针对孔隙介质中地下水流动问题提出了一种并行数值计算方法,并基于此设计了一套专用于求解大规模三维地下水流动方程的并行计算模块。计算模块基于区域分解的方法实现对模型区域的并行求解,采用了分布式内存和压缩矩阵技术解决大规模稀疏矩阵的存储及其计算,整合多种并行Krylov子空间方法和预条件子技术迭代求解大规模线性方程组。在Linux集群系统上进行了数值模拟实验,性能测试结果表明,程序具有良好的加速比和可扩展性。
This paper presents an efficient parallel computing method for large-scale groundwater flow simulation in porous media,and it is implemented in a prototype tool.Domain decomposition method is adopted to partition the modeling area.Distributed memory and compressed matrices technology are introduced for both the storage and evaluation of the large-scale sparse matrices.All kinds of Krylov subspace methods and preconditioners are integrated to assemble and solve linear systems of equations.Experiments are carried out on Linux clusters, and the results demonstrate that this method can achieve desirable speedup and scalability.