CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Phoenics (http://www.cfd-online.com/Forums/phoenics/)
-   -   fine grid volume (http://www.cfd-online.com/Forums/phoenics/51850-fine-grid-volume.html)

timur January 24, 2005 14:23

fine grid volume
 
Hello. Sorry my english. I try to use Fine Grid object in my calculation (in block regions). But earth send message (in result file): solvfg is in use for indvar= 149 lb index= 642510

nphi4 = 600

block-location index too big in gcslwl

execution aborted

error stop How can I find out the solution of this problem? Can anybody help me? Thanks in advance.

Mick Hughes February 3, 2005 12:50

Re: fine grid volume
 
I'm not sure what the problem is ! I wonder if you can simplify your problem. Try using a Null object instead of a Fine grid volume and refining the mesh density in the Null object region. Does the problem go away?

Try toggling various options to try and hone in on "the cause" . For Example if youre turb model uses LTLS try standard KE or running laminar.


timur February 3, 2005 17:52

Re: fine grid volume
 
Thank's for reply, Mike. I'm trying to use standart KE model and it variants but I couldn't obtain the convergence with FG. In the end I putting down the FG and refine grid without it.

Mick Hughes February 4, 2005 08:37

Re: fine grid volume
 
Wise decision

timur February 4, 2005 16:20

Re: fine grid volume
 
Thanks. Mike, can I ask you 1 question? In more of my problems FG (think not in my) doesn't work but very desirable. Can I hope that in Phoenics x.xx version FG will be work good (or almost good)? If ansver - yes, can I tell you - when? Thank you in advance.

Mick Hughes February 8, 2005 08:03

Re: fine grid volume
 
Yes, I'm not sure Timur, I think its pencilled in the pipe line for some work in the future, however I know that there are lots of other pressing matters with higher priority and limited resourses, so I am sorry I can not give you a straight answer.

One of the major problems with looking at FGV cases is that it is hard to find a simple case that is clearly wrong. With many cases some slight changing of conditions or reducing mesh density can cause the case to run or the results to seem realistic, so debugging is a problem.

If you have such a simple case it could be useful.


PattiMichelle April 3, 2005 14:45

Re: fine grid volume
 
Yes, I've noted problems when introducing a FGE - usually you have to relax things much more. Pressure mostly. Also, try turning on Automatic Convergence Control.

PattiMichelle April 3, 2005 14:47

Re: fine grid volume
 
I've also used nonuniform grids, with the fineness adjusted so that the fine regions coincide near objects that need grid refinement - that sometimes precludes the need for FineGridVolumes. You can make null objects which affect the grid to make the locations for your nonuniform grids.

timur April 4, 2005 02:50

Re: fine grid volume
 
Thank you Patti. Now I trying to use Null obj for grid refining. What about FGO and relaxation (and SARAH) of pressure etc - it does not helps me.

PattiMichelle April 4, 2005 15:04

Re: fine grid volume
 
I have trouble with FineGrids sometimes and SARAH isn't terribly helpful. If you have version 3.6 there is an Automatic Convergence Control which seems to be very helpful in many cases. You just have to start adding relaxation, especially of pressures and densities (especially for gases).

timur April 5, 2005 04:50

Re: fine grid volume
 
I have v. 3.6 with limitation of f-array 3 000 000 elements (32 Mb RAM). And I have v.3.5 without any limitation. So more of my problems I solve with v. 3.5 (unfortunately)

Patti Sheaffer April 7, 2005 15:03

Re: fine grid volume
 
Get More RAM

timur April 8, 2005 02:35

Re: fine grid volume
 
No, I have limitation version 3.6 (no more than 32 MB). My computer has 512 MB RAM.

Kevin Chow April 8, 2005 03:14

Re: fine grid volume
 
I'm just guessing here, but have you looked in one of the .ini files (Phoenics.ini?) - there might be a flag that you can change to increase the memory allocation space.

timur April 8, 2005 06:12

Re: fine grid volume
 
I bought ver 3.6 with limitations: Number of elements of f-array in my ver. Phoenics 3.6 - 3 000 000. That means I can't use more then 32 MB RAM. If I want to use more than 3 000 000 elements of f-array I have to recompile Earth, but my ver. non recompilible... And what is more - in systemdisc/phoenics/d_earth locations 1 file earth.exe (no earth.64, earth.256, earth.512 ...).

PattiMichelle May 3, 2005 13:09

Re: fine grid volume (Hi CHAM!)
 
Maybe CHAM would like to pipe in here??

mick August 21, 2005 05:23

Re: fine grid volume (Hi CHAM!)
 
I think kevin had a good idea. with V3.6 (PC) you could try and increase the EARTh size in the file \phoenics\d_allpro\cham.ini

Try experimentally increasing this timur.

[F-array] Satellite = 2000000 Earth = 10000000 <----- Photon = 100000 Pinto = 100000

Patti Sheaffer August 22, 2005 13:56

Re: fine grid volume (Hi CHAM!)
 
I wanted to mention that I had a problem with memory in 3.6. I received an error message from EARTH telling me I might want to increase swapfile size (however it's already at Win2000's max, 4 GIG) or else, as Mick suggested, the allocation in cham.ini. But this didn't work. I'm wondering if perhaps it didn't work because I only increased EARTH's allocation? i.e., [F-array] Satellite = 2000000 Earth = 200000000 Photon = 100000 Pinto = 100000



All times are GMT -4. The time now is 11:35.