Mesh faces selection for applying BCs and Loads

[yvanblanchard]

Hello,
I would like to know if it would be possible to select/pick some faces (pick on 3D view of geometry, or select faces by Identifiers), using pyVista or other lib (NB: faces could be retrieved from mesh using 'edges detection').
Then the vertices (or elements faces) linked to those faces would be used to apply boundary conditions and/or loads, for doing analysis in Felupe.
Thank you !

[adtzlr]

Hi @yvanblanchard, in FElupe, there exists only a method to extract cell faces which are on the surface. You have to create a boundary region to access the attribute RegionBoundary.mesh.cells_faces. However, the surface normal vectors are already calculated at the quadrature points - by identifying the neighbors and defining a threshold for the dot-products of normal vectors, we could split the surface in continuous feature faces. But there is no method available for that (yet). Would definitely be a good idea!

import felupe as fem
import numpy as np
import pyvista as pv

radius = 1.0
mesh = fem.Circle(n=4, radius=radius)
region = fem.RegionQuadBoundary(mesh)
faces = region.mesh.cells_faces

assert np.allclose(mesh.x[faces] ** 2 + mesh.y[faces] ** 2, radius)

plotter = pv.Plotter()
actor = plotter.add_points(
    np.pad(mesh.points[np.unique(faces)], ((0, 0), (0, 1))),
    point_size=20,
    color="red",
)

mesh.plot(plotter=plotter, opacity=0.7).show()

Here you may find some more examples: https://felupe.readthedocs.io/en/latest/felupe/dof.html#felupe.Boundary

I'm no PyVista expert, so you'd have to ask your question in their repo how to do that in PyVista.

See Also

• SolidBodyPressure
• PointLoad

[yvanblanchard]

Thank you !

Here is my part. Do you know if I can easily retrieve the regions (vertices or faces), using coordinates en length ?
Ex: green area (for load force) could be retrieved by filtering vertices of mesh with X coords range 0 to 10mm (size of left part) AND Z coords of greater or equal to 5mm (part thickness) AND the Y coords (depth) of the range -12.5 to +12.5 mm.

PATTE.STL.txt

Thank you again.

[adtzlr]

To mask the points you mentioned:

import felupe as fem
import numpy as np

mesh = fem.mesh.read("PATTE.STL")[0]

mask = np.logical_and.reduce(
    [mesh.x >= 0, mesh.x <= 10, mesh.z >= 5, mesh.y >= -12.5, mesh.y <= 12.5]
)
point_ids = np.arange(mesh.npoints)[mask]

However, there must be some problem with the model orientation or the units, because no points are selected. Could you please double-check your STL-file (or meshio has problems importing it)?

plotter = mesh.plot()
plotter.show_grid()
plotter.show()

P.S.: Your mesh contains triangles of a solid body in 3D space. FElupe requires solid cells to work, e.g. triangles in 2D and tetrahedrons in 3D space. You must use some external library to create a tet-mesh from the surface triangles.

[yvanblanchard]

yes, the STL mesh file has meters units (instead of mm), and the orientation is not fine for capturing vertices by spatial coords, sorry. I have to change this first...
But I have now the logical ; I just miss the part for applying BC & load on the 'point_Ids'.

I also have to fill the mesh skin with tets (after remeshing the skin with isotropic method), hopping pyVista could handle it, but I have doubt.

Thank you again Andreas!

[adtzlr]

You could have a look on scikit-gmsh but this package is in its early days. The mesh tools from nschloe (meshpro) are also great, some require a license. nschloe/pygmsh could do the job.

Back to FElupe: mesh.x is just a property which returns a read-only mesh.points[:, 0].

[adtzlr]

You can pass the point mask to a displacement boundary condition.

boundaries = dict(fixed=fem.Boundary(field, mask=mask))

A PointLoad requires the list of point ids.

load = fem.PointLoad(field_container, points=point_ids, ...)