Investigation of Atomization, Mixing and Pollutant Emissions for a Microturbine Engine

Christopher Bolszos researchbegan with a strong interest instudying thermodynamics,fluid dynamics, and combus-tion, and their roles in energygeneration and the environ-ment. This interest led him tohis research on reducing pollu-tant emission levels in turbinegenerators under the guidanceof Professor Samuelsen.Christopher, now a graduatestudent at UCI, hopes toexpand on this work and pro-duce results that can be appliedto advanced liquid fueled gasturbine systems. He describeshis undergraduate researchexperience as a very reward-ing opportunity to venture intothe forefront of engineering,a venture he hopes to buildupon in the future.

The majority of electric and motive power production in the worldtoday uses combustion to transform the chemical energy bound inthe fuel into thermal energy that can drive a piston, turn a turbine,or produce steam. Combustion is also responsible for the majorityof the air pollutant and global climate change gases emitted into thetroposphere. The reduction of pollutant impact from combustionis closely tied to the preparation of fuel and the mixing of the fuel

with air. This paper provides a basic understanding of the role of fuel air mixing ina liquid-fueled gas turbine engine and represents a major accomplishment by anundergraduate in the conduct of energy research.

K e y T e r m s

Airblast Plain Jet Atomizer Distributed Power

Generation

Equivalence Ratio Evaporation Time Lean Premixed

Prevaporized Combustion

Microturbine Generator Sauter Mean Diameter Weber Number

Investigation of Atomization,Mixing and Pollutant Emissionsfor a Microturbine Engine

Christopher D. BolszoAerospace Engineering and Mechanical Engineering

Small gas turbine engines, referred to as a microturbine generators (MTGs), pro-duce up to 500kW of electrical power and are ideal for distributed power gener-ation applications. By generating power where it is used (e.g., a commercial officebuilding), using MTGs can increase the reliability and quality of the electrical powerand allow the waste heat to be used to meet other energy requirements at the site.Combining electrical power generation with waste heat recovery, referred to as com-bined heat and power, substantially increases the overall efficiency of the unit andsignificantly reduces the mass emission of air pollutants per kW-hr of power gener-ated when compared to traditional reciprocating backup devices. This projectaddresses this issue experimentally by characterizing the pollutant emissions from aliquid fueled MTG (Capstone model C30), and establishing the extent to which thefuel preparation processes and operating parameters affect air pollutant emissions.The results reveal that the MTG selected produces low levels of pollutants comparedto other technologies currently used. Furthermore, the research critically examinesthe steps associated with preparing the liquid fuel for combustion to identify furtherpotential emissions reductions, demonstrates that emissions can be further reduced,and identifies a strategy to achieve the reduction.

1 3T H E U C I U N D E R G R A D U A T E R E S E A R C H J O U R N A L

A u t h o r

A b s t r a c t

F a c u l t y M e n t o r

Scott SamuelsenHenry Samueli School of Engineering

Introduct ion

As the demand for energy increases, the ability of tradition-al power generation strategies and power distribution infras-tructure is being challenged. In the traditional approach,central station power generation, power is generated at afew large plants and then transmitted over miles of wiresknown as a grid. Under an emerging concept, distributedpower generation, power is produced at the locations whereit is ultimately used, including commercial office buildings,schools, hospitals, and industrial plants. Although thisapproach reduces the loss of power across the grid, it alsoresults in the emission of air pollutants in the immediatevicinity of people. Small gas turbine engines, called micro-turbine generators (MTGs), which produce up to 500kW ofelectricity, provide an option for distributed power genera-tion (Borbeley et al., 2001; Smith, 2001).

Distributed generation is already accepted as a form ofbackup power. Backup generators are viewed strictly asinsurance against power failure with the hope they willnever be operated. As a result, it is difficult to justify largecapital investment in these devices. Reciprocating dieselengines are a common choice, since their large scale pro-duction allows for a lower cost. While these devices canprovide power when needed, they generate substantialamounts of pollution, require considerable maintenanceand are often unreliable.

