Techniques for Creating Geometries
Several techniques can ensure that a geometry results in a good mesh and gives reasonable solution times for the analysis. They include the use of symmetry and eliminating small details, gaps, holes, and singularities.
Using Symmetries
Using symmetry is one of the most effective ways to reduce the size of a model. For axially symmetric geometries, a 2D axisymmetric model is sufficient. You can easily visualize the results in a full 3D geometry using a Revolution 2D dataset. Other common cases of symmetry are sector symmetry and symmetry and antisymmetry planes, which can reduce the size of a 3D model.
Making the Geometry Match the Boundary Conditions
Sometimes the modeling domain is unbounded or too large for successful analysis. For those cases a suitable boundary condition can replace the exterior of the domain.
For outflows in fluid-flow models, for example, the boundary should be perpendicular to the fully developed flow. Inspections and modifications of the solved model might be necessary to verify the validity of the boundary condition. Also, for some applications, infinite elements or perfectly matched layers (PMLs) are available for modeling diffusion or wave propagation in unbounded domains.
Avoiding Excessively Small Details, Holes, and Gaps
Many geometries, especially those designed using a CAD system, contain small holes, details, and gaps. These small features can make the domain unbounded and must be removed before analysis. Small details and holes can lead to large meshes or even failure during mesh generation. Make sure the snapping feature is activated to avoid small gaps and mismatches between the geometry objects.
The CAD Import Module contains tools for automatic and interactive repair and defeaturing of 3D CAD data. For a 2D or 3D model you can also remove small details and prepare the geometry for efficient meshing using virtual geometry operations (see Virtual Geometry Operations).
Avoiding Singularities and Degeneracies in the Geometry
A singularity in a geometry is a sharp corner or angle that can create problems during meshing and analysis. In reality, a sharp reentrant corner leads to infinite stress values in a stress analysis of a perfectly elastic material. The stress value for a sharp corner is finite in the stress analysis, but refinement of the mesh increases the stresses in the corner without limit. To avoid a singularity, round sharp corners using fillets.
A degeneracy in the geometry can occur during solid modeling. For example, fillet areas that taper to a point and the apex of a cone can become degenerate points. These degeneracies might cause problems for the mesh generator and during the analysis. A common degeneracy in the geometry occurs when a 3D solid is created (for example, a cylinder) by rotation about an axis that touches the rotation area. It is then better to create the solid object by extruding a cross section or to use geometric 3D primitives.
About Finite and Infinite Voids
Primarily when working with the boundary element method, the geometry can contain one or more finite voids — closed empty regions (holes) in the geometry — and a single infinite void, the empty, unbounded region outside of the closed geometry. The finite voids have negative domain numbers (1, 2, and so on); the infinite void is always domain 0. These voids appear in the list of domains in the Selection List window, for example, but they are not highlighted when selected. You can use this method to highlight the boundaries of one or more voids:
1
Add an Explicit selection node under Definitions in the component with the geometry that includes the voids. Add the finite voids that you want to highlight to that selection by clicking the Paste Selection button (), where you enter the domain numbers for those voids (-2 -3, for example, to add finite voids 2 and 3). Enter 0 for the infinite void.
2
Add an Adjacent selection node under Definition, and then add the Explicit selection that you added in the previous step to the list under Input selections. With the default setting under Output Entities, to output the Adjacent boundaries, those boundaries are then highlighted in the Graphics window to show the finite voids inside or infinite void outside those boundaries.
As an alternative, you can turn finite void regions into solid domains and then make sure to skip meshing those domains.