Weld metal shrinks as it solidifies, and it shrinks against metal that is already cold and already holding its shape. The joint cannot contract freely, so it contracts against resistance, and the result is a locked-in elastic strain. That strain stays in the part indefinitely, sharing the load with whatever the part is asked to carry in service.
This is residual stress, and it is not a defect in itself. Almost every welded assembly has some. The engineering question is not how to eliminate it but whether the amount present matters for the way the part will be used, and whether relieving it would cost more than it is worth.
The two terms get used as if they were interchangeable and they are not. Stress relief is a thermal treatment intended to let the material relax internally, lowering the peak residual stress without changing the microstructure much. Annealing goes further and changes grain structure and properties, which is a different intention with different consequences.
For most fabricated exhaust and intake work, neither treatment is applied, and that is usually correct. The parts are thin, the restraint is modest and the material retains useful ductility. Where the distinction starts to matter is on parts that combine heavy sections with thin ones, or that are machined after welding.
A thin-wall tube welded to a thin-wall tube has very little restraint and very little stored energy. It will distort as it cools, which is visible and can be corrected, and the residual stress that remains is low relative to the strength of the section. Thermal treatment would add handling, cost and the risk of new distortion for a marginal gain.
There is also a metallurgical reason to be cautious. Heating some austenitic grades into the wrong range for the wrong time creates exactly the chromium depletion that ruins corrosion resistance. A stress relief treatment that is not properly specified can trade a stress problem for a sensitisation problem, which is a poor bargain on an exhaust part.
Look for geometry where a stiff region is welded to a flexible one. A thick machined flange welded to thin tube is the classic case: the flange cannot move, the tube wants to, and the weld toe carries the difference. A ring or a collar welded into a tube creates the same condition in a smaller space.
The other category is anything machined after welding. Cutting into a part that holds residual stress allows the material to redistribute, and the machined feature moves after the cut. A gasket face that is flat when it leaves the machine can be out of flat by the time it is bolted up, and the error is not the machinist's.
Titanium reacts with oxygen, nitrogen and hydrogen at elevated temperature, and a treatment that is meant to relax stress can instead contaminate the surface or the section if the atmosphere is wrong. That is why titanium thermal treatment is a controlled-atmosphere conversation rather than a simple oven schedule.
The practical consequence for a buyer is that any titanium part requiring thermal treatment should be described with the atmosphere as part of the requirement, not just the temperature. A shop that treats titanium like stainless in this respect will produce parts that look correct and behave unpredictably. Our own titanium work is planned around that constraint, which is why the treatment question gets settled at the quotation stage rather than at the oven.
Distortion is what happens when the shrinking weld pulls the part out of shape while it is still hot and soft. It is a geometry problem and it is addressed by sequencing, fixturing, tacking and heat balance. Residual stress is what is left afterwards, and it is a stored-energy problem.
Confusing the two leads to the wrong remedy. More fixturing controls distortion and can increase residual stress, because the part is held more rigidly while it tries to contract. Releasing the fixture then lets the part spring, which looks like a distortion failure and is really a stored stress being paid out.
For the great majority of fabricated parts, the useful work happens before any heat treatment is considered. Balanced welding sequences, backstep technique, symmetric passes, allowing a part to cool in its fixture rather than unclamping it hot, and avoiding excessive local heat input all reduce the stress that has to be dealt with later.
Fit-up does a lot of quiet work here as well. A joint with an even gap and consistent penetration needs less heat and produces a more uniform contraction. A joint that is forced closed with a clamp stores stress before the arc even starts, and that stress is still there when the clamp comes off. You can see how these habits show up in the finished work on our TIG stainless steel welding pages, and the titanium equivalent follows exactly the same logic.
A drawing that says nothing about stress relief is not wrong. It is simply leaving the decision to the shop, and the shop will default to whatever it usually does. That is acceptable when the part is ordinary and becomes a problem when the part is not.
The trigger for specifying something is usually one of three things: a machined feature that has to stay true after welding, a joint between very different section thicknesses, or a service duty that involves fatigue in a restrained location. If none of those apply, the honest answer is that the part does not need the treatment and adding it is a way of feeling careful rather than being careful.
The questions below separate parts that need a thermal step from parts that only look as though they might.
The reasonable position for most fabricated metalwork is that stress relief is a targeted tool rather than a default step. Used where the geometry demands it, it prevents movement and cracking that no amount of welding skill would have avoided. Applied everywhere, it adds cost and can introduce the very corrosion problems the material was chosen to avoid. Our titanium exhaust system and stainless steel exhaust system ranges are assembled with that judgement, the tube stock behind them sits under exhaust pipe, and the same logic runs through our titanium fabrication components and TIG titanium welding parts.
