Domain interface at different area

CFXer · July 21, 2020, 07:33

Hello!

I want to simulate an expansion pipe area.

When setting a domain interface at a different area(like the picture), can I use the normal GGI progress?
Does anybody know how the CFX domain interface operates when the domain interface at expansion or contraction areas?

Thank you in advance.

ghorrocks · July 21, 2020, 07:57

You could use a GGI to attach the cylinder to the rest of the domain. It will make meshing easier. Check the "Non overlap" conditions, you want to make sure areas of the GGI which do not overlap are treated as a non-slip wall.

But I would not use a GGI in this case if I was you. This geometry should be simple to mesh as a single domain with no GGI, and this will make a contiguous mesh through the joint. As there is likely to be a complex flow field at the joint you will want to resolution given by a contiguous mesh. A GGI interface is an interpolating interface, so cannot resolve the flow as accurately. It is preferable to put GGI interfaces in areas where the flow is simpler.

Stel · July 21, 2020, 08:39

Glenn is absolutely right, this geometry seems to be simple enough so that you will be able to mesh them together, thus making the cross-section area contiguous within the volume mesh.

If for some reason you still need to mesh both geometries separately, so what I suggest you is to create and mesh a circular surface boundary the same size as the pipe cross-section area at the wall of the bigger volume, just to be assigned as the side 2 of the interface (instead of the whole bigger area). This will considerably reduce the GGI error since the interface connection will have a 1.000 area ratio.

CFXer · July 21, 2020, 09:32

Thank you, Glenn and Stel.

Although I did not show you the whole geometry, there are other geometries connecting with one surface and 101 other surfaces.
So I wanted to know how to interface the whole surfaces.
And reply to Stel's post, I have a lot of difficulties in creating a circular surface on the interface surface because I'm using ICEM CFD for meshing.
And I have one more question.

Until now, I did not check the "Non overlap" condition, but the result shows zero velocity at wall. Is it possible to consider it as an appropriate result?

Stel · July 21, 2020, 10:43

This is not that difficult in ICEM. What you have to do is to create a circle on the surface, and then you Segment the surface using as a criteria the circle you have just created (Geometry tab > Surface). This will divide the whole surface you have now into two, the circular one and the remaining surface around it. Then assign the circular surface as a new part and create the mesh. If it is a structured mesh don't forget to split blocks accordingly and associate the related block face to the new surface. If it is an unstructured mesh than ICEM will automatically mesh it for you.

July 21, 2020, 07:57		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,872 Rep Power: 144	You could use a GGI to attach the cylinder to the rest of the domain. It will make meshing easier. Check the "Non overlap" conditions, you want to make sure areas of the GGI which do not overlap are treated as a non-slip wall. But I would not use a GGI in this case if I was you. This geometry should be simple to mesh as a single domain with no GGI, and this will make a contiguous mesh through the joint. As there is likely to be a complex flow field at the joint you will want to resolution given by a contiguous mesh. A GGI interface is an interpolating interface, so cannot resolve the flow as accurately. It is preferable to put GGI interfaces in areas where the flow is simpler. __________________ 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
Radiation in semi-transparent media with surface-to-surface model?	mpeppels	CFX	11	August 22, 2019 08:30
Periodic Pressure drop	cfd_begin	CFX	10	May 25, 2017 08:09
Rotating-Stationary Domain Interface Error	akalopsis	CFX	11	June 17, 2014 13:39
convergenceof natural convection prob. in cfx	cpkewat	CFX	15	January 31, 2014 07:29
Question about heat transfer coefficient setting for CFX	Anna Tian	CFX	1	June 16, 2013 07:28

July 21, 2020, 08:39		#3
Stel Member Henrique Stel Join Date: Apr 2009 Location: Curitiba, Brazil Posts: 93 Rep Power: 17	Glenn is absolutely right, this geometry seems to be simple enough so that you will be able to mesh them together, thus making the cross-section area contiguous within the volume mesh. If for some reason you still need to mesh both geometries separately, so what I suggest you is to create and mesh a circular surface boundary the same size as the pipe cross-section area at the wall of the bigger volume, just to be assigned as the side 2 of the interface (instead of the whole bigger area). This will considerably reduce the GGI error since the interface connection will have a 1.000 area ratio.

July 21, 2020, 09:32		#4
CFXer Member William Join Date: Jun 2020 Posts: 70 Rep Power: 6	Thank you, Glenn and Stel. Although I did not show you the whole geometry, there are other geometries connecting with one surface and 101 other surfaces. So I wanted to know how to interface the whole surfaces. And reply to Stel's post, I have a lot of difficulties in creating a circular surface on the interface surface because I'm using ICEM CFD for meshing. And I have one more question. Until now, I did not check the "Non overlap" condition, but the result shows zero velocity at wall. Is it possible to consider it as an appropriate result?

July 21, 2020, 10:43		#5
Stel Member Henrique Stel Join Date: Apr 2009 Location: Curitiba, Brazil Posts: 93 Rep Power: 17	This is not that difficult in ICEM. What you have to do is to create a circle on the surface, and then you Segment the surface using as a criteria the circle you have just created (Geometry tab > Surface). This will divide the whole surface you have now into two, the circular one and the remaining surface around it. Then assign the circular surface as a new part and create the mesh. If it is a structured mesh don't forget to split blocks accordingly and associate the related block face to the new surface. If it is an unstructured mesh than ICEM will automatically mesh it for you.