Sei sulla pagina 1di 13

15th AIAA/CEAS Aeroacoustics Conference (30th AIAA Aeroacoustics Conference) AIAA 2009-3368

11 - 13 May 2009, Miami, Florida

Acoustic Measurement and Prediction of


Solid Rockets in Static Firing Tests

Kota Fukuda1, Seiji Tsutsumi2, and Kozo Fujii 3


Japan Aerospace Exploration Agency, Sagamihara, Kanagawa, 229-8510, Japan

Kyoichi Ui4
Japan Aerospace Exploration Agency, Sengen, Tsukuba, 305-8505, Japan

Tatsuya ishii5, Hideshi Oinuma5, and Junichi Kazawa6


Japan Aerospace Exploration Agency, Chofu, Tokyo, 182-8522, Japan

Kenji Minesugi7
Japan Aerospace Exploration Agency, Sagamihara, Kanagawa, 229-8510, Japan

Acoustic measurements are executed in two series of static-firing tests of a solid rocket
motor. The obtained data are quantitatively compared with calculation results of an
empirical prediction method, NASA SP-8072 and CFD. According to the results, the
NASA SP-8072 overestimates the sound pressure levels at the 20° and 35° points from
the jet axis in the far field, although the SPLs at other measured points are reasonably
predicted. On the other hand, the CFD calculation can clearly explain the generation
and propagation mechanism of the acoustic wave and reasonably predict the SPLs at all
the measured points. From the results, it is confirmed that the prediction accuracy of the
CFD calculation is within 5 [dB] in overall sound pressure level, which is within the
experimental uncertainty involved in the measured data, and the CFD is effective for the
prediction of both the near and the far field acoustics generated from the rocket motors.

I Introduction

T he exhaust plume from a rocket engine generates severe acoustic waves. Since the acoustic wave causes the
acoustic load on the payload, prediction and reduction of the acoustics level around launch vehicles at lift-
off is quite important design factor taken into consideration early in the design process of the launch-pad.
Traditionally, Sound Pressure Level (SPL) around launch vehicles has been estimated with relatively simple
empirical method, NASA SP-8072[1] since 1971. The NASA SP-8072 is based on a large number of flight data
and results of static-firing tests taken in the United States. Acoustic parameters, such as source power
distribution, frequency distribution, and directivity, are modeled based on free jet characteristics. Since physical
mechanism of the acoustic generation is not necessarily considered in the method, the prediction accuracy is not
sufficient for the design of launch vehicles. Higher accuracy prediction methods need to be constructed. Several
studies already have been carried out to overcome those limitations of the NASA SP-8072[2-6], but those studies
are still an extrapolation of the NASA SP-8072 and the same problems still exist.
While, with the recent increase of the computational performance, numerical simulation of acoustic radiation
from rocket exhaust plumes is becoming feasible. The authors have applied Computational Fluid Dynamics
(CFD) technology to reveal the generation and propagation mechanism of the acoustic waves from Japanese
launch vehicles, such as the H-IIA, M-V solid rocket, and now studying the advanced solid rocket [7-8]. From
these results, the detail of the acoustic generation and propagation mechanism has been cleared.

1
Researcher, JAXA’s Engineering Digital Innovation (JEDI) center, AIAA Member..
2
Engineer, JAXA’s Engineering Digital Innovation (JEDI) center, AIAA Member.
3
Professor, Institute of Space and Astronautical Science (ISAS), AIAA Fellow.
4
Engineer, Space Transportation Mission Directorate / Advanced Solid Rocket Team.
5
Researcher, Aerospace Research and Development Directorate (ARD).
6
Researcher, Aerospace Research and Development Directorate (ARD), AIAA Member.
7
Associate Professor, Institute of Space and Astronautical Science (ISAS), AIAA Member..
1
American Institute of Aeronautics and Astronautics
092407

Copyright © 2009 by the American Institute of Aeronautics and Astronautics, Inc. All rights reserved.
In this study, acoustic measurements are conducted in two series of static-firing tests of a solid rocket motor
in order to investigate the detail of the acoustic phenomena. The obtained data are employed for the quantitative
comparison with the calculation results. The NASA SP-8072 is also applied in order to evaluate the prediction
accuracy and examine the difference between the acoustic phenomena and the empirical model.

II Acoustic Measurement
A. Outline of Static-Firing Test
Acoustic measurements were carried out in two series of static-firing tests of NAL-735 motors. The tests
were conducted at the JAXA’s Noshiro Testing Center in December, 2007. The tests were carried out as field
tests and the test site is located in front of shoreline. Figures 1 and 2 show the NAL-735 motor and a snapshot of
the 1st test, respectively. The specifications of the NAL-735 motor are shown in Table 1. The Reynolds number
based on the nozzle exit (Re=1.75×106) is one order of magnitude below the typical Reynolds number
for large size rocket motors and the exhaust plume is in the overexpanded condition (Pe/Pa=0.418). The
microphones were set as shown in Figure 3. In the 1st test, far-field acoustic signals were measured at 5 angles,
from 20° to 80°, in 15° increments on a far field arc as (M1-5). The radius of the measurement is 63.5 De (De:
diameter of nozzle exit) from the nozzle exit. In the 2nd test, measurements of the near field acoustics (M6-8)
were also carried out in order to examine the difference of acoustic phenomena between in the far and the near
fields. The analysis and comparison of the experimental data with calculation results of the CFD and the NASA
SP-8072 is based on data taken over 10 [s], when the time history of the chamber pressure was relatively flat
and the amplitude of the fluctuation was small.

