Sharp-interface Simulations of Energetic Multiphase Flows with Large Density and Viscosity Ratios

This paper proposes a Synchronized Donor-Region of Momentum fluxes (SynDRoM) framework combined with a viscosity limiter to enhance the numerical robustness and physical fidelity of sharp-interface simulations for energetic multiphase flows with large density and viscosity ratios.

The Gist

The Big Picture: Simulating a Stormy Sea

Imagine you are trying to create a computer simulation of a violent ocean storm. You want to see how waves crash, how air gets sucked into the water, and how bubbles form. This is tricky because water is heavy and thick, while air is light and thin. In physics terms, they have a massive difference in "density."

When computers try to simulate this, they often crash or produce weird, impossible results (like water suddenly turning into a ghost or air shooting through water like a bullet). This paper introduces a new set of rules (algorithms) to make these simulations stable, accurate, and physically realistic, even when the waves are crashing violently.

The Problem: The "Ghost" and the "Shock"

The authors explain that old methods for simulating these flows have two main flaws:

1. The "Ghost" Problem (Velocity Penetration):
   Imagine a heavy truck (water) and a feather (air) moving next to each other. In old simulations, the "wind" from the feather would sometimes blow the truck backward, or the truck would push the feather through its own body. This is called "velocity penetration." It creates fake, non-physical shapes in the water, like a "devil's horn" sticking out of the wave.
2. The "Shock" Problem (Momentum Spikes):
   To fix the ghost problem, scientists tried a new method called CMOM (Consistent Mass-Momentum). It's like keeping a strict ledger of how much "oomph" (momentum) every drop of water has. However, this method has a side effect. When a tiny bit of heavy water moves into a cell full of air, the math gets confused. It's like dividing a huge number by a tiny number, resulting in a massive, impossible spike in speed. This creates "velocity blobs"—fake pockets of air moving at supersonic speeds that shouldn't exist.

The Solution: The "SynDRoM" Method

The authors propose a new fix called SynDRoM (Synchronized Donor Region of Momentum flux). Here is how it works, using an analogy:

The Analogy: The Moving Conveyor Belt
Imagine a conveyor belt carrying boxes.

• The Old Way: You count the boxes (mass) and the weight of the boxes (momentum) separately. If a box moves, you might accidentally count its weight in a spot where the box hasn't actually arrived yet. This causes the "shock" or "spike" in speed.
• The SynDRoM Way: This method acts like a synchronized team. Before you move the weight, you look exactly at which part of the conveyor belt the weight is coming from.
  • It asks: "If I am moving this specific chunk of air, exactly which chunk of momentum is attached to it?"
  • It ensures that the momentum is only moved if the mass is actually there to carry it.
  • The Result: No more fake speed spikes. The air stays slow, and the water stays heavy, just like in real life. The simulation stays smooth and doesn't "blow up."

The Second Problem: The "Slippery" Viscosity

The paper also tackles a second issue: Viscosity (how thick or sticky a fluid is).

• The Problem: Water is sticky; air is slippery. When they mix at a sharp boundary (like a wave breaking), the computer tries to guess the "stickiness" in the middle. If it guesses wrong, the math becomes unstable, like trying to balance a pencil on its tip.
• The Fix: The authors introduce a Viscosity Limiter.
  • The Analogy: Imagine a speed limit sign. Even if the math tries to calculate a "stickiness" that would make the fluid move impossibly fast (unstable), the limiter says, "Nope, you can't go faster than the speed of the thinnest fluid here." It caps the calculation to keep the simulation from crashing, without changing the actual physics of the water or air.

The Proof: Does it Work?

The authors tested their new rules in three ways:

1. The Dam Break: They simulated a wall of water collapsing.
   • Old methods: The water looked distorted with fake spikes.
   • SynDRoM: The water crashed naturally, and the air didn't get sucked into the water in weird ways.
2. The Kelvin-Helmholtz Instability: This is when wind blows over water, creating rolling waves (like clouds).
   • Result: The simulation correctly showed the waves rolling up and growing, without the computer adding fake energy or damping the waves out. It proved the method respects the laws of physics.
3. The Breaking Wave: They simulated a massive, diagonal wave crashing.
   • Result: The wave broke, splashed, and created foam just like a real ocean. The total energy of the system stayed balanced (it didn't magically disappear or explode). Even when they added "stickiness" (viscosity), the simulation remained stable.

The Bottom Line

This paper presents a new "traffic cop" for computer simulations of water and air.

• It stops the air from ghosting through the water.
• It stops the water from creating impossible speed spikes.
• It keeps the "stickiness" calculations from breaking the math.

By synchronizing exactly what is moving with where it is moving, the authors have created a simulation tool that is much more robust and reliable for studying violent ocean events, like the kind naval engineers need to understand for ship design.

