-
StatusCompleted
-
Status date2013-06-20
-
Activity Code6A.029
A new software tool for the RF design of high-gain reflector antennas is being developed. The new tool targets rotationally symmetric reflector systems that cannot be accurately and/or efficiently modelled with existing tools. In particular, the tool will provide an extremely fast full-wave solution of rotationally symmetric reflector systems, even when 3D support structures or waveguide components are present. These new modelling capabilities will lead to a much shorter antenna design cycle without compromising the accuracy.
The objective of the project is to develop a software prototype with greatly enhanced EM modelling capabilities. The prototype will be able to solve a specific class of reflector problems with unforeseen speed and accuracy. In particular, the targeted reflector systems are composed of one or more rotationally symmetric reflectors, a feed horn with associated waveguide components, and a dielectric or metallic reflector support structure. Such a system is illustrated below.
The electrically large structure with 3D features is a challenging EM modelling problem, which is currently only solvable by software packages including general 3D algorithms, such the Method of Moments accelerated by the multilevel fast multipole method. However, even with a state-of-the-art 3D algorithm, the analysis time is typically in the order of 10-30 minutes, which leaves a numerical design optimization impossible or extremely time-consuming.
The goal of the present project is to develop a new hybrid algorithm, which allows 3D features to be accurately included, while reaching an analysis speed of 1-2 seconds. This fast algorithm will allow a full design optimisation, including the shape of the reflecting surfaces, without compromising the accuracy. This is expected lead to improved designs and a shorter development cycle.
The domain decomposition framework based on generalized admittance matrices for reflectors and support structures located in free space is a major development challenge. The boundary surfaces between the various domains are significantly larger than the surfaces typically found in closed waveguide problems where the admittance matrix approach is routinely applied. The large boundary surfaces, as well as the need for boundary surfaces of non-canonical shape, make the applied discretization scheme the most critical factor.
The new tool will be a complete design framework for the RF design of rotationally symmetric reflector systems, including modelling capabilities for 3D waveguide components and support structures. The tool will include a large toolbox for convenient problem definition, e.g, design wizards for feed horns and axially displaced reflector systems, a large number of built-in horn profiles as well as spline-based reflector and horn profiles suitable for optimization.
The toolbox will also include pattern templates for easy optimization of the radiation pattern to meet a specified pattern envelope. However, the distinct advantage of the new tool is the advanced analysis algorithms, including a rigorous combination of an extremely fast body-of-revolution solver and a general 3D solver. During the optimization of the reflector system, this algorithm will allow reflector shaping with full-wave accuracy, including 3D features, and the average analysis time per frequency point will be below one second. This modelling speed is currently not available in any commercial software package for EM modelling.
The software prototype will be built upon four existing analysis algorithms for electromagnetic analysis: Circular mode-matching available from TICRA’s CHAMP program, higher-order 3D Method of Moments (MoM) from TICRA’s GRASP program, a newly developed higher-order MoM for rotationally symmetric structures, and Physical Optics from GRASP. These four algorithms are rigorously combined by a new domain-decomposition framework, which will be developed in this project. This tool layout is illustrated in the chart below:
The 4 efficient solvers are combined by a domain decomposition framework based on generalized scattering and admittance matrices. This approach has proven effective for modelling of waveguide components but is extended in this project to the free-space region containing the reflectors.
The various regions of spaces are separated by port surfaces – denoted by the term “radiation ports”. In each region, the optimal solver can be chosen based on the geometric features inside the region and the electrical size. The generalized admittance matrix can then be obtained and stored for later use. The full solution is obtained by eliminating connected radiation ports and once the excitation is known, the radiated field can be found by solving for the field excitation coefficients at the radiation ports.
In addition to allowing the optimal solver to be chosen in each domain, the approach also allows very fast re-computation during the optimization of a complex system. A change in one geometric feature implies that the admittance matrix of that domain must be recomputed, while the admittance matrices of all other domains are available for fast reuse.
The project was divided in four phases: Requirements definition, electromagnetic modelling, software implementation, and validation. The duration of the project is approximately 14 months. Significant time is set aside for validation on real-life problems and subsequent design iterations.
Completed.