Noise & Vibration Group

[Link]

Traffic Noise Barriers

by

Dr G R Watts
TRL Annual Review 1995

abcdefghijklmn
TRL Limited
Crowthorne, Berkshire, RG45 6AU
United Kingdom
 TRAFFIC NOISE BARRIERS
by Dr G R Watts, Project Manager in Environmental Research

SUMMARY (ii) Angled barriers - i.e. barriers that are tilted away
or have contoured surfaces angled so as to
This paper describes the major types of noise barrier that disperse the noise, the aim being to prevent
have been employed alongside highways, the mechanisms significant sound reflections into the area where
involved in noise reduction and aspects of the research that screening is required.
has been carried out at TRL to improve their acoustic
performance. To aid optimisation a mathematical model- (iii) Capped barriers - i.e. barriers that have a specially
ling technique using the Boundary Element Method (BEM) shaped top section which reduces the sound
has been used to predict the noise reducing properties of a power contribution from acoustic waves dif-
wide range of barriers. Some of the most promising designs fracted over the top of the barrier.
have been tested at full scale which has involved the
This paper examines the acoustic performance of these
construction of a unique noise barrier test facility (NBTF)
barrier designs and describes the results of TRL research on
at TRL. It was concluded that the average noise screening
capped barriers.
of simple barriers can be increased significantly by rela-
tively simple modifications.
2. THEORY OF BARRIER ATTENUATION
1. INTRODUCTION Figure 1(a) shows the possible significant sound propaga-
tion paths over flat ground for sound radiating from a line
For many years barriers have been erected alongside major
or point source S to a receiver position R. In the absence of
roads to screen residential areas from high levels of traffic
a barrier the noise level at point R resulting from a traffic
noise. In Europe, North America, Japan and Australia
source at S is the sum of the contributions from the direct
many different types of barrier have been installed using a
path SR and the reflected path SGR. The direct and re-
wide variety of materials including wood, steel, aluminium,
flected rays destructively interfere creating excess attenu-
concrete and acrylic sheeting. Some of these designs have
ation over the free field propagation at 800-1000Hz. When
absorptive facings on the traffic side which reduce reflected
a 2m barrier is placed between source and receiver as
sound. Barriers over 8m in height have been used for some
shown in Figure 1(b) the direct contribution is replaced by
applications and novel capped barriers and angled barriers
the diffracted paths over the top of the barrier one of which
have been tested.
involves a reflection from the ground plane, while the
A review of literature world wide has identified a wide interfering reflected path SGR shown in Figure 1(a) is
range of noise barrier systems. Some of the more useful prevented by the barrier. Consequently, the barrier’s
summaries are given by Jackson (1979), Bar (1980), Rocchi effectiveness measured in terms of the change in sound
and Pederson (1985), Bowlby and Cohn (1986), Yamashita pressure levels (SPLs) when the barrier is erected (called
and Yamamoto (1987), Weiss (1988), and West (1989). the insertion loss) is reduced at these mid frequencies as can
Barriers that may offer improved performance over simple be seen in Figure 1(c). However, interference effects take
reflecting barriers can be grouped under the following place between the two diffracted paths shown in Figure 1(b)
broad headings: at lower frequencies (250-500 Hz) due to the greater
difference in path length, so the insertion loss is greatest in
(i) Absorptive barriers - i.e. barriers incorporating this frequency range.
elements on the traffic face that absorb a signifi-
cant proportion of incident sound and hence
reduce reflected sound which could contribute to
overall noise levels in the vicinity.

13
 material needed to effect the required attenuation can be
Receptor
Source R
estimated using the formula:
S
M = 3 . 10 -(A-10)/14 kg/m2 ( Department of Transport,1976)
hr
hs
For most practical cases the structural strength of the
G G r a s s c o ve r e d
material used for the barrier, rather than its acoustic prop-
d erty, is likely to determine the minimum mass required.
(a) Direct and reflected rays over open grassland

B
Receptor
3. ABSORPTIVE BARRIERS
Source R
Where plane vertical barriers exist on both sides of the road
S
Barrier they are normally parallel to each other, and in this situation
hr
sound is reflected back and forth between the barriers,
causing a reverberant build up of sound energy between the
(b) Direct and reflected rays over a barrier
vertical faces of the barriers. Figure 2 illustrates this effect.
In this case, it can be seen that a single reflection from
barrier A can lead to increased noise levels at a receptor
Insertion loss (dB)
position R located behind a second barrier B. Clearly,
10
multiple reflections between the barriers will tend to cause
an even greater deterioration in the screening performance.
Absorbing panels located on the sides of the barriers facing
0 the traffic can reduce this reflected contribution by absorb-
ing the sound energy from the incident wave.

