|March 9, 2010, 09:42||
Parallelization with icoFsiFoam
Join Date: Nov 2009
Posts: 57Rep Power: 7
I have a problem which might be very simple to solve - or very tricky.
I'm working with icoFsiFoam on OpenFoam-1.5-dev. So far i was quite content running the solver on my 1-core PC but as usual cases grow larger and more complex so finally I'm trying to perform the calculations on a computing cluster.
I did some preliminary tests using more basic solvers to learn about decomporsePar and reconstructPar / reconstructParMesh. So far, everything is fine and the cluster does a great job.
Now, the focus of my work is on FSI, thus I use icoFsiFoam. And here I ran into some problems with the decomposition for the parallelization. As I have two distinct mesh regions (fluid, solid) I'm basically not sure how to perform the decomposition in general. Shall I decompose both regions into let's say 6 SubDomains? or decompose only the fluid region? (I tried both without success). And how many processors do I need to assign to the task? The sum of solid and fluid subdomains?
I assume there is an elegant way to get this right, but I'm a little bit lost.
I would very much appreciate any help you could provide on that issue.
|Thread||Thread Starter||Forum||Replies||Last Post|
|parallelization calculation||SARU||OpenFOAM||0||December 15, 2009 20:24|
|icoFsiFoam case needed||vivekcfd||OpenFOAM||1||May 6, 2009 13:41|
|IcofsiFoam not supported under released 141 package||pi06jl6||OpenFOAM Running, Solving & CFD||4||June 26, 2008 13:59|
|Solid Large deformation for icoFsiFoam||fw407||OpenFOAM Running, Solving & CFD||2||June 13, 2008 10:58|
|Errors in running a icoFsiFoam case||jin_xu||OpenFOAM Pre-Processing||0||June 9, 2008 06:48|