Trailing Edge Noise (TEN)

• Trailing edge noise (TEN) is the broadband or tonal acoustic emission generated when turbulent boundary layer flows scatter at the sharp trailing edge of lifting surfaces.
• Its spectral properties include a high-frequency power-law decay and discrete tonal peaks influenced by factors like Reynolds number, angle of attack, and turbulence coherence.
• Control strategies such as serrated edges, porous treatments, and active flow control reduce TEN by disrupting coherent turbulent structures.

Trailing edge noise (TEN) is the broadband or tonal acoustic radiation generated as turbulent boundary-layer or separated flow convects past and scatters at the sharp or blunt trailing edge of a lifting surface, such as an airfoil, blade, or wing. TEN fundamentally constrains aeroacoustic performance in wind turbines, low-noise rotors, and airframe structures, governing both the broadband sound pressure spectrum and the radiated far-field sound power. The theory, prediction, experimental measurement, and control of TEN form a major research area bridging turbulence, instability theory, fluid mechanics, and computational aeroacoustics.

1. Physical Origin and Flow–Acoustic Coupling

The canonical mechanism for trailing edge noise is the conversion of near-wall turbulent pressure fluctuations into acoustic dipole radiation as they are convected past and scattered by the trailing edge. In modern formulations, this process is described by linear acoustic analogies (Lighthill, Ffowcs Williams–Hawkings), or by specialized theories such as Amiet's gust response approach, where the far-field power spectral density (PSD) depends on the wall-pressure spectrum Φ_pp and edge response function, with scaling

$Φ_{pp}(f)\propto U^5 (L / c₀^2) Φ_{uu}(f)$

where $U$ is freestream speed, $L$ a spanwise length scale (e.g. serration wavelength), and $c_0$ the speed of sound (Xue et al., 2023).

TEN is classified into distinct regimes:

• Broadband TEN: Results from turbulent boundary-layer wall-pressure fluctuations scattered at the sharp edge, yielding a continuous, high-frequency spectrum that decays as a power-law ($\sim f^{-n}$, $n\approx 1.7-2.0$ above $5$ kHz) (Lee et al., 2023).
• Tonal TEN: Generated when a laminar separation bubble (LSB) or shear-layer instability in transitional flows sets up a hydrodynamic–acoustic feedback loop, resulting in narrowband tones and discrete sidebands (Alva et al., 2024, Ricciardi et al., 2022, Ricciardi et al., 2019).
• Vortex Shedding TEN: Arises from coherent bluff body-like vortex streets near blunt trailing edges at lower frequencies.

Experiments confirm that only turbulent structures with large spanwise coherence (wavelengths $\lambda_z > 0.6 c$ at peak TEN frequencies, with $c$ the chord) radiate efficiently into the far field (Demange et al., 2024). The acoustic efficiency of scattered pressure is dictated by whether the spanwise wavenumber $|k_z|$ falls below the acoustic wavenumber $U$0.

2. Spectral Characteristics, Scaling, and Experimental Findings

Broadband TEN spectra exhibit several universal features (Xue et al., 2023, Lee et al., 2023, Yuan et al., 2024):

• Spectral Decay: TEN SPL decays monotonically at high frequencies, with the peak located at Strouhal $U$1 for a boundary-layer thickness $U$2.
• Amplitude range: Depending on airfoil design and Reynolds number, surface pressure fluctuations yielding TEN can have root-mean-square amplitudes of 50 Pa (as measured at $U$3) and radiate with spatially coherent streaks of comparable scale (Imai et al., 2024).
• Frequency bands: TEN dominates at $U$4 kHz. Lower frequencies are often associated with vortex shedding or LSB-induced tones, which may be superposed on the broadband spectrum.
• Influence of Reynolds Number and Angle of Attack: The frequency range and magnitude of reduction by trailing edge treatments depend strongly on $U$5 and angle of attack. E.g., in wind turbine airfoils, serrated edges yielded $U$6–$U$7 dB reduction (630–1600 Hz at $U$8), and up to $U$9 dB reduction at specific angles (Xue et al., 2023).
• Response to Free-Stream Turbulence: High inflow turbulence ($L$0–$L$1) can amplify boundary-layer velocity and wall-pressure fluctuations, increasing TEN by $L$2–$L$3 dB across broad bands (Botero-Bolivar et al., 2021).
• Directivity: Maximum acoustic emission occurs broadside to the trailing edge. Treatments such as serrations or porous edges typically attenuate absolute SPL but do not radically alter the underlying dipolar directivity pattern at frequencies of engineering interest (Xue et al., 2023, Zhang et al., 29 Dec 2025).

