An Experimental Study on the Relationship between Ballast-flying Phenomenon and Strong Wind
under High-speed Train
H. B. Kwon, C. S. Park
Korea Railroad Research Institute, Gyeonggi, Republic of Korea
Abstract
The mechanism of ballast-flying phenomena by strong wind induced by high-speed trains has extensively
been investigated by conducting wind tunnel test and field-measuring of wind velocity in the vicinity of the
track. The ballast gathered from the Seoul-Busan high-speed railway track have been classified by mass
and shape to find relationship between those properties and the characteristic of movement in high wind
and 16-channel Kiel-probe array has been used to examine the detailed flow structure near the surface of
the track. The probability of ballast-flying during the passage of the high-speed train has been assessed
comparing the results from wind tunnel test and those from field-measuring. The results shows that when
the G7 train as well as the KTX train runs at 300km/h, about 25m/s wind gust is induced just above the tie
and the probability for small ballast under 50g to fly is about 50% when it is on the tie. If the G7 train runs
at 350km/h, the wind gust just above the tie increases to 30m/s, therefore more radical countermeasure
seems to be needed.
1. Introduction
Since the beginning of the high-speed train operations in Korea at April, 2001, several unfavorable effects
to environment including ballast-flying phenomenon started to appear on the operating ballast-track lines.
Obvious evidence of ballast-flying such as damaged wheel, broken glasses at station, and injured
acoustic screen has frequently been reported[1] and the cause and mechanism of this phenomenon has
been investigated from two points of view[4]. The first one is ballast-flying by strong wind in the vicinity
between train and track and the other one is ballast-flying by impact of accreted snow and ice from train
under-floor.
When a train runs at high speed, a strong turbulent airflow is induced beneath the train because of the
boundary layer on the train surface as well as the perturbation by irregular shapes of train such as
protruding bogies and cavity of inter-car. At train speed from 200km/h to 300km/h, strong airflow about
between 30m/s and 50m/s is induced in the vicinity between train and track and it enforces the ballast on
the track to start roll along the track to bound away[4].
The drop of accreted snow or ice from train bottom, the other reason of ballast-flying, is seasonal event in
snowy weather in winter. When a train runs on the snow-covered track at high speed, the snow is blow
away to accrete to train underbody. The accreted snow on the train repeats to melt and freeze to be as
hard as ice, and is detached from train underbody to fall and hit the ballast on the track due to the
temperature change or impulsive vibration of train body.
In this study, the mechanism of ballast-flying phenomena by strong wind induced by high-speed trains
has extensively been investigated by conducting wind tunnel test and field-measuring of wind velocity in
the vicinity of the track. The ballast gathered from the Seoul-Busan high-speed railway track have been
classified by mass and shape to find relationship between those properties and the aerodynamic
characteristics. In wind tunnel test, the movement of ballast has been observed increasing wind velocity,
and the critical wind velocities in which the ballast start to move or totally blown have been yielded for
each ballast properties. For field test, 16-channel Kiel-probe array has been used to examine the detailed
flow structure near the surface of track. The wind velocities at 16 points which are distributed both
vertically and laterally have been measured during train passage and the spatial as well as temporal
characteristic of the high-speed wind above train has been analyzed. The probability of ballast-flying
during the passage of the high-speed train has been assessed comparing the results from wind tunnel
test and that from field-measuring. The definition of ballast-flying probability factor has been introduced by
using maximum and minimum critical wind velocity of ballast-flying and for various ballast-flying
probability factors have been calculated for ballast mass grades and train speeds.
2. Wind tunnel Tests
The ballast picked out on the Seoul-Busan high-speed line has been statistically investigated by mass
and aerodynamic shape. Individual characteristics of ballast against wind as well as the behavior of
ballast laid between tie models have been tested in wind tunnel increasing wind velocity.
2.1 Statistical investigation on the shape and mass of ballast
About 1000 ballast has been picked out from the Seoul-Busan high-speed line and 330 samples among
them has been classified by aerodynamic shape and mass. The aerodynamic shapes have been
classified into three types as presented in Table 1.
Type
Definition
Aerodynamic Property Illustration
A
Flat
Very variable as the
direction of wind
B
Half-spherical
Variable as the
direction of wind
C
Spherical
Slightly variable as the
direction of wind
Table 1: Classification of ballast shape
The numbers of ballast at each shapes and mass ranges are presented in Fig 1. As for the shapes, about
C and B type occupy about 44% and 40%, respectively. The mass of ballast distributes between 17g and
200g and the mean mass of A type is smallest and that of C type is largest.