Technical Summary

Technical Summary: Sharp-interface Simulations of Energetic Multiphase Flows with Large Density and Viscosity Ratios

Problem Statement
Simulating energetic multiphase flows with extreme density ratios (e.g., air-water systems with ratios $\sim 10^3$ to $10^4$) remains a significant challenge in naval hydrodynamics. Traditional velocity-based formulations often suffer from numerical divergence due to the misalignment of pressure gradients and density discontinuities, leading to spurious momentum transfer and energy blow-up. While Consistent Mass-Momentum (CMOM) transport frameworks improve robustness by enforcing momentum conservation and semi-discrete energy conservation, they introduce new difficulties when combined with sharp-interface methods. Specifically, CMOM can generate severe momentum shocks, resulting in non-physical velocity oscillations and overshoots (violating the Total Variation Diminishing, or TVD, condition) when sharp interfaces cross grid boundaries. Additionally, improper estimation of viscosity near interfaces, particularly when dividing dynamic viscosity by density in staggered grids, can introduce numerical instabilities at finite time steps, undermining solver robustness.

Methodology
The paper proposes a suite of physics-preserving numerical schemes designed to reconcile physical fidelity with numerical robustness for high-density-ratio flows.

1. Synchronized Donor-Region of Momentum flux (SynDRoM):
   To address the velocity oscillations inherent in standard CMOM, the authors introduce SynDRoM. This method enforces the monotonicity of the transported velocity field by synchronizing the momentum flux with the mass flux.

   • Mechanism: Unlike existing interface-aware slope limiters that focus solely on velocity reconstruction, SynDRoM reconstructs a density-velocity distribution pair within a cell to satisfy cell-averaged constraints and TVD bounds. It then evolves the momentum using a "donor region" concept determined strictly by the mass flux. The advected velocity is calculated as a weighted average over this specific donor region, ensuring that the updated velocity remains bounded.
   • Implementation: The method operates within a directionally split framework consistent with conservative Volume-of-Fluid (cVOF) advection. It utilizes a density-weighted second-order Runge-Kutta (RK2) temporal integration scheme to handle the non-linear evolution of velocity relative to density and momentum.

2. Bounded Kinematic Viscosity Limiter:
   To address instabilities arising from viscosity jumps at interfaces, the paper introduces a viscosity limiter.

   • Mechanism: The authors identify that standard interpolation (arithmetic or harmonic) of dynamic viscosity at staggered grid vertices, followed by division by density, can yield unbounded effective kinematic viscosity, violating viscous CFL constraints. The proposed limiter constrains the effective dynamic viscosity used in cross-difference terms by the minimum density of the surrounding momentum cells and a phase-occupancy-based upper bound for kinematic viscosity. This ensures the effective kinematic viscosity remains within a stability-preserving range.

Key Contributions

• SynDRoM Algorithm: A generalized method for CMOM that eliminates spurious velocity oscillations without sacrificing momentum conservation or interface sharpness. It explicitly accounts for the momentum distribution and the corresponding mass-flux-based donor region, avoiding the need for full interface reconstruction on staggered grids.
• Viscosity Stabilization: A simple yet effective viscosity limiter that prevents numerical divergence in high-viscosity-ratio flows by bounding the effective kinematic viscosity, thereby preserving stability without compromising physical fidelity.
• Directionally Split Consistency: A formulation that ensures momentum updates remain consistent with directionally split mass advection, preventing fictitious fluxes that could violate velocity boundedness.

Results
The proposed methods were validated through a series of test cases:

• 2D Dam Break: The SynDRoM formulation successfully resolved interface deformation and eliminated spurious velocity structures (such as "devil-horn" distortions seen in velocity formulations and "velocity blobs" in original CMOM) while maintaining energy boundedness.
• 1D Droplet Advection: In a quasi-one-dimensional test with a high density ratio ($10^{-3}$), SynDRoM maintained velocity monotonicity and TVD properties, whereas original CMOM and Favre-averaged variants produced non-physical overshoots.
• Kelvin-Helmholtz Instability: The scheme correctly captured the linear growth of perturbations and the subsequent nonlinear evolution (vortex pairing, roll-up) without artificial damping or energy blow-up. Horizontal momentum was conserved to solver tolerance, and high-velocity regions were correctly confined to the light-fluid phase.
• Stratified Poiseuille Flow: The viscosity limiter successfully prevented numerical divergence observed in standard arithmetic averaging and corrected the excessive velocity gradients seen in harmonic averaging, yielding stable and physically accurate solutions.
• 3D Standing Wave Breaker: Simulations of diagonal standing wave breakers demonstrated the solver's capability to handle extreme nonlinearity, jet formation, and air entrainment. The method maintained bounded mechanical energy even in the absence of explicit viscosity, with energy dissipation occurring naturally through the iLES (implicit Large Eddy Simulation) mechanism as the flow became irregular. Grid convergence studies confirmed consistent trends in energy evolution and dissipation rates.

Significance
The paper claims that the proposed modifications to advection and viscous schemes significantly enhance the fidelity and robustness of multiphase flow simulations, particularly for energetic free-surface flows encountered in naval hydrodynamics. By ensuring that CMOM methods satisfy the TVD condition near emptying interface cells and by bounding kinematic viscosity, the approach removes numerical artifacts that could trigger false break-up events. The authors emphasize that the SynDRoM and viscosity limiter are orthogonal to the specific interface capturing method used (demonstrated here with geometric VOF), making the approach broadly applicable. The work provides a pathway to simulate complex, high-density-ratio flows with physical consistency, momentum conservation, and empirical energy boundedness, addressing long-standing challenges in the field.