Fans are a fundamental element of engineering systems across a broad spectrum of applications, ranging from industrial and construction sectors (HVAC systems, heat pumps) to the automotive industry (cooling internal combustion engines and, increasingly, battery packs in electric vehicles) and household appliances. Their primary function is to induce the flow of air or other gases for ventilation, cooling, and heat transport. In the context of internal combustion engine (ICE) vehicles, the cooling fan is a key component, responsible for maintaining optimal engine temperature, especially at low vehicle speeds or during idling.
Despite their indispensable role in ensuring energy efficiency and the functionality of cooling and ventilation systems, fans are widely recognized as significant sources of noise. Particularly in the automotive sector, the noise generated by the radiator fan is one of the primary factors influencing the overall vehicle noise level. In certain cases, this noise can dominate other sources, such as the engine or the exhaust system, especially in high mass flow configurations. Increasing consumer expectations regarding acoustic comfort, both inside vehicles and in the external environment, combined with increasingly stringent noise emission regulations, make fan noise reduction an urgent engineering challenge. Fan noise directly affects perceived product quality and customer satisfaction, as well as brand image.
In the past, fan noise was often masked by other, louder components, such as internal combustion engines or exhaust systems. However, with technological progress and the growing market share of electric vehicles (EVs), where the absence of engine noise highlights other sources, fan noise becomes more dominant in the overall acoustic profile of the vehicle. This change in the acoustic landscape forces an evolution of the design paradigm: noise can no longer be a problem solved at a late stage of product development but must be integrally considered from the very beginning of the design process. Early inclusion of acoustic analysis in the design phase, often using advanced simulations, allows for the identification and elimination of potential noise sources before costly physical prototypes are built. Such an approach significantly shortens the product development cycle, lowers costs, and improves the overall quality and competitiveness of the product in the market.
Noise generated by fans can be classified into three main categories, each with different sources and characteristics
Aerodynamic Noise. This is the dominant form of noise in most fan applications. It arises as a result of complex interactions of airflow with the rotor (blades) and stationary fan components, such as the housing (shroud), struts, or guiding elements. Under typical operating conditions and at higher rotational speeds, aerodynamic noise constitutes the prevailing contribution to the overall fan noise level
Electromagnetic Noise. This type of noise has its source directly in the electric motor driving the fan. The fundamental mechanism of its generation is the mechanical excitation of the stator, resulting from fluctuations of the magnetic field in the air gap (between the rotor and the stator). It can also be induced by current harmonics generated by inverters, leading to characteristic switching noise.
Mechanical Noise. This is noise generated by the mechanical components of the fan and its drive system, such as bearings, gears, rotor imbalance, or viscous clutch wear
The contribution of electromagnetic and mechanical noise becomes significant and potentially unmasked only at low fan rotational speeds, typically below 500 to 1000 RPM. Above this range, aerodynamic noise decisively prevails, directing the majority of research and reduction efforts toward its mechanisms.
The complexity of the fan noise problem requires an integrated approach to NVH (Noise, Vibration, and Harshness) analysis, which encompasses not only aerodynamics but also vibroacoustics and electromagnetics. The dominance of aerodynamic noise in most fan operating conditions means that primary noise reduction efforts should focus on understanding and modifying the mechanisms related to airflow. However, in specific scenarios, such as low-speed fans or electric vehicles, electromagnetic and mechanical noise can become significant or even dominant issues. This requires engineers to have a holistic view of noise sources.
The use of advanced simulation tools - such as SIMULIA PowerFLOW for aeroacoustics, SIMULIA Manatee for electromagnetic noise and vibration analysis (e-NVH), and SIMULIA Wave6 for vibroacoustics - enables the modeling of complex interactions between different domains of physics. Such a multiphysics approach allows for comprehensive optimization, identifying and reducing all significant noise sources in the product, which is crucial for meeting growing market and regulatory requirements
Aerodynamic noise, the dominant component of fan emissions, is categorized into two main types: tonal (discrete) noise and broadband noise.
