Recommend FEM software.

Lou Pecora

unread,

Dec 15, 2009, 2:30:09 PM12/15/09

to

I have been looking around the web at various free FEM software
(libraries, frameworks, UI's, etc.) for solving PDEs like Helmholtz
equations. I found two that look rather complete, well supported, and
with good documentation. These are,

OFELI (http://www.ofeli.net/presentations.html) - a C++ framework which
takes some programming in C++ (which I know).

freefem++ - a framework with a UI and its own programming language on
top of C++ code- lots of documentation and looks very mature.

It is hard to tell by just looking at these if either is worth looking
further into. Before spending that time, I thought I ask to see if
anyone has used the code and what they would recommend.

Thanks for any insight and information.

--
-- Lou Pecora

Wes

unread,

Dec 30, 2009, 12:54:38 PM12/30/09

to

Lou,

Thank you for the links. I too have been searching for non-commercial
(i.e. low cost) finite element software. My list has included:
FEMM - GUI-based, electrostatics, magnetics, prewritten modules
FreeFEM++ - scripted, general formula input in approachable terms
FreeFEM3D - scripted, general formula input in approachable terms,
nice CSG input using the POV-Ray language (which I know)
OOMPH-LIB - C++ (I think more complete than OFELI)
OFELI - C++
Elmer - GUI-based, multiphysics, prewritten modules
GMsh+GetDP - scripted, general formula input in very technical/
abstract terms
NetGen+NGSolve - scripted, NetGen is used alone in many tools for
meshing, prewritten modules
OpenFOAM - specialized Navier-Stokes fluid flow - something not done
as generally well by the others listed

Each has many strengths. If anything, their weakness is only the lack
of a strength provided by another package. For instance, I think
Elmer to be a very nice GUI-based package, for solving multiphysics
problems. It can't be used for my purposes as its solvers are
prepackaged modules that don't allow new equations to be added without
Fortran coding, testing, validation, etc. However, FreeFEM++ does
allow new equations (in my case a restriction on the integral of
conductor charge density) to be added but it doesn't provide a GUI to
get started.

Coding in Fortran or C++ isn't a concern, it's the time required to
learn the system, try an idea, and validate that I've used the tool to
ask the model question properly in the first place so that the output
is worthy of consideration. In the end, I've decided to spend time
getting to know FreeFEM++.

Hope this helps,

-Wes

On Dec 15, 2:30=A0pm, Lou Pecora <pec...@anvil.nrl.navy.mil> wrote:
> I have been looking around the web at various free FEM software
> (libraries, frameworks, UI's, etc.) for solving PDEs like Helmholtz
> equations. I found two that look rather complete, well supported, and
> with good documentation. These are,
>

> OFELI (http://www.ofeli.net/presentations.html) - a C++ framework whic=
h

Lou Pecora

unread,

Dec 30, 2009, 6:21:12 PM12/30/09

to

In article <6eadnRr40OZDDqbW...@supernews.com>,
Wes <wes....@gmail.com> wrote:

Thanks for the insight, Wes. I haven't tried any of them. I'm at an
earlier stage than you. FreeFEM++ attracted me. I don't need a GUI,
but I really don't want to hard code in FORTRAN or C++ (like you said,

learn the system, try an idea, and validate that I've used the tool

properly -- that's steep curve to climb in a low level language). A
scripting approach sounds nice since it can provide a way to define the
system (boundary shape, potentials, PDEs, etc) + solving parameters
(level of refinement of the mesh, etc) without the compile, link, run,
debug stuff -- well less of it anyway. I also have to define my own
equations.

Next year I may get a chance to try FreeFEM++ or maybe something else.
If you get a chance to try a one or a few out, please post your
experience. Thanks.

--
-- Lou Pecora