Proceedings of the Combustion Institute, Volume 29, 2002/pp.

735–741

IN SITU FUEL CONCENTRATION MEASUREMENT NEAR SPARK PLUG

IN SPARK-IGNITION ENGINES BY 3.39 lm INFRARED LASER
ABSORPTION METHOD

EIJI TOMITA,1 NOBUYUKI KAWAHARA,1 SADAMI YOSHIYAMA,1 AKIHIKO KAKUHO,2 TERUYUKI ITOH2
and YOSHISUKE HAMAMOTO3
1
Department of Mechanical Engineering
Okayama University
Tsushima-Naka 3, Okayama 700-8530, Japan
2
Powertrain and Environment Research Laboratory,
Nissan Research Center, Nissan Motor Co. Ltd.
Natsushima-cho 1, Kanagawa 237-8523, Japan
3
Faculty of Environmental Engineering,
The University of Kitakyushu, Hibikino 1-1
Wakamatsu-ku, Kitakyushu 808-0135, Japan

Recently, improving the thermal efficiency and reducing the exhaust emissions of internal combustion
engines have become crucial. To this end, it is important to determine the fuel concentration in the vicinity
of the spark plug near the spark timing, because initial combustion affects the subsequent main combustion
in spark-ignition engines. In this study, a fiber optic system linked to an optical sensor installed in the spark
plug, by means of which light can pass through the combustion chamber, was developed to determine the
fuel concentration near the spark plug using an IR absorption method. A He-Ne laser with a wavelength
of 3.39 lm that coincides with the absorption line of hydrocarbons was used as a light source. By exchanging
an ordinary spark plug for this spark plug with the optical sensor, successive measurement of fuel concen-
tration before the spark timing near the spark plug was performed in a port-injection spark-ignition engine
fueled with iso-octane under the firing condition. The effects of pressure and temperature on the molar
absorption coefficient of fuel were clarified in advance. The air/fuel ratio averaged for many cycles near
the spark plug with this optical system agreed with that measured with a buret, which represented the
mean value averaged over a protracted period. Next, this sensor was applied to determine the air/fuel ratio
quantitatively in a direct-injection gasoline engine. As a result, it was clarified that the air/fuel ratio and
its standard deviation near the spark plug have a strong relationship to stable engine operation.

Introduction method for measuring the concentration of hydro-

carbons by means of a special valve in the spark plug
Recently, there have been intensive efforts to de- [1]. Collings et al. developed a fast-response flame
velop thermal efficiency and reduce exhaust emis- ionization detector to measure the cycle-to-cycle hy-
sions of internal combustion engines. Important to drocarbon concentration in the cylinder [2–4]. The
these efforts is an understanding of combustion phe- gas-sampling methods have often been used in com-
nomena including the correct mixture-formation mercial engines, though some degree of time delay
process in an actual engine. Cycle-to-cycle fluctua- occurs because of the length of the sampling pipe.
tion of combustion, especially in lean-burn spark- On the other hand, laser diagnostics such as Ray-
ignition engines, can deteriorate drivability. It is con- leigh scattering, laser induced fluorescence (LIF),
sidered that this is because the fuel concentration and Raman scattering methods have been devel-
near the spark plug fluctuates from cycle to cycle, oped. The application of these theoretical back-
affecting the fluctuation of the initial stage of com- grounds and experimental techniques to combustion
bustion. Therefore, it is very important to detect fuel research has been presented by Eckbreth [5]. Dur-
concentration near the spark plug at the spark tim- ing the past decade, Zhao and Ladommatos have
ing, particularly in direct-injection spark-ignition en- summarized results on many applications to mixture
gines. Currently, there are several methods by which formation in spark-ignition engines [6]. The Ray-
to measure the fuel concentration in spark-ignition leigh scattering method has been used to obtain con-
engines. centration measurement in a diesel engine [7], mix-
Gas sampling is considered to be the most reliable ture-formation process [8], and mixture strength
method. Matsui et al. developed a gas-sampling near the spark plug [9]. Raman scattering has been

