Académique Documents
Professionnel Documents
Culture Documents
1158 J. Acoust. Soc. Am. 104 (3), Pt. 1, September 1998 0001-4966/98/104(3)/1158/7/$15.00 © 1998 Acoustical Society of America 1158
Downloaded 02 Jan 2013 to 203.110.243.23. Redistribution subject to ASA license or copyright; see http://asadl.org/terms
calculations the train Mach number M 5U/c 0 (c 0 5 speed of
sound! is assumed to be smaller than about 0.2, but this
restriction can be relaxed by modification of the analysis,
although this is not discussed here.
The compression wave is generated by the gross dis-
placement of the air in the tunnel entrance by the incoming
train, and for a sufficiently streamlined train the influence on
this of flow separation can be ignored in a first approxima-
tion. When viscous effects at the tunnel walls are also ne-
glected, the displacement of the air by the moving train is
equivalent to that produced by a distribution of moving vol-
ume sources of constant strengths, and the unsteady motion
they produce can be represented by a velocity potential
f (x,t). The sources translate at the speed U of the train, and
their distribution at time t is denoted by q(x1Ut,y,z),
where q(x) may be taken to be the source distribution at the
instant t50 at which the nose of the train just enters the
tunnel.
The velocity potential is the solution with outgoing
wave behavior of
FIG. 1. ~a! Tunnel entrance with vented roof. ~b! Mapping the Z plane
bounded by the profile of the tunnel roof and its image in the ground onto
the right half of the z plane.
S 1 ]2
c 20 ] t 2
D
2¹ 2 f 52q ~ x1Ut,y,z ! , ~1!
1159 J. Acoust. Soc. Am., Vol. 104, No. 3, Pt. 1, September 1998 Michael Howe: High-speed train compression wave 1159
Downloaded 02 Jan 2013 to 203.110.243.23. Redistribution subject to ASA license or copyright; see http://asadl.org/terms
predicts the initial form of the compression wave profile sev- ] AT / ] x 8 Þ0. For a long train, the integral becomes constant
eral tunnel diameters ahead of the entrance, where the am- and equal to UA0 just after the passage of the nose through
plitude of the unsteady motion is small, and the perturbation the tunnel entrance into the region where ] f * / ] x51; this
pressure p52 r 0 ] f / ] t, where r 0 is the undisturbed air den- constant value is the maximum compression wave ampli-
sity. This initial profile is found with the aid of ~2! to be tude, equal approximately to r 0 U 2 A0 /A at low Mach num-
given by ~see Ref. 7 for details! bers. For a tunnel of constant cross-sectional area, or one
with a ‘‘flared’’ entrance, ] f * / ] x 8 decreases smoothly to
p ~ x,t ! [p ~ x,t ! '
r 0U
A
E q ~ x 8 1U @ t # ,y 8 ,z 8 ! zero in the direction out of the entrance along the path of the
train, and the pressure rise across compression wavefront
therefore tends to be monotonic.
]f*
3 ~ x8 ! d 3 x8 , x8 5 ~ x 8 ,y 8 ,z 8 ! ,
]x8
~3! II. THE POTENTIAL f * „x… FOR A VENTED TUNNEL
where the integration is taken over the region occupied by A. Integral formula for f * „x…
the sources, and @ t # 't1x/c 0 is the retarded time. The im-
pulsive micro-pressure wave p 8 (x,t), say, that radiates from Equation ~5! implies that the initial rise time of the com-
the far end of the tunnel is given in terms of the compression pression wavefront is governed by the behavior of
wave, and at large distance r from the tunnel exit at x ] 2 f * (x)/ ] x 2 , which is nonzero only in the vicinity of the
5x E , say, by the formula3,12,13 tunnel entrance. The magnitude of this second derivative
along the path of the train can be reduced by moderately
A ]p flaring the tunnel entrance, thereby causing the streamlines
p 8 ~ x,t ! ' ~ x E ,t2r/c 0 ! , r@h, ~4!
Vc 0 r ] t of the hypothetical potential outflow determined by f * to
diverge more gradually from the entrance. A similar result is
where V is the effective solid angle into which the wave achieved by the presence of a suitable arrangement of vents
radiates, determined by local conditions near the exit. Ac- near the tunnel entrance, through which a portion of the po-
cording to Eq. ~3! tential flow f * can ‘‘escape’’ before reaching the entrance.
