- Email: [email protected]
The performance of t-profile and associated noise barriers
Applied Acoustics 32 11991i 269-287
The Performance of T-Profile and Associated Noise Barriers
D. C. Hothersall, D. H. C r o m b i e & S. N. Chandler-Wilde Department of Civil Engineering, University of Bradford. Bradford, West Yorkshire BD7 IDP, UK Received I May 1990: revised version received 11 July 1990: accepted 25 July 19901
ABSTRACT .4 ret'iew ts presented on t/re results of pret'ious research, using experimental modelling and field measurement, qf the insertion loss produced hy T-. Y- and arrow-profile noise barriers. ,4 numerical model is discussed which enables tire ware field in the region of a barrier with a complex profile and surface cot'er to he determined. An brtegral equationJormulation of the problem is soh'ed us#rg the boundao" element approach. Spectra of insertion loss Jor T-. Y- and arrow-profile barriers calculated using ttre model are discussed. Results for the mean insertion loss orer a range of receirer positions for a broad-hand source are also presented. It is concluded that for harriers with reflecting surfaces the changes in #Tsertion loss for different profiles can be described using the concept of path difference. The intro~hwtion of absorbing upper surfaces produces a significant increase in insertion loss. br most conditions the Y- and arrow-profiles perform less efficiently than the T-profile. Tire marked increase b7 barrier efficiency when a rerv thin narrow cap is added to a t'ertical wall which has been reported by other" workers was not obsert'ed.
i INTRODUCTION The addition of a horizontal cap to a conventional barrier, thus creating a ~T-profile', has been the subject of a limited number of previous studies, t -'~ encompassing either acoustical scale modelling, direct field measurements, or both. In general, it has been established that a design o f this kind is likely 269 Applied Acoustics 0003-682X/9l/$03-50 ~ 199l Elsevier Science Publishers ktd, England. Printed in Great Britain
270
D. C. Hothersall, D. H. Crombie, S. N. Chandler-Wilde
to achieve a small gain in insertion loss over a conventional barrier of the same height. Although the gains reported from direct field measurement are modest, typically around 1"0-1"5 dB(A) for cap widths up to 1 m, there are a number of factors which suggest that T-profile barriers warrant further study. The terms "cap width" and "cap depth', used throughout this report, are illustrated in Fig. 1. Certain design factors seem to favour the addition of a horizontal cap, as opposed to increasing the height. Problems of wind-loading may be reduced in this way. Also, the situation is discussed in Ref. 3 where foundation design did not favour a further increase in height, thus necessitating the use of a Tprofile, or similar design, to increase insertion loss. The aesthetics of a barrier are also an important factor for nearby residents. Whilst a T-profile design is unlikely to improve barrier appearance, it is also unlikely to degrade it. The situation has arisen 5 where the demands of residents to retain a full view of their coastal surroundings has led to the erection of barriers having insufficient height to meet 'target' noise levels. This would seem to be an ideal situation to make use of a Tprofile design. Indeed, if predicted conventional barrier insertion losses are small, then the small increase that may be provided by a T-profile may help to justify the barrier in the first place. This paper commences with a review of previous research on T-profile and associated barrier forms. A numerical model has been developed which enables the insertion loss of these profiles to be calculated for a point source of sound, with a higher degree of accuracy than was previously possible. The effects of ground cover and absorptive treatment of the upper surface of the barrier cap can also be predicted. The remainder of the paper presents calculations using this model. In addition to examining spectra of insertion loss determined at third octave centre frequencies, the insertion losses for a broad-band source representative of A-weighted traffic noise are presented and discussed. The modelling technique is also used to obtain detailed information about the sound field close to the barrier.
2 PREVIOUS STUDIES Previous studies have suggested that a n u m b e r o f factors affect the acoustic efficiency o f a T-profile barrier. These include the width o f the horizontal cap, angling the cap from the centre to create either a Y-profile or an arrowprofile barrier, and absorptive treatment o f the upper surface o f the cap (which also varies the cap depth). The results are detailed in the following sections.
The performance of T-profile and associated noise barriers
271
2,1 Width of the horizontal cap (reflective) May and Osman 1 conducted a series of scale-model experiments covering a large number of barrier designs, including different T-profile barriers. They used mainly a broad-band point source of sound and quoted results in terms of the mean insertion loss over a range of representative receiver points. It was found that adding a horizontal cap (of effectively zero depth) would greatly enhance barrier insertion loss and that the improvement increased as the width of the cap increased. For a fixed height of barrier, the increase in insertion loss ranged from 2-2 dB(A) for a 0-4 m cap width to 6.4 dB(A) for 4.88 m (the cap width being equal to the barrier height in this case). Most of the figures and trends discussed in this section are summarised in Table 1, which also gives details of the geometries involved in the experiments. May and Osman also found that T-profile barriers performed significantly better than rectangular barriers of the same width and height. Also, creating a T-profile barrier with a 0-61m cap width was found to be 1-4 dB(A) more efficient than increasing the original height of the barrier by 0"61 m, the original height being 4.9 m. They also note that a cap width of 0.6 m coincides with the wavelength at 500 Hz. This is the region of the traffic noise spectrum that is thought to contribute most to human perception of road traffic noise. The insertion losses obtained strongly support the case for T-profile barriers, but, unfortunately, other modelling and field experiments, 2"3 whilst generally finding an improvement on creating a T-profile, found the increase in insertion loss to be well short of the magnitudes mentioned above. May and Osman conducted full-scale field measurements on a 4 m high barrier. 2 It was found that the addition of a 0.75 m horizontal cap increased insertion loss by about 1.0-1-5 dB(A) for the first row of houses. Beyond the first row of houses the barrier performance declined considerably. May and Osman's model results predict a figure of around 2-8 dB(A) (Ref. 1, Fig. 6) and they attribute the difference in results to high levels of background noise on site.
2.2 Absorptive treatment and depth of the cap May and Osman, in their scale-model experiments t found, first, that the insertion loss increased as the absorption of the material used to cover the upper surface of the horizontal cap increased and, secondly, that the insertion loss increased as the width of the absorptive cap increased. The rate of increase was greater than that observed for the reflective horizontal cap. However, they also found a rather surprising effect concerning the depth
D. C. Hothersoli. D. H. Cromhre, S. .I’. Chandler-CC~iltle
The perJbrmance of T-profile and associated noise barriers
273
ofthe horizontal cap. With cap widths of0-61 m it was tbund that a T-profile reflectice barrier with effectively zero cap depth pertbrmed 1-6 dBI A) better than one with a 0-2 m cap depth. Although the absorptive T-profile barrier with 0.2 m cap depth performed better than the reflective T-profile barrier of zero cap depth, the difference was only 0'3 dB(A). It was thus concluded that the absorptive coating should be kept as thin as possible. The only other cap treated with the 0-2 m deep absorbing material had a width of 2--*4m and this design gave an improvement in insertion loss of l'7dBiAt over a Tprofile reflective barrier with zero cap depth and the same ~vidth. It was suggested that this improvement would be approximately 1"6 dBIAI greater if the depth of the absorbent cap was reduced to near zero, by analogy with the above observation for the rigid cap. Hajek and Blaney 3 evaluated the performance ofa 5 m high barrier before and after the addition of a 1 m wide absorptive horizontal cap, using both direct field measurement and acoustical scale modelling. The improvement in insertion loss for the T-profile over the vertical wall was considerably less than expected from May and Osman's research. The 1 m absorptive horizontal cap increased insertion loss by about 1"0 dB(A). In this study the effects of the T-profile and conventional vertical carrier were measured simultaneously, thus ensuring the same traffic flow characteristics and meteorological conditions. Hajek and Blaney found that their direct field measurements agreed well with their scale-model experiments--which gave an average increase in insertion loss of 1-3 dB(A) for a 1 m cap ~vidth and 2' 1 dB(A) for a 2 m cap width, (Ref. 3, Figs 9 and 10). Bearing in mind May and Osman's findings, it should be noted that the T-profile absorptive cap had a depth of 0-08 m on the side nearest to the road, increasing gradually to 0.15 m on the residential side. Although not specified, it is assumed that these features have been simulated in the scale model of the site. 2.3 Y- and arrow-profile barriers In their scale-model experiments, t May and Osman also modelled the situation where the T-profile is angled from the centre at 14: to the horizontal to produce Y- and arrow-shaped designs. It was found, however, that the T-profile performed 0"7 dB(A) better than the Y-profile. which in turn performed 1.7 dB(A) better than the arrow-profile for reflective cap widths of 2-44 m and effectively zero depth. 2.4 Other studies Hutchins et al. ~ studied scale models ofT-, Y- and arrow-profiles. Although care was taken to reproduce the geometries used by May and Osman, ~ their
274
D. C. Hothersall, D. H. Crombie, S. N. Chandler-Wilde
models were only one-fifth of the scale, and only one receiver point, relatively close to the barrier, was considered. These limitations tend to hinder comparison. Nevertheless, some interesting spectra are presented, showing small gains in insertion loss in switching from a conventional to a T-profile barrier of the same height. In general, little difference was found between the T-, Y- and arrow-profile designs.
