1 Introduction

With regard to the climate crisis, the reduction of harmful environmental impacts is one of the most important goals of aviation research in the coming years. The European strategy paper “Flightpath 2050” sets targets for pollutant and noise emissions to be achieved by 2050 [1].

In this context, electric drive systems offer several advantages over combustion engines. On the one hand, electric motors can achieve a significantly higher efficiency and noise emissions are significantly lower than those of internal combustion engines. The latter was shown by Hallez et al. [2] on the base of fly-over measurements. Utilizing calculation models, Salucci et al. [3] demonstrated the potential to reduce noise emissions near airfields by establishing electrically and hybrid-electrically powered aircraft. On the other hand, storing electrical energy effectively, with respect to weight and volume, is still a challenge. Because of the low energy density of modern accumulators, compared to conventional fuels, only short-ranged missions are currently feasible for electrically powered aircraft [4]. The first series-produced, electrically powered aircraft is the Antares 20E by the manufacturer Lange Aviation GmbH [5]. Another example is the aerobatic plane Extra 330LE by Siemens, an electrified demonstrator, which was able to achieve a flight speed of 342.86  \(\frac{\mathrm{km}}{\mathrm{h}}\) with its 260 kW electric motor [6]. Utilizing the electric energy most efficiently is a crucial factor in achieving competitive practicality for this technology.

One step to achieve this goal is to employ the reversibility of electric drives. During descent phases the propeller can be driven as a turbine while the engine generates electric power that is stored in the accumulator. This technological approach was first adopted by Pipistrel with their Model Alpha Electro, a training aircraft which is able to recuperate energy in descent by means of a fixed pitch propeller [7, 8]. Further examinations of this technique by Barnes [9] and Carvalho [10] underline its applicability.

This practice, however, raises new questions to propeller design, noise emissions, and decent procedures. The work at hand is embedded in the research project E-DARIT (Electrical Drive And Recuperation In-flight Tests), which addresses these topics. Wind tunnel and flight tests are to be conducted to examine different propeller geometries with regard to their performance and acoustic characteristics. The aim is to identify propeller designs and settings that show good efficiency for both trust and turbine operation. In addition, the designs are analyzed in terms of their recuperation potential and the corresponding noise emissions. As stated before, noise emitted by electrical-driven aircraft and in-flight recuperation have been the subject of several studies. However, research regarding noise emissions from recuperative propellers is yet to be done.

As part of a preliminary test in the subsonic wind tunnel (USK) at the Institute of Aerospace Systems (ILR) of RWTH Aachen University, the performance of a conventional propeller was measured. Results have already been utilized for the validation of simulation outputs generated with the propeller design environment PropCODE [11] of the FH Aachen UAS in [12]. In the present work, the acoustic measurements performed with the reverberation chamber method are discussed. These serve as preparation and reference to the planned measurements on the optimized propeller.

The paper is organized as follows: First, the considered approaches to the turbine operation of propellers are introduced, followed by the documentation of the experimental setup. Subsequently, the method used to analyze the measurements is described before the results are discussed in conclusion. The acoustic values determined are compared with the flight performance data of the drive system to make differentiated statements about possible operating configurations.

2 Approaches to in-flight recuperation

Equivalently to its normal operation, depicted in Figure 1a, a propeller can be operated in turbine mode under several conditions and with various settings. Essential parameters are the rotational speed n, the inflow or flight velocity \(u_{\infty }\) and the blade pitch angle \(\beta\). Within the given project, three different approaches were identified, achieving a driving momentum acting from the propeller onto the shaft [12]. In the first case, the propeller is treated as a fixed pitch propeller, the setting of which can not be altered during flight. Hence, a change in circumferential force direction is only to be achieved by reducing the rotational speed, as shown in Figure 1b. In comparison to the trust mode, the resulting inflow vector w has now a negative resulting angle of attack \(\alpha\) (AoA) to the propeller blade and produces drag instead of thrust. For the second case, a negative AoA is achieved by altering the blade pitch to negative values to possibly exploit higher circumferential speeds, which is illustrated in Figure 1c.

