Consistent Interface Capturing Adaptive Reconstruction Approach for Viscous Compressible Multicomponent Flows

This paper presents a physically consistent numerical framework for viscous compressible multicomponent flows that adaptively combines THINC, MP, and WENO reconstruction schemes with a central scheme for tangential velocities to sharply capture material interfaces while minimizing dissipation and oscillations.

The Gist

Imagine you are trying to film a high-speed chase scene where two different types of fluids (like oil and water, or hot air and cold air) are crashing into each other. In the world of computer simulations, this is called a multicomponent flow.

The challenge for computer scientists is that these fluids behave differently depending on what's happening:

1. Shocks: Like a sonic boom or a car crash, where everything gets squished and changes instantly.
2. Interfaces: Like the boundary between oil and water, where they touch but don't mix immediately.
3. Smooth Flow: Like water flowing gently in a river.

The problem with current computer methods is that they often use a "one-size-fits-all" camera lens. If you use a lens designed for capturing fast crashes (shocks) to film the gentle boundary between oil and water, the image gets blurry. The computer "smears" the line, making the oil and water look like they are mixing when they shouldn't be. This is called numerical dissipation.

This paper introduces a new, smarter camera system called HY-THINC-D. Here is how it works, using simple analogies:

1. The Smart Detective (The Sensor)

Before the computer tries to draw the picture, it needs to know what it is looking at.

• Old Way: Some methods look at the "volume" of the fluids to decide if there is a boundary. This is like trying to find a wall by counting how many bricks are in a pile. If you have a messy pile of 100 different types of bricks, it's hard to tell where one wall ends and another begins.
• New Way: The author created a "Smart Detective" (a sensor). Instead of counting bricks, this detective looks at the entropy (a measure of disorder) of the flow.
  • If the detective sees a sudden, sharp jump in disorder, it knows, "Aha! There is a boundary here!"
  • Crucially, this detective is smart enough to ignore the "noise" of high-frequency waves (like ripples on a pond) so it doesn't get confused and draw fake lines where there are none.

2. The Specialized Tools (Reconstruction Schemes)

Once the detective identifies the scene, the computer switches tools based on the physics:

• For the "Crash Zones" (Shocks):
  The computer uses a standard, robust tool (like a heavy-duty hammer) called MP5 or WENO. These are great at handling the violent, messy crashes without breaking the simulation.

• For the "Gentle Boundaries" (Material Interfaces):
  This is where the magic happens. When the computer sees a boundary between two fluids (like oil and water), it switches to a specialized, ultra-sharp tool called THINC.

  • The Analogy: Imagine drawing a line with a thick, fuzzy marker (the old method). It looks blurry. Now, imagine switching to a laser-guided pen (THINC). It draws a razor-sharp line in just a few pixels.
  • The Catch: You can't use the laser pen for everything. If you try to use it on a shockwave, the simulation might explode. So, the computer only uses the laser pen for the specific variables that actually change at the boundary (density and volume), leaving the other variables alone.

3. The "Slippery" Rule (Viscous Flows)

The paper also tackles a tricky physics rule about viscous flows (flows with friction, like honey or real air).

• The Physics: In a real, sticky fluid, if two layers slide past each other, the speed of the fluid along the boundary (tangential velocity) must be smooth and continuous. It can't suddenly jump.
• The Problem: Old computer methods often treated these speeds like they were jumping (discontinuous), which caused the simulation to wiggle and shake (oscillate).
• The Solution: The author uses a "Central Scheme" for these specific speeds.
  • The Analogy: Imagine two people sliding past each other on ice. If they are holding hands (viscous), their hands must move at the same speed. The computer now forces the simulation to respect this "hand-holding" rule, ensuring the flow stays smooth and doesn't develop fake, unphysical swirls.

Why This Matters

Think of this new method as a chameleon camera.

• When it sees a violent crash, it acts like a rugged action camera.
• When it sees a delicate boundary between fluids, it switches to a high-definition macro lens to keep the line razor-sharp.
• When it sees smooth, sticky flow, it acts like a smooth operator to prevent shaking.

