The Most Expensive Minute Is the One When the Machine Stands Still

Intuition says the price of a CNC part comes from cutting: the more metal to remove, the more you pay. In a real machine shop, however, the tool is often cutting for less than half of the time you are billed for. The rest is clamping, changeover, tool setting, measurement. This article is about that other, invisible half of the price - and how to shrink it.

The calculation that surprises people

Downtime takes up nearly half the cycle, even in series production. In the SME handbook example of a simple shaft from Ø24×38 mm bar, the activities during which the tool is not cutting take 0.66 minutes against 0.75 minutes of cutting. At plant level it is worse: MachineMetrics measured average machine utilization of about 26% in 2022.

The cost estimating handbook published by SME contains a cost table for machining a simple shaft from Ø24×38 mm low-carbon steel bar. The time breakdown looks like this:

The activities during which the tool is not cutting add up to 0.66 minutes against 0.75 minutes of cutting - nearly half the cycle. And this is a series-production example, where the setup was spread over many pieces. For a single piece the proportions flip: downtime can cost more than ten times the cutting itself.

And even that is the optimistic picture, because it describes a single cycle. At plant level it gets harsher: MachineMetrics, a platform that collects data directly from the controls of thousands of machine tools across hundreds of US companies, measured average machine utilization of about 26% in 2022 (22-30% month to month). A typical machine tool spends three quarters of its time not running any program at all - waiting for material, a changeover, a program, or an operator.

The anatomy of downtime, measured to a hundredth of a minute

You might suspect this is theorizing. It is not - these times have been measured and tabulated for decades. In the Polish time-norm tables of 1972, non-productive times have chapters of their own, precise to hundredths of a minute. A few examples for lathes - historical values, but an unchanged mechanism:

• organizational activities when starting and closing a job: 10-14 minutes per batch;
• tooling up a lathe for single-tool work: from 6-11 minutes (between centers) to 12-27 minutes (on a faceplate with an angle plate and counterweight); every additional tool adds 3-10 minutes;
• first-piece inspection: 3-8 minutes, plus another 5-15 for complex work requiring a consultation with the foreman;
• clamping and unclamping one piece: from 0.14 minutes in a pneumatic chuck to 6.5 minutes in a four-jaw chuck with difficult alignment - a nearly fifty-fold difference, for the same part;
• every measurement with a caliper, micrometer, or gauge: 0.1-1.35 minutes, more for higher precision.

Add it up honestly: a simple one-off turning job, three tools, four-jaw clamping, first-piece inspection - that is 40-60 minutes before any cutting starts, cutting that might be worth 10 minutes.

An important caveat: these are numbers from the era of manual machines, and their components look different today. But the total for one-off production is still of the same order - on a CNC mill, preparing a new job realistically takes half an hour to two hours, because the time saved on handwheels and trial chips has been taken over by CAM programming, workpiece locating, and paperwork. The line items changed; the scale of the phenomenon did not.

What CNC killed, and what survived

CNC machines wiped out the small change from Wołk's tables: changing spindle speed (0.06-0.45 min "by handwheel" in 1972) no longer exists, a tool change (0.6-1.4 min) takes seconds in a tool magazine, and touch probes for parts and tools automate the setting that used to be done with trial chips and a micrometer.

Those touch probes were covered as a novelty in the trade press back in 2008.

But the three heaviest items survived, and one new item joined them:

1. Changeover and clampings - vises, fixtures, locating; every additional clamping of the part means a new setting cycle and a new chance of error.
2. First-article inspection - just as mandatory today as in 1972, especially with tight tolerances, which we wrote about recently.
3. Organizational work - the drawing, the job card, the material, the quality paperwork.
4. CAM programming - an item Wołk never knew: in one-off production the program can cost more time than the machining.

The practical conclusion in the SME handbook reads the same today as it did then: when planning machining, minimize the number of machines, the number of changeovers, and the number of clampings - because these, not the cutting parameters, make the difference in cost.

Designed by Magnific

Batch mathematics, or why one piece costs "a fortune"

A single piece is expensive because it carries the entire batch setup time on its own. With a 90-minute setup, one piece carries the full 90 minutes, ten pieces 9 minutes each, and one hundred just 54 seconds each. That is why a prototype costs hundreds, and two small batches instead of one cost nearly two setups.

