Aeroacoustic Study of Synchronized Rotors

Abstract

The main goal of the present study is to explore the noise mitigation potential using an active control strategy based on rotor phase synchronization. This work is focused on the effects of the inflow velocity on the noise interference effect. The inflow velocity does not affect the phase at which the interference phenomenon is observed, as expected. On the other hand, the intensity of the pressure fluctuations is influenced by the inflow velocity for all of the rotor phase shift conditions investigated. Specifically, as the inflow velocity increases, maintaining a constant rotational speed, in the Overall Sound Pressure Level graphs, a reduction of approximately 10 dB is observed. This effect also applies to cases of destructive interference, highlighting the remarkable versatility of this noise reduction technique.

1. Introduction

Multirotor drones, or UAVs (Unmanned Aerial Vehicles), have found extensive use in a variety of applications [1,2,3,4]. The scientific community faces new challenges due to the growing potential for these drones to operate in proximity to humans. The key objectives for UAV development traditionally include optimizing their performance and endurance, which are crucial for commercial interests. Additionally, issues related to safety, security, and privacy have been thoroughly explored [5,6,7]. Another central concern is their noise impact, widely recognized as a significant barrier to their widespread adoption [8,9,10,11]. The increasing need for noise reduction and its negative effects on human well-being provide strong motivation for this study, which presents an active noise control strategy. Specifically, a synchronization mechanism has been implemented with the goal of generating destructive interference between pressure waves, verifying its effectiveness not only in hovering but also when the propellers are in forward flight. This synchronization exploits the inherent characteristics of multirotor configurations, achieving effective noise reduction without compromising the drone’s structural integrity, aerodynamic shape, or overall performance. Due to the importance of noise reduction, the recent literature includes several studies on noise mitigation for the rotors in UAVs, with notable examples [9,10,11,12,13,14,15,16,17]. However, the aeroacoustic behavior of multirotor systems remains an open area for research. Zhou et al. [18,19] conducted experiments examining the impact of the rotor spacing, while Nargi et al. [20,21] analyzed the fluid dynamic and aeroacoustic interactions in side-by-side propellers, revealing increased noise and the presence of recirculation zones. Additionally, Tinney and Sirohi [22] assessed the aerodynamic and near-field acoustic performance of isolated, quad-, and hexacopter rotors, while Jia and Lee [23,24] investigated the interactions between coaxial and quadrotor rotors. Further studies by Ko et al. [25] explored the noise directivity in diamond and square multirotor arrangements, and Lee et al. [26] examined the rotor-to-rotor interactions by varying the separation distances. Moreover, research on phase control for noise reduction in multirotor UAVs is relatively limited. In 2019, NASA’s Langley Research Center [27,28] evaluated the influence of variables such as the number of propellers and the direction of rotation on the noise reduction potential of phase control. Moreover, Pagliaroli et al. [29] investigated the possibility of using the phase angle as an active control strategy and presented an innovative approach to studying the interaction. Guan et al. [30] proposed a systematic approach to noise reduction via phase synchronization, demonstrating a potential 12 dB reduction in the ground-level noise in the optimal configurations. Additionally, Shao et al. [31] found that phase synchronization, particularly with co-rotating propellers, can significantly reduce noise, aligning with the findings of the present study. Finally, a comparison between experimental tests and numerical simulations is presented in [32], showing that the load over the blade decreases when switching from an isolated propeller to a multirotor configuration. This effect is independent of phase control. In addition, the acoustic behavior is more sensitive to the coupling of the propellers. To our knowledge, most of the research on phase synchronization for noise reduction in multirotor drones has been theoretical, underscoring the need for experimental studies such as that presented in this work. Propellers’ sound emissions pose a significant challenge for noise characterization and prediction. In fact, the main noise sources remain consistent with those associated with helicopters; in the case presented in this work, numerous unknowns need to be investigated, such as the effect of a reduced size and the relationship, if one exists, between tonal noise, broadband noise, and the influence of forward flight.