MTGs offer an attractive option for reducing pollutantemissions in comparison to reciprocating systems (Miller,2004). Additionally, MTGs can produce power continuous-ly, increasing the reliability of the electrical power to thefacility. Furthermore, these devices can operate on liquidbyproducts to produce power from waste, an increasingvalue as the cost of natural gas rises.

A major issue with MTGs is that, because they are operatedcontinuously around people, their emissions must be mini-mized. The goals of this research are to characterize the airpollutant emissions from an MTG and to establish theextent to which the emissions are minimized by varying keyparameters (load, combustor fuel to air ratio, atomization ofthe liquid fuel, fuel and air mixing, and reaction tempera-ture) known to affect emission performance. The main pol-lutants of concern are nitrogen oxides (NOx) and carbonmonoxide (CO).

A 30 kW Capstone Model C30 MTG, a widely used MTG,was selected for this study (Figure 1). This model, like otherturbine engines, incorporates an injection method to pre-

vaporize and premix fuel and air prior to combustion.Within each fuel injector (three in the C30), the fuel dropletsare formed by an air-blast plain jet atomizer, which useshigh velocity gas (typically air) to break up a sheet or columnof liquid into a fine mist (Figure 2).

The air used for combustion enters primarily through fouropenings surrounding the air-blast nozzle, improving mix-ing before ignition. Downstream of the atomizer is theswirler, a collection of vertical slots that aids in mixing and

14 T h e U C I U n d e r g r a d u a t e R e s e a r c h J o u r n a l

I N V E S T I G A T I O N O F A T O M I Z A T I O N , M I X I N G A N D P O L L U T A N T E M I S S I O N S F O R A M I C R O T U R B I N E E N G I N E

Figure 1Capstone C30 Microturbine

Figure 2Capstone C30 liquid fired injector with enlarged cross-sectionshowing fuel spray

distributing the spray by inducing a helical spin to enteringair (Figure 3). Both of these air streams play an importantrole in the preparation of the fuel/air mixture and are there-fore expected to influence the formation of pollutants dur-ing combustion. The ratio of fuel to air governs the degreeof combustion heat release and, in turn, determines thefraction of pollutants formed in the exhaust products.

The fuel preparation process involves the atomization ofliquid fuel, followed by vaporization and mixing. Air-liquidatomization is used to enhance both the production of finedroplets and mixing (Georjon and Reitz, 1999; Lefebvre,1998). The rate of vaporization and mixing is determinedlargely by the amount of available energy capable of per-forming work (enthalpy) and the fluid dynamics of the

combustion air and swirler air. This study explores the roleof each step in the fuel preparation process and the subse-quent emissions performance to assure that the designresults in ultra low emission of air pollutants, and to identi-fy the degree and manner in which emissions can potential-ly be reduced even further.

Exper iment

Experiments were conducted with an MTG Test Rig, aMixing Test Rig, and an Atomization Test Rig. Diesel Fuel#2 (DF-2) was used in all the rigs.

MTG Test RigExhaust emissions were measured with a Horiba PG-250emissions analyzer via an extractive sample probe centeredat the exit plane of the exhaust stack. Emissions were char-acterized for 50 to 100 percent load operating conditions.The accuracy of the NOx and CO measurements is 0.25parts per million volume dry (ppmvd) and 2 ppmvd,respectively. The PG-250 was spanned and zeroed beforeeach measurement and the span was verified at the end ofeach test run. A special injector was fabricated to allow theconnection of pressure transducers and thermocouplersnear the air-blast nozzle exit of the injector during opera-tion.

Two sets of injectors were used in the MTG to study theeffect of the fuel-air ratio on emissions (Figure 4). BecauseNOx and CO emissions are a function of temperature, it isimportant to assess the sensitivity of the emissions to thereaction temperature. The increase in hole size is based onan anticipated reduction of NOx associated with lower tem-peratures produced by reducing the injector fuel-air ratio.

