Rayleigh-Taylor & Kelvin-Helmholtz: Why Fluid Interfaces Roll Into Spirals

Two fluids, one unstable interface

Put a dense fluid on top of a light one and gravity does not politely wait — the interface between them curls into mushroom-shaped plumes almost immediately. This is the Rayleigh-Taylor instability: whenever a heavier fluid is accelerated into a lighter one (or, equivalently, sits above it in a gravitational field), any tiny ripple in the interface grows exponentially rather than damping out. The growth rate for a sinusoidal perturbation of wavenumber k is:

γ = √(A · g · k)
A = (ρ₂ - ρ₁) / (ρ₂ + ρ₁)     Atwood number

The Atwood number A measures the density contrast: A = 0 means identical fluids and no instability, A → 1 means one fluid is vastly denser than the other (like water over air) and growth is fast. Shorter wavelengths (larger k) grow faster in the idealised inviscid theory, which is why real fluids need viscosity and surface tension to cut off the smallest scales — otherwise the growth rate would diverge and the interface would shatter into infinitely fine structure instantly.

Kelvin-Helmholtz: instability from shear, not gravity

A second, related mechanism needs no density difference at all: Kelvin-Helmholtz instability appears whenever two fluid layers slide past each other with a velocity difference ΔU across the interface. The classic cat's-eye vortices you see in wind blowing over water, or in Jupiter's banded cloud tops, are Kelvin-Helmholtz rollers. With gravity and density contrast added, the two effects compete: the flow is unstable when

(ρ₁ρ₂)/(ρ₁+ρ₂)² · ΔU² k  >  A · g
        (shear, destabilising)      (gravity + density gradient, stabilising if heavy fluid is below)

When the heavy fluid sits on the bottom, gravity stabilises the interface and a strong enough shear is needed to overcome it — the threshold above is the standard two-layer Kelvin-Helmholtz criterion. When the heavy fluid sits on top, Rayleigh-Taylor growth dominates regardless of shear, and the interface is unstable at every wavelength in the ideal (inviscid, unbounded) limit.

Surface tension sets the smallest scale

Surface tension σ resists curvature and therefore suppresses growth at short wavelengths, giving Rayleigh-Taylor a critical wavenumber below which perturbations actually decay instead of growing:

k_c = √( (ρ₂-ρ₁) g / σ )
fastest-growing mode ≈ √3 · k_c  (viscous cutoff, Chandrasekhar 1961)

This is why a layer of dense fluid a few millimetres thick can rest stably on a lighter one — surface tension pins the interface flat until the layer gets thick enough for gravity to win. Below the critical wavelength, tension always wins; above it, gravity takes over and the growth rate rises toward the inviscid √(Agk) result before viscosity clips the very shortest, fastest-growing modes.

From linear growth to turbulent mixing

The formulas above describe only the linear regime — small perturbations growing exponentially while they are still small. Once the amplitude becomes comparable to the wavelength, the mushroom caps roll up, secondary Kelvin-Helmholtz instabilities form along their stems, and the flow cascades into fully turbulent mixing. This late-time behaviour is famously hard to predict analytically and is why Rayleigh-Taylor mixing widths are usually fit to an empirical law, h(t) ≈ α·A·g·t², with the mixing constant α determined experimentally or numerically rather than derived from first principles.

Where these instabilities actually show up

Rayleigh-Taylor governs the fingering of a supernova's ejecta as it decelerates into slower surrounding gas, the collapse of an inertial-confinement-fusion capsule (where it is the single biggest obstacle to ignition), and the plumes rising from a lava lamp or a thunderstorm's cold downdraft. Kelvin-Helmholtz shapes the braided clouds seen in a strong wind shear layer, the flutter of a flag, and the turbulent mixing layer at the edge of a jet engine's exhaust. Both instabilities are also textbook cases for testing computational fluid dynamics codes, precisely because their linear growth rate is known exactly and any simulation can be checked against it before trusting its turbulent, non-linear phase.

Frequently asked questions

Try it live

Everything above runs in your browser — open Fluid Instability and change the parameters while it is running. Nothing is installed, nothing is uploaded, the whole model lives in one tab.