EFFICIENT NUMERICAL SIMULATION OF OFFSHORE STRUCTURES AND WIND TURBINES ON GENERAL PURPOSE GRAPHICS PROCESSING UNITS

被引:0
作者
Muskulus, Michael [1 ]
机构
[1] Norwegian Univ Sci & Technol, Dept Civil & Transport Engn, N-7491 Trondheim, Norway
来源
COMPUTATIONAL METHODS IN MARINE ENGINEERING IV (MARINE 2011) | 2011年
关键词
Computational Engineering; Marine Engineering; Offshore Wind Turbines; Multibody Dynamics; Parallel Computation; GPU; TIME-INTEGRATION METHODS; DYNAMICS; ALGORITHMS; SYSTEMS;
D O I
暂无
中图分类号
U6 [水路运输]; P75 [海洋工程];
学科分类号
0814 ; 081505 ; 0824 ; 082401 ;
摘要
Offshore wind turbines are complex systems that are usually simulated in the time-domain to take account of non-linear effects. The response of the structure is calculated by the finite-element method, or with the more general multi-body methods, and aerodynamic loads are calculated by specialized software which needs to interact with the structural code at each time step. This leads to problematic performance issues, and a ratio of simulated versus simulation time of 1:1 is about the current state-of-the-art. To tackle this issue that stands in the way of both industrial and research efforts requires a computational approach that fully integrates structural analysis with hydrodynamic and aerodynamic calculations, and recent developments in high-performance computing can be favourably employed here. Highly optimized graphics processor units were originally developed to speed up computer game experiences, but have now been adapted for computing tasks. This has been taken advantage of in many scientific and engineering contexts, and here we review the possibilities for marine engineering, with special regard for floating structures and offshore wind turbines. Structural response calculations rely on the solution of large sparse linear systems, and this can lead to drastic increases in performance when implemented in parallel on such general purpose graphics processing units. The main algorithms and remaining challenges are discussed, and we demonstrate the feasibility of such an approach.
引用
收藏
页码:438 / 449
页数:12
相关论文
共 45 条
[1]  
[Anonymous], 2011, NVIDIA CUDA C PROGRA
[2]  
[Anonymous], 2016, Programming massively parallel processors: a hands-on approach
[3]  
[Anonymous], 1997, Parallel programming with MPI
[4]  
[Anonymous], 2011, CUDA EXAMPLE INTRO G
[5]  
[Anonymous], 2003, ITERATIVE METHODS SP, DOI DOI 10.1137/1.9780898718003
[6]  
Bauchau OA, 2011, SOLID MECH APPL, V176, P3, DOI 10.1007/978-94-007-0335-3
[7]   A parallel nearly implicit time-stepping scheme [J].
Botchev, MA ;
van der Vorst, HA .
JOURNAL OF COMPUTATIONAL AND APPLIED MATHEMATICS, 2001, 137 (02) :229-243
[8]  
Brown D.T., 2005, HANDBOOK OF OFFSHORE
[9]  
Burton T, 2011, WIND ENERGY HDB, DOI DOI 10.1002/9781119992714
[10]  
Butenhof D. R., 1997, PROGRAMMING WITH POS