2. Methods

2.1. Drone Noise

In propeller-driven aircraft, the primary sources of noise are the engine and the propeller itself [10]. The noise produced by propellers has been thoroughly examined in the literature [9,14,16,33,34,35,36].

Typically, the aerodynamic noise from propellers is categorized into tonal and broadband contributions:

$$p^{\prime }(\mathbf{x},t)=p_{TN}^{\prime }(\mathbf{x},t)+p_{BB}^{\prime }(\mathbf{x},t)$$

(1)

where $p_{TN}^{\prime }(\mathbf{x},t)$ represents the tonal component of pressure fluctuations, while $p_{BB}^{\prime }(\mathbf{x},t)$ denotes the broadband counterpart.

For thin blades operating at subsonic Mach numbers ($M<1$), the narrow-band contribution of two primary sources

$$p_{TN}^{\prime }\left(\mathbf{x},t\right)=p_T^{\prime }\left(\mathbf{x},t\right)+p_L^{\prime }\left(\mathbf{x},t\right)$$

(2)

where $p_T^{\prime }$ represents the pressure fluctuations associated with the blade’s thickness, and $p_L^{\prime }$ pertains to aerodynamic loading.

The thickness term addresses the fluid displacement caused by the body, whereas the loading component considers the force distribution across the surface of the body.

Conversely, the broadband noise generated by the propeller is linked to the interaction between turbulent flow structures and the blade’s edge, as well as flow separation.

Typically, the broadband contribution can be divided into

$$p_{BB}^{\prime }(\mathbf{x},t)=p_{TE}^{\prime }(\mathbf{x},t)+p_{LE}^{\prime }(\mathbf{x},t)+p_S^{\prime }(\mathbf{x},t)$$

(3)

where $p_{TE}^{\prime }(\mathbf{x},t)$ denotes the trailing edge component associated with the interaction of the turbulent boundary layer over the blade’s surface, $p_{LE}^{\prime }$ refers to the leading edge component related to the incoming flow, and $p_S^{\prime }$ represents the separation term.

A schematic diagram illustrating the primary sources of broadband noise involved is shown in Figure 1.

In the present study, the primary focus is on the tonal component, as interference primarily affects this component. This can be associated not only with phase control but also with the effects on the blade loading, potentially altering the thrust produced by the propeller independently of synchronization. The literature indicates that in a side-by-side configuration, the thrust coefficient should decrease by approximately 5–6% [18,20]. The cited articles indicate a possible thrust reduction of 6.3% at 5200 RPM compared to that for a single rotor, along with a torque reduction of approximately 11%. This phenomenon is attributed to the wake interaction, which affects the torque values, resulting in a reduction in the absolute terms across the entire speed range. The torque fluctuations are similar to those observed in the case of a single propeller. Therefore, in the presence of a wake interaction between the two rotors, there is not only a decrease in thrust but also a reduction in torque, suggesting a potential decrease in blade drag as well.

2.2. The Experimental Setup

Investigations were conducted in the Acoustic Wind Tunnel Braunschweig (AWB). This is an open-jet wind tunnel (Göttingen-type) capable of reaching speeds of up to 65 m/s, where the environment surrounding the nozzle consists of an anechoic chamber with a cutoff frequency of 250 Hz. The nozzle is 1.2 m high and 0.8 m wide.

Far-field noise acquisitions were performed using 37 microphones arranged into four non-equidistant linear arrays. Among the four arrays, only array 1 and array 4 were used for this analysis, corresponding to the layout shown in Figure 2. The pressure time histories were sampled at a frequency of 100 kHz on a GMB Viper GmbH 48-channel data acquisition unit. For brevity, in the present study, only the case with the wind tunnel turned off is reported to simulate hovering conditions. For propulsion, two KDE 4012 XF-400 (KDE, San Antonio, TX, USA) motors with three-bladed 15.5″ × 6.3 TRIPLE-EDN propellers (KDE, San Antonio, TX, USA) were used, with the blades spaced equally at 120°. Both propellers were positioned within the potential core of the wind tunnel. In Figure 3, the blade characteristics are shown. The motors were managed using a custom-made phase synchronizer, implemented to control the rotor’s azimuthal position $\theta \left(t\right)$ and the phase between them $\psi \left(t\right)$. The convention used to denote the rotor phases and the phase delay is sketched in Figure 4.

