principal directions issue

Hello everyone,

I've considered a square plate in Karamba, supported on its four corners, and loaded by its self-weight(out-of-plane). When I consider the principal stress directions in the top (+1) and bottom(-1) layer get the following result (on green>bottom, on blue>top): 

But in ansys (considering the same boundary conditions, with the element shell63 and same meshing situation) the directions seem to be mantained switching from top to bottom layer (the signs of the vector reverse). This makes further sense, as long as the element formulation is based on the hypothesis of the linear distribution of stress on the shell section.

I've also noticed in Karamba that the principal stresses are linearly increasing from the mid- to the top layer, without changing direction (according to the theory above described). The discontinuity on the principal stress direction is on the very mid layer: from the mid to the bottom layer the principal stresses are once more linearly increasing, but with a different and not compatible direction with respect to the half-space above.

What's wrong (if it is)?

Thank you in advance

G.