Every crack has a reason. It may be a stress concentration at a bracket corner, a weld that lacked fusion, a part that was forced into place, or vibration over a long period. Welding the crack closes the symptom and leaves the reason in place, which is why a repaired part often cracks again a short distance away.
So the first job on a cracked component is not welding. It is working out where the load comes from and whether the repair should add material, change the shape, or simply restore the original section. A repair that ignores the cause is a scheduled repeat visit.
The visible crack ends somewhere the eye cannot follow. In most metals the crack tip runs ahead of the visible line, sometimes by a considerable distance, and it terminates in a small zone of deformed metal. Weld over the visible part and the tip stays there, ready to run again as soon as the part is loaded.
Finding the end means cleaning the surface and looking closely, sometimes with a light and a magnifier, sometimes with a penetrant check on the far side. On a painted or coated part, stripping the coating back well past the crack is part of the job, because the coating hides exactly the evidence you need.
A stop hole is drilled at the crack tip to remove the sharp termination and replace it with a round one. A round hole distributes stress around its circumference instead of concentrating it at a point, which slows or halts further growth. It is a stabilisation measure, not a repair.
Two things go wrong with stop-drilling. The hole is drilled short of the tip, which leaves the crack intact beyond it and does nothing. Or the hole is drilled and the part is returned to service with no further work, which stabilises the crack and leaves a component with a hole where sound metal should be. On anything structural, the hole is the first step of the repair, not the whole of it.
Once the crack is stopped, the crack path has to be removed. That means cutting a groove along the crack, through the full thickness, until the bottom and both walls are clean metal. Anything less leaves a remnant of the crack under the new weld, and the heat of welding will usually open it.
The groove profile should be wide enough for the torch to reach the root and for each pass to be cleaned. On thin material the groove may consume the whole section, which changes the repair from a groove-and-fill into a butt-weld with a backing arrangement. That is a different procedure and should be recognised as such before cutting starts.
A crack in a part that has carried exhaust gas, oil or coolant is full of contamination. The crack faces themselves hold deposits that no surface cleaning will reach, which is another argument for grooving rather than welding over. Once grooved, the faces can be cleaned, degreased and, where the material allows, ground bright.
Stainless and aluminium both need their own tooling for this. A brush or a grinding disc that has touched carbon steel will leave contamination that shows up as porosity or as a rust bloom on a stainless part within days. Keeping dedicated tools is a shop discipline, not a suggestion.
The fill should be built from the root outwards in small passes, each one allowed to cool before the next. A single heavy pass over a groove looks efficient and produces exactly the conditions that caused the original crack: a wide, weak, fast-cooling weld in a stiff section.
Peening between passes is sometimes used to relieve stress, and it has to be done properly or not at all. A light, uniform peen with a rounded tool redistributes residual stress. A heavy, localised hammering work-hardens the surface and can start a new crack in the weld itself.
A crack often sits in a region that has also lost thickness to corrosion or wear. Closing the crack restores continuity but not section, and a part that is continuous but thin will fail in the same place for a different reason. The repair should restore the original profile, including any radius at a corner that was there to spread load.
Rebuilding a radius matters more than it looks. A sharp internal corner concentrates stress, and a repaired corner that is left sharp is a crack waiting for its next opportunity. Filing or grinding a generous radius into the finished weld is a few minutes of work that changes the life of the part.
A repaired crack cannot be inspected the way a new weld can, because the geometry is irregular and the surface has been worked. What can be checked is continuity, profile, and the absence of undercut at the toes, plus a penetrant check on the finished surface where the material allows it.
Pressure-tight parts need a leak test, and the test should be done after the part has cooled and, ideally, after any machining. A repair that passes a visual and fails a leak test has told you something useful; a repair that is never tested has told you nothing.
Repair work is priced on the assumption that the damage is where it appears to be. The questions below are the ones that keep a repair from becoming a replacement.
Crack repair is routine work on exhaust sections, and the surrounding pipe fabrication is described under exhaust pipe work and exhaust pipe sections. The welding procedures behind it sit on our TIG stainless steel welding pages, finished stainless systems under stainless steel exhaust systems, titanium equivalents under titanium TIG welding parts, and the fabricated parts we build new rather than repair under fabricated components.