-10 A B Receptor
d path
Reflecte y R
l ra
S ipa
nc th
Traffic source Pri pa
-20

Figure 2: Significant sound propagation paths for

parallel barriers (ignoring ground reflections)
-30
125 250 500 1000 2000
There are several types of system that are used for sound
Frequency (Hz)
absorbing barriers. Briefly, they are:
(c) Insertion loss (ie difference in Sound Pressure Level
with and without barrier)
(i) Hollow box systems containing fibrous material.
For this system the barrier panels are designed so
Figure 1: Sound propagation over grassland that the side facing the traffic is perforated in
and barriers order to allow the transfer of acoustic energy into
the fibrous material contained within the box. The
To be efficient the barrier should largely prevent the direct
opposite side of the barrier is not perforated so
transmission of acoustic energy. In practical situations this
that sound cannot readily be transmitted through
is achieved if the sound energy that leaks through the
the panel. The fibrous material usually consists of
barrier is more than 10 dB below the energy that is dif-
fracted over the top of the barrier and round its edges. If the glass fibre or mineral wool, and is often protected
attenuation in dB(A) afforded by the barrier (A), based on by a thin layer of woven glass fibre to maintain
the path difference, is known then the minimum mass per dimensional stability. Figure 3(a) shows a cross-
section of such a barrier panel.
m2 (M) of a barrier constructed from a single homogenous

14
(ii) Systems which use panels constructed with open- A measure often used to characterise the absorptive prop-
textured porous materials. For these materials, erties of barrier panels is the random incident absorption
absorption is achieved by frictional losses in the coefficient measured in a reverberation room. The test
connected voids of the permeable layer. Exam- methods used by several countries include the use of tests
ples include panels constructed from specially derived from the International Standard ISO 354-1985
fabricated concrete which results in a relatively (Measurement of sound absorption in a reverberation room).
light open porous structure, panels made from A representative portion of the total barrier system, i.e.
treated and compacted wood shavings bonded panels, posts and seals, is constructed in a similar manner
with cement and panels constructed from com- to that used alongside the highway so that the acoustic
pressed coated flint particles. A solid imperme- performance that is likely to be achieved in practice can be
able backing is usually required with these measured. The absorption coefficients are determined for
systems to prevent sound being transmitted each third octave frequency band in the range from 100 Hz
through the panel, unless the panel is thick to 5 kHz. Figure 4 shows absorption spectra obtained for
enough without this. some of the barrier systems described above. It should be
noted that a value of unity for the absorption coefficient α
(iii) Systems containing resonant cavities. The traffic
indicates that the material is fully absorbing at that fre-
side of the barrier contains slots or holes which
quency, with all incident acoustic energy absorbed and no
connect with internal cavities. Resonance occurs
measurable reflection, though some of the α values plotted
at selected frequencies depending upon the
in Figure 4 are actually greater than unity due to measure-
dimensions of the cavities, causing relatively high
ment error and diffraction effects at the edge of the sample.
amplitude oscillations in the neck of the cavity
which lose energy through frictional damping.
Fibrous or foam fillers can be included in the Sound absorption coefficient
cavities to broaden the frequencies of the sound
1.2
absorbed. Figure 3(b) shows an example of a
slotted masonry block which can be used to Porous concrete

construct a load bearing wall. 1.0

Woodfibre cement
0.8
Metal/fibre fill
Solid wooden
panel
Perforations 0.6
Rockwool
Masonry cavity block
Rigid profiled 0.4
aluminum sheet
Traffic face

0.2

0
125 250 500 1000 2000 4000
Frequency (Hz)
(a) Perforated aluminium sheets with rockwool fill material

