A A A Study on Sound and Vibration Propagation Caused by External Flow Affecting Interior Noise of Railway Vehicles Gaku Minorikawa 1 Hosei University 3-7-2, Kajinocho, Koganei-shi, Tokyo, 184-0002, Japan Noboru Yamano 2 Hosei University 3-7-2, Kajinocho, Koganei-shi, Tokyo, 184-0002, Japan Kosuke Hotta 3 NIPPON SHARYO, LTD. 2-1-21, HONOHARA,Toyokawa-shi, Aichi, 442-0061, Japan Yuki Yamauchi 4 NIPPON SHARYO, LTD. 2-1-21, HONOHARA,Toyokawa-shi, Aichi, 442-0061, Japan ABSTRACT Among the aerodynamic noises generated by railway vehicles running at high speed, the noise generated by the structure on the roof not only affects the environment along the railroad tracks outside the vehicle, but also propagates inside the vehicle, damaging the comfort of the cabin. Exterior noise is a phenomenon in which aerodynamic noise generated by the structure propagates, and the source of the noise has been clarified and predicted. On the other hand, interior noise is a combination of flow-induced aerodynamic noise propagating acoustically and structural excitation caused by pressure fluctuation due to wake vortices around the structure, and these two phenomena have not been investigated quantitatively. In this study, the effects of aerodynamic noise generated by structures in the airflow on the acoustic environment inside and outside the vehicle were experimentally investigated using a small low-noise wind tunnel from the viewpoint of source and propagation characteristics, and practical modeling of these effects was attempted. In addition, the characteristics of the sound source were investigated in detail by computational aero acoustic analysis. 1 minori@hosei.ac.jp 2 noboru.yamano.8t@stu.hosei.ac.jp 3 k-hotta@n-sharyo.co.jp 4 y-yamauchi@n-sharyo.co.jp a Shea mar ce 21-24 AUGUST SCOTTISH BENT caso 1. INTRODUCTION When railway vehicles are running at high speed, equipment mounted around the vehicle body interrupts the flow and causes aerodynamic noise. Aerodynamic noise not only affects the environment along the tracks outside the vehicle, but also propagates inside the vehicle and impairs the cabin comfort. Aerodynamic noise has two components: a flow-induced component due to fluctuations in the fluid itself, such as vortex noise, and a flow-excitation component as the structural excitation force, which is caused by pressure fluctuations of vortices in the wake of equipment. At high-speed running condition, flow-induced noise with strong dependence on airflow velocity is the main source of exterior noise, while interior noise is a composite phenomenon of flow-induced noise and flow-excitation noise, and it is difficult to investigate the mutual relationship between them under actual conditions quantitatively, especially inside the vehicle during commercial operation. Therefore, a prediction method with sufficient accuracy for practical use is required. In this study, an experimental modeling of aerodynamic noise caused by external flow was attempted using a small low-noise wind tunnel from the viewpoints of source and propagation characteristics to investigate the effects on the sound environment inside and outside the vehicle. A simple aero acoustic analysis was also attempted to investigate aerodynamic noise source distribution and to validate the experimental results. 2. EXPERIMENT AND NUMERICAL SIMULATION 2.1. Experiment Figure 1 shows the experimental setup in a small low-noise wind tunnel. The purpose of this study is to investigate the effect on the noise propagated inside the box-shaped model with respect to the aerodynamic noise caused by the external flow due to various obstacles (such as equipment on the roof of a vehicle). The small low-noise wind tunnel used was a self-made semi-Göttingen-type wind tunnel with a blowing nozzle dimension of 130 mm x 130 mm, a maximum airflow velocity of 45 m/s, and background noise with airflow of 65 dB(A) at the maximum airflow velocity (1) . An obstacle was placed in line with this nozzle, and the bottom of the obstacle was supported by the top plate of the box that simulates the interior of a vehicle; the airflow passed over the surface of the top plate, and only the obstacle was exposed to the airflow. In order to reproduce how aerodynamic noise generated in the exterior space propagates to the interior space, microphones and accelerometers were installed inside and outside the box, and on the top panel of the box, respectively. The positions and symbols of the measurement points “S01-04” indicate microphones and “V01-05” indicate