Showing plsect plots in jupyter lab

[MrMechanics]

Hi

I've used the LatheCutter.ipynb example and noticed that you can plot elements or lines with "vtk=False" (mapdl.lplot(vtk=False) and mapdl.eplot(vtk=False)) in that notebook. This produces png images of the model as they would be displayed in ANSYS, and then shows these png-images directly in the notebook. But when I try to do this in other notebooks the png-files are created (although I usually have to run the command twice because the first png is just a 4 Bytes file that cannot be viewed), but don't show up in the notebook. I'm wondering if there is some setting I have to apply somewhere? The only thing I can think that I haven't tried must be some command hidden inside the LatheCutter.anf file.

The reason why I want this to work is because I want to use the mapdl.plsect('S','EQV',vtk=False) command, since I couldn't find any pyansys examples using PLSECT.

Edit: So it does seem to work inside PREP7 (at least mapdl.lplot(vtk=False) and mapdl.eplot(vtk=False) do work), but not in POST1. When I run the mapdl.plsect('S','EQV',vtk=False) command in POST1 I get this message:

---

DISPLAY LINEARIZED STRESS USING NODES DEFINED BY LPATH COMMAND.
ITEM=S COMP=EQV RHO= 0.0000

CUMULATIVE DISPLAY NUMBER 7 WRITTEN TO FILE file007.png - RASTER MODE.
DISPLAY TITLE=
crank_pin_with_contact

---

The notebook does generate the png file though. Any ideas why displaying the image works in PREP7 but not POST1?

[mikerife]

Hi @MrMechanics
Please see this section of the PyMAPDL documentation site:

https://mapdl.docs.pyansys.com/user_guide/plotting.html

The k/a/l/v/e/n plot commands in PyMAPDL are not the APDL plotting commands per se; rather they are implementations of those done via VTK written so that the plots can be embedded into the JL notebook.

Mike

[MrMechanics]

Hi @mikerife

Thank you for the link. I see there that the APDL plots (vtk=False) generate a png file which is then embedded into JL notebook using matplotlib somehow. With some trial and error I was able to create a workaround. I'll post it here for anyone searching for a way to include PLSECT plots into their JL notebooks (mapdl.version 22.2):

# first import these libraries
import matplotlib.image as img                          # used for showing plot images
import numpy as np                                      # advanced python math functionality
import time                                             # used to give mapdl extra time to generate png
from ansys.mapdl.core import launch_mapdl
import os                                               # used to get the current working directory
path = os.getcwd()

# second launch mapdl with run_location and jobname
title='crank_pin_with_contact'
mapdl = launch_mapdl(run_location=path, jobname=title)

# run analysis
mapdl.clear()
mapdl.title(title)
mapdl.prep7()

...

# put a plot somewhere inside PREP7 with vtk=False, which will generate a png file
# if you don't do this inside PREP7 first pngs are not generated later inside POST1
# for some reason
# you can use savefig='plot.png' if you don't want the plot embedded in the notebook
mapdl.vplot(vtk=False,savefig='vplot.png')

...

mapdl.slashsolu()
mapdl.deltim(0.05,0.01,0.1)
mapdl.antype('STATIC')
mapdl.solve()
mapdl.finish()

# start up POST1
mapdl.post1()

# create your path using two nodes
mapdl.path(name='linearized',npts=2)
mapdl.ppath(point=1, node='path_node1')
mapdl.ppath(point=2, node='path_node2')
linearized = mapdl.prsect(item='S', comp='EQV')
print(linearized)

# finally run PLSECT twice (the png generated last is unusable
# for me for some reason), wait 5 seconds for the png to be
# created, then get the filename using os.listdir(path), read it
# with matplotlib.image and plot it with matplotlib.plot
mapdl.plsect(item='S',comp='EQV',vtk=False,savefig='linearized_VonMises.png')
mapdl.plsect(item='S',comp='EQV',vtk=False,savefig='linearized_VonMises.png')
time.sleep(5) # wait 5 seconds for the plot to generate png file
files = os.listdir(path)
pngfiles = []
for file in files:
    if title+'0' in file:
        pngfiles.append(file)
im = img.imread(pngfiles[-2])
plt.imshow(im)

[mikerife]

Hi @MrMechanics & @germa89
I think you have found a logic bug in the PyMAPDL CAD plotting functions. In MAPDL the /SHOW command is used to redirect plots. When MAPDL is run interactively the default is to send the plots to the terminal/screen. But batch mode does not have a default. MAPDL as a Service treats outputs like MAPDL batch does (essentially). So in order to MAPDL plot something (in batch) in PyMAPDL we first need to issue the /SHOW command to tell MAPDL where to plot to i.e. a png file, or a jpeg file etc.

/SHOW,Close purges the graphics buffer, and so does doing some other plot.

What I think is happening is that the PyMAPDL vplot is not issuing /show,close at the end, so that is why the plsect command is producing a plot as opposed to a error message about the graphics display not being set by /show. But you need to purge the graphics buffer which is why you needed the second plsect. Normally, i.e. working in MAPDL batch and writing an APDL input file, we would have followed the /plsect with a /show,close. Or relied on some other plotting command we new was going to be used later.

You should be able to replace the two plsect commands with the following:

mapdl.show('png')
mapdl.plsect(item='s',comp='EQV')
mapdl.show('close')

Mike

[MrMechanics]

Hi @mikerife
You are correct. The commands you listed worked, so I don't need to run the plsect command twice.

Thanks again

[germa89]

I need to look into this.... creating an issue to have a look later.

Thank you @MrMechanics and @mikerife