The part passed every test — and broke after a year
Every manufacturer knows this scenario. The product passed its load tests, survived the lab, shipped in thousands of units — and a year later, units start coming back with the same fracture in the same place. The cause is almost always material fatigue: failure produced not by one large load but by the accumulation of millions of small ones. Fatigue is the leading cause of field breakage in products that were designed to be "strong enough," because static strength and long-term durability are two entirely different properties.
How a small load breaks a strong part
When a part carries a varying load — press and release, bend back and forth, continuous vibration — each cycle does a tiny amount of microscopic damage, concentrated wherever local stress is highest. After enough cycles a small crack initiates. Once it exists, it advances a little further with every subsequent cycle, until the remaining cross-section can no longer carry the load and the part fails suddenly, usually with no prior deformation to warn anyone.
That is why fatigue failures feel arbitrary to users: nothing unusual happened on the day it broke. The part simply finished accumulating its cycles.
The fracture surface tells the story. A smooth, often shiny region with concentric beach marks shows where the crack propagated slowly, and next to it a rough, granular region marks the final sudden overload. If you have field returns, photograph and examine those surfaces before anything else — they identify the initiation point, which identifies the design error.
The S-N curve, and the difference between metals
The governing relationship is the S-N curve: the higher the stress amplitude, the fewer cycles until failure. A part running at high stress may fail in thousands of cycles; the same part at low enough stress will survive millions.
Material class matters enormously here. Many steels exhibit an endurance limit — a stress level below which the part effectively lasts indefinitely. Aluminum alloys and most plastics do not: they keep accumulating damage even at low stress amplitudes, which means such parts must be designed to a defined finite life rather than to "forever." That single distinction changes how you specify a component, and it is one of the underappreciated inputs when comparing candidate materials — see aluminum vs steel.
Plastics add two more complications: they creep under sustained load, and their fatigue behavior is strongly affected by temperature and by internal heat generated during rapid cycling. A polymer part cycled quickly can heat itself into failure at a stress that would be safe if cycled slowly.
Where fatigue attacks: stress concentrations
- Sharp corners and abrupt section changes. Any sudden change in geometry multiplies local stress. The fatigue crack will start there, essentially without exception.
- Holes, slots, and thread roots. A thread root is a textbook stress concentrator. Bolts carrying cyclic load fail at the first engaged thread, which is why preload and thread engagement design matter so much — and why cutting threads directly into plastic under cyclic load is risky; see threads in plastic parts.
- Damaged surfaces. Scratches, coarse machining marks, tool witness lines, and wear scars are crack initiation sites. Improving surface finish in a critical region measurably extends fatigue life, and processes such as shot peening improve it further by putting the surface into compression.
- Welds and heat-affected zones. Welds contain micro-defects and residual tensile stresses, making them the weak point of any welded structure that vibrates. Where possible, keep welds out of the primary cyclic load path.
- Press fits, keyways, and corners of pockets. Anywhere a load is transferred abruptly between two bodies.
How to design a part that lasts years
Fatigue resistance is first of all geometric. Use generous radii at every section transition, move holes away from highly loaded regions, and spread load over a wider cross-section to reduce stress amplitude — halving the stress amplitude typically buys orders of magnitude more cycles, which is far more effective than upgrading the material.
In simulation, do not stop at peak static stress. Look specifically at the cyclic stress range at the concentration points, and evaluate mean stress as well as amplitude, since a tensile mean stress makes fatigue considerably worse. Guidance on setting that analysis up sensibly is in FEA simulation in product design.
In plastics you can deliberately harness the phenomenon rather than fight it: a properly proportioned living hinge in polypropylene is a part designed from the outset for many thousands of flex cycles, and the same logic applies to snap-fit features that are opened and closed repeatedly rather than once. And when the cyclic loading comes from vibration — in transit or in operation — the mitigation principles are the ones described in designing for drop and shock testing, applied over many more cycles at lower amplitude.
Fatigue and factor of safety are two separate checks
Do not conflate them. A static factor of safety answers "will the part survive the single largest load it will ever meet," and is set the way we describe in factor of safety in mechanical design. Fatigue answers a completely different question: "how many times can the part carry its ordinary working load." A component can pass the first check with a comfortable margin and fail the second badly — which is precisely what happens to products that break after a year. A complete design review runs both, against separate criteria.
Test it before the market tests it for you
Fatigue life is verified with accelerated cyclic testing: bending, pressing, or vibrating the part tens of thousands of times in a dedicated fixture during development, at load amplitudes representative of real use. Two cautions. Accelerate by cycle rate, not by raising load beyond the intended amplitude, or you will change the failure mode and learn nothing about real life. And in polymers, watch fixture and specimen temperature, because rapid cycling introduces heating the field product will never see.
A fatigue crack found in the lab costs a design revision. The same crack found across thousands of customers costs reputation, warranty, and sometimes the product line. More on durability engineering in our mechanical engineering hub.
If you have a part failing in the field and want the root cause diagnosed, or a new design expected to see cyclic loading and want it engineered properly the first time, contact Projects House through our form and we will take a look.