The objectives of this project were to:

1. establish a science-base needed to develop accurate models for particulate matter (PM), carbon monoxide (CO), N-O emissions (NOx), total unburned hydrocarbons (UHC) and individual hazardous air pollutants (HAPs) emissions from military gas turbine engines burning alternative fuels and
2. establish a science-based methodology for selecting practical alternative fuels that minimize emissions. 

The research approach involved mutually supportive, tightly coupled experimental and computational efforts. A series of well-controlled laboratory experiments were designed and conducted that systematically progressed in complexity in a way that permitted collective analysis of the data to develop a fundamental understanding of the production of emissions from burning alternative fuels under a wide range of conditions. Experimental facilities being used in the program included: a well-stirred reactor in which chemical kinetics are the controlling process; shock tube experiments focusing on kinetics at realistic pressures; premixed and non-premixed co-flow flame experiments with dependence on kinetics and molecular diffusion at variable pressure; and two combustor test rigs - a model combustor at Penn State and the Referee Combustor at Wright Patterson Air Force Base (WPAFB), which included the composite effects of all the processes occurring in a gas turbine combustor operated at pressure including spray atomization, vaporization, diffusion, turbulent mixing, stretch, kinetics, and turbulent/chemistry interactions. Early in the program, the systematic increase in complexity was extended by the addition of emissions experiments from a T-63 engine to the original program plan. In support of the program, the Air Force initiated a study of the emissions from a T-63 engine using the test fuels designed for this program as well as JP-8 and several practical alternative fuels. Thus, the test devices in this program systematically covered key chemical and physical processes from devices in which chemical kinetics were the controlling process to a small gas turbine.

In parallel to the experimental efforts, computations were performed for each of the experiments, except the Referee Combustor and the T-63 engine, using the two computational tools. The well-stirred reactor and shock-tube are modeled using the commercial CHEMKIN/SHOCKIN software package. A unique Navier-Stokes based computational fluid dynamic (CFD) code called UNICORN was used to simulate the co-flow flames and the model combustor at Penn State. The modeling studies resulted in a hypothesis for the linear behavior of cross-plotted UHC emissions from actual engines. A new set of laboratory experiments were conducted test the hypothesis.

To achieve the objective of evaluating and developing accurate models for predicting emissions from alternative fuels, a set of alternative fuels with known chemical kinetics was required. Unfortunately, there were no practical alternative fuels with known chemical kinetics at the time the program was initiated. As a result, a set of test fuels comprised of pure components, for which chemical kinetic mechanisms exist, was used to cover a range of compositions expected in future alternative fuels. Each of the fuels consisted of a binary mixture of n-dodecane and a second hydrocarbon representative of one of the four main classes of hydrocarbons that are present in alternative fuels – normal alkanes, iso-alkanes, cycloalkanes and aromatics. Normal and iso-alkanes are present in alternate fuels produced from Fischer-Tropsch (FT) processes, and cyclo-alkanes are present in alternative fuels produced from coal by liquefaction processes. The compounds selected to represent each of the major classes of alkanes present in alternative fuels were: n-heptane, iso-octane, and methyl-cyclohexane. Because the m-xylene/n-dodecane fuel is a simple surrogate for jet fuel, the combustion studies provide direct experimental evidence of the effects of changing jet fuel composition on emissions. The three purely paraffinic fuels, n-heptane/n-dodecane, iso-octane/n-dodecane, and methyl-cyclohexane/n-dodecane, represented possible alternative fuels created via a Fischer-Tropsch or coal liquefaction processes. The fuel set included a fifth fuel, n-dodecane/n-hexadecane, to investigate the effect of fuel volatility on emissions.

The experiments conducted with the T-63 engine yielded a wealth of results on fuel effects on emissions as engine power was varied. The results from the study established a basis for comparison between the results from the laboratory devices and an actual gas turbine engine. Experiments were conducted with the four binary fuel mixtures designed to study chemical effects of alternative fuels on emissions and covered the power range from idle to 85% of maximum power

CO2 emissions were lower by 1 to 3% for the purely paraffinic fuels compared to the JP-8 surrogate; this reduction correlates with the H/C ratio of the fuels. CO emissions were lower by approximately 10% for the purely paraffinic fuels compared to m-xylene fuel. Within measurement uncertainty, the NOx emissions were the same for the four test fuels. PM emissions for the three paraffinic fuel mixtures were substantially lower than the PM emissions for the m-xylene/n-dodecane fuel, which is a surrogate for JP-8.

The overall trends in NOx and PM are consistent with trends found in studies of prototype alternative fuels in military aircraft engines.