Fig.1 NAL-735 Motor. Fig.2 Snapshot of the 1st test.

Condenser microphone ICP microphone


39m (63.5De) 10m (16.3De)

20° 10m (16.3De)


35° 3m (4.9De) 3m (4.9De)
M5
50° M6 M7
65° M4 3m (4.9De)
80°
M3 M8
M2
M1

Piezoelectric microphone Piezoelectric microphone

(a) Far Field (the 1st and 2nd tests) (b) Near field (the 2nd test)

Fig.3 Schematic layout of measured points and microphone setting position.

2
American Institute of Aeronautics and Astronautics
092407
Table 1 Motor Specifications

Nozzle Exit Diameter: De 0.614 [m]

Nozzle Exit Mach Number: Me 2.92

Nozzle Exit Reynolds Number: Re 1.75 ×106

Pressure Ratio: Pe /Pa (Nozzle exit/ambient) 0.418

Thrust at the sea level 260 [KN]

B. Evaluation of Experimental Data


In the acoustic measurements, various factors which influence on the reliability of the acoustic measurement
should be considered in order to assure the reliability of the measured data. The principal factors are discussed
below.

B.1. Comparison of Microphone Type


By considering the frequency range, the temperature condition, and the sound pressure level, three types of
microphones were selected for the tests. The types of microphones are given in Table 2. The selected
microphones are Condenser Microphone, Piezoelectric Microphone, and Integrated Circuit Piezoelectric (ICP)
Microphone. The microphones were used in two different configurations; flash mounted to the ground and stand
as shown in Fig.3. The height of the stand (M1-5) and flash mount box (M6-8) are 1.45 [m] and 0.12 [m]
respectively. In the 1st test, the comparison of each microphone type was carried out in order to investigate the
characteristics of the microphones. Figure 4 shows the SPL measured by each type of microphone at the M1. In
this study, the measured acoustic data is analyzed for every 1/3 octave band. As can be seen in Fig. 4, the
frequency characteristics of the measured SPLs by three types of microphones are similar to each other up to 2
[kHz] and the maximum difference is about 6 [dB] at f=1250 [Hz]. While, for frequencies more than 2 [kHz],
the measured data by the condenser microphones should be used, since it is known that the sensitivity effect of
the microphones generally appears at high frequency range and the condenser microphone has the highest
sensitivity among all the microphones employed in the tests.

Table 2 Characteristics of microphones

Microphone Type Location Mount Type

1/4" Condenser Microphone (B&K4939) M1-M5 Stand mount


M1-M5,
Piezoelectric Microphone (Endevco2510) Flash mount
M6-M8 (only 2nd test)
1/4" ICP Microphone (ACO7016) M1-M4 Stand mount

In this study, the far field acoustic data were taken by the condenser microphone. On the other hand, the near
field acoustic data were measured by the piezoelectric microphone since the upper limit of the operational
temperature and SPL of the piezoelectric microphone is higher than other two microphone types.

B.2. Estimation of Experimental Uncertainty


In order to estimate the experimental uncertainty in the measured acoustic data, two series of the tests of the
same motors are carried out. In Figure 5, comparison of the 1st and 2nd test data measured at the M1 is shown.
The result indicates that the maximum difference between the 1st and the 2nd tests is about 4.5 [dB] at f=800
[Hz].

3
American Institute of Aeronautics and Astronautics
092407
On the other hands, the background noise is
important to evaluate the reliability of the acoustic 170
: Condenser Microphone
data. In this study, the background noise was 160 : Piezo. Microphone
: ICP Microphone
measured just before the static-firing test was 150
started. The background noise includes some
140
factors; wind speed, temperature, their fluctuation,

SPL [dB]
130
their gradient, and so on. The background noise
level was 40-75 [dB]. The wind speed measured 120

during the static-firing test was 3.9 [m/s]. The high 110
wind speed and the fluctuation are considered to be 100
the most significant cause of the background noise.
90
As stated below, however, the measured SPLs were 20 40 100 200 400 1000 2000 4000
(St=fDe/Ue) (0.01) (0.05) (1.0)
around 110-140 [dB] at almost all measured Frequency [Hz]
frequency range. So the background noise does not
significantly influence on the reliability of the Fig.4 Comparison of SPLs measured with
measured data since the signal-to-noise ratio (often various microphones. (1st test, M1 point)
abbreviated S/N) is sufficiently high.
Furthermore, the length/diameter ratio of the
motor case is high and the combustion instability is
observed from the pressure measurement inside of
the motor case. The correlation analysis between 170
: 1st test
the time histories of the internal pressure and the 160 : 2nd test
sound signal measured at each microphone point 150
was carried out since the combustion instability
140
may influence on the acoustic data. According to
SPL [dB]

the correlation analysis, it was confirmed that the 130

effect is not significant. 120


In conclusion of B.1 and 2, it can be concluded 110
that the maximum experimental uncertainty is 100
within 6 [dB] up to 2 [kHz]. In this study, the far
90
and near field acoustic data taken in the 2nd test 20 40 100 200 400 1000 2000 4000
(St=fDe/Ue) (0.01) (0.05) (1.0)
are used in the discussion below. Frequency [Hz]

