ANSYS vs. ABAQUS

[h_sadeghy]

Hi,

I have been asked to replace our FEA package with a more versatile and user-friendly one. I have seen this question (ABAQUS or ANSYS?) on many forums but I still could not find the right answer. I have limited knowledge about ANSYS but I have not used ABAQUS and I do not think I would want to learn how to use ABAQUS in order to choose one of them. I have seen people mentioned they are very similar in capabilities but the problem arose when I noticed that the price of purchasing and keeping ANSYS is almost double. I do not know how to justify the extra cost if they are very similar.
I have also seen people advising that the choice depends on the problem we are solving. My problem includes a nonlinear material in contact with a rigid body. I am interested in thermal and mechanical analysis. What software do you choose and why?

Thank you,

[Jorgen]

If those are your main problems that you will work on, and if Abaqus is indeed less pricey, then I would go with Abaqus.
Both Abaqus and ANSYS have similar feature sets related to non-linear modeling, and both are excellent FE packages.

-Jorgen

[h_sadeghy]

Thanks Jorgen,

It seems that I was wrong about the pricing. I was comparing the packages with two different licensing options. with similar licensing option, their price is not very different.

[FEM_Ben]

I have worked extensively with both, and I would indeed agree, depending on what problem you solve, you can either make an adequate choice or a quite bad one. My rule of thumb is as follows, if you solve standard problems with moderate nonlinearities, you will be happy with both programs. But if you have high nonlinearities especially constitutive one, but also geometrical ones, you will see that ANSYS often generates spurious results, which you can rectify if you know what the result will be.

From handling point of view, both have their advantages and disadvantages. I really loved the APDL of Ansys which is possible to use directly in input-file generation. Abaqus has a different philosophy which is very charming on the other hand if you are Python lover, which you will become very quickly if you once started. Python scripting enables you of things with ABAQUS that is just awesome.

Generally for my work (Subroutines, Nonlinearities, Multifield ... ) I do not have any choice, Abaqus is just the most versatile and robust tool available.

[h_sadeghy]

Thanks FEM_Ben,
The part I am working on does not experience high strain (less than 25% in most cases). Is this considered as high, miduim or low nonlinear?
We have decided to go for Ansys, since it also offer CFD which we need to include to the model some time down the road. I hope I made the right choice.
Thanks again

[FEM_Ben]

Hi h_sadeghy,

25% is a highly nonlinear setting. If Ansys does well with singularities and high strains in your setup, just run a mesh convergency study with your model. You can try h-method first and later on p-method if necessary. Do you have a 2d or 3d model? Do you have triangular or hexaedral meshes? CFD capabilities of ANSYS are very poor if you talk about CFX, Fluent is a bit better. If you just need k-eps models, you will be probably suffering a bit, but eventually reach your goals, unless you have FSI problems, this will be a hell of work to validate.

- Ben

[h_sadeghy]

That made me nervous :p. I always thought when people talk about high nonlinearities they mean strain above 50%. I have not got my model set up in Ansys yet. In fact I do not have full license and I am working with a temp license for a month until I get the approval from my manager for the software.
My model is 2D for now and later we want to try 3D. I mostly work with h-method and hexahedron mesh. Any suggestion will be highly appreciated :)