What is a fatigue crack, and where does it start in a welded socket?
Fatigue is progressive damage from repeated loading and unloading, often at stresses below the yield strength. A bracket that flexes every cycle can pick up microscopic damage over thousands or millions of cycles, and once a crack starts it can grow quickly. Many start in hard-to-see places: weld toes, bores and attachment points.
In a ball socket welded round its outside, the load goes from the pin through the ball and housing into the weld. Bending of the arm or a side load on the pin pulls on one side of the weld and pushes on the other, so the stress is highest at the weld toe on the loaded side, where the sharp change of shape concentrates it. Cracks there usually run along the toe, parallel to the weld. A crack that starts at the weld root, on the inside of the joint, cannot be seen from outside until it has grown through, which is one reason a visual check alone is not enough.
Why does the weld itself make a crack more likely?
Because welding builds in the two things fatigue needs: a notch and a tensile stress. The weld toe is a sharp change of shape, which concentrates stress. And the heat of welding leaves the metal near the weld stretched in tension, close to yield.
The size of that locked-in stress can be checked with arithmetic. Steel held rigidly while it heats up cannot expand, so it is pushed into compression; on cooling it tries to shrink back and is pulled into tension. For fully restrained steel the stress is E × α × ΔT. With E = 210,000 MPa and α = 12 × 10⁻⁶ per K, that is 2.52 MPa for every kelvin, so a swing of only about 100 K is enough to reach 250 MPa, the yield strength of an ordinary structural steel (illustrative values). Metal at the weld goes through far more than 100 K, so it yields while hot and, once cold, is left with a tensile residual stress near yield, balanced by compression farther away. That is why a weld toe can crack under a load that looks gentle on paper. The metal there already starts each cycle near yield, so what counts is the stress range of each cycle and the weld detail, not the peak load or the grade of steel. A cycle that is mostly compressive in the nominal calculation can still open a crack at the toe.
Misalignment makes it worse, and it is cheap to prevent. If the housing is tilted or offset in the fixture, or is pulled over by weld shrinkage, the load no longer goes straight through the weld. In a plate joint that carries load along its length, such as a pull plate or a lug, an offset e between the load line and the plate adds a bending stress. A plate 10 mm thick carrying a load with a 2 mm offset picks up a bending stress of 6e ÷ t = 6 × 2 ÷ 10 = 1.2 times the plain axial stress, so the stress at the toe is 2.2 times what a straight load path would give. Fatigue life of an as-welded detail falls roughly with the cube of the stress range (a slope of about 3 is the usual assumption, used here as an illustration), so 2.2 times the stress is about 2.2³ ≈ 10 times shorter life. A 1 mm offset gives 1.6 times the stress and about 4 times shorter life. The same sensitivity applies, less simply, to a tilted housing: a housing 60 mm wide that is out of square by only 0.5° puts one end of the bore about 0.5 mm off line, so small fit-up errors are worth checking. The same tilt also makes the pin bear on part of the ball width, which raises the contact pressure and the wear on that side (Q1-06).
Fixtures cut both ways. A rigid fixture stops the part from distorting, but the shrinkage that would have moved the metal is then locked in as residual stress, so a tightly clamped weld can leave a higher tensile stress than a freer one. Tack welds and weld start and stop points are small, hard-to-inspect starting places for cracks, and the last-welded side often holds the most stress. Re-welding a crack puts new heat and new residual stress into the same place, which is one more reason a quick repair rarely lasts (Q1-04).
For the inspection this gives you a few things to look for beyond the weld bead. Check for a gap or a step between the housing and the plate on one side, and for paint broken along that same side. Check whether the pin slides freely, and look for polishing or fretting debris on one side of the ball or pin only. Lay a square or straightedge across the housing to see whether it has pulled over. A crack that runs along the toe on the side with the gap or the polish points to a fit-up and alignment cause, not just a load that was too high.
Where do you look first?
Fatigue cracks commonly start at points of concentrated stress. Around a weld-on socket, that means:
- Weld toes. The edge of the weld where it meets the arm or bracket. A crack running from a weld toe deserves prompt attention, even if small.
