|
[Sponsors] |
November 12, 2004, 22:27 |
GRID refinement
|
#1 |
Guest
Posts: n/a
|
I'm working with PHOENICS 3.5.1 and I'd like to know what is the best way to refine the grid in only one domain region without influencing the other regions in the same axis.
What about the NULL object? Have you ever worked with it? Thanks Sileno |
|
November 13, 2004, 11:06 |
Re: GRID refinement
|
#2 |
Guest
Posts: n/a
|
Sileno,
Create a FineGrid object instead of null. Then double click it to set how many times in each direction you want to refine the mesh. Null objects will a cross od refinement what will lead to a mesh waste. If it does not bother you, it is possible to use it anyway. Regards, |
|
November 15, 2004, 10:54 |
Re: GRID refinement
|
#3 |
Guest
Posts: n/a
|
Dear Carlos,
How can I set the fine grid volume? Do I have to create a new object exactly where my model is or the model itself must be converted in a fine grid volume? Thanks. |
|
November 17, 2004, 22:36 |
Re: GRID refinement
|
#4 |
Guest
Posts: n/a
|
A fine grid volument is a type of object that can be created. The following assumes you are using the VR Editor
First create an object. Double click the object to bring up the propoerties manu. Change the object type from blockage to fine grid volume. Next select the attributes button and enter the number of refinments you need in each of the x,y,z directs. What this does is divide each of the cells with the fine grid volume by the number selected. Hope this helps, Leon |
|
November 19, 2004, 17:07 |
Re: GRID refinement
|
#5 |
Guest
Posts: n/a
|
I've been working quite easily with fine grid volume, but I noticed that the refinement works until certain limit and then the PHOENICS gives me an error message saying the following:
F-array too small. If your computer has sufficient memory, please increase NFDIM in d_earth\main.htm, which is currently nfdim= 33600000 by, say, 50%, and try again. How much is needed is not yet known. error stop Is it really a memory problem? My Pc is Petium4 with 1Gb of RAM memory. Thanks in advance. |
|
November 19, 2004, 22:43 |
Re: GRID refinement
|
#6 |
Guest
Posts: n/a
|
Sileno,
F array is a fixed size array. It does not change size with the amount of ram you have in your machine. To adjust look in POLIS / Encylopedia. From memory the easiest option is to copy a particluar file, CHAM has devoped, suited for the 1GB of ram. The F array limts the number of cells that can be modelled in any given domain. So if the number of cells either by fine grids or by other methods exceeds the limit then you will get this error. Hope this helps, Leon |
|
November 20, 2004, 02:37 |
Re: GRID refinement
|
#7 |
Guest
Posts: n/a
|
Hi, I do faintly remember coming across separate earexe.exe for diff RAM sizes. But for update sake, can you please let us know where & how to access/copy such files ? Thanks in anticipation.
|
|
November 27, 2004, 09:41 |
Re: GRID refinement
|
#8 |
Guest
Posts: n/a
|
Hi,
To increase NFDIM you must go to main.for. Open VRE, choose file>open file> main. Change NFIDM (use a number bigger than the suggested by PHOENICS) and close it. Then you must recompile your code (open build>earexe). Is your code recompilable? If it is not, ask your technical support to recompile an earexe and a satexe for you with this new NFDIM. But just a question Sileno: do you really need a mesh so refined? You will have a very heavy simulation... Just do it if it is really necessary, ok? Hope this helps, |
|
November 27, 2004, 11:54 |
Re: GRID refinement
|
#9 |
Guest
Posts: n/a
|
I really need a very refined mesh since I'm trying to study the influence of different architectural components of a certain type of building in the internal wind flow.
Related to the NFDIM increasing I managed to set it properly but unfortunately I got a new error message saying: lb index= 10559846 nphi4 = 600 block-location index too big in GXKESO execution aborted Could you help again? Thanks in advance. |
|
May 3, 2005, 17:26 |
Re: GRID refinement
|
#10 |
Guest
Posts: n/a
|
I have done nonlinearly spaced grids and null objects near objects in the domain that need finer grids. You can set the grid elements to get much finer near your buildings, and much coarser further away. Play around with setting the Power/ratio in the Distribution boxes in the lower half of the Grid Mesh settings dialog box. You can cause them to pile up near interesting regions. I have found it at least as useful as a Fine Grid Volume, and more numerically stable.
|
|
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
Grid refinement | shuo | Main CFD Forum | 4 | October 7, 2008 05:35 |
Grid Refinement | Yamuna | FLUENT | 0 | September 9, 2008 14:24 |
Grid refinement. | Sas | Main CFD Forum | 8 | September 2, 2005 09:36 |
Grid refinement 2 | Sileno | Phoenics | 0 | December 7, 2004 19:31 |
grid refinement | Gaz | FLUENT | 0 | March 7, 2002 15:01 |