Vorticity, Circulation, and Potential Flow

Take the velocity gradient of a flowing fluid and split it, exactly as for the displacement gradient of a deforming solid, into a symmetric rate-of-strain part and an antisymmetric rotation. The antisymmetric half is the vorticity field $\boldsymbol{\omega}=\nabla\times\mathbf{v}$ — and unlike its bookkeeping role in the Strain Tensor article, here it is a full dynamical actor. Vorticity is generated, transported, stretched, and (in viscous fluids) diffused; its behavior organizes everything rotational in fluid mechanics, from smoke rings to tornadoes to the lift on a wing. Its integral partner is circulation, the velocity summed around a closed material loop. Stokes' theorem makes circulation and vorticity flux two views of one quantity, but circulation is the more robust object: it survives even when vorticity collapses into singular lines. The central dynamical result is Kelvin's circulation theorem — in an ideal fluid, the circulation around any loop moving with the material never changes. Vortex lines are strings tied into the fluid: they can be stretched and bent, never cut. That single conservation law explains both why smoke rings persist and how a wing acquires the circulation that lifts it. The converse regime is equally powerful. A flow that starts irrotational stays irrotational, and an irrotational velocity field is a gradient, $\mathbf{v}=\nabla\phi$ . For incompressible flow $\phi$ obeys Laplace's equation — the whole nonlinear vector problem collapses to the linear theory of harmonic functions, with complex analysis handing out exact two-dimensional flows. This article builds the kinematics, proves Kelvin's theorem, and assembles potential-flow theory: the ideal-flow half of the story whose viscous completion is the Navier–Stokes article.

Key takeaways

Vorticity $\boldsymbol{\omega}=\nabla\times\mathbf{v}$ is twice the local angular velocity of a fluid element — it measures spin embedded in the field, not curvature of streamlines.
Circulation $\Gamma=\oint_C\mathbf{v}\cdot\mathrm{d}\boldsymbol{\ell}$ equals, by Stokes' theorem, the flux of vorticity through any surface spanning the loop.
Kelvin's circulation theorem: for inviscid, barotropic flow with conservative body forces, $\mathrm{D}\Gamma/\mathrm{D}t=0$ on material loops — vortex lines are frozen into the fluid and tube strength is conserved (Helmholtz).
Viscosity and baroclinicity ( $\nabla\rho\times\nabla p\neq0$ ) are the two great vorticity factories; without them an irrotational flow stays irrotational forever.
Irrotational flow has $\mathbf{v}=\nabla\phi$ ; adding incompressibility gives $\nabla^2\phi=0$ — linear, electrostatics-grade mathematics, with Bernoulli supplying the pressure and the Kutta–Joukowski theorem the lift $L=\rho U\Gamma$ .
Key idea: vorticity is the curl of velocity and equals twice the local spin; its flux is circulation, which ideal flow conserves on material loops (Kelvin) — so irrotational flow stays a gradient $\mathbf{v}=\nabla\phi$ and, if incompressible, solves $\nabla^2\phi=0$ .

The core insight

Vorticity is the curl of velocity and equals twice the local angular velocity of a fluid blob; its flux is circulation. In an ideal fluid circulation is materially conserved (Kelvin), so an initially irrotational flow stays a gradient $\mathbf{v}=\nabla\phi$ and — if incompressible — solves $\nabla^2\phi=0$ .

Where this lives today

Wake-vortex physics is air-traffic law: the counter-rotating trailing vortices required by Kelvin's ledger persist for minutes behind heavy aircraft, and ICAO separation minima — recently refined by the RECAT programs into finer weight categories — exist because a following aircraft that flies into a predecessor's $\Gamma$ can be rolled beyond its control authority. Airport arrival rates worldwide are set, in the end, by how fast trailing circulation decays. The same vortex-lattice and panel methods that discretize bound circulation across a wing remain the first design pass for every new airframe, including the current wave of electric-aviation and blended-wing prototypes, before Navier–Stokes solvers refine the answer. At the other end of the scale, circulation is literally quantized: in superfluid helium and Bose–Einstein condensates the "potential" is the phase of a macroscopic wavefunction, forcing $\Gamma=nh/m$ — Derivation C's multivalued potential with quantum mechanics choosing the winding number. Imaging and steering lattices of such quantized vortices is now routine in cold-atom laboratories, and tracking vortex-line reconnections there tests the one process Kelvin's theorem forbids ideal fluids, the process by which quantum turbulence decays.

The physical picture

Plant a tiny paddlewheel in the flow and let it drift. If it spins, the flow has vorticity there, and its angular velocity is exactly $\tfrac12\boldsymbol{\omega}$ . The wheel responds to the antisymmetric part of the velocity differences across it — to shear and swirl, never to whether its path happens to curve. That distinction defeats almost everyone's first intuition, so look at the three flows below before believing anything else about vorticity.

