turbulence

Watch smoke rise from an extinguished match. For a moment it climbs in a smooth ribbon, then suddenly it shatters into a chaotic, swirling, unpredictable mess. That breakdown into roiling eddies of every size is turbulence. It is the flow you see in rivers, behind aircraft, in stirred coffee, and in the atmosphere — and it is famously called the last great unsolved problem of classical physics.

Turbulence is not a separate equation; it is a behaviour of the Navier-Stokes equations at high Reynolds number. When inertia overwhelms viscosity (large Re), small disturbances no longer get damped — they grow, interact through the nonlinear (u dot grad) u term, and produce a hierarchy of eddies. Energy injected at large scales cascades down through ever-smaller eddies until, at the tiniest (Kolmogorov) scale, viscosity finally dissipates it as heat. This energy cascade gives Kolmogorov's celebrated prediction that the energy spectrum falls off like k^(-5/3) over the inertial range of wavenumbers k. Turbulent flow is irregular, mixes intensely, has greatly increased effective drag and diffusion, and is deterministic-yet-unpredictable in the chaos sense: tiny differences in initial conditions blow up.

Turbulence matters enormously and is handled honestly but imperfectly. We cannot solve it analytically; direct numerical simulation resolving every eddy is astronomically expensive at real-world Re, so engineers use turbulence models (RANS, large-eddy simulation) that average or filter and then approximate the unresolved small scales. These are useful and tuned, but they are models with closure assumptions, not exact solutions. Turbulence is the honest frontier where the Navier-Stokes equations are correct but our ability to extract answers from them runs out.