PERFORMANCE OF FHWA MODEL FOR PREDICTING TRAFFIC NOISE: A CASE STUDY OF METROPOLITAN CITY, LUCKNOW (INDIA)

. Industrial and transport activities are the two major sources of noise pollution in any metropolitan city. Lucknow city, the capital of the largest populated state Uttar Pradesh in India has an area of 310 sq. km and is rapidly growing as a commercial, industrial and trading centre of northern India. Th e population of Lucknow city as per census 2001 is 22.45 Lacs. It is expected that by the year 2021 it will make 45 Lacs. Th e total vehicle population in Lucknow city on 31 March 2008, was nearly 1 million with almost 80% two wheelers, 12% cars, 1.36% three wheelers, 0.45% buses etc. A study was carried out to assess the existing status of noise levels and its impacts on the environment with a possibility of further expansion of the city. Ambient noise levels were measured at diff erent locations selected on the basis of land use such as silence, heavy traffi c and residential and commercial zones. It was found that noise levels at all selected locations were much higher (75–90 dB) than the prescribed limits. Th e observed traffi c volume and data on road geometry were used to predict noise levels using Federal Highway Administration Agency (FHWA) model and the calculated noise levels were compared with the observed levels for checking the suitability of this model for predicting the future levels. It was established that the results obtained by FHWA model were very close to the observed noise levels and that the model was suitable to be used for other similar metropolitan cities in India.


Introduction
Vehicular noise pollution is increasing at an alarming rate in metropolitan cities with an increase in urbanization.
A rapid increase in population, unplanned urbanization and the development of transportation projects without environmental impact assessment may be listed as the main reason behind traffi c noise (Abdel Alim et al. 1983;Ayvaz 1994;Morillas et al. 2002;Homburger et al. 1992).
In India, noise levels in metropolitan cities have reached very high levels making 76-80 dB(A) and traffi c management technologies have failed due to a lack of enforcement and poor legislation. More than 55% of the total noise in our environment is due to vehicular movements. All India Institute of Medical Sciences New Delhi (AIIMS) has revealed that noise enlarges blood vessels of the brain thereby causing severe headaches. Even mild noise is enough to dilate the pupil of the eye. Repeated dilation makes it necessary to change the eyes' focus immediately, thus adversely aff ecting the ability to do delicate work. Th e aggravation of allergy, asthma, emotional breakdown, insomnia, hypertension, gastrointestinal problems, heart diseases, high blood pressure and malformation in foetal nervous system are only a few diseases caused by noise. Li et al. (2002) performed a study to analyze traffi c noise levels along three main roads in Beijing, China. Th ey discovered that the selected roads were overloaded by traffi c fl ow during daytime. Due to road traffi c, noise levels were above relevant environmental standards by 5 dB(A). Ali and Tamura (2003) conducted a traffi c noise study in Greater Cairo, Egypt and carried out an extensive measurement in 21 sites. Th ey measured the degree of annoyance by questionnaire. Th e received results revealed that there was a strong relationship between road traffi c noise levels and the percentage of highly annoyed respondents. Morrilas et al. (2002) carried out noise studies in Caceres, Spain and found out that noise levels were quite high with 90% of values higher than 65 dB(A) and the results were in coincidence with the results of other researchers. Koushki et al. (1999) conducted noise studies in Kuwait on the arterial roadway and observed that there was a strong correlation between traffi c volume and noise level. Bazaras (2006) performed studies of internal noise modelling problems of transport power equipment. Kliučininkas and Šaliūnas (2006) investigated problems of noise mapping for the management of urban traffi c fl ows. Gupta et al. (1986) studied problems of traffi c noise for various land uses for mixed traffi c fl ow. Rao (1991) carried out studies on prediction of road traffi c noise. Baltrėnas et al. (2007a) carried out effi ciency evaluation of a noise barrier. Baltrėnas et al. (2007b) performed a study to investigation of noise dispersion from seaport equipment on the enterprise territory and residential environment. Bazaras et al. (2008) performed studies in Lithuania at intersections and established interdependency between noise levels and traffi c fl ow. Vaisis et al. (2008) carried out noise prediction modeling nearby Siauliai railway station (Lithuania). Akgüngör and Demirel (2008) iinvestigated urban traffi c based noise pollution in the city of Kirikkale (Turkey). Paslawski (2009) investigated fl exibility in highway noise management.
Th erefore, it is the need of the hour to prevent noise before it reaches a dangerous level. Diff erent types of vehicles cause diff erent noise levels. Heavy vehicles, trucks in particular, are the most noise producing vehicles because of axle loads. If the axle load of a truck is reduced from nearly 2000 kg to 500 kg, a 15 dB(A) decrease in noise level is obtained. Vehicle speed is another major factor generating traffi c based noise. Th e faster a vehicle travels the more noise it generates because of the friction between tires and pavement. Actually, as the speed increases the friction noise surpasses the motor noise (Homburger et al. 1992). Apart from the type of a vehicle and its speed, the other factors are: • the volume of traffi c; • the number of heavy vehicles in the fl ow of traffi c. In addition, the following factors infl uence the noise level at a reception point at a reference distance from the highway: • distance between the source and the receiver; • ground absorption; • obstruction due to noise barriers; • obstruction due to a restricted angle of view; • refl ection eff ect. Generally, the loudness of traffi c noise is increased by heavier traffi c volume, higher speeds and larger numbers of trucks. Vehicle noise is a combination of the noises produced by the engine, exhaust and tyres. Th e loudness of traffi c noise can also be increased by defective muffl ers or other faulty systems of vehicles. Other condition (such as a steep slope) that causes heavy labouring of motor vehicle engines will also increase traffi c noise levels infl uenced by distance, terrain, vegetation and natural or manmade obstacles. Traffi c noise is not usually a serious problem for people who live more than 150 m away from heavily travelled freeways or more than 30 to 60m from lightly travelled roads.

