Environ. Sci. Technol. 1902, 26, 1581-1588
Effect of Fuel Structure on Emissions from a Spark-Ignited Engine. 2. Naphthene and Aromatic Fuels Edward W. Kaiser,' Waiter 0. Siegl, David F. Cotton, and Richard W. Anderson
Ford Motor Company, Scientific Research Laboratories, MD 3083,Dearborn, Michigan 48121-2053 A single-cylinder, production-type engine has been r u n at four operating conditions on two naphthene fuels (cyclohexane and methylcyclohexane), an aromatic blend (containing xylenes and ethylbenzene), and a fully blended gasoline. Engine-out hydrocarbon (HC) emissions for the naphthenes contain 12% 1,3-butadiene and 5% benzene. This represents the first observation of substantial benzene emission from a nonaromatic fuel. Butadiene emissions are large in comparison to the gasoline blend, and caution should be exercised when adding large quantities of the cyclohexanes to a fully blended fuel if butadiene emission is a concern. With the exception of benzene, the HC species emitted can be rationalized on the basis of the known high-temperature chemistry of cyclohexane. The aromatic blend produces significant quantities of both benzene (2-5%) and toluene ( 3 4 % ) in the engine-out emissions. This contributes to the observed enrichment (relative to the fuel composition) of these species in the exhaust of engines operating on gasolines containing heavier alkylated benzenes. Introduction
In a recent publication ( I ) , we demonstrated the critical role that fuel structure plays in determining both the total hydrocarbon (HC) emissions and the distribution of species in the engine-out exhaust from spark-ignited engines. During these previous experiments, a single-cylinderengine with no catalytic after-treatment was fueled with singlecomponent fuels. The fuels included gaseous straightchain alkanes, branched-chain liquid alkanes, toluene, and a five-component liquid-fuel blend. The current experiments extend those studies to include two cyclic alkanes (cyclohexane and methylcyclohexane),a mixture of heavier aromatics containing xylenes plus ethylbenzene, and a full-blend gasoline. The synthetic fuel blend in ref 1 contained primarily isooctane and toluene. The exhaust mole fractions of both unburned fuel and direct combustion products from this fuel mixture (e.g., isobutene and heptenes from isooctane or benzene and benzaldehyde from toluene) could be predicted to within 40% based on the emissions from the single-component fuels and the known fuel blend composition. Therefore, these limited data suggest that emissions studies of single-component fuels can be useful in estimating exhaust species mole fractions from fuel blends. The data from our engine, which is not equipped with a catalyst, represent engine-out rather than tailpipe emissions. However, a considerable fraction of the total urban driving cycle emissions from a catalyst-equipped vehicle exits prior to catalyst light-off, and these pre-light-off emissions should be essentially engine-out emissions. For this reason, we believe that measurements of engine-out emissions from single-component fuels can provide important information regarding the effect of fuel structure on tailpipe emissions from a driving cycle test. Finally, although a single-cylinder engine is used in these experiments, the head and piston have geometries typical of modern multicylinder engines and should provide useful information concerning production vehicles. 0013-936X/92/0926-1581$03.00/0
Cyclic alkanes (naphthenes) were chosen for testing because commercial and emission certification gasolines can contain significant naphthene content (2,3)and because cyclohexane, in particular, has been added in substantial amounts (18%) to test fuels (3). Because the distribution of many of the major exhaust species from a fuel blend can be roughly estimated from emission data obtained with single-component fuels, it is very important to obtain data from as many classes of single-component fuels that are present in gasolines as possible. Recent experiments in our laboratory using a pulse-flame combustor ( 4 ) showed that cyclohexane produces