A CFD-based high-order Discontinuous Galerkin solver for three dimensional electromagnetic scattering problems. (May 2015)

• Record Type: Journal Article
• Authors: Lu, Hongqiang; Xu, Yida; Gao, Yukun; Sun, Qiang; Qin, Wanglong
• Is Part Of: Advances in engineering software. Volume 83(2015)
• Page Start: 1
• Page End: 10
• Publication Date: 2015-05
• Subjects: Time-domain Maxwell's equations -- Discontinuous Galerkin method -- Computational efficiency -- Radar cross-section -- Quadrauture-free implementation -- Parallel computing; Computer-aided engineering -- Periodicals; Engineering -- Computer programs -- Periodicals; Engineering -- Software -- Periodicals; Periodicals; 620.0028553
• DOI: 10.1016/j.advengsoft.2015.01.001
• Languages: English
• ISSNs: 0965-9978

Abstract

Highlights: We develop a new high-order curved mesh generation method for high-order DG methods. Intersecting elements can not be observed even in highly-stretched cases. The quality of the generated mesh can be controlled by choosing proper modulus of elasticity. The method introduced in this paper is very easy to implement and to be extended to 3d case. Abstract: In this paper, a CFD (Computational Fluid Dynamics) based DG (Discontinuous Galerkin) method is introduced to solve the three-dimensional Maxwell's equations for complex geometries on unstructured grids. In order to reduce the computing expense, both the quadrature-free implementation method and the parallel computing based on domain decomposition are employed. On the far-field boundary, the non-reflecting boundary condition is implemented. Numerical integration rather than the quadrature-free implementation is used over the faces on the solid boundary to implement the electromagnetic solid boundary condition for perfectly conducting objectives. Both benchmark examples and complex geometry case are tested with the CFD-based DG solver. Numerical results indicate that highly accurate results can be obtained when using high order even on coarse grid and the present method is very suitable for complex geometries. Furthermore, the costs of CPU time and the speedup of the parallel computation are also evaluated.