Topological Mapping

It would be a real time saver if topological mapping (of loads and constraints) could be maintained with changes in geometry. I would like to propose a method of doing this.

In SolidWorks (and I assume other CAD packages) it is possible to name faces and edges.

This naming is preserved in the STEP file output.

In the screen shot below the custom face name created in SolidWorks can be seen in Salome.

It should be possible to use the names of geometric entities to maintain a link between geometry and topological entity sets (TES). As I see it there are two ways of doing this.
(1) Direct mapping - The mapping used in a previous TES definition is used to map new geometry to a new TES definition. So SimScale would attempt to find the geometric entities (by name) used previously and add them to the new TES group. Any changes that do not directly map (added or removed faces) would need to be made by hand.
(2) Rules based mapping - The user defines a set of rules to control which geometric entities get added to a TES. For example, any face with a name matching the string Fix* (with wildcard character) would be added to the new set (so Fix1, Fix2, Fix3 etc would all be added).

The final piece to the puzzle is maintaining the TES definitions when changing the mesh in a simulation. So if a mesh is changed and the new mesh has a TES with the same name as the previous mesh then all definitions using this TES should be maintained.

The advantage of this approach is that the naming only has to be done once in the CAD file. From then on even if the position, shape, size or quantities of geometric entities change the naming should maintain the links as required.

Is something like this a possibility?

2 Likes

Hi @BenLewis,

as always - great input! Thanks for the detailed screenshots! Supporting such associations/mapping between the topology of different domains is on the product roadmap. As always, it’s hard to share exact release dates but this is something that’s planned for rather sooner than later.

A few follow up questions/comments (as always :wink: ):

  • To confirm: What you’re describing is the use case of uploading another variant (or even just configuration) of the same design, right? You’ve set up the entire pipeline CAD->Mesh->Sim->Post and now you’re uploading a very similar design where you might have changed some dimensions and you want now to run sim of this as well.
  • The “Rules based mapping” is a super interesting approach - great input! That way, the number of e.g. fixed / loaded topological entities could be changed without ever touching the mesh / sim setup.
  • Another use case of such mapping would be a mesh study, where you generated different meshes for the same CAD and you want to run a mesh independence study, so the sim->mesh topology mapping would be preserved when switching the mesh. Since you didn’t raise this point, I assume it’s of lower priority for you?
  • The use of topological entity sets: You’re stressing the point that the “stable” / “preserved” referencing of topological entities within the domain would work via sets. While I do agree with this, I think referencing them directly should be possible as well. Say you’ve assigned a BC directly to some specific faces and now you’re uploading a variant of this design. If in the new CAD model these faces can be mapped to the old model (e.g. via their names, as you suggested) the BC association should be mapped as well. What do you think? Would you prefer the “indirect” mapping via the sets for some reason?

Best,

David

Hi @dheiny,

Exactly. I spend the majority of my time doing this.

Yes, I really like this approach because it requires no intelligence to match previous topology with new topology. The user is free to change what they like so long as the TEs are labelled. I got this idea from the workflow in Salome-Meca (the way mesh groups are made from geometric entities). The big thing for me was the discovery that faces and edges can be named in SolidWorks. So it seems almost everything needed already exists.

This is an interesting idea and one I would use but, you are right, it is a lower priority for me.

I only suggested the indirect method because it is close to the current work-flow in SimScale and therefore might be easier to implement. If it can be done directly, this is better. So long as the user has a point to manually tweak the settings (add/remove TEs) if required.

Regards, Ben

Hi @BenLewis,

got it! Will share more details and ask for feedback once we’re getting closer to this.

Best,

David