Platform / Maintenance

A breakdown doesn't cost what the repair costs.

The bearing was two hundred dollars. The machine was down nine hours waiting for it, two jobs moved, one shipped late, and nobody wrote any of that down anywhere. The repair is the only part of it that ever reaches a ledger.

Work the number yourself

There is no shortage of published figures for what an hour of downtime costs a manufacturer. Ignore them. They come from automotive and semiconductor plants where a stopped line is six figures an hour, and they tell you nothing useful about a four-machine cell in a twenty-person shop. Quoting them at you would be a sales tactic, not an argument.

Here is the arithmetic that does apply. Take one machine that went down unexpectedly in the last twelve months, and add up:

Most plants have never added that up for a single asset. Not because it is hard, but because the five pieces live in five places and none of them is the maintenance log. That is the real gap, and it comes before any software: not knowing what a PM is worth is precisely why the PM gets skipped.

The fire-fight burns your most expensive people

This is the line that never appears in any account, and in a small plant it is routinely the largest one.

When a machine stops unexpectedly, nobody dispatches a spare technician who happened to be idle. The shop pulls in whoever can actually solve it — the manufacturing engineer, the plant manager, often the owner. In a forty-person shop those are the same people who are already the buyer, the safety lead and the backup technician. They are also the only people who can do several other things, and those things now stop.

So an unplanned breakdown quietly converts your scarcest and highest-paid hours into phone calls chasing a part number. None of it gets coded to the repair. It comes out of the quote that went out late, the process nobody got to improve, the second shift that never got planned properly. A PM consumes an hour of the right person's time. The breakdown it would have prevented consumes a day from the three people you can least afford to lose — and it does it without warning, which is what makes it impossible to plan around.

Why PM programs quietly die

Nearly every shop has a PM schedule somewhere. Very few are running one.

It was usually built in a spreadsheet by someone who has since moved on. It triggers on calendar dates that take no account of whether the machine actually ran — so the saw that cut for sixty hours last month and the one that sat idle both come due on the same Tuesday. And when it does come due, the person who would do the work can't put a hand on the manual, doesn't know the part number, isn't sure the part is even in the building, and is already behind on something more urgent.

So it gets skipped. Skipping it costs nothing visible, right up until the day it costs a great deal. More on why that happens here.

What this does about it

• Triggers on run hours

  PMs come due on hours of operation, not on a calendar that doesn't know whether the machine ran. Belts, motors, pumps and filters on batch equipment wear by use, so that is what they are scheduled against. Machines that sat idle don't generate work nobody needs to do — which is how a schedule keeps its credibility with the people expected to follow it.

• Everything attached to the asset

  Manual links, part numbers, the manufacturer's suggested intervals and your own troubleshooting notes all live on the asset record. The person doing the job stops hunting for a binder, and the knowledge stops being resident in one head that eventually walks out of the building.

• The PM raises its own requisition

  A PM that calls for a belt, a filter or a pump creates the purchase requisition itself, against the right asset and the right work center. No separate request, no remembering to ask, no PM sitting complete-but-not-really because the part was never ordered. It is why maintenance and purchasing work best side by side, though each one runs on its own.

• History that accrues per asset

  Every PM, every breakdown, every part consumed, recorded against the machine. After a year you can answer the question that nobody in a small plant can answer today: which of these assets is quietly costing the most, and is it cheaper to keep fixing it or to replace it.

• Light enough to actually get logged

  A maintenance system that takes ten minutes of typing is a maintenance system with no data in it by March. Closing a PM is a few taps from a phone standing at the machine, with gloves on. If it isn't that, none of the rest of this page happens.

What it isn't

Worth being straight about the edges. This is not condition monitoring — there is no vibration analysis, no thermal trending, no predictive model, and no sensors on your machines. It does not need any connection to the machine tools at all, which is deliberate: it works on the day you install it, on equipment of any age, including the manual lathe in the corner. And it is not an EAM for a four-hundred-asset plant with a dedicated reliability team. If that is what you need, buy that.

Take the piece you need

Maintenance is a module, not a suite you have to swallow whole. It is not a slice of something that only starts working once you have bought the other six parts, and turning it on does not commit you to anything else.

If maintenance is what hurts, start there and leave the rest alone. If it turns out that what you really have is a purchasing problem wearing a maintenance costume — the PM was done on time, the part simply wasn't in the building — switch purchasing on beside it. Floor execution, quality and planning are there when you want them and out of your way when you don't. You add a module when it has a job to do, not because a license bundle said so.

All of it comes out of running plants rather than reading about them: built by people who have carried the pager, argued the capex and explained the late shipment. That is why it is sized and priced for a shop your size, instead of being an enterprise system with the expensive parts switched off.