You are on page 1of 9

TRAFFIC NOISE SOURCES IN

CUTTING DIPS ADJACENT TO


SCREENS
Andrew T Peplow

Marcus Wallenberg Laboratory for Sound and Vibration Research (MWL),


Department of Aeronautical and Vehicle Engineering, Kungl Tekniska Högskolan
(KTH), Stockholm, 100 44, Sweden. atpeplow@kth.se

ABSTRACT
In addition to considerations of acoustical efficiency, the design of road traffic noise
barriers depends on many other factors. Site conditions, planning restraints and visual
considerations will affect the final selected design, as will engineering considerations
related to the construction materials, the weather loading due to wind and
precipitation, the resistance to impact by vehicles, vandalism and repair and
maintenance. However, the use of a single vertical screen to obstruct the propagation
of noise from source to receiver is well known. The aim of this article is to show the
acoustical benefits and drawbacks of using a screen adjacent to a highway located in a
dip in an otherwise flat ground topography.

1 INTRODUCTION
Barriers are frequently applied as protectors against noise, where the so-called shadow
area is used. The barrier efficiency as a noise control measure depends on the barrier
itself and its interactions with the environment. All these factors cause degradation of
the barrier efficiency. In the free space the barrier efficiency is determined by barrier
acoustical features and the source-barrier-observation point geometry, and is defined
as the insertion loss (IL):
IL = 20 log10 ( Pg / Pb )
where Pg is the pressure at the receiver with the flat ground and cutting dip present,
and Pb is the pressure with both ground features and barrier present. The prediction
of acoustic propagation from a traffic noise source located in a dip over a noise-screen
onto surrounding flat grassland or rigd ground is presented. It is well known that over
flat ground the improvement of “insertion loss” for single noise screens have
significant marked differences for propagation over absorbing ground against rigid
ground, Hothersall, Chandler-Wilde and Hajmirzae [1.]. This is due to the screen
obstructing surface wave attenuation due to the absorbing ground cover. The aim here
is to extend the study of this phenomena for screens adjacent to small dips. A
numerical model using boundary element techniques enables the excess attenuation
and insertion loss for various noise barriers and cuttings of complex profile and
surface cover to be calculated. The model is applied to single-foundation noise
barriers to which additional absorbent or rigid side-panels may be added. Spectra of

1
insertion loss for a broadband traffic-source are presented. For a recent general review
of the application of boundary integral equation methods in modelling outdoor noise
problems see Chandler-Wilde [2.] and Morgan [3.]; for interference type multi-edge
profiles see Crombie, Hothersall, Chandler-Wilde and Watts [4.], [5.] [6.] and Peplow
and Chandler-Wilde [7.] for the problem without a noise screen. A review of
diffraction modelling for noise screens see Muradali and Fyfe [8.].

Figure 1: Measurement setup in semi-anechoic chamber for validation of model.


Screen and loudspeaker clearly visible.

2 NUMERICAL AND LABORATORY EVALUATION


In order to evaluate a number of alternative screen designs with respect to the
insertion loss, measurements was performed in a semi-anechoic room at MWL, KTH.
Also a boundary element model for screen insertion loss simulation was evaluated
using the screen test set-up.

2.1 TEST CASE


The test set-up as shown in Figure 1 consisted of a 6.0 m long section placed on the
floor along one side of the room. The screen was a 1.12 m high, 6 m wide and 0.02 m
thick chipboard panel that was mounted to the floor on cantilever beams. The
horizontal distance between the screen and the source center was 1.75 m. A monopole
source realized with a loudspeaker was used. Secondly two different screen designs
were tested, a plain screen, and a screen with a 0.08 m wide member mounted
perpendicular to the screen top. The modified inverted L-screen was introduced to
investigate any “standing waves” that may occur between the screen and the member.
Any successful minor modification to a barrier design may serve as a useful retro-fit
device. During measurements the source was excited with white noise. The sound

2
pressure level was measured, at a receiver point 7.5 m from the track center and 1.62
cm above ground, see Figure 1. The sound pressure level third octave spectra for the
different cases were estimated using proper averaging and signal analysis procedures.
Finally the screen insertion loss (IL) was estimated as the difference in sound pressure
level measured, without and with screen respectively.

4.5m

1.12m
1.5m

H
0.92m

1.5m
20m 40m 80m
1.0m

Figure 2: Barrier and dip geometry for traffic noise model

2.2 BOUNDARY ELEMENT MODEL


The coherent line source configuration modelled is not a practical one, as in reality a
number of different vehicles along the length of a road will contribute to the noise
field at a position behind the barrier (approximating an incoherent line source).
However, the results can be related to practical measurements. Specifically, the
numerical model predictions of insertion loss, at a particular frequency and receiver
position, agree very well with insertion loss results for a point source of sound. The
point source is located on the same line as the line source, and in the same vertical
plane as the receiver, perpendicular to the barrier. To confirm this observation,
comparisons have been made by Hothersall, Chandler-Wilde and Hajmirzae [1.] with
insertion loss spectra for point sources of sound, and model experiments in an
anechoic chamber and outdoor experiments, Watts [5.] for a variety of barrier cross-
sections.

