horse power
payload speed
H
p
W
p
V
: 4
The economic efciency of fuel used can be an index
indicating the transportation energy, and is, thus,
expressed by an inverse of the transportation efciency:
transport efficiencyp
W
p
V
H
p
W V
H
p
W
p
W
; 5
where W is the total payload, V the speed, and H
p
the
horse power.
Using the equations above, Fig. 2 shows a comparison
between the transportation energies necessary to carry
one person up to 1 km [2]. For three different types of
vehicles, i.e., train, bus and car, the number on the right
side indicates the transportation energy relative to the
train, in which the transportation energy of the train is
10
100
1000
1 0.1 10 100 1000 10000
1
SST
6
0
0
0
k
m
1
3
0
0
0
k
m
1
2
4
3
S
p
e
e
d
(
k
m
/
h
)
Distance (km)
Automobile
S
e
o
u
l

P
u
s
a
n
Train
Highspeed train (Shinkansen)
Jetplane
Supersonic plane
Superhigh speed train
New traffic system
Walking
I
n
t
e
r
c
o
n
t
i
n
e
n
t
Ship
I
d
e
a
l
s
p
e
e
d
i
n
t
h
e
2
1
t
h
c
e
n
t
u
r
y
I
d
e
a
l
s
p
e
e
d
i
n
t
h
e
p
r
e
s
e
n
t
t
i
m
e
Sea
M
a
x
i
m
u
m
d
i
s
t
a
n
c
e
Fig. 1. Relationship between speed and distance required for transportation vehicle.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 472
assumed to be 100 kcal. It is found that the transporta
tion energy of a car amounts to 6 times that of a train.
From the economic point of view of the fuel used,
Fig. 3 shows a comparison between each transportation
engine [3]. The solid line indicates the wellknown
KarmanGabriellis limiting line (KG line). The KG
line increases with the speed of transportation vehicle. It
is noted that the economic efciency of the fuel used in
transportation engine is improved as the transportation
vehicle approaches the KG line. Each transportation
vehicle has different speed ranges. For instance, the boat
is in the range below 50 km/h, the ground vehicles are in
between 50 and 200 km/h, the airplane in between 500
and 800 km/h.
These speed ranges can be classied, depending on
whether the vehicle is driven by the buoyancy force, the
reaction force, or the lift force. The KG line indicates the
lower limits of each speed range. For the range of the
buoyancy force support, the energy consumption in low
speed ranges is comparatively low. However, with the
speedup of the boat, the transportation efciency
becomes remarkably low due to the increased wave
drag on the boat. For instance, the value of P=W
p
V
increases up to several hundred times as the boat speed
increases from about 15 knot (28.7 km/h) to 30 knot
(55.6 km/h), resulting in an extremely low transportation
efciency. Therefore, a hydrofoil boat or a hovercraft of
the lift force support can be one of the alternatives for
higher speeds.
For the range of the reaction force support, the value
of P=W
p
V is low in the mid of the speed ranges. For
the speeds over this range, a highspeed railway train is
recommended. In this case, the value of P=W
p
V is on
an extended line for the existing conventional train
100kcal
172kcal
593kcal
Train
Bus
Automobile
Fig. 2. Energies necessary to carry one person up to 1 km.
0.001
0.002
0.003
0.005
0.004
0.01
0.02
0.03
0.04
0.05
0.1
0.2
0.3
0.4
0.5
1.0
2
3
10 20 30 50 100 200 300 500 1000 2000
TGV
ICE
Shinkansen
MAGREV
SST
B737
B727
B707
B767
B747
SR
KarmanGabrielli
HC
YS11
F27
0.001
0.002
0.003
0.004
0.01
0.02
0.03
0.04
0.05
0.1
0.2
0.4
0.5
1.0
0.005
0.3
3
1
2
Speed(km/h)
N
e
w
e
s
t
s
h
i
p
Large
tanker
Liner
Container
Warship
Small boat
Destroyer
Cruiser
H
B
Helicopter
Bus
Truck
Turboprop
Jet plane
Jet plane
(next generation)
Buoyancy
support
Reaction force
support
Lift support
Buoyancy support
Lift support
Reaction force
support
P
/
W
V
(
k
W
h
/
t
o
n
k
m
)
p
P
/
W
V
(
P
S
h
/
t
o
n
k
m
)
p
Automobile
Fig. 3. Comparison of each transportation.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 473
system. Further extension of this line approaches the
Magnetic levitation (Maglev) train system. For the
range of the lift force support, the value of P=W
p
V is,
in general, dependent on the liftdrag ratio, being
independent of the speed of the transportation vehicle.
2.3. Limiting factors to the speedup
In general, the transportation vehicle connecting from
city to city is required to meet the following conditions:
highspeed transportation, bulk volume transportation,
safe and comfortable transportation with less air
pollution and noise, highly reliable transportation with
low cost and maintenance, etc. The highspeed railway
train can be one of the alternatives to meet these
requirements.
Since 1940s, many countries have tried to speedup
the conventional train system. Fig. 4 shows the progress
of the speedup of train [4]. Symbol Jrefers to the TGV
in France, &the ICE in Germany, Kthe Shinkansen in
Japan and the HST in UK. It is found that in over a
half century, the train speed has increased more than
two fold.
Major limiting factors to the speedup of train result
from many different sources. Technical factors are
associated with train/rail systems, while geographical
factors are related to the tunnel system. For instance, in
Japan, the portion of the tunnel to the total railway line
amounts to about 60%, while in France it is at most
several per cent. Fig. 5 shows the technical limiting
factors to the speedup of train and associated factors.
These factors are mainly associated with the train body,
the track line, the electric devices around the track, etc.
For instance, the train speed along a curved track is
limited by the traveling performance, passengers
comfort and safety, which are again associated with
the train body and track line. Thus, to be able to
increase the maximum speed of trains, it is necessary to
take account of these limiting factors.
3. Aerodynamic problems of railway train
For the purpose of development of a faster and more
safe train system with lower air pollution and noise,
many researchers are paying much attention on the
aerodynamics of highspeed railway train. These works
have attention to the development of newgeneration
train body, rail and tunnel systems. The aerodynamic
phenomena with regard to highspeed railway train are
strongly dependent on the train speed. Thus, the
aerodynamic problems become more important as the
train speed increases.
In general, the train aerodynamics are related to
aerodynamic drag, pressure variations inside train,
traininduced ows, crosswind effects, ground effects,
pressure waves inside tunnel, impulse waves at the exit
of tunnel, noise and vibration, etc. The aerodynamic
drag is dependent on the crosssectional area of train
body, train length, shape of train fore and afterbodies,
surface roughness of train body, and geographical
conditions around the traveling train. The traininduced
ows can inuence passengers on the platform and is
also associated with the crosssectional area of train
body, train length, shape of train fore and afterbodies,
surface roughness of train body, etc.
The pressure variations, occurring as two trains
intersecting each other, are related to passengers
comfort and safe traveling of train. These are dependent
1935 1945 1955 1965 1975 1985 1995
100
200
300
400
500
DC
EL
Inter City Express(ICE)
Train de Grande Vitesse(TGV)
Shinkansen
High Speed Train(HST)
205
186km
215
243
331
256
253
286
318
380
319
317
345
406.9
336
325.7
EL
EL
482.4
515.3
ICE
Model
Shinkansen
TGV
TGV
TGVA
TGVA
ICE
ICE
S
p
e
e
d
(
k
m
/
h
)
Year
DC : diesel locomotive
EL : electric locomotive
Fig. 4. Progress of railway train.
Speed type
Maximum speed
Brake
performace
Power
Brake system
Body
Track
Tunnel
Electric lines
Electric facilities
Signal
communication
Travel equipment
Train
Railroad
Electricity
Environmental
problem
Passenger
comfort
Control
performance
Pantograph
performance
Traveling
performance
Retrogression
performance
Adjustable
velocity
Hardware
Limiting factors
Foundation Elements
of Railway system
Speed on
curved track
Speed on
branched track
Fig. 5. Factors limiting the speedup and related factors.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 474
on the shape of train fore and afterbodies, train width,
and the distance between track lines. The crosswind can
also inuence the safe traveling of the train, relating to
train height and perimeter, bridge system, etc.
The impulse wave at the exit of tunnel inuences the
surrounding area around the train track and is
dependent on the crosssectional area of train body,
the crosssectional area of tunnel, the shape of train
fore and afterbodies, the tunnel length, the kind of
track, etc. The pressure variations inuence the struc
tural strength of the train body, passengers comfort,
and are associated with the crosssectional area of the
train body, crosssectional area of tunnel, train length,
tunnel length, etc. Table 1 lists these major aerodynamic
problems of HST and associated factors. All of these
aerodynamic problems are closely related to the train
shape, which is required to produce aerodynamically
good characteristics.
4. Aerodynamic forces on railway train
4.1. Aerodynamic drag of train
The aerodynamic characteristics of HST are quite
different from those of airplane. There are many
characteristic features in the aerodynamics of the high
speed railway train, in the points that the train length is,
in general, very long, compared with the equivalent
diameter of it, the train runs close to adjacent structures,
passes through a conned tunnel, and intersecting with
each other, the train runs along a xed railway track,
always interacting with ground, and the train can be
inuenced by crosswinds. Thus, the aerodynamics,
which has been applied to airplane, may not be of help
for a detailed understanding of the HST aerodynamics.
In general, a desirable train system should be
aerodynamically stable and have low aerodynamic
forces. These aerodynamic characteristics are closely
associated with the aerodynamic drag of the running
train. The aerodynamic drag on the traveling train is
largely divided into mechanical and aerodynamic ones.
Of both, the aerodynamic drag can inuence the energy
consumption of train. Thus, detailed understanding on
the aerodynamic drag and its precise evaluation are of
practical importance.
It has been well known that the aerodynamic drag is
proportional to the square of speed, while the mechan
ical drag is proportional to the speed. Compared with
the mechanical drag, the portion of the aerodynamic
drag becomes larger as the train speed increases. Thus,
reduction of the aerodynamic drag on highspeed
railway train is one of the essential issues for the
development of the desirable train system.
In the open air without any crosswind effects, the
total drag on the traveling train can be expressed by a
sum of the aerodynamic and mechanical ones [5]:
D D
M
D
A
a bVW cV
2
; 6
where D
A
and D
M
are the aerodynamic and mechanical
drags, respectively, a; b and c are the constants to be
determined by the experiment, V the train speed and W
the train weight. In Eq. (6), the mechanical drag, being
proportional to the train weight, includes the sliding
drag between rails and train wheels, and the rotating
drag of the wheels.
The measurement of the total drag on train and its
precise prediction are not straightforward. The total
drag can be obtained by using a deceleration speed of
train or the consumed electric power, as will be
described later. Fig. 6 shows a typical example of the
measured total drag on train [5]. All of the data
Table 1
Aerodynamic problems and their related matters
Aerodynamic problems Related matters
1 Aerodynamic drag of train Maximum speed, energy consumption
2 Aerodynamic characteristics of train due to crosswinds Safety in strong crosswinds
3 Aerodynamic force due to passingby of two trains Running stability,
Quality of comfort for passengers
4 Winds induced by train Safety for passengers on platforms,
Safety for maintenance workers
5 Pressure variations in tunnels Quality of comfort for passengers,
(Ear discomfort)
Airtightness of vehicle,
Stress upon vehicle,
Ventilating system of vehicle
6 Micropressure waves radiating from tunnel exit Environmental problems near tunnel exit
7 Ventilation and heat transfer in underground station and tunnel Quality of comfort for passengers,
Prevention of disaster (re)
8 Aerodynamic noise Environmental problems
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 475
indicated refer to a given train with the same weight and
length, and these collapse onto a single line, given by a
curve t, D 12:484 0:04915 0:001654V
2
:
In order to reasonably estimate the total drag on the
trains with variable weight and length, it is necessary to
divide it into the mechanical and aerodynamic drags.
For instance, the leastsquares method can be used to
correlate the data, and consequently the term propor
tional to the square of speed can be considered as the
aerodynamic drag. However, it is quite difcult to
reasonably extract the aerodynamic drag from the total
drag, since it can contain natural wind effects, and
additionally, it can depend on the methods of how to get
the secondorder polynomials for the correlation as well.
4.2. Estimation of aerodynamic drag
Unlike the aerodynamics of airplane, the train runs
along a xed track, strongly interacting with surround
ing structures, ground, tunnel, platform, etc. Especially,
the presence of an intersecting train makes the analysis
of the train aerodynamics extremely difcult.
In order to speed up the train, it is necessary that the
electric motor power increases or the aerodynamic drag
reduces. Compared with the open air traveling, the
aerodynamic drag can considerably increase as the train
passes through a tunnel [6]. This is because the train
induced ows do work to increase the pressure by
interacting with the tunnel walls. A pantograph system
may produce the aerodynamic drag corresponding to
that caused by one train. In particular, the structures
underneath the train may produce more drag.
In the open air traveling, the aerodynamic drag on
train can be divided into two contributions; one is
dependent on train length and the other is independent
of it. The drag independent of the train length is the
pressure drag caused by the fore and afterbodies of
train. It is not easy to estimate the drag dependent on
the train length. This is because the friction drag on the
train body should involve all kinds of the drags
occurring in the connecting parts between trains,
photographs, the structures under the train, etc.
In this case, the aerodynamic drag can be expressed
as [7]
D
1
2
rA
0
V
2
C
dp
l
0
d
0
c
_ _
; 7
where V is the train speed, r the density of air, A
0
the
crosssectional area of train, C
dp
the coefcient of the
pressure drag caused by the fore and afterbodies of
train, d
0
the hydraulic diameter of train, l the train
length, and l
0
the hydraulic friction coefcient caused by
the connecting parts between trains, photographs, the
structures under the train, etc.
In general, a wind tunnel test for measuring the
aerodynamic drag on train, which is quite long
compared with the equivalent diameter, is highly
difcult. The use of a small model for the train wind
tunnel test causes several problems associated with lower
Reynolds numbers. In addition, ground effects on the
aerodynamic drag should be considered in the wind
tunnel test.
In Eq. (7) C
dp
can be obtained by wind tunnel
experiment. Using the real train entering into the tunnel,
Hara [8] reported that l
0
could be obtained by the
pressure rise on the train body, when it enters into
tunnel. A train entering into the tunnel compresses the
atmospheric air ahead of the train and the resulting
compression waves will propagate nearly at the speed of
sound towards the exit of the tunnel.
Meanwhile, the air displaced by the train entering into
the tunnel will discharge back from the entrance of
tunnel. In this case, the air ow should overcome the
frictional drag on the train body and tunnel walls,
resulting in a pressure gradient. Due to this fact, the
pressure on the train body will increase as the train
proceeds into the tunnel, as schematically shown
in Fig. 7.
The compression waves propagating along tunnel will
discharge from the exit of the tunnel, consequently
forming an impulse wave, as will be described later. At
the same instant as the compression waves discharge
0
50
100
150
200
0 50 100 150 200 250 300
D=12.484+0.04915V+0.001654V
2
Sinkansen 100series
Speed (km/h)
T
r
a
v
e
l
i
n
g
d
r
a
g
(
N
)
Fig. 6. Traveling drag on Shinkansen (series 100).
x
U
po
p2(AA')
v2
f
F
po(AA')
3 2
Fig. 7. Flow model for friction coefcient.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 476
from the exit of the tunnel, expansion waves will be
formed to meet the mass conservation law, and then
propagate back from the exit of the tunnel towards the
entrance of tunnel. In Haras analysis, it is assumed that
the expansion waves do not interact with the train body
inside tunnel and the afterbody of the train is still in the
open air.
