Finite Element Analysis? Any hints?

While I have taken a course on the Finite Element Method during my "master of engineering: architecture", I have never used this in practice, so it has faded considerably over the years. Since performance simulation and Building Information Modeling go hand in hand, I want to reload some of that knowledge. So I have been looking for some software tools to connect to a digital design workflow.

So far, it is cloudy... but I want to share some of it here.
  1. There are as many approaches as there are engineering problems... structural mechanics, acoustics, heat transfer and energy, ... They all can apply these methods. Most software seems to focus on one approach.
  2. It's a Windows world. Face it, most engineers work on Windows. Some of them use Linux. Most of them ignore Macs. So most engineering software is Windows-only.
  3. Most systems are commercial. Some Open Source systems are really open in the sense that they can be used as libraries to create a custom solution. This takes effort and requires experience and knowledge, which makes it less likely to lead to a finished usable product that even an architect can use.
That said, there are some solutions I'd like to mention here.
  1. OpenFoam: Open Source CFD (Computational Fluid Dynamics), to be used to solve fluid flows, chemical reactions, turbulence, heat transfer. Linux only.
  2.  JCAE : Java based Computer Aided Engineering, with volumes and finite elements modeler. Based on OpenCASCADE and cross-platform (currently supports Windows and Linux). No analysis, but can be used to prepare a mesh for analysis.
  3. Geometry Gym: a series of modules for Rhino Grasshopper (thus Windows-only) to link Grasshopper with a solver to assist interactive form-finding with engineering analysis.
  4. Karamba : a module for Rhino Grasshopper, where a direct analysis can be linked to an interactive Grasshopper construct. There are real limitations, e.g. rods are fixed to steel, have fixed profile, nodes are fixed etc... This makes it faster, but requires separate analysis if this does not fit your design. Made in collaboration with Bollinger-Grohmann-Schneider.
  5. Load Design : work-in-progress of a FEA software made specifically for OSX. While still a bit rough at the edges, it is already quite complete and usable. Input is text-based, which is not uncommon. Output is graphical, although import- and export are mostly non-existant right now. Free to use while under development.
  6. Salome is an Open Source platofrm for pre- and post-processing numerical simulations. It builds on OpenCASCADE and can be used as standalone software. Focus is on Linux (but binaries for Windows are available). Some parts it have been ported to OSX within the efforts around the Python wrappers for OpenCASCADE.
  7. FEAPpv : personal edition of a FEA toolset, as part of a course book. This is a free and open source edition that can be used for learning. You need a Fortran compiler. Available for Windows and Linux.
  8. Code Aster : series of tools for structural and thermodynamic simulations. Open Source and using Python (but also Tcl/Tk, GMSH). Can be used in combination with Salome.
  9. CAELinux : full Linux distribution packed with engineering tools, including Salome, Code_Aster, OpenFOAM...
Even though, many engineers will still rely on the well-known commercial solutions primarily, e.g.
  • Ansys
  • Scia products
  • BuildSoft products
  • Nastran
  • Abaqus
---- Added Section ---
  • miniFEA : another nice link, with a FEA program running inside a Web Browser (Java or Flash versions are available). This is targeted at education (students, instructors).

Comments

  1. Very glad you brought up this topic.

    Windows world - yes, definitely. All the Linux stuff is somewhere at the joint between engineering and proper science, but it fades quickly as you look at more practical tasks.

    Ansys, Abaqus, Nastran are all high-end and super expensive, and they are mostly into MCAD. As for AEC, they are basically either for research applications or for some really complex and innovative structures.

    The mainstream software for engineers is numerous and vendor lock-in is incredible here since the learning curve is usually crappy. Local standards support is crucial for software to get any attention (I'm talking about AEC FEA all the way here) and some sort of certification will likely exist, either in form of recommendations or some real certification like we have here in ex-USSR.
    It is crucial to the extent that me and my team had to lead the certification process (nuclear sector!) for the software we use all by ourselves with very little awareness from the vendor (let's not point fingers but its name begins with "Auto" and ends with "desk").

    The lack of some kind of a community-aware software recognized internationally by most (or at least some) engineers is a major pain.
    I'm drafting some crazy projects all the time to deal with that.

    ReplyDelete
  2. You might have a look at Elmer, see http://www.csc.fi/elmer. It is a true multiphysics code covering all major application areas, also supports Windows, and is Open Source.

    ReplyDelete

Post a Comment