Free meshing with quadrilateral and quadrilateral-dominated
elements
Free meshing with quadrilateral elements is the
default meshing technique for two-dimensional regions. The free meshing
technique with quadrilateral elements can be applied to any planar or curved
surface. An example of a mesh generated with this technique is shown in
Figure 1.
Free meshes are usually not symmetric, even if the part or part instance itself
is symmetric.
When you free mesh a complex region with quadrilateral elements using the
medial axis algorithm,
Abaqus/CAE
creates internal partitions that divide the region into simple structured mesh
regions and then seeds the smaller regions.
If you use the medial axis algorithm to mesh a region and then remesh the
region (for example, after modifying the seeds),
Abaqus/CAE
stores the internal partitions, and the new mesh is generated more quickly. In
addition, the internal partitions allow
Abaqus/CAE
to generate a fine mesh in a similar time to that required to generate a coarse
mesh. You cannot use the medial axis algorithm to mesh regions that contain
virtual topology, and it does not work well with imprecise parts.
In general, a medial axis-based free mesh with quadrilateral elements does
not match the mesh seeds exactly for the following reasons:
Abaqus/CAE
tries to balance the seeds between adjacent regions and the smaller regions
created by the internal partitioning.
Abaqus/CAE
tries to minimize element distortions.
However, when you apply fixed seed constraints,
Abaqus/CAE
matches the number of seeds exactly and attempts to match the seed positions
exactly.
Figure 2
illustrates a two-dimensional plate with fixed seeds and movable seeds and the
resulting all quadrilateral mesh.
You should specify fixed seeds on only a few edges, or
Abaqus/CAE
may not be able to generate a mesh. For example, if you specify fixed seeds
around one of the holes in the plate shown in
Figure 2,
the global seeding becomes overconstrained, and
Abaqus/CAE
cannot generate a mesh.
Using the medial axis algorithm, a free mesh generated with
quadrilateral-dominated elements is similar to a free mesh generated with all
quadrilateral elements and without transition minimization; however,
Abaqus/CAE
inserts a few isolated triangles in an effort to match the mesh seeds more
closely.
Abaqus/CAE
can generate a mesh with quadrilateral-dominated elements faster than it can
generate a mesh with all quadrilateral elements.
Advancing
front
When you free mesh a complex region with quadrilateral elements using the
advancing front algorithm,
Abaqus/CAE
generates quadrilateral elements at the boundary of the region and continues to
generate quadrilateral elements as it moves systematically to the interior of
the region.
When you choose the advancing front algorithm,
Abaqus/CAE
matches the seeds exactly (except when you are creating a three-dimensional
revolved mesh, and the profile being revolved touches the axis of revolution).
A quadrilateral mesh that matches the seeds exactly is shown in
Figure 3.
In general, the mesh transitions generated with the advancing front
algorithm are more acceptable than the transitions generated by the medial axis
algorithm; however, matching the seeds exactly in narrow regions may compromise
the mesh quality. In contrast with the medial axis algorithm, you can use the
advancing front algorithm in conjunction with imprecise parts and on regions
that contain virtual topology.
If you select the advancing front algorithm,
Abaqus/CAE
will also use mapped meshing where appropriate. (Mapped meshing is the same as
structured meshing but applies only to four-sided regions.) For more
information, see
What is mapped meshing?,
and
When can Abaqus/CAE apply mapped meshing?.
When mapped meshing is used,
Abaqus/CAE
makes minor adjustments to the mesh seeding. If you do not want the seeding to
change, you can use mesh controls to prevent the use of mapped meshing. For
more information, see
Assigning mesh controls.
By default,
Abaqus/CAE
minimizes the mesh transition when it generates a free quadrilateral mesh using
the medial axis algorithm. Minimizing the mesh transition results in a better
mesh that is generated more quickly; however, the generated nodes deviate
further from the mesh seeds.
Figure 4
illustrates the same planar part instance meshed using the medial axis
algorithm with and without minimizing the mesh transition and meshed using the
advancing front algorithm.