Computational & Technology Resources
an online resource for computational,
engineering & technology publications
PROCEEDINGS OF THE SECOND INTERNATIONAL CONFERENCE ON PARALLEL, DISTRIBUTED, GRID AND CLOUD COMPUTING FOR ENGINEERING
An External Code Coupler based on a Subdomain Decomposition Method Extended to Non-Linear Cases
A. Batti1, M. Brun2, A. Gravouil1 and A. Combescure1
1Laboratory for Contact and Structural Mechanics, UMR CNRS 5259,
A. Batti, M. Brun, A. Gravouil, A. Combescure, "An External Code Coupler based on a Subdomain Decomposition Method Extended to Non-Linear Cases", in , (Editors), "Proceedings of the Second International Conference on Parallel, Distributed, Grid and Cloud Computing for Engineering", Civil-Comp Press, Stirlingshire, UK, Paper 17, 2011. doi:10.4203/ccp.95.17
Keywords: sub-domain decomposition method, finite element code, non-matching time scales, transient non-linear dynamic problems, external code coupling.
The GC method has been proposed by Gravouil and Combescure for Newmark time schemes in linear dynamics  and, then, extended to non linear dynamics . Batti et al.  have proposed a sub-domain decomposition method called GC+ and also the GC+ external code coupler which is energy conserving; which permits parallel computation of free and link problems and which was limited to transient linear dynamic problems.
In this paper, we have presented the key equations of the GC+ sub-domain decomposition method in the case of a same time step used for all the sub-domains composing the structure. Then, we have described a linear external code coupler using the GC+ method in the case of multi-time scales. The extension of the GC+ method and of the GC+ external code coupling to the non-linear cases is proposed. The non-linear GC+ external code coupler manages to couple a macro non-linear implicit FE code with a micro non-linear explicit FE code. We have seen with an industrial numerical example, that the non-linear GC+ external code coupler is well used (with appropriate choice of the definition of the two macro implicit and micro explicit sub-domains of the global structure; and appropriate choice of values of macro and micro time steps imposed in each sub-domain during the analysis computation) and permits us to obtain good results in terms of maximal displacements and velocities of the structure considered, ensuring continuity of velocities at the interface between the two chosen sub-domains, ensures a weak energy dissipation at the interface, and needs very less CPU time than classical mono-domain solving.
purchase the full-text of this paper (price £20)