Considering the flow around the leading edge of the profile as well as the desirable deflection of the flow, it is obvious, that the orientation of the camber is unsuitable. Therefore, in the third case, the orientation of the profile is flipped, compared to the thrust pitch, while the direction of rotation is reversed. As shown in Figure 1d, the profile camber is now more appropriate for deflecting the inflow properly. However, a propeller blade’s twist over the radius r is oriented such that the profile is pitching down from root to tip \(\left( \mathrm {d}\beta (r) / \mathrm {d}r<0\right)\). In contrast, wind turbine designs exhibit the opposite. With respect to this geometrical inhibition, optimal flow conditions for the profile are limited to a certain radius range of the propeller blade. To generate as much torque as possible, the pitch angle should be set in a way such that the profile flow is smooth from the propeller tip inwards and then stalls at a certain point due to a steep AoA. Depending on the individual propeller, this point is located at around 60 % of the tip radius \(\left( D/2\right)\) [12].

Fig. 1
figure 1

Schemata of velocities, forces, and angles for a propeller blade profile in several operating modes

3 Experimental setup

3.1 Propeller and drive train

The investigated propeller, which was designed for paramotor aircraft by the manufacturer Helix Carbon GmbH, has the model designation H30V 1,00 m R-S-2. It is a two-bladed variable pitch propeller made of a fiber composite material and was chosen for these experiments due to its low overall blade twist (12.5\(^\circ\)) to achieve larger areas of attached flow during windmilling. The diameter of the propeller was shortened from the original value of overall 1.2 m to 1 m for the experiments due to wind tunnel size and ground clearance when mounted to the flying testbed. The operating limits of the propeller are at a speed of 4300 rpm and an engine power of 25 kW. For the wind tunnel experiments, a prototype framework was constructed to house the drive train for the propeller, shown in Figure 2.

Fig. 2
figure 2

Experimental setup of the drive train

An electric motor with the designation HPD20 by the manufacturer Geiger Engineering GmbH is used as the drive. It is a three-phase synchronous motor with external rotor, which can be used both as a prime mover and as a generator. According to the manufacturer, the motor delivers a rated continuous power of 20 kW at a speed of 2650 rpm and a maximum torque of 105 Nm. A suitable motor controller called MC300 from the same manufacturer is used to control the motor. The power supply and storage is provided by the lithium-ion accumulator 14S/20P of 3.1 kWh, also by Geiger Engineering GmbH.

3.2 Measurement equipment and facility

To assess the propeller performance, axial and torque measurements needed to be conducted in addition to the data collected by the motor controller. This ensured that thrust, respectively drag, and mechanical power could be recorded independently from any electric efficiency factors. As depicted in Figure 2, the compactness of the engine allows it to be mounted directly onto the load cells, which consequently measure the reaction forces without any bearing friction. These load cells are two combined strain gauge sensors by the manufacturer ME-Meßsysteme GmbH: the TS110a torque sensor which is designed for torques up to \(\pm {100}\,\mathrm{Nm}\) and the KR110a force sensor which has a nominal force of \(\pm {1000}\,\mathrm{N}\), both being of the accuracy class 0.1 % of their full scale. The sensors were operated with the GSV-4USB M12 measuring amplifier that is of the accuracy class 0.05 % and is equipped with a 16 bit A/D converter.

The entire drive train was fixed on the USK propeller test stand, positioned and aligned in the test section of the closed-loop wind tunnel, as shown in Figure 3. The circular nozzle outlet has a diameter of 1.5 m through which the facility is able to generate an open jet flow of up to 70  \(\frac{\mathrm{m}}{\mathrm{s}}\) with an turbulence level of around 2.5 %. Using a smoke probe, it was ensured that there is no interaction between the propeller and the turbulent shear layer of the open jet in any of the chosen operating points. With a shaft height of 2.34 m above the ground and the length of the test section of 3 m in a large plenum, the effects of the propeller’s surroundings on the flow and the measurements is considered to be negligible, even for static thrust tests.

Fig. 3
figure 3

Wind tunnel facility at ILR including experimental setup

Compared to the free jet, the plenum of the test section is a large chamber with irregular, sonically hard surfaces and is therefore considered suitable for the utilization of the reverberation method for the acoustic measurements. The procedure for performing measurements according to the reverberation chamber method is specified in the international standard DIN EN ISO 3741 [13]. Using this method, the sound power level \(L_W(f)\) of a frequency f emitted from an acoustic source can be calculated from the measured sound pressure level \(L_p(f)\) with the following equation:

$$L_W(f) = L_p(f) + 10 \, {\mathrm{lg}} \left( \frac{V}{T_s(f)}\right) - {6}{\mathrm{dB}}. $$
(1)

