Engine Combustion Research
NREL's research with actual engines allows the effects of fuel properties and fuel chemistry on engine performance and exhaust emissions to be quantified. We then use the results to validate predictive models or for insights leading to improved models describing how fuel properties impact engine performance and emissions.
Heat of Vaporization Effect on Engine Knock Resistance and Octane Appetite
NREL's single cylinder research engine experiments and engine simulations have revealed how heat of vaporization (HOV) and octane number interact to create a fuel's true knock resistance or octane index (OI). OI is a function of RON and OS:
RON – K . OS
Results show that the effect of increasing fuel HOV is to make K (an engine's octane appetite, a parameter previously thought to be a function of engine design and operating conditions only) smaller (i.e., more negative), compounding the benefit of high OS on OI at high engine loads. The research shows that for gas direct injection (GDI) engines, K is not only a function of engine design and operating conditions but is also critically affected by the fuel's HOV.
GDI engines now make up as much as 50% of new vehicle sales in the United States. A main advantage of direct injection is that it leads to evaporative cooling of the fuel-air mixture, effectively increasing a fuel's knock resistance.
For more information, see Effects of Heat of Vaporization and Octane Sensitivity on Knock-Limited Spark Ignition Engine Performance.
Ethanol Blending Impact on Gasoline Direct Inject Engine Aromatic Compound Evaporation and Particle Emissions
In a study, NREL examined the impact of ethanol and several aromatic compounds on aromatic species evolution during gasoline evaporation and its impact on particulate matter (PM) emissions from a single-cylinder gasoline direct injection engine. Dilution of the aromatic species in gasoline by ethanol blending may logically be expected to reduce PM emissions because aromatics are the primary species that form particles during gasoline combustion. However, increased evaporative cooling from the ethanol and ethanol's effect on when aromatics in the fuel droplet (or surface film or pool) evaporate can drive particles to increase.
The study's objective was to quantify these competing effects. The results show that a combination of ethanol's increased evaporative cooling and impact on aromatic compound vapor-liquid equilibrium can extend droplet lifetime and cause aromatics to evaporate later in the process than would be the case without ethanol. When fuels were burned in a DI single-cylinder engine at relatively high speed (2,500 rpm) and relatively high load, ethanol blending caused an increase in PM emissions for fuels containing low vapor pressure aromatics. At a lower speed (1,500 rpm), no statistically significant increase was observed, likely because more time was available for aromatics evaporation and mixing.
From the data, we developed a standard linear regression model, and coefficients for ethanol content, aromatic vapor pressure, and aromatic content were all significant (p < 0.05). However, the coefficient for aromatic content was roughly one order of magnitude larger than for the other variables, indicating that even in the presence of ethanol the aromatic content is the most important gasoline property parameter.
We also investigated whether nonlinear interactions between predictor variables might better explain the observed PM emissions results. We noted that since the presence of ethanol changes the droplets' distillation characteristics, some nonlinear interaction between ethanol and aromatic terms is likely required.
We used the LASSO (least absolute shrinkage and selection operator) regularized regression approach, and two features were selected based on their frequent appearance. The first was ethanol mole percent multiplied by aromatic mole percent divided by aromatic vapor pressure. The second was aromatic mole percent multiplied by aromatic yield sooting index (YSI) divided by aromatic molecular weight. The results of linear regression including only these two combined explanatory variables achieved an R2 = 0.959, substantially improved over the linear model without interaction terms (R2 = 0.839) and provides insight into how the particulate matter index formalism might be modified to include oxygenates.
To learn more about our work or explore partnership opportunities, contact Matthew Ratcliff.