiest of the three to make disappear on purpose.
Step one
Read the material
Three ideas. Why this fraction sees what nobody logged, the two losses hiding inside it, and the one setting that can erase them both.
Concept one
It counts what came out, not what happened
Availability is built from an account of events. Somebody noticed a stop, judged it long enough to matter, and wrote it down. That account is only ever as complete as the person keeping it.
Performance asks nothing about events. It takes the ideal cycle time, multiplies it by the number of parts that actually came out, and divides by the run time you actually had. Ideal cycle × total count, over run time. Nothing in that formula requires anyone to have observed anything.
So if the machine should have made 900 parts in the time it was running and made 720, performance is 80% — and it does not matter whether the missing 180 went to a jam, a slow feed, an operator waiting for material, or something nobody has yet identified. The loss is counted whether or not it was noticed.
That is the whole value of this fraction, and it is also its limitation. It is an excellent detector and a poor diagnostician. It tells you truthfully that four hours of capacity is gone. It cannot tell you where. Finding that out is floor work, not arithmetic.
One caution that catches people: total count means every part the machine produced, good and bad. Scrap took cycle time to make, so it belongs here. It is punished later, in quality. Counting only good parts here charges the same defect twice.
Concept two
Two losses live in here, and they multiply
Performance carries the two SPEED losses from lesson one: small stops, and reduced speed. They are genuinely different problems, and the fraction splits cleanly between them.
Small stops — idling and minor stoppages. A jam cleared by hand, a sensor blinked, a misfeed, a part reseated. Below the logging threshold and usually resolved without a technician. Individually trivial and collectively enormous: forty seconds, two hundred times, is over two hours.
Reduced speed — the machine runs, but slower than it was built to. It never stops, so nothing is ever recorded.
Here is the piece worth carrying: performance is itself the product of two fractions. The share of run time actually producing, and the ratio of ideal cycle to actual cycle. Multiply them and you get performance exactly, with nothing left over.
That matters because it lets you attribute the loss rather than guess. If performance is 72% and the split says small stops cost 12 points and slow running cost 16, you now know which of two very different projects to fund. One needs a mechanism fixed; the other needs a reason found.
Concept three
Ideal cycle time is where performance goes to die
Every figure in this fraction is measured except one. Parts are counted. Run time is clocked. Ideal cycle time is chosen — and it is therefore the softest number in the whole of OEE.
The temptation is obvious and almost never deliberate. Nobody can find the design rate; the manual is gone or the line has been modified. So somebody uses what the machine typically does. And performance immediately reads close to 100%, because you have divided the machine's actual speed by the machine's actual speed.
The loss has not been fixed. It has been defined out of existence, and every point of speed loss the plant had is now invisible and permanently unrecoverable, because the baseline moved with the machine.
Ideal cycle time must be the fastest rate the equipment can sustain producing good parts — the nameplate rate, or if that is genuinely unknown, the best rate ever demonstrated in a real run. The best demonstrated rate is defensible: the machine has done it, so it is not a fantasy. What is not acceptable is the average.
The tell is a plant reporting performance in the high nineties while its operators can name three things that slow the line down. When measurement and the people who run the machine disagree, the measurement is usually the one that is wrong.
Step two
Work it with your hands on it
Four hundred minutes of run time — availability has already had its share. Set the two speed losses and watch performance split between them. The last slider is the one from concept three: change what you call ideal, and see the loss vanish without anything improving.
Hands on it, off the screen. Take a tally sheet and one hour at a machine. Draw a single column, and put a mark in it every time the machine stops for any reason at all, however short — including the ones the operator clears without breaking stride. Write nothing else. Do not classify, do not ask why, do not interrupt.
Multiply by the length of the shift. Almost everywhere, that number is larger than anybody in the building expects, and it is the first evidence most plants have ever had that small stops are a real loss rather than an annoyance.
Step three
Prove you understood it
One question per idea, new numbers every time.
All three solid.
You can compute performance from output alone, split it between the two losses inside it, and spot the baseline that makes both disappear. Lesson five takes the last fraction — quality, and why rework counts against you even when the part ships.
Modern Total Productive Maintenance (TPM)
Lesson four of ten. This class is taught from the Total Productive Maintenance series — the eight pillars, the six big losses, and the roadmap into AI and predictive manufacturing. Written by our founder, Dr. Gene A Constant, and donated to the Foundation.
Read on Kindle The whole class AI + TPMGSU is a 501(c)(3) nonprofit and book sales are its only income. A purchase is not a tax-deductible donation. As an Amazon Associate, Global Sovereign University earns from qualifying purchases. This lesson is free and always will be. Global Sovereign University: Different by Design. Better by Mission.