scattering technique

ELSEVIER

JSAE Review 17 (1996) 231-237

Measurement of fuel vapor concentration in flash boiling spray by infrared extinction/scattering technique Masayuki Adachi a, Daisuke Tanaka b, Yoshiyuki Hojyo c, Marwan A1-Roub d, Jiro Senda b, Hajime Fujimoto b a Horiba, Ltd., Kyoto, Japan b Doshisha University, Kyoto, Japan c Sharp, Ltd., Nara. Japan d University of Wisconsin-Madison. Madison, WI 53706, USA

Received 5 January 1996

Abstract The MPI (multi-point injection) technique is widely used for gasoline engines due to its high responsiveness and high controllability. The pressure in the intake manifold becomes lower than the saturated vapor pressure of some components of gasoline in some operating conditions, and this causes the flash boiling phenomenon of the components. This phenomenon has great influence on spray atomization and vapor formation. In this study, the distribution of fuel vapor concentration of n-pentane was observed by applying the Infrared Extinction/Scattering method. The fuel vapor concentration can be calculated from the intensities of transmitted infrared light, under various spray conditions including the flash boiling spray.

1. Introduction The multi-point injection (MPI) technique, which has an electronically controlled injector installed slightly upstream of the inlet valve in each cylinder, has come to play a major role among the gasoline engine fuel injection techniques. The system is capable of controlling injection timing and the amount of fuel to be injected, based on engine operating conditions. This technique, however, has the potential to cause excessive pollutant formation in exhaust gas, such as unburnt hydrocarbons, due to the fuel-air mixing being not ideal when engine operating condition do not give enough time for the atomization to grow in the intake manifold. Also, the pressure in the intake manifold can be lower than the fuel vapor pressure depending on the engine operating conditions. This situation causes a flash boiling phenomenon of the fuel to take place and this accelerates the atomization process of the spray [ 1]. The authors investigated the atomization characteristics of an injector nozzle manufactured for the MPI system, changing the surrounding pressure from ambient pressure to a pressure lower than the vapor pressure of the fuel [2-4]. Also, a model of the atomization and vaporization characteristics of the flash boiling spray was performed

with a modeling vapor and cavitation bubble formation process in the fuel film and vaporization from the fuel surface [5,6]. The objective of this article is to measure vapor concentration distribution in the fuel spray injected into a constant volume chamber containing quiescent air at room temperature. The sample spray was created through an electronically controlled injector nozzle. N-pentane was employed as the test fuel since it is one of the major components of gasoline fuel. The measurement technique used here is the infrared extinction/scattering technique (IRES) modified for transient spray measurements [7]. In this technique, two different infrared wavelengths from a tunable diode laser, T D L , are employed to measure light extinction caused by fuel vapor and fuel droplets. The same concept has been utilized by Suzuki et al. [8] for liquid and vapor phase measurement in a vaporizing Diesel spray using two wavelength regions in the ultraviolet and visible ranges.

2. Vapor concentration measurement by two-wave-

lengths IRES technique In this technique, the infrared light is directly incident on the sample spray and the transmitted light reaches the

0389-4304/96/S15.00 © 1996 Society of AutomotiveEngineers of Japan, Inc. and Elsevier Science B.V. All rights reserved PII S0389-4304(96)00025-2

JSAE963I650

232

M. Adachi et aL / JSAE Review 17 (1996) 231-237

detector. The infrared wavelength used for vapor phase measurement is in the region where the vapor of the sample fuel has the same characteristic absorption. When the vapor concentration is high, the intensity at the detector decreases due to gas phase absorption. Therefore, the vapor concentration can be determined by the intensity read at the detector. Infrared light from the TDL can be modulated in a frequency of a several tens of kilo-Herz, with enough signal-to-noise ratio, in such a short period that an intensity measurement can be done in a few tens of microseconds. This quick response of measurement is well suited for the transient spray measurement carried out in this study. Figure 1 presents a concept of vapor concentration measurement in the flash boiling spray using this technique. The absorption spectrum of the vapor phase n-pentane which is diluted to 593 ppm by nitrogen is shown in Fig. 2. An absorption peak can be recognized in the vicinity of 1450 cm -~, and no absorption is seen at 1308 cm-~. Thus, 1450 cm-~ light is subject to (1) absorption by fuel vapor, (2) absorption by fuel liquid, and (3)

