hi, i read several mentioned in external forum about perfect and symetric section and geometry did not buckle. However, my personal expereinced in CalculiX of shell and solid element and large deformation analysis shown that’s not true. Buckling occurs and detected even the geometry is perfect, both in perpendicular and normal axes load cases.
any simple explanation about this condition? thank you.
eigenvalue extraction in CCX is not state of the art and there are many instances where the software will fail to find the right eigenvalue, please search the forum for more details on specific cases.
it’s related to nonlinear buckling with large deformation analysis, not linear eigenbuckling.
It may not buckle at all or overpredict the capacity so imperfections are almost always needed: Imperfections in buckling design - Enterfea
probably right for thin stocky structure with high stiffness or strength material e.g steel, but maybe not or less sensitive for thick stocky structure with low stiffness and almost elastic e.g plastic ABS or rubber like material. Also, i’m questioning how buckling detected in large deformation analysis of perfect geometry, mesh and loads.
I can only remember perfect geometry not buckling and just tried to set up an unstructured mesh, non-axis-aligned geometry Euler buckling column and it didn’t buckle with either mixed precision Pastix or MKL, direct or automatic incrementation.
Maybe you can share an example that does?
i’m not yet tested on slender and long members due to global buckling (Euler) and i will take a look into.
thanks for looking further about the cases, below some example of cylindrical shell. Buckling shape of sinusoidal wave detected and shown even the geometry, mesh and displacement loads are perfect symmetry.
attached picture and link of INP files.
This is a dubious case. It immediately starts barreling due to the Poisson effect and that barreled shape might act as the imperfection that triggers the shorter wavelength modes that subsequently appear. The first ones of them (like shown in your screenshots) still have the same symmetries as the initial barreled shape.
If you set Poisson’s ratio to 0, it doesn’t buckle.
thanks for some hints, but make zero values become unrealistic, specifically for rubber like materials.
Better a closer control of the load progress.
Cool to see how dimples rotate. That is correct. They are indistinguishable mode shapes of the same buckling mode (rotational invariance)
**
*Step, Nlgeom=on, Inc=10000
*Static, Solver=Pardiso
0.1,1,0,0.1
**
actually, i’m not yet interest in accuracy and correctness of result. Only question how buckling initiate detected in large deformation analysis when the geometry, mesh and load are perfect and symmetry? a simple math explanation behind the solver does.
In a perfectly symmetric system with ideal geometry, mesh, and loading, buckling initiation in large deformation analysis can be understood as the solver’s response to numerical imperfections. While there are no physical imperfections in your setup, numerical solvers introduce slight asymmetries due to factors like round-off errors in floating-point arithmetic operations, iterative solver tolerances, and interpolation schemes.
These numerical imperfections act as “perturbations”, causing the system to deviate from its symmetric equilibrium path. When the critical load is reached, these small perturbations grow exponentially, initiating buckling. This growth is consistent with the mathematical nature of bifurcation problems in stability analysis. In such cases, the governing equations have multiple solutions (e.g., the pre-buckling symmetric state and post-buckling asymmetric state), and numerical imperfections push the solution toward one of the buckling modes.
To summarize, buckling in a “perfect” setup is detected because solvers cannot achieve infinite precision. The mathematical explanation lies in the sensitivity of the stability problem to small perturbations, which inherently exist in any numerical computation.
that make a sense, i heard about this also, thank you for explained lengthen. Another question arise related to long member with global buckling (Euler) previously mentioned and i have taken some test, it almost undetected.
my understanding, a perfect geometry did not buckle is not always true, can be inappropriate for most case. It has specific and dependent: slenderness ratio, material strength or stiffness, boundary condition, element mesh uniformity, etc.
You’re right—perfect geometry not buckling is not always true and depends on factors like slenderness ratio, material properties, boundary conditions, and mesh quality. For long members prone to global (Euler) buckling:
Slenderness ratio & BCs: High slenderness ratios and proper modeling of boundary conditions (e.g., pinned vs. fixed) are critical for properly detecting buckling.
Mesh imperfections: A refined, uniform mesh and small geometric imperfections (e.g., using the first eigen-mode as a perturbation) help solvers capture instability reliably.
Perfect geometry might not show buckling in some cases because numerical imperfections alone may be insufficient to trigger instability, especially for global buckling. Adding controlled imperfections or reviewing boundary and mesh setups can often resolve this.
There is a simple math argument that would support the idea that in an ideal world (perfectly vertical geometry under perfectly vertical distributed load) column would not buckle.
Euler Critical Buckling load, which is surprisingly easy to derive, starts assuming there is an initial lateral displacement y.
Without it, there is no moment at the intermediate section and the solution of the second order differential equation can’t be a buckle shape.
So, without deviation from verticality there is no moment , no buckling. If we see buckling in real life or even in FEA, it’s due to that small “Y’s” emerged by means of numerical error, mesh imperfections, external disturbances,…
probably right, this consistent with modeling imperfection of geometry itself. It’s separated by local and global imperfection, usually local is the ratio with thickness of member related to flatness or perfect curved, but global is total length related to cambering or bowing (out of axes straightness)