E
]p r 0U 2 The distribution of vents along the tunnel wall must be cho-
~ x,t ! '2 q ~ x 8 1U @ t # ,y 8 ,z 8 ! sen carefully, however, since a rapid divergence of the
]t A streamlines will merely cause the tunnel to be acoustically
shorter, and the formation of the compression wave to be
] 2f *
3 ~ x8 ! d 3 x8 . ~5! delayed until the train passes from the vented to the unvented
]x8 2
section of the tunnel entrance, with no increase in rise time.
However, the actual value to be used on the right on ~4! will The influence of venting will be examined by assuming
usually be considerably different from that predicted by this the tunnel roof near the entrance to be perforated with iden-
formula because of nonlinear steepening of the wave in a tical apertures distributed with a number N (x) per unit area
long tunnel. Solution ~3! should be regarded as defining the ~Fig. 1!. The aperture cross section is assumed to be suffi-
initial waveform for use in a one-dimensional nonlinear ciently large that the unsteady aperture flows produced by
model of wave propagation in the tunnel. This aspect of the the passage of the train may be regarded as irrotational. This
problem is not pursued here, where attention is restricted to should be a good approximation during the period of wave
the mechanism of compression wave generation. formation, as the train enters the tunnel, since turbulence
diffusion from aperture walls and from the train occurs over
a much longer timescale. The aperture distribution function
B. The volume source strength N (x) is at our disposal, and can be adjusted to modify the
rise time of the compression wave.
Equation ~3! expresses the compression wave in terms
Introduce the representation
of a source distribution q that depends on the shape of the
train, but is nonzero only near the nose and tail. For small f * ~ x! 5 f 0* ~ x! 1 f A* ~ x! , ~7!
values of the area ratio A0 /A, it was shown in Ref. 7, by
comparison with the experiments of Maeda et al.,4 that q is where f *0 (x) is the potential function f * in the absence of
well approximated by the line source venting, and f A * (x) is the additional contribution from the
presence of the vents. To determine f A * (x), let G(x,x8 ) de-
] AT note the potential flow Green’s function that satisfies ¹ 2 G
q ~ x,y,z ! 5U ~ x ! d ~ y ! d ~ z2z T! , ~6!
]x 5 d (x2x8 ), and is defined such that G has vanishing normal
derivative on the unperforated tunnel walls, and ¹G→0 at
where A T(x) is the cross-sectional area of the train at dis- large distances from the point source, both within and out-
tance x measured from the nose, so that AT(L)[A 0 , and side the tunnel. Then
S D
the ground level line y50,z5z T lies in the vertical plane of
symmetry of the train.
The integration in ~3! is accordingly confined to the in-
tervals on the x 8 axis where the retarded value of
f A* ~ x! 5 R S
@ G ~ x,x8 !# x8
] f A*
] x 8n
~ x8 !
1
dS ~ x8 ! , ~8!
1160 J. Acoust. Soc. Am., Vol. 104, No. 3, Pt. 1, September 1998 Michael Howe: High-speed train compression wave 1160
Downloaded 02 Jan 2013 to 203.110.243.23. Redistribution subject to ASA license or copyright; see http://asadl.org/terms
where the integration is over the tunnel walls, and C. Rectangular tunnel with a vented roof
(]fA * / ] x n8 ) 1 denotes the normal velocity within an aperture
The determination of Q from ~11! requires a knowledge
directed out of the tunnel. The square bracket notation in this of both the unperforated velocity potential f * 0 (x) and
formula is defined such that, for any function f (x), @ f (x) # x is
Green’s function G(x,x8 ). Although f * 0 (x) can readily be
the exterior minus the interior limiting values of f (x) as the
calculated for a tunnel of circular or semi-circular cylindrical
point x approaches the tunnel wall.