735
736 ENERGY PRODUCTION—Combustion in Engines

used to measure the air/fuel ratio, residual gas ratio, with almost the same optical arrangement as that
and mixture-formation processes [10–14]. In partic- used by Koenig and Hall, and investigated fuel con-
ular, LIF measurement has been widely used be- centration history near the spark plug under the non-
cause the LIF signal is relatively strong and provides firing condition [34]. Iiyama et al. developed a di-
two-dimensional fuel concentration information at rect-injection gasoline engine and showed some
an indicated timing [15–21]. Individual studies have examples of concentration measurement near the
tried to determine the real fuel concentration with spark plug using a special optical sensor in the spark
various fluorescence dopants corresponding to the plug [35]. This optical sensor has a pair of sapphire
wavelength of the laser. However, these optical rods, the tips of which are cut at an angle of p/4
methods demand changes in the design of the com- radian so that a laser light can pass through the com-
bustion chamber of these engines because of instal- bustion chamber in an engine.
lation of optical windows. This means that these In this study, the optical sensor developed in the
methods cannot be applied to commercial engines. previous study [35] was revised. An optical system
Further, although these optical methods enable us for an in situ absorption method with a 3.39 lm laser
to obtain the fuel concentration distribution, its ac- for concentration measurement in a spark-ignition
curacy is relatively lower than that of the sampling engine was developed and tested under the firing
condition with iso-octane as fuel. The molar absorp-
method.
tion coefficient of iso-octane was determined under
With regard to an in situ optical method of ob-
various pressure and temperature conditions in ad-
taining fuel concentration near the spark plug, air/ vance. This sensor was also applied to determine the
fuel ratio fluctuations were determined with an op- air/fuel ratio in a direct-injection gasoline engine,
tical probe in a spark plug in spark-ignition engines and the relation between the air/fuel ratio near the
by detecting CH and C2 radicals in the burned gas spark plug and the combustion performance was in-
[22], and both temperature and air/fuel ratio were vestigated while changing injection timing.
measured with a special optical probe [23]. Witze et
al. developed a sensor having optical fibers, with Method of Infrared Absorption Measurement
which to detect the flame kernel and to determine
the rate of flame growth with the direction of flame Principle of Infrared Absorption
propagation [24]. Optical fibers have often been When light with only one wavelength such as a
used by researchers for the detection of flame and laser is used and the intensity of the light, I, decays
chemiluminescent emissions [25,26]. to the value of I0 through a gas along the measure-
The absorption of ultraviolet light was used by ment length, L, the transmissivity, I/I0, is expressed
Morishima and Asai for measuring mixture strength from Lambert-Beer’s law as follows:
near the spark plug in a spark-ignition engine [27].
log(I/I0) ⳱ ⳮ ecL (1)
IR absorption methods have often been applied to
detect hydrocarbons after the extraction of sample where e and c denote the molar absorption coeffi-
gas from the object of measurement. However, in cient and the molar concentration of the gas that
situ measurement is desirable for detecting the local absorbs the light, respectively. When the measure-
concentration of the fuel. Yoshiyama et al. showed ment length, L, is constant, the concentration can be
the possibility of concentration measurement near determined by measuring the transmissivity. When
the spark plug using an in situ IR absorption method hydrocarbons exist in the test section, the light is
with Lambert-Beer’s law [28]. A laser light with a absorbed strongly at about 3.4 lm. This strong ab-
sorbance is in conjunction with a C–H bond in the
wavelength of 3.39 lm was passed directly through
molecule structure of the hydrocarbon. When the
windows near a spark plug without optical fibers in
local concentration of the fuel in spark-ignition en-
a test engine. Propane was used as the fuel, and the gines is measured, it is desirable that the absorption
effects of pressure and temperature on the molar coefficient be comparatively strong, because the
absorption coefficient were obtained in advance. measurement length is limited.
The in situ local mixing ratio of methane was deter- The absorption coefficient depends on pressure
mined with fibers in a non-premixed burner by the and temperature in the test region. However, the
3.39 lm absorption method [29]. Recently, some ef- absorption coefficient of fuels such as iso-octane or
forts using an optical sensor in the spark plug and gasoline for a specified wavelength is not known at
an IR absorption method have been made by Koe- the present stage. Therefore, the absorption coeffi-
nig, Hall, and others [30–33]. They developed a cient of the fuel must be measured, in advance, up
measurement system using a lamp with IR wave- to the conditions in engines.
lengths and a band-pass filter of relatively wide
range. Various kinds of fuel concentration measure- Molar Absorption Coefficient
ments were performed in a spark-ignition engine In this study, the value of the molar absorption
even under the firing condition. This system requires coefficient was determined using a constant-volume
calibration for each optical arrangement. Kawamura vessel while changing pressure and temperature. A
et al. used a 3.39 lm laser instead of an IR lamp He-Ne laser with a wavelength of 3.392 lm was used
 SPARK-PLUG CONCENTRATION MEASUREMENT BY 3.39 lm LASER 737

