The inclusion of thin lossy, material layers, such as carbon based composites, is essential for many practical applications modeling the propagation of electromagnetic energy through composite structures such as those found in vehicles and electronic equipment enclosures. Many existing schemes suffer problems of late time instability, inaccuracy at low frequency, and/or large computational costs. This work presents a novel technique for the modeling of thin-layer lossy materials in FDTD schemes which overcomes the instability problem at low computational cost. For this, a 1D-subgrid is used for the spatial discretization of the thin layer material. To overcome the additional time-step constraint posed by the reduction in the spatial cell size, a Crank-Nicolson time-integration scheme is used locally in the subgridded zone, and hybridized with the usual 3D Yee-FDTD method, which is used for the rest of the computational domain. Several numerical experiments demonstrating the accuracy of this approach are shown and discussed. Results comparing the proposed technique with classical alternatives based on impedance boundary condition approaches are also presented. The new technique is shown to have better accuracy at low frequencies, and late time stability than existing techniques with low computational cost.

Abstract-Thin-sheet models are essential to allow shielding effectiveness of composite enclosures and vehicles to be modeled. Thin dispersive sheets are often modeled using surface-impedance models in finite-difference time-domain (FDTD) codes in order to deal efficiently with the multiscale nature of the overall structure. Such boundary conditions must be applied to collocated tangential electric and magnetic fields on either side of the surface; this is usually done on the edges of the FDTD mesh cells at the electric field sampling points. However, these edge-based schemes are difficult to implement accurately on stair-cased surfaces. Here, we present a novel face-centered approach to the collocation of the fields for the application of the boundary condition. This approach naturally deals with the ambiguities in the surface normal that arise at the edges on stair-cased surfaces, allowing a simpler implementation. The accuracy of the new scheme is compared to edge-based and conformal approaches using both planar sheet and spherical shell canonical test cases. Staircasing effects are quantified and the new face-centered scheme is shown have up to 3-dB lower error than the edge-based approach in the cases considered, without the complexity and computational cost of conformal techniques.Index Terms-Finite-difference time domain (FDTD), impedance network boundary condition, surface-impedance boundary condition (SIBC).

We present a new face centered approach to the collocation of the fields for the application of a surface impedance boundary condition (SIBC). This approach deals with the ambiguities in the surface normal that arise at the edges on staircased surfaces.. The accuracy of the new scheme is compared to edge based and conformal approaches using both planar sheet and spherical shell test cases. Stair-casing effects are evaluated and the face-centered scheme exhibits significantly less error than the edge based approach.

No abstract

Thin layer models are widely used in the finitedifference time-domain (FDTD) technique to efficiently model boundaries in multi-scale simulations as they significantly reduce simulation run-times and memory requirements. These models often utilise surface impedance boundary conditions (SIBCs) to represent the material of the boundary. Conformal meshes are a popular method of representing curved and non-aligned surfaces in FDTD. These meshes deform cells in the FDTD grid around the boundary between bulk materials so as to more accurately represent the shape of the material. Here we present an algorithm that combines the efficiency of a thin layer model with the accuracy of a conformal mesh. The algorithm is applied to three resonant cavity models and the accuracy verified using comparisons to non-conformal meshes and analytic solutions. Improvements are shown in the accuracy of the resonant frequencies and magnitude of the shielding effectiveness (SE) of the cavities. It is also shown to reduce the prevalence of extraneous features in the frequency response of the SE that are apparent when using a stair-cased mesh.