Symmetric structure-preserving discretization of N-phase incompressible fluid mixtures with arbitrary density ratios

This paper proposes a symmetric fully-discrete numerical method for incompressible N-phase Navier-Stokes-Cahn-Hilliary mixture models with arbitrary density ratios that rigorously preserves key physical properties, including exact mass and volume conservation, energy dissipation, and the phase saturation constraint.

The Gist

Imagine you are watching a pot of soup where oil, water, and vinegar are swirling together. In the real world, these liquids don't mix perfectly; they form distinct layers or droplets, and they push and pull against each other based on how heavy they are and how "sticky" they are. Simulating this on a computer is incredibly difficult, especially when you have more than two ingredients (like adding a third liquid) and when those ingredients have very different weights (like mixing heavy honey with light air).

This paper presents a new "recipe" for a computer program that simulates these complex fluid mixtures. Here is the breakdown of what the authors did, using simple analogies:

The Problem: The "Broken Scale"

When scientists try to simulate these fluids, they often run into a problem called "drift." Imagine a scale that is supposed to stay perfectly balanced. Over time, due to tiny computer rounding errors, the scale might slowly tip, making it look like mass is disappearing or appearing out of nowhere.

In complex mixtures with different densities, this is even worse. If the computer treats one liquid as the "main" character and the others as "sidekicks," the simulation can become biased. It might accidentally favor one liquid over another, breaking the symmetry of the real world. The authors wanted a method that treats every liquid exactly the same, like a democracy where every phase has an equal vote, ensuring that the total amount of "stuff" (mass and volume) never magically changes.

The Solution: A "Symmetric, Energy-Honest" Method

The authors created a new mathematical framework (a set of rules for the computer) that acts like a perfectly balanced ledger.

1. The "Equal Footing" Rule:
   Most old methods pick one liquid to be the "reference" (like picking a captain for a team). This paper's method treats all $N$ liquids as equal partners. It doesn't matter if you have 3 liquids or 10; the math treats them all symmetrically. This prevents the computer from accidentally favoring one liquid over another.

2. The "No-Drift" Guarantee:
   The authors proved that their method guarantees three things will never change, no matter how long the simulation runs:

   • Total Volume: The soup never expands or shrinks.
   • Total Mass: No liquid vanishes or appears from thin air.
   • Individual Mass: The amount of oil, water, and vinegar stays exactly the same (they can move around, but the total amount of each is locked in).

3. The "Energy Bank" Metaphor:
   Think of the fluid system as a bank account. The "energy" in the system is the money. In the real world, friction and mixing always cost money (energy is lost to heat). The authors' method ensures the computer simulation behaves like a strict bank: the energy balance sheet always goes down or stays the same; it never accidentally goes up. This is called "energy dissipation," and it keeps the simulation stable and realistic.

How They Did It

To achieve this, the authors had to rewrite the equations the computer uses.

• The "Saturation" Constraint: They ensured that at every single point in the simulation, the liquids fill up 100% of the space (no empty voids). If the liquids start filling the space perfectly, the math guarantees they will keep filling it perfectly forever, step-by-step.
• The "Arbitrary Density" Feature: Previous methods struggled when liquids had very different weights (e.g., a heavy metal liquid vs. a light gas). This new method works perfectly even when the density ratios are extreme.

The Proof: Running the Tests

The authors didn't just write the math; they tested it with three scenarios:

1. Convergence Test: They checked if the math gets more accurate as they made the computer's "grid" finer. It did, just as predicted.
2. Phase Separation: They simulated a messy mixture separating into distinct blobs. The computer correctly showed the blobs forming and the energy slowly decreasing, without any "ghost" mass appearing.
3. Rising Bubbles: They simulated a bubble rising through liquids. They compared their results to known benchmarks and found their method matched the physics perfectly, preserving the volume of the bubble exactly. They even simulated a bubble rising through two different liquid layers, showing it could handle complex, multi-layered interactions.

The Bottom Line

This paper provides a robust, "symmetric" tool for simulating complex fluid mixtures. It ensures that the computer simulation respects the fundamental laws of physics (conservation of mass and energy) at every single step, even when dealing with many different liquids that have very different weights. It's like upgrading from a leaky bucket to a sealed, perfectly balanced container for your fluid simulations.

Technical Summary

Technical Summary: Symmetric Structure-Preserving Discretization of N-Phase Incompressible Fluid Mixtures

Problem Statement
Diffuse-interface models, specifically the Navier–Stokes–Cahn–Hilliard (NSCH) family, are widely used to simulate interfacial dynamics in complex fluids by representing interfaces as smooth transition layers. While robust numerical methods exist for binary (two-phase) flows, the extension to $N$-phase incompressible mixtures with arbitrary density ratios presents significant challenges. Existing structure-preserving methods often rely on binary formulations or distinguish a reference phase, thereby breaking the inherent symmetry of the physical system.