In Fig. 7, the momentum equation can be applied to
the control volume between the crosssections 2 and 3,
as below
p
2
p
0
A A
0
f
0
F 0; 8
where p
2
is the pressure on the train body, p
0
the
atmospheric pressure, A the crosssectional area of
tunnel, A
0
the crosssectional area of train, f the
frictional force on train body, and F the frictional force
on tunnel walls. Eq. (8) can be changed to [8]
p
2
p
0
1 R
1
2
rV
2
c R
l
0
d
0
c
v
2
V
_ _
2
l
d
v
2
V
_ _
2
_ _
; 9
where R is the ratio of crosssectional areas of train to
tunnel, l the distance from the entrance of tunnel to
train, l the hydraulic friction coefcient on tunnel walls,
l
0
the hydraulic friction coefcient on the train body, d
the hydraulic diameter of tunnel, d
0
the hydraulic
diameter of train, V the train speed and v
2
the air ow
velocity occurring between train and tunnel walls. For a
known value of v
2
=V; the hydraulic friction coefcient l
0
on the train body can be obtained by measuring the
pressure rise (p
2
p
0
).
4.3. Comparison of the drags on different trains
The aerodynamic drag measurement results [9], which
were conducted using a wind tunnel test in France, are
summarized in Table 2, where each of the contributions
of the train body (TGV), the connecting part between
trains and the structures under the train on the
aerodynamic drag are indicated. The wind tunnel test
was carried out at a train speed of 260 km/h under
standard atmospheric conditions. The total drag is also
presented on the right side of Table 2.
Of the total drag, the aerodynamic drag only on the
train body is about 80%, the aerodynamic drag due to
the pantograph system and other devices over the train
is 17%, the rest drag of 3% is due to the mechanical
drag caused by the brake system, etc. From the
measured data above, the total traveling drag D on
train is given by D A BV CV
2
; where the
constants A and B are experimentally given by 250
and 3.256, respectively. Note that these values are 5%
and 17.1% of the aerodynamic drag on only the train
body, respectively.
Figs. 8 and 9 present the aerodynamic drag on the
Germany ICE [9]. The type of the ICE, its cross
sectional area and aerodynamic drag are also indicated
in Fig. 8. For example, the train of type a has a cross
sectional area of 14.61 m
2
and is assumed that its
aerodynamic drag is 100%. For the trains of different
types and crosssectional areas, relative aerodynamic
drag is given based on the train of type a. In the case of
the train of type k, the crosssectional area of the train is
11.39 m
2
and the aerodynamic drag relative to the type a
Table 2
Wind tunnel experiment for the traveling drag on TGV
Components of traveling drag Drag coefcient (151C,
1013 mbar)
Total drag
Drag in 260 km/h Power in
260 km/h (kW)
N/(km/h)
2
(%) N (%)
Aerodynamic component Drag of train 0.04595 80 3106 62.5 2243
Equipments on train
roof
0.00965 17 652 13.1 471
Total aerodynamic
components
0.05560 97 3758 75.6 2714
Disk brake 0.00170 3 115 2.3 83
Total 0.05730 100 3873 77.9 2797
Rolling drag (train
407 ton)
BV 3:256V 847 17.1 612
A 250 250 5.0 181
Total drag D A BV CV
2
4970 100.0 3590
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 477
is 78%. In Fig. 8, note that the trains of different cross
sectional areas have a train body mount system and a
skirt system to smooth the structures underneath the
train.
For the trains shown in Fig. 8, each portion of the
contributions of the fore and afterbodies of train, the
connecting part between trains, the train wall surfaces,
the pantograph system, etc. to the total aerodynamic
drag is given in Fig. 9 [9]. In the case of type a, it is
found that the aerodynamic drag caused in the
connecting part between trains is about 4%, the surface
friction drag 23%, the fore and afterbodies 8%, the
pantograph system 7%, and the underneath structures
of train 58%. For type f ; the total aerodynamic drag is
about 50% of that of type a; and each of the portions is
quite different from that of type a. For reference, all
these data refer to the same train length (200 m).
In the case of Japanese Shinkansen, the 0 series have
been known as l
0
0:018; and C
dp
0:2 [10,11]. An
empirical equation to predict the aerodynamic drag on
the traveling train is given by the following equation:
D 1:2 0:022VW 0:013 0:00029 cV
2
; 10
where D is the total drag (kg
f
), V the train speed (km/h),
W the total weight of train (ton), and l the train length
(m). The term that is proportional to the square of speed
is the aerodynamic drag. For l 400 m, corresponding
to the length of 16 trains, about 90% of the aerodynamic
drag is attributed to the friction drag on the middle part
of the train. Of the Japanese Shinkansen, series 100 has a
semibody mount system for the underneath structures,
and series 200 has an underneath coverage to prevent
snow accumulation on the train body. Table 3 lists some
major parameters inuencing on the aerodynamic
drag [10,11]. It is found that smoothing the under
neath structures of train by using the body mount
system or the skirt system reduces the hydraulic friction
coefcient.
4.4. Pressure drag
Of the aerodynamic drag components, the pressure
drag comes from the pressures on the fore and after
bodies of train, and, in the case with a double deck in
train series, it stems from the pressures due to the abrupt
change in the crosssectional area of the train. Assuming
that the coefcient of the pressure drag is C
dp
and in the
case of a double deck, it is C
dpd
; a wind tunnel test [12]
has been carried out to investigate the pressure drag. In
the experiment, each of the models for the fore and
afterbodies of train and the middle part of train has
been manufactured. Depending on the length of the
middle part of the model, the model train experiment of
different lengths could be done.
For a train model with a short length, the coefcient
of the aerodynamic drag can be written as
C
ds
D
s
1
2
rU
2
A
0
_ _; 11
15
10
5
0
a b c e
f
4%
23%
8%
7%
58%
2%
35%
10%
9%
44%
44%
5%
11%
40%
52%
14%
7%
27%
61%
15%
2%
22%
Type
C
r
o
s
s

s
e
c
t
i
o
n
a
l
a
r
e
a
(
m
2
)
Surface
friction
Pantograph
Head & tail
of train
Underneath
structures
Connecting
parts
Fig. 9. Aerodynamic drag components of ICE.
a
b
c
d
e
f
14.61 100%
11.39 78%
10.14 69%
8.70 60%
7.80 53%
7.10 49%
Type Crosssectional
area (m )
2
Drag(%)
Moving direction
(m
2
)
Fig. 8. Aerodynamic drag on ICE (the hatching area is the
device to smooth the structures underneath train).
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 478
where C
ds
is the coefcient of the aerodynamic drag for
the short train model, D
s
the aerodynamic drag, A
0
the
crosssectional area of the model train, U the wind
velocity, and r the density. Furthermore, the coefcient
of the aerodynamic drag can be divided into two
components: one depending on the length of the model
train and the other independent of the length of the
model train, as
C
ds
C
dp
l
0
m
c
s
d
0
; 12
where C
dp
is the coefcient of the pressure drag on the
fore and afterbodies of the model train, l
0
m
the
coefcient of the hydraulic friction on the model train,
l
s
the length of the model train, and d
0
the hydraulic
diameter of the model train.
Meanwhile, for a long model train, the coefcient of
the aerodynamic drag is similarly given
C
dl
C
dp
l
0
m
c
l
d
0
; 13
where l
l
is the length of the longer model train. From
Eqs. (12) and (13), the coefcient of the pressure drag
C
dp
can be obtained by
C
dp
C
ds
c
l
C
dl
c
s
c
l
c
s
: 14
In reality, the coefcient l
0
m
of the hydraulic friction
on the model train can be signicantly different from
that of real train, as will be described later.
4.5. Friction drag
Of the aerodynamic drag components, the estimation
of the friction drag is more complicated, compared with
that of the pressure drag. The friction drag comes from
the train walls, the pantograph system, the connecting
part between trains, other devices on train roof, etc.
As described previously, the friction coefcient can be
obtained by the pressure rise on the train body entering
into tunnel.
Fig. 10 shows the experimental data of the stagnation
pressure and the pressure rise on the train, which enters
into tunnel [8,10], where a refers to the state just after the
forebody of the train enters into tunnel, b just after the
afterbody of the train enter into tunnel, and c refers to
the instant that the compression waves occurring when
the train forebody enters into tunnel are reected back
from the exit of tunnel as the expansion waves. In order
to estimate the hydraulic friction coefcient of a train,
we can use the pressure rise from the instant a to b:
Fig. 11 represents the pressure distributions measured
on the train body for the time range between a and b
[8,10]. For the sake of simplicity, the forebody of the
train is assumed to be x 0; where x is the distance from
the entrance of tunnel to the forebody of the train. It is
found that the pressure increases nearly linearly. How
ever, in reality, the pressure near the forebody of the
train may deviate from the linear distribution due to the
forebody effects. In this case, we can eliminate the fore
body effects to precisely estimate the friction coefcient,
as schematically shown in Fig. 12, where the compression
waves due to the train forebody entering into tunnel do
not yet reach at the exit of the tunnel.
For the oweld shown in Fig. 12, we can again
assume that the ow is onedimensional. Neglecting the
viscous forces on the ows between the train forebody
and the compression waves, the isentropic relation and
the simple wave theory can be written as
p
1
r
g
1
p
0
r
g
0
; 15
2
g 1
gp
1
r
1
_
u
1
2
g 1
gp
0
r
0
;
_
16
Table 3
Parameters associated with aerodynamic drag (Shinkansen)
Series Crosssectional
area A
0
(m
2
)
Hydraulic
diameter d
0
(m)
Friction
coefcient of
train side body
(l
0
)
Pressure drag
coefcient C
dp
Train 0 12.6 3.54 0.017 0.20
200 13.3 3.64 0.016 0.20
100 12.6 3.54 0.016 0.15
Tunnel Line Crosssectional
area A (m
2
)
Hydraulic
diameter d (m)
Friction coefcient of tunnel wall (l)
Tokaido 60.5 7.8 0.02
Sanyo 63.4 8.1 0.02
Tohoku
Joetsu
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 479
where p is the pressure, u the velocity, r the density, g the
ratio of specic heats of air, and subscript 0 and 1
denote the states just before the compression waves and
just before the train forebody, respectively. In addition,
the ow properties just before and after the train fore
body can be obtained by the equation systems of mass
a
b
c
1
k
P
a
1.0sec
U
p
Train
Tunnel
Time(t)
Pressure distribution for the steady flow inside tunnel
Pressure on the side body of the 1st train
Pressure on the side body of the 2nd train
Stagnation pressure on the lst train
Time(t)
P
r
e
s
s
u
r
e
Pressure on the side body of the 4th train
Pressure on the side body of the 6th train
Fig. 10. Pressure rise due to the train entering into tunnel (U 220 km/h, and Shinkansen series 100).
3.0
2.0
1.0
0.0
400 300 200 100 0
Train speed, U=216km/h
Slab track
Tunnel
entrance
P
r
e
s
s
u
r
e
r
i
s
e
o
n
t
h
e
s
i
d
e
b
o
d
y
o
f
t
r
a
i
n
(
k
P
a
)
Train
head
Distance from tunnel entrance to train head, X(m)
Tunnel
entrance
Tunnel
entrance
Tunnel
entrance
Tunnel
entrance
Tunnel
entrance
Fig. 11. Pressure distributions on the side body of train.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 480
and energy conservation:
r
1
# u
1
r
2
# u
2
1 R; 17
g
g 1
p
1
r
1
1
2
# u
2
1
g
g 1
p
2
r
2
1
2
# u
2
2
g
g 1
p
s
r
s
; 18
p
1
r
g
1
p
2
r
g
2
p
s
r
g
s
; 19
# u
1
u
1
U; 20
# u
2
u
2
U; 21
where R is the ratio of crosssectional areas of train to
tunnel, U the train speed, and subscripts 2 and s denote
the state just after the train forebody and the stagnation
state, respectively. The ^ over the ow velocity u
indicates the ow velocity observed by the moving
coordinate system.
For the region between the train forebody and the
entrance of tunnel, the momentum equation reduces
to
p
2
p
0
1 R
1
2
r
2
U
2
x R
l
0
d
0
1
u
2
U
_ _
2
l
d
u
2
U
_ _
2
_ _
;
22
where x is the distance from the tunnel entrance to
the train forebody and R is the effective crosssectional
area ratio of train to tunnel, which can be obtained
from p
s
; p
2
; U; p
0
; and r
0
in the equations described
above.
Fig. 13 shows an experiment to obtain the effective
area ratio R from the Japanese Shinkansen test [13]. A
great deal of scattering is found in R; but it seems
that the experimental data have a certain trend
independent of x: Here we can get a mean value for R;
which is dened as the effective crosssectional area ratio
of train to tunnel. In order to obtain the hydraulic
friction coefcient of train, we can solve the coupled
equation systems from Eq. (15) to (22) for the known
values of U; p
0
; d
0
; d; l and r
0
; and the measured values
of p
2
and R:
Fig. 14 presents the value of the hydraulic friction
coefcient l
0
of train [13]. It is found that the scattering
in the data is quite big, but l
0
becomes nearly constant
with an increase in x: For a small x; the estimation of l
0
can lead to much error. Thus, we can use a mean value
of l
0
as the hydraulic friction coefcient for a larger x
range.
50 150 250 350 450
0.1
0.2
0.3
C
r
o
s
s

s
e
c
t
i
o
n
a
l
a
r
e
a
r
a
t
i
o
o
f
t
r
a
i
n
t
o
t
u
n
n
e
l
(
R
)
Distance from tunnel entrance to train head, X(m)
Fig. 13. Relationship between R and X (Shinkansen 0 series).
u1
x
xi
p2
u2
p1
po
p3
uo(=0)
1
2
3
U
po
u1
u2
0
po
Measuring point of pressure
Stagnation pressure of train
Head of waves
Tunnel entrance Tunnel
Time(t)
Time(t)
P
r
e
s
s
u
r
e
V
e
l
o
c
i
t
y
Fig. 12. Floweld made by the train entering into tunnel.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 481
5. Aerodynamic shape of railway train
5.1. Wind tunnel test
Detailed conguration of the fore and afterbodies of
train can signicantly inuence the aerodynamic char
acteristics. An example of typical wind tunnel test using
model trains is shown in Fig. 15, where the experiments
yield 16 different kinds of the fore and afterbodies of
model trains [14,15]. The forebody of the model train is
the same shape as the afterbody. The conguration of
the model train is characterized by number 1, 2, 3 and 4,
and the characters A, B, C and D. For the sake of
simplicity, here we dene the numbers as train series and
the characters as train types.
In addition, the number refers to the length of the
change in crosssectional area of the model train. For
instance, the length of the change in the crosssectional
area of the model train becomes longer with the number;
for further details, series 1, 2, 3 and 4 have the lengths of
the change in the crosssectional area of the model train
0.5 times, 1 time, 2 times, and 4 times the model width,
respectively. It is assumed that the length of the change
in the crosssectional area for the A type of series 1 is
zero.
Meanwhile, the model congurations of the A and B
types are nearly twodimensional, but those of the C and
D types are nearly threedimensional. For further
details, see Fig. 15. In real highspeed railway trains,
the ICE is close to A or B type, but the Shinkansen
(Series 100) close to C or D type.
These systematic train models were tested in a
subsonic wind tunnel, which is schematically shown in
Fig. 16 [14,15], where the fore and afterbodies are
installed onto the wind tunnel test section to measure the
aerodynamic forces on them. The fore and afterbodies
of the train model are always combined with the middle
part and the dummy fore and afterbodies, thus
forming a full model train. Fig. 17 shows the effects
of the forebody conguration on the aerodynamic
drag [14,15], where the middle part and dummy after
body are xed. Fig. 18 presents the inuences of the
50 150 250 350 450
0.0
0.02
0.04
Distance from tunnel entrance to train head, X(m)
H
y
d
r
a
u
l
i
c
f
r
i
c
t
i
o
n
a
l
c
o
e
f
f
i
c
i
e
n
t
o
n
t
r
a
i
n
(
)
'
Fig. 14. Relationship between l
0
and X (Shinkansen 0 series).
