AI Lab logo
menu MENU

Faculty Candidate Seminar

Computational Prototyping of Power Electronics and Energy Conversion Systems

Ali DavoudiUniversity of Illinois
SHARE:

Computational prototyping is generally defined as the process of substituting computer-generated models for physical prototypes. Accurate modeling algorithm and fast simulation techniques for power and energy systems are needed for dynamic characterization, controller design and analysis, real-time simulation, and probabilistic characterization of fault mechanisms and recovery processes. The promise is to allow designers to examine more radical, more efficient, and wider range of design alternatives. The challenge is in developing accurate modeling algorithms and simulation techniques which are fast yet accurate. Therefore, a futuristic approach demands revolutionizing design verification tools and analysis environments. An ideal computational prototyping environment should accurately model and simulate components, devices, and the overall integrated energy system. To address these issues, physics-based model-order reduction, parametric averaging, and multi-resolution simulation, respectively, are discussed.


Average-value modeling of switching systems is indispensable for computationally efficient frequency-domain characterization and/or for system-level time-domain transient studies. A simulation-based average-value modeling framework, based on a detailed switch-level simulation, is considered. This methodology avoids laborious analytical derivations and crude approximations. Moreover, it is seamlessly functional in all operational modes (continuous, boundary, discontinuous conduction modes), enables all control schemes (line-commutated rectifier, voltage- and/or current-mode converters), and provides design environments for different power electronics systems (e.g., dc-dc converters, brushless dc machine/synchronous machine-rectifier systems).


Physics-based models, e.g., extracted from finite-element methods and high-fidelity magnetic equivalent circuits, are highly accurate but computationally expensive. High-order dynamic physics-based models are developed that account for high-frequency eddy currents, laminations, 3-D effects, saturation, and relative motion. These models are later reduced with mathematically rigorous reduction techniques to a lower order model that includes only the dominant states in the desired bandwidth, thus preserving both simulation accuracy and computational efficiency.


Multi-resolution simulation offers rapid and accurate simulation of power electronics systems from a very general to a very detailed consideration. Numerically rigorous, order-reduction techniques are used to extract several levels of interconnected simulation resolutions from an originally high-order detailed model. The simulation resolution can be adjusted at the user’s discretion, even during a simulation run, yielding different simulation resolutions and speeds.
Ali Davoudi received the B.Sc. and M.Sc. degrees in Electrical and Computer Engineering from Sharif University of Technology, Tehran, Iran, and The University of British Columbia, Vancouver, Canada, in 2003 and 2005, respectively. Currently, he is working toward his Ph.D. at the Electrical and Computer Engineering Department, University of Illinois at Urbana-Champaign, and is expected to complete his degree requirements by July 2009. He worked for Texas Instruments Inc. and Royal Philips Electronics. He is also the recipient of UBC international graduate student fellowship, UIUC Henry Ford II Scholar award, and UIUC M. E. VanValkenburg graduate research award. He is the co-author of 40 journal articles, conference papers, and invention disclosures. His research interests are all aspects of modeling, simulation, dynamic characterization, and control of power electronics and energy conversion systems, biomechanical energy harvesting, and energy source diversification.

Sponsored by

EECS/ECE