;

Journal of Scientific & Industrial Research Vol . 62, January-February 2003 , pp 1 3- 1 9

Fueling System Control and Exhaust Emissions from Natural Gas Fueled Engines Keshav S Yarde

University of M ichigan-Dearborn, Dearborn, M ichigan, USA

Received: 07 Feb, 200 1 ; Revised received: 23 July 2002

The use of natural gas as a fuel for spark ignition engines is not new. But recently there has been an increased i nterest in natural gas as a fuel for lean-burn S I engines because of the potential to reduce exhaust emissions. While lean-burn technology has been used in heavy-duty, turbocharged natural gas engines in power industry, only recently it has been considered for l ight-duty applications. The performance of lean-burn engines for light-duty vehicles is highly dependent on fuel ing system and its control. This paper investigates the impact of natural gas fueling arrangement on engine stabi l ity and exhaust emissions of lean-burn S I engines. Two types of fueling arrangements were i nvestigated: a mixer type system and an electronic port injection. Port injection system resulted in better engine performance with regard to exhaust emissions and variabi l ity in engine load than the mixer system. Catalyst performance for exhaust treatment was affected by fueling arrangement, operating conditions and exhaust gas temperature.

Introduction Natural gas has better cold-start and warm-up

characteristics as wel l as wider flammabil ity l imits than gasol ine l -4 . The extended flammabil i ty of natural gas in the lean range can be used to operate spark ignition (SI ) engines at very lean equivalence ratios. The improvement in engine thermal efficiency and the potential to lower exhaust emissions have provided i mpetus to investigate lean-burn natural gas engines for automotive application. However, two major technical issues need to be considered when contemplating operation of a S I engine on lean mixtures5.6 .

( i ) Gasoline fueled S I engines are often converted to natural gas use by using a mixer type (or similar) fueling system. Such systems may produce large variations in equivalence ratio from cylinder-to-cyl inder and from cycle-tocycle. Such variations in equi valence ratio may lead to poor engine performance and may limit engine operation.

(i i) Lean-bum operation could produce higher levels of hydrocarbons in engine exhaust. A significant portion of the engine-out hydrocarbons would be in the form of methane, an important species contributing to greenhouse effece. Although not currently regulated, there are concerns that future regulations may include l imits on tailp ipe emissions of methane.

In addition, variations In equivalence ratio could lead to an increase in engine-out emissions, including those of methane (MHC) and non-methane hydrocarbons (NMHC).

Fuel injected natural gas engines provide a tighter control on variations in air-to-fuel ratio (AIF) . This would help in extending the lean operating limit of the engine and may help to reduce hydrocarbon levels . Furthermore, if the engine could be operated at very lean equivalence ratios the levels of oxides of nitrogen (NOx) may be sufficiently low to avoid using after treatment for this species8.

The work presented in this paper examines the effect of natural gas fueling systems on SI engine performance with respect to exhaust emissions and variabi l i ty in other performance parameters. Experimental

Three different automotive SI engines were lIsed in the experimental work. Two of the three were 4-cylinder, 8-valve, 1 .6 L automotive type engines that were converted to operate on natural gas. One of these was equipped with a natural gas fuel mixer, a device similar to a gasoline carburetor. The fuel mixer (a single point device) supplied mixture to all the four cylinders. A metering valve upstream of the mixer was used to meter fuel flow rate to the engine. The metering valve controlled the average AIF ratio supplied to the engine. An on-line fuel compressor supplied pressurized fuel (90- 1 20 kPa) to the pressure regulator upstream of the

1 4 J SCI IND RES VOL 62 JAN UARY-FEBRUARY 2003

mixer. The regulator maintained outlet pressure at slightly above the atmospheric.

The other 4-cylinder engine was equipped with an asynchronous electronic port fuel injection system. In this case, two of the four injectors were energized simultaneously while the other two were out of phase in relation to the first bank of injectors. Thus the timing of fuel injectors were not synchronized with their respective cylinders. Pressurized natural gas from storage bottles was supplied to the engine fuel rail via a pressure regulator. The rail pressure was maintained at 800-900 kPa. The desired equivalence ratio was achieved by controll ing the injector pulse width.

