Highly parallel time integration of viscoelastic flows

2001 ◽  
Vol 100 (1-3) ◽  
pp. 191-216 ◽  
Author(s):  
A.E. Caola ◽  
Y.L. Joo ◽  
R.C. Armstrong ◽  
R.A. Brown
Author(s):  
Shilei Han ◽  
Olivier A. Bauchau

Traditionally, the time integration algorithms for multibody dynamics are in sequential. The predictions of previous time steps are necessary to get the solutions at current time step. This time-marching character impedes the application of parallel processor implementation. In this paper, the idea of computing a number of time steps concurrently is applied to flexible multi-body dynamics, which makes parallel time-integration possible. In the present method, the solution at the current time step is computed before accurate values at previous time step are available. This method is suitable for small-scale parallel analysis of flexible multibody systems.


2005 ◽  
Vol 45 (1) ◽  
pp. 197-217 ◽  
Author(s):  
Bernhard A. Schmitt ◽  
Rüdiger Weiner ◽  
Helmut Podhaisky

2021 ◽  
Author(s):  
D King ◽  
C Hills ◽  
M Kielstra ◽  
M Torrence

2019 ◽  
Vol 27 (0) ◽  
pp. 822-830
Author(s):  
Akihiro Fujii ◽  
Shigeo Kaneko ◽  
Teruo Tanaka ◽  
Takeshi Iwashita

Author(s):  
Thomas Richter ◽  
Nils Margenberg

We present a parallel time-stepping method for fluid-structure   interactions. The interaction between the incompressible   Navier-Stokes equations and a hyperelastic solid is formulated in a   fully monolithic framework. Discretization in space is based on   equal order finite element for all variables and a variant of the   Crank-Nicolson scheme is used as second order time integrator. To   accelerate the solution of the systems, we analyze a parallel-in   time method. For different numerical test cases in 2d and in 3d we   present the efficiency of the resulting solution approach. We also   discuss some challenges and limitations that are connected   to the special structure of fluid-structure interaction problem.   In particular, we will investigate stability and dissipation     effects of the time integration and their influence on the     convergence of the Parareal method. It turns out that especially     processes based on an internal dynamics (e.g. driven by the vortex     street around an elastic obstacle) cause great     difficulties. Configurations however, which are driven by     oscillatory problem data, are well-suited for parallel time     stepping and allow for substantial speedups.


PAMM ◽  
2014 ◽  
Vol 14 (1) ◽  
pp. 951-952 ◽  
Author(s):  
Robert Falgout ◽  
Stephanie Friedhoff ◽  
Tzanio Kolev ◽  
Scott MacLachlan ◽  
Jacob B. Schroder

Sign in / Sign up

Export Citation Format

Share Document