Year: 2006
Type: dissertation
Abstract: A rising trend in science and engineering is in the utilization of increasingly highfidelity and accurate analysis codes in the design analysis and optimization process.In many application areas such as photonics, electromagnetics, aerospace, biomedical, micro-electro-mechanical systems and coupled-field multidisciplinary system design processes, simulation procedures involving Computational Structural Mechanics (CSM), Computational Fluid Dynamics (CFD) or Computational Electronics & Electromagnetics (CEE) are an important step in the design process.The time taken for these processes generally varies from many minutes to hours or days of supercomputing time.This often leads to high computing costs in the design optimization process, hence a much longer design cycle time to locate a near optimum design solution.This thesis presents a Gridenabled scalable parallel evolutionary framework for solving computationally expensive optimization problems under limited time budget.The essential backbone of the approach leverages the de facto standards of Grid computing to facilitate secure access to heterogeneous processing nodes and software resources that may be dispersed across many geographical locations in the evolutionary optimization search.By exploiting the Grid evolution framework and a multi-level parallelization strategy of hierarchical parallel GAs, we present the evolutionary optimization of a realistic 2D aerodynamic airfoil structure in a Grid environment having multiple heterogeneous clusters, i.e., clusters with differing specifications and processing nodes.From our results obtained, it is possible to conclude that the Grid enabled hierarchical parallel evolutionary algorithm is not a mere hype but offers a credible alternative, providing significant time reduction to complex engineering design optimization.
DOI: 10.32657/10356/35751
Author: Hee Khiang Ng
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W2544811845