A steel bracket holds 10,000 lbs in a static test. You load it to 4,000 lbs in service — well under half its strength. Six months later, it snaps without warning. This is fatigue failure, and it accounts for an estimated 80-90% of all mechanical failures in service. If you design or select anything that moves or vibrates, you need to understand it.

What's happening at the micro level: Cyclic loading — repeated stress/unstress or stress reversal — nucleates tiny cracks at stress concentrations (notches, holes, sharp corners, surface scratches). Each cycle grows the crack a microscopic amount. The part looks fine until the remaining cross-section can't carry the load, and it fractures suddenly. The failure surface is distinctive: a smooth "beach mark" region where the crack propagated slowly, and a rough granular region where it finally broke.

The S-N Curve: The relationship between stress amplitude (S) and number of cycles to failure (N) is captured in an S-N curve (also called a Wöhler curve). For ferrous metals like steel, there's typically an endurance limit — a stress below which the part theoretically survives infinite cycles. For most steels, this endurance limit is roughly:

S_e ≈ 0.5 × S_ut (for ultimate tensile strengths below ~200 ksi)

So a steel with 80 ksi ultimate tensile strength has an endurance limit around 40 ksi. Load it cyclically below 40 ksi and it should last indefinitely. Aluminum and copper have no true endurance limit — they'll eventually fail at any cyclic stress, which is why aircraft have mandatory inspection intervals.

What kills your endurance limit in practice:

• Surface finish — A machined surface might reduce S_e by 20%. A rough as-forged surface by 60%+.
• Stress concentrations — A sharp-cornered keyway or a drilled hole without a fillet creates local stress 2-4× the nominal value.
• Corrosion — Corrosion pitting creates crack initiation sites. A corroded steel part may have essentially zero endurance limit.
• Temperature — Elevated temps reduce fatigue life.

Real-world example: The 1954 de Havilland Comet disasters. The world's first commercial jet suffered explosive fuselage failures. Investigation found fatigue cracks originating at the sharp corners of the rectangular cabin windows. Each pressurization cycle stressed those corners. The fix for all subsequent aircraft: oval or round windows with generous radii — eliminating the stress concentration.

Rule of thumb for design: Use generous fillets (radius ≥ 1/8 of section thickness) at all transitions. Specify surface finish on fatigue-critical parts. And if your part sees cyclic loads, design to the endurance limit, not the yield strength.