1A
1B
1C
1D
2A 3A 4A
2B 3B 4B
2C
3C
4C
2D 3D 4D
Fig. 15. Model train conguration.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 482
afterbody conguration on the aerodynamic drag
[14,15], where the middle part and dummy forebody
are xed. Note that the length of the forebody is
normalized by the width of the train model.
For series 1 of a comparatively short forebody, the
aerodynamic drag on types B and C is nearly the same,
and their C
d
values are relatively low, compared with
types A and D. It is found that the forebody of type A
has the highest C
d
value. For a given type of the train
model, it is interesting to note that the aerodynamic drag
does not change for L=W values larger than 1.0.
It seems that the afterbody effects on the aerody
namic drag are more sensitive to L=W; compared with
those of the forebody. For a given type of the model
train, series 4 has the lowest C
d
value. In the case of the
same series of the model train, the aerodynamic drag on
type D is the lowest. The fore and afterbodies effects
on the aerodynamic drags are associated with ow
separations, which can inuence the traininduced ows,
as will be described next.
5.2. Traininduced ows
The winds induced by a traveling HST can affect
passengers at platform and the structures around the
railway lines. Thus, estimation of the traininduced ows
1500
3
8
5
8
0
0
4
1
5
1
3
1
332 400 613
1
3
.
5
300
1
0
3
50m/s
347
141
347
1
3
1
(a) Aerodynamic drag measurement of train head
1500
3
8
5
8
0
0
4
1
5
1
3
1
332 400 613
1
3
.
5
300
1
0
3
50m/s
347
(b) Aerodynamic drag measurement of train tail
Head
Middle
Tail(dummy)
Exit of windtunnel
Head(dummy)
Middle
Tail
Pitot tube
Exit of windtunnel
Pitot tube
Pitot tube
Balance meter
Balance meter
Fig. 16. Test rig for aerodynamic drag on train.
:1A,2A,3A,4A
:1B,2B,3B,4B
:1C,2C,3C,4C
:1D,2D,3D,4D
0.0 1.0 2.0 3.0
4.0
0.0
0.2
0.8
0.4
0.6
3C type
Middle(dummy)
Head
D
r
a
g
c
o
e
f
f
i
c
i
e
n
t
(
C
)
D
L / W
Fig. 17. Effect of L=W on aerodynamic drag.
:1A,2A,3A,4A
:1B,2B,3B,4B
:1C,2C,3C,4C
:1D,2D,3D,4D
0.0 1.0
2.0
3.0 4.0
0.0
0.2
0.8
0.4
0.6
3C type
Middle
Tail
(dummy)
L / W
D
r
a
g
c
o
e
f
f
i
c
i
e
n
t
(
C
)
D
Fig. 18. Effect of L=W on aerodynamic drag.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 483
should be included in a structural design of the platform
and surrounding facilities around the railway lines. For
a given speed of train, the traininduced ows are
strongly dependent on the forebody conguration, and
train length. In particular, the traininduced ows can
differ, depending on whether the forebody shape is two
or threedimensional.
Fig. 19 shows an example of the experimental test rig
for the traininduced ows [14,15]. In order to simulate
the traininduced ows occurring at a platform, a model
platform is placed on one side of the model train, in
which a hotwire system is installed at a height of 1.20 m
from the ground surface, and the traininduced ows at
a certain distance away from the model train is
measured. The traininduced ow data are presented,
subtracted the wind tunnel air velocity from the
measured velocities.
The traininduced ows due to the forebody are like a
source ow, while those due to the afterbody are like a
sink ow. This phenomenon is more striking for two
dimensional shapes rather than threedimensional ones.
The effects of the fore and afterbodies on the train
induced ows are shown in Fig. 20 [14,15]. It is found
that the traininduced ows become small with an
increase in L=W: For a given series, types A and B
produce lower traininduced ows, compared with types
C and D. Note that types C and D are close to three
dimensional shapes. In types A and B, a more train
induced ows will pass over the model train roof, while
in types C and D, the ow that spills over the model
train can be of the same magnitude as it passes beside
the train.
For series 13, the traininduced ows produced in
type D are larger than those in type C. This is attributed
30m/s
(a) Measurement of train head
(b) Measurement of train tail
(c) Train and platform
158
5
0
1
1
2
70
30m/s
Pitot tube Pressure measurement
Head Middle
(dummy)
Model platform
Exit of windtunnel
Model platform
Middle
Tail
Pressure measurement
(dummy)
Exit of windtunnel
Ground plate
Measuring point of pressure
Model platform
Fig. 19. Test rig for traininduced ow measurements.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 484
to a large wall curvature existing between the forebody
of type D and the middle part.
Fig. 21 shows the effects of the afterbody of the
model train on the traininduced ows [14,15]. It is
found that the traininduced ows are not strongly
dependent on the type of the model train used, but the
series of it somewhat affects the traininduced ows,
which are the lowest for series 4. It is worthwhile noting
that for series less than L=W 4:0; the traininduced
ows increase with an increase in L=W: In the case of
series 1, the traininduced ows are nearly the same,
regardless of the types used, since for all of the types, the
ow separates from the connection part between the
afterbody and middle section. However, for series 2
and 3, the ow goes along the model surface without
appreciable separation, resulting in relatively larger
traininduced ows.
In real highspeed trains, the traininduced ows can
be inuenced by the train afterbody shape as well as
boundary layer ows. Thus, the model test results can be
somewhat different from the real circumstances.
A further study is needed to resolve these problems.
5.3. Aerodynamic forces due to trains passing each other
When a train travels in the open air, it induces a very
complicated oweld. The implications of this oweld
on trackside structures and passing trains are much
more important for trains than for other vehicles, since
trains operate much closer to adjacent structures or
other trains.
In order to investigate the trains passing each other in
the open air, using the wind tunnel, we can simulate the
passing trains using a xed side plate. Here, we discuss
about the oweld forming on a vertical at plate placed
beside the model train. Fig. 22 schematically illustrates
the experimental test rig to simulate the trains passing
each other [14,15]. The side plate has a multiple of
pressure taps. When a train passes another train in the
open air, considerably large pressure uctuations occur
on the sidewall of the train. It is known that a positive
negative pulse pressure is generated as the forebody of a
train passes another train, while a negativepositive
pulse as the afterbody passes.
In order to compare the model test of the side plate
with the pressure variation on real trains which pass
each other in the open air, the spatial axis can be
changed into the time axis. Let us focus on the peak
pressure values of the pressure variations on the side
body of the trains, occurring when the fore and after
bodies of the train pass each other. Fig. 23 presents the
peak pressures produced when the forebody of train
passes each other [14,15]. It is found that the positive
peak value of the pressure variation reduces as the
length of the forebody of train increases, and type A
produces the highest peak pressure, compared with
other types of train model. This is because type A
generates the biggest stagnation region on the forebody
of the model train.
Meanwhile, the absolute value of the negative peak
pressure decreases as the length of the forebody
increases, and it is relatively low in types A and B,
compared with types C and D. This can be understood
by the fact that the curvature of the streamlines along
the side body of the model train is low in types A and B
which are close to the twodimensional shape.
Fig. 24 shows the negative peak pressures on the side
body of the model train produced when the afterbody
of the model train passes each other [14,15]. It is found
that the absolute value of the negative peak pressure
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.0 1.0 2.0 3.0
4.0
:1A,2A,3A,4A
:1C,2C,3C,4C
:1D,2D,3D,4D
L / W
N
o
n
d
i
m
e
n
s
i
o
n
a
l
f
l
o
w
v
e
l
o
c
i
t
y
(
u
/
U
)
Fig. 21. Effect of train tail on traininduced ows.
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.0 1.0 2.0 3.0 4.0
:1A,2A,3A,4A
:1C,2C,3C,4C
:1D,2D,3D,4D
:1B,2B,3B,4B
L / W
N
o
n
d
i
m
e
n
s
i
o
n
a
l
f
l
o
w
v
e
l
o
c
i
t
y
(
u
/
U
)
Fig. 20. Effect of train head on traininduced ows.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 485
50m/s
(a) Measurement of train head
(b) Measurement of train tail
(c) Model train and side plate
105
5
0
1
0
0
70
18
50m/s
1
1
Pressure measurement
Head
Middle
(dummy)
Side plate
Pitot tube
Ground
Exit of windtunnel
Middle
Tail
Pressure measurement
Side plate
Ground
(dummy)
Exit of windtunnel
Side plate
Pressure measurement
Ground
Fig. 22. Test rig for pressure variation measurement.
0.4
0.2
0.0
0.2
0.4
0.6
0.8
1.0
0.0 1.0 2.0 3.0 4.0
:1A,2A,3A,4A
:1B,2B,3B,4B
:1C,2C,3C,4C
:1D,2D,3D,4D
L / W
P
r
e
s
s
u
r
e
c
o
e
f
f
i
c
i
e
n
t
(
C
)
p
Fig. 24. Effect of train tail on pressure variation.
0.4
0.2
0.0
0.2
0.4
0.6
0.8
1.0
0.0 1.0 2.0 3.0 4.0
:1A,2A,3A,4A
:1B,2B,3B,4B
:1C,2C,3C,4C
:1D,2D,3D,4D
L / W
P
r
e
s
s
u
r
e
c
o
e
f
f
i
c
i
e
n
t
(
C
)
p
Fig. 23. Effect of train head on pressure variation.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 486
produced when the afterbodies of the model train pass
each other is lower than that created when the fore
bodies of the model train pass each other. The length of
the afterbody somewhat affects the negative peak
pressures, but it does not inuence the positive peak
pressure. It is, thus, concluded that the model trains
which are close to the twodimensional shape produces
the positive and negative peak pressures less than those
of the threedimensional shape.
In the open air, the pressure variations on the side
bodies of real trains passing each other are shown in
Fig. 25 [5], where two cases are compared with each
other; The speed V of one train is zero while the speed U
of the other train is 260 km/h, traveling towards the
right side, as shown in Fig. 25(a). In Fig. 25(c), the speed
V of one train is 210 km/h, while the speed U of the
other train is 260 km/h. The pressure variations are
measured on the middle part of each train.
It is found that the positivenegative pressure varia
tion like a pulse wave is produced as the forebodies of
real trains pass each other, while the negativepositive
pressure variation is created as the afterbodies pass
each other. These pressure variations on the side body of
trains are strongly dependent on the detailed shape of
the fore and afterbodies and the speed of each train,
and can cause the yawing motions of the traveling trains.
It is known that the peak pressures produced by the
trains passing each other is proportional to the square of
the speed of trains and the timewidth of the peak
pressures is proportional to an inverse of the sum of the
speeds of each train. Here it should be noted that in
the open air, the pressures, produced opposite side to the
trains passing each other, nearly remain constant at
atmospheric pressure without any appreciable uctua
tions, as shown in Fig. 25(b).
p
(
k
P
a
)
0
0.5
1.0
0.5
1.0
U=260km/h, V=210km/h
0.5s 0.0s 1.0s 1.5s
U=260km/h, V=0km/h
0
0.5
1.0
0.5
1.0
p
(
k
P
a
)
0.5s 0.0s 1.0s 1.5s 2.0s 2.5s
0
0.5
U
V
A
B
(c)
(b)
(a)
Passing train head
Passing train tail
Pressure fluctuation
Pressure fluctuation (opposite side of the train)
Pressure fluctuation
Time(sec)
Time(sec)
Fig. 25. Pressure variations occurring when two trains pass each other.
1500
2
5
0
0
1
8
7
1
2
1
12
693 333
1390
39
333
Dummy
Dummy
Measuring trains
Rail
Bridge
Rotating end plate
Rotating end plate
End plate End plate
T
h
r
e
e
c
o
m
p
o
n
e
n
t
f
o
r
c
e
m
e
t
e
r
Fig. 26. Test rig for crosswinds.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 487
5.4. Crosswind effects
The crosswind effects on the traveling train can
closely be associated with the traveling safety. The cross
winds can be more seriously inuence when the train
runs over a bridge [1618]. Fig. 26 shows the typical
example of a 1/30 model test to investigate the three
component forces on the train running over a bridge
[19]. The wind tunnel has a dimension of 1.5 m2.5 m.
In tests, the wind speed and its angle of attack are
changed, the resulting Reynolds number being in the
range of 6 10
4
8 10
4
. Table 4 summarizes the results
of the wind tunnel tests for the coefcients of drag (C
d
),
lift (C
l
) and pitching moment (C
M
), compared with a
computational prediction [19].
For the wind speed of 10 m/s, the coefcients of
drag (C
d
), lift (C
l
) and pitching moment (C
M
) are,
respectively, 1.4, 0.2 and 0.2 as the train runs in
the open air, while these are, respectively, 1.7, 0.6, and
0.3 as the train travels over the bridge. It is believed
that the drag and lift coefcients on the train become
much higher when it travels over the bridge. The
computations of the threecomponent forces only
qualitatively predict the measured aerodynamic forces
on the train.
6. Aeroacoustic problems of railway train
6.1. Aerodynamic noise due to train
For the assessment of aerodynamic noises produced
by a traveling train in the open air, it can be often
convenient to classify the noise sources. In addition to
the aerodynamic noises due to the ows around the
traveling train, there are many different noises which are
caused by train wheels, structures around track,
pantograph system, etc. In order to reduce these noises,
it is required to know how extent is each contribution to
the noises. In general, aerodynamic noises are strongly
dependent on the train speed U [2022], being approxi
mately proportional to U
6
2U
8
: Thus, the noise allevia
tion is of more practical importance when the train
speed increases.
Fig. 27 shows the aerodynamic noises produced by a
traveling train in the open air [2325]. It can be found
that the aerodynamic noises due to the traveling train
are largely generated by the forebody of the train, the
connection part between trains, and the panto
graph system. In practice, the pantograph system is
composed of many bars with small diameters, which can
play a musical instrument to create the aerodynamic
noises. The pantograph system creates a number of
vortices behind it. A pantograph cover can be used to
reduce the aerodynamics noises generated by the
pantograph system [24,25], but it can be of an additional
1 2 3 4 5 6 7 8 9 10 11 12
1
0
d
B
1.0s
(a) Typical noise level of highspeed train
(b) Outlook of highspeed train
U=235km/h
Time(t)
S
o
u
n
d
p
r
e
s
s
u
r
e
l
e
v
e
l
(
d
B
)
Downward pantograph
Upward pantograph
Pantograph
Pantograph cover
Air conditioning unit
Highvoltage cable connector
Fig. 27. Aerodynamic noise level of HST.
Table 4
Comparison of wind tunnel experiment and computational results
Train Train+bridge
Experiment Computation Experiment Computation
Drag coefcient C
d
1.4 1.4 1.7 1.7
Lift coefcient C
l
0.2 0.2 0.6 0.2
Pitching moment C
M
0.2 0.3
Reynolds number Re 56,000 5000 84,000 5000
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 488
aerodynamic drag and can be a source of additional
noise as well. There may be a big separation region
downstream of the pantograph cover, and consequently
it can cause the train body to vibrate, consequently
inuencing the traveling safety of train and passengers
comfort.
The forebody of a train is one of the noise sources. In
usual, there are a lot of roughness on the forebody
surface. The aerodynamic noises are strongly dependent
on the detailed conguration of the surface roughness
and the entire shape of train forebody as well. These
geometrical congurations are associated with the wind
speed along them and separation.
Fig. 28 shows a typical example of the aerodynamic
noise measurement at a location of 25 m away from
a traveling train [24,25], where the peak frequency
components generated by the forebody of train are
presented. It is found that the aerodynamic noises are
largely composed of highfrequency components. From
the point of view of the aerodynamic noises, it is
desirable that the forebody conguration of train
should have a long nose to reduce aerodynamic noises.
6.2. Wind tunnel test
The aerodynamic noises are almost always associated
with the aerodynamic drag. Reducing the aerodynamic
noises should be done without increasing the aerody
namic drag. In addition to the aerodynamic noises
generated by the forebody of train, the connection part
between trains has lots of component structures such as
ventilation system, pantograph system, etc. Of them, a
majority part of noises are generated by the pantograph
system.