C. Ground Effect Fig.5 Comparison of SPLs between 1st and


Since the rocket motor was horizontally 2nd tests. (M1 point, condenser microphone)
installed and the nozzle center was located at 1.8
[m] height from the ground surface, the measured
acoustic signals include the ground effect. The
ground surface of the test site is acoustically hard
concrete and the authors recently have developed
170
an acoustic impedance model to evaluate the : M1
ground effect on hard surfaces. Figure 6 shows the 160 : M2
: M3
SPLs taken at the M1-5 in the 2nd test. In Fig.6, 150 : M4
: M5
the SPLs slightly decrease at around f=1[kHz] at 140
SPL [dB]

all the points and the ground effect is a possible


130
cause of the phenomena. In this study, the ground
120
effect is evaluated using the acoustic impedance
model proposed by the authors in order to remove 110

the effect from the measured data. 100


For the evaluation of the ground effect, some 90
20 40 100 200 400 1000 2000 4000
acoustic impedance models have been proposed (St=fDe/Ue) (0.01) (0.05) (1.0)
(for example, [9]-[11]). Among the models, Delany Frequency [Hz]
and Bazley model [9] (D-B model) is the most
widely used for various ground surfaces. However, Fig.6 Comparison of measured SPLs.
the application of the D-B model to acoustically
hard surfaces is disputable, since the model is
based on the test data for the fibrous absorbent

4
American Institute of Aeronautics and Astronautics
092407
materials. We compared the ground effect estimated by the D-B model with experimental data by Watanabe et al.
[12]
, which are measured on the acoustically hard surface, asphalt. According to the results, it is confirmed that
the prediction accuracy of the D-B model becomes low as the distance between the source and the receiver point
becomes long. From these results, the authors proposed the following new acoustic impedance model.

ζ = ζRe + i⋅ζIm (1)

ζRe = 1 + (25.0 × tan-1 (0.18 / fc,n - 0.52) + 22.0) × (σ / fc,n ) 0.09 (2)

ζIm = (0.01 × tan-1 (0.05 / fc,n - 1.0) + 1.8) × (σ / fc,n ) 0.10 (3)

fc,n = f ⋅ (r2-r1) / c1 (4)

where σ is flow resistivity of the ground. r1 and r2 is the path length from the source to the microphone (direct
path) and the path length from the image of the source to the microphone (reflected path). c1 is the speed of
sound in the atmosphere. Since the phase change of the sound wave due to the reflection on the ground is small
for acoustically hard surfaces, the frequency coefficient, fc,n, is introduced into the new model. The values of
coefficient and exponent are selected to minimize the difference between the analyzed ground effect in dB and
the experimental data by Watanabe et al. [12]. The comparison result between the analyzed ground effect by the
D-B model and the new model and experimental data indicate that the new model can reasonably predict the
ground effect compared with other models. More detail on the proposed acoustic impedance model and the
comparison results can be found in Fukuda et al. [13].
In the static-firing test, the far field acoustics were measured with the stand and the near field acoustics were
measured with the flash mount box. The height of the stand (M1-5) and flash mount box (M6-8) are 1.45 [m]
and 0.12 [m] respectively, while the center of nozzle exit is positioned at 1.8 [m] height from the ground surface.
Figure 7 shows the analyzed sound attenuation due
to the ground effect for the M1 and M6 points
estimated by the new acoustic impedance model. 15
:M1
Here, the sound source distribution is expressed by 12
:M6
placing point sources and the sound source 9
6
location is set based on the NASA SP-8072 model1
3
SPL [dB]

stated below. The figure shows that the


0
interference of the direct and reflected waves
-3
occurs at around 1, 2, 3, 4 [kHz] and the highest
-6
attenuation is observed at 1 [kHz] with the value of
-9
about 7 [dB] at the M1. While, at the M6, the first
-12
cancellation occurs at a higher frequency, around 8
-15
[kHz] with the value of about 10 [dB], because the 20 40 100 200 400 1000 2000 4000 10000
(St=fDe/Ue) (0.01) (0.05) (1.0)
lower position of the microphone causes the first Frequency [Hz]
cancellation to occur at a higher frequency than
with the higher microphone. By using the analyzed Fig. 7 Estimated sound attenuation
sound attenuation data, the ground effect is due to ground effect. (M1 point)
removed from the measured acoustic data. The
obtained data is employed for the quantitative
evaluation of the SP-8072 and the CFD results in
this paper.