Figure

The paddlewheel test in three flows — solid-body rotation spins the wheel everywhere, the line vortex carries it in circles without spinning it, and straight parallel shear spins it despite dead-straight streamlines

Circulation is the loop-integrated "tendency to swirl" of the fluid threading a closed curve, and Stokes' theorem says swirl measured on the boundary equals total vorticity crossing any membrane spanning it — vorticity is a flux density and circulation its flux, exactly as magnetic field stands to magnetic flux. Kelvin's theorem is best pictured with vortex lines as physical strings knotted into the fluid. Ideal flow drags them along, stretches them — which spins them faster, a skater pulling in her arms — but can never cut or reconnect them. Viscosity and baroclinic torques are the only scissors. This freezing-in is how a wing gets its lift: as the wing starts moving, it sheds a starting vortex, and Kelvin's ledger forces an equal-and-opposite bound circulation onto the wing itself.

Figure

Kelvin bookkeeping at takeoff — the big material loop starts with zero circulation and must keep it, so the starting vortex shed from the trailing edge is exactly balanced by the bound circulation that lifts the wing

Potential flow is the regime with no vortex strings at all: $\mathbf{v}=\nabla\phi$ , all dynamics collapsed into one linear scalar equation with boundary conditions — the same mathematics as electrostatics, bought entirely by the no-vorticity assumption.

The core equations

Start with pure kinematics. Near any point, expand the velocity field to first order; the gradient tensor splits into symmetric and antisymmetric parts, precisely as the displacement gradient did for solids:

\partial_j v_i = \underbrace{\tfrac12(\partial_j v_i+\partial_i v_j)}_{e_{ij}\ \text{(rate of strain)}} \;+\; \underbrace{\tfrac12(\partial_j v_i-\partial_i v_j)}_{-\tfrac12\epsilon_{ijk}\omega_k\ \text{(rotation)}} \qquad (1)

where: $e_{ij}$ — rate-of-strain tensor, which stretches and shears parcels; $\omega_k$ — vorticity components; $\epsilon_{ijk}$ — the Levi-Civita symbol. A fluid element simultaneously translates, distorts at rate $e_{ij}$ , and rotates rigidly at rate $\tfrac12\boldsymbol{\omega}$ — Derivation A extracts that last statement cleanly. The viscous stress of the next article will be built from $e_{ij}$ ; everything rotational lives in $\boldsymbol{\omega}$ . The rotational content gets its own field. Define

\boldsymbol{\omega}\equiv\nabla\times\mathbf{v} \qquad (2)

where: $\boldsymbol{\omega}$ — the vorticity, equal to twice the local angular velocity of the fluid element. It is a local quantity: solid-body rotation has $\boldsymbol{\omega}=2\boldsymbol{\Omega}$ everywhere, straight shear flow $u(y)\hat{\mathbf{x}}$ carries $\boldsymbol{\omega}=-u'(y)\hat{\mathbf{z}}$ , and the swirling line vortex has $\boldsymbol{\omega}=0$ almost everywhere. The paddlewheel figure is equation (2) made visible. Integrate the swirl around a closed loop and you get the circulation, which Stokes' theorem equates to a vorticity flux:

\Gamma=\oint_C \mathbf{v}\cdot\mathrm{d}\boldsymbol{\ell}=\int_S \boldsymbol{\omega}\cdot\mathrm{d}\mathbf{S} \qquad (3)

where: $C$ — a closed loop; $S$ — any surface spanning it; $\mathrm{d}\mathbf{S}$ — the oriented area element. Circulation is the more durable object: for a line vortex all the vorticity hides in a singular filament, yet $\Gamma$ around it remains finite and measurable (Derivation C). The magnetostatic analogy is exact — vorticity plays magnetic field, circulation plays the enclosed current. Dynamics enters through Euler's equation from the previous article, $\rho\,\mathrm{D}\mathbf{v}/\mathrm{D}t=-\nabla p-\rho\nabla\Phi$ with a conservative body-force potential $\Phi$ . Taking its curl kills both gradients and leaves an evolution law for vorticity alone (the viscous term is included here for later reference):

