## General Mechanical

#### How to use line bodies in 2D structural analysis?

• crumberg
Subscriber

I am attempting to do a 2D analysis in Workbench using Fluent and Mechanical. The problem is similar to the common airfoil example, but I need to incorporate line bodies into my structure. Inside mechanical, the editor seems not to accept any type of line body for 2D analysis, i.e. it rejects the model types Beam, Pipe, Link/Truss, Cable, and Reinforcement for line bodies. There are no other structural options given. Is there any workaround for this problem? Thanks for any help.

• SaiD
Ansys Employee

Hello! If you have trusses or beams in your problem, I suspect that the problem is not truly 2D i.e. it cannot be classified as a plane strain or a plane stress problem. What do the lines represent in real life? Are they beams that reinforce the airfoil? Is the physical situation that you are modeling such that either strain or stress in the third direction is zero?

May be you can describe the physical geometry a bit more in detail.

But if the line bodies actually represent beams for reinforcements, then you may have to do a 3D analysis with the complete 3D geometry.

• crumberg
Subscriber

Thanks for the response. I was afraid that it might be the case that the software will not allow me to model in 2D. The geometry represents an endothelial cell in a parallel plate flow chamber. The reinforcements represent cytoskeletal structures that are crucial to the question that the model is investigating. They are very thin compared to the other structures (about 1000x thinner than the height of the cell in this model) so I was hoping to justify the use of 2D beams/cables for this model on that ground.

Unfortunately, the complexity of the 3D shape has led to never-ending convergence issues over the past year of work on this model, so I was trying to simplify it into a 2D approximation so that I would at least have some kind of limited result.

• SaiD
Ansys Employee

That sounds like a very interesting problem!

About these cytoskeletal structures: in an actual 3D model, are these structures 2D? What I mean is that if you extrude this structure in the third direction, the lines would get converted to a 2D geometry. Does that represent your 3D geometry?

If not, then the assumption of a 2D analysis is wrong (even to get preliminary results) because the geometry changes in the third direction.

May be you can create another post about your convergence issues and we can see if the community has any ideas that might help.

• crumberg
Subscriber

That's a good point. The line structures represent very thin beam-like or cable-like elements rather than planes in the third dimension. A periodic structure with a thin repeating geometry might be a more appropriate simplification than the 2D analysis, but I don't think periodicity is an available feature for mechanical structures in Workbench.

I went ahead and put some info regarding the 3D model in my reply to Peter along with the archived project files. I can elaborate on the fixes I've attempted and the types of errors I've gotten if that would be of use.

• peteroznewman
Subscriber

Hello @crumberg

I am interested in helping you get your 3D model to converge in Static Structural. What version of ANSYS are you using? Are you able to share the archive so I can take a closer look? If so, clear the results and the mesh to minimize the file size, save the project then do File Archive to save a .wbpz file.  Upload that file to a sharing site such as Google Drive, Dropbox or Jumpshare, which is what I use. Put a link to the file in your reply.

• crumberg
Subscriber

Thank you. I really appreciate your willingness to take a look. I cleared all generated data in the modules to reduce the file sizes. Here is the link to the folder: https://drive.google.com/drive/folders/1ksJo4iIcEyQKl1i5d2oBqnQIHBlnuqw7?usp=sharing

I have included two of the most recent 3D models in the folder. The first ("gaussian_endothelial_cell") builds the geometry using 3D curve import from points and uses hydrostatic fluids for the cell interior. I eventually came close to convergence with this model, but the forces are not being transferred through the cell to the nuclear membranes as intended. I also built it at 1000x scale to try to avoid geometric idiosyncrecies. (If I understand correctly, APDL Mechanical is dimension agnostic but Workbench is not.)

In the second file ("dome_endothelial_cell") I tried to avoid some of the geometric complexities of the other model by manually connecting nodes from point imports to avoid geometric gaps and changing the shape somewhat to avoid any concavity in the cross-sectional geometry of the cell. I also switched from hydrostatic elements to solid elements with a low elastic modulus and high bulk modulus to represent the viscoelastic regions (cytoplasm and nucleus). I hoped that this would allow pressures to properly transfer through from the outer membrane to the nucleus of the cell. The scale was switched from millimeter to micron since I was optimistic about the model and that is the preferred scale for my final results. I left the contact types as frictionless because I was able to get them to close acceptably under those conditions but I also tried running the model using bonded contacts. I have had a lot of trouble with meshing in this model, particularly on the nucleus. For troubleshooting, I tried suppressing the nucleus and replacing the void with a fixed displacement. The solver then had trouble with excessive distortion errors in the plasma membrane.

Thanks for taking a look, and I appreciate any insight you can offer.

• peteroznewman
Subscriber

• crumberg
Subscriber

Sorry for forgetting to include that info. I am using Workbench 2021 R2

• peteroznewman
Subscriber

Sorry for the delay, I was on vacation for a long weekend. Now I am back and find I don't have Workbench 2021 R2 installed, so I will need some more time to get that installed.

• crumberg
Subscriber

No worries. After a lot of tuning with the mesh, I was able to get the 3D "dome" model to converge at nearly every time step before the solver ran out of memory at 99%. It required an excessive stabilization force, however (0.1), making the results unrealistic; so I would still be very glad of any insight you can offer.