Vorticity and Circulation

A tornado, a smoke ring puffed across a room, the tight little whirlpool that forms as your bath drains, the invisible corkscrew of air trailing off the tip of an airliner's wing, the swirl you set up when you stir milk into coffee — all of these are the same phenomenon wearing different clothes. Fluid is spinning. Not just flowing along, but rotating locally, element by element. The quantity that measures that local spin — how fast a tiny blob of fluid is turning on its own axis — is called the vorticity, and it is one of the most powerful ideas in all of fluid dynamics.

This page builds vorticity from the ground up as \boldsymbol{\omega} = \nabla\times\mathbf{u}, shows that it equals exactly twice the angular velocity of a fluid element, and ties it — through Stokes' theorem — to the circulation \Gamma = \oint \mathbf{u}\cdot d\boldsymbol{\ell}, the net "swirl" measured around a closed loop. Along the way we meet the single most treacherous misconception in the subject — that curved streamlines mean spin and straight ones don't — and put it firmly right. We finish with Kelvin's circulation theorem, the conservation law that makes vortices so astonishingly persistent: smoke rings that drift the length of a hall, tornadoes that live for an hour, the wingtip vortices that force jets to take off a safe distance apart.

Vorticity: twice the local spin of a fluid element

Imagine dropping a microscopic paddle wheel — a tiny cross with vanes — into the flow and asking: does it turn? If it does, the fluid there has vorticity; if it merely translates without rotating, the fluid is locally irrotational. To make this precise, look at how the velocity field varies from point to point. The vorticity is the curl of the velocity field:

\boldsymbol{\omega} = \nabla\times\mathbf{u}.

It is a vector: its direction is the local axis of spin (by the right-hand rule) and its magnitude is the rate of that spin. In two dimensions the flow spins only about the z-axis, and vorticity collapses to a single component:

\omega_z = \frac{\partial v}{\partial x} - \frac{\partial u}{\partial y},

where \mathbf{u} = (u,\,v). The deep fact is what this number is. Decompose the local velocity gradient into a symmetric part (pure stretching — the strain rate) and an antisymmetric part (pure rotation). The antisymmetric part rotates a fluid element as a rigid little body with angular velocity \boldsymbol{\Omega}, and working it through gives

The relation. A fluid element spins with angular velocity \boldsymbol{\Omega} = \tfrac{1}{2}\,\boldsymbol{\omega} = \tfrac{1}{2}\,\nabla\times\mathbf{u}. Equivalently, \boldsymbol{\omega} = 2\,\boldsymbol{\Omega} — vorticity is twice the local angular velocity.
Solid-body check. Rotate a whole fluid rigidly at angular speed \Omega, so \mathbf{u} = \boldsymbol{\Omega}\times\mathbf{r}. Then every element turns at \Omega and indeed \omega_z = 2\Omega everywhere — uniform vorticity.
Irrotational flow. A flow with \boldsymbol{\omega} = \mathbf{0} everywhere is called irrotational: paddle wheels translate and deform but never turn. Such flows admit a velocity potential, \mathbf{u} = \nabla\phi.

Keep that paddle-wheel test in mind — it is the physical meaning of the curl, and it is what the figure further down puts on screen.

Circulation, and its link to vorticity by Stokes' theorem

Vorticity is a local quantity — it lives at each point. Its global partner is the circulation: pick any closed loop C drawn in the fluid and add up the component of velocity along the loop, all the way around:

\Gamma = \oint_C \mathbf{u}\cdot d\boldsymbol{\ell}.

Circulation measures the net tendency of the fluid to run around the loop rather than across it — a single number capturing the swirl enclosed. And here is the bridge that makes vorticity so useful: circulation around a loop equals the total vorticity threading the surface it bounds. That is Stokes' theorem applied to the velocity field.

The identity. \Gamma = \oint_C \mathbf{u}\cdot d\boldsymbol{\ell} = \iint_S (\nabla\times\mathbf{u})\cdot d\mathbf{A} = \iint_S \boldsymbol{\omega}\cdot d\mathbf{A}.
Reading it. The circulation around a loop is the flux of vorticity through any surface capping that loop. Vorticity is the "circulation per unit area".
Uniform-vorticity shortcut. If the vorticity is constant over a flat region of area A, then simply \Gamma = \omega\,A.

Reveal the figure below one step at a time. First a shear flow — perfectly straight, parallel streamlines — where a dropped paddle wheel nonetheless spins, because the fluid slides past faster on one side than the other. Then a free vortex — streamlines that are perfect circles — where the paddle wheel, remarkably, does not spin at all. Finally a rectangular loop illustrating \Gamma = \oint\mathbf{u}\cdot d\boldsymbol{\ell}.

This is the trap almost everyone falls into, and getting it right is the whole point of the subject. Vorticity is about the local rotation of a fluid element, not the curvature of the streamlines. The two have nothing directly to do with each other.

Straight streamlines, yet spinning. Take a simple shear flow \mathbf{u} = (\alpha y,\,0) — every streamline is a dead-straight horizontal line. But the fluid slides faster the higher (or lower) you go, so a little paddle wheel gets pushed harder on one side than the other and turns. Indeed \omega_z = \partial_x v - \partial_y u = -\alpha \neq 0. Straight lines, genuine vorticity.