Figure 4: Random incidence sound absorption coeffi-

cients by one-third octave band frequency for different
barrier materials

Clearly, to be effective the barrier material must be highly

absorptive at frequencies that are significant in highway
Cavity with sound absorptive
material traffic noise spectra. Figure 5 shows a typical A-weighted
(b) Masonry cavity block
spectrum which was obtained from measurements taken
close to a rural dual carriageway road surfaced with a
Figure 3: Sound absorptive barrier systems conventional hot rolled asphalt. It is clear that the frequency

15
spectrum is broad band, i.e. it does not exhibit any signifi- 4. ANGLED BARRIERS
cant tonal characteristics. The highest sound levels occur at
frequencies close to 1 kHz. For effective performance, An alternative to using sound absorptive barriers is to angle
therefore, traffic noise barriers should absorb strongly over the barrier or parts of the barrier away from the road such
a wide range of frequencies between, say, 250 to 5000 Hz. that the reflected wave from the traffic face of the barrier is
deflected upwards. This reduces the contribution to noise at
A-weighted SPL (dB)
receptor positions relatively close to the ground.
70
Menge (1978) carried out scale model tests on parallel
barriers using plywood to represent hard reflecting sur-
60 faces. Insertion loss was measured with the barriers erected
vertically and angled at 5, 10 and 15° to the vertical. At an
50 angle of 10° or greater the improvement in insertion loss
was very similar to that produced by parallel barriers
40 having fully absorptive faces.

30 The performance of angled barriers was also measured at

full scale at TRL’s Noise Barrier Test Facility and the
results of Menge were confirmed. We also demonstrated,
20
however, that it was not necessary to tilt the whole face of
10 the barrier (Watts, 1994). Tilting a 1m wide central strip of
the barrier produced an average reduction in noise levels
only 0.3dB(A) higher than if the entire face of the barrier
0
125 250 500 1000 2000 4000 had been tilted to the same degree. It should be feasible to
Frequency (Hz) produce a contoured panel with this degree of tilt which
would be as effective as fully coating the traffic face of the
barrier with absorptive material, yet at lower cost. One
Figure 5: A-weighted spectrum for traffic noise pro- potential problem with tilting barriers is that under adverse
duced on a typical bituminous surface meteorological conditions the reflected sound waves could
be refracted such that they return to ground level.
While most of the absorptive materials perform adequately
it can be seen that the cavity block absorber exhibits
absorption coefficients which peak at relatively low fre- 5. CAPPED BARRIERS
quencies and therefore would not be expected to have a
significant effect on reducing typical highway traffic noise. In addition to reducing reflected noise, barriers have been
It should be possible, however, to extend the performance altered in cross-section in an attempt to reduce the noise
of this material to higher frequencies by adjusting the size diffracted into the shadow zone. Potentially this is the most
of cavity and fill material. fruitful area of research since it can be shown that relatively
minor changes made at the diffracting edge can have
The performance of absorptive barriers were measured at significant effects on screening performance. For this rea-
full scale at TRL’s Noise Barrier Test Facility as part of a son much research has been carried out to optimise the
research programme on novel designs of barrier (Watts, design of the barrier profile. Many designs have been
1994). Details of the facility are given in section 5 below. examined using mathematical and scale modelling, and the
For these tests 2m high reflective barriers were erected 34m more promising designs have been tested at full scale.
apart on flat ground and noise measurements were taken out
to 80m behind the barrier and up to a height of 4.5m. Results For the purpose of obtaining reliable information on dif-
indicate that the addition to the barrier faces of absorptive fraction effects a unique full scale noise barrier test facility
material with the correct acoustic characteristics can re- has been constructed at TRL (Figure 6). The noise source
duce noise levels in the screened area by 4dB(A). This is consists of an 800W speaker which can be positioned to
equivalent to removing over half the traffic from the road represent the traffic source on motorways and all purpose
. dual carriageway roads as given in the UK standard traffic

