CFX Polyhedral elements vs Other meshes (Fluent, Star CCM, etc.)

u2berggeist · May 21, 2018, 11:58

So as discussed here, CFX essentially interprets whatever mesh it's given into polyhedral control volumes.

Is there any appreciable difference between this (letting CFX "create" a polyhedral mesh) and making a Polyhedral mesh and solving it with a code that can actually use it (Fluent, Star CCM, etc.)?

The only major thing I can think of is that starting with a polyhedral mesh gives you more direct control over the definition of the control volumes themselves?? I'm not sure what you'd do with that more direct control...

Second, but kinda related question, when creating a tet mesh for CFX (who interprets the mesh into polyhedral control volumes), is there anything in particular to look out for? Ie. any mesh quality parameters to look out for or any specific meshing methods that are preferable for this kind of application?

ghorrocks · May 21, 2018, 20:16

There is a LOT more to a CFD solver than the structure of the mesh it uses. You won't get the same results on different solvers even if you make the polyhedral meshes the same due to these myriad of other differences.

For tet meshes the issues are the same for any mesh - avoid small angles and high aspect ratios, and make mesh size transitions smooth.

u2berggeist · May 26, 2018, 16:01

Thanks! I'll be looking to actual numerical computation then I guess.

ghorrocks · May 27, 2018, 07:07

Sorry, I do not understand that comment.

u2berggeist · May 27, 2018, 12:09

Poor phrasing on my part. I meant to say "I guess I'll have to look into more detail on how CFD solvers implement their codes onto meshes (namely CFX in this case)"

ie. The generally touted benefits of a polyhedral mesh may not always be applicable, so I need to figure out more details as to why they're generally seen as quite good in their specific application as opposed to just applying those results onto any solver with a polyhedral mesh.

ghorrocks · May 28, 2018, 01:21

That is correct. Not only are there differences in the numerical approach meaning that some solvers work better in some cases than others, but then there is the other differences between software for things like usability, available physics models, quality of support and of course, price.

u2berggeist · May 28, 2018, 10:30

For sure. I'm mainly focused on CFX because that's what I'm intimately familiar with (as far as the general solving under workings and scripting goes), my school has a license for it so price isn't a concern, and I'm focusing on a turbo-esque research project, so I'll probably be taking advantage of the turbo tools inside CFX.

May 21, 2018, 11:58	CFX Polyhedral elements vs Other meshes (Fluent, Star CCM, etc.)	#1
u2berggeist Member James Wright Join Date: Oct 2015 Posts: 40 Rep Power: 11	So as discussed here, CFX essentially interprets whatever mesh it's given into polyhedral control volumes. Is there any appreciable difference between this (letting CFX "create" a polyhedral mesh) and making a Polyhedral mesh and solving it with a code that can actually use it (Fluent, Star CCM, etc.)? The only major thing I can think of is that starting with a polyhedral mesh gives you more direct control over the definition of the control volumes themselves?? I'm not sure what you'd do with that more direct control... Second, but kinda related question, when creating a tet mesh for CFX (who interprets the mesh into polyhedral control volumes), is there anything in particular to look out for? Ie. any mesh quality parameters to look out for or any specific meshing methods that are preferable for this kind of application?

May 21, 2018, 20:16		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,870 Rep Power: 144	There is a LOT more to a CFD solver than the structure of the mesh it uses. You won't get the same results on different solvers even if you make the polyhedral meshes the same due to these myriad of other differences. For tet meshes the issues are the same for any mesh - avoid small angles and high aspect ratios, and make mesh size transitions smooth. __________________ Note: I do not answer CFD questions by PM. CFD questions should be posted on the forum.

May 27, 2018, 07:07		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,870 Rep Power: 144	Sorry, I do not understand that comment. __________________ Note: I do not answer CFD questions by PM. CFD questions should be posted on the forum.

May 28, 2018, 01:21		#6
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,870 Rep Power: 144	That is correct. Not only are there differences in the numerical approach meaning that some solvers work better in some cases than others, but then there is the other differences between software for things like usability, available physics models, quality of support and of course, price. __________________ Note: I do not answer CFD questions by PM. CFD questions should be posted on the forum.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
[Commercial meshers] Using starToFoam	clo	OpenFOAM Meshing & Mesh Conversion	33	September 26, 2012 05:04
[Commercial meshers] Importing Fluent polyhedral meshes	fra76	OpenFOAM Meshing & Mesh Conversion	2	October 11, 2008 05:44
[Commercial meshers] Trimmed cell and embedded refinement mesh conversion issues	michele	OpenFOAM Meshing & Mesh Conversion	2	July 15, 2005 05:15
STAR, Fluent, CFX and conj. heat transfer	star-user	Main CFD Forum	8	January 21, 2003 01:07
Fluent vs. CFX vs. Star CD	Sam	Main CFD Forum	2	November 22, 2000 05:36

May 26, 2018, 16:01		#3
u2berggeist Member James Wright Join Date: Oct 2015 Posts: 40 Rep Power: 11	Thanks! I'll be looking to actual numerical computation then I guess.

May 27, 2018, 12:09		#5
u2berggeist Member James Wright Join Date: Oct 2015 Posts: 40 Rep Power: 11	Poor phrasing on my part. I meant to say "I guess I'll have to look into more detail on how CFD solvers implement their codes onto meshes (namely CFX in this case)" ie. The generally touted benefits of a polyhedral mesh may not always be applicable, so I need to figure out more details as to why they're generally seen as quite good in their specific application as opposed to just applying those results onto any solver with a polyhedral mesh.

May 28, 2018, 10:30		#7
u2berggeist Member James Wright Join Date: Oct 2015 Posts: 40 Rep Power: 11	For sure. I'm mainly focused on CFX because that's what I'm intimately familiar with (as far as the general solving under workings and scripting goes), my school has a license for it so price isn't a concern, and I'm focusing on a turbo-esque research project, so I'll probably be taking advantage of the turbo tools inside CFX.