Hereafter, the results obtained from the performance qualification campaign for the phase control system are shown and discussed. The standard deviation of the phase delay, reported as a horizontal error bar in Figure 5, is very small, highlighting the good stability of the controller. In the same figure, the standard deviations of the rotational speed of the two propellers are represented as vertical error bars. It is worth noting that only minor variations in the rotational speed are required to maintain a stable phase shift between the two propellers. This indicates that the test cases can be considered to operate at a constant rotational speed. Additionally, using two markers (a red dot for the master propeller and a blue circle for the slave propeller), the mean rotational speed values for both rotors were plotted, showing a perfect overlap between the two markers. Finally, the mean rotational speed values upon $\psi$, for the seven different test cases, remain fairly constant around the nominal angular velocity. Hence, it can be concluded that the performance of the synchronization system is quite satisfactory. This graph refers to a rotational speed of 5200 RPM, with a wind tunnel flow velocity of 10 m/s. Each color refers to a different test.

2.3. Test Cases

Several test cases were investigated by varying the propeller distances, inflow velocity, and rotor phase delay, as reported in

Table 1. Test matrix of the experimental campaign.

Test Case n.	$\mathit{\psi }$ [deg]	$\mathbf{\Omega }$ [RPM]	d/D	U∞ [m/s]
0–4	0	5200	1.20	0, 5, 10
5–9	15	5200	1.20	0, 5, 10
10–14	30	5200	1.20	0, 5, 10
15–19	45	5200	1.20	0, 5, 10
20–24	60	5200	1.20	0, 5, 10
25–29	75	5200	1.20	0, 5, 10
30–34	90	5200	1.20	0, 5, 10

.

Due to space and time constraints, only the cases listed in the table will be shown out of the 104 cases analyzed; however, the presented results were consistent across all of the cases in the database. Additionally, the data analysis will focus on microphones 6 and 33, which were positioned in the rotor disk plane.

3. Results

In Figure 6, the Overall Sound Pressure Level (OASPL) for microphone 6 is shown, with variations in the inflow velocity and rotor phase delay; the frequency limits used for the OASPL calculation correspond to those imposed by the applied digital filter, with $f_{min}=150$ Hz and $f_{max}=2000$ Hz. In this figure, two OASPL maxima occur at $\psi =$ 15° and $\psi =$ 75°, with a minimum at $\psi =$ 45°.

The phase at which these maxima and minima occur remains consistent across different inflow velocities. These maximum and minimum points are linked to the interaction between coherent components of the pressure fluctuations: the phase shift and the distance between the rotors introduce a time delay, leading to either destructive or constructive interference. Increasing the forward velocity reduces the propeller noise by decreasing the aerodynamic loading, yet this does not affect the effects that phase control has on the acoustic interaction of the two propellers in any way. Notably, in scenarios where the phase delay induces destructive interference, this effect remains consistent.

For microphone 33, positioned on the opposite side of the rotor plane, an inverse OASPL trend is observed (see Figure 7): two minima appear at $\psi =$ 15° and $\psi =$ 75°, and a maximum appears at $\psi =$ 45°.

It is worth noting that a phase shift of 75° produces constructive interference at microphone 6 and destructive interference at microphone 33. This interference phenomenon significantly influences the polar distribution of the noise, impacting the noise’s directivity. In contrast, the noise directivity for an isolated rotor would be symmetrical. The most critical aspect of this noise mitigation technique is that acoustic mitigation can only be achieved within specific, limited areas using two rotors. To highlight the effect of the inflow velocity on the noise interference, two phase delays were selected to focus the following data analysis, $\psi =$ 45° and $\psi =$ 75°, corresponding to the minimum/maximum OASPL observed in Figure 6 and Figure 7. In Figure 8, the OASPL for $\psi =$ 45° and $\psi =$ 75° appears to be evenly spaced, showing that the effect of speed on constructive or destructive interference is similar; therefore, the flow velocity reduces the aerodynamic load in the same way for the two different interferences. Indeed, for constructive interference, the noise generated by the two rotors is spatially combined. Conversely, destructive interference should cancel out the noise generated by the propellers regardless of the inflow velocity.

