Parallel Computation of a Dam-Break Flow Model Using OpenACC Applications

Abstract

Two key factors in dam-break modeling are accuracy and speed. Therefore, high-performance calculations are of great importance to the simulation of dam-break events. In this study, we develop a two-dimensional hydrodynamic model based on the finite volume method to simulate the dam-break flow routing process. Roe’s approximate Riemann solution is adopted to solve the interface flux of grid cells and accurately simulate the discontinuous flow. A graphics processing unit (GPU)-based parallel method, OpenACC, is used to realize parallel computing. Because an explicit discrete technique is used to solve the governing equations, and there is no correlation between grid calculations in a single time step, the parallel dam-break model can be easily realized by adding OpenACC directives to the loop structure of the grid calculations. To analyze the performance of the model, we considered the Pangtoupao flood storage area in China using a Nvidia Tesla K20c card and four different grid division schemes. By carefully studying the implementation method and optimization of data transportation in the parallel algorithm, a speedup factor of 20.70 can be achieved. This acceleration is better than that of the OpenMP method with a 16-kernel computer. Further analysis reveals that models involving a larger number of calculations exhibit greater efficiency and a higher speedup rate. In addition, the OpenACC parallel mode has good portability, making it easy to realize parallel computation from the original serial model. This GPU-based parallel computation has the advantages of high performance and easily available required hardware.