A Note on Hyperbolic Relaxation of the Navier-Stokes-Cahn-Hilliard system for incompressible two-phase flow

Jens Keim, Hasel-Cicek Konan, Christian Rohde · 2024

Read A Note on Hyperbolic Relaxation of the Navier-Stokes-Cahn-Hilliard system for incompressible two-phase flow by Jens Keim, Hasel-Cicek Konan, Christian Rohde (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-11904.txt.

Introduction

Jens Keim, Hasel-Cicek Konan, and Christian Rohde offer a note on the flow of two fluids that do not mix. Their paper is from two thousand twenty-four. The model they use joins the Navier-Stokes equations with the Cahn-Hilliard equation. Its solutions show strong non-local effects, which make the math hard to solve.

They build a simpler first-order system for part of the flow. It uses a form of fake squeeze for one part and a friction-like step for the other. They prove the new system is hyperbolic in the one-way case, with a well-shaped entropy pair. Since it is hyperbolic, standard solver tools apply, and early tests in one way are shown.

Catalog record

CREATORJens Keim, Hasel-Cicek Konan, Christian Rohde
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE64 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-11904

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

A Note on Hyperbolic Relaxation of the Navier-Stokes-Cahn-Hilliard system for incompressible two-phase flow Jens Keim, Hasel-Cicek Konan, Christian Rohde Abstract . We consider the two-phase dynamics of two incompressible and immiscible fluids. As a mathematical model we rely on the Navier-Stokes-Cahn-Hilliard system that belongs to the class of diffuse-interface models. Solutions of the Navier-Stokes-Cahn-Hilliard system exhibit strong non-local effects due to the velocity divergence constraint and the fourth-order Cahn-Hilliard operator. We suggest a new first-order approximative system for the inviscid sub-system. It relies on the artificial-compressibility ansatz for the Navier-Stokes equations, a friction-type approximation for the Cahn-Hilliard equation and a relaxation of a third-order capillarity term. We show under reasonable assumptions that the first-order operator within the approximative system is hyperbolic; precisely we prove for the spatially one-dimensional case that it is equipped with an entropy-entropy flux pair with convex (mathematical) entropy. For specific states we present a numerical characteristic analysis. Thanks to the hyperbolicity of the system, we can employ all standard numerical methods from the field of hyperbolic conservation laws. We conclude the paper with preliminary numerical results in one spatial dimension. Keywords: two-phase flow, diffuse-interface modelling, Navier-Stokes-Cahn-Hilliard system, first-order hyperbolic relaxation 1. Introduction The incompressible motion of a viscous one-phase fluid is governed by the Navier–Stokes system. In absence of external forces and setting the density of the fluid to be 1 it is given in d\in\{2,3\} space dimensions by the equations (1.1) \begin{array}[]{c}\begin{array}[]{rcccl}&&\divergence{\bf{u}}&=&0,\\[6.45831pt] {\bf{u}}_{t}&+&({\bf{u}}\cdot\gradient){\bf{u}}+\gradient p&=&\nu\Delta{\bf{u}}\end{array}\quad\text{ in }\Omega\times(0,T).\end{array} Here \Omega\subset{\mathbb{R}}^{d} is an open set and T>0 some end time. By \nu>0 we denote the viscosity parameter. The unknowns are the velocity {\bf{u}}={\bf{u}}({\bf{x}},t)\in{\mathbb{R}}^{d} and the pressure (perturbation) p=p({\bf{x}},t)\in{\mathbb{R}} . Classical solutions of ( 1.1 ) dissipate the kinetic energy. The divergence constraint \eqref{NS}_{1} leads to a strong non-locality for solutions of the Navier–Stokes system such that the pressure is governed by an elliptic differential equation. A classical approach to avoid this non-locality is the artificial-compressibility approximation, see [ 5 , 4 , 6 ] and [ 27 , 28 ] . In this case, the Navier–Stokes system is approximated by the set of evolution equations (1.2) \begin{array}[]{c}\begin{array}[]{rcccl}\alpha p_{t}^{\alpha}&+&\displaystyle\divergence{\bf{u}}^{\alpha}&=&0,\\[8.61108pt] {\bf{u}}_{t}^{\alpha}&+&\displaystyle({\bf{u}}^{\alpha}\cdot\gradient){\bf{u}}^{\alpha}+\frac{1}{2}(\gradient\cdot{\bf{u}}^{\alpha}){\bf{u}}^{\alpha}+\gradient p^{\alpha}&=&\nu\Delta{\bf{u}}^{\alpha}\end{array}\quad\text{ in }\Omega\times(0,T).