A two-dimensional boundary element model was designed to simulate the test setup
shown in Figure 2. The grassland region, was modelled with a frequency dependent
absorption coefficient whereas the screen, the ground on the screen shadow side and
road model 4.0m from the screen were modelled as rigid. In all simulated cases the
sources were assumed to be harmonic and located 0.92m above a rigid highway. The
approach adopted in this study is to solve accurately the governing wave equation by
using the boundary element method which allows complete flexibility in the
specification of noise barrier design, external members, and distribution and type of
ground surface and absorbent barrier elements. In eq. (1) below γ denotes those parts
of the boundary for which the specific surface admittance β ≠ β c , reciprocal of

3
surface normal impedance, or which lie above the admittance ground denoted by β c .
The boundary element method is a numerical method applied to the reformulation of
the Helmholtz equation as a boundary integral equation. In this study the integral
equation employed is similar to the approximation for a barrier above flat ground
Crombie, Hothersall and Chandler-Wilde [4.].:

 ∂Gβc (r , rs ) 
ε (r ) p(r ) = p0 (r ) + ∫  − ik β (rs )Gβc (r , rs )  p (rs ) ds (rs ) (1)
γ 
∂n(rs ) 

the true formulation is derived and stated in Peplow [9.] and [10.]. For the purpose of
this study the boundary element method is implemented using a source located at one
side of the screen and exhibiting monopole radiation characteristics. The numerical
model makes the assumption that the material is locally reacting and, to describe the
surface impedance, the empirical model of Delany and Bazley, [11.] is used with a
flow resistivity of 250 000 Nsm-4 and a layer depth 0.1m for grassland behind the
screen,

25

20
Insertion Loss (dB)

15

Rigid Screen
10
Rigid L-Screen

5 Rigid Screen
(simulation)
L-Screen
(Simulation)
0
100 1,000 10,000
Freq (Hz)

Figure 3: Measured and calculated insertion loss of vertical screen and L-shaped
screen with loudspeaker excitation.

2.5 RESULTS – LABORATORY TESTS


Both simulated and measured results clearly showed that for this test set-up selecting
a source with monopole characteristics. The measured screen insertion losses with
loudspeaker excitation for the two different screen alternatives are plotted in Figure 3.
Clearly the difference between the two alternatives is smaller than 1 – 2 dB. It seems
that an additional member only increases the effective height of the vertical noise

4
screen as no standing waves, expected near 1000 Hz, were observed. The inverted L-
shaped barrier was not modeled in the simulations but the increased “effective height”
of the screen reappears later in the analysis when there is a dip in the highway.

Figure 4: Barrier insertion loss for flat rigid ground and dip cutting over rigid
ground, H=0.1m

The boundary element model was evaluated as a tool for predicting the insertion loss
of low screens placed close to the source with interaction between the screen and the
source is taken into account. In Figure 3 the simulated plain screen insertion loss is
plotted together with the measured data. The agreement is good considering the
simplifying assumptions. The frequencies for insertion loss maxima and minima are
accurately predicted with the insertion loss peaks and troughs are exaggerated by
around 2 dB. Notice that at lower frequencies, around 100 Hz, measured and
predicted disagree. This is partly due to the different source signals; white noise in the
measurements and pure tone in the simulation and partly to the fact that the measured
results are third octave band results. Note that the results for the simulations are for
one receiver position only. The results are encouraging in terms of noise control
against the standard set for traffic noise screen design. However, sound reduction at
much larger distances, over 100m say, is not as significant. The following section
discusses the situations at larger distances, with and without sources in dips, and
results where a grassland model is included.

3 NUMERICAL SIMULATIONS
Figures 4 and 5 illustrate the spectra of insertion loss at two receiver positions for a
vertical noise screen in rigid ground and grassland environments respectively. As
shown in Figure 2 the edge of the rigid highway is 4.0m from the screen. It is clear
from both figures that a slight perturbation in the geometry, H=0.1m, only produces a

5
significant effect above 1000 Hz. Note that there is no significant difference in
insertion loss across the frequency range for rigid ground and grassland results. Over
the entire frequency range and six receiver positions an average insertion loss value
may be obtained. This is evident as columns A in Figure 6.

The attenuation of a barrier for a broadband noise source, with a spectrum


representative of A-weighted road traffic noise, was also obtained at each of the
receiver positions by combining the calculated results at third octave centre
frequencies for propagation with and without the presence of the barrier. The effect of
this combination is in smoothing out the complicated interference effects in the
spectra for receiver positions above the ground and is presented as contour diagrams
in Figure 7.

