# Usage of Total Mesh Displacement in CEL for Movement Computation

 Register Blogs Members List Search Today's Posts Mark Forums Read

August 30, 2016, 08:32
Usage of Total Mesh Displacement in CEL for Movement Computation
#1
New Member

Join Date: Aug 2013
Posts: 23
Rep Power: 9
Hi,
I'm trying to solve a bending case, causing problems with mesh stiffness. . For visualization: 2 is bending and the subdomains 1 should move downwards according to the Total mesh displacement.

In 2D the application of:
Code:
`maxVal( Total Mesh Displacement Y  )@REGION:FSI_y_2`
is fine for adjusting the connected subdomain.

Coming to my question: If I'm trying this in 3D, there are curved shapes leading to unsuitable shifting values for the subdomains. Is there a possible way for copying the movement of a 3D area to another?

Attached Images
 Unbenannt.PNG (53.6 KB, 52 views)

 August 30, 2016, 20:54 #2 Super Moderator   Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 16,409 Rep Power: 127 You could use the probe function to get the displacement at one point and use it to define the motion of another boundary or subdomain. In 3D for complex shapes I would consider using the default mesh displacement diffusion method.

August 31, 2016, 01:41
#3
New Member

Join Date: Aug 2013
Posts: 23
Rep Power: 9
Hi,
thanks for your input. I am actually using the displacement diffusion method (but with a value function). The problem is the stiffness. Due to the small gaps and close walls (my bending geometry as well as the enclosing one) none of the built-in settings works properly. Thats why I'm stuck with this workaround. The outer region 1 needs to be moved downwards so the inner one can be deformed with a suitable stiffness. I enclosed part of my example studies with deformations.
Attached Images
 Unbenannt.jpg (90.7 KB, 19 views)

September 2, 2016, 03:39
#4
New Member

Join Date: Aug 2013
Posts: 23
Rep Power: 9
Hi,
sorry for double posting ...

I start again with a motion undefined region close to my moving element and with an enclosing domain, that I'll force to move. As the second picture clearly shows: one problem is the continuity of movement. Is it somehow possible to take the Mesh Displacement Y of an edge (with a named selection for addressing) and create a CEL / UDF function (close to the probe thing ghorrocks mentioned) and create a function dependant on a coordinate (lets say x) for application as a movement input?

Someone did something like this?
Attached Images
 CFX2.png (54.1 KB, 14 views) CFX3.png (69.3 KB, 12 views)

 September 2, 2016, 07:13 #5 Super Moderator   Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 16,409 Rep Power: 127 If you can't get the mesh deformation stuff working you can do dynamic remeshing. The tip of your device covers quite a large size range and dynamic remeshing would help here. Having said that I think with some careful planning you should be able to get this working with mesh deformation only. Have you considered a GGI at the end of the paddle so the mesh at the end hardly moves, but the paddle can slide up and down with a GGI?

September 2, 2016, 07:37
#6
New Member

Join Date: Aug 2013
Posts: 23
Rep Power: 9
Quote:
 Originally Posted by ghorrocks If you can't get the mesh deformation stuff working you can do dynamic remeshing. The tip of your device covers quite a large size range and dynamic remeshing would help here.
As I plan to implement 2way FSI, additional dynamic remeshing is quite time consuming ...

Quote:
 Originally Posted by ghorrocks Having said that I think with some careful planning you should be able to get this working with mesh deformation only. Have you considered a GGI at the end of the paddle so the mesh at the end hardly moves, but the paddle can slide up and down with a GGI?
Currently I don't use interfaces ... I thought about the sliding of meshes but didn't manage to get it working correctly so far