Noise Investigation and Modelling of Urban Traffic Under Dynamic Conditions

Abstract

Noise pollution as an environmental concern has lately found a lot of attention amongst environmentalists, engineers and general public. There are different sources of noise in the environment like traffic noise: rail, road and aircraft, use of loudspeakers during weddings and religious functions and machinery noise: industry or construction machinery. Vehicular traffic noise (emanating from vehicles plying on the roads) is a major contributor to the overall noise in the environment to which the general population is exposed. The number of vehicles on the roads (especially two-wheelers and cars) has been increasing relentlessly with growing population, demand, affordability and purchasing power. The high traffic noise levels have adverse health effects on the human population. Thus, it is imperative for the town planners and transport engineers to consider the noise aspect at the conceptual stage of the design of urban roads, residential and commercial buildings and market areas. Traffic noise prediction models help in the assessment of noise levels at the initial design stages and later stages when modifications are planned. They are also useful for environmental impact assessment (EIA) studies. The present work describes the urban traffic noise problem with respect to the high noise levels present on the Indian roads, by taking the Patiala city as a representative case. The existing noise levels were assessed at various locations on the city roads by using the standard measuring equipment and procedure. The measurements revealed that the existing noise levels are much higher than the prescribed limits. A Graph Theoretic (GT) approach which is a systems approach has been used to develop a new traffic noise model and also to propose a new traffic noise descriptor. An Urban Traffic Noise System (UTNS) has been proposed which consists of the subsystems: road traffic subsystem, human subsystem, environment subsystem, traffic network subsystem and urban prosperity subsystem. The first subsystem (road traffic subsystem) has been studied in detail in the present work. The different steps involved in the formulation of a traffic noise model have been illustrated. ‘Permanent function’ in the form of a composite noise index (‘p’) has been formulated and its calculation and correlation (in terms of ‘log p’) with the existing noise level descriptors like Leq, have been illustrated with examples. In the first example, the traffic situation presented in Nicol and Wilson’s work (2004) has been taken to establish and illustrate the proposed methodology. The parameters considered v in the model are traffic volume, street width, height of measuring point above street level and average speed. GT approach is applied for developing the model. A permanent noise index ‘p’ which takes into account the effect of different traffic noise parameters and the interactions between them, is proposed and verified with the experimental data. The variation of ‘p’ with Leq, L90 and L10 for the different measuring locations has been plotted, and equations for the three noise level descriptors and ‘log p’ have been formed using regression analysis. The values of multiple R, R2 and error for the testing dataset show good correlation between the measured and predicted values of the traffic noise level, which match fairly well. The second illustrative example is that of an urban traffic noise scenario in Patiala city, India. An experimental study has been carried out at the Fountain roundabout in the city in order to further validate the methodology developed using the GT approach. Data related to the road traffic subsystem (consisting of four parameters traffic volume, percentage of heavy vehicles, acceleration of vehicles and road width) has been collected. The permanent noise index ‘p’ is formulated using the GT approach and verified with the experimental data. The values of R, R2 and error in the predicted noise level (Leq dB(A)) indicate that the results are matching fairly well. Another gap that was identified from a study of the literature was the use of soft computing methods in traffic noise modelling only in a limited way. Genetic algorithms (GAs) and artificial neural networks (ANNs) seem to have been used so far for the modelling purpose. There are many other state-of-the-art methods/techniques which intuitively were felt to have a potential of application in this area. Four different soft computing methods (Generalized Linear Model, Decision Trees, Neural Networks and Random Forests) have been explored and used in developing the models for typical road traffic conditions in Patiala. The traffic noise prediction models are developed, with equivalent sound pressure level, Leq as the output (dependent variable) and the traffic noise variables: hourly traffic flow, percentage of heavy vehicles and average speed of vehicles, as the independent variables. The results obtained were analyzed and it was seen that the ‘Random Forests’ method excels over all the other techniques on the performance criteria of correlation, R2 (coefficient of determination), RMSE (root mean square error) and accuracy. The effect of honking on the noise level was also studied, as it is an important parameter in the urban Indian traffic scene. An experimental investigation was carried out at five different locations in Patiala city with different levels of traffic volume and honking. It has been seen, vi as expected, that the overall traffic noise levels increase due to the presence of heavy honking. Though it is quite obvious, but some quantification of the effect of the horn noise was required, which has been addressed. Also, the improvement in the model by including the honking effect has been validated in an objective way. A systematic acoustic design methodology for a noise barrier as an abatement measure is presented thereafter. The design methodology proposed is useful for determining the height of a barrier to achieve the desired attenuation based purely on acoustic considerations. It is hoped that the work presented can be useful to urban town planners, pollution control bodies/authority, civil engineers working on road and infrastructure projects, acoustic engineers and officials working on environment impact assessment (EIA) studies with an objective of vehicular traffic noise prediction and abatement.