Tonal noise is characterized by the presence of clearly audible, individual tones in the frequency spectrum. These tones appear at the Blade Passing Frequency (BPF) and its harmonics. The BPF is defined as the product of the number of fan blades and the rotor's rotational frequency (expressed in Hz). Under unfavorable conditions, these tones may be perceived as an unpleasant "howl" or "drone"
The primary cause of tonal noise is the periodic fluctuation of blade loading, triggered by the non-uniformity of the mean flow velocity field at the rotor inlet. As a blade rotates, it encounters varying flow conditions (e.g., along the azimuthal direction), causing periodic changes in the vector angle and the amplitude of the relative air velocity at the leading edge. These velocity fluctuations generate periodic changes in the aerodynamic forces acting on the blades, which are radiated as tones at the BPF and its multiples. Sources of inlet flow non-uniformity can include various structural elements or environmental conditions, such as:
Another significant mechanism for tonal noise generation is the interaction of rotating aerodynamic blade wakes (so-called turbulent wakes) with stationary elements located downstream of the rotor. These elements include:
The magnitude of BPF tones and their harmonics is strongly related to the geometry of the entire fan assembly, its housing, and its mounting - for example, the distance between the blades and stationary components, as well as the azimuthal periodicity of the system.
In addition to BPF tones, subharmonic peaks may also appear in the noise spectrum, particularly in axial fans. These result from the interaction of vortex structures forming in the gap between the blade tip and the housing (the so-called "tip clearance") with the succeeding fan blades. These vortices rotate at a lower angular velocity than the fan shaft, leading to the generation of narrow-band peaks at subharmonic blade passing frequencies. The vortex structures are often organized into a series of spirals (up to three), leading to multiple blade-vortex interactions per blade passage, which explains the occurrence of higher subharmonic peaks. Acoustic beamforming studies confirm that the primary sources of both tonal and broadband noise are localized in the region of the tip gap and the blade tip.
Broadband (continuous) noise is characterized by a smooth frequency spectrum devoid of distinct tonal peaks. It is frequently the dominant factor in the overall sound pressure level, especially at operational speeds below 1500 RPM.
A primary mechanism is turbulence ingestion noise. Similar to tonal noise, this arises from flow disturbances; however, instead of periodic discontinuities, the source is the random velocity fluctuations of the turbulent inflow. This leads to stochastic variations in the angle of attack and relative velocity, resulting in fluctuating blade loads and broadband emissions. A parallel phenomenon occurs when downstream stationary components (e.g., guide vanes) "chop" the rotating turbulent rotor wakes. The resulting spectral shape depends on the ratio of the characteristic integral length scales of turbulence to the blade spacing. If these scales are significantly smaller than the spacing, the noise is purely broadband due to uncorrelated sources. Conversely, if the turbulent scales exceed the blade spacing, the sources become correlated, and the spectrum exhibits "broadband humps" centered around BPF multiples. This is primarily observed in the low-to-medium frequency range.
Another significant noise source arises from aerodynamic phenomena around the blade. It is referred to as Blade-Vortex Interaction (BVI) noise, which occurs as a result of the blade interacting with vortices forming in its immediate vicinity. This is a specific and often dominant source of what is de facto tonal noise. It occurs when fan blades interact with turbulent vortices generated either upstream of the fan (e.g., from a radiator or non-uniform inflow) or by preceding blades. This interaction creates periodic pressure fluctuations that radiate as sound. The resulting noise can be divided into two distinct mechanisms:
Another source of noise is Tip Vortex Noise, which applies exclusively to axial and mixed-flow fans. The flow in the gap between the blade tip and the housing is turbulent and complex due to local phenomena, such as induction vortices (Tip Leakage Vortices) forming at the blade tips caused by cross-flow transverse to the blade movement, resulting from the pressure difference on both of its sides. This noise is the result of the interference of induction vortices with the housing and the adjacent blade, and its intensity increases as the radial distance between the blade tip and the housing increases.
The characteristics of aerodynamic fan noise are heavily dependent on the fan's architecture and geometry. Below are the primary sources of aerodynamic noise categorized by fan type:
Tonal noise in axial fans stems primarily from inlet flow non-uniformity and the interaction with stationary components downstream of the rotor, such as Outlet Guide Vanes (OGV). The dominant broadband noise sources include the interaction of turbulent inflow with the rotor blades, blade self-noise, and tip vortex noise. In configurations featuring guide vanes, the impingement of turbulent blade wakes onto the OGV also significantly contributes to the broadband spectrum.
