Frequency and Strain Rate in Fatigue Testing: Does Test Speed Matter?

Introduction

A question that comes up regularly in fatigue testing and vibration qualification programs is whether the frequency of cyclic loading — or the associated strain rate — influences the fatigue life of a component. The honest answer is: it depends very much on the material and the dominant failure mechanism.

For most structural metal fatigue work, the standard engineering assumption is that frequency has little or no effect over typical laboratory test frequencies (roughly 1–100 Hz), provided the specimen does not self-heat and the environment is not chemically aggressive.

Polymers, elastomers, and composites are a different story; their stiffness and damping are inherently rate dependent because the underlying deformation mechanisms are viscoelastic. At elevated temperature, where creep mechanisms become active, frequency can become a dominant variable because slower cycling allows more time for creep damage to accumulate per cycle. This post works through each of these regimes and the physical reasons behind them — including the extreme end of the frequency range, ultrasonic fatigue testing at roughly 20 kHz.

Metals: Frequency as a Second-Order Variable

For conventional high-cycle fatigue (HCF) of metallic alloys — aluminum, steel, titanium — the dominant variables are well established:

• Stress amplitude is the primary driver of life, as captured by the classical Basquin relation:

where \(\sigma_f’\) is the fatigue strength coefficient and \(b\) is the fatigue strength exponent.

• Mean stress is important and is typically corrected for using a Goodman, Gerber, or Walker formulation.
• Surface finish, residual stress, and environment are significant secondary variables.
• Frequency, by contrast, is usually only a second-order influence on the underlying S-N behavior.

This is why a rotating-beam fatigue machine running at 50–100 Hz and a servo-hydraulic load frame running at 5–20 Hz on the same alloy will generally produce comparable S-N curves. The dislocation mechanisms that drive cyclic plasticity and crack initiation in metals operate on time scales far faster than the mechanical loading frequency itself, so the cycle count — not the rate at which cycles are applied — is the controlling damage parameter.

That said, frequency does matter indirectly whenever it changes one of the following:

• Specimen self-heating. At high frequency and high stress amplitude, hysteretic energy dissipation per cycle, multiplied by a high cycling rate, can raise the bulk specimen temperature enough to alter material properties and accelerate damage. This is a heat-generation effect, not a fundamental rate effect on the fatigue mechanism itself.
• Corrosion and oxidation kinetics, discussed further below.
• Creep, relevant primarily at elevated temperature.
• Dynamic material response, relevant only at strain rates far above the quasi-static regime.

Strain Rate: A Related but Distinct Variable

Strain rate is conceptually different from cyclic frequency, though the two are obviously coupled through the strain amplitude and waveform shape. Many metals exhibit classic strain-rate sensitivity: as strain rate increases, flow stress increases and ductility decreases, a behavior captured by constitutive models such as Johnson-Cook and Cowper-Symonds.

For ordinary fatigue testing, the imposed strain rates are far below the threshold where this effect becomes significant — typically several orders of magnitude below the \(10^2\)–\(10^4\ \text{s}^{-1}\) range where dynamic strengthening becomes appreciable. Consequently, strain-rate effects on the underlying material flow behavior are negligible for conventional fatigue qualification.

The picture changes substantially at very high strain rates: impact loading, crash loading, explosive loading, and projectile impact. Under these conditions, the strain rate can shift the material into a regime where flow stress is elevated and ductility is reduced, which in turn affects crack initiation behavior, fracture toughness, and the residual fatigue properties of a structure that has survived an impact event. This is a separate engineering problem from conventional HCF qualification and is addressed using dynamic constitutive models in hydrocode or explicit FE simulations rather than standard S-N curves.

Ultrasonic Fatigue Testing: The Extreme End of the Frequency Range

Conventional testing at 1–100 Hz becomes impractical once the goal is characterizing very high cycle fatigue (VHCF) behavior beyond \(10^8\)–\(10^9\) cycles. At 100 Hz, reaching \(10^9\) cycles takes roughly 115 days of continuous machine time; reaching \(10^{10}\) cycles — increasingly relevant for turbine blades, valve springs, and other high-frequency rotating or vibrating hardware — takes over three years. This is the practical motivation for ultrasonic fatigue testing.

Ultrasonic fatigue machines drive the specimen at its own longitudinal resonance, typically near 20 kHz, using a piezoelectric transducer and a horn that amplifies displacement to the target strain level. At 20 kHz, \(10^9\) cycles takes about 14 hours and \(10^{10}\) cycles takes about 6 days — a speedup of roughly 200 to 2000 times relative to conventional servo-hydraulic or rotating-beam testing. The method is standardized in ASTM E2948 and has become common practice for VHCF material characterization.

Agreement with conventional-frequency data tends to be good when the fatigue mechanism is dislocation-based crack initiation in an inert or dry environment — many aluminum alloys and steels show consistent S-N behavior across the 20 Hz to 20 kHz range in this regime, since the underlying slip mechanisms are effectively rate-insensitive at the strain rates involved.

