MAPDL Element Size Transition Example

This example shows you how to control the mesh density using PyMAPDL.

There are many cases in which you will want to control the density of the mesh nearby a region of local high stresses (e.g. simulated crack, filleted bracket, etc.). This usually introduces a sharp element size transition in the mesh. This example demonstrates one way to reduce this effect using a simple geometry.

First, start MAPDL as a service.

from ansys.mapdl.core import launch_mapdl

mapdl = launch_mapdl()
print(mapdl)

Product:             Ansys Mechanical Enterprise Academic Student
MAPDL Version:       24.2
ansys.mapdl Version: 0.69.dev0

The Geometry (a box)

This example uses very simple geometry. In this case, a 3D box with 4 sides. The two remaining sides are open, and we have a rectangular tube. Its dimensions are 5 x 5 x 1. We create this using 8 keypoints and then building 4 areas.

mapdl.prep7()

k0 = mapdl.k(1, 0, 0, 0)
k1 = mapdl.k(2, 0, 5, 0)
k2 = mapdl.k(3, 5, 5, 0)
k3 = mapdl.k(4, 5, 0, 0)

k4 = mapdl.k(5, 0, 0, 1)
k5 = mapdl.k(6, 0, 5, 1)
k6 = mapdl.k(7, 5, 5, 1)
k7 = mapdl.k(8, 5, 0, 1)

a0 = mapdl.a(1, 2, 3, 4)
a1 = mapdl.a(5, 6, 7, 8)
a2 = mapdl.a(3, 4, 8, 7)
a3 = mapdl.a(1, 2, 6, 5)

Mesh size

We set the global mesh size to 0.7, but we’re particularly interested in face a2, so we set the element size of that face to 0.1. Then we just need to specify an element type (we use SHELL181 here) and mesh our geometry.

Additionally, we specify that we want triangular elements using mshape. This is purely to demonstrate the effect. This will work just as well for a quadrilateral mesh.

mapdl.esize(0.7)
mapdl.aesize(a2, 0.1)
mapdl.mshape(1, "2D")

mapdl.et(1, "SHELL181")
mapdl.amesh("ALL")
mapdl.eplot(show_edges=True, show_axes=False, line_width=2, background="w")

Smoothing the transition

The mesh mostly looks fine, but we have quite a sharp transition in element size along the edges of our focus area. This is likely undesirable because these are corners and generally require a higher degree of accuracy. This is simple to remedy, however.

To do this, we need to take the keypoints that bound a2 and set their element size to be 0.1 as well. This sets all elements near those keypoints to a size of 0.1 including elements not on a2 itself. This spreads the transition away from the edges.

To demonstrate this we first clear our existing mesh using aclear. Then we use kesize to set the keypoint element sizes, and finally we remesh. The results are self-evident.

mapdl.aclear("ALL")
for k in [k2, k3, k6, k7]:
    mapdl.kesize(k, 0.1)
mapdl.amesh("ALL")
mapdl.eplot(show_edges=True, show_axes=False, line_width=2, background="w")

Smoothing the transition into a2

Similarly, if we were happy to reduce the mesh size at the edges of a2, we could set the keypoint element sizes to be an intermediate value, which would shift the main location of the size transition to within the face of a2.

mapdl.aclear("ALL")
for k in [k2, k3, k6, k7]:
    mapdl.kesize(k, 0.2)
mapdl.amesh("ALL")
mapdl.eplot(vtk=True, show_edges=True, show_axes=False, line_width=2, background="w")

Stop mapdl

mapdl.exit()