Environmental Science and Pollution Research (2022) 29:62330–62346

https://doi.org/10.1007/s11356-022-20074-8

RESEARCH ARTICLE

Acoustic performance of near‑rail low‑height noise barriers installed

on suburban railway bridges
Lizhong Song1,2 · Kui Gao1,2 · Quanmin Liu1,2 · Liangtao Zhang3 · Qingsong Feng1,2 · Wenjie Guo1,2

Received: 14 December 2021 / Accepted: 30 March 2022 / Published online: 9 April 2022
© The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2022

Abstract
With the large-scale construction of suburban railway viaducts, the noise problem along the viaducts is becom-
ing more and more prominent. Conventional vertical noise barriers have been widely used to alleviate the noise
problem along the suburban railway viaducts. However, conventional vertical noise barriers often have an adverse
effect on the urban landscape and also block the view of train drivers and passengers. A type of near-rail low-
height noise barrier was planned to install on the viaducts of Wenzhou Rail Transit Line S1 to reduce the impact
of noise on residents along the railway lines. To assess the acoustic performance of the near-rail low-height noise
barrier, a numerical procedure for railway viaduct comprehensive noise considering wheel-rail noise and structure-
borne noise of the bridge and noise barriers was proposed and then verified by a field test. On this basis, numeri-
cal models were established to compare the acoustic performances of the near-rail low-height noise barrier and
conventional vertical noise barrier. The influences of the height and top shape of the near-rail low-height noise
barrier on the acoustic performance were discussed. Based on the numerical analysis results, it is found that both
the near-rail low-height noise barrier and conventional vertical noise barrier have good acoustic performances.
The noise reduction effect of the near-rail low-height noise barrier is slightly better than that of the conventional
vertical noise barrier. The acoustic performance of the near-rail low-height noise barrier gradually improves, but
the improvement rate gradually slows down as the height of the noise barrier increases. The noise reduction effects
of both the inverted L-shaped and Y-shaped near-rail low-height noise barrier are obviously better than that of
the vertical one, while the noise reduction effects of the inverted L-shaped near-rail low-height noise barrier are
slightly better than the Y-shaped one.

Keywords Near-rail low-height noise barrier · Acoustic performance · Suburban railway · Wheel-rail noise · Structure-
borne noise · Numerical analysis

Introduction

The suburban railway is a passenger rail transit system

within a metropolitan area. The service range of suburban
Responsible Editor: Philippe Garrigues railways is generally within 50 to 100 km. The station dis-
tance is short, and the density is high. The design speed
* Quanmin Liu should be 100 to 160 km/h. In principle, the average station
[email protected]
spacing is no less than 3 km.
1
State Key Laboratory of Performance Monitoring Compared with underground lines, the construction of
and Protecting of Rail Transit Infrastructure, East China viaduct lines is easier and faster. The construction cost of
Jiaotong University, Nanchang 330013, China viaduct lines is only 1/4 to 1/3 of that of underground lines
2
MOE Engineering Research Centre of Railway with the same length. Therefore, suburban railways are
Environmental Vibration and Noise, East China Jiaotong mostly built in the form of viaduct lines. Taking Wenzhou
University, Nanchang 330013, China
Rail Transit Line S1 as an example, the total length of the
3
China Railway Siyuan Survey and Design Group Co., Ltd., line is 53.5 km, among which, the length of the viaduct line
Wuhan 430063, China

13
Environmental Science and Pollution Research (2022) 29:62330–62346 62331

is 39 km, accounting for 72.9%. With the large-scale con- assess the acoustic performances of the low-height noise
struction of suburban railway viaducts, the noise problem barriers.
along the viaducts is becoming more and more prominent. When a train passes through suburban railway viaducts,
Noise barriers have been widely used to alleviate the excited by rail roughness, wheels and rails vibrate and radi-
noise problem along the suburban railway viaducts (He ate “wheel-rail noise.” Rail vibration is transmitted through
et al. 2021). Conventional vertical noise barriers, as track structure to viaduct deck and induce viaduct to vibrate,
shown in Fig. 1a, are usually installed at the edge of the generating “bridge-borne noise” (Thompson 2009). Because
viaducts, with heights of 2.15 to 3.15 m. The acoustic wheel-rail noise accounts for a large proportion in the total
performances of conventional vertical noise barriers can noise, it has attracted people’s attention for a long time.
be improved by increasing height, introducing sound- Remington (1976a, b) comprehensively explained the gen-
absorbing materials, and optimizing cross-sectional pro- eration mechanisms of wheel-rail rolling noise from the per-
files (Laxmi et al. 2021). However, conventional vertical spective of wheel-rail interaction and developed an analyti-
noise barriers often have an adverse effect on the urban cal model for the prediction of wheel-rail rolling noise by
landscape and also block the view of train drivers and treating the wheel as a monopole source with uniform direc-
passengers. tivity and considering the rail as an Euler beam supported
In recent years, low-height noise barriers have been grad- continuously on a layer of elastic foundation. Thompson
ually used to attenuate tramway and highway noise (Koussa (1990) improved Remington’s wheel-rail noise prediction
et al. 2013; Jolibois et al. 2015). However, there has been model by treating the wheel as a finite element (FE) model
very limited research on the acoustic performances of low- and extending the rail model to the track model, i.e., con-
height noise barriers installed adjacent to railways. Com- sidering the vibration and noise radiation of the sleeper and
pared with conventional vertical noise barriers, noise barri- simulated as a Timoshenko beam supported on two layers of
ers can be installed closer to railways and thus can achieve a continuous foundations. Based on this model, a wheel-rail
similar noise reduction effect with smaller heights. noise prediction software named “TWINS” was developed
Wenzhou Rail Transit Line S1 is the first rail transit line (Thompson et al. 1996a, b). Zhang et al. (2016b, c, 2017,
in operation in Wenzhou, China. It is planned to install 2019) improved Thompson’s wheel-rail noise prediction
low-height noise barriers adjacent to rails on the viaducts model and software TWINS by considering the influences of
of Wenzhou Rail Transit Line S1, as shown in Fig. 1b, to noise radiation from sleeper, ballast, and the ground reflec-
reduce the impact of noise on residents along the railway tion, thus raising the prediction accuracy of wheel-rail noise.
lines. However, low-height noise barriers are rarely used in Bridge-borne noise is another important noise source of
China. Therefore, it is necessary to establish a prediction suburban railway viaducts. Bridge-borne noise is mainly
model of noise radiated from suburban railway viaducts to concentrated on low frequency. Although bridge-borne

