As building designers, chances are that you will be called upon to design a building next to a busy roadway. At a very early stage in the design you will need to be able to estimate the sound levels being generated by the traffic on this road and come up with some solutions for the control of any resulting noise.

Predicting traffic noise is no simple procedure. It involves a fair amount of assumption and, to some degree, simplification - so an understanding of the process behind the formulae used is just as important as an understanding of the formulae themselves. This topic concentrates on a method originally developed for the English Department of the Environment and used in Australia by the Environment Protection Agency, the Main Roads Department and most consultants.

Calculation Methods

Obviously the best method of determining the sound levels being generated at a specific site are to go there and physically measure them. However, the time you spend there may not be truly representative of the levels actually experienced (i.e.: don't go there at 2:00pm on a Sunday afternoon) or it may not even be possible to obtain real measurements (i.e.: when assessing the impact of a proposed road). It is obvious, therefore, that some method of both predicting and averaging traffic noise is required.

Estimating Noise Levels - L10(18 hr)

The designation L10 denotes the level, in dB(A), which is exceeded for only one-tenth of any specified hour. In order to calculate L10, levels need to be recorded periodically, perhaps as often as every 5 seconds. These levels are then plotted on a frequency distribution graph to form a distribution curve. The L10 can then be read directly as the point on the dB(A) axis below which 90% of the area under the graph falls. This is simply a matter of dividing up the measured values into groups, usually each integer dB, counting down one-tenth of the total values measured and reading off the x-axis.

L10(18hr), the level used in most regulations, is simply the arithmetic mean of the 18 separate one-hourly values of L10 covering the period 6:00am to 12:00pm on a normal working-day.

Estimating Vehicle Noise Levels

When using traffic noise as the source, the method we are going to use begins by predicting the level that would be experienced 10m from the near-side of the road. For the moment, consider the major factors in the prediction of noise levels generated by a line of traffic.

Number of Vehicles per Hour

If the number of vehicles using the road is known, the noise level at 10m from the near-side of a road can be estimated from:

L10(18hr) = 28.1 + 10 log Q dB(A)

or

L10(1hr) = 41.2 + 10 log q dB(A)

Where:
Q = total number of vehicles between 6:00am and 12:00pm, and
q = total number of vehicles per hour.

Vehicle Speeds and Size

The general sound level given above has to be corrected for the average speed of vehicles using the road as well as for the number of heavy vehicles (usually classes B and C, however, a heavy vehicle is defined as one with an unladen weight of 1525kg). Obviously, the faster the vehicles are travelling, the more noise they will generate from tyre-noise through to the general air displacement. Similarly, heavy vehicles generate higher sound level than lighter vehicles, thus the percentage of heavy vehicles using the road must be found. The correction is then given by:

Cusage = 33 log (v + 40 + (500/v))
       + 10 log (1 + (5p/v)) - 68.8 dB(A)

Where:
v = average speed (km/h), and
p = percentage of heavy vehicles (0 - 100%).

Type of Road	Speed / Size	Use
Rural roads	110 km/h	108
Urban Freeway	90 km/h speed limit	92
Urban Highway	70 km/h speed limit	65
Urban Street	Dual Carriageway	60
Urban Street	Single Carriageway	55
Urban Street	Single Congested	50

The Gradient of the Road

Obviously engines must work harder to climb steep hills, thus the gradient must be considered. If using the actual average speed along the road (i.e.: from Main Roads Dept), then the correction is given by:

Cg = 0.3G

Where:| G = percentage gradient.

If the design speed of the road is used, the correction becomes:

Cg = 0.2G

The Road Condition

In normal circumstances no correction is required for either sealed of gravel roads. If, however rare, the road is constructed of concrete with deep random grooving greater than 5mm in width, then the following correction must be applied; Ccond = 4 - 0.03p where p is the percentage heavy vehicles.

