Navier-Stokes Equations — The Mathematics of Fluid Flow

Newton's second law, written for a fluid

Treat a fluid as a continuous medium — valid at scales far larger than the ~70 nm mean free path of air molecules — and the incompressible Navier-Stokes equations are just Newton's second law applied to a fluid parcel: ρ(∂u/∂t + u·∇u) = −∇P + μ∇²u + f, paired with the continuity equation ∇·u = 0 stating that fluid is neither created nor destroyed. Every term has a clear job: local acceleration, the nonlinear advective term u·∇u that is the source of nearly all the complexity, pressure gradient force, viscous diffusion, and body forces like gravity. Combine the time derivative and advection into the material derivative D/Dt and the equation reads simply ρ Du/Dt = −∇P + μ∇²u + f — mass times acceleration equals net force.

The Reynolds number: one ratio explains chaos

In 1883, Osborne Reynolds injected dye into pipe flow and found a sharp transition from straight streamlines to chaotic mixing as speed increased. The controlling parameter, Re = ρUL/μ = UL/ν, is the ratio of inertial forces (which sustain motion and spin up vortices) to viscous forces (which damp disturbances). Below roughly Re ≈ 2300 pipe flow stays laminar; above about 4000 it's fully turbulent. Everyday examples span an enormous range: a swimming bacterium sits at Re ~ 10⁻⁴, where viscosity totally dominates and stopping is instantaneous; a commercial airliner wing operates near Re ~ 10⁷–10⁸; ocean currents reach Re ~ 10¹⁰.

Re = ρUL/μ = UL/ν
Kolmogorov microscale:  η = (ν³/ε)^(1/4)
Energy spectrum (inertial subrange):  E(k) ∝ ε^(2/3) · k^(−5/3)

The Kolmogorov cascade and boundary-layer separation

Turbulence looks random but has a precise statistical architecture. Andrei Kolmogorov's 1941 cascade theory pictures large eddies, injected at the integral scale, breaking down into progressively smaller ones, transferring energy downward without loss until they hit the Kolmogorov microscale η = (ν³/ε)^(1/4), where viscosity finally converts kinetic energy to heat. In the inertial subrange between those scales, the energy spectrum follows the famous −5/3 power law, confirmed everywhere from wind tunnels to solar wind data. Separately, at high Reynolds numbers viscosity's effect is confined to a thin boundary layer near solid surfaces (Ludwig Prandtl, 1904); when pressure rises in the flow direction, this layer can reverse and separate from the surface, creating a recirculation zone and a jump in drag — the mechanism behind wing stall and why golf balls have dimples, which trip the boundary layer early to delay separation.

A million-dollar open question

In 2000 the Clay Mathematics Institute named the Navier-Stokes existence and smoothness problem one of seven Millennium Prize Problems, each worth $1,000,000 — and it remains unsolved. The question: given smooth starting data, do smooth 3D solutions exist for all future time, or can the nonlinear advection term concentrate energy into arbitrarily fine structures faster than viscosity dissipates it, producing a finite-time blow-up? In two dimensions, global smoothness is proven; in three, it is genuinely open, even though local existence for a short time is known (Leray, 1934) and weaker "Leray-Hopf" solutions exist globally without guaranteed uniqueness or smoothness. In practice this doesn't stop engineers: Computational Fluid Dynamics (finite volume, finite difference, spectral and lattice-Boltzmann methods) solves the equations numerically every day, with turbulence modelled via DNS, Large Eddy Simulation, or the industrial workhorse RANS — the unsolved question is purely whether the exact equations, taken literally, could ever misbehave.

Frequently asked questions

Try it live

Everything above runs in your browser — open Real-Time Fluid Simulation, inject vortices, shear or plumes with a click, and watch Jos Stam's stable-fluids solver render the Navier-Stokes equations live. Nothing is installed, nothing is uploaded.