The Result:
The paper proves this method works by testing it on difficult scenarios, like a shockwave hitting a bubble of helium inside a shell of gas, or complex mixing layers. The results show that this new method captures the boundaries much sharper than existing methods, without the annoying "fuzziness" or "shaking" that usually ruins these simulations. It's a more accurate, stable, and physically consistent way to simulate how different fluids interact at high speeds.

Technical Summary

1. Problem Statement

Simulating viscous compressible multicomponent flows presents significant numerical challenges, primarily due to the coexistence of different types of discontinuities:

• Shock Waves: Discontinuities in density, pressure, and normal velocity.
• Contact Discontinuities/Material Interfaces: Discontinuities in density and volume fractions, but with continuous pressure and normal velocity.
• Tangential Velocity Behavior: In inviscid flows, tangential velocities are discontinuous across contact interfaces. However, in viscous flows, tangential velocities must remain continuous across material interfaces due to the no-slip condition and viscous diffusion.

Key Issues with Existing Methods:

• Excessive Dissipation: Standard high-order schemes (like WENO or MP) often smear contact discontinuities and material interfaces over many grid cells due to numerical dissipation.
• Inconsistent Reconstruction: Applying interface-sharpening schemes (like THINC) to all variables across all discontinuities can lead to physical inconsistencies. For instance, applying THINC to pressure or tangential velocities across a contact interface (where they are continuous) can generate spurious oscillations or cause simulation crashes.
• Detection Limitations: Many existing sensors rely on volume fractions to detect interfaces, which becomes computationally expensive and unreliable in domains with many species. Furthermore, some sensors (like TENO-based ones) fail to distinguish between shocks and contact discontinuities, leading to the inappropriate application of sharpening schemes to continuous variables.

2. Methodology

The paper proposes a physically consistent, adaptive reconstruction approach that selects the reconstruction scheme based on the specific physics of the wave structure and the flow regime (viscous vs. inviscid).

A. Governing Equations

The study utilizes the quasi-conservative five-equation model (Allaire et al.) for two-fluid mixtures, including viscous terms. The system solves for:

• Two continuity equations (for partial densities $\alpha_1\rho_1, \alpha_2\rho_2$).
• One momentum equation.
• One energy equation.
• One advection equation for the volume fraction ($\alpha_1$).

B. Adaptive Reconstruction Strategy

The core innovation is a hybrid approach that switches between MP5/WENO (for shocks and smooth regions) and THINC (Tangent of Hyperbola for INterface Capturing) specifically for contact discontinuities.

1. Contact Discontinuity Detection (The Sensor):

• A new sensor is developed based on a variable $s = \sqrt{p/\rho^\gamma}$, which is loosely related to entropy.
• Why $s$? Unlike density, $s$ exhibits a sharp jump across material interfaces but remains smooth in high-frequency acoustic regions (unlike density-based sensors which can trigger falsely in smooth oscillating flows).
• The sensor uses smoothness indicators inspired by WENO but adapted for $s$. It effectively distinguishes contact discontinuities from shocks and high-frequency noise.

2. Variable-Specific Reconstruction:
The reconstruction strategy differs based on the variable type and flow physics:

• In Characteristic Space:
  • Acoustic Waves (Shocks): Reconstructed using MP5 or WENO.
  • Entropy/Contact Waves ($W_2, W_3$): Reconstructed using THINC to sharpen the interface.
  • Volume Fraction ($W_5$): Reconstructed using THINC.
  • Shear Waves ($W_4$, tangential velocity):
    • Inviscid: Reconstructed using MP5/WENO (discontinuous).
    • Viscous: Reconstructed using a Central Scheme (continuous).
• In Primitive Variable Space:
  • Phasic Densities & Volume Fractions: Reconstructed using THINC in contact regions.
  • Pressure & Tangential Velocities: Reconstructed using a Central Scheme across material interfaces in viscous flows to maintain continuity and prevent oscillations.