Fig. 1  Schematic of noise

barriers installed on railway
viaducts. a Conventional verti-
cal noise barrier. b Low-height
noise barrier (unit: mm)

13
62332 Environmental Science and Pollution Research (2022) 29:62330–62346

noise is not as harsh as high-frequency wheel-rail noise, it passes by. Wheels and rails are induced to vibrate by the
can make people feel agitated and uncomfortable and may wheel-rail forces and then radiate noise. The rail vibrations
cause chronic diseases of the neuroendocrine system and are transmitted to the bridge through track structures and
cardio-cerebrovascular system (Gorai and Pal 2006). There- cause the bridge to vibrate and radiate noise.
fore, bridge-borne noise has been paid more and more atten- The numerical procedure for railway viaduct comprehen-
tion in recent years. Li et al. (2012) established a prediction sive noise is shown schematically in Fig. 2. First, wheel-rail
model of structure-borne noise from an urban rail transit forces are calculated by a wheel-rail interaction model. Next,
concrete U-shape bridge based on mode superposition a wheel-rail noise model and a bridge-borne noise model are
method and modal acoustic transfer vectors (MATVs) and established, respectively, by the SEA and BEM. Finally, the
verified by a field test. Li et al. (2014) improved the predic- comprehensive noise is obtained by superimposing wheel-
tion model of structure-borne noise from urban rail transit rail noise and bridge-borne noise.
concrete U-shape bridges by calculating MATVs based on
the 2.5-dimensional boundary element method (2.5D BEM), Wheel‑rail interaction model
thus raising the prediction efficiency of the prediction model.
Zhang et al. (2013) established a prediction model of struc- The wheel-rail interaction model, as shown in Fig. 2, was
ture-borne noise from a high speed railway (HSR) concrete established based on dynamic receptance method (Reming-
box girder bridge based on train–track–bridge dynamic ton 1987). The wheel-rail interaction force can be written as
interaction theory and BEM and verified by a field test. Song
et al. (2020) improved the prediction model of structure- R(𝜔)
F(𝜔) = − (1)
borne noise from concrete box girder bridges based on wave- AW + AR + AC
guide finite element method (WFEM) and 2D BEM, which
where R(ω) denotes rail roughnesses and AW, AR, and AC
significantly raised the prediction efficiency of the prediction
denote the receptances of wheel, rail, and contact spring.
model. Li et al. (2015a) established a prediction model of
Rail roughnesses refer to the ISO 3095 (2005) standard,
structure-borne noise from a railway concrete-steel compos-
as shown in Fig. 3, which can be expressed as (Li and Wu
ite bridge based on train–track–bridge dynamic interaction
2013):
theory and statistical energy analysis (SEA) and verified by
a field test. Liu et al. (2020) improved the prediction model
( ) {
R 18.45lg𝜆 + 27.20 𝜆 ≥ 0.01m
of structure-borne noise from a railway concrete-steel com- 20lg =
−9.70 𝜆 < 0.01m (2)
r0
posite bridge using a hybrid FE-SEA method, thus solving
the contradiction between prediction accuracy and predic- where r0 is the reference rail roughness whose value is
tion efficiency. ­10−6 m and λ is the wavelength of the rail roughness.
The mentioned noise prediction models were all consider- Other calculation parameters are listed in Table 1.
ing a single noise source. There were few noise prediction
models which considered both wheel-rail noise and struc- Wheel‑rail noise prediction model
ture-borne noise. In order to accurately assess the acous-
tic performances of the low-height noise barriers installed The FE models of wheel and rail, as shown in Fig. 2, were
adjacent to railways on viaducts, this paper establishes a established by FE analysis software ANSYS. The wheel
prediction model of noise from a suburban railway viaduct was modelled with 8-node solid elements. Firstly, the cross-
considering wheel-rail noise and structure-borne noise of the section of half-wheel and quarter-axle was meshed freely,
bridge and noise barriers. Then, the prediction model is veri- and then the wheel FE model was obtained by rotating the
fied by a field test. On this basis, the acoustic performances meshed cross-section 360°. The wheel FE model had a total
of the low-height noise barriers and conventional vertical of 11260 elements and 14006 nodes, and its maximum ele-
noise barriers are compared, and the influences of height ment size did not exceed 0.0675 m. The rail was also mod-
and top shape on the acoustic performance of the low-height elled with 8-node solid elements. Firstly, the cross-section
noise barriers are analyzed. of rail was divided into 132 elements, and then the rail FE
model was obtained by stretching the meshed cross-section.
The element size along the rail was 0.01 m. The calculated
Numerical procedure parameters of wheel and rail are listed in Table 1.
The wheel-rail force was applied to the FE models of
Wheel-rail noise and bridge-borne noise are two major noise wheel and rail as excitation, respectively, to solve the
sources of the comprehensive noise radiated from suburban vibration responses of wheel and rail. Then, the vibration
railway viaducts induced by passing trains. Due to track responses were taken as acoustic boundary condition, and
irregularities, wheel-rail forces are generated when a train the wheel-rail noise was obtained based on BEM. Finally,

