Effets of Inter-Coahing Spaing on Aerodynami Noise Generation Inside High-speed Trains 1 J. Ryu, 1 J. Park*, 2 C. Choi, 1 S. Song Hanyang University, Seoul, South Korea 1 ; Korea Railroad Researh Institute, Uiwang, South Korea 2 Abstrat Experiments were performed in order to investigate the effets of inter-oah spaing on the aerodynami noise generation inside high-speed trains. The avity in the inter-oah spae indues the flow separation and self-sustained osillation whih results in signifiant inrease in the wall pressure spetra on the struture of the high speed train. From the propagation of the wall pressure along the struture, it results in flow indued vibration of side glass windows and onsequent interior noise generation. To ondut the experiments in the ontrolled environment, the wind tunnel tests were performed. The model of the avity whih has same physial dimensions as the atual interoah spaing of the train was ustom-built and installed in the wind tunnel test setion. The wall pressure spetra were measured using mirophones. The effets of avity spaing, flow speeds, and different geometries of the mud-flap were investigated. To further analyze the harateristis of noise generation from the flow over the avity, the numerial analysis was performed using a ommerial CFD ode. Using the numerial model, the effets of avity opening length on the flow around the avity was investigated. 1. Introdution Vibration and noise generation of high speed trains should be minimized for safety and omport ride ondition of passengers. There are several important soures that ontribute to interior noise generation. Compared to other noise soures, the aerodynami noise soure inreases more rapidly, and beome one important ontributor to the interior noise. (1)(2)(3)(4) To further inrease the operating speeds, the noise generation mehanism of the aerodynami soures should be understood. From the identified mehanism, effetive methods of noise ontrol an be proposed. The primary goal of this investigation was to assess the influene of the geometries of the interoah spaing on the noise generated from side glass windows of high-speed trains. The turbulent boundary layer flow is separated at the mud-flap, and its intensity inreases due to flow feedbak at the avity. The wall pressure fields generated from turbulent flow over the avity exite the window of the high-speed train (Fig. 1). From the fored vibration of the window, interior noise is generated. The mud-flap is installed at the inter-ouh spaing. The mud-flap is made of rubber supported by sheet metals. The mud-flap protets equipments in the inter-oah spaing and redues the avity opening length. In this paper, wind tunnel was used for analyzing the flow indued exitations under ontrolled experimental setup. The model of the avity whih was designed to enable easy hange of the avity opening length and mud-flap shapes was ustom-built. The mirophone was installed flush with the wind tunnel floor and was used in measuring the pressure generated by the flow over the avity. To further investigate the harateristi of flow over the avity, two-dimensional numerial model was built using a ommerialized CFD ode. By omparing numerial model to the measured results, the effets of avity opening length on the wall pressure exitation were investigated. Pantograph Sound radiation Fig. 1 Aeroaousti noise generation inside high speed trains * Corresponding author
2. Measurement of aero-aousti noise with wind tunnel 2.1 Wind tunnel Experiments were performed using a ustom-built test fixture installed in a wind tunnel (open iruit type), as depited in Figure 2. The test model has the same avity dimension as the atual interoah spaing. It was installed at the test setion of the wind tunnel. The mud-flap and avity opening was flush with the wind tunnel test setion floor. The test model was struturally isolated from the vibration of the wind tunnel walls using a foamed rubber. The test setion size of the wind tunnel is 0.8 0.8 1.6 m, and the allowable flow speed ranged from 5 to 70 m/s. Mirophones (B&K Type 4951) was mounted flush with the test setion floor. Manometer was used for measuring free-stream flow veloity. 0.84 m 0.07 m Mud flaps L 0.66 m avity Mirophones Wind tunnel floor 15 mm 1 mm 20 mm 1.27 m Fig.2 Wind tunnel, model of inter-ouh spaing, and installed mirophones 2.2 Wall pressure spetral harateristis Figure 3(a) shows the measured wall pressure spetra for various avity opening lengths. Sine the wind tunnel did not have the muffling devies, it showed a large low frequeny resonane around 10 Hz. This low frequeny ontents orresponds to the half wavelength resonane of the wind tunnel itself. Above that frequeny, there were two frequeny ontents: resonane frequeny ontents and the broadband frequeny ontents. With inreasing avity opening length, the measured wall pressure spetra inreased espeially at low frequenies. The half wavelength resonane did not showed dependene of the avity opening length. Figure 3(b) shows the dependene of sound pressure level on the avity opening