15T H E U C I U N D E R G R A D U A T E R E S E A R C H J O U R N A L

C h r i s t o p h e r D . B o l s z o

Figure 3Capstone C30 fuel injector and spray phenomena. The arrowsshow the manner in which high temperature combustion air entersthe injector and vaporizes, further atomizing and mixing the fuel.The premixer length, Lpremixer, is defined as the injector tubelength the atomized fuel droplets have to vaporize and mix with airbefore they ignite. The mean droplet residence time, tresidence, andevaporation time, tvap, in the injector tube are the two key param-eters that determine whether prevaporization is attained.

Figure 4Baseline and modified C30 injectors

Mixing Test RigThe mixing test rig, a full-scale 1/3 section of the C30engine, simulates the air flow characteristics in the engine(Figure 5). The design of this test rig was developed andverified by engine measurements and computational fluiddynamics (CFD) analysis using CFD-ACE+ software.

Gaseous methane was used as fuel in a computer aidedmodel of the Capstone injector and surrounding flowgeometry, which is shown in Figure 6. The inlet conditionsfor the air entering the annulus shown in Figure 6 were

determined by measure-ments made in the MTGtest stand, and verified thefully developed flowassumption used in themodel. The fuel inlet condi-tion was set as a constantmass flow at the injectorcenterline. The outlet ofthe geometry was set toconstant pressure.

In the mixing test rig con-figuration, a Horiba FIA-236 hydrocarbon analyzer(flame ionization detector)was used to map fuel con-centration over the exitplane of the injector.Rather than using liquidfuel for these tests, naturalgas was flowed through the

fuel line. Natural gas was used as a trace in the air flow toidentify the airs role in mixing. Results obtained with thistest rig allowed mixing performance of different injectorconfigurations and/or operating conditions to be quantita-tively compared.

Atomization Test RigThe single injector test rig was designed to characterize C30injector atomization performance. Tests were conducted onthe fuel injector atomization independent of the fuel-airpremixer tube (Figure 7).

The premixer was removed from the setup to isolate therole of the air-blast on the atomizer performance (Figure7a). A Malvern laser diffraction system with a 100 mm focallength lens was used to measure and quantify the character-istics of the fuel spray (Figure 7b). A calibration reticule(Laser Electro-Device Ltd Model RR-50.0-2.0-0.030) wasused, and measurements of the spray were taken at oneinjector tube diameter (2.67 cm) from the injector exit plane.Further traversing of the flow was performed to ensuremeasured spray characteristics remained invariant through-out and a representation of the overall flow was established.

Results and Discussion

EmissionsEmission measurements were obtained for the C30 MTGoperated on DF-2 with both sets of injectors. Results versusload setting are presented for the baseline injector as the



Figure 5Single injector mixing test rig with hydrocarbon measurementprobe

Figure 6CFD geometry of C30 injectorand surrounding flow path;shown with downward orienta-tion and with air flow enteringalong far right side of arc

Figure 7(a) Injector without premixer tube and (b) injector in test configu-ration

open points in Figures 8 and 9 for NOx and CO emissions,respectively. The results are presented as corrected to a con-stant level of O2 (15%) to ensure that differences in theamount of air in the exhaust stream is not simply diluting theconcentration of pollutants. For the baseline injectors, COemissions decrease with load and are below 10 ppmvd cor-rected at 15% O2 for 50100% load operation. NOx emis-sions increase with load and are approximately 20 ppmvd at15% O2 at maximum power output. The results demonstratethat the MTG produces orders of magnitude less NOx thanthe current reciprocating backup generation devices.

Results are also presented for the modified injector, shownas filled symbols in Figures 8 and 9. The results for themodified injector are higher than those produced by thebaseline, yet are still low compared to those produced by atypical reciprocating engine. The modified MTG producesabout 2 g of NOx and CO per kilowatt hour (kW-hr) at fullload, compared to modern reciprocating diesel generatorsthat easily produce 20 g of NOx and CO per kW-hr (Miller,2004).