III Method of Analysis


A. NASA SP-8072
There are two acoustic source allocation methods for predicting acoustic loads. Sound generation sources are
allocated along the exhaust stream line as shown in Figure 8. In the first method, each frequency band is
assigned at a unique location along the flow axis. The relation between the source axial position and the flow
axis is shown in Figure 9. Here, the sound source location is decided for each center frequency of 1/3 octave
band. While, the second method assumes that the sound in each frequency band is generated through the flow
axis, not at a unique point. In the second method, the plume length is divided into a number of slices as shown

5
American Institute of Aeronautics and Astronautics
092407
in Fig.8 (b). In this study, the 2.5 times length of the potential core is split into 60 slices. Here, the potential core
length is estimated by the following equation [1].

Xcore /De=3.45⋅ (1+0.38Me)2 (5)

Figure 10 shows that location of each slice and the


overall acoustic power for each slice. Although the
acoustic parameters, such as source power each allocated source point
distribution, frequency distribution, and directivity,
is modeled for the far field acoustics in the SP-
8072, the near field acoustics, such as M6-8 θ
discussed in this paper (Fig.3 (b)), were analyzed Microphone
by the same model as in the far filed. When (a) Model 1
considering application of the SP-8072 to the
actual design of launch pads, the field near the each slice
launch vehicles must be analyzed, since the sound
loads near the vehicle and launch pad are important
θ
for the design phase. In this paper, the first and Microphone
second methods are called as model 1, 2,
(b) Model 2
respectively and the acoustic efficiency, which is
the ratio of the acoustical power to the whole Fig.8 Sketch of acoustic generation source allocation.
mechanical power, was set to be 0.4 percent in
order to make the peak value of the analyzed SPLs
at the M1-3 to be close to the experimental data.

100 180

40
170
Overall Acoustic Power
for each slice [dB]

20
X/De

10 160

4
150
2

1 140
10 100 1000 10000 2 4 10 20 40
Frequency [Hz] Axial position of each slice center X/De

Fig.9 Sound source allocation position Fig.10 Overall acoustic power for each slice
for model 1. for model 2.

B. CFD
Based on the numerical simulation, we aim to design launch pad for better acoustic environment. Since our
target is acoustics, numerical techniques with higher order both for the spatial and temporal discretization are
required [14]. However, such numerical methods are not satisfactory for the designers in terms of robustness and
computational cost. Generation of smooth computational grid required for the higher-order spatial techniques is
another important issue. The acoustic level measured in typical launchers exceeds 140dB (OASPL), which
means the dominant acoustic wave originates in the large-scale motion of the turbulent shear layer, even if the
Reynolds number based on the nozzle exit is the order of 106 or 107. Therefore, application of implicit Large-
Eddy Simulation (LES) and the qualitative discussions were carried out by the authors to reveal the generation
and propagation of the acoustic wave of the H-IIA launch vehicle and the M-V former solid rocket [7-8]. In this
study, the quantitative validation is conducted by using the noise data obtained from the static-firing tests of the
real solid rocket motor. Both of the exhaust plume of the motor and the ambient air are assumed to have an ideal
mixed property of the combustion gas inside the motor. The simple low-dissipation AUSM (SLAU) [15] is
6
American Institute of Aeronautics and Astronautics
092407
employed in order to evaluate the numerical fluxes for
the convective terms. Second-order spatial accuracy
was achieved by applying monotonic upstream
centered schemes for conservation laws (MUSCL)
interpolation based on the primitive variables.
Symmetric Gauss-Seidel alternate directional implicit
factorization scheme is used for time integration with
Newton-Raphson sub-iteration. Second-order
temporal accuracy is guaranteed by three-point
backward differencing formula. The viscous terms are
evaluated by the second-order central differencing
scheme.
Total number of the grid points used in this study Fig.11 Grid distribution around nozzle exit with
are 349 (streamwise) × 130 (circumferential) × 210 time-averaged density plot.
(radial) = 9.5 million. The grid spacing at the jet shear
layer at the nozzle exit is 0.001De, and then, the fine
mesh is extended along the jet shear layer as observed
in Fig.11. We carried out the grid convergence study,
but one of the cases is discussed here. PSD of axial 100
: -5/3 Slope
velocity component at the fluctuating shear layer is : X=5De
indicated in Fig.12. The PSD decays with the slope of 10-1 : X=6De
: X=7De
-5/3 until St=fUe/De=0.2, so that the turbulent : X=8De
structure of the shear layer can be resolved up to this 10-2

