The weld shrinks as it cools — and the part goes with it. Distortion is not avoidable; it is controllable. The control starts with the numbers: heat input, weld size, and sequence.
Distortion is the permanent change of shape a fabrication takes because of welding. It comes in five flavors — longitudinal shrinkage along the weld, transverse shrinkage across it, angular distortion where the joint rotates, bowing, and buckling — and they usually arrive together.
Distortion is not a metallurgical defect and not a weld-metal defect — it is a dimensional one. It matters because out-of-tolerance geometry means fit-up problems downstream, residual stress locked into the structure, and expensive straightening. It is the defect you can't grind out, only work around.
Weld distortion, shrinkage, warpage, buckling, angular distortion, longitudinal shrinkage, transverse shrinkage, bowing.
Distortion is the one defect you can't photograph close — you have to stand back and measure.
Distortion is shrinkage redistribution. Heat input, weld size, restraint, and sequence decide how much the part moves — and where.
The weld metal melts, expands, and then contracts as it solidifies and cools — and it contracts more than it expanded. That gives you shrinkage in three directions at once: transverse across the weld, longitudinal along it, and angular, where the top of the joint gets more heat than the bottom, shrinks more, and rotates the joint like a hinge.
How much of that shrinkage shows up as movement depends on restraint. A part clamped tight doesn't move — it stores the load as residual stress instead. A part free to move bends or buckles. The goal is never to fully prevent shrinkage: it is to make it show up where you can afford it. Buckling is the dramatic case — thin plate under compressive residual stress near the weld goes wavy when the stress exceeds the plate's buckling strength.
Prevention is a sequence decision made before the first arc — plus a heat input you can defend with a number.
Plan the sequence — backstep or block welding, alternate sides, weld toward the neutral axis. Set weld size to the design minimum, not the maximum. Pre-bend (preset) where the direction of movement is predictable.
Keep heat input at the minimum consistent with fusion — the heat input calculator puts a number on it. Weld both sides of a joint in alternation; keep travel speed steady.
Control interpass temperature and let the part cool. Measure your progress with a straightedge — you can correct a trend before it becomes a twist.
"Weld it all and straighten it after" is the most expensive habit in fabrication. The correction cost is usually five times the prevention cost — plan the sequence first.
Once distortion is in the part, you have three tools: mechanical straightening, thermal straightening, and stress relief. Order matters.
| Step | Action | Why |
|---|---|---|
| 1 | Measure the distortion and decide what's actually out of tolerance. | You fix what the drawing says, not what you feel. |
| 2 | Mechanical straightening — press, jacks, or straightening clamps, working the part past flat and letting it spring back. | Controlled cold forming; no metallurgical risk. |
| 3 | Thermal straightening — controlled torch heating of specific zones. | Heat expands the zone; on cooling it contracts and pulls the part straight — a skill, not a shortcut. |
| 4 | Stress relief where the code requires it. | Sometimes the residual stress, not the shape, is the real problem. |
| 5 | Verify against the drawing — flatness, squareness, dimensions. | Acceptance is dimensional; prove it. |
Thermal straightening is the dangerous tool — too much heat, or heat on the wrong material (aluminum, hardened steel), ruins the plate or starts cracks. Practice on scrap first, and check the code before stress-relieving: it changes mechanical properties.
See the Flame Straightening Technical Guide for spot, line, and wedge heating patterns, sequencing, and temperature control.
The process decides how much heat lands per inch — and therefore how the shrinkage shows up.
A good distortion-control process — low heat input per pass versus stick. Use small passes and let the part cool; the temptation is long, hot continuous runs.
Precisely controllable heat, but slow — many small passes. Fine for thin stainless and aluminum where you can manage the sequence; the risk is heat buildup on intricate parts.
High heat input per pass. Watch weld size — an oversized 7018 fillet shrinks like it's trying to. Keep the weld to the design minimum and control interpass.
High deposition means fewer passes and often less total heat — but each pass is hot. Sequence and interpass control matter more than the raw heat input number.
All distortion control is parameter control. Here's what to check — the heat input calculator ties them together.
| Parameter | Check | Typical Fix |
|---|---|---|
| Heat input (kJ/in) | Amps × volts ÷ travel speed | Keep it at the minimum that gives sound fusion — calculate, don't guess |
| Weld size vs. design minimum | Compare to the drawing | Size to the minimum; every extra 1/16" of leg adds shrinkage |
| Pass sequence | Backstep, block, alternating sides | Balance shrinkage on both sides and along the joint |
| Interpass temperature | Pyrometer or temperature stick | Let the part cool between passes; a hot part moves more |
| Clamping / restraint | Fixtures, tacks, clamps | Restraint trades movement for residual stress — balance the two |
| Travel speed | Part of the heat input equation | Faster travel = less heat per inch = less shrinkage |
The Heat Input Calculator is the core tool for distortion control. Compute kJ/inch before you weld, and keep it at the minimum that gives fusion.
Distortion is controlled with hardware as much as with technique — positioners and automation are the two big levers.
Positioners are the distortion-control workhorse: they put the joint in a position where you can weld symmetrically and downhand, and let you alternate sides without fighting gravity. Automation is sequence control — the same bead, same direction, same interpass, weld after weld — which is exactly what shrinkage balance needs.
Distortion is a dimensional inspection: measure before, measure during, measure after.
Acceptance depends on the code and project: AWS D1.1, ISO 5817, API 1104 and project specifications each set their own limits for this discontinuity. Check the governing document before judging a weld acceptable — there is no universal pass/fail rule.
The questions welders actually ask about distortion — answered straight.
Both. Some distortion is inevitable in every fusion weld — that is the fact. It becomes a defect when the part exceeds the drawing or code tolerance. The skill is making the inevitable shrinkage show up where you can afford it.
Sequence and size: weld both sides in alternation or use backstep, keep the weld at the design minimum, keep heat input low, and consider preset — pre-bend the corner slightly past square so shrinkage lands it square. Clamping helps, but it trades movement for residual stress.
Usually the opposite. Preheating slows cooling and gives shrinkage more time to accumulate, so distortion can increase. Preheat is for crack and hydrogen control — if the code requires it, use it, and offset the extra shrinkage with sequence and restraint.
Yes, with care. Controlled torch heating of limited zones — staying well below the transformation temperature on carbon steel — expands the zone and pulls the part straight as it cools. It is a skill: practice on scrap, watch the temperature, and never do it on aluminum or hardened steel without experience.
Sequence. Clamping is restraint: it reduces movement but stores the load as residual stress, which can resurface as cracks or later movement. Sequence spreads the shrinkage so less of it lands in any one place. Clamp to stabilize, sequence to control.