Fig. 1. Molar absorption coefficient of iso-octane under

changing pressure and temperature. (Double circles de-
note the data obtained by Tsuboi et al. [36]).

Fig. 3. Optical sensor in a specially designed spark plug.

The molar absorption coefficient, determined

from equation 1, increases with increasing tempera-
ture and decreases with increasing pressure, as
shown in Fig.1. The degrees of increase and de-
crease are less under conditions of higher pressure
and temperature. For iso-octane, there are few data;
those by Tsuboi et al. are indicated with a double
circle [36]. They measured the absorption coeffi-
cient from 300 to 1200 K using a shock tube, and
the absorption coefficients of iso-octane and several
other hydrocarbons were almost constant below 800 K,
and the data on iso-octane showed great fluctuation
from 1.8 to 2.8 ⳯ 105 cm2/mol, as shown in Fig. 1.
On the other hand, the standard deviation of each
value obtained in this study is about 2%. These re-
Fig. 2. Schematic diagram of experimental apparatus for sults confirm the reliability of our method.
a spark-ignition engine and fiber optic system from lasers
to detectors, (BPF, band-pass-filter; HM, half mirror; M, Fuel Concentration near the Spark Plug in
mirror). Spark-Ignition Engines
Optical Arrangement
Figure 2 shows the optical arrangement and sche-
as incident light. The light, which is made intermit-
matic diagram of the experimental apparatus for a
tent with a chopper, passes through an optical win-
spark-ignition engine. A light source with a wave-
dow into the test region in the vessel and is absorbed length of 3.392 lm was introduced into an optical
by the hydrocarbon therein. The light passes fiber through which IR light could pass. The light
through another window to reach an IR detector was guided to a spark plug through a connector. The
(PbSe) with constant temperature. After nitrogen spark plug had a pair of rods set 5 mm apart from
had been supplied to the vessel up to atmospheric each other and whose tips were cut at an angle of
pressure, iso-octane was aspirated with a micro- p/4 radian so that the light passed through one of
syringe. Nitrogen was added to a certain pressure, the rods and went back to the other rod, as indicated
and temperature was controlled with heaters on the by the dashed line in Fig. 3. The measurement
side and bottom of the vessel. Iso-octane was vapor- length was 5 mm, whereas that of the older sensor
ized and mixed very well with nitrogen by stirring. was 3 mm [35]. Therefore, the sensitivity of concen-
The experimental conditions of temperature and tration measurement increased because of greater
pressure were independently set between 100 and absorption. The laser light was detected with an IR
2000 kPa and between 300 and 600 K, respectively. detector (InSb) through a band-pass filter. The IR
738 ENERGY PRODUCTION—Combustion in Engines

