Low-Mach-number limit for multiphase flows

Cassandre Lebot · 2025

Read Low-Mach-number limit for multiphase flows by Cassandre Lebot (2025) 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-2512-16286.txt.

Introduction

Cassandre Lebot reviews a limit in fluid theory in their arXiv paper, posted in 2025. When flow speeds are far below the speed of sound, the Mach number is tiny, and compressible flow equations should simplify to nearly incompressible ones. Taking that low-Mach limit cleanly is a classic mathematical problem.

Lebot first surveys what is known for single-phase flows governed by the compressible Navier-Stokes or Euler equations. The focus then shifts to two-phase flows, mixtures like gas bubbles in liquid. Several model families are compared: pressure given by an algebraic rule or by its own equation, one shared velocity or two separate ones, and versions with or without entropy. The paper is a formal map of which limits have been justified and how the choices interact.

Catalog record

CREATORCassandre Lebot
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE68 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2512-16286

Browse the full library · About the Tower of Babel

Full text

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

Low-Mach-number limit for multiphase flows Cassandre Lebot Abstract This paper is devoted to the study of the low-Mach-number limit for solutions of the compressible Navier-Stokes or Euler equations for different types of fluids. We first review the different results obtained in the case of flows consisting of one phase. Then, we focus on the low-Mach-number limit for two-phase flows, considering different types of systems: with an algebraic closure or a PDE closure for the pressure, with one single or two different velocities, without or with entropy. 1 Introduction The aim of this paper is to introduce the concept of low-Mach-number limit for compressible multiphase flows. The Mach number is a dimensionless quantity that represents the ratio of the flow velocity to the speed of sound in the fluid. A low Mach number is a criterion for incompressible flow. In the past, several authors have worked on the passage from compressible equations to an incompressible system when the Mach number tends to zero. The motivations are several. First, we better understand incompressible equations and we may use the passage to the limit and comparison to understand more properties on compressible equations. Moreover, low-Mach-number equations have several advantages in numerical analysis: simplification of the algorithms, smaller time step for the numerical integration. For more details on the numerical challenges for low-Mach-number systems, we refer the reader to [ 4 ] . However, this question of the zero-Mach-number limit for multiphase flows is still to be examined. To the best of our knowledge, [ 43 ] and [ 36 ] are two of the very few articles that studied this question. The present contribution aims at giving an overview of this subject by discussing the formal derivation of the zero-Mach-number limit for several types of multiphase flow models. This paper is organized as follows. In Section 2 , we give a brief review of the low-Mach-number limit for one-phase flow. In Section 3 , we focus on two-phase flows system with the same velocity for both phases and derive the low-Mach-number equations for two-phase systems considering two types of pressure closure: an algebraic closure and a PDE closure. We discuss first the isentropic case and then the non-isentropic case. In Section 4 , we derive the low-Mach-number equations for two-phase flows system with two distinct velocities. Finally, Section 5 contains some open problems and remarks about topics that we have not considered in this paper. 2 Low-Mach number limit for one-phase flow We give here a very short review of some results on the low Mach number limit for compressible flows. We refer the interested reader for more detailed reviews to [ 40 ] , [ 22 ] or [ 3 ] . 2.1 The isentropic case In this part, we restrict ourselves to the case of isentropic fluids, caracterized by a constant entropy, and we assume that the flow evolves in the domain \Omega (that we will precise later) from the d -dimensional space {\mathbb{R}}^{d} , with d\geqslant 2 . We write t\in{\mathbb{R}}^{+} for the time variable and x\in\Omega denotes the space variable. We consider the following governing equations of a compressible flow: \left\{\begin{aligned} &\partial_{t}\rho+\operatorname{div}(\rho u)=0,\\ &\partial_{t}(\rho u)+\operatorname{div}(\rho u\otimes u)+\nabla p(\rho)-\mu\Delta u-(\lambda+\mu)\nabla\operatorname{div}u=0\end{aligned}\right. (2.1) with initial conditions \displaystyle\rho_{\lvert t=0}=\rho_{0}\geqslant 0\quad\text{and}\quad u_{\lvert t=0}=u_{0} (2.2) and where \rho=\rho(t,x)\geqslant 0 represents the mass density, u=u(t,x) the velocity vector, p=p(\rho(t,x)) the pressure. The viscosity coefficients \lambda and \mu satisfy \mu>0 and \lambda+2\mu>0 and are assumed to be constant (this can be relaxed). We consider a barotropic fluid with a \gamma pressure law: p(\rho)=\rho^{\gamma}\text{, with }\gamma>1. There are several ways on how to introduce the Mach number in the compressible flow equations. First, we can consider a varying equations of state and also varying transport coefficients, like viscosity coefficients for viscous fluids (see [ 18 ] , [ 39 ] ). In that approach, a particular fluid is viewed as “nearly incompressible”. The alternative way to derive the compressible Navier-Stokes equations in the small Mach number regime is to consider flows on varying time or spatial scale. This amounts to use a dimensional analysis, by rewriting the equations in their dimensionless form. One way to do it is by scaling each variable by its characteristic value (see [ 30 ] ). The other way is to use a change of independent variables (see [ 2 ] , [ 31 ] ). This is the approach we take in the following development (for more details, see the introduction in [ 32 ] ). We denote by \varepsilon the Mach number. The incompressible limit is the limit where the speed of the fluid becomes negligible compared to the speed of sound. Thus we reset the flow speed to the scale by replacing u with \varepsilon u^{\varepsilon} . A particle will move a distance of order 1/\varepsilon during a time of order 1/\varepsilon^{2} , meaning that the relevant time scale is 1/\varepsilon . This is why we can use a change of variable like t^{\varepsilon}=\varepsilon^{2}t and x^{\varepsilon}=\varepsilon x , with t^{\varepsilon} and x^{\varepsilon} of order 1 (see [ 15 ] ). Thus, we scale u and \rho in the following way \displaystyle u(t,x)=\varepsilon u^{\varepsilon}(t^{\varepsilon},x^{\varepsilon}),\quad\rho(t,x)=\rho^{\varepsilon}(t^{\varepsilon},x^{\varepsilon}). (2.3) We can rewrite System ( 2.1 ): \left\{\begin{aligned} &\partial_{t}\rho^{\varepsilon}+\operatorname{div}(\rho^{\varepsilon}u^{\varepsilon})=0,\\ &\partial_{t}(\rho^{\varepsilon}u^{\varepsilon})+\operatorname{div}(\rho^{\varepsilon}u^{\varepsilon}\otimes u^{\varepsilon})+\frac{1}{\varepsilon^{2}}\nabla p(\rho^{\varepsilon})-\mu^{\varepsilon}\Delta u^{\varepsilon}-(\lambda^{\varepsilon}+\mu^{\varepsilon})\nabla\operatorname{div}u^{\varepsilon}=0.\end{aligned}\right. (2.4) The initial data are \rho^{\varepsilon}_{\lvert t=0}=\rho^{\varepsilon}_{0} and u^{\varepsilon}_{\lvert t=0}=u^{\varepsilon}_{0} . We will see that it can be relevant to use initial data that depend on \varepsilon . Formally, when we take the limit \varepsilon\to 0 in System ( 2.4 ), we cancel the terms with the highest factor in \varepsilon in the equations. Therefore from the momentum equation we obtain when \varepsilon goes to zero \displaystyle\nabla p(\rho^{\varepsilon})\to 0. (2.5) Let’s call C a constant in space such that the limit density \rho^{0} satisfies, because of ( 2.5 ): \displaystyle p(\rho^{0})=C(t), \displaystyle\rho^{0}=C(t)^{1/\gamma}. From the mass equation we get \displaystyle\frac{1}{\gamma}\frac{C^{\prime}(t)}{C(t)}+\operatorname{div}u^{0}=0. We now integrate in space over \Omega : \displaystyle\int_{\Omega}\frac{1}{\gamma}\frac{C^{\prime}(t)}{C(t)}=0. In the bounded or in the periodic case, we deduce that C^{\prime}(t)=0 and thus that \rho^{\varepsilon} must converge to a positive constant. In the case of the whole space, this depends on the condition that we use on \rho^{0} at infinity. Now passing to the limit in the mass equation we obtain \displaystyle\operatorname{div}u^{\varepsilon}\to 0. (2.6) Finally, coming back to the momentum equation, we find that the limit velocity u^{0} must satisfy the incompressible Navier-Stokes equations \left\{\begin{aligned} &\operatorname{div}u^{0}=0,\\ &\partial_{t}(\rho^{0}u^{0})+\operatorname{div}(\rho^{0}u^{0}\otimes u^{0})-\mu\Delta u^{0}+\nabla\pi=0.\end{aligned}\right. with \pi being the “limit” of \frac{(\rho^{\varepsilon})^{\gamma}-\bar{\rho}^{\gamma}}{\varepsilon^{2}} . Remark 2.1 . We can also rewrite the momentum equation as \rho^{0}\Big(\partial_{t}u^{0}+(u^{0}\cdot\nabla)u^{0}\Big)-\mu\Delta u^{0}+\nabla\pi=0. We end up with a compatibility condition on the initial data: it is more or less difficult to pass to the limit in System ( 2.4 ) if the initial data verifies itself the conditions ( 2.5 )-( 2.6 ) or not. Clearly, it is easier to pass to the limit if it does. Hence, we introduce the notion of “well–prepared” initial data. Definition 2.2 . We say that the initial data (u_{0}^{\varepsilon},\rho_{0}^{\varepsilon}) are well-prepared if they satisfy \displaystyle\operatorname{div}u_{0}^{\varepsilon}=\mathcal{O}(\varepsilon),\quad\rho_{0}^{\varepsilon}=\bar{\rho_{0}} with \bar{\rho_{0}} a constant. Otherwise, we say that the initial data are “ill-prepared” or we call them “general” initial data. In that case, it actually means that sounds waves are generated initially and propagate with high speed in the domain (speed of order 1/\varepsilon ).

Read the complete text