0.04 .g

0,02

/

e~

0 i

i

1550

1500

02 0 --J

~Injector apor oplets i

i

i

i I

500

Fig. 2. Absorptionspectrumof n-pentanevapor(concentration:593 ppm). scattering by fuel droplets. Therefore, the intensity at the detector can be expressed as:

1450Vab

i450cmt

i

1450 1400 1350 wavenumbers [cm-rl

, - - _ _ ~ 3000 2500 2000 1500 1000 wavenumbers [cm"r]

= log(~)

I.~ iOi(~a~i'~50:1~glI1(')/ff~b)+'0g(I1~):i"ab+]Og(jl~)I's~.c...... r:~

\

/

¢.,

+ l o g ( I° ] + log( I° t I l ]Lab ~ I /Lsc

(1)

where log(lo/l)145ova b is the amount of absorption by the vapor phase n-pentane at 1450 cm -I, log(lo/l)La b is the amount of absorption by the liquid phase n-pentane at 1450 cm - t , and log(lo/l)Ls c is the amount of scattering by the fuel droplets. In the same manner, infrared light of 1308 cm 1 is also subject to the above three extinction reasons. Although no absorption can be seen at 1308 cm-1 in Fig. 2, the absorption may not be negligible at very high vapor concentrations which are likely to exist in the spray studied here. Therefore, the intensity of the 1308 cm-1 light at the detector can be expressed as:

1308

t.

lOg(T)13os= log(T)13o,~+log( [)i,abq-log(T)L~~ '"'"

))

Io

It is reported elsewhere that the term log(lo/I)La b is negligibly small compared to the term log(10/l)wb in Eqs. (1) and (2) when methanol is considered as a sample fuel [8]. Also, the amount of scattering from droplets, expressed as log(lo/l)l,~c in Eqs. (1) and (2), are found to be identical, due to the fact that both probe wavelengths are close. As a result, Eq. (3) can be derived from Eqs. (1) and

1308cm-1 I

I0

=,oJ'o/1 ]1308Vab +loj'ot ~ 1 ]Lab

~

(2): I

I, I,, I, "h Iog(T)145olog(T)13os= iog(~I )145o-l°g(T)t3o8 r

r

Vab

;

Vab

.,~)

Fig. 1. Two infrared absorption and scattenng methods.

l°g( ~-~) 1450- l°g( ~ ) ,308 = I0 = l o g ( / ) 1450Vab -- log(@- ) ,308 Vab

(3)

M. Adachi et al./JSAE Review 17 (1996) 231-237 As shown in Eq. (3), the amount of absorption due only to the vapor phase can be calculated by subtracting the measured intensity at 1308 cm-~ from the intensity at 1450 cm-1. The absorption from the fuel vapor can be expressed by the Lambert-Beer law:

where e is the molar absorption coefficient (m3/mol • m), C is the vapor concentration (mol/m3), and L is the optical path length (m). Now, let the molar absorption coefficients at the two wavelengths, i.e. 1450 cm-~ and 1308 cm -1, be e~ and E2 respectively. Then the following equations can be obtained:

log(@)1450Vab = g ; C L '

log(@),3o

8

3. Experimental apparatus and procedure Prior to the measurement, calibration was performed in order to determine the molar absorption coefficients. Fig. 3 shows the setup for the calibration. Nitrogen gas from a bottle (~) first travels through a needle valve and a flow meter @ so that its flow ratio is controlled. Then the flow

l

I

/

N2 gas

,,

/

+ 1. 2. 3. 4. 5.

N2 bottle Flow meter Water bath(313K) Copper pipe Flask

Constant volume chamber Controller Diode laser CaF2 optical window Injector

i

6. 7. 8. 9. 10.

Detector Pre-amplifier Oscilloscope Transient memory CPU

Fig. 4. Schematic diagram of experimental apparatus for measurement of vapor concentration distribution.

= 82CL.