2.5 Acoustic efficiency: T-profile against increased height Hajek and Blaney a compared increasing the height with adding a T-profile of the same width as the increase. They found that increasing the height was more effective at larger distances behind the barrier (30-40 m), but that the T-profile was more efficient at shorter distances (10-20 m). Following this observation, a more extensive comparison was made, using many more receiver positions, from which it was found that the T-profile (for a 1 m cap width) performed more efficiently only at points close to the ground and less than 10m from the barrier. Elsewhere, the increased height barrier performed significantly better. It was thus concluded that it was acoustically more efficient to increase the height of a barrier rather than build a T-profile. This contradicts the result found by May and Osman t for a 0-61 m top width, mentioned earlier.
3 N U M E R I C A L MODEL The numerical model has been described in detail elsewhere.6"T It is a twodimensional model which assumes an infinite, coherent line source of sound, parallel to an infinite noise barrier of uniform cross-section and surface covering along its length. The source configuration is not a practical one, but the results can be related to practical measurements. Specifically, the numerical model predictions of insertion loss, at a particular fequency and receiver position, agree very well with results for a point source of sound, with the point source located on the same line as the line source, and in the same vertical plane, perpendicular to the barrier, as the receiver. To confirm this observation comparisons have been made with insertion loss spectra for point sources of sound obtained from other numerical models and from model experiments in an anechoic chamber and outdoors, for a variety of barrier cross-sections. 6"7 Figure 1 shows the two-dimensional configuration which is considered for the rest of the paper. The barrier is situated on a flat plane of uniform admittance (both the ground and the barrier surfaces are assumed to be locally reacting). The coordinates of the corners of the barrier are input as
The performance of T-profile and associated noise barriers
275
c . p w ~ (w)
fl
Ai
-
i !
..-'
(t)
J"
i
f 1Sin
3
>
Fig. 1. The two-dimensional model, illustrating the concept of "effective height" of the barrier.
data for the model and the surface characteristics in the form of the acoustic admittance of each linear surface segment can be defined separately. For a given source position, using a boundary element method of solution of a developed form of the Kirchoff-Helmholtz integral equation, it is possible to calculate approximately the pressure at points on the surface of the barrier. The points selected are at intervals of )./5 along the segments, where 2 is the wavelength of the source. By proper formulation of the problem (use of an appropriate fundamental solution when developing the boundary element method) only elements over the barrier surface are necessary and none are required in the ground surface. Having obtained the pressure in the surface of the barrier the pressure at a receiver point in or above the ground surface can be obtained from the integral equation. For short wavelengths and large barriers the expense in terms of computing time and storage to solve the problem can be considerable. Results are presented in terms of insertion loss, defined by I L = 20 logto ( p ~ ) d B
where P~ is the acoustic pressure at the receiver for the given source position with the flat ground present and Pb is the pressure when the barrier is introduced. The wave solution obtained by this method results in a complicated pressure field when the effects of diffraction by the barrier are combined with reflection from the ground surface if the source and receiver are above the ground. The source was positioned in the ground surface at 15 m from the centre line for every barrier considered. The pressure was calculated at third octave centre frequencies between 63 Hz and 3.16KHz for nine receiver positions at heights of 0, 1-5 and 3 m above the ground and at distances of 20,
D. C. Hothersall, D. H. Crombie, S. N. Chandler-Wilde
276
50 and I00 m from the centre line of the barrier, on the opposite side to the source. This provided representative results over a practical range of positions. The third octave spectra of insertion loss were examined in detail {see Section 4.1) at one of these positions, i.e. at 50 m from the barrier centre line in the ground surface. Since for this case both source and receiver were in the ground, interference effects due to rays reflected from the ground were eliminated. The shape of the spectra could thus be ascribed only to diffraction around the barrier. The insertion loss for a broad-band noise source, with a spectrum representative of A-weighted road traffic noise was obtained at each of the receiver positions by combining the calculated results at third octave centre frequencies for propagation with and w-ithout the presence of the barrier. The effect of this combination was to smooth out the complicated interference effects occurring in the spectra for receiver positions above the ground. The result was that simple, smooth trends in broad-band insertion loss were observed between the receiver positions (see Section 4.2). Two ground types were considered, "hard ground' with zero admittance and "soft ground' with admittance defined by the empirical relations of Delany and Bazley s derived for fibrous materials with a flow resistance parameter of a = 2 5 0 0 0 0 N s m -~. This value is c o m m o n l y used to characterise grassland surfaces. All the barrier shapes were symmetrical about the centre line and the effects of various absorbent treatments to the upper surfaces were considered. Assuming a typical barrier surface element to consist of rigidly backed layer of depth D, of material of normalised admittance fi~ and propagation constant /% and assuming a large refractive index, the normalised surface admittance is given by fi = - tan
Sur/ace treatment
Flow resistance {or, Nsm -~)
Depth {O. m)
Mean statistical absorption coefficient 1~)
I 2 3 4
Infinity 200 000 20 000 200
00 O"1 O"1 10-0
0-00 0'40 0"80 0"95
The per tbrmance o f T-profile and associated noise barriers
277
1
J o 1a,~
Fig. 2.
am
~
m
1.1~
~
Statistical absorption coefficientas a function of frequency for the four different surface treatments.
k calculated in terms of the flow resistance, a, using the Delany and Bazley expressions) The values of D and a selected for each absorbent treatment are given in Table 2. These can be related to the statistical absorption coefficient 9 which is a commonly quoted parameter for such materials. The absorption coefficients for the four different treatments described in Table 2 are plotted in Fig. 2 as a function of frequency.
4 N U M E R I C A L M O D E L RESULTS It has been established that an important factor affecting the efficiency of a barrier in attenuating sound is its height. For complex profiles use has commonly been made of the concept of "effective height' of an equivalent vertical barrier. The equivalent barrier is obtained by projecting rays from source and receiver to graze the edges of the barrier: the intersection of these rays define its position and height. In the present investigation some adjustment in height of the T- and Y-shaped barriers was carried out so that the comparison was between barriers with virtually the same effective height. As results were calculated simultaneously for nine receiver positions, it was not possible to readjust the height and position of the barrier for each receiver point. As a compromise the height of the barrier was adjusted so that for every profile considered, the ray from the source, grazing the upper corner of the barrier, intersected its centre line at 3.02 m above the surface (see Fig. 1). For the T-shapes this meant that the physical heights were 2-92,
278
D. C. HothersaU, D. H. Crombie, S. N. Chandler-Wilde
2"82 and 2"72 m for cap widths of 1-0, 2'0 and 3"0m, respectively. This produced some variation in path differences from those which would occur if the equivalent barrier was based on a rigorous calculation for each receiver position. However, the errors in the path difference were never greater than 0-03 m, which can be related to a maximum change in insertion loss of about 0"5 dB for the broad-band results reported in the following section. For all the barrier shapes considered in Sections 4.1 and 4.2 the thickness of the wall and cap elements was 0.2 m.