The compression ratio was 9.2:1. The iso-octane fuel

was supplied from the injector equipped with an in-
take port. The molar absorption coefficient is rela-
tively high, which means that the absorption is strong
and leads to higher accuracy. The ignition timing was
14⬚ before top-dead center (BTDC). One of the
spark plugs was exchanged for the special spark plug
equipped with the optical arrangement. Assuming
that liquid droplets exist in the gas, light at 3.392 lm
detected with a sensor decreases due to the scatter-
ing of the droplets. The light is not only absorbed
but also scattered under such conditions. When light
with another wavelength, one that is not absorbed
by hydrocarbons, is used, only the effect of scattering
can be extracted. The HITRAN database [37] indi-
cates that a wavelength of 632.8 nm has no absorp-
Fig. 4. Transmissivity of light from a laser with a wave-
tion line of hydrocarbons. The transmissivity of the
length of 3.392 lm with combustion near idling condition.
light from a visible He-Ne laser of this wavelength
showed almost the same level around unity with the
crank angle near idling operation. It was found that
the existence of liquid droplets could not be ob-
served. As a result, it was confirmed that the 632.8 nm
signal indicated no droplets, or that scattering from
the droplets present was negligible under this con-
dition.
When the engine was operated at 850 rpm under
the near-idling condition, the transmissivity was ob-
tained, as shown in Fig. 4, for a crank angle from
50⬚BTDC to 20⬚ after top-dead center (ATDC). The
gas was compressed when the crank angle reached
near the top-dead center (TDC) due to the piston
motion, and the laser light was absorbed because of
the increase in molar density, as presented in Fig. 5.
After the flame passed through the measurement re-
gion, there were no hydrocarbons because the un-
burned gas changed into burned gas. The standard
Fig. 5. Molar concentration of fuel near spark timing deviation shows a larger value between 5⬚BTDC and
obtained from the data in Fig. 4. 10⬚ATDC. This means that each value changed to
unity rapidly after the pass of the flame, though the
arrival time of the flames is different from cycle to
detector signal was recorded at every crank angle.
cycle, so that the averaged data of transmissivity in
According to Lambert-Beer’s law presented in equa-
Fig. 4 returns to unity gradually. Information on fuel
tion 1, the fuel concentration near the spark plug concentration near the spark plug at the spark timing
was determined when the transmittance signal was is very useful in engine design in terms of perfor-
measured. In this study, the data were obtained suc- mance. After a flame front has passed through the
cessively during one experiment and during about measurement region, the fuel concentration cannot
5 s. Rather than using a chopper, the light was cut be determined. This study focuses on the concen-
off before and after the data were obtained. The sig- tration measurement before the ignition timing; in-
nal levels before and after the data were almost the formation after the ignition timing is not discussed
same, so it was concluded that this method is useful here. However, the signal might imply a lot of infor-
if no emission lights are added to the signal during mation (i.e., as a subject of future work).
the experiment. Figure 6 shows the final result of the air/fuel ratio
(A/F) from 50 to 10⬚BTDC. Here, the mass flow of
Experiment in a Port-Injection Spark-Ignition air was determined with the averaged data for four
Engine cylinders from an air flowmeter. The concentration
just before the spark timing near the spark plug al-
A four-stroke-cycle spark-ignition engine with four most agreed with the mean value that was obtained
cylinders in line was prepared in order to investigate from the measurement of the flow rate with a buret,
the measurement method of this technique. The which represented the mean value averaged over
bore and stroke were 89 and 96 mm, respectively. many cycles. At the crank angle before 35⬚BTDC,
 SPARK-PLUG CONCENTRATION MEASUREMENT BY 3.39 lm LASER 739

toward the TDC. The absolute values of the stan-

dard deviations of transmissivity shown in Fig. 4 and
the molar concentration shown in Fig. 5 do not ap-
pear to be large, although the relative values divided
by the average value are very large before 35⬚BTDC
because of poor signal-to-noise ratio due to the vi-
bration of the optical fiber. This explains why the
standard deviation of A/F is very large. In light of
these findings, the design of the optical sensor
should be robust enough to minimize any effect of
vibration on accuracy.

