Thin plate shield

Applications that use a region to act as a shield can take advantage of MagNet's Thin Plate boundary condition. Shield regions are most often very thin when compared to the extents of the geometry being modelled and require a very fine mesh for accurate results. These fine meshes in 3D will increase the time the engineer has to wait for results. Since the thin plate is a boundary condition it does not require a volume all to itself. All that this boundary condition needs is to be applied to the interface of the two regions.

The problem presented here is a spherical shield inside of a cylindrical coil. The shield's thickness can vary from 100 to 1000 times smaller than the extents of the model. Due to the symmetry of the problem only a 10 degree sweep of half the problem is studied. The model can be seen on the left, a section was removed so the spherical shield can be clearly seen.

Results

Click image to see full size version

The shaded field plot of the magnitude of the flux density |B| on the right shows how there is no penetration in the spherical region because of the Thin Plate boundary applied.

This particular image has a thin plate thickness of 2mm. The sphere's diameter is 27.2 cm and the model's extents are 80cm. The field displayed is on the xy plane, and the y-axis is aligned with the axis of the cylinder.

Click image to see full size version

This second shaded field plot displays the magnitude of the flux density |B| field on the boundary itself. This is a rotated view which shows the top of the 10 degree slice.

Click image to see full size version

Using Infolytica's powerful parameterization engine, the engineer can enter different thicknesses for the shield. Once solved the user can then decide on which configuration best suits the application.

The graph on the right shows the field inside and outside of the shield for a number of thicknesses, the smaller the thickness the more the field penetrates.

Click image to see full size version

Before thin plates were implemented in MagNet 6 the user was required to create an actual volume to model the shield. This method worked very well, but does it work better than the thin plate approximation? The answer is no as the graph on the right demonstrates. The pairs of lines represent the field inside and outside of the shields for varying thicknesses. When the shield is an actual volume, after it becomes too thin an excessive amount of mesh elements are required. With such an amount of elements, solving time will increase and it no longer seems practical to create a component to act as a shield, when there is the stress of deadlines and the option of thin plates.

This application shows that thin plates not only give accurate results, but also they are more powerful than the previous method used. As well, thin plates solve much faster. In the benchmarks run at Infolytica, when 10 parameterized problems were solved, thin plates finished more than 4 times quicker. This is due to Infolytica's optimized meshing engine which recognizes that the thin plates do not modify the mesh. The mesh does not depend on the thin plate thickness, so in subsequent solves no re-meshing is required.