Electromagnetic Solvers in Ansys HFSS

Hi, I'm Mark with Ozen Engineering and this video will cover an overview of the advanced electromagnetic field solvers available in ANSYS HFSS. These solvers are available in the traditional interface, also referred to as the fully arbitrary 3D or the MCAD design type.

The capability of HFSS has tremendously increased over time and now includes a full set of powerful solvers that can be used to simulate a wide variety of applications. The timeline shows how the solvers in HFSS have been continually improving for over 30 years.

The first approximately 20 years were focused on optimizing the performance of the finite element method by developing advanced matrix solvers, element types, and distributed solve technologies.

Additional solvers were then added, including fast moment method implementations, time domain finite element solvers, and asymptotic methods, which use physical optics and shooting and bouncing rays.

There's also been great improvements in the use of computational resources such as GPUs and distributed computing systems. To better understand the strengths of each solver, it helps to use a diagram of electrical size versus complexity of the geometry and the materials.

The original solver in HFSS uses the finite element method and is considered the gold standard for electromagnetics because of its automatic adaptive meshing algorithm and conformal mesh elements.

This full wave method solves the fully coupled set of Maxwell's equations and is used to design many electromagnetic devices, including RF microwave components, antenna arrays, PCB packages, and transitions.

It uses a volumetric mesh that accurately conforms to any type of geometry and can model complex materials, including those having anisotropic or spatially variant dielectric and magnetic properties.

It's well suited for electrical sizes of the problems shown here, and new capabilities in meshing and distributing solutions using HPC resources are always increasing the size of problems that can be simulated with this approach.

HFSS includes frequency domain, time domain, and eigenmode solvers to further extend the applications of the finite element method. The next full wave solver available in HFSS uses an equation technique and is referred to as the IE solver.

This type of solver is considered a fast method of moments and was first introduced in HFSS in 2010. It also uses an adaptive mesh refinement process to create an optimized mesh on the surfaces of three-dimensional objects.

The IE solver is well suited for open region problems with electrically large structures that are primarily conductive or with thin dielectric layers. Since this type of solution solves for the unknowns related to surface currents, it can be used to solve those larger models.

The last solver shown here is based on the shooting and bouncing rays method and is referred to as SBR+.

It was first added into HFSS in 2016 and is based on a ray tracing technique that combines physical optics, geometrical optics, and advanced deflection physics to accurately model edge scattering and creeping waves.

It's an excellent way to efficiently simulate very electrically large structures and create dynamic images. It's also a great tool for modeling large models, such as antennas installed on vehicles and aircraft.

It excels at modeling large metallic structures, multilayer dielectric and magnetic materials, and can also handle arbitrary dielectrics with a non-uniform thickness.

The SBR plus solver also includes an advanced Doppler processing algorithm, which can simulate the response of high-frequency radar systems, such as those used in the automotive industry. In HFSS, they can be linked together as different domains within the same design.

This ability to combine the solvers in a coupled hybrid approach significantly increases the power of HFSS, since each solver can be applied to do what it does best.

For example, a detailed antenna design can be modeled using finite elements, and the larger surrounding structure can be modeled with an IE or SBR plus region. This allows us to solve complex and electrically large simulations, all within the same HFSS.

As an example of how the solvers can be used together, we can model an advanced antenna such as this millimeter wave waveguide slot array using the full wave finite element method.

The HFSS finite element solver includes advanced domain decomposition methods to efficiently represent the array using a repeating set of unique unit cells. Each unit cell can include different geometries to model the different waves of the element.

The HFSS solver can be used to model the different parts of the array in the radome housing, and each one can be independently meshed, which significantly speeds up the solution.

We can combine the finite element model with SBR plus or the IE solver to capture the effects of placing the antenna array in an enclosure on the front of a vehicle.

This allows us to predict the impact of the surrounding structure on the installed antenna performance, such as the far-field radiation patterns. The size of this model may be on the order of hundreds of wavelengths.

And for very electrically large scenarios, we can use the SBR plus solver to calculate coupling between antennas, radar cross-section returns, or the performance of an automotive radar system in a dynamic full-scale environment.

This simulation can include the surrounding buildings, trees, terrain, and other vehicles in the scene. Finally, let's look at an example of how the development of these solvers has dramatically improved our ability to model the SBR plus and the IE solver.

We've already improved our ability to model a system such as this reflector antenna with two offset feed horns. The first way we can model this is to simulate the full volume with finite elements using a perfectly matched layer as an absorbing boundary.

Because this meshes the entire volume, it requires over 140 gigabytes of memory and about two hours of runtime. This type of simulation would need to be performed on an engineering workstation, and the runtime would limit the number of design iterations we can investigate.

If we use a hybrid approach, we can also use the SBR plus solver to simulate the volume of the system. To couple the finite element region with a conformal boundary integral and replace the PML, this reduces the model volume, which reduces the memory and runtime by around 20%.

Moving on to the third approach, we can use separate conformal finite element domains to enclose each component, since the fields in the space in between are accurately represented by the IE solver.

This further reduces the memory requirement by an order of magnitude and the runtime by a factor of five compared to our original model. Continuing with the hybrid approach, we can replace the reflector antenna's finite element region with an IE region, which only needs to mesh the surface.

This solution only takes a small fraction of the original resources in runtime. And finally, since the reflector antenna is an electrically large metal structure, we can use a physical optics or SBR plus region to represent that part of the model.

That now brings us to a solution that uses less than one gigabyte of memory and runs in less than one minute. So we can see that using the hybrid approach allows this large problem to be solved on essentially any computer and enables us to study many variations to optimize the performance.

Thank you very much for watching this video. Please visit us at ozoninc.com for more information about how we can help you use these simulation capabilities for your application. Thank you.