The Machine Was Never Off, and Still Lost a Shift: CNC Utilisation and What It Does to Your Quoting

Quoting a machining job means deciding how many hours the machine will need, and that decision is almost always made from memory. The machine ran fine last week. It has been busy. Nobody remembers it stopping. So the quote assumes it runs, and when the job comes in late the explanation is that this particular job was awkward. It usually was not. #Manufacturing #RemoteMonitoring #Utilisation

The installation

A CNC machine tool, monitored for spindle load and utilisation. The owner says it is mainly useful when quoting, which is a less common reason to instrument a machine than watching for a breakdown, and a more interesting one.

Over the three days behind this post the spindle ran for 63.7 hours out of 72.3:

By the standards of most workshops that is very good. It is also eight and a half hours short, and where those hours went is not where anybody would guess.

Reading the utilisation chart

Two readings that sound like the same thing and are not.

Utilisation is the share of each reporting window the spindle was actually turning. Load is how hard it was working while it turned, as a percentage of what the machine can deliver. A machine can be at 100% utilisation and 20% load, which means it is busy doing nothing much, and that is a different problem from being idle.

The utilisation line here is pinned at 100% for most of its length. Just under two thirds of all the readings sit at a flat hundred. That is the shape of a machine that is genuinely working, and it is why the answer from memory is always “it runs all the time.”

But the line is not solid. It is cut, over and over, by notches down to 80, to 60, to 40. Each notch is a slice of a ten minute window where the spindle was not turning. Individually every one of them is trivial. There are enough of them that they add up to more than a shift.

Meanwhile the load line holds a tight band whenever the spindle is actually turning: 48% on average, never below 45.5% and never above 51.5%. It drops only where utilisation drops. Steady load with a stable ceiling is a healthy signal: the machine is not being asked for more than it has, and it is not degrading. The problem is not how hard it works. It is how often it stops.

Where the hours went

This is one day of raw power draw, which makes the texture easier to see than three days does.

The plateau is around 5.3 kW, and it is interrupted constantly. Drops to 4 kW, to 2 kW, a couple all the way to zero. Note what this chart does not have: a night. There is no overnight shutdown, no weekend cliff, no block of hours where the machine is off. Over the full three days only 50 minutes of readings were at zero power.

So the eight and a half missing hours are not downtime in any form a person would report. They are 123 short stoppages spread across the window, landing in 27% of every ten minute sample taken. Roughly one sample in four contained an interruption.

That is the finding, and it is the reason this machine is monitored for quoting rather than for failure. A short stoppage is a tool change that took longer than it should, a part that did not seat, an operator waiting on material, a program pause. None of them are faults. None of them get written down. Every one of them is invisible by the end of the shift, and together they are a full working day a week.

What this does to a quote

The numbers below are a worked example, not a promise. Put your own figures in.

Take the ordinary version of this mistake. You quote a job at 40 machine hours, priced on the assumption that the machine delivers 40 hours of cutting in 40 hours of calendar time. At the measured rate it delivers

leaving a shortfall of

You are 4.8 hours short, which either comes out of margin as overtime or comes out of the delivery date.

Do that across a year of quotes and the shape of the business changes. Not dramatically, which is the difficulty. Twelve per cent is small enough to absorb, blame on individual jobs, and never fix.

Against that:

• Quote on the measured number, not the nameplate. A machine that runs at 88% should be quoted at 88%. The quote gets slightly less competitive and considerably more accurate, and accurate is what makes a workshop money over a year.
• Knowing the rate is worth something even if you never improve it. It converts an unexplained overrun into a budgeted, predictable cost.
• Short stoppages are the cheapest capacity you will ever buy. Recovering a third of them adds about four per cent utilisation with no new machine, no new shift, no capital. You cannot recover what you have not counted.
• The record also answers the question a customer asks when a job is late, which is what actually happened. Being able to answer it precisely is worth more than being able to argue about it.

The honest framing is that this device will probably never prevent a catastrophe. It just stops a workshop from quoting against a machine that does not exist.

What to measure if you run one of these

Related installations

Power draw, load, bearing temperature and vibration run across the rest of the plant, and what differs is the reason each machine is watched. A production line drive motor is instrumented for the vibration trend and almost nothing else. A plant air compressor is there because compressed air costs more than lighting and heating together, and its owner needed somewhere to prove it. An injection moulding machine counts cycles and short stops for the same quoting reason this machine does, because quoted capacity and actual capacity have never been the same number.

Two more are worth opening for the failures behind them. A steam boiler feed pump threw a bearing and took the boiler with it, and a materials test rig once stopped on a Friday night and was not found until Monday, which is the same lesson this post makes about short stoppages, scaled up to a whole weekend.

See it running

The device page shows the live readings, the charts above in their interactive form, and the alert history, including the short stoppage clusters that triggered rules of their own. If the device link below ever stops resolving, the public device gallery lists everything currently shared, and devices come and go from it as owners decide.

To put your own machine on a chart, you can send readings without creating an account and see them arrive on a live chart in a couple of minutes, or read how to connect a first device.

Open this device Monitor your own equipment

Comments

Loading comments...