4.1 Spectra of insertion loss Figure 3 shows spectra of insertion loss, calculated at third octave centre frequencies for a vertical wall 3.0 m in height and a T-shaped barrier with reflective surfaces of cap width 3.0 m and height 2.72 m. The receiver position is 50m from the barrier, in the hard ground surface. The spectra show similar trends and close agreement, to within 2 dB. Similar agreement was observed for 1.0 and 2.0 m cap widths when the barrier heights were 2.92 and 2"82 m, respectively. These results confirm that the effect of the width of a hard top can be described in terms of a change in effective height. The spectra in Fig. 4 show the effect of applying different absorbent materials to the upper surface of a T-profile barrier. The dimensions of the barrier are indicated, and the receiver is 50 m from the barrier in the hard
.°
- .......
...
s
10
3.0m ILl (10
_z
so
o . ~u2 m aS
~
T . 4 q ~ I q ~ I OLQRWJ
r
16
m
T
"~
Vlqlr.N. WaLJ.~
I
;
100
am
,
,
i
,
;
~
;
,
i i
4
1,4oo
~, _t~m
FRSOtmNCY (Hz) Fig. 3.
Spectra of insertion loss for a thin wall barrier and T-profile barrier.
The performance of T-profile and associated noise barriers
i
//I
279
....... .'~.
I
;ira
M
!.
",
\
",
T
IL~mA T
I 2Jl~m
mOamlq N J
M
,
e,a~
I
I
U
3
~
I
i
C
~
\\ ", \'\ \\\. \
2
Oamlq 4 i
I
,
i
,
4 ~
I
F
I
I8
FREaUENCV (Hz) Fig. 4.
"
Spectra of insertion loss showing the effect of applying the four different absorbent materials to the upper surface of the cap of a T-profile.
T T" T / -
ml
T41~mmo I
Fig. 5.
,
,
i
Y-II~olo i
,
,
i
,
,
,
i
I
Spectra of insertion loss for T-, Y- and arrow-profile barriers. In each case the cap has absorbent treatment 3.
280
D. C. Hothersall, D. H. Crombie, S. N, Chandler-Wilde
ground surface. The numbers on the plots refer to the surface conditions defined in Table 2 and Fig. 2. As the absorption coefficient of the surface increases the spectra show an increase in insertion loss. For treatments 2 and 3 the frequencies at which the spectra deviate from the results for a perfectly reflecting upper surface are approximately those at which the statistical absorption coefficient reaches 0-6 (see Figs 2 and 4l. The strongly absorbing treatment, 4. produces greatest insertion losses at all frequencies. In Fig. 5 spectra of insertion loss at a receiver position in the ground, 50 m from the centre line of the T-, Y- and arrow-profile constructions, are compared. The length of the upper section in each case is 2"0 m and the upper surt:ace has treatment 3, as defined in Table 1. The slope of the arms of the Yand arrow-profiles is 32 ~. The arrow shape generally produced the lowest insertion losses over all frequencies. The T-profile was more efficient than the Y-profile at high frequencies, but the reverse was true below a b o u t 630 Hz.
4.2 Broad band insertion loss The insertion losses calculated at the nine receiver positions for a broadband point source with a spectrum characteristic of A-weighted road traffic noise are shown in Fig. 6. The abscissa represents the horizontal distance
.... 55 - - . . . . 10
SOFT GROUND
dlr ~ - -
g Z
lS
...............
•
HARD GROUND
O E
RECEIVER
EFFECTIVE HEIGHT = 3.02 m
m
z_
V
HEIGHT h--3 m
2m
m
o.2m A T
=''t"
h=1.5 m
i2.a2m
i
h=O m ~
D
2S It)
90
1~
D, DISTANCE TO RECEIVER (m) Fig. 6. Insertion loss for a broad-band source spectrum with the barrier configuration shown. Hard and soft ground surfaces are considered. D is the horizontal distance of the receiver from the barrier and h is its height above the ground.
The performance of T-profile and associated noise barriers
281
from the centre line of the barrier to the receiver and the height of the receiver above the ground is also shown. The barrier geometry is indicated and all the surfaces of the barrier were reflecting. Results were calculated for both uniform 'hard" and "soft" ground surfaces. There was considerable variation between results at different receiver positions. The most obvious systematic effect which was observed was in the hard ground case where the insertion loss in the ground was between 1-3 dB less than for positions above the ground. For all the barrier shapes considered the results followed similar trends. In order to derive results which provided information about the overall effectiveness of the different barrier shapes, the mean insertion loss was calculated for the six receiver positions above the ground. These results are plotted in Fig. 7 and show the variation with the width of the upper surface of a T-profile form and the effect of absorptive treatment. The mean insertion loss for a conventional, reflective wall barrier with width W = 0-2 m is also indicated, as is that for a similar barrier which tapers to a point over the upper 0"5 m, giving W = 0-0 m. For both the reflecting upper surface and surface treatment 2, the effect of increasing the cap width was negligible. It must be remembered that the physical height of the barrier was adjusted to maintain a constant effective height of 3-02 m for all these results. For
.......
SOFT GROUND
•
---, 1
~
CO to
. . . . . . . .
. . . .
d~- . . . . . . . . . . . . . . .
• 3
-,t 4
HARD GROUND ........................
a...................
Q
• 1 i
T
3.
" ~ - ~W
~_..~,_
*a'AT m
Ikl
1-.M
,tJemb~ Cmlfq I
r
i
................
HEIGHT : 3.02 m
~
¢m=l~:
,tl=~tm~ (:mull 3
4
~mm~nt C~n~I 4
J
i 1
2
3
W , WIDTH OF CAP (m) Fig. 7. Mean insertion loss for a broad-band source spectrum for T-profile barriers as a function of the cap width (W). The effect of the tour different absorptive treatments (detailed in Table 2) are also shown for both hard and soft ground surfaces.
D. C. Hothersall, D. H. Crombie, S. N. Chandler-Wilde
282
~,
S
el •
=, R E F L E C T I N G
THIN SCREEN
- 2
SOFT
O
GROUND
:
~ls
HA~D GROUND
; ........ m-.-.~.
'
"
,
• 4
3 "'"
: ',
. . . . . . .
.-e
-a ~ " ~
~
-eQ
qlO
"-~
~d.~
--,,.
:____ .-
_Jl____
:
I
;
-7'0
I
I
.410
I
-SO
I
;
..40
I
I
-30
I
q
4
. . . . .
_j [
,
[
-20
I
i
-10
I
I
0
i
I
I
10
e , SLOPE O F C A P A R M S
i
2Q
I
I
30
I
I
I
40
I
SO
I
:
(IO
,
;
~,
I
I
II0
;
IO
(DEGREES)
Fig. 8. Mean insertion loss for a broad-band source spectrum plotted as a function of the slope of the cap arms (0). The four different absorptive treatments are also shown for both hard and soft ground cover. Values for a reflecting thin screen are also indicated. treatments 3 and 4 a progressive improvement in insertion loss is observed as the cap width is increased. On comparing the results for hard and soft ground surfaces there is a significant absolute difference in results of around 6.5dB. However, otherwise, the trends for the two sets of results are remarkably similar. In Fig. 8 the mean broad-band insertion loss over the six receiver positions above the ground is plotted as a function of the slope of the cap arms, 0. The total width of the upper surface of the barrier is 2 m. Again, adjustment has been made to the physical height in order to maintain a constant effective barrier height, as described in Section 4.1. As 0 approaches + 90: the results appear to converge to values very close to the value for a reflecting thin screen. F o r a totally reflecting ground surface the arrow-profile performs less efficiently than the T-profile for all values of 0. F o r the Y-profile some marginal improvement in insertion loss is observed when the upper surface is strongly reflective, but not for efficiently absorbing surface treatments. Fewer points were calculated for the case of soft ground but they indicate a similar trend.
4.3 Thin caps As outlined in Section 2 and shown in Table 1, the results of model experiments t provide evidence that a significant improvement in insertion
The performance o f T-profile and associated noise barriers
A
y
m
W
t & TEFFECTIV 0.2
O -I
283
.-p
E HEIGHT = 3.02 m
.
Z
o_ I,ILl
A
•
_ _ ~
-'~.
......
..~
i
Z is Ul
t " 0.2 m ui >
t - 0.01 m --
......... O
~
~
;
t
L
;
•
i
I,
1
I
I
I
[
2
I
I
I
~
--
-dk--
L
~
I
I
r
3
('at), WIDTH OF CAP (m) Fig. 9.
