Guides / Design

The Part Was Flat Until the Clamps Came Off

September 21, 2026 by The Alu CNC Parts editorial desk / how this is checked

A shop machines a plate, everything measures correctly in the fixture, the clamps come off and the part rocks on the surface plate. This is one of the most common and most frustrating failures in aluminum machining, and it is rarely anyone’s fault in the way people assume.

Where the stress comes from

Aluminum arrives at a shop already loaded with internal stress, from two sources.

Rolling. Plate and sheet are squeezed between rollers, which stretches the surface relative to the core.

Quenching. Heat treatment to a T6 temper ends with a rapid quench. The outside cools and contracts first, the core cools later, and the part ends up with compression on the outside and tension inside.

While the plate is intact, all of this cancels out and it sits flat. It is a spring that is holding itself.

What machining does

Cut a pocket and you remove material that was carrying part of that load. The remaining metal moves until the forces balance again.

Three things make movement worse:

1. How much you remove. Taking half the starting volume out moves far more than skimming a face.
2. How asymmetric the removal is. Pocketing one side of a plate is the worst case. Removing the same volume equally from both faces is much safer.
3. How thin the result is. A thin part has less stiffness to resist the stress it is left holding.

The fixes, roughly in order of effectiveness

Use stress relieved material. T651 plate is stretched after heat treatment specifically to even out rolling and quench stress. It should be the default on any plate part. It reduces the problem substantially and costs almost nothing.

Use cast tooling plate. MIC-6 and its equivalents are cast and stress relieved, so there is very little stress to release. For jig plates and fixture bases this is the right answer, as long as you can live with the lower strength.

Rough, rest, finish. Rough machine the part leaving 0.5 to 1 mm of stock, unclamp it, let it move overnight, then re-fixture and take the finishing cut. The roughing releases most of the stress, the rest happens while it sits, and the finish pass cuts a now-stable part true. This is the standard professional answer and it costs a second setup plus a day.

Machine both sides evenly. Where the design allows, take material from both faces alternately rather than hollowing out one side.

Clamp gently. Over-clamping bends the part before machining even starts. It measures perfectly in the fixture and springs out afterwards. Vacuum fixtures and low clamping force solve this.

Design symmetrically. Ribs on both sides, pockets balanced about the neutral axis. This is free if it is done early and impossible if it is done late.

Say it on the drawing

The single thing that prevents this problem commercially is telling the shop that flatness matters, before they quote.

FACE 'A' FLAT WITHIN 0.05 IN THE FREE STATE.
MATERIAL: 6061-T651 OR MIC-6.

“In the free state” is the important phrase. It means unclamped, sitting on its own, which is how the part will be used. Without it, a shop can legitimately measure the part while it is still held down.

A drawing with that note gets quoted with the rough, rest and finish sequence built in. It will be more expensive than the naive quote, and it will produce parts that are actually flat. The cheaper quote is only cheaper until the first batch is rejected.

Sources

Figures in this guide are checked against the following. Published values are typical ranges for test material, not a guarantee for your part.

• Residual stress
• Quenching
• Stress relief and annealing

Frequently asked questions

Keep reading

• Tolerance Every Feature and You Will Pay for Every Feature
• The Wall Is Not Too Thin Until It Starts Moving