length. At otave bands where the tonal omponents have effets the measured sound pressure level inreased signifiantly with inreasing avity opening length. At higher frequeny otave bands, the effets of the avity opening length were negligibly small. Sound Pressure Level (db, re. 2E-5 Pa) 100 95 90 85 500Hz 1000Hz 2000Hz 80 0 5 10 15 20 25 30 35 Cavity opening length [ m ] Fig. 3 Sound pressure level for various avity opening lengths in (a) narrow bands, and (b) otave bands 100Hz A. Tonal omponents The resonane frequeny ontents beome more signifiant with inreasing avity opening length. This resonane frequenies dereases with inreasing avity opening length. The disturbane generated as the boundary layer flow enter the avity travels and amplifies as it travels along the avity. The amplified disturbane exites the mud-flap and the sound wave is generated. The generated sound wave travels bak to the upstream mud-flap, and enhane the disturbane generation. From this feedbak loop, the tonal omponents as shown in Figure 1 is generated. This
flow feedbak frequeny is predited by Rossiter s equation as L L n β + =, n=1,2,3, (1) UC fn Where L is the avity opening length, U is the flow onvetion veloity ( 0.6 U, U is the free stream flow veloity), is the speed of sound, β ( 0.25) is the phase lag, and f n is the tonal frequeny. (5) Figure 4 shows the omparison of the predited resonane frequenies to the measured values. It showed lose agreement between measured and predited values, whih verifies that the tonal omponents our due to resonane due to flow feedbak yle. The avity aousti modes were not important for the sound generation harateristis of the inter-oah spaing. The blade passage frequeny ourred at 152 Hz (fan rotation speed of 760 rpm and 12 blades) regardless of the avity opening length. 2.5 fl / U 2.0 1.5 1.0 0.5 measured :, n = 1, n = 2, n = 3, n = 4 predited :, n = 1, n = 2, n = 3, n = 4 0.0 0.0 0.1 0.2 Mah M=U/ Fig. 4 The measured and predited tonal frequenies and its dependene on flow veloity B. Broadband omponents Exept the tonal omponents in Figure 3(a), the wall pressure spetra is generated from the boundary layer flow. When there is no avity, the generated wall pressure is dependent on the flow onvetion speeds (U ), the frition veloity (U τ ), boundary layer thikness (δ). (6)(7) The empirial relationship is summarized as 2 4 2 Φ pp ( ω) ~ ρ0 Uτ ( δ / U )( ωδ / U ) ( ωδ / U << 1), (2) 2 4 1 Φ pp ( ω) ~ ρ0uτ ( ω) ( 1 < ωδ / U < Uτ δ / 30υ ), (3) 2 4 1 4 Φ pp ( ω) ~ ρ0uτ ( ω) ( ωδ / U ) ( ωδ / U > Uτ δ / 30υ ). (4) This frequeny dependent variation in the pressure level is plotted in Figure 3(a) as dotted lines. It shows lose agreement with the measured variations exept lowest frequeny ranges (f < 30 Hz) in whih the aousti resonane inside the wind tunnel dominates the measured values. This broadband sound pressure level inreased with inreasing avity opening length. For avity opening length of 0.24 m, the effets of flow veloity was measured, Figure 5. With inreasing flow veloity, the measured wall pressure spetra inreased. Also, the tonal frequenies inreased. Fig. 5 Sound pressure level dependene on free stream flow veloity
3. Noise Control Methods To investigate the effets of mud-flap geometries, various shapes of mud flap were built using aryl. The atual mud flap is made of rubber. For this parametri study, the aryl of lager elastiity was used to represent the atual mud-flap. When the geometries are similar, there was no signifiant hange in the measured wall pressure spetra. The purpose of making the orrugated mud-flap (ases 2-5) rather than the flat, retangular shape (ase 1) in Figure 6, was to redue the magnitude of the tonal omponents. For the mud-flap shapes tested in this study, there was no signifiant variation in the measured wall pressure spetra. This suggests that the tonal omponents our regardless of the spanwise variation of the avity opening length. For minimal generation of the wall pressure spetra, the simple, mud flap is more advantageous than the orrugated one. (8) As an alternative approah to redue the pressure generation, the avity walls were treated by aousti foams. The aousti foams were ut in the shape of triangle with maximum height of 0.15 m and ompletely overed avity walls. The sound pressure inside the avity was also measured. Figure 7(a) shows the omparison of the measured sound pressure before and after the treatments. With the aousti treatments, there was signifiant redution in the pressure inside the avity, about 4.2 db in the entire frequeny ranges. Although there was also redution in the generated pressure, its impat was less signifiant for the pressure measured outside the avity (Fig. 7(b)). Case 1 Case 2 Case 3 Case 4 Case 5 Fig. 6 Sound pressure level variation with mud-flap geometries Fig. 7 Sound pressure level (a) in avity and (b) out of avity before and after the treatments 4. Numerial modelling of flow over the avity For predition of the flow over the avity, the two dimensional model was built using ommerial software FLUENT. The analysis was performed using k-ω model. The number of mesh was 22400. The length from flow inlet to outlet was 10 m (Fig. 8). Figure 9 shows the alulated kineti energy. High turbulene is generated in the avity opening, and irulation of the flow inside the avity was