The Inverse Square Law and Ground Cover

Simple flat ground situations are uncommon and only very rarely will there be an unobstructed view of a straight, level and uniform road. However, because the effects of barriers, vegetation and gradients are considered separately, the idealised situation is used in order to consider surface effects. These effects take into account both the distance from the road edge and the height of the observer. If d is the horizontal distance between the near-side of the road and the receiver, h is the height of the receiver, then the total distance between the source and receiver is given as dtotal = ((d + 3.5)2 + (h - 0.5)2 )^1/2, then the correction for hard ground is given by:

Chard = - 10 log (dtotal / 13.5) dB(A)

Assuming that the source is 0.5m above the ground and 3.5 m away from the near-side of the road. If the intervening ground is soft or grassland and [1 < h < (d+3.5)/3], then the correction becomes:

Csoft = -10 log (dtotal / 13.5) + 5.2 log (3h / (d + 3.5)) dB(A)

NOTE: It is important that the hard ground correction always be used whenever there is any screening involved by barriers as it truly represents the sound propagation from a line source. The grassland formula is simply a correction to the base hard-ground situation and the inclusion of barrier effects will invalidate results. If the intervening ground is a combination of hard ground and grassland and no barriers are involved, then the correction corresponding to whichever type of ground dominates is to be applied.

Simple Barriers

As discussed in a previous lecture, the effect of a barrier depends upon the frequency of the sound and the path difference (a + b - c) between source and receiver. As traffic noise is a complex sound and it has already been estimated in dB(A) at 10m, rather than calculate a correction for each frequency band, a method for deriving the overall correction can be used. This uses a table of coefficients and the following formula:

Cbarrier = S (An xn) dB(A)

where:
x = log (a + b - c) and the coefficients A0, A1, ... An are as follows;

Coefficient	Shadow Zone	Illuminated Zone
A0	-15.4	0.0
A1	-8.26	+0.109
A2	-2.787	-0.815
A3	-0.831	+0.479
A4	0.198	+0.3284
A5	+0.15390	+0.04385
A6	+0.12248	-
A7	+0.02175	-

NOTE: Barriers can still have some effect even if they do not obscure the sight-line between the source and receiver. This is known as the 'illuminated zone' (as opposed to the 'shadow zone') and is used to evaluate the small screening correction for receiver points that can just see the source over the top of a barrier.

It is not necessary to use constants beyond A7 in the shadow zone and beyond A5 in the illuminated zone. This table is only valid for (-3 <= x <= +1.2). Below -3 (-4 in the illuminated zone), all A values should be taken as -5 whereas above +1.2, the method should not be used.

Problems with Low Barriers

Often the effects of very low barriers turn out to be less than the effect of using soft ground. Such is the case with the twin-beam metal crash barriers some of you will be so familiar with. In situations where a crash barrier is used and the intervening ground is soft, the noise level should evaluated as for grassland, ignoring the barrier, and also using the hard-ground correction with the barrier. The result yielding the lowest value should be used.

Embankments and Elevated Roads

When buildings or earth mounds screen the reception point from the road, the height and position of the equivalent barrier should be determined. Diagram 1 shows the method used. Basically, a wide barrier is approximated by a thin barrier situated at the intersection point of the line joining the source to the top of the barrier and the line joining the receiver to the top.

NOTE: Then multiple barriers of different heights screen the receiver from the road, they should be evaluated separately and only the correction resulting in the lowest noise level used.

Reflections

As well as the attenuation corrections, allowances have to be made for the reflection of sound off vertical surfaces. Therefore, in many situations, further corrections must be made as detailed below.

• If the receiver is within 1m of a building facade, then the noise level is increased by 2.5dB(A).
• Measurements taken down side streets increase the level by 2.5dB(A) due to reflections from adjacent houses.
• Reflective surfaces on the far side of the road increase the level by 1dB(A).

Complex Situations

Corrections for the Angle of View

In many cases the angle of view of the road will include a range of different configurations (i.e.: bends in the road, intersections, short barriers, etc.). To accommodate this, the overall field of view must be divided into a number of segments of uniform propagation conditions. The overall sound level can be found by treating each segment as if it filled the entire 180o angle of view. The following correction is then applied to each segment.

Cv = 10 log (f / 180)

Where:
f = actual field of view in degrees.

NOTE: In such segments, the road is always projected right along the field of view and the distance from the segment is measured perpendicular to the extended road.

A Regularly Spaced Series of Barriers.

When considering a situation in which a road is partially screened by a regular series of uniform barriers, rather than strictly apply the segment principle, it is possible to use only two segments. One of these covers the gaps and the other the screened areas. To do it this way, both the average gap, b, and average building length, B, is required. If Z = (B / (B+b)), then the two covering angles can be taken to be:

 fgaps = 180 * Z 

 fbldgs = 180 * (1-Z)

Widely separated dual-carriageways

Treat as two separate roads and combine the results using logs.

Multiple sources

Once again, treat each source/road separately and combine the results using logs.