Near-Field Combustion-Noise Source Dynamics in a Reacting Supersonic Temporal Mixing Layer

This study utilizes high-fidelity direct numerical simulation to characterize the near-field combustion-noise dynamics of a supersonic reacting hydrogen-air temporal mixing layer, revealing that broadband pressure fluctuations are modulated by combustion intermittency and exhibit weak, low-frequency coherence with heat release rather than collapsing onto a single dominant mode.

The Gist

Imagine a high-speed river of air where two streams are sliding past each other: one is hot, one is cold. In this specific experiment, the streams are made of hydrogen and air moving so fast that they are supersonic (faster than the speed of sound). When they mix, they ignite, creating a chaotic, dancing flame. This paper is a deep dive into the "noise" this fire makes, but not the noise you hear with your ears; it's a study of the invisible pressure waves and energy bursts happening right next to the fire.

Here is the story of what the researchers found, explained simply:

The Setting: A Noisy Kitchen vs. a Wild Forest

Usually, when people study engine noise (like in a jet or a rocket), they look at confined spaces, like a pipe or a chamber. Think of this like a kitchen: if you clap your hands, the sound bounces off the walls, creating a specific echo or a loud, steady hum. In these "kitchen" environments, the heat and the sound waves get locked in a feedback loop, creating a predictable, musical tone.

However, this study looked at a wild forest (an open, boundary-free space). There are no walls to bounce sound off. In this open environment, the noise doesn't hum a single note. Instead, it's more like a storm: a chaotic, broadband roar that covers a wide range of frequencies, with sudden, loud bursts of activity.

The Main Characters

The researchers used a super-powerful computer simulation (like a high-definition movie of the physics) to track three main things:

1. Heat Release: The actual fire burning.
2. Pressure: The "push" or "squeeze" of the air.
3. Dilatation: How much the air is expanding or shrinking.

The Big Discovery: A Chaotic Dance, Not a Waltz

In the "kitchen" (confined engines), the heat and the pressure dance in perfect step, like a waltz. They lock into a rhythm.

In this "forest" (open supersonic flow), the researchers found that the heat and pressure do not dance in step.

• The Analogy: Imagine a drummer (the heat) and a bassist (the pressure) playing together. In a confined room, they might lock into a steady beat. In this open space, the drummer plays a burst of fast notes, and the bassist responds a split second later, but then they drift apart again. They are loosely connected, but there is no single, dominant rhythm holding them together.
• The Result: The noise is "broadband," meaning it's a mix of many frequencies, not a single tone. The connection between the fire and the sound is intermittent—it happens in short, sporadic bursts rather than a continuous flow.

The "Burst" Phenomenon

The paper highlights that the noise isn't constant. It comes in bursts.

• Think of a firework or a popcorn kernel. For a long time, nothing happens. Then, suddenly, a kernel pops (a burst of heat), creating a loud crack (a burst of pressure).
• The researchers found that when these "pops" happen, the pressure waves get stronger. But these pops are random and short-lived. They don't form a steady cycle.

Where Does the Sound Go?

The researchers also looked at which direction this noise "wants" to go.

• The Analogy: Imagine a sprinkler head. Some sprinklers spray water in a perfect circle; others spray mostly to one side.
• The Finding: The heat sources in this flow act a bit like a sprinkler that sprays slightly more to the sides (cross-stream) than straight ahead or behind. It's not a perfect beam, but there is a slight preference for shooting energy sideways.

The "Fingerprint" of the Noise

To understand the relationship between the fire and the sound, the researchers used some advanced math tools (like looking at the "fingerprint" of the data):

• Statistical Structure: They found that the pressure waves share more "personality traits" (statistical structure) with the heat release than they do with the air expanding and shrinking.
• Predictability: Knowing what the pressure did a split second ago helps you guess what the heat will do next, but the relationship is weak and messy. It's not a perfect prediction.
• No Simple Loop: The data showed no signs of a simple, repeating loop (like a pendulum swinging back and forth). Instead, the system is complex, chaotic, and driven by these random bursts.

The Bottom Line

This paper tells us that in a fast-moving, open fire (like a supersonic jet engine without a nozzle), the noise is not a steady hum caused by a locked-in rhythm. Instead, it is a chaotic, broadband roar driven by random, high-energy bursts of burning. The fire and the sound waves interact, but they do so in a loose, sporadic way, creating a complex soundscape rather than a simple musical tone.

The study focuses entirely on describing this "near-field" behavior (what happens right next to the fire) and does not claim to solve how to silence these engines or how this applies to medical devices. It simply maps out the chaotic physics of the noise at its source.

Technical Summary

Technical Summary: Near-Field Combustion-Noise Source Dynamics in a Reacting Supersonic Temporal Mixing Layer

Problem Statement
Combustion noise in reacting flows arises from the interaction of unsteady heat release, compressibility, and turbulent dynamics. While the theoretical foundations of combustion noise (e.g., Lighthill's analogy, Rayleigh criterion) are well-established for confined systems like ducts and combustors—where acoustic feedback often leads to narrowband thermoacoustic instabilities—the behavior of noise in boundary-free, open shear flows remains distinct. In open configurations such as free jets and mixing layers, noise signatures are predominantly broadband, distributed across frequency, and driven by stochastic heat release fluctuations rather than self-sustained resonant modes.

Existing literature often focuses on spatially developing flows, lower-Mach-number regimes, or restricted views of source-response coupling. There is a need to characterize the near-field organization of pressure and heat release in supersonic, boundary-free reacting flows to distinguish between localized source dynamics and global acoustic feedback. This study addresses this gap by examining a supersonic reacting hydrogen-air temporal mixing layer to understand how localized reacting structures, compressive disturbances, and intermittency interact to shape the near-field pressure response.

