S.C.F. discrepancies in FEA programs?

[gmmonkey]

Hello all,

Ok, what I’m doing is, calculating the stress concentration factor for a plate with a hole getting pulled uniaxially in the Cartesian 2-direction. I’m first of all calculating the theoretical, and then using my results from ABAQUS to calculate a stress concentration factor.

When I compared the 2 results, the S.C.F. calculated from ABAQUS was much higher at the edge of the hole. I know it’s something to do with ABAQUS calculating the stress at gauss points, and then extrapolating to get the value at the node. Can anyone point me in the right direction or give me an in-depth explanation to the exact reason of why the maximum stress is much higher at the edge of the hole in AQBAQUS than in reality?

Also, another thing I noticed is, when looking at the output .dat file of ABAQUS the stresses at the edge of the hole in the Cartesian 1-direction were quite large! Should they not, theoretically, be zero? Also, shouldn’t my stresses (S11 and S22) at the node points, in the 1 and 2 directions be the exact same as my principal stresses (SP1 and SP2)? It’s also calculated shear stresses when it shouldn’t have. I’m a bit confused. I’m aware this is a fundamental flaw in the method FEA (program) uses and would be eternally grateful if anyone can point me to some website or give a clear explanation. Thank you!

 

[SWComposites]

There is no reason why you should not be able to get the SCF to match for a simple loading case and material. Some questions:
What are the dimensions of the plate and hole?
How are you loading the plate?
What boundary conditions are used?
What element type are you using?
How many elements between the hole and edge of plate?
What is the material and what elastic properties are you using?
What is the theoretical SCF?
What is the SCF from the model?
How are you calculating the SCF from the model? i.e. what far field stress are you using?
Note that you will have to correct the theoretical SCF for finite plate width, as most SCF formula are for infinite width plats.

 

[johnhors]

With a fine enough mesh and correct BC's , there WILL be an almost exact match between theory and FE. Gauss point stress extrapolation is a red herring, with a good quality mesh there will be little stress variation across any one element and errors from extrapolation will be infinitesimal.