Acoustic Rig Testing Successfully Maps Complex Modal Sound in Engine Nacelles

Aeroacoustic engineers have successfully demonstrated a comprehensive experimental procedure to detect and separate both circumferential and radial acoustic modes within a ducted fan nacelle, establishing an empirical framework directly applicable to full-scale commercial aircraft engines. As modern high-bypass turbofan engines continue to enlarge their front fan diameters to improve propulsive efficiency, the noise generated by the fan stage has become the dominant acoustic source of civil transport aircraft. Strict international noise airworthiness regulations demand deep reductions in engine acoustic emissions, making the targeted design and optimization of internal acoustic liners indispensable. However, identifying how sound energy is actually partitioned across hundreds of propagating acoustic modes inside a confined nacelle duct has long remained a notorious measurement challenge in the aeronautical industry. High-speed internal airflow, complex duct geometry, and severe acoustic interference from turbulent boundary layers have historically prevented engineers from acquiring clean modal power data under realistic operating conditions.

To overcome these persistent experimental bottlenecks, a research team led by Kangle Xu investigated acoustic modal propagation using a scaled rig fan modeling a modern civil transport engine. The experimental nacelle featured an 18-bladed fan measuring roughly 1.0 meter in diameter, operating within an annular duct designed to replicate representative downstream outlet guide vanes, bypass ducts, and liner configurations. Testing focused particularly on high-power operating conditions of approximately 93 percent engine power, an operational regime where fan blade tips rotate at supersonic velocities. Under these demanding conditions, the duct environment produces extreme acoustic sound pressure levels dominated by multiple blade passage frequency tones alongside intense turbulent boundary layer noise. The researchers demonstrated that accurately mapping acoustic energy in such environments requires a unified methodology combining non-uniform sensor geometry, precision rotational tracking, and advanced spatial filtering to isolate true sound waves from aerodynamic pressure fluctuations.

The mathematical foundation of ducted sound propagation rests on classical aeroacoustic duct theory, in which acoustic pressure within an annular cylinder is represented as an infinite orthogonal summation of azimuthal and radial eigenmodes. In a cylindrical coordinate system with mean axial flow Mach number, propagating sound waves split into distinct downstream and upstream traveling components characterized by specific axial wavenumbers. Hard-walled duct boundary conditions require radial eigenfunctions composed of Bessel functions of the first and second kind, weighted by duct hub-to-tip ratios. When an axial wavenumber is purely real, the corresponding modal pair becomes cut-on and radiates freely out of the intake or exhaust duct, whereas imaginary wavenumbers indicate cut-off modes that decay exponentially with axial distance. Integrating the acoustic intensity across the duct cross section yields the sound power transmitted by each individual acoustic mode, providing the crucial quantitative metric needed to tune absorbent acoustic wall treatments.

Capturing the azimuthal modes within the intake and bypass sections required the strategic deployment of circular wall-flush microphone arrays designated as CMD1 and CMD3. Conventional circular arrays with uniform sensor spacing are severely constrained by spatial Nyquist limits, requiring an impractical quantity of physical sensors to avoid spatial aliasing at high acoustic frequencies. The investigators solved this physical constraint by optimizing the angular spacing of 100 flush-mounted transducers around the duct circumference. By minimizing an array sidelobe response objective function, the non-equidistant sensor spacing preserved an array dynamic range of at least 20 decibels while extending mode detection capabilities well beyond half the total channel count. This optimized geometric arrangement allowed the stationary intake array to cleanly measure azimuthal mode orders reaching from negative 70th to positive 70th order, ensuring that all propagating acoustic tones were captured without spatial ambiguity.

Because physical engines and drive shafts exhibit minute rotational unsteadiness during extended testing, raw acoustic pressure signals recorded by wall transducers suffer from continuous phase drift across consecutive shaft trigger intervals. This phase jitter causes higher-order engine harmonics and blade passage frequency tones to become smeared into the underlying broadband noise floor. To counteract this loss of spectral clarity, the researchers implemented an adaptive resampling algorithm that tracked individual shaft revolutions in real time. Each trigger interval was resampled into a fixed number of digital points, dividing the instantaneous shaft rotational phase evenly and linearly interpolating pressure records onto precise angular steps. By synchronizing all recorded time-domain data to the physical rotation of the rotor, the procedure suppressed acoustic components incoherent with the shaft rotation, sharply reducing phase errors for engine order harmonics and significantly boosting the signal-to-noise ratio of pure tones.

Following rotational phase reconstruction, the team separated the complex pressure spectra into discrete tonal components and underlying broadband sound. At supersonic blade tip speeds, blade-to-blade non-uniformities and rotational speed fluctuations create numerous distinct tones, with pure tones at blade passage frequencies exceeding broadband acoustic levels by more than 30 decibels. The automated separation algorithm successfully extracted the broadband acoustic spectrum, which was subsequently analyzed across standard one-third octave frequency bands. Experimental measurements across varied operational rotational speeds revealed that broadband acoustic sound pressure levels increased systematically with increasing fan shaft speed. This positive correlation confirmed that ducted broadband noise is primarily generated by turbulent flow fluctuations over the blade surfaces and turbulent wake shedding downstream of the rotor blades, exhibiting random phase distribution among duct propagation modes.

