large scale inflow distortions as a source mechanism for ... · a bell mouth type inlet nozzle with...

19th AIAA/CEAS Aeroacoustics Conference (34th AIAA Aeroacoustics Conference), 27 - 29 May 2013, Berlin

American Institute of Aeronautics and Astronautics

1

Large Scale Inflow Distortions as a Source Mechanism for Discrete Frequency Sound from Isolated Axial Fans

Michael Sturm1 and Thomas H. Carolus2 Instiut für Fluid- und Thermodynamik, Universität Siegen, 57068 Siegen, Germany

The acoustic performance of isolated axial fans is strongly depending on the quality of the inflow. Current standards for the acoustic measurement of industrial fans require a certain undisturbed volume upstream of the fan. However, unexpected discrete frequency sound indicates that this may not be appropriate for reliable acoustic measurements. In this study, the flow field in the immediate vicinity of the fan rotor as well as in the entire surrounding room that is much larger than the undisturbed volume recommended by the standards is examined. Smoke visualizations and RANS simulations reveal that large scale flow structures develop in the room. This moderate flow in the room and not e.g. rotor self-induced or tip clearance vortex-type structures give rise to the inflow distortion in the fan intake observed. This hypothesis is confirmed by measurements with a hemispherical flow conditioner placed relatively far upstream of the fan rotor which homogenizes the inflow and reduces the fan's tonal noise considerably.

Nomenclature Aenvl = enveloping surface Aref = reference surface a = speed of sound BPF = blade passing frequency ci,ideal = ideal value of the respective velocity component cm = axial velocity cr = radial velocity cref = reference velocity cu = circumferential component of absolute velocity cvol = volumetric averaged velocity d1 = hub diameter d2 = duct diameter dHFC = diameter of Inflow Control Device dFC = diameter of Flow Conditioner element dwire = diameter of coated wires Ei = energy spectrum of respective velocity component f = frequency fs = sampling frequency I = integral time scale LEi = energy spectrum of respective velocity component in level form Lp5 = sound pressure level inside the anechoic chamber LW4 = sound power level inside the duct LW45 = sound power level averaged from LW4 and LW5 LW5 = sound power level inside the anechoic chamber lFC = length of Flow Conditioner element Ma = circumferential Mach number MSE = mean squared error N = number of measurement points 1 Ph.D. student, Institute for Fluid- and Thermodynamics, michael.sturm@uni-siegen, corresponding author 2 Professor, Institute for Fluid- and Thermodynamics, [email protected]



2

n = rotational speed p0 = reference pressure r = radius

m mc cR = autocorrelation of the axial velocity cm

Re = Reynolds number Str = Strouhal number based on BPF s = tip clearance TI = turbulence intensity t = time u = circumferential velocity V = flow rate w = relative velocity wdist = disturbed relative velocity wup = upwash velocity y+ = dimensionless wall distance z = number of blades β = flow angle βdist = disturbed flow angle Δf = frequency resolution Δfref = reference frequency resolution ΔL = distance between the blades Δr = distance between the measurement points in radial direction Δρ = distance between the measurement points in circumferential direction ax = axial turbulent length scale υ = kinematic viscosity ρ = circumferential coordinate φ = flow rate coefficient ρc = impedance (ρc)0 = characteristic impedance σ = solidity of wire mesh τ = time lag

I. Introduction he acoustic spectrum of an industrial axial fan is often dominated by tonal sound at blade passing frequency (BPF). Figure 1 shows the acoustic spectrum of a typical fan rotor operated with upstream struts, downstream

guide vanes or isolated, i.e. without any disturbing devices, and running with very low circumferential Mach number. The fan is balanced to high precision and installed on a standardized test rig with a perfectly symmetric inlet nozzle and an undisturbed intake domain of the size much larger as recommended by the current standards (two fan diameters in each direction1,2), see Figure 2. While the existence of BPF tones of a fan with guide vanes or struts up- or downstream is explained by the classical rotor/stator interaction mechanism, the origin of the tones in case of the isolated rotor is not expected. According to well known models3-5, tones at BPF should not occur, when (i) the rotor is completely isolated, i.e. struts and guide vanes are non-existent or very far up- or downstream, (ii) the inlet geometry is symmetric and (iii) the rotor is rotating at a low characteristic circumferential Mach number. Then the volume displacement by the moving blades and the unavoidable steady loading which cause monopole and Gutin dipole sound, respectively, are irrelevant. It rather requires additional unsteady forces on the blades to explain the existence of tonal dipole sound. One potential reason for that kind of unsteady forces is a disturbed inflow.