16
 noise prediction model “Calculation of Road Traffic Noise”
(Department of Transport, 1988). Microphones can be
positioned to measure the noise level at any point on a wide
flat mowed grassland area free of reflecting objects. To
measure the performance of the barrier under test condi-
tions, recorded noise is broadcast in four broad frequency
bands sequentially, producing in turn an approximate level
of 110 dB in each of the one-third octave bands from 100
Hz to 3.15 kHz at a reference microphone positioned 1m
from the speaker. The one-third octave band levels are
weighted using an averaged A-weighted highway traffic
Figure 6: View of Noise Barrier Test facility at TRL, noise spectrum (Watts et al, 1994), and corrections can be
showing test barrier with rounded absorptive cap made for variations in speaker output and wind speed and
(barrier k in Figure 7) and speaker system direction. In this way the screening performance of the
for noise source barriers for a typical traffic noise source can be evaluated.

Most of the designs chosen for testing

Simple resulted from mathematical modelling
reflective
barriers of the insertion losses of a wide range of
3 barrier shapes using a boundary ele-
2 2.5 ment method. The designs included T-
0 1.7 3.6
shaped barriers and multiple edged bar-
riers as well as commercially available
(a) (b) (c)
designs (Figure 7). Option k consists of
1 2 a rounded absorptive cap 0.5m in
T - shaped
barriers diameter (Figure 6) and option l is a
device designed to exploit the principle
2 of sound interference. The inventors of
1.4 2.0 3.1
device l claim that the sound entering
(d) (e) (f) the ducts on the traffic face of the device
is channelled down through the ducts
1 2
Multiple and then destructively interferes with
edge 0.5
the sound that has propagated directly
barriers 1
across the top edge of the device (Iida et
2.4 2.5 2.6 2.7 al, 1984).

(g) (h) (i) (j) The overall performance of the barriers

was calculated by computing the aver-
Commercial
designs age sound level (corrected to zero wind
All dimensions in metres speed) over six microphone and two
2.5 Absorptive material speaker conditions. Figure 7 shows the
2.2
1.9 Relative average
0.8 0 average reduction in noise levels for the
insertion loss
different barrier profiles compared with
(k) (l) a simple reflecting barrier of identical
overall height. It can be seen that gains
Figure 7: Cross-section dimensions and relative of 2.5 - 3 dB are possible with both T-shaped and multiple
insertion losses of simple and capped barriers edge barriers. Adding these profiles has the same effect as
raising the height of a simple plane barrier by nearly 1m.
Measurements were repeated along a line 30° to the perpen-
dicular to the barrier to determine whether relative insertion

17
loss varied with angle. The results for measurements at 30° reformulation of the Helmholtz wave equation in terms of
to the barriers were similar to those recorded along the an integral equation. For this purpose the barrier and
perpendicular (Watts et al, 1994). This indicated that the ground surfaces are divided into boundary elements of
gains for a line source such as a traffic stream would be length no greater than L/2 (where L is the wavelength). The
close to those measured in these tests. Such barrier profiles effects of ground cover and absorptive surface treatment of
might therefore be useful for screening traffic noise in the barrier can also be included. Homogeneous, still air
situations where the maximum height of barriers needs to conditions are assumed in all computations.
be limited because of environmental considerations (e.g.
visual intrusion, reduction in sunlight) or where extra Calculations were carried out using the BEM for the same
screening is required from an existing barrier and the costs source, receiver and barrier configurations as in the experi-
of increasing the height would be excessive. mental investigations and the same methodology was used
to process the results. Although the model assumes cylin-
drical spreading and the experimental results are for a point
6. THEORETICAL MODELLING source, the different assumptions do not affect comparisons
of insertion loss, though of course predictions of the vari-
The numerical model used is based on the boundary ele- ations of SPL with distance are affected.
ment method (BEM). It is a two-dimensional model which
in three dimensions is equivalent to an infinite line source, In the BEM it is necessary to define the characteristics of
and an infinitely long barrier parallel to the source. Along the absorbing surfaces of the barrier and of the grassland.
its length the barrier has a uniform cross-section and The Delany and Bazley model (Delaney and Bazley, 1970)
surface covering. The model calculates the wave field was used to describe approximately the impedance of each
behind the barrier at a particular frequency by solving a of these surfaces. For the barrier surfaces a flow resistivity