3. Structure and Coherence of Acoustic Sources

High-resolution experimental and numerical studies establish that TEN is radiated primarily by boundary-layer structures with large spanwise coherence and low spanwise wavenumber (Demange et al., 2024, Yuan et al., 2024):

• Modal Decomposition: Spanwise Fourier decomposition and spectral proper orthogonal decomposition (SPOD) identify streamwise-traveling wavepackets, with phase velocity $L$4, as the principal acoustic sources (Yuan et al., 2024).
• Low-Rank Acoustic Field: The acoustic region is dominated by the leading A-SPOD mode, which captures $L$5–$L$6\% of acoustic energy. This implies that a reduced-order model retaining only a few dominant modes per spanwise wavenumber can reconstruct the far-field spectrum with $L$7 dB error.
• Cut-on Condition: Only spanwise modes with $L$8 contribute to radiated sound. This enforces a critical spanwise scale for efficient sound generation: $L$9 at typical mid-chord frequencies (Demange et al., 2024).
• Implication for Control: Devices that break spanwise coherence—such as serrations or spanwise-varying porous elements—can disrupt the wavepackets responsible for TEN (Yuan et al., 2024).

4. Trailing Edge Noise Control: Passive and Active Strategies

Multiple strategies have been devised for TEN reduction, exploiting physical insights into the underlying source mechanisms.

4.1 Serrated Trailing Edges

Sawtooth or sinusoidal serrations alter the edge geometry, promoting destructive interference of scattered pressure and reducing spanwise coherence (Lyu et al., 2015, Zhang et al., 29 Dec 2025, Tian et al., 2023, Xue et al., 2023):

• Mechanism: The phase variation induced by the serrated profile (parameterized by half-height $c_0$0 and wavelength $c_0$1) causes out-of-phase contributions from adjacent segments, attenuating the radiated field.
• Effectiveness Criteria: For efficient suppression, the serration must satisfy $c_0$2 and $c_0$3, where $c_0$4 and effective height $c_0$5 accounts for the spanwise coherence length of turbulence (Lyu et al., 2015). The recent three-dimensional Wiener–Hopf models enable efficient parametric optimization of serration shapes by accurately capturing source-region and far-field effects (Zhang et al., 29 Dec 2025).
• Role of Turbulence Coherence: If the streamwise coherence length $c_0$6 is short, as in real non-frozen turbulence, serrations are less effective—optimal suppression occurs for $c_0$7 (Tian et al., 2023).
• Experimental Reduction: Serrations produce up to $c_0$8–$c_0$9 dB noise reduction across mid-to-high frequency bands, with minor sensitivity to $\sim f^{-n}$0 in $\sim f^{-n}$1–$\sim f^{-n}$2 range (Xue et al., 2023). Geometric flexibility is possible with negligible loss in noise reduction.

4.2 Porous Trailing Edge Treatments

Porous materials applied to the trailing edge (e.g., sintered metals, foams) mitigate TEN by damping wall-pressure fluctuations via Darcy–Forchheimer drag, disrupting pressure jumps, and attenuating scattered acoustic energy (Fassmann et al., 2018, Rottmayer et al., 2023, Kisil et al., 2017):

• Modeling: Volume-averaged acoustic perturbation equations incorporate porosity $\sim f^{-n}$3, permeability $\sim f^{-n}$4, pore size $\sim f^{-n}$5, and Forchheimer inertial effects.
• Optimized Performance: Gradient-free surrogate optimization yields broadband reductions of $\sim f^{-n}$6–$\sim f^{-n}$7 dB (300–5000 Hz), with attenuation sensitive to pore geometry, porosity, and placement (Rottmayer et al., 2023, Fassmann et al., 2018). Finite porous extensions yield maximum low-frequency gains, while high frequencies may be limited by source–source interference at permeable-impermeable junctions (Kisil et al., 2017).
• Physical Limitation: The benefit saturates for porosity extensions exceeding the relevant turbulent length scale, and partial porosity underperforms full trailing edge porosity by $\sim f^{-n}$8–$\sim f^{-n}$9 dB at higher frequencies.

4.3 Geometric Modifications

Sinusoidal trailing edge deformations of modest amplitude ($n\approx 1.7-2.0$0–$n\approx 1.7-2.0$1) in flat plates or jet–wing configurations can reduce OASPL by up to $n\approx 1.7-2.0$2 dB, primarily targeting low-frequency TEN without significant impact on aerodynamic performance (Horner et al., 2021).

4.4 Flow Control and Targeted Forcing

Active approaches leverage phase-locked or feedback-controlled interventions:

• Extremum Seeking Control (ESC): Model-free algorithms dynamically adjust suction/blowing actuator intensity or position, achieving up to $n\approx 1.7-2.0$3 dB suppression for tonal TEN or $n\approx 1.7-2.0$4 dB for broadband/multi-tone cases (Oliveira et al., 2021).
• Phase-Targeted Thermal Forcing: Laser-induced localized heating disrupts Tollmien–Schlichting wavepackets and the hydrodynamic–acoustic feedback loop, suppressing or delaying TEN as mapped via phase-locked PSP surface pressure measurements (Imai et al., 2024).
• Spanwise-periodic Streaks: Arrays of roughness elements placed to generate optimal streak amplitudes can weaken Kelvin–Helmholtz instability in the separation bubble, fully suppressing tonal TEN in transitional flows for carefully matched Reynolds number, array geometry, and bubble location (Alva et al., 2024).