The relationship between the BPF tones and spatial inflow inhomogeneities has been studied for many years. Most authors studied the impact of artificial inflow distortions caused by upstream obstructions like a simple rod6-8, struts9, turbulence screens10-11, rectangular obstructions12 or stator/rotor interaction13. On the other hand, Hanson6 found that eddies sucked into the fan from the free atmosphere may act as acoustic sources alike. The motivation of his study was the existence of the BPF tones when aero engines are measured acoustically under static conditions on the ground, while the tones do not occur under in-flight conditions. One consequence was the development of hemispherical inflow control devices which should reduce the incoming disturbances and reproduce in-flight conditions at the test rig on the ground. However, the current standard ISO 13347 for acoustic measurements of airborne noise emitted by industrial fans does not suggest any inflow conditioner at all1. By referring to the

T



3

appropriate revised standard DIN EN ISO 5801 for aerodynamic test facilities, it rather specifies an undisturbed volume of two fan diameters in each direction upstream of the fan2. As suggested by the unexpected tones in fan spectra this might not be sufficient for comparable and reliable acoustic measurements.

10.8

d 2

15d2

11.67d2

flow inlet

fan section withdiameter d

2

Figure 2. Fan test rig in the University Siegen with large room (anechoic chamber); the red box indicate the recommended undisturbed volume of two fan diameter in each direction of the inlet according to ISO 133471.

The hypothesis of this study is that the "free" inflow to the fan from a large-sized room as depicted in Fig. 2 already shows features of a 'turbulent" atmosphere with large scale flow structures as described in Ref. 6. Hence, the objective of this paper is to study the flow field not only in the immediate vicinity of the fan rotor but in the entire large-sized room. A further concern is the clarification of how possible large scale flow structures in the room affect the inflow inside the comparably small duct, in which the fan is installed. A disturbed inflow to the fan is a potential source of tonal sound at blade passing frequency.

For an initial picture of the flow field at the intake a simple experimental flow visualization technique is conducted. By simulating the flow in the large-sized room via a Reynolds-Averaged Navier-Stokes approach, the possible generation of large scale flow structures and its effect of the flow in the fan's intake is studied. The resulting inflow profile in the intake is validated experimentally by three dimensional hot wire measurements. Other candidates for BPF tones are secondary flows such as the rotor tip clearance flow or the vortex system in the hub

0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

50

60

70

80

90

Strouhal number

Lp,

dB

isolated fanupstream strutsdownstream guide vanes

Figure 1. Acoustic spectra of a typical fan operated in a large room according to Figure 2; three configurations: isolated fan, rotor with upstream struts and rotor with downstream guide vanes.



4

region. In order to distinguish ingested inflow distortions from those phenomena, the effect of two upstream inflow conditioners is examined. If these flow conditioners have a remarkable impact on the emission of tonal sound, it is rationally concluded that the disturbed inflow, induced by the large-scale environment, is the relevant source mechanism for discrete frequency sound at blade passing frequency.

II. Methodology

A. Test rig and fundamental experimental techniques The standardized test rig used in this study is shown schematically in Fig. 3. The fan takes the air from a large

room (here the semi-anechoic chamber at the University Siegen) and exhausts into a duct with an anechoic termination. The flow inlet is located off-center in the reverberant floor. The operating point is controlled by a throttle downstream of the termination, while the flow rate is determined by a calibrated hot film probe in the duct. The low pressure fan unit contains five cambered and swept blades which are designed with an in-house design software for axial fans. The rotor is manufactured and balanced with very high precision to avoid any non-aeroacoustic sound sources. A bell mouth type inlet nozzle with a ¼ rotor diameter radius is employed. Thin supporting struts are positioned one rotor diameter downstream of the rotor. No other obstructions are present. Table 1 shows the main fan characteristics.

semi-anechoic chamber(reverberant floor)

nozzle

microphonposition

flow inlet

anechoic termination

throttle

in-duct microphoneflow

straightener

fan

duct (d2 = 0.3 m)

Figure 3. Top view of the standardized test rig for acoustic measurements (not to scale).