Figure 5 : Insertion loss for flat grassland and dip cutting over grassland,
H=0.1m

Computed average insertion loss results are shown in Figure 6. These show that there
is a significant change in the physics of the problem if a source is located in a dip
cutting perturbation compared to sound propagation attenuation especially over flat
grassland. The cutting changes the source characteristics significantly especially if the
source is located within the cutting or in its immediate vicinity. It seems that the
effective height of the 1.12m noise screen increases, up to 3dBA as the depth of the
cutting, H, increases. Hence the average insertion loss values increase but then
decrease significantly as the depth of the dip reached H=1.6m, decreasing by 3dBA.
For rigid ground the addition the sides and edges of the cutting create an extra
screening effect which may disturb the shadowing or diffraction qualities of a noise
barrier. These two effects are not easily quantifiable but the following show the
overall qualitative behaviour. However, the right hand figure shows average insertion
loss at the six receiver positions as the cutting dip increases in depth. It is clear that in

6
all but two receivers the insertion loss decreases with increasing depth of cutting dip.
This is a significant result as it appears that not only does the screen interrupt any

Figure 6: Average insertion losses over rigid geometry (LH) and grassland
geometry (RH)
surface waves that may propagate along the grass ground cover but the cutting dip
also plays a significant role in the disturbance of attenuated sound over grassland.
Without the screen any noise source located in a cutting dip sound propagation may
be attenuated significantly over grassland environments. It may be concluded that a
noise screen in this environment is not necessary. Contours of average insertion loss
for various geometries over rigid ground are illustrated in Figure 7.

4 CONCLUSIONS
The following conclusions may be drawn from the analysis:

1. The ground geometry in the immediate vicinity of a noise source, such as a


vehicle sitting in a cutting, can significantly effect the sound attenuation
beyond a traffic highway.

2. The effective height of a noise screen may increase if a source is located in a


cutting dip.

3. Obstruction of sound propagation, by screens for example, from sources in


cuttings can have a negative effect on attenuation.

Controlled laboratory measurements to test the efficiency of screens, with sources


located in small cuttings, are being planned to be validated against field measurements.
In the future stronger conclusions may be drawn for numerical predictions that have
been validated against physical data.

7
(A) (B)

(C) (D)
Figure 7 : Contours of weighted insertion loss (dBA) for four geometries over
rigid ground with barrier. (A) rigid flat ground, (B) cutting dip, H=0.1m, (C)
cutting dip, H=0.8m, (D) cutting dip, H=1.6m.

ACKNOWLEDGEMENTS
This project is a part of the framework project Gröna Tåget. The author is most
grateful to Kent Lindgren, Alexander Häggström, who performed the laboratory
measurements and Ulf Carlsson, Leping Feng and Mats Åbom who cooperated on the
project.

REFERENCES
[1.] Hothersall DC, Chandler-Wilde SN, Hajmirzae MN. Efficiency of single
noise barriers. Journal of Sound and Vibration 146(2), 1991; 303-322.
[2.] Chandler-Wilde SN. Tyndall Medal Lecture : The boundary element method
in outdoor noise propagation, Proc. of the Inst. of Acoustics 19, 1997; 27-50.
[3.] Morgan PA, `Boundary element modelling and full scale measurement of the
acoustic performance of outdoor noise barriers,' Ph. D. thesis, Brunel University,
Uxbridge, UK, 1999.
[4.] Crombie DH, Hothersall DC, Chandler-Wilde SN. Multiple-edge noise
barriers. Applied Acoustics 44(4), 1995; 353-367
[5.] Watts GR, Acoustic performance of multiple edge noise profile at motorway
sites, Applied Acoustics, 47, 1996; 47-66.

8
[6.] Watts GR, Morgan PA. Acoustic performance of an interference-type noise-
barrier profile, Applied Acoustics, 49, 1996; 1-16.
[7.] Peplow AT, Chandler-Wilde SN. Noise propagation from a cutting of
arbitrary cross-section and impedance, Journal of Sound and Vibration, 146, 1999;
303-322.
[8.] Muradali A, Fyfe KR. A study of 2D and 3D barrier insertion loss using
improved diffraction-based methods, Applied Acoustics, 53(1-3), 1998;49-75.
[9.] Peplow AT, Numerical predictions of sound propagation from a cutting over a
surrounding flat ground, Journal of Computational Acoustics, 13(1), 2005; 145-162.
[10.] Peplow AT, On the efficiency of screens near roadside cuttings
Journal of Sound and Vibration, 298(4-5), 2006; 982-1000.
[11.] Delany ME, Bazley EN, Acoustical properties of fibrous absorbent materials,
Applied Acoustics, 3, 1970; 105-116.

You might also like