Hertz to Cycles per Minute

Divide ten million by the cycles-per-minute figure. At 5 Hz — a rate a mid-size hydraulic frame holds comfortably on a coupon — that is 300 per minute and 23.1 days of unbroken running. Lift the rate to 30 Hz and the same runout takes 3.86 days; a resonant machine at 100 Hz finishes in 27.8 hours. The gap becomes brutal in the very-high-cycle regime: a gigacycle at 5 Hz would run for roughly 6.3 years, which is exactly why ultrasonic rigs at 20 kHz exist — they cover the same billion cycles in under fourteen hours.

They create load in completely different ways. A servo-hydraulic frame forces the piston to follow a commanded waveform, and each cycle demands enough oil flow to move the actuator through the full displacement twice, so valve capacity and specimen compliance cap the rate. A resonant machine instead tunes the moving mass until the assembly wants to vibrate at the chosen frequency, then tops up only the energy lost to damping. The trade-off is control. Hydraulics reproduce any waveform, including slow strain-controlled ramps and replayed service histories; a resonant rig is largely restricted to constant-amplitude sine near its own tuned frequency, and changing specimen geometry moves that frequency.

It drifts downward, and that drift is the most useful signal the test produces. A dwell parks the excitation on a resonance located during a sweep, so the article is driven at its own natural frequency and the response is amplified many times over the table input. As a crack initiates and extends, local stiffness falls and the natural frequency falls with it. A controller running closed-loop tracking follows the peak to keep the dwell centred, and the tracked frequency becomes a crack-growth trace — a drop of a fraction of one percent is commonly enough to trigger an inspection or an end-of-test. Hold a fixed frequency instead and the specimen simply walks out from under the excitation, so the response quietly collapses while the counter keeps climbing.

Enough to change the result, which is why the rate cannot simply be raised until the schedule fits. Energy lost to hysteresis on every reversal is deposited in the gauge section, and the power going in scales with the count per minute while convection off the surface barely changes. Polymers and thick laminates suffer worst because they damp heavily and conduct badly — plastics work is often held to a few hertz for that reason alone, with a thermocouple or an infrared spot watching the surface and a temperature rise limit agreed in advance. Metals tolerate far more, yet ultrasonic rigs still run in pulse-and-pause bursts with compressed air on the specimen. If a fast run gives a shorter life than a slow one at the same stress amplitude, suspect heating before you suspect the material.

Constant-amplitude testing makes it trivial: one sine period is one cycle, so the per-minute rate and the machine counter agree exactly. A measured service history has no such rhythm — it is a jagged sequence of reversals with small excursions riding on large ones. Rainflow counting is the standard way of decomposing it, pairing each reversal with its matching partner to extract closed hysteresis loops as whole cycles and leaving unmatched residuals as half counts. The resulting histogram of ranges is what feeds a damage sum, so an hour of road-load data may reduce to a few thousand counted cycles spread over a dozen amplitude bins. When that history is replayed on a rig, the cycles-per-minute figure describes how fast the block repeats rather than how fast damage accrues, and the two have to be tracked as separate numbers.