Application to a Direct-Injection Gasoline Engine

Next, this sensor was installed in a direct-injection
Fig. 6. Air/fuel ratio near the spark timing obtained gasoline engine with a single cylinder, as shown in
from the data in Fig. 4. Fig. 7, to discuss the relationship between the mix-
ture-formation process and combustion in firing cy-
cles. The bore and stroke were 86 and 86 mm, re-
spectively. The engine was operated at an engine
speed of 1400 rpm, and the air/fuel ratio in total was
set to 40. The indicated mean effective pressure was
314 kPa. The injector was installed in the indicated
region, between both intake ports; the spark plug is
located near the center of the cylinder head. In this
study, gasoline was injected in the cylinder before
the compression TDC, so that the experiments were
performed under the stratified charge condition.
The end of the fuel injection timing was set to 40,
50, and 60⬚BTDC, while the ignition timing was set
to 10⬚BTDC. According to the results of gas flow
obtained from a three-dimensional simulation [38],
it was revealed that the fuel impinges on the bowl
in the piston and the evaporated fuel is directed to
the spark plug. The special spark plug with an optical
sensor was installed in the engine. The value of the
molar absorption coefficient for gasoline was mea-
sured in advance as was that for iso-octane. Here, a
diode laser with a wavelength of 1.3 lm, which does
Fig. 7. Schematic diagram of test engine (direct-injec- not have absorption lines of hydrocarbons, was used
tion gasoline engine). to monitor for the existence of liquid droplets with
a 3.39 lm laser, simultaneously, as shown in Fig. 2,
although no compensation was performed for the
the air/fuel ratio showed a value much larger than data in this case. In the normal and stable condition,
the preset value of 16. The fuel is induced from an no decay in signal from the 1.3 lm laser could be
early stage of the open period of the intake valve, seen.
because the engine operation is in the idling condi- Figure 8 shows the history of averaged air/fuel
tion and a small amount of fuel is injected. The fuel ratio values and the standard deviation in the end of
cloud flows with the intake air and exists near the the fuel injection period, hiend. The data were sam-
piston side, and the air/fuel ratio near the spark plug pled for every 2.6⬚ change in crank angle based on
shows that fuel is less before 35⬚BTDC. As the pis- the frequency of the chopper, which was 3.2 kHz.
ton moves up toward the cylinder head, the fuel is The air/fuel ratio near the spark plug became
mixed homogeneously, so that the air/fuel ratio de- smaller rapidly after 20 crank-angle degrees from
creases gradually. However, this explanation may be the end of the injection and larger and larger with
uncertain at the present stage because the molarity the elapsed time because of the effect of turbulent
is smaller before 35⬚BTDC, as shown in Fig. 5; thus, diffusion. In hiend ⳱ 60⬚BTDC, the air/fuel ratio be-
the standard deviation of A/F appears very large. comes about 20 at the spark timing. In this case,
Therefore, smaller molarity may give low accuracy there was no injection timing for stable engine
before 35⬚BTDC; however, the accuracy increases operation, under which the coefficient of variance
740 ENERGY PRODUCTION—Combustion in Engines

Fig. 8. Effect of fuel injection timing on cycle-to-cycle fluctuation of the air/fuel ratio.

