Repository logo

Environmental Hydrology

Permanent URI for this communityhttps://hdl.handle.net/10413/6541

Browse

Browsing Environmental Hydrology by Subject "Air--Pollution--Environmental aspects."

Filter results by typing the first few letters
Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    Predicting emissions using an on-road vehicle performance simulator.
    (2002) Govindasamy, Prabeshan.; Lyne, Peter William Liversedge.
    South Africa is coming under increasing pressure to conform to the rest of the world in terms of emissions regulations. The pressure is caused by a number of factors: international organisations requiring local companies to adhere to environmental conservation policies, evidence from within South Africa that efforts are being made to reduce environmental pollution in line with other countries and keeping abreast of the latest technologies that have been incorporated into vehicles to reduce emissions. In light of these problems associated with emiSSions, a study was initiated by the Department of Transport and the School of Bioresources Engineering and Environmental Hydrology at the University of Natal to investigate and develop a method of predicting emissions from a diesel engine. The main objective of this research was to incorporate this model into SimTrans in order to estimate emissions generated from a vehicle while it is travelling along specific routes in South Africa. SimTrans is a mechanistically based model, developed at the School, that simulates a vehicle travelling along a route, requiring input for the road profile and vehicle and engine specifications. After a preliminary investigation it was decided to use a neural network to predict emissions, as it provides accurate results and is more suitable for a quantitative analysis which is what was required for this study. The emissions that were predicted were NOx (Nitric oxide-NO and Nitric dioxide-N02), CO (carbon monoxide), HC (unbumt hydrocarbons) and particulates. The neural netWork was trained on emissions data obtained from an ADE 447Ti engine. These neural networks were then integrated into the existing SimTrans. Apart from the neural network, an algorithm to consider the effect of ambient conditions on the output of the engine was also included in the model. A sensitivity analysis was carried out using the model to prioritise the factors affecting emissions. Finally using the data for the ADE 447Ti engine, a trip with a Mercedes Benz 2644S-24 was simulated using different scenarios over the routes from Durban to Johannesburg and Cape Town to Johannesburg in South Africa to quantify the emissions that were generated.