M e a n insertion loss for T-profile barriers plotted as a function of cap width (W) a n d cap d e p t h (t) for a b r o a d - b a n d source spectrum.
loss could be obtained by the application of a horizontal reflecting cap of limited width (about 0"6 m), to the top of a vertical wall barrier. It was also argued that the improvement would be greatest if the depth of the cap was kept to a minimum, or effectively zero. Although this treatment appeared very cost effective, full-scale trials proved disappointing, giving only marginal increases in insertion loss of about 1-0-1"5dB(A) for a 0.019m deep, 0.75 m wide cap 2 and about 1.0dB(A) for a 0-08m deep, 1.0m wide absorptive cap. 3 A series of calculations was carried out to investigate any possible benefits that may be obtained from very thin horizontal caps. Figure 9 shows the mean insertion loss for the six standard, above-ground receiver positions for the source with the broad-band spectrum. In every case the height of the barrier was adjusted to allow for the path difference effect. Insertion loss is plotted against cap width for horizontal caps which were 0-2 m and 0.01 m thick. The results indicated only a marginal increase in insertion loss associated with the reduced cap thickness at these receiver positions. Similarly, the difference found between a conventional wall barrier of width 0-2 m and one of width 0.01 m was negligible. 4.4 The acoustic field close to the barrier
The modelling method enables detailed information about the acoustic field close to the barrier to be obtained. Figure 10 shows the broad-band insertion
284
D. C. Hothersall, D. H. Crombie, S. N. Chandler-14ilde
U
.0
L1
-2
.1 -4
i
.1" ~
;
i
i
i
i
i
i
i
i
i
;
i
i
i
|.. X
2.7
i
.,3 -4 -2" ..4" ,.6" i ,,6 -6" ~
1i 17 17 1? 17 li 1; 1;
Lll
~
(.i) (a)
-1 ol .0 -0 S
-1" -i .i -1" .
-3" .
-2
.
.
-4
,
i
~
i
~
i
"
.
-3
.
o
-4
-8
4
4
i
;
i
"
4"
i
~ ;
;,
;
•
4
£. 1
ZJI
B441WUER ~
=, 2 . N m
2.4 ~U
i
~
I
;
-0.4
r
l
I
J
L
-0.11
,
l
Z
t
~
-4.1
I
,
,
I
~1
i
0
i
t
F
0.1
J
i
J
0,2
•
i
0,3
i
;
'
0.4
0.5
wao'rH(m)
(b) Fig. I0. Insertion loss in the upper region of three barriers. (a) T-profile with cap depth of 0"2 m and absorptive treatment 3 applied to the upper surface; (b) reflective T-profile with cap depth of 0-01 m; (¢) vertical wall.
285
The performance of T-profile and associated noise barriers U -0
U
*0 -0
Ikl
.0 |
,1"-,"
j.. ~,..U
.," .," .,"
~
~ .1" ~ i
~ i
i
i
. , ' . ¢ .2" .2" g
i
i
.i ,2" ,¢ .~" a" ~"
i
U
i
m
i
i
i
i
i i
; i
HIBm,l f f . , 2 J I m
U 8.4
,4,$
WU)TI'I{ ~
(c) Fig. lO--contd.
loss near three barriers (each of height 2"96 m) calculated for the A-weighted traffic noise spectrum between 63Hz and 3"16 kHz. As all the points considered were between 14.0 and 16.0m from the source, the free-field pressure used in calculating P~ (the pressure with the ground present) is roughly constant and therefore, P~ (which is exactly twice the free-field pressure as the source is in the hard ground surface) is roughly constant. Thus, the magnitude of the change in insertion loss from one point to another in Fig. 10 is the same as the change in the sound pressure level, to within a maximum error of I dB over the range of points. An insertion loss of zero dB indicates no change on introducing the barrier and, for the vertical wall (Fig. 10(c)), figures near to this value are observed close to the upper surface. The insertion loss changes to - 3 dB close to the front surface due to reflection. In Fig. 10(b) results are presented for a T-profile barrier with a 0-5 m wide and 0-01 m thick cap. Low values of insertion loss (and hence, high values of pressure) are observed below the cap on the source side. Close to the upper surface a smaller change than might have been expected occurred across the width, from 0 to i dB. When an absorbing cap was used (Fig. 10(a)) the change in insertion loss over the width of the cap was greatly increased (from i to 7 dB) and it is to this that the increased efficiency of absorbing caps can be attributed.
286
D. C. HothersalL D. H. Crombie, S. :V. Chandler-Wilde
It is also possible to study the monofrequency sound field near the barriers using the numerical model. Further study should provide information for optimising the shape and distribution of absorber for maximum efficiency.
5 CONCLUSION Detailed information has been obtained regarding the performance of Tprofile and associated barriers in attenuating sound. The computer model is two-dimensional but, as has been discussed, the results obtained predict with some accuracy the insertion loss for an infinite straight barrier of uniform section with a point source of sound. 6"~ The results from the model are not, in terms of absolute values, applicable to the incoherent line source configuration which would model a traffic flow. However, it is expected that the relative performance of different barrier shapes, as predicted by the model, will be similar to that observed with an incoherent traffic stream.l~ From this study the following conclusions are drawn: (1)
(2)
(3)
(4)
(5)
For reflective T-profile barriers the insertion loss is constant with respect to the cap width provided that the barrier height is adjusted so that the effective barrier height remains approximately constant. When comparing results with those of May and Osman L it must be remembered that the barrier configurations that they considered had a constant height, and thus differing effective heights. The introduction of a weakly absorbing upper surface to a T-profile barrier produced no significant improvement in insertion loss, whereas more strongly absorbent surfaces gave progressively greater increases. Also, the effect increased with the width of the cap. The portion of the insertion loss spectrum in which increases occur can be related to the frequency dependence of the absorption coefficient of the cap material. When the T-profile was modified to an arrow-profile, a significant reduction in insertion loss was observed. This trend also occurred for the Y-profile when the upper surfaces were strongly absorbing (Fig. 8). Comparison of the results obtained in this paper with previously reported results for solid barriers in the form of 'wedges' and 'flattopped mounds '6 indicate that, for the same broad-band source over hard ground, the arrow- and T-profiles perform significantly better than their counterparts with similar upper sections, the difference being around 1-2 dB. For a reflective T-profile barrier, changing the depth of the cap from 0.2 m to 0-0! m had a negligible effect upon the insertion loss, at the
The performance of T-profile and associated noise barriers
287
receiver points monitored, using a broad-band point source of sound. {6) Detailed investigation of the acoustic field around wall and T-profile barriers indicated the existence of an extended area o f high acoustic pressure along the upper surface of the reflective T-profile, and a very much lower pressure beneath the T on the side opposite the source, than was found at the same positions for the wall barrier.
REFERENCES 1. May, D. N. & Osman, M. M., Highway noise barriers: New shapes. J. Sound Vibration, 71 (1980) 73-101. 2. May, D. N. & Osman, M. M., The performance of sound absorptive, reflective and T-profile noise barriers in Toronto. J. Sound Vibration, 71 (1980) 65-71. 3. Hajek, J. J. & Blaney, C. T., Evaluation ofT-profile noise barriers. Transportation Research Record, 983 (1984) 8-17. 4. Hutchins, D. A., Jones, H. W. & Russell. L. T., Model studies of barrier performance in the presence of ground surfaces. Part II--Different shapes. J. Acoustical Soc. America, 75(6) (1984) 1817-26. 5. Knauer, H. S., Noise barriers adjacent to 1-95 in Philadelphia. Transportation Research Record, 740 (1980) 13-21. 6. Hothersall, D. C., Chandler-Wilde, S. N. & Hajmirzae, N. M., Efficiency of single noise barriers. J. Sound Vibration (in press). 7. Chandler-Wilde, S. N. & Hothersall, D. C., The boundary integral equation method in outdoor sound propagation. Proc. Inst. Acoustics, 9 (1987) 37-44. 8. Delany, M. E. & Bazley, E. N., Acoustical properties of fibrous absorbent materials. App. Acoustics, 3 (1970) 105-16. 9. Bies, D. A. & Hansen, G. H., Flow resistance information for acoustical design. App. Acoustics, 13 (1980) 357-91.
