Two-component inner--outer scaling model for the wall-pressure spectrum at high Reynolds number

This paper proposes two new semi-empirical models for the wall-pressure spectrum in high-Reynolds-number turbulent flows that overcome the limitations of conventional models by combining inner- and outer-scaled spectral components to accurately capture low-frequency growth and variance across a wide range of friction Reynolds numbers.

The Gist

Imagine you are standing next to a speeding train or a jet engine. You hear a low, rumbling roar mixed with a high-pitched hiss. That sound is caused by turbulent air rushing over the surface of the vehicle. This rushing air creates tiny, invisible "bumps" in pressure against the skin of the plane or ship.

Engineers need to predict exactly how strong these pressure bumps are. Why? Because if they get it wrong, the metal skin of the plane could eventually crack from fatigue, or the noise could become unbearable for passengers.

For decades, scientists have tried to write a "recipe" (a mathematical model) to predict these pressure bumps. The most famous recipe, called the Goody model, worked well for slow-to-moderate speeds. But as we push vehicles to higher speeds (higher Reynolds numbers), the old recipe started to fail. It missed the deep, low-frequency rumble and overestimated the total shaking force.

This paper introduces two new, improved recipes that work for everything from small pipes to massive high-speed aircraft, covering a huge range of speeds.

Here is how the authors solved the problem, using simple analogies:

1. The Problem: Two Different Types of Noise

Think of the air rushing over a surface like a busy highway.

• The Inner Layer (The Small Cars): Close to the road (the wall), the air moves in tiny, frantic, chaotic little swirls. These are fast, high-pitched, and happen very close to the surface.
• The Outer Layer (The Big Trucks): Further out, the air moves in giant, slow, rolling waves. These are deep, low-frequency rumbles that carry a lot of energy.

The old models tried to describe the whole highway with a single, simple curve. But at high speeds, the "Big Trucks" (outer layer) get much bigger and louder, creating a second, distinct rumble that the old models couldn't see.

2. The Solution: The "Two-Component" Approach

The authors realized they couldn't use one curve. Instead, they built a model that adds two separate ingredients together:

1. Ingredient A (The Inner Scale): A fixed recipe for the tiny, fast swirls near the wall. This part stays the same regardless of how fast the vehicle goes.
2. Ingredient B (The Outer Scale): A flexible recipe for the giant, slow waves. This part grows and changes as the speed increases.

By adding these two ingredients together, they can recreate the full sound of the turbulence, capturing both the high-pitched hiss and the deep rumble.

3. The Two New Recipes

The paper offers two ways to mix these ingredients, depending on what you need:

Recipe A: The "Log-Normal" Mix (The Quick & Easy Version)

• How it works: Imagine the energy of the noise is shaped like a bell curve (like a hill). This recipe uses two bell curves—one for the small swirls and one for the big waves—and stacks them on top of each other.
• Best for: Engineers who need a quick, simple calculation that fits the data perfectly without needing complex physics equations. It's like using a pre-made cake mix: it's fast and reliable.

Recipe B: The "Modified Lorentzian" Mix (The Physics-First Version)

• How it works: This recipe is built on deeper theoretical rules about how fluids behave. Instead of just guessing the shape of the curves, it uses math that describes how the energy should behave at the very lowest and highest frequencies (the edges of the spectrum).
• Best for: Engineers who need to predict what will happen at speeds even higher than anything we have measured yet. It's like baking a cake from scratch using first principles; it's harder, but it gives you confidence that the cake will taste right even if you change the oven temperature.

4. Why This Matters

• Accuracy: The old models thought the "Big Trucks" (outer layer) were quieter than they actually are. These new models show that at high speeds, those big waves carry a massive amount of energy.
• Variance (The Total Shake): The models correctly predict that as speed increases, the total shaking force (variance) grows logarithmically (slowly but steadily). The old models got this wrong, predicting too much shaking.
• Universality: These models work for different shapes: flat wings (boundary layers), round pipes, and flat channels. It's a "one-size-fits-all" solution for smooth surfaces.

The Takeaway

Think of the authors as audio engineers who finally figured out how to separate the "hiss" from the "rumble" in a noisy room. By realizing that the noise comes from two distinct sources (tiny fast swirls and giant slow waves) and modeling them separately, they created a tool that helps engineers design quieter, stronger, and more durable aircraft and ships.

Instead of guessing the noise, we can now listen to the "two voices" of the turbulence and predict exactly how loud the future will be.

Technical Summary

1. Problem Statement

Accurate prediction of wall-pressure fluctuations in turbulent boundary layers is critical for mitigating structural fatigue and radiated noise in aerospace and marine applications. While the three-dimensional wavenumber-frequency spectrum is the complete description, most experimental data is limited to the one-dimensional frequency spectrum, $\phi_{pp}(f)$.

Existing semi-empirical models, most notably the widely used Goody model (2004), fail at high Reynolds numbers ($Re_\theta$ or friction Reynolds number $\delta^+$). Specifically:

• They fail to capture the energetic growth in the low-frequency range associated with large-scale outer motions.
• They over-predict the variance of wall-pressure fluctuations.
• They rely on a single modified Lorentzian distribution to match inner and outer timescales, which forces the spectral peak to rise artificially to maintain gradients, leading to discrepancies at high $\delta^+$.

The authors aim to develop models that correctly capture the spectrum's evolution across a wide range of Reynolds numbers ($\delta^+ \in [180, 47,000]$) and accurately predict the logarithmic growth of variance.

