Introduction
This Getting Started Guide describes how to use Skin Depth Seeding in Ansys Electronics Desktop.
This guide answers the following questions:
- What is Skin Depth seeding?
- Why use seeding?
It also covers the following topics:
- Skin Depth mesh plot visualization
- Improvement compared to older implementation
- Using simple models
- Choosing settings
This guide does not discuss when a user should use Skin Depth Seeding.
What Is Skin Depth Seeding?
Skin Depth seeding is a mesh operation feature of solvers that use the 3D Modeler, including HFSS and Maxwell 3D.
The Length Based mesh operations refine the specified objects or surfaces. Skin Depth Based mesh refinement lets you calculate or specify a skin depth for mesh refinement, as well as the number of layers of elements to generate within the specified skin depth.
The skin depth is the total depth of all layers combined. The Skin Depth setting provides an easy, alternative approach to creating physical models of each layer using pseudo-sheet bodies. Creating and adjusting a complex, layered physical model is difficult. Correcting errors is difficult. But changing the skin depth, surface triangle length, and the number of layers is simple and has a quick turnaround time. With Skin Depth refinement, the model itself is much easier to adjust.
For example, consider a rotor, drawn as a stack of seven objects:
Skin Depth Layering on a single object can provide comparable or superior accuracy in the solution, and is much easier to create and adjust:
To test various stackings with the Skin Depth Layering Method for Mesh Seeding:
- Revert to initial mesh
- Change settings
- Apply mesh ops
To test various stackings using a pseudo body method:
- Recreate the model with different body thicknesses
- Remake the entire initial mesh
- Tau mesher might not be available for all heights
Skin Depth Layering is much easier and faster. In both cases, remember that a layered mesh or pseudo body modeling is a means to a good solution, not a goal in itself. Using layered elements for relevant designs is one way to get to a good solution.
For example, where a pseudo model design had 85K tets, the following table shows some different results where the Surface Triangle Length target is given different values.
The layer reporting counts diagonal shifting as failure, even if there are layers.
Why Use Seeding?
In general, Ansys recommends the use of adaptive meshing. However, in some cases an adaptive solution is not possible. Skin Depth Seeding helps in some of these cases. Additionally, some users demand Skin Depth Seeding for situations in which they create layered structures to force the mesh. This affects the robustness of meshing.
Use of Skin Depth Seeding in appropriate projects is a means to arriving at a converged solution sooner.
In the following figure:
- Green – no seeding, no sheets; raw model pure adaptive solution
- Red – sheet body version
- Blue – new Skin Depth Seeding, but slightly older result; latest result is much better than the blue as shown before
Skin Depth Seeding Implementation Change
Prior to 2018, Skin Depth Seeding utilized layered points. The current implementation uses layered elements.
On Selection > Length Based mesh seeding occurs on the surface. Inside Selection > Length Based occurs within the interior volume of the body. These mesh operations do not involve Skin Depth seeding. Compare the results of On Selection > and Inside Selection > Length Based mesh operations:
While the surface mesh is similar, the use of a clip plane shows that the interior mesh for Inside Selection is different.
Layered Elements Implementation
Electronics Desktop requires less layering than other EM tools, and other physics. In contrast with other tools, the Electronics Desktop:
- Does not just look at layered elements
- Pays attention to element count, number of points/elements in the interior
- Offers incredibly small element counts
Layered Elements can be applied to select faces of solid bodies. Elements are stretched parallel to the faces and compressed in the normal direction.
General Guidance for Skin Depth Based Refinement
To best understand how Skin Depth refinement works, experiment with simple models to understand refinement behavior. You can test bodies with skin depth seeding in isolation using scratch projects.
Particularly for models with true curved surfaces (for example, true cylinders), the Curved Surface Meshing setting has a large impact on the layering success rate. To change this or other settings, right-click Mesh in the Project Manager and select Initial Mesh Settings.
Selecting a very coarse initial mesh results in a low (60%) success rate. Finer meshes easily hit 80%, surging to 95%.
For example, the following figure shows a faceted cylinder with triangles and segments on a face. The open triangles show imperfect layering, but a very good success rate.
A true cylinder with similar mesh settings (including a medium resolution for curved surface meshing) shows a lower success rate.
Moving the slider two ticks to the right improves the success rate for the true cylinder much more effectively than a change to Skin Depth Refinement would. In this case, the overall success goes from 55% to 81%.
Note that the different layers, 0 through 4 in this example, will have different success rates.
In addition to the options in the Initial Mesh Settings dialog box, you can use Skin Depth Based Refinement settings to reach a suitable success rate supporting a good solution. To access these options, right-click Mesh in the Project Manager and select Assign Mesh Operation > On Selection > Skin Depth Based.