for the indicated mean effective pressure is under Conclusions

5%. The fuel injected in the cylinder does not im-
An optical sensor in a spark plug was specially de-
pinge on the bowl in the piston because the piston
signed to determine the fuel concentration history
is located too far from it to catch the fuel. As a result,
in spark-ignition engines. First, the effects of pres-
the vaporized fuel is not transported to the spark sure and temperature on the molar absorption co-
plug. efficient of fuel were independently determined by
In hiend ⳱ 40⬚BTDC, the ignition timing of using a constant-volume vessel. The molar absorp-
10⬚BTDC is suitable for stable engine operation, as tion coefficient of iso-octane increases with increas-
shown in Fig. 8c. When the engine operated under ing temperature, and decreases with increasing pres-
the condition of changing ignition timing at hiend ⳱ sure. The degrees of increase and decrease are lower
40⬚BTDC, the ignition timing could be changed in at higher pressure and temperature. Second, the fuel
the region between 2 and 20⬚BTDC and still allow concentration history near the spark plug was deter-
smooth operation of the engine. In hiend ⳱ mined quantitatively in a spark-ignition engine of the
50⬚BTDC, the engine could operate with stable port-injection type with combustion. The fuel con-
combustion at ignition timing between 8 and centration just before the spark timing almost agreed
28⬚BTDC, as shown in Fig. 8b. In both cases, the with the mean value that was obtained from mea-
stable operation regions correspond to the averaged surement of the flow rate with a buret for many cy-
air/fuel ratio between 10 and 17, with small standard cles. Next, this sensor was applied to measure the
deviation. air/fuel ratio in a direct injection spark-ignition en-
After the flame passes through the laser beam, gine. As a result, it was found that injection timing
there are no hydrocarbons in the measurement re- strongly affects air/fuel ratio and its standard devi-
gion because the fuel has been consumed. The air/ ation. It is necessary to maintain a moderate air/fuel
fuel ratio is expected to increase suddenly for each ratio near the spark plug and to control the cycle-to-
cycle. Therefore, when the standard deviation of the cycle fluctuation of air/fuel ratio to achieve stable
air/fuel ratio is small and the averaged value in- operation.
creases rapidly, as shown in Fig. 8c, the cycle-to-
cycle fluctuation of the combustion will be small. In Acknowledgement
contrast, when the standard deviation of the air/fuel
ratio increase in hiend ⳱ 60⬚BTDC and the cycle-to- This study is partially supported by Tanikawa Founda-
cycle fluctuation of the combustion is large, slow in- tion in fiscal year 2000.
crease in air/fuel ratio, as shown in Fig. 8a, occurred.
In a direct-injection gasoline engine, it is necessary REFERENCES
to maintain a moderate air/fuel ratio near the spark
plug and to control the cycle-to-cycle fluctuation of 1. Matsui, T., Tanaka, T., and Ohigashi, S., SAE technical
air/fuel to achieve stable combustion. This sensor paper 79-0483.
will be useful in the development of a new engine 2. Collings, N., SAE technical paper 88-0517.
as a tool with which to confirm the air/fuel ratio at 3. Cheng, W. K., Galliot, F., and Collings, N., SAE tech-
a local region. nical paper 89-0579.
 SPARK-PLUG CONCENTRATION MEASUREMENT BY 3.39 lm LASER 741