\end{array} Adding a time derivative for the pressure and scaling it with a small parameter \alpha>0 , the pressure is governed now by an evolution equation. For \alpha\to 0 solutions {\bf{u}}^{\alpha}({\bf{x}},t) and p^{\alpha}=p^{\alpha}({\bf{x}},t) of the system ( 1.2 ) recover the solutions {\bf{u}} and p of the original Navier–Stokes system ( 1.1 ). The additional correction term \frac{1}{2}(\gradient\cdot{\bf{u}}^{\alpha}){\bf{u}}^{\alpha} in ( 1.2 ) 2 ensures that classical solutions dissipate the sum of kinetic energy and the potential contribution \alpha(p^{\alpha})^{2}/2 . Most important, the structure of the artificial compressibility approximation allows to use methods from the field of hyperbolic evolution equations to understand the first-order sub-system (obtained by setting \nu to be zero) in ( 1.2 ). This has in particular consequences for the numerical simulation. One can employ numerical methods for systems of hyperbolic evolution laws to discretize this sub-system of ( 1.2 ). In such a way also convection-dominated flow regimes can be handled on the numerical level, see e.g. [ 15 , 23 , 22 ] . In this note, we focus on the dynamics of incompressible two-phase flows and aim at developing a framework that transfers the artificial-compressibility approach into this realm. We consider as a model problem a simplified version of the Navier–Stokes–Cahn–Hilliard (NSCH) equations as originally proposed in [ 17 ] . The NSCH model couples the Navier–Stokes equations with the fourth-order convective Cahn–Hilliard equation for the phase dynamics. Thus, it belongs to the class of diffuse-interface (or phase-field) approaches, see [ 2 ] . The major challenge to approximate the NSCH system using a first-order system is therefore the treatment of the coupled convective Cahn–Hilliard equation. We refer to [ 8 , 11 ] for approximation techniques for the pure Cahn–Hilliard evolution. In this work, we rely on a friction-type approximation that mimics the structure of Euler-Korteweg systems which have been considered as approximations of the Cahn–Hilliard equations in, e.g., [ 21 , 25 ] . Roughly speaking, the fourth-order operator in the Cahn–Hilliard equation is substituted by a third-order Korteweg term. The last step to obtain a first-order evolution system (always ignoring the viscous part of the stress tensor which is not touched) is the use of the relaxation technique from [ 24 ] . We present our new approximative system with a first-order sub-system in ( 4.1 ). The remainder of the paper is structured as follows: In Section 2 , we give a short outline of the NSCH system and its thermodynamical structure. In Section 3 , we propose the friction-type approximation of the NSCH system and we show that it is a thermodynamically consistent system of evolution equations. In Section 3.1 , we introduce our new approximative system and show that it obeys a natural energy dissipation rate (see Theorem 4.2 ). Moreover, we show under fair conditions in one spatial dimension that it involves a hyperbolic first-order sub-system similar to the one in the artifical-compressibility approximation ( 1.2 ). For the one-dimensional sub-system, we present a numerical characteristic analysis to understand the qualitative behaviour of the eigenvalues of the flux Jacobian. In Section 5 , we present extensive numerical results for the one-dimensional version, i.e., the system ( 4.7 ) below. 2. The Navier-Stokes-Cahn-Hilliard system In this section, we review a simplified version of the Navier–Stokes–Cahn–Hilliard (NSCH) system as originally introduced in [ 17 ] . In view of the approximations which we will introduce in the sequel, we recall the energy dissipation equation for solutions of the initial boundary problem for the NSCH system. Finally, we mention a result on the existence of strong solutions in two space dimensions. The domain of the flow is the bounded open set \Omega\subset\mathbb{R}^{d} , d\in\{2,3\} , with boundary \partial\Omega . The outer normal of \partial\Omega will be denoted by {\bf{n}}\in{\mathcal{S}}^{d-1} . For some end time T>0 , we define the space-time domain \Omega_{T}=\Omega\times(0,T) . 2.1. Formulation of the NSCH system Let us consider the dynamics of two incompressible and immiscible fluids at constant temperature. There are two approaches to model such two-phase flows: the sharp-interface approach and the diffuse-interface approach. In the sharp-interface ansatz, one obtains a free-boundary value problem which separates the two bulk domains by a well-defined phase boundary as codimension- 1 manifold. We focus here on the diffuse-interface (or phase-field) approach, where the phase boundary is smeared out with an interfacial width \gamma>0 , see Figure 2.1 . Figure 2.1 . Sketch of a diffuse-interface configuration with phase boundary of width \gamma . If the parameter \gamma tends to zero, one expects that an appropriate sharp-interface model is attained in the limit. Here, we are not interested in the sharp-interface limit and therefore fix the parameter \gamma throughout the entire note. As a meanwhile well-known diffuse-interface approach, we start from the classical Navier–Stokes–Cahn–Hilliard (NSCH) system, which can be traced back to [ 17 ] . It is a combination of the incompressible Navier–Stokes equations for the hydromechanical dynamics and the Cahn–Hilliard equation for the two-phase dynamics. The governing partial differential equations are given by (2.1) \begin{array}[]{c}\begin{array}[]{rcccl}&&\divergence{\bf{u}}&=&0,\\[6.45831pt] {\bf{u}}_{t}&+&({\bf{u}}\cdot\gradient){\bf{u}}+\gradient p&=&-c\gradient\mu(c)+\nu\Delta{\bf{u}},\\[6.45831pt] c_{t}&+&\divergence(c\vecu)-\divergence(\grad\mu(c))&=&0,\\[6.45831pt] &&\mu(c)&=&W^{\prime}(c)-\gamma\Delta c\end{array}\quad\text{ in }\Omega_{T}.\end{array} The hydromechanical unknowns

Read the complete text