Make waves.
A little chaos. A lot of fluid dynamics. Drag through the field and see what moves.
Start a little trouble.
THREE QUICK EXPERIMENTSReal equations.
Room to play.
A 2D numerical playground. Not a calibrated engineering tool.
What am I looking at?+
Two passive dyes move through a computed velocity field. The fluid bends around obstacles; your pointer adds dye and momentum. Switch to Speed to see fast paths, or Rotation to distinguish clockwise and counterclockwise motion.
All controls use grid units. Dye is illustrative, not a calibrated concentration. This coarse two-dimensional model still has some numerical smoothing and approximate pressure projection.
How is the flow calculated?+
A staggered-grid solver combines limited second-order MacCormack advection, implicit diffusion, and iterative pressure projection. Plain semi-Lagrangian advection blurred away more motion than the viscosity setting did. The optional WebGPU experiment still uses it. The channel has a fixed inlet, open outlet, and impermeable walls and obstacles. CPU computation runs in a worker; optional WebGPU computation is experimental. A larger grid is not a physical refinement study.
- Divergence RMS
- —
- Inlet / outlet flux
- —
Inspect the motion+
Peak / RMS speed shows how concentrated the motion is. Lower energy participation means a smaller fraction of the field contains most of the motion. These are numerical diagnostics, not measured physical accuracy.
- Peak / RMS speed
- —
- Energy participation
- —
- Kinetic energy
- —
- Peak vorticity
- —
- Momentum residual RMS
- —
Residual term norms: acceleration / advection / viscosity / pressure
—