13
Environmental Science and Pollution Research (2022) 29:62330–62346 62333

Wheel-rail interaction model

Vehicle Suspension system

Wheel-rail Track irregularity
Rail contact
Fastener Wheel spring

Track slab

Bridge

Wheel-rail forces

Wheel-rail vibro-acoustic numerical model Bridge vibro-acoustic numerical model

Wheel FE model Rail FE model Bridge FE model

Wheel vibration Rail vibration

Wheel noise Rail noise

Bridge vibration
Wheel-rail noise
Bridge-borne noise BE model
Wheel-rail noise SEA model

Comprehensive noise

Fig. 2  Schematic of numerical procedure for railway viaduct comprehensive noise

a SEA model of wheel-rail noise propagation considering subsystems, each acoustic cavity was coupled with the
the noise insulation, absorption, and reflection of ground, adjacent structure subsystem, and the acoustic cavities
bridge, and sound barrier, as shown in Fig. 2, was estab- were also coupled with each other. To consider the noise
lished using vibro-acoustic analysis software VA One. The insulation of the train, no acoustic cavity was established
SEA model included structure subsystem and acoustic in the train body position. Damping loss factor of concrete
cavity subsystem. To ensure the energy transfer between was set as 0.015 (Li et al. 2015a) and that of acoustic

13
62334 Environmental Science and Pollution Research (2022) 29:62330–62346

Fig. 3  Rail roughness spectrum 30

25
20

Roughness Level (dB)

15
10
5
0
-5
-10
-15 0.63
0.5
0.4
0.315
0.25
0.2
0.16
0.125
0.1
0.08
0.063
0.05
0.04
0.0315
0.025
0.02
0.016
0.0125
0.01
0.008
0.0063
0.005
0.004
0.00315
1/3 octave wavelength (m)

taken as a noise source, and the spatial distribution char-

Table 1  Calculation parameters of the train, rail, and bridge
acteristics of the wheel-rail noise could be predicted.
Component Parameter Unit Value

Train Mass of bogie t 2.43

Bridge‑borne noise model
Mass of wheel set t 1.744
Stiffness of primary suspension kN/m 1252
The bridge FE model, as shown in Fig. 2, was also estab-
Length between bogie centers mm 15700
lished by FE analysis software ANSYS. The box girder
Wheelbase mm 2500
bridge was modelled by 4-node shell elements in order
Wheel diameter mm 860
to accurately simulate the medium- and high-frequency
Wheel density kg/m3 7850
local vibration characteristics of the bridge. The calculated
Young’s modulus of wheel N/mm2 2.06 × ­1011
parameters of the bridge are listed in Table 1.
Poisson’s ratio of wheel - 0.3
The wheel-rail force was applied to the bridge FE
Loss factor of wheel - 0.0001
model as excitation to solve the vibration responses of
Rail Cross-section area of rail mm2 7745
the bridge. Then the bridge FE model was imported into
Rail density kg/m3 7850
vibro-acoustic simulation software LMS Virtual.lab to
Young’s modulus of rail N/mm2 2.06 × ­1011
generate the bridge BE model, as shown in Fig. 2. Finally,
Poisson’s ratio of rail - 0.3
the bridge vibration responses were inputted to the bridge
Loss factor of rail - 0.01
BE model as acoustic boundary condition, and the spatial
Fastener stiffness MN/m 60
distribution characteristics of the bridge-borne noise could
Fastener spacing m 0.625
be predicted.
Bridge Height of box girder m 2.15
Width of bridge deck m 10.9
Width of bridge bottom m 5.4 Comprehensive noise
Bridge density kg/m3 2420
Young’s modulus of bridge N/mm2 3.45 × ­1010 The comprehensive noise is the superposition of wheel-
Poisson’s ratio of bridge - 0.2 rail noise and bridge-borne noise. The sound energy of
Loss factor of bridge - 0.02 wheel-rail noise is mainly concentrated in medium- and
high-frequency band of 500 to 2000 Hz (Thompson 2009),
while the sound energy of bridge-borne noise is mainly
concentrated in the low-frequency band below 200 Hz (Li
cavity referred to Li et al. (2020). The sound absorption et al. 2015b; Zhang et al. 2016a; Song et al. 2018). There-
coefficient of the ground used in the SEA model is shown fore, taking 200 Hz as a boundary, the wheel-rail noise
in Fig. 4. The wheel-rail noise calculated by BEM was model was used to predict the noise in the frequency band

13
Environmental Science and Pollution Research (2022) 29:62330–62346 62335

Fig. 4  Sound absorption coef- 1.0

ficient of ground
0.9

Sound absorption coefficient

0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
100
125
160
200
250
315
400
500
630
800
1000
1250
1600
2000
2500
3150
4000
5000
1/3 octave band center frequency (Hz)

Meanwhile, a numerical model for predicting noise dur-

ing a train passing by was established. Then, the numerical
model was verified by comparing measured and calculated
results.

