Abaqus, Imperfection

[Entherm]

In Abaqus, when using imperfection from buckle analysis, how do we determine the scale factor?
*IMPERFECTION, FILE = (my odb file from buckle), STEP=1
Mode number, scale factor
What does the scale factor represent? For example, I want my out of plane displacement(U3) to have an imperfection of 0.5, is there a way I can specify imperfection for just U3?

 

[BJI]

You have normalised displacement in the buckling analysis, so use a scale factor of 0.5 to get a maximum displacement of 0.5mm (or whatever your units system is). It is a linear superposition of buckling eigenmodes, so I don't see how you could bias it in one direction. Maybe select a buckling mode that is most suited to the direction of imperfection you are trying to assess (i.e. the first buckling mode might not always be the most appropriate). The alternative is to directly define an imperfection using node coordinate perturbations.

 

[Entherm]

Thank you for the clarification. So as per your explanation, if I am modeling my static analysis model in inches and I place a scale factor of 0.2, the normalized displacement will be multiplied by 0.2 inches. But in my analyses, I used the first mode shape and factor of 0.2 but the results do not look good. Instead, if I use a factor of 4, the results (OOP displacement) looks fine. However, for my model 4 inches, imperfection is too much in a real-life scenario.

 

[BJI]

What do you mean by the results don't look good?

 

[Entherm]

My OOP displacement decreased at a time step which clearly is not supposed to happen. But as I increase the imperfection scale factor, the U3 was no longer decreasing.

 

[FEA way]

When choosing scale factor for imperfections a rule of thumb is to use order of magnitude lower than the structural dimension (like shell thickness or beam cross-section) and then descrease each subsequent scale factor by half.

 

[BJI]

It is an initial perturbation, so any subsequent displacements could only be caused by the step parameters. Are you running with nlgeom on? Could there be snap-through type buckling if your initial displacement is smaller? Hard to know without some images or the input files.

 

[Entherm]

Yes, the NLGeom is on. But the interesting thing is that the problem is seen only for the OOP displacements and not for U1 or U2. Therefore, if I plot a graph of force vs in-plane displacement in Riks analysis I do not see any snap-through.