Passed the Pull Test, Cracked After Months of Vibration

On vehicle parts, power tools and appliances, the most common field failure of a welded plastic assembly is not being pulled apart but cracking under vibration. Every one of those parts passed a static pull test at despatch.

How fatigue differs from static loading

A pull test measures how much force the part carries once. Fatigue is an alternating load far below that value under which a microcrack starts at a stress concentration and grows with cycles until the remaining section fails suddenly.

Two consequences: the failure load can be far below the static strength, and failure is abrupt, with almost no warning beforehand.

So on parts that see vibration, raising the static pull figure achieves nothing. What has to fall is the stress concentration.

Where the concentrations are

Sharp roots at the joint. The root of an energy director or the step of a shear joint, left sharp, is where the crack starts. Adding a radius on the drawing is the most direct improvement to fatigue life available.

The edge of the flash. Where flash meets the base material there is a geometric notch, equivalent to a pre-made crack. Controlling flash volume and adding a trap is not only cosmetic.

Discontinuous sections of the weld. Where fusion is uneven, the end of an unfused region is a crack source. Here uniformity matters more than average strength: a joint that is strong on average but locally weak has the fatigue life of its weak section.

Residual stress. Stress locked in by welding adds to the service stress, raising the mean stress level and reducing fatigue life accordingly. Lower force, longer hold and less forced fixture constraint all reduce it.

Abrupt wall transitions. Do not route the joint across a change in wall thickness.

The material layer

Fatigue behaviour varies widely by material. Semi-crystalline grades such as PA and POM generally fatigue better than amorphous ones. Glass reinforcement raises static strength substantially, but fibres do not cross the interface, so fatigue cracks propagate along it readily, and the fatigue gain at a weld is far smaller than the static strength gain.

This is routinely overestimated in design: the strength data for a filled grade doubles, so the weld is assumed to double too. It does not.

Validate against the real spectrum

Fatigue cannot be extrapolated from a pull test. Run a rig test to the product's actual vibration spectrum, in frequency range, acceleration level and direction, and inspect after the target cycle count: whether the joint has cracked, whether a sealed part has begun to leak, whether dimensions have moved.

Then section the part and find where the crack started. That location points straight at what to change: root radius, flash, seating or residual stress.

Consistency on the line

Fatigue failure is especially sensitive to individual weak parts. A few under-fused parts in a batch become a few early field failures once assembled. Per-part monitoring of energy, power and post-weld height with tightened limits, taking those parts out before despatch, is the most effective control on field failure rate.

Post-weld height is particularly useful, since it reflects the amount of fusion directly and exposes local shortfalls better than energy does.

Root radii, flash traps and residual stress control are in the design handbooks on the downloads page. Hold control and per-part monitoring are on the ultrasonic welding machine pages.

Send the vibration conditions, the material and the drawings to 1427498429@qq.com, or call +86 769 8202 9510 / +86 137 6001 0932.