Centrifugal fans are a type of fan in which the air changes its flow direction by $90^{\circ}$—it enters in a direction parallel to the fan axis and exits in a direction perpendicular to it. The primary cause of tonal noise in centrifugal fans is the non-uniformity of the flow field at the rotor inlet, and in centrifugal fans with a spiral housing (volute)—the interaction of the flow in the blade wakes with the volute cut-off (tongue).
The mechanisms of broadband noise generation are similar to those in axial fans, with the exception of tip vortex noise, which does not occur in centrifugal fans. A significant source of broadband noise, however, is the interaction of the turbulent flow at the rotor discharge with the volute cut-off and the entire housing.
Centrifugal fans with forward-curved blades have unique acoustic characteristics. In these fans, broadband noise is related to the flow structure throughout the entire fan rather than just the flow between the blades. The noise level in these fans increases with the flow rate, without reaching a minimum at the Best Efficiency Point (BEP), which is typical for other types of fans.
In a cross-flow fan, air passes through the rotor twice, entering the suction zone, traversing the interior, and being expelled in the discharge area. This results in a two-stage acceleration of the fluid. Although generally characterized by relatively quiet operation, their aeroacoustic profile is highly sensitive to the formation and stability of internal vortices. Similar to forward-curved centrifugal fans, the noise level increases continuously with the flow rate, lacking a distinct minimum at the BEP. Their acoustic signature is governed more by the large-scale flow topology throughout the assembly than by the fine-scale flow structures on the individual blade surfaces.
Precise determination of fan noise levels requires specialized measurement and analytical methods to quantitatively assess acoustic emissions and identify their sources. In acoustics, two fundamental parameters are used to describe noise: Sound Pressure Level (SPL) and Sound Power Level (SWL).
The Sound Pressure Level (SPL) is a value directly measurable using a sound level meter. It results from periodic changes in air pressure caused by the compression and rarefaction of an elastic medium (typically air). Acoustic pressure is defined as the root mean square (RMS) value of these fluctuations over a given period. Due to the vast dynamic range of human hearing (from 20 μPa to 200,000,000 μPa), SPL is expressed on a logarithmic scale relative to a reference pressure of 20 μPa. SPL depends heavily on the distance from the source, the propagation medium, and environmental factors such as reflections or absorption. For instance, doubling the distance from a point source typically results in a 6 dB decrease in SPL. Since human sensitivity to sound is non-linear across the frequency spectrum - peaking between 1–5 kHz (specifically around 3000–4000 Hz) - sound level meters utilize frequency-weighting filters. The A-weighting filter is most commonly used for environmental noise, as it best approximates the human ear's response to moderate sound levels.
The Sound Power Level (SWL) cannot be measured directly; it represents the total acoustic energy emitted by a source per unit of time, independent of its surroundings. SWL is an intrinsic characteristic of the sound source and does not vary with distance or room acoustics. For a reliable comparison between different devices, SWL values should be used exclusively.
Fan noise measurements are conducted using either laboratory methods (in acoustic chambers) or field methods (in-situ).
Laboratory measurements of fan sound power are often conducted in specially designed acoustic chambers (anechoic and reverberation chambers) that minimize environmental influence on the measurement results. An anechoic chamber is a room designed to eliminate sound reflections, creating conditions similar to a free acoustic field. This allows for the precise determination of the sound power level of a noise source according to standards such as ISO 3745. Measurements in a semi-anechoic chamber (with a reflecting floor) are frequently used for large sources. The anechoic chamber must meet specific requirements, verified by comparing the spatial decay of the sound pressure level from a test source with the decay predicted by the inverse square law. In the chamber, microphones are arranged on a measurement surface (e.g., a hemisphere) surrounding the tested object at points corresponding to different surface areas. The EN ISO 3744:2010 standard recommends six microphone positions on a hemisphere if the largest dimension of the reference parallelepiped does not exceed 8 m. The background noise level in the chamber should be at least 10 dB lower than the noise level generated by the tested source. If the difference is between 10 dB and 20 dB, a correction is applied. A reverberation room is a room with highly sound-reflective walls that creates a diffuse acoustic field. This also allows for the determination of the sound power level, especially for small sources, according to ISO 3743 standards.