3. Viscous Flow Handling (The "HY-THINC-D" Algorithm):

• A specific algorithm is introduced for viscous flows where tangential velocities are continuous.
• The Ducros sensor (a shock detector that does not detect contact discontinuities) is employed.
• If the Ducros sensor indicates a shock ($\Omega_d > 0.01$), tangential velocities use the nonlinear MP5 scheme.
• If no shock is detected (i.e., across a material interface), tangential velocities are computed using a central scheme. This prevents the generation of unphysical vortices and oscillations at the interface.

C. Numerical Discretization

• Spatial: Finite Volume Method on a Cartesian grid.
• Temporal: Third-order Strong Stability-Preserving (SSP) Runge-Kutta.
• Fluxes: HLLC approximate Riemann solver for convective fluxes; $\alpha$-damping approach for viscous fluxes.

3. Key Contributions

1. Physically Consistent Adaptive Scheme: The paper establishes that different variables require different reconstruction strategies. Applying THINC to continuous variables (pressure, tangential velocity) across contact interfaces is physically inconsistent and leads to failure; the proposed method avoids this.
2. Novel Contact Discontinuity Sensor: A sensor based on the variable $s = \sqrt{p/\rho^\gamma}$ is developed. It robustly detects material interfaces without relying on volume fractions (making it suitable for multi-species flows) and avoids false triggering in high-frequency smooth regions.
3. Viscous Interface Treatment: The development of an algorithm that uses a central scheme for tangential velocities across material interfaces in viscous flows. This resolves the issue of spurious oscillations and unphysical vortices often seen in high-resolution viscous simulations.
4. Extension to Multicomponent Flows: The method is successfully extended from single-species to multi-species compressible flows, handling complex interactions between different gases and phases.

4. Results and Validation

The method (termed HY-THINC or HY-THINC-D for viscous cases) was validated against several benchmark cases:

• Multi-species Shock Tube: The method captured the material interface within a few grid cells with zero oscillations, significantly outperforming standard MP5 and WENO schemes in resolution.
• Material Interface Advection: Demonstrated sharp interface preservation without smearing.
• Shock/Density Wave Interaction (Shu-Osher): The sensor correctly identified the shock but did not activate the THINC scheme in high-frequency sinusoidal regions, preserving the wave profile. In contrast, other sensors (like TENO or density-based detectors) falsely identified these regions as discontinuities, distorting the solution.
• Periodic Double Shear Layer (Viscous): The proposed method prevented unphysical "braid vortices" that appeared in upwind schemes (MP5) and TENO-THINC schemes. The central scheme for tangential velocities was crucial here.
• Kelvin-Helmholtz Instability (Viscous): Successfully simulated the mixing of two species. The method showed that even pressure could be computed via a central scheme across the interface without oscillations.
• Compressible Triple Point: Captured complex shock-interface interactions and fine-scale vortical structures. The sensor correctly identified contact discontinuities ($C_1, C_2, C_3$) while ignoring shockwaves, allowing THINC to sharpen only the contacts.
• Richtmeyer-Meshkov Instability (Viscous): Showed improved resolution of roll-up vortices compared to MP5, with a thinner interface thickness.
• Shock-Multi-Bubble Interaction (Hypersonic): The method remained robust and stable in a complex, high-speed multi-material scenario where standard MP schemes failed (due to negative density/pressure).

5. Significance

This work addresses a critical gap in computational fluid dynamics: the simulation of viscous multicomponent flows with high resolution at material interfaces.

• Robustness: It prevents simulation crashes caused by the physical inconsistency of applying sharpening schemes to continuous variables.
• Efficiency: By avoiding volume-fraction-based detection, it scales better for flows with many species.
• Accuracy: It achieves "sharp" interface capturing (reducing numerical diffusion) without introducing the spurious oscillations that typically plague high-order schemes near contact discontinuities.
• Physical Fidelity: The explicit treatment of tangential velocity continuity in viscous flows via central schemes ensures that the numerical solution adheres to the underlying physics of viscous shear, a feature often overlooked in inviscid-focused literature.

In summary, the paper provides a comprehensive, wave-appropriate reconstruction framework that dynamically adapts to the local flow physics, offering a superior alternative to existing uniform reconstruction strategies for complex compressible multicomponent flows.