Keyboard shortcuts

Press ← or → to navigate between chapters

Press S or / to search in the book

Press ? to show this help

Press Esc to hide this help

Mesh slicing

Figures use PyVista; verbose plotting boilerplate is omitted.

UMesh.slice(origin, normal) cuts a mesh with a hyperplane and returns a mesh of one dimension lower: a 3D volume becomes a 2D polygonal (PGON) surface embedded in 3D, and a 2D region becomes a 1D SEG2 line. In both cases the hyperplane passes through origin and is normal to normal.

import numpy as np

import mefikit as mf

Slicing a 3D volume

Let us build a box and cut it with a general oblique plane, neither through a mesh node nor aligned with an axis. The slice is a 2D mesh made of polygons, but its coordinates remain in 3D so it can be plotted with the source volume.

axis = np.linspace(0.0, 1.0, 11)
volume = mf.build_cmesh(axis, axis, axis)

origin = np.array([0.37, 0.42, 0.51])
normal = np.array([0.7, 1.1, 1.3])
surface = volume.slice(origin, normal)

surface.block_types(), surface.coords().shape
(['PGON'], (240, 3))
_, offsets = surface.blocks()["PGON"]
sizes, counts = np.unique(np.diff(offsets), return_counts=True)
dict(zip(sizes.tolist(), counts.tolist()))
{3: 92, 4: 53, 5: 29, 6: 33}

The output only contains the intersection of the volume with the plane. The cells are PGONs, even when the section is a quadrilateral. Here the general cut produces triangles, quadrilaterals, pentagons and hexagons.

plotter = pv.Plotter()
plotter.add_mesh(volume.to_pyvista(), style="wireframe", opacity=0.15)
plotter.add_mesh(surface.to_pyvista(), show_edges=True, color="tab:orange")
plotter.show()

Slicing a 3D volume along a line

slice_line(origin, direction) cuts the volume with the line through origin along direction. Here we use a general oblique line that does not pass through any mesh node. The result is a 1D SEG2 mesh whose coordinates remain in 3D, so it can be drawn on top of the volume.

line3d = volume.slice_line([0.23, 0.37, 0.11], [0.2, -0.3, 1.0])
line3d.block_types(), line3d.coords().shape
(['SEG2'], (16, 3))
plotter = pv.Plotter()
plotter.add_mesh(volume.to_pyvista(), style="wireframe", opacity=0.1)
plotter.add_mesh(
    line3d.to_pyvista(dim="all"),
    color="tab:green",
    line_width=6,
    render_lines_as_tubes=True,
)
plotter.show()

Transferring a field onto the slice

Slicing does not copy fields, because the target cells are not the source cells. The mf.transfer operators remap them. ConstantPiecewise samples the volume cell containing each slice-cell centroid, which is a natural choice for a piecewise-constant field; MovingLeastSquares or InverseDistance can be used when a smoother result is wanted.

volume.fields["T"] = mf.X + 2.0 * mf.Y + 3.0 * mf.Z

surface_transfer = mf.transfer.ConstantPiecewise(volume, surface)
surface.fields["T"] = surface_transfer("T")

line_transfer = mf.transfer.ConstantPiecewise(volume, line3d)
line3d.fields["T"] = line_transfer("T")

surface.fields["T"].values()["PGON"][:5]
array([2.1, 2.2, 2.1, 2.2, 2.3])

The slice now carries T, and the values follow the analytic field T(x, y, z) = x + 2y + 3z sampled in the volume cells:

plotter = pv.Plotter()
plotter.add_mesh(volume.to_pyvista(), style="wireframe", opacity=0.1)
plotter.add_mesh(surface.to_pyvista(), scalars="T", show_edges=True)
plotter.add_mesh(
    line3d.to_pyvista(dim="all"),
    scalars="T",
    line_width=6,
    render_lines_as_tubes=True,
)
plotter.show()

Slicing a 2D mesh

In 2D the cut is a line. Here we use a general oblique line that does not pass through mesh nodes; the result is a SEG2 mesh. Its field is transferred in the same way.

domain = mf.build_cmesh(axis, axis)
line = domain.slice_line([0.13, 0.27], [1.0, 1.7])

domain.fields["T"] = mf.X**2 + mf.Y
line.fields["T"] = mf.transfer.ConstantPiecewise(domain, line)("T")

line.block_types(), line.coords().shape
(['SEG2'], (16, 2))
plotter = pv.Plotter()
plotter.add_mesh(domain.to_pyvista(), style="wireframe", opacity=0.35)
plotter.add_mesh(
    line.to_pyvista(dim="all"), scalars="T", line_width=6, render_lines_as_tubes=True
)
plotter.show()