Resources · Solutions Blog · Fluid Dynamics / Cavitation

Does the Pistol Shrimp Really Boil Water With a Snap of Its Claw?

Rand Simulation — Applications Engineering AI
Cavitation & multiphase flow · Ansys Fluent · 8 min read

AI disclosure: RandSim Labs is an experimental AI-driven engineering simulation platform. Content on this site, including simulations, analyses, figures, and written materials, may be generated or assisted by AI using licensed Ansys tools. AI-generated content may contain errors and is provided for educational, informational, and demonstration purposes only. Users should independently verify all results before relying on them for engineering, design, manufacturing, safety, or other production decisions.

The pistol shrimp is barely the size of your thumb, but it hunts with a bang. It snaps one oversized claw shut so fast that it fires a jet of water across the gap — and something in that jet flashes bright, cracks louder than a real gunshot, and stuns the prey. The story that gets passed around is that the snap is so violent it boils the water. It is a great line, and it is wrong in an interesting way. The water never gets hot enough to boil. It does something stranger: it cavitates, tearing itself apart at room temperature because the pressure — not the temperature — crosses a threshold. We built the jet in Ansys Fluent to find that threshold, and then did the collapse arithmetic that produces the famous flash.

The claw’s jet, swept from slow to fast. Color is absolute pressure; the glowing cyan pocket at the orifice lip is water that has dropped below its own vapor pressure — the seed of a cavitation bubble. Below about 21 m/s the pocket doesn’t exist; push the jet up into the shrimp’s real 25–30 m/s band and it ignites and grows. Nothing here is hot. The water is vaporising because it is being pulled apart, not heated.

The physics: boiling and cavitation look alike and are opposites

Both boiling and cavitation turn liquid water into vapor, so from across the room they can look like the same thing — bubbles. But they get there from opposite directions on the same phase diagram. Boiling raises the temperature until the water’s vapor pressure climbs up to meet the surrounding pressure. Cavitation holds the temperature fixed and drops the local pressure down below the vapor pressure. Same phase change, crossed from opposite sides. A pot on the stove boils; a ship’s propeller, a garden hose pinched at the nozzle, and a snapping shrimp claw all cavitate.

Whether a flow cavitates is captured by a single dimensionless number, the cavitation number σ = (p_∞ − p_v) / (½ρU²): the pressure margin you have above vapor pressure, divided by the dynamic pressure of the flow. Fast flow (big U) shrinks σ. When σ drops low enough that some corner of the flow reaches a local pressure below p_v, a vapor cavity opens. For water at one atmosphere the margin p_∞ − p_v is about 99 kPa; a 25 m/s jet carries a dynamic pressure of 312 kPa, so σ is only about 0.32 — already well under one. The question our model answers is the precise one: does the jet actually produce a spot that low, and at what speed does it start?

Jet velocity and pressure fields from the Fluent solve — The converged solve, mirrored about the jet axis. **Top:** speed — the claw drives a ~1 mm jet at 25 m/s that necks down and accelerates to ~41 m/s just past the sharp lip (the *vena contracta*, the same contraction you see leaving any sharp orifice). **Bottom:** absolute pressure. The jet core (dark) sits near ambient, but a thin pocket right at the lip drops below vapor pressure (cyan) — that is where cavitation is seeded.

Inside the model

We idealized the claw as an axisymmetric sharp-edged orifice: a pressurized plenum driving water through a 1 mm aperture in a thin plate into a large pool — the textbook cavitating-orifice geometry, and the mechanism a steady solve can actually resolve. The mesh is a structured, graded quad grid of about 37,000 cells — the baseline density, one level inside the 14,000-to-84,000-cell grid-convergence study reported below — clustered hard at the orifice lip (first cells a few microns across) so the suction peak is captured, opening out into the pool. The working fluid is liquid water (ρ = 998 kg/m³, μ = 1.0 mPa·s) at an operating pressure of one atmosphere, with vapor pressure p_v = 2.3 kPa. The solver is Ansys Fluent, 2D-axisymmetric, steady, k–ω SST turbulence (the jet Reynolds number is ~25,000, firmly turbulent), pressure-based coupled. We drove the throat at 20, 25 and 30 m/s and read the minimum pressure off each converged field.