Therefore, the terms El and e 2 can be determined when log(lo/l)450Vab and log(lo/l)13OSVa b are measured with known value for the terms C and L. In the present study, calibrations for each molar absorption coefficient were carried out by changing the optical path length, L, with the constant saturated vapor concentration, 23.2 m o l / m 3, at a pressure of 101 kPa and a temperature of 293 K. The vapor concentration in the spray can be calculated from (1) log( lo/1)teSova b and log( lo/l)13OSVa b from the measurement, (2) ~ and a 2 from the calibration, and (3) L from the "onion peeling" calculation technique [8].

n-pentane gas (313K)

1. 2. 3. 4. 5.

t

Vab

(5)

Saturated n-pentane gas (293K)

I

233

6. Bubbler 7. Water bath(293K) ~. Copper pipe t). Flask 10. Test cell

Fig. 3. Schematic diagram of experimental apparatus for calibration of molar absorption coefficient.

goes through the first copper coil tubing @ in a heated water bath @ kept at 313 K and the first flask @ with a bubbler @ placed in the same bath. The flask has adequate n-pentane liquid being vaporized by the nitrogen flow. After the first flask, the nitrogen gas containing n-pentane vapor is again introduced in the second copper coil tubing @ located in the second heated water bath (Z) and the second flask (~). The temperature for the second bath is kept at 20 K lower than the first bath so that some of the vapor condenses back into liquid phase. Thus, securely prepared vapor of the saturated n-pentane at 293 K can be produced and introduced to the optical cell @ . The optical path length of the calibration cell can be changed to 4, 5, 10, 17, 22, and 30 mm, and the molar absorption coefficient was determined from the transmitted light intensity for each wavelength. Figure 4 shows a schematic diagram of the spray measurement system using the TDL. The fuel spray is injected into the quiescent air inside the constant volume chamber @ . The dimension of the chamber, 150 W × 250 H × 170 D, is large enough to prevent the spray behavior being affected by the wall. A pintle type nozzle was employed as an injector. The nozzle has a 0.931 mm diameter injection orifice, 0.798 mm pintle diameter and 26 degree pintle angle. The TDL manufactured by Laser Photonics Inc. is controlled by the unit (~), and emits light from the laser element @ installed at a cryogenically controlled dewar. The light is introduced to the spray chamber having calcium fluoride windows @ , and incident on the sample spray. The transmitted light finally reaches the mercury cadmium telluride detector @ on the dewar. The detector signal is swept into the oscilloscope @ via a preamplifier @ or swept into the transient memory @ , triggered at the start of the injection. The digitizing rate of the transient memory is 2 microseconds and data collection takes 20 milliseconds. The pressure inside the spray chamber, Pb, is set at 48, 35, 21, and 14 kPa. The injection pressure is set at 298, 285, 271 and 264 kPa, keeping the pressure difference

234

M. Adachi et al./JSAE Review 17 (1996) 231-237 1.2

between Pb and injection pressure Pinj at 250 kPa. The total of 48 measurement points are located every 2 m m in radial distance up to 14 mm, and every 5 m m in axial distance up to 30 mm. An assumption was made for the spray to be azimuthally homogeneous. After obtaining the accumulated absorption through the fuel vapor using the two-wavelength IRES technique, the concentration distribution was calculated by the onion peeling method. A laser sheet scattering picture was taken for comparison of the concentration distribution and the spray structure. Doubled frequency N d - Y A G laser light, 532 nm in wavelength, is used for this purpose. The droplet distribution is the only information that can be acquired from the scattering picture and the vapor phase does not affect the picture at all.

log(~)~ =1.24CL

/

1.0 .41'...-, v

o

0.8

0.6

oe-

tS~-10lkPa ~:'=250kPa T~Tb=298K

/O / ~

O 1450cml _ • 1308cm-~ (saturated n-pentsne gas)

0.4 f

<

0.2 0.0

OJ

10

20

Optical length

30

40

L x 10-3[m]

Fig. 6. Relation between absorbance and optical length. m e a s u r e m e n t are carried out to determine one vapor concentration is negligibly small.

