Ansys Advice

[DHKpeWI]

I am considering purchasing Ansys and would like some advice from someone who has used it in the past. What is your view of the software: Easy to use? Easy to learn?

I need to be able to detect Euler buckling and "local" buckling in "slender" compression elements similar to stiffeners in a plate girder. Has anyone used Ansys for this or similar applications?

In this particular application the stiffeners are connected with intermittent welds. Since the stiffeners are thin gage steel and the clear span between welds is +/-12", the stiffeners can buckle between the welds.

Thanks,
DHK

 

[rb1957]

i thought we cleared crippling with a hand calc ?

 

[rstupplebeen]

I have found Ansys to be the least intuitive main stream FEA program that I have used. Here is a list of front ends that I have used in my preferred order.

Abaqus CAE
Femap
Comos/Simulation (inside SolidWorks)
Patran
Ansys workbench
Mechanica
Ansys classic
MARC Mentat

Remember this is just my opinion and you should probably get some free trials and demos to make you own judgment. I hope this helps.

Rob Stupplebeen

 

[spongebob007]

I would add a vote for NX/NASTRAN. Your CAD and analysis tool are all in one and it is much more powerful than that crap that you can get as an add-on to solid works. NX has some very powerful tools for CAD geometry simplification, and if you use NX as your CAD tool, you can have design,drafting, and analysis all in one tool.

I like ANSYS Workbench and at the time I thought it was the best thing going, but NX is so much better on the model perperation side of thing. NX will also allow you to work with FEA models generated in other codes and you actually don't have to use NASTRAN as a solver. You could use NX as the front end and ANSYS as the solver if you wanted.

 

[BlasMolero]

Dear DHK,
Please consider to take a look to FEMAP V10.1.1 & NX Nastran 7.0 bundle, very professional, and relation price/performance is the best of the market, in the basic bundle you have linear static, frequency analysis, buckling, heat transfer and nonlinear analysis, you have CWELD elements to perform spot weldings, preloaded bolts, composites, and a complete & professional finite element library to mesh in FEMAP any real structure.
Best regards,
Blas.

 

[FeX32]

Check out Comsol.
If you need the flexibility of being able to write sub codes within the solver for a particular application, I find this one best. It's coding is essentially based on Matlab code.

Ansys is a good software as well.


Fe

 

[flash3780]

Ansys is good software (and widely used). Their Workbench program is much more intuitive than their previous GUI (although there is some functionality that hasn't been transferred over yet).

I prefer ABAQUS/CAE over Ansys Workbench since it has a command line interface, and I feel that Workbench is a bit too detached from the math that's actually going on behind the scenes. Although, it's still a good program, don't get me wrong.

I've also used Pro/Mechanica, which is an interesting piece of software. They use a p-method solver, which increases the order of the elements in areas of higher stress. So, rather than using more elements, they use smarter elements. The problem that I have with it is, it's a bit too much of a "black box". The users don't have much control over the mesh. A bad mesh is a bad mesh, even if you have 24th order elements.

 

[spongebob007]

flash,

Out of curiosity, why do you prefer a command line interface? I used ABAQUS over ten years ago and I hated it because of the command line interface. I don't want to have to type in lines of commands to do simple things. To get around this, I used HYPERMESH for pre/post and only used ABAQUS as a solver.

I also don't get how you feel that Workbench is a bit too detached from the math. Is it because it is so simple to use that you don't need a PHD in analysis? I would say that statement you made applies to Pro/Mechanica. I agree with your assesment about it being too much of a black box. We demoed it and I hated it for the exact reason that you had no control over the mesh.

One thing I didn't like about Workbench (and it may be better now) is that it relied too heavily on the CAD geometry and if you wanted to use anything other than 3D tets it was somewhat of a pain to do and you needed to purchase the design modeler program. I work extensively with shell, beam, and lumped mass elements and the thing I think that stands out about NX is that it is very easy to create 1D and 2D elements from 3D geometry. I can also create elements manually without underlying geometry if I need to. You can also manually edit the input deck from within NX.

