An exhaust system is asked to do two jobs at once, and they pull in different directions. It has to carry hot gas away from the engine while staying sealed, and it has to do that while being shaken continuously by the engine, the road and the chassis it is bolted to. A part can be strong enough for the first job and still fail at the second, which is why vibration deserves to be treated as its own subject rather than as a footnote to heat.
Engine vibration arrives at a steady frequency that rises and falls with engine speed. Road input is irregular and unpredictable. Body flex adds a slow, large movement on top. A fabricated exhaust sits at the junction of all three, held by a handful of mounts, and every one of those inputs finds its way into the same welds.
A fatigue failure starts where stress concentrates, and welds concentrate stress by their geometry. The toe of a weld, where the bead meets the parent metal, is a shape change. Any undercut, any slight mismatch in fit-up, any place where two tubes meet at an angle rather than lining up, adds to that concentration.
There is a second factor that has nothing to do with shape. The material beside a weld has been through a different thermal history from the tube further along, and on thin-wall stainless or titanium that difference is significant. The result is a narrow band that is slightly harder, slightly different in structure, and located right where the load transfers from one section to the next. Fatigue cracks in fabricated exhausts overwhelmingly begin in that band or just outside it, rather than in the middle of a straight run.
Length changes the picture. A long unsupported section of tubing behaves like a beam, and it moves most in the middle of its span. If a system is supported only at its ends, the middle flexes with every bump, and the small angle changes at each end become the repeated strain that the joints have to absorb.
The practical consequence is that the first crack usually appears at a fixed point rather than a random one. It appears where a bracket, a hanger rod or a flange prevents the pipe from moving with its neighbours, which is also where the pipe is thinnest or most worked. A bend just after a mount is a common location, and a short stub of pipe between two rigid points is another.
A good mount carries weight and lets the pipe expand lengthwise at the same time. A bad mount carries weight and stops the pipe from moving at all. The second type looks more secure on the bench and fails sooner in service, because every expansion cycle is being converted into stress at that point instead of being allowed to travel along the run.
Hanger position matters for the same reason. Moving a support a small distance along a pipe can change which joint takes the strain without changing any material or weld on the part. It is one of the cheapest changes available, and it is often the right answer when a system has already cracked once. The tubing itself carries work of this kind on our exhaust pipe pages.
A fatigue crack leaves a story behind. The surface near the origin is usually smoother, having been worked back and forth many times, and it grows in steps that leave a pattern of fine lines spreading outward. The final separation, when the remaining section lets go, is rougher and duller. Read in that order, the two zones tell you where the crack started and which direction it travelled.
An overload failure looks different. It shows a single rough surface, often with a necked or stretched region, and no evidence of slow growth at all. The distinction is not academic: a fatigue crack points at restraint, support or a detail that needs changing, while an overload failure points at something that was struck, over-pressured or badly fitted.
Where a part has been through our titanium welding parts route, the origin is often visible in relation to the weld toe, and the location alone is usually enough to suggest the next iteration.
A bellows or braided flexible section is often described as a cure for cracking, and it is a useful component rather than a cure. It absorbs lengthwise movement and small angular deflection, which relieves the pipe that would otherwise be pushed and pulled by expansion. What it does not do is remove vibration from a badly supported system, and a flexible section that is installed under tension, or bent through a tight angle, will fatigue itself quickly.
The same caution applies to soft joints in the charge side of the engine bay. A silicone connection tolerates movement that a rigid pipe cannot, but it has to be routed so the movement it accepts is the movement the system produces. Our performance silicone hose range covers exactly that role, and the routing decisions around it matter as much as the hose itself.
Most fabricators do not have a shaker table, and most do not need one. Useful information arrives from simpler places. A part that has run on a car for a season can be inspected at every joint and bracket, and the pattern of where fretting, polished contact marks or loose fasteners appear will point at the areas that are moving.
Listening helps as well. A resonance that shows up as a drone or a buzz at one particular engine speed is telling you that a section of the system has a natural frequency the engine can excite. Changing the length of an unsupported span, or adding a mount in the right place, moves that frequency away from the range the car actually uses.
The important limit is that road and track use validates one installation, on one vehicle, over a limited time. It does not prove that a part will survive elsewhere without the same support arrangement, and it does not predict how much longer a joint that has already begun to fretting would have lasted.
A custom exhaust or intake part is normally supplied as a component, not as a complete installation, so the support arrangement belongs to the customer. That makes the following points worth settling early.
This is the kind of information that changes a design rather than a figure on a sheet. It is the same set of questions we ask on made-to-order work across our customization pages and our shop by material section.
When something cracks, the temptation is to make the obvious place thicker. That sometimes works, and it sometimes moves the failure a short distance along the pipe to a point that is now weaker by comparison. A better first step is to ask what the failed part was being asked to do, and whether the restraint, the span or the joint detail was the real cause.
The answer is usually a combination. A support in a better place, a joint prepared to sit flush rather than gap, a length of tubing made from a stock with more tolerance for movement, and a flexible section doing a job it was actually designed for. Fatigue is patient, and it rewards designs that give it nothing to work on.
Work of this kind at twin turbocharger installations shows how much of the answer lies in layout rather than in a single material choice.