Table 1. Main characteristics of the fan.

Duct diameter d2 0.3 m Hub diameter d1 0.135 m

Number of blades z 5 Tip clearance ratio s/d2 0.1 %

22Re /d n at d2

9.36·105 Circumferential Ma at d2

2 /Ma d n a 0.139

Rotational speed n 3000 min-1 Design flow rate optV 0.65 m3/s

Design flow rate coefficient 2 3

24 / ( )opt optV d n 0.195

The acoustic measurements are conducted by using four microphones located as indicated in Fig. 3. The time signal of the sound pressure is measured by three microphones (Brüel & Kjaer type 4190) with a radial distance of 1.3 m and an angular spacing of 35° from the axis of rotation and captured with a sampling frequency fs = 25.6 kHz. The signal analysis is based on the power spectral density which was obtained by the function pwelch in MATLAB

TM R2012a. The parameters chosen for pwelch were window = hann(nfft), noverlap = 0, nfft = length(time signal)/30. The spectra from the windows were averaged, their final frequency resolution is Δf = 1 Hz. For all levels, the reference pressure is p0 = 2·10-5 Pa. All measurements were performed at a rotational speed of n = 3000 min-1 and at the fan design operating flow rate coefficient φopt = 0.195.



5

The sound power radiated into the free field on the suction side of the fan is derived from the averaged sound pressure measured at several positions on an enveloping surface around the intake, i.e.

5,

30.1 ( ) 0

51

( )1( ) 10log 10 10log 10log

3p iL f envl

wi ref

A cL f

A c

. (1)

The duct sound power is measured by a microphone downstream the fan section. In order to compensate for pseudo sound and duct mode effects, the microphone is equipped with a slit tube and a nose cone (as recommended in Ref. 14). The sound power inside the duct Lw4 is in principle calculated analog to Eq. (1), but with just one sound pressure signal and corrected according to DIN ISO 513614. The averaged sound power level then becomes

4 50.1 ( ) 0.1 ( )45 ( ) 10log 10 10w wL f L f

wL f . (2)

All acoustic spectra are depicted in terms of the Strouhal number

/Str f BPF , (3)

where BPF n z is the blade passing frequency.

B. Flow visualization To visualize the flow at the intake of the fan, an array of six horizontal wires has been stretched in front of the

nozzle, see Fig. 4. The wires were coated with a glycerin/water mixture and heated electrically. The wires are made of brass alloy and have a diameter of dwire = 0.25 mm. The visualization device is positioned 0.1 m upstream the nozzle. At this position the Reynolds number based on the wire diameter and the axial velocity is

Re 30m wired

c d

(4)

and hence is in the range of laminar flow in typical flows around cylinders15. Thus, no wakes should be induced by the wires and the flow disturbance arising from wires is minimized. A high quality camera was used to record a film of the generated streamlines. To capture a larger area around the intake and not only single streamlines, a smoke generator was also used to produce smoke over a wide region around the intake.

U

Figure 4. Experimental set-up for flow visualization at the intake of the fan section.

C. Numerical set up To examine the flow conditions inside the large-sized room, simulations were conducted using the commercial

solver ANSYS FLUENT 14.0. The geometric situation in its original dimensions of the test rig was used to set up the



6

computational domain, see Fig. 5. In contrast to Fig. 3, the duct is considered only 3.5 diameters downstream the nozzle and the hub is neglected for the sake of simplicity. Only the effect of the empty semi-anechoic chamber is assessed. Other structural elements such as supports for instruments etc. are potential candidates for more vortex-like flow structures, but are not considered in this study. The computational domain is meshed with a block structured grid of 1.45 million cells. At the flow inlet the mass flow rate is set such it corresponds to the fan design point. At the outlet the static pressure is set to ambient conditions, i.e. relative pressure of 0 Pa. All other boundaries are no slip walls. All simulations use the statistically steady RANS equations with the realizable k-ε turbulence model and the Enhanced Wall Treatment to allow the present range of the dimensionless wall distance 1.5 < y+ < 155. The solver uses a 2nd order discretization scheme for all variables. The flow is treated as incompressible. Convergence is assumed if either the RMS residuals with respect to conservation of mass, momentum and turbulence are lower than 10-6 or if flow field quantities remain constant over numerous iterations (within a tolerance of +/- 5%).

air suckingduct

flow inlet

Figure 5. Computational domain (left) and numerical grid (right) for the simulations of the large room.

