3d condensation

Hi guys,

I am running a d3q15 BGK simulation with attractive particle forces to simulate condensation. The model is based on my 2dimensional model which works fine and produceses a circle at convergence. I am currently using Matlab (due to my supervisor’s preference :/)

My problem with my 3dsimulation is:

  1. Depending on the initial density, either a sphere or an infinitely long cylinder forms. Note I am using a continous boundary. Due to this I am suspecting there could be some error with my code as I believe due to surface tension that a sphere should always form. Any ideas from where this problem could arise? When I replace the interparticle forces with a global unit x-y-z force and add a bounceback boundary condition, keeping the f_eq the same, as expected the fluid gathers up in a corner nicely…

  2. I am having problems visualising this. Currently I plot points with size and color depending on the density and I only plot above a certain density. This works as I can see the high density water forming but I feel it is not optimal. I am considering saving the data and postprocessing it with openGL. How is a 3d density field usually plotted in Matlab?


Now I managed to reproduce the problem for 2 dimensions as well :confused:

Here are some pictures:

Yeah the 3d visualization is horrible :stuck_out_tongue: still pondering how I am gonna do that but atleast it’s able to show what’s going on. My initial instict suggests the problem is involved with initial density distribution (which is exactly the same as used in the multi-component model on this site) as well as small grids. However when I run a 100x100x100 simulation I still manage to get a cylinder. I do not manage to get a rectangle for the 2d case though with grids larger than 50x50.

Really stumped on this so any help would be appreciated.

edit: Also yeah they are obviously perodic boundaries which definately have something to do with the problem.

for visualization I use ParaView. Jonas has also reported a positive experience with this tool. Once your data is in the correct file format, you can do almost anything, e.g. contour plots, slices, streamlines etc. It is simple to learn how to use it, and ParaView is free. There is also a book available.

You, Sir, just made my day :smiley:


Still I’m wondering about the appearance of the cylinder, something anyone else experienced?

Yes you can get cylinders if you use periodic boundary condition …


Yeah realised a bit after this post that, given our previous discussion, as long as all particles along the interface experience the same force then the fluid is in equilibrium.