Fig 1: Mass and shape of ballast
2.2 Aerodynamic characteristics of individual ballast
The aerodynamic characteristics of individual ballast have been investigated in the middle-scale wind
tunnel of Seoul National University with 1.35m(W) × 0.95m(H) × 2m(L) test section and 50m/s maximum
operational wind velocity. To prevent danger caused by the ballast flying into the tunnel, screen of 40cm
height has been and installed at the downstream end of test section and the test wind velocity has been
restricted to 35m/s. A row of ballast has been set on the floor of wind tunnel perpendicular to the wind
direction and total 12 test conditions identified by the ballast mass and shape have been tested. The
pictures of ballast before and after test are presented in Fig 2.
(a) Before test
(b) After test
Fig 2: Ballast in wind tunnel
Fig 3 shows the critical wind velocity at which the ballast starts to move on various ballast mass and
shape conditions. On each conditions of mass and shape, the wind velocity at which any of the ballast
start to move is plotted by filled symbol and the wind velocity at which all of the ballast stat to move is
plotted by unfilled symbol. In all case, the ballast started to move at about 20m/s wind velocity. Due to the
limitation of wind tunnel test velocity, the totally blown out wind velocity were not actually measured,
however, we could observe that spherical ballast with small mass under 50g were blown out at about
25m/s. As the mass increases, the critical wind velocity at which the ballast is totally blown out increased
and guessed to be between 25m/s and 45m/s as indicated as dotted line in Fig 3.
50
Totally blown out
Critical Wind Velocity(m/s)
40
30
Start to move
20
Type A
Type B
Type C
10
0
0
50
100
150
200
Ballast Mass(g)
Fig 3 Critical wind velocity of ballast-moving
2.3 Aerodynamic characteristics of ballast between ties
To simulate the ballast lying on the track, models of tie and ballast between ties have been installed in the
wind tunnel as presented in Fig 4. The dimension of model is very similar situation as Seoul-Busan highspeed line and 1000 ballast is set between the ties.
Fig 4 Ballast Track Model in Wind Tunnel
After installing the model of ballast and ties, wind tunnel has been operated increasing the wind velocity
slowly. The ballast on the tie started to move at the wind speed of 20m/s in accordance with the first wind
tunnel test for individual ballast. When the wind velocity reached to 33m/s, the first flying of ballast
between tie has been observed. The Fig 5 presents the video frames of ballast flying instance with white
circle around flying ballast and the speed of flying ballast seems as high as wind velocity regarding the
dimension of tie and the time between of video frames.
(a) t = 0 sec
(b) t = 0.03 sec
(c) t = 0.06 sec
(d) t = 0.09 sec
Fig 5 Ballast-flying Phenomena (Wind Velocity=33m/s)
3. Field measurements
To investigate the actual wind characteristic under the high-speed train, Kiel-probe array system has
been developed and measured the flow field above the track during the passage of the KTX train for test
run at the speed of 300km/h.
3.1 Measurement system
Kiel-probe array which can measure absolute wind velocity for up to 15° oblique wind direction has been
implemented to build probe array to measure spatial as well as temporal wind velocity above the track
during the train passage. The Kiel-probe array system is composed of 16 channels with 10 horizontal
measuring points covering 72cm width and 6 vertical measuring points covering 20cm height.
To transform the measured pressure into electric signal, multi-channel pressure scanning system made
by Scanning System Inc. has been used. The sampling rate of pressure is 100Hz and the resulting data is
transferred to laptop computer via TCP/IP protocol.
The wind velocity at each measuring points are calculated by the static pressure and corresponding total
pressures using equation (1)
Ui =
2
( Pi − Ps )
ρ
(1)
Where, Ui is wind velocity at ith measurement point, ρ is air density, Pi is total pressure at ith measuring
point, and Ps is static pressure measured at common static pressure hole.
3.2 Test condition
Kiel-probe array system has been installed on the track of Seoul-Busan high-speed line at 109km from
Seoul to Busan where is open field without any disturbance of tunnel or bridge. Fig 6 shows the photo of
Measurement system installed.
Fig 6 Kiel-Probe Array Installed in Seoul-Busan High-speed Line(KP=109km)
Table 2 presents the description of train passed during the tests. The speed of train is from 279.4km/h to
295.2km/h and the corresponding passing time distributes from 4.73sec to 5.00sec considering the train
length 388.1m composed by 20 car.
