Sliding interface- ACMI

khedar · January 15, 2017, 18:23

Because in the real geometry my inlet and outlet sections are not rectangular channels but they are somewhat elliptical...

i.e. In the first picture (which is a test case before simulating the real one) the inlet and outlet sections are rectangle which is a simplification of the real one.

http://www.pic-upload.de/view-32496605/case1.png.html

Inlet and Outlet sections for the real one are like shown in the second link.

http://www.pic-upload.de/view-324966...ction.png.html

This allows me two mesh the three parts separately without any dependence on the other. The reason is I would need an O-grid mesh for the circular inlet and outlet section to get a quality mesh, whereas for the main channel i would like to use normal mesh.

Tobi · January 15, 2017, 18:30

Ah okay, got the point. However, I do not agree with your statement, that you would get not a good mesh if you mesh everything together. But that is up to you and your meshing procedure. However, your question does finally not fit into this thread.

khedar · January 15, 2017, 19:46

Why not. Is it not possible to use ACMI for non-sliding problems when the mesh has non-overlapping patches?

arjun · January 16, 2017, 01:24

Quote:

Originally Posted by khedar

Why not. Is it not possible to use ACMI for non-sliding problems when the mesh has non-overlapping patches?

rotate the mesh little bit arbitrarily and check if it runs now.

Sometimes the neighbour search fails if the patch is flat along one of the coordinate. Once the interface is calculated it seems like simple enough problem to run.

If the intersection took place properly then the problems could come up due to skew. At the interface the skew of newly generated cells could be extremely bad. So start the calculation with first order schemes and more non orthoginal correctors. Then later upon convergence change the options to higher order terms.

Hope it helps.

khedar · January 21, 2017, 20:30

thanks arjun, The problem might have been with ACMI patch. I found the ACMI implementation of 4.x to be the most recent and I tested with that version. The simulation ran for 10-12 iterations before blowing up. But at least it was running. Second change what I did was the Velocity internal field initialised to 0 instead of inlet velocity earlier. This did the trick and the simulation ran flawlessly. Also I ran the simulation for first few iterations using PCG solver for p_rgh equation and then switched to GAMG for better stability. Now everything is running perfectly.

There might also be a problem with mergeMeshes/createBaffles, since after i run these in 2.3 i was getting open boundaries but in 4.x it was okay. So basically i have switched to 4.x now.

January 15, 2017, 18:30		#42
Tobi Super Moderator Tobias Holzmann Join Date: Oct 2010 Location: Bad Wörishofen Posts: 2,711 Blog Entries: 6 Rep Power: 52	Ah okay, got the point. However, I do not agree with your statement, that you would get not a good mesh if you mesh everything together. But that is up to you and your meshing procedure. However, your question does finally not fit into this thread. __________________ Keep foaming, Tobias Holzmann

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Wind turbine simulation	Saturn	CFX	60	July 17, 2024 06:45
Question about heat transfer coefficient setting for CFX	Anna Tian	CFX	1	June 16, 2013 07:28
An error has occurred in cfx5solve:	volo87	CFX	5	June 14, 2013 18:44
Error finding variable "THERMX"	sunilpatil	CFX	8	April 26, 2013 08:00
RPM in Wind Turbine	Pankaj	CFX	9	November 23, 2009 05:05

January 15, 2017, 18:23		#41
khedar Senior Member khedar Join Date: Oct 2016 Posts: 111 Rep Power: 10	Because in the real geometry my inlet and outlet sections are not rectangular channels but they are somewhat elliptical... i.e. In the first picture (which is a test case before simulating the real one) the inlet and outlet sections are rectangle which is a simplification of the real one. http://www.pic-upload.de/view-32496605/case1.png.html Inlet and Outlet sections for the real one are like shown in the second link. http://www.pic-upload.de/view-324966...ction.png.html This allows me two mesh the three parts separately without any dependence on the other. The reason is I would need an O-grid mesh for the circular inlet and outlet section to get a quality mesh, whereas for the main channel i would like to use normal mesh.

January 15, 2017, 19:46		#43
khedar Senior Member khedar Join Date: Oct 2016 Posts: 111 Rep Power: 10	Why not. Is it not possible to use ACMI for non-sliding problems when the mesh has non-overlapping patches?

January 21, 2017, 20:30		#45
khedar Senior Member khedar Join Date: Oct 2016 Posts: 111 Rep Power: 10	thanks arjun, The problem might have been with ACMI patch. I found the ACMI implementation of 4.x to be the most recent and I tested with that version. The simulation ran for 10-12 iterations before blowing up. But at least it was running. Second change what I did was the Velocity internal field initialised to 0 instead of inlet velocity earlier. This did the trick and the simulation ran flawlessly. Also I ran the simulation for first few iterations using PCG solver for p_rgh equation and then switched to GAMG for better stability. Now everything is running perfectly. There might also be a problem with mergeMeshes/createBaffles, since after i run these in 2.3 i was getting open boundaries but in 4.x it was okay. So basically i have switched to 4.x now.