TY - GEN
T1 - Parallelization of a 3-D coast circulation model for continental shelf circulation - QUODDY
AU - Shao, Zhiyu S.
AU - Yost, Scott A.
AU - Chen, Chunfang
PY - 2005
Y1 - 2005
N2 - Numerical simulation of large scale problems has a sharply growing computation demand. One of the solutions of this problem is to apply parallel computing technology. In this paper, the dominant parallel strategy used in engineering application, Domain Decomposition, is discussed and implemented in three parallelization applications. The main focus of this study is a 3-D finite element model, QUODDY, which solves the linearized shallow water equations forced by tidal or other barotropic boundary conditions, wind and/or density gradients. A 2.5 degree turbulent closure model by Mellor Yamada is utilized in the simulation. The sequential QUODDY code is studied first. Then Domain Decomposition approach implemented in the sequential code. By dividing the task into several smaller portions, the computation time was reduced dramatically. Message Passing Interface (MPI) protocols are used for the communications in all processors. Implementation of MPI improves the portability of the paralleled code. The performance of the parallel code is studied and shows good performance. Copyright ASCE 2005.
AB - Numerical simulation of large scale problems has a sharply growing computation demand. One of the solutions of this problem is to apply parallel computing technology. In this paper, the dominant parallel strategy used in engineering application, Domain Decomposition, is discussed and implemented in three parallelization applications. The main focus of this study is a 3-D finite element model, QUODDY, which solves the linearized shallow water equations forced by tidal or other barotropic boundary conditions, wind and/or density gradients. A 2.5 degree turbulent closure model by Mellor Yamada is utilized in the simulation. The sequential QUODDY code is studied first. Then Domain Decomposition approach implemented in the sequential code. By dividing the task into several smaller portions, the computation time was reduced dramatically. Message Passing Interface (MPI) protocols are used for the communications in all processors. Implementation of MPI improves the portability of the paralleled code. The performance of the parallel code is studied and shows good performance. Copyright ASCE 2005.
UR - http://www.scopus.com/inward/record.url?scp=37249013854&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=37249013854&partnerID=8YFLogxK
U2 - 10.1061/40792(173)278
DO - 10.1061/40792(173)278
M3 - Conference contribution
AN - SCOPUS:37249013854
SN - 0784407924
SN - 9780784407929
T3 - World Water Congress 2005: Impacts of Global Climate Change - Proceedings of the 2005 World Water and Environmental Resources Congress
SP - 278
BT - World Water Congress 2005
T2 - 2005 World Water and Environmental Resources Congress
Y2 - 15 May 2005 through 19 May 2005
ER -