Energy error exceeds limit when only a fraction of part is rigid

[wgm096350]

Describe the bug
I have a very large electronic device model (size of rad file> 900MB) which is classified. During the solving proces, RADIOSS suggests the energy error exceeds limit and stops. I have simplified the large, classified model to the following unclassified, small model which resembles the large model and also has the same energy error issue.
There is two parts in the model. One of them is a elastic body but a fraction of its nodes are constrained by /RBODY simulating MPC BEAM constraints. The other is a rigid floor made of shell elements. The elastic body is shifted to ensure its bottom surface contact the upper surface of the shell at the beginning of the simulation. The inital velocity of the elastic body is 1000 mm/s.
As expected, the elastic body bounced back into the air. But at the time of 0.1534E-02s, the energy error suddenly exceeds the +90% limit as shown. Once the /RBODY is removed, everthing is normal. Thus, I think the /RBODY is the source of the issue.

To Reproduce
With the following uploaded model, the issue could be reproduced
CubeRigidImpactNew.zip

Expected behavior
Energy error exceeding the +90% limit is not expected.

Output
Here is the *.out file.
CubeRigidImpactNew_0001.zip

Desktop (please complete the following information):

• Version Build tag: 1109567_128532022_2020_0090417_10
• Hardware 12th Gen Intel(R) Core(TM) i9-12900H 2.50 GHz
• Configuration -nt 10

[wgm096350]

Although there are some workarounds to this issue in the simplified model, such as

• change Itetra10 = 1000 to Itetra10 = 3
• Change 10-node tetrahedral to 4-node tetrahesdral
• etc.
  But the workarounds is not applicable to the large, original electronic device model. So, I sincerely hope the developing team could help me to locate the root cause of the issue so that the large model could also computed without error.

[PaulAltair]

I think if setting an RBODY on 2nd order elements, you should include the midside nodes in the secondary group (Both workarounds you mention would have the effect of supressing or removing the midside nodes so that could be why they work too): If you don't include the midside nodes you didn't rigidise those elements, only their corners: If I add the midsides (roughly, not sure I got all the exact correct nodes for the rigid elements required), it runs to termination:

I also made a couple of other changes, your /BCS card had no ID, actually surprised it was running ok like that! And there was an empty node group (16) that I removed

PS_001_Midsides_Added.zip