4. C a l i b r a t i o n result 4.2. Correlation between the amount o f vapor and the absorption

4.1. Spray reproducibility Reproducibility has to be recognized sufficiently for the spray in this study because the two probe wavelengths can not be used simultaneously. Figures 5(a) and 5(b) show ten absorption data for each wavelength measured at 1, 2, and 3 milliseconds after the injection. In each case, the deviation is within 5% and the reproducibility is satisfactory. Therefore, it can be concluded that the error being generated by the technique in which two different intensity

The measured relation between the optical length of the calibration cell and the absorption for each probe wavelength is shown in Fig. 6. The pressure inside the cell is 101 kPa and the temperature is 293 K. According to this relation Eqs. (6) and (7) can be derived:

(,o) (,0)

log /

1450Vab

log -7 1308Vab

0.3 P'v=48kPa A p=250kPa Tf=-Te=293K

0.2

= 1.24 CL,

(6)

0.05 eL.

(7)

From these equations Eq. (8) is also derived. For the practical measurement of the spray, Eq. (8) is used to

0.1 r~

<

O I

I

.~eo

I

1 2 3 Time from injection start t (ms)

1

(a) 1450 cmj

0.2

3

Time t (ms) (a) 1450cm I

0.3 Pb=48kPa A p=250kPa Tf=Tb=293K

2

~}

(~

g.z o.~

0.1

o

<

0

0

0 I i I 1 2 3 Time frominjectionstart t (ms)

o

4

(b) 1308cml Fig. 5. Repeatability of absorbance for ten measurements.

1

2

3

Time t (ms) (b) 1308clni Fig. 7. Temporal change in intensity of transmitted infrared light (Pb = 21 kPa, Ap = 250 kPa, x = 0, z = 20 mm).

M. Adachi et a l . / JSAE Review 17 (1996) 231-237

determine the line-of-sight absorption for each measurement point:

(,0)

log

- log -7 1450

= 1.19 CL.

235

5. Vapor concentration distribution

Figures 7(a) and 7(b) present raw transmitted intensity measurement data for each wavelength measured at radial and axial distance of zero and 20 mm respectively when the pressure of the chamber, Pb, is 21 kPa, the pressure difference, A p, is 250 kPa, and the temperature of the fuel and the chamber is 293 K. The infrared light intensity is modulated at 20 kHz. After about one microsecond, a decrease in intensity can be observed due to the absorption through the vapor and liquid n-pentane and scattering by the n-pentane droplets. The same measurements were carried out for several pressure conditions with the 48 measurement points as described above. Figures 8(a), 8(b), 9(a), and 9(b) show the typical vapor concentration distributions along with the laser scattering picture taken for the same spray condition. The measurements for these figures were taken when the chamber

(8)

1308

In general, the molar absorption coefficient is dependent on the temperature and the pressure. For propane, however, it is reported elsewhere that the change in the absorption coefficient caused by temperature is very small [9]. It can be deduced that the above characteristics can also apply for n-pentane because the molecular structures of the two compounds are very similar. Moreover, change in the temperature in the present case is not sufficient to affect the molar absorption coefficient, and also the change in the pressure is thought to be negligible. Therefore, the concentration for practical measurement is calculated based on the calibration results under room temperature and ambient pressure.

Distance from spray axis x (ram) 10 0 10 i

"~

i

I

I

[

z

|

I

I

I

0

o=~" 2 0 ~

E

v

M

•e ~

20

10t-"-z

10

=

0

10

<~.0

0

E

% 4X

20 8

6X-"

~5

"-" ~'~ 10

="""~"~0 .3

c

15

" zleOut~et

~Io~ ' oL

Oista~C z (.ram)

30 (a) t=l.0ms Distance from spray axis x (ram) 10

,

,

,

i

O I

i

,

I

i

10 I

='-" 20 r ' ~ Om

i

E v

~o~10 o= ba ba O

10

20

!

00

E O

20

~

%0.

0

0a

11 ~

30

5 lc~ Distar~u z (ra~)

(b) t=2,0ms Fig. 8. Concentrationdistribution of n-pentanevapor(Pb = 48 kPa, Ap = 250 kPa, Te = Tb = 293 K).

