Exporting to ParaView

Exporting to ParaView

pyfdstools.paraview converts every FDS output type to VTK, so the results can be explored interactively in ParaView.

Installing

The export routines need pyevtk and numpy-stl, which are not core dependencies:

python -m pip install "pyfdstools[paraview]"

Until they are installed, reaching for one of these names raises an ImportError naming the extra:

ImportError: pyfdstools.paraview requires optional dependencies which are
not installed (No module named 'evtk'). Install them with
'pip install pyfdstools[paraview]'.

Nothing else about pyfdstools changes; importing the package works either way.

Exporting

Each routine takes the same arguments and writes one file per timestep plus a .pvd index. Open the .pvd in ParaView and it loads the whole time series with the timeline populated.

import os
import pyfdstools as fds

workingDir = os.path.join(
    os.path.dirname(fds.__file__), 'examples', 'data', 'case001.zip')
chid = 'case001'
outDir = './vtk'

fds.exportSl3dDataToVtk(chid, workingDir, outDir=outDir)   # 3-D slices
fds.exportSl2dDataToVtk(chid, workingDir, outDir=outDir)   # 2-D slices
fds.exportBndfDataToVtk(chid, workingDir, outDir=outDir)   # boundary files
fds.exportBndeDataToVtk(chid, workingDir, outDir=outDir)   # &GEOM boundary
fds.exportPrt5DataToVtk(chid, workingDir, outDir=outDir)   # particles
fds.exportS3dDataToVtk(chid, workingDir)                   # smoke3D

Argument	Effect
outtimes	times to export; every output time when omitted
outDir	where to write; alongside the results when omitted
binary	appended binary data rather than ASCII, default True
dtype	VTK datatype name, chosen from the data when omitted
ftype	'ImageData' or 'RectilinearGrid', slices only

Uniform or stretched meshes

ftype='ImageData' (the default) stores only an origin and a spacing, which loads faster and uses less disk. It is only correct for uniformly spaced meshes. For a case using &TRNX, &TRNY or &TRNZ, use 'RectilinearGrid', which stores the coordinates explicitly:

fds.exportSl3dDataToVtk(chid, workingDir, outDir=outDir,
                        ftype='RectilinearGrid')

Exporting a subset of times

A long run produces a lot of files. Restrict them:

import numpy as np

fds.exportSl3dDataToVtk(chid, workingDir, outDir=outDir,
                        outtimes=np.arange(0, 601, 30))

Smoke3D

Smoke3D data are stored as single bytes and are exported that way by default, which keeps the files small. Pass decode=True for physical units:

fds.exportS3dDataToVtk(chid, workingDir, decode=True, dtype='Float32')

Geometry

ParaView cannot read FDS obstructions, so export them as an STL to show alongside the field data:

fds.obstToStl(workingDir, chid, outDir=outDir)

This writes <chid>.stl covering every rectangular obstruction in the case. &GEOM surfaces are already triangulated; export them with exportBndeDataToVtk.

In ParaView

1. Open the .pvd file. ParaView loads every timestep and enables the timeline.
2. Add the .stl with File > Open and set it to a solid color to show the geometry.
3. For a 3-D slice, Volume representation gives a smoke-like render; Slice and Contour filters cut through it.
4. For particles, apply a Glyph filter and scale by a particle quantity.
5. Colour by a quantity from the toolbar, then Rescale to Custom Data Range and fix the range so it does not rescale on every timestep.

VTKHDF

Recent FDS versions can write VTKHDF directly. To read that back into the same dictionary layout query2dAxisValue returns:

python -m pip install "pyfdstools[vtk]"

data, units = fds.query2dAxisValue_vtkhdf(
    workingDir, chid, 'TEMPERATURE', axis=1, value=2.55, time=30)

The result has the same x, z, datas and times keys, so downstream code does not need to know which source it came from.

The standalone script

pyfdstools/example_paraview_export.py predates the paraview module and duplicates part of it. It is kept as a worked example of the VTK file format; use pyfdstools.paraview for real work.