Homogenization of a pseudo-parabolic system via a spatial-temporal decoupling: upscaling and corrector estimates for perforated domains (Q2305088)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Homogenization of a pseudo-parabolic system via a spatial-temporal decoupling: upscaling and corrector estimates for perforated domains
scientific article

    Statements

    Homogenization of a pseudo-parabolic system via a spatial-temporal decoupling: upscaling and corrector estimates for perforated domains (English)
    0 references
    0 references
    0 references
    0 references
    10 March 2020
    0 references
    The authors start with the pseudo-parabolic problem for unknown functions \(V_{\alpha }^{\epsilon }\), \(U_{\alpha }^{\epsilon }\) with \(\alpha \in \{1,\ldots ,N\}\) posed in \((0;T)\times \Omega ^{\epsilon }\) as: \( \sum_{\beta =1}^{N}M_{\alpha \beta }^{\epsilon }V_{\beta }^{\epsilon }-\sum_{i,j=1}^{d}\frac{d}{dx_{i}}(E_{ij}^{\epsilon }\frac{dV_{\alpha }^{\epsilon }}{dx_{j}}+\sum_{\beta =1}^{N}D_{i\alpha \beta }^{\epsilon }V_{\beta }^{\epsilon })=\mathcal{H}_{\alpha }^{\epsilon }+\sum_{\beta =1}^{N}(K_{\alpha \beta }^{\epsilon }U_{\beta }^{\epsilon }+\sum_{i=1}^{d} \widetilde{J}_{ij}^{\epsilon }\frac{dU_{\beta }^{\epsilon }}{dx_{i}})\), \( \frac{dU_{\alpha }^{\epsilon }}{dt}+\sum_{\beta =1}^{N}L_{\alpha \beta }U_{\beta }^{\epsilon }=\sum_{\beta =1}^{N}G_{\alpha \beta }V_{\beta }^{\epsilon }\). The domain \(\Omega ^{\epsilon }\) is deduced from a simply connected bounded domain \(\Omega \subset \mathbb{R}^{d}\), \(d=2,3\), with a \(C^2\)-boundary \(\partial \Omega \) when removing periodically distributed identical perforations of size \(\epsilon \). The boundary of \(\Omega \) is divided in two parts, its inner boundary \(\partial _{int}\Omega ^{\epsilon }\), which corresponds to the boundary of the inner holes, and its outer boundary \( \partial _{ext}\Omega ^{\epsilon }\). The initial condition \(U_{\beta }^{\epsilon }=U_{\beta }^{\ast }\) is imposed on \(\{0\}\times \Omega ^{\epsilon }\) together with the Dirichlet and Neumann boundary conditions \(V_{\alpha }^{\epsilon }=0\) on \((0;T)\times \partial _{ext}\Omega ^{\epsilon }\) and \(\frac{ dV_{\alpha }^{\epsilon }}{d\nu _{D^{\epsilon }}}=\sum_{i=1}^{d}( \sum_{j=1}^{d}E_{ij}^{\epsilon }\frac{dV_{\alpha }^{\epsilon }}{dx_{j}} +\sum_{\beta =1}^{N}D_{i\alpha \beta }^{\epsilon }V_{\beta }^{\epsilon })n_{i}^{\epsilon }=0\) on \((0;T)\times \partial _{int}\Omega ^{\epsilon }\). The authors define the notion of weak solution to this problem introducing a variational formulation. The coefficients of this problem satisfy appropriate boundedness and smoothness properties, which imply the existence of a unique weak solution \((U^{\epsilon }V^{\epsilon })\in H^{1}((0;T)\times \Omega ^{\epsilon })^{N}\times L^{\infty }((0;T);\mathbb{V}_{\epsilon }\cap H^{2}(\Omega ^{\epsilon }))^{N}\) to the problem, with \(\mathbb{V}_{\epsilon }=\{v\in H^{1}(\Omega ^{\epsilon })\mid v=0\) on \(\partial _{ext}\Omega ^{\epsilon }\}\). The authors build an extension operator \(\mathcal{P} ^{\epsilon }\in \mathcal{L}(L^{2}(\Omega ^{\epsilon });L^{2}(\Omega ))\cap \mathcal{L}(\mathbb{V}_{\epsilon };H_{0}^{1}(\Omega ))\) such that, for any \( v\in \mathbb{V}^{\epsilon }\), the bounds \(\left\Vert \mathcal{P}^{\epsilon }v\right\Vert _{L^{2}(\Omega )}\leq C\left\Vert v\right\Vert _{L^{2}(\Omega ^{\epsilon })}\) and \(\left\Vert \nabla \mathcal{P}^{\epsilon }v\right\Vert _{L^{2}(\Omega )^{d}}\leq C\left\Vert \nabla v\right\Vert _{L^{2}(\Omega ^{\epsilon })^{d}}\)\ for some constant \(C>0\). They prove uniform estimates for the solution \(\left\Vert U^{\epsilon }\right\Vert _{H^{1}(\Omega ^{\epsilon })^{N}}(t)\leq Ce^{\lambda t}\) and \(\left\Vert V^{\epsilon }\right\Vert _{(\mathbb{V}_{\epsilon })^{N}}\leq C(1+\widetilde{\kappa } e^{\lambda t})\) for some positive constants \(C\), \(\widetilde{\kappa }\) and \( \lambda \) and that the extensions of this solution two-scale converge to some limit together with some of their derivatives. This allows to upscale the solution through this two-scale convergence process. The authors build the limit problem satisfied by the two-scale limits. They also upscale the problems through asymptotic expansions and this lead to the same limits. A further result deals with correctors that the authors build and for which they prove a convergence rate. The paper ends with an application to concrete corrosion.
    0 references
    0 references
    0 references
    0 references
    0 references
    periodic homogenization
    0 references
    Neumann boundary conditions
    0 references
    two-scale convergence
    0 references
    upscaled system
    0 references
    corrector estimates
    0 references
    concrete corrosion
    0 references
    0 references
    0 references
    0 references