The Performance of T-Profile and Associated Noise Barriers
D. C. Hothersall, D. H. C r o m b i e & S. N. Chandler-Wilde Department of Civil Engineering, University of Bradford. Bradford, West Yorkshire BD7 IDP, UK Received I May 1990: revised version received 11 July 1990: accepted 25 July 19901
ABSTRACT .4 ret'iew ts presented on t/re results of pret'ious research, using experimental modelling and field measurement, qf the insertion loss produced hy T-. Y- and arrow-profile noise barriers. ,4 numerical model is discussed which enables tire ware field in the region of a barrier with a complex profile and surface cot'er to he determined. An brtegral equationJormulation of the problem is soh'ed us#rg the boundao" element approach. Spectra of insertion loss Jor T-. Y- and arrow-profile barriers calculated using ttre model are discussed. Results for the mean insertion loss orer a range of receirer positions for a broad-hand source are also presented. It is concluded that for harriers with reflecting surfaces the changes in #Tsertion loss for different profiles can be described using the concept of path difference. The intro~hwtion of absorbing upper surfaces produces a significant increase in insertion loss. br most conditions the Y- and arrow-profiles perform less efficiently than the T-profile. Tire marked increase b7 barrier efficiency when a rerv thin narrow cap is added to a t'ertical wall which has been reported by other" workers was not obsert'ed.
i INTRODUCTION The addition of a horizontal cap to a conventional barrier, thus creating a ~T-profile', has been the subject of a limited number of previous studies, t -'~ encompassing either acoustical scale modelling, direct field measurements, or both. In general, it has been established that a design o f this kind is likely 269 Applied Acoustics 0003-682X/9l/$03-50 ~ 199l Elsevier Science Publishers ktd, England. Printed in Great Britain
270
D. C. Hothersall, D. H. Crombie, S. N. Chandler-Wilde
to achieve a small gain in insertion loss over a conventional barrier of the same height. Although the gains reported from direct field measurement are modest, typically around 1"0-1"5 dB(A) for cap widths up to 1 m, there are a number of factors which suggest that T-profile barriers warrant further study. The terms "cap width" and "cap depth', used throughout this report, are illustrated in Fig. 1. Certain design factors seem to favour the addition of a horizontal cap, as opposed to increasing the height. Problems of wind-loading may be reduced in this way. Also, the situation is discussed in Ref. 3 where foundation design did not favour a further increase in height, thus necessitating the use of a Tprofile, or similar design, to increase insertion loss. The aesthetics of a barrier are also an important factor for nearby residents. Whilst a T-profile design is unlikely to improve barrier appearance, it is also unlikely to degrade it. The situation has arisen 5 where the demands of residents to retain a full view of their coastal surroundings has led to the erection of barriers having insufficient height to meet 'target' noise levels. This would seem to be an ideal situation to make use of a Tprofile design. Indeed, if predicted conventional barrier insertion losses are small, then the small increase that may be provided by a T-profile may help to justify the barrier in the first place. This paper commences with a review of previous research on T-profile and associated barrier forms. A numerical model has been developed which enables the insertion loss of these profiles to be calculated for a point source of sound, with a higher degree of accuracy than was previously possible. The effects of ground cover and absorptive treatment of the upper surface of the barrier cap can also be predicted. The remainder of the paper presents calculations using this model. In addition to examining spectra of insertion loss determined at third octave centre frequencies, the insertion losses for a broad-band source representative of A-weighted traffic noise are presented and discussed. The modelling technique is also used to obtain detailed information about the sound field close to the barrier.
2 PREVIOUS STUDIES Previous studies have suggested that a n u m b e r o f factors affect the acoustic efficiency o f a T-profile barrier. These include the width o f the horizontal cap, angling the cap from the centre to create either a Y-profile or an arrowprofile barrier, and absorptive treatment o f the upper surface o f the cap (which also varies the cap depth). The results are detailed in the following sections.
The performance of T-profile and associated noise barriers
271
2,1 Width of the horizontal cap (reflective) May and Osman 1 conducted a series of scale-model experiments covering a large number of barrier designs, including different T-profile barriers. They used mainly a broad-band point source of sound and quoted results in terms of the mean insertion loss over a range of representative receiver points. It was found that adding a horizontal cap (of effectively zero depth) would greatly enhance barrier insertion loss and that the improvement increased as the width of the cap increased. For a fixed height of barrier, the increase in insertion loss ranged from 2-2 dB(A) for a 0-4 m cap width to 6.4 dB(A) for 4.88 m (the cap width being equal to the barrier height in this case). Most of the figures and trends discussed in this section are summarised in Table 1, which also gives details of the geometries involved in the experiments. May and Osman also found that T-profile barriers performed significantly better than rectangular barriers of the same width and height. Also, creating a T-profile barrier with a 0-61m cap width was found to be 1-4 dB(A) more efficient than increasing the original height of the barrier by 0"61 m, the original height being 4.9 m. They also note that a cap width of 0.6 m coincides with the wavelength at 500 Hz. This is the region of the traffic noise spectrum that is thought to contribute most to human perception of road traffic noise. The insertion losses obtained strongly support the case for T-profile barriers, but, unfortunately, other modelling and field experiments, 2"3 whilst generally finding an improvement on creating a T-profile, found the increase in insertion loss to be well short of the magnitudes mentioned above. May and Osman conducted full-scale field measurements on a 4 m high barrier. 2 It was found that the addition of a 0.75 m horizontal cap increased insertion loss by about 1.0-1-5 dB(A) for the first row of houses. Beyond the first row of houses the barrier performance declined considerably. May and Osman's model results predict a figure of around 2-8 dB(A) (Ref. 1, Fig. 6) and they attribute the difference in results to high levels of background noise on site.
2.2 Absorptive treatment and depth of the cap May and Osman, in their scale-model experiments t found, first, that the insertion loss increased as the absorption of the material used to cover the upper surface of the horizontal cap increased and, secondly, that the insertion loss increased as the width of the absorptive cap increased. The rate of increase was greater than that observed for the reflective horizontal cap. However, they also found a rather surprising effect concerning the depth
D. C. Hothersoli. D. H. Cromhre, S. .I’. Chandler-CC~iltle
The perJbrmance of T-profile and associated noise barriers
273
ofthe horizontal cap. With cap widths of0-61 m it was tbund that a T-profile reflectice barrier with effectively zero cap depth pertbrmed 1-6 dBI A) better than one with a 0-2 m cap depth. Although the absorptive T-profile barrier with 0.2 m cap depth performed better than the reflective T-profile barrier of zero cap depth, the difference was only 0'3 dB(A). It was thus concluded that the absorptive coating should be kept as thin as possible. The only other cap treated with the 0-2 m deep absorbing material had a width of 2--*4m and this design gave an improvement in insertion loss of l'7dBiAt over a Tprofile reflective barrier with zero cap depth and the same ~vidth. It was suggested that this improvement would be approximately 1"6 dBIAI greater if the depth of the absorbent cap was reduced to near zero, by analogy with the above observation for the rigid cap. Hajek and Blaney 3 evaluated the performance ofa 5 m high barrier before and after the addition of a 1 m wide absorptive horizontal cap, using both direct field measurement and acoustical scale modelling. The improvement in insertion loss for the T-profile over the vertical wall was considerably less than expected from May and Osman's research. The 1 m absorptive horizontal cap increased insertion loss by about 1"0 dB(A). In this study the effects of the T-profile and conventional vertical carrier were measured simultaneously, thus ensuring the same traffic flow characteristics and meteorological conditions. Hajek and Blaney found that their direct field measurements agreed well with their scale-model experiments--which gave an average increase in insertion loss of 1-3 dB(A) for a 1 m cap ~vidth and 2' 1 dB(A) for a 2 m cap width, (Ref. 3, Figs 9 and 10). Bearing in mind May and Osman's findings, it should be noted that the T-profile absorptive cap had a depth of 0-08 m on the side nearest to the road, increasing gradually to 0.15 m on the residential side. Although not specified, it is assumed that these features have been simulated in the scale model of the site. 2.3 Y- and arrow-profile barriers In their scale-model experiments, t May and Osman also modelled the situation where the T-profile is angled from the centre at 14: to the horizontal to produce Y- and arrow-shaped designs. It was found, however, that the T-profile performed 0"7 dB(A) better than the Y-profile. which in turn performed 1.7 dB(A) better than the arrow-profile for reflective cap widths of 2-44 m and effectively zero depth. 2.4 Other studies Hutchins et al. ~ studied scale models ofT-, Y- and arrow-profiles. Although care was taken to reproduce the geometries used by May and Osman, ~ their
274
D. C. Hothersall, D. H. Crombie, S. N. Chandler-Wilde
models were only one-fifth of the scale, and only one receiver point, relatively close to the barrier, was considered. These limitations tend to hinder comparison. Nevertheless, some interesting spectra are presented, showing small gains in insertion loss in switching from a conventional to a T-profile barrier of the same height. In general, little difference was found between the T-, Y- and arrow-profile designs.