form,7 a convenient analytical representation of the Green’s
Now let Q̄(x j )5r S j ( ] f A* / ] x 8n ) 1 dS(x8 )be the volume function is not generally available. We may avoid this diffi-
flux directed out of the tunnel through the jth aperture S j culty for the tunnel of rectangular cross section in Fig. 1 by
whose centroid is at x j . When the apertures are small rela- adopting a two-dimensional approximation to the potential
tive to the tunnel diameter, and the distance of the point x flow from the entrance, by assuming that the local irrota-
from the nearest aperture exceeds the aperture diameter, tional motion in the entrance is uniform in the z direction
G(x,x8 )'G(x,x j ) may be regarded as constant when per- ~transverse to the direction of motion of the train!. This ap-
forming the integration in ~8! over S j . Since the length scale proximation cannot give a complete description of compres-
of variation of Q̄(x j ) between neighboring apertures must be sion wave generation, but it yields predictions that are fully
comparable to that of the potential function f * 0 , which may consistent with analytical and numerical results obtained pre-
be supposed to drive the motion through the apertures, and viously for an unvented tunnel.4,5,7,14,15
there are N apertures per unit area of the tunnel wall, it Consider the conformal transformation16
follows that the integrand of ~8! can be approximated by
pZ
N (x8 )Q̄(x8 ) @ G(x,x8 ) # x8 , and therefore that 215 z 2 1ln z 2 , ~12!
h
f * ~ x! 5 f *
0 ~ x! 1 R
S
Q ~ x8 !@ G ~ x,x8 !# x8 dS ~ x8 ! , which maps the Z5x1iy plane bounded by the tunnel roof
and its image in the ground @Fig. 1~b!# onto the right half
~Re z .0) of the z plane, with the entrance points Z56ih
Q ~ x8 ! [N ~ x8 ! Q̄ ~ x8 ! . ~9! mapping onto z 56i, and the point at infinity within the
tunnel mapping onto z 50.
In this formula Q(x) is the volume flux per unit surface area The potential f 0* is the real part of w 0 [ f 0* 1i c 0*
of the tunnel wall, and the integration is over the region 5(2h/ p )ln z. The perforations are assumed to be confined
where QÞ0. to the roof of the tunnel and distributed uniformly in the z
direction ~out of the plane of the paper in Fig. 1!, so that
N [N (x). The complex potential of flow from the vented
B. Calculation of Q „x… tunnel entrance is then given by the following analog of Eq.
~9!
The volume flux Q̄(x j ) through the jth aperture can be
expressed in terms of the potential difference @ f̄ * # x j across
the aperture produced by the unperforated potential f * o and
w[ f * 1i c * 5w 0 1
1
p
E 2`
0
Q ~ x 8 ! $ ln~ z 2 1 j 21 ~ x 8 !!
U U
has the dimensions of length, and depends only on the shape
and size of the aperture; it is equal to 2R for a circular
aperture of radius R in a thin wall, and approximately equal Q ~ x ! 52 a ~ x ! ln
j 1~ x !
j 2~ x !
1
a~ x !
h
E 2`
0
Q~ x8!
to A (aperture area/ p ) for a nonelongated, thin wall aperture
of arbitrary shape. The potential difference @ f̄ * # x j can actu-
ally be approximated by the representation ~9! of f * (x),
because, although @ G(x j ,x8 ) # x8 is singular as x8 →x j , the
HU
3 ln
j 21 ~ x 8 ! 2 j 21 ~ x !
j 22 ~ x 8 ! 2 j 21 ~ x !
U
singularity is integrable, and its contribution to the surface
integral from a vanishingly small region around x8 5x j
shrinks to zero. Thus Q(x) satisfies the integral equation
1ln U j 22 ~ x 8 ! 2 j 22 ~ x !
j 21 ~ x 8 ! 2 j 22 ~ x !
UJ dx 8 , ~14!
1161 J. Acoust. Soc. Am., Vol. 104, No. 3, Pt. 1, September 1998 Michael Howe: High-speed train compression wave 1161
Downloaded 02 Jan 2013 to 203.110.243.23. Redistribution subject to ASA license or copyright; see http://asadl.org/terms
FIG. 2. Volume flux distribution function Q(x) for a vented tunnel roof
with a (x)[(hK / p )N (x)50.35ex/h .
