Elsevier

Journal of Computational Physics

Volume 357, 15 March 2018, Pages 324-337
Journal of Computational Physics

A multi-mesh finite element method for phase-field based photonic band structure optimization

https://doi.org/10.1016/j.jcp.2017.12.031Get rights and content

Highlights

  • We develop a multi-mesh finite element method for phase-field model of photonic band gap optimization.

  • The state equation is solved on a coarse mesh and the phase-field evolving on a fine mesh.

  • The proposed multi-mesh approach saves up to 60 percent computational efforts.

  • The adaptive meshes are investigated for saving computational costs.

Abstract

A novel finite element method with multiple meshes is proposed, which is applied to solve the phase-field models for photonic band structures optimization. In our approach, fine meshes are used for the phase field evolution, which allows fine resolution for shape representations. The coarse meshes are adopted for the finite element analysis of the state equation. Such a multi-mesh approach could save a considerable amount of computational costs. Numerical convergence is illustrated through comparisons between our computational results and benchmarks. The efficiency and robustness of the multi-mesh approach are also shown.

Introduction

Photonic crystals are periodic dielectric structures designed to prevent the propagation of electromagnetic waves with specific frequencies. The first studies could be dated back to Rayleigh's work on one-dimensional layered structures in 1887. However, the concept of photonic band gaps in multi-dimensions was introduced by Yablonovitch [1] and John [2] in 1987. With the development on the theory of shape and topology optimization [3], [4], [5], [6], a basic problem of photonic band structure optimization is to find a medium structure that exhibits a band gap as large as possible.

There has been plenty of contributions to maximization problems of band gaps for two-dimensional photonic crystals. In [7], [8], the authors proposed a generalized gradient ascent algorithm to maximize the band gaps iteratively, both for the transverse magnetic field (TM mode) and the transverse electric field (TE mode). In [9], Kao and Osher used the level set methods for maximizing band gaps. In this framework, the Binary Level Set Method (BLSM) is a constantly used strategy for topology and shape optimization in the design of photonic crystal [10] and two-density drum [11]. Various numerical optimization techniques, such as the genetic algorithm, are applied to photonic crystal design by many other researchers [12], [13], [14].

The feasibility of phase-field model for shape and topology optimization was first investigated by Bourdin and Chambolle [15]. In the literature of computational sciences, the phase field method is assumed to be an alternative way to track the interface efficiently [16], [17]. It is always served as a easy-to-use methodology for numerical simulations of the phase transition phenomenon [18], [19], as well as the two phase flow simulations [20], [21], [22]. Recently, Takezawa et al. proposed a phase field model for shape and topology optimization, with applications to structural optimization [23] and design of photonic devices [24]. The main contribution of their work is a new double-well function including the shape sensitive analysis, which is both theoretically featherable and numerically efficient for solving the optimization problem. However, a piecewise constant approximation is assumed for the design problems with two kinds of materials, so that the staircase phenomena could not be avoided. The work of Nguyen T et al. [28] suggests a multi-resolution scheme to obtain high resolution designs for structure, which is a novel idea for topology optimization and significant computational costs are saved.

In this paper, we propose a multi-mesh approach for the finite element calculations of the phase field method [24], a piecewise linear approximation is used for a continuous approximation to the phase field variable. The motivation of multi-mesh strategy is to reduce the overall computational costs without losing the numerical accuracy. The rest of this paper is organized as follows. The phase field models for optimization of photonic band structures are introduced in the next section. Section 3 introduces the details of multi-mesh approach, where the finite element analysis on the coarse mesh and the phase field evolution on the fine mesh are both formulated. In section 4, numerical results on some benchmark problems are reported. Finally, we made some conclusions and discussions.

Section snippets

Photonic band structure optimization

The photonic band structure optimization is a well-known model in computational material science. The work of Cox and Dobson [7] described the problem mathematically as a band gap maximization. For the ease of representation, let us consider the case of TM mode. When a scattered problem is considered, the purpose of the design is to maximize the gap between the j-th band λ(j) and the (j+1)-th band λ(j+1). It means to find an optimal shape Ω for the scattered object in the admissible shape set Da

The multi-mesh finite element methods

The multi-mesh approach can potentially reduce the time costs for solving the partial differential equation system. In [25], Li introduced how the multi-mesh working and built a new adaptive mesh strategy in the framework of multiple meshes settings. It was then proved to be effective in the finite element calculations for the simulations of dendritic growth [26], [29], [30].

Numerical experiments

Let us consider the problem of band gap maximization for photonic crystal design, which is served as the benchmark problem [7], [9], [10], [24]. The computational domain is set to be a square Ω=[0.5,0.5]2. In all the numerical examples presented in this section, the dielectric constants for the void and material are taken as ϵ0=1 and ϵ1=8.9 respectively. In our calculations, 30 equi-distributed wave number α located at the boundary of the Brillouin zone are calculated for plotting the

Conclusion

We propose a multi-mesh finite element method for the phase field model arising in photonic crystal design. The key contribution is to reduce the cost of finite element analysis on the state equation by solving on a much coarser meshes. The proposed approach has several advantages. The first one is the considerable savings in computational efforts compared with the single mesh scheme. The second advantage is the flexibility of approximation to the phase field variable, where the nodal finite

Acknowledgements

This research work is supported by the Fundamental Research Funds for the Central Universities 2016XZZX005-01, and the National Natural Science Foundation of China (Grant 11471092, 11471284 and 11201153).

References (32)

  • A. Madzvamuse

    Time-stepping schemes for moving grid finite elements applied to reaction–diffusion systems on fixed and growing domains

    J. Comput. Phys.

    (2006)
  • E. Yablonovitch

    Inhibited spontaneous emission in solid-state physics and electronics

    Phys. Rev. Lett.

    (1987)
  • S. John

    Strong localization of photons in certain disordered dielectric superlattices

    Phys. Rev. Lett.

    (1987)
  • O. Pironneau

    Optimal Shape Design for Elliptic Systems

    (1984)
  • J. Sokolwski et al.

    Optimal Shape Design for Elliptic Systems

    (1984)
  • B. Mohammadi et al.

    Applied Shape Optimization for Fluids

    (2001)
  • Cited by (20)

    • Unconditionally energy stable schemes for fluid-based topology optimization

      2022, Communications in Nonlinear Science and Numerical Simulation
    • A two-grid method for the phase-field model of photonic band gap optimization

      2021, Computers and Mathematics with Applications
      Citation Excerpt :

      Takezawa et al. extended the phase field method into photonic crystal band gap problem in [21]. Moreover, Wu et al. [23] proposed a multi-mesh approach for the finite element calculations by using the phase field method. Two-grid(also called two-mesh) method is an effective strategy in many applications of finite element methods.

    • A two-grid method for level-set based topology optimization with GPU-acceleration

      2021, Journal of Computational and Applied Mathematics
      Citation Excerpt :

      In 2010 Huang [20] proved that two-grid methods have optimal convergence rates for finite element solution to the semi-linear elliptic equations. In recent years, two-grid methods have been applied to other problems by many researchers, Wu et al. [21,22] proposed to apply two-grid methods to phase-field based dendritic growth simulation and photonic crystal design, in which considerable computational costs are saved in both applications due to the two-grid settings. Shao et al. [23] proposed a two-grid method based on modified Newton’s iteration for the Navier–Stokes equations.

    View all citing articles on Scopus
    View full text