Outline of the field test

A field test of noise during a train passing by was carried out

on a viaduct of Wenzhou Rail Transit Line S1. Two carbon
fiber rods were erected at 7.5 m and 25 m away from the
center of the right track, respectively. Brüel and Kjær micro-
phones were fixed on the rods; a DEWESoft data acquisition
system was used to record the noise when a train passed by.
As shown in Fig. 6, N1 and N2 were fixed 1.2 m and 3.5 m
above the top of the rail on the rod 7.5 m away from the
center of the right track, while N3 and N4 were fixed 1.2 m
and 3.5 m above the top of the rail on the rod 25 m away
Fig. 5  Photo of the field test from the center of the right track.

of 200 to 5000 Hz; meanwhile, the bridge-borne noise Comparison of the measured and calculated noises
was used to predict the noise in the frequency band below
200 Hz. Test data of noise when a train running on the right track
at a speed of 94 km/h passed through the test section were
chosen to be compared with the data calculated by a numeri-
Validation of the numerical procedure cal model for predicting noise during a train passing by,
as shown in Fig. 7 and Table 2. The measured and calcu-
To verify the above numerical procedure, a test section lated noises agree well, indicating that the numerical model
was selected on a viaduct of Wenzhou Rail Transit Line meets the precision requirement of practical engineering
S1 to carry out a field test of noise when a train passes by, and can accurately predict the noise radiation when the train
as shown in Fig. 5. passes through the viaduct. However, there are still some

13
62336 Environmental Science and Pollution Research (2022) 29:62330–62346

Fig. 6  Layout of measuring

points

90 90
A-weighted sound pressure level (dB(A))
A-weighted sound pressure level (dB(A))

80 Measured 80 Measured
70 Calculated 70 Calculated
60 60
50 50
40 40
30 30
20 20
10 10
0 0
20 31.5 50 80 125 200 315 500 800 1250 2000 3150 5000 20 31.5 50 80 125 200 315 500 800 1250 2000 3150 5000
1/3 octave band center frequency (Hz) 1/3 octave band center frequency (Hz)

(a) (b)
90 90
A-weighted sound pressure level (dB(A))

A-weighted sound pressure level (dB(A))

80 Measured 80 Measured
70 Calculated 70 Calculated
60 60
50 50
40 40
30 30
20 20
10 10
0 0
20 31.5 50 80 125 200 315 500 800 1250 2000 3150 5000 20 31.5 50 80 125 200 315 500 800 1250 2000 3150 5000
1/3 octave band center frequency (Hz) 1/3 octave band center frequency (Hz)

(c) (d)
Fig. 7  Comparison of measured and calculated noises. a N1. b N2. c N3. d N4

13
Environmental Science and Pollution Research (2022) 29:62330–62346 62337

Table 2  Measured and calculated overall SPL [dB(A)] higher above the top of the rail, respectively. These two
Location Measured SPL Calculated SPL
types of noise barriers are mainly composed of H-shaped
steel columns and composite sound-absorbing plates. The
N1 89.36 88.20 distance between the two adjacent H-shaped columns is
N2 88.67 88.39 2 m, and the length of the composite sound-absorbing
N3 80.98 82.12 plate is 1.96 m. The composite sound-absorbing plates
N4 80.82 79.63 are inserted between the two adjacent H-shaped columns,
and the gap between the plate and the column is filled with
rubber strips.
differences between the measured and calculated noises in To compare the acoustic performances of conventional
a couple of frequency bands, which is mainly due to the vertical noise barrier and near-rail low-height noise bar-
unavoidable differences between the calculated parameters rier, the noises when a train passes through viaducts at
and boundary conditions of the numerical model and the a speed of 120 km/h without noise barriers, with con-
actual situation. ventional vertical noise barriers and with near-rail low-
height noise barriers, were calculated, respectively, by
the numerical models. To consider the noise absorption
Numerical analysis and discussions effect of the noise barrier, the sound absorption coef-
ficient of the composite sound absorption plate of the
Numerical models were established based on the verified near-rail low-height noise barrier measured in laboratory
numerical procedure. Acoustic performances of the near- was set in the model, as shown in Fig. 8. For convenience,
rail low-height noise barrier and conventional vertical the difference between the sound absorption coefficients
noise barrier were compared. The influences of the height of the near-rail low-height noise barrier and vertical noise
and the top shape of near-rail low-height noise barrier on barrier was ignored.
the acoustic performances were discussed. Figure 9 shows the spectra of A-weighted sound pressure
levels (SPLs) at different field points. Both the conventional
vertical noise barrier and the near-rail low-height noise bar-
Comparison with the conventional vertical noise rier have obvious noise reduction effects. However, install-
barrier ing the conventional vertical noise barrier will increase the
noise in the frequency range of 63 to 80 Hz, which is mainly
The vertical noise barriers (Fig. 1a) and near-rail low- caused by the secondary structure-borne noise radiated from
height noise barriers (Fig. 1b) are 3.73 m and 0.94 m the conventional vertical noise barrier.

Fig. 8  Sound absorption coef- 1.0

ficient of composite sound-
absorbing plate 0.9
Sound absorption coefficient

0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
100
125
160
200
250
315
400
500
630
800
1000
1250
1600
2000
2500
3150
4000
5000

1/3 octave band center frequency (Hz)

13
62338 Environmental Science and Pollution Research (2022) 29:62330–62346

A-weighted sound pressure level (dB(A)) 90 90

A-weighted sound pressure level (dB(A))

80 80
70 70
60 60
50 50
40 40
30 30
without noise barrier without noise barrier
20 with conventional vertical noise barrier 20 with conventional vertical noise barrier
10 with near-rail low-height noise barrier 10 with near-rail low-height noise barrier
0 0
20 31.5 50 80 125 200 315 500 800 1250 2000 3150 5000 20 31.5 50 80 125 200 315 500 800 1250 2000 3150 5000
1/3 octave band center frequency (Hz) 1/3 octave band center frequency (Hz)

(a) (b)
90 90
A-weighted sound pressure level (dB(A))

A-weighted sound pressure level (dB(A))

80 80
70 70
60 60
50 50
40 40
30 30
without noise barrier without noise barrier
20 with conventional vertical noise barrier 20 with conventional vertical noise barrier
10 with near-rail low-height noise barrier 10 with near-rail low-height noise barrier
0 0
20 31.5 50 80 125 200 315 500 800 1250 2000 3150 5000 20 31.5 50 80 125 200 315 500 800 1250 2000 3150 5000
1/3 octave band center frequency (Hz) 1/3 octave band center frequency (Hz)

