|
Researchspace >
General science, engineering & technology >
General science, engineering & technology >
General science, engineering & technology >
Please use this identifier to cite or link to this item:
http://hdl.handle.net/10204/6351
|
| Title: | An accelerated, fully-coupled, parallel 3D hybrid finite-volume fluid–structure interaction scheme |
| Authors: | Malan, AG Oxtoby, OF |
| Keywords: | 3D fluid-structure-interaction Vertex-centered finite volume Algorithms Modelling technologies |
| Issue Date: | Sep-2012 |
| Publisher: | Elsevier |
| Citation: | Malan, AG and Oxtoby, OF. 2012. An accelerated, fully-coupled, parallel 3D hybrid finite-volume fluid–structure interaction scheme. Computer Methods in Applied Mechanics and Engineering, vol. 253, pp 426-438 |
| Series/Report no.: | Workflow;9739 Workflow;9888 |
| Abstract: | In this paper we propose a fast, parallel 3D, fully-coupled partitioned hybridunstructured finite volume fluid–structure-interaction (FSI) scheme. Spatial discretisation is effected via a vertex-centered finite volume method, where a hybrid nodal-elemental strain procedure is employed for the solid in the interest of accuracy. For the incompressible fluid, a split-step algorithm is presented which allows the entire fluid-solid system to be solved in a fully-implicit yet matrix-free manner. The algorithm combines a preconditioned GMRES solver for implicit integration of pressures with dual-timestepping on the momentum equations, thereby allowing strong coupling of the system to occur through the inner solver iterations. Further acceleration is provided at little additional cost by applying LU-SGS relaxation to the viscous and advective terms. The solver is parallelised for distributed-memory systems using MPI and its scaling efficiency evaluated. The developed modelling technology is evaluated by application to two 3D FSI problems. The advanced matrix-free solvers achieve reductions in overall CPU time of up to 50 times, while preserving close to linear parallel computing scaling using up to 128 CPUs for the problems considered. |
| Description: | Copyright: 2012 Elsevier. This is the pre-print version of the item. The definitive version is published in Computer Methods in Applied Mechanics and Engineering, vol. 253, pp 426-438 |
| URI: | http://www.sciencedirect.com/science/article/pii/S0045782512002794 http://hdl.handle.net/10204/6351 |
| ISSN: | 0045-7825 |
| Appears in Collections: | Aeronautic systems High performance computing Advanced mathematical modelling and simulation General science, engineering & technology
|
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.
|