Document Type

Thesis - University Access Only

Award Date

2011

Degree Name

Master of Science (MS)

Department / School

Mechanical Engineering

Abstract

Despite the great growth in capability of computer hardware, the system size, complexity of structures, and time scales present in virtual prototyping of multi-body dynamical systems will continue to challenge the field of computational multi-body dynamics for the foreseeable future. In this thesis, the scientific problems in parallel virtual prototyping of multi-body dynamical systems are articulated. An efficient hybrid parallelizable algorithm is then introduced to address these problems. The procedure has been developed through hybridization between sequential order Nor O(N) procedure (N: total number of bodies in a multi-body system (MBS)) and parallel computing techniques, and hybridization between direct methods and iterative methods. Implementation of the algorithm on TeraGrid computing systems for large-sized MBS is further discussed. The algorithm is coded in C and MPI. Some MPI message passing techniques such as buffered message passing strategy have been used to reduce communication overhead. Correctness of the codes has been verified and compared with baseline bench mark cases. Various simulation cases and computing results are presented to demonstrate impact of the TeraGrid super computers to the performance of the algorithm. TeraGrid has provided computing facilities and resources for implementation of the algorithm for virtual prototyping of large-sized multi-body dynamic systems, which is not possible with the cluster-based computers at author's institute. TeraGrid offers higher computing efficiency than the cluster computing systems. Specifically, simulation cases and results are further presented in detail for the demonstration of computing efficiency and flexibility of the algorithm on TeraGrid super computers. An N body tether has been used as a case study to carry out simulation with number of bodies ranging from 16 to 512. The method is based on cutting system inter-body joints so that the system of largely independent multi-body sub chains is formed. These sub chains interact with one another through associated constraint forces at the cut joints. The increased parallelism is obtained through cutting joints and the explicit determination of associated constraint forces combined with a sequential O(n) procedure. The bodies with computational loads were distributed among processors of a TeraGnd computing machine. In one extreme case, all bodies were assigned to one processor so that the sequential O(n) performance was achieved. In another extreme case, each body was assigned to one processor individually for concurrent computing so that a parallel O(log 2 n) performance was achieved. Generally, the algorithm will give a computational performance between 0( n) and O(log 2 n). The algorithm has been implemented on the TeraGrid supercomputing machines such as Big-Ben and Big Red. Big-Ben is a Cray XT3 super-computer system with 2,068 nodes. Each node contains two 2.6 GHz AMD Opteron single core processors which share 2 GB of memory and network connection. Big-Ben is located in Pittsburgh Supercomputing Center (PSC) [73], and Big Red is located in Indiana University. Big Red is a distributed shared-memory cluster, consisting of 768IBM JS21 Blades, each with two Dual-Core PowerPC 970 MP Processors, 8GB of memory, and a PCI-X Myrinet 2000 adaptor for high bandwidth, low-latency MPI applications.

Library of Congress Subject Headings

Dynamics -- Computer simulation

Parallel algorithms

Format

application/pdf

Number of Pages

139

Publisher

South Dakota State University

Download

Share

COinS