\frac{\mathrm{D}\boldsymbol{\omega}}{\mathrm{D}t}=(\boldsymbol{\omega}\cdot\nabla)\mathbf{v}-\boldsymbol{\omega}(\nabla\cdot\mathbf{v})+\frac{1}{\rho^2}\nabla\rho\times\nabla p+\nu\nabla^2\boldsymbol{\omega} \qquad (4)

where: $(\boldsymbol{\omega}\cdot\nabla)\mathbf{v}$ — vortex stretching and tilting, the skater effect that intensifies stretched tubes; $\boldsymbol{\omega}(\nabla\cdot\mathbf{v})$ — dilatation, active only in compressible flow; $\rho^{-2}\nabla\rho\times\nabla p$ — the baroclinic source, generating vorticity wherever isobars and isopycnals misalign (sea breezes are born here); $\nu\nabla^2\boldsymbol{\omega}$ — viscous diffusion, zero for ideal fluid. Strike the last two terms in ideal barotropic flow and vorticity is purely advected and stretched: no source, no sink — the differential shadow of Kelvin's theorem, which Derivation B proves in integral form. When there is no vorticity anywhere and the domain is simply connected, the velocity is a pure gradient:

\boldsymbol{\omega}=0\;\Rightarrow\;\mathbf{v}=\nabla\phi \qquad (5)

where: $\phi$ — the velocity potential. Add incompressibility $\nabla\cdot\mathbf{v}=0$ from the kinematics article, and the potential is harmonic:

\nabla^2\phi=0 \qquad (6)

where: $\nabla^2$ — the Laplacian. This is the entire dynamics of incompressible irrotational flow: a linear boundary-value problem ( $\partial\phi/\partial n=0$ on rigid surfaces, prescribed flow at infinity), with the pressure recovered afterwards from the unsteady Bernoulli relation $\partial_t\phi+\tfrac12v^2+p/\rho+\Phi=f(t)$ of the previous article. Nonlinearity has not vanished — it has been quarantined inside Bernoulli's quadratic term. In two dimensions the machinery becomes complex analysis. Pair the potential with the stream function $\psi$ of the kinematics article:

w(z)=\phi(x,y)+\mathrm{i}\,\psi(x,y),\qquad \frac{\mathrm{d}w}{\mathrm{d}z}=v_x-\mathrm{i}v_y \qquad (7)

where: $w$ — the complex potential, an analytic function of $z=x+\mathrm{i}y$ ; $\phi,\psi$ — harmonic conjugates obeying the Cauchy–Riemann equations, their level curves orthogonal. Every analytic function is an exact flow: $w=Uz$ is a uniform stream, $w=(\Gamma/2\pi\mathrm{i})\ln z$ the line vortex, $w=U(z+a^2/z)$ flow past a cylinder — and conformal maps carry these onto airfoils, which is how the first wings were designed.

Derivation AVorticity as twice the local angular velocity

We make equation (2)'s physical reading exact: the antisymmetric part of the velocity gradient rotates fluid elements rigidly at angular velocity $\tfrac12\boldsymbol{\omega}$ . The strategy is a first-order Taylor expansion of the velocity field about a point, followed by rewriting the antisymmetric contribution as a cross product. This three-line calculation is what licenses every "paddlewheel" statement in the subject, so it pays to see the indices work once.

Step 1. Expand the velocity at a nearby point $\mathbf{x}+\delta\mathbf{x}$ :

v_i(\mathbf{x}+\delta\mathbf{x})=v_i(\mathbf{x})+\partial_j v_i\,\delta x_j+\mathcal{O}(\delta x^2)

Step 2. Split $\partial_j v_i$ as in equation (1). The symmetric part $e_{ij}\,\delta x_j$ strains the element; isolate the antisymmetric part and rewrite it with the vorticity, using $\tfrac12(\partial_j v_i-\partial_i v_j)=-\tfrac12\epsilon_{ijk}\omega_k$ :

\left[\tfrac12(\partial_j v_i-\partial_i v_j)\right]\delta x_j=-\tfrac12\epsilon_{ijk}\,\omega_k\,\delta x_j=\tfrac12\left(\boldsymbol{\omega}\times\delta\mathbf{x}\right)_i

Why the identity holds: an antisymmetric 3×3 matrix has three independent entries, exactly a vector's worth, and $\epsilon_{ijk}$ is the canonical dictionary between the two.

