CFD Online Discussion Forums

CFD Online Discussion Forums (
-   FLOW-3D (
-   -   Insufficent memory (

RY December 30, 2008 11:57

Insufficent memory

I am wondering how much memory F3D needs? Is there any limitation on the total/active number of mesh?

My computer has Intel Xeon Quad-core 2.0GHz CPU and 3.25G RAM. However, after completing pre-processor, F3D stopped with an error message - "cannot allocate memory for vvect, possibly due to insufficient memory". My model has 4,524,435 (total) and 3,292,093 (active) cells.

I would appreciate any suggestions.

Thanks, RY

CK December 30, 2008 21:25

Re: Insufficent memory
RY I guess you are using windows 32-bit system right? the maximum allowable total cell count with 32-bit system is ~ 1,300,000. so, pls either reduced cell count or upgrade to 64-bit OS.

michael barkhudarov December 30, 2008 23:44

Re: Insufficent memory

The total memory used by FLOW-3D varies with the selected physical and numerical options. Generally, I would say the limit is rather around 4 million for a 32-bit system, that is, it will use 2 Gb of memory with around 4 million active cells.

One possibility to increase the maximum cell count (up to 7.0-7.5 mln) is to use the single precision version. Or, of course, switch to a 64-bit system, as CK suggests.

RY December 31, 2008 09:50

Re: Insufficent memory
Thanks CK and Michael.

Yes, I am using 32-bit WIN XP...will try to ask IT to upgrade the system.

However, it may not be possible to further reduce the cell number because cell size is limited by the small components in a much bigger geometry.


CK January 1, 2009 20:34

Re: Insufficent memory
RY sound like you are solve problem which involved combination of very small object to very large objects. so do you mind to share what kind of problem that you are working at? do you faced any multi-block mesh error issue with your problem?

RY January 5, 2009 15:26

Re: Insufficent memory
Happy New Year CK.

I am studying the hydaulics in a flocculation basin, which includes two perforated baffles. Orifices on the baffles are much smaller than the basin geometry. As head loss through the orifices on baffles is very important, I have to limit the mesh size to resolve small orifices.

I try to avoid using multiblock meshing, as, to my experience, it would introduce errors if not being well designed, and would not be necessarily reduce computing time.

Any ideas?

michael barkhudarov January 9, 2009 01:22

Re: Insufficent memory

Can the perforated baffles be represented with the porous baffle model with appropriate flow losses to avoid resolving the small openings?


All times are GMT -4. The time now is 18:39.