Table 1: Average noise levels behind barriers and relative insertion losses at normal incidence
Option Average Relative insertion Theoretical
A-weighted SPL* loss compared predicted
(number of repeat with 2m high barrier relative
measurements) insertion loss

Simple reflective
a) 2m high 64.8 (14) ± 0.1 - -
b) 2.5m high 63.1 ( 8) ± 0.1 1.7 1.9
c) 3m high 61.2 (16) ± 0.1 3.6 3.4

T-shape, 2m high
d) 1m wide reflective 63.4 ( 4) ± 0.2 1.4 0.4
e) 1m wide absorptive 62.8 ( 8) ± 0.1 2.0 2.5
f) 2m wide absorptive 61.7 ( 6) ± 0.1 3.1 3.7

Multiple edge, 2m high

g) 1m wide reflective - 0.5m panels 62.4 ( 8) ± 0.1 2.4 1.4
h) 1m wide absorptive - 0.5m panels 62.3 (10) ± 0.1 2.5 2.1
i) 1m wide absorptive - 1m panels 62.1 ( 6) ± 0.1 2.6 2.6
j) 2m wide absorptive - 0.5m panels 62.1 ( 6) ± 0.2 2.7 1.4

Commercial systems **
k) Absorptive cylinder 62.3 (12) ± 0.1 0.8 -
l) Interference device 62.1 (24) ± 0.1 1.9 1.5

* Sound Pressure Level

** For commercial systems comparisons were made with a simple barrier of identical height

18
 Insertion loss (dB) Insertion loss (dB)
10 10

0 0

-10 -10

-20 -20

Expt Expt
Boundary Element Method (BEM)
-30 -30
Boundary Element Method (BEM)

-40 -40

-50 -50
125 250 500 1000 2000 125 250 500 1000 2000
Frequency (Hz) Frequency (Hz)

(a) Simple 2 metre barrier (b) Multiple edge barrier

Figure 8: One-third octave band insertion loss spectra at a distance of 40m, with source 5.5m from barrier, measured
experimentally (blue line) and calculated by Boundary Element Method (BEM) (red line)

of 20,000 Nsm-4 and a depth of 0.1m were used. In order to measurements on the NBTF, and from BEM modelling. As
determine suitable parameters to describe the grassland explained above, peaks in the experimental results at mid
surface separate experiments were carried out throughout frequencies are a result of the ground absorption effect in
the testing period to determine spectra of the sound at a the spectrum of the unobstructed sound field. There is very
range of receiver positions. A fitting procedure enabled good agreement between the position of these peaks in the
best values for the parameter to be derived. Values of flow experimental and theoretical curves, indicating accurate
resistivity between 200,000 and 400,000 Nsm-4 were meas- modelling of the ground and barrier configurations.
ured depending on the moisture content of the ground. A
value of 250,000 Nsm-4 (best fit for the dry condition) was
chosen as representative of the majority of conditions. 7. FUTURE DEVELOPMENTS
Further details of the model used are given by Hothersall et
The tests at full scale and modelling work indicated the
al (1991).
potential benefit of multiple edge barriers for improving
Table 1 shows the measured average reductions of the acoustic performance. Further tests were carried out at the
profiles compared with a simple plane barrier of identical roadside by attaching a 1m wide multiple edge profile to
height together with the results predicted from the BEM motorway barriers. Figure 9 shows a view of an asymmet-
analysis. It can be seen that the results are generally similar ric design attached to a 2m tall barrier on M25. The decrease
and indicate the usefulness of the model for predicting the in noise was similar to that predicted from tests at full scale.
performance of untested profiles. These encouraging results have led to a decision to develop
commercial products. Further modelling is being carried
To obtain a further understanding of the way in which the out and it should be possible to tune the barrier profile to
various profiles improved screening performance and to achieve the most cost-effective means of reducing noise in
further validate the BEM predictions, the spectra of inser- a given situation. For example, closing the space between
tion loss for the various profiles were examined. In Figure the lower edge of the side panels and the main barrier in
8 are shown spectra of insertion loss for a simple 2m high Figure 7(h) gives better noise reduction than the open
barrier and a multiple edge design, as obtained both from design, particularly when attached to a tall barrier. The