Step 3. Compare with rigid rotation. A rigid body spinning at $\boldsymbol{\Omega}$ has relative velocity $\boldsymbol{\Omega}\times\delta\mathbf{x}$ between points separated by $\delta\mathbf{x}$ . Matching,

\boldsymbol{\Omega}_{\text{local}}=\tfrac12\boldsymbol{\omega} \qquad\Longleftrightarrow\qquad \boldsymbol{\omega}=2\,\boldsymbol{\Omega}_{\text{local}}

Why the paddlewheel measures this: a small rigid cross averages the surrounding fluid's velocity differences; the strain part pulls its arms symmetrically and produces no net torque about the center, while the rotation part turns it bodily at $\tfrac12\boldsymbol{\omega}$ . *Verified: for solid-body rotation $\mathbf{v}=\boldsymbol{\Omega}\times\mathbf{r}$ , direct curl gives $\nabla\times(\boldsymbol{\Omega}\times\mathbf{r})=2\boldsymbol{\Omega}$ , matching the general result; for shear flow $u(y)\hat{\mathbf{x}}$ it gives $\boldsymbol{\omega}=-u'\hat{\mathbf{z}}$ , a wheel spinning clockwise when the upper fluid moves faster. Matches Batchelor §2.3.*

Go deeper · Derivation B · 1 minKelvin circulation theorem

Derivation BKelvin circulation theorem

We prove that ideal flow conserves the circulation around every material loop — the master conservation law of vortex dynamics. The strategy: differentiate $\Gamma=\oint\mathbf{v}\cdot\mathrm{d}\boldsymbol{\ell}$ following the fluid, which produces two terms (the velocity changes, and the loop itself is carried and stretched); show the stretching term integrates to zero exactly, then let Euler's equation reduce the rest to the closed-loop integral of a gradient. Watch which hypothesis retires which term — inviscid, barotropic, conservative forcing each earn their keep, and each names a way real fluids evade the theorem.

Step 1. Differentiate the loop integral materially, product rule on both factors:

\frac{\mathrm{D}\Gamma}{\mathrm{D}t}=\oint_{C(t)} \frac{\mathrm{D}\mathbf{v}}{\mathrm{D}t}\cdot\mathrm{d}\boldsymbol{\ell}+\oint_{C(t)} \mathbf{v}\cdot\frac{\mathrm{D}(\mathrm{d}\boldsymbol{\ell})}{\mathrm{D}t}

Step 2. Dispose of the stretching term. A material line element evolves by the velocity difference across it, $\mathrm{D}(\mathrm{d}\boldsymbol{\ell})/\mathrm{D}t=(\mathrm{d}\boldsymbol{\ell}\cdot\nabla)\mathbf{v}=\mathrm{d}\mathbf{v}$ , so

\oint_C \mathbf{v}\cdot\mathrm{d}\mathbf{v}=\oint_C \mathrm{d}\!\left(\tfrac12 v^2\right)=0

Why zero: the integrand is an exact differential of the single-valued function $\tfrac12v^2$ , and any exact differential integrates to zero around a closed loop.

Step 3. Insert Euler's equation with conservative body force, $\mathrm{D}\mathbf{v}/\mathrm{D}t=-\rho^{-1}\nabla p-\nabla\Phi$ :

\frac{\mathrm{D}\Gamma}{\mathrm{D}t}=-\oint_C\left(\frac{1}{\rho}\nabla p+\nabla\Phi\right)\cdot\mathrm{d}\boldsymbol{\ell}

Step 4. Spend the barotropic hypothesis. If $\rho=\rho(p)$ , then $\rho^{-1}\nabla p=\nabla\mathcal{P}$ with $\mathcal{P}=\int\mathrm{d}p/\rho$ a single-valued function of position:

\frac{\mathrm{D}\Gamma}{\mathrm{D}t}=-\oint_C \nabla(\mathcal{P}+\Phi)\cdot\mathrm{d}\boldsymbol{\ell}=0

Why barotropy is essential: if surfaces of constant $p$ and constant $\rho$ misalign, $\rho^{-1}\nabla p$ is not a gradient and its loop integral survives — that residue is precisely the baroclinic source of equation (4), the term that manufactures sea breezes and ocean eddies.

Step 5. Conclude $\mathrm{D}\Gamma/\mathrm{D}t=0$ for every material loop. Helmholtz's vortex theorems follow at once: applying the result to small loops around a vortex tube shows tube strength is constant along its length and in time, and vortex lines move with the fluid — they cannot end in the interior, and in ideal flow they are never created or destroyed. Verified: the theorem implies an initially irrotational flow ( $\Gamma=0$ on all contractible loops) stays irrotational, consistent with the vorticity equation (4) whose ideal barotropic form has no source term; matches Landau–Lifshitz Vol. 6 §8 and Batchelor §5.1.

Go deeper · Derivation C · 1 minThe line vortex and its radius-independent circulation

Derivation CThe line vortex and its radius-independent circulation

We analyze the fundamental singular solution of vortex dynamics — the flow that swirls in circles yet carries zero vorticity — and expose why potential flow in a non-simply-connected domain can still hold circulation. The strategy: posit the azimuthal $1/r$ velocity field, compute its vorticity and circulation directly in cylindrical coordinates, and then examine the multivalued potential that makes it all consistent. This one solution is the hydrogen atom of the subject: wings, tornado far-fields, and quantized superfluid vortices are all built on it.

