Multivariate analysis of heavy-duty diesel engine emissions

Emissions data from experiments using three heavy-duty diesel engines, two engine operation modes, ten fuel types, and three aftertreatment device types were analyzed using positive matrix factorization (PMF) and principal component analysis (PCA) to determine the applicability of these two techniques in identifying possible sources, source compositions, and source strengths. Another analysis was performed using multiple regression techniques to correlate the chemicals, particle size distributions, and mutagenicity of the particle-associated soluble organic fractions (SOF) and the semi-volatile organic fractions (XOC) with fuel properties and engine operating conditions. The results showed that both PMF and PCA results were comparable. The three-obtained possible sources, based on the knowledge of diesel engine combustion, were: (1) pyrolysis of unburned fuel hydrocarbons under high temperature condition in which fuel molecules were pyrolized to form soot particles associated with total particulate matter (TPM), (2) in-cylinder high temperature combustion, the condition that promotes the formation of oxides of nitrogen (NO_x), and (3) unburned fuel and lubricating oil, the major sources of total hydrocarbons. Source strength or emission strength of each source exhibited high variability due to the various types of engines used, operating conditions, fuels, and aftertreatment devices. The lowest emission strengths of all sources were observed in the 1995 engine operated with 10–16% exhaust gas recirculation (EGR), ultra low sulfur fuel (0.57 ppm), and a catalyzed particulate filter (CPF). The most significant influential parameters for estimating the emissions, with the exception of sulfate (SO₄) and polynuclear aromatic hydrocarbons (PAH), were engine parameters such as engine power, speed, load, and fuel injection pressure. The most significant influential parameters for estimating SO₄, PAH, particle size distribution, and mutagenicity of the SOF and XOC were fuel parameters such as olefin, sulfur content, specific gravity, 10%, and 50% distillation points. The results suggested that if the increased combustion temperature from high load could be controlled to reduce NO _x formation, increased load, fuel injection pressure, and lowered fuel sulfur content could reduce all NO_x, TPM, and HC emissions. Most results are consistent with the known knowledge of the compression-ignition engine combustion and other relevant studies concerning fuel and engine effects on the emissions, thus showing that the PMF and multiple regression analysis models have the ability to describe diesel exhaust data.