Count the setups before you name a price

What does one more setup cost?

A setup is one clamping of the part in a known orientation. Each one brings three kinds of time. Setting up: mounting and aligning the workholding, loading and checking the tools, setting the work origin and measuring the first part. Programming: a program or program section of its own, with its own origin. And on every part, clamping and unclamping, often with a check measurement.

Setup time falls once per batch and programming time once per program; clamping comes back with every part. That difference decides what an extra setup does to the price at 1, 10 or 50 off. So keep the three times apart, in the estimate as well.

The number of setups and the time per setup are two separate levers. A zero-point clamping system or a fixture made ready in advance shortens setting up, yet the part is still reclamped as often. Lean calls this turning internal setup, work that can only happen with the machine stopped, into external setup, work done while the machine is still running. So book the gain in the time per setup, not in the count.

How many setups does this part need?

Start with the features, not the machine. For each feature, note the machining direction the tool can reach it from: +Z for the pocket in the top, -X for the two threaded holes in the side face, -Z for the face on the underside. Features that share a direction belong together.

In 3-axis milling one side faces the spindle, and every other direction is a new setup. A feature that can be reached from two directions, such as a through hole, moves into a setup that already exists. Whatever is left after that shuffle is the lowest count you can expect.

What makes a side unreachable?

Accessibility is more than a direction. The tool has to reach the feature without the holder or the spindle touching the part or the workholding. A deep, narrow pocket needs a long cutter, and on a long overhang that cutter starts to chatter before it reaches depth. The side is reachable on paper and not at the machine.

The workholding takes up room as well. The face held in the vice jaws cannot be machined in that setup. A clamping margin under the part gets round that. It costs material, and the margin has to come off again in a later setup.

Does a 5-axis machine save setups?

Often, yes. In 5-axis milling the machine tilts the part or the spindle, so the top and four side faces come within reach in one setup. The housing in the example goes from three setups to two: everything from above and from the sides in the first, the underside in the second.

A step in between is a 3-axis machine with a fourth axis: a rotary table that turns the part to a new position between two operations. The side faces around that axis then come within reach in one setup, the end faces do not. Each added axis makes the machine dearer, and each reclamping that falls away is one less moment for an error between features to creep in.

A 5-axis machine has limits of its own. The usable work envelope can shrink once the table tilts, and part and workholding have to clear the spindle in every position. A 5-axis machine also usually carries a higher hourly rate. So the question is whether the setup time saved outweighs that rate across the whole batch.

Which features have to share a setup?

Tolerances between features pull them together. Two holes with a position tolerance of 0.02 mm relative to each other pick up the repeatability of the workholding and the measuring error of a new work origin every time the part is reclamped. If that stack-up is larger than the share of the tolerance left for it, they go in one setup, even when that means a less convenient direction.

A shop without a 5-axis machine can sometimes hold the same requirement across two setups on a 3-axis machine, with an accurate fixture and tight inspection. Both routes are valid: which one fits depends on what your shop owns and runs reliably. How a requirement like that steers the route is covered in what a 0.02 position tolerance does to the route.

Then work out the sequence back to front: the last setup clamps on faces that are already machined. How to plan it is set out in the setup plan that starts with the last setup.

What does an extra setup do to the unit price?

A worked example with the same housing. The third setup on the 3-axis machine takes 45 minutes to set up and 30 minutes to program, once. Every part goes in and out of it one more time: 3 minutes. The hourly rate is €80. What one part carries of the one-off time is its one-off share per part.

As a formula: extra cost per part = (setup time + programming time) × hourly rate ÷ quantity + clamping time per part × hourly rate. At 10 off, the extra setup is mostly one-off time: €10 of the €14. At 50 off, the one-off share has dropped to €2, and the €4 of clamping per part is still there. A bigger batch thins out the setup time, not the work on each part.

On a large batch the route itself can change. A fixture or a machine with more axes that does not pay at a handful of parts earns its keep once the clamping time saved per part comes back often enough. So at every quantity break, cost the other route as well.

That is why the setup count belongs in every quantity break. An estimate that uses one average setup time per part underprices small batches of many-sided parts, and overprices simple ones. How one-off time shapes the whole set of quantity breaks is covered in why a small batch costs so much per part.

How do you keep the count on record?

Write the count down per setup: the machining direction, the workholding, the features it covers and the setup time. A colleague can then check it, and on a repeat part you can see straight away which setup goes faster when the fixture from last time has been kept.

A count that lives only in the head of the manufacturing engineer walks out of the door with that colleague. The ERP does not catch it: it keeps the hours booked on the last order, not why there were three setups and not two. Count two on the repeat order and, at 10 off, the quote is €14 per part short, and that comes out of the margin. So record the count in the estimate itself, setup by setup, beside the price it carries.