The third engine was a modified version of an 8-cyl inder, 4.6 L, 1 6-valve automotive engine. The engine was equipped with a synchronous type fuel 1I1Jection system that had injection tlmmg synchronized with the cyl inder events. The fuel rai l , designed to handle gaseous fuel , was equipped with fuel pressure and temperature sensors. Pressurized fuel from fuel tanks was supplied to the rai l via a high-flow pressure regulator, which was heated by the engine coolant to avoid freezing. The rai l pressure was maintained at 800-900 kPa during the tests. This engine had a h igher compression ratio than the other two, as shown in Table I .

The average equivalence ratio, , of the mixture was determined by moni toring the output of a widerange exhaust oxygen (UEGO) sensor'9. The sensor was cal ibrated for the typical natural gas composition in the test cel l s . CylifJder pressure-crank angle histories of 1 00 consecutive cycles were stored and analyzed in each case for variabil ity in the combustion process and its relationship to engine performance.

Exhaust gas samples were continuously monitored for oxides of n itrogen (NOx), carbon monoxide (CO), total hydrocarbons (THC) , oxygen (02), and MHC and NMHC. The gas species were analyzed m an analyzer cart consisting of

Table I - Engine specifications Engine Cyl inders and Compress. Fueling

displacement ratio system

A 4-Cyl. 1 .6 I 8.5 : I Mixer

B 4-Cyl. 1 .6 I 9.2 : 1 Fuel inj. asynchr

C 8-Cyl. 4.6 1 1 0.6: 1 Fuel i nj . synchr.

chemiluminescent, non-di spers ive, flame ionization and paramagnetic analyzers, respectively. In addition, a gas chromatograph was used to differentiate hydrocarbons in the engine exhaust.

The engine and the dynamometer systems were ful ly i nstrumented with load cel ls , temperature and pressure sensors, flow meters for air and fuel flow rates and a data acquisition system. The pressure transducer output was integrated with the crank angle markers and analyzed for indicated mean effective pressure ( imep) , coefficient of variation (COV) in imep, partial misfi re cycles, etc. Natural Gas Composition

Unlike gasol ine, where fuel properties are better control led, natural gas has no such requirement in most of the countries. I ts composition can vary because of several factors including geograph ical location, supplier, season, etc . What is general ly control led is the energy content of the fuel . Natural gas consists mostly of methane but the methane composition in some places can vary from as high as 98 per cent to as low as 80%. This could impact engine performance and would impose l imitations on design of dedicated natural gas engines. The qual i ty and composition of natural gas used in the tests was monitored on a periodic basis . Methane levels in the fuel were between - 94 - 97 per cent, as shown in Figure I . Ethane and propane made up between I to 3 per cent of the gas composition . The remaining portion of the fuel was made up of nitrogen, carbon dioxide, water vapor and other trace gases. No sensitivity analyses were made with regard to gas composition because of the nearly uniform composi tion of the fuel 10.

Variability A gasol ine fueled S I engine, when converted to

operate on natural gas, general ly produces lower

50 tf 40 :a 30 c.. 20 e 1 0

0 r 90 92 94 96 98

Average methane content, %

Figure I - Distribution of methane in natural gas samples

1 00

V ARDE: NATURAL GAS FUELED ENGINES IS

brake mean effective pressure (bmep) at wide open throttle l -2 . The level of reduction may depend on engine des ign, type of fueling system and operating parameters. A further reduction in bmep would occur as the mixture equivalence ratio is reduced . Tests were conducted on al l three natural gas engines at various bmep values and engine speeds. In thi s paper, however, only benchmark values of bmep and engine speed are used for discussion . The NF ratios were changed from stoichiometric to lean, nearly to a point where I per cent of the monitored cycles showed partial misfire.

