How does turbulence work?

Stir cream into coffee and you are watching the problem: smooth flow shatters into eddies, the eddies spawn smaller eddies, and prediction collapses into statistics. Turbulence is everywhere ordinary, in rivers, wingtips, arteries, and stars, and it is often called the last great unsolved problem of classical physics. The embarrassment is precise: the governing equations have been written down since the 1840s. They simply refuse to be understood.

Why it matters

Most of the fluid motion in the universe is turbulent. Weather forecasts, aircraft drag, blood flow through a stenosed artery, fuel mixing in engines, the formation of stars from collapsing gas clouds: all run on turbulence, and all are handled today with approximations tuned to experiment rather than derived from first principles. A real theory would improve every one of them. There is also a pure-mathematics stake. Whether the Navier-Stokes equations always yield smooth, well-behaved solutions is one of the seven Millennium Prize problems, with a million dollars waiting on the answer.

What has been tried

Osborne Reynolds showed in 1883 that a single number, now bearing his name, predicts when smooth flow in a pipe goes turbulent, which organized the phenomenon without explaining it. Lewis Fry Richardson described the energy cascade in the 1920s, big whirls feeding little whirls, and in 1941 Kolmogorov turned that picture into quantitative predictions about how energy distributes across eddy sizes; those predictions hold up remarkably well, and remain the field’s crown jewel. Since then the main tools have been statistical theories that hit a wall called the closure problem, where every equation for one average drags in an unknown deeper average, and brute-force computer simulation, whose cost grows so violently with flow speed that fully resolving the air over a real aircraft wing remains out of reach.

Where the edge is

Turbulence is not mysterious the way dark matter is; nothing unknown is hiding in the fluid. The mystery is that deterministic equations produce motion whose statistics we can measure but not derive. Small corrections to Kolmogorov’s 1941 picture, driven by the flow’s intermittent violent bursts, are measured precisely and still not predicted from the equations.

What would count as an answer

A theory that derives the measured statistics of turbulence from the Navier-Stokes equations rather than fitting them, or a proof settling whether those equations can break down. Either would retire a scandal that has outlasted quantum mechanics, relativity, and everyone who called it temporary.