For both injectors, the NOx and CO emissions levels werebelow 30 ppm for all loads studied. This illustrates that,even with a significant modification in combustion condi-tions (by decreasing the equivalence ratio by 12%), theMTG still demonstrates superb emissions performancewhen compared with the 1000+ ppm emissions of thereciprocating engines currently in use as backup generators.

The results also suggest that the fuel-air mixture producedby the injectors is not fully premixed and prevaporized. Ifthe fuel and air were ideally mixed, the larger air holes in themodified injectors would result in reduced reaction temper-atures and, therefore, reduced NOx emissions. To illustrate,the expected behavior with assumed premixed/prevapor-ized operation is represented by the NOx and CO trendlines derived from MTG measurements (Figure 10). Sincethe actual NOx emissions increased with more injector airflow, it can be concluded that the assumption of pre-mixed/prevaporized is not valid.

Mixing PerformanceIdeal premixed/prevaporized conditions were not attained;therefore, the level of premixing achieved is of interest. Tohelp establish the degree to which the injectors were per-forming, the baseline emissions measurements were com-pared to results determined for a near perfect, prevaporizedand premixed combustion system (Figure 11) (Leonard andStegmaier, 1994). This comparison presents NOx as a func-tion of average reaction temperature for the baseline results



Figure 8NOx emissions versus load with modified injector

Figure 9CO emissions versus load with modified injector

Figure 10Expected NOx and CO behavior for two injectors assuming idealprevaporization and premixing

along with the generalized data obtained from previousmeasured emissions using high sulfur diesel (HSD) and nat-ural gas. The HSD line represents a theoretical NOx limitfor a heavier distillate diesel fuel compared to DF-2 withperfect premixing and is used as a worst case scenario interms of emissions (Lee et al., 2001). According to theresults, it appears that mixing in the C30 liquid fuel injectorcan be improved (Figure 11). Note that emission results fora natural gas fired at 60 kW MTG operating at similar con-ditions and with a similar fuel injection approach is able toachieve excellent premixing (Phi et al., 2004). By plottingthese results (Figure 11), its position on the line affirms theopportunity to improve mixing. As a result, an opportunityto improve the emissions performance through an improve-ment in the premixing of the injector is apparent.

The modified injector showed increased NOx emissions,meaning that either the atomization or mixing performanceof the modified injector was inferior to that of the baseline.Any NOx reductions potentially achieved by reducing theaverage fuel/air ratio are offset by inferior mixing.

To further assess the mixing characteristics of each injector,tests were conducted at atmospheric conditions using thesingle injector mixing test rig. Natural gas was flowedthrough the liquid fuel passage as fuel and a flame ioniza-tion detector (FID) based hydrocarbon analyzer was used tomap the injector exit mixing profile by measuring themethane concentration at discrete grid positions at theinjector exit plane.

The single injector mixing measurements were carried outfor both the baseline and modified injectors. The pressuredrop of the engine was used to scale the fuel and total airflow rates for atmospheric conditions. Normalized concen-tration profiles derived from the experiments are presented

in Figure 12 with orange being higher methane concentra-tion. To quantify this variation, the coefficient of variationof the exit plane fuel/air ratio was calculated as 11.6% and13.6% for the baseline and modified injectors, respectively.The results show that variation in mixing is 20% higher thanthe baseline. This helps verify and quantify the inferior mix-ing performance of the modified injector. Variation in thedegree of mixing would lead to regions of higher local tem-peratures and, therefore, higher NOx emissions.

A computational fluid dynamic (CFD) model1 was devel-oped to further assess the mixing performance of the fuelinjectors. The assumptions for this model were thoseexplained earlier in the experiments-mixing test rig sectionof this paper. The results show that, by enlarging the airhole size and maintaining system flow conditions, addition-



Figure 11Effects of nonuniform fuel and air mixing on NOx formation

1. CFD-Ace+ (ESI) was used to compute the flowfield and mixing. A Reynolds AveragedNavier Stokes (RANS) approach was used, with standard k-e turbulence model, and a turbu-lent Schmidt number of 0.5 (Qing, 2005).

