有限元水流计算中内存和运行效率初探

Exploration on memory requirement and operation efficiency of finite element method in flow calculation

  • 摘要: 有限元在非恒定流计算中耗内存大、耗机时多的问题一直困扰着工程师,限制了有限元的应用.采用分步有限元方法离散二维浅水方程,并初步探索了行指标矩阵压缩存储稀疏矩阵的方法和效率以及预条件双共扼梯度法在求解有限元方法中形成的大型线性方程组中的效率,取得了满意的结果.在一定程度上缓解了有限元在非恒定流计算中存在的耗内存大、耗时多的问题.

     

    Abstract: The problem of bulky memory requirement and low efficiency in calculation of unsteady flow with finite element method(FEM) has troubled the engineers for long time.It also restricts the applications of FEM to fluid dynamics The method of time-splitting and FEM were combined to solve the 2-D shallow water equations.The row-indexed sparse storage mode was used to store the sparse coefficient matrix,and the preconditioned bi-conjugate gradient method was used to solve the huge linear system.The results are satisfactory both in memory and efficiency The problem has been solved to a certain extent.

     

/

返回文章
返回