4. Galliot, F., Cheng, W. K., Cheng, C., Sztenderrowicz, 22. Ohyama, Y., Ohsuga, M., and Kuroiwa, H., SAE tech-
M., Heywood, J. B., and Collings, N., SAE technical nical paper 90-1712.
paper 90-0485. 23. Sohma, K., Yukitake, T., Azuhata, S., and Takaku, Y.,
5. Eckbreth, A. C., Laser Diagnostics for Combustion SAE technical paper 91-0499.
Temperature and Species, 2nd ed., Gordon and Breach 24. Witze, P. O., Hall, M. J., and Wallace, J. S., SAE tech-
Publications, Amsterdam, The Netherlands, 1996. nical paper 88-1638.
6. Zhao, H., and Ladommatos, N., in Engine Combustion 25. Spicher, U., Schmitz, G., and Kollmeiner, H.-P., SAE
Instrumentation and Diagnostics, Society of Automo- technical paper 88-1637.
tive Engineers, 2001, pp. 337–394. 26. Shoji, H., Shimizu, T., Toshida, K., and Saima, A., SAE
7. Arcoumanis, C., Green, H. G., and Whitelaw, J., Exp. technical paper 95-2407.
Fluids 3:270–276 (1985). 27. Morishima, R., and Asai, K., SAE technical paper 92-
8. Zhao, F., Taketomi, M., Nishida, K., and Hiroyasu, H., 0235.
SAE technical paper 93-2641. 28. Yoshiyama, S., Hamamoto, Y., Tomita, E., and Minami,
9. Lee, K. H., and Foster, D. E., SAE technical paper 95- K., JSAE Rev. 17(4):339–345 (1996).
0814. 29. Mongia, R., Tomita, E., Hsu, F., Talbot, L., and Dib-
10. Johnston, S. C., SAE technical paper 79-0433. ble, R. W., Proc. Combust. Inst. 26:2749–2755 (1996).
11. Grünefeld, G., Beushausen, V., Andresen, P., and 30. Hall, M. J., and Koenig, M., Proc. Combust. Inst.
Hentschel, W., Appl. Phys. B 58:333–342 (1994). 26:2613–2618 (1996).
12. Knapp, M., Luczak, A., Beushausen, V., Hentschel, W., 31. Koenig, M., and Hall, M. J., SAE technical paper 97-
Manz, P., and Andresen, P., Proc. Combust. Inst. 1644.
26:2589–2596 (1996).
32. Koenig, M. H., Stanglmaier, R. H., Hall, M. J., and
13. Miles, P. C., and Dilligan, M., SAE technical paper 96-
Matthews, R. D., SAE technical paper 97-2982.
0828.
33. Koenig, M. H., and Hall, M. J., SAE technical paper
14. Egermann, J., Koebcke, W., Ipp, W., and Leipertz, A.,
98-1053.
Proc. Combust. Inst. 28:1145–1152 (2000).
34. Kawamura, K., Suzuoki, T., Saito, A., Tomoda, T., and
15. Andresen, P., Meijer, G., Schluter, H., Voges, H.,
Kanda, M., JSAE Rev. 19(4) 305–310 (1998).
Koch, A., Hentschel, W., Oppermann, W., and Rothe,
35. Iiyama, A., Itoh, T., Muranaka, S., Takagi, Y., Iriya, Y.,
E., Appl. Opt. 29:2392–2404 (1990).
16. Baritaud, T. A., and Heinze, T. A., SAE technical paper Noda, T., Urushihara, T., Naitoh, K., paper F98T048,
92-2355. Proceedings of the Twenty-Eighth FISITA World Au-
17. Zhao, F., Taketomi, M., Nishida, K., and Hiroyasu, H., tomotive Congress, 1998.
SAE technical paper 94-0988. 36. Tsuboi, T., Inomata, K., Tsunoda, Y., Isobe, A., and
18. Neij, H., Johansson, B., and Aldén, M., Combust. Nagaya, K., Jpn. J. Appl. Phy. 24(1):8–13 (1985).
Flame 99:449–457 (1994). 37. Rothman, L. S., Rinsland, C. P., Goldman, A., Massie,
19. Berckmüller, M., Tait, N. P., Lockett, R. D., Green- S. T., Edwards, D. P., Flaud, J.-M., Perrin, A., Camy-
halgh, D. A., Ishii, K., Urata, Y., Umiyama, H., and Peyret, C., Dana, V., Mandin, J.-Y., Schroeder, J.,
Yoshida, K., Proc. Combust. Inst. 25:151–156 (1994). McCann, A., Gamache, R. R., Wattson, R. B., Yoshino,
20. Reboux, J., Puechberty, D., and Dionnet, F., SAE tech- K., Chance, K. V., Jucks, K. W., Brown, L. R., Nemtch-
nical paper 94-1988. inov, V., and Varanasi, P., J. Quant. Spectrosc. Radiat.
21. Itoh, T., Kakuho, A., Hishinuma, H., Urushihara, T., Transfer 60:665–710 (1998).
Takagi, Y., Horie, K., Asano, M., Ogata, E., and Ya- 38. Naitoh, K., Takagi, Y., and Kuwahara, K., SAE tech-
masita, T., SAE technical paper 95-2465. nical paper 98-1440.

COMMENTS
Sergei Labnda, National Academy of Science of Belarus, and the signal-to-noise (S/N) ratio affect the accuracy of
Belarus. How long can your probe operate? What is the the measurement. When this sensor detects the fuel drop-
accuracy of your concentration measurements? lets with a reference beam, we know that there are drop-
lets, and then the engine does not operate under stable
conditions. We did not compensate in the data for the ef-
Author’s Reply. The sensor plug worked well for 10 to
fect of the droplets. The existence of other gases including
20 min during one experiment. In this period, we did not water vapor, carbon dioxide, carbon monoxide, and oxygen
encounter problems with dust under stable operation. This was confirmed to be negligible by investigating each ab-
sensor is not yet a commercial product; rather, it is a trial sorption line near 3.392 lm using the HITRAN database.
one for concentration measurements. One of the sensors The molar absorption coefficient was measured carefully
has operated for more than 6 months with careful treat- within a few percent of accuracy. One of the problems that
ment. make the S/N ratio worse is the vibration of the fiber trans-
It is difficult to describe the accuracy at the present mitted from the engine. Therefore, we need to improve
stage. We believe that the existence of fuel droplets and the sensor in order to discuss cycle-to-cycle fluctuations in
other gases, the value of the molar absorption coefficient, concentration.