D. Inflow conditioner and hot wire anemometry Two inflow conditioners were used (see Table 2) to homogenize the inflow and allow an investigation of the

influence of the inflow conditions on the acoustic emission. The Hemispherical Flow Conditioner (HFC) consists of a combination of a fluid conditioner structure, which reduces the lateral turbulences, and a downstream layer of wire mesh to reduce axial disturbances16. Other geometric dimensions are shown in Table 2 and based on recommendations in the literature16-18. The hemispherical device can be flange mounted to the nozzle of the test rig, see Table 2. A second inflow conditioner is a plane layer of small tubes, inserted into the duct upstream the fan (Tubular Flow Conditioner, TFC). The axial position of the TFC with respect to the rotor is shown in Fig. 6. Preliminary acoustic studies showed that the self noise of the inflow conditioners is negligible.

The flow field in the intake is measured in the measurement plane according to Figure 6 by hot wire anemometry. To capture spatial variations of the velocities a cross section of the duct 0.96d2 upstream the leading edge of the fan is scanned by a 3D hot-wire probe (Model 1299A from TSI Incorporated). The probe consists of three wires arranged crosswise and is connected with a universal anemometry system Streamline 90N10 from Dantec Dynamics. The anemometer operates in a constant-temperature mode. The measured signals are temperature corrected by using a temperature probe A8B from Dantec Dynamics. A 24-bit data acquisition system (Module PXI-4495 from National Instruments) is used, with which up to 16 channels can be measured synchronously. The probe is calibrated in a low turbulence wind tunnel.

A slot extending over the half of the circumference is cut in the duct, so that the probe can protrude into the duct. Preliminary investigations have shown that the influence of this slot on the flow is small. The orientation of the polar coordinate system for the hot wire measurements is indicated in Fig. 6 (right). The probe is attached to an automatic three axis positioning system from Isel, which has a positioning accuracy of 0.02 mm. The measurement points in the cross sectional plane of the duct have an angular distance of Δρ = 5° and radial distance of Δr = 5 mm, respectively. Together with the center of the plane, this leads to 2017 measurement points. All measurements are conducted over a time interval of one second with a sampling rate of fs = 25.6 kHz.

The inflow velocity profiles will be evaluated in terms of the axial velocity cm, the circumferential velocity cu and the radial velocity cr, according to the coordinate system indicated in Fig. 6. Due to the fact that the data of the hot wire measurements are not recorded simultaneously, only time averaged data will be evaluated assuming a



7

statistically stationary state. All velocity components are made non-dimensional by the volumetric averaged velocity inside the duct,

2

4vol

Vc

D

. (5)

Table 2. Inflow conditioners.

HFC

diameter ratio of the device

2/ 2.84HFCd d

wire mesh

Flow Conditioner

anechoic foam

flow direction

dH

FC

solidity of the wire mesh 2 21 / ( ) 0.44b a b

length ratio of the flow conditioner element

/ 10FC FCl d

TFC

length ratio of the flow conditioner element

F/ 10FC Cl d

supporting struts

inlet nozzle

flowdirection

fan blades

d 2

X Y

hot wire probe

connection to positioning

device

r

rx

measurement plane for hot wire measurements

position of tubular flow conditioner

Figure 6. Lateral view of the axial fan section investigated (left; X = 0.96d2, Y = d2) and top view (right).

dFC

l FC

lFC

dFC



8

Furthermore, the so called upwash velocity wup will be evaluated. Commonly, a fan blade is designed for a specific angle of attack , which varies in spanwise direction. When the blade moves through a spatially inhomogeneous inflow velocity field, the blade experiences changes in the angle of attack Δwhich leads to periodic blade force fluctuation due to the rotation of the blade and hence tonal sound. The change of the angle of attack is directly proportional to a change in the flow angle , i.e. Δ=Δ=dist. Due to the mismatch between the disturbed flow angle βdist and the blade angle, the upwash velocity occurs, which is the component of the relative velocity wdist perpendicular to the blade. In Fig. 7, this is shown for the case of an inflow with a co- and counter-rotating swirl.

u u

cm

cm

cu

cu

w ww

up

wup

wdist

wdist

dist

dist

Figure 7. Velocity triangles at an arbitrary radial section showing the occurrence of the upwash velocity wup

in case of a co-rotating swirl (left) and a counter-rotating swirl (right).

