In this paper, a composite grid method (CGM) for finite element (FE) analysisof an electromagnetic field with strong local interest is proposed. The method is based on theregular finite element method in conjunction w...In this paper, a composite grid method (CGM) for finite element (FE) analysisof an electromagnetic field with strong local interest is proposed. The method is based on theregular finite element method in conjunction with three basic steps, i.e. global analysis, localanalysis, and modified global analysis. In the first two steps, a coarse finite element mesh is usedto analyze the global model to obtain the nodal potentials which are subsequently used asartificial boundary conditions for local regions of interest. These local regions with theprescribed boundary conditions are then analyzed with refined meshes to obtain more accuratepotential and density distribution In the third step, a modified global analysis is performed toobtain more improved solution for potential and density distribution. And iteratively, successivelyimproved solutions can be obtained until the desired accuracy is achieved. Various numericalexperiments show that CCM yields accurate solutions with significant savings in computing timecompared with the regular finite element method.展开更多
文摘In this paper, a composite grid method (CGM) for finite element (FE) analysisof an electromagnetic field with strong local interest is proposed. The method is based on theregular finite element method in conjunction with three basic steps, i.e. global analysis, localanalysis, and modified global analysis. In the first two steps, a coarse finite element mesh is usedto analyze the global model to obtain the nodal potentials which are subsequently used asartificial boundary conditions for local regions of interest. These local regions with theprescribed boundary conditions are then analyzed with refined meshes to obtain more accuratepotential and density distribution In the third step, a modified global analysis is performed toobtain more improved solution for potential and density distribution. And iteratively, successivelyimproved solutions can be obtained until the desired accuracy is achieved. Various numericalexperiments show that CCM yields accurate solutions with significant savings in computing timecompared with the regular finite element method.