Median barriers as a portion of a divided highway are provided to minimize the cross-median crashes. Moreover, median barriers similar to roadside noise barriers could protect people from transportation noise. Thus, there is a need to investigate various median barrier models to identify changes of insertion loss over a simple rigid barrier. In order to estimate the acoustical influence of median barrier's profile in the shadow zone, different median barrier models are presented and their insertion losses are calculated over a frequency range from 50 to 4000 Hz using a two-dimensional boundary element method. The present investigation has clearly revealed that among the profiled median barriers, T-shape, Y-shape, and L-shape provide better performance than that of the other shapes. It is also found that among inclined barriers, V-shape barrier significantly presents higher values of attenuation. Based on the calculation of different geometrics, it has been shown that a further 2 dB (A) in efficiency could be obtained by a better design of the median barrier which is labeled model "L."
Keywords: Boundary element method, environmental noise pollution, insertion loss, median barrier, transportation noise control
|How to cite this article:|
Monazzam MR, Fard SB. An investigation on the noise reduction performance of profiled rigid median barriers at highways. Noise Health 2012;14:106-12
| Introduction|| |
Noise barriers as an environmental noise control are used to reduce traffic noise. Median barriers like road side noise barriers can be used to intercept the noise between the source and the receiver, although road side barriers are the most common type for the noise mitigation of transportation systems. Extensive research has been carried out to predict the performance of the road side noise barriers. Crombie et al, in 1995 by proposing multiple-edge noise barriers, showed an increase in their efficiency at the deep shadow zone.  Hothersall et al reviewed the researchers performed on barriers with caps having T, Y, and arrow profiles.  Complete reviews of noise barrier on their performances can be found in Refs. ,,,,
Median barriers are originally designed in streets, freeways, and highways to minimize the chance of cross-median crashes as a result of errant vehicle encroachment.  However, considering median barrier type and its preferred placement location various opinions are existed.  Median barriers similar to road side barriers are able to substantially reduce the noise in the shadow zone. However, in order to estimate the efficiency of the barrier, it is important to consider aesthetic (geometry and placement) and cost points of view. On average, environmental noise barriers reduce A-weighted noise levels by 3-7 dB, depending on their design and height. 
In 2002, Martin and Hothersall presented the experimental data of a median noise barrier and informed that the differences between their insertion loss values at receiver positions were within 1 and 2 dB.  In contrast, there is yet no study to evaluate the influence of median barrier with different shapes.
In this study, after describing briefly the shapes of median barriers, insertion loss at 1/3 octave center frequencies of these models are compared, and the main aspects of the results obtained are discussed. The general purpose of this article is to demonstrate the improvement in median noise barrier efficiency that can be obtained by using new profiles.
| Methods|| |
Numerical modeling formulation
The boundary element method (BEM) is a type of numerical schemes with good accuracy, which has been used commonly for predicting the insertion loss of different shapes of barriers. ,, The applied method is a two-dimensional method which takes into account the effects of the ground plane. Configuration used for the numerical analysis of BEM is represented in [Figure 1].
The x-axis lies in the ground plane and y-axis is placed along the longitudinal direction. -axis is placed along the longitudinal direction. γ also represents the surface of the barrier above the ground plane. The geometrical variables of the barrier are assumed to be constant in one direction (z-axis). Note that this method is applicable since the ground is assumed to be reacting with normalized surface admittance β (rs ) at the point rs = (xs , ys ) on γ.
A boundary integral equation in which the integral is taken over the barrier's surface is obtained by reformulating the Helmholtz equation via Green's theorem. The boundary element technique is used to numerically solve this equation. , The barrier surface is divided into "n" elements. Inside each element the acoustic pressure p (rn , r0 ) is held to be constant, where rn is the mid-point of the nth element. Considering this approximation, the integral equation reduces as follows: 
the normal derivative at rs which guide into the propagation normal; and k the wave number.
ε(r) = 1 when r lies in the propagation medium exclude on γ and ε(r) =1/2 when r is a point on γ. In this field, only on a rigid ground, the sound pressure G(r, r0 ) is the acoustic pressure at a point r due to the source at r0 . G(r, r0 ) is given by:
where ro' indicates the image of the source in the ground (ro'= (x 0 -y 0 )) and H 0 (1) is the Hankel function of the first type and order 0.
