From the Analysis Reference Manual: Degrees of Freedom
The Shell element always activates all six degrees of freedom at each of its connected
joints. When the element is used as a pure membrane, you must en sure that
restraints or other sup ports are provided to the degrees of freedom for normal translation
and bending rotations. When the element is used as a pure plate, you must ensure
that restraints or other sup ports are provided to the degrees of freedom for in-
plane translations and the rotation about the normal.
The use of the full shell behavior (membrane plus plate) is rec om mended for all
three- dimensional structures.
Note that the “drilling” degree of freedom (rotation about the normal) is not used
for the layered shell and should not be loaded.
The three translation degrees of freedom are pretty basic. That's just going to be the X, Y and Z deflections for each joint of the shell. The rotations are a little more complicated. Assuming a shell in the XY plane, the rotations about the X and Y plane are your basic plate rotations and impart normal bending moments into the plate / shell.
The rotation about the Z axis (for a shell in the XY plane) is something that I've always been a little skeptical of. This is the drilling degree of freedom for plates and shells. And, older plate element formulations never include this degree of freedom. But, if you want to test it. Then you connect a column to the center of a flat mat foundation (maybe something like a 10x10 mesh) and apply a torque to that column. Then you could do the same thing to a model where you use a rigid link to transfer that torque instead of relying on the drilling degree of freedom. If your results are pretty close, then you've proven that this degree of freedom is accurately modeled.