M. Adachi et al./JSAE Review 17 (1996) 231-237

236

Distance from spray axis x (ram) 10

0 !

i

i

1

I

10 i

I

i

)

~" 2 0 ~

I

0

g N

~-~ l 0 t - I - _,v.-

20

°FT)

10

10 O (D

~ 2\/ 4'~/ 6\

O

f'.>,;

20

0

aX., ~': %

]0

%:

30

(b)

~rora ~°zz~e °~tlet D~St,a~ce z (r~na')

t=l.Oms

Distance from spray axis x (ram) 10

0 i

~.

E

0

i

i

i

I

10 i

i

1

i

I

f

20

N

10

0

8o 10

o. o

N 0

",>> 2 \ /

<\/

2o

0 15

e o~t~e

30 (b) t=2.0ms Fig. 9. Concentration distribution o f n-pentane v a p o r ( P b = 14 kPa, A p = 2 5 0 kPa, Tf = T b = 293 K).

pressure, Pb, was 48 kPa and 14 kPa. When the chamber pressure is 48 kPa, where entrainment is evident, and when the pressure is 14 kPa, where flash boiling is evident [5], the fuel vapor existence can be recognized in regions where the fuel droplets can be seen in the scattering picture. These discussions prove the capability of this technique to some extent. Other measurements, which are not shown here, exhibited a tendency that the local vapor concentration does not change much in a short period of time. A high vapor concentration can be observed at the vicinity of the axis for the chamber pressure of 48 kPa shown in Fig. 8. This is attributed to the fact that the vapor exists just around the spray due to the contracting spray formation [3]. In the case when t = 1.0 millisecond shown in Fig. 8(a), the vapor concentration is nearly zero at a radial distance of 4 mm and an axial distance of 10 mm, while the vapor existence can be recognized at the same

radial distance with an axial distance of 15 mm. Again the vapor disappears at a radial distance of 8 mm and an axial distance of 15 mm. Hence, the vapor concentration is distributed along the outer edge of the spray and can be recognized as a conical shape distribution of the vapor due to the configuration of the pintle in the injector. However, the vapor apparently spreads out rather uniformly in the radial direction in the case of the chamber pressure 14 kPa, as shown in Fig. 9. This is attributed to the flash boiling effect of the fuel and less entrainment compared to Fig. 8. Further experiment shows that this tendency is enhanced by lowering the chamber pressure. This is because the flash boiling effect and the atomization are accelerated with decreasing pressure, and the surface area of the droplets and vaporizing velocity increase. The existence of the vapor also corresponds to the existence of the droplets as observed in the case with the chamber pressure at 48 kPa.

M. Adachi et al./JSAE Review 17 (1996) 231-237

237

6. Conclusion

Acknowledgments

The two-wavelength IRES technique was applied for the flash boiling spray. The effect of flash boiling on the vapor distribution in the spray was discussed. Conclusions drawn from the present study are as follows. (1) Feasibility of the vapor concentration technique using two infrared wavelengths is demonstrated. (2) Along with the laser sheet droplet measurement, correlation was shown between the vapor concentration and the scattering intensity by the droplets, and between the existence of vapor and droplets. (3) Increase in the amount of vapor and uniform vapor distribution for a large volume are observed when the chamber pressure is reduced to cause the flash boiling effect.

The contribution from Mr. K. Yamamoto who helped this study as an undergraduate is greatly appreciated.

References [1] [2] [3] [4] [5] [6] [7] [8]

Sudo, et al., JSAE Trans., Vol. 16 (1978), p. 17. Tsukamoto, et al., JSME Trans. (B), Vol. 58, No. 547 (1992), p. 977. Senda, et al., JSME Trans. (B), Vol. 58, No. 553 (1992) p. 2919. Senda, et al., SAE Paper, 920382. Senda, et al., JSME Trans. (B), Vol. 60, No. 578 (1994), p. 3551. Senda, et al., JSME Trans. (B), Vol. 60, No. 578 (1994), p. 3556. Suzuki, et al., JSME Trans. (B), Vol. 59, No. 563 (1993), p. 2334. M. Adachi, et al., Combustion Science and Technology, Vol. 75 (1991), p. 179. [9] Yoshiyama, et al., JSME Trans. (B), Vol. 60, No. 572 (1994), p. 1486.