CNC Machining Distortion and Stress Relief: The Complete Process Chain from Blank to Finish Machining

A part measures good on the machine, then goes out of tolerance the moment the fixture is released and it reaches the inspection room. Or it ships within tolerance and the dimensions drift slowly once it is running on the customer's assembly line. Nine times out of ten the cause is not insufficient machine tool accuracy — it is residual stress being released layer by layer during machining. Distortion control is not about "machining more carefully". It is about releasing stress at the correct point in the routing and matching the machining allowance to that sequence. This article follows the process chain in order and sets out where stress comes from, how to identify it, the stress relief parameters, the allowance plan, and the countermeasures for the two part families where problems concentrate.

1. Where Distortion Comes From: The Four Sources of Residual Stress

Before discussing distortion, draw a distinction: what you are looking at may be elastic springback, plastic deformation, or redistribution of stress inside the material. Each calls for a completely different response — and the category most often misdiagnosed on the shop floor as "the machine is not accurate" is in fact the fourth one, stress release.

Stress built into the raw material

Rolling, forging, casting and welding all leave stress inside the material. When a thick plate is rolled, the surface and the core deform by different amounts; cut it open or remove a large area of material from one side and the balance is broken, so the part bows towards the weakened side. Bar peeling, slab roughing and face milling of forgings are all classic cases of "it moves the moment you open it up". This type of stress is directional and follows the material grain, and it normally calls for one stress relief treatment before roughing and another after it.

Stress introduced by cutting

Cutting is not simply "removing material". It does three things at once: it causes plastic deformation, it produces work hardening in the surface layer, and it inputs cutting heat. A dull tool, an excessive radial depth of cut or insufficient coolant will all build a tensile stress layer at the surface, and thin-walled parts show it most clearly. This stress is concentrated in the outer few tens to a few hundred micrometres; it shows up as bowing on the machined face, burring at the edges, or a plate arching as soon as one slot has been milled.

Clamping stress and thermal distortion

Clamp too hard and the part is "bent" into tolerance while it is held, then springs back when released. Vacuum chucks, soft jaws and multi-point clamps produce the same false conformance if the pressure is unevenly distributed. Thermal distortion is a separate mechanism: the spindle and the cutting zone heat up, the workpiece expands with them, and the dimension that was correct during cutting comes out undersize after cooling. The dimensional drift caused by day-to-night temperature swings in batch production belongs here as well.

Heat treatment and phase transformation stress

Quenching, carburising and nitriding all introduce transformation stress and thermal stress. If a quenched part goes straight into finishing, grinding or finish turning removes part of the hardened layer, the stress rebalances, and the dimensions move. The correct approach is to place stress relief or stabilisation between heat treatment and finishing, not to skip it.

2. Diagnose First: Stress Distortion, or Clamping and Thermal Distortion

Get the cause wrong and every downstream process change is wasted effort. There are three checks that can be run at the machine, and they give an answer within ten minutes.

Check one: measure springback with the jaws released. Measure a critical dimension while the part is still clamped on the machine, then measure again after releasing it. If the difference exceeds one third of the tolerance, clamping stress is almost certainly the dominant cause.

Check two: measure again after thermal soak. Measure immediately off the machine, then again after 2–4 hours at controlled temperature. Dimensions that continue to change indicate internal stress being released; purely thermal expansion and contraction normally settles within 30 minutes.

Check three: check symmetry. Roundness, ovality and wall thickness variation at symmetrical positions are the classic fingerprint of stress release. If the distortion runs in one direction along the material grain, blank stress can be confirmed almost without doubt.

Only after the cause is clear should the process be changed: clamping problems are fixed with the fixture and the clamping sequence, thermal problems with cooling and in-process measurement, and only genuine stress problems require a stress relief operation.

3. Stress Relief Annealing: Temperature, Time and Furnace Loading

The objective of stress relief annealing is not to change the microstructure but to let residual stress relax by creep below the recrystallisation temperature. Choose the parameters badly and you either achieve nothing or damage the material condition.

Reference parameters for common materials

How the load is stacked matters just as much: parts should be supported off the hearth, never stacked under load, and the thermocouple must be attached to the workpiece itself rather than to the furnace chamber. For thick-walled parts, keep the heating rate at or below 100 °C/h. Skip these details and the furnace will run its programme while only half the stress is actually released.

Is a second cycle needed after semi-finishing?

For parts with tight accuracy requirements, the answer is yes. The usual sequence is: rough machining → stress relief → semi-finishing → stress relief again (at a slightly lower temperature) → finishing. The purpose of the second cycle is not to remove stress inherent in the material but to remove the surface stress newly introduced by roughing and semi-finishing. Omit it and nothing shows up in the short term; six months later the customer reports that "the accuracy has drifted", and the rework cost is far higher than one heat treatment cycle.

4. How to Sequence the Routing: Allowance Allocation for Roughing, Semi-Finishing and Finishing

More allowance is not safer. Too much allowance means finishing has to remove more material, cutting force and cutting heat both rise, and the part distorts more easily. Too little allowance and the distortion and surface defects from the previous operation cannot be removed.

