Parallel Processing Approach For Crash Dynamic Analysis
Chiang, Kuo-ning, (1989) Doctoral Thesis, Georgia Institute of Technology, Atlanta.
In the area of nonlinear transient analysis, research is critically needed on the development and evaluation of parallel methods for nonlinear dynamic analysis of complex nonlinear finite element and/or finite difference structural problems. The development of general-purpose finite element structural analysis computer programs have provided the capability to address a wide range of structures problems (Ref. 2-4). However, these software systems are severely limited for nonlinear transient calculations because of the available speed on current sequential computers (Ref. 10).
Projected advances in computer technology indicate that significant increases in effective calculation speed will be available in the 1990's, through fifth generation supercomputer achitectures consisting of arrays of processors operating in parallel on different tasks (see e.g., Ref. 1 for a survey). Such advanced supercomputers, denoted as MIMD (multiple instruction, multiple data) computers, have the potential for increasing effective calculation speeds by several orders of magnitude. But this potential increase in speed cannot be effectively utilized without the development and implementation of appropriate numerical algorithms for structures which take advantage of the parallel computation features of this new generation of computers. Use of existing conventional algorithms and software will not realize the full potential of these new MIMD computers, and research is needed in the development of parallel structural analysis/design algorithms for these computers. The objective of this research has been to develop and evaluate parallel methods for crash dynamic analyses of complex nonlinear finite element and finite difference structure problems.
Manuscript: order via UMI