Here, V accounts for the plenum volume, while \(T_s(f)\) is the determined reverberation time of sound with the considered frequency within the room. The reverberation times used were taken from an established data set for this facility. For the acoustic measurement, a microphone of the model MI-17 from the manufacturer ROGA-Instruments was used. This is an \(\frac{1}{4}''\) IEPE (Integrated Electronics Piezo-Electric) free-field microphone of accuracy class 1. The maximum sound pressure level is 130 dB with an inherent noise of 36 dB. The frequency response is \(\pm {0.5}\,\mathrm{dB}\) for 20 to 4000 Hz and \(\pm {1.5}\,\mathrm{dB}\) for 420 kHz. The measuring microphone was driven by the Plug.n.DAQ data acquisition system with the maximum sampling rate 48 kHz with a resolution of 16 bit. Its signal-to-noise ratio is \(>{90}\,\mathrm{dB}\) and the absolute accuracy is 2 %. The microphone calibration was performed regularly with a SV 30A pistonphone by SVANTEK. To cancel out the influence of room modes, the microphone is mounted onto a motor-driven rotating boom. Installed about 4 m next to the tunnel axis, compare Figure 3, it allows vertical circular movement with a diameter of 6 m and a circulation time of 50 s. Due to the measurement method in the diffuse sound field of the reverberation chamber, the distance to the sound source is irrelevant.

3.3 Procedure

The operating points for the measurement campaign were defined by the aforementioned four modes: thrust, recuperation with fixed blade pitch (case 1), recuperation with reduced pitch (case 2), and recuperation with flipped blades and reversed direction of rotation (case 3). A quick overview is given by Figure 4, were all points are characterized through the inflow velocity \(u_{\infty }\), the rotational speed of the propeller n and the nominal blade pitch angle \(\beta\). The latter was measured at the tip of the blade as a distinct reference. During the measurements of the recuperation cases, no predefined rotational speed could be targeted because solely a breaking torque could be set in the engine control system. Therefore, the resulting speed was the outcome of the respective force equilibrium. To adapt this, the procedure was such that the breaking torque, which could be set through the motor controller, was increased successively for a constant inflow velocity. Hereafter, the latter was altered for the next set of operating points with several breaking torques.

Fig. 4
figure 4

Totality of the approached measuring points

To account for background noise, several measurements were conducted without mounted propeller blades for all targeted inflow velocities and rotational speeds.

4 Data postprocessing

The presented results within this contribution focus on the acoustic measurements; therefore, a detailed insight of the data processing procedure is given in the following section.

In addition to the reverberation room processing, the measurement sound pressure levels had to be filtered selectively for background noise. The latter is mainly characterized by the wind tunnel, which emits a high level of broadband noise coupled with distinct tonal peaks, as displayed in Figure 5. However, the tonal peaks of the actual measurement of a propeller operating point, designated as “signal”, significantly overtop the noise levels so that they can be used for further evaluation. Close peaks, like those at around \(f = {175}\,\mathrm{Hz}\) in the example given, are clearly distinguishable, whereas the signal peak at \(f = {330}\,\mathrm{Hz}\) will be downgraded by the following corrections. Since a sound level is dominated by its loudest component, it is considered sufficient for this comparative studies to solely include these tonal peaks from the signal for the summation of the overall sound power level for the considered operating point.

Fig. 5
figure 5

Comparison of typical signal and noise frequency spectra (here: case 3, \(u_{\infty } = {22.5}\,\frac{\mathrm{m}}{\mathrm{s}}\), \(\beta = {163}^\circ\), \(n={2266}\,\mathrm{rpm}\))

The resulting data processing is depicted in Figure 6. In an upstream step, the frequency spectra of the measurements were calculated, following Welch’s method with a Hann window, an overlap of 50 % and a frequency resolution of 1 Hz. The provided spectra of signal and noise are imported and compared. The processing script detects maxima in the signal of the measurement and selects them as potential fundamental tones. To identify the correct peaks, the noise distances of the levels at these fundamentals and their harmonics are calculated and summed up. The peak group with the largest sum of noise level distance is selected as belonging to the genuine fundamental peak within the measurement signal. The levels of the tonal peaks are corrected for the influence of the extraneous level and the reverberation chamber according to DIN EN ISO 3741, represented with equation 1. However, it must be stated that the extraneous noise requirements of this standard have not been met, which will be accounted for using error bars in the result figures. After A-weighting the corrected peaks, the final step of level summation is carried out to form the overall sound power level of the source as a comparison value of the measurements.

