Computational & Technology Resources
an online resource for computational,
engineering & technology publications
PROCEEDINGS OF THE SIXTH INTERNATIONAL CONFERENCE ON PARALLEL, DISTRIBUTED, GPU AND CLOUD COMPUTING FOR ENGINEERING
Edited by: P. Iványi and B.H.V. Topping
Smoothed Particle Hydrodynamics for water wave generation
S. Lee, J.W. Jung, K. Ko and J.W. Hong
Department of Civil and Environmental Engineering, Korea Advanced Institute of Science and Technology, Daejeon, Republic of Korea
S. Lee, J.W. Jung, K. Ko, J.W. Hong, "Smoothed Particle Hydrodynamics for water wave generation", in P. Iványi, B.H.V. Topping, (Editors), "Proceedings of the Sixth International Conference on Parallel, Distributed, GPU and Cloud Computing for Engineering", Civil-Comp Press, Stirlingshire, UK, Paper 12, 2019. doi:10.4203/ccp.112.12
Keywords: smoothed particle hydrodynamics, wavemaker, water wave generation.
In this study, the optimal parameters for Smoothed Particle Hydrodynamics (SPH) simulations are investigated, and the parallelization efficiency for the wavemaker simulation is evaluated. An indoor wavemaker has been commonly used to simulate offshore waves because it can accurately produce linear or nonlinear water waves described by various theoretical equations. However, it is still difficult to conduct large-scale experiments using indoor wavemakers due to the excessive construction costs. This limitation stimulates the development of numerical techniques to reduce time and cost. A number of wavemaker simulations have been conducted to solve fluid mechanics problems, and the application area has been extended to fluid-structure interaction (FSI), free surface flows, etc.
The conventional Eulerian method has been widely used to solve free surface wave problems by discretizing the fluid domain withmeshing. However, the mesh-based methods have several disadvantages in analyzing free surface problems due to the large movement of the free surface and irregular deformation. Unlike the Eulerian method, the SPH method is able to describe the fluid with a finite number of particles, which classifies the approach as a Lagrangian method. Since the fluid consists of discrete particles interacting with neighboring particles, a large number of particles are required, and the computation is very expensive. Therefore, it is important to find optimal parameters to ensure accurate simulation and an appropriate number of threads for parallel computing.
In order to determine the optimal values, we conduct wavemaker simulations using a commercial code LS-DYNA by controlling the parameters that affect the computational accuracy and cost. Through the use of mass-weighted damping to suppress unwanted wave reflection at the end of a water flume, a uniform regular wave is generated, and the numerical results are compared with analytical and experimental results. After determining the optimal values of SPH simulation parameters, the computation times are evaluated by increasing the thread number up to 32. When the number of threads increases above a certain level, the speedups converge. Optimizing parameters and corresponding thread numbers for SPH simulations through parametric studies that ensure numerical accuracy and efficiency would be beneficial for a large number of SPH simulations in a batch mode.
purchase the full-text of this paper (price £22)