I am never can import a previous work in abaqus if the mesh is not the same. So perhaps you can try to import in abaqus with the same mesh for take into account previous work. And only after in a second step in Abaqus I think you can refine the mesh in specefic areas.

Good morning !

To my knowledge there might me no "simple solution" ... ?!

My suggestion: You have to apply / realize a "transfer" - e.g. by interpolate for the "new" nodes in between the displacements of "old" nodes.

But that sounds like an awful data handling ... ;-)

Dear Marco, if you are using Cae i dont think you can do this. If you are an .inp user then your problem may be resolved via *map solution. This command maps the solution from a previous analysis (with x-mesh density) to a new analysis (with Y-mesh density). This is particularly usefull in large deformation problems. However you need to have 2 different input files. Study the *map solution in the keyword manual and you may find the answer to your problem.

I haven't seen anyone doing this. Nevertheless, you can tell abaqus which set of nodes you want to save in the node file (.fil). I must stress that i'm not sure of this, but maybe you have a way to preserve those nodes numbering (ID). In the new model, you would refine the mesh, but those nodes should stick in place, with the same ID.

On other hand, are you having trouble in catching the desired buckling mode? Can't you model it directly in the geometry by using sine shape assumption? I'm not saying to generate this via .inp file directly, but rather from modifying the python script. If you can produce the surface via CAE, you sure can generate it by script modification.

My main question is why would your mesh for the BUCKLE analysis (i.e. an LBA) have to be coarser than for any subsequent analyses, which I am assuming must be nonlinear (i.e. GNIA). I can see no merit in having a LBA run on a coarser mesh that then GNIA. The LBA is always less computationally intensive analysis unless you are extracting many eigenmodes, which I assume is not the case since you just need the critical one, so why would you need to do this?

I seem to remember that ABAQUS can import shapes calculated on one mesh and impose it on another mesh by interpolation, but this is never a good idea especially if you are modelling imperfection sensitive structures as the interpolation may produce unexpected results.

Just run the same analyses on the same mesh.

One might cast doubt on addressing imperfection by buckling mode shapes. While this technique is the most accurate one, it is not the best way to capture imperfections due to the aforementioned concerns.

I propose the following procedure:

Impose some imperfections explicitly in your model (asymmetric boundary conditions, or some small out of plane loads). If your model was really sensitive to imperfection, it would automatically accounts for the governing buckling mode shape (make sure your model can capture Geometric Nonlinearities). Care should be exercised in defining imperfections. For example if your model is a column with single curvature behavior, apply some lateral load on its top or if your model is a rather flat shell, apply some out-of-plane pressure on it.

This simple technique is well accepted for Civil Engineering structures and have been acknowledged by many building codes.

Almost sure you can't do this in Abaqus. Instead you could export a report file and the interpolate the data with Matlab. By the wat, why do you need a coarser mesh in the less computational intensive analysis. By far the *BUCKLING procedure is faster that the postbuckling analysis you may be performing with *IMPERFECTION.

Dear Marko,

I am not sure if that can be done using Abaqus or other Abaqus tools. On the other hand I think you can use some other pre/post processors to achieve same result. I believe that Hyperworks by Altair Engineering or Ensight by CEI have such capability.

I agree with Seyed,

If you know the desired (critical) buckling shape, the best solution is the use of model without imperfections and apply some small lateral load to "help" in buckling (of course with geometrical non-linearity turned on).

In some cases, with nonlinear solver, it is possible to capture buckling behavior even with NO imperfections and NO artificial loads in model, I did such analysis for big models successfully. The question is, if the desired buckling mode will be captured.

Just try, maybe your problem will disappear. If not, apply small lateral loads.

I had updated the perfect FE model of a shell by using the result file from the eigen value buckling analysis by using update geometry command in ANSYS. Check there may be something similar in Abaqus. You can read my paper for further detail: Buckling of thin-walled cylindrical shells under axial compression, Int. J. Numer. Meth. Engng 2009; 79:1332–1353

Good question.. Please share me the best answer might you trust...

Regards…

Good question.. Please share me the best answer might you trust...

Regards…