[snappyHexMesh] A DMZ for snappyHexMesh

ConnClark · September 19, 2012, 20:19

Hello,

I have been meshing with some large meshes (at least for my laptop) and have about 6GB of data swapped out before snappyHexMesh starts eliminated mesh cells that are in enclosed surfaces. Is there a way to designate a Don't Mesh Zone to tell sHM not to bother with refining or meshing in volume before it gets to the point of eliminating cells in an enclosed surface? (if not it would be a nice feature to save time meshing )

Thank You

Conn

SirWombat · September 20, 2012, 04:23

Here's an idea: you could use a "coarse" snappyHexMeshDict with a smaller surface than the original one. Then use your "fine" snappyHexMeshDict on that mesh. This will only work, if all of the "fine" surfaces is still inside a mesh region. Otherwise snappy won't snap at all.

SirWombat · September 20, 2012, 04:29

just tried it with my setup, 'cause I really like the idea of saving memory and time!!

for the coarse mesh (snappyHexMeshDict_coarse):

Code:

castellatedMesh true;
snap            false;
addLayers       false;

then turn all your surface refinements to "0" and delete all refinementRegions.

for the fine mesh (snappyHexMeshDict_fine):
just take you original snappyHexMeshDict.

I used the following script "fastermesh.sh"

Code:

#!/bin/sh
cp system/snappyHexMeshDict_coarse system/snappyHexMeshDict
blockMesh
snappyHexMesh -overwrite
cp system/snappyHexMeshDict_fine system/snappyHexMeshDict
snappyHexMesh -overwrite

Seems to work very well in my case.

Greetings,
Jan

ConnClark · September 20, 2012, 18:56

Thank You, That helped a lot!!

Conn

ConnClark · September 21, 2012, 12:58

Oops,

I spoke too soon. After looking at the simulation data something wasn't right. The Mercedes I was simulating has been turned into a rice-boy neo-cubist car.

The coefficient of lift is an impressive value of 1.4 . I don't think anybody has achieved that in auto design before.

back to swapping to disk.

September 19, 2012, 20:19	A DMZ for snappyHexMesh	#1
ConnClark New Member Conn Clark Join Date: Dec 2010 Posts: 11 Rep Power: 15	Hello, I have been meshing with some large meshes (at least for my laptop) and have about 6GB of data swapped out before snappyHexMesh starts eliminated mesh cells that are in enclosed surfaces. Is there a way to designate a Don't Mesh Zone to tell sHM not to bother with refining or meshing in volume before it gets to the point of eliminating cells in an enclosed surface? (if not it would be a nice feature to save time meshing ) Thank You Conn

September 20, 2012, 04:23		#2
SirWombat Member Jan Join Date: Dec 2009 Location: Berlin Posts: 50 Rep Power: 19	Here's an idea: you could use a "coarse" snappyHexMeshDict with a smaller surface than the original one. Then use your "fine" snappyHexMeshDict on that mesh. This will only work, if all of the "fine" surfaces is still inside a mesh region. Otherwise snappy won't snap at all. ConnClark likes this. __________________ ~~~_/)~~~

September 20, 2012, 04:29		#3
SirWombat Member Jan Join Date: Dec 2009 Location: Berlin Posts: 50 Rep Power: 19	just tried it with my setup, 'cause I really like the idea of saving memory and time!! for the coarse mesh (snappyHexMeshDict_coarse): Code: castellatedMesh true; snap false; addLayers false; then turn all your surface refinements to "0" and delete all refinementRegions. for the fine mesh (snappyHexMeshDict_fine): just take you original snappyHexMeshDict. I used the following script "fastermesh.sh" Code: #!/bin/sh cp system/snappyHexMeshDict_coarse system/snappyHexMeshDict blockMesh snappyHexMesh -overwrite cp system/snappyHexMeshDict_fine system/snappyHexMeshDict snappyHexMesh -overwrite Seems to work very well in my case. Greetings, Jan ConnClark likes this. __________________ ~~~_/)~~~

September 20, 2012, 18:56		#4
ConnClark New Member Conn Clark Join Date: Dec 2010 Posts: 11 Rep Power: 15	Thank You, That helped a lot!! Conn

September 21, 2012, 12:58		#5
ConnClark New Member Conn Clark Join Date: Dec 2010 Posts: 11 Rep Power: 15	Oops, I spoke too soon. After looking at the simulation data something wasn't right. The Mercedes I was simulating has been turned into a rice-boy neo-cubist car. The coefficient of lift is an impressive value of 1.4 . I don't think anybody has achieved that in auto design before. back to swapping to disk.