It’s not just penetration but also significant element distortion. It resembles hourglassing but it can be just a matter of wrong setup - too coarse mesh and large inertia forces. You are running explicit dynamics, you have to be very careful with mass scaling and check kinetic energy whenever possible. We discussed this in your previous threads, for example here: How to apply rotational speed correctly - #5 by FEAnalyst
I still think that it would be better to use implicit solver for this but of course then you will struggle with convergence (especially when using force/moment loading instead of prescribed displacement/rotation).