Measurement of Black Carbon and Particle Number Emission Factors from Individual Heavy-Duty Trucks

TitleMeasurement of Black Carbon and Particle Number Emission Factors from Individual Heavy-Duty Trucks
Publication TypeJournal Article
Year of Publication2009
AuthorsGeorge A Ban-Weiss, Melissa M Lunden, Thomas W Kirchstetter, Robert A Harley
JournalEnvironmental Science and Technology
Volume43
1419
Issue5
Pagination1419-1424
Date Published03/2009
Abstract

Emission factors for black carbon (BC) and particle number (PN) were measured from 226 individual heavy-duty (HD) diesel trucks driving through a 1-km-long California highway tunnel in August 2006. Emission factors were based on concurrent increases in BC, PN, and CO2 concentrations (measured at 1 Hz) that corresponded to the passage of individual HD trucks. The distributions of BC and PN emission factors from individual HD trucks are skewed, meaning that a large fraction of pollution comes from a small fraction of the in-use vehicle fleet. The highest-emitting 10% of trucks were responsible for ∼40% of total BC and PN emissions from all HD trucks. BC emissions were log-normally distributed with a mean emission factor of 1.7 g kg−1 and maximum values of ∼10 g kg−1. Corresponding values for PN emission factors were 4.7 × 1015 and 4 × 1016 # kg−1. There was minimal overlap among high-emitters of these two pollutants: only 1 of the 226 HD trucks measured was found to be among the highest 10% for both BC and PN. Monte Carlo resampling of the distribution of BC emission factors observed in this study revealed that uncertainties (1σ) in extrapolating from a random sample of n HD trucks to a population mean emission factor ranged from ± 43% for n = 10 to ± 8% for n = 300, illustrating the importance of vehicle sample sizes in emissions studies. When n = 10, sample means are more likely to be biased due to misrepresentation of high-emitters. As vehicles become cleaner on average in the future, skewness of the emissions distributions will increase, and thus sample sizes needed to extrapolate reliably from a subset of vehicles to the entire in-use vehicle fleet will become more of a challenge.

DOI10.1021/es8021039