Fig. 6
figure 6

Simplified processing procedure for acoustic measurements to yield overall sound power levels

In addition, an error estimation is performed as a function of the extraneous level distance. This adds up to the aforementioned uncertainties of the measurement equipment and accounts for lower signal-to-noise ratios. The resulting uncertainties are \(\pm {0.8}\,\mathrm{dB}\) for larger separation from extraneous levels, but reach higher absolute values for the negative error limits of up to \({- 10}\,\mathrm{dB}\) for measurements with very small extraneous level separation.

5 Results and discussion

In this section, the results of the measurement campaign are presented, focusing on the acoustic aspects. The propeller performance, also in comparison to a numerical calculation model, is considered in more detail in [12].

5.1 Thrust mode

As a reference, the more common thrust case was tested with a constant rotational speed of 2600 rpm. The results showed good agreement with established propeller theory and the numerical performance calculations for this specific propeller [12]. Summarized in Figure 7, the calculated sound power levels \(L_W\) and the propeller efficiency \(\eta _P\) are plotted over the advance ratio J on the left-hand side and over the power coefficient \(c_P\) on the right-hand side for several blade pitch angles \(\beta\).

Fig. 7
figure 7

Thrust mode: Comparison of the determined sound power level \(L_W\) with the propeller efficiency \(\eta _P\), related to the advance ratio J and the power coefficient \(c_P\) at \(n = {2600}\,\mathrm{rpm}\)

A pitch of \(16^\circ\) being more of a climbing setting, shows higher propeller efficiencies at the lower advance rates, while the \(22^\circ\) setting has a higher efficiency for larger advance ratios. The latter setting is able to reach higher power coefficients as it transfers the most power into the propulsion of the fluid. Intuitively this setting turns out to be the loudest, considering the dependence on the advance ratio. Here, the different plots of \(L_W\) are quite clearly distinguished from each other. With the value of J decreasing, hence with decreasing inflow, respectively flight velocities, the sound power level rises above 100 dB and reaching the highest value for static thrust, as the strongest velocity gradients are to be expected in this point of operation.

With decreasing \(\beta\), the sound power level is reduced. As can be seen here, with the lower signal-to-noise ratio, the measurement uncertainty increases in the direction of lower values.

Considering the dependence of the sound level on the power coefficient, it is recognizable that the pitch angle does influence how much power the propeller is able to transfer; however, it has no significant impact on how loud the propeller operates.

5.2 Recuperation case 1

By reducing the rotational speed, down to between 1070 to 2310 rpm as depicted in Figure 4, the propeller was transferred into turbine operation. Recuperation case 1 uses the same pitch angle range as the reference trust tests and is documented in Figure 8. Here, the transferred power is represented in the power coefficient \(c_p\) on the lower ordinate. Furthermore, the influence of the tip Mach number \({\mathit{Ma}}\) is considered on the right-hand side, since the rotational speed would not be kept constant anymore.

From an acoustical point of view, the maximum values of the sound power level are 5 dB lower for this case than for the thrust mode. The pitch angle setting of \(20^\circ\) stands out in particular, since its sound power level is around 10 dB lower than that of the other settings, which means a halving of the sound pressure. A first hypothesis for this significant reduction is that for this propeller and with this pitch angle setting, the flow around the entire blade is well attached and at the same time creates a uniform, smooth loading of the propeller disc. However, this needs to be confirmed with further investigations using qualitative and/or quantitative flow visualization techniques.

Fig. 8
figure 8

Recuperation case 1: Comparison of the measured sound power level \(L_W\) with the power coefficient \(c_P\), related to the advance ratio J and the blade tip Mach number \({\mathit{Ma}}\)

Along with this lower noise emissions, a similar amount of power is recuperated, as can be seen in the lower graph. This indicates some optimization potential with regard to the noise impact of aircraft in decent with recuperating propellers and is to be studied in more detail during the planned flight tests within this project.

Just as the tip Mach number itself, its impact on the emitted sound power is quite low the in the considered range. However, a slight proportional tendency can be observed.

5.3 Recuperation case 2

As can be seen in Figure 4, the measurement points obtained with a further reduced pitch angle for case 2, led to an operation at lower rotational speeds (780 to 1600 rpm) than for case 1. In comparison, the advance ratios stay similar, however the power coefficient does not reach values as high as before, which can be observed in Figure 9. Consequently, within a certain interval around an advance ratio of 2.5 the noise emissions are small as well, comparable to the results for \(\beta = {20}^\circ\) in case 1. Nevertheless, values rise for lower advance ratios, while for larger values of J, \(\beta = {1}^\circ\) yields an operation point of relatively low noise and comparatively good power coefficient.