For measurements in real-world conditions (in-situ), where the acoustic environment is complex and numerous reflections occur, special field methods are used, such as sound intensity mapping. This method allows for the identification and localization of noise sources on the surface of the tested object. Sound intensity (i.e., the flow of acoustic energy) is measured at many points on a surface surrounding the source, allowing for the creation of a contour map showing the noise distribution. Sound intensity mapping is particularly useful for determining sound power under conditions similar to a free field over a reflecting plane, according to the PN-EN ISO 3744 standard. These tests are performed using a sound intensity probe consisting of a pair of microphones.
Specialized equipment is used for fan acoustic measurements:
One of the key methods for analyzing noise measurement results is spectral analysis, which allows the noise signal to be decomposed into its frequency components, revealing the presence of tones and broadband noise characteristics. The Fast Fourier Transform (FFT) is commonly used - an algorithm for efficiently calculating the discrete Fourier transform (DFT) of a signal from the time domain to the frequency domain. The FFT measurement process includes signal acquisition (digital sampling at a rate at least twice the highest frequency component, according to the Nyquist theorem), optional windowing (e.g., Hanning window to minimize spectral leakage), applying the FFT algorithm, analysis of results (amplitude and phase spectrum), and visualization. A significant measurement parameter is frequency resolution, defined as the ratio of the sampling frequency to the number of samples ($f_{res} = \frac{f_s}{N}$), defining the smallest detectable difference between two spectral components. FFT analysis is essential for identifying tonal peaks (e.g., BPF and its harmonics) and assessing broadband noise characteristics. It allows for understanding the causes of excessive noise and directing reduction efforts.
Numerical simulations, particularly Computational Fluid Dynamics (CFD) and Computational AeroAcoustics (CAA), are invaluable tools in the design and optimization of fans for noise reduction. Various approaches are used in fan noise simulations, frequently combining CFD methods with CAA techniques. CFD calculation is a fundamental tool for simulating fluid flow patterns and serves as an essential preliminary step for CAA. It allows for the detailed analysis of how air flows through the fan, ducts, heat exchangers (radiators, heaters), inlets, outlets, diffusers, vents, or filters, identifying areas of turbulence, flow instability, and pressure fluctuations, which are direct noise sources. Without an accurate solution of the flow field, the precise prediction of sound generation and propagation would be impossible.
Simulations enable engineers to quickly analyze and compare numerous design variants in a virtual environment. This includes studying different blade geometries, blade counts, shroud designs, and entire air flow system configurations. By detecting and correcting potential noise-related design errors early in the development process - before physical prototypes are built - significant time and cost savings can be achieved. This iterative digital prototyping drastically reduces the need for expensive physical prototypes and associated testing, thereby shortening the product development cycle. The economic and temporal benefits of CFD-CAA methods make these tools not only analytical instruments but strategic elements of the product development process. This represents a fundamental shift from traditional "build-test-fix" cycles - where noise problems are identified and solved late in development - to a "simulate-optimize-validate" paradigm. In this modern approach, simulation proactively guides initial design choices regarding acoustic characteristics, while physical tests serve primarily for final validation and fine-tuning. Such a proactive approach minimizes costly late-stage design changes. Discovering noise-related errors only after physical prototypes exist can involve significant costs for re-tooling, materials, and project delays. Consequently, simulation becomes a primary design tool, enabling engineers to explore a wider design space and achieve an optimal, quiet solution much faster and more cost-effectively than through physical prototyping alone.
The characteristic features of fan noise require the application of various modeling approaches and analysis techniques. Tonal noise is often associated with deterministic, periodic interactions (e.g., blade-wake interaction or blade-radiator interaction), requiring transient simulations that accurately capture these events. Broadband noise, on the other hand, results from the stochastic nature of turbulence, requiring the use of advanced turbulence models—such as Large Eddy Simulation (LES)—that capture transient flow fluctuations. This means that a comprehensive aeroacoustic simulation must precisely capture both the mean flow and its transient components to correctly predict the full noise spectrum. If a simulation relies solely on a steady-state flow model (e.g., RANS), it will inherently filter out the time-varying components responsible for generating tonal and most broadband noise. Therefore, for accurate fan noise prediction, the simulation must be transient. Furthermore, because tonal and broadband noise originate from different physical mechanisms, the chosen CAA models must be capable of capturing both phenomena simultaneously.