Figure 12Normalized exit plane concentration profiles: (A) Baseline injector,(B) Modified injector

al mixing interactions with fuel are required for optimalpremixing. Therefore, in agreement with the experimentalmixing rig results, a less uniform concentration of fuel andair is produced at the injector exit plane when compared tothe baseline injector (Phi et al., 2004).

It is also possible that the mixing is impacting CO emissionsby cooling the reaction in the relatively fuel lean region nearthe injector wall in the upper left portion of Figure 12b. Theinferior fuel distribution seems to be more directly associat-ed with the enlarged holes, but a slight misalignment of thefuel tube within the atomizing air passage may have someimpact on fuel distribution. However, testing different injec-tors of the same type with different alignments has shownto produce repeatable mixture distributions and emissions.

Measurement Results with Instrumented InjectorAlthough the mixing tests suggest significant differences inthe performance of the two injectors, the tests were con-ducted with natural gas, resulting in near instantaneousvaporization. In a normal engine, liquidnot natural gasis injected. As a result, further analyses of the role of atom-ization and vaporization were carried out.

Atomization tests were conducted under standard condi-tions. Since these conditions differ from the elevated tem-peratures and pressures inside the engine, the operation ofthe injector had to be scaled to match the key parametersthat control atomization. Injector conditions during engineoperation were measured to establish the scaling. In air-blastatomization, pressure, not temperature, plays the dominantrole in determining droplet formation and overall spraybehavior. Therefore, atomization studies can be performedat atmospheric conditions by matching pressure drops inthe nozzle. Table 1 shows the associated MTG pressuresand pressure drops for full power conditions. Equation 1defines the pressure drop of combustion and atomizationairflows. The pressure drop of the combustion airflow isdetermined from the calculated total mass flow of the C30and the measured injector area.

Atomization and Evaporation StudyMeasurements of the mean fuel drop size (Sauter meandiameter) produced by the airblasted plain jet atomizer forthe baseline injector were obtained using laser diffractionunder scaled engine conditions using DF-2. All of theexperimental data were obtained using air supplied via anair-compressor at standard conditions. Test conditions werederived as follows:

1. Determine combustion air flow using the pressure dropthat was obtained from engine measurement as a scal-ing parameter.

2. Calculate fuel flow rate matching fuel to combustion airratio at the maximum load.

3. Determine atomization flow using the pressure dropthat was obtained from engine measurement.

Atomization air varied such that the atomization air to liq-uid mass flow ratio (ALR) spanned from 0.2 to 0.7. Theresults are plotted as the Sauter Mean Diameter (SMD) ver-sus ALR and as in Figure 13 along with a comparison of anempirical expression (Equation 2) that was developed todescribe drop sizes for a similar fuel injector design (Rizkand Lefebvre, 1984).

In Equation 2, D32 is the Sauter Mean Diameter, do is the liq-uid discharge opening diameter, is the liquid surface ten-sion, UR is the relative coflowing velocity of the two streams,A is density of air, L is the liquid density, and L is the liq-



Pressure of combustor 0.335 MPaPressure drop of combustion air flow 5.6 %

Pressure of atomization air inlet 0.344 MPa

Pressure drop atomization air flow 2.6 %

Table 1Air flow pressure drops in MTG

100= pressure of combustorpressure drop

(%)Ppressure drop (1)

Figure 13Influence of ALR on SMD from calculation using Equation 2 andmeasured values using laser diffraction

+ 0.15

+

ALRdoL

L 11

0.52

+

=

ALRdUdD

oRAo

110.480.40.4

232

(2)

uid viscosity. The error bars illustrate the lack of achieve-ment of stable atomization, which is undesired since thepresence of large droplets causes variations in local equiva-lence ratio and, therefore, emissions. Given that Equation 2was derived for ALRs between 2 and 8, finding good agree-ment between the equation and the experimental data forALRs as low as 0.30 provides a validation for the expansionof the range of ALRs in which Equation 2 can be applied.

