Dr.-Ing. Wolfgang Ackermann
Office: S2|17 231
work +49 6151 16-24021
fax +49 6151 16-24027
- High-Frequency Engineering
- Analytical and Numerical Methods
- High-Performance Computing
- Beam-Dynamics Simulation
Within the working team “high-frequency engineering” of the computational electromagnetics laboratory, classical type of electrical engineering problems are processed covering a broad range of high-frequency electromagnetic field computations using state-of-the-art numerical and analytical techniques on modern supercomputer technology. The observable shift of the application field to ever-increasing frequencies requires together with the continuously growing interest in higher modeling accuracy, a permanent further development of the employed computer programs to cope with the changing demands in many fields of applications.
Main emphasis of the work
Numerical calculation of high-frequency electromagnetic fields in time and frequency domain based on volume-discretizing approaches like the finite integration technique, the finite element and the finite volume methods as well as the surface-discretizing techniques like the boundary element methods. Special focus is put on practical applications with complex geometrical structures and material distributions where precise and robust solutions are searched for.
- Efficient assembly and solution of (non-linear) eigenvalue problems for waveguides (2-D) or resonators (3-D) as well as their combinations employed for example for the modeling of resonators with explicit waveguide couplings. In addition to the characteristic eigenvalues which specify important physical quantities like the cut-off frequencies, the propagation constants or the resonance frequencies, also the corresponding eigenvectors to represent the desired eigensolutions are of fundamental importance.
- Determination of the eigenvalue distribution of regular or chaotic resonators where a large amount of eigenfrequencies (>1000) inevitably have to be considered. The desired eigenvalues are calculated either with the help of digital signal processing techniques applied to the data obtained by fast time-domain calculations (GPU based) or with the direct solution of large (generalized) eigenvalue formulations.
- Development of fast beam-dynamics simulation techniques by means of efficient solving the time-dependent Vlasov equation. In order to avoid the expensive discretization of the observed six-dimensional phase space, the direct time evolution of statistical moments of the underlying particle-density distribution is utilized without the necessity to determine explicitly the corresponding density distribution. Favorable numerical variables for the aspired beam-dynamics simulations are given by the lowest-order raw moments as well as the centralized second-order moments. Incorporating additional centralized higher-order moments enables to increase the approximation quality successively.
- Modeling of cavities used within circular particle accelerators (synchrotrons) which are partially filled with non-linear magnetic materials to enable a continuous adjustment of the resonance frequency to the circulation frequency of the moving particles. With the help of an adjustable magnetic bias current, the cavity can be selectively operated in a wide frequency range.
- Research and development of new cavities used for linear accelerators with the main emphasis on an efficient feed-in and feed-out of external and beam-driven electromagnetic fields. The new design has to take into account the high demands put on the field quality but simultaneously has to guarantee the reliable extraction of parasitic modes.
- Application and further development of parametric model-order-reduction methods to enable the fast evaluation of transfer functions which are based on the solution of Maxwell's equations formulated on the systems level. The originally large systems are carefully reduced to deliberately maintain the model-dependent parameters also in the reduced model. Apart from the frequency, an appropriate consideration of the utilized materials is of fundamental importance where geometrical parameters are also included in the evaluation process explicitly.
- Simulation of the propagation of electromagnetic waves in tunable non-isotropic materials which can be realized for example in classical waveguide configurations as partial liquid-crystals fillings. In a first step, the static or the dynamic orientation of the non-isotropic liquid-crystal molecules have to be determined by explicit consideration of the external controlling fields. Once the director configuration is known, the simulation of the high-frequency response of the configuration can be determined.
- Determination of a large amount of eigenvalues located in the low-frequency part of the real spectrum of large symmetric (generalized) eigenvalue formulations. Due to the high demands on the required computational resources in terms of computing capability and storage capacitance, efficient algorithms utilizing parallel high-performance computers have to be developed and employed to solve real-life applications.
- Creation of practical types of problems emerging from the field of micro systems engineering on the abstract system level which can be further processed in the framework of a subsequent mathematical model order reduction. On this occasion, unnecessary degrees of freedom can be advantageously eliminated already during the initial modeling on account of known physical properties to further support the mathematical motivated reduction process.
- Development of model order reduction methods used for the extraction of physical meaningful equivalent circuit diagrams from three-dimensional passive conductor arrangements. Because here many practical applications do not rely on the wave nature, in favor of the numerical robustness the modeling is performed with the help of the Darwin formulation instead of the more general Maxwell's equations.