5. Analytical and Computational Models

Analytical, semi-analytical, and data-driven models form the basis for both understanding and practical TEN prediction.

• Amiet-Type Theories: Central to broadband TEN prediction; compute far-field PSD via convolution of wall-pressure statistics and edge-scattering response.
• Iterative/Matrix Wiener–Hopf Methods: Employed for serrated or porous trailing-edge models, capturing multi-point coupling between acoustic sources and extended geometries (Zhang et al., 29 Dec 2025, Kisil et al., 2017, Lyu et al., 2015).
• Hybrid CFD/CAA and LES: Combined RANS or LES for flow/turbulence statistics, followed by acoustic analogies or direct propagation for spectrum evaluation.
• Surrogate and Optimization Frameworks: Parameterized design (shape, porosity distribution) is tractable through Kriging/Efficient Global Optimization (Rottmayer et al., 2023) or ensemble Kalman methods regularized for geometric smoothness (Luo et al., 17 Sep 2025).
• Reduced-Order Models: SPOD-based acoustic reconstruction enables minimal-mode, high-fidelity predictions (error $n\approx 1.7-2.0$5 dB) over full spectra (Yuan et al., 2024).
• Rotating Blade Extensions: Convected Green’s functions and proper Doppler-weighted averaging formalize TEN prediction for wind turbine or propeller blades at high frequency ($n\approx 1.7-2.0$6), with Doppler exponent $n\approx 1.7-2.0$7 required for energy conservation (Sinayoko et al., 2013).

6. Design Principles and Performance Limits

Critical factors, design criteria, and operational limits include (Xue et al., 2023, Tian et al., 2023, Lyu et al., 2015, Zhang et al., 29 Dec 2025):

• To maximize TEN reduction:
  • Serration height $n\approx 1.7-2.0$8 should exceed the local displacement thickness $n\approx 1.7-2.0$9 for the operational $5$0.
  • Spanwise period-to-height ratio $5$1 in the range 0.2–0.8 yields robust performance.
  • Porous extensions or streak-inducing roughness arrays should cover the spanwise correlation length of dominant wall-pressure fluctuations.
  • For serrations or porous edges, match $5$2 to the streamwise coherence length $5$3 for the target frequency band.
  • Minimize source–source interference for partial porosity; extend the treated region to encompass all coherent source locations.

Performance is bounded at high frequencies by the loss of coherence in turbulence. Overly aggressive serrations or roughness can promote premature transition, increasing broadband turbulence and potentially sacrificing aerodynamic efficiency or increasing far-band noise.

Table: Empirical TEN Reduction Results for Representative Passive Measures

Treatment	Frequency Band (Hz)	Typical Reduction (dB)	Optimized Parameters
Serrated TE (airfoil)	500–6300	2–15	$5$4, $5$5 (Xue et al., 2023)
Porous Edge	300–5000	6–10	Porosity $5$6, $5$7mm (Rottmayer et al., 2023)
Sinusoidal TE (jet)	St $5$8	up to 1.5	$5$9–$\lambda_z > 0.6 c$0, $\lambda_z > 0.6 c$1–10 (Horner et al., 2021)
Streak roughness	Tones (0.8–6.5 St)	up to 20; full suppression	$\lambda_z > 0.6 c$2, $\lambda_z > 0.6 c$3 (Alva et al., 2024)

7. Outlook and Challenges

Ongoing research aims to:

• Unify turbulence–acoustic models under non-frozen, finite-coherence frameworks for direct model–experiment reconciliation (Tian et al., 2023).
• Generalize predictive tools to complex flow environments (high inflow turbulence, urban wind application, unsteady loading) (Botero-Bolivar et al., 2021).
• Exploit data-driven and reduced-order models (SPOD, resolvent analysis) for real-time optimization and feedback (Yuan et al., 2024).
• Enable high-fidelity, computationally efficient optimization for practical engineering design with parameter uncertainties and manufacturing tolerances (Zhang et al., 29 Dec 2025, Luo et al., 17 Sep 2025).
• Develop integrated control strategies combining passive (geometry, porosity) and active (thermal, flow control) elements for next-generation low-noise aerodynamic systems.

The field remains active at the intersection of fundamental fluid mechanics, aerodynamic design, and computational acoustics, with rapid advances in experimental diagnostics and simulation capabilities enabling new theoretical paradigms and control strategies for TEN.