CNC Processing Deformity: Why Parts Move After the Cut
CNC processing deformity is any deviation from the drawing that appears during or after machining: bow, twist, taper, wall collapse or a bore that closes up overnight. This page is for engineers and buyers who need to know which mechanism is bending their part and what actually fixes it. We cover heat, residual stress, clamping, tool load and thin-wall behavior, then give the judgment rules we use in our own shop.
In this article
- 1
- 2
- 3
- 4
- 5
- 6
What actually causes CNC processing deformity
Metal does not bend because the machine is weak. It bends because energy entered the workpiece and did not leave evenly. Three things carry that energy: cutting heat, the stress already frozen inside the raw stock, and the force of the fixture holding the part still. Each one behaves differently, and each one needs a different countermeasure.
Cutting heat is the easiest to see. A 12 mm carbide end mill at 3,000 rpm in 6061 aluminum pulls most of the heat into the chip, but in 316 stainless or Ti-6Al-4V the tool edge and the workpiece take a much larger share. The top surface expands, the bottom stays cool, and the part bows toward the tool. After it cools, the bow reverses direction. This is why a part can measure in tolerance on the machine and fail inspection two hours later.
Residual stress is harder to see because it is invisible until material is removed. Rolled plate and extruded bar are not stress-free. They are balanced: tension on one side, compression on the other. Machine away 60 percent of one face and that balance breaks. The part moves to find a new equilibrium. No amount of coolant stops this. Only stress relief before roughing, or symmetric material removal, does.
Clamping force is the third mechanism and the most common in thin parts. A vise or chuck holds the part in a deformed state. The tool cuts a true circle. Release the clamp and the part springs back, so the circle becomes an oval. The error is not in the cut. It is in the release. If your out-of-round only shows up after unclamping, this is the cause.
- 1Thermal bowReverses after cooldown; shows up as taper or flatness drift
- 2Residual stressAppears as movement after each material removal pass
- 3Clamping spring-backError only visible after the fixture is released
How material choice sets the risk level
Some materials forgive you. 6061-T6 aluminum machines fast, conducts heat well, and holds a wall down to roughly 1 mm without much drama if the passes are light. Magnesium AZ31B and AZ91D are even more thermally stable but bring chip-fire risk, so they need different handling, not different geometry.
Stainless 304 and 316L sit in the middle. They work-harden, they hold heat at the edge, and they move when you take a heavy roughing pass on one side. 17-4PH in the H900 condition is worse because it is strong and has a low thermal conductivity. Expect more cutter pressure, more deflection and more spring-back on unsupported walls.
Titanium TC4 (Ti-6Al-4V) is the hardest common case. Thermal conductivity is roughly 7 W/m·K, so heat stays at the cut. The material also has a low elastic modulus, about 114 GPa, which means the wall deflects away from the tool and then springs back. Thin titanium ribs are the classic deformity part. They need low radial engagement, high feed per tooth and a rigid, well-supported setup.
Plastics behave on a different axis. POM and PA move with humidity and temperature, so a bore that fits at 20 °C may not fit at 35 °C. PEEK and carbon fibre are abrasive and stiff, and they can delaminate at the edge. For these, deformity is often a measurement problem first and a machining problem second.
- 1Low risk6061, 6082, 2024, brass C36000, magnesium
- 2Medium risk304, 316L, 4140, 17-4PH, POM
- 3High riskTi-6Al-4V, Inconel, thin-wall 7075, PEEK
Thin walls, long bores and open sections
Geometry decides how much the material can move. A solid block has nowhere to go. A 1.5 mm wall on a 100 mm deep pocket has a lot of room. The rule we use is simple: if the wall thickness is less than 10 times the tool diameter, treat it as thin and plan the sequence before the first cut.
Long bores behave like tubes. Boring a Ø25 mm hole through 200 mm of steel removes a ring of material and lets the remaining wall relax inward. The bore measures on size when the boring bar leaves, then closes 0.01 to 0.03 mm after the part cools. A sizing pass after a stress-relief pause is the usual fix.
Open sections, such as a C-channel or a frame with one face removed, twist rather than bow. The twist follows the toolpath direction. Cutting both sides in the same operation, or alternating passes left and right, keeps the stress field closer to balanced and cuts twist noticeably.
Asymmetry is the amplifier. Any part where one face has 80 percent of the material removed and the other face has almost none will move. Designers can reduce this by adding a balancing pocket on the light side, even a non-functional one. It costs a little cycle time and saves a straightening operation later.
Machining parameters that hold the part still
Roughing is where deformity is created. A heavy radial depth of cut, 60 to 70 percent of the tool diameter, pushes the part hard and leaves a thick layer of stressed material. High-efficiency toolpaths with 10 to 15 percent radial engagement and a deeper axial cut spread the load along the flute instead. The part sees less peak force, and the wall has less reason to deflect.
Climb milling on the finishing pass leaves a compressive surface and a cleaner edge. Conventional milling on a finish pass can pull the wall into the tool and leave a witness mark where the flute entered. On thin walls, a single finishing pass with a sharp tool beats three light passes with a dull one.
Coolant strategy matters more than most people expect. Flood coolant removes heat from the part, but it also creates a temperature gradient if it only hits one side. Through-spindle coolant keeps the cut zone stable. For titanium, high-pressure through-tool coolant is close to mandatory on deep pockets.
