A variational front-tracking method for multiphase flow with triple junctions

This paper proposes a robust, unconditionally stable, and volume-preserving variational front-tracking finite element method for simulating multiphase flows with complex topological features like triple junctions and boundary contacts.

The Gist

Imagine you are trying to simulate a complex dance involving several different types of liquids—like oil, water, and honey—all swirling together in a container.

The challenge is that these liquids don't just mix; they push against each other, forming distinct boundaries. Where two liquids meet, you have an interface. But where three liquids meet, you get a triple junction—a tiny, high-tension point where three different "walls" collide.

Simulating this is a mathematical nightmare. If your simulation isn't perfect, the liquids might "leak" into each other, the bubbles might shrink for no reason, or the math might simply "explode" (become unstable).

This paper presents a new, smarter way to track these moving boundaries. Here is the breakdown of how they did it.

1. The "Two-Map" Strategy (The Eulerian vs. Parametric approach)

Most simulations try to use one single "map" (a grid) to track everything. But the researchers here use two different types of maps working together:

• The Bulk Map (Eulerian): Imagine a fixed grid of streets in a city. We use this to track the "flow" of the liquids (the velocity and pressure) through the streets. The streets don't move, but the cars (the liquid) do.
• The Boundary Map (Parametric): Instead of just looking at the streets, we use a separate, flexible "rubber band" to trace the exact edges where the liquids meet. This rubber band can stretch, bend, and move independently of the street grid.

By combining these, they can track the heavy, swirling movement of the liquid and the delicate, thin edges of the bubbles at the same time.

2. The "Smart Rubber Band" (Handling Triple Junctions)

The hardest part is the triple junction (where three liquids meet). In older methods, these points often get "stuck" or cause the math to break.

The researchers used a clever trick called the BGN approach. Imagine the boundary is a rubber band. Usually, if you pull a rubber band, it only moves in the direction you pull. But the researchers gave the points on the rubber band "tangential freedom." This means the points can slide along the edge of the bubble without changing the bubble's shape.

This "sliding" allows the triple junctions to move smoothly and naturally, preventing the "rubber bands" from getting tangled or distorted. It’s like allowing the dancers in our liquid dance to shuffle their feet to stay in formation without breaking the dance pattern.

3. The "No-Leak" Guarantee (Volume Preservation)

In many simulations, if you run it long enough, a bubble of air might slowly "evaporate" simply because of rounding errors in the math.

The authors created a "Structure-Preserving" version of their method. They added a mathematical "policeman" that checks the volume of every liquid phase at every step. If a bubble starts to shrink due to a calculation error, the math automatically adjusts the boundary to ensure the volume stays exactly the same. It’s like having a digital scale that constantly corrects the weight of your ingredients to ensure your recipe stays perfect.

4. Why does this matter? (The Results)

The researchers tested their method on everything from rising bubbles to complex "triple bubbles" (three bubbles joined together).

• It’s Stable: It doesn't crash or produce "spurious velocities" (fake, jittery movements that aren't actually happening in real life).
• It’s Accurate: It can simulate how a heavy bubble can actually "drag" a lighter bubble upward, just like in real physics.
• It’s Robust: It works in both 2D (flat surfaces) and 3D (real-world volumes), making it useful for real-world engineering, like designing better ink-jet printers or understanding how oil moves through underground reservoirs.

Summary Metaphor

If simulating multiphase flow is like trying to film a high-speed car chase through a crowded city, this paper provides:

1. A high-speed camera that tracks the cars (the liquid).
2. A flexible GPS tracker that follows the exact path of the cars (the interfaces).
3. A specialized sensor that ensures the cars don't magically disappear or grow larger during the chase (volume preservation).

Technical Summary

Technical Summary: A Variational Front-Tracking Method for Multiphase Flow with Triple Junctions

1. Problem Statement

The paper addresses the numerical simulation of multiphase flows (defined here as flows with three or more distinct phases) governed by the incompressible Navier–Stokes equations. The primary mathematical and computational challenge lies in the presence of triple junctions—points (in 2D) or lines (in 3D) where three distinct fluid interfaces meet—and the evolution of these junctions alongside moving contact lines at fixed boundaries.

