Error analysis for the finite element approximation of the Darcy-Brinkman-Forchheimer model for porous media with mixed boundary conditions (Q2195894)

The authors consider a viscous flow inside a porous medium embedded in a computational domain \(\Omega\subset\mathbb R^d\), (\(d=2,3\)). The domain is assumed to be a bounded open set with Lipschitz boundary with outward unitary normal \(\vec n\). It is assumed the porous medium has a spatially varying porosity \(\varepsilon: x\in\Omega\longmapsto\varepsilon(x)\in(0,1]\) and that it is modeled by the Darcy-Brinkman-Forchheimer equation (DBF). The latter can be formulated in dimensionless form as follows \[ \begin{cases} -\mathrm{div}(2Re^{-1}\varepsilon S(\vec u)-\varepsilon\vec u\otimes\vec u))+\varepsilon\nabla p+\alpha(\varepsilon)\vec u+\beta(\varepsilon)\vec u|\vec u|=\varepsilon\vec f, &\text{in } \Omega,\\ \mathrm{div}(\varepsilon\vec u)=0, &\text{in } \Omega, \end{cases} \tag{1} \] where \(Re\) is the Reynolds number and \(\vec f\) is an external force field, the tensor \(S(\vec u)\) is the symmetric part of the Jacobian matrix of the velocity field \(\vec u\). The following boundary conditions are considered \[ \begin{cases} \vec u=0, &\text{on }\Gamma_w,\\ \vec u=\vec u_{in}, &\text{on }\Gamma_{in},\\ \varepsilon(2Re^{-1}S(\vec u)-p)\vec u=0, &\text{on }\Gamma_{out} \end{cases} \tag{2} \] where \(\partial\Omega=\Gamma_w\cup\Gamma_{in}\cup\Gamma_{out}\) and each part corresponds respectively to the walls, the inlet and the outlet. Under some assumptions on \(\varepsilon,\alpha,\beta\) the authors reduce the problem to nonlinear variational problem in the spaces \[ \mathbf{X}_1=\{\vec v\in H^1(\Omega)^d, |\vec v|_{\Gamma_w=0}\},\quad \mathbf{X}=\{\vec v\in H^1(\Omega)^d, |\vec v|_{\Gamma_w\cup\Gamma_{in}=0}\}. \] They proved the well-posedness of the DBF model (1) with mixed boundary conditions (2) as well as the convergence of the Taylor-Hood finite element method when using a discrete porosity. They also provided a fixed point iteration, and proved its convergence to solve the discrete non-linear problem and gave some numerical experiments to illustrate the optimal error estimates obtained theoretically.