ID
train
type
train
length
train
speed
time of
passin
(m)
1
2
3
4
5
(km/h)
g (sec)
289.07
4.83
291.08
4.80
KTX 388.104 295.18
4.73
279.43
5.00
295.18
4.73
Table 2 Descriptions of Train passed
3.3 Test results
Fig 7 shows the typical history of wind velocity above track during train passage calculated by equation
(1) in case 2 defined in table 2. The passage of train nose makes a small peak of wind velocity and the
wind sharply increase to make high wind fluctuation until the passage of train tail. The delay from the
nose passage to the start of high wind could be explained by the characteristic of viscosity and turbulence
of air. When a train runs at high speed, a laminar boundary layer is formed on the train underbody
becouase of the viscous characteristic of air. As it goes downstream, the boundary layer fully develops
and the laminar boundary layer transforms to turbulent boundary layer affecting far above the train
underbody, in other words, just above track surface. The large fluctuation after the development seems to
be originated from the disturbance of bogie system and inter-car because the period and number of
fluctuation approximately agrees with those of the passage by one vehicle.
tail passing
fully developed
Wind Velocity(m/s)
40
nose passing
-20
20
60
100
140
180
(mm high
from tie
upper surface)
20
0
15
20
time(sec)
Fig. 7 Wind velocity fluctuation during train passage (train speed = 291km/h)
The starting locations of fully developed flow from nose end are calculated for all cases and presented in
Table 3. The location ranges from 29.52m to 49.02m that corresponds between the second vehicle and
the third vehicle. The wide variation relative to that of train speed seems to be originated from the
characteristic of turbulence flow. Deducing from the above results, the turbulent boundary layer and the
disturbance by the bogie system and inter car are the main origins of the high wind above track during the
passage of the train.
ID
1
train speed Location of Fully
(km/h)
development (m)
289.07
37.23
2
291.08
40.83
3
295.18
29.52
4
279.43
45.11
5
295.18
49.02
Table 3 Start location of fully developed flow (from nose end)
To access the intensity of strong wind under train, average wind velocity of train gust, U is introduced as
defined in equation (2).
t tai l
U =
pas sin g
∫
t fully developed
U (t )
300
dt
ttai l pas sin g − t fully developed U train
(2)
Where, ttail passing and tfully developed is the time when the tail of train passes and the time when the flow is fully
developed, respectively and Utrain is the train speed in km/h. Therefore, the average wind velocity of train
gust is the time average of the velocity of fully developed flow over the track normalized at train speed of
300km/h.
The vertical and horizontal distributions of averaged wind velocities have been plotted in Fig 8 and Fig 9,
respectively. The vertical distribution shows that the average wind velocities at 200mm height were
maximum about 40m/s and as the measurement points get lower, the average wind velocities decrease.
The average wind velocity 20mm lower than the tie top(6.49m/s~18.56m/s) shows dramatic decrease
relative to that 20mm higher than the tie top(24.59m/s~30.60m/s). This result implies that the blockage
effect of the tie to the airflow induced by train is very large and higher tie or lowered ballast surface can
be good countermeasures to prevent ballast flying due to the strong during the passage of high-speed
train.
The horizontal distribution of average wind velocities shows large value at the center of track and as the
measurement point goes to side, the wind velocities decreases to more than the half of that at center of
track. From the above result, we can deduce that the airflow is strongest at the center of the track
therefore, primary consideration should be taken into the center of track to develop countermeasures
against ballast flying problem due to the strong wind by trains.
train underbody
500
Height from tie top(mm)
400
300
200
100
0
0
10
20
30
40
50
60
Average wind velocity(m/s)
70
80
90
300km/h
Fig. 8 Vertical Distribution of Average Wind Velocity (at center of track)
Rail center
Distance from track center(mm)
700
600
500
400
300
200
100
direction of train
0
0
5
10
15
20
25
30
35
Average wind velocity(m/s)
Fig. 9 Horizontal distribution of Average Wind Velocity (at 20mm high from the upper top of tie)
4. Assessment of ballast-flying probability
The probability of ballast to fly has been accessed based on the behavior of ballast in high wind examined
from wind tunnel test results and the average wind velocity measured from the field test for KTX train
running on Seoul-Busan high-speed line about at the speed of 300km/h. To access the probability of
ballast-flying quantitatively, Ballast-Flying Probability Factor(BFPF) is introduced defined as;
i) Vtrack < Vmin,
ii) Vtrack > Vmax,
BFPF = 0
BFPF = 1
iii) Vmin < Vtrack < Vmax,
BFPF =
m2
Vtrack − Vmin dm
max − Vmin m2 − m1
m
∫V
(3)
1
Where, Vmax and Vmin mean the maximum and minimum wind velocity at which the ballast moves and the
Vtrack is the average wind velocity just above the track. When Vtrack is lower than Vmin(i), no ballast could be
supposed to fly because the intensity of wind is not strong enough to blow ballast that BFPF could be set
as 0. When Vtrack is higher than Vmax(ii), all of the ballast would be flown out and the BFPF could be set as
1. When Vtrack is between Vmin and Vmax(iii), the BFPF could be calculated by equation (3) integrating the
overlapped portion of velocities over the interval between specific ballast masses.