19
average difference in noise level with and without this DELANY, ME and E N BAZLEY (1970). Acoustical properties of
fibrous materials. Applied Acoustics 3, p 105.
profile fitted to a 4m tall barrier is 3.9 dB(A) compared with
1.1 dB(A) for the original design with an absorptive panel. DEPARTMENT OF TRANSPORT and WELSH OFFICE (1988). Calcu-
lation of Road Traffic Noise. London: HMSO.
This may be because any sound leakage between the side
panels and main barrier is prevented by the presence of the DEPARTMENT OF TRANSPORT (1976). Noise barriers - standards
and materials. Technical Memorandum H14/76. London: HMSO.
base plate.
HOTHERSALL D C, D H CROMBIE and S N CHANDLER-WILDE
(1991). The performance of T-profile and associated noise barriers.
Applied Acoustics, vol 32, p 269-287.

IIDA K, Y KONDOH and Y OKADO (1984). Research on a device for

reducing noise, Transport Research Record 983, Washington DC. pp 51-
54.

INTERNATIONAL ORGANIZATION FOR STANDARDIZATION

(1985). Measurement of sound absorption in a reverberation room. ISO
354-1985. London: BSI.

JACKSON G M (1979). A review of highway noise barriers. Highway

Engineer. November, pp 13-16.

MENGE C W (1978). Sloped barriers for highway noise control. Proceed-

Figure 9: Asymmetric multiple edge barrier profile ings of Internoise 78, San Francisco, 1978, pp509-512
attached to a 2m high barrier alongside the M25
ROCCHI S E and S PEDERSON (1985). Feasibility of Transparent noise
barriers. Transportation Research Record 1255. pp 87-93.

8. CONCLUSIONS WATTS G R (1994). Acoustic performance of parallel noise barriers - full

scale tests. Paper presented at Inter-noise ’94, Yokohama. Proceedings of
1994 International Conference on Noise Control Engineering, 1, pp 538-
The following conclusions can be drawn from this state-of- 586.
the-art review of noise barriers.
WATTS, G R, E CROMBIE and D HOTHERSALL (1994). Acoustic
performance of new designs traffic noise barrier - full-scale tests. Journal
1. Absorptive barriers can be designed to eliminate re- of Sound and Vibration, 177, pp 289-305.
flected noise but it is necessary to choose the absorptive
WEISS M (1988). Summary of highway noise barrier construction in the
material to eliminate the most important frequencies of United States. Transportation Research Record 1176, Washington DC.
traffic noise at the site of application. pp 1-4.

WEST P J (1989). The development of innovative traffic noise barriers

2. Tilted or contoured barriers should be considered as an 1989. National Transport Conference, Melbourne. The Institution of
alternative to absorbing barriers, but account needs to be Engineers.
taken of the possibility of the reflected sound being re-
YAMASHITA, M. and K YAMAMOTO (1987). Sound barriers - how are
fracted back to ground level under certain adverse meteoro- they used, Proceedings of the International Conference on Noise Control
logical conditions. Engineering, Inter-noise’87, Beijing. pp 399-402.

3. Multiple edge barriers and other barrier shapes are a

solution to enhancing the acoustic performance of barriers ABOUT THE AUTHOR
without raising the overall height of the barrier system.
Greg Watts gained a first class degree in Physics at Man-
4. The usefulness of the Boundary Element Method for chester University and joined TRL in 1969. He is a Fellow
designing barrier profiles to enhance performance has been of the Institute of Acoustics and has published over 60
demonstrated. technical papers. He currently specializes in mathematical
modelling of sound propagation to develop cost effective
means of mitigating traffic noise. This has involved mod-
REFERENCES elling both novel designs of noise barrier and road surfaces.
He has carried out fundamental research into sound percep-
BAR P (1980). Materiaux utilises dans les ecrans acoustiques en bordure
de voies routieres. Bulletin PCM, v4, p19-23. tion using TRL’s unique listening room facilities. He
currently chairs BSI and CEN committees concerned with
BOWLBY W and L M COHN (1986). Sound-absorptive highway noise
barriers. Federal Highway Administration, FHWA-TS-86-214.
developing standards for noise barriers.

20