How to expedite dynamic implicit quasi static analysis

[Tinni1]

Hello,

I am performing a numerical simulation of the compression test of the cold-formed steel stud and track assembly.

In a practical scenario, there is a small gap between the stud and track, due to the rounded corner. And the experimental curve has

To simulate the gap closing behavior, I have defined contact conditions.

The geometric imperfections have been modeled, by conducting an initial linear buckling analysis and then scaling the buckling mode shapes with the appropriate scaling factor.

For post-buckling analysis, I am using the dynamic implicit quasi-static step (Initialy tried the static risk method, but that is giving a convergence problem). I have conducted a frequency analysis and the time period of the quasi-static step is 10 times the natural frequency of the system as suggested in the Abaqus user manual. I have applied a displacement-controlled loading where 100 mm displacement has been applied in 2.5 sec time.

The top and bottom support arrangement of the test assembly has been idealized as discrete rigid plate elements.

My problem is: my analysis is very slow. The gap between the stud and track is 3 mm (considering center line modeling) and is the gap is not yet closed due to slow analysis.

could you please advise, on how could I expedite this analysis?

The picture of my numerical model showing the initial gap is attached below:

[URL unfurl="true"]https://res.cloudinary.com/engineering-com/raw/upload/v1664196655/tips/slow_quasi_static_analysis_irvuvo.docx[/url]

 

[FEA way]

Maybe there are still convergence issues, check the status file. Some highly-nonlinear problems can be solved more efficiently using quasi-static analyses in Abaqus/Explicit.

 

[Tinni1]

I checked the status file, and it shows that in many time steps, the solution is unattended, but then after that, it converges.

Let me give some more details of the problem:

Initially, I modeled the columns with measured geometric imperfection with a 5 mm x 5 mm mesh. In this model, I was able to simulate the gap-closing behavior by defining contact conditions and using the dynamic implicit quasi-static analysis.

Analysis was not slow at all. I achieved the peak load and failure pattern, the same as testing.

But the analytical load Vs stiffening curve was very stiff as compared to testing.

Since, I used measured geometric imperfections, the imperfection profile, and hence the numerical model is not a smooth profile. I thought this could be one of the reasons for the stiff curve.

Hence I am exploring the buckling analysis method, which can enable me to obtain a smooth profile. In this case, I have refined the mesh to 2mmx2mm, with S4R elements.

But then analysis has become very slow in this case.

I am attaching the test curve and analytical curve below:
[URL unfurl="true"]https://res.cloudinary.com/engineering-com/raw/upload/v1664198267/tips/Abaqus_analysis_problem_i60xnv.docx[/url]


Could you please provide any direction?

 

[FEA way]

Abaqus/Explicit may also be used for postbuckling analyses involving imperfections. We usually resort to it when there are large nonlinearities causing convergence issues or slow convergence in Abaqus/Standard. You may find some comparisons and benchmarks showing the performance of static (with or without stabilization), Riks, dynamic implicit and dynamic explicit procedures in nonlinear buckling problems.

 

[oldNail]

You cannot expedite convergence until you know the cause of the slow convergence. The web tool

https://msgfile.info

can help summarize issues from the .msg file so you can have some ideas. As FEA way suggested, sometimes Explicit is a more efficient option for buckling but it brings its own challenges.