In the field of fan noise simulation, the integrated Dassault Systèmes SIMULIA software suite holds a critical position, allowing for comprehensive acoustic analysis: aeroacoustic, vibroacoustic, and even electromagnetic (e-NVH). For fan applications, the key tools include PowerFLOW/PowerACOUSTICS, Manatee, and Wave6.
SIMULIA PowerFLOW forms the basis of this integrated approach. It is advanced CFD software distinguished by its unique transient calculation method based on the Lattice-Boltzmann Method (LBM). This innovative approach to solving fluid flow equations offers significant advantages in fan noise simulation. LBM is inherently a transient and compressible calculation method, making it ideal for aeroacoustic calculations. PowerFLOW directly simulates multi-scale turbulent flows, allowing for the accurate prediction of both tonal and broadband noise. This fundamental feature enables the direct solution of fluid flow and accompanying acoustic phenomena without the need for additional approximations. This is particularly important when studying transient phenomena characteristic of fan applications. Due to these properties, LBM simulations can be used to perform Direct CAA (Direct Computational AeroAcoustics). Although LBM is globally second-order accurate, it exhibits acoustic dissipation properties corresponding to sixth-order Navier-Stokes finite difference schemes. Additionally, its dispersion properties are equivalent to second- or third-order schemes. The low numerical dissipation in the LBM scheme allows acoustic waves to propagate accurately from the flow area and flow-induced noise sources to the far field. This makes direct acoustic estimation possible even in the far field, although it remains computationally intensive and requires high temporal and spatial resolution (discretization). Consequently, accurately separating acoustic propagation from the source area in the near field to microphones in the far field within the PowerFLOW computational domain can be impractical, as sound waves propagating to the far field are very small pressure disturbances compared to turbulent pressure fluctuations in the near-field source area.
Visualization of FIND noise sources in standard vehicle HVAC systems (color scale). Visualization of flow and cabin acoustics using the time derivative of pressure (grayscale) (Dassault Systèmes).
Therefore, for large distances, acoustic analogy methods are commonly used to solve the Ffowcs Williams–Hawkings (FW-H) equation. The PowerACOUSTICS tool is dedicated to this purpose, serving as an aeroacoustic post-processor. PowerACOUSTICS consists of four modules key to fan noise analysis:
PowerACOUSTICS analyzes transient PowerFLOW simulation results, enabling the calculation of acoustic field variables at practically any distance from the source. This approach effectively decouples the noise generation problem from its long-distance propagation. CFD simulations performed in PowerFLOW first capture potential noise sources in the near field; subsequently, the acoustic analogy method is applied within the Far-Field Analysis Module of PowerACOUSTICS.
SIMULIA Wave6 is a software suite for vibroacoustic and aero-vibroacoustic simulation that offers unique methods for noise and vibration analysis across the entire audible frequency range. In the context of fans, Wave6 integrates results from PowerFLOW to model noise transmission through the structure. It calculates both air-borne and structure-borne noise using Finite Element Methods (FEM) for low frequencies (e.g., to capture global structural modes) and Statistical Energy Analysis (SEA) for medium and high frequencies (to model wave propagation through complex structures). Wave6 facilitates the efficient transfer of geometry, meshes, and results from flow and structural models to vibroacoustic models.
Acoustic analysis of a vehicle HVAC system (Dassault Systèmes)
For electrically driven fans, SIMULIA Manatee is dedicated to electromagnetic noise (e-NVH) analysis. Manatee calculates the distribution of magnetic forces generated by the fan's electric motor and the resulting vibrations and noise. It offers a set of weakly coupled multi-domain solvers that account for the mutual interaction of electrical, magnetic, structural, and acoustic phenomena. Manatee provides tools for analyzing magnetic force signatures, spatiograms, and contribution charts to help pinpoint noise sources.
The effectiveness of PowerFLOW and PowerACOUSTICS in practical noise analysis has been confirmed by numerous studies comparing simulation results with experimental measurements. Over the past decade, such research has been conducted by academic centers and leading companies such as Valeo, MAHLE, and Delphi/Aptiv.