substantial exhaust emissions of benzene and smaller amounts of butadiene. Because these species are of considerable health concern even in small amounts (5-7)and the sources of butadiene emissions are not well established (€9,engine tests of naphthene fuels are important. Our naphthene data can also be interpreted in light of known high-temperature reaction mechanisms of cyclohexane (9),shedding additional light on engine emission processes. Benzene is typically enriched in engine exhaust relative to i h percentage in the initial gasoline fuel (7,10-13). In previous experiments (I), we demonstrated that pure toluene fuel generates a substantial amount of benzene emissions (8% of total identified HC emissions). This conversion contributes to the benzene enrichment in the exhaust gas since gasoline normally contains appreciable toluene. The magnitude of benzene emission from toluene is similar to that observed in earlier studies of fuel mixtures containing toluene (IO). It is possible that a similar enrichment in exhaust toluene occurs because of the presence of more highly substituted benzenes in gasoline fuels. To explore enrichment of the exhaust benzene and toluene further, emissions data from an aromatic fuel containing only xylenes and ethylbenzene were also obtained. Finally, the single-cylinder engine was operated on a gasoline fuel to provide baseline data for comparison with the singlecomponent fuel experiments. Experiment
The single-cylinder engine has been described in detail (I). As in the earlier experiments, the baseline engine condition was a fuel-air equivalence ratio (a) of 0.9 (fuel-lean), MBT spark timing [defined as the spark advance required to yield peak cylinder pressure at 13" (crank angle) after top dead center], 1500 rpm, 90 "C coolant temperature, 73 "C oil temperature, and a load of 3.8 bar IMEP. No exhaust gas recirculation was added to the intake mixture in these experiments. This steady-state condition is typical of a midspeed, midload cruise. In addition, experiments were carried out at either 2500 rpm, MBT-12" spark advance (retarded timing), or CP = 1.15 while the other conditions were kept as defined for baseline. The liquid fuels were introduced through an injector located in the intake port. As in the previous experiments, the start of injection was at 660 crank angle degrees (0 CAD is top dead center of compression), and the fuel injection continued for approximately 45 crank angle degrees. Thus, injection occurred during the compression stroke with the intake valve closed.
0 1992 American Chemical Society
Environ. Sci. Technol., Vol. 26, No. 8, 1992
1581
Table 11. Measured Species in Exhaust at Baseline Condition"
Table I. Measured Species in Gasoline Fuel" % carbon
% carbon
species
in fuel
species
in fuel
isobutane n-butane 1,3-butadiene n-pentane isopentane methylpentanes dimethylpentanes
0.5 2.1 0.0 7.6 11.3 3.3 3.6
n-hexane isooctane benzene toluene ethylbenzene xylenes trimethylbenzenes
1.3 10.0 1.8 14.5 1.1 6.3 4.0
"Percentage of total carbon in the liquid gasoline fuel determined by GC. Only the major classes of species and others of interest in this experiment are included.
The gas chromatographic (GC) analysis was identical to that described earlier (1). Only hydrocarbons were measured; oxygenated organics were not included in the analysis. The sampling technique, in which exhaust gas was withdrawn into an evacuated Pyrex flask covered with foil to exclude light and diluted 1O:l with nitrogen, was also identical with two exceptions: (1)the line connecting the sample flask to the exhaust pipe was heated to prevent water condensation and to minimize sample loss; (2) the exhaust sample pressure in the flask prior to dilution was reduced to 70 Torr. These changes provided 10-20% better GC sample recovery than was observed earlier using the heavier liquid fuels when compared to the total HC emissions measured by a hot FID instrument, which was connected to the exhaust pipe by a heated sample line. However, the distribution of the exhaust species changed by less than 10% when the GC sample line was heated for either