2.5 Acoustic efficiency: T-profile against increased height Hajek and Blaney a compared increasing the height with adding a T-profile of the same width as the increase. They found that increasing the height was more effective at larger distances behind the barrier (30-40 m), but that the T-profile was more efficient at shorter distances (10-20 m). Following this observation, a more extensive comparison was made, using many more receiver positions, from which it was found that the T-profile (for a 1 m cap width) performed more efficiently only at points close to the ground and less than 10m from the barrier. Elsewhere, the increased height barrier performed significantly better. It was thus concluded that it was acoustically more efficient to increase the height of a barrier rather than build a T-profile. This contradicts the result found by May and Osman t for a 0-61 m top width, mentioned earlier.
3 N U M E R I C A L MODEL The numerical model has been described in detail elsewhere.6"T It is a twodimensional model which assumes an infinite, coherent line source of sound, parallel to an infinite noise barrier of uniform cross-section and surface covering along its length. The source configuration is not a practical one, but the results can be related to practical measurements. Specifically, the numerical model predictions of insertion loss, at a particular fequency and receiver position, agree very well with results for a point source of sound, with the point source located on the same line as the line source, and in the same vertical plane, perpendicular to the barrier, as the receiver. To confirm this observation comparisons have been made with insertion loss spectra for point sources of sound obtained from other numerical models and from model experiments in an anechoic chamber and outdoors, for a variety of barrier cross-sections. 6"7 Figure 1 shows the two-dimensional configuration which is considered for the rest of the paper. The barrier is situated on a flat plane of uniform admittance (both the ground and the barrier surfaces are assumed to be locally reacting). The coordinates of the corners of the barrier are input as
The performance of T-profile and associated noise barriers
275
c . p w ~ (w)
fl
Ai
-
i !
..-'
(t)
J"
i
f 1Sin
3
>
Fig. 1. The two-dimensional model, illustrating the concept of "effective height" of the barrier.
data for the model and the surface characteristics in the form of the acoustic admittance of each linear surface segment can be defined separately. For a given source position, using a boundary element method of solution of a developed form of the Kirchoff-Helmholtz integral equation, it is possible to calculate approximately the pressure at points on the surface of the barrier. The points selected are at intervals of )./5 along the segments, where 2 is the wavelength of the source. By proper formulation of the problem (use of an appropriate fundamental solution when developing the boundary element method) only elements over the barrier surface are necessary and none are required in the ground surface. Having obtained the pressure in the surface of the barrier the pressure at a receiver point in or above the ground surface can be obtained from the integral equation. For short wavelengths and large barriers the expense in terms of computing time and storage to solve the problem can be considerable. Results are presented in terms of insertion loss, defined by I L = 20 logto ( p ~ ) d B
where P~ is the acoustic pressure at the receiver for the given source position with the flat ground present and Pb is the pressure when the barrier is introduced. The wave solution obtained by this method results in a complicated pressure field when the effects of diffraction by the barrier are combined with reflection from the ground surface if the source and receiver are above the ground. The source was positioned in the ground surface at 15 m from the centre line for every barrier considered. The pressure was calculated at third octave centre frequencies between 63 Hz and 3.16KHz for nine receiver positions at heights of 0, 1-5 and 3 m above the ground and at distances of 20,
D. C. Hothersall, D. H. Crombie, S. N. Chandler-Wilde
276
50 and I00 m from the centre line of the barrier, on the opposite side to the source. This provided representative results over a practical range of positions. The third octave spectra of insertion loss were examined in detail {see Section 4.1) at one of these positions, i.e. at 50 m from the barrier centre line in the ground surface. Since for this case both source and receiver were in the ground, interference effects due to rays reflected from the ground were eliminated. The shape of the spectra could thus be ascribed only to diffraction around the barrier. The insertion loss for a broad-band noise source, with a spectrum representative of A-weighted road traffic noise was obtained at each of the receiver positions by combining the calculated results at third octave centre frequencies for propagation with and w-ithout the presence of the barrier. The effect of this combination was to smooth out the complicated interference effects occurring in the spectra for receiver positions above the ground. The result was that simple, smooth trends in broad-band insertion loss were observed between the receiver positions (see Section 4.2). Two ground types were considered, "hard ground' with zero admittance and "soft ground' with admittance defined by the empirical relations of Delany and Bazley s derived for fibrous materials with a flow resistance parameter of a = 2 5 0 0 0 0 N s m -~. This value is c o m m o n l y used to characterise grassland surfaces. All the barrier shapes were symmetrical about the centre line and the effects of various absorbent treatments to the upper surfaces were considered. Assuming a typical barrier surface element to consist of rigidly backed layer of depth D, of material of normalised admittance fi~ and propagation constant /% and assuming a large refractive index, the normalised surface admittance is given by fi = - tan
Sur/ace treatment
Flow resistance {or, Nsm -~)
Depth {O. m)
Mean statistical absorption coefficient 1~)
I 2 3 4
Infinity 200 000 20 000 200
00 O"1 O"1 10-0
0-00 0'40 0"80 0"95
The per tbrmance o f T-profile and associated noise barriers
277
1
J o 1a,~
Fig. 2.
am
~
m
1.1~
~
Statistical absorption coefficientas a function of frequency for the four different surface treatments.
k calculated in terms of the flow resistance, a, using the Delany and Bazley expressions) The values of D and a selected for each absorbent treatment are given in Table 2. These can be related to the statistical absorption coefficient 9 which is a commonly quoted parameter for such materials. The absorption coefficients for the four different treatments described in Table 2 are plotted in Fig. 2 as a function of frequency.
4 N U M E R I C A L M O D E L RESULTS It has been established that an important factor affecting the efficiency of a barrier in attenuating sound is its height. For complex profiles use has commonly been made of the concept of "effective height' of an equivalent vertical barrier. The equivalent barrier is obtained by projecting rays from source and receiver to graze the edges of the barrier: the intersection of these rays define its position and height. In the present investigation some adjustment in height of the T- and Y-shaped barriers was carried out so that the comparison was between barriers with virtually the same effective height. As results were calculated simultaneously for nine receiver positions, it was not possible to readjust the height and position of the barrier for each receiver point. As a compromise the height of the barrier was adjusted so that for every profile considered, the ray from the source, grazing the upper corner of the barrier, intersected its centre line at 3.02 m above the surface (see Fig. 1). For the T-shapes this meant that the physical heights were 2-92,
278
D. C. HothersaU, D. H. Crombie, S. N. Chandler-Wilde
2"82 and 2"72 m for cap widths of 1-0, 2'0 and 3"0m, respectively. This produced some variation in path differences from those which would occur if the equivalent barrier was based on a rigorous calculation for each receiver position. However, the errors in the path difference were never greater than 0-03 m, which can be related to a maximum change in insertion loss of about 0"5 dB for the broad-band results reported in the following section. For all the barrier shapes considered in Sections 4.1 and 4.2 the thickness of the wall and cap elements was 0.2 m.