Fig. 29 shows a typical measurement example of the
aerodynamic noises which are caused by the pantograph
system [24,25], where the pantograph system and
microphone array are schematically illustrated. The
locations of the microphone are indicated by M1M7.
The wind tunnel used has a test dimension of
3.0 m5.0 m, and its maximum wind speed and the
turbulence intensity are estimated by 270 km/h and
0.2%, respectively.
In general, the aerodynamic noises generated by the
pantograph system have some directivity towards the
ow direction, normal to the direction of the panto
graph length. Thus, the measurement locations M7 and
M3, which are, respectively, just over and beside the
pantograph, are employed to assess the aerodynamic
200 500 1000 2000 5000
1
0
d
B
U=270km/h
U=230km/h
Shinkansen (100 series)
1/3 Octave band frequency (Hz)
S
o
u
n
d
p
r
e
s
s
u
r
e
l
e
v
e
l
(
d
B
)
Fig. 28. Aerodynamic noise due to train head.
4000mm
1
1
6
0
m
m
5435mm
1
0
0
0
m
m
2
0
0
0
m
m
5
1
6
0
m
m
5000mm
4000mm
1
0
0
0
m
m
2
0
0
0
m
m
M1
M2
M3
M4
M5
M6
M7
M2,4,6
M1,3,5
M7
Exit of windtunnel Side view Front view
Fig. 29. Test rig for aerodynamic noise measurement.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 489
noises, as shown in Fig. 30. The wind speed is changed
between 100 and 260 km/h. It is found that the sound
pressure level (SPL) of the aerodynamic noises is closely
related to the wind speed, and the aerodynamic noises
generated by the pantograph system are composed of
large amplitude components of a wide band frequency.
This is because the pantograph system is very compli
cated and thereby, the noises being caused by many
different sources. For instance, the pantograph system is
composed of many bars of different diameters. Thus, the
different characteristic lengths to the aerodynamic
noises can be responsible for the wide band frequency.
6.3. Reduction of aerodynamic noise
In order to reduce the aerodynamic noises produced
by a traveling train, the forebody conguration of train
is needed to be of the long nose with a smooth surface,
and the middle part of train to be designed without any
sizable roughness. An aerodynamically welldesigned
shape of train can have such a noise level as low as in the
boundary layer shear ows. In practice, there is a limit in
making the strain surface aerodynamically smooth.
Fig. 31 presents a measurement example of the
aerodynamic noises which are generated by a two
dimensional body, like the crosssectional area of the
pantograph system [24,25]. A circular cylinder with a
diameter of 50 mm (see Fig. 31(a)) and a square cylinder
with the same equivalent diameter (see Fig. 31(b)) are
employed as the aerodynamic noise sources. The wind
speed is 200 km/h. The noise measurement is done at a
location beside the cylindrical body. In order to
investigate the surface roughness effects, the cylindrical
body is coated with a sponge, a carpet, and a thick cloth,
respectively. It is found that the SPL of the aerodynamic
noises in the square cylinder is higher than that
produced in the circular cylinder. For the circular
cylinder, it seems that coating the body surface some
what reduces the aerodynamic noises.
In the case of the square cylinder, the peak SPL and
its peak frequency for the cylinder with an equivalent
diameter of 100 mm are higher than those for the
cylinder of an equivalent diameter of 50 mm. In
addition, coating the cylinder surface reduces the
aerodynamic noise level.
63 125 250 500 1000 2000 4000 8000
50
60
70
80
90
110
100
50
60
70
80
90
110
100
63 125 250 500 1000 2000 4000 8000
100km/h
150km/h
260km/h
200km/h
100km/h
150km/h 260km/h
200km/h
1
/
3
O
c
t
a
v
e
b
a
n
d
s
o
u
n
d
p
r
e
s
s
u
r
e
l
e
v
e
l
(
d
B
)
Central frequency (Hz)
(a) M3
1
/
3
O
c
t
a
v
e
b
a
n
d
s
o
u
n
d
p
r
e
s
s
u
r
e
l
e
v
e
l
(
d
B
)
Central frequency (Hz)
(b) M5
Fig. 30. Aerodynamic noise due to pantograph system.
63 125 250 500 1000 2000 4000 8000
50
60
70
80
90
110
100
63 125 250 500 1000 2000 4000 8000
50
60
70
80
90
110
100
(a) Circular cylinder
(b) Square cylinder
U=200km/h
U=200km/h
1
/
3
O
c
t
a
v
e
b
a
n
d
s
o
u
n
d
p
r
e
s
s
u
r
e
l
e
v
e
l
(
d
B
)
Central frequency (Hz)
Central frequency (Hz)
1
/
3
O
c
t
a
v
e
b
a
n
d
s
o
u
n
d
p
r
e
s
s
u
r
e
l
e
v
e
l
(
d
B
)
: Wind tunnel noise level
: No coating
: Sponge coating
: Carpet coating
: Thick cloth coating
: Wind tunnel noise level
: No coating
: Sponge coating
: Carpet coating
: Thick cloth coating
Fig. 31. Aerodynamic noise in ows over circular and square
cylinders.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 490
Fig. 32 shows the effects of the shape of the cross
sectional area of the cylindrical body on the aerody
namic noises [24,25]. Several shapes of the cross
sectional area of pantograph are investigated for the
purpose of design of the pantograph system. The overall
SPL seems to be the lowest in the cylindrical body like
an elliptic shape. Fig. 33 presents the measurement
example of the aerodynamic noises generated by several
streamlined bodies [24,25]. It is found that the stream
lined body B has a peak frequency at 500 Hz, but the
body C, a peak frequency at 3.15 kHz. These peak
frequencies are due to the trailing edge vortices of the
streamlined bodies.
Compared with the overall SPLs generated in the
previous square and circular cylinders, for the stream
lined bodies the overall SPLs are signicantly lower. It
is, thus, believed that the aerodynamic noises can be
reduced if the pantograph system is designed as a
streamlined body.
7. Vibration of railway train
A considerable amount of the lateral vibration of a
train can be often found when the train travels at high
speeds in the open air. Such a lateral vibration becomes
more signicant near the trail of the train or in the train
equipped with the pantograph system [2628]. The
lateral vibration of the train can be an important factor
to the traveling performance. The Karman vortices
downstream of the train can be responsible for the
lateral vibration. The vertical ows are closely asso
ciated with the train length and detailed conguration of
the afterbody of train. For long trains, some structural
vibrations occurring in the leading coach can be one of
the reasons for the trailing coach to appreciably vibrate.
The study on the lateral vibration of train has not been
sufcient in the past.
8. Aerodynamics of railway train/tunnel systems
8.1. Aerodynamic analysis of train/tunnel systems
The aerodynamic problems occurring when train travels
at high speed in tunnel are more complicated and serious,
compared with the open air traveling. The aerodynamic
drag and noises on the train are strongly dependent on the
pressure waves in the tunnel. The aerodynamic drag on a
train traveling in a tunnel can signicantly increase,
compared with that in the open air [6,29,30].
When a HST enters a tunnel, a compression wave is
formed ahead of the train which propagates along the
tunnel at a nearly sonic speed. A part of the compression
wave is reected back from the exit of the tunnel as an
expansion wave. A complex wave interaction occurs
inside the tunnel due to successive reections of the
pressure waves at the exit and entry to the tunnel. These
pressure waves cause large pressure transients resulting
in uctuating loads on the train causing discomfort to
passengers. It is necessary to predict these pressure
transients to design trains and tunnels, and to improve
the passenger comfort.
Further, a part of the compression wave leaving the
tunnel exit gives rise to an impulse noise, as will be
described later. Such an impulse noise was not an
important issue in the past when the speed of trains was
not so high. But in recent years, with the increase in the
speed of trains the noise and vibration due to impulse
waves have become a new type of environmental noise
problem. According to some measurements conducted
near the exit of the tunnel, the noise is known to be of
low frequency of short duration, and its magnitude
50
60
70
80
90
100
110
63 125 250 500 1000 2000 4000 8000
U=200km/h
a) b)
c) d)
: a)
: b)
: c)
: d)
:
1
/
3
O
c
t
a
v
e
b
a
n
d
s
o
u
n
d
p
r
e
s
s
u
r
e
l
e
v
e
l
(
d
B
)
Central frequency (Hz)
Cross section
Windtunnel noise level
Fig. 32. Aerodynamic noise in ows over several cylinders.
40
50
60
70
80
90
100
63 125 250 500 1000 2000 4000 8000
U=200km/h
Central frequency (Hz)
1
/
3
O
c
t
a
v
e
b
a
n
d
s
o
u
n
d
p
r
e
s
s
u
r
e
l
e
v
e
l
(
d
B
) : Cross section a)
: Cross section b)
: Cross section c)
: Windtunnel noise level
a)
b)
c)
Fig. 33. Aerodynamic noise in ows over streamlined bodies.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 491
being approximately proportional to V
3
(where V is
train velocity). Moreover the impulse noise is closely
related to the detailed characteristics of the compression
waves inside the tunnel [3135].
Here, let us focus on the aerodynamics of the train
traveling inside tunnel. In nature, the aerodynamics of
train/tunnel systems is governed by a threedimensional,
unsteady, turbulent, compressible ow. Very frequently,
this method requires an extremely timeconsuming work
of high cost and much effort to solve the governing
equations as well. It is more helpful to deal with the
aerodynamics of train/tunnel systems with several
reasonable assumptions.
We here describe a simple analytical approach to
understand such complicated aerodynamic problems
occurring inside tunnel. Assume that the crosssectional
areas of train and tunnel are constant, and their equivalent
diameters are much larger than tunnel length, and train
speed V is very low compared with the speed of sound
corresponding to the atmospheric conditions, and the
propagation speed of pressure wave is the same as the
speed of sound. Under these assumptions, it is reasonable
to prescribe that the continuity equation involves variable
density of air but the compressibility effect is not
considered in the momentum and energy equations.
Assuming that u is the air velocity, and p the pressure,
the continuity equation is written as [36]
a
2
qu
qx
1
r
qp
qt
g 1j 23
and the momentum equation is given by
qu
qt
1
r
qp
gx
f ; 24
where a is the speed of sound, the density r is assumed
to be constant, x the distance along tunnel, t the time, g
the ratio of specic heats (g 1:4) and f and j are the
frictional force and energy dissipation, respectively, as
given in Eqs. (25) and (26). For the sake of simplicity,
here we divide the ows into three regions: ahead of
train, behind train, and in train. For the regions ahead
of and behind train, the frictional forces are generated
on tunnel walls:
f
l
2d
ujuj; 25
j
l
2d
juj
3
: 26
For the region in train, the frictional forces stem from
both train body and tunnel wall surfaces. Thus, the
frictional force and energy dissipation can be expressed as
f
l
2d
1
1 R
u
0
ju
0
j
l
0
2d
0
1
1 R
u
0
Vju
0
Vj; 27
j
l
2d
1
1 R
ju
0
j
3
l
0
2d
0
R
1 R
ju
0
Vj
3
; 28
where d and d
0
are, respectively, the hydraulic diameters
of tunnel and train, and l and l
0
the friction coefcients of
tunnel wall and train body surfaces, respectively, R the
ratio of the crosssectional areas of train to tunnel, and u
0
the ow velocity occurring between train and tunnel.
The compatibility conditions should be used to
connect the owelds in the three regions. Using the
coordinate system moving with train, the conservation
laws of mass and energy are given by
1 Ru
0
V u V; 29
p
0
1
2
ru
0
V
2
p
1
2
ru V
2
p
0
; 30
where p
0
is the stagnation pressure on the forebody of
train. In general, a wake ow is formed behind the trail
of train; but far away from it, the ow can be regarded
to be uniform across tunnel crosssectional area. For the
region closed by the trail of train and the uniform ow
area, the conservation laws of mass and momentum are
expressed as
u V 1 Ru
0
V; 31
p ru V
2
p
0
1 Rru
0
V
2
C
dp
R
1
2
ru
0
V
2
; 32
where C
dp
is the coefcient of the pressure drag on train
in the open air. At the entrance and exit of tunnel, it is
assumed that the ow discharges at atmospheric
pressure and when the ow comes into tunnel, it is also
assumed that the pressure reduces as much as the
dynamic pressure. At the entrance and exit of tunnel, the
boundary conditions can be given by
tunnel entrance p
1
2
ru
2
; uX0;
0; uo0;
_
_
_
33
tunnel exit p
0; uX0;
1
2
ru
2
; uo0:
_
_
_
34
The aerodynamic drag on the train traveling in the
tunnel is given by
D
A
0
1
2
r
R
0
_ _
u
0
V
2
1
2
rC
dp
u
0
V
2
u
2
_ _
1
2
r
l
0
d
0
_
in
V u
0
jV uj dx
_
V
2
_
out
dx
_
Dp
0
; 35
where D is the aerodynamic drag and A
0
the cross
sectional area of train. R and 0 in the rst term on the
right side are, respectively, taken for the forebody
of train to be inside and outside tunnel. In similarly
u
0
v
2
and v
2
in the second term are selected for the
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 492
afterbody of train to be inside and outside tunnel,
respectively. The integral of the third term is performed
along train length, and the subscripts in and out indicate
the inside and outside tunnel, respectively. The last term
p
0
on the right side is the pressure difference between the
fore and afterbodies of train. In the case where the
train is fully outside tunnel, Eq. (35) reduces to Eq. (7)
mentioned previously. Before train enters tunnel, the
ow velocity and pressure both remain zero. Using these
initial conditions, the equations above can be solved
after some calculations.
8.2. Pressure wave due to the train entering into tunnel
The train entering into tunnel at high speed plays a
role of the piston motion against the air inside tunnel
and thus compresses the air in front of the train. The
resulting compression wave propagates along tunnel at a
nearly sonic speed. The compression wave will be
distorted or attenuated by viscous actions and heat
transfer inside tunnel [37]. To understand the compres
sion wave generated at the entrance of tunnel, here we
introduce a simple onedimensional approach again.
When a train of the crosssectional area of A
0
enters a
tunnel (crosssectional area: A) at speed V; a compres
sion wave is formed at the vicinity of the entrance of
tunnel, as schematically shown in Fig. 34. It is assumed
that the compression wave form and the resulting
pressure rise are expressed as Dpx and Dp
21
p
2
p
1
; respectively. Here p
1
and p
2
are the static pressures
just upstream and downstream of the compression wave,
respectively. Assuming that the value of Dp
21
is very
small, compared with the atmospheric pressure p
1
; and
that isentropic ow is formed between states 2 and 3
[38], Dp
21
can be given as a function of the train Mach
number M
t
and the crosssectional area ratio f:
Dp
21
1
2
gp
1
M
2
t
1 f
2
f
2
1 f
2
M
t
M
2
t
; 36
where g is the ratio of specic heats of air, and f is
A A
0
=A: If we want to use the blockage ratio R of
train to tunnel instead of f; R is replaced with (1 f).
Fig. 35 shows the relation of Dp
21
and train velocity
V; together with the measured data (R 0:216) at real
tunnels [38]. The calculated curve for R 0:216 is well
agreed with the measured data. The value of Dp
21
is
nearly proportional to V
2
and decreases with a decrease
in R:
The Mach number M
s
of shock wave can be obtained
if we assume the compression wave as a weak shock
wave. It is noted from Fig. 35 that M
s
is very close to
unity. For instance, the compression wave produced by
the train, which enters into tunnel at a speed of 400 km/
h, propagates along the tunnel at a nearly sonic speed.
It can be deduced from the above results that the
strength Dp
21
of the compression wave formed at the
entrance of tunnel is almost independent of the shape of
the forebody of train. However, it is expected that the
compression wave form Dpx should be dependent on
train velocity, shape of forebody, blockage ratio, etc.
