Wednesday, March 13, 2013

Playing with Some Fluid Simulation

A recent project has me playing around with some fluid simulation.  I got some nice looking results and thought I would post a few.

The first is a buoyancy-driven flow simulated using a Marker-And-Cell (MAC) grid with the Boussinesq approximation.  This is simply a source term applied to the vertical velocity based on the fluid density, making it trivial to add to a basic solver.  Although I have no explicitly simulated viscosity, numerical viscosity is enough to cause some interesting unsteady flow effects

The second is a simulation of smoke in box, using a 3D 'Stable Fluids' solver with the first order advection replaced by the second-order 'BFECC' scheme.  This gives -way- better results than standard first order semi-Lagrangian discretizations:

The results compared to the first order scheme (shown below) are pretty impressive:

That's it..

No comments: