Learning Forum

[jridao2]

Hi everyone,

This forum's been very helpful to me in the past but I am (sort of) stuck on an issue that I haven't directly found an answer for yet. Suppose I have the following structure of a blue/green exterior tub that has a red panel that goes across the tub's width as shown below:

Currently the model is all solids but since all the objects are composite laminates, I will be using ACP and thus have the following shell/surface geometry to port into ANSYS Workbench from my CAD package:

You'll notice that in the pictures above and below, modeled as all surfaces/shells, the red panel is not in contact with blue/green tub anymore like it had been when everything was a solid with thickness.This is because the tub shell model has been taken from the tub's external geometry because, for all intents and purposes, I can't model the tub from its internal geometry.

The above image is a cross section view normal to the panel showing the gap between the panel and the sides and bottom of the tub.

 

My first question is after building the model in ACP such that the gap between the tub and the panels is 0 (or close to it) by defining laminates to be the exact width of the gap, how can I set up the model (in Static Structural for example) so that the panel is bonded to tub? I'm making the assumption that the panel and tub are very rigidly bonded.

What I have tried:

• I've read and tried implementing a shared topology in SpaceClaim to no avail because the panel and tub aren't inherently touching as in this video
  . I was considering trying to merge the nodes between the panel and the tub, thus forcing them to act combined. But per this forum post, topology would theoretically do a better job as it is automatic.


• Used a bonded connection between the panel and the tub as in this article. I was only able to get the thickened shell models to bond when they shells were parallel to each other. In my scenario, the panel is effectively perpendicular to the tub sides and bottom.


• Joining the tub to the panel as a fixed joint between the two. This did actually get the model to run, though I'm skeptical of the results as I am unfamiliar with the general difference between a contact and a bonded connection.

Again the assumption I'd like to model is that the gap is a joint that is very rigid as in reality, a strong epoxy will be filling the very small gap between the panel and the tub.

Any help would be greatly appreciated.

 

Thanks,

Rafael

[jridao2]

As an update, I am still unable to solve this problem. I've simplified the investigation into a geometry of 2 perpendicular, but offset, surfaces that I import into ANSYS from PTC Creo. After properly defining the surfaces/shells in ACP, the resulting mesh is achieved once the ACP results are transferred to a Static Structural analysis:

 

 

Where red highlights the initial surface geometry. The base panel is 20mm thick. As such, its perpendicular panel is model as a square surface that is offset 20mm above the base surface geometry. When using a bonded contact between the two as shown below, the model is ill constrained as expected due to the fact that the surfaces aren't touching.

 

 

When I use the same faces and constrain them to be connected via a fixed joint however, I am able to set boundary conditions as shown below and get results though I am skeptical of the accuracy of these results and therefore am still seeking help with improving my model.

 

 

Thanks,

Rafael

[peteroznewman]

Rafael,

I don't use ACP so I don't know the "rules" for modeling in that environment, but I will suggest something which may or may not be appropriate.

1) For Bonded Contact, if you increase the Pinball Radius to a sufficiently large value, will the contact between the shells show as "Closed"?

2) Is it legitimate to edit the surface geometry and manually close the gap in SpaceClaim?  Then you could use Shared Topology.

3) In the same vein as #2 but without editing geometry, you can define a mesh connection that would extend the shell elements on the vertical plane by picking that edge as the slave to join the horizontal surface as the master.  You might have to edit a tolerance similar to a Pinball.

Regards,
Peter