D. Potential flow from the vented tunnel entrance ground indicates that ] 2 f * / ] x 2 will be correspondingly
Figure 2 illustrates the variation of the predicted volume smaller along the train track. Equation ~5! then suggests that
flux density Q as a function of x/h when the fractional open the rise time of the compression wave will be larger for the
area of the vents decreases exponentially from the tunnel vented tunnel.
entrance according to a (x)50.35e x/h . Q vanishes at the tun-
nel entrance, where the potential jump across the roof is null, III. INFLUENCE OF VENTING ON COMPRESSION
and attains a maximum near x52h. The dramatic effect of WAVE RISE TIME
these vents on the potential flow from the entrance can be A. Front of train modeled by a point source
calculated from ~13! and is depicted in Fig. 3, in which
streamlines c * (x)5constant are plotted for both the un- Preliminary estimates of the influence of roof venting
vented and vented tunnel entrances. The more gradual diver- can be obtained by assuming that the source distribution ~6!
gence of the streamlines from the vented tunnel in the vicin- that models fluid displacement by the nose of the train is
ity of the center line of the tunnel plus its image in the concentrated in one point, i.e., by taking
q ~ x! 'A0 U d ~ x ! d ~ y ! d ~ z ! , ~16!
where the net source strength A 0 U5 * L0 U( ] AT / ] x)dx,
and
the vertical plane of symmetry of the train is taken to coin-
cide with z50. This approximation for q is formally identi-
cal to that obtained when the train is modeled by a semi-
infinite cylinder of cross-sectional area A0 .
Equations ~3! and ~5! now supply the following approxi-
mations to the pressure and pressure gradient of the compres-
sion wave generated as the train enters the tunnel,
r 0 U 2 A0 ] F *
p ~ x,t ! ' ~ 2U @ t # /h,0,0 ! ,
A ]X
~17!
]p r 0 U A0 ] F *
3 2
~ x,t ! '2 ~ 2U @ t # /h,0,0 ! ,
]t hA ]X2
where the following dimensionless notation has been intro-
duced:
x f * ~ x!
X5 , F * ~ X,y/h,z/h ! 5 . ~18!
h h
Equations ~17! are applicable within the tunnel at large dis-
tances from the entrance, but before nonlinear steepening of
the wavefront becomes important.
The dashed and solid curves in Fig. 4, respectively, rep-
FIG. 3. Streamlines of uniform flow from the tunnel entrance defined by the
resent the predicted pressure and pressure gradient when the
velocity potential f * (x): ~a! unvented tunnel; ~b! vented roof with a (x) behavior of f * near the tunnel entrance is determined by the
[(hK / p )N (x)50.35e x/h . two-dimensional approximation ~13! for ~i! an unvented tun-
1162 J. Acoust. Soc. Am., Vol. 104, No. 3, Pt. 1, September 1998 Michael Howe: High-speed train compression wave 1162
Downloaded 02 Jan 2013 to 203.110.243.23. Redistribution subject to ASA license or copyright; see http://asadl.org/terms
FIG. 6. Compression wave pressure gradients ( ] p/ ] t)/( r 0 U 3 A0 /hA) for
L/h52: ~a! unvented entrance; ~b! vented with a (x)50.35e 0.6x/h .
FIG. 5. ~a! Volume flux distribution Q(x) for a (x)50.35e 0.6x/h . ~b! Com-
pression waves p/( r 0 U 2 A0 /A) ~— — — —! and pressure gradients
]p 22 r 0 U 3 A0
( ] p/ ] t)/( r 0 U 3 A0 /hA) ~————! for the unvented and vented tunnels. ~ x,t ! '
]t LA
nel, and ~ii! a tunnel with the roof venting of Fig. 2 ~see Sec.
II D!, for which a (x)[hK N(x)/ p 50.35e x/h . Venting in-
creases the rise time of the wave by a factor of about 25 , and
3 E 0
L/h
F ~ X,h/L !
] 2F *
]X2
~ X2U @ t # /h,0,0 ! dX,
1163 J. Acoust. Soc. Am., Vol. 104, No. 3, Pt. 1, September 1998 Michael Howe: High-speed train compression wave 1163
Downloaded 02 Jan 2013 to 203.110.243.23. Redistribution subject to ASA license or copyright; see http://asadl.org/terms
are applicable to a tunnel of rectangular cross section with a shape of train nose on compression wave generated by train entering tun-
variably vented roof near the entrance, and permit the distri- nel,’’ Proceedings of the International Conference on Speedup Technol-
ogy for Railway and Maglev Vehicles ~Yokohama, Japan, 22–26 Novem-
bution of the vents to be optimized to maximize rise time.
ber 1993!, pp. 315–319.
