Elmer validation/limitations

It's been a while since I've been on these forums, wanting to get back in the swing of things. Nowadays I need to do multiphysics simulations, and so it seems the best and only choice for me is Elmer. However, I'm wondering what the limitations of it are, if any. Also, I noticed that software like Code_Saturne is validated against physical results, has Elmer been validated in this way? I've already looked on Elmer's website a bit, wondering if someone can give me some straight answers. I may ask this on Elmer's forums directly as well. Thanks in advance for any answers.

I thought that Code_Aster was mostly geared towards mechanics, whereas Elmer was literally geared specifically towards multiphysics simulation. Upon reading a little bit, I see that Code_Aster can do multiphysics coupling, so basically you can simulate things you can write code for. But can Code_Aster do things like electromagnetics out of the box? Code_Aster and Code_Saturne have the the advantage that they are validated and backed by one of the largest electricity producers on the planet though. This is what my question is about; wondering if Elmer has any such spectacular credentials.

Hi Jake

Elmer has been verified always when new models have been introduced. The verification may be againts simple analytical solutions, or against results from other codes. The verificiation is not complete in the sense that all possible choices by the user would have been verified. That would probably be impossible but generally also the user is assumed to know what she is doing.

The philosophy of Elmer is quite different from Code_Aster and Code_Saturne since these codes are the backbone of the simulation activity of EDF and verification and validation cannot be taken lightly if you have nuclear power to consider. This on the other hand, makes the codes more stale. It is more difficult to introduce new features upstream in a catheedral type of development model. In Elmer we emphasize agility and strive to have a more bazaar type of development model. Sometimes good ideas are implemented the same day, and people may even be given right to commit code. Combined with the extreme modularity of Elmer I would guess that Elmer is more easily adaptable, but on the other hand, if you're doing what EDF is doing you'll find better validated models there.

It is quite difficult to say that Elmer would not have any limitations. There are so many fields: fluids, structures, electromagnetism, acoustics, transport, etc. Usually people will have some specific field in their mind and do not maybe think of the number of fields. The main strength in Elmer is that these fields may be coupled without an a priori defined way. In individual fields among OS codes OpenFOAM is certainly more able in high Re flows. Elmer excells more in FSI, and recently also perhaps in acoustics and electromagnetism where there are quite modern models. From developers point of view there are actually quite little limitations as you can always write new code, but the time-to-solution may just become impractical if there are existing solutions in some other codes.

Peter