Methods of Reducing Highway Noise
Road traffi c noise in most of the urban areas is increasing at an alarming rate which is a cause of concern for the residents living along the highways. Noise levels must be controlled in order to reduce its societal impacts. Reducing noise levels may be produced following Harris (1979) who suggested some noise control measures like: • motor vehicle control; • land use control; • highway planning and design; • buff er zones; • noise barriers; • using dead end streets for residential complexes; • depressing freeways and arterial roads below the ground level; • creating more gap between road and buildings; • constructing high rise buildings along the roads providing barrier for low rise buildings; • making external and internal sound insulated walls; • making double glazed windows.

Noise Reduction on New Roads
All of the above described measures can be employed on both existing and new roads. However, the following additional measures may be introduced on new roads: • A new road can be located away from noise sensitive areas, such as schools or hospitals and placed near less-sensitive areas such as business centres or industrial areas. New roads can also be located in developed areas. • New roads can be constructed below ground level. A large amount of noise from vehicles travelling on a similar type of the road is defl ected into the air by embankments on the side of the road. Th erefore, these embankments function as noise barriers. • A new road can be designed and constructed as level as possible. Th e elimination of steep slopes helps with reducing traffi c noise. Although there are a huge number of noise reduction measures possible, however all of those have certain limitations. At the same time, there are many situations where none of these noise reduction measures can be employed successfully. In such situation, the only option left with the local authorities is to provide adequate muffl er devices for the vehicles producing louder noise.

Traffi c Noise and Traffi c Variables
Noise from interrupted fl ow traffi c in urban areas has diff erent characteristics from traffi c noise generated by free fl ow on rural highway. Noise levels in urban areas depend on surrounding conditions such as: • carriageway width; • building on road side; • road intersection. Th e noise generated by traffi c under interrupted fl ow condition may also be regarded as the aggregation of individual vehicle noise. Vehicle operation under such condition is predominately due to acceleration and braking. Th e interrupted fl ow condition occurs at intersection, congested roads and other road geometrics where common acceleration and braking manoeuvres exist.

About FHWA (Federal Highway Administration Agency) model
Over the past decade, considerable progress has been made in developing techniques for predicting noise level for road traffi c. Much of early work concentrated on forecasting noise level from freely fl owing traffi c and the result of these studies has been used to calculate noise emanating from the highway and a similar type of a new road carrying traffi c travelling at moderate and high speed.
Pamanikabud and Vivitjinda (2002) where: L o -basic noise level of a stream of vehicles; ΔL i -adjustment applied. Th e basic noise level is the noise emitted by a particular class of the vehicle at a distance of 15 m from the centre of the inner lane at the given speed and for the given road surface. FHWA model calculates noise level through a series of adjustments to the reference sound level measured through fi eld measurements. Th e actual FHWA model is in the form given below: where: L eq -hourly equivalent sound level; L o -reference energy mean emission level; A vs -volume and speed correction; A D -distance correction; A B -barrier correction; A F -fl ow correction; A G -gradient correction; A S -ground cover correction. Volume and speed correction: where: D o -reference distance given as 10 m; D -distance of measurement from the centre of each lane; αground cover coeffi cient. Barrier correction can be estimated using the expression: 2 cos 5 20 log 2 cos where: N oi -Fresnel number for the specifi c category; N o = 2δλ (where: δ -path diff erence). Traffi c fl ow adjustment: where: V -volume for the category in veh/h; S -speed in km/h; D o -reference distance. Grade correction is taken as the percentage of grade.

Limitations on FHWA model:
1. In this model, vehicles are classifi ed into three categories namely light commercial vehicles, medium trucks and heavy trucks. 2. Adjustments are applied for the calculation of hourly L eq . 3. Reference distance for measurement is taken as 15 meters from the centre of the near site lane and the actual distance of measurement is recorded. 4. No separate lane concept for acceleration or deacceleration is considered. Noise Standards. Th e ambient noise standards of diff erent types of zones being followed in India are given in Table 1.

