浅水方程大规模并行计算模拟城市洪水演进
Simulation of Flood Propagation in Cities Using Large Scale Parallel Computation of Shallow Water Equations

摘 要：基于有限体积法建立求解溃坝水流运动的数学模型，利用 Riemann 问题求解 Godunov 格式计算通量，通过 空间分块划分的大规模并行计算来加速运算，实现了城市洪水演进的精细数值模拟．数值模拟结果表明，建筑物在溃坝波波前到达初期所受水流荷载最大；受上游建筑物消波、蔽护作用，溃口下游建筑物所受荷载随至溃口的距离的增大而减小，建筑物的存在使得溃坝波在城区传播速度减慢．并行测试表明，缓冲模式比非缓冲模式更适用于浅 水方程大规模并行求解．目前大规模并行计算能够实现对面积达 400,km2 的城区洪水演进进行实时仿真模拟，为滨 河、滨海城市在受到溃坝、海啸等洪水威胁时的快速决策和应急疏散提供科学依据．
Mathematical model was established for dam-break flow based on finite volume method. Godunov scheme was adopted to evaluate the flux of the Riemann problem. Spatial decomposition using blocks was used for large scale parallelization and computation acceleration. With these，fine simulation of the flood propagation in urban cities was realized. Numerical simulation results show that：buildings experience the peak load from flow at the initial stage of the arrival of the dam-break wave front；owing to the wave dissipation and shielding effect by the upstream buildings，the load decreases with the increase of the distance to the dam-break gate；the existence of buildings slows down the propagation of the flood. Parallel computation tests show that the buffered data mode is more suitable for large scale parallel computation of shallow water equations than the non-buffered mode. Large scale parallel computation with current technique can fulfil real-time simulation of flood propagation in urban areas as large as 400,km2，which can scientifically support the fast decision and emergency evacuation when riverside or coastal cities experience flood threats from dam-break or tsunamis