자료유형 | 학위논문 |
---|---|
서명/저자사항 | Efficient Tsunami Simulation at Local and Global Scales. |
개인저자 | Qin, Xinsheng. |
단체저자명 | University of Washington. Civil and Environmental Engineering. |
발행사항 | [S.l.]: University of Washington., 2019. |
발행사항 | Ann Arbor: ProQuest Dissertations & Theses, 2019. |
형태사항 | 175 p. |
기본자료 저록 | Dissertations Abstracts International 81-05B. Dissertation Abstract International |
ISBN | 9781687947444 |
학위논문주기 | Thesis (Ph.D.)--University of Washington, 2019. |
일반주기 |
Source: Dissertations Abstracts International, Volume: 81-05, Section: B.
Advisor: Motley, Michael R. |
이용제한사항 | This item must not be sold to any third party vendors.This item must not be added to any third party search indexes. |
요약 | Tsunami hazard evaluation and mitigation is of great importance to coastal communities around the world, especially after the frequent occurrence of large tsunamis in the past two decades. Many physical phenomena need to be modeled during a tsunami event, e.g. tsunami wave generation and propagation, coastal inundation, and forces on structures. Most of them are nonlinear and involve a wide range of length scales, and thus are challenging to model. In this dissertation, the ability of three-dimensional (3D) and two-dimensional (2D) models to capture tsunami forces on structures and flow through a constructed environment is first analyzed. Then the development of a GPU-accelerated hyperbolic partial differential equation (PDE) solver with adaptive mesh refinement (AMR), with application to solving several PDEs that govern different physical processes arising in tsunamis, is presented and discussed. Tsunami inundation is the final and most destructive phase of tsunami evolution that comes after tsunami wave propagation in the ocean. The numerical modeling of this phase that incorporates the constructed environment of coastal communities is challenging for both 2D and 3D models. Inundation and flooding in this region can be too complex for 2D models to capture properly, while for 3D models a very fine mesh is required to properly capture the physics, dramatically increasing the computational cost and rendering impractical modeling of some problems. To evaluate the capability of the current tsunami inundation models, comparisons are made between GeoClaw, a depth-integrated 2D model based on the nonlinear shallow water equations (NSWE), and the interFoam solver in OpenFOAM, a 3D model based on Reynolds Averaged Navier-Stokes (RANS) equations for tsunami inundation modeling. The two models are first validated against existing experimental data of a bore impinging onto a single square column. Then they are used to simulate tsunami inundation in a physical wave tank model of Seaside, Oregon. The resulting flow parameters from the models are compared and discussed, and these results are used to extrapolate tsunami-induced force predictions and give guidance for the use of numerical models in other similar situations. Numerical modeling of tsunami processes is computationally expensive. Being able to do this faster means we can simulate a problem with higher resolution to potentially get more accurate result, simulate the same problem faster to send out tsunami warning earlier, or perform more tsunami simulations within a given time budget when doing probabilistic hazard assessment or studying the uncertainties of the process. Using Adaptive Mesh Refinement (AMR) as implemented in GeoClaw speeds up the process by greatly reducing computational demands, while accelerating the code using the Graphics Processing Unit (GPU) could do so through faster hardware but has not previously been implemented in GeoClaw. The second part of this dissertation presents an efficient CUDA implementation of the GeoClaw code. The code can model transoceanic tsunami simulation by using AMR and solving the shallow water equations in spherical coordinates. Numerical experiments of the 2011 Japan tsunami and a local tsunami triggered by a hypothetical Mw 7.3 earthquake on the Seattle Fault illustrate the correctness and efficiency of the code. The GPU implementation, when running on a single GPU, is observed to be 3.6 to 6.4 times faster than the original model running in parallel on a 16-core CPU. Three metrics are proposed to evaluate performance of the model, which shows efficient usage of hardware resources. |
일반주제명 | Civil engineering. Applied mathematics. Computer science. |
언어 | 영어 |
바로가기 | ![]() |