A practical allocation rule: leave 1–3 mm for roughing (adjusted to part size and rigidity), 0.3–0.5 mm for semi-finishing and 0.1–0.2 mm for finishing. Thin-walled parts should be tightened further, with symmetrical material removal written into the routing: rough and bore alternately so that wall thickness stays as uniform as possible, and one side is never thinned first and left with unbalanced rigidity.

Another point that is often overlooked is how parts are stored between operations. Left flat on a rack, a semi-finished thin-walled part will gradually sag under its own weight; it should be stood on end or supported in dedicated tooling, and this should be stated on the operation card.

5. Countermeasures for the Two High-Risk Part Families

Thin-walled sleeves and rings

The core problem with thin-walled parts is insufficient rigidity: any radial clamping force produces ovality. Countermeasures in order of effectiveness: replace radial clamping with axial clamping; use soft jaws or an expanding mandrel to increase the contact area; increase the number of operations and take light cuts each time; where necessary, fill the bore with low-melting-point alloy or wax before machining; and schedule a stress relief cycle before finishing. When boring and turning the outside alternately, remove only 0.2–0.3 mm per pass so the stress is released progressively rather than all at once.

Slender shafts

The problem with a slender shaft is tool deflection and bending. A follow rest and a steady rest are the basic measures, but changing the direction of the cutting force is more effective: use reverse feed (the tool travels from the chuck towards the tailstock) so that the axial cutting force pulls the workpiece straight instead of bowing it. Clamping between two centres with a driving dog eliminates the bending introduced by chuck clamping. For parts with a very high length-to-diameter ratio, machining in stages is more stable than one continuous pass. On the heat treatment side, a slender shaft must be stress relieved after rough turning, otherwise it will bend progressively during finish turning.

6. How to Prove Distortion Is Actually Under Control

Distortion control cannot rest on "it feels much better now". It needs a closed data loop. Three things are worth standardising:

First, full dimensional inspection of the first part plus a 24-hour re-check. Measure the same part when machining is complete and again after 24 hours at controlled temperature, and record the difference in the first-article report. This is the most direct evidence of internal stress.

Second, process capability on critical characteristics. Run CPK studies on characteristics prone to drift, such as wall thickness, roundness and concentricity. Anything below CPK 1.33 means the process window is too narrow and the allowance or the stress relief needs to be adjusted.

Third, retained samples by batch. Keep 1–2 parts from each batch and re-measure them three months later. Stress release happens over time, and re-measuring retained samples uncovers the "passed at despatch, out of tolerance six months later" risk before the customer does.

If residual stress genuinely needs to be quantified, X-ray diffraction or the hole-drilling method can be used, but both require expensive equipment and carry high cost; for most production lines the indirect methods above are enough to locate the problem.

7. Common Mistakes and What They Cost

• Treating distortion as a machine problem: the machine is adjusted and the tools are changed over and over, and the problem stays exactly as it was. A clamp-release springback test first would save several days.
• Copying someone else's stress relief temperature: running aluminium alloy at steel parameters over-ages it outright; taking stainless steel above 400 °C reduces its corrosion resistance.
• Remembering stress relief only after finishing: by then the stress has already been released as distortion, and the part can only be scrapped or downgraded.
• Applying one allowance to everything: leaving 2 mm on every part means thin-walled parts distort even more because of the higher cutting force.
• Storing parts randomly between operations: semi-finished parts left flat and loaded lose their accuracy while they wait for the next operation.

What these mistakes have in common is that the cost does not appear in machining. It appears in assembly, on the customer's line, and in the after-sales claim six months later. Investment in distortion control essentially moves cost forward, out of after-sales.

8. Frequently Asked Questions

Q1: Does every part need stress relief? No. Parts with ordinary accuracy that are neither thin-walled nor slender can be controlled through sensible allowance allocation and clamping practice. Stress relief is needed for precision parts, thin-walled parts, slender parts, and parts where a large volume of material is removed from thick stock.

Q2: Natural ageing or stress relief annealing — which one? Natural ageing relies on long-term storage to let stress release slowly; the cycle is measured in months, it consumes floor space and it is not controllable. Stress relief annealing takes a few hours and gives repeatable results. Apart from a very small number of large castings, industry uses artificial ageing almost exclusively.

Q3: Does the part still need straightening after stress relief? Stress relief can introduce new bending, particularly in slender parts. Straightening should be scheduled after stress relief and before finishing, and because straightening itself introduces stress, a low-temperature stabilisation treatment may be needed afterwards.

Q4: Why must the stress relief temperature for stainless steel be kept low? Austenitic stainless steel precipitates chromium carbides in the 450–850 °C range, which makes it susceptible to intergranular corrosion. The stress relief temperature must therefore be held below 400 °C; where a higher temperature is required, only low-carbon or stabilised grades such as 316L or 321 can be used.

Q5: Can stress relief be done on a machining centre? No. Stress relief requires uniform heating of the whole part and slow cooling; it is a heat treatment operation and must be done in a furnace. What can be done on the machine is to control cutting heat and clamping force so that less new stress is introduced.

If you are trying to trace the cause of a distortion problem on a specific part, send us the drawing, the material grade, the current process routing and the measured distortion data. We will apply the diagnostic methods above and come back with process recommendations — not with a suggestion to buy new equipment.