accelerometers. Sound-absorbing materials were placed inside the model so that resonance due to the internal acoustic modes would not dominate the actual phenomena. The generated noise and vibration were analyzed at the measurement frequency up to 2000 Hz. The experimental setup is shown in Figure 2. Figure 2 and Table 1 show a list of the obstacles tested in this experiment. The obstacle shapes were based on a rectangular shape of L=40mm, W=80mm, H=40mm, with a 20mm fillet added to the edge. In addition, obstacles with a width and height that would have a two-dimensional shape with respect to the wind tunnel nozzle were tested and compared. The actual obstacle installation is shown in Figure 3. S04 Collector Nozzle 390 Obstacle 570 130 170 60 300 60 800 50 S01 150 V05 V04 150 300 600 V01 V02 V03 130 130 S02 400 Aluminum model Aluminum plate Type ‘Shape cow HR REWwo80 | RectangularW-8onm | 40 80 40 Tewoso | TopfiletweGomm | 40 80 40 20 ‘SFwoa0 | Sidefiletw=80nm | 40 80 40 20 REWI60 | Rectanguiorwatéonm | 40 160 40 ~ TFwi6o | Toofiletw=160mm | 40 160 40 20 HF wi60 | Side filet w=i60nm | 40 160 40 20 REWI6O | RectangularH=160mm | 40 40 160 ~ SFHIGO | Sie filet H=t60mm | 40 40 160 20, co S03 Sound absorbing material 50 210 290 Figure 1: Experimental setup and measurement points 500 Microphone Collector Microphone Obstacle Nozzle Accelerometer Aluminum model Figure 2: Setup of obstacle and sensors Table 1: Type of obstacles Figure 3: Arrangement of obstacle 2.2. Numerical simulation OpenFOAM was used as the CFD (Computational Fluid Dynamics) tool and Actran as the CAA (Computational Aero Acoustic) tool. And SNGR (Stochastic Noise Generation and Radiation (3), (4) ) was used as the coupling method between CFD and CAA. SNGR is one of the simple CAA methods, which generate aerodynamic noise from the average turbulent kinetic energy obtained from RANS (Reynolds-Averaged Navier-Stokes) using random functions. Table 2 and Figure 4 show an overview of the aerodynamic noise analysis and the analytical model, respectively. For the CFD model, the experimental model was reproduced only the area above of the top plate, where airflows mainly passed, and the calculated airflow velocity was 30 m/s. For the CAA model, the aerodynamic noise source was extracted in two regions: around the obstacle and around the collector of the wind tunnel. The outside field was modeled as free field condition and the sound absorbing material inside the box was modeled with a simple 2D element that could give the sound absorption coefficient. na a Table 2: Numerical Method CFD-CAA Coupling SNGR CAA Method Direct frequency response Frequency Domain 5-1000 [Hz] Method RANS Turbulent model k-ω SST Pressure-Velocity Coupling SIMPLE CFD a) CFD Model b) CAA Model Figure 4: Numerical Model 3. RESULT AND DISCUSSION 3.1. Experiment result 1) Noise and vibration characteristics caused by obstacles Firstly, the basic characteristics of noise and vibration generated by a simple rectangular obstacle (Type RE_W080) are discussed. Figure 5 shows the 1/3 octave band level of the internal noise at measurement points S01 and S02 in the box, the 1/3 octave band level of the external noise at measurement point S04 directly above the obstacle, and the 1/3 octave band level of the vibration acceleration at measurement point V01 near the obstacle on the box top panel. In each case, the level increases as the airflow velocity increases. While the overall level of noise increases in proportion to approximately the sixth power of the airflow velocity, the overall level of vibration acceleration increases below that. The sound pressure level of S01 near the top panel has a peak at 160 Hz, which is the same as the peak of the vibration acceleration of the top plate. Therefore, it can be seen that the pressure fluctuation caused by the airflow passing through the obstacle excites the top plate and becomes a sound source. On the other hand, S02 at the center in the box shows a large decrease at around 200 Hz, so it indicates that the influence of internal acoustic modes. In addition, since there is no clear peak in the external noise, the aerodynamic noise of the obstacle is considered to be the main source. S01 [RE_W080] S02 [RE_W080] 80 80 25m/s 30m/s 35m/s 25m/s 30m/s 35m/s 70 70 A-weighted SPL[dBA] A-weighted SPL[dBA] 60 60 50 50 40 40 30 30 20 20 OA OA 80 80 100 125 160 315 100 125 160 315 200 250 400 500 630 800 200 250 400 500 630 800 1000 1000 a) Interior sound pressure level (S01) b) Interior sound pressure level (S02) Frequency[Hz] Frequency[Hz] S04 [RE_W080] V01 [RE_W080] 80 140 25m/s 30m/s 35m/s 25m/s 30m/s 35m/s 130 70 A-weighted SPL[dBA] 120 60 VAL[dB] 110 50 100 40 90 30 80 20 70 OA OA 80 80 315 100 125 160 100 125 160 315 200 250 400 500 630 800 200 250 400 500 630 800 1000 1000 c) Outside sound pressure level (S04) d) Top plate vibration acceleration level (V01) Frequency[Hz] Frequency[Hz] Figure 5: Comparison of noise and vibration acceleration levels with airflow velocity 2) Effect of obstacle size In order to investigate the effect of obstacle shapes on noise characteristics in wind tunnel experiments, internal noise of RE_W080, which is the standard condition for obstacles, is compared with RE_W160 (W = 160mm) and RE_H160 (H = 160mm), which are sufficiently large than the wind tunnel nozzle size (130mm×130mm) . Figure 6 shows comparison of the noise for W=80mm, W=160mm, and H=160mm at u=30m/s. The generated noise is larger when the obstacle is wider relative to the top plate, while the trend remains almost the same. In addition, the frequency characteristics are different when the size of the obstacle changes in the vertical direction. It is considered that this is due to the difference in the vibration state of the top plate due to the directions of flow separation from the obstacle. On the other hand, the difference with and without the fillet is more significant in the case of H=160mm, so it seems to be due to the effective length of the fillet. S01 [u=30m/s] S01 [u=30m/s] 80 80 RE_W080 TF_W080 RE_W160 TF_W160 70 70 A-weighted SPL[dBA] A-weighted SPL[dBA] 60 60 50 50 40 40 30 30 20 20 OA OA 80 80 315 100 125 160 100 125 160 315 200 250 400 500 630 800 200 250 400 500 630 800 1000 1000 a) W=80mm b) W=160mm Frequency[Hz] Frequency[Hz] S01 [u=30m/s] 80 RE_H160 SF_H160 70 A-weighted SPL[dBA] 60 50 40 30 20 OA 80 100 125 160 315 200 250 400 500 630 800 1000 c) H=160mm Figure 6: Comparison of internal sound pressure levels due to the size of obstacles Frequency[Hz] (Measurement point S01) 3) Effect of fillet at obstacle Figure 7 shows comparison of noise and vibration due to the different edge shapes of the obstacle. By providing a fillet at the top, a reduction of up to 3 dB can be seen in internal and external noise and the top plate vibration. This indicates that aerodynamic noise and pressure fluctuations are reduced by suppressing the separation of the flow passing through the edge of the obstacle. S01 [u=30m/s] S02 [u=30m/s] 80 80 RE_W080 TF_W080 RE_W080 TF_W080 70 70 A-weighted SPL[dBA] A-weighted SPL[dBA] 60 60 50 50 40 40 30 30 20 20 OA OA 80 80 315 100 125 160 100 125 160 315 200 250 400 500 630 800 200 250 400 500 630 800 1000 1000 a) Interior sound pressure level (S01) b) Interior sound pressure level (S02) Frequency[Hz] Frequency[Hz] S04 [u=30m/s] V01 [u=30m/s] 80 140 RE_W080 TF_W080 RE_W080 TF_W080 130 70 A-weighted SPL[dBA] 120 60 VAL[dB] 110 50 100 40 90 30 80 20 70 OA OA 80 80 100 125 160 315 100 125 160 315 800 200 250 400 500 630 200 250 400 500 630 800 1000 1000 c) Outside sound pressure level (S04) d) Top plate vibration acceleration level (V01) Frequency[Hz] Frequency[Hz] Figure 7: Comparison of noise and vibration acceleration levels by obstacle shape 4) Effect of cavity Figure 8 shows comparison of the cases of a single rectangular-shaped obstacle and two obstacles placed at an interval to generate noise due to cavity. At the external noise, the SPL in the case of with the cavity is larger at all frequencies and the effect of the downstream obstacle is apparent, however the resonance of the cavity is not clearly observed. On the other hand, there is not clearly difference in the internal noise and the top plate vibration. These indicates that the difference in excitation force due to airflow pressure fluctuations is not pronounced. Cavity noise radiates toward the top of the cavity aperture, so it is assumed to have little effect on the internal noise. S01 [u=30m/s] S02 [u=30m/s] 80 80 RE_W080 RE_W080s with cavity RE_W080 RE_W080s with cavity 70 70 A-weighted SPL[dBA] A-weighted SPL[dBA] 60 60 50 50 40 40 30 30 20 20 OA OA 80 80 100 125 160 315 100 125 160 315 200 250 400 500 630 800 500 630 800 200 250 400 1000 1000 a) Interior sound pressure level (S01) b) Interior sound pressure level (S02) Frequency[Hz] Frequency[Hz] S04 [u=30m/s] V01 [u=30m/s] 80 140 RE_W080 RE_W080s with cavity RE_W080 RE_W080s with cavity 130 70 A-weighted SPL[dBA] 120 60 VAL[dB] 110 50 100 40 90 30 80 20 70 OA OA 80 80 100 125 160 315 100 125 160 315 200 250 400 500 