CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM Running, Solving & CFD (http://www.cfd-online.com/Forums/openfoam-solving/)
-   -   How to combine mixer-Ggi with translational motion (http://www.cfd-online.com/Forums/openfoam-solving/93614-how-combine-mixer-ggi-translational-motion.html)

mneben October 20, 2011 10:08

How to combine mixer-Ggi with translational motion
 
1 Attachment(s)
Hi OpenFoamers,

I'd like to know if it is possible to combine a topological mesh manipulation with cell stretch/morphing meshes.
For example the mixer-ggi-tutorial case: The standard case is with rotating motion of the inner mesh and the outer mesh is static.
Now I'd like to move the whole inner part (which still rotates) and the cells of the outer (static) part have to stretch/shrink (the cell sizes of the inner rotating mesh stays constant).

Can you tell me how can I solve this problem?

SailorLiu January 29, 2015 13:29

Hi Matthias,

After almost four years, have you finally solved this problem? I am working on an exactly same question as yours. Could you kindly give me some hints?

Best wishes,

Yuanchuan

Quote:

Originally Posted by mneben (Post 328753)
Hi OpenFoamers,

I'd like to know if it is possible to combine a topological mesh manipulation with cell stretch/morphing meshes.
For example the mixer-ggi-tutorial case: The standard case is with rotating motion of the inner mesh and the outer mesh is static.
Now I'd like to move the whole inner part (which still rotates) and the cells of the outer (static) part have to stretch/shrink (the cell sizes of the inner rotating mesh stays constant).

Can you tell me how can I solve this problem?


mneben January 30, 2015 05:02

Hello Yuanchuan,

Yes you are right, I finally solved this problem. Depending on your concrete scope I can show you two solutions.
Both solutions work with OF 2.3 and in both cases I use the GGI technique (which is called AMI in OF23).
The two solutions combine rotation and translation in a similar way, but the first one works with lagrangian smoothing solver and the second one uses an analytic mesh deformation technique.
With the first solution I’ve never reached an eccentricity more than 70%, but I was simulating a journal bearing so I needed much more (like 95%) so I developed the second one.
I hope this helps.

Greetings

Matthias

SailorLiu January 30, 2015 07:12

Hi Matthias,

Thank you very much for your reply and also confirming this could be done!

My project is to simulate floating wind turbines like this:

http://newsimg.bbc.co.uk/media/image...873795_-18.jpg

Since the upper turbine will revolve, I guess I will have to adopt AMI or GGI. In the meanwhile, due to the movement of its supporting platform in waves, the while body will translate and rotate as well. That is why I want to combine solid body motion of the turbine using AMI or GGI and mesh deformation of the platform using dynamic mesh technique.

I am not familiar with the "lagrangian smoothing solver" in your first way. As to the "analytic mesh deformation technique", do you mean the the same technique I mentioned above i.e., using the dynamicFvMesh library?

For now, I am thinking about creating a new library which inherits from sixDoFRigidBodyMotionSolver but adds some functions like solidBodyMotion. Unfortunately I have barely started. It will be really helpful if you can kindly shed some light upon how you achieve it.

By the way, do you think it is OK to think of AMI or GGI as just a boundary condition like fixedValue rather than something related to mesh motion? As far as I know, they are kind of like an interpolation method. So as long as I can make sure the two interfaces always overlap with each other, it should be all right no matter how the mesh moves. Please correct me if I am wrong.

Thanks again!

Best wishes,

Yuanchuan

Quote:

Originally Posted by mneben (Post 529640)
Hello Yuanchuan,

Yes you are right, I finally solved this problem. Depending on your concrete scope I can show you two solutions.
Both solutions work with OF 2.3 and in both cases I use the GGI technique (which is called AMI in OF23).
The two solutions combine rotation and translation in a similar way, but the first one works with lagrangian smoothing solver and the second one uses an analytic mesh deformation technique.
With the first solution Iíve never reached an eccentricity more than 70%, but I was simulating a journal bearing so I needed much more (like 95%) so I developed the second one.
I hope this helps.

Greetings

Matthias


mneben February 9, 2015 10:00

Hello Yuanchuan,

laplacian smoothing for mesh is a technique which is described here http://www.google.de/url?sa=t&rct=j&...85464276,d.d24

Additionally I implemented an analytic solution of the point displacement for a concentric TC-system with variable eccentricity.
Due to the fact that the movement in comparison to my case is 3D I recommend using the mesh motion via solving a Laplace equation with variable diffusivity.

As you already mentioned you should combine the AMI or GGI algorithm with a six-degree-of-freedom-solver, so you first have to think about an appropriate algorithm that combines these two motions. Therefore you have to take care, that the inner and outer sliders of the interface always stays concentric.



Greetings

Matthias

SailorLiu February 9, 2015 11:02

Hi Matthias,

Thanks again for the paper and also guiding me through this. I always use the quadratic model for diffusivity and it seems rather robust so I guess I will stick with it. You are totally right that the biggest challenge lies in how to maintain the concentricity for the inner and outer interface.

My preliminary thought is to first define a subset of the mesh which covers both the interface around the turbine and the floating structure supporting the turbine. This subset of the mesh will exhibit the same 6DoF motion as imposed on the whole structure using the Laplacian smoothing method to diffuse the point displacement to the region outside it while the topology inside it will remain unchanged. Then AMI or GGI will be called to rotate the turbine, or specifically the inner interface region along with the turbine will move using the solidBodyMotionFunction.

This is quite a project for me and I guess for now I need to at least understand the dynamicFvMesh class as well as its inheritance diagram (mainly dynamicMotionSolverFvMesh and solidBodyMotionFvMesh) before I start. Do you happen to have or know any other relevant or helpful materials I can take advantage of to help me understand these classes better? Cheers.

Best wishes,

Yuanchuan Liu

mneben March 12, 2015 05:52

You know this one (Mesh motion alternatives in OpenFOAM)?
http://www.google.de/url?sa=t&rct=j&...87920726,d.ZWU

SailorLiu March 12, 2015 06:12

Hi Matthias,

Thank you so much for showing me this reference! I will definitely read it carefully.

Best wishes,

Yuanchuan Liu


All times are GMT -4. The time now is 07:56.