In other words, the OASPL for destructive interference is expected to be independent of the aerodynamic load. The acoustic decay in this image appears to be linear, but there are not enough cases to demonstrate this.

To explain this behavior, the Sound Pressure Spectral Levels (SPSLs) related to microphone 6 for the different inflow velocities are calculated and reported in Figure 9; the resolution used for the spectra shown is $\Delta f$ = 1 Hz.

Comparing Figure 9a,b, it is noted that the largest difference between constructive and destructive interference conditions is associated with the second harmonic, which is reduced by approximately 10 dB. This reduction is observed at all wind tunnel velocities and also for microphone 33, as shown in Figure 10.

Destructive and constructive interference only affect the second harmonic in all cases, maintaining independence from the free-stream velocity. This phenomenon is particularly interesting because the interference being limited to the second harmonic is not immediately intuitive. The blue dashed line in both Figure 9 and Figure 10 represents the acoustic spectrum of both motors in a side-by-side configuration without the propeller, clearly highlighting the difference in noise emissions.

A phenomenon already observed in the work of Sinibaldi [14] occurs, namely that the acoustic behavior of the motor changes when the propeller is removed, even though the motor’s rotational speed remains unchanged. This change in behavior is due to the difference in the load on the motor shaft. Indeed, it can easily be observed that in the case with the propeller, two peaks appear among the tonal components, which are attributed to the acoustic characteristics of the motor with the propeller. In contrast, when analyzing the motor without the propeller, the same result is not obtained.

To investigate this effect more thoroughly, Figure 11 displays four low-pass-filtered time series: (a) an isolated rotor, (b) a synchronized rotor ($\psi =$ 0°), (c) rotors under constructive interference ($\psi =$ 45°), and (d) rotors under destructive interference ($\psi =$ 75°).

The acoustic signature of the isolated rotor blade’s passage is a negative–positive peak, representing an expansion wave followed by a compression wave, as shown in Figure 11a. In the figure, “c” denotes compression, “e” denotes expansion, and a multiplicative coefficient indicates when the amplitude is a multiple of that observed in the isolated rotor.

In the case of the synchronized rotor, a“c-e” pair appears, with a time delay fully attributable to the spatial separation of the two rotors, amounting to 1.52 ms, matching the delay expected from the rotor spacing analytically calculated, which corresponds to approximately 1.4 ms (see Figure 11b).

Under constructive interference, a “2e-2c” signature is observed, with an expansion–compression wave similar to that for the isolated rotor but with a double amplitude (see Figure 11c).

Finally, under the destructive interference condition, the interaction between the two “e-c” sequences results in the cancelation of one expansion and one compression wave in Figure 11d. Specifically, the second “e” and the third “c” are canceled, leaving a single expansion and compression with increased delay. This selective cancelation explains the effect on the second harmonic of destructive interference.

In this dimensionless space–time domain of Figure 12, the pressure fluctuations are shown, simultaneously sampled using all of the microphones in the same array. For the plot, time was non-dimensionalized using the blade passing frequency, $f^{\ast }$, while the space was normalized according to the corresponding wavelength, $\lambda =c/f^{\ast }$, where c represents the speed of sound. This visualization aims to reveal the wavefront’s topology. The time series were filtered prior to the plotting.

A comparison is made between the pressure fluctuations of an isolated rotor positioned at the center of the tunnel and those of a perfectly synchronized, side-by-side configuration ($\psi =$ 0°). In Figure 12a, we have the isolated rotor, where from the graph, we can observe that only one pressure wave advances at a time, at each blade passing period ($\Delta t{f}^{\ast }=1$). This means that the rotor emitting the wave is a single rotor. Of the two dashed rectangles, the larger one indicates the pressure wave emitted by the blade, while the smaller one focuses on a much less intense wave visible only for $x/{\lambda }^{\ast }<0.5$ (highlighted in the dotted rectangle). This is due to an installation effect linked to the upstream reflection of the tonal component caused by the pylon’s presence. In Figure 12b, in the side-by-side case, we see two waves for each pass instead of just one. The waves do not overlap, as seen here, due to ${\tau }_d{f}^{\ast }$, the delay caused by the distance between the rotating propellers, since the pressure wave emitted by the more distant propeller has to cover a greater distance.