The core difficulty in the $N$-phase setting lies in simultaneously satisfying several competing requirements at the discrete level:

1. Symmetry: The numerical scheme must treat all phases on equal footing without eliminating variables via the saturation constraint or introducing a privileged reference phase.
2. Conservation: The method must preserve phase-wise volume and mass, as well as total volume and total mass.
3. Saturation Constraint: The sum of volume fractions must equal unity ($\sum \phi_\alpha = 1$) at every point and time step to prevent non-physical voids or overlaps.
4. Energy Dissipation: The discrete solution must satisfy a discrete energy-dissipation law consistent with the continuum thermodynamics.
5. Arbitrary Density Ratios: The method must remain stable and accurate even when phases have significantly different densities, which introduces strong coupling between the momentum and mass balance equations.

Methodology
The authors propose a fully-discrete finite element method for the $N$-phase incompressible NSCH mixture model. The approach is built upon a unified mixture-theory framework where the system is governed by a single mixture momentum equation and $N$ phase mass balance equations.

• Equivalent Reformulation: To facilitate structure preservation, the authors first derive an equivalent strong formulation of the continuum equations. This reformulation utilizes specific test functions (weights) to derive the energy evolution. Crucially, they modify the momentum equation to eliminate nonlinear kinetic terms from the mass balance weights, ensuring that all test functions belong to standard Sobolev spaces suitable for finite element discretization.
• Symmetric Treatment: The scheme is formulated directly in terms of the full set of phase variables ($\phi_\alpha$) and a single mixture velocity ($v$), without eliminating any phase or introducing a reference phase. This maintains the permutation symmetry of the phases.
• Discretization Strategy:
  • Time Discretization: A fully-discrete scheme is constructed using a time-stepping approach. The gradient-free part of the free energy is treated using time-averages (inspired by [37]) to ensure convexity and stability. Regularizations are introduced for the density and mobility matrix to handle cases where volume fractions might temporarily leave the physical interval $[0, 1]$.
  • Spatial Discretization: The method employs a finite element approximation using $H^1$-conforming spaces. The velocity space is chosen to be inf-sup stable with the pressure space.
  • Key Structural Features: The scheme utilizes a skew-symmetric form for the convection term and specific discrete weights for the mass and momentum equations. This allows for the derivation of exact discrete conservation laws and a discrete energy-dissipation inequality.

Key Contributions

1. Symmetric $N$-Phase Discretization: The paper presents the first fully-discrete method for $N$-phase incompressible NSCH models that is genuinely symmetric with respect to all phases, avoiding the asymmetry introduced by reference-phase formulations.
2. Exact Structural Preservation: The authors prove that every discrete solution satisfies:
   • Exact phase volume conservation.
   • Exact phase mass conservation.
   • Exact total volume and total mass conservation.
   • A discrete energy-dissipation law.
   • Preservation of the volume-saturation constraint ($\sum \phi_\alpha = 1$) pointwise, provided it holds for the initial data.
3. Arbitrary Density Ratios: The method is robust for mixtures with arbitrary density contrasts, a regime where many existing schemes struggle due to the coupling of density and mass balances.
4. Mixture-Aware Consistency: The formulation respects the "mixture-aware" properties of the continuum model, ensuring that the system reduces correctly when phases are merged or when a phase is absent (reduction on simplex faces).

Numerical Results
The authors validate the theoretical properties through several numerical experiments:

• Convergence Test ($N=3$): A test with three phases and arbitrary densities demonstrates that the scheme achieves the expected order of convergence for the chosen finite element spaces, with errors decreasing as the mesh is refined.
• Phase Separation ($N=3$): A simulation of ternary phase separation shows the formation of interconnected structures and triple junctions. The results confirm that the total energy is non-increasing and that mass/volume conservation errors are at the level of machine precision.
• Rising Bubble ($N=2$): The method is tested against the standard Hysing et al. benchmark for a rising bubble. The results for bubble center of mass and rise velocity agree well with reference data from sharp-interface codes and other diffuse-interface formulations. The scheme maintains volume conservation exactly.
• Three-Fluid Rising Bubble ($N=3$): A complex scenario involving a gas bubble rising through two stratified liquid layers is simulated. The method correctly captures the physical behavior of the bubble either remaining trapped at the interface or penetrating the upper layer, depending on the bubble radius relative to the critical capillary radius.

Significance and Claims
The paper claims to provide a robust computational framework for incompressible $N$-phase mixture flows with complex interfacial dynamics and arbitrary density contrasts. The primary significance lies in the ability to retain the defining thermodynamic and conservation structures of the continuum equations at the fully-discrete level without sacrificing symmetry.

The authors emphasize that previous methods often broke symmetry or failed to preserve the saturation constraint exactly in the presence of arbitrary density ratios. By constructing a scheme that treats all phases symmetrically and preserves the saturation constraint pointwise, this work addresses a critical gap in the numerical simulation of multiphase flows. The resulting method is shown to be stable and accurate for representative multiphase flow problems, offering a reliable tool for simulating systems where long-time integration and strong coupling are required.

The paper concludes by noting that while the current work establishes the foundation for symmetric structure-preserving discretization, future work could extend these ideas to higher-order temporal discretizations, more general mixture-aware closures, and compressible $N$-phase flows.