For this study, in the BEM formulation, dimension of elements was taken to be less than λ/5 in order to ensure a higher accuracy of constant surface pressure over an element.  Hence, the barrier efficiency is evaluated through insertion loss at 1/3 octave frequency intervals between 50 and 4000 Hz. In order to avoid the problem of singular frequencies which can lead to wrong results for some frequencies, the computations are done in more detailed frequencies at 1/15 octave band and then these results are averaged into 1/3 octave band based on the simple energy averaging method. The insertion loss at each frequency is defined as
In this equation, Ph and Pg , respectively, are the pressure from the receiver with and without the presence of the barriers. Although this method can be time consuming, it was proved to be very accurate to calculate the sound propagation over complex shapes in a homogeneous atmosphere. 
| Result and Discussion|| |
Results of acoustic characterization of a few different shapes using two-dimensional BEM considered in this study are discussed as follows.
[Figure 2] describes the cross-sectional designs of the barriers which are examined in this investigation. In this study, to provide a similar condition for better comparison among different designed models, the width of all different barriers is taken as 0.03 m. The main focus for reference barrier was to design a barrier with negligible thickness where the cable median barrier is very thin in a real environment.
The overall height is 1 m to be more realistic and consistent with a previous study.  The length of each part of the caps in the relevant shapes is 15 cm and the angle for the sloped barriers is 10 degrees, which has been shown to be most effective. 
All surfaces were acoustically reflective which means the surface admittance is zero. In order to omit the interference effect between the direct wave and the ground reflection, the source is aligned in a plane perpendicular to the barrier edge and it is located on the rigid ground 8.5 m away from the base of the barrier. Thus, the source coordinate is fixed at (-8.5, 0). The source can simulate well the passing by vehicle noise, and the selected distance of noise source from center line of the median barrier can simulate the average distance of crossing vehicle in typical highways. Receivers are assumed to be located at various distances from the barrier (20, 50, and 100 m) and on the rigid ground and heights of 1.5 and 3 m above the ground.
It is worth adding that the IL at the mentioned different receiver positions were computed from 50 to 4000 Hz and then for broadband IL the results are converted to the A-weighted insertion loss, while in the case of shadow zone, only 500 Hz as an example was considered due to the importance of this frequency in traffic noise. In fact this article examines these distances to show the efficiency of the median barriers in both near and far shadow zones.
Efficiency of single noise barriers
The performance of each barrier was evaluated using the above mentioned method and then the comparisons of insertion loss separately were done in the shadow zone. [Figure 3] depicts the results obtained from barrier model "I" at receiver point (50, 0). Although at higher frequency range, above 80 Hz, barrier model "I" led to larger improvements up to 14 dB, it is shown that it has negative effect at frequencies 50 and 63 Hz.
|Figure 3: Performance of a rigid plain median barrier at receiver location (50, 0)|
Click here to view
[Figure 4] shows the change in the insertion loss at the receiver (50, 0) by increasing the thickness of the barrier model "I," which is called reference barrier in this article. It can be seen that the peaks of insertion loss achieve by barrier model "H" are at 80, 250, and 400 Hz, while at 800 Hz the lowest value of insertion loss is found. Furthermore, the performance of barrier model "H" is higher than that of in the reference barrier at frequencies less than 630 Hz. It has been shown by many researchers that the insertion loss of a wide barrier can be significantly higher than that of a thin barrier with the same height.  This can be explained by higher effective height in wide barriers compared with that of in thin barriers with the same overall height. Despite the above fact, the amount of performance improvement is not significantly high at frequencies above 800 Hz, which is related to the low height and small thickness changes in barrier model "H" compared with the reference barrier. It is worth remembering that the stem thickness in barrier model "H" is only 15 cm, taking the 1 m overall height and the tested source and receiver locations, the amount of increase in effective height is only 0.7 cm, which is not a significant improvement. That is why the overall amount of improvement is not considerable. However, it can be predicted that by increasing the effective height of median barrier, the acoustic performance similar to roadside barrier is improved.
|Figure 4: Performance improvement by barrier model "H" compared with the reference barrier at receiver location (50, 0)|
Click here to view
In the case of multiple edged barriers, it appears that the insertion loss is different depending on the diffraction edge shapes. [Figure 5] and [Figure 6] represent the insertion loss improvement of, respectively, the top modified and branched median barriers compared with the reference barrier. In this article, the barriers model "L" and "T" are called as top modified barriers and barrier models "A," "Y," and "Z" are known as branched barriers.
|Figure 5: The effects of top modified barrier for two models (L-shape and T-shape) along with a simple reactive barrier (I shape) at receiver point (50, 0)|
Click here to view
|Figure 6: Scatter field around barrier model "L" and its equivalent plain median barrier (90 degree is above; 0 and 180 degrees are, respectively, receiver and source sides of the barriers) at 200 Hz|
Click here to view
It can be seen from [Figure 5] that the new designed diffraction edges in barrier model "L" are better than that of in reference barrier at the entire tested. As can be seen in [Figure 5], the highest improvement of insertion loss in model "L" is at very low frequencies and the lowest improvement is 1.75 dB at 125 Hz. However, in T-shaped model, among frequency ranges of 630-1250 Hz, the performance of barrier model "T" is low compared with the reference barrier. At the receiver side in barrier model T, the incident and reflected sounds that occurred under this edge make the construction effect jointly where it is responsible for a considerable increase in those wavelengths.