Step 1. Take the planar azimuthal field

\mathbf{v}=\frac{\Gamma}{2\pi r}\,\hat{\boldsymbol{\theta}}

Step 2. Compute the vorticity off the axis using the cylindrical curl:

\omega_z=\frac{1}{r}\frac{\partial}{\partial r}\left(r\,v_\theta\right)=\frac{1}{r}\frac{\partial}{\partial r}\left(\frac{\Gamma}{2\pi}\right)=0\qquad(r>0)

Why zero despite circular streamlines: the $1/r$ profile is exactly tuned so a paddlewheel's outer arm, moving on the slower larger circle, is compensated by its inner arm on the faster smaller circle — it orbits without turning, like a gondola on a Ferris wheel.

Step 3. Compute the circulation around a circle of any radius $r$ :

\oint\mathbf{v}\cdot\mathrm{d}\boldsymbol{\ell}=v_\theta\cdot 2\pi r=\frac{\Gamma}{2\pi r}\,2\pi r=\Gamma

Why radius-independent: by Stokes' theorem the circulation counts enclosed vorticity, and all of it sits in a delta-function filament on the axis, $\boldsymbol{\omega}=\Gamma\,\delta^2(\mathbf{r})\,\hat{\mathbf{z}}$ — the exact analogue of the magnetic field around a current-carrying wire.

Step 4. Exhibit the potential. $\mathbf{v}=\nabla\phi$ with

\phi=\frac{\Gamma}{2\pi}\,\theta

which increases by $\Gamma$ on each circuit of the axis. On the punctured plane — not simply connected — a locally valid potential may be globally multivalued, so "irrotational" fails to force "zero circulation." The loophole is load-bearing: it is precisely what lets a wing in irrotational flow carry the bound circulation that lifts it. Verified: dimensional check $[\Gamma]=\mathrm{m^2/s}$ throughout; superposing this solution with a uniform stream and evaluating the Kutta–Joukowski integral reproduces $L=\rho U\Gamma$ ; matches Batchelor §2.5 and Landau–Lifshitz Vol. 6 §10.

Limiting cases

Case	Result	Meaning
Solid-body rotation	$\mathbf{v}=\boldsymbol{\Omega}\times\mathbf{r}\Rightarrow\boldsymbol{\omega}=2\boldsymbol{\Omega}$	Maximally rotational; circulation grows with enclosed area, $\Gamma=2\Omega\,\pi r^2$ ; no potential exists.
Line (irrotational) vortex	$v_\theta=\Gamma/2\pi r$ , $\boldsymbol{\omega}=0$ for $r>0$	Circular streamlines, zero vorticity; the same $\Gamma$ threads every enclosing loop — all spin hides in the singular axis.
Parallel shear	$\mathbf{v}=u(y)\hat{\mathbf{x}}\Rightarrow\boldsymbol{\omega}=-u'(y)\hat{\mathbf{z}}$	Straight streamlines with nonzero vorticity — the cleanest refutation of the curvature intuition.
Ideal barotropic flow	$\mathrm{D}\boldsymbol{\omega}/\mathrm{D}t=(\boldsymbol{\omega}\cdot\nabla)\mathbf{v}$	No sources: vorticity only advects, tilts, and stretches — Kelvin/Helmholtz frozen-in dynamics.
Irrotational and incompressible	$\nabla^2\phi=0$	Potential flow: linear, electrostatics-grade; pressure from Bernoulli afterwards.
Two dimensions	$(\boldsymbol{\omega}\cdot\nabla)\mathbf{v}=0$	No stretching in 2D — vorticity is materially conserved scalar; 2D turbulence cascades energy upscale, opposite to 3D.

Common trapVorticity is not curved streamlines — the vortex that swirls with zero vorticity and the straight flow that spins

The irrotational line vortex $v_\theta=\Gamma/2\pi r$ moves fluid in perfect circles with $\boldsymbol{\omega}=0$ everywhere off-axis, while plain parallel shear $\mathbf{v}=u(y)\hat{\mathbf{x}}$ — streamlines straight as rails — carries $\boldsymbol{\omega}=-u'(y)\hat{\mathbf{z}}\neq0$ . Judging rotationality by eye from streamline shape gets both cases wrong. Always compute $\nabla\times\mathbf{v}$ , or imagine the drifting paddlewheel: it reads the local antisymmetric velocity gradient, not the curvature of its path. The same trap sinks Stokes'-theorem arguments around wings: the region outside an airfoil is not simply connected, so zero vorticity in the fluid does not force zero circulation around the body.