4.1 Spectra of insertion loss Figure 3 shows spectra of insertion loss, calculated at third octave centre frequencies for a vertical wall 3.0 m in height and a T-shaped barrier with reflective surfaces of cap width 3.0 m and height 2.72 m. The receiver position is 50m from the barrier, in the hard ground surface. The spectra show similar trends and close agreement, to within 2 dB. Similar agreement was observed for 1.0 and 2.0 m cap widths when the barrier heights were 2.92 and 2"82 m, respectively. These results confirm that the effect of the width of a hard top can be described in terms of a change in effective height. The spectra in Fig. 4 show the effect of applying different absorbent materials to the upper surface of a T-profile barrier. The dimensions of the barrier are indicated, and the receiver is 50 m from the barrier in the hard
.°
- .......
...
s
10
3.0m ILl (10
_z
so
o . ~u2 m aS
~
T . 4 q ~ I q ~ I OLQRWJ
r
16
m
T
"~
Vlqlr.N. WaLJ.~
I
;
100
am
,
,
i
,
;
~
;
,
i i
4
1,4oo
~, _t~m
FRSOtmNCY (Hz) Fig. 3.
Spectra of insertion loss for a thin wall barrier and T-profile barrier.
The performance of T-profile and associated noise barriers
i
//I
279
....... .'~.
I
;ira
M
!.
",
\
",
T
IL~mA T
I 2Jl~m
mOamlq N J
M
,
e,a~
I
I
U
3
~
I
i
C
~
\\ ", \'\ \\\. \
2
Oamlq 4 i
I
,
i
,
4 ~
I
F
I
I8
FREaUENCV (Hz) Fig. 4.
"
Spectra of insertion loss showing the effect of applying the four different absorbent materials to the upper surface of the cap of a T-profile.
T T" T / -
ml
T41~mmo I
Fig. 5.
,
,
i
Y-II~olo i
,
,
i
,
,
,
i
I
Spectra of insertion loss for T-, Y- and arrow-profile barriers. In each case the cap has absorbent treatment 3.
280
D. C. Hothersall, D. H. Crombie, S. N, Chandler-Wilde
ground surface. The numbers on the plots refer to the surface conditions defined in Table 2 and Fig. 2. As the absorption coefficient of the surface increases the spectra show an increase in insertion loss. For treatments 2 and 3 the frequencies at which the spectra deviate from the results for a perfectly reflecting upper surface are approximately those at which the statistical absorption coefficient reaches 0-6 (see Figs 2 and 4l. The strongly absorbing treatment, 4. produces greatest insertion losses at all frequencies. In Fig. 5 spectra of insertion loss at a receiver position in the ground, 50 m from the centre line of the T-, Y- and arrow-profile constructions, are compared. The length of the upper section in each case is 2"0 m and the upper surt:ace has treatment 3, as defined in Table 1. The slope of the arms of the Yand arrow-profiles is 32 ~. The arrow shape generally produced the lowest insertion losses over all frequencies. The T-profile was more efficient than the Y-profile at high frequencies, but the reverse was true below a b o u t 630 Hz.
4.2 Broad band insertion loss The insertion losses calculated at the nine receiver positions for a broadband point source with a spectrum characteristic of A-weighted road traffic noise are shown in Fig. 6. The abscissa represents the horizontal distance
.... 55 - - . . . . 10
SOFT GROUND
dlr ~ - -
g Z
lS
...............
•
HARD GROUND
O E
RECEIVER
EFFECTIVE HEIGHT = 3.02 m
m
z_
V
HEIGHT h--3 m
2m
m
o.2m A T
=''t"
h=1.5 m
i2.a2m
i
h=O m ~
D
2S It)
90
1~
D, DISTANCE TO RECEIVER (m) Fig. 6. Insertion loss for a broad-band source spectrum with the barrier configuration shown. Hard and soft ground surfaces are considered. D is the horizontal distance of the receiver from the barrier and h is its height above the ground.
The performance of T-profile and associated noise barriers
281
from the centre line of the barrier to the receiver and the height of the receiver above the ground is also shown. The barrier geometry is indicated and all the surfaces of the barrier were reflecting. Results were calculated for both uniform 'hard" and "soft" ground surfaces. There was considerable variation between results at different receiver positions. The most obvious systematic effect which was observed was in the hard ground case where the insertion loss in the ground was between 1-3 dB less than for positions above the ground. For all the barrier shapes considered the results followed similar trends. In order to derive results which provided information about the overall effectiveness of the different barrier shapes, the mean insertion loss was calculated for the six receiver positions above the ground. These results are plotted in Fig. 7 and show the variation with the width of the upper surface of a T-profile form and the effect of absorptive treatment. The mean insertion loss for a conventional, reflective wall barrier with width W = 0-2 m is also indicated, as is that for a similar barrier which tapers to a point over the upper 0"5 m, giving W = 0-0 m. For both the reflecting upper surface and surface treatment 2, the effect of increasing the cap width was negligible. It must be remembered that the physical height of the barrier was adjusted to maintain a constant effective height of 3-02 m for all these results. For
.......
SOFT GROUND
•
---, 1
~
CO to
. . . . . . . .
. . . .
d~- . . . . . . . . . . . . . . .
• 3
-,t 4
HARD GROUND ........................
a...................
Q
• 1 i
T
3.
" ~ - ~W
~_..~,_
*a'AT m
Ikl
1-.M
,tJemb~ Cmlfq I
r
i
................
HEIGHT : 3.02 m
~
¢m=l~:
,tl=~tm~ (:mull 3
4
~mm~nt C~n~I 4
J
i 1
2
3
W , WIDTH OF CAP (m) Fig. 7. Mean insertion loss for a broad-band source spectrum for T-profile barriers as a function of the cap width (W). The effect of the tour different absorptive treatments (detailed in Table 2) are also shown for both hard and soft ground surfaces.
D. C. Hothersall, D. H. Crombie, S. N. Chandler-Wilde
282
~,
S
el •
=, R E F L E C T I N G
THIN SCREEN
- 2
SOFT
O
GROUND
:
~ls
HA~D GROUND
; ........ m-.-.~.
'
"
,
• 4
3 "'"
: ',
. . . . . . .
.-e
-a ~ " ~
~
-eQ
qlO
"-~
~d.~
--,,.
:____ .-
_Jl____
:
I
;
-7'0
I
I
.410
I
-SO
I
;
..40
I
I
-30
I
q
4
. . . . .
_j [
,
[
-20
I
i
-10
I
I
0
i
I
I
10
e , SLOPE O F C A P A R M S
i
2Q
I
I
30
I
I
I
40
I
SO
I
:
(IO
,
;
~,
I
I
II0
;
IO
(DEGREES)
Fig. 8. Mean insertion loss for a broad-band source spectrum plotted as a function of the slope of the cap arms (0). The four different absorptive treatments are also shown for both hard and soft ground cover. Values for a reflecting thin screen are also indicated. treatments 3 and 4 a progressive improvement in insertion loss is observed as the cap width is increased. On comparing the results for hard and soft ground surfaces there is a significant absolute difference in results of around 6.5dB. However, otherwise, the trends for the two sets of results are remarkably similar. In Fig. 8 the mean broad-band insertion loss over the six receiver positions above the ground is plotted as a function of the slope of the cap arms, 0. The total width of the upper surface of the barrier is 2 m. Again, adjustment has been made to the physical height in order to maintain a constant effective barrier height, as described in Section 4.1. As 0 approaches + 90: the results appear to converge to values very close to the value for a reflecting thin screen. F o r a totally reflecting ground surface the arrow-profile performs less efficiently than the T-profile for all values of 0. F o r the Y-profile some marginal improvement in insertion loss is observed when the upper surface is strongly reflective, but not for efficiently absorbing surface treatments. Fewer points were calculated for the case of soft ground but they indicate a similar trend.
4.3 Thin caps As outlined in Section 2 and shown in Table 1, the results of model experiments t provide evidence that a significant improvement in insertion
The performance o f T-profile and associated noise barriers
A
y
m
W
t & TEFFECTIV 0.2
O -I
283
.-p
E HEIGHT = 3.02 m
.
Z
o_ I,ILl
A
•
_ _ ~
-'~.
......
..~
i
Z is Ul
t " 0.2 m ui >
t - 0.01 m --
......... O
~
~
;
t
L
;
•
i
I,
1
I
I
I
[
2
I
I
I
~
--
-dk--
L
~
I
I
r
3
('at), WIDTH OF CAP (m) Fig. 9.