This time delay is defined as

$${\tau }_d{f}^{\ast }={\displaystyle \frac{d}{c}}{f}^{\ast }$$

(4)

In the space sub-domains $x/{\lambda }^{\ast }<0$ and $x/{\lambda }^{\ast }>1$, a reduction in the wavefront intensity is observed due to the tonal noise’s directivity and at greater distances from the microphones to the source (image blurring). The wavefronts are perfectly equispaced in time. This aspect is linked to the stability of the active control system for the rotational speed and phase.

In Figure 13, the pressure fluctuation fields generated by side-by-side rotors under interference conditions with phases $\psi =$ 45° and $\psi =$ 75° are presented.

Across all four fields, a single blue-red band is observed for each blade passing period, similar to that for an isolated propeller. This phenomenon arises because under interference conditions, the noise emitted by the two sources (the rotors) synchronizes, causing the setup to behave as a unified system.

In Figure 13a,d, the wavefronts are barely visible due to destructive interference, whereas in Figure 13b,c, the wavefronts are significantly more pronounced due to constructive interference.

In Figure 13c, a slight temporal delay reveals a thin, less intense red stripe near each wavefront. This feature is associated with the pylon’s reflection, explaining its temporal lag, as it is detected solely by microphone array 4. This effect, classified as an installation effect, exhibits pronounced directivity, consistent with the findings reported by [37].

4. Discussion

This experimental research investigated the potential of phase synchronization as an active noise control strategy for rotors, focusing on the impact of the incoming flow velocity. The results indicate that phase synchronization can effectively reduce noise regardless of the incoming flow velocity.

From a practical perspective, these findings suggest that phase synchronization holds significant potential for noise reduction in real-world multirotor systems. For example, in drones, where the interaction between the propellers is high and accounts for much of the acoustic emissions, implementing this control system could significantly reduce their noise emissions. Phase synchronization is a relatively simple mechanism and would not negatively impact flight performance, as it could be deactivated during specific maneuvers if deemed detrimental and then reactivated autonomously during steadier flight conditions. Another application where this system could bring notable benefits is within wind farms. Since wind conditions would not affect its effectiveness, it could help reduce the acoustic emissions from wind turbine blades, making them more acceptable to local populations and enabling the installation of wind turbines closer to inhabited areas.

Phase synchronization is a promising technique for noise reduction in multirotor systems.

Future research will delve deeper into the analyzed database, which contains numerous configurations yet to be studied. Additionally, investigating the effects of airflow from different directions could be a future direction for evaluating the system’s effectiveness under varying conditions.

5. Conclusions

In conclusion, an experimental campaign was conducted to examine the impact of the inflow velocity on the acoustic field for various rotor phase delays. It should be noted that the measurements were performed on the rotor plane, where the tonal component of thickness noise was dominant, in order to examine the effects on this tonal part; however, this may have limited our ability to highlight other phenomena.

The primary effect of an increased velocity is a reduction in the OASPL, as expected, due to the decreased aerodynamic load. However, regardless of the flow velocity, certain conditions led to the maximum or minimum OASPL values, corresponding to constructive or destructive interference, respectively. The noise reduction ranged from 3 to 6 dB, depending on the observer’s position, and the type of interference (constructive or destructive) varied with the observer’s location for a given phase angle. Notably, the phase at which these interferences occur remains unaffected by the wind tunnel velocity.

The interference effects were especially significant for the second harmonic, which increased with constructive interference and decreased with destructive interference, while the first harmonic remained unaffected.

Overall, these findings suggest that phase synchronization holds potential as a noise reduction strategy, independent of the inflow velocity, allowing its behavior to be studied in hover conditions without considering advance ratio effects.