0

I have 3 1D arrays (node x-coordinates, node y-coordinates and Von-Mises stress scalar) exported from an FEA solver.

I want to create 2D contour plots as shown below in Python:

Stress plot example

I have managed to create such plot as shown below: Stress plot result

import matplotlib as mpl
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.tri as tri

for orient in ['top', 'bot', 'side']: 
    x = []
    y = []
    z = []
    stress = []
    with open('data.txt') as file:
        for line in file:
            cur_line = line.split('\t')
            cur_x_old = cur_line[0]
            cur_y_old = cur_line[1]
            cur_z_old = cur_line[2]
            cur_s_old = cur_line[3]
            
            if cur_x_old == 'X Location (mm)':
                pass
            else:
                cur_x = cur_x_old.replace(",",".")
                cur_y = cur_y_old.replace(",",".")
                cur_z = cur_z_old.replace(",",".")
                cur_s = cur_s_old.replace(",",".")
                
                x.append(float(cur_x))
                y.append(float(cur_y))
                z.append(float(cur_z))
                stress.append(float(cur_s))
                
    stress = np.array(stress)
    x = np.array(x)
    y = np.array(y)
    z = np.array(z)
    
    levels=np.linspace(stress.min(), stress.max(), num=100)
    triang = tri.Triangulation(x, y)
    
    if orient == 'side':
        plt.figure(figsize = (max(x)/50, abs(min(y))/50))
        plt.tricontourf(triang, stress, cmap = 'jet', norm = mpl.colors.Normalize(0, 100), levels = levels, extend = 'max')
        plt.scatter(x, y, color = 'k')
    else:
        plt.figure(figsize = (max(x)/50, max(z)*2/50))
        plt.tricontourf(x, z, stress, cmap = 'jet', norm = mpl.colors.Normalize(0, 100), levels = levels)

My problem is that by triangulating the data, unwanted triangles are generated at the edge of the mesh (see Stress plot result). The black dots are the scatter plot from x and y coordinates. I want the colour plot to be only inside the boundaries of the grid. Is there a way to remove these unwanted triangles?

Ian Venter
  • 1
  • Hi. Why are you using triangulation? You can simply use `plt.scatter(x,y,c)`. If you really wanted to plot good stuff, tell us what FEA solver are you using? You can probably export `vtk` and plot using `pyvista`. These packages are tailored for such plot and they have evolved over years. – Prakhar Sharma Aug 22 '22 at 18:26
  • Hi. I want a smooth contour plot rather than small dots with colours everywhere. I am using ANSYS and I am exporting nodal x,y,z and stress values into a text file. – Ian Venter Aug 23 '22 at 08:24
  • Ok. It is easy to export the VTK file if you know APDL. Otherwise Ansys is proprietary software so there is no direct way to do this. – Prakhar Sharma Aug 23 '22 at 08:30
  • The easiest way is I think using [pyMAPDL](https://mapdl.docs.pyansys.com/api/plotting.html). Or may also try [this](https://stackoverflow.com/questions/41722822/how-to-read-ansys-data-files-in-paraview) – Prakhar Sharma Aug 23 '22 at 08:32
  • or the native way which I used months ago was to export the output file, which contains the nodal coordinates and the mesh with solutions. You just need to write a complex python script to generate a vtk file. Otherwise it is better to switch to open source FEM solvers such as FeniCS. – Prakhar Sharma Aug 23 '22 at 08:38

0 Answers0