Applying Circumferential displacement (torsional) to a hollow cylinder

[pranav04]

Hello,
I have modeled a 3D hollow circular cylinder and wish to simulate Static General step for torsional displacement of small shear strains. The cylinder is clamped on one side and I'm assigning value of u_r3 (along axis) to the reference point (center of the edge circle) on other end (using coupling constraint and selecting center reference point as control point and the nodes on boundary as region).
Problem faced: On visualizing the odb file, the radius of the circular edge (on which circumferential strain is applied) is getting increased. I have tried constraining its u_1, still it gets displaced in radial direction. Please suggest where i might be making mistake.

Thanks in advance

 

[corus]

Is it increasing or is it that the remainder of the cylinder is shrinking relative to where you have applied the displacement and the exaggerated distorted shape shows it to have expanded. Plot contours of the displacement to check.

 

[pranav04]

If I see it laterally, it can be visualized as a frustum with zero displacement at clamped end and uniformly increasing radius till the other end at which circuferential displacement is applied. I am not able to understand why the radius is getting changed in spite of arresting the displacement in that direction.
If you need any file or other information, please let me know.

 

[corus]

I applied torsion to an edge of a cylinder by applying a shear load along the edge around the circumference and saw a similar effect. In fact the radial and axial displacement is zero but the circumferential displacement increases as you'd expect when applying a twist/torque. If you think of the cylinder lying on the XY plane with its axis long the Z direction, then if you take a point on the X axis, for example, then that will displace in the Y (circumferential) direction but not in the X (radial direction). It would appear, therefore, that there is an apparent radial displacement when viewing the distorted shape although the only displacement is in the circumferential direction. Sorry, but that's the best explanation I can come up with, but it is very curious.

 

[pranav04]

Corus, I'm glad you thought of this. This thing clicked to me a few days back. So, I started applying kinematic coupling constraint. Now, since the center point is reference point and all the edge nodes are connected with rigid links to the reference point, then applying rotation to center point should not change the radius of outer radius.
I understand that giving circumferential displacement will definitely change the radius (as you suggest) but giving rotation about a point should not change it. I am sure there should be a way to study displacement controlled torsional deformation of hollow cylinder (by keeping both the ends rigidly clamped).
By the way, today I also tried applying moment in the same way, and still the radius is increasing at that end.

Do you know how can I connect a rigid plate/ring at one end of cylinder with rigid links and then rotate that plate/ring with specified displacement? Please suggest steps as I have never done assembly of two parts in Abaqus.

 

[corus]

A rigid plate, with rigid elements is created as with any other part though how this would differ from coupling the nodes to a central point I don't know. You could also create a single part of a closed cylinder and define the flat plate as having rigid elements. As I said before, the radial displacement doesn't change with torsion, it's just the circumferential displacement and what you're viewing is deceptive. To view the radial displacement try defining a local cylindrical coordinate system and view contours of displacement in that system.