Presets
Playback
Fluid and flow
Elastic mount (vortex-induced vibration)
View
Camera
Layers
Quality
Model check
The science
Reynolds number and the onset of shedding
A cylinder of diameter D in a stream U of a fluid with kinematic viscosity nu is governed by Re = U D / nu. For Re < 47 the wake is a steady pair of attached eddies. At the critical value Rec ≈ 47 the wake becomes unstable (a Hopf bifurcation) and sheds vortices alternately from the two sides: the Karman vortex street. Between about 47 and 190 the street is laminar and two-dimensional; above that three-dimensional instabilities appear, and for Re > 300 the wake is turbulent although the shedding stays visible as a coherent peak.
Strouhal number
The shedding frequency is fs = St U / D. We use Roshko's correlation St = 0.212 (1 - 21.2 / Re) for 47 < Re < 300 and St ≈ 0.21 above. Water at 1 cm/s past a 10 mm rod gives Re = 100, St = 0.167 and fs = 0.17 Hz; a 3 mm wire in a 5 m/s wind sings at about 350 Hz (the Aeolian tone).
The street as point vortices
Each shed vortex is a Lamb-Oseen vortex of circulation G: velocity utheta = G / (2 pi r) (1 - exp(-r2/s2)) and vorticity w = G / (pi s2) exp(-r2/s2), with a core s that grows slowly. The cylinder is the potential-flow doublet u = U(1 - R2(x2-y2)/r4), v = -2 U R2 x y / r4. Every vortex is advected by that flow plus the Biot-Savart velocity of all others (midpoint Runge-Kutta). The dye particles use the same velocity field.
For an infinite staggered street with spacing a between same-sign vortices and row separation h, Karman showed the street moves relative to the fluid with u = G / (2a) tanh(pi h / a) and is stable only for h / a = 0.281. We choose the circulation so that the vortices drift at 0.86 U (as measured behind real cylinders) and shed them at spacing a = 0.86 U / fs.
Forces and the elastic mount
Each shed vortex flips the sign of the side force: CL(t) = CL0 sin(2 pi fs t) with the empirical amplitude CL0 ≈ 0.9 sqrt(1 - 47/Re); the drag fluctuates at 2 fs around CD ≈ 1 + 4/sqrt(Re). Force per unit length is F = ½ rho U2 D C. A cylinder on springs obeys m y'' + c y' + k y = FL(t); at resonance, fs = fn, the steady amplitude is F0 / (m sqrt((wn2 - w2)2 + (2 zeta wn w)2)). This is what shakes chimneys, risers, power lines and periscopes.
Independent checks run in the page
1) A hidden 300-time-unit run records a probe velocity downstream; its Hann-windowed spectrum must peak at the Roshko St (within 3.5 %). The frequency comes from vortices passing the probe, not from the shedding clock. 2) The measured drift speed of mid-street vortices must match Karman's formula evaluated with the street's measured a, h and G (within 8 %; the real street is finite). 3) The mount oscillator is integrated for 120 periods and compared with the closed-form resonance amplitude (within 3 %).
Honest simplifications
The model is two-dimensional and inviscid except for core growth; vortices are inserted at the rate and strength the physics demands rather than being born in a boundary layer, there is no feedback of the cylinder motion on the wake (so no true lock-in frequency capture), and the wake broadening toward h = 0.281 a is an empirical relaxation. Above Re 190 real wakes are three-dimensional. Lift/drag amplitudes are empirical fits, not computed from the flow. Playback is slowed enormously for fast flows; the readouts are in real units.
Things to try
- Drag the flow speed across Re = 47 and watch the steady wake switch to a street.
- Use the lock-in preset, then change U slightly: the oscillator response collapses away from resonance.
- Click anywhere to place the probe; compare its local vorticity with the colours.
Educational model of an idealised laboratory flow.