The probe says the pocket floors are flat. Then the rib comes off the fixture, goes onto the inspection table, and one end sits off the surface. Nobody misread anything. The first number was taken with the clamps holding the part straight; the second, after they let go.
That gap between the clamped part and the free part is the whole problem. Before you change feeds or blame the machine, check four things, in this order of convenience, and expect more than one of them to be involved.
Researchers at Tecnalia in Spain measured this directly. They mounted a laser sensor in a five-axis machine and scanned thin 2024-T3 aluminum plates before machining, while clamped, during unclamping and after.
On small plates, unclamping barely changed the result. On larger plates, with fixturing points more than 139 mm apart, unclamping alone changed distortion by up to 250 µm and altered the final shape. Their conclusion: treat unclamping as a process step and plan it before you cut.
First, sort out what moved it
- Blank stress. The distortion follows the plate lot or where the part was cut from the plate, and it changes as more material comes off.
- Clamping. The big jump happens between the roughed part clamped and the same part released, and it changes when you change clamp force or support.
- Heat. Clamped readings drift with time or temperature, and the free part settles back after it cools to room temperature.
- Cutting stress. One thin wall pulls in during the finishing pass and is still out after the tool has moved away, and it changes with tool condition or cutting parameters. If the wall springs back once the tool leaves, that immediate movement was elastic deflection under the cutting load.
1. Stress that was already in the plate
Thick aluminum plate is quenched and then stretched to relieve stress. Los Alamos researchers measured 75 mm thick 7050 plate before and after that stretch: it cut residual stress by about a factor of ten. What is left is small, but a rib can remove most of the plate. As you take material off, the stress that is left loses its balance, and a thin part has little stiffness left to resist it.
Weber and colleagues at TU Kaiserslautern and UC Davis modeled both the plate stress and the stress from cutting, using the real tool path, and checked it against test parts.
On parts with low initial plate stress and 3 mm walls, changing only the milling path cut distortion by 42%. Their simulation showed up to 69% for 2 mm walls. On parts with high initial stress, milling the inverse of the predicted shape onto the back face reduced distortion by up to 77%.
Their comparison was between milling paths inside the pockets, which is a different change from alternating faces, but it shows how much the milling path alone can move the result.
2. Change the removal sequence before the feeds
The usual shop answer is to rough both faces, let the part move, then finish. In one eMastercam thread, a machinist described reaching 0.02 mm flatness on an aluminum part with two roughing operations, stress relief with the part free, and finishing on both sides.
For a new rib, run it as a small trial. Take parts from the same plate lot. Keep the fixture, tools and parameters fixed, and change only the sequence. Measure the same points on every part after roughing, once clamped and once released. If the alternating route keeps shrinking that jump, sequence is your lever. If a repeat with a second plate lot moves the results more than the sequence did, look at the stock first.
Thin walls need the same thinking at a smaller scale. Sandvik Coromant recommends light, fast cuts with small depths of cut for thin walls, and step-support machining, working the wall down in overlapping passes so that uncut material stays beside it for support.
3. What the fixture was holding
If most of the movement shows up when you release the clamps, the fixture was holding the part straight against its own stress. Strap clamps press hard at a few points. A part finished in that state takes on the shape it was clamped in.
Two changes usually help. After roughing, release the part, let it settle and clamp it again with lower force before finishing. And spread the hold over larger, well-supported areas: vacuum fixtures do this, but only where the suction zones and supports sit, so lay them out under the thin sections.
4. Heat and cutting stress
If clamped readings change between morning and afternoon, or the part reads differently warm than cold, compare the same points only after the machine and the part have reached a repeatable thermal state, before you change anything else. If one wall pulls in during the finishing pass and stays displaced after the tool leaves, look at the tool and the cut: a worn edge or a heavy finishing pass on a thin wall adds stress right at the surface.
When it is the machine
Machine a simple test block, thick and symmetric, with the same program structure. If it comes off flat and the rib still bows, start with the rib’s stock, support and sequence. If the test block also drifts, or readings change with the time of day, check thermal growth and axis geometry.
An in-machine probe helps here: you can measure the part clamped, released and re-clamped without moving it to the inspection table between steps. Before you call the machine builder, keep those readings with the machine temperature and the program that cut the part.
How we’d set up a machine for long aluminum ribs
For ribs one to three meters long, we’d quote a machine from our gantry series. The LP1614 to LP3014 models run X travel from 1,600 to 3,000 mm, with tables up to 3,000 mm long, so a long part can be roughed on both faces and finished in one envelope.
For aluminum we’d specify a BT40 or motor spindle running at 12,000, 15,000 or 18,000 rpm instead of the 6,000 to 8,000 rpm belt-driven BT50. Those speeds suit the light, fast finishing cuts thin walls need.
For thin parts we’d add a vacuum table, and a workpiece probe so you can measure clamped and released states on the machine. We’ve supplied machines to customers making aerospace structural parts and large thin-walled aluminum parts.
Our article on aerospace machining on gantry centers covers the larger parts, and the one on cold plate flatness after joining covers a different kind of distortion. If you’re starting a new rib, send us the drawing with the plate size, alloy and temper, and the flatness requirement. We can then discuss the machine and workholding configuration.
Sources
- M. Casuso et al., Laser integration and novel nomenclature for multistage distortion measurement and geometry analysis in thin aluminum plates: applications in aerospace manufacturing, International Journal of Advanced Manufacturing Technology 143 (2026) 2939–2952
- D. Weber et al., Simulation based compensation techniques to minimize distortion of thin-walled monolithic aluminum parts due to residual stresses, CIRP Journal of Manufacturing Science and Technology 38 (2022) 427–441
- Los Alamos National Laboratory, residual stress in 7050 aluminum plate
- Sandvik Coromant, machining of aluminium wing ribs
- eMastercam forum, Let work piece settle after roughing?, October 2006
- CHANSIN CNC gantry series specifications
- Featured image: Magellan Aerospace, cropped