frequency. From the experimental results discussed


10-3
later in Fig.14, The peak SPL appears at around
PSD

St=0.2 or lower, depending on the measurement


10-4
locations. It is concluded that this simulation can
resolve the dominant fluctuation of the turbulence,
10-5
although the numerical method applied here has low
accuracy both in the space and time as mentioned 10-6
above. The present study employs the Kirchhoff
method [16] to calculate the far field sound. The 10-7
integral surface is located as shown in Fig.13 so as not 0.01 0.1 1
Strouhal number
only to avoid the impingement of the non-linear flow,
but also to minimize the grid points. Maximum grid Fig.12 PSD of axial velocity in the shear layer. (The
spacing at the acoustic propagation region inside the measurements are executed at the red points in
integral surface is 0.05De. Since the present numerical Fig.19)
scheme requires approximately 20 grid points to
compute a propagating wave, the cut-off frequency in
this study is St=0.05. The computational time step is
lower enough to keep the Courant-Friedrich-Levy
(CFL) number less than unity at the fluctuating jet
shear layer. While, the computational time is required Kirchhoff surface
to be longer enough for the Fast Fourier Transform
(FFT) analysis. The lower limit of the resolved
frequency range is set down to St=0.01, and it takes
about 2 days by using the single node of the NEC SX-
5De
6 supercomputer at JAXA.
The Reynolds-Averaged Navier-Stokes (RANS) 30De
simulation including the nozzle of the motor is
conducted firstly. Then, the result at the nozzle-exit
plane is mapped on the implicit LES domain as an
inflow condition. Fig.13 Position of Kirchhoff surface. (Upper:
snapshot of Lighthill source, lower: entropy)

7
American Institute of Aeronautics and Astronautics
092407
IV Results and Discussions
Figure 14 indicates comparison of the experimental and the analyzed 1/3 octave band SPLs at each
measurement point shown in Fig.3. The experimental data shown in Fig.14 as blue solid lines is modified data
by considering the ground effect. The SPLs increase rapidly at around St=0.4 (f=1 [kHz]) at the M1-5. This
range corresponds to the frequency where the slightly decrease in SPLs appears in the original experimental data
without the ground effect (Fig.6). These results imply that the frequency range where ground effect appears is
reasonably analyzed, but the first dip in the sound attenuation shown in Fig.7 was possibly overestimated. The
ground effect is influenced by the environmental condition, such as temperature, wind speed, wind direction,
fluctuation of them and so on. Furthermore, in this study, the sound source location is set based on the SP-8072
model 1, because the real source distribution cannot be known. However, since the increase of the SPLs at the
M1-5 is within the estimated experimental uncertainty (6 [dB]), the data can be used for the evaluation of the
SP-8072 and the CFD calculation up to 2 [kHz] within the experimental accuracy of 6 [dB]. Considering the
SPLs at the range below St=0.4 (f=1 [kHz]), at the M1, 2, the peak of the SPL appears at around St=0.05-0.1
(f=200-400 [Hz]), while the one at M4, 5 is at around St=0.02 (f=80[Hz]). The result indicates that the frequency
at the peak value becomes low as the angle from the nozzle axis becomes small. In the near field (M6-8), the
peaks appear at around St=0.05-0.1 (f=200-400 [Hz]), similarly to the ones at the M1, 2.

The 1/3 octave band SPLs analyzed by the SP-8072 models 1 and 2 are shown in Fig.14 as green line and
brown broken line, respectively. The comparison between the analyzed SPLs at the M1, 2, 3 and the
experimental data (Fig.14 (a)-(c)) indicate that the SP-8072 reasonably predicts the experimental data and the
maximum prediction error is about 10 [dB]. On the other hand, at the M4, 5 (Fig.14 (d)-(e)), the SP-8072
overestimates the energy spectra at almost all frequency range. Especially at the M5, the SPLs at St=0.03-0.06 is
more than 15 [dB] higher than the experimental data. On the near field (M6, 7, 8) data, the SPLs are reasonably
predicted by the SP-8072 and the maximum prediction error is about 6 [dB] as shown in Fig.14 (f)-(h). The
difference between the models 1 and 2 appears at high frequency range (St>0.05). When compared to the model
2, the SPLs predicted by the model 1 are closer to the experimental results at the M1, 2, while at the M3, 6, 7, 8,
the model 2 predicts closer value to the experimental data than the model 1.
The M7 is located at 38° from the jet axis as shown in Fig.3 (a) and the M4 is at 35°. Note that the SP-8072
overestimates the SPL at the M4 and the one at the M7 point is estimated reasonably, even though the angles are
almost the same. Considering the results, the directivity model in the SP-8072 is discussed below. In this paper,
only the model 1 is discussed for the sake of simplicity, since the same directivity model is used in the model 2.
In the model 1, the sound source is distributed along the jet axis as shown in Fig.8 (a) and the sound power at
each frequency is decided based on the normalized sound power spectrum as shown in Figure 15. Figs.9 and 15
indicates that the maximum sound power is at St=0.02 (f=80[Hz]) and the location is 23.6 De from the nozzle
exit. When considering the angle θ in Fig.8, the angle for the M7 is larger than the one for the M4 as shown in
Figure 16, since the M7 is located upstream from the M4. For example, the angles θM4 and θM7 from the source
with the maximum sound power (St=0.02) are 52.1° and 98.7°, respectively, while the angles θM3 and θM5 from
the source are 70.6° and 31.1°. This result implies that the directivity model for the small angles, i.e. θM4 (52.1°),
θM5 (31.1°), is considered to be inappropriate for the real supersonic free jet. This discussion is very important
for the design of the launch-pad, since the prediction of the near field acoustics is inevitable in the design phase.
On the other hand, the acoustic efficiency was set to be 0.4 percent in the SP-8072, in order to make the peak
value of the analyzed SPLs at the M1-3 to be close to the experimental data as. The efficiency value cannot be
set appropriately in advance of the acoustic prediction and the results are totally dependent on the set value. For
example, if the efficiency is set for the M4-5, the SPLs at the M1-3 become lower than the experimental data.
The overall sound pressure levels (OASPL) in the far field (M1-5) are analyzed over the resolved frequency
range (St=0.01-0.05) for the CFD as shown in Figure 18. The result indicates that the OASPLs are
overestimated by the SP-8072 at the M4, 5 (θ=35°, 20°) and the maximum prediction error at the M5 is more
than 10 [dB].