 

[flash3780]

Spongebob,
The ABAQUS that you used 10 years ago and the ABAQUS of today are different animals. Their interface CAE has improved immensely.
As far as Ansys Workbench goes, it's a great product. The only thing that can be a little frustrating is if you're trying to look at specific nodes or elements, sum nodal forces, etc. Things that could be easily accomplished in a command line require scripting. If you want that sort of flexibility, you have to use Classic Ansys (which doesn't have nearly as nice of an interface).

 

[DHKpeWI]

Guys:

Thanks to all for the input on Ansys and other FEA software.
Has anyone ever done an Euler buckling or local buckling analysis using FEA? A local buckle is a phenomenon in a cross section with elements that are "thin" with respect to their width. When these "thin" elements are in compression a ripple or localized buckle will occur before the material can reach its yield stress or before the Euler buckling load is reached.

 

[BlasMolero]

Dear DHK,
Euler Buckling is the term used to denominate clasic linear buckling analysis, is the type that all FEA packages perform when running a linear buckling analysis, OK?. Here you are an example I did prepare a months ago to show linear buckling analysis:

http://www.iberisa.com/soporte/femap/pandeo/pandeo_euler.htm

When I have free time I will include more examples meshed with Shell elements, I have a few in the "pipeline".

Buckling is a nonlinear phenomenun (by the geometry), then use linear buckling Euler results with caution, performing a nonlinear analysis will give you the correct load vs. displacements curve.

Best regards,
Blas.

Best regards,
Blas.

 

[ajose]

DHK,

Answering your question, yes ANSYS Multiphisycs Workbench, the one i have worked with, perform the plate buckling analisys.

By now there are several ANSYS on market, the classical one and the workbench. Besides those you have Multiphisics, Mechanical, Civil, etc.

The classical one has on line commands.

The wokbench is like a windows program and you have the options of in line commands (inserting on problem tree).


Ajose

So have to be carefull on choosing process.

 

[rybose]

Generally speaking there are two ways to analyze instability (buckling) in FEA: linear eigenvalue analysis and geometric non-linear analysis.

Linear eigenvalue analysis is typically the first step. It returns the buckled mode shape ("Euler" or local depending on which is more critical) and the expected buckling load in terms of a multiplier of the applied load. Linear eigenvalue analysis does not return the magnitude of the displacements nor does it give you any information about postbuckling load carrying capability.

Linear eigenvalue analysis is typically a conservative estimate of the buckling load because real structures have geometric and material imperfections and load eccentricities. Generally speaking, the sensitivity to these imperfections depends on the stability of the postbuckled structure.

Compression panels with thin facesheets and stout stiffeners are known to have the ability to carry substantial load after local "facesheet" buckling (stable postbuckling behavior). Geometric non-linear analysis can be used to load a model past its 1st buckling (bifurcation) load to see how much more load it can take before material failure, collapse etc.

Running the geometric non-linear analyses are very difficult and expensive to run. Their results are often highly dependent on (usually) arbitrary parameters - artificial damping, applied imperfections, boundary conditions etc.

Analytical calcs for local buckling and postbuckling for compression panels can be founded in any aircraft stress analysis textbook (Niu, Bruhn etc.).

 

[DHKpeWI]

Thanks to all for yor thoughts.

rybose:
Which Michael C. Y. Niu book would you recommend?

 

[trainguy]

rybose,

Don't you mean -

Linear eigenvalue analysis is typically a NON-conservative estimate of the buckling load?

i.e. Given imperfections etc., the actual load required to buckle a structure will be lower than the value calculated by linear analysis.

tg

 

[rybose]

trainguy - I stand corrected. Thanks.

DHKpeWI - I'd recommend Niu's "Airframe Stress Analysis & Sizing"

 

[DHKpeWI]

rybose:

Thanks