Point Vortices
Source lecture(s): PC316 Ch. 8
Intuition
Shrink a Rankine vortex core to zero and you get a point vortex: all the circulation on a single point, inducing \(v_\theta = \Gamma/2\pi r\) around itself. The crucial property is what it does not do — a point vortex has no self-induced motion. It cannot push itself. So a collection of them obeys a rule of almost embarrassing simplicity:
Each vortex is advected by the velocity field of all the others.
That single sentence generates orbiting pairs, translating pairs, leapfrogging rings, chaos, and a respectable model of two-dimensional turbulence. Drive it in the point vortex sandbox.
The equations
This is a Hamiltonian system with \(H = -\frac{1}{4\pi}\sum_{i<j}\Gamma_i\Gamma_j\ln r_{ij}\), where \(\Gamma_i x_i\) and \(y_i\) are conjugate variables — an unusual and elegant structure in which position coordinates are canonically conjugate to each other. Conserved: \(H\), the linear impulse \(\sum\Gamma_i\mathbf{r}_i\), and the angular impulse \(\sum\Gamma_i|\mathbf{r}_i|^2\).
Three exact results
Co-rotating pair (equal \(\Gamma\), separation \(d\)). Each drives the other at \(\Gamma/2\pi d\) perpendicular to their line, so the pair orbits its midpoint:
Merger of co-rotating vortices is the elementary coarsening step of 2-D turbulence.
Counter-rotating pair (\(+\Gamma\) and \(-\Gamma\), separation \(d\)). Each advects the other in the same direction, so the pair translates in a straight line at
This is a wingtip vortex pair seen head-on, and the cross-section of a vortex ring. It explains at once why smoke rings move: the ring propels itself by its own induction.
Vortex near a wall. The no-penetration boundary condition is satisfied by an opposite-sign image vortex mirrored in the wall (the same method of images used in electrostatics). The real vortex is then advected by its image, gliding parallel to the wall at
The dynamics gets rich fast
- Two vortices: always integrable. They orbit or translate, forever.
- Three vortices: integrable in general (enough conserved quantities), but capable of intricate motion including collapse in finite time for special initial conditions.
- Four or more: chaotic. This was one of the first recognised examples of Hamiltonian chaos.
- Leapfrogging: two counter-rotating pairs in line alternately pass through one another — the 2-D cross-section of two vortex rings chasing each other, and a genuinely beautiful demonstration that these simple rules produce complex behaviour.
Common mistakes
- Including the self-term. The \(j \neq i\) is essential; a point vortex does not move itself. Including it gives a division by zero, or worse, a plausible-looking wrong answer if you regularise it.
- Forgetting that images are slaved. The image is not an independent degree of freedom; it must be recomputed from the real vortex at every stage of the integrator, not advanced separately.
- Expecting energy to be the Hamiltonian. \(H\) is not the kinetic energy (which diverges for point vortices); it is the interaction energy, and it is the generator of the dynamics.
Related concepts
- Rankine vortex — the finite-core version
- Vorticity · Helmholtz theorems
- Wingtip vortices — the counter-rotating pair in the wild
- Conformal mapping & images — the wall boundary condition
- Point vortex sandbox — all of the above, live
Knowledge graph position
Prerequisites: vorticity, potential flow, Helmholtz theorems. Leads to: vortex-ring dynamics, 2-D turbulence, vortex methods in CFD.
Quiz
Q1 (computational). Two vortices of \(\Gamma = 4\) m²/s sit 2 m apart. What is their orbital period?
Answer
\(T = 2\pi^2d^2/\Gamma = 2\pi^2(4)/4 = 2\pi^2 \approx 19.7\) s.
Q2 (conceptual). Why does a counter-rotating pair translate while a co-rotating pair orbits?
Answer
Each vortex is advected by the other's field, which is perpendicular to the line joining them. For equal signs the two induced velocities are anti-parallel, producing rotation about the midpoint. For opposite signs they are parallel, so both move the same way and the pair translates with fixed separation.
Q3 (MCQ). A vortex at height \(h\) above a wall moves because:
- (a) the wall exerts a viscous drag
- (b) its image vortex, of opposite sign at depth \(h\), advects it parallel to the wall
- (c) of self-induction
- (d) it does not move
Answer
(b). The image enforces no penetration at the wall, and the real vortex sits in the image's field — a counter-rotating pair of separation \(2h\), giving glide speed \(\Gamma/4\pi h\).