Referring to Fig. 7, the upwash velocity wup becomes

sinup dist distw w (6)

with the flow angle β

1tan mc

u

, (7)

the disturbed flow angle βdist

1tan mdist

u

c

u c

(8)

and the disturbed relative velocity wdist

2 2( )dist m uw c u c . (9)

In addition to the mean velocity components, the statistic parameters of the turbulent inflow will be determined experimentally. These are the turbulence intensity

' 2 ' 2 ' 21

3 m u r

vol

c c cTI

c

, (10)



9

where ... denotes the time-average and the turbulent axial length scale ax, is determined by auto correlation method taking into account the assumption of frozen turbulence, i.e.

mct z

. (11)

The autocorrelation of the axial velocity becomes

' '

' 2

( ) ( )m m

m m

c c

m

c t c tR

c

. (12)

Here, τ is the lag of the correlated time signal. By integrating the autocorrelation m mc cR over the time lag τ one obtains

the integral time scale I

0

m mc cI R d

(13)

and finally yields the turbulent axial length scale ax by multiplying the integral time scale I with the time-averaged axial velocity mc ,

ax mI c . (14)

In order to analyze the frequency content of a measured velocity signal, a 1-D energy spectrum is defined such that

'2,( )

( ) i rmsi

d cE f

df , (15)

which is equivalent to the one-sided spectrum of the power spectral density calculated with the MATLAB routine pwelch. The energy spectra are depicted in level form using the definition

2

( ) 10log i refEi

ref

E fL f

c

[dB]. (16)

The reference velocity cref is chosen as 1 m/s and the reference frequency band width Δfref 1 Hz.

III. Results

A. Flow conditions inside the large-sized room In Fig. 8, streamlines indicated by smoke released from (i) an arbitrary position in the room (left) and (ii) from

the heated wires (right) are shown at arbitrary time instances. Helical streamlines indicate vortex type flow structures entering the fan section. Position and size of these structures vary with time. However, closer observation proves that the characteristic time scale of this variation is very small as compared to the rotational speed of the rotor, i.e. the flow structures are quasi-stationary for many revolutions of the rotor. Blades cutting through these flow structures would encounter periodic blade forces fluctuations and hence radiate tonal sound.

The RANS computed streamlines, Figure 9 confirm these findings. Flow particles starting at inlet on the ground of the chamber (with less than 1 m/s flow velocity!) circle around in the room until they are caught by the fan intake. Due to the deflection of the flow at the top of the room a pair of large scale vortices develops. This type of flow pattern has also been observed by other tests with flow visualization techniques. While the helical streamlines enter the duct-type fan inlet in a helical way, the streamwise structures are stretched, resulting in a reduction of their



10

cross-section. Due to the conservation of momentum this leads to an amplification of streamwise vorticity and thereby an increased circulation19. Thus, even minor inhomogeneities far upstream of the intake can have a major impact on the quality of the inflow to the fan rotor.

Figure 8. Flow visualizations at the intake of the fan section at different instances of time.

Figure 9. Streamlines inside the large room from two different points of view.

B. Flow conditions inside the duct The large-scale room induced flow structures now enter the intake of the fan. Figure 10 shows the RANS-

computed velocity profiles as well as the hot wire measurements in terms of contour plots of the averaged velocity in polar coordinates. As expected, both the simulation as well as the hot wire measurements reveal a large scale left-turning vortex flow while a smaller counter rotating vortex exists in the right part of the duct. Based on this qualitative agreement, it can be assumed that the flow conditions inside the large room are reproduced correctly by the conducted simulation.

