Non-linear Buckling analysis with ANSYS 1

[koenigsegg]

Hi all,

I'm currently having some troubles analising a thin walled conical conduct subjected to the atmospheric pressure on the outside (vacuum inside). The problem is that no matter what I do I can't find the buckling load (or mode) with a non-linear analysis: the structure simply deforms like there was no imperfections and never buckles.

If anyone could help me I'd appreciated.

Thanks in advance.

Here is the code that I used to perform the analysis:

/SOLU ! Enter the solution mode
ANTYPE,STATIC

PSTRES,ON

!boundary conditions

SFA,ALL,1,PRES,-0.1, !Apply atmospheric pressure

LSEL,S,LOC,z,0
DL,ALL,,UX,0
DL,ALL,,UY,0
DL,ALL,,UZ,0
LSEL,S,LOC,z,900
DL,ALL,,UX,0
DL,ALL,,UY,0
lsel,all
asel,all
nsel,all

SOLVE
FINISH


!************************Eigenvalue buckling analysis ***************************

/SOLU ! Enter the solution mode again to solve buckling
ANTYPE,BUCKLE ! Buckling analysis
BUCOPT,LANB,6
SOLVE
FINISH

/SOLU ! Re-enter solution mode to expand info - necessary
EXPASS,ON ! An expantion pass will be performed
MXPAND,6 ! Specifies the number of modes to expand
SOLVE
FINISH

/POST1 ! Enter post-processor
SET,LIST
SET,FIRST ! Read in data for the desired mode
PLDISP ! Plots the deflected shape



!****************************Nonlinear buckling analysis *************************
*GET,MYLOADF,ACTIVE,,SET,FREQ
/PREP7
/INQUIRE,MYJOBNAME,JOBNAME
UPGEOM,0.005,1,1,%MYJOBNAME(1)%,RST
!F,1254,FX,1
/SOLU
ANTYPE,STATIC
NLGEOM,ON
TIME,MYLOADF*1.1
NSUBST,20,1E5,20
OUTRES,ALL,ALL
LNSRCH,ON
SFSCALE,ALL,1.1*MYLOADF
/SOLU
SOLVE

 

[Stringmaker]

So it sounds like there is an imperfection present in your structure? That's only the first step. You need to introduce some form of perturbation to induce buckling. This could be a small point load at a node. Figuring what to make this and where to put is one of the most difficult things to determine for buckling.

 

[koenigsegg]

Hi Stringmaker,

Theoretically there is no imperfection in the structure. But to analyse buckling with the non-linear analysis, I´ve indroduced a small geometric perturbation using the command: UPGEOM,0.005,1,1,%MYJOBNAME(1)%,RST

This updates the geometry by introducing a perturbation that is 0.5% (0.005) of the buckling mode obtained by the eigenvalue buckling analysis. I've tested this approach in other test cases and it always worked.

 

[koenigsegg]

Does anyone have any idea about what might be wrong?

Any help is appreciated.

Thanks

 

[crisb]

I dont know the details of your model but is the combination of the vacuum and external pressure having a stabilising effect, so that if any imperfections arise the structure wants to revert back to a regular section?

 

[koenigsegg]

I thought of that too, but if the structure doesn't buckle, how come the eigenvalues analysis returns acceptable loads and modes?

 

[crisb]

Have you tried the model with varying amounts of pressure differential to see if there is a cut off point when buckling doesnt occur?
I am getting out of my depth here but presumably eigenvalue buckling is based on elastic theory and will not reflect nonlinear behaviour. Maybe the nonlinear behaviour is dominant and not secondary, as in some sort of membrane effect.
I remember seeing some kids science program on tv that tried to explain why soap bubbles always end up spherical even when they start off random shapes, and was wondering if there might be some sort of analogy!

 

[koenigsegg]

>>Have you tried the model with varying amounts of pressure differential to see if there is a cut off point when buckling doesnt occur?

Yes, I've tried different loads and none seems to be enough to cause buckling with a non-linear analysis.

 

[koenigsegg]

Hi all,

To all of you that have been following this thread, I just wanted to tell you that I have found what was wrong with my analysis. Apparently, to perform this kind of analysis, the boundary conditions need to be aplied on the nodes rather than on lines and areas (go figure ).

Thanks to all