Manifestation of spurious currents and interface regularization in wind turbulence over fast-propagating waves

This study systematically evaluates how spurious currents and interface regularization techniques impact the accuracy of wind turbulence simulations over fast-propagating waves, identifying curvature estimation and flux discretization as critical factors in minimizing numerical errors that otherwise compromise predicted turbulence statistics.

The Gist

Imagine you are trying to film a high-speed race between a gust of wind and a crashing ocean wave. You want to capture every detail: how the wind pushes the water, how the water sprays up, and how the air swirls around the crest.

To do this on a computer, scientists build a digital twin of the ocean and atmosphere. They use a grid (like a giant 3D checkerboard) to calculate how the water and air move. But here's the problem: the line where the water meets the air (the interface) is incredibly tricky to draw on a computer grid.

This paper, by Hanul Hwang and Catherine Gorlé, is essentially a quality control report on three different ways computers try to draw that wavy line. They found that if you draw the line poorly, the computer starts "hallucinating" physics that don't exist, ruining the simulation.

Here is a breakdown of their findings using simple analogies:

1. The "Ghost Wind" Problem (Spurious Currents)

Imagine you are balancing a heavy ball on a flat table. If the table is perfectly level, the ball sits still. But if the table is slightly tilted (even by a tiny, invisible amount), the ball starts rolling.

In the computer simulation, the "table" is the mathematical calculation of surface tension (the skin-like force holding water together).

• The Issue: Some methods (called isoPhi) are like a wobbly table. They calculate the curve of the wave slightly wrong. Because the math thinks the surface is curved differently than it really is, it creates a fake force.
• The Result: This fake force creates "ghost winds" or spurious currents. These are tiny, fake whirlwinds that appear right at the water's surface.
• The Analogy: It's like trying to listen to a quiet whisper (the real wind) while someone next to you is constantly tapping a spoon on a glass (the ghost wind). The tapping drowns out the whisper, making it impossible to hear what's actually happening.

2. The "Smearing" Problem (Interface Regularization)

Now, imagine you are trying to paint a sharp line between blue (water) and white (air) on a canvas.

• The Issue: Another method (called gradPhi) tries to keep the line sharp by using a "compression" tool. It's like using a heavy brush to force the paint to stay in a straight line.
• The Result: While this stops the "ghost winds," it introduces a new problem. The heavy brush accidentally smears a little bit of blue paint into the white area and vice versa. In physics terms, this creates an artificial flux.
• The Analogy: It's like a conveyor belt moving boxes. If the belt moves too fast and the boxes are slightly squished, they might push extra weight onto the next station. The computer thinks there is more momentum (push) moving from the water to the air than there actually is. This makes the wind look stronger than it really is.

3. The "Perfect" Solution (plicRDF)

The researchers tested a third method called plicRDF.

• How it works: Instead of a wobbly table or a heavy brush, this method uses a "laser-guided ruler." It calculates the curve of the wave with extreme precision.
• The Result: It eliminates the ghost winds without smearing the paint. The line between water and air stays sharp and accurate.
• The Takeaway: When they used this method, the simulation matched real-world experiments much better. The "ghosts" disappeared, and the "smearing" stopped.

Why Does This Matter?

You might think, "So what if the computer has a tiny fake wind?"

The authors explain that in the real world, wind and waves are constantly fighting and dancing together.

• In slow waves: The fake winds are small enough that you might not notice them.
• In fast, high-energy waves (like a storm): The fake winds become as strong as the real wind. This completely messes up the prediction.

If you are a weather forecaster or an engineer designing an offshore wind farm, you need to know exactly how much force the wind exerts on the waves.

• If you use the wobbly table method, you might think the wind is weaker than it is because the fake currents cancel out the real ones.
• If you use the heavy brush method, you might think the wind is stronger than it is because of the artificial smearing.

The Bottom Line

This paper is a warning to scientists: Don't just trust the numbers.

When simulating the ocean, the way you draw the line between water and air changes the physics of the whole system. To get accurate weather forecasts and safer engineering designs, we need to use the "laser-guided ruler" (the plicRDF method) to ensure we aren't measuring ghosts or smearing the data.

In short: To understand how the wind and waves talk to each other, the computer needs to see the boundary clearly, without any digital static or blur.

Technical Summary

1. Problem Statement

Accurate simulation of wind turbulence over fast-propagating water waves is critical for weather forecasting and coastal engineering. However, high-fidelity simulations (DNS/LES) of wind-wave interactions face significant challenges in two-phase flow modeling, specifically regarding the air-water interface.

• Numerical Artifacts: In high wave-age regimes (where wave speed is high relative to wind speed), numerical errors at the interface can reach magnitudes comparable to physical flow velocities.
• Spurious Currents: Errors in curvature estimation and surface tension discretization generate non-physical "spurious currents" that contaminate turbulence statistics.
• Interface Regularization: Methods used to maintain a sharp interface (e.g., compression terms) can introduce artificial mass flux, altering momentum transfer across the interface.
• Gap: Existing studies often omit explicit interface resolution or rely on simplified boundary conditions, failing to capture critical two-phase processes like wave breaking and sea spray, or they lack a systematic assessment of how specific interface-capturing schemes affect wind-wave coupling.