M e a n insertion loss for T-profile barriers plotted as a function of cap width (W) a n d cap d e p t h (t) for a b r o a d - b a n d source spectrum.
loss could be obtained by the application of a horizontal reflecting cap of limited width (about 0"6 m), to the top of a vertical wall barrier. It was also argued that the improvement would be greatest if the depth of the cap was kept to a minimum, or effectively zero. Although this treatment appeared very cost effective, full-scale trials proved disappointing, giving only marginal increases in insertion loss of about 1-0-1"5dB(A) for a 0.019m deep, 0.75 m wide cap 2 and about 1.0dB(A) for a 0-08m deep, 1.0m wide absorptive cap. 3 A series of calculations was carried out to investigate any possible benefits that may be obtained from very thin horizontal caps. Figure 9 shows the mean insertion loss for the six standard, above-ground receiver positions for the source with the broad-band spectrum. In every case the height of the barrier was adjusted to allow for the path difference effect. Insertion loss is plotted against cap width for horizontal caps which were 0-2 m and 0.01 m thick. The results indicated only a marginal increase in insertion loss associated with the reduced cap thickness at these receiver positions. Similarly, the difference found between a conventional wall barrier of width 0-2 m and one of width 0.01 m was negligible. 4.4 The acoustic field close to the barrier
The modelling method enables detailed information about the acoustic field close to the barrier to be obtained. Figure 10 shows the broad-band insertion
284
D. C. Hothersall, D. H. Crombie, S. N. Chandler-14ilde
U
.0
L1
-2
.1 -4
i
.1" ~
;
i
i
i
i
i
i
i
i
i
;
i
i
i
|.. X
2.7
i
.,3 -4 -2" ..4" ,.6" i ,,6 -6" ~
1i 17 17 1? 17 li 1; 1;
Lll
~
(.i) (a)
-1 ol .0 -0 S
-1" -i .i -1" .
-3" .
-2
.
.
-4
,
i
~
i
~
i
"
.
-3
.
o
-4
-8
4
4
i
;
i
"
4"
i
~ ;
;,
;
•
4
£. 1
ZJI
B441WUER ~
=, 2 . N m
2.4 ~U
i
~
I
;
-0.4
r
l
I
J
L
-0.11
,
l
Z
t
~
-4.1
I
,
,
I
~1
i
0
i
t
F
0.1
J
i
J
0,2
•
i
0,3
i
;
'
0.4
0.5
wao'rH(m)
(b) Fig. I0. Insertion loss in the upper region of three barriers. (a) T-profile with cap depth of 0"2 m and absorptive treatment 3 applied to the upper surface; (b) reflective T-profile with cap depth of 0-01 m; (¢) vertical wall.
285
The performance of T-profile and associated noise barriers U -0
U
*0 -0
Ikl
.0 |
,1"-,"
j.. ~,..U
.," .," .,"
~
~ .1" ~ i
~ i
i
i
. , ' . ¢ .2" .2" g
i
i
.i ,2" ,¢ .~" a" ~"
i
U
i
m
i
i
i
i
i i
; i
HIBm,l f f . , 2 J I m
U 8.4
,4,$
WU)TI'I{ ~
(c) Fig. lO--contd.
loss near three barriers (each of height 2"96 m) calculated for the A-weighted traffic noise spectrum between 63Hz and 3"16 kHz. As all the points considered were between 14.0 and 16.0m from the source, the free-field pressure used in calculating P~ (the pressure with the ground present) is roughly constant and therefore, P~ (which is exactly twice the free-field pressure as the source is in the hard ground surface) is roughly constant. Thus, the magnitude of the change in insertion loss from one point to another in Fig. 10 is the same as the change in the sound pressure level, to within a maximum error of I dB over the range of points. An insertion loss of zero dB indicates no change on introducing the barrier and, for the vertical wall (Fig. 10(c)), figures near to this value are observed close to the upper surface. The insertion loss changes to - 3 dB close to the front surface due to reflection. In Fig. 10(b) results are presented for a T-profile barrier with a 0-5 m wide and 0-01 m thick cap. Low values of insertion loss (and hence, high values of pressure) are observed below the cap on the source side. Close to the upper surface a smaller change than might have been expected occurred across the width, from 0 to i dB. When an absorbing cap was used (Fig. 10(a)) the change in insertion loss over the width of the cap was greatly increased (from i to 7 dB) and it is to this that the increased efficiency of absorbing caps can be attributed.
286
D. C. HothersalL D. H. Crombie, S. :V. Chandler-Wilde
It is also possible to study the monofrequency sound field near the barriers using the numerical model. Further study should provide information for optimising the shape and distribution of absorber for maximum efficiency.
5 CONCLUSION Detailed information has been obtained regarding the performance of Tprofile and associated barriers in attenuating sound. The computer model is two-dimensional but, as has been discussed, the results obtained predict with some accuracy the insertion loss for an infinite straight barrier of uniform section with a point source of sound. 6"~ The results from the model are not, in terms of absolute values, applicable to the incoherent line source configuration which would model a traffic flow. However, it is expected that the relative performance of different barrier shapes, as predicted by the model, will be similar to that observed with an incoherent traffic stream.l~ From this study the following conclusions are drawn: (1)
(2)
(3)
(4)
(5)
For reflective T-profile barriers the insertion loss is constant with respect to the cap width provided that the barrier height is adjusted so that the effective barrier height remains approximately constant. When comparing results with those of May and Osman L it must be remembered that the barrier configurations that they considered had a constant height, and thus differing effective heights. The introduction of a weakly absorbing upper surface to a T-profile barrier produced no significant improvement in insertion loss, whereas more strongly absorbent surfaces gave progressively greater increases. Also, the effect increased with the width of the cap. The portion of the insertion loss spectrum in which increases occur can be related to the frequency dependence of the absorption coefficient of the cap material. When the T-profile was modified to an arrow-profile, a significant reduction in insertion loss was observed. This trend also occurred for the Y-profile when the upper surfaces were strongly absorbing (Fig. 8). Comparison of the results obtained in this paper with previously reported results for solid barriers in the form of 'wedges' and 'flattopped mounds '6 indicate that, for the same broad-band source over hard ground, the arrow- and T-profiles perform significantly better than their counterparts with similar upper sections, the difference being around 1-2 dB. For a reflective T-profile barrier, changing the depth of the cap from 0.2 m to 0-0! m had a negligible effect upon the insertion loss, at the
The performance of T-profile and associated noise barriers
287
receiver points monitored, using a broad-band point source of sound. {6) Detailed investigation of the acoustic field around wall and T-profile barriers indicated the existence of an extended area o f high acoustic pressure along the upper surface of the reflective T-profile, and a very much lower pressure beneath the T on the side opposite the source, than was found at the same positions for the wall barrier.
REFERENCES 1. May, D. N. & Osman, M. M., Highway noise barriers: New shapes. J. Sound Vibration, 71 (1980) 73-101. 2. May, D. N. & Osman, M. M., The performance of sound absorptive, reflective and T-profile noise barriers in Toronto. J. Sound Vibration, 71 (1980) 65-71. 3. Hajek, J. J. & Blaney, C. T., Evaluation ofT-profile noise barriers. Transportation Research Record, 983 (1984) 8-17. 4. Hutchins, D. A., Jones, H. W. & Russell. L. T., Model studies of barrier performance in the presence of ground surfaces. Part II--Different shapes. J. Acoustical Soc. America, 75(6) (1984) 1817-26. 5. Knauer, H. S., Noise barriers adjacent to 1-95 in Philadelphia. Transportation Research Record, 740 (1980) 13-21. 6. Hothersall, D. C., Chandler-Wilde, S. N. & Hajmirzae, N. M., Efficiency of single noise barriers. J. Sound Vibration (in press). 7. Chandler-Wilde, S. N. & Hothersall, D. C., The boundary integral equation method in outdoor sound propagation. Proc. Inst. Acoustics, 9 (1987) 37-44. 8. Delany, M. E. & Bazley, E. N., Acoustical properties of fibrous absorbent materials. App. Acoustics, 3 (1970) 105-16. 9. Bies, D. A. & Hansen, G. H., Flow resistance information for acoustical design. App. Acoustics, 13 (1980) 357-91.