Not "how many cycles until failure" — that's the symptom. The real split is whether the material plastically deforms every single cycle.
Fatigue itself is cyclic-stress-driven crack growth — a completely different mechanism from creep, which is steady-stress deformation. But fatigue itself splits into two regimes that behave, and are analyzed, very differently. High-cycle fatigue (HCF) runs at low stress amplitudes, well within the material's elastic range, and takes a large number of cycles — typically more than 104–105 — to initiate and grow a crack to failure. Low-cycle fatigue (LCF)runs at high stress amplitudes that push the material past yield on every cycle, generating a small amount of plastic strain each time, and it fails in far fewer cycles as a direct result. The dividing question isn't "how many cycles will this take" — that's the outcome. It's "is the strain each cycle elastic, or does it include a plastic component," and that determines which design method actually applies.
In the elastic regime, stress and strain are locked together by Hooke's law, so it doesn't matter whether you track stress amplitude or strain amplitude — they carry the same information, and the simpler one (stress) is what Basquin's stress-life relationship uses. That's the entire high-cycle-fatigue design method: keep stress amplitude below a safe curve (often approaching a flat endurance limit for steels around 106–107 cycles), and life follows. The moment stress amplitude gets high enough to cause yielding every cycle, that link breaks — stress plateaus near the material's cyclic yield strength even as strain keeps growing, so stress amplitude alone stops distinguishing a part that's barely damaged from one that's about to crack. The Coffin-Manson strain-life relationship exists precisely to cover that regime: it correlates fatigue life against the plasticstrain amplitude directly, which is what actually accumulates the damage — each cycle's worth of dislocation motion and microstructural rearrangement — when the material can no longer stay elastic.
Wrong, and it's backwards for the low-cycle regime specifically. High-cycle fatigue does involve low stress — that's the part people remember, because it's the classic "rotating shaft fails after millions of cycles below its static strength" story. But low-cycle fatigue happens under high stress — high enough to cause real plastic deformation on every single cycle — and that's exactly why it fails in so few cycles. A pressure vessel nozzle cycled through large thermal or pressure transients, or a landing gear component loaded near yield on every takeoff-landing cycle, can fail in the hundreds or low thousands of cycles specifically because each cycle does substantial plastic damage, not despite the stress being high. The number of cycles to failure and the stress level move in oppositedirections across the whole fatigue spectrum: more stress means fewer cycles needed to fail, not more. "Fatigue" as a category spans both ends — it is not synonymous with "low stress."
Explains the real dividing line between high-cycle and low-cycle fatigue — not simply 'how many cycles' but whether the material's strain each cycle stays elastic (high-cycle, analyzed with Basquin's stress-life relationship) or includes a real plastic component (low-cycle, analyzed with the Coffin-Manson strain-life relationship). Illustrated with an S-N curve split into both regimes and side-by-side stress-strain hysteresis loops.
High-cycle vs. low-cycle fatigue is a split within fatigue itself — both regimes involve cyclic, repeated stress, and neither has anything to do with sustained constant loading at elevated temperature (that's creep). The question here is not 'is the load cycling,' it's 'how much of each cycle's strain is plastic.' A rotating shaft under a small, purely elastic bending stress millions of times is high-cycle fatigue. A pressure vessel nozzle yielding a little on each thermal startup-shutdown cycle, only a few thousand times over its life, is low-cycle fatigue. Both are fatigue; neither is creep.
High-cycle fatigue design (the stress-life or Basquin approach) plots stress amplitude against cycles to failure on a log-log S-N curve and, for many ferrous alloys, identifies an endurance limit — a stress amplitude below which fatigue life is effectively infinite (conventionally taken around 10^6-10^7 cycles). Low-cycle fatigue design (the strain-life or Coffin-Manson approach) instead correlates the plastic strain amplitude per cycle against cycles to failure, because once yielding happens every cycle, stress amplitude plateaus near the cyclic yield strength and stops being a useful predictor — strain keeps distinguishing a lightly-loaded part from a heavily-loaded one when stress no longer can.
High-cycle fatigue governs most rotating and vibrating machinery — shafts, bearings, springs, turbine blades under steady operating loads — where millions of small stress reversals accumulate over a service life. Low-cycle fatigue governs components that see relatively few but severe load excursions: pressure vessel and piping nozzles cycled by startup/shutdown thermal transients, aircraft fuselage skin cycled by cabin pressurization once per flight, and structural connections cycled by seismic loading. Correctly classifying which regime a component sits in determines which fatigue design curve and which material property data (S-N curve vs. strain-life coefficients) actually apply.
It's a conventional guideline rather than a hard physical boundary — commonly cited as roughly 10^4 to 10^5 cycles. Below that range, the stress amplitude needed to fail in so few cycles is usually high enough to cause meaningful plastic strain each cycle (low-cycle behavior); above it, stress amplitudes are typically low enough that strain stays elastic (high-cycle behavior). The real, physical dividing line is whether plastic strain per cycle is significant, not the cycle count itself — the count is just the usual symptom.
No. Many ferrous alloys (plain carbon and low-alloy steels) show a fairly distinct S-N curve plateau, treated as an endurance limit below which life is effectively infinite. Most non-ferrous metals — aluminum, copper, and their alloys among them — do not show a true plateau; their S-N curves keep sloping gently downward indefinitely, so fatigue design for these materials specifies an allowable stress at a defined finite life (e.g., 10^8 cycles) rather than a true infinite-life limit.
Once a material yields every cycle, stress amplitude stops varying much with additional damage — it saturates near the material's cyclic yield strength regardless of how much plastic strain is actually being imposed. Strain amplitude, particularly the plastic strain component, keeps increasing with more severe loading and correlates far better with the actual cycle-by-cycle damage (dislocation motion, microstructural degradation) that drives crack initiation in this regime.
Yes, and this combination — often called high-cycle fatigue superimposed on low-cycle fatigue, or HCF/LCF interaction — is a real design concern in turbomachinery. A turbine blade sees relatively few large low-cycle stress excursions from engine startup and shutdown, plus millions of small high-cycle vibratory stress cycles superimposed on top from blade resonance during steady operation. Each mechanism has to be evaluated, and their combined effect assessed, rather than treating the component as purely one regime or the other.
The modified Goodman criterion is a mean-stress correction built for the high-cycle, stress-life framework, and it assumes the material response stays elastic. It is not the appropriate tool for low-cycle fatigue, where the analysis needs to account for cyclic plasticity directly — strain-life methods (Coffin-Manson, or a combined strain-life equation with both elastic and plastic terms) are used instead.
Try our Materials Science & Metallurgy Studio
More calculators, simulators, and guides for this discipline.