2. Methodology

The authors conducted a systematic assessment using the OpenFOAM multiphase-flow solver, comparing three widely used interface-capturing methods:

1. isoPhi: A geometric Volume-of-Fluid (VOF) method using isoAdvector (interface reconstruction based on phase indicator $\phi$).
2. plicRDF: A geometric VOF method using isoAdvector but with Reconstructed Distance Function (RDF) for improved curvature estimation.
3. gradPhi: An algebraic VOF method using the MULES (Multidimensional Universal Limiter with Explicit Solution) limiter, which employs a compression term to sharpen the interface.

Simulation Framework:

• Governing Equations: Incompressible Navier-Stokes equations with the Continuum Surface Force (CSF) model for surface tension.
• Data Analysis: A triple decomposition technique was employed to separate the flow field into:
  1. Phase-independent mean flow.
  2. Wave-coherent fluctuations.
  3. Turbulent fluctuations.
• Test Cases:
  • Static Droplet: To quantify curvature errors and spurious current magnitudes.
  • Moving Droplet: To assess the impact of density/viscosity ratios and interface advection on velocity fields.
  • Solitary Wave: To isolate interface advection effects and flux discretization errors.
  • Monochromatic Waves: Simulations at reference and large scales (high wave age) with turbulent wind inlet conditions, validated against experimental data from Buckley and Veron [46].

3. Key Contributions

• Quantification of Error Sources: The study isolates and quantifies two primary numerical error mechanisms: (1) curvature-induced spurious currents and (2) artificial flux from interface regularization (compression terms).
• Method Comparison: It provides a rigorous benchmark comparing geometric (isoAdvector) vs. algebraic (MULES) VOF approaches in the context of wind-wave coupling.
• Scaling Analysis: The paper demonstrates how these numerical errors scale with grid resolution and wave propagation speed, revealing that errors become critical when spurious current magnitudes approach the bulk flow velocity.
• Experimental Validation: The study bridges the gap between numerical artifacts and physical observations by validating simulation results against high-quality experimental wind-wave data.

4. Key Results

A. Static and Moving Droplet Benchmarks

• Spurious Currents: The isoPhi method produced spurious currents on the order of $O(10^{-1})$, while plicRDF reduced this to $O(10^{-4})$ and gradPhi to $O(10^{-3})$.
• Curvature Accuracy: plicRDF achieved exact pressure jumps and curvature values due to the RDF-based reconstruction. isoPhi suffered from significant curvature errors that scaled linearly with grid refinement ($\Delta x^{-1}$), whereas plicRDF remained largely insensitive to grid resolution.
• Moving Droplet: In moving cases, gradPhi exhibited velocity differences driven by the compression term, particularly at coarse resolutions. plicRDF showed velocity differences primarily driven by curvature errors, which became significant only when the interface moved slowly relative to the spurious current magnitude.

B. Solitary Wave (Interface Advection)

• Artificial Flux: The gradPhi method (MULES) introduced an artificial momentum flux layer near the wave crest, aligned with the wave propagation direction. This flux originated from the compression term and scaled with the interface advection speed.
• Energy Transfer: This artificial flux led to unphysical momentum transfer from water to air, increasing kinetic energy in the air phase and causing rapid decay in the water phase, particularly when the compression coefficient ($C_\alpha$) was high.

C. Monochromatic Waves (High Wave Age)

• Phase-Averaged Statistics:
  • isoPhi: Produced distorted phase-independent mean velocities and underestimated wave-coherent stresses due to persistent spurious currents.
  • gradPhi: Predicted higher wave-coherent stresses than plicRDF due to the artificial flux enhancing momentum transfer above the wave crest.
  • plicRDF: Provided the most physically consistent results, showing smooth convergence with grid refinement.
• Grid Convergence: isoPhi failed to converge asymptotically due to curvature errors. gradPhi showed convergence but with a bias toward higher velocities. plicRDF demonstrated the best asymptotic convergence behavior.
• Scale Effects: In large-scale simulations (higher wave celerity), the impact of spurious currents diminished, but the artificial flux from gradPhi remained significant on coarser grids, creating thickened velocity layers.

D. Comparison with Experimental Data

• Validation against Buckley and Veron's data confirmed that spurious currents (in isoPhi) significantly distort both mean and fluctuating flow components.
• plicRDF yielded the best agreement for wave-coherent stress magnitudes, though all methods underestimated turbulent stress levels (likely due to insufficient grid resolution near the viscous sublayer).
• gradPhi slightly overpredicted wave-coherent stress, attributed to the artificial flux mechanism.

5. Significance and Conclusions

• Curvature is Critical: For high wave-age regimes and laboratory-scale simulations, accurate curvature estimation is paramount. The RDF-based geometric VOF (plicRDF) is superior to standard geometric (isoPhi) and algebraic (gradPhi) methods because it eliminates curvature-induced spurious currents.
• Regularization Trade-offs: While algebraic methods like gradPhi (MULES) maintain sharp interfaces, their compression terms introduce artificial momentum fluxes that can dominate physical effects in under-resolved simulations or fast-propagating waves.
• Practical Implications:
  • Researchers must be cautious when using algebraic VOF methods for wind-wave coupling, as compression terms can artificially enhance stress and momentum transfer.
  • Grid refinement alone cannot fix curvature errors in standard geometric VOF; advanced reconstruction techniques (like RDF) are necessary to achieve asymptotic convergence.
  • Accurate prediction of wind-wave interaction requires a balance between suppressing spurious currents and minimizing artificial flux, with plicRDF emerging as the most robust choice for the tested configurations.

This study provides a definitive guide for selecting interface-capturing methods in environmental fluid dynamics, emphasizing that numerical artifacts can be as influential as physical parameters in determining simulation fidelity.