Model Order Reduction Techniques for Electro-Quasistatic Simulation Methods in Electrical Power Transmission Technology

点击翻译按钮获取中文摘要

The goal of this research project (following up a successful research project of the applicant with the same title supported under DFG-grant is the development of efficient simulation methods for dimensioning high voltage system components with special consideration of electric field grading materials. These include metal oxide varistor materials or hybrid polymeric materials with granular microvaristors based on silicon carbide and zinc-oxide as filler materials used for non-linear resistive electric field grading. Currently, these materials are studied intensively in order to verify their operational capability in power engineering devices such as high-voltage energy cables, cable accessories, insulator coatings, terminal corona protections or compact high-voltage bushings. Due to the special non-linear characteristics of these field grading materials computer simulations in time-domain are necessary in the component design of such systems. This leads to time intensive numerical calculations of electro-quasistatic field distributions which need to take into account the non-linear characteristics of these materials. The research results of the applicant on mathematical model-order-reduction techniques (MOR techniques) developed under DFG grant already allow to successfully reduce the computational time requirements of these electro-quasistatic time-domain methods and for first time even enabled uncertainty quantification analysis simulations for 2D FEM models of high-voltage devices with nonlinear field grading. In this follow up research project these mathematical techniques are to be refined and extended with novel approaches and alternative problem formulations. These novel techniques will be studied with respect to numerical efficiency and accuracy. The aim is to improve the efficiency of the nonlinear electro-quasistatic time-domain simulation methods such that parameter optimization and uncertainty quantification analysis methods can be used even for high resolution 3D FEM models within reasonable computational times.