Estimating the distortion before hand and applying it before welding in the opposite direction, so that it counter acts the distortion post welding finally nullifying it.
We had a supplier come back on a re-procurement contract and "explain" how the part could not be made as they had repeatedly failed to make a suitable weldment.
Their welder was trained in the Air Force for structural repairs; all his welds were defect free. They also were huge. Ginormous. Fat. Since nearly 100 had previously been made by a different supplier, we know it was possible - eventually the original welder was willing to go and show their guy how to put down a suitable bead, fast, defect free. The fat welds? 2-3 days per unit. The desired welds - 2-3 units get welded in a day.
Tackling deformation due to welding is a tough thing - it depends on how the part is restrained, how much the surrounding metal is heated, how much the melt zone shrinks with the dropping temperature which depends on how big the melt zone is. It very much depends on if the welding is symmetric or not, it depends on how fast the welder goes (affecting the size of the melt zone and the surround material expansion). Sometimes it depends on the crystalline state of the various components, going from face-centered cubic to body-centered cubic in the case of carbon steel.
Set up trial pieces and see how much deformation occurs with different approaches and how uniform the results are from some single approach.
My old roommate is a metal fabricator. He would routinely estimate the deflection of steel and stainless steel parts when he was going to weld them. Like, to better than 20 thou when dealing with 6ft lengths. Usually, the fab process would focus on tack weld location and layup configuration to minimize distortion. IE: If you have a 2 sided fillet weld holding a wall that needs to be straight... you'd tack both sides in similar spots, and you don't weld 4 passes on one side before hitting the other side. A lot more goes into the fabrication than just "make the weld called out on the drawing".
In theory, you could take the material's initial condition (assuming it's heat treatable) and the distortion coefficient for different phases, and use that with the estimated size of the HAZ to guess how much material is going to phase change and pull things. In reality, you also have the thermal expansion to deal with and the welder skill/weld quality. If a fat fillet weld gets TIGed in slowly, the whole joint gets super hot, fuses, and will pull like crazy as it cools. A faster welder, smaller bead, or colder process can mitigate all this.
Long story short: trial and error. This is prime example of the benefits of skill and experience. A skilled welder with their head screwed on straight, and varied experience? Will be able to figure out a great process in short order that any monkey could then follow. But if you ask the monkey to weld things straight the 1st time, you might be disappointed. Expect a slower production rate for the first few pieces, with marginal results on the accuracy... but then things should straighten (heh) themselves out. Most of the welders I have worked with like doing good work and absolutely nailing the dimensions. It's a source of pride. But every joint needs some practice if you want perfection.
My old roommate is a metal fabricator. He would routinely estimate the deflection of steel and stainless steel parts when he was going to weld them. Like, to better than 20 thou when dealing with 6ft lengths. Usually, the fab process would focus on tack weld location and layup configuration to minimize distortion. IE: If you have a 2 sided fillet weld holding a wall that needs to be straight... you'd tack both sides in similar spots, and you don't weld 4 passes on one side before hitting the other side. A lot more goes into the fabrication than just "make the weld called out on the drawing".
In theory, you could take the material's initial condition (assuming it's heat treatable) and the distortion coefficient for different phases, and use that with the estimated size of the HAZ to guess how much material is going to phase change and pull things. In reality, you also have the thermal expansion to deal with and the welder skill/weld quality. If a fat fillet weld gets TIGed in slowly, the whole joint gets super hot, fuses, and will pull like crazy as it cools. A faster welder, smaller bead, or colder process can mitigate all this.
Long story short: trial and error. This is prime example of the benefits of skill and experience. A skilled welder with their head screwed on straight, and varied experience? Will be able to figure out a great process in short order that any monkey could then follow. But if you ask the monkey to weld things straight the 1st time, you might be disappointed. Expect a slower production rate for the first few pieces, with marginal results on the accuracy... but then things should straighten (heh) themselves out. Most of the welders I have worked with like doing good work and absolutely nailing the dimensions. It's a source of pride. But every joint needs some practice if you want perfection.
OP
All good post as well.
Welding procedure depends on the configuration of the welded assembly. Heat treat condition.
Higher hardness produces more stress related issues. Some times welding fixtures are or maybe required. To restrained condition.
Preheat in an oven or by torch. The conditions of thick or thin sections. Thin section will heat and cool faster. Preheat mitigates stress, thus distortion., as said before actual welding procedure and the skill , and type of welding.
On complex weld assemblies tack welds are required to hold in place. Welds are completed.