Based on the agreement of Equation 2 to the experimentaldata shown in Figure 13, it is reasonable to apply Equation2 to estimate the atomizers performance at actual engineconditions. If Equation 2 is applied using the properties atactual engine conditions, the D32 at full load is estimated tobe 50 m. Equation 2 can also be used to explore how toalter the injector operation to change the droplet sizes pro-duced. The insight from using the above equation for theC30 atomizer is illustrated in Figure 14.

If the evaporation rate is insufficient, liquid fuel enters theprimary zone, potentially contributing to higher NOx emis-sions. Subsequently, an analysis of the vaporization charac-teristics was carried out to explore whether the atomizationquality is sufficient to lead to full prevaporization.

The analysis was performed using an effective evaporationconstant, eff (Lefebvre, 1998). This concept simplified cal-culations of the evaporation characteristics of fuel dropletsby allowing the determination of eff algebraically fromflow properties and operating conditions, and avoids solv-ing systems of nonlinear partial differential equations. Theaverage droplet lifetime was determined by the D2 Law withthe adaptation of a modified evaporation constant, eff,where D0 is the initial drop size, which is equal to the SMDin this case, eff is the effective evaporation constant, and teis the evaporation time (Equation 3).

Figure 15 shows the results of this analysis in the context ofC30 emissions and suggests that droplet lifetime is greaterthan residence time within the C30 injector. This indicatesthat droplets will persist in the combustor, which is notdesired for low emissions. This could show why measuredemissions are well above the levels associated with well-pre-mixed systems (Figure 11).

The role of the air flowing through the combustion airholes on atomization also needs to be considered. While it

is expected that the atomizing air is the primary flow usedin air-blast atomization of the liquid fuel, the interaction ofthe combustion air jets with the spray droplets also plays arole. By computing the Weber Numberthe ratio of iner-tial to liquid surface tension forces (Equation 4)for theatomization and combustion air at full operating load, it isfound that the atomization air plays a dominant role in fuelbreakup.

In Equation 4, A is the density of air, UR is the relativevelocity of the coflowing streams, D32 is the Sauter MeanDiameter, and is the liquid surface tension. It is assumedthat the combustion air flows parallel with DF-2 at the noz-zle exit.



Figure 14ALR and UR effect on SMD for C30 airblast atomizer. The arrowillustrates finer drop sizes at full load condition.

Figure 15Effect of SMD on drop lifetime

322DU

We RA= (4)

effe

Dt

20

= (3)

Based on the Weber numbers, shown in Table 2, the com-bustion air aids atomization. Since the modified injector hasa lower Weber number, larger droplet sizes (and thereforelonger vaporization times) are a significant factor in theincrease in NOx emissions observed for the modified injec-tor.

Conclusion

NOx and CO emissions for a commercial Capstone C30 liq-uid fuel fired microturbine generator have been character-ized. The sensitivity of the emissions performance to load,combustor fuel to air ratio, atomization of the liquid fuel,fuel and air mixing, and reaction temperature were estab-lished.

The results affirm the ability of a gas turbine engine toachieve low emissions of both NOx and CO. The enginehas superior emissions performance compared to recipro-cating engine technology currently being used for backuppower.

