'Noisy' load-displacement curve for beam with large corner radii

[isok89]

Hi,



I'm performing a three point bending test with a contact plate, displacement controlled. I use shell elements and my cross section is basically a thin walled hat shaped section susceptible to local buckling.

For cross sections with a large corner radii (using hat sections)I have an extremely noisy load-displacement curve. The ultimate loads I get are fine
(0.1% difference with experimental output). For the common corner radii(=used in practice), I have a smooth load-displacement curve. Anyone knows why I am not having a smooth load-displacement curves for larger corner radii?

I am performing it implicit dynamically without stabilization.

'large corner radii' = 10 mm a 15 mm at between 40 a 90 degrees with the connecting elements having
a length of 50- 100 mm

'common corner radii' = 2 a 5, idem

 

[Mustaine3]

Maybe some local buckling first. But with only that description it is impossible to say.