(c) (d)
Fig. 9  Sound pressure levels at different field points. a N1. b N2. c N3. d N4

Insertion loss (IL) is the difference in sound pressure near-rail low-height noise barrier is slightly better than that
level before and after installation of noise barriers, which of the conventional vertical noise barrier at all field points.
directly reflects the noise reduction effect of the noise bar- To compare the influences of the conventional vertical
rier. Spectra of IL of the conventional vertical noise bar- noise barrier and the near-rail low-height noise barrier on
rier and the near-rail low-height noise barrier are plotted in the spatial distribution of noise during a train passing by,
Fig. 10. There is little difference in noise reduction effect contour maps of A-weighted SPLs at the mid-span of the
between the conventional vertical noise barrier and the near- bridge without noise barriers, with conventional vertical
rail low-height noise barrier above 200 Hz. However, the noise barriers, and with near-rail low-height noise barriers
noise reduction effect of the near-rail low-height noise bar- were drawn, respectively, as shown in Fig. 11. The noise
rier is obviously better than that of the conventional verti- near the bridge without a noise barrier reaches 98 dB(A)
cal noise barrier, because the structure-borne noise of the and gradually decreases with the increase of the distance
conventional vertical noise barrier is more significant than from the noise sources. The noise drops down to below
that of the near-rail low-height noise barrier. 64 dB(A) at 120 m away from the center of the bridge.
To quantitatively compare the noise reduction effects of Whether the conventional vertical noise barrier or the
the conventional vertical noise barrier and the near-rail low- near-rail low-height noise barrier can obviously reduce
height noise barrier, the values of overall A-weighted SPL the noise around the bridge. The noises at 60 m away from
and IL of different noise barriers at field points N1-N4 are the bridge with conventional vertical noise barriers and
summarized in Table 3. Values of IL of the conventional with near-rail low-height noise barriers reduce to about
vertical noise barrier at field points N1-N4 range from 67 dB(A) and 66 dB(A), respectively, indicating that the
6.63 to 9.57 dB(A), while values of IL of the near-rail low- noise reduction effect of the near-rail low-height noise bar-
height noise barrier at field points N1-N4 range from 7.33 to rier is slightly better than that of the conventional vertical
9.63 dB(A). In other words, the noise reduction effect of the noise barrier.

13
Environmental Science and Pollution Research (2022) 29:62330–62346 62339

15 15

10 10
Insertion loss (dB(A))

Insertion loss (dB(A))

5 5

0 0

-5 -5

-10 conventional vertical noise barrier -10 conventional vertical noise barrier
near-rail low-height noise barrier near-rail low-height noise barrier
-15 -15
20 31.5 50 80 125 200 315 500 800 1250 2000 3150 5000 20 31.5 50 80 125 200 315 500 800 1250 2000 3150 5000
1/3 octave band center frequency (Hz) 1/3 octave band center frequency (Hz)

(a) (b)
15 15

10 10
Insertion loss (dB(A))

Insertion loss (dB(A))

5 5

0 0

-5 -5

-10 conventional vertical noise barrier -10 conventional vertical noise barrier
near-rail low-height noise barrier near-rail low-height noise barrier
-15 -15
20 31.5 50 80 125 200 315 500 800 1250 2000 3150 5000 20 31.5 50 80 125 200 315 500 800 1250 2000 3150 5000
1/3 octave band center frequency (Hz) 1/3 octave band center frequency (Hz)

(c) (d)
Fig. 10  Insertion losses at different field points. a N1. b N2. c N3. d N4

Table 3  Overall SPL and IL of different measures at different field To intuitively compare the influence of the conventional
points vertical noise barrier and the near-rail low-height noise bar-
Location Measure SPL [dB(A)] IL [dB(A)] rier on the spatial distribution of noise during a train passing
by, a contour map of ILs of two types of noise barriers was
N1 Without noise barrier 91.37 - drawn, respectively, as shown in Fig. 12. The noise reduc-
Conventional vertical noise 81.80 9.57 tion effect of the near-rail low-height noise barrier is better
barrier
in most areas beside the bridge and is worse in some areas
Near-rail low-height noise 81.74 9.63
barrier under the bridge than that of the conventional vertical noise
N2 Without noise barrier 90.65 - barrier.
Conventional vertical noise 82.81 7.84
barrier Height of the near‑rail noise barrier
Near-rail low-height noise 82.20 8.44
barrier To compare the acoustic performances of near-rail noise bar-
N3 Without noise barrier 82.60 - riers with different heights, values of IL of near-rail noise
Conventional vertical noise 75.58 7.02 barriers 0.84 m, 0.94 m, and 1.04 m, respectively, above the
barrier
top of the rail at a train speed of 120 km/h were calculated,
Near-rail low-height noise 74.98 7.63
as shown in Fig. 13. The noise reduction effect of the near-
barrier
rail noise barrier gradually improves as the height increases.
N4 Without noise barrier 82.45 -
To quantitatively compare the noise reduction effects
Conventional vertical noise 75.82 6.63
barrier of the near-rail noise barrier with different heights, the
Near-rail low-height noise 75.12 7.33 values of overall A-weighted SPL and IL of near-rail
barrier noise barrier 0.84 m, 0.94 m, and 1.04 m, respectively,

13
62340 Environmental Science and Pollution Research (2022) 29:62330–62346

Fig. 11  Contour maps of

SPLs at mid-span of bridge. a
Without noise barrier. b With
conventional vertical noise bar-
rier. c With near-rail low-height
noise barrier (unit: dB(A))

above the top of the rail at field points N1-N4 are sum- 10.4 dB(A). In other words, when the height of the near-
marized in Table 4. Values of IL of the near-rail noise rail noise barrier increases from 0.84 to 0.94 m and from
barrier 0.84 m, 0.94 m, and 1.04 m, respectively, above 0.94 to 1.04 m, respectively, the noise reduction effect of
the top of the rail at field points N1-N4 range from 6.55 the near-rail noise barrier increases by 0.78 to 1.34 dB(A)
to 8.29 dB(A), from 7.33 to 9.63 dB(A) and from 7.73 to and 0.4 to 0.77 dB(A).

