Fixing end of flexible pipe in axial direction (still moving in z?)

1

Hello, I am trying to model the FSI problem of an elastic pipe, which is subjected to a pressure pulse at one end (in the fluid). The classic Moens-Korteweg problem.

The problem is based on the preCICE tutorial case elastic-tube-3d (but does use shell elements).

This is following Bazilevs2006 and Greenshield2005.

Pipe is 20 mm diameter, 2 mm wall thickness and 100mm long. The solid model is made from second order, triangular shell elements.

I use

*BOUNDARY
 Ninlet,3,0
 Noutlet,3,0

where 3 is the z-direction, i.e. axial direction.

I would expect this to fix all nodes on the ends of the tube to be fixed in that direction and allow expansion in radial (x and y) direction.

It does do that somewhat, but there is still considerable displacement in z.

Am I using the wrong BCs here?

2

Try defining it this way:

*BOUNDARY
 Ninlet,3,3,0
 Noutlet,3,3,0
3

Immediately after posting the question I re-checked the documentation. And while it says, that the second DOF can be ommited if only one is constrained:

 First line:

    *BOUNDARY
    Enter any needed parameters and their value.

Following line:

    Node number or node set label
    First degree of freedom constrained
    Last degree of freedom constrained. This field may be left blank if only one degree of freedom is constrained.

all the examples i could find have the duplicate DOF present.

And, indeed:

*BOUNDARY
 Ninlet,3,3,0
 Noutlet,3,3,0

does behave as expected.

But did I read that snippet from the docs wrong? Or is there a bug in CCX? I’ll stick with the traditional method for now.

Thanks for the help.

4

Here’s what is meant in the documentation by leaving the third field blank:

*BOUNDARY
 Ninlet,3,,0
 Noutlet,3,,0

And this would work too.