A stress-relief pause between roughing and finishing is cheap insurance. Let the part sit at room temperature for several hours, or run a low-temperature thermal cycle if the material allows it, then take the finishing cuts. The part has already moved once. The finishing pass then cuts the final geometry, not a transient shape.
- 1Roughing10–15% radial engagement, deep axial cut, high feed per tooth
- 2FinishingClimb milling, sharp tool, single pass on thin walls
- 3CoolantThrough-tool, high pressure, symmetric coverage
- 4SequencingRough, relax, finish; alternate sides to balance stress
Workholding that does not imprint itself on the part
Every fixture applies force. The question is whether that force stays after the part leaves the machine. A standard vise with 20 kN of clamping pressure on a 6 mm wall will deform it. The cut is accurate. The release is not. Soft jaws machined to the part profile spread the load over more area and cut the peak pressure by a large margin.
For thin plates, vacuum chucks and magnetic chucks distribute force across a face instead of a point. They are not always strong enough for heavy roughing, so a common pattern is vacuum for finishing and mechanical clamping for roughing. The part gets the best of both.
Support is as important as clamping. An unsupported wall vibrates, and vibration shows up as chatter and then as a tapered or wavy surface. Filling a pocket with low-melt wax or a water-soluble compound supports the wall from the inside without adding clamp marks. The filler is removed after machining with warm water or a controlled melt-out.
For long shafts and thin tubes, a steady rest or a tailstock reduces sag. A Ø40 mm shaft with 500 mm of unsupported length will deflect under its own weight, and the deflection changes as material is removed. Supporting both ends and the middle keeps the axis straight.
Matching the deformity symptom to the likely fix
Use the symptom you can measure, not the one you suspect.
| Symptom | Likely mechanism | First fix to try | When it is not the answer |
|---|---|---|---|
| Bow reverses after cooling | Cutting heat | Through-tool coolant, lower speed | Part is already stress-free stock |
| Movement after each pass | Residual stress | Stress relief before roughing | Thin wall deflects under tool load |
| Oval bore only after unclamping | Clamping spring-back | Soft jaws, lower clamp pressure | Bore is truly out of round on the machine |
| Twist along the long axis | Asymmetric removal | Alternate passes on both faces | Fixture is not flat to begin with |
| Chatter marks on a thin wall | Low stiffness | Fill pocket, reduce radial engagement | Tool overhang is the real problem |
| Bore closes 0.01–0.03 mm later | Thermal contraction | Sizing pass after a cooldown pause | Material has high residual stress |
| Flatness drifts overnight | Stress relaxation | Rough, relax, finish sequence | Part is too thin for any stable setup |
Pick the fix by what the part does after it leaves the machine
If the part measures correctly on the machine and moves after unclamping, fix the workholding and the stress-relief sequence first. If it measures wrong on the machine, fix the toolpath, the tool and the parameters. Do not chase a clamping problem with a parameter change, and do not chase a thermal problem with a new fixture.
Questions engineers ask about CNC processing deformity
How do I know whether the deformity is thermal or stress-related?
Measure the part in three states: on the machine with the clamp on, on the machine with the clamp released, and at room temperature two hours later. If the error appears only in the third measurement, it is thermal. If it changes after each roughing pass, it is residual stress.
Thermal movement is usually 0.005 to 0.03 mm on a 100 mm feature and it reverses direction as the part cools. Stress-driven movement is often larger and does not reverse.
Can a finishing pass correct a part that already moved?
Yes, if there is enough stock left. This is the reason we leave 0.3 to 0.5 mm on thin walls and critical bores for the finishing pass after a relaxation pause.
It does not work if the part has already twisted across a large flat face. In that case, the finishing cut follows the twisted surface and the flatness error stays.
Does a slower spindle speed always reduce deformity?
No. Lower speed reduces heat but raises the chip load per tooth, which raises cutting force. On a thin wall, force is often the bigger problem. The useful move is to keep the surface speed in a reasonable range for the material and reduce radial engagement instead.
Which materials are the worst for thin-wall deformity?
Ti-6Al-4V and Inconel are the hardest common cases because of low thermal conductivity and high strength. 7075 aluminum in a thin wall is also difficult because it is strong and less ductile than 6061.
If a design can use 6061-T6 instead of 7075, the deformity risk drops noticeably with almost no strength penalty for many brackets and housings.
Do you stress-relieve raw stock before machining?
For parts with tight flatness or thin walls, we rough machine, let the part relax, then finish. Raw stock certified as stress-relieved is available for some aluminum and steel grades. We check the drawing and the material certificate before choosing the sequence.
When the part allows, symmetric removal on both faces is a cheaper alternative to a full thermal stress-relief cycle.
How is deformity checked before shipment?
We inspect 100 percent of parts before shipment, with raw material checks, in-process monitoring and a final dimensional inspection. For thin or tight-tolerance parts, the final check happens after the part has reached room temperature, not straight off the machine.
Inspection reports are available on request. Tolerances are held to ±0.005 mm where the drawing requires it.
Send the drawing before the deformity starts
Upload your part and get a quotation plus a free DFM analysis within 12 hours. We will flag thin walls, asymmetric removal and tight flatness before the first cut, not after.
12-hour quoteFree DFM analysis100% inspection
More machining notes from the shop floor
We publish setup notes, tooling trials and inspection data from the factory floor.
