Prof. Jin-Fa Lee
ElectroScience Laboratory, ECE Department
The Ohio State University
1320 Kinnear Rd., Columbus, OH 43212
Jin-Fa Lee received the B.S. degree from National Taiwan University, in 1982 and the M.S. and Ph.D. degrees from Carnegie-Mellon University in 1986 and 1989, respectively, all in electrical engineering. From 1988 to 1990, he was with ANSOFT (later acquired by ANSYS) Corp., where he developed several CAD/CAE finite element programs for modeling three-dimensional microwave and millimeter-wave circuits. From 1990 to 1991, he was a post-doctoral fellow at the University of Illinois at Urbana-Champaign. From 1991 to 2000, he was with Department of Electrical and Computer Engineering, Worcester Polytechnic Institute. He joined the Ohio State University at 2001 where he is currently a Professor in the Dept. of Electrical and Computer Engineering. Prof. Lee is an IEEE fellow and is currently serving as an associate editor for IEEE Trans. Antenna Propagation. Also, he is a member of the Board of Directors for Applied Computational Electromagnetic Society (ACES).
Prof. Lee’s main research interests include electromagnetic field theories, antennas, numerical methods and their applications to computational electromagnetics, analyses of numerical methods, fast finite element methods, fast integral equation methods, hybrid methods, domain decomposition methods, and multi-physics simulations and modeling.
CEM Algorithms for EMC/EMI Modeling: Electrically Large (Antenna Placements on Platforms) and Small (SI in ICs and Packaging) Problems
Modern antenna engineering often involves the use of metamaterials, complex feed structures, and conformally mounting on large composite platforms. However, such antenna systems do impose significant challenges for numerical simulations. Not only do they usually in need of large-scale electromagnetic field computations, but also they tend to have many very small features in the presence of electrically large structures. Such multi-scale electromagnetic problems tax heavily on numerical methods (finite elements, finite difference, integral equation methods etc.) in terms of desired accuracy and stability of mathematical formulations.
Another important electromagnetic application is the study of signal integrity in ICs. Recent advances in VLSI interconnect and packaging technologies, such as the increasing number of metal layers and the 3D integration, have paved the way for higher functionality and superior performances. During the reduction of the size, power, and cost in today’s advanced IC interconnects and packaging, the signal integrity (SI) has become more crucial for system designers. The previous common practice adopted by industries, such as using only static parasitic RC or RLC equivalent networks for physical designs, are gradually abandoned. It has come to use full-wave computational electromagnetic (CEM) methods for the ultimate accuracy check.
In this lecture, I will present our on-going efforts in combating the multi-scale electromagnetic problems, both electrically large (antenna placements on platforms) and electrically small but complex (SI in ICs) through the use of non-conventional PDE methods that are non-conformal. The non-conformal numerical methods relax the constraint of needing conformal meshes throughout the entire problem domain. Consequently, the entire system can be broken into many sub-problems, each has its own characteristics length and will be meshed independently from others. Particularly, our discussions will include the following topics:
- Integral Equation Domain Decomposition Method (IE-DDM): A very significant breakthrough that has been accomplished is the IE-DDM formulation. For example, we show an electromagnetic plane wave scattering from a mock-up fighter –jet with thin coatings at the X- and Ku-bands by dividing the platform into many closed objects, and noting that they will be touching each other through common interfaces.
- Non-Conformal DDM with Higher Order Transmission Conditions and Corner Edge Penalty: By introducing two second-order transverse derivatives, one for TE and one for TM, the derived 2nd order TC provides convergence for both the propagating and evanescent modes. Moreover, on the corner edges sharing by more than two domains, an additional corner edge penalty term needs to be added in the variational formulation. Consequently, the robustness of the non-conformal DDMs is now firmly established theoretically and numerically.
- Multi-region/Multi-Solver DDM with Touching Regions: Many multi-scale physical problems are very difficult, if not impossible, to solve using just one of the existing CEM techniaues. We have been pursuing a multi-region multi-solver domain decomposition method (MS-DDM) to effectively tackle such problems. Various CEM solvers are now integrated into a MS-DDM code and collectively, it emerges as the only alternative for solving many real-life applications that were thought un-solvable before.
- Discontinuous Galerkin Time Domain (DGTD) Method with Hierarchical MPI-CUDA GPU Implementation: We shall also discuss in some details, our on-going efforts in the time domain simulation, the DGTD algorithm. Particularly, the use of graphical process unit (GPU) in speeding up the DGTD computations.
Science and Applications of CEMs and Multi-Physics Computations
Computational Electromagnetics (CEM) techniques, such as FDTD, BEM (or Method of Moments), FEM, are playing increasingly important roles in many electromagnetic applications. In this lecture, I shall first describe the fundamental principles behind the popular CEM methods, and elucidate their corresponding strengths and weaknesses.
The second part of the lecture focuses on how to combine a suite of CEM techniques and couplings of multi-physics modelling to solve challenging real-life engineering applications. Particularly, the following three main topics will be emphasized:
- Full wave solutions of EM radiations and scatterings in the vicinity of large composite platforms (with various thin coatings and exotic metamaterials). The multi-scale nature and fine geometrical features of this application tax significantly on engineering ingenuity. As a consequence, a plethora of novel CEM techniques, including multi-solver DDM, DDM for general integral equations for both PEC and lossy dielectric materials, PDE methods using polyhedral elements instead of conventional brick and tetrahedral elements, and the material homogenization, have been developed to mitigate these technical challenges directly.
- Co-design suite for modelling antenna systems (with front end electronics) and full IC packages taking into account both EM and thermal effects. In many real-life antenna systems, the temperature distributions in the environment as well as within the antennas affect greatly the overall system performance. Moreover, the signal and power integrity analyses for full IC packages usually require considerations of conductor losses, which are in general functions of the temperature. The multi-physics coupling, both in the frequency and time domains, between EM and thermal effects are the main concern in this topic.
- Large land vehicles on lossy rough surfaces, microwave imaging, and modelling 3D rough surface and random multi-layer media for applications in LCDs, organic LEDs. New simulation and modelling techniques are pursued diligently to address this application area, and preliminary results are promising.