A welding procedure specification is not a description of how a skilled welder works. It is the written record of the variables that were frozen at the moment a joint was qualified, so the same joint can be produced again by somebody else, on a different day, without repeating the experiment. Read it as a set of constraints rather than as an instruction sheet and it becomes useful.
That distinction explains most of the friction on a shop floor. A welder who has run thin-wall stainless for years will see values tighter than his own habit. It describes a narrower corridor than his experience covers, and the corridor exists because a coupon was bent and cut inside it.
So the useful question is never whether the procedure matches your instinct. It is which variables the document has frozen, and which ones it left open because they made no difference to the qualification result.
Every procedure starts with a joint description, and that description carries an assumption about preparation that is easy to lose. The groove form, root gap and land together decide how much metal the welder places and how much base metal has to melt to make the root tie in.
If the preparation in the shop is not the preparation in the document, the current range no longer means what it meant on the coupon. A gap that crept wider because the saw blade wore will swallow a root pass qualified on a tight fit, and the welder will compensate by slowing down, which raises heat input, which changes the backside profile. The procedure did not fail. The preparation drifted. That is why we treat cut quality and deburr as qualified variables on our TIG stainless steel welding work.
The most misread line in any procedure is the current line, and it is written as a range on purpose. Whoever qualified it knows the correct value depends on how fast the welder can travel, how thick the wall actually came out, and how much heat a fixture or chill bar is pulling away. A range is honest. A single figure would be fiction.
Inside that range the welder manages a balance. More current with faster travel gives a narrower bead and a tighter heat-affected zone. Less current with slower travel gives a wider, flatter bead and more total heat in the part. Both can pass a bend. They do not produce the same distortion, and they do not produce the same colour on stainless.
What the range forbids is drifting outside it to solve a fit-up problem. If a joint needs more current than the document allows in order to fuse, the current is not the problem. The fit is wrong, or the tungsten is dressed wrong, or the gas coverage is failing, and each of those has its own fix.
The filler callout is usually short and almost always under-read. It names a classification and a diameter, but what actually decides success is whether the rod was dry, clean and stored away from grinding dust. A rod left open on a bench picks up moisture and shop air, and moisture in the arc is one of the quiet routes to porosity.
Shielding is written as a gas and a flow range, and flow is where most floor arguments happen. Too little and the pool fogs, too much and the gas turns turbulent and drags shop air into the envelope. Neither failure announces itself. Both show up later as discolouration or as pinholes that only appear after polishing. The procedure should also state whether a purge or trailing shield is required, and on titanium that is not optional, since contamination can sit below a weld that looks acceptable. Our notes on titanium fab components go further into how that shows up on real parts.
Position looks administrative and is not. A procedure qualified flat does not automatically transfer to a vertical or overhead joint on the same thickness, because gravity changes how the puddle sits and how much current the welder can carry before the pool sags. Direction of travel matters for the same reason: forehand and backhand put different heat into the joint ahead of the pool.
Multi-pass work adds a variable single-pass work does not have, which is what the surface of the previous pass looks like before the next goes on. Oxide, silicon islands, spatter and grinding residue all sit between passes and all can be trapped. A procedure that requires brushing is not being fussy, it is closing a route to lack of fusion.
Here is the part welders rarely hear stated plainly. A procedure fixes the variables a test proved to matter. It does not fix torch angle, arc length within reason, the rhythm of filler addition, or the exact instant the welder pauses to let the pool catch up. Those belong to technique, and technique is what the qualification was really testing when it passed.
That is also why a procedure cannot rescue someone who has never run the material. The document narrows the corridor, and the welder still has to walk it. On thin-wall tube the difference between a good root and a burned-through one sits in a fraction of a second of dwell that no document can specify.
Training and documentation are not substitutes. A shop with a rigorous procedure and no practice produces paperwork. A shop with practice and no procedure produces parts that cannot be repeated. Both halves show up in what you can browse under stainless steel exhaust systems.
Most bad procedures in circulation were not written for the joint they are applied to. They were copied from a similar job, adjusted from memory and never re-qualified. The tell is usually a mismatch between the joint description and the drawing, or a current range that makes no sense for the wall in front of you.
Copying is not automatically wrong. Variables travel well when the material family, thickness band and joint form are the same. They stop travelling the moment the section gets thinner, the alloy changes, or the part gains a closed volume that has to be purged.
A procedure also only functions if the welder can reach it. The version that matters is the one at the bench, with the joint number on it, matching the drawing revision in the welder's hand. When a joint fails a bend or a penetrant check, the first thing to establish is which variables were used, not which were specified, and that requires recording what actually happened.
If you buy welded parts rather than make them, the procedure is part of what you are paying for whether or not it appears on the quotation. Ask for it and read the joint description against your drawing.
None of that needs a metallurgist. It needs the first page of a document read against the drawing on your desk, and a question about what happens when the two disagree. Tigindustry has been welding in Shanghai since 2009, and you can see the range on our titanium exhaust system and exhaust pipe pages, or browse by alloy under shop by material.
