Knowledge / Ship systems

Propeller cavitation: how a non-uniform wake changes blade loading

Connect vapour formation, wake harmonics, blade incidence and pressure pulses through a worked cavitation-number example.

On this page

Propeller cavitation is vapour formation where local liquid pressure becomes sufficiently low. The hull wake matters because a rotating blade repeatedly crosses different inflow velocities, changing its incidence, loading and cavity volume. Mean ship speed alone therefore cannot explain whether cavitation appears, how strongly it fluctuates, or whether it damages a surface.

Start with pressure, not visible bubbles

A blade produces thrust by changing water momentum and establishing a pressure difference across its surfaces. In a sufficiently low-pressure region, vapour cavities can develop from available nuclei. When those cavities enter higher pressure, they can collapse. Cavitation is different from ventilation, where atmospheric air reaches the propeller, although both can occur in disturbed near-surface operation.

A photograph of a white region does not identify its composition or damage potential. Stable attached sheet cavitation, a trailing tip-vortex cavity and a collapsing cloud have different dynamics. Surface erosion depends on collapse close to material and repeated local loading; the mere presence of visible cavitation is neither a quantitative erosion rate nor proof of an unacceptable propeller.

A wake fraction is an average, not a map

An illustrative axial wake fraction is w = 1 − V_A/V_S, where V_A is representative propeller advance velocity and V_S is ship speed. It condenses a complicated field into one number. Two stern arrangements can have the same average w yet place very different velocity deficits near the upper blade passage. Radial and tangential flow components also change the velocity triangle.

The nominal wake describes the hull-generated flow without the propeller acting in it; the effective wake includes interaction with the working propeller. These should not be substituted without explanation. The ITTC numerical-prediction procedure discusses full-scale and effective-wake corrections. A computed average inflow that matches the thrust point can still miss a local wake peak.

Why a slow patch can increase suction loading

For a simplified section with fixed pitch and negligible induced-velocity changes, the inflow angle is approximately arctan[V_A/(2πnr)], with n in revolutions per second and r the section radius. Reducing axial velocity reduces that angle and can increase blade incidence. The suction-side pressure can then fall, even though the local axial speed became smaller. This is why reasoning only from ship speed is misleading.

The blade subsequently leaves the wake deficit and its cavity may shrink rapidly. Blade skew distributes the encounter in phase along the radius, but it does not erase the wake or guarantee freedom from cavitation. At off-design pitch, manoeuvring inflow or high loading, the favourable behaviour of a design point may change.

Cavitation number requires a stated reference

Define σ = (p_ref − p_v)/(0.5ρV_ref²), where p_ref is absolute reference pressure, p_v is vapour pressure at the stated temperature, ρ is density and V_ref is a declared reference velocity. A lower σ generally represents less pressure margin relative to that velocity scale. It is not a universal onset threshold: geometry, incidence and nuclei still matter.

The ITTC model-test procedure allows several velocity conventions and requires the selected reference to be identified. A value based on nD cannot be compared directly with one based on local resultant section velocity. Gauge pressure also cannot replace absolute pressure in the numerator unless the atmospheric reference is consistently restored.

Worked example: immersion changes the available margin

Assume seawater density 1 025 kg/m³, gravitational acceleration 9.81 m/s², atmospheric pressure 101.3 kPa, vapour pressure 2.3 kPa and a reference point 4.0 m below a calm surface. Take V_ref = 25.0 m/s. These are illustrative inputs, with hydrostatic pressure and no stern-wave correction. Then p_ref = 101.3 + 1 025 × 9.81 × 4.0/1 000 = 141.521 kPa.

The dynamic-pressure denominator is 0.5 × 1 025 × 25.0²/1 000 = 320.3125 kPa, giving σ = (141.521 − 2.3)/320.3125 = 0.4346. At 2.0 m immersion, keeping every other input fixed, p_ref becomes 121.4105 kPa and σ becomes 0.3719. The roughly 14.4% reduction describes this chosen nondimensional margin, not a 14.4% increase in erosion.

If V_S = 10.0 m/s and mean V_A = 7.0 m/s, w = 0.30. For n = 2.0 s⁻¹ and diameter D = 5.0 m, the advance coefficient J = V_A/(nD) = 0.70. These additional values locate an operating point; they do not specify the azimuthal wake distribution or a cavitation-inception boundary.

Blade frequency explains tones, not every pulse

A five-bladed propeller rotating at 2.0 s⁻¹ has blade-passing frequency Zn = 10.0 Hz. Repeated blade events can produce components at that frequency and its harmonics. A change in cavity volume adds pressure excitation to the non-cavitating blade contribution; irregular shedding also contributes broadband fluctuations.

The ITTC pressure-fluctuation procedure distinguishes phase-related harmonic analysis from broadband analysis. A single amplitude at 10 Hz cannot describe a short, steep collapse event fully. Hull pressure measured near the propeller is also a different quantity from far-field underwater radiated noise: position, propagation and structural response intervene.

What an informative test must preserve

A useful comparison records draught, speed, shaft rate, pitch where relevant, thrust or torque, pressure reference and water condition. Images need blade position and a time basis, so cavity growth can be connected with wake passage. Otherwise two attractive photographs may represent different operating points rather than better and worse geometry.

Model tests cannot match every full-scale similarity parameter simultaneously. Reynolds effects, nuclei population, model accuracy and wake reproduction require attention; the ITTC testing guidance explicitly treats these issues. Numerical output likewise needs temporal and spatial convergence and validation appropriate to the quantity being claimed.

Match the remedy to the observed mechanism

Improving inflow uniformity, redistributing blade loading, changing geometry or adjusting an operating point can address different mechanisms. Each option can affect efficiency, structural loading or another cavitation form. An efficient open-water propeller therefore need not be the quietest or most erosion-resistant choice behind a particular hull.

A defensible diagnosis connects where the cavity forms, when it collapses and which pressure or damage signature accompanies it. Repeat observations under comparable conditions are more useful than assigning every stern vibration to cavitation. Final acceptance belongs to the specified vessel, operating envelope and applicable contractual or class criteria, rather than to one photograph or calculated σ.

Sources