C. Impact of flow conditioners The simulation and the measurements show a disturbed inflow in case of free inflow. If this disturbed inflow is

the determining source mechanism for the tonal sound at blade passing frequency, one should observe a significant impact on the tones when using inflow conditioners to homogenize the flow. Figure 11 shows the averaged acoustic spectrum for the case of free inflow and with HFC and TFC. The amplitudes of the BPF tones are highlighted by dots. As anticipated, the spectrum contains distinctive peaks at the blade passing frequency and higher harmonics, especially when no inflow conditioners are used. Clearly, any tone but not the broadband sound is affected by both inflow conditioners. Most notably, the HFC decreases the amplitude of the tones the most - the 2nd harmonic nearly vanishes. The TFC still reduces the BPF sound but not as significantly as the HFC. A moderate difference in the



11

amplitude of the tones between the HFC and the TFC can be detected. This result is a strong indicator that the disturbed inflow induced by the large scale environment is responsible for the tonal sound at blade passing frequency.

0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5

40

50

60

70

80

Strouhal number

Lw

45,

dB

free inflow

HFC

TFC

Figure 11. Acoustic spectra for different inflow conditions.

To further verify the actual influence of the inflow conditioners on the flow, Figure 12 shows contour plots of

the measured averaged velocity in polar coordinates of all configurations. For the orientation of the polar coordinate system see Fig. 6, i.e. positive values of the circumferential velocity correspond to a left-turning swirl. The velocity profiles show a high degree of non-uniformity in case of free inflow. In case of the HFC and the TFC the profiles of the circumferential and radial velocity component are much more uniform as compared to free inflow. In the case of free inflow the counter clockwise vortex structure can be observed, especially in the large area of positive circumferential velocity in the upper half of the plane. The circumferential and radial velocity components amount up to 20% of the volumetric averaged velocity. Consequently, this spatially non-uniform inflow is most likely to result in the observed distinctive peaks at blade passing frequency. From the fact that vortex structures are discernable even in the averaged flow field it becomes evident that they are more or less stationary in time. Compared to the free inflow, the high circumferential and radial velocities and hence the vortex structure can hardly be observed in case of both the HFC and the TFC. Here, circumferential and radial velocity components are ±5% of cvol at maximum. An essential difference between the two flow conditioner is that the HFC ensures a much more uniform axial velocity distribution than the TFC. The TFC homogenizes the circumferential and radial velocities but not the axial component. Presumably this is due to the specific arrangement of the flow conditioner with the downstream wire mesh of the HFC on the one hand and the flow conditioning of the HFC already at the intake outside of the duct on the other hand.

Figure 10. Velocity profiles; left: RANS, right: hot-wire measurements; axial velocity is color encoded, circumferential velocity indicated by vectors.



12

As a measure for the non-uniformity of the single velocity profiles, the mean squared error MSE is calculated,

i.e.

2

,1

1 N

i i ideali

MSE c cN

, (17)

where N is the number of measurement points, ci is one of the velocity components and ci,ideal is the corresponding ideal velocity. For the ideal velocity a block profile of the inflow is assumed as it is also used in the blade design, i.e. cm,ideal = cvol and cu,ideal = cr,ideal = 0. Table 3 shows the values of the MSE for the different configurations and velocity components. From the MSE values it becomes evident that in case of the axial velocity the velocity profile is not improved when the flow conditioners are used - the TFC even increases the non-uniformity. In contrast to that the velocity profiles of the circumferential and radial velocities are much more uniform in case of both the HFC and the TFC. Here, the TFC is even the most effective.

Table 3. MSE of the particular velocity components for different inflow conditions

cm cu cr

free inflow 0.19 0.50 0.32

HFC 0.18 0.14 0.13

TFC 0.26 0.08 0.09

Figure 12. Measured velocity profiles upstream the fan for different inflow conditions: free inflow (top), HFC (middle) and TFC (bottom); from left to right: axial velocity, circumferential velocity and radial velocity.



