Concept of Multi-Frequency Adaptive Meshing

The multi-frequency adaptive meshing method, implemented in Ansys HFSS, provides you with an efficient and convenient method to automatically obtain a mesh appropriate for broadband frequency simulation where the adaptive algorithm uses errors from multiple time-harmonic frequency simulations when refining the mesh instead of using only one frequency. Note that for each adaptive pass, the same mesh is used for all frequencies in order to generate errors for mesh refinement for the next adaptive pass. This process is described by the flow chart in the following figure. At a high level, the differences compared with single frequency adaption include generation of errors from multiple frequencies at each adaptive pass and combining of the errors before refining the mesh.

There already exist adaptive techniques to tailor a mesh for broadband frequency analysis. Those techniques amount to adapting a mesh at N frequencies sequentially via mesh links in HFSS. If the ordered frequencies are f1, f2,...,fN the adaptive refinement process starts with f1as the first frequency using a coarse initial mesh. The process then follows the steps shown in the Single Frequency Adaptive Mesh Refinement Flow Chart until convergence. Next frequency f2 is solved using the converged mesh from f1, as the initial mesh and once again the steps shown in the Single Frequency Adaptive Mesh Refinement Flow Chart are applied by the adaptive process. This process continues until the convergence of the final frequency fN. The ability for all adapt frequencies to influence mesh adaption at each pass is a major benefit of using the new multi-frequency adaptive meshing algorithm in HFSS compared to the sequential multi-frequency adaptive meshing approach. It typically produces a smaller final mesh and improves the adaption performance by solving the frequencies in parallel.

The multiple-frequency adapt algorithm has three potential performance benefits compared with the standard sequential algorithm. The benefits are as follows:

• Solves adapt frequencies in parallel.
• Reuses the solutions of all the adapt frequencies for subsequent frequency sweeps. In contrast the sequential method can reliably reuse only the solution of the last frequency since it solves other frequencies on different meshes.
• Obtains a smaller mesh that yields similar accuracy.