Field Studies
A study was carried out to assess the existing status of noise levels and its impacts on the environment with a possibility of a further expansion of the city. Ambient noise levels were measured at diff erent locations selected on the basis of land use such as silence, heavy traffi c and residential and commercial zones. Th is study was mainly intended to measure the noise level in urban and semiurban locations and hence the locations were chosen so as to represent diff erent zones within an urban area like residential zone, commercial zone, silence zone and heavy traffi c zone. Th e details of the selected location are given in Table 2. Keeping in view the objective of the study, a fi eld data collection programme was designed to collect data regarding the following parameters: • classifi ed traffi c volume; • classifi ed traffi c speed; • ambient noise level; • geometric parameters like road width, the number of lanes, lane width, shoulder width, the presence of median and its width, the presence of pedestrian sidewalk and its width and the details of roadside developments; • longitudinal section parameters like the distance of a receptor point from the intersection; • adjoining land use and the presence of bus stops etc. which would aff ect the continuous fl ow of traffi c; • miscellaneous information regarding the type and condition of roadway etc. Traffi c fl ow was generally uninterrupted in character. Th e basic noise data was taken by placing noise level meter 1.2 m above the ground level.
Classifi ed traffi c speed. Th e classifi ed traffi c speed was recorded for both directions at each of the selected locations. Th e classifi ed traffi c speed study was carried out for the same duration as the noise level study and the traffi c volume study. Th e spot speed of vehicles was recorded using the traffi c hand held radar. Th e identifi ed vehicles categorized for spot speed study are listed below: • car/jeep/van; • scooter/motorcycle; • light commercial vehicle; • bus; • truck; • tractor trailer. Ambient noise level. Ambient noise levels for the selected locations were collected using the Leutron make noise level meter. Ambient noise data was taken at varying distances from the pavement edge to incorporate the eff ect of distance in noise dissipation in the model development process. Ambient noise pollution data was collected continuously for a period of twelve hours for both directions at all identifi ed locations. For data collection, each hour was divided into the intervals of 15 minutes and observations were taken at an interval of 15 seconds. Th us, a total of 240 observations were taken in an hour. Th e data is organized to calculate L 10 , L 50 , L 90 and L eq . Geometric parameters. Th e parameters of road like road width, the number of lanes, lane width, shoulder width, the presence of median and its width, the presence of pedestrian sidewalk and its width and the details of roadside developments as well as miscellaneous information regarding the type and condition of roadway were also recorded etc. Adjoining land use and the presence of bus stops etc. would aff ect the continuous fl ow of traffi c.
Traffi c volume. As the directional classifi ed traffi c volume is the basic data requirement of this study, traffi c volume studies were carried out at all locations identifi ed for the detailed study. At all selected locations, traffi c volume studies were conducted continuously for a period of 12 hours (8 am-8 pm). Directional classifi ed traffi c volume data was manually recorded in pre-designed, hourly traffi c volume recording proforma subdivided into 15 minute intervals. Since the diff erent classes of vehicles use the common roadway facilities without segregation on the highway, traffi c fl ow becomes heterogeneous, and hence it is required to convert all the categories of vehicles into a single unit called Equivalent Passenger Car Units (EPCU).
Traffi c volumes at all locations have been presented, both in the form of total vehicles per hour as well as converted into PCUs and expressed in terms of equivalent passenger car units. Th e conversion factors are given in Table 3. Traffi c volume count in EPCU is given in Table 4.

Analysis, Results and Discussion
Th e noise levels recorded at each location were taken on an excel sheet and worked out.
Th e traffi c volume and average traffi c speed data of each location were also tabulated on an excel spreadsheet and from the equations given for each correction, the corresponding value for distance and speed corrections were worked out.
Th e L eq obtained for near and far lanes were combined to get the fi nal hourly L eq values for each location the calculation of which has been compared with the observed L eq values of each location.
A regression equation between the observed and FHWA calculated L eq values for each location has been drawn and the regression coeffi cient R 2 has also been worked out to assess the performance of the model.
Th e correlation equation obtained between observed and calculated L eg values for each location has been given in Table 5.
To check the sensitivity of the model, t-test was conducted and good results were yielded.

Conclusions and Comments
A combined regression equation for all locations selected on the basis of land use is then combined together to validate FHWA (Federal Highway Administration Agency) noise prediction model for Lucknow city taking 120 observed L eq and 120 calculated L eq values. Th ese are plotted on the Cartesian co-ordinates and a liner equation is developed between these two values as shown in Fig. Th e developed noise prediction model is: Y = 69.864 · ln(X) -233.59 and the model has R 2 = 0.9075, where: Y -predicted L eq noise level value; X -observed L eq noise level value. Hence, it can be concluded that FHWA model may be used for predicting noise pollution levels in metropolitan cities like Lucknow.