Adaptive mesh refinement in two-phase flow (Wigley Hull)

jantheron · October 5, 2011, 13:19

I have been able to run the Wigley hull test case and would now like to improve the quality of the solution by better resolving the fluid-air interface.

Does anybody have experience with setting up adaptive grid refinement for such a case (using the interface alpha1 value as adaptive parameter)?

Thank you!

Dr. Johannes N Theron
www.kanoefabrik.com

daveatstyacht · October 6, 2011, 13:33

Johannes,
While dynamic mesh refinement is readily achievable using alpha as the refinement parameter (I there is a dam break tutorial that does just that under interdymfoam tutorials), it is ridiculously slow to remesh. In the case of a free surface problem where the wave height isn't too large I would recommend just using a refinement zone to increase the resolution of the interface and save the huge overhead of the mesh refinement process as it will take longer to refine than to just run a denser mesh to begin with. I can think of only certain cases such as bubble columns and tank sloshing where the mesh refinement process makes sense compared the use of a refinement zone approach.
Regards,
Dave

jantheron · October 6, 2011, 13:41

Dave

Thanks for your reply. I was not aware AMR was that slow. I already have a fine grid across the interface but at the moment it is resolved over two or three cells and when there are waves it goes into the coarser region. Perhaps I'll just add more buffer cells.

Thanks again.

Jan

daveatstyacht · October 6, 2011, 16:15

Johannes,
I only ran the tutorial with a refinement at every time step (ridiculously slow). I suppose it is conceivable to refine far less frequently. I stopped investigating the refinement capability because of the need to include mesh motion for vessel motions and the apparent incompatibility of the adaptive refinement with mesh motion (I would love to be mistaken about it).
Dave

jantheron · October 6, 2011, 16:52

Dave

Hmmm. That is a shame. It would be really great to have an overset grid capability for OpenFOAM so one can easily do arbitrary body motions using a fixed body-fitted grid around the vessel and at the same time do grid refinement in the areas of the background grid where the ship is located. Gerris has very nice AMR, but they use Cartesian grids that are far easier to refine and coarsen.

Have you done any 6DOF work on floating bodies?

Jan

daveatstyacht · October 6, 2011, 19:55

Yes, though interDyMFoam becomes increasingly unstable as the mesh resolution is increased. Better success is achievable with shipFoam as it is more stable. Consider joining the OF ship hydrodynamics group: http://www.cfd-online.com/Forums/gro...mechanics.html

jantheron · October 7, 2011, 13:59

I did, but it seems their 1.7 version was too unstable and it was retracted. The latest version is 1.6 and I'm working on 2.0.1. Perhaps I'll try to incorporate 1.6 ideas into 2.0.1.... Lots of work though. I'm under no illusions. Thanks for the pointer!

October 5, 2011, 13:19	Adaptive mesh refinement in two-phase flow (Wigley Hull)	#1
jantheron New Member Johannes N Theron Join Date: Feb 2010 Location: Hamburg Posts: 25 Rep Power: 16	I have been able to run the Wigley hull test case and would now like to improve the quality of the solution by better resolving the fluid-air interface. Does anybody have experience with setting up adaptive grid refinement for such a case (using the interface alpha1 value as adaptive parameter)? Thank you! Dr. Johannes N Theron www.kanoefabrik.com

October 6, 2011, 16:15		#4
daveatstyacht Senior Member Dave Join Date: Jul 2010 Posts: 100 Rep Power: 16	Johannes, I only ran the tutorial with a refinement at every time step (ridiculously slow). I suppose it is conceivable to refine far less frequently. I stopped investigating the refinement capability because of the need to include mesh motion for vessel motions and the apparent incompatibility of the adaptive refinement with mesh motion (I would love to be mistaken about it). Dave Ralph M likes this.

October 6, 2011, 19:55		#6
daveatstyacht Senior Member Dave Join Date: Jul 2010 Posts: 100 Rep Power: 16	Yes, though interDyMFoam becomes increasingly unstable as the mesh resolution is increased. Better success is achievable with shipFoam as it is more stable. Consider joining the OF ship hydrodynamics group: http://www.cfd-online.com/Forums/gro...mechanics.html Ralph M likes this.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
mass flow in is not equal to mass flow out	saii	CFX	12	March 19, 2018 06:21
[snappyHexMesh] Boundary layer in a pipe	Clementhuon	OpenFOAM Meshing & Mesh Conversion	6	March 12, 2012 13:41
[snappyHexMesh] non-smooth mesh	Svensson	OpenFOAM Meshing & Mesh Conversion	11	January 18, 2012 10:13
adaptive mesh refinement in STARCCM+	abdullahkarimi	Siemens	1	November 16, 2010 19:26
Mesh for 3 dim Geometry	Phil	FLUENT	9	July 12, 2000 05:39

October 6, 2011, 13:33		#2
daveatstyacht Senior Member Dave Join Date: Jul 2010 Posts: 100 Rep Power: 16	Johannes, While dynamic mesh refinement is readily achievable using alpha as the refinement parameter (I there is a dam break tutorial that does just that under interdymfoam tutorials), it is ridiculously slow to remesh. In the case of a free surface problem where the wave height isn't too large I would recommend just using a refinement zone to increase the resolution of the interface and save the huge overhead of the mesh refinement process as it will take longer to refine than to just run a denser mesh to begin with. I can think of only certain cases such as bubble columns and tank sloshing where the mesh refinement process makes sense compared the use of a refinement zone approach. Regards, Dave

October 6, 2011, 13:41		#3
jantheron New Member Johannes N Theron Join Date: Feb 2010 Location: Hamburg Posts: 25 Rep Power: 16	Dave Thanks for your reply. I was not aware AMR was that slow. I already have a fine grid across the interface but at the moment it is resolved over two or three cells and when there are waves it goes into the coarser region. Perhaps I'll just add more buffer cells. Thanks again. Jan

October 6, 2011, 16:52		#5
jantheron New Member Johannes N Theron Join Date: Feb 2010 Location: Hamburg Posts: 25 Rep Power: 16	Dave Hmmm. That is a shame. It would be really great to have an overset grid capability for OpenFOAM so one can easily do arbitrary body motions using a fixed body-fitted grid around the vessel and at the same time do grid refinement in the areas of the background grid where the ship is located. Gerris has very nice AMR, but they use Cartesian grids that are far easier to refine and coarsen. Have you done any 6DOF work on floating bodies? Jan

October 7, 2011, 13:59		#7
jantheron New Member Johannes N Theron Join Date: Feb 2010 Location: Hamburg Posts: 25 Rep Power: 16	I did, but it seems their 1.7 version was too unstable and it was retracted. The latest version is 1.6 and I'm working on 2.0.1. Perhaps I'll try to incorporate 1.6 ideas into 2.0.1.... Lots of work though. I'm under no illusions. Thanks for the pointer!