© Ihor Mirzov, 2019-2022
Distributed under GNU General Public License v3.0

Downloads | How to use | Screenshots | Your help | For developers | TODO

CalculiX to Paraview converter (frd to vtk/vtu)

Converts CalculiX ASCII .frd-file to view and postprocess analysis results in Paraview. Generates von Mises and principal components for stress and strain tensors.

Creates separate file for each output interval - it makes possible to animate time history. Caution! If you have 300 time steps in the FRD, there will be 300 Paraview files. If you need one file - write output only for one step in your CalculiX model.

Converter is tested on CalculiX examples. Here is how some test log looks like.

FRD reader is tested to reduce processing time as much as possible. Now it's quite optimized and fast, but Python itself is slower than C/C++. Here we can do nothing, so, for example, Calmed converter must be faster - another question is if it's able to read and convert any CalculiX results.

How to use

To run this converter you'll need Python 3 and optionally pipx:

pip3 install ccx2paraview
# or
pipx install ccx2paraview

Please, pay attention that .frd-file type should be ASCII, not binary! Use keywords *NODE FILE, *EL FILE and *CONTACT FILE in your INP model to get results in ASCII format.

Run converter with command (both in Linux and in Windows):

ccx2paraview yourjobname.frd vtk
ccx2paraview yourjobname.frd vtu

Also you can pass both formats to convert .frd to .vtk and .vtu at once.

It is recommended to convert .frd to modern XML .vtu format - its contents are compressed. If you have more than one time step there will be additional XML file created - the PVD file. Open it in Paraview to read data from all time steps (all VTU files) at ones.

Starting from ccx2paraview v3.0.0 legacy .vtk format is also fully supported - previously there were problems with component names.

A snippet for Paraview programmable filter to convert 6 components data array to full tensor:

import numpy as np 
res = np.array([])
pd = inputs[0].PointData['S']
for xx,yy,zz,xy,yz,xz in pd:
    t = np.array([[xx,xy,xz],[xy,yy,yz],[xz,yz,zz]])
    res = np.append(res, t)
tensor = dsa.VTKArray(res)
tensor.shape = (len(pd), 3, 3)
output.PointData.append(tensor, 'S_tensor')

A snippet for Paraview programmable filter to calculate eigenvalues and eigenvectors:

import numpy as np
eigenvalues = np.array([])
eigenvectors = np.array([])
pd = inputs[0].PointData['S']
for xx,yy,zz,xy,yz,xz in pd:
    t = np.array([[xx,xy,xz],[xy,yy,yz],[xz,yz,zz]])
    w, v = np.linalg.eig(t)
    w_ = np.absolute(w).tolist()
    i = w_.index(max(w_))
    eigenvalues = np.append(eigenvalues, w[i]) # max abs eigenvalue
    eigenvectors = np.append(eigenvectors, v[i]) # max principal vector
eigenvectors = dsa.VTKArray(eigenvectors)
eigenvalues = dsa.VTKArray(eigenvalues)
eigenvectors.shape = (len(pd), 3)
eigenvalues.shape = (len(pd), 1)
output.PointData.append(eigenvectors, 'S_max_principal_vectors')
output.PointData.append(eigenvalues, 'S_max_eigenvalues')

Attention! While developing this converter I'm using latest Python3, latest VTK and latest Paraview. If you have problems with opening conversion results in Paraview - update it.

Screenshots

Converted von Mises stress field with Turbo colormap:

Converted translations field with Viridis colormap:

Your help

Please, you may:

• Star this project.
• Simply use this software and ask questions.
• Share your models and screenshots.
• Report problems by posting issues.
• Do something from the TODO-list as a developer.
• Or even become a sponsor to me.

For developers

Install package:

pip3 install ccx2paraview

In your code use ccx2paraview package in this way:

    import logging
    import ccx2paraview.common
    logging.basicConfig(level=logging.INFO, format='%(levelname)s: %(message)s')
    c = ccx2paraview.common.Converter(frd_file_name, ['vtu'])
    c.run()

If you have Python version >= 3.8 create binary with nuitka:

pip3 install nuitka

In Windows:
set CC=C:\\MinGW64\\mingw64\\bin\\gcc.exe
python3 -m nuitka --follow-imports --python-flags=-m ccx2paraview

In Linux:
python3 -m nuitka --follow-imports --python-flags=-m ccx2paraview

If you have Python version < 3.8 create binary with pyinstaller:

pip3 install pyinstaller
pyinstaller __init__.py --onefile

Read how to create packages for pypi.org:

python3 -m pip install --upgrade build twine
python3 -m build
python3 -m twine upload dist/*

Read about VTK file formats and VTK unstructured grid. Remember that FRD file is node based, so element results are also stored at nodes after extrapolation from the integration points.

TODO

Test CALMED binary.

Log memory consumption.

Read binary .frd files.

Read DAT files: it would be a killer feature if Paraview could visualize results in Gauss points. Use CCXStressReader.

Contribute to meshio. FRD writer. Use meshio XDMF writer: https://github.com/calculix/ccx2paraview/issues/6

Add element’s material tangent stiffness tensor. Easiest for the paraview user would be to provide it in the (deflected) global cartesian frame. This dataset is useful for checking input data for anisotropic materials, as well as for the stuff with inverse design of fields of this tensor. But it’s a lot more work to produce, especially with nonlinear materials. It’s almost as useful to see the highest principal value of the stiffness, as a scalar or a vector. (but for the vector you need to do the transformation)