High-performance multiprocessor computers provide new and interesting opportunities to solve large-scale structural engineering problems. However, the development of new computational models and algorithms that exploit the unique architecture of these machines remains a challenge.
High Performance Computing in Structural Engineering explores the use of supercomputers with vectorization and parallel processing capabilities in structural engineering applications. The book focuses on the optimization of large structures subjected to the complicated, implicit, and discontinuous constraints of commonly used design codes and presents robust parallel-algorithms for analysis of these structures.
The authors apply the algorithms to and analyze the performance of minimum weight designs of large, steel space trusses and moment-resisting frames, with or without bracings, consisting of discrete standard shapes. They clearly show that adroit and judicious use of vectorization techniques can improved the speedup of an optimization algorithm, and that parallel processing can lead to even further speedup.
With its review of the necessary background material, generous illustrations, and unique content, this is the definitive resource for the analysis and optimization of structure on shared-memory multiprocessor computers. By extension, High Performance Computing in Structural Engineering will prove equally valuable in distributed computing on a cluster of workstations
Table of Contents
Solution of Simultaneous Linear Equations
Parallel-Vector Algorithms for Analysis of Large Structures
Impact of Vecorization on Large-Scale Structural Optimization
Optimization of Large Steel Structures Using Standard Cross Sections
Optimum Load and Resistance Factor Design of Large Steel Structures
"Someone actively working in this field would find High-Performance Computing in Structural Engineering useful"…This book contains many 'step-by-step' algorithms and numerous high quality figures."
--B. H. Thacker,Applied Mechanics, review (Vol. 54, No. 3 March 2001).