A prominent example is the study "Validation of the Lattice Boltzmann Method for Simulation of Aerodynamics and Aeroacoustics in a Centrifugal Fan," which aimed to validate the Lattice Boltzmann Method (LBM) for centrifugal fan applications. Authors Rebecca Schäfer and Martin Böhle used SIMULIA PowerFLOW to analyze a centrifugal fan with a spiral housing, comparing the simulation results with experimental data from a physical test stand.
Tested centrifugal fan without a volute, schematic top view of the fan test stand, and the measurement setup inside an anechoic chamber (Schäfer and Martin Böhle, 2020).
The research evaluated four different operating points, ranging from partial load to overload, demonstrating excellent agreement between simulation and measurements for both aerodynamic and acoustic parameters. Aerodynamic performance - specifically pressure rise and efficiency - was confirmed with deviations of only 0.5–3.5% for the pressure rise coefficient and 0.7–2.5% for efficiency. In the aeroacoustic domain, the study showed excellent correlation in acoustic spectra and total sound pressure levels across a frequency range from 100 Hz to approximately 2400 Hz, with differences between simulation and measurement staying within 0.2–2.4%.
Total sound pressure level $L_{p,e}$ compared to the first and second peak BPF levels $L_{p,e} BPF1$ and $L_{p,e} BPF2$ extracted from the acoustic spectra averaged over measurement positions M1 to M11 in the experiment (EXP) and simulation (SIM) (Schäfer and Martin Böhle, 2020).
Furthermore, LBM simulations accurately predicted sound pressure levels for both tonal components (Blade Passing Frequency – BPF) and broadband noise. Because PowerFLOW enables direct visualization of the flow and acoustic fields, it provides deep engineering insights by identifying four primary noise generation areas:
Sound pressure level of the rotor and housing filtered around the BPF (top) and in the broadband range of 100–2500 Hz (bottom) (Schäfer and Martin Böhle, 2020).
This publication serves as an excellent confirmation of the effectiveness of the LBM method as an efficient approach to designing low-noise fans.
A very interesting example of the acoustic analysis of an engine cooling fan module (Condenser, Radiator, and Fan Module – CRFM), typical for the automotive industry - encompassing the fan, electric motor, shroud, and heat exchangers (condenser and radiator) - can be found in the article by Lallier-Daniels et al. (2016) . The study compared acoustic field values obtained in PowerFLOW and PowerACOUSTICS using both the direct method and the FW-H acoustic analogy with experimental measurements
.
Model of an engine cooling fan module (Condenser, Radiator, and Fan Module – CRFM) (Lallier-Daniels et al., 2016)
In the simulation, due to the complexity of the meshing, the heat exchangers are modeled as porous media regions whose resistance reflects real-world pressure losses. Additionally, a blocking plate was included behind the fan module to simulate the proximity of an engine block in a car. The simulation mirrors the measurement setup in a semi-anechoic chamber, utilizing a large domain and artificially increased viscosity in the outer layers to prevent acoustic wave reflections. The fan operated under free-flow conditions at a rotational speed of 2535 RPM, and acoustic data were collected using 20 measurement points placed 0.5 m in front of the module, reflecting the experimental microphone setup.
The results obtained from the PowerFLOW simulation compared with experimental data show a good overall correlation between the simulation and the experiment. Acoustic spectra revealed the presence of tonal noise around the first blade passing frequency (BPF) and a broadband "hump" in the frequency range below the second BPF. This broadband "hump" could be linked to the presence of rotating turbulent structures in the flow interacting with the fan blades. In the flow field analysis, the $\lambda_2$ vortex identification criterion, proposed by Jeong and Hussain (1995), was utilized. The $\lambda_2$ coefficient is defined based on the eigenvalues of a tensor that is the sum of the squares of the symmetric and antisymmetric parts of the velocity gradient tensor. A vortex (or vortex core) is identified as a coherent region where this tensor has two negative eigenvalues. $\lambda_2$ visualizations allow for the visualization of characteristic patterns in the vortex field: a ring of vortex structures near the gap between the fan housing and the blades, and larger structures appearing where the Condenser, Radiator, and Fan Module (CRFM) frame is closest to the fan. Such vortex structures, especially those rotating at a fraction of the fan speed, can interact with subsequent blades, leading to increased tonal and broadband noise generation. Determining the $\lambda_2$ coefficient in PowerFLOW is an effective method for identifying and visualizing vortex cores in complex flows, which is crucial for understanding aeroacoustic noise generation mechanisms. Furthermore, the analysis of pressure fluctuations on the surfaces of the fan, stator, and electric motor showed high levels in areas such as the suction side of the blades near the leading edge, correlating with the appearance of vortex structures.