Considering the influence of the tip Mach number, the impact seems more distinct, since the emitted sound power level rises quite clearly for the two higher pitch angles of \(7^\circ\) and \(9^\circ\). In the \(c_P\) plot, the several steps of successively increasing the breaking torque setting in the motor controller are clearly visible: for the same inflow velocity, the this leads to lower rotational speeds, hence lower Mach numbers and an increasing power coefficient and more recuperated power. Increasing the inflow speed shifts the curves almost horizontally toward higher Mach numbers, while revealing that the higher \(\beta\) settings are more suitable for higher flight speeds.

Fig. 9
figure 9

Recuperation case 2: Comparison of the measured sound power level \(L_W\) with the power coefficient \(c_P\), related to the advance ratio J and the blade tip Mach number \({\mathit{Ma}}\)

5.4 Recuperation case 3

In case 3, only a few stable operating points could be established. The reason for this was the motor control system, which only allowed braking torques to be set up to a certain limit until they collapsed, particularly at high speeds. As can be seen in Figure 10, it was possible to carry out series of measurements for the setting angles 163\(^\circ\) and 165\(^\circ\).

It is particularly noticeable that the speeds achieved here are considerably higher, up to 2600 rpm, respectively, the corresponding advance ratios are lower. As far as the sound power level is concerned, it resides again in the range of case 1. Due to the lack of data points, no clear tendency for the dependence on rotational speeds and Mach numbers is determined. Nevertheless, higher absolute powers are converted. The power coefficient turns out to be lower due to the high rotational speeds. This power conversion matches the expectations, since with a two-blade rotor the optimum operating range is at high tip-speed ratios. At lower tip-speed ratios, three- or four-blade propellers have an advantage, which is why they can probably be more suitable for cases 1 and 2 and should be considered for further research in this field.

Fig. 10
figure 10

Recuperation case 3: Comparison of the measured sound power level \(L_W\) with the power coefficient \(c_P\), related to the advance ratio J and the blade tip Mach number \({\mathit{Ma}}\)

6 Conclusion and outlook

In the present study, a conventional propeller with an electric drive system was investigated in a subsonic wind tunnel. The focus lied on identifying differences in performance and noise emissions for several points in thrust and recuperation operation. Three approaches to recuperation were investigated: Fixed pitch propeller at speed reduction (case 1), controllable pitch propeller with negative angle of attack (case 2), and inverted, backward running propeller (case 3).

Using the reverberation chamber method, the emitted acoustic powers were determined and related to the values collected using the load cells and the motor control system. Although the acoustic measurements in the subsonic wind tunnel were prone to considerable uncertainties due to the background noise, it was nevertheless possible to obtain results relating to the emitted sound power by selecting the tonal peaks of the signal.

From these results, it can be seen that for the propeller under investigation, the pitch angle in thrust mode has no significant influence on the acoustic emission as a function of the power converted. However, with respect to the recuperated power and the acoustic power, tendencies could be derived for the three cases investigated. It became clear that the setting angle has a significant influence on the acoustics in the first case and must therefore be selected correctly when designing or choosing a fixed pitch propeller that is also intended for recuperation operation. This implies that a compromise may be required between the acoustics in that state and the propeller thrust performance in the design operating point. In the second case, there was hardly any difference between the investigated setting angles in terms of sound power. These levels were lower than for the remaining cases, but the converted power was also the lowest. The third case shows potential for the highest energy recovery for a two-bladed propeller, while sound emissions are also at a high level. Compared to the investigated thrust case, the recuperation operating points enabled with this experimental setup overall showed sound power levels smaller by at least 5 dB.

The distinct difference of recuperation case 3 from the others reveals the significant influence of the tip-speed ratio on the recuperated power. By varying the number of blades, another parameter for optimizing the recuperation operating points shall be investigated in the future, since the pitch angles of the first and second cases tend to provide good conditions for higher blade numbers. As a next step within the given project, the presented wind tunnel measurements will be complemented with flight tests and fly-over measurements, beginning with two-bladed propellers. In addition, the acoustic measurements will be utilized to validate the newly implemented acoustic modeling within the PropCODE environment to further improve propeller design capabilities.