SolidPrep ("Solid Preprocessor") is a collection of methods developed to automate and optimize the design of continuum structures. It is currently under development, but is usable.
It serves mainly as research software, and contains experimental methods alongside established methods. It is designed to be extensible and flexible, in order to allow new methods to be added and compared to established ones.
Currently features:
- Gmsh/OneLab meshing and GUI
- JSON-based problem specification file
- Topology optimization:
- SIMP
- Convolution density filtering
- Compliance and volume minimization
- MMA and GCMMA
- Beam-based design automation (experimental)
- Fast design of continuum structures based on methods used for sizing beams
- Uses OpenCascade for solid geometry
Dependencies:
- OpenCascade
- LAPACK, LAPACKE, CBLAS
- Gmsh
- VTK (Dependency for OpenCascade, may not be bundled with it)
- RapidJSON (included as submodule)
- MPI (currently only MPICH)
- MUMPS (CMake version)
- Eigen
- PETSc
- GNU Scientific Library
- Catch2 v2 (for tests)
Also includes code from jdumas/mma.
Installation consists on installing the dependencies somewhere CMake can find them (which should be done automatically when building them with CMake or using a Linux package manager). Then, load submodule with:
git submodule update --initAfter that, the program can be built as usual:
mkdir build
cd build
cmake ..
makeThe program has not been tested on Windows nor MacOS. It should work, but it is possible that some modifications may be needed.
To run an optimization problem, just execute the program in a terminal environment with the path to the problem's file as argument, for example:
./build/SolidPrep ./examples/cantilever.json
The JSON files defines the analysis to be done as well as all of the characteristics of the problem (except for the geometry of the ground structure, which should be supplied as a STEP file).
See the examples folder for more information.