Methodology
The study utilizes high-fidelity, time-resolved Direct Numerical Simulation (DNS) data of a supersonic reacting temporal mixing layer. The flow is simulated using RAPTOR, a massively parallel, fully compressible reacting DNS solver. The configuration involves a hydrogen-air mixing layer that evolves temporally, allowing for continuous growth via a rescaling procedure. Autoignition occurs due to the mixing of hot oxidizer and fuel streams, driven by high temperatures from the oxidizer stream and supersonic shock effects.

The analysis focuses on postprocessed fields of pressure ($p$), heat release ($\dot{q}$), and dilatation ($\nabla \cdot \mathbf{u}$) within the sampled mid-plane ($x, y$). Key methodological components include:

• Normalization: Fields are nondimensionalized based on the averaged shear-layer thickness ($\delta_0$) and convective velocity ($U_c$).
• Source Characterization: The study employs source-side indicators derived from linearized wave equations, including a heat-release source proxy ($S^*$), the Rayleigh index ($R^*$), and dilatation forcing.
• Spatial Analysis: Compactness measures (autocorrelation-based length scales) are used to determine the spatial extent of source patches relative to the shear layer thickness. A planar source-radiation-potential projection is computed to assess angular bias.
• Spectral and Time-Frequency Analysis: Power spectral densities (PSD), cross-spectra, and coherence are calculated. Short-time Fourier transforms (STFT) and Continuous Wavelet Transforms (CWT) are used to resolve transient spectral bursts and intermittent activity.
• Nonlinear and Information-Theoretic Metrics: Bicoherence is used to detect quadratic phase coupling. Mutual information and transfer entropy quantify statistical dependence and directional coupling between pressure and heat release.
• Dynamical Systems Analysis: Phase portraits, recurrence plots, delay embeddings, Poincaré sections, and finite-time divergence metrics are employed to distinguish between low-dimensional resonant behavior and high-dimensional, turbulence-driven dynamics.
• Intermittency Analysis: Conditional probability density functions (PDFs) are computed for pressure and dilatation during "burst" events defined by high heat-release activity.

Key Results

1. Spatial Organization: Heat release and scalar-gradient activity are spatially localized within the reacting shear layer. In contrast, pressure and dilatation fluctuations span a much larger portion of the domain, influenced by large-scale shock-expansion patterns and freestream compressibility. The Rayleigh index shows spatially intermittent positive and negative spots, indicating localized, rather than uniform, pressure–heat-release exchange.
2. Spectral Characteristics: The flow exhibits broadband spectral energy distributed continuously across frequency, with no collapse onto a single dominant mode. Pressure fluctuations possess the highest spectral energy density, followed by dilatation and then heat release.
3. Weak and Intermittent Coupling: Coherence analysis reveals weak pressure–heat-release coherence concentrated in selected low-frequency bands (approximately $f^* \approx 0.017–0.037$). There is no persistent phase-locked relationship. Time-frequency analysis (STFT/CWT) confirms that energetic content is associated with transient, burst-like events rather than sustained oscillations.
4. Intermittency and Bursts: Combustion noise is dominated by sporadic, high-amplitude events. During heat-release bursts, pressure and dilatation excursions intensify. Conditional PDFs show that burst events are associated with statistically distinct flow states, with dilatation shifting toward positive values (volumetric expansion) and pressure biasing toward positive fluctuations.
5. Dynamical Behavior: Dynamical systems analysis (phase portraits, recurrence plots, embeddings) reveals diffuse, aperiodic dynamics. The heat-release signal exhibits intermittent revisitation of similar states (burst-driven organization) but lacks the compact, closed geometry of a low-dimensional resonant oscillator. Finite-time divergence metrics indicate transient sensitivity without sustained exponential growth.
6. Information-Theoretic Findings: Pressure fluctuations share more statistical structure with heat release than with dilatation ($I(p; \dot{q}) > I(p; \nabla \cdot \mathbf{u})$). Transfer entropy indicates a directional asymmetry where past pressure fluctuations contain more predictive information about future heat release than vice versa, though this is interpreted as a short-lag predictive measure rather than a causal propagation pathway.
7. Radiation Potential: The planar source-radiation-potential projection indicates a moderate low-frequency angular bias in the heat-release source distribution, with a directivity index of approximately 2.8 dB and a stronger cross-stream bias.

Significance and Claims
The paper claims to provide a comprehensive characterization of source-side organization and near-field pressure response in a boundary-free, supersonic reacting flow. The primary significance lies in establishing that, unlike confined thermoacoustic systems, the near-field dynamics of open reacting shear layers are:

• Broadband and Intermittent: Driven by stochastic heat release and compressibility rather than global instability modes.
• Locally Organized: Strongest source activity is tied to mixing-layer dynamics and localized source patches rather than large-scale acoustic feedback loops.
• Statistically Coupled but Not Phase-Locked: Pressure and heat release exhibit weak, transient coupling modulated by combustion intermittency.

The authors explicitly state that these conclusions are specific to the source-side organization within the sampled DNS plane. The study does not calculate far-field radiation or observer-dependent acoustic fields; rather, it provides the necessary foundation (source reconstruction, phasing, and spectral organization) for subsequent observer-based or acoustic-analogy radiation calculations. The work reinforces the interpretation of the Rayleigh index in open flows as a measure of local, intermittent energy exchange rather than a global instability criterion.