TY - GEN

T1 - Towards a parallel time integration method for nonlinear systems

AU - van der Valk, Paul L.C.

AU - Rixen, Daniel J.

N1 - Publisher Copyright:
© The Society for Experimental Mechanics, Inc. 2014.

PY - 2014

Y1 - 2014

N2 - Obtaining the forced dynamic response of large nonlinear structural models is in practice computationally expensive. As time integration involves solving a static-like nonlinear problem at each time steps, these simulations could take up to several days to solve. In a lot of cases however, the global nonlinearity of the model could be relatively mild or parts of the model can be assumed to behave linearly and the strong nonlinearities, that require many iterations to solve, are localized in a small number of regions of the model. Normal approaches to solve this more efficiently require one to reduce the linear and/or mildly nonlinear parts of the system. In this paper however, a different approach is taken. Here we decompose the total time integration by separating the iterations required, into iterations on the global (linearized) interface problem and iterations on the (local) substructure level. It will be shown that this approach leads to a method that can be efficiently implemented in a parallel computing environment.

AB - Obtaining the forced dynamic response of large nonlinear structural models is in practice computationally expensive. As time integration involves solving a static-like nonlinear problem at each time steps, these simulations could take up to several days to solve. In a lot of cases however, the global nonlinearity of the model could be relatively mild or parts of the model can be assumed to behave linearly and the strong nonlinearities, that require many iterations to solve, are localized in a small number of regions of the model. Normal approaches to solve this more efficiently require one to reduce the linear and/or mildly nonlinear parts of the system. In this paper however, a different approach is taken. Here we decompose the total time integration by separating the iterations required, into iterations on the global (linearized) interface problem and iterations on the (local) substructure level. It will be shown that this approach leads to a method that can be efficiently implemented in a parallel computing environment.

KW - Dual assembly

KW - Newmark time integration

KW - Nonlinear models

KW - Parallel computations

KW - Simulations

UR - http://www.scopus.com/inward/record.url?scp=84988732068&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-04501-6_12

DO - 10.1007/978-3-319-04501-6_12

M3 - Conference contribution

AN - SCOPUS:84988732068

SN - 9783319007700

T3 - Conference Proceedings of the Society for Experimental Mechanics Series

SP - 135

EP - 145

BT - Dynamic Behavior of Materials - Proceedings of the 2013 Annual Conference on Experimental and Applied Mechanics

PB - Springer New York LLC

T2 - 32nd IMAC Conference and Exposition on Structural Dynamics, 2014

Y2 - 3 February 2014 through 6 February 2014

ER -