STATISTICAL APPROACH TO THE DEVELOPMENT OF A MICROSCALE MODEL FOR ESTIMATING EXHAUST EMISSIONS OF A LIGHT DUTY GASOLINE VEHICLE

摘要

Pronounced interest has focused on mobile source emissions in recent decades, as the number of vehicles in use has rapidly increased, with a increase in vehicle miles traveled (VMT), especially in urban areas. According to the 2003 EPA Trends Report estimates, in 2003 on-road transportation sources emitted 36% of nitrogen oxides (NO_x), 63% of carbon monoxide (CO), and 29% of volatile organic compounds (VOCs) in the US.At the state and regional levels, transportation and air quality engineers are developing various models to help predict vehicle exhaust emissions. This research involves the development of a statistical model for estimating vehicle tailpipe emissions of NO_x. The data used in developing the model was collected under actual driving conditions, and should thus address driving dynamics well.Second-by-second emissions data was collected using a Horiba On-Board Measurement System (OBS-1300) installed in a 2007 Dodge Charger Car. The OBS-1300 collects second-by-second measurements of NO_X, hydrocarbons (HC), CO, carbon dioxide (CO_2), exhaust temperature, exhaust pressure, and vehicle position. 40 hours of emissions data were collected, including 20 hours arterial and 20 hours highway, and 20 hours during peak traffic conditions and 20 hours during off peak conditions.A preliminary model was developed, followed by diagnostic tests, Modified Levene (normality) test and constant variance test, to evaluate the validity of the dataset. Following these tests, the best subset regression, stepwise regression and backward deletion model building methods were used to develop an overall best model. SAS software was used in the model development process.A model was developed to predict NO_x in ppm as a function of vehicle velocity, power demand, time of day, and velocity/time of day interaction. The model could be used in the future to evaluate proposed emissions control strategies.