Fluid Dynamics & Turbulence — Navier-Stokes, Reynolds Number and the Energy Cascade

The Navier-Stokes Equations

All of classical fluid dynamics rests on two conservation laws applied to a continuum fluid: conservation of mass (continuity equation) and conservation of momentum (Newton’s second law for a fluid parcel). For an incompressible Newtonian fluid these give the Navier-Stokes equations:

The term ( u · ∇) u is the nonlinear convective acceleration — the origin of turbulence. It represents the stretching and tilting of fluid vorticity as faster fluid elements overtake slower ones. The viscous term μ∇²u (dynamic viscosity μ) smooths velocity gradients, dissipating kinetic energy as heat. Proving that smooth solutions to the 3D Navier-Stokes equations always exist is one of the seven Millennium Prize Problems (Clay Mathematics Institute, $1 million prize).

Physical interpretation

Each term has a direct physical meaning. The pressure gradient −∇p drives flow from high to low pressure (Poiseuille pipe flow, meteorological winds). Viscosity resists deformation: water has μ ≈ 10³ × 10−³ Pa·s at 20 °C; air has μ ≈ 1.8 × 10−&sup5; Pa·s. The density ρ multiplies inertia: heavier fluids need more force to accelerate.

Reynolds Number & the Laminar-Turbulent Transition

Whether a flow is smooth (laminar) or chaotic (turbulent) depends on the competition between inertial and viscous forces, quantified by the dimensionless Reynolds number:

For pipe flow (Hagen-Poiseuille): laminar for Re < 2300, transition 2300–4000, fully turbulent Re > 4000. For external flow over a flat plate: transition near Re ≈ 5 × 10&sup5;. A commercial aircraft cruises at Re ∼ 10&sup7;; a bacterium swims at Re ∼ 10−&sup4; (dominated by viscosity; inertia is negligible). At low Re, flow is reversible — G.I. Taylor’s famous 1966 ink-reversal experiment demonstrated this dramatically.

Transition mechanisms

Turbulence does not appear spontaneously. Transition is triggered by instabilities: Tollmien-Schlichting waves in boundary layers (linear instability, Re_δ ≈ 520), secondary instabilities, and bypass transition under real-world freestream turbulence. The Orr-Sommerfeld equation governs the linear stability of parallel shear flows and predicts critical Reynolds numbers for channel and pipe flows.

Bernoulli Equation & Applications

For steady, inviscid, incompressible flow along a streamline, the Bernoulli equation is a statement of energy conservation:

Applications are ubiquitous: Venturi flowmeters, Pitot tubes on aircraft, carburettors, chimney draught, the Coandã effect, and most critically aerodynamic lift. A wing (aerofoil) is shaped to accelerate air over the upper surface (shorter path through a pressure trough) more than the lower surface, creating a pressure difference Δp that produces lift L = Δp × area. Bernoulli strictly applies only along a streamline; the full story involves circulation Γ (Kutta-Joukowski theorem: L = ρ U Γ per unit span).

Pitfall: equal transit time fallacy

A common misconception claims air splits over a wing and must “meet again” at the trailing edge (implying upper air moves faster because the path is longer). This is false: air parcels do not meet. The real mechanism is the Kutta condition (smooth trailing-edge departure) combined with circulation enforced by the no-slip condition and viscosity.

Boundary Layers

The no-slip condition requires that the fluid velocity at a solid wall equals the wall velocity. This creates a thin zone — the boundary layer — where velocity transitions from zero (at the wall) to the freestream value U∞. The Blasius solution for a laminar boundary layer over a flat plate gives boundary layer thickness:

Within the turbulent boundary layer the velocity profile follows the log-law of the wall: u+ = (1/κ) ln y+ + B, with von Kármán constant κ ≈ 0.41 and B ≈ 5.0. The viscous sublayer (y+ < 5) is dominated by molecular viscosity; the log layer (30 < y+ < 300) and outer region dominate overall drag. Modern aircraft spend enormous engineering effort reducing the turbulent boundary layer’s contribution to skin friction drag, which accounts for roughly 50% of total drag.

Vortex Dynamics & the Von Kármán Vortex Street

Vorticity ω = ∇ × u measures the local rotation of fluid elements. Helmholtz’s theorems (1858) show that a vortex tube in an inviscid fluid convects with the flow, maintains its strength, and cannot end in the fluid. Kelvin’s circulation theorem: DΓ/Dt = 0 in an inviscid barotropic fluid.

