ABAQUS explicit reaction force results not matching hand calculations 1

[SAK123]

This is the first time I have run ABAQUS explicit. Naturally, I want to check the ABAQUS results against hand calculations when I can. In this particular model, I have a simple rectangular plate, pinned boundary conditions on all 4 sides and a non-dynamic uniform load applied (load is applied and held through all time steps. Pressure does not vary with time). very small weight density is applied - to removed the weight effect on the reactions. I made the model like a simple static model to check that I am using the explicit parameters correctly.

The sum of the reactions should equal the pressure X plate surface area.

The FEA results gives me about 85% of the total reaction load from hand calculations. I am checking the last load step reactions, so I am not checking any reaction solution before the final load step.

Any suggestions as to why the FEA reactions are low from hand calcs ?

Is there any simple model as I described above that I can run to verify my approach to running the model ? Perhaps I am setting an ABAQUS parameter incorrectly.

Thanks for any thoughts.

 

[Dave442]

Hi,

You are trying to run a quasi-static analysis. As such, you should compare the kinetic energy in your model to the internal energy to ensure that inertial effects are negligible. If inertial effects are non-negligible your results may vary from the static solution.

Check out the section in the user manual that covers quasi-static explicit analyses.

Good luck,
Dave

 

[Denial]

I have never used Abaqus, so can only offer a suggestion based on my usage of various other FE suites.

Some suites ignore loads that are applied directly into supported degrees of freedom.[ ] The results will then appear not to be in equilibrium, but the deflections and stresses will still be correct.[ ] Since you are performing a simple test analysis, you have probably modelled your plate with a minimal number of elements[ ] If the UDL(?) you have applied to your plate elements has been converted (by Abaqus) to nodal loads before it begins the main analysis, then this might be the cause of your apparent discrepancy.

Try a further test.[ ] If this is what is happening, then if you modelled your rectangular plate with 8x8 elements you would expect, very roughly, to "lose" about 12% of the applied UDL (depending upon the element formulation). If you use 16x16 elements instead, you would expect to "lose" about 6%.