This continues from an old shells thread where I found issues I couldn't explain at the time.
While looking for something else I found this entry in help docs about shells stiffness. See the paragraph about "COEF_RIGI_DRZ = “CTOR”.
It turns out that CA hard wires the drill stiffness value on shells to 1e-05, regardless of units and/or materials used in a model. I then search for this issue, and found several old entries about shells in forums: shells_drill_K_1, shells_drill_K_2, shells_drill_K_3.
I then re-run my shell test model with a TIED contact that was giving me the wrong moments in statics and wrong rotational RBModes (not 0Hz) in a free-free modal. Defaults in CA: COEF_RIGI_DRZ = 1e-05 | new test value: COEF_RIGI_DRZ = **1e-09**.
This is the model with 2 soft couplings (LIAISON_RBE3) for both load and constraints + 1 tied contact on shells:
Then when solving linear statics with 9 unit loads: 3 forces, 3 moments, 3 Gravity loads we can check on input vs sum of reactions for LCs 4,5,6; the ones with moments (2000N*mm) as inputs. We can see in LHS tables with defaults the moments are out, even on a simple model, see in red. RHS shows the same results after playing with the shells drill K value and moment results are now correct on all tables:
If we then solve a free-free modal; i.e. no constraints to ground, we expect 6 RBModes at near 0Hz. With defaults the rots modes 4,5,6 are out, but with non-default drill K they are correct at near 0Hz:
