I
GEORGE J. MULLANEY General Electric Research Laboratory, Schenectady, N. Y.
A Shock Tube I s Used to Study
...
Autoignition of Liquid Fuel Sprays An innovation to technique offers a new approach to studying an important process in compression ignition engines
IN
THE PAST, bomb experiments and continuous flow systems have provided ignition lag data for liquid fuel sprays. I n the present work, the shock tube technipe, applied recently to autoignition research with gaseous fuel mixtures (7, 72, 73), is extended to liquid fuel sprays. Investigations were made with air pressures at fuel injection ranging from 10 to 30 atm., with most of it done at about 15 atm. Air temperatues at the time of fuel injection ranged from 600" to 1000" K.
Experimental For most of the work, measurements were made with a Fastax camera and pressure transducers (Figure 1). These had a rise time of 60 microseconds, with their output signals displayed on oscilloscopes. Oscilloscope sweeps were1 initiated by the incident shock wave in the same way that fuel injection was timed. In a typical experiment, the cellulose acetate diaphragm separating high pressure driver gas in 1 from low pressure driven gas in 2 was broken manually. When the incident shock wave passed the face of a pressure transducer, the resulting output signal was fed to a variable time delay circuit, C, through amplifier A and triggering circuit B. One part of the variable delay circuit was always used (through power supply D) to energize solenoid valve 4 which abruptly dropped the pressure a t fuel nozzle 3 to a level preset by manual hydraulic pump 5. This action caused
a pulse of a known quantity of fuel to be injected into the shock tube. A second variable delay was used, for example, for triggering oscilloscope E and short duration flash lamp F. Pressure transducer G near the combustion zone measured pressure rise during the explosive reaction of the liquid fuel spray. The flash lamp marked one frame of the high speed pictures synchronizing film and pressure transducer records. High speed camera H was focused on the window section on one side of the shock tube. A built-in timing lamp exposing the film edge at 120 cycles per second permitted measurement of
I 2 3 4 5
A B C D
DRIVER SECTION DRIVEN SECTION FUEL NOZZLE SOLENOID VALVE HYDRAULIC PUMP AMPLIFIER TRIGGERING CIRCUIT TIME DELAY UNIT POWER SUPPLY
Figure 1.
frame speed. A partial mask and floodlight on the window opposite the camera permitted photographic observation of the beginning of fuel injection and evaporation of the fuel. Windows were 2 cm. high and 24 cm. long. Liquid fuel was sprayed along the longitudinal axis of the tube at about 10" total cone angle; spray boundaries were visible in the window section of the shock tube. The fuel nozzle injected about 108 mg. of fuel in the end of the shock tube when initial fuel pressure was 10,000 p.s.i. Typical fuel jet penetration distance (from high speed photography) was 12 to 14 cm. The window section of the shock tube
I I E F G H
OSCILLOSCOPE FLASH LAMP PRESSURE TRANSDUCER HIGH SPEED CAMERA
Diagram shows shock tube and auxiliary equipment
Measurements were m a d e with a high-speed camera and pressure transducers
VOL. 51, NO. 6
JUNE 1959
779
O 500
I
L
I
I
k
'
I
600 700 800 AIR TEMPERATURE,
'
l 900
,
I1 1000
OK.
Figure 2. These are typical curves of ignition lag and evaporation time obtained for iso-octane
could be replaced with a metal plate drilled and tapped for five pressure transducers along its length about 5.5 cm. apart. These were useful in studying initial rates of pressure rise due to the explosive reaction and the shock wave formation which followed under some conditions. Air used in the driven part of the shock tube was oil free and had a dew point temperature of -60' F. Liquid fuels studied included iso-octane (2,2,4-trimethylpentane), benzene, and several 99 mole yo n-paraffins. Air temperature was measured by a sound velocity technique (9) to determine its influence on induction times and the explosive reaction. Pressure transducers were calibrated at frequent intervals.
rate, necessary in determining initial air pressure, was approximately 5 p.s.i. per millisecond. One limitation involves fuel-air ratio in the shock tube experiment. Neither local nor over-all fuel-air ratios are known with any degree of certainty. Approximate over-all fuel-air ratios were calculated using the volume of shocked air through which fuel penetrated (air density was obtained from pressure and temperature) and the quantity of fuel injected. Other experiments (5, 70) showed that induction time was not strongly dependent on over-all fuel-air ratio but minimum time occurred near stoichiometric ratio. The explosive reaction following ignition lag does depend on fuel air ratio to a marked degree. Schlieren pictures after shock reflection show that the air is turbulent prior to fuel injection. Temperature at a given point in the combustion zone may thus be higher than average. If local fuel-air ratio is nearer stoichiometric than average, rapid chemical reaction, as evidenced by growth of luminosity, may begin. Correction for the effect of heat absorption of the liquid fuel spray on initial air temperature was not included because of uncertainties involved. Rough calculations indicate that local air temperature drop due to the fuel spray might be about 100" K. at initial air temperatures of 1000° K. Results and Discussion
Fuel Evaporation. Air pressure and temperature (from velocity of sound measurement) prior to fuel injection could be selected so that, for a given fuel, ignition lag was consideiably
-
20,
longer than evaporation time of the total amount injected. A diffuse light on one window of the shock tube with the camera looking through the opposite window permitted photographing fuel injection and evaporation. Fuel injection time was essentially constant (1.7 milliseconds). Evaporation time was defined as the interval from fuel injection to the time when spray was no longer visible on the film. Calculations and experiments (74,75) have shown that evaporation time was progressively less sensitive to fuel structure and air temperature above 600' K. As evaporation time (circa 2-3 millisecond) was not markedly different for a wide range of fuels, comparison of other kinetic processes would be more meani ngful Ignition Lag. Ignition lag is defined as the time from first appearance of liquid fuel in the shock tube until the beginning of luminosity on high speed pictures. Figure 2 shows typical ignition lag data for liquid fuel sprays. At the same air pressure, ignition lag for iso-octane is many times longer than that of n-hexadecane for the air temperature range studied (Figure 3). Induction time measurements above 1000' K. were discarded in many cases because small luminous points were observed in the combustion zone prior to fuel injection, and they would act as pilots thus drastically reducing ignition lag. These particles, perhaps fine dust particles or microscopic bits of cellulose acetate, persisted in spite of carefully cleaning the driven part between firings. Although fuel spray duration was 1.7 milliseconds, shorter ignition lags could be obtained because ignition began
.
I
I
I
1
Errors and limitations
For an ideal shock tube without friction, reflected shock pressure at the nozzle end of the driven tube would remain constant (except for possible disturbance from wave reflections at the hot-cold gas interface) until the rarefaction wave arrived. For a long shock tube (6-meter driver section; 10meter driven) with considerable wall friction, pressure measured at the wall near the tube end rises until the rarefaction wave reaches that point. The earlier part of this pressure rise occurs at a time rate typically twice that which occurs later. The latter part of the pressure-time curve was used for autoignition experiments. Here pressure rise
780
Figure 3. Ignition lag was much longer for isooctane than for n-hexadecane
INDUSTRIAL AND ENGINEERING CHEMISTRY
I 1
I ! . AIR PRESSURE 250 PSI. I
I
I
FUEL S P R A Y A U T O I O N I T I O N
Figure 4.
Pressure records and pictures of autoignition combustion were taken simultaneously Air temperature a t fuel injection, 1000" K.
A.
on the fuel spray side a t high initial air temperatures and pressures (Figure 4). A lower limit on ignition lag might eventually be reached because of the finite time required to vaporize the fuel and to mix fuel and oxidizer after injection. Fuel injection, evaporation, mixing, and low temperature reactions which lead to final rapid combustion can overlap each other to various degrees. In autoignition experiments with liquid fuel sprays, just as with autoignition of premixed vapors (7), instantaneous, homogeneous, explosive reaction of the whole charge is not observed even at lower initial air temperatures (600' to 700' K.) and long ignition lags (15 to 20 milliseconds). Over-all fuel-air ratio, usually 0.5 stoichiometric, was determined from the measured quantity of fuel injected and the volume of high pressure, high temperature air into which it penetrates.
n-Heptane
B.
Benzene
Differences in method of measuring ignition lag should be considered in comparing these results with bomb experiments (2, 3) on autoignition with liquid fuel sprays. Bomb tests showed longer ignition lags. Induction time for liquid sprays using a flow method (70) has been measured a t higher initial air temperatures (1050' to 1300" K.) but low pressure (around 1 atm.). Figure 3 shows a typical measurement for iso-octane. Very little is known about ignition lag for chemical reactions at still higher initial temperatures. Gaseous detonation studies ( 4 ) with gas temperatures 1500" to 1800" K. and gas pressure 1 atm. and above indicate that induction times are in the fraction of a microsecond range. Pressure effect on induction time has been studied by the shock tube technique for only one hydrocarbon, iso-
octane. When air temperature at fuel injection was 600" to 1000" K., pressure dependence was p-1.48 at 15 to 30 atm. I n compression ignition machine studies with premixed vapors (8, 77) the exponent varied markedly with the hydrocarbon fuel used. Explosive Reaction. Multistage autoignition studies (6) using premixed fuelair vapors have defined temperature limits for cool and hot flames at fixed charge densities in engine and compression ignition machine experiments. In this study pressure-time measurements of liquid spray autoignition were made at pressure-temperature combinations resulting in cool flames. An early pressure rise which may indicate cool flame reaction occurred in some experiments, but detailed study of this aspect was not made. Because a homogeneous reaction was not observed in fuel spray combustion, VOL. 51, NO. 6
JUNE 1959
781
STATION 0
I
2
3
4
5
Figure 5. Six locations were available for w a I I moun t e d pressure transducers
-
SQUARE CROSS SECTION
pressure-time histories were examined a t various points in the combustion zone. Simultaneous oscilloscope measurements were made using three of four wall-mounted pressure transducers with a maximum of six locations available near the fuel injection point (Figure 5). A survey of pressure rise as a function of time a t various stations showed that, for initial pressure and temperature, pressure rise began almost simultaneously at stations 3, 4, and 5. Station 3 corresponds to maximum penetration distance for the liquid spray shown by high speed photography. With constant air pressure at fuel injection, pressure-time curves at this station for one fuel show that as air temperature a t fuel injection is increased, rate of pressure rise becomes more severe. Increasing pressure rise rate continues with increase in initial air temperature until ignition lag becomes smaller than fuel injection and evaporation time. Because the explosive gases were not confined, rate of pressure rise and peak pressures from the rapid chemical reaction were considerably loiver than expected in a n autoignition reaction a t constant volume. Significance of the pressure measured
a t the shock tube wall is of interest. In the liquid spray combustion process, nonhomogeneous autoignition begins near the tube center. Gradually steepening compression waves may travel from the combustion zone a path of 1 to 2 cm. before reaching the pressure transducer a t the wall. Sound wave travel time from initial autoignition point to the wall is about 20 microseconds, and a very rapid growth of the luminous zone is noted by high speed photography. Therefore the pressure transducer reading is believed to indicate directly the course of the combustion process. Compression Wave Travel and Shock Wave Formation. Pressure pulse caused by explosive energy release from the fuel must originate where fuel is available (stations 3 to 5, Figure 5). T h e resulting pressure wave travels through high temperature air away from the fuel nozzle. For a given fuel and fuel-air ratio, travel time before a shock wave is formed depends on air pressure and temperature at fuel injection. However, in these experiments it was necessary that fuel injection and evaporation time be equal to or less than ignition lag. If it is not, the small
300
‘2501
/
STATION 3 AND. 5 u??
a
[ w
150
INITIAL PRESSURE 400PS.I loo- INITIAL 800’K. TEMPERATURE
STATION 0 SHOCK WAVE NEAR COMPLETE FORMATION -
TIME, MICROSECONDS
782
INDUSTRIAL AND ENGINEERING CHEMISTRY
Figure 6. Pressure-time curves for n-heptane show shock wave near complete formation as it passes station 0 (Figure 5)
amount of fuel available initially limits the extent of rapid reaction; fuel flow rate then controls energy release rate. Shock wave formation distance is more than 40 cm. when 180 mg. of n-heptane is injected into air at 800” K. and 400 p.s.i. (Figure 6). Ratio of peak pressure to initial pressure is 1.75. Stoichiometric constant volume combustion for the same conditions gives a theoretical pressure ratio of 3.5. Increasing the air temperature a t fuel injection reduces shock xvave formation time. Acknowledgment
T h e interest and assistance of G. E. Moore, C. W. Moon, and R. B. LVilkerson are gratefully acknowledged. Literature Cited
(1) Fay, J. A4.,“Fourth Symposium on Combustion,” p. 501: Williams & Wilkins. Baltimore. 1953. (2) H k n , R. W.’, Chas., J. O., Ellis, C. F., Hughes, I
