shell element stresses 1

[Mohamed Maher]

Hi all

I have one model done by our sub contractor and he didn't reply for us any more

he provide us sap model contain steel column with shell elements means, its pipe 610 x 30 so he define all the column length by shell elements(shell -thin 30 mm) , my question if i want to check this column I should check stresses S1 AND S2 OR SVM(Von Mises)


Thanks
Maher

 

[FEA way]

It depends on what you want to do with the results. Sometimes stress components are needed, for example for verification with analytical solution. But in most cases, von Mises stress is used since we want to compare the results directly with yield or tensile strength.

 

[Mohamed Maher]

Thanks
yes i want to check the yield limit , other question please what is the difference between stress on top of the shell and bottom of the shell , in 30 mm is it should big difference between both stresses? as i have on top is not pass as a stresses but in bottom is pass as stresses for the same shell element for the Von Mises

 

[FEA way]

Shell elements have multiple integration points in the through-thickness direction and stresses from each of them can be displayed. Usually an envelope is used to check just one value of the stress for each material point. Anyway, you will be looking for the maximum value, regardless of its location.

 

[Insert_Smurf_Here]

I hope you are checking for buckling of the columns as well. The Von Mises will not capture that check assuming your analysis is linear-elastic.

 

[Mohamed Maher]

thanks for the information
the analysis is non-linear analysis is that enough for considering buckling.
I think if the column is shell element it should be checked through buckling analysis and make sure that the factor of mode 1 is bigger than 1 do you agree with that?

 

[FEA way]

With geometric nonlinearity enabled, buckling can occur in the simulation. However, nonlinear buckling analyses are usually performed with imperfections - often in form of scaled linear buckling mode shapes. Anyway, the first step would be to carry out a linear buckling analysis.

 

[centondollar]

The questions you ask indicate that you do not understand the contents of the basic plate model (from which the shell model is derived), i.e., what normal stresses the model provides, and that you do not properly understand buckling analysis. The element type has nothing to do with the requirement to do a buckling check, and furthermore - as "FEA way" mentioned - nonlinear buckling analyses does not provide eigenvalues and eigenmodes, but rather requires some initial mode shape to estimate the imperfection that, when applied to iterative solution of the problem at hand, at some point produces a rapid change in the force-deflection curve.

I suggest that you consult your textbooks and colleagues for advice on how to proceed with dimensioning a steel column (strength and buckling) using non-linear analyses.

PS. The straightforward method to perform this analysis and dimensioning is to start with a beam model. Analyze and dimension it against bending and shear (which is usually uniaxial, and probably is in your case too), and then perform an eigenvalue calculation to determine the linear buckling load. Note that FEM is not necessarily needed for the aforementioned steps. If the beam passes both checks, it is probably adequately sized, but if you want to do more in-depth analysis, you may use the first buckling mode as an imperfection and then perform a non-linear analysis (increase load in steps and trace the force-deflection curve in the transverse and axial direction) to identify a non-linear limit load, which provides a more realistic (and always smaller than the eigenvalue solution!) buckling load.

 

[centondollar]

As another important note, if your subcontractor does not reply to your calls or emails, the odds of their model being correctly made and interpreted are almost zero, so you should throw their results in the trash and either perform the analysis and design yourself or find another subcontractor to perform it.