13

In Fig. 13 the turbulence intensity TI, the turbulent length scale ax and the upwash velocity wup are depicted in terms of contour plots in the same plane as in Fig. 12. A turbulence intensity TI of up to 5% can be observed in case of free inflow. When the two conditioning devices are used, the turbulence can be clearly reduced - in case of the HFC down to lower than 1%. The TFC reduces the turbulence intensity to a value of about 2%. Regarding the contour plots of the axial turbulent length scale, it becomes clear that the HFC reduces also the turbulent length scale, while there is hardly any difference between the TFC and the case of free inflow. A high value of the axial turbulent length scales indicates a large axial size of the eddies. Comparing the profiles of the upwash velocity wup in Fig. 13 with the plots of the single velocity components in Fig. 12, it becomes obvious that the circumferential velocity is the decisive quantity for the occurrence of the upwash velocity wup. Consequently, the upwash velocity is dramatically reduced in case of the TFC as well as the TFC, while the TFC seems to reduce the upwash velocity even a little bit more.

The upwash velocity is often an input parameter for acoustic models, since it is thought to induce blade force

fluctuations. For the occurrence of tonal sound, the blade pressure fluctuations and hence the upwash velocity have to contain periodic parts. Thus, the influence of the inflow conditioners on the spectrum of the upwash velocity should be studied. With the results of the hot wire measurements, an artificial time dependent signal of the upwash velocity can be generated by transferring the data into the moving system of blades. Due to the measurement period of one second and given the design fan speed of n = 3000 min-1, the artificial time signal corresponds to 50 revolutions. With this manipulation of the asynchronously at the various positions measured hot wire data the effect of the spatially inhomogeneities can be illuminated. Figure 14 shows the energy spectrum of the incident upwash velocity at an arbitrary circumferential position averaged over the radius for the different inflow configurations. In all cases, sharp peaks at shaft speed (Str = 0.2) and higher harmonics are clearly detectable. These peaks indicate that the blades undergo a periodic fluctuation, i.e. running through a stationary distortion each revolution. This stationary distortion is in case of free inflow the large vortex structure shown by the time-averaged velocities in Fig. 12. The occurrence of sharp peaks in the spectrum of the incident upwash velocity is somewhat expectable, but the

Figure 13. Contour plots at the measurement plane upstream the fan for different inflow conditions: freeinflow (top), inflow control device (middle) and flow conditioner (bottom); from left to right: turbulence intensity, axial turbulent length scale and upwash velocity.



14

attenuation in case of the HFC correlates to the behaviour of measured blade pressure fluctuations in a previous study20. This is an indicator that the observed blade pressure fluctuations in Ref. 20 actually originate from the inhomogeneity of the inflow and not e.g. from blade induced secondary flows such as tip leakage flows. Thus, the flow conditions inside the large-sized room far upstream have the dominant impact on the discrete frequency sound at blade passing frequency of the isolated fan rotor.

IV. Conclusion The undisturbed intake domain of the size as recommended by the current standards proved to be insufficient to

eliminate the affect of the intake environment on the generation of fan noise. Visualization by smoke and simulation by RANS showed that the flow in a room much larger than the recommended intake domain already shows features like a 'turbulent' atmosphere. The observed large scale flow structures, although associated with rather small flow velocities, cause a spatially inhomogeneous intake velocity profile which is the reason for tones at blade passing frequency produced by the isolated fan rotor. Other candidates for BPF tones such as secondary flows from the rotor tip clearance in the hub region are of negligible relevance at the operating point of the fan investigated (i.e. design point). This has been proven by the installation of upstream inflow conditioners and their remarkable effect on the inflow distortion and eventually the tones.

Two strategies seem promising in order to reduce the effect of the intake environment. On the one hand the room from where the fan intake ingests the air must be very large, which in most cases is impractical. On the other hand, a more feasible technique would be a careful conditioning of the inflow to the rotor. The hemispherical inflow conditioner, similar to those used for aero engine test, proved to be most effective.

Acknowledgments This study is funded by the “Deutsche Forschungsgemeinschaft (DFG)”. The authors gratefully acknowledge this support.