13
Environmental Science and Pollution Research (2022) 29:62330–62346 62341

Fig. 12  Contour map of ILs of 50

different barriers. a Conven-
tional vertical noise barrier. 3.5 4.5
b Near-rail low-height noise
40 5.5 6.0
3.0 4.0
barrier (unit: dB(A)) 5.0
30

Height (m)
2.0
6.5 6.0
20 2.5

10 8.0
9.5 7.5 7.0
10.0 8.5 6.5
9.0 6.0
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Distance from bridge center (m)

(a)
50
7.5
5.5 6.5
4.0 4.5
40
6.0 7.0
3.5
30
Height (m)

3.0
20
5.0

10 8.5 9.0
8.0
0 95
0 10 20 30 40 50 60 70 80 90 100 110 120
Distance from bridge center (m)

(b)

To intuitively compare the influences of the near-rail noise inverted L-shaped, and Y-shaped near-rail noise barriers
barrier with different heights on the spatial distribution of 0.94 m high above the top of the rail at a train speed of
noise during a train passing by, a contour map of differences 120 km/h, as shown in Fig. 15, were calculated, respectively.
in ILs between the near-rail noise barriers of different heights Insertion losses of near-rail low-height noise barriers with
was drawn, as shown in Fig. 14. Although the noise reduction different top shapes are shown in Fig. 16. The noise reduc-
effect of the near-rail noise barrier increases with the increase tion effects of the inverted L-shaped and Y-shaped near-
of the height of the noise barrier, the improvement rate of the rail noise barriers are obviously better than that of vertical
noise reduction effect gradually slows down. For example, one, but there is little difference between the noise reduction
when the height of the near-rail noise barrier increases from effect of the inverted L-shaped and the Y-shaped near-rail
0.84 to 0.94 m, the noise reduction effect of the noise barrier noise barriers.
increases by more than 0.4 dB(A) at 100 m away from the To quantitatively compare the noise reduction effects
center of the bridge, whereas when the height of the near- of the near-rail noise barrier with different top shapes,
rail noise barrier increases from 0.94 to 1.04 m, the noise the values of overall A-weighted SPL and IL of vertical,
reduction effect of the noise barrier increases by less than inverted L-shaped, and Y-shaped near-rail noise barriers at
0.2 dB(A) at 100 m away from the center of the bridge. field points N1-N4 are summarized in Table 5. Values of
IL of the vertical, inverted L-shaped, and Y-shaped near-
Top shape of the near‑rail noise barrier rail noise barriers at field points N1-N4 range from 7.33
to 9.63 dB(A), from 8.07 to 11.16 dB(A), and from 7.98 to
To investigate the influence of the top shape of the near- 10.96 dB(A). In other words, the noise reduction effects of
rail low-height noise barrier, the values of IL of vertical, the inverted L-shaped and Y-shaped near-rail noise barriers

13
62342 Environmental Science and Pollution Research (2022) 29:62330–62346

15 15

10 10
Insertion loss (dB(A))

Insertion loss (dB(A))

5 5

0 0

-5 0.84 m -5 0.84 m
0.94 m 0.94 m
-10 1.04 m -10 1.04 m

-15 -15
20 31.5 50 80 125 200 315 500 800 1250 2000 3150 5000 20 31.5 50 80 125 200 315 500 800 1250 2000 3150 5000
1/3 octave band center frequency (Hz) 1/3 octave band center frequency (Hz)

(a) (b)
15 15

10 10
Insertion loss (dB(A))

Insertion loss (dB(A))

5 5

0 0

-5 0.84 m -5 0.84 m
0.94 m 0.94 m
-10 1.04 m -10 1.04 m

-15 -15
20 31.5 50 80 125 200 315 500 800 1250 2000 3150 5000 20 31.5 50 80 125 200 315 500 800 1250 2000 3150 5000
1/3 octave band center frequency (Hz) 1/3 octave band center frequency (Hz)

(c) (d)
Fig. 13  Insertion losses of near-rail noise barriers with different heights. a N1. b N2. c N3. d N4

are obviously better than that of the vertical one, while the
noise reduction effect of the inverted L-shaped near-rail
Table 4  Overall SPL and IL of near-rail noise barrier with different noise barrier is slightly better than that of the Y-shaped one.
heights To intuitively compare the influences of the near-rail low-
Location Measure SPL [dB(A)] IL [dB(A)] height noise barrier with different top shapes on the spatial
distribution of noise during a train passing by, the contour
N1 Without noise barrier 91.37 -
map of differences in ILs between the vertical and inverted
0.84-m near-rail noise barrier 83.08 8.29
L-shaped near-rail noise barriers was drawn in Fig. 17a, and
0.94-m near-rail noise barrier 81.74 9.63
the contour map of differences in ILs between the vertical
1.04-m near-rail noise barrier 80.97 10.4
and Y-shaped near-rail noise barriers was drawn in Fig. 17b.
N2 Without noise barrier 90.65 -
The noise reduction effect of the inverted L-shaped near-rail
0.84-m near-rail noise barrier 83.31 7.34
noise barrier is slightly better than Y-shaped near-rail noise
0.94-m near-rail noise barrier 82.20 8.44
barrier in most areas beside the bridge.
1.04-m near-rail noise barrier 81.61 9.04
N3 Without noise barrier 82.60 -
0.84-m near-rail noise barrier 75.81 6.79
0.94-m near-rail noise barrier 74.98 7.63
Conclusions
1.04-m near-rail noise barrier 74.55 8.05
N4 Without noise barrier 82.45 -
To assess the acoustic performance of near-rail low-height
0.84-m near-rail noise barrier 75.90 6.55
noise barriers, a numerical procedure for predicting com-
0.94-m near-rail noise barrier 75.12 7.33
prehensive noise of suburban railway bridges considering
1.04-m near-rail noise barrier 74.72 7.73
wheel-rail noise and structure-borne noise of the bridges

