Particle explosion, contact algorithm, and normal vector definition

[sunny12138-9527]

Hello everyone,

I'm encountering severe particle explosion during an Anura3D simulation. Shortly after the start, particles near the contact interface fly apart with unrealistic velocities, causing the computation to abort.

I suspect the issue lies in the contact algorithm. I’ve seen a previous suggestion that modifying the normal vector definition in the .cps file could resolve similar problems, and that one needs to "determine the node positions" to set the normal direction. However, I'm unclear about how to do this in practice. My main confusions are:

Which two nodes should be chosen to define the contact normal? (e.g., nodes on the master/slave surfaces, or specific boundary nodes?)

How can I obtain the exact node IDs and coordinates from my mesh, and what is the correct syntax to specify the normal in the .cps file (e.g., using the NORMAL keyword)?

Are there any other critical settings to check, such as contact stiffness, contact type, or penalty factors?

I would greatly appreciate any guidance on:

The step-by-step procedure or rules for selecting the nodes to define the normal vector;

A correct example or snippet of the normal definition in the .cps file;

General advice for debugging particle explosion related to contact normals.

Thank you very much for your help!

[lcheng83]

Could you describe it more clearly, or provide the compressed package of GiD calculation files.

[sunny12138-9527]

My current issue is that I cannot locate the exact positions of the nodes/particles that are causing the simulation to blow up. I tried changing the obtuse-angled corner at the slope toe to a rounded fillet, and the situation improved considerably, but I still haven't found a way to identify where those exploding particles/nodes are. I am a student from China. Could I possibly add your contact information and discuss with you some questions about using Anura3D? Thank you.