In order to eliminate the edge effect, the mentioned edge is covered by an absorption material (flow resistivity = 20,000) and shown that the negative effect is declined.
Also, to study the negative effect of barrier model T in certain frequencies, length of the cap is firstly decreased (from 15 to 10 cm) and then increased (from 15 to 20 cm). It has been found that by increasing the cap span, the effective frequency is shifted toward lower frequencies, while in the occasion of short cap, the effective frequency is moved to higher frequencies.
The worst performance of barrier model "T" is found to be at 1250 Hz. Barrier model "L" has one extra edge at the source side compared with barrier model "T." It seems that the barriers with more source side edges provide better performance almost at entire frequency ranges. Increasing the number of source side edges can provide a situation to more wave cancellation in the source side of the barriers before flowing sound wave energy to the shadow zone.
The scatter sound field around the above two mentioned barriers are also examined at 200 Hz for more clarity of the top surface contribution of barrier model "L." The results are shown in [Figure 6]. As one clearly sees, barrier model "L" could provide lower scatter energy especially at both barrier sides. It means the energy flows toward receiver (right hand side of the graph) is significantly lower in barrier model "L" than that of the reference barrier model "I," which obviously causes higher insertion loss in the shadow zone.
[Figure 7] compares three different branched median barriers with their equivalent reference barrier. The figure shows that although the arrow shape improves slightly the barrier performance at very low frequencies, this barrier at frequencies higher than 250 Hz could not improve the performance compared with the reference barrier. This can be explained by the angle of the source branch of the barrier. This branch sloped upward at the source side, which causes sound wave reflected upward rather than downward. The efficiency of Y shape improves almost at entire frequency ranges except at some few frequencies. The weak effectiveness of Y-shape barrier is due to constructive effect of reflective wave from source side branch of the barrier and the reflective wave from its ground image. This can move to different frequency by change in the geometry and also it could be removed by using either passive or reactive absorbing surface on the source side branch, particularly its reflective edge. The performance barrier model "Z" at frequency higher than 315 Hz starts to become less effective. It is worth adding that smother trends are identified by Z shape which could be explained by having less diffraction edge compared with two other mentioned barriers.
|Figure 7: The amount of improvement made by three different branched barriers compared with the reference barrier at receiver point (50, 0)|
Click here to view
Generally, it can be seen that the trends of Y shape and Z shape across low-frequency noise spectra are almost the same, which is related to their similar source side branches. Although the Y shape having an extra branches at the receiver side provides better performance at higher frequencies. While the trend of acoustic efficiency of a branched barrier above 200 Hz can be improved by changing the branch's angle from Y to arrow shape which provides better extra sound pressure attenuation. Thus, in a large frequency band, it seems that it is possible to build a barrier much more efficient than the simple usual ones. 
In order to investigate the effect of sloped barrier's stem, different barrier designed shapes have been studied. [Figure 8] compares the amount of improvements made by three different sloped barriers including barriers model "K", "N," and "X" over the reference barrier. All three barriers have been tilted toward receiver by 10 degrees. The only differences are about their receiver side designs. It can be seen from the figure that barrier model "X" with highest construction materials has almost the worst performance among these three designed barrier models. It seems the receiver side of this barrier and its angles provide such a weakness for this barrier shape. In contrast, barrier model "N" having a vertical receiver side construction has slightly better performance compared with barrier model "X." This can lead us to explain this phenomenon by the angle of receiver side construction. In barrier model "X," the receiver side of barrier reflects the diffracted wave upward, which can introduce construction effect with either the incident or diffracted waves, while this is reduced in barrier model "N." Barrier model "K" having no extra receiver construction has smoother and better performance than that of two other mentioned sloped barriers. In fact, in this barrier the reflection from receiver side of barrier hits the ground and its reflection moves upward.
|Figure 8: The effect of improvement made by three different sloped barriers (barrier models "K", "N," and "X") compared with the reference barrier at receiver point (50, 0)|
Click here to view
The amount of improvement made by barrier models "V" and "R" over reference barriers is shown in [Figure 9]. Both barriers have the same overall heights but different facing angles. Barrier model "R" reflects the wave upward, while barrier model "V" inversely reflects the incident wave downward.
