Stress relieving welded machine bases: thermal, vibratory, or neither
By Ryan, independent sourcing agent, Ho Chi Minh City, Vietnam
Every weld pass pulls. The molten bead contracts as it cools while the cold parent metal around it resists, and what remains is locked-in tension near the weld balanced by compression further out. In a bracket this is irrelevant. In a machine base where you are about to remove 6 mm of material from one face to hit a 0.02 mm flatness, it is the whole problem — because cutting that face unbalances the system, and the part moves to find a new equilibrium.
That movement is why a frame can pass inspection in the shop and be out of tolerance when it arrives.
What the methods actually do
Thermal stress relief
The part goes into a furnace, is brought up at a controlled rate to roughly 580–620 °C for carbon steel, soaked for about an hour per 25 mm of section thickness, then cooled slowly in the furnace to around 300 °C before coming out into still air.
At soak temperature the yield strength of the steel has fallen far enough that residual stresses above the reduced yield relax by local plastic flow. The mechanism is well established and the result is predictable: residual stress typically drops to 10–20% of its as-welded level.
The costs are real but not primarily financial. It needs a furnace big enough for the part, which is the binding constraint in most regions. It consumes about two weeks of calendar time, most of it the controlled cool. It puts scale on the part, so machining allowance has to survive the treatment. And it can distort thin or unsymmetrical fabrications if fixturing is careless.
Vibratory stress relief
The part is instrumented, a mechanical exciter is clamped to it, and it is driven at or near resonance for twenty to forty minutes. Proponents describe local plastic flow at stress concentrations; the treatment is applied in the shop, takes under an hour, needs no furnace, and leaves no scale.
The honest summary of the literature is that vibratory treatment reliably improves dimensional stability — the part moves less in subsequent machining and service — while the evidence for bulk residual stress reduction comparable to thermal treatment is much weaker. It is a real effect and a useful one. It is not thermal relief with the inconvenience removed, and a shop presenting it that way is telling you something about the shop.
Where it genuinely fits: parts too large for any available furnace, parts that cannot take furnace scale, and as a supplementary treatment between roughing and finishing after a thermal cycle has already been done.
Natural aging
Leave the weldment outdoors through some months of thermal cycling and it relaxes on its own. Old machine tool builders did this with cast iron beds and the practice has a real basis.
It is defensible when the shop has a genuine aging yard, the part is scheduled far enough ahead to sit in it, and the schedule is documented. It is not defensible as a retroactive justification for having skipped a process step. If a quote comes back conspicuously cheap and the answer to “what is your stress relief route?” is “we age it naturally,” ask which yard, and for how long, and where that is recorded.
Choosing, in practice
| Situation | Route |
|---|---|
| Machined flatness ≤ 0.05 mm over 1 m | Thermal, no substitution |
| Precision frame, part exceeds furnace envelope | Rough machine → vibratory → rest → finish machine |
| Structural weldment, no tight machined geometry | Often none required; say so explicitly |
| High-value frame, long service life expected | Thermal after welding, vibratory between rough and finish |
| Thin-wall or strongly asymmetric fabrication | Thermal with a fixture plan agreed in advance |
The second row is the compromise worth understanding, because “too big for the furnace” is common on machine bases over about 6 m. Rough machining first removes most of the material and therefore most of the future movement; vibratory treatment then improves stability through the finishing cut. It is weaker than thermal relief and it is far better than proceeding straight to finish.
Writing it into the drawing
Vague notes produce vague compliance. “Stress relieve before machining” is an invitation to do the cheapest thing that can be called stress relief.
A specification that holds:
Process note SR-1. After welding and before rough machining, thermally stress relieve per the following: heating rate ≤ 100 °C/h above 300 °C; soak 590 ± 20 °C; soak time 1 h per 25 mm of maximum section, 2 h minimum; furnace cool at ≤ 100 °C/h to 300 °C, then still air. Supply the time–temperature chart with the part. Machining allowance to remain ≥ 3 mm on all surfaces at time of treatment.
Four things make this work. It names the temperature band, so it cannot be done at 400 °C and called compliant. It ties soak time to section thickness. It requires the chart, which converts the promise into evidence. And it fixes the sequence relative to machining, which is the part buyers most often leave open — relieving after finish machining destroys the geometry you just paid for.
For welded structures generally, EN 1011 and the post-weld heat treatment provisions of the pressure-equipment codes are the usual sources to cite if your customer needs a standard reference rather than a written-out cycle.
The questions to ask a shop
- What is the internal envelope of your furnace, and is it yours or subcontracted?
- Do you chart every cycle, and can I see a chart from a recent similar part?
- Where in your routing does stress relief sit relative to rough and finish machining?
- What machining allowance do you leave on before treatment?
- If you are proposing vibratory treatment, why — and what is the fallback if the part still moves?
A shop that answers these fluently has done it many times. A shop that answers the first one with a number it cannot immediately confirm is quoting your job and planning to subcontract the furnace, which is fine if it is disclosed and a problem if it is not.
Every PO placed through this network carries the treatment route, the chart requirement and the machining sequence as line items rather than assumptions — see how the sourcing process runs. If you want a frame drawing reviewed before it goes out to quote, send it through. The companion piece on what drives fabrication cost covers where stress relief sits in the total price, and the drawing checklist puts the process note in sequence with the rest of the callouts. When a frame still moves after installation, see alignment drift after installation.
Have a frame drawing to price?
Send it through for a manufacturability review and a matched quote from the heavy-fabrication network. Reply within 24 business hours.