Worked examples

Example 1 — reading a tornado with the Rankine vortex (confidence builder). A tornado is modeled as a Rankine vortex: solid-body rotation inside a core of radius $r_c=50\ \mathrm{m}$ , irrotational $1/r$ flow outside, with peak wind $v_{\max}=60\ \mathrm{m/s}$ at the core edge. Find the circulation, the core vorticity, and the wind speed 200 m from the axis.

Step 1. Circulation from the core-edge speed: $\Gamma=2\pi r_c\,v_{\max}=2\pi\times50\times60=1.88\times10^{4}\ \mathrm{m^2/s}$ . By Kelvin this $\Gamma$ threads every loop enclosing the core.

Step 2. Core vorticity (solid-body part, $v_\theta=\Omega r$ with $\Omega=v_{\max}/r_c=1.2\ \mathrm{s^{-1}}$ ): $\omega=2\Omega=2.4\ \mathrm{s^{-1}}$ — every parcel in the core turns full circle in $2\pi/\Omega\approx5$ seconds.

Step 3. Outside, the flow is the line vortex of Derivation C: $v_\theta(200\ \mathrm{m})=\Gamma/(2\pi\times200)=1.88\times10^4/1257=15\ \mathrm{m/s}$ .

Step 4. $\Gamma=1.9\times10^4\ \mathrm{m^2/s}$ , core vorticity $2.4\ \mathrm{s^{-1}}$ , and the wind at 200 m is 15 m/s — the $1/r$ tail is why tornadoes are felt far beyond their visible funnel, and the vorticity confinement is why the funnel edge is so sharp.

Example 2 — circulation that holds an airplane up (exam level). A light aircraft of mass $m=1000\ \mathrm{kg}$ cruises at $U=50\ \mathrm{m/s}$ in air of density $\rho=1.2\ \mathrm{kg/m^3}$ ; its wing has span $b=11\ \mathrm{m}$ and mean chord $c=1.5\ \mathrm{m}$ . Estimate the bound circulation per wing section, and the implied velocity split between upper and lower surfaces.

Step 1. In steady level flight the wing carries the weight: $L=mg=1000\times9.81=9.81\times10^3\ \mathrm{N}$ , or per unit span $L'=L/b=9.81\times10^3/11=892\ \mathrm{N/m}$ .

Step 2. Kutta–Joukowski gives the circulation each section must hold: $\Gamma=L'/(\rho U)=892/(1.2\times50)=14.9\ \mathrm{m^2/s}$ .

Step 3. Model the circulation as a velocity difference between upper and lower surfaces along the chord: $\Gamma\approx\Delta v\times c$ , so $\Delta v=\Gamma/c=14.9/1.5=9.9\ \mathrm{m/s}$ — upper-surface flow roughly $55\ \mathrm{m/s}$ , lower roughly $45\ \mathrm{m/s}$ , a $\pm10\%$ asymmetry.

Step 4. Cross-check with Bernoulli from the previous article: the pressure difference implied is $\Delta p\approx\rho U\,\Delta v=1.2\times50\times9.9=594\ \mathrm{Pa}$ , and $\Delta p\times c=594\times1.5=891\ \mathrm{N/m}=L'$ — the ledger closes.

Step 5. Each meter of wing carries $\Gamma\approx15\ \mathrm{m^2/s}$ of bound circulation, equivalent to a ~10 m/s upper/lower velocity split. Kelvin's theorem adds the epilogue: an equal and opposite starting vortex with $\Gamma\approx-15\ \mathrm{m^2/s}$ per section was left on the runway at takeoff, and the wingtip vortices connecting them trail behind the aircraft — which is why heavy jets impose wake-separation minutes on the traffic behind them.

Origins · 1 minHermann von Helmholtz and how this was discovered

Hermann von Helmholtz

1821–1894 · Berlin, Königsberg, Heidelberg

Helmholtz came to fluids from physiology — he was an army surgeon who measured the speed of nerve impulses — and brought a physiologist's respect for what is conserved. His 1858 paper on vortex motion founded the subject: vortex lines move with the fluid, vortex tubes keep their strength, and they can never end in the interior. The results so impressed William Thomson (Kelvin) that he built a theory of atoms as knotted vortex rings in the ether — wrong about atoms, but the mathematics it spawned became knot theory, and Kelvin's own 1869 circulation theorem gave Helmholtz's laws their cleanest proof. Helmholtz meanwhile invented the ophthalmoscope, formulated the general law of energy conservation, and taught half the next generation of German physics; the vortex theorems remain the working language of everyone who studies tornadoes, wingtips, or superfluid helium.

