AAAR 32nd Annual Conference
September 30 - October 4, 2013
Oregon Convention Center
Portland, Oregon, USA
Abstract View
Predicting Transient Particle Number Emissions from Different Blends and Feedstocks of Biodiesel Using an Artificial Neural Network
TYLER FERALIO, Britt Holmén, University of Vermont
Abstract Number: 42 Working Group: Combustion
Abstract Ultrafine particles (UFP, diameter < 100nm), which dominate diesel engine emissions, are known to cause adverse health effects in humans (aggravated asthma, decreased lung function, irregular heartbeat, and nonfatal heart attacks). Previous studies have shown changes in both particle mass (PM) and particle number (PN) emissions when diesel engines are fueled by different blends of biodiesel and petro-diesel as opposed to neat petro-diesel.
The objective of this research was to model the measured UFP emissions for a single engine running on 9 different fuels. Particle number distributions were measured from a 1.9L Volkswagen diesel engine coupled to an eddy current dynamometer in real-time for neat and blended petro-diesel, soybean oil based biodiesel, and waste cooking oil based biodiesel (B0, B10, B20, B50, and B100) with a TSI Engine Exhaust Particle Sizer (EEPS, 32 channels, 5.6-560nm range). For these experiments the engine was run through a transient drive cycle developed from on-road data collected from a Volkswagen TDi vehicle while it was operated through urban, highway, rural, and suburban driving conditions.
Data collected for modeling included fuel properties, standard onboard diagnostics (OBD-II) parameters, and PN emissions data. Principal component analysis (PCA) was performed on all of the potential model inputs to determine which parameters were the predominant factors in UFP emissions. The parameters ‘selected’ using PCA were used to develop an artificial neural network to model PN emissions from the Volkswagen engine. Preliminary data suggest that PN emissions, specifically in the 10nm range, increase as the amount of biodiesel in the fuel increases. Model results will be presented for both soy and waste cooking oil biodiesel blends.