Hey everyone,
I'm modeling a CFS steel shear wall assembly made from back to back columns and a thin steel sheet. I'm using S4R elements and mesh independent fasteners to simulate the screws. General Static step is defined with the Newton Raphson solver. A displacement control loading is applied at the top of the wall to simulate the experiment. My main issue is, I can never get the results of the simulation to match the test results. I'm using a relatively fine mesh (0.5 inch global seed) and the deflected shape and failure modes are in good agreement with experimental results. However, when I compare the force-displacement plots, the model is too stiff and it does not show any material softening at high displacements. The material model used in my model is an elastic-plastic model with isotropic hardening. The stress/strain values were obtained from a coupon test.
After trying many different approaches, a colleague of mine suggested I add geometric imperfections. I performed linear buckling first using a linear perturbation step, I also modified the keyword file to add the displacement variables to the results files using:
*NODE FILE
U
After the buckling analysis, I copied the model in the same directory and performed the displacement controlled static analysis. I included the imperfections by also editing the keyword editor using:
*IMPERFECTION, FILE=Buckle, STEP=1
1,0.8
Where 1 refers to the 1st buckling mode and 0.8 refers to the sheet thickness (scale to the 1.0t of the sheet). The weird part is that the results were exactly the same. I checked the dat file and the imperfections were indeed included in the 2nd analysis however there was 0 effect. I tried using multiple pairs of buckling modes and I got the same results.
I have a feeling I'm doing something wrong, all the literature I checked used a riks step after the linear perturbation step to perform post buckling analysis. What I want is to include geometric imperfections to run a displacement controlled pushover static analysis. Anyone has any experience dealing with such issues? Any comment or suggestion that would shed light would be greatly appreciated.
Mr. Ghaith
All steps are correct as mentioned in your question. In my opinion, the procedure of analysis may be changed to dynamic explicit, or please used the quadratic element (S8R).
Check that NLGEOM is on. Also be sure that controlled displacements are sufficiently high for buckling.
Safaa,
I tried using S8R elements, I'm getting the same results unfortunately.
Victor,
NLGEOM is on, the model is still behaving the same as if imperfections were not included.
Thanks for the input though, the model seems too stiff
Hi Ghaith. Does your model react on the additional small load which mimics geometric imperfection in General Static step?
Victor,
Not sure I'm following your point, what do you mean does the model react to the additional small load?
I mean that you will apply some additional small load (say, concentrated force) to make the model geometrically imperfect. Often this method helps to cause buckling.
Oh, I get what you mean now.
No I didn't try this method, I'm trying to apply imperfections through superposition of buckling modes.
But this method should work as well, I'll read on it.
Thanks!
I suspect that only Static Riks step is needed for superposition of buckling modes.
Have you ever tuned the parameter referring to the quantity of imperfection (0.8 in your question)? I guess you could get the effect of imperfection by adding this parameter to an amount comparable to the thickness. The maximum displacement of the 1st buckling mode in the Buckle.fil is unit 1. It's dimensionless. As a result, the quantity of imperfection in your .inp file must be dimensioned.
If you mean that 0.8 is an imperfection factor, you should multiply it with the thickness：
*IMPERFECTION, FILE=Buckle, STEP=1
1,0.8*
Yachuan,
The quantity of imperfection was calculated as 1*sheet thickness. Sheet thickness in this case is 0.8 so thats where the number came from.
To answer your question, I did try magnifying the imperfection factor by 10, so:
*IMPERFECTION, FILE=Buckle, Step=1
1,8
Still didn't get any difference in results.
Mr. Ghaith,
I performed Static, General analysis of notched steel components using ABAQUS by following the same procedure as you explained. Imperfection you are introducing into your model is 0.8 (1 times of sheet thickness). According to different literature, it can be taken as up to 10% of structural thickness. I think if you vary this parameter, you can find significant changes in your analysis results.
Thank You.
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Materials Science
- Materials
- Inorganic Materials
- Metals
- Alloys
- Steel