Why parts break well below their yield strength — and why the number of times a load is applied matters as much as how big it is.
A steel bracket can hold a one-time load ten times over and survive with room to spare. Bolt that same bracket into a machine that flexes it a small amount a few hundred times a minute, and it can snap in half within a year — at a stress level far below anything that would have bent it, let alone broken it, in a single application. That is fatigue failure: damage that accumulates from repeated, cyclic loading, not from any single load being too large. It is one of the most common causes of mechanical failure in real service, and it is governed by an entirely different set of rules than static strength.
Fatigue starts as a microscopic crack, almost always at a stress concentration — a sharp corner, a notch, a fillet, a hole, a weld toe, a machining mark, or any other surface flaw where the local stress runs far higher than the nominal stress calculated from σ = F/A. Each load cycle opens and closes that crack a tiny amount, extending it by a barely measurable distance. Nothing looks wrong from the outside for a long time. Then, once the crack has eaten away enough of the cross-section, the remaining material can no longer carry the load and fractures suddenly — often with almost no warning and very little visible deformation, which is what makes fatigue failures so dangerous compared to a slow, visible static overload.
A static strength check asks one question: is the applied stress below the yield or ultimate strength? A fatigue check asks a completely different question: at this stress amplitude, how many cycles will it take before a crack initiates and grows to failure — and does the part actually see that many cycles over its life? The S-N curve answers that question experimentally, by cycling identical specimens at different stress amplitudes and recording cycles to failure. Steel and titanium show a genuine flattening of that curve — a stress amplitude below which fatigue cracks essentially stop propagating, giving a true endurance limit, often cited as roughly 40-50% of the ultimate tensile strength for steel. Aluminum, magnesium, and most other non-ferrous alloys never flatten out; their fatigue strength is always quoted at a specific number of cycles (commonly 5×10⁸) rather than as a limit, because the curve keeps sloping downward no matter how far out you test it.
False, and it is one of the most consequential misunderstandings in mechanical design. Static yield strength and fatigue strength are entirely different material properties, measured by entirely different tests, and there is no reliable way to derive one from the other except through established empirical relationships and testing. A part can be loaded to a fraction of its yield strength — comfortably "safe" by static logic — and still fail in fatigue once it accumulates enough cycles, especially at stress concentrations that static-only thinking routinely ignores: a sharp fillet radius, an unfinished hole edge, or a weld toe can locally raise the true stress several times higher than the nominal calculation suggests. Surviving one big load proves nothing about surviving a million small ones. That is precisely why fatigue analysis is a required, separate check in the design of anything that sees repeated or vibratory loading — shafts, springs, aircraft structures, pressure vessel nozzles, weldments, and rotating machinery among them.
Explains why fatigue failure happens under repeated, cyclic loading at stress levels far below a material's static yield strength — driven by microscopic crack initiation at stress concentrations and slow crack growth — and why steel and titanium exhibit a true endurance limit on the S-N curve while aluminum and most non-ferrous alloys do not.
Most static-strength thinking treats a part as either safe or unsafe based on a single stress-versus-strength comparison. Fatigue doesn't work that way: it is a damage-accumulation process driven by the number of load cycles as much as the stress level. A component can be loaded to only a fraction of its yield strength on every cycle and still fail, because a microscopic crack — almost always starting at a stress concentration — grows a tiny amount every cycle until the remaining cross-section can no longer carry the load.
The S-N curve (stress amplitude, S, versus cycles to failure, N, typically plotted on a log cycle axis) is generated by cycling identical specimens at different stress amplitudes and recording how many cycles each survives before fracture. For steel and titanium, this curve flattens into a horizontal asymptote, usually somewhere around 10⁶–10⁷ cycles — the endurance limit (Se) — below which the material can theoretically endure an unlimited number of cycles without fatigue failure. Aluminum, magnesium, and most other non-ferrous alloys show no such flattening; their curve keeps sloping downward indefinitely, so their fatigue behavior is instead reported as a fatigue strength at a specified number of cycles (often 5×10⁸), since failure will eventually occur at any nonzero cyclic stress amplitude given enough repetitions.
Fatigue is the dominant failure mode in rotating shafts, springs, fasteners, weldments, aircraft structures, pressure vessel nozzles, and any component subjected to vibration or repeated load cycles in service. Designing against it requires identifying and mitigating stress concentrations (generous fillet radii, smooth surface finishes, shot peening, avoiding sharp threads or undercuts at highly loaded sections), and applying appropriate safety factors against the S-N curve or endurance limit for the actual expected stress amplitude and cycle count — not just against static yield or ultimate strength.
Fatigue strength is the stress amplitude a material can withstand for a specified number of cycles, and it always applies. The endurance limit (or fatigue limit) is a special case: a stress amplitude below which the material can theoretically survive an infinite number of cycles. Steel and titanium typically have a true endurance limit; aluminum and most non-ferrous alloys do not, so their fatigue behavior is always expressed as a fatigue strength at a stated cycle count rather than a true limit.
The commonly cited metallurgical explanation is that in steel and titanium, interstitial atoms (like carbon and nitrogen in steel) pin dislocations, effectively 'locking' microstructural slip below a certain stress amplitude and halting crack propagation. Aluminum's crystal structure and alloying behavior don't produce this pinning effect in the same way, so microscopic damage keeps slowly accumulating at any nonzero cyclic stress, no matter how small, given enough cycles.
Yes, and this is routine in practice. Fatigue strength at a given number of cycles is often a small fraction of the static yield or ultimate strength — for many steels, the endurance limit is roughly 40-50% of ultimate tensile strength, and local stress concentrations can push the effective stress at a critical location far higher than the nominal calculated stress, making fatigue failure possible at loads that would never come close to yielding the part in a single application.
A near-dominant one. Almost all fatigue cracks initiate at a stress concentration — a sharp corner, notch, fillet, hole, keyway, weld toe, or surface defect — because the local stress there (nominal stress × the stress concentration factor, Kt) is far higher than the nominal stress used in a basic strength calculation. Reducing stress concentrations (larger fillet radii, polished surfaces, eliminating sharp re-entrant corners) is one of the most effective ways to improve fatigue life.
Typically via rotating-beam or axial fatigue testing (per standards like ASTM E466 or the classic R.R. Moore test): a batch of identical specimens is cycled at a chosen stress amplitude until each fractures, and the stress amplitude versus cycles-to-failure data points are plotted, usually with stress on a linear or log axis and cycles on a log axis, then fit with a curve or, for endurance-limit materials, a flattening asymptote.
Try our Mechanical Studio
More calculators, simulators, and guides for this discipline.