isooctane or gasoline fuel. Selected exhaust samples from cyclohexane fuel were analyzed several times over a period of 18 h to assess their stability. None of the species with the exception of 1,3butadiene and cyclopentadiene showed a concentration change. Butadiene remained constant for approximately 1h and then decreased at a rate of 5-7 % /h in the fuel-lean samples but showed no decrease in the CP = 1.15 (fuel-rich) sample. Because both the NO, and the O2in the exhaust decreased substantially under rich conditions, we also ran a test at CP = 1.0. This condition reduces the O2in the engine exhaust by a factor of 4 relative to the lean condition while maintaining the same NO, concentration. At = 1.0, the butadiene loss rate was reduced to 1.5%/h, indicating that while NO, may play a role, the O2concentration seems important to the loss process. The cyclopentadiene loss rate in samples obtained during lean operation was much faster (25%/h following a 1-h period during which little decrease was visible). Rich operation reduced the cyclopentadiene loss rate to 2%/h, while at stoichiometric fuel-air ratio, a 10% /h loss rate occurred. Loss of butadiene (14, 15) and cyclopentadiene (16) in exhaust samples has been observed previously. To minimize this effect, analyses were carried out within approximately 1-2 h after sampling, and while a significant loss of cyclopentadiene might still occur, the butadiene results should be accurate to within 10%. For both of these species, however, the results under fuel-lean operation must be regarded as lower limits. High-purity solvent-grade cyclohexane and methylcyclohexane were used in these experiments. The carbon content of the aromatic blend consisted of 69.5% m-and p-xylene (not separated in our GC analysis), 10.8% oxylene, and 19.6% ethylbenzene. Any impurities were present at less than 0.05%. The gasoline was a blend having 91 research octane and a T,,distillation point of 157 "C. The carbon content of this fuel consists of 69% 1582
Environ. Scl. Technol., Vol. 26, No. 8, 1992
exhaust species
fuel c-CGH12 c-C7H14 i-CsHls
methane 15 acetylene 60 ethylene 260 ethane 4 allene 2 5 propyne propylene 58 1,3-butadiene 135 isobutene 2 n-butane n-pentane isopentane 1,3-pentadiene 9 cyclopentene 11 cyclopentadiene 22 2-methyl-l,3-butadiene cyclohexane 335 cyclohexene 69 methylcyclohexane 1-methylcyclohexene isooctane benzene 54 to1uene ethylbenzene xylenes styrene sum of GCb (ppm C1) 1115 unburned fuele (%) 30 1185 total HCd (hot FID) 2295 NO," (ppm) 730 CO' (ppm) co2e(%) 11.6 2.0 0 2 " (%) MBT spark timing' 20
20 53 211 10 5 9 102 120 8
25 40 70 18 23
D
E
15 40 20
7
9 145
21
180 10 472
25 6 25
20 44 170 10
5
28
100 19 55 90 11
22
12
5 58 540 23 60 12
1500 36 1485 2260 660 11.5 2.2
23
970 5
90 55 48 70 198 313 20 1250 83 60 9 2080 1950 1650 47 80 2630 2424 610 565 595 11.0 13.5 11.8 2.2
27
2.1
26
2.1 24
"These data are species mole fractions (ppm C1) in undiluted exhaust samples. ppm Cl(species i ) = ppm(i) X carbon no.(& Only species with mole fractions in excess of 1% of the total in one of the fuels are included in the table. Isooctane, 2,2,4-trimethylpentane. Fuel D, aromatic fuel (80.3% xylenes, 19.6% ethylbenzene). Fuel E, gasoline (see text). Baseline condition: 1500 rpm, MBT, = 0.9,3.8 bar IMEP. *Sum of the carbon content of all GC peaks contained in the chromatogram, assuming that the carbon response of all species is the same. Includes peaks not included in table. Contribution of unburned fuel to sum of GC (for gasoline see text). dTotal HC emissions (ppm CJ measured by the hot FID emission instrument connected to the exhaust pipe by a heated sample line. 'Mole fractions measured by engine emission console instruments. /Spark advance for MBT operation (degrees before top dead center).
saturates, 30% aromatics, and less than 1% olefins; the fractions of the total carbon number represented by selected major components as determined by GC analysis of the gasoline are presented in Table I. Prior to running each new fuel, the engine fuel system was purged with that fuel to remove any contamination from the previous sample.
