FluidLab

Flow past a cylinder, with the numbers showing

A real lattice-Boltzmann solve, not a shader trick: 400×160 grid, bounce-back walls, and a live drag coefficient. Push the Reynolds number past 47 and watch the wake lose its symmetry.

St and the shedding period appear once a few vortices have detached — give it 30–60 seconds.

Drag Cd
—
Lift Cl (RMS)
—
Strouhal St
—
Shedding period
—
Cd (momentum check)
—
Stability
—

What you are looking at

The colour is vorticity: red for counter-clockwise, blue for clockwise. A cylinder in a stream grows a boundary layer on each shoulder; past a certain Reynolds number those layers separate and roll up into vortices that peel off alternately — the Kármán vortex street.

The two numbers underneath are the point. Cd is the time-averaged drag coefficient, integrated from the momentum the walls exchange with the fluid. St is the Strouhal number, the non-dimensional shedding frequency, found by timing the lift signal's zero crossings.

What the Reynolds number changes

Re < 47

Steady and symmetric

The wake is a pair of fixed recirculating eddies. No lift oscillation, no street — the simulation settles into a still picture.

47 – 200

Clean shedding

Vortices detach alternately and the wake is two staggered rows. This is the regime the readouts are calibrated for; St sits near 0.18 here.

200 – 250

Thinner, faster, noisier

Shedding gets quicker and the wake becomes three-dimensional in reality — something a 2D solver cannot reproduce. Treat the numbers as trends, not predictions.

Checked against published values

Measured with this solver at Re = 100, D = 17 cells, 400×160 grid, run for 14,000 steps (about six shedding cycles).

QuantityThis pageLiterature
Drag coefficient Cd1.40≈ 1.35
Strouhal number St0.181≈ 0.164
Shedding period1,175 steps—
Blockage ratio D/H10.6%0% (unbounded)
Mach number / relaxation0.139 / 1.849—

Both quantities read high by roughly the amount you would expect from a 10.6% blockage. Close the gap by shrinking the cylinder relative to the channel — the “Cylinder size” slider — at the cost of resolving the wake with fewer cells.

The momentum check figure in the panel is a second, independent estimate of the drag: flux of ρu² + p across two cross-sections. On the default settings it lands within about 7% of the integrated value; a large disagreement is the quickest sign that the run is not yet converged or the grid is too coarse.

FAQ

Why is the drag coefficient slightly higher than the textbook 1.35?

Because of blockage. The cylinder takes up 10.6% of the channel height, which squeezes the flow around it and raises both drag and shedding frequency by roughly 5–10%. In an unbounded domain the same solver lands much closer to the reference values.

What is the second drag number?

A momentum balance across two cross-sections, upstream and downstream of the cylinder. It is computed from the flow field alone, with no force integration, so agreement between the two is a check that the simulation is healthy rather than a claim of extra accuracy.

Why does nothing happen below Re ≈ 47?

Below that threshold the wake is steady and symmetric: the recirculation bubble sits still behind the cylinder. The shedding instability sets in around Re 47 and the familiar alternating street develops through Re 100–200.

Can I run several cylinders?

Yes — raise the “Cylinders” slider up to five. Spacing is fixed at three diameters, so you can watch the wake of one body feed the next.