The lowest level of PM emissions was observed for the test fuel comprised of normal alkanes, n-heptane/n-dodecane almost 10 times less than the m-xylene/n-dodecane fuel at 85% power. The iso-octane and methyl-cyclohexane fuels had similar PM emissions that were approximately twice as high as the PM from the n-heptane fuel. (These same relative trends in PM were observed in the co-flow flame experiments.) For all four test fuels, total UHC emissions decreased by more than an order of magnitude between idle and 85% power. The UHC emissions did not show any consistent trend among the four test fuels. Speciation of the UHC emissions showed that the major components comprising the UHC emissions were similar among the fuels and included: ethylene, formaldehyde, acetaldehyde, and methane. However, speciation of the UHC emissions also showed some direct effects of fuel composition on hazardous air pollutants (HAPs). The m-xylene and methyl-cyclohexane fuels produced approximately twice as much benzene as the n-heptane and iso-octane fuels. (The higher production of benzene was also observed in the well-stirred reactor studies.) Cross-plots of the speciated UHC emissions showed the same linear behavior that was observed in the field studies of aircraft emissions, although slopes may be different.

Another key finding from the detailed speciation of the UHC emissions was that a substantial fraction of the UHC emissions at idle and low power was from unreacted fuel. The presence of unreacted fuel in the UHC means that the composition of any alternative fuel will have a substantial, direct effect on UHC emissions when engines are operated at idle and near-idle power levels.

The findings from the experimental and modeling work conducted in this program and reported in 10 peer review journals and 30 conference papers support the following statements when comparing the binary mixtures of paraffinic compounds representative of alternate fuels to the m-xylene fuel, which is a surrogate for JP-8:

• Effects on NOx, CO and total hydrocarbons were small, generally 10% or less.
• Effect on PM emissions was substantial, with reductions in PM emissions by one order of magnitude.
• Low molecular weight components of UHC emissions were not strongly affected. 
• Some high molecular weight UHC species increased in concentration depending on the fuel composition, e.g., benzene emissions were higher for the m-xylene fuel and the methyl-cyclohexane fuel.
• Unique UHC species were present for some of the fuels, e.g., iso-butene for the iso-octane fuel and toluene for the m-xylene fuel.

The experimental results on the impact of fuel volatility while holding chemical composition fixed support the following statements:

• Decreasing fuel volatility resulted in substantially increased PM emissions for the Referee Combustor operating at condition near engine idle although these PM emissions are likely due to increased UHC (or fuel) emissions rather than soot.
• Decreasing fuel volatility did not affect soot emissions for a model combustor operating at rich fuel-air ratios corresponding to high power engine operation. Due to the unique design of this combustor, however, this conclusion should be considered as tentative pending tests in other types of combustors.

In addition to these specific findings on alternative fuel effects, the program led to three major insights that advance the state of the art understanding of engine emissions:

• The experimental work in the T-63 engine and the Referee Combustor demonstrated that substantial fractions (30 to 50%) of the UHC emissions can be traced to unreacted fuel. This finding means that the composition of any fuel directly affects the UHC emissions, including emissions of HAPs that are present in the fuel.
• The experimental and modeling studies conducted to understand the linear behavior in UHC emissions at low power led to the important new insight that UHC emissions form in a gas turbine combustor when reacting premixed fuel-air mixtures are quenched. Thus the emissions are not determined by reaction kinetics alone, but rather by coupling of chemical kinetics, fluid mechanics, and flame extinction processes. Furthermore, multiple quenching mechanisms can lead to UHC emissions, including mixing to very lean fuel-air ratios, quenching due to flame propagation into regions of low fuel-air ratio, and flame stretch. Accurate modeling of the emissions from a gas turbine combustor will require that each of the multiple quenching mechanisms be modelled accurately. The current program included modeling and simulations of quenching via mixing.
• The experimental and modeling work related to chemical effects on PAH and soot formation led to substantial progress toward resolving the longstanding issue within the combustion community related to modeling the effect of aromatic species on soot concentration and the spatial distribution of soot. Enhanced PAH reactions in the chemical kinetic mechanism and improvements in the soot model to track particle size distribution led to substantially improved ability to predict soot concentrations and distributions in all of the co-flow flames.

In addition to these fundamental insights the program also made the following important fundamental contributions:

• The kinetic model with detailed PAH chemistry that was developed in this program is a major contribution from this work because it was a critical element in the advancement of the soot modeling capability. Development of this model was not a part of the original program design, but became necessary when kinetic model development within the combustion community did not evolve as quickly as anticipated. In addition, the modeling studies have demonstrated that this mechanism captures many key trends in gas-phase emissions from gas-turbine engines.
• The data base from the various experiments made using the same set of test fuels is a unique resource for researchers who are simulating gas turbine combustion because of the breadth of chemical and physical processes encompassed across the experimental devices and because of the wide range of experimental conditions used.
• To our knowledge, no research teams other than this team and the SERDP-funded team led by Venkat Raman (WP-2151) are attempting to simulate speciated emissions from gas turbine engines. So the work is unique in the field.
• The key findings and contributions from this program have substantially extended the fundamental knowledge required to evaluate a science-based methodology for selecting alternative fuels based on their emissions characteristics.