scientific article; zbMATH DE number 1286325

From MaRDI portal
Publication:4242827





The article is devoted to studying solvability and properties of solutions to the following boundary value problem: \[ -\text{div} \sigma = f - \nabla p,\qquad \text{div} u = 0\quad\text{in }\Omega,\qquad u = 0\quad\text{on }\partial\Omega, \] where the velocity \(u:\Omega\to\mathbb{R}^2\) and the pressure \(p\) are unknown, whereas the exterior force \(f\) is a given function. Here \(\Omega\) is a bounded domain in \(\mathbb{R}^2\) with smooth boundary \(\partial\Omega\in C^2\). The stress tensor \(\sigma\) is assumed to be determined by the rule \[ \sigma = \frac{\partial g}{\partial\tau}\bigl(\varepsilon(u)\bigr), \quad g(\tau) = g_0(| \tau|)\quad\forall\tau\in\mathbb{M}^2_s, \] where \(\varepsilon(u)=\frac{1}{2}\bigl(\nabla u + (\nabla u)^T\bigr)\) is the strain velocities tensor, \(\mathbb{M}^2_s\) is the space of real symmetric second-order matrices. The function \(g_0\in C^2(\mathbb{R})\) is supposed to be positive and such that \[ \frac{L_0}{(1+t)^{2-r}}\leq g_0(t) \leq\frac{L_1}{(1+t)^{2-r}} \quad\forall t\in\mathbb{R}_+,\qquad 0\leq\frac{g_0'(t)}{t} - g_0(t)\quad\forall t>0, \] where \(L_0\) and \(L_1\) are some positive constants, \(r\in (1,2)\), and \(g_0'(0) =0\). For instance, \(g_0(t) = (1+t^2)^{r/2}\) satisfies the above conditions. The author studies the regularity near \(\partial\Omega\) of generalized solutions to the problem under consideration. Estimates are given in the Maurey space \(L^{r,\lambda}\) for \(r\approx 2\). The interior estimates are obtained by the regularity technique for weak solutions to variational problems in elastoplasticity [\textit{G. A. Seregin}, Appl. Math. Optimization 28, 307-335 (1993; Zbl 0779.73086)]. The main results are exposed in the form of the following conditions: \[ \sigma\in W^1_2(\Omega_0; \mathbb{R}^2),\quad p\in W^1_2(\Omega_0),\quad (1 + | \varepsilon (u)|)^{r-2}| \nabla\varepsilon(u)| \in L^2(\Omega_0), \] for \(\Omega_0\subset\Omega\) and \(\overline{\Omega}_0\cap\partial\Omega\subset\partial_0\Omega\), where \(\partial_0\Omega\) is a linear open part of the boundary of \(\Omega\). Moreover, if \(f\in\L^{\tilde r,\mu_0}(\Omega; \mathbb{R}^2)\), \(\mu_0 >0\), \(\tilde{r} = 2r/(3r - 4)\) then there exists a \(\nu\in (0,1)\) such that \(\varepsilon(u),\sigma\in C^{\nu}(\overline{\Omega}_0; \mathbb{M}_s^2)\).











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4242827)