History

1752

— d'Alembert finds that ideal potential flow past a body exerts zero drag, a paradox unresolved for 150 years

1858

— Helmholtz publishes the vortex theorems: vortex lines are material, tube strength is conserved

1867

— Kelvin, inspired, proposes vortex atoms and begins the mathematics that becomes knot theory

1869

— Kelvin proves the circulation theorem, the integral conservation law behind Helmholtz's results

1902

— Kutta and, independently, Joukowski (1906) tie circulation to lift, making wings calculable

1955

— Feynman and Onsager's quantized vortices are observed in superfluid helium, circulation in units of h over m

Going further

References. Landau & Lifshitz, Fluid Mechanics (Vol. 6), §§2, 8–11 · Batchelor, An Introduction to Fluid Dynamics, §§2.3–2.5, 3.1, 5.1 · Goldstein, Classical Mechanics, 3rd ed., Ch. 13 · Taylor, Classical Mechanics, Ch. 16 · Morin, Introduction to Classical Mechanics, fluids chapter

Provenance. A verifier should recompute: (i) the gradient decomposition and $\boldsymbol{\omega}=2\boldsymbol{\Omega}$ (Derivation A, checked on solid-body rotation); (ii) the Kelvin cancellations — the $\oint\mathbf{v}\cdot\mathrm{d}\mathbf{v}=0$ stretching term and the closed-loop gradient integral (Derivation B); (iii) the line-vortex results $\omega_z=0$ , radius-independent $\Gamma$ , multivalued $\phi=\Gamma\theta/2\pi$ (Derivation C); (iv) Example 1 ( $\Gamma=1.88\times10^4\ \mathrm{m^2/s}$ , $\omega=2.4\ \mathrm{s^{-1}}$ , $v(200\,\mathrm{m})=15\ \mathrm{m/s}$ ) and Example 2 ( $\Gamma=14.9\ \mathrm{m^2/s}$ , $\Delta v=9.9\ \mathrm{m/s}$ , ledger closing to $891\ \mathrm{N/m}$ ). Kutta–Joukowski $L=\rho U\Gamma$ , added-mass and quantized-circulation facts are standard: Landau–Lifshitz Vol. 6 §§8–11, Batchelor §5.1. The Rankine tornado numbers are illustrative model values, not measurements. The full story — boundary-layer separation, vortex-induced vibration, turbulence and its cascades, superfluid vortex dynamics — lives in the Fluid Mechanics track (migrates later as its own subject); this chapter carries only the ideal-flow vortex structure the continuum thread needs.

Builds on: Ideal Fluid Kinematics and the Continuity Equation · Euler's Equations and Bernoulli's Theorem · Strain Tensor and Deformation of Continua Leads to: The Navier–Stokes Equations and Viscous Flow · Elastic Waves in Solids · The One-Dimensional Wave Equation

If the paddlewheel does not spin in a line vortex, what happens to it exactly at the axis?

The idealized line vortex is singular there: all the circulation's vorticity is compressed into a delta function on the axis, so "the paddlewheel at $r=0$ " spins infinitely fast — a sign the model has been pushed past its jurisdiction. Real vortices resolve the singularity with a core: viscosity smooths the innermost region into near-solid-body rotation, giving the Rankine profile of Example 1 (or its smooth cousin, the Lamb–Oseen vortex). Outside the core the $1/r$ irrotational solution is excellent; inside, vorticity is finite and the paddlewheel spins at $\Omega=v_{\max}/r_c$ . The composite structure — rotational core, potential far field — is the generic anatomy of every real vortex from bathtub drains to hurricanes.

Kelvin says circulation around a material loop is conserved and starts at zero. So where does a wing's circulation actually come from?

From a bookkeeping split, not from violating the theorem. Take a huge material loop enclosing the wing and the air behind it: its circulation is zero before takeoff and stays zero, exactly as Kelvin demands. But as the wing accelerates, viscosity in the millimeter-thin boundary layer at the sharp trailing edge — where Kelvin's inviscid hypothesis genuinely fails — sheds a concentrated starting vortex of circulation $-\Gamma$ . The big loop still totals zero, so the sub-loop around the wing must now carry $+\Gamma$ . The magnitude is fixed by the Kutta condition: $\Gamma$ adjusts until the flow leaves the trailing edge smoothly. Viscosity's role is catalytic — it acts in a tiny region, sets $\Gamma$ , and the resulting lift $L=\rho U\Gamma$ is then computable entirely within ideal-flow theory.

What is d'Alembert's paradox, and how is it resolved?

