MODELLING OF TRAFFIC NOISE FOR MID-BLOCK LOCATIONS IN URBAN AREAS

Abstract

Rapid motorization has brought many environmental issues, i.e., noise pollution, air pollution, which challenges the sustainability of mid-sized cities in India. Technological advancements have improved human living, but they have also resulted in many negative consequences. Pollution is one of those major concerns for urban communities. Traffic noise is one of the most irritating and harmful components of environmental pollution for the residents residing near busy roads in urban areas. Many countries have regulated vehicle noise emission standards and framed regulations to reduce urban traffic noise starting with the London Noise Survey from UK. Noise induces physiological and psychological effects, notably annoyance and sleep disturbance, hearing impairment, speech intelligibility, physiological dysfunctions, mental illness, headache, hypertension, performance reduction, and cardiovascular diseases. Research on traffic noise has significantly contributed to understanding of noise pollution situations both in urban and interurban context. Traffic noise studies are undertaken through development of a reference energy mean emission level model, a prediction model, and subsequent research on mitigation measures. For the development of REMEL model, experiments were conducted in US on a test track to develop the REMEL models for three vehicle categories (Automobiles, Medium trucks, and Heavy Trucks). In Indian context, there is a need to develop models for vehicle categories that are currently plying on in our cities and reflect the real maintenance and mechanical condition of these vehicles. A track-based study is likely to miss these effects. Many literatures show the traffic noise prediction models for developed countries like FHWA model for US, CoRTN model for the UK, and other countries that have their own noise standards and traffic noise prediction models. Nevertheless, these models are not applicable in Indian context due to the vehicle categories, their dimensions are different, road conditions, weather variation, engine noise, and different traffic flow patterns such as heterogeneous traffic conditions rather than homogeneous traffic conditions. There is a need to develop a specific traffic noise model for mid-sized cities in Indian context, which captures the multiplicity of modes. With these model developments, there is a need for re-evaluation of policies and regulations for mitigation and to provide effective and feasible mitigative measures. Most countries are focusing on environmental emission factors (air, noise, water, etc.) for sustainable development. To regulate the pollution factor (Traffic noise) they have developed policies and regulations to mitigation of traffic noise pollution in their countries as per their condition for mitigation. The Central Pollution Control Board (CPCB), India has developed standards of noise levels for different land use areas and intervals of time, and same is follo wed by all the agencies in the country. For the mitigation of traffic noise, several studies suggested different materials and methods. All these methods might be feasible for some countries to use those materials-based noise barriers but may not be feasible for many countries due to the high cost and unavailability of techniques. In Indian context, there is a need to study feasible and cost-effective solutions for noise mitigation in form of height of walls, noise barriers on flyovers, and creating a curtain wall of vegetation for the mid-sized Indian city. The present study is focused on the analysis and development of the traffic noise model in the mid-sized city of Kanpur. The reference energy mean emission level (REMEL) equations have been developed by using the single event of vehicle noise with their corresponding vehicle speed. Those REMEL equations have been developed for eleven vehicle categories with 10800 vehicle samples for bituminous pavement as well as ten vehicle categories with 4390 vehicle samples for cement concrete pavement. By using these REMEL equations and different adjustment factors like volume and speed, distance, and ground section adjustment factors, the traffic noise prediction model has been developed for mid-sized Indian city. This prediction model is developed under heterogeneous traffic condition, and it is also applicable to Indian vehicle categories, road conditions, and traffic flow conditions. Traffic noise level, traffic volume, speed, and geometry data (road width, median size, distance from source or roadway to receiver, shoulder size, etc.) have been collected from fifteen mid-block locations that cover different land-use areas (residential, commercial, industrial, and silent) in the mid-sized city of Kanpur. The accuracy of the developed modified FHWA prediction model has been executed by comparing the observed noise level with a predicted noise level in terms of mean absolute percentage error and root mean square error. This observed noise level is also compared with Central Pollution Control Board (CPCB) standards at each location of different land-use areas and found that most of the locations exceeded the prescribed limits. After the development of modified FHWA Prediction model, a new Artificial Neural Network (ANN) based traffic noise model has also been developed. In this model, an Improved Artificial Neural Network-based African Vulture Optimization Algorithm (IANN-AVOA) has been used for the noise prediction for the mid-block section. The IANN-AVOA prediction technique overcomes the challenges related to weight initialization and error minimization in Artificial Neural Networks (ANNs) by leveraging the African Vulture Optimization Algorithm (AVOA). The CI-WHO feature selection method addresses the challenge of selecting relevant features from a large set of variables, ensuring optimal input for the noise prediction model. The method of data collection was the same as previous prediction model, but meteorological parameters have been added to this ANN-based model. The accuracy of this IANN-AVOA model has been executed by comparing it with other existing methods in terms of accuracy, precision, and mean square error.