Volatile organic compound emission factors from roadside measurements

Abstract

Volatile organic compounds (VOCs) play a significant role in the generation of urban photochemical smog. In addition, some VOCs, such as benzene, are harmful to human health. In Japan, motor vehicles are the dominant source of VOCs. Therefore, it is important to determine the emission of VOCs from vehicles in order to estimate human risk and the production mechanisms of photochemical smog. In this study, we estimated emission factors with a methodology that considered the following points: (1) real-world emissions, (2) individual VOCs, (3) low vehicle speeds, (4) low investigation cost, and (5) user-friendly methodology. Samples were collected approximately 5 m from each side of National Route No. 467 in Kanagawa Prefecture, Japan. Sampling consisted of twelve 1-h sampling periods at three points on three dates: 21 February 2003 (7:00–19:00), 13 May 2003 (7:00–19:00), and 13 September 2003 (8:00–20:00). The samples were analyzed using GC/FID and GC/MS. In addition, information on vehicle types, traffic volumes, and weather conditions was collected from beside the road. Emission factors of individual VOCs were estimated from the measured data by running the CALINE4 dispersion model as an inverse model. The average speed of all vehicles was 22 km h −1; 81.3% of all vehicles were light-duty vehicles, 12.3% were heavy-duty vehicles, and 6.5% were motorcycles. We estimated the emission factors of 34 individual VOCs. The emission factors for all vehicles combined averaged over all sampling days ranged from 0.25 to 51 mg vehicle −1 km −1. The emission factors of benzene and toluene were 5.2 and 17 mg vehicle −1 km −1, respectively. In addition, the estimated emission factors were compared with those estimated from other recent studies. The emission factors for light-duty vehicles (LDVs), heavy-duty vehicles (HDVs), and motorcycles separately were also estimated by using a non-negative least squares method. However, these emission factors were found to be unreliable for the current sample size; therefore, the sample size needed to estimate reliable emission factors was calculated.