Simulation methods are an important part of today’s product development process. At later stages of the product development process, it is necessary to utilize more complex simulation models which lead to increased computational times. In order to realize efficient simulation processes, computational times have to be reduced. Since nearly every personal computer provides more than one core, parallelization methods are one of the most promising ways. In this thesis, the finely-, middle- and coarsely-structured level of parallelization methods are investigated regarding their potential for non-smooth multibody dynamics. Concerning the middle-structured level, two new internal parallelization methods for the parallel evaluation of the equations of motion are presented. On the coarsely-structured level, two Time-Stepping schemes with inexact Jacobian matrices and parallelization methods within the numerical integration schemes are proposed. The huge potential of the developed methods is presented by academic and complex industrial examples.
«
Simulation methods are an important part of today’s product development process. At later stages of the product development process, it is necessary to utilize more complex simulation models which lead to increased computational times. In order to realize efficient simulation processes, computational times have to be reduced. Since nearly every personal computer provides more than one core, parallelization methods are one of the most promising ways. In this thesis, the finely-, middle- and coars...
»