[3941]. From some computational and experimental
works it is found that the compression wave form is a
function of train velocity, Dp
21
; and the equivalent
diameter of tunnel D; as indicated in an empirical [41]:
Dp
comp
Dp
21
1
2
1
p
tan
1
V
0
t
0:3 D
_ _ _ _
; 37
where t is the time from the center of the compression
wave form. Eq. (37) is often employed to simulate the
compression wave produced at the entrance of real
tunnels.
8.3. Pressure variation and aerodynamic drag inside
tunnel
According to the previous work, the ratio of the cross
sectional areas of train to tunnel (=blockage ratio R)
has been known to be one of the most important
Entrance
Train
Compression
wave
Distance
P
r
e
s
s
u
r
e
P2
P1
P(X)
Crosssection area of train (A)
Train speed
V
u3
3
2
u2
1
U
Compression wave
generated by train
Crosssectional area of tunnel (A)
'
Fig. 34. Onedimensional ow model of compression wave.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 493
parameters inuencing the aerodynamics of train/tunnel
systems [3943]. As R increases, the pressure variation,
ow velocity and aerodynamic drag increase. Addition
ally, the coefcients of friction forces on tunnel wall and
train body can signicantly inuence the aerodynamics
of train/tunnel systems.
Using the theoretical equations aforementioned,
Fig. 36 presents the pressure variation and aerodynamic
drag on a train traveling inside the tunnel with a
comparatively short length [36,39]. The thick solid line
indicates the measured pressure variation and the thin
solid lines the theoretical predictions, where tunnel
length L is 471 m, the crosssectional area A of tunnel
58.2m
2
, train length l 100 m, the crosssectional area of
train A
0
13.7 m
2
, the coefcient of friction l on tunnel
wall 0.02, the coefcient of on friction l
0
on train body
0.018, the speed of train U 249 km/h, and C
dp
0.12. It is
found that the pressure strongly uctuates with time,
and the pressure uctuations are qualitatively similar to
the fore and afterbodies of train. The theoretical
results predict the measured pressure uctuations
comparatively well.
For a longer tunnel of L 3264 m, Fig. 37 presents
both the measured and predicted pressure variation and
aerodynamic drag on a train of length 400 m [36,39]. It is
again found that the theoretical results predict the
measured pressure variation and aerodynamic drag
comparatively well, in considering the highly compli
cated oweld. Thus, it is believed that the theoretical
equations can be used to predict the aerodynamics of
train/tunnel systems.
For the trains passing each other inside tunnel,
the pressure variation is shown in Fig. 38, where
L 5305 m. The measured and predicted pressure
variations [36,39] are obtained under the situation that
29 s later after one train (l 400 m) enters into tunnel at
a speed of 207 km/h, the other train (l 300) enters at a
speed of 201 km/h opposite to the tunnel. Unlike a single
train traveling inside the tunnel described above, it is
found that a negative pressure of its peak value of
3000 Pa occurs when two trains intersect each other
inside tunnel, and a high pressure of about 3000 Pa just
before passing each other. This is due to the pressure
0
1
2
3
60
40
20
0
0 2 4 6 8
1
0
1
2
Train head
P
r
e
s
s
u
r
e
(
k
P
a
)
Time(sec)
Time(sec)
Time(sec)
P
r
e
s
s
u
r
e
(
k
P
a
)
D
r
a
g
(
k
N
)
Train tail
Aerodynamic drag
Computation
Measurement
Computation
U=249km/h, C DP=0.12
L=471m, A=58.2m, =0.020
2
l=100m, A'=13.7m , '=0.018
2
(a)
(b)
(c)
Fig. 36. Pressure variations and aerodynamic drag on train
(U 249 km/h, C
dp
0:12; L 471 m, A 58:2 m
2
, l
0:020; l 100 m, A
0
13:7 m
2
, l
0
0:018).
0 100 200 300 400
Train speed, V (km/h)
0
1
2
3
4
5
M
a
g
n
i
t
u
d
e
o
f
C
o
m
p
r
e
s
s
i
o
n
w
a
v
e
(
2
1
)
R=1.0
R=0.7
R=0.5
R=0.216
R=0.1
Measurement
(a) V vs P
21
0 100 200 300 400
Train speed, V (km/h)
1.00
1.02
1.04
1.06
1.08
1.10
S
h
o
c
k
M
a
c
h
n
u
m
b
e
r
(
M
s
)
R=1.0
R=0.7
R=0.5
R=0.216
R=0.1
(b) V vs Ms
p
)
Fig. 35. Effect of train speed on the magnitude of compression
wave and shock Mach number.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 494
waves made ahead of the two trains. From these
pressure variations with time, it can be inferred that
strong side forces act on the two trains when passing
each other, and their signs are changed just before and
after passing each other. Consequently, these forces can
cause strong yawing motions of the train.
8.4. Pressure variation inside train
In practice, complete sealing of a train body is not
possible because the train should have a lot of air
conditioning devices, ventilation systems, etc. Thus, the
large amplitude pressure variations on the train body
during traveling inside tunnel can penetrate into
passengers room, often leading to ear discomfort in
passengers [4446]. This phenomenon will be more
serious with the increase in train speed. Systematic
study is required to alleviate the pressure variations
inside train.
The passengers ear discomfort inside train is, in
general, associated with the magnitude of pressure
variation, the rate of pressure variation, the sign of
0 10 20 30 40 50 60 70
80
2
1
0
1
90
100
110
120
130
140
150
Head
Tail
Computation
Measurement
Measurement
Time(sec)
T
r
a
v
e
l
i
n
g
d
r
a
g
(
k
N
)
S
t
a
t
i
c
p
r
e
s
s
u
r
e
(
k
P
a
)
( L=3264m, A=60.4m , =0.02
l =400m, A'=13.7m , '=0.018
U=200km/h, C DP=0.15 )
2
2
Fig. 37. Pressure variations and aerodynamic drag on train (L 3264 m, A 60:4 m
2
, l 0:02; l 400 m, A
0
13:7 m
2
, l
0
0:018;
U 200 km/h, C
dp
0:15).
4
2
0
2
4
20
40
60
80
100 120
P
r
e
s
s
u
r
e
(
k
P
a
)
Entering into tunnel
Measurement
Computation
Leaving from tunnel
Time(t)
Tunnel length (L)=5305m
Train 1( l )=400m, train speed (U)=207km/h
Train 2( l )=300m, train speed (U)=201km/h
(Train 2 enters into tunnel 29 second later after train 1 enters into tunnel)
Fig. 38. Pressure variations on trains passing each other in tunnel.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 495
pressure variation, etc. [4749]. In real highspeed
railway trains, a special ventilation system is usually
adopted to alleviate the pressure variations. This system
controls the ow rate supplied into and exhausted from
train, using a fan blower, according to the pressure
variations occurring inside train. These pressure varia
tions are proportional to the square of train speed
[48,49]. The ventilation system may not be enough to
reduce the pressure variations as the train speed
increases.
In order to alleviate the pressure variation inside
train, several control methods have been investigated
using a damper system, a continuous ventilation system,
and a continuous ventilation control system [48,49].
All of these methods are to control the magnitude
and the rate of the pressure variation occurring
inside train. Fig. 39 schematically shows a ventilation
system controlling the ow rate [48,49]. It controls
the ow rate checking the pressure variations inside
train.
Figs. 40 and 41 present an example of experi
mental and computational works to alleviate the
pressure variations inside train [48,49]. It is found
that the pressure variation inside train can reduce using
the ventilation system above, and computational
results represent the experimental ones with a good
accuracy.
9. Impulse wave at the exit of tunnel
The compression wave generated by the train entering
into the tunnel propagates along tunnel. When it meets
the open end of the tunnel, it is reected by the
boundary conditions of the open end and propagates
50 0 150 100
4
2
0
2
4
Time (sec)
P
r
e
s
s
u
r
e
(
k
P
a
)
External pressure (experiment)
Internal pressure (experiment)
Internal pressure (computation)
Fig. 40. Pressure variations inside train.
Tunnel information
C
o
n
t
r
o
l
u
n
i
t
Pressure transducer
outside car
Pressure tramsducer
inside car
Electronic valve
Car body
Electronic valve
Flow meter (troattling device)
Supply fan
Troattling device
Exhaust fan
Fig. 39. Ventilation system for alleviating pressure variations inside train.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 496
back along the tunnel, with an inverse phase to the
incident compression wave. The resulting expansion
wave reects again from the entrance of the tunnel.
Thus, the complex wave reections are repeated inside
tunnel, leading to strong pressure transients. If the
incident wave were an expansion wave, then it would
reect back from the exit of the tunnel, as a compression
wave. Such a situation is again encountered as the
afterbody of the train enters into the entrance of
tunnel.
These pressure waves inside tunnel essentially experi
ences attenuation and distortion during the propagation
processes until it fades out due to viscous friction and
heat transfer effects, as schematically illustrated in
Fig. 42. A compression wave may transition to a weak
shock wave due to the nonlinear wave characteristics.
This transition process is not yet well understood. Much
work has been devoted to the attenuation and distortion
mechanisms of the compression wave occurring during
the propagation process.
Some researchers [5052] have calculated the propa
gation processes of the compression waves with a proper
treatment of viscous friction and heat transfer effect and
their prediction results are compared with the measured
results in real highspeed railway tunnels. They have
discussed the distortion problem of the compression
wave occurring ahead of a highspeed railway train and
showed that theoretical and computational predictions
represent well the experimental results associated with
the wave attenuation and distortions. But the transition
characteristics from a compression wave to a weak
shock wave during the propagation processes are,
however, not well known.
Some part of the compression wave, which reaches
the open end of tunnel, is discharged from the open
end, as shown in Fig. 42. The resulting wave usually
reduces to a pulse wave, leading to an impulsive noise,
which is recently being a new type of environmental
noise problem. The impulsive waves emitted from
the open end of the tunnel have some characteristics
like a direct current component of the incident pressure
wave, while the pressure transients inside tunnel are
alike an alternative current component of the incident
wave.
(a) Train No.1
(b) Train No.16
0 20 40 60 80 100
4
2
0
2
4
2
0
2
0 20 40 60 80 100
U=350km/h
Time (sec)
P
r
e
s
s
u
r
e
(
k
P
a
)
Time (sec)
P
r
e
s
s
u
r
e
(
k
P
a
)
CFD
Exp.(pressure outside train)
Exp.(pressure inside train)
CFD(pressure inside train)
CFD
Controlled pressure
inside train
Pressure inside train
(no control)
Pressure outside
train
Fig. 41. Pressure variations inside and outside train (computa
tion).
Impulse noise emitted
from tunnel exit
P
r
e
s
s
u
r
e
Distance
Attenuation,
or Distortion
S
o
u
n
d
p
r
e
s
s
u
r
e
Time
Entrance
Tunnel
Train
Compression wave
generated by a train
entering a tunnel
Compression wave
near exit of tunnel
Distance
P
r
e
s
s
u
r
e
Exit
a b
c
Fig. 42. Wave diagrams in highspeed railway train/tunnel systems.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 497
9.1. Stateoftheart of impulse wave
Here we discuss on the impulse wave discharged from
the exit of tunnel. Until now, the railway tunnel impulse
wave has been largely investigated in Japan, since they
have a number of tunnels in Shinkansen lines. Many
aspects with regard to the impulse wave have been
known [53,54], they are as follows.
The wave form of a railway tunnel impulse wave is
like a pulse shape with a high peak of short duration.
The magnitude of the impulse wave is closely associated
with detailed shape of the compression wave which
reaches the exit of tunnel, and is nearly proportional to
the time derivative of the compression wave form. For a
short tunnel below 1 km, the magnitude of the impulse
wave is a function of the cubic of train speed, regardless
of whether the railway line is made of slab or ballast
track. The peak pressure of the impulse wave attenuates
with distance, being inversely proportional to distance,
except for the near eld very close to the exit of tunnel.
For a comparatively long tunnel, the impulse wave is
signicantly inuenced by whether the railway line is
made of slab or ballast track. For ballast tracks, the
magnitude of the impulse wave decreases with an
increase in tunnel length. For long slab tracks, it is no
longer proportional to the cubic of the train speed. The
nonlinear wave effects cause the compression wave front
to be steeper during its propagation inside tunnel, and
consequently leading to the increased magnitude of the
impulse wave. In the slab tracks shorter than 10 km, the
impulse wave becomes stronger with tunnel length, but
over 10 km, the magnitude of the impulse wave
decreases with tunnel length.
Fig. 43 presents the pressure histories measured at
several locations inside and outside tunnel, where train
speed is tested by 167 and 197 km/h [53,54]. In
Fig. 43(a), where the pressure history is measured at a
location of 90 m upstream of the exit of tunnel, the
pressure suddenly increases as the compression wave
reaches the measurement point, and then, after 0.5 s it
decreases due to the expansion wave reected from the
exit of tunnel. Fig. 43(b) shows the impulse wave
measured at a location of 20 m outside away from
the exit of tunnel. It is found that the peak pressure of
the impulse wave increases more than 4 times as train
speed increases from 167 to 197 km/h.
In order to investigate the frequency contributions to
the impulse wave shown in Fig. 43, a typical result of the
frequency analysis is presented in Fig. 44 [53,54]. For the
train speed of 197 km/h, the impulse wave is mainly
composed of the frequency components below 100 Hz,
but below 167 km/h the frequency components over
20 Hz seem to be close to the background noise level.
For a slab track tunnel of 13 km, Fig. 45 presents the
pressure histories measured at several locations indi
cated [53,54]. It is obvious that the compression wave
front generated at the entrance of tunnel considerably
changes during propagation inside tunnel. Such a
distortion in the wave front is directly associated with
the impulse wave occurring at the exit of tunnel.
9.2. Theory of impulse wave
The railway tunnel impulse wave can be well under
stood from the aeroacoustical theory [20], as next.
Assume that the air is discharged from an open end of a
tube with a constant crosssectional area and that the
90m 22m
a
b
0.1s
0
50
100
150
U=197km/h
U=167km/h
0
5
10
15
Compression wave at point a
Impulse wave at point b
0.1s
U=197km/h
U=167km/h
Time (sec)
P
r
e
s
s
u
r
e
(
k
g
/
m
2
)
P
r
e
s
s
u
r
e
(
k
g
/
m
2
)
Time (sec)
(a)
(b)
Fig. 43. Compression wave and impulse wave in slab track
tunnel (L 8:9 m).
f=1Hz
1
0
d
B
0
50 100
U=167km/h
U=197km/h
Frequency (Hz)
S
p
e
c
t
r
a
l
a
m
p
l
i
t
u
d
e
Fig. 44. Frequency analysis of impulse wave.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 498
volume of the space outside the tube is V
0
; and the mass
per unit time [q kg/(m
3
s)] is added to a unit volume of
the air in this space, the resulting pressure uctuation at
a point of far eld from tube exit can be obtained from
the wave equation of Lighthill:
1
a
2
q
2
Dp
qt
2
r
2
Dp
qq
qt
: 38
As an approximate solution of the above equation, the
pressure uctuation Dpr; t at a point of distance r away
from the tube exit is expressed as
Dpr; t
1
4pr
qm
qt
_ _
; 39
where m is the mass (kg/s) per unit time discharged from
the tube. The following equation relates m to q of
Eq. (38):
m
_ _ _
V
0
q t
r
a
_ _
dV 40
It is also assumed that a compression wave propagates
from the left of the tube towards the open end, the
pressure rise Dp
comp
due to the compression wave
induces the ow velocity u
comp
inside tube:
Dp
comp
rau
comp
: 41
The discharge velocity u
e
of air from the open end of the
tube is approximately obtained from such a condition
that an expansion wave reects from the exit of the tube:
u
e
E2u
comp
: 42
Meanwhile, the mass ow rate discharged from the open
end of the tube is given as Eq. (43), if we assume that the
crosssectional area of the tube is A
t
:
m ru
e
A
t
: 43
By using Eqs. (41)(43), Eq. (39) reduces to
Dpr; t
A
t
2pra
qDp
qt
_ _
comp
: 44
The above equation refers to such a case that the sound
produced from a source radiates into a threedimen
sional innite space. Therefore, Eq. (44) is corrected as
Eq. (45), if the sound radiates into a semiinnite space
like the exit of tunnel:
Dpr; t
A
t
pra
qDp
qt
_ _
comp
: 45
From Eq. (45), we see that the sound pressure Dp at the
far eld is inversely proportional to the distance r: For a
given distance of r away from the sound source, the
sound pressure Dp is also proportional to the time
variation qDp=qt
comp
of the pressure rise due to the
compression wave.
