[StructuralMechanics] Post-simulation von Mises stress calculation for shells returns zeros

[takeshimg92]

Description
For some modal analysis applications, I need to calculate quantities derived from displacement a posteriori.

Essentially, I use (1) the ModelPart coming from the solver and (2) the automatic VTK output for the displacement of each mode. For a few reasons, I need the displacement field to be normalized before calculating von Mises, so I normalize it (using Pyvista, reading from the VTK files) and then calculate von Mises stress as follows:

for mode in mode_names:
    displacement = modal_fields[mode] # this is a normalized field, as a Pyvista UnstructuredGrid

    # set artificially this as the displacement field in the model part  
    model_part.ProcessInfo[KM.STEP] = 0 # this is the ModelPart coming from my running of the simulation
    for node, u in zip(model_part.Nodes, displacement):
        node.SetSolutionStepValue(KM.DISPLACEMENT, u.tolist())

    # von Mises as a dictionary from id to value (density, so no multiplication by volume)
    dic = {}
    for element in model_part.Elements:
        id = element.Id
        value = element.CalculateOnIntegrationPoints(SMA.VON_MISES_STRESS, model_part.ProcessInfo)[0]
        dic[id] = value

This works just fine for solid elements (SmallDisplacementElement), but the same code for ThinShells returns zero.
My questions are:

1. Is there a way to bypass this?
2. Is there anything I need to be careful with, or change in the source code so this will work for shells?

Scope
StructuralMechanics

[loumalouomega]

Probably VON_MISES_STRESS is not implemented in the shell. Which shell are you using ?

[takeshimg92]

@loumalouomega thin shells 3D3N. From here it seems it is calculated - for steady state analysis I can just get the von Mises stress in the usual VTK outputs.

[loumalouomega]

Huum strange, a priori if implemented should return values different from 0.

[loumalouomega]

Huum strange, a priori if implemented should return values different from 0.

Any idea @KratosMultiphysics/structural-mechanics ?

[AlejandroCornejo]

IDK exactly but could it be that with shells they use VON_MISES_TOP or bottom?