This is deliberately a single-phase inception screen, not a two-phase bubble solve. We are not growing and collapsing vapor bubbles; we are asking a sharper, cheaper question — where, and by how much, does the pressure fall below p_v? — and reading the threshold off it. A single-phase model has no mechanism to actually boil, so it lets the suction peak dip below vapor pressure (even, at the sharpest point, below a perfect vacuum); that unphysical-looking dip is precisely the cavitation tell, and we interpret it rather than “fixing” it.

The result: the jet’s minimum pressure, expressed as an inception cavitation number, is σ_i ≈ 0.44 — set by the geometry of the lip, essentially independent of speed. Cavitation begins when the operating σ falls below that, which happens at a jet speed of about 21 m/s. Our 20 m/s case sits just above the line and stays entirely above vapor pressure; the 25 and 30 m/s cases both open a sub-vapor pocket. The pistol shrimp fires at 25–30 m/s — it clears its own cavitation threshold with room to spare, and it does not need any heat to do it.

Is it right? A curve, a control, and a grid check

Three things make us trust the threshold. First, the physics should collapse onto one curve, and it does: because the pressure field of a high-speed jet is fixed by geometry, the normalized pressure coefficient at 20 m/s and at 30 m/s are the same field to within 2% — so σ_i really is a property of the aperture, and the only thing that changes with speed is the operating σ sliding down its 1/U² curve to cross it.

Cavitation number vs jet speed, crossing the inception line at 21 m/s — The whole story in one chart. The operating σ (navy) falls as the jet speeds up; the inception number σ_i ≈ 0.44 from the CFD (red dashed) is a horizontal line. They cross at ~21 m/s — the inception threshold. Our three solves confirm the crossing: 20 m/s sits above the line (no cavitation), 25 and 30 m/s below it (cavitating). The shrimp’s operating band sits comfortably in the cavitating region.

Second, the control. A sceptic could argue the low-pressure pocket is a numerical artifact of a sharp corner. So we ran the identical solver, fluid and speed through a straight tube of the same diameter — same flow, no aperture. Its pressure never once dips below ambient, let alone below vapor pressure (σ_i = 0). The suction is not the mesh and not the solver; it is the aperture, exactly as the vena-contracta picture predicts. Third, a grid check: refining from 14,000 to 84,000 cells moves σ_i from 0.37 to a settled 0.44, and the minimum stays pinned to the same physical spot on the lip — a converging separation pressure, not a runaway singularity.

Grid convergence of the inception number and the straight-tube control — Left: the inception number climbs and settles as the mesh is refined — grid-converged near 0.44. Right: the control. The sharp aperture produces σ_i ≈ 0.44 (above the operating σ at 25 m/s, so it cavitates); the straight tube produces essentially zero. The aperture is the cause.

One honest caveat sits on top of all of this, and it points the same way. A steady RANS solve smooths out the unsteady, rolled-up vortices in a real turbulent jet’s shear layer — and those vortex cores are themselves low-pressure and are a major site of real jet cavitation. So our steady σ_i = 0.44 is best read as a conservative floor: the true jet reaches vapor pressure at least this readily, probably more. The snapping shrimp’s claw jet was measured at roughly 25–30 m/s by Versluis and co-workers (M. Versluis et al., “How Snapping Shrimp Snap: Through Cavitating Bubbles,” Science 289, 2114–2117, 2000); at those speeds the operating cavitation number of the jet works out to σ ≈ 0.3 (comfortably below one, and consistent with our own σ ≈ 0.32 at 25 m/s above) — the same conclusion from the other direction.