The impulse wave form Dp
pulse
is obtained from the
time derivative of a compression wave form Dp
comp
; as
schematically illustrated in Fig. 46. It is known that the
peak pressure of the impulse wave is proportional to the
maximum pressure gradient in the compression wave
front, as indicated below:
Dp
pulse;max
A
t
pra
qDp
qt
_ _
comp;max
: 46
Fig. 47 shows the result of a total variation diminishing
(TVD) numerical calculation for the axisymmetric,
unsteady, compressible ow [55,56]. In the numerical
calculation, it is assumed that an innite bafe plate is
installed at the exit of a tube with its diameter of D; and
a compression wave propagates towards the tube exit
and discharges into still air. It can be known that
impulse wave is produced outside the tube and then
attenuated with the increase in its propagation distance.
According to Eq. (46), the strength of the impulse
wave, i.e., the peak pressure in the impulse wave form, is
proportional to a maximum value of time derivative of
compression wave front that reaches the exit of tunnel.
P
r
e
s
s
u
r
e
(
k
g
/
m
2
)
A B
C
D
38.5m
7064m
32.5m
13030m
58m 20m
U=203km/h
Compression wave at point A
0
100
50
150
P
r
e
s
s
u
r
e
(
k
g
/
m
2
)
P
r
e
s
s
u
r
e
(
k
g
/
m
2
)
P
r
e
s
s
u
r
e
(
k
g
/
m
2
)
Time (sec)
U=206km/h
Compression wave at point B
0
100
50
150
0.1s
Time (sec)
0
100
50
U=206km/h
Compression wave at point C
Time (sec)
0.1s
0
2
4
6
8
10
12
U=206km/h
Time (sec)
Impulse wave at point D
(a)
(b)
(c)
(d)
Fig. 45. Compression wave and impulse wave (L 13:03 m).
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 499
Besides this, it is known from experience that the
strength of impulse wave can be dependent on the
following factors:
(1) the velocity of train entering into tunnel;
(2) the ratio of crosssectional area of train to tunnel;
(3) the characteristics of railway track (ballast or slab
track);
(4) the length of tunnel;
(5) the shape of walls inside tunnel;
(6) the topography at the vicinity of the exit and
entrance of tunnel, etc.
The sound pressure of a spherical sound wave
radiating from a point source attenuates, being inversely
proportional to the distance from the sound source.
Assuming that the SPL at a point of the distance r
0
away
from a sound source is expressed as L
p0
(dB), the SPL L
p
(dB) at a point of the distance r; being further away from
that point, is given by
L
p
L
p0
20 log
10
r=r
0
: 47
It can be noted from the above equation that the second
term of the right is 6:02 dB for r=r
0
2:0: In other
words, a spherical sound wave is attenuated by 6 dB for
the double distance.
9.3. Slab and ballast track tunnels
In order to fully understand the generation mechan
ism and detailed characteristics of the impulse wave
occurring at the exit of tunnel, it is necessary for us to
investigate the relationship between the compression
wave fronts at both the entrance and exit of tunnel. To
do this, the detailed distortion and attenuation of the
compression wave propagating along tunnel should be
understood. Here we discuss about the impulse wave
occurring at the exit of the slab and ballast track tunnels.
A compression wave given at a location of x away
from the entrance of tunnel can be modeled by Eq. (48)
[57], when it propagates along a tunnel with the length
of L;
du
dt
1
2
l
2d
H
u
2
n
_
p
_
d
H
1
g 1
Pr
_
_ _ _
_
t
N
qu
qt
z
dz
t z
_
_
; 48
Particle
path
Expansion
wave
t
Impulsive
wave
c
x
pmax (x)
p(x)
t
0
(e)
(d)
pe, max (x)
pe
t
0
D
L
K
A
B
0
b
p
Compression
wave
a
pi (t=0)
p
*
(b)
(a)
0
(c)
Infinite
baffle plate
Compression
wave
Exit
x
Tube
Fig. 46. Discharge of compression wave from tunnel exit.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 500
where the differential operator d=dt is dened as
d
dt
q
qt
u c
q
qx
; 49
where t is the time, u the ow velocity induced by the
compression wave, c the speed of sound, l the coefcient
of hydraulic friction, and d
H
the hydraulic diameter of
tunnel, n the kinematic viscosity, and Pr the Prantl
number. In this case, the pressure P of the impulse wave
at far eld from the exit of tunnel is given by a function
of the pressure p of the compression wave form at the
exit of tunnel, as described in Eq. (45).
In the case where the compression wave front is very
steep like a shock wave, the relationship between the
pressure P of impulse wave and the pressure p of
compression wave can be simulated as a vibration plate
with a radius of a [58]. The resulting sound pressure
P at a location of r away from the exit of tunnel can be
given by
P
1
2p
_
_
N
N
4Ao
R2Ka 1 iX2Ka
expiot
sin
Ka
2
r
a
_ _
2
1
_
r
a
_ _ _ _
exp i
p
2
Ka
2
r
a
_ _
2
1
_
r
a
_ _ _ _ _ _
do; 50
where k is the wave number, w the angular frequency, I
the complex number, and Ao is the Fourier transform
of the pressure of the compression wave at the exit of
tunnel, as given by
Ao
1
2p
_
_
N
N
pt expiot dt: 51
In Eq. (50), Rx and Xx are, respectively, expressed as
Rx 1 2J
1
x=x; 52
Xx 2S
1
x=x; 53
where J1x and S1x are Bessel function and Struve
function, respectively.
Eqs. (50) and (51) can be obtained by using a Fast
Fourier transform. Figs. 48 and 49 present calculation
results of the compression and impulse wave forms and
its frequency components [57], using the impulse wave
theories above, where the train speed U is 200 km/h, and
tunnel lengths are 1 and 5 km. It is found that after the
impulse wave, the uctuations of about 13 Hz are
qualitatively similar to the characteristic features of the
impulse wave occurring at the exit of a long slab track
tunnel, as shown in Fig. 48. For longer tunnels, the
impulse wave seems to be stronger and its frequency
band is broader. This implies that tunnel length should
be an important factor to inuence the impulse wave.
In ballast track tunnels, the pressure gradient of
compression wave front can be reduced due to the
ballast effects, which produce ow turbulence and
suction actions on the compression wave [59]. In high
speed railway tracks, the relative roughness of ballasts is
very small, being estimated by about 1/250. In consider
ing the fact that the ballast part on the track occupies
only
1
4
times the equivalent diameter of tunnel, it is
Fig. 47. Impulse wave discharged from tube exit.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 501
difcult to conclude that the increased resistance due to
the ballasts drastically changes the major characteristics
of impulse wave. A more systematic work is needed to
correctly assess the ballast effects on the impulse wave.
9.4. Short and long tunnels
In very short tunnels, the compression wave front may
not appreciably change during its propagation inside
tunnel. In such a case, it may be reasonable to believe
that the time needed for the pressure to rise from
atmospheric state to a level of compression wave is
determined by the equivalent diameter of tunnel and
train speed [57]:
p
2
p
0
1=2ru
2
2R R
2
1 R
2
: 54
Thus, the time derivative qp
i
=qt of the pressure p on
the righthand side of Eq. (54) may be determined by
U
3
U
2
=U
1
; where U is train speed.
Meanwhile, the magnitude of the impulse wave at the
exit of tunnel can be inuenced by the geography around
the tunnel exit. The peak pressure P
max
t of impulse
wave can be obtained by an empirical equation [57]:
p
max
k
U
3
r
; 55
where r is the distance away from the exit of tunnel.
As tunnel length increases, the impulse wave in slab
tracks can be signicantly different from that in ballast
track. For train speeds of 180 and 200 km/h, Fig. 50
presents the tunnel length effects on the peak pressure of
the impulse wave, which is measured at a location 20 m
away from the exit of tunnel [57]. It is found that for a
comparatively short tunnels, less than 1.5 km, the tunnel
length does not inuence the peak pressure of the
impulse wave; but for the tunnels over 1.5 km, the tunnel
0 50 100
50
40
30
20
10
0
0
50
150
0.5 0 0.5
0
2
4
6
8
10
12
14
0.5 0 0.5
100
V=200km/h
L=5000m
V=200km/h
L=5000m
Compression wave
Impulse wave
Frequency anaysis
R=20m
V=200km/h
L=5000m
P
r
e
s
s
u
r
e
(
k
g
/
m
2
)
P
r
e
s
s
u
r
e
(
k
g
/
m
2
)
Time (sec)
Time (sec)
Frequency (Hz)
S
P
L
(
d
B
)
(a)
(b)
(c)
Fig. 49. Computed compression and impulse waves in ballast
track tunnel.
0.1 1 10 100
Distance, X(m)
0.1
1
10
100
P
m
a
x
(
k
g
/
m
2
)
V=200km/h
V=180km/h
Fig. 50. Relationship between tunnel length and impulse wave.
0.0 0.5 0.5
50
100
150
0
V=200km/h
L=1000m
Compression wave
P
r
e
s
s
u
r
e
(
k
g
/
m
2
)
P
r
e
s
s
u
r
e
(
k
g
/
m
2
)
Time (sec)
0.0
0
2
4
6
8
10
12
14
0.5 0.5
Impulse wave
R=20m
V=200km/h
L=1000m
Time (sec)
50
40
30
20
10
0
0 50 100
Frequency anaysis
V=200km/h
L=1000m
S
P
L
(
d
B
)
Frequency (Hz)
(a)
(b)
(c)
Fig. 48. Computed compression and impulse waves in slab
track tunnel.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 502
length seems to inuence the peak pressure. This is
because the compression wave attenuates due to the
viscous and heat transfer effects on tunnel walls or its
front changes due to nonlinear effects. It is believed that
the attenuation of the compression wave would be
greater in ballast tracks than that in slab tracks, and that
the distortion of the compression wave due to the
nonlinear effect is strongly dependent on the strength of
the entry compression wave and consequently it would
be more remarkable as the train speed increases.
9.5. Control methodologies of impulse wave
Most of the research for reducing the tunnel impulse
noise have been concentrated on alleviation of the
pressure gradient of the compression wave propagating
inside tunnel. These efforts are largely applied to the
train body or tunnel. If the blockage ratio of train to
tunnel R decreases by using available means, the impulse
noise can be reduced. However, such a method will
result in a reduction in vehicle efciency or an increased
cost of construction of a larger tunnel. Optimization of
train shape can also reduce the tunnel impulse noise.
Similar to an airplane body, this method has started
receiving much attention in recent years.
Control techniques can also be applied to the tunnel
at the entrance of the tunnel, the inside of the tunnel, or
the exit of the tunnel. A hood at the entrance of the
tunnel can reduce the pressure gradient of the initial
compression wave. Inside the tunnel, branch tunnels,
water curtains, spray mists, porous/cavity walls and
ballast on the track can be utilized to alleviate the
compression wave propagating along the tunnel. More
over, an active control method by using the impulse
wave with an inverse phase against the tunnel impulse
noise or a passive silencer can be applied to the exit of
the tunnel. These control methodologies have not yet
been investigated sufciently, and more systematic
research is needed for application to real tunnels.
9.5.1. Train body
The control techniques applicable to the train body
can be considered as both the reduction of the trains
crosssectional area and the optimal design of the train
forebody. Fig. 51 presents the effects of the blockage
ratio of train to tunnel R on the compression wave
which is generated at the entrance of tunnel, where train
Mach number M
t
is 0.286 [60]. The detailed shape of the
train forebody is given by a revolutional ellipsoidal
body with the major and minor axes of a and b;
respectively, and the entry compression wave is obtained
at x=D 6:8: It is found that the overpressure of the
compression wave and its pressure gradient increases as
R increases. The maximum pressure gradient of the
entry compression wave is an increasing function of the
blockage ratio R; and is nearly independent of the train
Mach number (except at very high speeds), when
it is normalized by the train speed, as presented in
Fig. 52 [60].
Fig. 53 presents the effects of the shape of train fore
body on the entry compression wave, where the train
forebody shape is given by both the major and minor
axes of the revolutional body, and train Mach number
M
t
is 0.184, R is 0.116 and a=b is 7.0 [60]. For the three
shapes of train forebody used, it seems that the over
pressure of the entry compression wave is nearly
independent of the forebody shape, but the entry
1 0 1 2 3 4 5
0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
tV/d
M=0.286
a/b=3.0
R=0.30
1 0 1 2 3 4 5
tV/d
0.2
0.0
0.2
0.4
0.6
0.8
1.0
M=0.286
a/b=3.0
R=0.22
R=0.12
R=0.30
R=0.22
R=0.12
(a)
(b)
Revolutional
ellipsoidal body
(
p
/
t
)
/
(
V
/
2
d
)
3
8
(
p

p
)
/
(
V
/
2
d
)
2
8
8
Revolutional
ellipsoidal body
V
/
2
d
)
3
8
Revolutional
ellipsoidal body
p
/
t
)
/
(
V
3
/
2
d
)
Fig. 53. Effect of train head on compression wave.
0.6
0.7
0.8
0.9
1.0
1.1
2 3 4 5 6 7
a
b
Length of train head/radius (a/b)
R=0.116
M=0.184
:
:
Revolusional paraboloidal
body
Revolusional
ellipsoidal
body
Circular cone
R
e
d
u
c
t
i
o
n
r
a
t
e
o
f
p
r
e
s
s
u
r
e
g
r
a
d
i
e
n
t
Computational results
Experimental results
Fig. 54. Effect of train head on pressure gradient of compres
sion wave.
Head
C
r
o
s
s

s
e
c
t
i
o
n
a
l
a
r
e
a
o
f
t
r
a
i
n
Head of train
Effective
Noneffective
Close to double carsp type
Close to aerodynamic wedge type
Fig. 55. Optimum shape of train head to reduce the impulse
wave.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 504
90 m away from the entrance of tunnel, 90 m back from
the exit of tunnel, and 20 m away outside from the exit
of tunnel, respectively. In the case of the entry hood
system, its crosssectional area is about 1.57 times the
tunnel crosssectional area. The hood length is about
15 m and several windows are installed on the hood
walls. These windows play a role of the bleed ventilation
hole to exhale the compressed air. For the entry hood
system, the entry compression wave and the impulse
wave are considerably reduced, compared with no entry
hood system.
Fig. 56. Entry hoods installed at the tunnel entrance.
C
A B
20m 90m 90m
U=197km/h
0
50
100
U=197km/h
0
50
100
150
0.1s
0.1s
5
0
0.1s
10
15
Compression wave at point A
Compression wave at point B
Impulse wave at point C
P
r
e
s
s
u
r
e
(
k
g
/
m
2
)
P
r
e
s
s
u
r
e
(
k
g
/
m
2
)
P
r
e
s
s
u
r
e
(
k
g
/
m
2
)
Time(sec)
Time(sec)
Time (sec)
(a)
(b)
(c)
Fig. 57. Compression and impulse waves in slab track tunnel
(no entry hood, L 5389 m).
0
50
100
U=200km/h
0
50
100
150
0.1s
0.1s
5
0
0
0.1s
A B
C
20m 80m 100m
U=200km/h
Compression wave at point A
Compression wave at point B
Impulse wave at point C
P
r
e
s
s
u
r
e
(
k
g
/
m
2
)
P
r
e
s
s
u
r
e
(
k
g
/
m
2
)
P
r
e
s
s
u
r
e
(
k
g
/
m
2
)
Time(sec)
Time(sec)
Time(sec) (a)
(b)
(c)
Fig. 58. Compression and impulse waves in slab track tunnel
(entry hood, L 5389 m).