$\lambda_2$ field isosurfaces – the contour map represents the normalized horizontal velocity component (Lallier-Daniels et al., 2016)
To quantify dominant noise sources, the Ffowcs Williams–Hawkings (FW-H) acoustic analogy in PowerACOUSTICS was applied. Surfaces were isolated (fan blades, hub, rings, stator, motor, frame, and blocking plate) to enable precise source identification. The results closely matched measurements, confirming that the noise primarily results from air-structure interaction. Analysis of Power Spectral Density (PSD) and directivity patterns for the 1st, 2nd, and 3rd BPF showed very good agreement between direct CAA analysis, FW-H calculations, and measurements. Source contribution analysis ultimately revealed that the blocking plate significantly contributes to noise across the spectrum due to the disturbed outlet flow, while the rotor and stator remain the primary factors for the first BPF.
Comparison of acoustic pressure Power Spectral Density (PSD) calculated in the simulation (CAA) for analysis times of 0.35 s (Sim 1) and 1.0 s (Sim 2) with experimental values (EXP).
Analysis of the obtained values, such as the Power Spectral Density (PSD) of acoustic pressures as well as the directivity patterns for the 1st, 2nd, and 3rd BPF, showed very good agreement between the results of the direct CAA analysis in PowerFLOW, calculations using the FW-H acoustic analogy in PowerACOUSTICS, and direct measurements, especially for higher frequencies. Source contribution analysis revealed that the blocking plate significantly contributes to the noise level across the entire frequency spectrum, which is a result of its exposure to the disturbed outlet flow from the fan. PSD levels from the stator elements also closely matched the levels from the module frame. Ultimately, for the first BPF, the primary factors contributing to the noise were found to be the rotor and stator elements.
Fan noise analysis is a complex issue requiring a holistic approach that accounts for various acoustic emission sources alongside advanced measurement and simulation methods. Fans, which are critical components in systems ranging from industrial applications to the automotive sector, are also significant noise sources. This presents engineers with the ongoing challenge of optimizing designs for both aerodynamic performance and acoustic comfort.
The primary sources of fan noise are categorized into aerodynamic, electromagnetic, and mechanical components.
Understanding these fundamental mechanisms is essential for effective noise reduction.
Fan noise measurement relies on two key quantities: Sound Pressure Level (SPL), which is an environment-dependent measurable effect, and Sound Power Level (SWL), which is an intrinsic property of the source, independent of the environment. Precise measurements require professional acoustic chambers (anechoic and reverberation) and advanced instrumentation, including Class 1 sound level meters, precision measurement microphones, and high-fidelity data acquisition systems. Frequency spectrum analysis using the Fast Fourier Transform (FFT) remains indispensable for identifying specific tonal and broadband components.
In parallel with experimental methods, numerical simulations using Computational Fluid Dynamics (CFD) and Computational AeroAcoustics (CAA) play an increasingly vital role. Advanced methods, such as the Lattice-Boltzmann Method (LBM) coupled with Very Large Eddy Simulation (VLES) and the Ffowcs Williams–Hawkings (FW-H) acoustic analogy, enable noise prediction at the earliest design stages. This reduces the reliance on costly physical prototypes and provides high-fidelity data on noise sources. Integrating these simulations with vibroacoustic (NVH) analysis allows for a comprehensive, system-wide approach to noise reduction.
The integrated SIMULIA solution - encompassing PowerFLOW, PowerACOUSTICS, Manatee, and Wave6 - constitutes a powerful toolkit for manufacturers of cooling and HVAC systems. It enables the design of quieter, more efficient fans that meet increasingly stringent market and regulatory NVH requirements.