A critical barrier to flush-mounted acoustic measurements in high-speed ducts is the presence of turbulent boundary layers along the nacelle walls. The hydrodynamic pressure fluctuations within these boundary layers do not represent propagating sound waves, yet their intense energy can completely obscure true acoustic signals, especially at lower frequencies. To resolve this interference, the authors positioned an axial array designated as RMD1 containing 60 equidistant microphones spaced 1.4 centimeters apart downstream of the outlet guide vanes in the bypass duct. Unlike stationary circumferential rings, the RMD1 array rotated continuously around the central axis of the bypass duct, recording wall-pressure measurements at 2-degree increments across an entire 360-degree sweep. This sweeping protocol generated 10,800 distinct spatial coordinate measurements, providing the spatial data density necessary to perform two-dimensional spatial Fourier transforms and generate comprehensive wavenumber-frequency spectra.

Through spatial discrete Fourier transformation of cross-power spectra between axial microphone pairs, the researchers separated convective aerodynamic turbulence from acoustic waves based on their distinctive propagation velocities and axial wavenumbers. Convective boundary layer fluctuations travel with the mean flow at an empirical convection speed of approximately 80 percent of the mean axial velocity, concentrating hydrodynamic energy into a well-defined diagonal zone on the wavenumber plane below 5,000 Hertz. In contrast, acoustic waves propagate upstream and downstream relative to the moving medium at the local speed of sound, defining entirely separate wavenumber trajectories. Integrating the wavenumber-frequency spectrum across these bounded wavenumber domains enabled the researchers to isolate pure hydrodynamic pressure, downstream-traveling acoustic power, and upstream-traveling acoustic power into independent frequency spectra, preventing low-frequency hydrodynamic contamination from distorting acoustic liner evaluation.

The power integration revealed essential aeroacoustic behavior across the audible frequency range in the bypass channel. Below 1,000 Hertz, hydrodynamic boundary layer noise dominated the wall-pressure environment, demonstrating why raw transducer measurements at low frequencies fail to reflect acoustic absorption treatments. Furthermore, below 2,000 Hertz, upstream-propagating acoustic energy significantly exceeded downstream acoustic energy due to turbulent jet noise generated at the nacelle exhaust nozzle propagating back up the bypass channel. Without wavenumber decomposition, this upward jet noise and hydrodynamic turbulence would cause engine engineers to severely undervalue the actual transmission loss performance of bypass acoustic liners. Above 2,000 Hertz, downstream acoustic waves rapidly became the dominant component of the duct sound field, while turbulent boundary layer noise dropped off sharply and cut off entirely above 5,000 Hertz.

Direct modal decomposition confirmed that acoustic power distributions in both the intake and bypass ducts were strongly governed by rotor blade geometry. At the first blade passage frequency, corresponding to the 18th engine order harmonic of the 18-bladed fan, the 18th azimuthal mode dominated both sound pressure and sound power distributions. In the nacelle intake, although mode orders spanned broadly up to order 70, only modes between negative 20th and positive 20th order were theoretically cut-on and capable of radiating outward. In the bypass duct, simultaneous axial and circumferential measurements by the rotating array resolved individual radial orders for each azimuthal mode, proving that mode (18, 0) was the primary carrier of downstream acoustic power while mode (18, 1) dominated upstream propagation. Despite higher upstream sound pressure levels driven by incoherent jet mixing, downstream sound power levels were nearly 20 decibels higher than upstream power, establishing a robust, physically verified benchmark for industrial aeroengine acoustic design.

Subject of Research: Aeroacoustic mode detection and decomposition in ducted fan engine nacelles

Article Title: Acoustic mode detection in an engine nacelle with a scaled rig fan

Article References: Xu, K., Lyu, Y., Feng, L., Liu, C., & Ji, J. (2025). Acoustic mode detection in an engine nacelle with a scaled rig fan. Advances in Aerodynamics, 7(1), Article 13. https://doi.org/10.1186/s42774-024-00198-w

Image Credits: AI Generated

DOI: 10.1186/s42774-024-00198-w

Keywords: Acoustic Mode Detection, Engine Nacelle, Turbofan Fan Noise, Azimuthal Modes, Radial Modes, Adaptive Resampling, Wavenumber Decomposition, Turbulent Boundary Layer, Blade Passage Frequency, Non-Equidistant Array, Rotating Linear Array, Acoustic Liners, Duct Aeroacoustics, Sound Power Level, Acoustic Pressure Spectrum, Civil Aircraft Noise