The research also clarifies the distinction between emissionsreduction strategies for a liquid fueled versus a well-pre-mixed gasous fueled turbine engine, and demonstrates thatair pollutant emissions can be further reduced. Whileincreasing the airflow to the injector can, in a simplifiedmodel, reduce reaction temperatures and NOx emissions,the observed increase in emissions indicates that the keyassumptions of complete vaporization and perfect mixingbehind this model do not hold. This result is significant inthat it indicates that the reasonable and intuitive approachof improving the emissions by simply reducing the injectorequivalence ratio and ensuring sufficient mixing do notapply in the current system. This is attributed to the com-plexity of the phenomena involved in the preparation ofthe fuel and air prior to combustion (i.e., atomization,vaporization, and mixing). Measured mixing performanceof the two injectors reveals that NOx correlates with injec-tor mixing performance. As a result, strategies that furtherimprove fuel-air mixing while maintaining a reasonable levelof atomization can potentially reduce emissions further.

Acknowledgements

A special thanks is due to my research advisors, ProfessorScott Samuelsen and Dr. Vince McDonell, whose uniqueinsight, guidance and encouragement have contributed tothe research presented herein. I would also like to acknowl-edge Josh Mauzey of the UCI Combustion Laboratory andSosuke Nakamura of Mitsubishi Heavy Industries, whosepatience, persistence, and professional experience has pro-vided insight and perspective to the research. I would like toacknowledge the California Energy Commission (Contract500-00-020) and Capstone Turbine Corporation for theirfunding and support. Finally, I would like to thank my fam-ily for their unconditional love and support.

Works Ci ted

Bittlinger, G., and N. Brehn. High Pressure Combustion Test ofLean Premixed Prevaporized (LPP) Modules in an AxiallyStaged Combustor Using a Multisector Rig. 14th ISABEConference, Florence, Italy Sept. (1999).

Borbeley, A.M., and J.F. Kreider. Distributed Generation: ThePower Paradigm for the New Millennium. Boca Raton: CRCPress, 2001.

Georjon, T.L., and R.D. Reitz. Atomization and Sprays 9 (1999):23154.

Lee, J.C.Y., P.C. Malte, and M.A. Benjamin. Low NOxCombustion for Liquid Fuels: Atmospheric PressureExperiments Using a Staged Prevaporizer-Premixer. Paper2001-GT-0081, Turbo Expo. June 2001, New Orleans, LA.

Lefebvre, A.H. Atomization and Sprays. Hemisphere Publishing,1998.

---. Gas Turbine Combustion. 2nd ed. Philadelphia: Taylor andFrancis, 1999.

Leonard, G., and J. Stegmaier. Development of anAeroderivative Gas Turbine Dry Low Emission CombustionSystem. Journal of Engineering Gas Turbines and Power116 (1994): 54246.

Lorenzetto, G.E., and A.H. Lefebvre. Measurements of DropSize on a Plain-Jet Airblast Atomizer. American Institute ofAeronautics and Astronautics Journal 15.7 (1977): 100610.



Atomization Air Combustion Air

Baseline 30 7.6Modified 30 5.8

Table 2Weber numbers for baseline and modified injectors using 50micron SMD

Miller, Wayne. Presentation. Criteria of Emission from BackupGenerators California Energy Commission, October 6,2004, Sacramento, California.

Phi, V.M., J.L. Mauzey, V.G. McDonell, and G.S. Samuelsen. FuelInjection and Emissions Characteristics of a CommercialMicroturbine Generator. Paper GT-2004-54039, TurboExpo. June 2004, Vienna, Austria.

Qing, Wang. Evaluation of a Gas Turbine Premixer Performance.Thesis. University of California, Irvine, (2005).

Richards, G.A., M.M. McMillian, R.S. Gemmen, W.A. Rogers, andS.R. Cully. Issues for Low-Emissions, Fuel Flexible PowerSystems. Progress in Energy and Combustion Science 27(2001): 14169.

Rizk, N.K., and A.H. Lefebvre. Spray Characteristics of Plain-JetAirblast Atomizers. Journal of Engineering Gas Turbinesand Power 106 (1984): 63944.

Smith, Douglas J. Distributed Generation: The Power toChoose. Power Engineering 105.3 (2001): 32.



Documents

Investigation of Atomization, Mixing and Pollutant Emissions for a Microturbine Engine