Only the very simplest case in which the aperture distribu- 5
M. Iida, T. Matsumura, K. Nakatani, T. Fukuda, and T. Maeda, ‘‘Opti-
tion decreases exponentially with distance into the tunnel has mum nose shape for reducing tunnel sonic boom,’’ Institute of Mechanical
been considered, but our predictions indicate that at least Engineers Paper C514/015/96 ~1996!.
6
sixfold increases in the rise time are then possible. Applica- N. Sugimoto, ‘‘Shock-free tunnel for future high-speed trains,’’ Paper
PS3-3, Proceedings of the International Conference on Speedup Technol-
tion of the method to tunnels with more complicated en- ogy for Railway and Maglev Vehicles ~Yokohama, Japan, 22–26 Novem-
trance geometries is straightforward in principle. This is cur- ber 1993!.
7
rently being done for a tunnel of circular cylindrical cross- M. S. Howe, ‘‘The compression wave produced by a high-speed train
section, of the type frequently used in model scale tests, and entering a tunnel,’’ Proc. R. Soc. London, Ser. A 254, 1523–1534 ~1998!.
8
M. S. Howe, ‘‘The generation of sound by aerodynamic sources in an
which should therefore permit experimental validation of the inhomogeneous steady flow,’’ J. Fluid Mech. 67, 579–610 ~1975!.
theory. 9
M. C. Swarden and D. G. Wilson, ‘‘Vehicle-tunnel entry at subsonic
speeds. Final Report—Part 1,’’ Rept. No. DSR 76111-3, Engineering
ACKNOWLEDGMENTS Projects Laboratory, Department of Mechanical Engineering, Massachu-
setts Institute of Technology ~1970!.
10
The author expresses his gratitude to Dr. Tatsuo Maeda, A. E. Vardy, ‘‘Ventilation approach regions for railway tunnels,’’ Trans-
Head of the Aerodynamics and Noise Reduction Group of port Engineering Journal, American Society of Civil Engineers 101, 609–
619 ~1975!.
the Railway Technical Research Institute, Tokyo, for his 11
A. E. Vardy, ‘‘Reflection of step-wavefronts from perforated and flared
help and advice during the preparation of this paper. extensions,’’ J. Sound Vib. 59, 577–589 ~1978!.
12
Lord Rayleigh, The Theory of Sound ~Dover, New York, 1945!, Vol. 2.
13
1
S. Ozawa, Y. Morito, T. Maeda, and M. Kinosita, ‘‘Investigation of the A. D. Pierce, Acoustics, An Introduction to its Principles and Applications
pressure wave radiated from a tunnel exit,’’ Railway Technical Research ~American Institute of Physics, New York, 1989!.
14
Institute Rept. 1023 ~in Japanese, 1976!. T. Ogawa and K. Fujii, ‘‘Prediction and alleviation of a booming noise
2 created by a high-speed train moving into a tunnel,’’ paper in Proceedings
S. Ozawa, ‘‘Studies of the micro-pressure wave radiated from a tunnel
exit,’’ Railway Technical Research Institute Rept. 1121 ~in Japanese, of the European Community Conference on Computational Methods in
1979!. Applied Sciences ~Wiley, New York, 1996!.
3 15
S. Ozawa, T. Maeda, T. Matsumura, K. Uchida, H. Kajiyama, and K. T. Ogawa and K. Fujii, ‘‘Numerical investigation of three dimensional
Tanemoto, ‘‘Countermeasures to reduce micro-pressure waves radiating compressible flows induced by a train moving into a tunnel,’’ Comput.
from exits of Shinkansen tunnels,’’ in Aerodynamics and Ventilation of Fluids 26, 565–585 ~1997!.
Vehicle Tunnels ~Elsevier Science, New York, 1991!, pp. 253–266. 16
G. K. Batchelor, An Introduction to Fluid Dynamics ~Cambridge U.P.,
4
T. Maeda, T. Matsumura, M. Iida, K. Nakatani, and K. Uchida, ‘‘Effect of Cambridge, 1967!.
1164 J. Acoust. Soc. Am., Vol. 104, No. 3, Pt. 1, September 1998 Michael Howe: High-speed train compression wave 1164
Downloaded 02 Jan 2013 to 203.110.243.23. Redistribution subject to ASA license or copyright; see http://asadl.org/terms