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 505
For the entry hood, Fig. 59 presents the frequency
components of the impulse wave [63,64]. It is known
that the entry hood system reduces highfrequency
components of the impulse wave. Fig. 60 presents the
entry hood effects on the impulse wave, where the entry
hood lengths used are 30 and 40 m [63,64], and the
tunnel length is 6.64 km. The impulse wave becomes
stronger as train speed increases, and can be given by the
law of U
3
: For the entry hood with the length of 40 m,
the impulse wave strength is only about 10% of that of
no entry hood and for the 30 m entry hood, it reduces to
about 50% of that of no entry hood.
It is likely that the length of the entry hood can
signicantly inuence the impulse wave strength. Fig. 61
shows the effects of the entry hood length on the impulse
wave, where the entry hood effects are given by terms of
the reduction ratio of the pressure gradient of the entry
compression wave and the reduction ratio of the train
speed corresponding to the reduced pressure gradient
[63,64]. For instance, the reduction ratio is about 0.92,
which is equivalent to a reduction in the train speed of
8%, as the reduction ratio of the pressure gradient of the
entry compression wave is 0.8.
It is known that the pressure gradient of the entry
compression wave reduces as the hood length increases,
but it no longer reduces over a certain hood length, and
that the entry hood leads to a reduction effect of the
train speed. For a given train speed and tunnel length,
the entry hood length can be optimized to reduce the
impulse wave strength.
9.5.3. Inside tunnel
In general, there are some branch tunnels inside real
tunnels. These are used as a passage to maintain tunnel
and track system, or to store some tools necessary for
the maintenance. These branch tunnels can be used to
reduce the impulse wave at the exit of tunnel because
those can bleed out the compressed air inside tunnel.
Additionally, soundabsorption materials are applied to
the tunnel walls, playing the same role of the ballasts on
track. Water curtain and mist spray which are applied
inside tunnel are used to reduce the impulse wave
[59,6568]. Here we discuss several methods to reduce
the impulse wave.
When the compression wave propagates through tunnel
with a branch tunnel, some part of the compression wave
190 200 210 220 230 240 250 260 270
5
10
20
50
100
200
500
Train speed (km/h)
M
a
x
i
m
u
m
p
r
e
s
s
u
r
e
o
f
i
m
p
u
l
s
e
w
a
v
e
,
P
m
a
x
(
P
a
)
Entry hood (length : 40m)
Entry hood (length : 30m)
No entry hood
Fig. 60. Effect of entry hood length on the peak pressure of
impulse wave.
U=197km/h
0 50 100
No entry hood
U=200km/h
0 50 100
1
0
d
B
Entry hood
1
0
d
B
Frequency (Hz)
S
p
e
c
t
r
a
l
a
m
p
l
i
t
u
d
e
Slab track tunnel (length : 5389m)
Frequency (Hz)
S
p
e
c
t
r
a
l
a
m
p
l
i
t
u
d
e
(a)
(b)
Fig. 59. Effect of entry hood on the frequency of impulse wave
(slab track tunnel, L 5389 m).
0.0
0.2
0.4
0.6
0
10 20 30
40
50
0.8
0.1
0.6
0.5
0.9
0.8
0.7
Entry hood length (m)
R
e
d
u
c
t
i
o
n
r
a
t
e
o
f
p
r
e
s
s
u
r
e
g
r
a
d
i
e
n
t
(
)
R
e
d
u
c
t
i
o
n
r
a
t
e
o
f
t
r
a
i
n
s
p
e
e
d
(
1
/
3
)
Fig. 61. Entry hood length effect.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 506
goes into the branch tunnel. In this case, a reected wave
propagates back to the main tunnel, consequently
inuencing the compression wave propagating into the
main tunnel. For the branch tunnel shown in Fig. 62, the
pressures p
1
t and p
2
t before and after the compres
sion wave propagating through the branch tunnel are
expressed by using a low frequency approximation
[65,67]:
p
2
t
2
2 n
p
1
t
4n
2 n
2
p
1
t 2T; 56
where t is the time, n is the ratio of the crosssectional
area of the branch and main tunnels, and T ( l=c) is a
characteristic time associated with the branch tunnel
length l and the speed of sound (c). From Eq. (56), it is
known that for long branch tunnels p
2
t is 2=n 2
times p
1
t: Thus, the pressure gradient of the compres
sion wave reduces to 2=n 2:
For a sufciently long branch tunnel, several branch
tunnels can considerably reduce the impulse wave. For
instance, in the case of eight branch tunnels of n 0:2;
the pressure gradient of the compression wave propa
gating through tunnel reduces to about 47% of that of
the entry compression wave. However, there are many
unknown problems for the branch tunnel effects. For
instance, how long is the branch tunnel length to obtain
practical benets? Of what effect is the branch tunnel
inclination? etc. The resonance effects which can be
generated in the branch tunnels should be investigated
prior to the practical application.
Fig. 63 presents measured compression and impulse
waves for no branch tunnel, where train speed is 200 km/
h [65,67]. Fig. 64 presents the branch tunnel effects using
a computational analysis [65,67]. It is known that the
branch tunnel considerably reduces the impulse wave.
The computational results are compared with real tunnel
tests in Fig. 65, where train speed is 210 km/h [65,67].
The computed result seems to overpredict the strength
of the impulse wave and fails in predicting the impulse
wave frequency. More elaborate computational work is
needed to solve this problem. With the branch tunnel,
the peak pressure of the impulse wave reduces from
about 7 to 5 Pa. These branch tunnel effects should be
further investigated for higher speeds of train.
Fig. 66 presents a practical application of the water
spray to reduce the impulse wave [66]. An experiment is
carried out using the compression wave propagating
through a branch tunnel. The water spray region is
0 0.5 0.5
0
0.5
1.0
140
120
100
80
60
40
20
0
1
0
d
B
0 50 100
R=20m
U=200km/h Compression wave
Impulse wave
Frequency anaysis of impulse wave
0 0.5 0.5
U=200km/h
f=1Hz
Time(sec)
P
r
e
s
s
u
r
e
(
P
a
)
P
r
e
s
s
u
r
e
(
P
a
)
Time(sec)
Frequency (Hz)
S
p
e
c
t
r
a
l
a
m
p
l
i
t
u
d
e
(a)
(b)
(c)
Fig. 63. Measured compression and impulse waves (no branch
tunnel).
Tunnel
Vertical branch Inclined branch
p1(t) : Pressure before branch
p2(t) : Pressure after branch
Fig. 62. Branch installed inside highspeed railway/tunnel.
U=200km/h
p
(
P
a
)
0.1sec
0
5
10
Time (sec)
Fig. 64. Computed impulse wave in branch tunnel.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 507
about 30 m over the branch tunnel. Fig. 67 presents the
change in the compression wave form before and after
the water spray region [66], where train speed is 210 km/
h, the water spray is operated under the supply pressure
of 2.0 atm (abs). It is known that the pressure gradient of
the compression wave considerably reduces after the
water spray region. This method may highly cost to
obtain practical reduction effect, and further study is
needed.
9.5.4. Tunnel exit
The control methods to reduce the impulse wave are
also applied to the exit of tunnel. These are classied
into passive and active controls. The passive controls
yield a silencer system at the exit of tunnel, while the
active controls use a negative impulse wave with an
inverse phase to the impulse wave.
Fig. 68 schematically shows the silencer system
installed into the exit of a duct [69,70]. A simple shock
tube with an open end can be used as the tunnel impulse
wave simulator. A weak shock wave propagates into the
lowpressure chamber of the shock tube, and it
discharges from the open end of the shock tube,
producing the impulse wave at the exit of the shock
tube. A silencer system is installed at the exit of the
shock tube to control the impulse wave, as shown in
Fig. 68(b), where the silencer has a height of H and a
length of L: A series of bafe plates having a height of h
are inserted into the silencer. In experiment, the
operating pressure ratio of the shock tube is changed
below 1.40 to obtain the weak shock wave, which is
similar to the compression wave at the exit of tunnel.
Fig. 69 presents the major characteristics of the
impulse wave using the shock tube experiment, where
the sound pressure level DSPL of the impulse wave is
taken relative to that at r
ref
4D:
DSPL SPL
rref
SPL
rr
; 57
where D is the diameter of tube and r is the radial
distance from the exit of tube. According to the linear
aeroacoustics, the strength of the impulse wave reduces
to a half for a double distance, 6.0 dB/a double
distance. It is found that the absolute value of DSPL
increases as r=D increases and the shock Mach number
decreases. For a location far away from the exit of tube,
the impulse wave with a weak incident shock wave is
alike the acoustic wave, but for a stronger incident shock
wave it differs from the linear acoustic waves. This is
associated with the nonlinear effects of the impulse wave
with a strong incident shock wave.
0.0 0.1 0.2 0.3 0.4 0.5
0.0
0.5
1.0
1.5
B
C
Time (sec)
P
r
e
s
s
u
r
e
(
k
P
a
)
Fig. 67. Effect of water spray on compression wave.
B C
115m 287m
30m
115m 85m 30m
S S
A
D
632m
T
u
n
n
e
l
Water spay region
Fig. 66. Experiment of water spray.
U=210km/h
0.2s
5
0
5
10
Time (sec)
P
r
e
s
s
u
r
e
(
P
a
)
Fig. 65. Measured impulse wave in branch tunnel.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 508
In order to investigate the detailed conguration of
the silencer system, the height h of the bafe plates and
the length L of the silencer are changed, as shown in
Fig. 70 [69,70]. For no bafe plates inside the silencer,
the reduction of the impulse wave increases with L=D: It
seems that the bafe plates lead to more reduction of the
impulse wave. In this case, reduction of the impulse
wave is not strongly dependent on L=D: The maximum
reduction of the impulse wave is obtained for h=H
0:75:
Fig. 71 presents the effect of the height H of the
silencer on the impulse wave [69,70]. For no bafe
plates, the reduction of the impulse wave slightly
increases with H; but for bafe plates the reduction of
the impulse wave is again obtained at h=H 0:75: In
this case, an H over D does not produce the practical
reduction of the impulse noise. Figs. 72 and 73 present
the effect of the silencer conguration on the impulse
H
L
D
h Box
Baffle plate
Diaphragm
Needle
Driver section Driven section
1320 2130
x
y
z
6060
zm
Needle
Microphone
r
Driven section Driver section
253 30
Measurement field
1 2
Diaphragm
(a) Openended shock tube
(b) Exit box and baffle plate
Exit box
Crosssection
Fig. 68. Schematics of experimental facility and measurement system.
1.00 1.01 1.02 1.03 1.04 1.05 1.06 1.07
Me
10
6
4
2
0
r/D=5
r/D=10
r/D=15
S
P
L
8
12
Fig. 69. Propagation characteristics of impulse wave.
'
S
P
L
(
d
B
)
h/H
0
2
4
6
8
0.0 0.5 1.0
r/D=15
zm/D=0.5
=45
Me=1.03
L/D=0.5
=1.0
=1.5
=2.0
Flow
D
H
L
h
Fig. 70. Effects of silencer and bafe plate on impulse noise
reduction.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 509
noise, where the volume of the silencer remains constant
at D
3
: It is found that the reduction of the impulse noise
is somewhat dependent on the silencer conguration.
Figs. 74 and 75 show the computational results to
investigate the impulse noise, where the incident
compression wave is given by [70,71]
Dp
p
a
Dp
p
a
1
2
1
p
tan
1
px
L
16pD
3L
_ _ _ _
: 58
In Eq. (58), Dp is the overpressure of the incident
compression wave, p
a
the atmospheric pressure, x the
distance from the tunnel exit, and L is the wave length of
the incident compression wave. Eq. (58) has been known
as an empirical relation of the compression wave
occurring inside the real highspeed railway tunnels. In
Fig. 75, the impulse waves at x=D 2:0 are presented
for three different incident compression waves with and
without the silencer [70,71]. For L=D 0:1 the peak
pressure of the impulse wave slightly reduces, but the
second peak is generated. For L=D 1:0 and 2.0, it
seems that the silencer does not reduce the impulse wave.
It is concluded that such a passive control method
using the silencer and bafe plates above is dependent
on the wave length of the incident compression wave. In
reality, the wave length of the compression wave which
is generated inside highspeed railway tunnels is
estimated to be several to several 10 times the equivalent
diameter of the tunnel. Thus, the size of the silencer to
produce the practical reduction of the impulse noise
should be of a comparable order as the wave length of
the compression wave.
Meanwhile, an active noise control (ANC) concept
can also be applied to the impulse wave. This concept
yields a negative impulse wave having the same
characteristics as the tunnel impulse wave [72]. On the
verge of the discharge of the compression wave from the
exit of tunnel, the negative impulse wave is discharged
from an additive sound source, as schematically shown
in Fig. 76. A pressure transducer detects the com
pression wave P
1
; and its signal having an impulse
response X
ss
drives the additive sound source S through
a signal processor. Then the wave form P
c
at the
location 2 is made of both the impulse wave P
p
discharged from the exit of tunnel and the negative
impulse wave P
a
: The resulting wave form can be
expressed as [72]
p
p
p
1
r
1e
r
e2
; 59
p
a
p
1
r
1s
r
s2
X
ss
; 60
where r
1e
is the impulse response from point 1 to the exit
of tunnel, r
e2
the impulse response from the exit of
tunnel to the reference point, r
1s
the impulse response
against the output signal from the pressure transducer 1,
and r
s2
is the impulse response from the additive sound
source to the reference point. Neglecting the acoustic
combination between the output signal from the
additive sound source and the electric signal from the
pressure transducer 1, the reduction of the tunnel
impulse wave is obtained if Eq. (61) is satised as
X
ss
r
1e
r
e2
=r
1s
r
s2
: 61
In this case, the strength of the tunnel impulse wave is
ideally zero. For the ANC applications like this, it is,
thus, of practical importance to know the impulse
response against the tunnel impulse wave, and in
especially, to properly predict the tunnel impulse wave
from the compression wave inside the tunnel. The
directivity of the negative impulse wave should be
fully understood with the practical installation of the
additive sound source at the exit of tunnel. This ANC
'
S
P
L
(
d
B
)
h/H
0
2
4
6
8
0.0 0.5 1.0
r/D=15
zm/D=0.5
=45
Me=1.03
H/D=0.5
Flow
D
H
L=1.5D
h
=1.0
=1.5
Fig. 71. Effect of silencer height on SPL.
'SPL(dB) Crosssection Exit box
L=D
H=D
D
D/2
D/2
D/2
D/2
D/4
(a)
(b)
(c)
(d)
(e)
(f)
(g)
0 dB
2.5 dB
3.1 dB
2.4 dB
2.4 dB
2.1 dB
3.1 dB
r/D=15, zm/D=0.5, =45
Me=1.03,
Fig. 72. Reduction of impulse noise using a passive control.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 510
control methods above have not yet been investigated
sufciently, and a more systematic research is needed for
application to real tunnels.
10. Concluding remarks
The aerodynamic and aeroacoustic problems accom
panied by the speedup of train system are, at present,
receiving a considerable attention as a practical en
gineering issue that should be urgently resolved. With
the speedup of trains, many engineering problems
which have been reasonably neglected at low speeds,
are being raised with regard to aerodynamic noise and
vibrations, impulse forces occurring as two trains
intersect each other, impulse wave at the exit of tunnel,
ear discomfort of passengers inside train, etc. These are
the major limiting factors to the speedup of train
system. Such factors all are closely associated with the
ows occurring around the railway train. However,
much effort to speed up the train system has to date
been paid on the improvement of electric motor power
rather than understanding the ow physics around the
train and thereby nding a proper control method. This
has led to larger energy losses and performance
deterioration of the train, since the ows around train
are more disturbed due to turbulence of the increased
speed, consequently the ow energies being converted to
aerodynamic drag, noise and vibrations.
