A novel decoupled and stable scheme for an anisotropic phase-field dendritic crystal growth model (Q2274716)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | A novel decoupled and stable scheme for an anisotropic phase-field dendritic crystal growth model |
scientific article |
Statements
A novel decoupled and stable scheme for an anisotropic phase-field dendritic crystal growth model (English)
0 references
1 October 2019
0 references
The authors propose a numerical scheme adapted to the resolution of an anisotropic phase-field dendritic crystal growth model written as: \(\tau (\phi )\phi _{t}=\nabla \cdot (\kappa ^{2}(\nabla \phi )\nabla \phi +\kappa (\nabla \phi )\left\vert \nabla \phi \right\vert ^{2}\mathbf{H}(\phi ))- \frac{f(\phi )}{\epsilon ^{2}}-\frac{\lambda }{\epsilon }p^{\prime }(\phi )T\), \(T_{t}=D\Delta T+Kp^{\prime }(\phi )\phi _{t}\), posed in \(\Omega \times (0,\infty )\), where \(\Omega \) is a smooth, open, bounded and connected domain in \(\mathbb{R}^{d}\), \(d=2,3\). Here, \(\phi \) is a scalar phase-field function equal to \(1\) in the solid phase and to \(-1\) in the liquid phase, \(T\) is the temperature, \(\epsilon \), \(K\) and \(\lambda \) are positive parameters, \(f\) is the derivative of the Ginzburg-Landau double well potential \(F(\phi )=(\phi ^{2}-1)^{2}\), \(\kappa (\nabla \phi )\) is associated to the anisotropy property: \(\kappa (\nabla \phi )=1+\epsilon _{4}\cos (m\theta )\), where \(m\) is the number of folds of anisotropy, \(\epsilon _{4}\) is the anisotropy strength, \(\theta =\arctan (\frac{\phi _{y}}{\phi _{x}})\), \(\tau (\phi )\) is the mobility parameter, \(D\) is the diffusion rate of the temperature, \(\mathbf{H}(\phi )=4\epsilon _{4}\frac{4}{\left\vert \nabla \phi \right\vert ^{6}}(\phi _{x}(\phi _{x}^{2}\phi _{y}^{2}-\phi _{y}^{4}),\phi _{y}(\phi _{x}^{2}\phi _{y}^{2}-\phi _{x}^{4}))\) and \(p(\phi )=\frac{1}{5}\phi ^{5}-\frac{2}{3}\phi ^{3}+\phi \). The authors introduce the auxiliary function \(U(x,t)=\sqrt{\frac{1}{2}\left\vert \kappa (\nabla \phi )\nabla \phi \right\vert ^{2}+\frac{1}{4\epsilon ^{2}}F(\phi )+B}\), where \(B\) is a constant which ensures the positivity of the radicand. The problem is then rewritten as: \(\tau (\phi )\phi _{t}=-Z(\phi )U-\frac{ \lambda }{\epsilon }p^{\prime }(\phi )T\), \(U_{t}=\frac{1}{2}Z(\phi )\phi _{t} \), \(T_{t}=D\Delta T+Kp^{\prime }(\phi )\phi _{t}\). The initial conditions \( \phi (t=0)=\phi _{0}\), \(T(t=0)=T_{0}\), \(U(t=0)=\sqrt{\frac{1}{2}\left\vert \kappa (\nabla \phi _{0})\nabla \phi _{0}\right\vert ^{2}+\frac{1}{4\epsilon ^{2}}F(\phi _{0})+B}\)\ are imposed. The authors introduce the totally decoupled and linear scheme: \(\tau (\phi ^{n})\frac{\phi ^{n+1}-\phi ^{n}}{ \delta t}+\frac{S_{1}}{\epsilon ^{2}}(\phi ^{n+1}-\phi ^{n})-S_{2}\Delta (\phi ^{n+1}-\phi ^{n})=-Z^{n}U^{n+1}-\frac{\lambda }{\epsilon }p^{\prime }(\phi ^{n})T_{\ast }^{n}\), \(\frac{T_{\ast }^{n}-T^{n}}{\delta t}=Kp^{\prime }(\phi ^{n})\frac{\phi ^{n+1}-\phi ^{n}}{\delta t}\) \(U^{n+1}-U^{n}=\frac{1}{2 }Z^{n}(\phi ^{n+1}-\phi ^{n})\), where \(S_{1}\) and \(S_{2}\) are two positive stabilizing parameters and \(T^{n+1}-T_{\ast }^{n}=D\Delta T^{n+1}\). The authors prove that this scheme is unconditionally energy stable and satisfies the discrete energy dissipation law: \(\frac{1}{\delta T} (E^{n+1}-E^{n})\leq -\left\Vert \sqrt{\tau (\phi ^{n})}\frac{\phi ^{n+1}-\phi ^{n}}{\delta t}\right\Vert ^{2}-\frac{\lambda D}{\epsilon K} \left\Vert \nabla T^{n+1}\right\Vert \) with \(E^{n+1}=\int_{\Omega }(\left\vert U^{n+1}\right\vert ^{2}+\frac{\lambda }{2\epsilon K}\left\vert T^{n+1}\right\vert ^{2}-B)dx\). In the last part of the paper, the authors present results obtained through the numerical scheme and they prove its accuracy.
0 references
phase-field function
0 references
anisotropic dendritic growth
0 references
unconditionally energy stability
0 references
decoupled linear scheme
0 references
0 references
0 references
0 references
0 references
0 references
0 references
0 references