Feedback wanted: Chrono adapter boundary for an open-source CAE application layer

[Shlomo]

Hi,

I’m working on an early RFC-stage open-source CAE application-layer initiative.

The proposed first public MVP is MBD-first, and Chrono is one possible backend through a solver-adapter layer.

The goal is not to replace Chrono, build a new MBD solver, or create a thin GUI wrapper around Chrono.

The goal is to define a solver-independent application layer that could eventually connect CAD/geometry, assemblies, MBD, FEA, CFD-ready workflows, meshing, visualization, results, and automation.

For the Chrono side, I’m trying to define the correct boundary between:

- a solver-independent Core Model
- an Application / Command Layer
- a Chrono MBD Adapter
- Simulation Cases
- time-based Results
- visualization / replay

The key question I’d appreciate feedback on is:

What is the biggest mistake to avoid when designing a solver-independent application model and adapter boundary around Chrono?

Repository:
https://github.com/Shlomo365/open-cae-application-layer

Main RFC discussion:
https://github.com/Shlomo365/open-cae-application-layer/discussions

Thanks,
Shlomo

[Jose]

Hello Shlomo. 

I have read your project design. It is very interesting to me, because we are working on very similar objectives. Your architecture proposal is quite similar to the software I am developing. Here is a quick description of what I have:

A CAD-MBD-FEM integration software centered around FreeCAD. It is solver-agnostic. So far I can solve the same problems using Chrono, MBDyn and MuJoCo for Mbd, and Calculix and Mystran for Fem. 

It offers a weak coupling between Mbd and Fem, by reducing the dynamic problem to a set of static problems, one per mbd frame. Same procedure for any mbd solver- fem solver pair.

We have created a set of joints capable of distributing the mbd forces/torques as nodal loads at the adequate mesh nodes. The joints take into consideration the shape of the contact faces in CAD, and carefully select an adequate formulation to ensure force and moment equilibrium throughout all the frames. We have written and coded these formulations, wich are the core of our project.

It is a user-friendly CAD-mbd-fem integration. Users can create joints, mesh parts, create imposed motions, add fem, and post-process all from within FreeCAD. No need to write code or input files.

It is CAD-centered. If the user changes the dimensions of the CAD parts, these changes trigger a cascade of recomputations, affecting assembly, then mbd and fem. Make a cad part bigger. The joints addapt. The mbd solver sees the new mass and inertia. The fem solver sees the new fem loads. The user sees the new stresses. All happens automatically.

Currently we are still testing and refining the software. Duri g the last few months I have been improving its multi-solver capabilities, and this is why I got into Chrono (before I was using only MBDyn and Ccx). 

Hope you find it interesting. You can read more here: www.mbdfem.com, although the webpage is almost a year outdated now. Im working on a new version.

Kind regards.

--
You received this message because you are subscribed to the Google Groups "ProjectChrono" group.
To unsubscribe from this group and stop receiving emails from it, send an email to projectchron...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/projectchrono/b1f4a84e-17e6-4ae9-a35c-d220e03f731an%40googlegroups.com.