The highspeed railway train system connecting
between cities is required to have aerodynamically
and aeroacoustically good performance to meet safe
and comfortable transportation with less air pollution
and noise, and reliable transportation with low cost and
maintenance. Although, recently, Schetz [73] has deeply
reviewed the aerodynamics of highspeed railway train,
information available on the train aerodynamics is
still lacking, and a great deal of work should be
made to solve the aerodynamic and aeroacoustic
problems of highspeed railway trains. The major
objectives of this review article are to enhance the
current understanding of the aerodynamic phenomena
with regard to the highspeed railway trains, and to
provide help in designing the railway train systems of
'SPL(dB)
4.9 dB
3.9 dB
3.8 dB
3.7 dB
3.6 dB
3.4 dB
3.3 dB
r/D=15, zm/D=0.5, =45 Me=1.03,
(a) (h)
(i) (b)
(c)
(j)
(k)
(d)
(e)
(f)
(g)
(l)
(m)
(n)
L=1.5D
H=D
D h=0.75D
L=1.5D
H=D
D h=0.75D
Flow Flow
Exit box & Baffle plate Exit box & Baffle plate 'SPL(dB)
3.2 dB
3.1 dB
4.3 dB
3.8 dB
4.1 dB
3.1 dB
2.3 dB
Fig. 73. Impulse noise reductions at far eld.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 511
safe and comfortable transportation with less air
pollution and noise.
Acknowledgements
The second author of this paper would like to thank
the postgraduate students at Andong National Uni
versity Gasdynamics Lab for drawing gures and typing
manuscript.
References
[1] Bouladon G. Future 1970;3(15).
[2] Akaki S. Some aspects of transport vehicles speeds
their energy consumption. Bull JSME 1992;95878 (in
Japanese).
[3] Gabrielli G, von Karman T. Mech Eng ASME
1950;72(10):775.
[4] Yomiuri Newspaper. Issued on 11 May 1991 (in Japanese).
[5] Ozawa S. Aerodynamic forces on train. JSME 1990;90037
(in Japanese).
[6] Kim HD. Aerodynamic analysis of a train running in a
tunnel(1). Korean Soc Mech Eng (KSME), 1997;
21(8):96372.
[7] Hara T, Nishimura B. Aerodynamic drag on train. RTRI
Rep 1967;591 (in Japanese).
[8] Hara T. Method of measuring of aerodynamic drag of
trains. Railway Technical Research Institute, Japan
1965;6(2).
[9] Guihew C. Resistance to forward movement of TGVPSE
trainsets: evaluation of studies and results of measure
ments. Fr Railw Rev 1983;1(1).
[10] Hara T. Measurement method of aerodynamic drag on the
train running in tunnel. RTRI Rep 1967;608 (in Japanese).
Without exit box
With exit box
1
0
0
0
P
a
100ms
L/D=0.1
L/D=1.0
L/D=2.0
1
0
0
0
P
a
100ms
1
0
0
0
P
a
100ms
Without exit box
With exit box
Without exit box
With exit box (silencer)
(a)
(b)
(c)
Fig. 75. Computation of passive control for impulse wave.
Fig. 74. Compression wave forms for computation.
P1
Xss S
Pc
Pa
Pp
Exit
Tunnel
1
e
2
r1s
rs2
re2
Impulse wave
Negative
impulse wave
r1e
Fig. 76. Positive control for impulse wave.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 512
[11] Maeda T, Kinoshita M, Kajiyama H, Tanemoto K.
Aerodynamic drag of Shinkansen electric cars (series 0,
series 200, series 100). RTRI JNR 1987;1371 (in Japanese).
[12] Maeda T. Aerodynamic drag of Shinkansen electric cars.
RTRI JNR 1987;1(3) (in Japanese).
[13] Maeda T, Kinoshita M, Kajiyama H, Tanemoto K.
Estimation of aerodynamic drag of Shinkansen trains
from pressure rise in tunnel. In: Proceedings of the Sixth
International Symposium on the Aerodynamics and
Ventilation of Vehicle Tunnels, Durham, UK, 1988. p.
6178.
[14] Maeda T, Kinoshita M, Kajiyama H, Yanemoto K.
Aerodynamic drag of Shinkansen electric car; series 0,
series, 200, series 100. RTRI Rep Inst 1989;30(1):4856 (in
Japanese).
[15] Ido A, Iida M, Maeda T. Wind tunnel tests for nose and
tail of train. RTRI JNR 1993;7(7) (in Japanese).
[16] Baker CJ. The wind tunnel determination of cross wind
forces, moments on a high speed train. In: Schulte
Werning B, Gregoire R, Malfatti A, Matschke G. editors.
In: Proceedings of the Brite/Euram Project Symposium on
Transient Aerodynamics and Railway Systems Optimiza
tion, Paris, 1999, Wiesbaden, 2000.
[17] Baker CJ, Humphreys ND. Assessment of the adequacy of
various wind tunnel techniques to obtain aerodynamic
data for ground vehicles in cross winds. J Wind Eng
Aerodyn 1996;60:4968.
[18] Suzuki M, Tanemoto K, Maeda T. Wind tunnel tests on
aerodynamic characteristics of vehicles under crosswinds.
In: Proceedings of the World Congress on Railway
Research, Tokyo, 1999.
[19] Maeda T. Numerical study of ow around car on bridge at
crosswind conditions. RTRI JNR 1990;4(2) (in Japanese).
[20] Lighthill MJ. On sound generated aerodynamically.
1.General theory. Proc R Soc A 1952;211:564.
[21] King III WF. A pr! ecis of developments in the aeroacous
tics of fast trains. J Sound Vib 1996;193(1):34958.
[22] Hardy AEJ. Railway passengers and noise. Proc World
Congr Inst Mech Eng 1999;213F:17380.
[23] Moritoh Y, Zenda Y, Shimizu Y, Nagakura K. Aero
dynamic noise of high speed railway cars. In: International
Conference on Speedup Technology for Railway and
Maglev Vehicles, Yokohama, Japan, 1993.
[24] Morikawa T. Experimental investigations on reduction of
aerodynamic noise generated by pantographs using anec
hoic wind tunnel. RTRI JNR 1988;2(9) (in Japanese).
[25] Morikawa T. Fundamental experiments far a low aero
dynamic noise pantograph. RTRI JNR 1990;4(11) (in
Japanese).
[26] Koyanagi S. A method of lateral vibration analysis
of the tailvehicle of a train. RTRI JNR 1988;2(12): (in
Japanese).
[27] Suzuki M. Computational study on ow induced vibration
of highspeed train in tunnel. In: Proceedings of the
International Conference Flow Induced Vibration, 7th,
Lucerne, 2000.
[28] Fujimoto H, Miyamoto M. The lateral vibration analysis
of the tail car in a train. RTRI Rep 1988;2(5) (in Japanese).
[29] Vardy AE. Aerodynamic drag on trains in tunnels. 1.
Synthesis and denitions. Proc World Congr Inst Mech
Eng 1996;210:2938.
[30] Vardy AE. Aerodynamic drag on trains in tunnels. 2.
Prediction and validation. Proc World Congr Inst Mech
Eng 1996;210:3949.
[31] de Wolf WB, Emmenie EAFA. A new test facility for the
study of interacting pressure waves and their reduction in
tunnels for highspeed trains. In: Proceedings of the
International Symposium on the Aerodynamics and
Ventilation of Tunnels, 9th, Aosta Valley, Italy, London,
1997.
[32] Gr! egoire R, R! ety JM, Masbernat F, Morin" ere V,
Bellenoue M, Kageyama T. Experimental study (scale 1/
70th) and numerical simulations of the generation of
pressure waves and mircopressure waves due to high
speed traintunnel entry. In: Proceedings of the Interna
tional Symposium on the Aerodynamics and Ventilation of
Vehicle Tunnels, 9th, Aosta Valley, Italy, 1997.
[33] Howe MS. Review of the theory of the compression wave
generated when a highspeed train enters a tunnel. Proc
World Congr Inst Mech Eng 1999;213F:89104.
[34] Mancini G, Violi AG. Pressure wave effects of high speed
trains running parallel on large and medium sized tunnels
of Italian high speed lines. Proc World Congr Railw Res,
Tokyo: Railr Tech Res Inst, 1999.
[35] SchulteWerning B, Matschke G, Gregoire R, Johnson T.
RAPIDE: a project of joint aerodynamics research of the
European highspeed rail operators. Proc World Congr
Railw Res, Tokyo: Railr Tech Res Inst, 1999.
[36] Yamamoto A. Aerodynamics of train and tunnel. RTRI
JNR 1983:1230 (in Japanese).
[37] Sockel H. The aerodynamics of trains. In: Schetz JA, Fuhs
AE, editors. Handbook of uid dynamics and uid
machinery. New York: Wiley, 1996. p. 172141.
[38] Matsuo K, Aoki T, Mashimo S, Nakatsu E. Entry
compression wave generated by a highspeed train entering
a tunnel. In: Proceedings of the International Symposium
on the Aerodynamics and Ventilation of Vehicle Tunnels,
9th, Aosta Valley, Italy. 1997.
[39] Ozawa S. Numerical simulation of aerodynamic problems
in train/tunnel systems. RTRI Rep 1990; 48 (in Japanese).
[40] Ogawa T, Fujii K. Numerical simulation of compressible
ows induced by a train moving into a tunnel. J Comp
Fuilds Dyn 1994;3(1).
[41] Ogawa T, Fujii K. Aerodynamics on train/tunnel systems.
JSME 1994;970940 (in Japanese).
[42] Yamamoto A. Pressure variations, aerodynamic drag, and
tunnel ventilation in shinkansen type tunnel. RTRI Rep
1973;871 (in Japanese).
[43] Iida M, Maeda T. Numerical calculation of pressure
transients in railway tunnels. RTRI Rep 1990;4(7) (in
Japanese).
[44] Schultz M, Sockel H. Pressure transients in railway
tunnels. In: Schneider W, Troger H, Ziegler F, editors.
Trends in applications of mathematics to mechanics.
BHRA Fluid Engineering, Harlow, UK, 1989. p. 339.
[45] Komatsu N, Yamada F. The reduction of the train draft
pressure in passing by each other. Proc World Congr
Railw Res, Tokyo: Railr Tech Res Inst, 1999.
[46] Gawthorpe RG. Pressure comfort criteria for rail tunnels
operations. In: Haerter A, editor. Aerodynamics and
ventilation of vehicle tunnels. New York: Elsevier, 1991.
p. 17388.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 513
[47] Sato K, Ikada M, Nakagawa M. Effects of pressure
changes on pain sensation of human ears. RTRI JNR
1989;3(3) (in Japanese).
[48] Zenda Y. Study on the ventilating system of Shinkansen
vehicle by simulating the internal pressure. RTRI JNR
1988;2(12) (in Japanese).
[49] Kobayashi M, Suzuki Y, Akutsu K. Alleviating ear pains
by controlling air pressure in ventilating system of
Shinkansen car. RTRI JNR 1990;4(7) (in Japanese).
[50] Schultz M, Sockel H. The inuence of unsteady friction on
the pressure waves in tunnels. In: Sixth International
Symposium on the Aerodynamics and Ventilation of
Vehicle Tunnels, BHRA Fluid Engineering, 1988.
[51] Kim HD. Numerical study on attenuation and distortion
of compression waves propagating into a straight tube.
Korean Soc Mech Eng (KSME) 1996;20(7):231525.
[52] Kage K, Kawagoe S. Numerical study on nonlinear effects
of compression waves propagated in a tube, Trans. Jpn
Soc Mech Eng (JSME) 1992;58(516):33842 (in Japanese).
[53] Ozawa S. Studies of micropressure wave radiated from a
tunnel exit. RTRI JNR 1979;1121 (in Japanese).
[54] Ozawa S, Moritoh Y, Maeda T, Kinoshita M. Investiga
tion of pressure wave radiated from a tunnel exit.
Quarterly Report of Railway Technical Research Institute,
Japan. 1976;1023 (in Japanese).
[55] Matsuo K, Aoki T, Kashimura H. Computers in railway 3.
In: Murthy TKS, et al., Computational mechanics, Vol. 2,
1992. p. 455.
[56] Kim HD, Setoguchi T. Study of the discharge of weak
shocks from an open end of a duct. J Sound Vib
1999;226(5):101128.
[57] Ozawa S. Micropressure wave at the exit of tunnel. RTRI
Rep 1980;37(1) (in Japanese).
[58] Blake WK. Mechanics of owinduced sound and vibra
tion. New York: Academic Press, 1986 [Chapter 89].
[59] Ozawa S, Maeda T, Matsumura T, Uchida K. Effect of
ballast on pressure wave propagating through tunnel. In:
International Conference On Speedup Technology for
Railway and Maglev Vehicles, Yokohama, Japan, 1993.
[60] Maeda T, Matsumura T, Iida M, Nakatani K, Uchida K.
Effect of shape of train nose on compression wave
generated by train entering tunnel. In: International
Conference on Speedup Technology for Railway and
Maglev Vehicles, Yokohama, Japan, 1993.
[61] Dayman B, Vardy AE. Alleviation of tunnel entry pressure
transients (report 1); experimental program. In: Proceed
ings of the Third International Symposium on the
Aerodynamics and Ventilation of Vehicle Tunnels, 1979.
p. 34362.
[62] Kage K, Miyake H, Kawagoe S. Numerical study of
compression waves produced by highspeed trains entering
a tunnel (2nd report, effects of shape of hood). Bull JSME
1993;59(560).
[63] Ozawa S, Maeda T. Model experiment on reduction of
micropressure wave radiated from tunnel exit. Interna
tional Symposium on Scale Modeling, 1822 July, 1988,
Tokyo, pp. 3337.
[64] Ozawa S, Maeda T, Matsumura T, Uchida K, Kajiyama
H, Tanemoto K. Countermeasures to reduce micro
pressure waves radiating from exits of Shinkansen tunnels.
In: Proceedings of the Seventh International Symposium
on the Aerodynamics and Ventilation of Vehicle Tunnels,
Brighton, UK, 1991.
[65] Yamamoto A, Ozawa S, Maeda S. Counter measures of
micropressure waves using branch tunnels. RTRI Rep
1984;41(3) (in Japanese).
[66] Woodhead CA, Fox JA, Vardy AE. Analysis of water
curtains in transient gas ows in ducts. In: Proceedings of
the Second International Conference on Pressure Surges,
BHRA Fluid Engineering, Craneld, UK, 1976.
[67] Yamamoto A, Maeda S. Reduction of tunnel micro
pressure wave using branches. RTRI Rep 1981;38(8) (in
Japanese).
[68] Wakai K, Ochiai T, Sumida I. Control of shock wave
intensity traveling in a tunnel through density gradient
region (effect of distribution pattern). Trans Jpn
Soc Mech Eng (JSME) 1995;61(581):295302 (in
Japanese).
[69] Kim HD, Setoguchi T. Experimental study on reduction of
impulsive noise generating at the exit of highspeed railway
tunnel. Korean Soc Mech Eng (KSME) 1996;20(7):
237585.
[70] Kim HD, Setoguchi T. Passive control of unsteady
compression wave using vertical bleed ducts. Korean Soc
Mech Eng (KSME) 1997;21(9):1095104.
[71] Kim HD, Setoguchi T. Reduction of impulsive noise
caused by unsteady compression wave. JSME Int J Series
B, 1997;40(2):2239.
[72] Kanai H, Abe M, Kido K. A new method to arrange
additional sound source used in active noise control.
Acoustica 1990;70:25864.
[73] Schetz JA. Annu Rev Fluid Mech 2001;33:371414.
R.S. Raghunathan et al. / Progress in Aerospace Sciences 38 (2002) 469514 514