Please use this identifier to cite or link to this item:
|Title:||Towards interactive steering of a very large floating structure code by using HPC parallelisation strategies|
message passing paradigm
shared memory concept
very large floating structure
|Citation:||Frisch, J., Gao, R., Mundani, R.-P., Wang, C.M., Rank, E. (2012). Towards interactive steering of a very large floating structure code by using HPC parallelisation strategies. Proceedings - 14th International Symposium on Symbolic and Numeric Algorithms for Scientific Computing, SYNASC 2012 : 473-480. ScholarBank@NUS Repository. https://doi.org/10.1109/SYNASC.2012.15|
|Abstract:||Very large floating structures (VLFSs) have been used for broad applications such as floating storage facilities, floating piers, floating bridges, floating airports, entertainment facilities, even habitation, and other purposes. Owing to its small bending rigidity, VLFS deforms elastically when subjected to wave action. This elastic deformation due to wave is called hydro elastic response and it can be obtained by solving the interaction between the surface wave and the floating structure in the frequency domain. In solving the fluid-structure interaction, the floating structure can be modelled by applying the finite element method, whereas the fluid part may be analyzed by using the Green's function method. When using the Green's function which satisfies the boundary condition on the free-surface, the sea bottom and that at infinite distance from the floating structure, the unknown parameters to be determined for the fluid part can be minimized to be only those associated with the wetted surface of the floating structure. However, in the evaluation of the Green's function, extensive computation time O(N2) is needed (N is the number of unknowns). Therefore, acceleration techniques are necessary to tackle the computational complexity. Nowadays, standard multi-core office PCs are already quite powerful if all the cores can be used efficiently. This paper will show different parallelisation strategies for speeding up the Green's function computation. A shared memory based implementation as well as a distributed memory concept will be analysed regarding speed-up and efficiency. For large computations, batch jobs can be used to compute detailed results in high resolution on a large computational cluster or supercomputer. Different speed-up computations on clusters will be included for showing strong speed-up results. © 2012 IEEE.|
|Source Title:||Proceedings - 14th International Symposium on Symbolic and Numeric Algorithms for Scientific Computing, SYNASC 2012|
|Appears in Collections:||Staff Publications|
Show full item record
Files in This Item:
There are no files associated with this item.
checked on Jul 27, 2018
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.