• DocumentCode
    1428843
  • Title

    Computation of Second Order Capacitance Sensitivity Using Adjoint Method in Finite Element Modeling

  • Author

    Ren, Zhuoxiang ; Qu, Hui ; Xu, Xiaoyu

  • Author_Institution
    UPMC, Paris, France
  • Volume
    48
  • Issue
    2
  • fYear
    2012
  • Firstpage
    231
  • Lastpage
    234
  • Abstract
    The sensitivity of capacitances to the design parameters can be computed either by the derivative of state variables (the derivative method) or by the adjoint variable method. This paper proposes a second order sensitivity computation method that combines the two methods in exploring the solutions of the derivative of state variables and of the adjoint variables. By using the variational method for the electric flux computation, it can be shown that the right hand side of the adjoint system is identical to the right hand side of the original system in capacitance matrix extraction. The computational complexity of the method is hence identical to that of the first order sensitivity using the derivative method, which is linear to the number of design parameters. The first and second order derivatives of the material matrix necessary in the implementation are obtained with the help of the first and second order local jacobian derivatives. The method is validated through an example of a comb-drive microelectromechanical system (MEMS).
  • Keywords
    Jacobian matrices; capacitance; finite element analysis; micromechanical devices; Jacobian derivative; MEMS; adjoint variable method; capacitance matrix extraction; comb-drive microelectromechanical system; electric flux computation; finite element modeling; second order capacitance sensitivity; Capacitance; Conductors; Equations; Finite element methods; Jacobian matrices; Materials; Sensitivity; Adjoint variable method; finite element method; first and second order jacobian derivatives; second order sensitivity;
  • fLanguage
    English
  • Journal_Title
    Magnetics, IEEE Transactions on
  • Publisher
    ieee
  • ISSN
    0018-9464
  • Type

    jour

  • DOI
    10.1109/TMAG.2011.2172194
  • Filename
    6136740