13
Environmental Science and Pollution Research (2022) 29:62330–62346 62343

Fig. 14  Differences in ILs 50

between near-rail noise bar- 0.4
riers of different heights. a 0.4
0.94–0.84 m. b 1.04–0.94 m
40
(unit: dB(A))
30

Height (m)
0.3
0.5
20
0.6
1.0 0.4
10 1.4
1.2
1.1
0.9 0.8
1.3 0.7
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Distance from bridge center (m)
(a)
50

0.1
40

30
Height (m)

0.2
0.1
20 0.1
0.3
0.7 0.1
10
0.8
0.5 0.4 0.1
0 06
0 10 20 30 40 50 60 70 80 90 100 110 120
Distance from bridge center (m)
(b)

(a) (b) (c)

Fig. 15  Near-rail low-height noise barriers with different top shapes. a Vertical. b Inverted L-shaped. c Y-shaped

and noise barriers was proposed in this paper and then veri- (1) Near-rail low-height noise barriers have the same good
fied by a field test. On this basis, numerical models were acoustic performance as conventional vertical noise
established to compare the acoustic performances of near- barriers. Meanwhile, the noise reduction effect of the
rail low-height noise barrier and conventional vertical noise near-rail low-height noise barrier is slightly better than
barrier. Influences of height and top shape of the near-rail that of the conventional vertical noise barrier.
low-height noise barrier on the acoustic performance were (2) The acoustic performance of the near-rail low-height
also discussed. The main conclusions are as follows: noise barrier improves with the increase of the height

13
62344 Environmental Science and Pollution Research (2022) 29:62330–62346

15 15

10 10
Insertion loss (dB(A))

Insertion loss (dB(A))

5 5

0 0

-5 vertical -5 vertical
inverted L-shape inverted L-shape
-10 Y-shape -10 Y-shape

-15 -15
20 31.5 50 80 125 200 315 500 800 1250 2000 3150 5000 20 31.5 50 80 125 200 315 500 800 1250 2000 3150 5000
1/3 octave band center frequency (Hz) 1/3 octave band center frequency (Hz)

(a) (b)
15 15

10 10
Insertion loss (dB(A))

Insertion loss (dB(A))

5 5

0 0

-5 vertical -5 vertical
inverted L-shape inverted L-shape
-10 Y-shape -10 Y-shape

-15 -15
20 31.5 50 80 125 200 315 500 800 1250 2000 3150 5000 20 31.5 50 80 125 200 315 500 800 1250 2000 3150 5000
1/3 octave band center frequency (Hz) 1/3 octave band center frequency (Hz)

(c) (d)
Fig. 16  Insertion losses of near-rail low-height noise barriers with different top shapes. a N1. b N2. c N3. d N4

Table 5  Overall SPL and IL Location Measure SPL [dB(A)] IL [dB(A)]

of near-rail noise barrier with
different top shapes N1 Without noise barrier 91.37 -
Vertical near-rail noise barrier 81.74 9.63
Inverted L-shaped near-rail noise barrier 80.21 11.16
Y-shaped near-rail noise barrier 80.41 10.96
N2 Without noise barrier 90.65 -
Vertical near-rail noise barrier 82.20 8.44
Inverted L-shaped near-rail noise barrier 81.03 9.62
Y-shaped near-rail noise barrier 81.18 9.47
N3 Without noise barrier 82.60 -
Vertical near-rail noise barrier 74.98 7.63
Inverted L-shaped near-rail noise barrier 74.18 8.42
Y-shaped near-rail noise barrier 74.28 8.32
N4 Without noise barrier 82.45 -
Vertical near-rail noise barrier 75.12 7.33
Inverted L-shaped near-rail noise barrier 74.38 8.07
Y-shaped near-rail noise barrier 74.47 7.98

of the noise barrier; however, the improvement rate (3) The acoustic performances of the inverted L-shaped and
gradually slows down. Y-shaped near-rail low-height noise barriers are obviously

13
Environmental Science and Pollution Research (2022) 29:62330–62346 62345

Fig. 17  Differences in ILs 50 0.3 0.3

between vertical near-rail noise 0.3
barrier and a inverted L-shaped
near-rail noise barrier and b
40
Y-shaped near-rail noise barrier
0.4
(unit: dB(A)) 30

Height (m)
0.3
20
0.9 0.7
1.5 0.8
10 1.41.2 0.6
1.7 1.11.0
1.6 1.3 0.5 0.3
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Distance from bridge center (m)
(a)
50
0.3

40 0.3

30
Height (m)

0.4
20 0.5

1.4 0.6
1.0
10 1.2
1.5 0.9 0.7
1.1 0.8
0 1.3
0 10 20 30 40 50 60 70 80 90 100 110 120
Distance from bridge center (m)
(b)

better than that of the vertical one, while the acoustic per- Declarations
formance of the inverted L-shaped near-rail low-height
noise barrier is slightly better than the Y-shaped one. Ethics approval and consent to participate Not applicable.

Consent for publication Not applicable.

Competing interests The authors declare no competing interests.