Circular streamlines, yet NOT spinning. The free vortex u_\theta = k/r (the bath-drain swirl, the tip vortex) has streamlines that are perfect circles — and yet \omega = 0 everywhere except at the very centre. A paddle wheel carried around the loop keeps pointing the same way, like a chair on a Ferris wheel: it orbits without rotating. The inner face, closer to the centre, moves faster; the outer face slower; and for the special 1/r profile these two effects cancel exactly, leaving zero net spin. Curved streamlines, zero vorticity.

The moral: don't eyeball the streamlines. Compute \nabla\times\mathbf{u}, or imagine the paddle wheel. That is the only reliable test.

Worked examples

Example 1 — vorticity of a shear flow. Take \mathbf{u} = (u,\,v) = (3y,\,0). Then

\omega_z = \frac{\partial v}{\partial x} - \frac{\partial u}{\partial y} = 0 - 3 = -3\ \text{s}^{-1}.

Non-zero, despite the ruler-straight streamlines. Every fluid element rotates at \Omega = \omega_z/2 = -1.5\ \text{s}^{-1} (clockwise).

Example 2 — a free vortex is irrotational. In Cartesian form the free vortex is \mathbf{u} = \left(-\dfrac{k\,y}{x^2+y^2},\ \dfrac{k\,x}{x^2+y^2}\right). Grinding out the derivatives (away from the origin),

\omega_z = \frac{\partial v}{\partial x} - \frac{\partial u}{\partial y} = 0 \quad (r \neq 0).

Circular streamlines, yet zero vorticity. All the swirl is concentrated in a singular core at r=0.

Example 3 — circulation by Stokes. A region of fluid has uniform vorticity \omega_z = 4\ \text{s}^{-1}. What is the circulation around a rectangular loop 2\ \text{m} \times 3\ \text{m} lying in it? Using \Gamma = \omega\,A,

\Gamma = \omega_z\,A = (4)\,(2\times 3) = 24\ \text{m}^2/\text{s}.

We got the loop integral without doing a loop integral — Stokes turned it into a trivial area times a constant.

Example 4 — circulation from the four sides. Suppose you walk a rectangular loop and measure the accumulated \mathbf{u}\cdot d\boldsymbol{\ell} along each side as +8, +1, -5 and +2 (in \text{m}^2/\text{s}). The circulation is just the sum:

\Gamma = 8 + 1 - 5 + 2 = 6\ \text{m}^2/\text{s}.

A positive net circulation, so by Stokes there is a net counter-clockwise vorticity threading the loop.

Two archetypes: solid-body rotation vs. the free vortex

Two idealised swirling flows anchor the whole subject, and their tangential speed profiles v_\theta(r) could hardly be more different.

Solid-body rotation, v_\theta = \Omega\,r: the fluid spins as a rigid disc (think of a bucket of water spun up until it turns with the bucket). Speed grows linearly outward, and the vorticity is uniform, \omega = 2\Omega everywhere. This is the vortex core.
Free vortex, v_\theta = k/r: speed falls as 1/r, screaming fast near the middle and gentle far out — the bathtub drain, the tornado's outer region, the wingtip vortex. It is irrotational everywhere except a singular core at the centre.

A real tornado or bath vortex is a Rankine vortex: solid-body rotation in the core (the linear part), smoothly matched to a free-vortex 1/r tail outside. The graph shows the two building blocks.

Even though the free vortex has zero vorticity everywhere off-axis, a loop that encloses the core has non-zero circulation: all the vorticity is packed into the singular centre, and Stokes' theorem faithfully reports it as \Gamma = 2\pi k for any loop around the axis. Vorticity and circulation are two views of the same swirl.

Kelvin's circulation theorem: why vortices last

The final piece explains the sheer stubbornness of vortices. Follow a loop that moves with the fluid — a "material loop", made always of the same fluid particles. How does the circulation around it change in time? For an ideal fluid the astonishing answer is: it doesn't.

The statement. For a material loop C(t) convected with the flow, \frac{D\Gamma}{Dt} = \frac{D}{Dt}\oint_{C(t)} \mathbf{u}\cdot d\boldsymbol{\ell} = 0. The circulation around any such loop is conserved.
The conditions. It holds for an inviscid fluid (no viscosity), subject to conservative body forces (like gravity), and that is barotropic (pressure a function of density alone, p = p(\rho)).
The consequence. Vorticity cannot be created or destroyed inside such a fluid: a flow that starts irrotational stays irrotational, and existing vortices are "frozen" into the fluid and simply carried along. Smoke rings persist; a tornado holds together; the wingtip vortex trails for kilometres.

Real fluids have a little viscosity, so vorticity does slowly diffuse and vortices eventually fade — which is exactly why the theorem's conditions matter, and why it is viscosity (at the wing's surface) that generates the circulation responsible for lift in the first place. But over the timescales of a smoke ring crossing a room, Kelvin's theorem is a superb approximation, and it is the reason the ring survives the journey at all.

Lift, by the Kutta–Joukowski result, is L = \rho\,U\,\Gamma per unit span — proportional to the circulation around the wing. But Kelvin's theorem says circulation is conserved, and the air started at rest with \Gamma = 0. So where does the wing's circulation come from without breaking the conservation law?

The trick is to shed an equal and opposite vortex. As a wing starts moving, a starting vortex peels off the trailing edge and is left behind; the wing keeps an equal and opposite bound circulation around itself. Draw one big material loop around both and the total is still zero — Kelvin is satisfied — but the loop hugging just the wing now carries the circulation that generates lift. And that trailing air, curling up off the wingtips into two powerful tip vortices, is why air-traffic control spaces landing aircraft minutes apart: the swirl is still there, faithfully conserved, long after the plane has passed.