Designing for Fatigue? Your Steel Choice Might Be Wrong

Why do my parts keep failing even though the stress is below yield?

Because you're designing for static strength, not fatigue. Since 1830, engineers have known that a metal under repetitive stress fails far below its yield point (FAMU-FSU metal fatigue course report). Fatigue is the quiet killer: about 90% of all service failures from mechanical loading are fatigue failures. You do the stress calculation, pick a material with yield to spare—and then the part snaps after a few hundred thousand cycles. That's not bad luck; that's a design flaw.

Fatigue starts with a crack at a stress riser—a hole, a keyway, corrosion, a sharp radius—then propagates with every cycle until sudden fracture (Iowa State ME 325 fatigue notes). The stress that causes this can be far below yield. So if you're only checking static stress, you're missing the real enemy.

Is fatigue strength just a percentage of tensile strength?

For steels, yes—roughly. The uncorrected fatigue strength is about half the ultimate tensile strength, capped at 100 ksi for steels stronger than about 200 ksi UTS (Iowa State ME 325 fatigue notes). So if you choose a high-strength steel thinking it's automatically fatigue-resistant, you might be disappointed. A 150-ksi steel gives you an endurance limit near 75 ksi, but a 250-ksi steel still gives you only 100 ksi. That's a plateau, not linear gain. For aluminum and other nonferrous metals, there's no endurance limit at all—no stress below which they never fail. You have to design for a finite life, typically using the fatigue strength at 5x10^8 cycles (Iowa State ME 325 fatigue notes). So yes, the rule of thumb works for steels, but only below 200 ksi UTS, and it's meaningless for aluminum.

Does a stronger material always mean a better fatigue life?

Not necessarily. Yield strength is not fatigue strength. Take 7075-T6 aluminum: yield about 73 ksi, which is more than double the yield of 304 stainless steel at roughly 30–35 ksi (Engineers Edge). But 7075 has no endurance limit, so under millions of cycles it will eventually fail if any alternating stress is present. In contrast, a milder steel like 1018 (yield ~54 ksi) has a true endurance limit—if you keep stress below about half its UTS, it can last indefinitely (Iowa State ME 325 fatigue notes; Engineers Edge). So for a component that sees millions of cycles, a lower-strength steel can outlast a high-strength aluminum. The trick is to match the material to the load spectrum. If the load is steady, strength wins. If it's cyclic, you want a material with a high endurance limit, and that often means steel over aluminum unless weight is critical.

What's the real difference between stiffness and strength for fatigue?

Stiffness is not strength. Young's modulus E is the ratio of stress to strain—a measure of how much a material deflects under load (ETB Young's Modulus). Steel's E is about 210 GPa; aluminum's is about 70 GPa (Engineers Edge). So a steel part is three times stiffer than the same geometry in aluminum. But fatigue is driven by stress amplitude, not deflection. A flexible part can have lower stress because it can bend without concentrating strain. However, if you replace steel with aluminum in a fatigue-critical part, you'll need a larger cross-section to keep deflection and stress similar. The lower stiffness of aluminum means more elastic strain for the same stress, which can be forgiving, but the lack of an endurance limit is a killer. In short, stiffness affects how loads distribute; strength and fatigue limits determine where cracks start.

Are all steels alike when it comes to fatigue?

No. The carbon content changes the whole game. Low-carbon steels (