Stiffness Mismatch

[Naus]

Hi everyone,

I am using ABAQUS to model the response of a finite-sized, repeating 2D truss structure under compressive loading. The trusses are very slender. I have conducted tests on metallic specimens and I'm having trouble correlating the experimental data to the ABAQUS predictions, specifically the stiffness (N/m). Although the peak load values correlate well (within 3%), ABAQUS is predicting approximately 1.8 times the experimental stiffness of my 304 Stainless structure, and 1.4 times the stiffness of my Titanium structure. Elastic-plastic material properties were taken from ASTM E8 tension tests done in-house.
I'm using B21 elements, enough of them that the stress-strain behavior has converged nicely.

One of my colleagues suggested the element itself may be too stiff? Any ideas would be great.

Thanks,
Tony

 

[corus]

Try using more elements. Finite elements show a monotonic rate of convergence tending to be over stiff with fewer elements.

corus

 

[Naus]

Corus,

Thanks for the reply. I just reran the simulation after doubling the element density and saw about 0.03% reduction in peak load and no real discernable change in stiffness. Do you have any other ideas? Thanks in advance,

Tony

 

[corus]

If the trusses are very slender then they may be buckling. Check with your model what the buckling load will be. This is best done with a large displacement analysis and some discontinuity in either your geomtry or loading rather than an Euler buckling calculation.

corus