|Figure 9: The effect of improvement made by two different sloped wide barriers (barrier models "V" and "R") compared with the reference barrier at receiver point (50, 0)|
Click here to view
Barriers model "R" and "K" have the same source side slope shape, but the top edge is different. In fact, barrier model "R" has two diffraction edges while barrier model "K" has just one top diffraction edge. Despite the fact that these two barriers have the same overall heights, the effective height in barrier model "R" is slightly higher due to its wider top surface. By comparing the results of [Figure 8] and [Figure 9], it can be seen that the amount of reduction of insertion loss in low frequencies is higher in barrier model "X" compared with that in barrier model "R". This could be explained by the top surface contribution of barrier model "R" in this frequency ranges.
Among all sloped barrier, it seems that for frequencies lower than 2000 Hz, the V-shaped barrier with a flat top surface is more efficient than the others. This can be explained by both downward reflection effect of source side of this barrier and its higher effective heights due to its wider top surface span. These results are similar to that of Ishizuka and Fujiwara, who reported that with the barriers under different surface conditions, the performance of barrier can be considerably affected by the acoustic properties of the surfaces covering the upper edge of the barrier. 
Broadband insertion loss
To gain a better understanding of the effects of the tested barriers, the A-weighted traffic noise spectrum is calculated by combining the results for insertion loss at one-third octave band center frequencies for each barrier at nine receivers over the range 50-4000 Hz. 
[Table 1] summarizes the differences in the mean insertion loss of all designed barriers and the reference barrier (barrier model "I").
|Table 1: Comparison of the mean A-weighted insertion loss of straight edge barrier along with some multiple edges and sloped barriers|
Click here to view
Barrier models "H", "T", "Y", "L." and "V" could improve the overall A-weighted performance of the reference barrier. The common character among these barriers can be summarized by having higher effective heights and having more than one diffraction edges. Among sloped barriers also, the best barriers are among those with sloped stem toward source. Comparing all the results in the table, one can easily notice that the extra diffraction edge in the studied geometry is more effective than effective height and even slopping the barriers. The reason behind this finding can be described by low height and source/receiver distances from the median barriers. In this case, the highest improvement is made by barrier model "L" having four source side edges.
Sound field at the shadow zone
In this part of the study, a set of calculation in a wide area (2500 receiver points) behind three barriers model "I," "L," and "V" is done at 500 Hz. This frequency is used to take in account the traffic noise low-frequency character. The studied area was from 20 to 270 m distance from center line of median barriers on the ground extended to 10 m above ground. The near-field performance is excluded from this investigation because of the nature of median barrier position in highways. The amount of performance improvement provided by barrier model "L" compared with the reference barrier (barrier model "I") is shown in [Figure 10]. In the areas closer to the barrier with higher height, the amount of improvement is not significant, but at far field beyond 100 m the average improvement is more than 2 dB. However, this amount of improvement as it was expected is not provided by using barrier model "V" according to results, which is shown in [Figure 11]. The overall pattern of improvement in barrier model "V" is similar to the barrier model "L," but the amount of improvement almost every where in the tested field is significantly higher in barrier model "L."
|Figure 10: Insertion loss improvement of barrier model "L" relative to the reference barrier (barrier model "I") at 500 Hz|
Click here to view
|Figure 11: Insertion loss improvement of barrier model "V" relative to the reference barrier (barrier model "I") at 500 Hz|
Click here to view
| Conclusion|| |
In this study, the BEM is applied in order to predict the efficiency of several types of noise barriers over nine receiver points in 1/3 octave band frequencies from 50 to 4000 Hz. A theoretical study on the performance of different profiled median barriers compared with the selected barrier as a reference model in wide areas in shadow zone was also conducted. Based on the above investigations, the following results have been achieved:
It is worth noting that the above presented results were obtained purely by numerical modeling and all surfaces are assumed to be rigid (all surface admittances are zero, which is the Neumann boundary condition), and environmental factors such as atmospheric turbulence are ignored in the model. Therefore, further work needs to be done experimentally in a real application. Furthermore, more works also need to be done on improving the acoustical efficiency of the median barriers by using different surface materials and profiles. An extensive investigation is being done to study the above and will be subjects of future papers.
- Increasing the width of the barrier from model I to model-H improves the overall broad band A-weighted performance of the barrier by 0.081 dB (A). This amount of improvement is not a significant improvement. The controversy between these findings and the results presented by Hothersall et al. for the single barriers installed in one side of roadways can be explained by lower heights and the small possible stem thickness in the tested median barriers. 
