Hang off shoulder nonlinear: plasticity-geometric-

I am engaged in an analysis where I am trying to determine the plastic collapse load for a mandrel (gray) and split collar which serves as a hang off shoulder (orange). The mandrel is loaded on the internal bore with pressure and also has a corresponding end load due to pressure on a seal diameter. There is also an externally applied hang off force. Both the pressure end load and external end force are applied to the bottom face of the gray part. The orange split collar is supported by a thrust bearing at the bottom. (The orange part is actually a split half which is held together with “keys”. The orange split collar is supported by a thrust bearing at the bottom.

An asymmetric model was used to approximate the part and increase the run time.

Assumptions:

Yield Function: Von Mises and associated flow rule

Material Curve: ASME Section VIII Div. 2 – Annex 3-D – Section 3-D.3 (monotonic)

True stress vs true strain curve starting at yield (very small plastic strains)

“Tail” of stress strain curve has zero slope (elastic plastic) as per ASME criteria (at 163 ksi per code)

Material Hardening Rule: Isotropic

Nonlinear Type 1: Contact (No penetration (Surface to surface) with 0.16 coefficient of friction)

Nonlinear Type 2: Geometric (Large Displacement/Large Strain)

Nonlinear Type 3: Material (Plasticity)

See below (section view for clarity):

iso view.png

Elemental stress plot at final step (collapse):

(Close up at mandrel to collar hang off shoulder)

Global Mesh: 0.25”

Local Mesh 0.0125”

True stress true strain curve:

stress strain curve.png

Load curve:

time curve.png

Question:

1. The software in my analysis is outputting stresses above the final stress on the stress strain curve (163 ksi) entered into the software. It is my understanding that element stresses should be limited to the final stress points on the stress strain curve, at which point the element stiffness is zero. The output from my analysis seems incorrect. Am I missing something? (A setting or input parameter in the curve perhaps)

FYI: Solidworks Knowledge base - Solution S-027928:

For plasticity or non-linear elastic material definition it takes the last couple of data points and uses that to extrapolate linearly.

Note: stresses probed at corner radius - most exceed 163 ksi:

stress with curve at radius.png

2. Is using the mesh refinement study from a linear analysis sufficient to prove the nonlinear analysis mesh is acceptable? If not what is an acceptable convergence criteria?