Fast and iterative solution techniques for PEEC modeling in automotive EMC simulation
Dipl.-Ing. Martin LudwigApril 13, 2005, noon HF 136
New highly sophisticated automotive electronic systems combined with high operating frequencies lead to an increased risk in terms of problems in the field of electromagnetic compatibility (EMC). Mastering potential EMC problems in the early design phase of cars is a main technical issue for automotive manufacturers. Even if all sub-systems fulfill the given EMC standards a combined integration into the car result in electromagnetic interferences within the total system. Using measurement techniques only this disturbances can not be detected before the first complete prototype is available. At this late stage modifications are usually complex and expensive to execute. Numerical simulation techniques are an important key to predict the system EMC quality in the early design phase. A numerical simulation method suitable for automotive EMC simulation ist the partial element equivalent circuit method (PEEC). The PEEC method transforms a 3D complex conducting object to a equivalent linear network with basic electrical elements. The electrical behaviour of this circuit can be evaluated by conventional circuit simulators (e.g. SPICE) based on the solution of linear equations. System matrices for realistic problems in the PEEC approach are of dimension 105 - 108 which results in enormous complexity and storage demands. Moreover the density of the system matrices makes this approach improper for iterative solving methods. By using sophisticated matrix sparsification techniques in conjunction with the well known nodal-analysis (NA) formulation it is possible to obtain a sparse symmetric system of linear equations suitable for iterative solvers.