Boundary selection of 3D geometric boundaries using MPh coding

[vczut]

Hi!

I am developing a model with two components:

1. Component 1 performs a deformation simulation based on a 3D parametric geometry (e.g., varying strut length).
2. Component 2 imports the deformed configuration from Component 1 to perform a subsequent stress analysis.

Using the MPh Python interface, I am able to automate much of the process — including geometry updates, running the deformation simulation, exporting the resulting mesh, and importing it into Component 2. However, a key issue is that boundary selections in Component 2 do not persist across mesh imports. As a result, critical boundary conditions (such as applied loads) become undefined or incorrect when the mesh is updated (e.g. the domain selection disappears).

My question: is there a way within the MPh API to ensure consistent boundary selections (e.g., using named selections or tags) to ensure the boundary conditions are applied to a certain boundary selection of the geometry?

If needed, I can share a minimal working example to illustrate the workflow and current limitations.

Thanks for your great work on MPh, it’s been an excellent tool for integrating COMSOL into automated pipelines.

Best regards,
Vera

[john-hen]

Hi Vera, thanks for the kind words. I don't think I can be of much help here though. I guess MPh can help to check that an explicit isn't empty, but that only detects the problem and does not solve it. In the past, I've always just used what Comsol itself offers: coordinate-based selections such as "box" or "cylinder". When the geometry is already parametrized, that's usually quite straightforward. As far as I know, that's what we're supposed to do to make the selections robust.

[vczut]

Thank you for fast reply and your suggestion, this solved my problem!

Since I am using MPh-Py for my research, I was also wondering if you have any publications on MPh-Py that I could cite?

[john-hen]

No publications, no. There's only the DOI for the software itself, as found on the front page: https://doi.org/10.5281/zenodo.11539226.