Research in electromagnetic engineering at the Silesian University - - PowerPoint PPT Presentation
Silesian University of Technology Gliwice, Poland Research in electromagnetic engineering at the Silesian University of Technology, Gliwice, Poland Andrzej Karwowski Silesian University of Technology Department of Electronics 3 rd Cost
Silesian University of Technology Gliwice, Poland
Andrzej Karwowski Silesian University of Technology Department of Electronics
3rd Cost Meeting Action IC0603 “ASSIST” Limassol, Cyprus April 9 – 11, 2008
Silesian University of Technology Gliwice, Poland
Who am I ?
Andrzej Karwowski, PhD, DSc Member IET, IEEE (AP-S, EMC-S, MTT-S) Professor at SUT Leader of Radioelectronics and Electromagnetic Engineering (EE) Research Team at SUT
e-mail: a.karwowski@theiet.org, a.karwowski@ieee.org
Silesian University of Technology Gliwice, Poland
Outline Silesian University of Technology (SUT) Key facts Electromagnetic engineering research team What are we doing and how? “Fast” hybrid MoM-PO method Numerical example
Silesian University of Technology Gliwice, Poland
Silesian University of Technology (SUT)
Upper Silesia Region (Katowice) the greatest industrial region in Poland population 4.9 million Gliwice main location of STU (9 Faculties; 3 in Katowice and Zabrze)
Silesian University of Technology Gliwice, Poland
Silesian University of Technology (SUT) Statistics
Number of students 32000 Number of academic 1750 teachers Number of employees 3300 Number of graduates 125000 Number of PhD 3500 graduates
Silesian University of Technology Gliwice, Poland
Silesian University of Technology (SUT) 12 Faculties
Architecture Automatic Control, Electronics and Computer Science Civil Engineering Chemistry Electrical Engineering Mining and Geology Material Science and Metallurgy Energy and Environmental Engineering Mathematics and Physics Mechanical Engineering Organization and Management Transport
Silesian University of Technology Gliwice, Poland
Silesian University of Technology (SUT)
Faculty of Automatic Control, Electronics and Computer Science Department of Automatic Control Department of Electronics Department of Computer Science Department of Electronics Courses in Electronics and Telecommunications with honors in Electronic equipment, Biomedical electronics, Microelectronics, Telecommunications, Radioelectronics Radioelectronics Electromagnetic Engineering Research Team (5 PhDs permanent)
Silesian University of Technology Gliwice, Poland
Electromagnetic Engineering Research Team What are we generally interested in?
General Research Interests Computational Electromagnetics (CEM) Development and application of numerical modeling techniques and tools for predicting electromagnetic behavior of radiating and scattering structures. Emphasis placed on the so-called “full-wave” methods. Applications mainly in antennas and electromagnetic compatibility.
Silesian University of Technology Gliwice, Poland
Electromagnetic Engineering Research Team What methodologies and tools we employ?
MoM – Frequency-domain Integral Equation (MPIE) approach combined with Method-of-Moments for non-penetrable conducting objects involving arbitrary combinations of wires and surfaces FDTD – classical Yee/Taflove approach with necessary modifications/improvements for problems involving non- homogeneous structures with lossy dielectrics Asymptotic Techniques – Asymptotic Waveform Evaluation (AWE), High-Frequency techniques (Geometrical Optics (GO), Physical Optics (PO), Geometrical Theory of Diffraction (GTD)
Silesian University of Technology Gliwice, Poland
Electromagnetic Engineering Research Team What methodologies and tools we employ?
MBPE – Model Based Parameter Estimation Tools – Commercial full-wave 3D electromagnetic simulators from CST, AWR, selected packages from Mentor Graphics, Matlab. Many in-house codes (the team members are fluent in Fortran programming (Lahey Fortran 77, Lahey Fortran 95).
Silesian University of Technology Gliwice, Poland
Electromagnetic Engineering Research Team What are we currently involved in?
Computationally Efficient Wideband “Fast” Hybrid Techniques MoM-PO – MoM combined with PO, the impedance matrix interpolation in the frequency domain, and adaptive sampling of the observable MoM-FDTD – …
Silesian University of Technology Gliwice, Poland
Fast MoM-PO Method
MoM-PO – this hybrid formulation implies decomposition of the structure into MoM region and PO region. The MoM region is handled by MoM, and the effects from the secondary sources in PO region are incorporated in boundary conditions for the em field in MoM region (Jakobus, Landstorfer; 1995) MoM-PO impedance matrix interpolation – the impedance matrices [Z] are precomputed (and stored) at several relatively distant nodes and then interpolation is used to find [Z] at in intermediate frequencies (Newman; 1988) Dynamic adaptive sampling of the observable – initially, the
frequency nodes and then the adaptive sampling based upon bisection is used to find the observable value elsewhere
Silesian University of Technology Gliwice, Poland
Fast MoM-PO Method Numerical example
Geometry of the antenna: 5 loops; diameter 5.1 cm, pitch angle 14o , wire radius 1mm) The entire structure is subdivided into 4056 triangular patches and 83 linear segments Structure decomposition: MoM region – 179 basis functions PO region – 5910 basis functions
conducting objects”, Electronics Letters, vol. 43, no. 9, pp. 486-487, Apr. 2007.
bandwidth”, Microwave and Optical Technology Letters, Vol. 58, No. 3, pp. 738-741, March, 2008
Helical antenna driven at the base against a corner PEC platform
Silesian University of Technology Gliwice, Poland
Fast MoM-PO Method Numerical example
b) a)
2
g t t
9 9
7.772 10 3.33 10 1.2 10
Vs s s A g t
derivative of the Gaussian pulse
Silesian University of Technology Gliwice, Poland
Fast MoM-PO Method Numerical example
Current magnitude at the antenna driving-point versus frequency
Silesian University of Technology Gliwice, Poland
For matrix interpolation, the frequency nodes were taken at 200 MHz (11 nodes for 100 MHz to 2100 MHz frequency range). Only 3 nodal matrices and currently evaluated [Z] matrix are stored For adaptive current interpolation, an initial frequency step of 20 MHz was taken. The convergence tolerance of 1% was taken. The adaptive algorithm generated 776 non-uniformly spaced current samples with the initial frequency resolution locally refined by a factor of 512 resulting in frequency step of 39.0625 kHz. To meet the convergence criterion with uniform sampling, 51201 samples would be required.
Fast MoM-PO Method Numerical example
Silesian University of Technology Gliwice, Poland
The total execution time required to compute 776 + 11 current samples
Fast MoM-PO Method Numerical example
MoM MoM-PO+I+S
number of basis functions MM region PO region 6089
5910 RAM 600 MB 35 MB time 113 min 6.5 min
time required to compute a single current sample !!!
Computer resources needed to perform computations of the current at the driving-point of the antenna
Silesian University of Technology Gliwice, Poland
Silesian University of Technology Gliwice, Poland
Time required to compute 776 samples and 11 interpolation nodes Time of computations of a single sample directly by MM-PO – 32.8s Time of computations of a single sample by MM-PO+I – 0.04s
Numerical results
MM MM-PO+I+S
number of basis functions MM region PO region 6089
5910 RAM 570 MB 35 MB time 113 min 6.5 min
Computer resources needed to perform computations
time of computations