Development of JMAG Designer
Before outlining some of the recent improvements in JMAG-Designer I would like to explain some of the motivations behind the creation of JMAG-Designer. When we started JMAG-Designer we could see several trends in the use of CAE tools which would be hard to follow just by improvements to JMAG-Studio. A significant increase in the popularity of 3D CAD programs meant we would need to have much better support for 3D models and links to CAD systems. Also, with 3D CAD now ubiquitous we would need our own 3D geometry editor for JMAG. One of the first features we added to JMAG-Designer to support 3D CAD models was the cut plane analysis feature. Full 3D simulations take time, so early in the design process it is preferable to run 2D simulations, but that used to require creating a whole new model. Using the cut plane analysis feature you can import an existing 3D geometry and automatically create a 2D model from a slice though the model.
Support for 3D geometry extends beyond the interface as it puts extra demands on the mesh generator. Creating meshes for 3D geometries poses a challenge for the mesher since the accuracy of standard tetrahedral meshes has been insufficient for applications such as cogging torque calculations. So far it has been necessary to rely on hand crafted manual meshes for such simulations. However, JMAG now includes the world leading mesher specifically created for rotating machines which enables accurate cogging torque calculations to be performed using the automatically created mesh from 3D CAD geometries (Fig 1).
Fig. 1 Automatically created extruded mesh for rotating machines
The reuse of 3D data for 2D simulations is part of a general theme for JMAG-Designer as a user we do not want you to have to enter the same information more than once. So, for example, we have a material database allowing materials data to be entered once and then re-used. Studies may be copied so that the same settings can be re-used. The analysis template function lets you create the settings for a simulation and save it to the template which can be re-used with different geometries. This concept is being taken much further with the new JMAG Virtual Test Bench which provides the templates for complex calculations involving multiple simulations.
Another trend we noticed was the increase in the number of analyses performed. Faster computers and improvements in the analysis speed mean that many more results can be obtained. When this is coupled with JMAG-Designer's parametric analysis function a very large number of analysis runs can be created. This makes handing of the result data important. JMAG-Designer introduced project files for grouping multiple analysis studies together and has some basic data management functions for organizing the result files. The new Virtual Test Bench and JMAG Super Express extend this by providing a searchable database of previous analysis results. An overview of these results can then be displayed in a dashboard (Fig 2).
Fig. 2 Dashboard display from the Virtual Test Bench
Early versions of JMAG-Designer concentrated on the CAD models and ease of use for common operations. However, with the retirement of JMAG-Studio it was important that JMAG-Designer also provide the detailed manual control over the mesh and element level results (Fig 3) that was available in Studio. Now with the release of JMAG-Designer 11.1 it is possible to have both the convenience of directly using 3D geometries while also having the fine manual mesh control when required. The geometry editor in version 11.1 supports many more functions for directly manipulating the mesh. Furthermore, the feature based meshing capabilities of the geometry editor allow the manual meshing functions to be used with the parametric analysis features. This makes it easier than ever to create accurate simulations for varying geometries.
Fig.3 Per-element results in JMAG-Designer 11.1
With JMAG we want to be able to solve large models quickly and recent additions to the solvers have added new capabilities to speed up the analysis. For example the Explicit Error Correction (EEC) method significantly reduces the length of the initial transient period for certain time domain simulations so significantly reducing the simulation time. Our magnetic solver can now use the power of the GPU to speed up the calculation and the existing parallel solver has been incorporated into the thermal and electric solver modules.