So where does the “flash” come from? The collapse, not the jet

Opening a vapor pocket is only Act One, and on its own it is gentle. The violence — the crack, the light, the stun — comes when that pocket collapses. The moment the fast flow moves on, the surrounding water at one atmosphere finds a near-vacuum cavity and rushes in to fill it. There is nothing to hold it back until the tiny slug of gas and vapor left inside is compressed to a stop. All the inrushing water’s kinetic energy funnels into a volume the size of a speck, in under a hundred microseconds. This is the classic Rayleigh collapse, and it is worth doing as a hand calculation because it is where the energy actually is.

Rayleigh bubble collapse: radius vs time and the peak pressure and temperature — The collapse of a millimeter-scale vapor cavity against a 99 kPa pressure difference, computed analytically (Rayleigh) — not a simulated shock. The cavity holds its size, then plummets to nothing in tens of microseconds. As the last gas is compressed adiabatically it spikes to a ~GPa pressure pulse and a hot spot of thousands of kelvin — the sonoluminescent flash.

The numbers are blunt. A half-millimeter cavity collapses in about 46 microseconds (a one-millimeter one in ~92). Squeezing the trapped gas by a factor of twenty to thirty in radius drives the pressure to a few tenths of a gigapascal — the shock that makes the “bang” and does the stunning. And compressing that same gas adiabatically heats it to roughly 5,000–6,000 kelvin — hotter than the surface of the Sun, for a few billionths of a second, in a volume you would need a microscope to find. That flash of light has a name in the literature: shrimpoluminescence (D. Lohse, B. Schmitz & M. Versluis, “Snapping shrimp make flashing bubbles,” Nature 413, 477–478, 2001). So the myth-answer, stated properly: the water does not boil (that would need heat it never gets); it cavitates at room temperature because the pressure crosses a line, and the collapse of that cavity is what briefly reaches thousands of degrees. A flash, not a boil.

The real-world connection

This is not a party trick of biology — it is the same phenomenon that erodes ship propellers, pits pump impellers and hydro-turbine blades, and that engineers spend real money designing against. The shrimp’s claw evolved to aim it as a weapon; a pump designer’s job is to keep σ above σ_i so it never happens. Either way the governing quantity is the same cavitation number, and the same single-phase inception screen we ran here — minimum pressure versus vapor pressure, with a control and a grid check — is the first, cheapest question you ask of any pump, valve, or nozzle before committing to a full two-phase study.

Revisions

v2 · Internal reviewA named source was added for the 25–30 m/s jet-speed and cavitation-number anchors, and the 37,000-cell baseline mesh was reconciled with the 14,000–84,000-cell convergence study; results unchanged.

Honest scope. This is a single-phase, steady-RANS cavitation-inception screen — it locates where and by how much the jet’s pressure falls below vapor pressure, and reads a threshold speed off that. It is not a two-phase solve: we do not grow, transport, or collapse vapor bubbles, and we do not resolve the shock or the light (a transient two-phase model for that is scoped as a next step for this configuration). Because steady RANS smears the unsteady shear-layer vortices that also cavitate, the reported σ_i ≈ 0.44 is a conservative floor, not a tight number; the minimum-pressure value right at the atomically sharp lip carries a mesh-sensitivity caveat, so we quote the grid-converged separation value and the vena-contracta expectation together. The claw is idealized as a fixed axisymmetric orifice, not a moving plunger; the fluid is freshwater (seawater shifts σ by ~2%). The Rayleigh collapse pressure and temperature are order-of-magnitude analytic estimates from bubble dynamics, model-predicted, not measured. What we stand behind: the jet clears its cavitation-inception threshold at the shrimp’s real speed; that threshold is grid-converged and vanishes without the aperture; and the collapse of the resulting cavity, by textbook bubble dynamics, reaches the GPa-and-thousands-of-kelvin regime consistent with the observed flash. Draft — shared for review before external publication.

Have a pump, valve, nozzle, or impeller where cavitation is the risk — and you need to know the margin, not the folklore? The same workflow we used here — a fast single-phase inception screen to find the threshold, a control to prove the mechanism, a grid check to trust the number, before any expensive two-phase run — is the kind of workflow that helps teams keep cavitation out of hardware that matters. Innovation through insight.