observed. The magnitude of this irulation inreased with inreasing avity opening length. Figure 10 2 shows the pressure oeffiient, p = 2( P Pref ) / ρu, at the leading edge of the mud-flap where P ref is the atmospheri pressure. (9)(10) With inreasing avity opening length, there was signifiant variation of the pressure near the leading edge. From this disturbane, the sound is generated more strongly and propagate to the trailing edge of the mud-flap. It is expeted that this higher magnitude disturbane inreases separation of the flow, and inreased generation of tonal omponents. The CFD model onsidered in this study does not take into aount of the hange of turbulene over time, pressure osillation effet whih is harateristi of turbulene flow in avity. For this purpose, it requires methods suh as LES (Large Eddy Simulation) or DNS (Diret Numerial Simulation). Fig. 8 The dimension of model used in CFD analysis Fig. 9 Distribution of kineti energy for opening length of 1.0 m Fig. 10 Pressure oeffiient variation with opening length 5. Conlusion To investigate the effets of avity spaing of high speed trains, experiments were performed using a ustom-built test fixture installed in the wind tunnel. The apparatus has the similar avity as the one found in the train. Due to the omplex nature of the turbulent flow, statistial desriptions of the wall pressure fields were used. When the surfaes of the struture are irregular, the turbulene intensity of the grazing flows is signifiantly inreased relative to flow grazing over smooth and flat surfaes. For the urrent study, the avity ated as the irregularity. Inreased wall pressure spetra were measured for the larger avity opening length. In the wall pressure spetra, several pure tones were measured whih is not present for turbulent flows over smooth surfae. These tonal omponents in the measured pressure ourred due to the flow feedbak at the avity opening. These frequenies were predited by Rossiter s equation and agreed well to the measured values. The tonal omponents at speifi frequenies were signifiantly larger than those of the wall pressure ontribution from turbulent flows alone. Signifiant variation was observed for the magnitude of these pure-tones from the variation of the length of the avity. As the avity opening length dereases, the tonal frequenies inreased, but its amplitude dereased. CFD model of the avity was built and was used in investigating the effets of avity spaing on the magnitude of the turbulene and pressure amplitude. Sine there is a limitation in minimizing the avity opening length, methods to redue wall pressure
exitation was investigated. First approah was to orrugate the mud-flap. This orrugation resulted in redution of the wall pressure measured at speifi frequenies. However it yielded larger exitation in the overall frequeny ranges. As an alternative approah, the aousti treatment in the inter-ouh spaing was onsidered. When the avity was treated by the aousti materials, redution of the aousti pressure was observed in the pressure inside the avity and also on the wall pressure spetra. Aknowledgements This work was supported by the researh fund of Korea Railroad Researh Institute. Referenes [1] K. H. Park, J. H. Park, S. M. Song, T. H. Kim, T. J. Lee, S. H. Choi, 2006, Wind tunnel tests for analyzing noise generation form the inter-oah spaing of a high-speed train, Proeedings of the KSNVE Annual Autumn Conferene. [2] H. I. Koh, J. C. Kim, C. W. Lee, 2006, A Study on Noise Charateristis of High speed Trains, Proeedings of the KSNVE Annual Spring Conferene. [3] S. H. Choi, J. H. Park, C. K. Park, 2006, Noise generated form the inter-oah spaing of a highspeed train, Proeedings of the KSNVE Annual Spring Conferene. [4] K. H. Park, 2007, " Wind tunnel tests for analyzing noise generation from the inter-oah spaing of a high-speed train", M.D. Thesis, Department of Mehanial Engineering, Hanyang University, Seoul, Korea. [5] J. E. Rossiter, 1964, Wind-tunnel experiments on the flow over retangular avities at subsoni and transoni speeds, ARC R&M 3438. [6] W. K. Blake, 1986, Mehanis of Flow Indued Sound and Vibration.(vol. 2), Bethesda, Maryland. [7] G. C. Koh, H. S. Jung, D. H. Kim, J. S. Joe, 2003, Study on the Aerodynami Charateristis of Hanyang Low Speed Wind Tunnel, Journal of the Korean soiety for aeronautial & spae sienes. [8] J. M. Ryu, J. H. Park, K. H. Park, S. M. Song, S. H. Choi, 2007, Effets of mud-flap parameters on aeroaousti noise generation inside high speed trains, Proeedings of the KSNVE Annual Spring Conferene. [9] H. S. Song, 2008, Numerial Analysis of Flow Charateristis for a Cavity with Flaps, M.D. Thesis, Department of Mehanial Engineering, Hanyang University, Seoul, Korea. [10] P. K. Kundu & I. M. Cohen, Fluid Mehanis(3/E.), Elsevier Aademi Press