Agreement breaks down, and ultrasonic data should be treated with caution, in three situations:

• Environmentally assisted fatigue. Corrosion fatigue and stress corrosion cracking are controlled by time at the crack tip, not cycle count. At 20 kHz the crack tip is held near peak stress for roughly 1/1000th the time per cycle compared to a 20 Hz test, starving time-dependent chemical attack of the exposure it needs. Ultrasonic testing in ambient air can substantially overstate fatigue life relative to a corrosive or humid service environment.
• Specimen self-heating. Sustained 20 kHz cycling generates significant hysteretic heating, particularly at higher stress amplitudes or in metals with limited thermal conductivity. Standard practice is intermittent burst-mode operation (a fraction of a second on, several seconds off) with active air or water cooling and infrared monitoring, keeping specimen temperature rise below roughly 5–10°C. Undocumented self-heating is one of the more common sources of disagreement in the published ultrasonic-versus-conventional literature.
• Dwell-sensitive titanium alloys. For alloys susceptible to cold dwell fatigue, ultrasonic testing cannot reproduce dwell damage at all, since there is no meaningful hold time at peak stress at 20 kHz. Ultrasonic S-N data for these alloys characterizes the non-dwell mechanism only and should not be used as a stand-in for dwell-corrected life estimates.

Frequency Effects on Fatigue Crack Growth

Frequency has a more clearly documented influence once a structure transitions from crack initiation to crack propagation, because crack growth rate depends on processes that occur at the crack tip on a real-time basis rather than purely on cycle count. Three mechanisms are typically responsible:

• Crack closure. The degree to which a fatigue crack remains partially closed during the loading cycle affects the effective stress intensity range \(\Delta K_{\text{eff}}\) driving growth, and closure behavior can be rate-sensitive.
• Environmental attack at the crack tip. Many environmentally assisted cracking mechanisms require a finite time for the environment to interact chemically with the freshly exposed crack-tip material.
• Oxidation kinetics. Oxide formation and oxide-induced crack closure both depend on exposure time, not cycle count.

A classic illustration is aluminum alloy crack growth in humid air. Lower test frequencies give the environment more time per cycle to interact with the crack tip, and crack growth rates can increase measurably even though the nominal stress intensity range \(\Delta K\) is held constant. This is fundamentally a time-dependent environmental effect superimposed on the mechanical (Paris law) crack growth process:

where the coefficient \(C\) itself becomes frequency- and environment-dependent under aggressive conditions, even though \(m\) is largely a mechanical material constant.

Composites

Carbon fiber reinforced polymer (CFRP) and glass fiber reinforced polymer (GFRP) composites behave quite differently from metals because the polymer matrix governs much of the rate-dependent response:

• Matrix viscoelastic behavior is inherently rate dependent — stiffness and damping (loss factor \(\tan\delta\)) both vary with loading frequency.
• Hysteretic heat generation can become significant at typical composite fatigue test frequencies, since polymer matrices have far lower thermal conductivity and higher internal damping than metals.
• Fatigue life often changes measurably with test frequency as a result.

For this reason, composite fatigue testing is commonly restricted to roughly 5–10 Hz, well below the rotating-beam or servo-hydraulic frequencies typical of metal testing, specifically to avoid artificial self-heating that would not be representative of the in-service loading environment.

Elastomers

Rubber and other elastomeric materials show the strongest frequency dependence of any class discussed here. Increasing frequency simultaneously increases:

• dynamic stiffness, due to the viscoelastic storage modulus increasing with rate,
• internal heating, due to hysteretic loss,
• and hysteresis itself, which is the source of the heat generation.

The combination can dramatically reduce fatigue life at high frequency relative to low frequency, even at nominally identical strain amplitudes. Elastomer fatigue and durability testing protocols must therefore specify frequency carefully and account for thermal equilibration time.

Implications for Vibration Qualification Testing

This subject is directly relevant to aerospace vibration qualification practice. Random vibration tests are commonly compressed into 1–3 minutes per axis — far shorter than the actual service-life loading duration the hardware will experience. The underlying assumption that makes this acceptable is that fatigue damage for metallic structures depends primarily on the number and amplitude of stress cycles (captured through Miner’s rule and the component’s S-N curve), not on calendar time or the rate at which those cycles are applied.

This assumption is generally valid for metallic aerospace structures, provided the test does not introduce thermal or environmental effects that would not occur in the actual service environment. For satellite structures, launch vehicle hardware, and aircraft components, engineers therefore typically treat frequency as a non-factor in metal fatigue damage accumulation — only the stress spectrum and the accumulated cycle count matter. This is the physical basis for accelerated test time compression methods (e.g., scaling test duration via Miner’s rule equivalence) that are standard practice in MIL-STD-810 and NASA-HDBK-7004 type qualification programs.

The same assumption becomes much less reliable for composite or polymer-rich hardware, where self-heating and viscoelastic rate effects mean that compressed-duration, elevated-frequency testing may not faithfully reproduce the damage mechanism that would occur under the slower, longer-duration service environment.

Practical Rule of Thumb

Summary

Whether cyclic frequency or strain rate matters in fatigue testing depends fundamentally on which mechanism controls the damage process. For conventional high-cycle fatigue of metals at room temperature, frequency is a second-order variable and accelerated testing at higher frequency — including ultrasonic testing at 20 kHz for VHCF characterization — is generally an acceptable and increasingly standard practice, provided the fatigue mechanism is initiation-dominated, the environment is inert, and self-heating is controlled. Once the failure mechanism involves environmental attack at a crack tip, dwell-sensitive alloy behavior, viscoelastic matrix response, hysteretic self-heating, or creep, frequency moves from a second-order nuisance variable to a primary driver of fatigue life, and test frequency — however fast — must be selected and validated with care to avoid introducing non-representative damage mechanisms.

References