Reduce Surface Triangle Length target to improve layering success rate, where success is consistent layering.
Consider the Surface Triangle Length to Mean Skin Depth ratio. For guidance:
- Avoid stretched triangles on the surface
- Get a few points on the interior of faces
- Test a wide range of ratios from 2.5 to 800
- Set a ratio of 100 or above to contribute to a good success rate
- Above 800, mesh tolerance Quality issues come in to play
Remember that layered meshing is the means to a solution, not the end goal. Layering elements is one way to get a good solution. Always check the correctness and accuracy of solutions.
Guidance for HFSS for Skin Depth Based Refinement
- For conducting material, you must enable Solve Inside or the feature has no effect.
- Some spiral inductors might obtain a better Q value when Solve Inside is enabled.
- For projects where you solve inside conductors, there may be a subset of models that benefit.
User Inputs for Skin Depth Based Refinement
For Skin Depth Based refinement to have an effect, the Solve Inside property for the selected 3D object's material must be checked.
Select the faces of the solid body to which you want to apply Skin Depth Based refinement, and right-click Mesh in the Project Manager. Select Assign Mesh Operation > On Selection > Skin Depth Based... to display the Skin Depth Based Refinement dialog box.
The Skin Depth is the total depth of all layers combined. Very thin layers may cause a reduction in mesh quality or unnecessarily cause the generation of a very large mesh. Further regions refined under this operation and its close neighbors do not participate in solution adaptive refinement. This is another reason to use this seeding operation with caution.
To calculate the skin depth based on the object's material permeability and conductivity and the frequency at which the mesh will be refined, click Calculate Skin Depth...
The Calculate Skin depth dialog box appears.
Initial values for Relative Permeability and Conductivity are taken from the Materials library. For HFSS, the Frequency value is taken from the Solution Setup.
All three values must be non-zero to obtain a calculated value using this feature.
When you click OK, the Calculate Skin Depth dialog box disappears and the calculated value appears in the Skin Depth field of the Skin Depth Based Refinement dialog box.
You can edit the calculated value or provide your own without using calculation. Check whether the calculated value is appropriate to fit within the dimensions of the object.
Next, specify a Number of Layers of Elements.
For the number of layers, choose 2, 3, 4 or (at most) 5 layers. First layer thickness = 1/(2n-1) = 0.333, 0.143, 0.0667, 0.0322. The elements are stretched parallel to face, and are compressed in the normal direction.
Optionally, provide the maximum edge length of the surface mesh in the Surface Triangle Length text box. The default value is set to 20% of the maximum edge lengths of the bounding boxes of each selected face.
The solver will refine the surface triangle mesh (the faces of the tetrahedra touching the surface) until their edge lengths are less than or equal to the specified value.
By default, the Restrict Number of Surface Elements setting is cleared. This allows the mesher to use symmetry more effectively. If you restrict the number of surface elements, symmetry may be affected. However, in certain cases, you can restrict the number of surface elements to prevent runaway refinement.
To restrict the number of elements added during refinement of the faces, first observe the estimated number of surface elements in the Maximum Number of Surface Elements text box. Then, select the Restrict Number of Surface Elements option to enable the Maximum Number of Surface Elements option. Adjust the specified number of surface elements as desired.
You cannot edit the Maximum Number of Surface Elements, and the specified number has no effect on the resulting mesh, until you select the Restrict Number of Surface Elements option immediately above it.
When the mesh is generated, the refinement criteria you specify are used. This operation is approximately the same as having slabs of tetrahedra, but it is not guaranteed to prevent tetrahedra from crossing slab interfaces. Use caution with this mesh operation, as very thin layers may cause a reduction in mesh quality or unnecessarily cause the generation of a very large mesh. Further regions refined under this operation and its close neighbors do not participate in solution adaptive refinement. This is another reason to use this seeding operation with caution.
Generate Initial Mesh
After defining mesh operations for a model (including Initial Mesh Settings, Skin Depth Based, or other operations), you can generate an initial mesh without solving the analysis. To do so, right-click Analysis in the Project Manager and choose Generate Mesh from the shortcut menu.
After generating a mesh, you can overlay mesh plots and use the Cut Plane features to examine the appearance of the mesh. Also, right-click Analysis in the Project Manager and choose View Mesh Feedback to access the Mesh Feedback tab of the Model Analysis. Here, you can evaluate the success of the mesh and view any problems. The Mesh Error Description box on this tab provides a statistical evaluation of the overall body's meshing success (expected versus found triangles) as well as success percentages by count and by area.