Two Dimensional Hydrodynamic Numerical Model Based On Unstructured Grid Gpu Acceleration Technique

In recent years,with the frequent occurrence of extreme weather such as rainstorm and hail in the world,secondary disasters such as debris flow and flood are becoming more and more common.As the main means to study flood problems,flood numerical model has made important contributions in flood control forecast.In response to the national policy on flood disaster and the compilation of national flood risk map,aiming at the shortcomings of the current two-dimensional numerical model,a set of hydrodynamic numerical model with more accurate terrain description,higher simulation accuracy and faster calculation speed is developed.The unstructured grid is used in the model,which has strong flexibility for complex configuration,can adapt to various complex terrain,reduces the number of unnecessary grids,and can be effectively applied to various complex terrain and boundary conditions.The finite volume method(FVM)of Godunov scheme is used to solve the two-dimensional shallow water equations in the flood routing process,and the GPU technology is introduced to accelerate the calculation,which greatly improves the calculation speed without reducing the accuracy of the grid.The model can simulate the flood process when the boundary conditions are complex and the river channel is winding and rugged,and the calculation results are accurate and the acceleration effect is remarkable.Based on modeling,the accuracy,harmony,water conservation and stability of the model are verified by five classic examples,which shows that the model can deal with complex mixed flow and has high-resolution shock wave capture ability.In addition,through the simulation calculation of the actual flood routing process,it shows that the model can be effectively used to simulate the actual flood routing process,including the ordinary flood routing process,flood floodplain process and dam break flood,and the simulation results are in good agreement with the literature value and the measured value.Therefore,through the above examples,the model can simulate flood routing well,and the conclusion is effective and reliable.The model can provide strong technical support for flood prevention and early warningFinally,different devices and models are used to simulate and compare the calculation time.After using GPU to accelerate the calculation process,the calculation time of the model is greatly reduced;The ERMS of the simulation results is less than half of the standard deviation of the simulation values,and the ENS are all above 0.78.The simulation results are in good agreement with the literature values and the measured values.At the same time,the calculation efficiency of the unstructured GPU model is higher.Compared with the CPU model,the acceleration ratio of the unstructured GPU model is 38.4 or above;Compared with the structural GPU model,the computational efficiency of the unstructured GPU model can be increased by two times or more.For complex terrain,due to its excellent adaptability,unstructured grid can better describe the terrain details and reduce the calculation of redundant grid,which makes the calculation results more accurate and faster.This model has important application value to guide the flood risk map compilation of flood control coordination construction.