Simulation problem of compression test of cold formed steel columns

[Tinni1]

Hello,

I am trying to perform a simulation of a compression test of a lipped channel cold-formed steel column using Abaqus static risk analysis. The CFS column is connected with a small piece of track at the top and bottom with the help of a self-tapping screw.

In my test setup, I have a heavy hot-rolled steel support assembly at the top and bottom of the CFS column.

This support assembly I have simulated with a discrete rigid plate at the top and bottom. The rigid plate is connected to the tracks with tie constraint. Between the stud and the track, I have defined node-to-surface contact with tangential behavior friction less and normal behavior hard contact.

My test is displacement controlled, hence I have applied a displacement-controlled loading.

By performing the analysis I got the peak load close to that obtained from testing. However, my test load Vs axial shortening curve has a much lower slope whereas the analysis curve is having a much higher slope.

Could anyone provide any direction, on how I should approach to solve this issue? Any suggestion will be much appreciated!

The attachment shows the test support condition and also the Abaqus model.

[URL unfurl="true"]https://res.cloudinary.com/engineering-com/raw/upload/v1661021153/tips/Testing_simulation_query_e2yxko.docx[/url]

 

[FEA way]

Do you account for imperfections in the analysis ? Try refining the mesh and changing the element type. Also, make sure that boundary conditions correctly represent test supports. Those are the most common reasons for too stiff models.

 

[Tinni1]

Hello,

Thanks for your response.

Yes, I have taken measured geometric imperfections in the analysis.

Since the imperfections are measured manually, the overall imperfection profile is not a smooth curve.

Do you think, that could be a reason for a stiff model?

 

[FEA way]

You can try applying different imperfections but this would likely influence the limit load as well. Can you share both curves (from the test and from the simulation) ?

 

[Tinni1]

Thanks,

Can we get into a meeting through a consulting service?

 

[g.alshamsi]

What material model are you using? Are the material properties (young's modulus, Fy, Fe etc...) from a coupon test or did you use typical values found in the literature?

For Cold-formed steel, it's proven that the S9R5 element yields the best results, I'm assuming you're using the S4R element. If S9R5 is not an option you can try the S8R.

 

[Tinni1]

Hello,

Thanks for your suggestion. I used the material properties as per coupon test data and I have used an elastic perfectly plastic material model. I also tried a two-stage stress-strain model based on literature. The result was not much different.

As suggested by you I am trying S8R (Instead of S4R), but the slope of the analytical curve is still much higher than the test curve. The peak load and the column failure pattern is close to that obtained through testing.

 

[g.alshamsi]

Can you please post the experimental curve vs FEM curve here? Also are there any connections in your test setups that use self-tapping screws?

 

[Tinni1]

Thanks for your response.

Yes, I have self-tapping screws at the stud to track the connection.
I have modeled a tri-linear stiffness of these screws based on the literature.

The test Vs FEM curve is below:[URL unfurl="true"]https://res.cloudinary.com/engineering-com/raw/upload/v1661920305/tips/simulation_problem_exyfr3.docx[/url]

 

[g.alshamsi]

Looks very off, I'd start by re-checking the boundary conditions.

good luck,