The variational multiscale element free Galerkin method for MHD flows at high Hartmann numbers

详细信息    查看全文

• 作者：Lin Zhang^a ; Jie Ouyang^b ; Xiaohua Zhang^c ; ^{zxyxclzmrzln@163.com} ; ^{zhangxiaohua07@163.com}
• 关键词：MHD flows ; EFG ; VM ; Hartmann number
• 刊名：Computer Physics Communications
• 出版年：2013
• 出版时间：April, 2013
• 年：2013
• 卷：184
• 期：4
• 页码：1106-1118
• 全文大小：1253 K

文摘

The aim of the paper is the development of an efficient numerical algorithm for the solution of magnetohydrodynamics (MHD) flow problems with either fully insulating walls or partially insulating and partially conducting walls. Toward this, we first extend the influence domain of the shape function for the element free Galerkin (EFG) method to have arbitrary shape. When the influence factor approaches 1, we find that the EFG shape function almost has the Delta property at the node (i.e. the value of the EFG shape function of the node is nearly equal to 1 at the position of this node) as well as the property of slices in the influence domain of the node (i.e. the EFG shape function in the influence domain of the node is nearly constructed by different functions defined in different slices). Therefore, for MHD flow problems at high Hartmann numbers we follow the idea of the variational multiscale finite element method (VMFEM) to combine the EFG method with the variational multiscale (VM) method, namely the variational multiscale element free Galerkin (VMEFG) method is proposed. Subsequently, in order to validate the proposed method, we compare the obtained approximate solutions with the exact solutions for some problems where such exact solutions are known. Finally, several benchmark problems of MHD flows are simulated and the numerical results indicate that the VMEFG method is stable at moderate and high values of Hartmann number. Another important feature of this method is that the stabilization parameter has appeared naturally via the solution of the fine scale problem. Meanwhile, because this proposed method is a type of meshless method, it can avoid the need for meshing, a very demanding task for complicated geometry problems.