LBA: Different Results with Internal VS CCX Solver

Aftab

Hi There
I have analysed below beam for buckling using Internal and CCx solver. CCX solver values are not right, i have crossed checked with another solver. Could someone kindly point out why CCX values are out by a margin.

disla

HI Aftab,

I'm getting both the same result after some arrangements.

-I normally introduce the load as 1 so the buckling factor becomes the buckling load.
-Symmetry as far as I can to reduce computation time.
-Work some minutes on axis orientation of the elements to be more uniform. There was an alert of axis orientation in the details of the calculation process.

¿Does 200/205 KN has sense to you.?

Edited: Total spam 400KN/410KN

Regards

disla

I'm sorry Aftab,

Hope you didn't have a heart attack with my small result.
I have applied symmetry, but your model has one fixed and one sliding support. My fault did not see it.

764KN ccx against 740KN Internal Solver.
760KN ccx once refined.

I think shifting point smaller than 1 is giving some issues.
I still suggest small initial loading ( 1KN) and use a shifting point = 1.

Fix some orientations as there is a warning " *INFO in gen3dnor: in some nodes opposite normals are defined"

Regards

Aftab

Many thanks Disla for your input. Much appreciated, I will check your model in the morning. Many thanks again.

Victor

Beware that some versions of CCX have a bug that excludes modes with a buckling much below 1. The version of CCX 2.16 that comes with Mecway 13.1 is OK but other 2.16's might not be. Disla's way of adjusting it so shift point is 1 avoids this bug.

Using Mecway 13.1's CCX 2.16, I get mode 1 as 0.2748 with the internal solver and 0.2799 with CCX which looks like pretty good agreement. The mode shapes aren't quite the same but I suspect that's because one or both of them have some mixing with nearby modes that have similar buckling factors.

Aftab

Much appreciated Victor , I am using version 2.17 pradiso.

Aftab

Hi Disla
I am trying to verify above LBA by doing a non-linear analysis by exciting the web with a small out of plane load of 100N, i am getting a relatively higher capacity than predicted by the LBA? Could you kindly check whether i am doing this right? Looking from the deflection and checking against the load graph , the web starts buckling from 1000kN t0 1200kN.

Use below link as the file size was bigger than 50MB.

https://drive.google.com/file/d/1kNQ5ZTJjAorDQjr3bb8fd_SmbAkinaZ3/view?usp=sharing

disla

Buckling Load

Loading conditions have been changed as load is now applied to the web.

887KN ccx against 822KN Internal Solver. Without 100N
872KN ccx against With 100N

To be fair the 100N should be applied incrementally as 100*t. (*)

First mode appears at t=0.3s. See Picture.

That is 2700KN*0,3=810KN in line with the LBA.


(*)This is why most of the codes request a minimum of 3 for the buckling factor. Small differences , imperfections or loading conditions can drive the system to a completely different buckling mode.

Aftab

Many Thanks For your assistance, yes you are right, I changed the load for LBA, and could not recall that. Much appreciated.