Figure 2 shows a comparison between the coefficient of variation in indicated mean effective pressure of the 4-cyl inder engine with mixer arrangement and the 8-cylinder fuel-injected engine. Although a portion of these variations may be attributed to normal combustion patterns in the cyl inder, it is bel ieved that the large differences observed between the two engines are associated with the type of fueling systems employed. The synchronous injection system performed much well over the mixer type system. The COY in imep" of the 8-cyl inder engine operated on natural gas was about the same as when it was operated on gasoline with values of , in both cases, between 1 .0 and 0.9 ( the gasoline version of the engine had a compression

c:. '"

.5

.5 ;;. 0 u

1 0 -,-----,.-----------,

5

0 +---r------_r--4

05 0.6 0.7 0.8 0.9 E(luivalcnce Ratio

250 kPa, Mixer 250 kPa. FI - - - - 450 k Pa, Mixer - - - - 450 kPa, FI

Figure 2 - COY in IMEP for two fuel systems at 1 500 RPM (4-Cyl. mixer and

8-Cyl. FI)

ratio of 9 .4 : I) . When the gas mixture was made leaner the two engines showed diverging trend in COV. Operation of the engine with the mixer was l imited to 0.6 equivalence ratio due to occasional partial misfire. Attempts to operate the engine at sti l l leaner equ ivalence ratio were accompanied by unsteady engine behaviour, more frequent misfires and large variations in imep. On the other hand the synchronous fuel injection system provided an acceptable control of air-to-fuel ratio and the engine could operate at equivalence ratio as lean as 0.56 at COY of less than 8 per cent.

Exhaust gas analyses from two of the four cyl inders revealed that the mixer arrangement produced cylinder-to-cyl inder variations In of as much as 1 5 per cent . Variations in $ of this magnitude would cause deterioration in engine performance, particularly when operated on lean mixtures. S imi lar analyses of the 8-cyl inder engine showed much lower average variations in cylinder to cyl inder $. The results of the four-cylinder engine with asynchronous injection system were more in line with those of the synchronous system. Exhaust Emissions

The effect of fue l ing system on NOx emissions is shown in Figure 3 at two engine loads and 1 500 rpm. The 4-cylinder engine, with its mixer type fuel ing arrangement, produced s l ightly lower levels of NOx, which can be attributed to its lower compression ratio.

Both the engines produced some variations in engine-out NOx but the mixer arrangement had larger, particularly in the 0.6 to 0.7 equivalence ratio range. When operated in the very lean range

14

1 2

"- 1 0 .c :: x >< 6 0 Z 4

2

0 . . . , 0.5 0.6 0.7 0.8 0.9

Equivalence Ratio - - . . - -4-Cyl, Mixer. 250 kPa - 0() - 4-Cyl. Mixer, 450 kPa ---+- S-Cyl, FI, 250 kPa ---'- S-Cyl, Fl. 450 kPa

Figure 3 - Engine-out NOx at 1 500 rpm

1 6 J SCI IND RES VOL 62 JANUARY-FEBRUARY 2003

... ..c :: en ,.; 0 :z

1 4

1 2

1 0

8

6

4 -

2

0 0.5 0.6 0.7 0 .8 0 .9

Equivalence Ratio

- - + - - 4-Cyl, Mixer, 250 kPa 4-Cyl, FI, 250 kPa 8-Cyl, FI, 250 kPa

Figure 4 - Engine-out Nox at 2500 rpm

(equivalence ratio - 0.6) the fuel injected engine produced very low levels of specific NOx whi le maintain ing good combustion stabi l ity.

Figure 4 shows a comparison of engine-out NOx trends at 250 kPa bmep and 2500 rpm for the three engines. As expected, the trends are very similar i rrespective of the fuel system used. The two 4-cylinder engines show almost s imi lar levels of specific NOx except for the magnitude of variations.

The 8-cyl inder engine, with i ts h igher compression ratio, produced sl ightly h igher NOx. Although it is difficult to decipher from the plots the results of the tests show the variations in specific NOx emissions with the mixer type fuel system was higher than those realized with the injection systems.