Fig 10 presents the probability of ballast-flying relative to train speed and ballast mass. The critical wind
velocities of ballast-moving measured by wind tunnel tests in Fig 3 are implemented to determine the Vmin
and Vmax, which are indicated as solid line in Fig 10. The average wind velocities obtained from the filed
test on KTX train using equation (2) are employed to determine the Vtrack at the speed of 300km/h and
that at 350km/h has been estimated using linear extrapolation by train speed. Geometrically explained,
BFPF in equation (3) means the portion of lower area than Vtrack to the area enveloped by Vmin and Vmax,
over specific ballast masses. If we assume Vtrack as 25m/s which is the average wind velocity at 20mm
over track center at the train speed of 300km, Vtrack is larger than Vmin over all mass range which means
that there is probability of ballast-flying over all mass ranges. As mass of ballast gets smaller, in addition,
the more probability of ballast-flying is expected. If Vtrack is assumed 29.5m/s which corresponds to the
train speed 350km/h, overall probability of ballast-flying is expected to increase referring to the that at
300km/h train speed.
50
Vmax
Wind Velocity(m/s)
40
30
Vtrack (U train = 350km/h)
V track (U train= 300km/h)
Vmin
20
10
0
0
50
100
150
200
Ballast Mass(g)
Fig 10 Probability of ballast-flying relative to ballast mass
Table 4 presents the BFPF calculated based on the velocity profiles in Fig 10. At the train speed of
300km/h, ballast of 0~50g mass shows most large BFPF of 39.8% and in other mass ranges, the BFPF
distributes between 11.5% and 24.2%. At the train speed 350km/h, the BFPF increases in all mass
ranges. Total BFPF at the train speed 350km/h is about double of that at 300km/h. From the results, we
can deduce that the speed-up of high-speed train can cause serious ballast-flying problem and
appropriate countermeasure should accompany with the speed-up of high-speed train.
Ballast mass(g)
0~50
Train speed(km/h)
300
350
39.8%
64.3%
50~100 100~150 150~200
Total
19.7%
43.3%
24.2%
46.1%
11.5%
32.8%
14.7%
33.6%
Table 4 Ballast-flying Probability Factor(BFPF) Relative to Train Speed and Ballast Mass
5. Conclusion
To investigate the mechanism of ballast-flying phenomena by strong wind induced by high-speed trains,
wind tunnel test on the behavior of ballast in high wind and the field-measuring of wind velocity in the
vicinity of the track for KTX train running on Seoul-Busan high-speed line has been performed
From the results, we can conclude that;
- The ballast laid on the wind tunnel floor started to move at the wind velocity of 20m/s and is guessed to
totally blown out at the wind velocity between 25m/s and 45m/s. As for the ballast stacked between ties,
the first ballast-flying has been observed at the wind velocity of 33m/s
- The critical wind velocity at which the ballast start to move becomes lower as the mass of ballast
becomes smaller and more rounded shaped ballast has lower critical wind velocity than flatter ballast.
- The turbulent boundary layer and the disturbance by the bogie system and inter car are the main origins
of the high wind above track during the passage of the train and the effect of nose shape seems
negligible.
- The average wind velocities at 200mm height were maximum about 40m/s and as the measurement
points get lower, the average wind velocities decrease. However, the wind velocity just above the tie is
over 25m/s that it is larger than the critical wind velocity of ballast-flying.
- The blockage effect of the tie to the airflow induced by train is very large and higher tie or lowered
ballast surface can be good countermeasures to prevent ballast flying due to the strong during the
passage of high-speed train.
- The airflow is strongest at the center of the track therefore, primary consideration should be taken into
the center of track to develop countermeasures against ballast flying problem due to the strong wind by
trains
- The probability of ballast-flying at the train speed 350km/h is about double of that at 300km/h, then
appropriate countermeasure should accompany with the speed-up of high-speed train
Acknowledgements
The authors gratefully acknowledge that this study is a result of “R&D Project for High speed Railway
system” with the financial support from the Korean government.
References
[1] H. B. Kwon, C. S. Park, “Substructure flow analysis and experiments of high speed train for
researching the mechanism of ballast dispersion”, Proceeding of Korean Society for Railway 2003
fall, book 3, pp. 275-280, (2003).
[2] J-L Peters, , “Aerodynamics of very high speed trains and maglev vehicles: State of the art and future
potential”, Int. J. of Vehicle Design, Special Publication Sp3, (1983).
[3] E. Wilemsen, "High Reynolds number wind tunnel experiments on trains", Journal of Wind Eng. and
Ind. Aerodynamics, 69-71 pp437-447, (1997).
[4] M. Yhshida, M. Uchida, N. Yaguchi, N. Mifune, "Countermeasures for Ballast-flying Phenomena
Caused by High-speed Trains", RTRI Report, Vol.6, No.6, pp27-36, (1992). (In Japanese)