Downtime cost obeys a simple formula, which SME writes as $T_s = \frac{T_b}{Q}$: batch setup time divided by the number of pieces. That single fraction bar explains most of the "strange" phenomena in quotations:

• with a 90-minute setup: one piece carries the full 90 minutes, ten pieces carry 9 minutes each, one hundred carry 54 seconds each;
• that is why a prototype costs hundreds, even though "it is just five minutes of milling";
• that is why ordering 5 pieces now and 5 in a month realistically costs nearly two setups - more than 10 pieces at once;
• and that is why, as the SME authors note, there is no good setup formula in job-shop production - the owner of a small shop estimates it from experience. And experience, like any knowledge kept in one head, is not always available on Friday at 3 p.m.

How to shrink the invisible half of the price

Downtime shrinks mainly through fewer clampings, faster locating, offline tool presetting and CAM programming, grouping similar jobs into part families, and pricing setup explicitly in the quote. The shop's customer can help by consolidating orders into larger batches, asking for price breaks at different quantities, and not splitting one batch across several delivery dates.

For the shop, the levers have been known for decades; only the tools have changed:

• fewer clampings - multi-side machining, done-in-one setups, a sensible operation sequence;
• faster locating - modular fixtures, zero-point clamping systems, touch probes instead of dial-indicator hunting;
• offline tool presetting and offline CAM programming - the machine cuts while the programmer prepares the next job;
• part families - grouping similar jobs so one setup serves several items;
• honest setup pricing - not "included", not "roughly", but explicit and separate, because only then can you see what pays off.

For the shop's customer the conclusions are even simpler: consolidate orders into larger batches, ask for price breaks at different quantities, and do not split one batch across several delivery dates without a reason - every date is a separate setup, and someone has to pay for it.

Calculated, not guessed

In a good quote, downtime is a separate, computable line item, as in the 1972 tables. That is exactly how the MetronQ engine calculates it: setup driven by part complexity and the number of clampings, CAM programming, first-article inspection - all spread over the batch alongside the cutting cost itself, with price breaks visible for different quantities.

Those minutes are priced from the machine-hour rates and operation times the shop sets itself. In MetronQ quoting, the customer can see the unit price fall as the quantity grows - and stops wondering why one piece costs so much. Times and rates are calculated for the machines the shop enters and calibrates itself, as described under machines and calibration.

Because the most expensive minute in a machine shop really is the one when the machine stands still. The second most expensive: the one in which someone calculates all this downtime by hand.

PS What it costs to automate that calculation - check the pricing.

Frequently asked questions about setup time and CNC part price

Why is a single CNC part so expensive?

Because the whole batch setup time is divided by the number of pieces, following the formula $T_s = \frac{T_b}{Q}$. With a 90-minute setup, one piece carries the full 90 minutes, ten pieces carry 9 minutes each, and one hundred carry 54 seconds each. That is why a prototype costs hundreds, even though the milling itself takes minutes.

How much of CNC machining time is actual cutting?

Often less than half of the time you pay for. In the SME example, non-cutting activities take 0.66 minutes against 0.75 minutes of cutting, and for a single piece downtime can cost more than ten times the cutting itself. MachineMetrics data from 2022 show average machine utilization of only about 26%.

How long does it take to prepare a new job on a CNC mill?

Realistically half an hour to two hours. The time saved on handwheels and trial chips has been taken over by CAM programming, workpiece locating and paperwork. For comparison, the 1972 tables give 40-60 minutes of preparation for a simple one-off turning job before any cutting starts, cutting that might be worth 10 minutes.

Which non-productive times survived the move to CNC?

Changeover and clampings, first-article inspection, and organizational work survived, and CAM programming joined them; in one-off production the program can cost more time than the machining. Small items disappeared: changing spindle speed no longer exists and a tool change takes seconds. That is why planners minimize the number of machines, changeovers and clampings.

How can a customer reduce setup costs?

By consolidating orders into larger batches, asking for price breaks at different quantities, and not splitting one batch across several delivery dates without a reason. Every date is a separate setup that someone has to pay for. Ordering 5 pieces now and 5 in a month realistically costs nearly two setups, more than 10 pieces at once.