Learning Forum

[zhihuizou1988]

 When I use Ansys analyzes the classical benchmark, hemispherical shell problem as shown in the first picture below, I got an error shown as the second picture below.

The displacement boundary conditions and the force boundary conditions are shown in the following picture. More specifically, the displacement boundary conditions are applied as,

the left boundary is fixed with translation in y-direction, i.e., u_y = 0, and the rotations around x and z-axis are zero, i.e., r_x = r_z = 0;

the right boundary is fixed with translation in x-direction, i.e., u_x = 0, and the rotations around y and z-axis are zero, i.e., r_y = r_z = 0; 

the lower-left corner point is used to remove the rigid body motion in z-direction, i.e., u_z = 0.

The force boundary conditions are specified at node 1 and node 2 as shown in the following picture, i.e., a constant force along x-direction at node 1, and a constant force along y-direction at node 2. 

Theoretically, there is no translation displacement boundary condition applied at node 1 along x-direction and no translation displacement boundary condition applied at node 2 along y-direction. Therefore, there should not be conflicting DOF between the displacement constraints and the loading.

Is anybody know how should I solve this conflict but still apply the boundary conditions correctly? THanks very much.

[jj77]

At least in apdl if you mix boundary conditions on FE entities (nodes), and geometry a warning is issued.

(It might also be another source for the problem)

I would like in the benchmark, apply all BC only on nodes as nodal displacements, rotations and forces, and follow exactly as the benchmark.

 

I will do it later and show a screenshot of the WB BC.

 

[jj77]

So if we use only nodal BC then of course it is OK. Below are the settings, and with this it solves fine (the command snippet is to use same shells as in verification example VM-R029-T9 181).

The named  selections contain the nodes at the symmetry (X and Y) planes and where the forces  (FX and FY)are applied.

[zhihuizou1988]

Thanks jj77 so much. According to your suggestions, I just tried to apply all boundary conditions on geometry boundaries. And it worked very well. Thank you so much.

[jj77]

Good, both, methods (FE, or on geometry) work.

 

If this discussion has provided an answer to your original question, could you please mark the post with Is Solution,

many thanks,

jj77 

[zhihuizou1988]

Hi jj77,

 

One more question for you please. Do you know how to get the stress controu plot in the middle surface rather than in the whole body? Thanks a lot.

 

[jj77]

You can choose it under settings, see below. (have in mind that the middle surface does not have the bending stress, only membrane action).

[zhihuizou1988]

The middle surface should include shear stress as well, right?

[jj77]

Shear is more tricky - I am not familiar with ansys shell formulation, I would check the theory or element reference (say shell 181 if that is what you use), and look at the formulation for shear.

 

I would recommend to start a new post with the shear question on shells.

[zhihuizou1988]

Thank jj77 a lot. Shell 181 is based on Reissner-Mindlin shell theory. I am sure it includes shear stress. 

[jj77]

Typically for shell elements, at least for the software I work with (Strand7), the shell and beams both assume a parabolic transverse shear stress, which is zero on the top/bottom and maximum in the mid plane. 

It could be though that shell 181 gives something else since there is a special treatment of shear in it. I can find anything about the distribution on the shell 181 ref. If you need to know exactly post another question, and perhaps some of the ansys guys on the forum might have some feedback (or do some simple tests to find out)

[zhihuizou1988]

Got you jj77. It is very helpful.