Taylor-Green Vortex Decay: Turbulence's Favorite Benchmark

A rare exact solution to a notoriously hard equation

The Navier-Stokes equations governing fluid flow almost never have a clean, closed-form solution once viscosity and nonlinearity are both in play - which is exactly why turbulence remains one of the great unsolved problems of classical physics. The Taylor-Green vortex, published by G. I. Taylor and A. E. Green in 1937, is a rare and treasured exception: a fully explicit analytic solution of the incompressible Navier-Stokes equations in two dimensions, describing an array of counter-rotating vortices that decay smoothly under viscosity alone.

The 2D solution

The classic two-dimensional Taylor-Green flow starts from a simple sinusoidal velocity field and decays exponentially in time, with the exact time dependence coming directly from substituting the ansatz into the Navier-Stokes equations and confirming it satisfies them exactly, viscous term included:

u(x,y,t) =  cos(x) sin(y) * exp(-2*nu*t)
v(x,y,t) = -sin(x) cos(y) * exp(-2*nu*t)
p(x,y,t) = -(1/4)(cos(2x) + cos(2y)) * exp(-4*nu*t)
// nu = kinematic viscosity; this satisfies incompressible Navier-Stokes EXACTLY at every t

Everything decays with the same simple exponential: the velocity amplitude falls as exp(-2*nu*t) and the kinetic energy, proportional to velocity squared, falls as exp(-4*nu*t). No turbulence, no cascade - just an orderly array of vortices spinning down as viscosity dissipates their kinetic energy into heat, forever preserving the same spatial pattern. It is exactly this simplicity that makes it valuable: any numerical solver can be pointed at the Taylor-Green initial condition and its output compared, point for point and digit for digit, against the known analytic answer.

The 3D version: a turbulence benchmark

The three-dimensional Taylor-Green vortex uses a similar sinusoidal initial condition but with no analytic solution - and that absence of an exact answer is precisely what makes the 3D case famous. Starting from smooth, large-scale sinusoidal vortices, the flow's nonlinear term stretches and tilts vortex lines, breaking the big, orderly structures into progressively smaller ones through vortex stretching, the fundamental mechanism by which three-dimensional turbulence transfers energy from large scales to small ones (a mechanism that simply does not exist in two dimensions, which is why the 2D case above stays laminar and orderly).

Two diagnostics track this transition. Kinetic energy decreases monotonically and smoothly throughout, dissipated ultimately by viscosity. Enstrophy - the volume integral of vorticity squared, a direct measure of how much rotational, small-scale structure is present - does the opposite: it grows as the flow develops finer and finer vortical structures, reaches a sharp peak around a dimensionless time of about 9 (for the standard initial Reynolds number used in benchmark studies), and only then starts to fall as viscous dissipation, which scales with enstrophy, finally outpaces the cascade's production of new small scales.

Why every CFD code gets tested on it

The 3D Taylor-Green vortex has become the de facto standard first benchmark for any new computational fluid dynamics (CFD) solver claiming to handle transitional or turbulent flow, because it offers something rare: a fully specified, reproducible, geometry-free initial condition (a periodic cube, no walls, no inflow/outflow boundary conditions to get wrong) that nonetheless develops the full physics of vortex stretching, transition to turbulence, and an energy cascade down to a direct numerical simulation (DNS)'s smallest resolved scales. Because the setup is so clean, results from different codes and different numerical methods - finite volume, spectral, lattice Boltzmann - can be compared against each other and against high-resolution reference DNS data almost point by point, isolating numerical-method errors from any ambiguity in boundary conditions or geometry.

What the peak enstrophy time actually measures

A solver that gets the enstrophy-peak time and magnitude right is correctly resolving the point at which the flow is generating the smallest, most demanding turbulent structures it will ever produce in this benchmark - a strong indication that the numerical scheme's dissipation and dispersion errors are small enough not to distort the energy cascade. A solver that peaks too early, too late, or at the wrong magnitude is telling you, quantitatively, how much numerical error its discretization scheme, grid resolution, or time-stepping is injecting into a genuinely difficult nonlinear regime - which is exactly the information a CFD developer needs before trusting the same code on a harder, real-world turbulent flow.

Frequently asked questions

Try it live

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