Total Emissions Results Table I1 presents the mole fractions of the major species determined by GC analyses of single experiments for all four fuels at the baseline operating condition. Repeat baseline experiments were carried out for cyclohexane (c-C6H12),methylcyclohexane (c-C7HI4),and gasoline. The mole fractions from the two experiments agreed to better than *lo% from the mean of the experiments for all species. A measurement using isooctane fuel is also presented to provide a cross comparison with the earlier data obtained using this engine (1). This table contains the GC mole fractions of the major HC species in the engine-out exhaust, as well as the engine emission console measurements of total HC (using the hot FID), NO,, CO,
C02,and O2 (all corrected to wet exhaust condition). The MBT spark timing for each fuel is included for reference. Also included in the table is the percentage contribution of unburned fuel to the total emissions determined by the GC. Because gasoline contains many components, we estimated the fuel contribution by subtracting the s u m of the major products (CH4,C2H2,C&, C2H6, C3H6,allene, propyne, isobutene, and 1,3-butadiene) from the GC total. This estimate is strictly an upper limit but is useful for comparing the effect of engine operating conditions on the unburned fuel contribution to the exhaust for gasoline. The hot FID emissions measurement of total exhaust HC for isooctane in Table I1 (1970 ppm C,) is similar to two measurements obtained 1year earlier (1950 and 1680 ppm C,). When the heated sample line is used, the sum of the HC mole fractions determined by GC agrees well (f5%) with the hot FID value for the alkane fuels. This is an improvement over the observed recovery (80%)using the cold sample line with isoodane in our earlier study (I). However, the distribution of HC species in the exhaust gas from isooctane fuel is identical to within 10% with the results obtained with a cool sample line in our earlier experiments. The typical recovery for gasoline was observed to range from 80 to 90%. Recovery for the aromatic fuel blend was somewhat poorer even with the heated line (65-80%), probably because of loss of these lower vapor pressure species even in the heated sampling lines. The trends in HC emissions with fuel composition are similar to those observed in ref 1. Total HC emissions for the aromatic blend are the largest and are close to those observed for toluene. Gasoline fuel emissions are lower than either isooctane or the aromatic blend, agreeing with measurements made using a synthetic multicomponent mixture ( I ) and with data from a multicylinder engine (17). Cyclohexane and methylcyclohexane HC emissions are lower, falling near to those observed for isopentane (I). Table I11 presents data obtained during fuel-rich operation, while Table IV contains the results for both a retarded spark timing (MBT-12O) and a higher engine speed (2500 rpm). Rich operation results in higher total HC emissions, with most of the increase arising from higher concentrations of unburned fuel, methane, and acetylene. This agrees with several other fuel-rich engine experiments (see ref 1and citations therein). The methane and acetylene mole fractions increase by factors of 10 and 5, respectively, while the unburned fuel contribution increases by a factor of 2-3 for the naphthene fuels and an estimated 1.7 for gasoline. For all fuels, the mole fractions of the olefinic partial combustion products typically change by much smaller amounts and can even decrease when @ is increased from 0.9 to 1.15. Both retarded spark and high-speed operations reduce the total HC emissions by 2 0 4 0 % depending upon the fuel. At these two engine conditions, the fractional contribution of unburned fuel to the exhaust emissions is reduced, particularly for the naphthenes, falling to 12% for cyclohexane. The magnitude of reduction in unburned fuel mole fraction depends upon the fuel used and is greatest in these experiments for the naphthenes (a reduction of a factor of 3 4 , intermediate for gasoline, and least for the aromatic fuel. This effect is indicative of increased postcombustion burnup of crevice or oil-film stored fuel. Increased burnup is expected for retarded spark and high-speed operations because these conditions produce hotter late-cycle cylinder and exhaust gas temperatures (18). A t all conditions, the cyclohexane HC emissions are approximately 25% smaller than those observed for me-
Table 111. Measured Species in Exhaust for @ = 1 . W exhaust species
fuel D C-CBHlZ c-C~HII
125 167 methane 267 268 acetylene 268 219 ethylene 21 14 ethane 3 6 allene 4 8 propme 66 43 propylene 113 86 1,s-butadiene 7 isobutene n-butane n-pentane isopentane 7 14 1,3-pentadiene 1 3 cyclopentene 15 17 cyclopentadiene 24 2-methyl-l,3-butadiene 16 1100 cyclohexane 33 47 cyclohexene 1388 methylcyclohexane 18 1-methylcyclohexene isooctane 24 33 benzene 4 toluene ethylbenzene xylenes styrene 2120 2490 sum of GCb (ppm C,) 52 56 unburned fuelc (%) 1960 2456 total HCd (hot FID) 725 730 NO,‘ (ppm) 4.22 4.28 COe (%) 10.2 10.1 coze (%) 0.15 0.18 0 z e (%I 17 18 MBT spark timing’
117 188 30 3 5
E 200 221 202 26 7 10 114 17 82 29 96 146
161 156 119 337 194 32 481 147 2097 28 9 3388 2665 76
