For inspecting the data, these packages are suggested:
paraview
hdf-compass | hdfview
To use the Makefile you need hdf5-tools as well as the ParOSol package
installed.
Use this repository: reox/parosol-tu-wien and follow the build instructions there.
This version is also slightly different in it's dataformat than the original version or the tu-wien version from Elan.
It's probably best, to check the output format and adjust it to your needs. A fork on bitbucket also contains code to switch different output formats.
The Makefile contains the bare minimum to run an analysis on the samples from
the ParOSol repository.
First, adjust the paths in Makefile. We assume that you checked out the
parosol-tu-wien repository one folder below this repository and already
successfully build ParOSol.
Otherwise you have to adjust the paths of parosol and createxmf.py binaries
too.
By default, we configure mpirun to use all available physical CPU cores. Be
aware of that!
The example will use the mesh sphere.h5.
If you run make, ParOSol will make the calculations and write the results back
into results/sphere.h5. Also a results/sphere.xmf file will be created.
Open this file using paraview.
If everything was successful, you should see something like this:
Using the tool h5dump, you can inspect existing data.
A description of the dataformat is available in the userguide.
Beware: This userguide is wrong in some places! Use it with caution!
Depending on the version you use, a /Parameters group is required (it is never
used though).
The Makefile adds it before passing the data to ParOSol, but you can also add
it already beforehand.
The Image_Data/Image dataset contains the voxels, where each value corresponds
to the elastic modulus of this voxel.
Each voxel will be a single element in the resulting mesh.
Note, that this Data is given in Z, Y, X, which means if you are writing data
using h5py, you need to provide a numpy array using the shape (z, y, x).
Also the boundary conditions (displacements or loads) are given in this format.
But, the resulting Mesh will be again in X, Y, Z format! That means, if you
are reading the mesh and write back to the image, you need to swap X and Z.
A last note: ParOSol works directly on the given file. That means, if you do not want your input data to be changed (allthough ParOSol will not write in the Image data), you need to copy the file.
There are two types of boundary conditions known to ParOSol: Loadings and Displacements/Fixed nodes. BC only act on nodes, and therefore node coordinates are used in the creation of boundary conditions.
The creation is a little bit tricky, as the format how the node coordinates is
not the same as in the Mesh output.
Basically the coordinate given here, is not the absolute coordinate in the unit
of voxelsize, but the number of the node along an axis.
What does this mean? Here is a short example:
If you have an image with x * y * z voxel, you will end up with
(x+1) * (y+1) * (z+1) nodes. The coordinates for your boundary condition are exactly
these numbers.
In the output mesh, the coordinates of the nodes are in the unit of voxelsize!
Beware, that you need to give the boundary condition also in the (z, y, x)
format - the same as the image data!
Again, the mesh data is then in (x, y, z) format.
Units are not used inside ParOSol. But If you use a consistent unit schema, all your results will also be consistent.
TO BE CONFIRMED
Input Data:
- Voxelsize: [mm]
- Loading on nodes: [N] ?
- Displacements of nodes: [mm]
- E-Module (Value in
/Image_Data/Image): [MPa]
Output Data for elements:
- Strains: all six sym. components of strain tensor
- Stresses: all six sym. components of stress tensor
- von Mises stress
- Strain energy density (
1/2 * sigma_ij * epsilon_ij) - Effective Strain (
sqrt(2*SED/YoungsMod)
Output Data for nodes:
- displacements, as vector
- forces, as vector