Simulating the Demolding of a Flexible Mold with Physical Thickness

Hi everyone,

I’m trying to simulate the demolding process of a flexible silicone mold using Grasshopper and Kangaroo 2, and I’m having some difficulties getting the material to behave realistically. I’d appreciate any advice or suggestions.

I have two objects: a flexible silicone mold (A) and a rigid object (B) that is enclosed by the mold.

Object A is a closed Brep with inner and outer surfaces and a constant physical thickness of approximately 1–5 mm. Object B is also a closed Brep and represents the rigid piece being cast.

What I want to achieve is quite simple in principle: I would like to grab the silicone mold and pull it away from the rigid object, as happens during a real demolding process. The mold should be able to stretch and bend around the geometry, while maintaining its thickness and recovering its original shape as much as possible. Object B should remain rigid and stationary.

So far, I’ve managed to get a basic simulation working with Kangaroo 2.

For the mesh preparation, I’m using QuadRemesh with a target count of around 2,000, followed by Combine&Clean, Triangulate and Weld Mesh. I apply this process to both objects.

For the simulation itself, I’m currently using:

  • Show

  • EdgeLengths (LengthFactor = 1.0, Strength = 500)

  • HingePoints and Hinge (Strength = 10)

  • Grab

  • Solid Point Collide, using the mesh of Object B as a fixed obstacle

  • Entwine and Kangaroo Solver

The good news is that I can already grab the mold and pull it away from the rigid object, and the collision with B works reasonably well.

However, there are still a few problems I haven’t been able to solve.

First, the mold behaves more like a piece of fabric or a thin membrane than a silicone rubber body. Increasing the EdgeLengths and Hinge strengths helps, but it doesn’t really give me the material behavior I’m looking for.

Second, when I pull the mold using Grab, different parts of the mesh can intersect each other. Solid Point Collide prevents intersections with object B, but obviously doesn’t handle self-collision.

Most importantly, although the original geometry is a closed Brep with actual thickness, my current simulation is based on a triangulated surface mesh. I suspect this is one of the main reasons why the result doesn’t behave like a real solid.

I’ve been looking into tetrahedral meshes, finite element approaches and the SoftBodyCollide component, but I’m not sure which direction would be best.

Has anyone tried something similar in Kangaroo?

Would it be possible to simulate the mold as a volumetric deformable solid, rather than a surface membrane? And what would be the best approach for handling self-collision while keeping the physical thickness?

I’m also wondering whether Kangaroo 2 is suitable for this kind of simulation or if I should consider a different FEM solver.

I’ve attached screenshots of my current Grasshopper definition and mesh preparation setup.

Any suggestions or example definitions would be really helpful.

Thanks!

Demold_Mold.gh (393.2 KB)

See here

I have continued working on this and have achieved better performance with this configuration.

I’ve discovered that applying an anisotropic tetrahedral analysis to irregular meshes is not as straightforward as it might seem. I have set that method aside and have chosen to continue exploring approaches that use springs and hinges.

Thank you very much; I’ll see how things work with what you’ve just sent me.

I am posting my progress here so that it remains on record in the forum.

Thanks. :slight_smile:

Demold_Mold_2.gh (398.1 KB)

Teacher @DanielPiker, Could you provide some leads on how to better simulate this behavior? What might it be most comparable to? As Quan Li suggests, should it be modeled like an inflatable system—perhaps with spring lengths set to 1—or is a hinge-based system not the best approach?