630 800 200 250 400 500 630 800 1000 1000 c) Outside sound pressure level (S04) d) Top plate vibration acceleration level (V01) Frequency[Hz] Frequency[Hz] Figure 8: Comparison of noise and vibration acceleration levels with and without cavity 3.2. Numerical result Figure 8 shows a comparison between the experiment and numerical results. In the external observation point S04, the phenomenon could be reproduced within about 10 dB in all frequency bands, and the reduction due to the difference in the upper edge shapes is almost the same. On the other hand, in the internal point S01 and S02, there are large difference in the distributions. However, the effect of reducing SPL by the fillet could be reproduced. So it is confirmed that the qualitative evaluation of the difference in the obstacle shape is possible even with the accuracy of this numerical model. Especially for quantitative differences at the internal point is thought that the cause is that the numerical model is very simply modeled (especially modeling of the sound absorbing material), therefore it can be improved by making them in detail. Since the consistency between the experimental model and the numerical model is confirmed mutually, the sound source distribution is investigated by CFD. Figure 9 shows TKE (turbulent kinetic energy) distribution. The effect that the fillet suppresses the generation of TKE from the upper edge of the obstacle to the downstream is observed and it is the reason that the aerodynamic noise is reduced. These result confirm that this experimental model can evaluate the difference in sound sources formed by the external flow. S01 S02 80 80 Exp:RE_W080 Exp:TF_W080 CAA:RE_W080 CAA:TF_W080 Exp:RE_W080 Exp:TF_W080 CAA:RE_W080 CAA:TF_W080 70 70 A-weighted SPL[dBA] A-weighted SPL[dBA] 60 60 50 50 40 40 30 30 20 20 100 125 160 315 100 125 160 315 200 250 400 500 630 800 200 250 400 500 630 800 1000 1000 a) Interior sound pressure level (S01) b) Interior sound pressure level (S02) Frequency[Hz] Frequency[Hz] S04 80 Exp:RE_W080 Exp:TF_W080 CAA:RE_W080 CAA:TF_W080 70 A-weighted SPL[dBA] 60 50 40 30 20 125 160 100 315 200 250 400 500 630 800 1000 c) Outside sound pressure level (S04) Frequency[Hz] Figure 8: Comparison Experiment with Numerical data Turbulent Kinetic Energy Turbulent Kinetic Energy a) RE_W080 b) TF_W080 Figure 9: Sound source survey with CFD 4. CONCLUSIONS In order to establish an experimental model that can evaluate the effects on the sound environment inside and outside railway vehicles in terms of source and propagation characteristics, the box- shaped model with a top plate and obstacles was installed in a small low-noise wind tunnel, so that experimental modelling of the sound source and noise propagation caused by external aerodynamic noise was attempted. A simple aero acoustic analysis was also attempted to verify the experimental results by investigating sound source distribution, etc. From the experimental results, it was possible to reproduce the propagation of aerodynamic noise generated by obstacles into the box through the top plate, and to observe the transmission of aerodynamic noise, the propagation of noise due to the vibration of the top plate caused by airflow pressure fluctuation, and the effect of acoustic resonance in the box. In addition, the differences in the noise and vibration characteristics of the aerodynamic noise generated from the obstacle shape and obstacle cavity were observed. SNGR was used in the numerical analysis. It was found that this method could be used to evaluate the reproduction of broadband aerodynamic noise qualitatively. In the future, it is necessary to use this experimental model to clarify quantitatively the mechanism of propagation of flow-induced noise and flow-excitation noise caused by external flow to internal noise, and to improve the accuracy of SNGR and other CAA methods. 5. REFERENCES 1. Yuta Kato and Gaku Minorikawa, “Study on design and prototyping of small low noise wind tunnel”, inter-noise 2015(san Francisco, California, USA),2015 2. Yoshiyuki Maruta and Gaku Minorikawa, “Background Noise in Measuring Section”, Transactions of the Japan Society of Mechanical Engineers, Series (B) , Vol.(61),No(590) ,1995 3. C. Bailly and D. Juve. A stochastic approach to compute subsonic noise using linearized euler’s equations. AIAA-paper 99-1872, 1999. 4. W. Bechara, C. Bailly, and P. Lafon. Stochastic approach to noise modelling for free turbulent flows. AIAA Journal, 32 (3):455–463, 1994. Previous Paper 493 of 769 Next