The High Frequency Structure Simulator (HFSS™) is a software tool for 3D full-wave electromagnetic field simulation. HFSS provides E- and H-fields, currents, S-parameters, and near and far radiated field results. Intrinsic to this engineering design tool is an automated solution process where users are only required to specify geometry, material properties and the desired output. HFSS automatically generates an appropriate, efficient and accurate mesh for solving the problem using the proven finite element method (FEM).
HFSS includes many Ansoft-pioneered research and development innovations. These breakthroughs have made HFSS a widely used software product for solving 3D full-wave electromagnetic field simulations. The invention of tangential vector basis functions enabled the highly accurate finite element method for electromagnetic field solution. The transfinite element method, another Ansoft invention, allows the 3D finite element solution to couple to port solutions for fast and accurate multi-mode S-parameter extractions. Finally, the development of automatic mesh generation and adaptive refinement was a key innovation for reliable, repeatable and efficient results.
New in HFSS 12.0
HFSS 12.0 is a major step forward for three-dimensional full-wave electromagnetic field simulation with new innovations for engineering simulation and design. The software includes key updates in mesh generation, solver technologies, and enhancements to the user interface and the modeler. The most significant solver technology enhancement is domain decomposition, a technique that allows the tool to exploit high-performance computing (HPC) capabilities to solve electromagnetic field problems of unprecedented size and scope. With domain decomposition, a single HFSS job can be divided into smaller pieces and then distributed across a network of computers. The use of all of the memory across a network allows for truly giant simulations with the rigor, accuracy and reliability of HFSS.
A faster and more robust meshing algorithm generates higher-quality, more efficient tetrahedral meshes. This meshing technology is particularly effective on complex geometries, especially those imported from external 3D CAD tools. Other important enhancements include mixed element orders, curvilinear elements and adjoint derivative computation. Ease of use and automation in the user interface has been improved and include additional modeler capabilities such as sheet wrapping and imprinting. Advanced integration with load-sharing utilities and an Ansoft-developed Remote Solve Manager (RSM) provide integration within popular computing environments. These advances in HFSS 12.0 enable electrical engineers to expand their solution capability, exploit HPC hardware and fully integrate electromagnetics analysis into their design processes.
HFSS HPC Option
The HPC Option in HFSS 12.0 enables new solver technology using Ansoft’s breakthrough implementation of the domain decomposition method. With the HPC option, large-scale simulations can be solved across a network of machines using all of their available memory.
HFSS 12.0 has seamless, out-of-the-box integration with industry standard schedulers (computer queuing systems) to allow engineering organizations to combine the power of HFSS with Load Sharing Facility (LSF), from Platform, Windows HPC Scheduler, from Microsoft, Portable Batch System (PBS) from Altair, and Sun Grid Engine (SGE).
These enhancements allow engineers to simulate and design at a scale and speed never before possible. Users of this latest version of HFSS software can achieve a dramatic reduction in development time and costs while at the same time realizing increased reliability and design optimization.
High-Performance Computing (HPC) Technologies
The HFSS High-Performance Computing (HPC) options enable intra- and inter-machine parallel solving and processing that distribute and speed the solution.
Domain Decomposition Method (DDM)
Figure 1 Domain decomposition divides mesh across network to solve large problems.
DDM enables the simulation of very large models by accessing the memory of a network of machines. Figure 1 shows how DDM splits the finite-element mesh of the geometry automatically into a number of smaller mesh sub-domains. HFSS determines the optimum number of domains, depending on the mesh size and the number of computers and processors available. The domains are analyzed separately on a single machine or on a network of machines, after which an iterative procedure on the domain interfaces reconstructs the full solution. This network memory access allows the simulation of very large models for which one machine might not have enough memory. It also reduces simulation time and overall memory load, offering in some cases better than linear speed improvements with each additional processor.
Figure 2 Seven-element helix array (a) showing uniform excitation at S-band with mutual coupling and ground plane edge effects (b).
Figure 3 Ansoft HPC with domain decomposition solved the helix array on a spacecraft with other antennas nearby (1.3 M tetrahedra, 25 M unknowns, 35 computer cores).
Figure 2 shows a seven-element phased array of helix elements simulated in HFSS 12.0 across the 3 to 4 GHz S-band. The array is excited uniformly to provide broadside radiation. Mutual coupling among the elements and edge effects of the finite-sized ground plane are included because the array was simulated with all the elements and ground plane present. Such a model is useful for understanding these effects and for preliminary design. Of course antennas are ultimately installed on some larger system. Figure 3 shows the helix array placed on a satellite. Real estate is precious on a satellite platform and other services must share space on the spacecraft. Another high-gain antenna is also mounted on the spacecraft with the potential of disrupting radiation from the helix array. HFSS 12.0 with domain decomposition was used to simulate both antennas and the spacecraft in a single simulation. As can be seen in the figure, field interaction is significant when the array beam is steered 10° off bore-sight. The radiation of the array interacts with the back side of the high-gain reflector antenna. The size and scope of this simulation is unique in full-wave simulators.
Multiprocessing (MP)
The multiprocessing option is used for solving models on a single machine with multiple processors or multiple cores that share RAM. Operations during the solution process are parallelized across the available cores on a machine thus significantly reducing the simulation time.
Distributed Solve Option (DSO)
The DSO option allows users to distribute parametric sweeps to explore variations in geometry, materials, boundaries and excitations. Additionally, users can distribute frequency sweeps to generate responses over a broad frequency band of interest. This time-saving capability splits multiple pre-defined parametric design variations and/or frequency points, solves each simulation instance on a separate machine and then reassembles the data. This dramatically accelerates parametric studies and design optimization.
Volumetric Meshing
A new highly robust volumetric meshing technique results in even more efficient and higher-quality meshes that reduce memory and simulation time. The new technique meshes the 3D volume with uniformly distributed tetrahedra, followed by boundary refinement and surface meshing. By following this procedure, HFSS can reliably mesh geometries imported from other CAD tools while creating an even higher quality mesh for finite element simulation.
New Element Technologies
Curvilinear elements and mixed element orders allow for higher accuracy and more efficient distribution of computational resources. Curvilinear elements model the fields exactly on curved surfaces and in these cases provide higher accuracy even with a coarser mesh discretization. Mixed element orders allow for an automated and judicious localized application of element order based on geometry and electromagnetic requirements. Smaller features are solved more efficiently by lower-order elements while large homogenous regions benefit from higher-order elements, all element orders being automatically and appropriately “mixed” in one mesh. Mesh refinement is now performed on both the size and order of the elements.
Adjoint Derivatives
Adjoint derivative computation provides a highly efficient and accurate procedure to evaluate the derivatives of S-parameters with respect to geometric and material model parameters. This technique provides sensitivity information for use in device tuning, tolerance evaluation and optimization. These derivatives are employed to speed up the sequential nonlinear programming (SNLP) optimizer included with the Optimetrics™ add-on program.
Conclusion
HFSS has been used the world over to design electronic and microwave products from smart phones to high-speed computer backplanes to high-performance antenna systems. The HPC options in HFSS allow engineers to solve giant electromagnetic simulations on a network of computers using all of the available memory. A new meshing engine provides even more reliable and high-quality FEM meshes that simulate even faster. New element technologies enable true curved surface simulation representing the fields along those surfaces perfectly without approximation. Mixed element orders provide numerical efficiency and accuracy from very concentrated regions of the problem to very open regions. Adjoint derivatives provide engineers data on the local sensitivity of their designs with respect to project variables like material properties, geometry and boundary conditions.
Ansoft
Pittsburgh, PA
(412) 261-3200
www.ansoft.com/hfss
RS No. 304