When a fabricated part breaks, it leaves a surface behind that describes how it broke. The information is not subtle or mysterious; it is geometric, and it can usually be read with good light and a low-power magnifier. A fabricator who looks at a break properly can normally separate a part that was overloaded from a part that was worked to death, and can often point at the exact spot where the trouble started.
The reason it matters is that the two diagnoses lead to completely different actions. If a part was overloaded, the question is what put so much load through it. If a part fatigued, the question is what allowed repeated movement to concentrate in one place. Getting the diagnosis wrong means making a change that does not address the cause, and watching the next version fail the same way.
The origin is where the fracture began, and it is almost always identifiable. On a fatigue break, the origin end of the surface is smoother and often darker or discoloured, and fine lines spread away from it in the direction of travel. The last part of the section to separate is rougher, sometimes with a ragged lip, because it failed suddenly once the remaining material could no longer carry the load.
Origins cluster in predictable places. The toe of a weld, where the bead meets the parent metal. The edge of a heat-affected zone just outside the weld. A sharp internal corner at a bracket, a cut edge left with a burr, or a scratch or tool mark on the surface. Machined edges, drilled holes and sheared plate all concentrate stress, and any of them can become the starting point if the part is cycled often enough.
An overload break looks entirely different. There is no slow-growth zone, no progression of lines, no obvious dark area. The whole surface is fresh, rough, and often shows signs of deformation: a part that was stretched before it tore, or a section that necked down before separating. Bending overload leaves one region compressed and the other stretched, and the two look different from each other.
Mixed cases exist and are common in exhaust work. A part may crack by fatigue over a period, then tear the remaining ligament in a single overload event when the crack reaches a critical length. The reading is still possible: the fatigue zone is there, the overload zone is there, and the boundary between them shows how much section was left when the final failure happened. What that boundary tells you is how much warning the part gave before it let go.
Material behaviour shapes the appearance of a break. A ductile material deforms visibly before separating and shows a dull, fibrous surface with tearing. A material that fails in a more brittle way separates with less warning, and the surface may show flat facets and a brighter, more crystalline look.
Titanium adds a complication, and a useful one. The colour of the surface records how much oxygen the metal saw at temperature. Heat tint that runs from gold through to blue and beyond tells a fabricator how hot that region got and how well the shielding covered it when the part was welded. On a broken titanium part, discolouration near the origin often points at a weld that was already compromised before the part ever entered service. That is a piece of information only the person who built it can fully use, and work of this kind is grouped under our titanium welding parts pages.
Some defects announce themselves when a joint is broken apart. Lack of fusion, where the filler never merged with the parent metal on one side, leaves a smooth boundary with none of the tearing that surrounds it. Porosity shows as rounded voids, singly or in clusters. Incomplete penetration leaves a notch at the root of the joint, which acts as a built-in stress riser from the day the part was made.
None of these are normally visible from outside, which is why a broken part is such useful evidence. A weld that looked perfectly acceptable and passed a visual inspection can reveal a root that never fused, and the location of that defect usually corresponds exactly with where the crack started. That correspondence is the most actionable thing a fracture reading produces.
A photograph of a broken part gives useful information and should not be treated as a complete diagnosis. Lighting, angle and image quality all change how a surface reads, and features that are clear in the hand disappear in a picture. Colour banding on titanium, in particular, is easy to misjudge on screen.
What can be said from a photo is usually at the level of direction: this looks like fatigue rather than overload, the origin appears to be at this corner, the part seems to have bent before it broke. What cannot be said is how much of the section was left, what the internal surfaces look like, or whether the material matched specification. Those need the part itself, or a cut section from it, examined by someone who can measure. There is no substitute for getting the failed component back, which is why a returned part is worth more to a project than a description of it.
Once the origin is known, the change is usually straightforward to choose. An origin at a sharp internal corner asks for a radius. An origin in a heat-affected zone asks for a different joint preparation, a different heat input or a different filler selection. An origin where a bracket joins a tube asks whether the bracket should be there at all, or whether the tube should carry the load alone.
An origin in the parent metal, well away from any weld, is the most interesting result, because it points at something other than the fabrication: restraint, support, or a coincidence between the shape of the part and the frequency the engine produces. Those cases are answered by changing the layout rather than the process, and the properties of the stock involved are worth checking at the same time through our shop by material listings.
Failure history is one of the most useful things a customer can bring to a custom job, and it is often left out. The points below are the ones worth passing on.
With those details, the next version can be designed to answer the actual cause. The same discipline applies across our exhaust pipe and other parts work, and to charge-side assemblies such as intercooler units.
A fracture reading is most valuable to the person who built the part, because they know what was done to it. They know whether the shielding slipped, whether the fit-up had a gap, whether the tube was annealed or as-supplied, whether the mount was added late in the job. A customer can describe a failure well, but only the fabricator can match the description to the sequence of decisions that produced the part.
That is also why returned parts improve later work more than any single specification change. Each one closes a loop between what was designed and what actually happened, and those loops accumulate into better judgement on the next job. Customers who want a design reviewed after a failure can raise it through our customization process.