- It is well known that the overall improvement of the T-shaped barrier is greater than that of the reference barrier. This aspect was confirmed in this study by the observed mean A-weighted insertion loss by 0.26 dB (A). However, the extra efficiency value about 0.32 dB (A) was found at nine receiver points for Y-shape. This shape has a side sloped branch, which provide the acoustic performance advantages for this barrier model.
- In prediction of the performance of different inclined barrier, model K results in fewer frequency peaks in the insertion loss spectrum, although among these shapes (K, N, R, V, and X model), the overall excess attenuation belongs to V model.
- It is found that in same conditions the influence of the acoustic barrier (model L) is much better than the other edged barriers. It is explained by the redirection of the sound wave upward by the edges, which is provided by the novel presented shape with some extra diffraction edges.
| Acknowledgments|| |
The authors would like to thank Tehran University of Medical Sciences for financial support of this study. The authors declare that they have no conflicts of interest.
| References|| |
|1.||Crombie DH, Hothersall DC, Chandler-Wild SN. Multiple-edge noise barrier. Appl Acoust 1995;44:353-67. |
|2.||Hothersall DC, Crombie DH, Chandler-Wilde SN. The performance of T-shape profile and associated noise barriers. Appl Acoust 1991;32:269-81. |
|3.||Watts GR, Crombie DH, Hothersall DC. Acoustic performance of new designs of traffic noise barriers: Full scale test. J Sound Vib 1994;1771:289-305. |
|4.||Fujiwara K, Hothersall D C, Kim Ch. Noise barriers with reactive surfaces. Appl Acoust 1998;53:255-72. |
|5.||Shao W, Lee H P, Lim S P. Performance of noise barriers with random edge profiles. Appl Acoust 2001;62:1157-70. |
|6.||Ishizuka T, Fujiwara K. Performance of noise barriers with various edge shapes and acoustical conditions. Appl Acoust 2004;65:125-41. |
|7.||Monazzam MR, Lam YW. Performance of profiled single noise barriers covered with quadratic residue diffusers. Appl Acoust 2005;66:709-30. |
|8.||Bligh R, Miaou SH. Lord D, Cooner S. Evaluation of Median Barrier Guidelines. 2006; Report number: 0-4254-1. Available from: http://tti.tamu.edu/documents/0-4254-1.pdf. [Last accessed on 2010 Dec 23] |
|9.||Kim TG, Donnell ET, Lee D. Use of cultural consensus analysis to evaluate expert feedback of median safety. Accid Anal Prev 2008;40:1458-67. |
|10.||Arenas JP. Potential problems with environmental sound barriers when used in mitigating surface transportation noise. Sci Total Environ 2008;405:173-9. |
|11.||Martin SJ, Hothersall DC. Numerical modeling of median road traffic noise barriers. J Sound Vib 2002;251:671-81. |
|12.||Godinho L, Antonio J, Tadeu A. Sound propagation around rigid barriers laterally confined by tall buildings. Appl Acoust 2002;63:595-609. |
|13.||Baulac M. Defrance J, Jean P. Optimization of multiple edge barriers with genetic algorithms coupled with a Nelder-Mead local search. J Sound Vib 2007;300:71-87. |
|14.||Baulac M, Defrance J, Jean P. Optimization with genetic algorithm of the acoustic performance of T-shaped noise barriers with a reactive top surface. Appl Acoust 2008;69:332-42. |
|15.||Seznec R. Diffraction of sound around barriers: use of boundary elements technique. J Sound Vib 1980;73:195-209. |
|16.||Hothersall DC, Chandler-Wilde SN, Hajmirzae MN. Efficiency of single noise barriers. J Sound Vib 1991;146:303-22. |
|17.||Watts GR. Acoustics performance of parallel traffic noise barriers. Appl Acoust 1996;47:95-119. |
|18.||Lam YW. A boundary element method for the calculation of noise barrier insertion loss in the presence of atmospheric turbulence. Appl Acoust 2004;65:583-603. |
|19.||Duhamel D. Shape optimization of noise barriers using genetic algorithms. J Sound Vib 2006;297:432-43. |
|20.||BS EN. Road traffic noise reducing devices- Test method for determining the acoustic performance, Part 3. Normalized traffic noise spectrum. 1998. p. 1793-3. |
Mohammad Reza Monazzam
Occ Hyg Department, School of Public Health, Tehran University of Medical Sciences, Tehran 13155 119
[Figure 1], [Figure 2], [Figure 3], [Figure 4], [Figure 5], [Figure 6], [Figure 7], [Figure 8], [Figure 9], [Figure 10], [Figure 11]