Solve potential flow past any closed body and integrate the Bernoulli pressure over its surface: the net force parallel to the stream is exactly zero — no drag. d'Alembert derived this in 1752, and since real bodies obviously feel drag, it discredited ideal-flow theory for over a century. The resolution (Prandtl, 1904) is that the no-slip condition breeds a thin vortical boundary layer that potential theory omits; at high Reynolds number this layer may separate, leaving a broad wake whose pressure never recovers to the front-side values, and the fore-aft pressure asymmetry is the drag. The paradox is thus a precision statement about where ideal theory lives: excellent outside boundary layers and wakes, structurally blind inside them. The Navier–Stokes article picks up exactly here.

Why can a vortex line not simply end in the middle of the fluid?

Because vorticity is a curl, and the divergence of a curl vanishes identically: $\nabla\cdot\boldsymbol{\omega}=0$ . Vorticity is a solenoidal field, exactly like the magnetic field, so its field lines admit no sources or sinks — they must close into loops (smoke rings), extend to boundaries (a stirred cup: the vortex ends on the bottom and the free surface), or go to infinity. Helmholtz's theorem adds the dynamical half: in ideal flow those lines are also material, dragged with the fluid at fixed tube strength. When you see a tornado "end" in mid-air, the vorticity has not terminated — it has spread into a diffuse, invisible continuation that no longer condenses cloud.

The vortex stretching term looks like magic — how does stretching a tube speed it up, and why does this matter for turbulence?

It is angular-momentum conservation wearing continuum clothes. Take a vortex tube of cross-section $A$ and vorticity $\omega$ : Kelvin fixes the circulation $\Gamma\sim\omega A$ , and incompressibility fixes the volume $A\ell$ . Stretch the tube to double length: $A$ halves, so $\omega$ doubles — the skater pulling in her arms, with $(\boldsymbol{\omega}\cdot\nabla)\mathbf{v}$ as the differential statement. This is the engine of three-dimensional turbulence: chaotic advection statistically favors stretching, so vorticity intensifies at ever-smaller scales, cascading energy downscale until viscosity burns it off. In two dimensions the term vanishes identically (vorticity points out of the plane, gradients lie in it), turbulence cascades energy upscale instead, and that inverse cascade is why Jupiter's Great Red Spot and terrestrial jet streams are so stubbornly long-lived.

How do pressure and density misalignment actually create rotation? The baroclinic term feels abstract.

Picture a shoreline on a summer afternoon. Pressure surfaces are nearly horizontal (hydrostatics), but the air over hot land is less dense than over cool sea, so density surfaces tilt. Now consult the torque balance: the pressure-gradient force per unit mass is $-\rho^{-1}\nabla p$ , which is stronger in the light air than the dense air at the same height. Columns of light air get pushed up more effectively than dense columns resist — the differential push is a torque, and a circulation spins up: onshore breeze at the surface, offshore return aloft. That is $\rho^{-2}\nabla\rho\times\nabla p$ doing its work, and the same misalignment drives ocean eddies, the global overturning circulation, and the baroclinic instability that makes mid-latitude weather.

Is the complex-potential trick just a curiosity, or did it matter historically?

It designed the first airplanes. Because any analytic $w(z)$ is automatically an incompressible irrotational flow, and conformal maps preserve analyticity, Joukowski could map the exactly solvable cylinder-with-circulation flow onto realistic curved, cambered airfoil shapes — the Joukowski airfoils — and read off lift from $L=\rho U\Gamma$ with $\Gamma$ set by the Kutta condition. Through the 1930s, conformal mapping was theoretical aerodynamics. Its modern descendants are panel and vortex-lattice methods (numerical superposition of the same singular solutions), still the first tool applied to every new wing, and the mathematics returns in unexpected places: Laplacian growth, electrostatics, and even the conformal field theories of statistical mechanics.

In superfluids the circulation is quantized in units of h over m. What does Planck's constant have to do with Kelvin's theorem?

The superfluid's velocity is the gradient of the phase $S$ of a macroscopic wavefunction, $\mathbf{v}=(\hbar/m)\nabla S$ — potential flow made literal, with $\phi=(\hbar/m)S$ . The wavefunction must be single-valued, so around any closed loop the phase can change only by $2\pi n$ , forcing $\Gamma=\oint\mathbf{v}\cdot\mathrm{d}\boldsymbol{\ell}=n\,h/m$ : Derivation C's multivalued potential, but with quantum mechanics restricting the winding to integers. Kelvin's theorem then becomes exact rather than approximate — the superfluid has strictly zero viscosity, so circulation cannot decay continuously; quantum turbulence relaxes only by discrete vortex reconnections. Onsager and Feynman predicted this in the 1940s–50s; imaging the vortex lattices in rotating condensates is now a standard cold-atom experiment.