Hydrocarbon emissions were h ighly influenced by the fuel ing arrangement. The mixer system (Figure 5) produced higher levels of THC over most of the equivalence ratio. But the major difference between the two fuel systems was observed for < 0.7. Hydrocarbon levels increased very rapidly when the engine with the mixer was operated in this region . The cyl inder-to-cylinder variabi l ity in mixture quali ty produced high levels of THe. In addition the engine with the mixer arrangement experienced detectable partial misfire around - 0.6. Early flame quenching in those cylinders of the engine that experience leaner than average air-to-fuel ratio could contribute to h igh levels of THC between 0.6 and 0.7 equivalence ratio.

Simi lar quali tative patterns were seen at higher engine speed, as shown in Figure 6. S ince engines A and B were similar, except for the fuel system, the differences in THC levels between the two engines are attributed to their respect ive fuel systems.

140

... 1 20 ..c 1 00 80 U 60 :: 40 '"

20 0

0.5 0.6 0.7 0.8 0.9 Equivalence Ratio

- - + - - 4-Cyl. Mixer. 250 kPa - - 0 - - 4-Cyl. Mixer. 450 kPa -+-- 8-Cyl. A. 250 kPa 8-Cyl, A. 450 kPa

Figure 5 - Engine-out THC at 1 500 rpm

90 ----------------------,

80

70

.a 60 :: 50 U 40 30

20

1 0

I .

o t-== 0.5 0.6 0.7 0.8 0.9

Equivalence Ratio

4-Cyl, FI, 250 kPa - - + - - 4-Cyl. Mi xer, 250 kPa

J'"

8-Cyl, Fl. 250 kPa

Figure 6 - Engine-out THC at 2500 rpm

The difference in THC in the two fuel injected engines is attributed to several factors including the operating mode of the injection systems. Synchronizing injection system to produce better mixing, lower residual gas fraction and higher compression ratio helped the 8-cyl inder engine achieve lower emissions of THe. Methane made up between 68 to 85 per cent of the THC in the exhaust, while ethane and propane comprised small portions. At present, MHC emission is unregulated but there are concerns that it could be part of future automotive emission regulations. The possibi l ity of MHC emission regulations wi l l necessitate gas engine manufacturers to control its exhaust level, a point discussed later in the paper.

Carbon monoxide emissions in the gas engines were general ly low. The mixer type fuel system produced higher levels of specific CO in the very lean equivalence ratio region than the other two systems. Figure 7 shows CO emissions from the three engines

t

.>

YARDE: NATURAL GAS FUELED ENGINES 1 7

25

.. 20 " .c 1 5 , 1 0 0 U

5

0 0.5 0.6 0.7 O.S 0.9

Equivalence Ratio - - + - - 4-Cyl, Mixer, 250 kPa ----.-. SCyl, FI, 250 kPa _____ 4-Cyl, Fl, 250 kPa

Figure 7 - Engine-out CO levels at 2500 rpm

at 2500 rpm. The steep increase in CO level around -0.6 for the 4-cyl inder engines i s partly the result of poor combustion of the lean mixtures. Emissions Control

The catalytic control of the three chemical species in engine exhaust was not much affected by the fueling system. The major influence of the system was the increased levels of engine-out pol lutants caused by variabi l ity in mixture qual ity. Several three-way catalysts (TWC), specially formulated for natural gas engine exhaust, were used in the present study. The catalysts employed had ceramic and metal substrates and uti l ized Pt, Pd and Rah noble metals . Exhaust gas was sampled before and after the converter to evaluate conversion efficiency of the specIes.

