[Vinzent]

Hi,
I want to calculate mesh deformations with CFX12.0. The very simple 2D test modell is a square with a square hole. The hole moves within the field from bottom left to top right. Please look at the images.
Image 2 shows the calculated deformed mesh with stationary walls. Image 3 is the same exept the top wall with "unspecified mesh motion".
My question ist, how can I keep the outer square shape while the top wall nodes can move just in x-direction (like in image 3) to avoid the small angles at the walls?
Compared with structural mechanics you can call it a frctionless support.
I know, for the shown test case there ist no need to let CFX calculate the deformations. The main problem is much more complex.
Thanks in advance.
Vinzent

[singer1812]

For this particular problem (I dont know how much more complex your real one is), you could create a small area(s) on the outer walls that corresponds to the projection of the wall(s) from the middle item to the outer wall(s). This effective cuts each of the outer walls into 3 walls each. Use specified mesh displacement (or location, depending on how you write your CEL) on the outer walls. The top and bottom walls will have zero mesh displacement in the Z and Y direction, the side walls will have have zero mesh displacement in the Z and X direction. For the top and bottom X displacement use a CEL function that ramps a weghting from zero at the corners to 1 near the middle area, and is one in the middle area. Multiply this by the xdisplacement that the inner feature moves at.
Do the same for the side walls Y displacement except multiply this by the ydisplacement that the inner feature moves at.

[Vinzent]

Thank you very much.
This is a good idea. But unfortunately it is not exactely what I want, because of the real model properties. I don´t want to determine the displacement distribution along the bondaries, resp I can´t. For example the top displacement in x direction should be solved by cfx, y and z have to be zero.
Any other ideas?
Vinzent

[singer1812]

Are you sure you cant specify the top and side displacements? If you are using CFX to determine the movement of the inner thingy based on flow (i.e. 6DOF), you can write a cel term that gets the movement (generated from the 6DOF) amount of the inner thingy for the x and y direction, and then just use that in your ramping function movement for the edges.

[stumpy]

You don't need to pre-determine the x displacement along the top, you just need it to be driven by the motion of the square cut-out. So create those ramp functions as Edmund described (0 to 1, hold constant at 1, then 1 to 0 as you move along the boundary) then multiply those ramp functions by the displacement of the cut-out. You can query the current displacement using CEL functions, e.g. minVal(Total Mesh Displacement X)@cutOutWall. Those ramp functions will be a function of the local x coordinate value, so you'll also need to query where the transition points are using, for example, maxVal(Global X)@TopWallPart1 or minVal(Global X)@TopWallPart3.