8
American Institute of Aeronautics and Astronautics
092407
170 170
Resolved frequency range : Modified experimental data Resolved frequency range : Modified experimental data
160 : NASA SP-8072 model1 160 : NASA SP-8072 model1
: NASA SP-8072 model2 : NASA SP-8072 model2
150 : CFD (Present) 150 : CFD (Present)

140 140
SPL [dB]

SPL [dB]
130 130

120 120

110 110

100 100

90 90
0.005 0.01 0.05 0.1 0.5 1 0.005 0.01 0.05 0.1 0.5 1
(Frequency [Hz]) (40) (200) (4000) (Frequency [Hz]) (40) (200) (4000)
Strouhal number Strouhal number

(a) M1 point (e) M5 point


170 170
Resolved frequency range : Modified experimental data Resolved frequency range : Modified experimental data
160 : NASA SP-8072 model1 160 : NASA SP-8072 model1
: NASA SP-8072 model2 : NASA SP-8072 model2
150 : CFD (Present) 150 : CFD (Present)

140 140
SPL [dB]

SPL [dB]
130 130

120 120

110 110

100 100

90 90
0.005 0.01 0.05 0.1 0.5 1 0.005 0.01 0.05 0.1 0.5 1
(Frequency [Hz]) (40) (200) (4000) (Frequency [Hz]) (40) (200) (4000)
Strouhal number Strouhal number

(b) M2 point (f) M6 point


170 170
Resolved frequency range : Modified experimental data Resolved frequency range : Modified experimental data
160 : NASA SP-8072 model1 160 : NASA SP-8072 model1
: NASA SP-8072 model2 : NASA SP-8072 model2
150 : CFD (Present) 150 : CFD (Present)

140 140
SPL [dB]

SPL [dB]

130 130

120 120

110 110

100 100

90 90
0.005 0.01 0.05 0.1 0.5 1 0.005 0.01 0.05 0.1 0.5 1
(Frequency [Hz]) (40) (200) (4000) (Frequency [Hz]) (40) (200) (4000)
Strouhal number Strouhal number

(c) M3 point (g) M7 point


170 170
Resolved frequency range : Modified experimental data Resolved frequency range : Modified experimental data
160 : NASA SP-8072 model1 160 : NASA SP-8072 model1
: NASA SP-8072 model2 : NASA SP-8072 model2
150 : CFD (Present) 150 : CFD (Present)

140 140
SPL [dB]

SPL [dB]

130 130

120 120

110 110

100 100

90 90
0.005 0.01 0.05 0.1 0.5 1 0.005 0.01 0.05 0.1 0.5 1
(Frequency [Hz]) (40) (200) (4000) (Frequency [Hz]) (40) (200) (4000)
Strouhal number Strouhal number

(d) M4 point (h) M8 point

Fig. 14 Comparison of 1/3 octave band SPL.

9
American Institute of Aeronautics and Astronautics
092407
20
Normalized relative Sound Power [dB]

15
each source point
10

5
θM7
0 θM4
-5 M7
-10

-15 θM7 > θM4 M4


-20
0.01 0.02 0.04 0.1 0.2 0.4 1
(Frequency [Hz]) (40) (200) (4000)
Strouhal number

Fig. 15 Normalized sound power distribution Fig. 16 Schematic image of source and receiver
in SP-8072 model 1. point in SP-8072.

10 170
: St=fD e/Ue=0.004 : Experimental data (2nd test)
: St=fD e/Ue=0.0125 : NASA SP-8072 model1
: St=fD e/Ue=0.04 160 : NASA SP-8072 model2
5 : St=fD e/Ue=0.125 : CFD
: St=fD e/Ue=0.4
150
0
Directivity [dB]

OASPL [dB]

140
-5
130

-10
120

-15 110

-20 100
0 20 40 60 80 100 120 140 160 180 10 20 30 40 50 60 70 80 90
Angle from jet axis [deg] Angle from jet axis [deg]

Fig. 17 Directivity model of SP-8072. Fig. 18 Comparison of OASPL.