- Weld start and stop points and any repair patches, where the shape of the weld changes.
- Undercut. A groove melted into the metal beside the weld.
- Holes, threads and sharp corners, and changes of thickness.
Look for other clues too:
- Rust stains or lifting paint along a weld. They may expose a crack that opens and closes in service. A narrow, repeating break in the paint is worth investigating, not touching up.
- Fine red-brown or black debris around pins and clamped joints. That is fretting, a sign that parts are moving under load. It does not prove fatigue, but it can start a crack.
- Distorted gussets, elongated holes, loose fasteners, local dents. These show the load path has changed.
- Changes in alignment, sound or vibration. These are not proof of fatigue, but they are a reason to inspect.
How do you confirm a crack?
Looking shows only what is visible, not the depth of a crack, its direction or whether there are others. The common non-destructive tests are:
| Test | What it finds |
|---|---|
| Magnetic particle | Surface and near-surface cracks in magnetic materials, such as many carbon steels |
| Dye penetrant | Cracks that reach the surface of a clean, non-porous surface |
| Ultrasonic | Internal flaws and loss of thickness in suitable sections |
| Eddy current | Certain conductive materials and complex surfaces |
On a ferromagnetic carbon steel weld, magnetic particle testing is usually the first choice for the toe, because it also finds cracks just under the surface. Dye penetrant finds only cracks that open to the surface, and it needs the paint and dirt removed first. No single method suits every job, and heavy coatings, rough fabricated surfaces and poor access can limit results. The practical plan is: clean the area, make a visual and size check, apply a suitable test, and compare with earlier inspection records. The same methods are described for weld-on rod eyes in K03-Q1-27.
When a crack is confirmed, record its position, length and direction, with photographs that show scale. Compare the direction with the direction of the load. A crack that runs along a toe on the loaded side points to bending of the arm or side load on the pin, which tells you where the cause is, not only where the damage is.
Is a cracked weld acceptable?
Not normally. Cracks have sharp tips that concentrate stress more severely than most other weld flaws, which is why most fabrication codes do not accept a crack of any size (American Welding Society, Inspection Trends). They must be removed, and if the dug-out area goes deeper than the minimum design thickness, it has to be filled with sound weld metal.
Should you stop, watch or repair?
It depends on the part's job, material, loading, crack position, how fast it grows and what happens if it fails. For a safety-critical item, a lifting point or a primary structural member, isolating the machine and getting an engineering assessment is generally the prudent course. Watching a crack instead of repairing it is only reasonable when the risk is understood, the crack is known not to be growing, and an engineer has agreed the interval between checks.
A weld-on socket often carries a lift arm or a hitch load, so treat a cracked one as a safety part until someone qualified has said otherwise.
What are the risks of a quick repair?
- A crack stop hole, welding over the crack, or adding a plate can look like a quick solution. Welding adds heat-affected zones, residual stress and new stress concentrations. A reinforcing plate can stiffen one section while moving load to a weaker area next to it.
- A repair on the visible crack alone can put the machine back to work without removing the cause. The result is often a repeat failure at the same place or right beside it.
You must also know the steel before you weld it, and the first question is what the material is. A hand-held PMI gun cannot measure light elements such as carbon and sulfur, which matter for weldability. A laboratory chemical analysis is used to set the minimum preheat and choose a filler metal, and whether the part is cast, forged or rolled also matters. Repair details follow a qualified welding procedure.
Why did it crack?
The best repair starts with finding out why the part failed:
- Was the section too small for the real load, or was the stress range larger than the weld detail can take (Q1-03)?
- Has vibration risen because of worn bearings or play in the joint (Q1-06)?
- Is the weld shape creating a notch?
- Has corrosion cut the effective section?
- Is the bracket being used as a lifting or access point it was not made for?
The answer may point to a different weld detail, a different material, more support, corrected alignment or a changed operating routine. Where the damage is widespread, or the weld stays hard to inspect, replacing the part may cost less than a complicated repair. For replacement steps, see Q1-08.