Ghost Nodes

ewissmach · September 9, 2022, 11:06

I am getting an error message saying:

*** warning: there are 9 u ghost nodes of component 1 shared by two flow regions in mesh block 1 ***
*** refine the mesh for accurate velocity at these ghost nodes ***

From the manual it is my understanding that ghost nodes are applicable in the immersed boundary method and work to make force and velocity head predictions more accurate at wall boundary conditions. This warning, however, makes me think that my ghost nodes are adding more error to the simulation. Obviously the message suggests I should refine the mesh, but I would like to understand the problem first and would prefer to not refine my mesh further if I can avoid it.

Does anyone have any input on this?

silveriver · September 14, 2022, 13:51

Just imagine that you are modeling a water tube submerged in a pool. Water flows inside and outside of the tube. So there are two separated flow regions and the flow field could be quite different in these two regions. If the mesh resolution for the wall of the tube is low, it is possible that there is only one ghost node at some locations of the wall. This would be difficult to choose a region to do velocity interpolation for the ghost node.

September 9, 2022, 11:06	Ghost Nodes	#1
ewissmach New Member Emilee Wissmach Join Date: Mar 2022 Posts: 7 Rep Power: 4	I am getting an error message saying: * warning: there are 9 u ghost nodes of component 1 shared by two flow regions in mesh block 1 * * refine the mesh for accurate velocity at these ghost nodes * From the manual it is my understanding that ghost nodes are applicable in the immersed boundary method and work to make force and velocity head predictions more accurate at wall boundary conditions. This warning, however, makes me think that my ghost nodes are adding more error to the simulation. Obviously the message suggests I should refine the mesh, but I would like to understand the problem first and would prefer to not refine my mesh further if I can avoid it. Does anyone have any input on this?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
StarCCMS+ on AWS Parallel Cluster not distributing workload across multiple nodes	dwagoner	STAR-CCM+	3	May 25, 2021 03:39
Adding compute nodes - new nodes faster than existing ones?	SLC	Hardware	3	November 14, 2019 16:35
32 Cores Mini-Cluster, 2 Nodes vs. 4 Nodes performance	Courageous	Hardware	3	July 7, 2016 12:37
meshing F1 front wing	Steve	FLUENT	0	April 17, 2003 13:37
CFX4.3 -build analysis form	Chie Min	CFX	5	July 13, 2001 00:19

September 14, 2022, 13:51		#2
silveriver New Member Join Date: Sep 2022 Posts: 1 Rep Power: 0	Just imagine that you are modeling a water tube submerged in a pool. Water flows inside and outside of the tube. So there are two separated flow regions and the flow field could be quite different in these two regions. If the mesh resolution for the wall of the tube is low, it is possible that there is only one ghost node at some locations of the wall. This would be difficult to choose a region to do velocity interpolation for the ghost node.