Designing Welded Fabrications to Minimise Heat Distortion
Reading Time: 5 minutes
A welded frame can leave the fabrication table looking very different from the CAD model, even when its components were cut correctly and the welds meet the drawing. The usual response is to change the welding sequence or restrain the assembly more heavily. Those measures can help, but the workshop may already be dealing with distortion created by the design.
During welding, a small area is heated while the surrounding material remains relatively cool and restrains it. The joint then contracts as it cools, pulling against an assembly that is no longer at a uniform temperature. Because the heat and stiffness are rarely distributed evenly, the resulting movement is rarely even either.
Some movement is unavoidable, particularly in large bespoke fabrications. A sound design keeps it within useful limits and provides a realistic route to the final dimensions. The design review is the right place to address this, while the joint layout and manufacturing route are still open to change.
Give the heat somewhere to go
Distortion tends to become visible in the least stiff part of an assembly. When a thin plate is welded to a much heavier member, the lighter component heats quickly and offers less resistance as the joint contracts. The heavy section may remain reasonably stable while the plate beside it bows, which can make a structurally acceptable connection difficult to assemble or finish.
The position of the weld matters as much as the difference in thickness. Shrinkage acting well away from the neutral axis has more leverage over the fabrication, so a long weld placed on one side of a flexible section can produce a pronounced pull. Bringing the joint closer to the neutral axis may reduce that effect, while opposing welds can sometimes balance one another when the geometry allows.
This does not make symmetry a design rule. Bespoke machinery is rarely symmetrical, and load paths should not be compromised merely to make a neater weld layout. The useful habit is to look at the proposed joints and ask which part of the assembly will be free to move as each weld cools. If the answer is always the same thin panel or unsupported flange, the layout deserves another look.
Stiffeners can help, although their effect is easy to overestimate. A rib adds local rigidity, but the welds needed to attach it also introduce heat and shrinkage. Before adding more welded parts, consider whether a folded return or a standard section could provide the required stiffness with fewer joints. The simplest way to reduce distortion is sometimes to remove a weld from the design altogether.
Material behaviour also changes the scale of the problem. Austenitic stainless steel generally expands more than carbon steel when heated, which makes thin stainless fabrications particularly sensitive to an arrangement that concentrates heat on one side. A joint layout that has worked well in mild steel should therefore be reviewed before it is repeated in a different material.
TWI's guidance on distortion prevention by design reaches the same practical conclusion: shrinkage is inevitable, so the design should keep weld metal to what the joint requires and position it where its effect can be controlled. Correction remains possible, but it is a poor substitute for a joint that was easier to fabricate in the first place.
Bigger isn’t always better
Oversized welds are common on fabrication drawings because a larger fillet can feel like a safe choice. In reality, the excess metal increases deposition time and introduces more shrinkage without improving the connection in proportion. Once the required throat has been established from the load case and the relevant design requirements, adding weld metal for reassurance is poor value.
This is especially important when the design uses a continuous weld by default. Some joints need a full-length connection for strength or sealing, while others can perform properly with intermittent welds. The decision must account for the service conditions, including any corrosion or fatigue concerns, but it should still be a decision rather than a habit carried from one drawing to the next.
Joint preparation deserves the same attention. A generous root gap or wide bevel increases the volume that must be filled, which adds heat before any distortion-control technique has been considered. On thicker material, a double-sided preparation can reduce weld volume and help balance shrinkage through the plate, provided the reverse side is accessible and the workpiece can be handled safely.
The handling plan can overturn what appears efficient in a joint section. A double-sided preparation may become expensive when a large frame has to be turned repeatedly or when the welder cannot reach the reverse side in a sensible position. Weld design belongs to the complete fabrication method, so the theoretically neatest joint is not always the best one to manufacture.
Over-welding often begins with a weld symbol that was added too casually. Once an unnecessarily large weld appears on the drawing, the fabricator is obliged to price and produce it unless the design is formally changed. The symbol deserves the same care as the plate thickness or bolt specification elsewhere on the assembly.
Leave room for a sensible welding sequence
A drawing can define every joint correctly while leaving no workable order in which to make them. Parts may close off access as the assembly develops, or an early weld may lock the frame before the remaining joints can be aligned. By the time the conflict reaches the workshop, solving it can require an improvised fixture or a departure from the intended build method.
The broad sequence should be considered while the model is still flexible. Breaking a large fabrication into sensible subassemblies may allow critical joints to be made in a better position, after which the completed sections can be brought together with fewer difficult welds. The designer should also check that the assembly can be turned and supported without placing a partially welded structure in an unstable condition.
Alternating between opposite sides or using a backstep sequence can help control how shrinkage accumulates. Neither is a universal instruction, because the suitable approach depends on the specific joint and the qualified procedure available in that workshop. A designer who writes a detailed sequence without speaking to the fabricator can create a different problem while trying to solve the first.
This is where early fabrication input earns its place in design. An experienced welder or production engineer can often identify where a joint will pull and whether the proposed access is realistic. That conversation is most useful before release, when the model can still respond to what the workshop knows.
The drawing should make the functional requirements clear without prescribing workshop technique unnecessarily. Where a sequence is essential to structural performance or dimensional control, it needs to be agreed and documented. Elsewhere, the fabricator should have room to use a qualified procedure and a method suited to the equipment available.
Welding is not a precision process
Large weldments should not be treated as precision-machined components. Even with a thoughtful joint layout and a controlled sequence, critical faces may move enough to affect alignment. Expecting every mounting surface to remain flat after welding can turn an unrealistic drawing tolerance into hours of straightening and inspection.
Where accuracy matters, the design needs a deliberate finishing route. Bearing locations or equipment mounting pads may require machining after welding, with enough allowance left to clean up the surface once the fabrication has settled. The datum arrangement should also let the machinist set up the assembly and verify the finished relationship without relying on surfaces that may have moved.
The same judgement needs to carry through to the tolerance scheme. A blanket dimensional tolerance across a large frame may be easy to place on the drawing, but it says little about which features govern the function of the machine. Identifying the dimensions that must be checked after welding gives the fabricator a clearer target and avoids spending time correcting movement that has no practical consequence.
Pre-setting and restraint can support that dimensional plan. On a repeat fabrication, measured experience may show how far a part needs to be set in the opposite direction so that shrinkage brings it towards the required position. For one-off work, a trial joint or an agreed fabrication allowance is usually more credible than a confident estimate made from the CAD model alone.
Heavy restraint needs care because prevented movement does not disappear; it remains in the assembly as residual stress. In crack-sensitive materials, that can create a more serious problem than the visible distortion the fixture was intended to stop. Clamping arrangements should therefore support the qualified welding procedure and material behaviour rather than simply hold the work as rigidly as possible.
Distortion becomes much easier to control before the plates are cut. Designers rarely need to predict every millimetre of movement from first principles, but they do need to give the fabricator a workable joint layout and a clear definition of the dimensions that matter. When the fabrication and finishing route are considered together, the finished assembly has a far better chance of matching its purpose without an expensive correction stage.
ATLUS considers welded fabrication as part of the wider design process, with joint arrangement and workshop reality influencing the model before drawings are released. Our design engineering and design for manufacture services help clients develop industrial equipment that can be built as intended. Talk to ATLUS if an existing design needs review or a new fabrication needs developing from first principles.