(0.01≤St≤0.05, M1-M5 points)

Since the plume ejected from the motor is in the overexpanded condition (Pe/Pa=0.418) as described in
Table.1, it is observed from Fig.11 that the Mach disk is formed at the exit of the nozzle. The vorticity contour-
line shown in Fig.19 indicates that the slip line appearing at the back of the Mach disk shows unsteady motion.
The jet shear layer also starts to fluctuate at about 2De downstream from the nozzle exit. Unsteadiness of the
slip line and the jet shear layer produces large-scale structures.
From the divergence of velocity plot shown in Fig.19, generation of the acoustic wave is clearly observed.
Since the large-scale structures propagate downstream at the supersonic speed, the Mach wave is generated and
radiated obliquely downstream as observed in Fig.19. Distribution of the OASPL displayed in Fig.20 indicates
the intense Mach wave radiation between 5De and 10De where the supersonic jet shear layer fluctuates, which
means that the fluctuation of the jet shear layer generates dominant noise.
The analyzed 1/3 octave band SPLs at the measurement points are compared in Fig.14. From the far-field
results, the peak SPL at upstream positions such as the M1, 2, 3 appears at higher frequency as compared with
M4, 5. On the other hand, the results in the near field (M6, 7, 8), the peak SPL appears at higher frequency
which is similar to the M1, 2, 3. These results correspond to the experimental results as stated above. From the
directional characteristics indicated by the OASPL in the far field (Fig.18), the intense Mach-wave propagation
appears at 50° from the jet axis. In order to discuss in detail, 1/3 octave band filtered OASPLs are compared in
Figure 21. The widespread distribution of the low frequency noise is observed from Fig.21 (a)-(d). On the other
hand, the higher frequency acoustic wave has a strong directivity at about 50° from the jet axis (Fig.21 (e)-(h)).

10
American Institute of Aeronautics and Astronautics
092407
The result discussed in Fig.21 clearly explains the features shown in Figs.14 and 18. In comparison of 1/3
octave band SPLs and OASPL shown in Fig.14 and Fig.18, the CFD reasonably predict the SPLs at all the
measured points over the resolved frequency range (St=0.01-0.05). From the results, it is confirmed that the
prediction accuracy of the CFD calculation is within 6 [dB] in 1/3 octave band SPLs and 5 [dB] in OASPL,
which is within the experimental uncertainty involved in the measured data, and the CFD is effective for both
the near and the far field acoustics generated from the rocket motors.

The CFD results indicate that the fluctuation of the supersonic jet shear layer generates dominant noise. As a
result of comparison with the experimental data, it can be concluded that the adequate expression of the sound
generation and propagation mechanism is essential in the acoustic prediction for supersonic free jets and
reasonably calculated in the CFD calculation. On the other hand, in the results of the SP-8072, the SPLs at the
M4, 5 (20°, 35°) in the far field are overestimated. These results imply that the directivity of the dominant noise
generated from the sound source is not appropriately modeled in the SP-8072. According to the comparison with
the near and far field SPLs analyzed by the SP-8072, in particular the directivity model for small angles is
inadequate for the real supersonic jet and should be further examined to improve the prediction accuracy. This
discussion is very important for the design of the launch-pad, since the prediction of the near points from the jet
axis is inevitable in the design phase. Furthermore, the efficiency value cannot be set appropriately in advance
of the acoustic prediction in the SP-8072.

2De

Mach disk
Slip line

Fig.19 Divergence of velocity vector plot with vorticity contour-line.

OASPL
OASPL (0.01≤St≤0.05)

150 155 160 165 170 175 180 185

5De 10De

Vorticity magnitude

Fig.20 Distribution of OASPL (upper) and vorticity (lower).

11
American Institute of Aeronautics and Astronautics
092407
130 [dB] Filtered OASPL 160 [dB]

(a) 0.009 ≤ St ≤ 0.011 (35.6 ≤ f ≤ 45.0)

(b) 0.011 ≤ St ≤ 0.014 (44.5 ≤ f ≤ 56.1)

(c) 0.014 ≤ St ≤ 0.018 (56.1 ≤ f ≤ 70.7)

(d) 0.018 ≤ St ≤ 0.023 (71.3 ≤ f ≤ 89.8)

(e) 0.023 ≤ St ≤ 0.029 (89.1 ≤ f ≤ 112.3)

(f) 0.029 ≤ St ≤ 0.036 (111.4 ≤ f ≤ 140.3)

(g) 0.036 ≤ St ≤ 0.046 (142.5 ≤ f ≤ 179.6)

(h) 0.046 ≤ St ≤ 0.057 (178.2 ≤ f ≤ 224.5)


Nozzle 5De 10De 15De 20De 25De 30De
Exit

Fig.21 1/3 octave band filtered OASPL distribution.

