FEA / FEM coding

[peto24ap]

Hi all,

I am looking for some textbook, lecture notes or online resource about how to efficiently organize FEM code (from the programming point of view).
I have already wrote some small FEM codes for specific problems, but this time I need to make it bigger (coupling different element types - 1D, 2D, 3D, different solvers for built in material and geometric nonlinearity, and so on..). I am not looking for theory of FEM, just some guideline, how to make the structure of the code as general, clear and transparent, as possible.

Any advice will be greatlly apreciated.

Thank you

 

[bkal]

I am not sure if this is significant, but have you decided which programming language are you going to use? When I was developing the FE codes (last century), I used Fortran 77, and I would imagine that with all the OO etc, the structure would be much different now.

 

[ShellsPlatesMeshes]

Hi,
Try this book; it comes with a CD containing FEA codes, pretty advanced and well-commented:

http://members.ozemail.com.au/~comecau/quad_shell.htm

(It is also available from the Amazon:

http://www.amazon.com/Computational-Geometry-Surfaces-Application-Analysis/dp/0646930818

)

 

[Guest262653]

I've been using the following book:

http://eu.wiley.com/WileyCDA/WileyTitle/productCd-1119973341.html

 

[peto24ap]

Thank you for you suggestions gentleman,

bkal: until present day I was using Matlab, but my current work deals with form finding, nonlinear analysis and structural optimization of shell and tensile membrane structures. I need quiet strong pre-processor for that, so I decided to build a Python application for a Rhinoceros CAD software. My work until today was relatively easy, since Matlab has many useful options and procedures already build in. With Python I have to start from the ground. Had you been developing FE codes in a commercial level ? What was your specialization ?

ShellsPlatesMeshes: Thank you for your suggestion. I saw this book couple of months ago somewhere on the web, but I was affraid that the stuff in there will be more mathematical then practical, and pretty much advanced for my relativelly common problems. I was affraid to spend 200 dollars for something I will not be able to use. Do you have some experience with the book ?

avscorreia: Thank you. It looks like a good beggining !

 

[IDS]

I can recommend:
Programming the Finite Element Method, 5th Edition by Smith Griffiths and Margetts

http://au.wiley.com/WileyCDA/WileyTitle/productCd-1119973341.html

Also my blog has a fair bit of open source frame analysis code, mainly in VBA with links to compiled code in Fortran, C++, C# and more recently Python.

Doug Jenkins
Interactive Design Services

http://newtonexcelbach.wordpress.com/

 

[ShellsPlatesMeshes]

peto24ap: The very reason why I suggested that book is that the CD-ROM contains really working codes for the FEA of shells. You just copy them to you hard disk, compile/link into an executable (using any C/C++ compiler, like MSVC 2008 Express Edition, which is downloadable from the Microsoft website for free). It takes just 2-3 minutes. Of course, the codes are large and rather complex; but so is the problem itself: FEA of shells. I am not aware of any other book that would contain working FEA codes for arbitrarily complex shells.

 

[bkal]

I was doing my programming in late 80-es, and I was at Swansea at the time, so influenced by authors such as Zienkiewicz, Owen and Hinton, but also Bathe and Wilson. In my case, the main area of interest was development of material constitutive models and I used a commercial, off-the-shelf pre- and post-processor. Because of that, Fortran seemed like a good choice for a number crunching application.

 

[IceBreakerSours]

Python has tonnes of freely available/downloadable libraries, just in case you weren't aware. One just has to search a little bit to find the right library.

Are you new to this forum? If so, please read these FAQ:

http://www.eng-tips.com/faqs.cfm?fid=376

http://www.eng-tips.com/faqs.cfm?fid=1083