HiI'm working on a 3D CFD model with Comsol 5.4 and tring to increese de number of elements but im runing low on RAM (i have 32GB of RAM working with an i7-4790). Im using P1+P1 (first order discretization for pressure and velocity fields) the mesh elements and DOF reported en the messege window are:
- Complete mesh consists of 697500 domain elements, 91570 boundary elements, and 2682 edge elements.
- Number of degrees of freedom solved for: 2787596 (plus 5292006 internal DOFs).
And the log reports at the end of the simultion:
- Solution time: 4356 s. (1 hour, 12 minutes, 36 seconds)
- Physical memory: 26.61 GB
- Virtual memory: 28.37 GB
I`m using the default solver configuration. Fully coupled, algebraic multigrid, GMRES solver.It seems to me that it's using more RAM than it should, and i wonder if it has anything to do with the internal degrees of freedom. I don't really understand what they are, and I can't find anything about internal DOF. I woud like to know if they use same amount of memory as regular DOF. And if is a way to reduce them.i'll appreciate any information about this, any other thoughts on the subject or if you can tell me if i having a wrong idea about it and my memory usage is normal
Dear Felipe Suarez,
the internal degrees of freedom are used by COMSOL to store information during solution, but they do not have equations so do not make the system matrix larger. I report here a quote from COMSOL user guide:
"When a solver starts working, it prints the name of the solvers used and then prints the number of degrees of freedom in the linear systems to be solved to the log. For certain problems, there are additional degrees of freedom involved in the discrete problem formulation that do not affect the size of the matrices assembled by the solver. These are called internal degrees of freedom and are displayed separately from the actual degrees of freedom in the log."
I hope this will help,
CA
Thanks Ciro
I read the comsol guide. It's quite clear, but still find odd my memory usage.
Dear Felipe Suarez
Have you tried to gradually increase the mesh refinement and monitor the memory allocation?
I suggest to look to this brief article that may be interesting for you:
https://www.comsol.com/blogs/much-memory-needed-solve-large-comsol-models/
Have a nice day,
CA
- Badges
- Science topic