Boundary conditions for dynamic mesh interface patch between two mesh regions
 

Hello all

I am trying to solve a problem with dynamic mesh.
Consider concentric cylinders, both having a oscillatory motion in opposite directions. But the far end of the domain in fixed.
I created two different meshed for both the cylinders and merged the meshed (just like oscillatingACMI tutorial).
The case runs fine accept that the quality of mesh degrades gradually because the node points at the ACMI patch remain fixed to the patch. Can anyone suggest an appropriate boundary condition at the patch so that the nodes slip at the patch.
I tried using fixedNormalSlip (used in potentialFreeSurfaceDyMFoam tutoria), but dose't solve the propose.

Hi

I am struggling to understand how this BC works in pointDisplacement for a 3D case. Any Help ?
For 2D case I figured out that the if the movement is in y axis either n as (1,0,0) or as (0,0,1) causes slip along y axis.
I have two concentric cylinders dynamicaly moving and I have an ACMI patch between them. How can I use this BC at the patch so that slip taked place in z axis on the patch.
How to modify the code for (r,theta) coordinate.

Greetings Virag and welcome to posting on the forum!

I've moved your second post from this thread: http://www.cfd-online.com/Forums/ope...ormalslip.html - mostly so that it's easier to keep track of your problem/requests for this case.

I've read your posts a few times and I'm still confused. Can you provide an image that shows what you're trying to mesh? I ask this, because it's possible that ACMI is not the correct way to solve your case.

Best regards,
Bruno

Hi

Thanks a lot for replying

I am attaching the mesh which would explain the case. So the case has two different bodies one or both oscillating along the z axis. And one is inside the other. The grid is a cut plane view. Plane passing through the centre of the bodies (XZ or YZ plane). I meshed the bodies in two separate mesh files and later merged the meshes and all other proceduers for ACMI like topoSet, creatBaffles, createPatch. I am looking for BC so that the node points on the patch (which is b/w two bodies) slid in the patch so that non-orthogonality of the mesh is under control. The difference between this case and the oscillating ACMI case in tutorials is that here the farfield boundaries are fixed and the oscillating displacement is provided on the bodies rather that the fluid volume.

Hi Virag,

OK, this certainly looks interesting. But I'm still having a hard time fully understanding what you're trying to do.

My first interpretation is that the outer body is closing the gap on the top, therefore no longer allowing fluid go over the outer body. If this is the case, then ACMI was not designed for this in specific, at least not as far as I know.

My second interpretation would be that the ACMI patch would be a cylinder patch between the two bodies, with the axis aligned with Z... but if this was the case, then there would be no need at all for ACMI in the first place!? AMI would probably be more than enough.


From what I investigated some days ago, the ACMI feature was designed as indicated here: http://www.openfoam.org/mantisbt/view.php?id=1450#c3770
In other words, for each side of a cyclic patch, the ACMI has 2 parts: the AMI cyclic (slave or master) and the fixed wall (A and B side). The two slide over each other, so that whenever the cyclic mode is not turned on, it uses the fixed wall settings.

If I understood you correctly in the first interpretation, then you have the deletion of cells at the top and that will therefore not work as you expect it to work.

If it's neither one of these interpretations, then I really need a simple schematic design of what you're trying to achieve.

Best regards,
Bruno

Hello Bruno

Thanks for the prompt reply and the effort to understand the case. My apologies if I was not clear. Let me try to explain it again. Lets take you second interpretation and start from there. You are right in understanding accept that the motion is not a rotational motion about Z axis rather it is translation along the y axis. I am attaching another Image the first image is the top view of the second image which is a cut plane. Isn't AMI only applicable for rotational motion like stator-rotor. In my case the motion is more like a piston-cylinder but a wider gap between them. Imagine the ACMI patch to be a cylindrical patch b/w the cylinder and piston. Both the bodies are floating. I understand that ACMI should not be a first thought, but I tried a single domain with no patch b/w two bodies, but due to the sharp corners in the inner body after some time steps the mesh non-orthogonality goes beyond 80 and case diverges. I got inspired by the ACMI tutorial and the fixedNormal BC in potentialFreeSurfaceDyMFoam. Any suggestions? I can't make the fluid domains oscillate as the flow fields on the farfields should remain fixed walls (Assume them to be water test tank).
Thanks in advance.

Hi Virag,

I think there was a misunderstanding on what AMI stands for... it was first introduced in OpenFOAM 2.1.0: http://www.openfoam.org/version2.1.0/ami.php - quoting only the relevant excerpts:
In other words: the rotating regions is the most common application shown throughout OpenFOAM's tutorials, but in fact, AMI is a very powerful way of exchanging mass/energy flow between two non-conformal patches. In other words, it doesn't matter much what is the motion between the patches, as long as they are always properly overlapping each other! :)

ACMI was developed for contemplating a missing feature in AMI: the ability to allow exposed parts of a patch in the AMI patches. This is explained here: http://www.openfoam.org/version2.3.0/ami.php - look for the section "Arbitrarily Coupled Mesh Interface (ACMI)" on that page.


In your case, since the far field is fixed, this means that you can simply use the AMI patches for overlapping the inner region and the outer region, because the far field is fixed.


Let's use for example the tutorial "incompressible/pimpleDyMFoam/movingCone":

1. t = 0.0 s:
   http://www.cfd-online.com/Forums/att...1&d=1424725489
2. t = 0.002 s:
   http://www.cfd-online.com/Forums/att...1&d=1424725497

Now, by using ParaView, we can do some neat geometrical transformations and create 2 examples:

1. AMI example, where the cones do not intersect each other. The AMI cyclic patches are the middle axis:
   http://www.cfd-online.com/Forums/att...1&d=1424725963
2. ACMI example, where the cones are moving over each other, therefore having parts of the patches connecting and not connecting over time:
   http://www.cfd-online.com/Forums/att...1&d=1424725970

(Note: will edit this post to add the images in a few minutes...)
(Edit: I've added the images ;))

Best regards,
Bruno

Thanks Bruno

Implemented AMI on the translation motion.

Hi, Bruno,

I am interested in your second picture. You have done that with paraview, but will it work in actual ACMI? I am testing a similar case, and I am not able to solve the problem of distorted mesh between cellZones. If you put slip boundary condition (as it is in the moving cone tutorial), the points are moving with an intermediate velocity, calculated from both meshes.

What should be the correct boundary condition in the interface?

With best regards

Robert