AAAR 33rd Annual Conference
October 20 - October 24, 2014
Rosen Shingle Creek
Orlando, Florida, USA
Abstract View
Quantifying Variability in Molecular Markers Used for Vehicle Source Profiles: Effects on PM Source Apportionment Results
ALBERT A. PRESTO, Andrew Hix, Christopher Hennigan, Allen Robinson, Carnegie Mellon University
Abstract Number: 156 Working Group: Source Apportionment
Abstract PM source apportionment results obtained with the chemical mass balance (CMB) model can be highly sensitive to the choice of source profiles. Source profiles for gasoline and diesel vehicle emissions have often been constructed from relatively small vehicle fleets (N<10), and in some cases published source profiles represent the emissions of a single vehicle. Source apportionment studies cope with the variety of available source profiles by considering averages of published profiles, or by using graphical tools such as ratio/ratio plots to select source profiles that bound ambient data. Additionally, many published source profiles are dated, and may not be representative of emissions from modern vehicles. For example, the majority of the vehicles tested in the large Kansas City Vehicle Emissions Study and Gasoline/Diesel split study were produced >10 years ago.
This study addresses source profiles for gasoline vehicle emissions in two ways. First, we present emissions measurements of molecular markers such as hopanes and polycyclic aromatic hydrocarbons (PAHs) for a fleet of 65 in-use vehicles spanning model years 1987-2012. This dataset represents the first large-scale measurements of molecular marker emissions and marker/OC ratios from Tier 2/LEV-2 (e.g., model years 2003-2013) vehicles for use in source apportionment.
While the vehicle fleet tested here is larger than many previous studies, it is still insufficient to describe the variability of molecular marker emissions across the entire on-road vehicle fleet in the U.S. We therefore used bootstrap resampling to generate probability distributions of marker/OC ratios. These probability distributions are used as inputs to a Monte Carlo implementation of CMB in order to determine the impact of gasoline source profile variability on CMB predictions.