References 1ISO 13347. "Industrial Fans - Determination of Fan Sound Power Levels Under Standardized Laboratory Conditions",

International Organization for Standardization, 1st Edition, Switzerland, 2004. 2DIN EN ISO 5801. "Industrial Fans - Performance Testing Using Standardized Airways", Deutsches Institut für Normung

e.V., Beuth-Verlag, Berlin, Germany, 2011. 3Goldstein, M. E. "Aeroacoustics," McGraw-Hill, USA, 1974. 4Roger, M. "Noise in Turbomachines – Noise from Moving Surfaces" VKI Lecture Series 2000-02, von Karman Institute for

Fluid Dynamics, Belgium, 2000. 5Zhu, S.-J. and Wu, X.-J. "Research on the Farfield Tonal Fan Noise at Small Mach Number "Proceedings of ASME

International Mechanical Engineering Congress and Exposition, Chicago, Illinois, USA, 2006.

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

-70

-60

-50

-40

-30

-20

Strouhal number

LE

wup

, dB

free inflow

HFC

TFC

Figure 14. Energy spectrum of the incident upwash velocity averaged over the radius.



15

6Hanson, D. B. "Spectrum of Rotor Noise Caused by Atmospheric Turbulence" J. Acoust. Soc. Am., Vol. 56, No. 1, 1974, pp. 110-126.

7Kobayashi, H., Groeneweg, J. F. "Effects of Inflow Distortion Profiles on Fan Tone Noise" AIAA Journal, Vol. 18, No. 8, 1980, pp. 899-906.

8Chiu, W.-S., Lauchle, G. C., Thompson, D. E. "Subsonic Axial Flow Fan Noise and Unsteady Rotor Force" J. Acoust. Soc. Am., Vol. 85, No. 2, 1989, pp. 641-647.

9Lakshminarayana, B., Moiseev, N., Thompson, D. E. " Noise Due to Interaction of Inlet Turbulence and Guide Vane Secondary Flow with Propulsor" Technical Memorandum, 1975, Pennsylvania State University URL: http://www.dtic.mil/dtic/tr/fulltext/u2/a020096.pdf [cited 12 May 2013].

10Scharpf, D. F., Mueller, T. J. "An Experimental Investigation of the Sources of Propeller Noise due to the Ingestion of Turbulence at Low Speeds" Experiments in Fluids, Vol. 18, No. 4, 1995, pp. 277-287.

11Signor, D. B., Yamauchi, G. K., Mosher, M. J., George, A. R. "Effects of Ingested Atmospheric Turbulence on Measured Tail Rotor Acoustics" Journal of the Helicopter Society, Vol. 41, No. 1, 1996, pp. 77-90.

12Washburn, K. B., Lauchle, G. C. "Inlet Flow Conditions and Tonal Sound Radiation from a Subsonic Fan" Noise Control Engineering Journal, Vol. 31, No.2, 1988, pp. 101-110.

13Woodward, R. P., Elliott, D. M., Hughes, C. E., Berton, J. J. "Benefits of Swept-and-Leaned Stators for Fan Noise Reduction" Journal of Aircraft, Vol. 38, No. 6, 2001, pp. 1130-1138.

14ISO 5136. "Acoustics – Determination of Sound Power Radiated into a Duct by Fans and Other Air-Moving Devices – In-Duct Method" International Organization for Standardization, Switzerland, 2003.

15Schlichting, H. "Boundary-layer theory", McGraw-Hill, USA, 1979. 16Scheimann, J., Brooks, J. D. "A Comparison of Experimental and Theoretical Turbulence Reduction from Screens,

Honeycomb and Honeycomb-Screen Combinations" 11th Aerodynamic Testing Conference, Colorado Springs, USA, AIAA, Technical Papers, 1980, pp. 129-137.

17SAE International "Methods of Controlling Distortion of Inlet Airflow During Static Acoustical Tests of Turbofan Engines and Fan Rigs" Aerospace Information Report, 1985.

18Groth, J., Johansson, A. V. "Turbulence Reduction by Screens" Journal of Fluid Mechanics, Vol. 197, 1988, pp. 139-155. 19Greitzer, H., Tan C. S., Graf M. B. "Internal Flow: Concepts amd Applications", Cambridge University Press, USA, 2004. 20Sturm, M., Carolus, T. H. "Tonal Fan Noise of an Isolated Axial Fan Rotor due to Inhomogeneous Coherent Structures at

the Intake" Noise Control Engineering Journal, Vol. 60, No.6, 2012, pp. 669-706.

large scale inflow distortions as a source mechanism for ... · a bell mouth type inlet nozzle with...

Documents