Convolution finite element method: an alternative approach for time integration and time-marching algorithms (Q2666124)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Convolution finite element method: an alternative approach for time integration and time-marching algorithms
scientific article

    Statements

    Convolution finite element method: an alternative approach for time integration and time-marching algorithms (English)
    0 references
    0 references
    22 November 2021
    0 references
    The basis of the analysis is an alternative approach for time integration and time marching algorithms for wave propagation in solid elastic media. The problem states as follows: \(\Omega \subset \mathbb{R}^3\) is an open domain of the linear elastic body with Lipschitz boundary \(\Gamma\) and open subsets \(\Gamma_{gi}\) and \(\Gamma_{hi}\) for \(i = 1,2,3\) such that \(\Gamma_{gi} \cap \Gamma_{hi} = \emptyset\) and \(\Gamma_{gi} \cup \Gamma_{hi}\) is dense in \(\Gamma\). The governing equations are \begin{itemize} \item[1.] Equation of motion: \(\sigma_{i j,j} + \rho b_i = \rho \ddot{u}_i\) on \(\Omega \times(0,T)\), \item[2.] Kinematic equations: \(\varepsilon_{ij} = u_{(i,j)} = \frac12 (u_{i,j} + u_{j,i})\) on \(\Omega \times(0,T)\), \item[3.] Constitutive equations: \(\sigma_{ij} = C_{ijkl}\varepsilon_{kl}\) on \(\Omega \times(0,T)\), \item[4.] Boundary conditions: \(u_i = g_i(x,t)\) on \(\Gamma_{gi} \times(0,T)\), \(\sigma_{ij} n_j = h_i(x,t)\) on \(\Gamma_{hi} \times(0,T)\), \item[5.] Initial conditions: \(u_i(x,0) = u^0_i(x)\) on \(\bar\Omega\) and \(\dot{u}_i(x,0) = v^0_i (x)\) on \(\bar\Omega\), \end{itemize} where the components of \(u\), \(\varepsilon\) and \(\sigma\) are sufficiently smooth for a strong solution for the so defined initial boundary value problem. It is proven that the following set of equations is equivalent to the equation of motion and the initial conditions: \[ \rho u(x,t) = [t *(\nabla.\sigma + \rho f)](x,t) + \rho (u^0(x) + tv^0(x)), \] and furthermore that weak and strong solutions are equivalent under certain conditions. Furthermore, the Galerkin formulation corresponding to the convolution weighted residual is established and several properties of the method are discussed culminating in a solution procedure for elastodynamic problems for which numerical results are presented.
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    finite element method
    0 references
    initial-boundary value problem
    0 references
    convolutional variational principle
    0 references
    structural elastodynamics
    0 references
    linear elasticity
    0 references
    wave propagation
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references