12
American Institute of Aeronautics and Astronautics
092407
V Concluding Remarks and Future Studies
In this study, acoustic measurements are carried out in two series of static-firing tests of a solid rocket motor.
The principal factors which influence on the measured acoustic data are examined and the experimental
uncertainty is estimated to be within 6 [dB] up to 2 [kHz]. Furthermore, the ground effect is removed from the
measured data by using an acoustic impedance model proposed by the authors. The obtained data are employed
for the quantitative evaluation of the empirical method of NASA SP-8072 and the CFD.
The SP-8072 overestimates the sound pressure level at the 20° and 35° points from the jet axis in the far
field, although the SPLs at other observation points are reasonably predicted and the maximum difference is
within 10 [dB]. From the comparison between the far and the near field data, it is confirmed that the directivity
model for small angles from the jet axis is inadequate for the real supersonic jet and should be further examined
to improve the prediction accuracy.
On the other hand, the CFD calculation can clearly explain the sound generation phenomena. The CFD
results show that the fluctuation of the supersonic jet shear layer generates dominant noise, the Mach wave.
Furthermore, the CFD can reasonably predict SPLs for the resolved frequency range (St=0.01-0.05) at all the
measurement points. The prediction accuracy of the CFD is within 5 [dB] in OASPL, which is within the
experimental uncertainty involved in the measured data. In the acoustic prediction by the CFD, any models like
directivity and efficiency models are not necessary and that is great advantage for the design of the launch-pad.
From these results, it is confirmed that the CFD methodology is effective for the prediction of both the near
and the far field acoustics generated from the rocket motors.

References
1
Eldred,K.M., and et al., “Acoustic Loads Generated by the Propulsion System”, NASA SP-8072, June 1971.
2
Varnier, J., “Experimental Study and Simulation of Rocket Engine Freejet Noise”, AIAA Journal, Vol. 39,
No. 10, Oct. 2001, pp. 1851-1859.
3
Varnier, J., and Raguenet, W., “Experimental Characterization of the Sound Power Radiated by Impinging
Supersonic Jets”, AIAA Journal, Vol. 40, No. 5, May 2002, pp. 825-831.
4
Koudriavtsev, V., Varnier, J., and Safronov, A., “A Simplified Model of Jet Aerodynamics and Acoustics”,
AIAA Paper 2004-2877, May 2004.
5
Koudriavtsev, V., and Safronov, A., “Acoustic Model for Supersonic Jet Interaction with a Complex
Deflector”, Proc. of the 6th European Symposium on Aerothermodynamics for Space Vehicles, Nov., 2008.
6
Casalino, D., Barbarino, M., Genito, M., and Ferrara, V., “Improved Empirical Methods for Rocket Noise
Prediction through CAA Computation of Elementary Source Fields”, AIAA Paper 2008-2939, May 2008.
7
Tsutsumi, S., Takaki, R., Shima, E., Fujii, K., and Arita, M., “Generation and Propagation of Pressure waves
from H-IIA Launch Vehicle at Lift-off”, AIAA Paper 2008-390, Jan.2008.
8
Tsutsumi, S., Fukuda, K., Takaki, R., Shima, E., Fujii, K., and Ui, K., “Numerical Study on Acoustic
Radiation for Designing Launch-Pad of Advanced Solid Rocket”, AIAA Paper 2008-5148, July, 2008.
9
Delany, M. E., and Bazley, E. N., “Acoustical Properties of Fibrous Absorbent Materials”, Applied
Acoustics, 3, pp.105-116, 1970.
10
Miki, Y., “Acoustical Properties of Porous Materials”, Journal of Acoustical Society of Japan, 11, 1, pp.19-
28, 1990.
11
Attenborough, K., “Acoustical Impedance Models for Outdoor Ground Surfaces”, Journal of Sound and
Vibration, 99, 4, pp.521-544, 1985.
12
Watanabe, M., Takeda, K., and Kotake, S., “Measurement Method for Pressure Reflection Coefficients and
Ground Reflection Correction”, Technical Report of National Aerospace Laboratory, TR-460, 1976. (in
Japanese)
13
Fukuda, K., Tsutsumi, S., Ui, K., Ishii, T., Takaki, R., and Fujii, K., “Examination of Methodology for
Evaluating Ground Effect of Static-Firing Tests on Rocket Motors”, Proceedings of the 41th Fluid Dynamics
Conference / Aerospace numerical Simulation Symposium 2009, June, 2009. (in Japanese, to be published)
14
Bodony, D. J. and Lele, S. K., “Review of the Current Status of Jet Noise Predictions Using Large-Eddy
Simulation”, AIAA Paper 2006-486, Jan., 2006.
15
Shima, E., Kitamura, K., “On New Simple Low-Dissipation Scheme of AUSM-Family for All Speeds”,
AIAA Paper 2009-136, Jan., 2009.
16
Lyrintzis, A. S. and Uzun, A., “Integral Techniques for Jet Aeroacoustics Calculations”, AIAA Paper 2001-
2253, May, 2001.

13
American Institute of Aeronautics and Astronautics
092407

Potrebbero piacerti anche