Past a circular cylinder at moderate Re (47–200) the wake becomes periodic: vortices shed alternately from upper and lower sides, forming the von Kármán vortex street. The Strouhal number St = f d / U ≈ 0.2 governs shedding frequency f for cylinder diameter d. This vortex-induced vibration famously caused the 1940 Tacoma Narrows bridge collapse (though the full mechanism was aeroelastic flutter, not pure vortex resonance). Offshore oil risers and power lines are also vulnerable and require helical strakes to disrupt periodic shedding.

Turbulence & the Kolmogorov Energy Cascade

Fully turbulent flow contains eddies spanning a huge range of scales. Andrey Kolmogorov (1941) proposed that turbulence is self-similar in the inertial subrange: energy fed at large scales (L, integral scale) cascades down through eddy-splitting without dissipation until the Kolmogorov microscale η where viscosity finally converts kinetic energy to heat.

Here ε is the turbulent kinetic energy dissipation rate per unit mass (W/kg), k is the wavenumber. The −5/3 power law has been verified across atmospheric turbulence (10−1000 m scales), ocean currents, pipe flow, and even on other planets. The ratio L/η scales as Re^3/4: at Re ∼ 10&sup6; the range spans four decades of scale. This is why direct numerical simulation (DNS) of high-Re turbulence requires ∼Re^9/4 grid points and is computationally intractable above Re ∼ 10&sup4; even on today’s supercomputers.

Turbulence models

Because DNS is too expensive, engineering uses turbulence models. Reynolds-averaged Navier-Stokes (RANS): time-average the equations, yielding the Reynolds stress tensor −ρ<u′u′>, closed by eddy viscosity (k-ω, k-ε models). Large Eddy Simulation (LES): resolve eddies larger than a filter scale, model only sub-grid stresses (Smagorinsky model). Hybrid RANS/LES (DES): cheap near walls, accurate in separated regions. All models introduce free parameters requiring calibration against experiments.

Compressible Flow & Shock Waves

When flow speed approaches the local speed of sound the incompressibility assumption breaks down. The Mach number Ma = U/a (a = √(γRT) for an ideal gas) characterises the regime: Ma < 0.3 incompressible; 0.3–0.8 subsonic compressible; 0.8–1.2 transonic (with local supersonic pockets); 1.2–5 supersonic; >5 hypersonic.

At Ma > 1 a normal shock forms: flow decelerates abruptly, density and pressure jump discontinuously, and entropy increases (oblique shocks in 2D/3D). The Rankine-Hugoniot jump conditions relate upstream/downstream properties:

Key application: a supersonic aircraft must accelerate through the transonic sound barrier (drag coefficient peaks at Ma ≈ 1 due to wave drag) and operate supersonically where wave drag is manageable by swept wings and area ruling. The Prandtl-Glauert singularity (infinite lift at Ma = 1 in linearised theory) is a model artefact — real flows are continuous but non-linear.

Dimensional Analysis & Similarity

The Buckingham Π theorem states that a physically meaningful equation relating n dimensional quantities with k independent dimensions can be written in terms of n−k dimensionless groups (Π-parameters). Fluid dynamics uses this constantly: Re, Ma, Fr (Froude), We (Weber), Eu (Euler) and Strouhal numbers. Two flows are dynamically similar if all relevant dimensionless groups match — the basis for wind-tunnel testing (scale models at matched Re) and ship towing-tank tests (matched Fr).

Superfluids & Quantum Turbulence

At temperatures below the λ point (Tλ ≈ 2.17 K), liquid ⁴He becomes a superfluid (Bose-Einstein condensate). Its viscosity vanishes identically and flows without resistance. Vorticity is quantised: any circulation must be an integer multiple of κ = h/m (the quantum of circulation). Vortex cores are ∼1 Å thick and carry exactly one quantum κ. Quantum turbulence is a tangle of these quantised vortex filaments; at large scales it exhibits a Kolmogorov −5/3 spectrum, hinting that the classical cascade emerges from quantum vortex dynamics.

Simulations on This Platform

Our fluid dynamics category includes hands-on tools across the full Reynolds-number spectrum:

• Bernoulli Simulator — venturi meter, Pitot tube, aerofoil pressure
• Aerofoil & Lift — NACA profiles, stall angle, L/D polar
• Bath Waves — shallow-water wave dispersion and reflections
• Blood Flow — Poiseuille profile, stenosis, aneurysm
• Cloud Formation — moist convection, lapse rate instability
• Cyclone & Coriolis — geostrophic balance, Hadley cells
• Bénard Convection — Rayleigh-Bénard cells, Nusselt number
• Bubble Dynamics — Rayleigh-Plesset equation, cavitation