Availability, and what really stops a machine

Concept one

A stop is not a breakdown, and not every stop counts

Stops come in three kinds, and only two of them touch availability.

Unplanned stops — the breakdown. Something failed, nobody chose it, somebody got called. This is the one everyone already counts.

Planned but productive stops — changeover, setup, tooling, first-piece approval. You chose to do it, and you are not making parts while it happens. These count against availability. Choosing a stop does not make it free. This is where plants most often flatter themselves: the changeover was on the schedule, so it goes in the excused pile, and forty minutes a day disappears from the number.

Excused stops — breaks, shift handover, planned maintenance, no demand. These come out of the denominator, not the numerator. They do not lower availability; they shrink the window it is measured over. Lesson two showed what that choice is worth, and it was often twenty points.

One test settles almost every argument: could this machine have been making good parts during that time if you had wanted it to? If yes, it counts against you. If the answer depends on who you ask, write the rule down before you next report a number.

Concept two

Changeover is the stop you can shrink this month

Breakdowns are the loud loss and the hard one. Fixing them means condition monitoring, spares policy, root-cause work — months of it. Changeover is the quiet loss and the easy one, and it is usually the larger of the two.

The method is SMED, and its whole insight is one distinction. Every task in a changeover is either internal — it can only be done while the machine is stopped — or external — it could be done while the machine is still running the previous job.

Fetching the next die is external. Finding the right spanner is external. Warming the tool, staging the material, printing the setup sheet, checking the first-off gauge is calibrated: all external. Most plants do all of it internally, with the machine standing idle, because nobody ever drew the line.

So the first pass is not to work faster. It is to watch one changeover with a stopwatch, write down every task, and mark each one internal or external. Then move every external task outside the stop. Halving a changeover on the first attempt is ordinary, and it costs nothing but attention.

Only after that do you streamline what is genuinely internal — quick-release clamps instead of bolts, shims cut to a stop, settings recorded instead of dialled in by feel.

And keep the clock honest, as lesson one insisted: the changeover starts at the last good part of the old run and ends at the first good part of the new one. Not when the machine turns over.

Concept three

The stops too short to log never reach availability at all

Every plant has a threshold, whether or not anyone wrote it down. Below some length — three minutes, five, ten — a stop is not recorded. The operator clears it and carries on.

So a machine can stop two hundred times in a shift and still report ninety-four per cent availability, because none of those stops was long enough to become an event. The number is not lying. It is answering a narrower question than the one you asked.

Those minutes are not gone. They surface in performance, as fewer parts than the run time should have produced, which is lesson four. But they arrive there anonymously — as a gap between expected and actual output, with no cause attached — and a loss with no cause attached does not get fixed.

This is also the difference between sporadic and chronic failure. A sporadic breakdown is dramatic, obvious and gets a project. A chronic fault — the jam that happens every twenty minutes and takes forty seconds — is tolerated, worked around, and eventually stops being seen as a fault at all. It is almost always the more expensive of the two.

The defence is not a better formula. It is a tally sheet at the machine and one hour of somebody's attention.