2. Methodology

The authors propose a two-component scaling model based on the premise that wall-pressure fluctuations arise from the superposition of two distinct spectral components:

1. Inner-scaled motions ($g_1$): Associated with near-wall turbulence, invariant at high $\delta^+$ when scaled by viscous units.
2. Outer-scaled motions ($g_2$): Associated with large-scale structures, whose amplitude and bandwidth grow with $\delta^+$ when scaled by outer units.

The total pre-multiplied spectrum is modeled as a linear sum:
$$f\phi_{pp}^+ = g_1(T^+; \delta^+) + g_2(T^o; \delta^+)$$
where $T^+$ is the inner-scaled period and $T^o$ is the outer-scaled period.

Two specific model formulations were developed and calibrated against a comprehensive dataset comprising:

• Boundary Layers: High-resolution Large Eddy Simulations (LES) and experiments ($\delta^+ \approx 500 - 11,000$).
• Pipe Flows: Experimental data from the CICLoPE facility ($\delta^+ \approx 4,800 - 47,000$).
• Channel Flows: Direct Numerical Simulations (DNS) ($\delta^+ \approx 180 - 5,200$).

Model A: Log-Normal Approach

• Formulation: Both $g_1$ and $g_2$ are modeled as log-normal distributions in the period domain.
• Features:
  • $g_1$ includes a viscous damping term ($r_v$) that ensures the inner component approaches an asymptotically invariant state at high $\delta^+$.
  • $g_2$ amplitude broadens smoothly with $\delta^+$.
• Calibration: Parameters (amplitudes, peak locations, widths) are fitted directly to the data for each flow type (boundary layer, pipe, channel).

Model B: Modified Lorentzian Approach

• Formulation: Uses a modified Lorentzian spectral shape (similar to Goody but separated into two components) to enforce known asymptotic behaviors.
• Features:
  • Inner Component ($g_1$): Fixed parameters based on theoretical constraints (e.g., $f^{-1}$ overlap region, steep high-frequency decay $\sim f^{-6}$). It is treated as $\delta^+$-invariant for high Reynolds numbers.
  • Outer Component ($g_2$): Parameters (amplitude, high-frequency decay, transition sharpness) vary with $\delta^+$ using logistic (sigmoidal) functions. This ensures smooth transitions between low-$\delta^+$ and high-$\delta^+$ asymptotic limits.
• Calibration: Primarily trained on high-$\delta^+$ pipe data, then adapted for boundary layers by adjusting the outer break period and amplitude.

3. Key Contributions

1. Bimodal Spectral Representation: The paper establishes that the wall-pressure spectrum is best described as the sum of two overlapping spectral components (inner and outer) rather than a single unified distribution.
2. High-Reynolds Number Validity: Both models successfully reproduce the emergence of a distinct outer-scaled peak at low frequencies (high $T^o$), a feature missed by the Goody model.
3. Variance Prediction: The models accurately recover the observed logarithmic growth of wall-pressure variance ($\langle p_w^2 \rangle^+ \propto \ln \delta^+$), correcting the significant over-prediction seen in previous models.
4. Continuous Asymptotic Formulation: Model B provides a continuous formulation that varies smoothly with $\delta^+$, allowing for confident extrapolation to extremely high Reynolds numbers (e.g., $\delta^+ = 5 \times 10^5$) relevant to engineering applications.
5. Unified Framework: The approach unifies data from boundary layers, pipes, and channels, identifying both universal behaviors and flow-specific differences (e.g., slower growth of low-frequency energy in pipes compared to boundary layers).

4. Results

• Spectral Fit: Both Model A and Model B show excellent agreement with experimental and numerical data across the entire frequency range and Reynolds number spectrum.
  • Model A offers a compact representation with minimal inputs, ideal for reproducing specific datasets.
  • Model B enforces physical asymptotic limits (e.g., specific power-law slopes), making it more robust for extrapolation.
• Peak Evolution: The models correctly capture the shift from a single inner peak at low $\delta^+$ to a bimodal spectrum with a distinct inner peak ($T^+ \approx 10-20$) and an emerging outer peak ($T^o \approx 0.2-0.8$) at high $\delta^+$.
• Variance: The calculated variance from the models aligns with established correlations (e.g., Schlatter & Örlü for boundary layers; Lee & Moser for channels) and significantly outperforms the Goody model, which over-predicts variance by a large margin.
• Extrapolation: Model B was extrapolated to $\delta^+ = 5 \times 10^5$, predicting that the outer peak would contain the majority of the energy, centered around 2,000 Hz, while the inner peak moves out of the human hearing range.

5. Significance

This work provides a physics-informed, compact empirical representation of wall-pressure fluctuations that addresses critical gaps in current predictive capabilities.

• Engineering Impact: By accurately predicting the low-frequency energy growth and variance, these models enable more reliable design of aircraft and marine structures against noise and fatigue.
• Theoretical Insight: The separation of the spectrum into inner and outer components validates the "attached eddy" hypothesis in the context of wall pressure and clarifies the scaling laws governing the transition from inner to outer dominance.
• Future Utility: The framework is designed to be extensible. The authors suggest that by adjusting the $g_1$ and $g_2$ components, the model could be adapted to flows with pressure gradients, compressibility, or surface roughness, offering a versatile tool for future turbulence research.

In summary, Massey et al. have successfully replaced the "one-size-fits-all" Goody model with a dual-component framework that captures the complex, Reynolds-number-dependent physics of wall-pressure spectra, offering superior accuracy for both current high-Reynolds data and future engineering extrapolations.