The first repair held. The part ran, and then a crack appeared not through the new weld but immediately beside it, along the edge of the heat-affected zone, often as a fine line growing from the weld toe outward. That location is the diagnosis. It says the weld itself is sound and the material around it has been changed.
A crack through the weld metal means something different: bad fusion, contamination, insufficient penetration or a filler mismatch. A crack running parallel to the weld, just outside the fusion line, is about restraint and about what the heat did to the parent metal beside the pool.
Welding heats a band of parent metal beside the pool and cools it quickly. That band ends up harder and less ductile than the metal either side, and it carries residual stress from contracting against colder material. The result is a stiff strip with a sharp transition, and strain in the part concentrates at that boundary instead of spreading.
If the repaired area is also restrained, because it sits between a flange and a hanger or because it was clamped hard while it cooled, the strain has nowhere to go but into that boundary. The crack re-initiates there for the same reason the first one did, except the material beside the weld now absorbs less of it.
Fatigue from vibration produces cracks at a different location: at the toe of a bracket, at a hanger root, or at the end of a straight run where the section changes. Corrosion produces cracking from a pit, and the pit is visible. A crack that was never fully removed simply carries on from its original tip.
The distinguishing check is where the tip sits. Grind or blend the area and inspect with dye penetrant. If the indication continues from the old crack line under the new weld, the repair never reached sound metal. If it starts fresh at the heat-affected zone boundary and runs away from the weld, the mechanism is restraint rather than an incomplete repair.
Ask how the crack was prepared. A crack welded over without being ground out to sound metal will always return, because the tip is still there and it is the sharpest feature in the part. A repair that was ground back, dressed and then welded has a chance of holding.
Look at the profile. A convex bead with a sharp toe creates a stress concentration at exactly the point where the new crack appears. A flatter, blended finish spreads that transition. Dressing the toe is not cosmetic on a part that flexes; it changes where the strain lands in service.
On titanium the question is sharper. The metal absorbs gases when it is hot and unprotected, and the result is a brittle layer that cracks rather than yields. A repair welded without adequate shielding behind and around the joint can look clean on the outside and be contaminated beneath, and the first crack appears at the edge of the affected zone.
Colour is the accessible clue. A titanium weld and its surroundings should show a limited, controlled tint. Chalky white, dull grey or a powdery flaking surface beside the bead points at contamination. The TIG-welded titanium parts range covers the joints where this matters most in service.
The second repair should start further back than the first. Cut out the whole affected band rather than welding beside it, so the new joint sits in material that has not already been hardened once. Support the section so it does not move while it cools, and let it cool without quenching.
Where the same area keeps cracking, the geometry is usually the real cause and the weld is only where it shows. A gusset, a change of section, a hanger moved along the run, or a flex section that lets the pipe move: those change the strain. The fabricated titanium components range covers the pieces used to make that kind of change.