Existing numerical methods (such as level-set, diffuse-interface, or volume-of-fluid) often struggle with the accurate representation of sharp interfaces, the prevention of spurious velocities, the maintenance of energy stability, and the exact conservation of mass (volume) for each phase. Specifically, for front-tracking methods, representing the topological complexity of triple junctions while ensuring the force balance condition (Young-Laplace law at the junction) is a significant hurdle.

2. Methodology

The authors propose a variational front-tracking framework that unifies the treatment of 2D and 3D multiphase flows. The methodology is built on several key pillars:

• Hybrid Formulation: The model combines an Eulerian formulation for the bulk fluid equations (Navier–Stokes) and a parametric formulation for the evolving interfaces (curve networks in 2D, surface clusters in 3D).
• BGN Framework: They leverage the "Barrett-Garcke-Nürnberg" (BGN) approach, which introduces tangential velocity freedom in the interface parameterization. This allows the triple junctions to move naturally and ensures that the mesh quality of the interfaces remains high (asymptotically equidistributed) without the need for manual remeshing.
• Unfitted Finite Element Method (FEM): The bulk equations are solved using an unfitted mesh approach, meaning the interface mesh is independent of the bulk mesh. To improve accuracy and volume preservation, they employ an XFEM (Extended Finite Element Method) enrichment.
• Variational Weak Formulation: The authors derive a weak formulation that incorporates:
  • Continuity of velocity across interfaces.
  • Balance of normal stresses and capillary forces (surface tension).
  • Kinematic conditions (interfaces move with the fluid).
  • Force balance at triple junctions (the weighted sum of conormals vanishes).
• Structure-Preserving Variants:
  • Energy Stability: A linear fully discrete scheme is developed that is unconditionally stable with respect to the discrete energy.
  • Volume Preservation: A nonlinear, structure-preserving variant is introduced using time-weighted discrete surface normals. This ensures that the enclosed volume of each phase is preserved exactly at the discrete level.

3. Key Contributions

• Unified Multiphase Framework: A robust mathematical framework capable of handling $N$ phases and complex junction topologies in both 2D and 3D.
• Mathematical Rigor: The paper provides a proof of existence and uniqueness for the linear discrete scheme and establishes an unconditional energy stability result.
• Triple Junction Treatment: A novel discrete treatment of curvatures at triple junctions, motivated by the continuous force-balance condition, which ensures the well-posedness of the numerical system.
• Exact Mass Conservation: The introduction of a scheme that achieves exact volume preservation for each phase, a critical requirement for long-term physical accuracy in multiphase simulations.

4. Results and Numerical Validation

The authors validate their methods through extensive 2D and 3D simulations:

• 2D Results:
  • Steady-state bubbles: Demonstrated that the XFEM-enriched scheme captures pressure jumps exactly and eliminates spurious velocities in standard double and triple bubble configurations.
  • Dynamic flows: Successfully simulated the movement of triple junctions and the rise of bubbles trapped between different fluids.
  • Volume Conservation: Compared the linear scheme against the structure-preserving scheme, showing that the latter maintains constant bubble volumes even in complex rising-bubble scenarios.
• 3D Results:
  • Simulated the rise of a trapped lens and a triple bubble in 3D.
  • Confirmed that the method accurately captures the deformation of interfaces and the rise velocity of bubbles, maintaining high mesh quality and volume conservation in three dimensions.

5. Significance

This work represents a significant advancement in the computational fluid dynamics (CFD) of multiphase systems. By providing a method that is simultaneously unconditionally stable, volume-preserving, and mesh-robust, the authors offer a powerful tool for industrial and scientific applications (e.g., microfluidics, oil extraction, and ink-jet printing) where the precise tracking of multiple interfaces and their junctions is essential for physical fidelity.