The conversion of CO was successful at all operating conditions with a conversion efficiency of 90-99 per cent. In general, conversion of CO at stoichiometric or lean condition is not difficult provided the temperature in the converter is above a threshold value. NOx conversion was high near stoichiometric AIF ratio but decreased rapidly as the mixture was made leaner (Figure 8) for the synchronous injection system. Reduction of NOx in the exhaust of a lean-bum gas engine is difficult due to the oxidizing environment in the converter. Aftertreatment of NOx may require use of different approaches such as NOx traps, lean-NOx catalyst or the addition of reducing agent . But if the NOx levels are sufficiently low at lean equivalence ratios then further reduction of NOx may not be necessary. The NOx level real ized in the fuel injected gas engines at - 0.6 was about I g/kWh. Spark ignition engines that are specifically designed and optimized for

'. '.

____ . _____ o, L--- __ __ .... . _ _ 100 . ,

id' SO c 60 0 'Vi 40 .. .. c 20 0 U 0

0.5

. .

0.6

'-"'"'> - _ .. - -, ....

0.7 0.8 0.9 Equivalence Ratio

--+-- NOx conv. 250 kPa _____ NO' conv, 450 kPa - - - - THC conv, 250 kPa - - 'X' - - THC COil v, 450 kPa

Figure 8 - Conversion of Nox and THC at 1 500 rpm, 8-cylinder

natural gas could lower the level further. CO in the exhaust helps in the reduction process of NOx at stoichiometric NF ratio. If the CO level is low then the conversion of NOx would be l imited because of the reaction

2 CO + 2 NO

The CO levels in the fuel-injected engine were indeed low. High conversion efficiency of the species was achieved by biasing the fuel control system to operate at a s l ightly richer than stoichiometric equivalence ratio, typically around - 1 .03.

Conversion of hydrocarbons in the natural gas engines proved to be a chal lenging task. Oxidation of methane involves several chain init iating and chain branching reactions. Complete conversion of CH4 to CO2 and H20 requires suffic iently high temperature . h f I 1 2- 1 , S ' I even 111 t e presence 0 a cata yst . . II1ce a arge portion of THC was made up of methane, conversion efficiency of THC wou ld be low at lower gas temperatures (Figure 8) .

The exhaust gas temperatures for a given speed and load for the three test engines varied within a band, the higher temperature being for the 8-cyl inder engine (Figure 9) . A comparison of species conversion efficiency and exhaust gas temperature revealed that the l ight-off temperature (a temperature for 50 per cent conversion) for MHC in the converters was about 850 K. Exhaust gas temperature in excess of 850 K would be needed for catalytic control of MHC emission from lean-bum, automotive gas engines. A control led, fine-tuned natural gas injection system, combined with an optimized location and design of the catalyst, could address these problems.

1 8 J SCI IND RES YOL 62 JANUARY-FEBRUARY 2003

900 ---------------------- :::.:: - 800 = -; .. 700 S

VARDE: NATURAL GAS FUELED ENGINES

1 0 King S R , The impact o f natural gas compositIOn o n fuel metering and engine operational characteristics, SAE Paper 920593, 1 992.

I I Heywood J B, I nternal combustion engine fundamentals, Chap. 9 (McGraw-Hil i , Inc., New York) 1 988.

1 2 Glassman I , Combustion, 2nd ed (Academic Press, Orlando, FL, USA) 1 987.

1 3 Rudy W M, Catalytic control of exhaust emissions from CNG fueled engines, SAE Optimized Methanol and Natural Gas Fueled Vehicles Workshop, Knoxvi l le, TN, USA, 1 990.

Nomenclature

(j COY - llnep - __ lmep (Jirrep = Standard deviation in imep

c:I> = Stoichiometric A I F Actual A I F

Dr Keshav S Yarde is a Professor of Mechanical Engineering and Associate Dean of the Col/eRe of Engineering and Computer Science of the University of Michigan, Dearborn Campus, Michigan. U.S.A. Dr. Yarde has been working in the area of combustion and exhaust emissions of spark and diesel engines. high-pressure fuel sprays, conventional and aLternate fuels and combustion and flow diagnostics. He has published over 75 technical papers in various journaLs, transactions and engineering society conferences. His projects have been sponsored by the US National Science Foundation, Department of Energy. Department of Defense, US Department of Transportation, and severaL automotive and energy cOlllpanies.

1 9