Author contribution Lizhong Song established the numerical model

and was the major contributor in writing the manuscript. Kui Gao,
Quanmin Liu, and Liangtao Zhang performed the measurement. Qing- References
song Feng and Wenjie Guo analyzed and reviewed the measured data.
All authors read and approved the final manuscript. Gorai AK, Pal AK (2006) Noise and its effect on human being-a
review. J Environ Sci Eng 48(4):253–260
Funding This work was financially supported by the National Natu- He W, He K, Zou C, Yu Y (2021) Experimental noise and vibra-
ral Science Foundation of China (52008169, 52068030) and the open tion characteristics of elevated urban rail transit considering the
funding project of the State Key Laboratory of Performance Monitor- effect of track structures and noise barriers. Environ Sci Pollut
ing and Protecting of Rail Transit Infrastructure (HJGZ2021108). Res 28:45903–45919
ISO 3095 (2005) Railway applications - acoustics - measurement of
Data availability The datasets used and analyzed during the current noise emitted by railbound vehicles
study are available from the corresponding author on reasonable Jolibois A, Defrance J, Koreneff H, Jean P, Duhamel D, Sparrow
request. VW (2015) In situ measurement of the acoustic performance of

13
62346 Environmental Science and Pollution Research (2022) 29:62330–62346

a full scale tramway low height noise barrier prototype. Appl Song LZ, Li XZ, Zheng J, Guo M, Wang XX (2020) Vibro-acoustic
Acoust 94:57–68 analysis of a rail transit continuous rigid frame box girder bridge
Koussa F, Defrance J, Jean P, Blanc-Benon P (2013) Acoustic per- based on a hybrid WFE-2D BE method. Appl Acoust 157:107028
formance of gabions noise barriers: numerical and experimental Thompson DJ (1990) Wheel-rail noise: theoretical modelling of the
approaches. Appl Acoust 74:189–197 generation of vibrations, PhD thesis. Southampton: University
Laxmi V, Thakre C, Vijay R (2021) Evaluation of noise barriers of Southampton
based on geometries and materials: a review. Environ Sci Pollut Thompson DJ (2009) Railway noise and vibration: mechanisms, mod-
Res. https://​doi.​org/​10.​1007/​s11356-​021-​16944-2 elling and means of control. Elsevier Ltd., Oxford
Li Q, Wu DJ (2013) Analysis of the dominant vibration frequencies of Thompson DJ, Hemsworth B, Vincent N (1996a) Experimental
rail bridges for structure-borne noise using a power flow method. validation of the TWINS prediction program for rolling noise,
J Sound Vib 332:4153–4163 part 1: description of the model and method. J Sound Vib
Li Q, Xu YL, Wu DJ (2012) Concrete bridge-borne low-frequency 193(1):123–135
noise simulation based on train-track-bridge dynamic interaction. Thompson DJ, Fodiman P, Mahé H (1996b) Experimental validation of
J Sound Vib 331(10):2457–2470 the TWINS prediction program for rolling noise, part 2: results. J
Li Q, Song XD, Wu DJ (2014) A 2.5-dimensional method for the pre- Sound Vib 193(1):137–147
diction of structure-borne low-frequency noise from concrete rail Zhang X, Li XZ, Liu QM, Wu JF, Li YD (2013) Theoretical and exper-
transit bridges. J Acoust Soc Am 135(5):2718–2726 imental investigation on bridge-borne noise under moving high-
Li XZ, Liu QM, Pei SL, Song LZ, Zhang X (2015a) Structure-borne speed train. Sci China Technol Sci 56(4):917–924
noise of railway composite bridge: numerical simulation and Zhang X, Li XZ, Song LZ, Su B, Li YD (2016a) Vibrational and acous-
experimental validation. J Sound Vib 353:378–394 tical performance of concrete box-section bridges subjected to
Li XZ, Zhang X, Zhang ZJ, Liu QM, Li YD (2015b) Experimental train wheel-rail excitation: field test and numerical analysis. Noise
research on noise emanating from concrete box-girder bridges Control Eng J 64(2):217–229
on intercity railway lines. Proc IMechE Part f: J Rail and Rapid Zhang XY, Squicciarini G, Thompson DJ (2016b) Sound radiation
Transit 229(2):125–135 of a railway rail in close proximity to the ground. J Sound Vib
Li XZ, Hu XH, Zheng J (2020) Statistical energy method for noise 362:111–124
reduction performance of the vertical noise barrier alongside rail- Zhang XY, Thompson DJ, Squicciarini G (2016c) Sound radiation from
way bridges. Appl Acoust 170:107503 railway sleepers. J Sound Vib 369:178–194
Liu QM, Thompson DJ, Xu PP, Feng QS, Li XZ (2020) Investigation of Zhang XY, Thompson DJ, Jeong H, Squicciarini G (2017) The effects
train-induced vibration and noise from a steel-concrete compos- of ballast on the sound radiation from railway track. J Sound Vib
ite railway bridge using a hybrid finite element-statistical energy 399:137–150
analysis method. J Sound Vib 471:115197 Zhang XY, Thompson DJ, Quaranta E, Squicciarini G (2019) An engi-
Remington PJ (1976a) Wheel/rail noise part I: characterization of the neering model for the prediction of the sound radiation from a
wheel/rail dynamic system. J Sound Vib 46(3):359–379 railway track. J Sound Vib 461:114921
Remington PJ (1976b) Wheel/rail noise part IV: rolling noise. J Sound
Vib 46(3):419–436 Publisher's note Springer Nature remains neutral with regard to
Remington PJ (1987) Wheel/rail rolling noise, I: theoretical analysis. jurisdictional claims in published maps and institutional affiliations.
J Acoust Soc Am 81:1805–1823
Song LZ, Li XZ, Hao H, Zhang X (2018) Medium-and high-frequency
vibration characteristics of a box-girder by the waveguide finite
element method. Int J Struct Stab Dy 18(11):1850141

13