Tension field time pronounced zigzag pattern

Hello there,

I have worked out a plastic calculation, and the stress field of the first principal stress shows a distinct zig-zag pattern. The model consists of Quadratic Hexahedral Elements (C3D20). The pattern occurs because of discontinuous results on corner and midpoint nodes.

How can this happen and what can I do about it? Linear elements produce a continuous pattern.

I am using version 2.20 on Windows 10.

Can you share the input file ? If not, please show the whole model and describe how it’s loaded and constrained. Have you tried with other element types ? The choice of element type in CalculiX is particularly important when plasticity is involved: Elasto-plastic bending - elements, accuracy

The model is a simple notched tensile specimen. By using the symmetry I have modeled only 1/8 and given the symmetry surfaces with corresponding boundary conditions (x=0, y=0, z=0). At the upper surface I give up a displacement (y=10)

Here is my input file:

*INCLUDE, INPUT=all.msh
*INCLUDE, INPUT=sym_x.nam
*INCLUDE, INPUT=sym_y.nam
*INCLUDE, INPUT=sym_z.nam
*INCLUDE, INPUT=force.nam

*MATERIAL, Name=steel
*ELASTIC
200e3, .3
*PLASTIC, HARDENING=ISOTROPIC
400.000000, 0.000000
500.000000, 0.042179
600.000000, 0.170387
700.000000, 0.496688
800.000000, 1.225562
900.000000, 2.700146
1000.000000, 5.460853


*SOLID SECTION, Elset=Eall, Material=steel

*STEP, NLGEOM
*STATIC
    .001, 1

*BOUNDARY
    Nsym_x, 1, 1, .0
    Nsym_y, 2, 2, .0
    Nsym_z, 3, 3, .0
    Nforce, 2, 2, 10.0

*NODE FILE
U
*EL FILE
S, E, PEEQ
*NODE PRINT, NSET=Nsym_y
RF

*END STEP

Here is a picture of the whole model. Nsym_y is “under” the model, and Nsym_z is “behind” the model.

you can try using reduced integration type (C3D20R) to see the differences since full integration is not preferred to be use in plastic analysis.

1 Like

Thank you very much. I changed the elements to C3D20R and now the result looks reasonable.

2 Likes