Computational Fluid Dynamics, or CFD, has interested me since I originally heard about it in a class at MIT for high school students in 2004. Since, I’ve been working hard trying to teach myself enough math to solve the navier-stokes equations on my own. As I write this I’ve just begun solving partial differential equations, so I’ve got a little ways to go yet.
Here is a collection of pictures created by the cluster, running code written by Andrei Chernousov. His code can be found here at the time of writing. I have mirrored a few of the codes at the bottom. A short description follows each image.
MPI_layer2
Here are two frames of a sequence showing the vorticity in a fluid in which a single jet enters from the left. You can see the Karmon Vortex Street: a phenominon where “eddies” spiral out to eaither side behind a flow, generally occuring as a flow goes around either side of a circular object.
MPI_jet
One can see the interaction between two streams of a fluid. Really quite beautiful.
MPI_duct
This shows fluid flow through a duct. (It’s different than the others because the visualization too I used was GnuPlot, rather than Winfield) Generally people think fluid through an enclosed space just stays together at a constant rate: not true! In fact, the friction with the edges of the confinment causes that fluid to go slower, creating these complicated paterns at the edges.
Local copies of Andrei’s code: layer2.tar.gz, mpi_layer2.tar.gz, duct.tar.gz, mpi_duct.tar.gz, mpi_jet.tar.gz.