ILASS-Europe 2002
Zaragoza 9 –11 September 2002
GAS-DISSOLVED HIGH QUALITY GASOLINE SPRAY
BY USING CO2-DISSOLVED MIXTURE
A. Rashkovan*, B. Rivin*,
V. Kholmer**, and E. Sher*([email protected])
*Department of Mechanical Engineering
Ben-Gurion University of the Negev, Beer-Sheva, Israel
**The Negev Academic College, Beer-Sheva, Israel
Abstract
An experimental study of the atomization of a steady spray of fuel containing dissolved CO2, under
atmospheric conditions, is presented. A high-pressure fuel injection system has been designed to produce a
spray having a lower SMD than that obtained typically with a common-rail fuel injection system for the same
injection pressure. In the present design, a mixture of fuel and dissolved CO2 is introduced to an injector unit.
The downstream part of the injector consists of an inlet orifice, an expansion chamber, a swirl duct, and a
discharge orifice. When the mixture enters the expansion chamber, a part of the dissolved gas is transformed
into tiny bubbles that grow inside the expansion chamber. When the bubbly mixture is discharged through
the discharge orifice, these bubbles undergo a rapid flashing process while the liquid bulk disintegrates into
small droplets. We have reported previously [12] about the effect of the design parameters (geometrical
proportions and injection pressure) on the spray characteristics (as measured with a laser particle size
analyser Malvern X-Mastersizer), and on the liquid disintegration process. It was concluded that the
atomization of gasoline fuel containing dissolved CO2, is significantly promoted by the flash-boiling
phenomenon to result low SMD and D90 sprays.
In the present paper, a parametric study of the droplets volume fraction distributions is presented. It was
found that the spray structure of a fuel/dissolved gas mixture is essentially different from that of a singlecomponent fuel.
Introduction
In order to limit unburned hydrocarbons emissions to an acceptable level, the injected fuel must be vaporized
before the spark is introduced. Moreover, the complete evaporation of the fuel can make the ignition process
more robust [1]. Previous studies showed that fuel droplet SMD of 25µm and less is an essential operating
condition for gasoline direct injection engines. The most common technique for gasoline direct injection
combustion systems involves an elevated fuel pressure in combination with a swirl nozzle. The required fuel
injection pressure for achieving SMD of 15 to 25 µm is typically between 5 and 13 MPa [2].
Recently, a unique technique for achieving the desired fuel spray characteristics has been proposed. It
involves atomization of fuel containing dissolved gas and is based on the flash boiling phenomena. Because
it enhances atomization and increases initial spray angle for rapid fuel-air mixing along with reduction of
spray penetration [3], the flash boiling fuel atomization has a promising potential application to direct
injection engines. A number of studies have been performed on atomization and combustion of diesel fuel
containing dissolved CO2 [4, 5, 6]. It was found that the characteristics of the spray for the liquefied CO2
mixed fuel are better than those of a pure diesel fuel spray. This has been attributed to the flash boiling
process. The effect is promoted as the ambient pressure decreases and the mole fraction of the CO2 increases
[4]. Zhen H. et al. [5] studied the flow pattern of diesel fuel containing dissolved CO2 in a hole-type injector
with different L/D ratios. They found that cavitations phenomena at the sharp edges of the orifice affect the
atomization performance of a given injector by changing the inside orifice flow pattern. They also concluded
that increasing the nozzle L/D ratio improves the spray quality of the fuel-CO2 mixture [6].
Zeigerson-Katz, M., and Sher, E. [7,8,9], and Solomon et al. [10,11], studied the effect of installing an
expansion chamber downstream the discharge orifice, for liquid containing dissolved gas. They have
observed a positive effect of the expansion chamber on the SMD of the final spray. Geometrical parameters
of the chamber were optimised with respect to their maximum effect on the atomization of the mixture.
In the present paper, a parametric study of the droplets volume fraction distributions is presented when an
expansion chamber is installed.
Discussion
The distributions of the droplets volume fraction for different mixtures under various injection pressures (P)
are presented in fig. 1. It may be concluded that the pure fuel sprays and lean CO2 mixture sprays do not
exhibit a bimodal nature for the entire range of injection pressures (3-4.5 MPa). An increasing of the
injection pressure and/or CO2 content in the mixture, results in the appearance of a second mode in the region
of the smaller droplets (here PDF designates Probability Density Function).
14
14
P=3.1[MPa]
0% CO2
7.6% CO2
9.8% CO2
12.5% CO2
17.3% CO2
12
10
8
PDF
PDF
P=4.2[MPa]
10
8
6
6
4
4
2
2
0
0% CO2
7.6% CO2
9.8% CO2
12.5% CO2
17.3% CO2
12
1
10
0
100
1
10
D [µ m]
100
D [µ m]
Figure 1. Experimental results
The main goal of the present study was to parameterize the experimentally received bimodal volume
distributions. For this purpose, each distribution curve was fitted by a sum of two lognormal distributions as
described by (1,2).
PDF = C1 ⋅ PDF1 + C 2 ⋅ PDF2
PDF1,2
(1)
 (lnX − lnD1,2 )2 
=
⋅ exp −

2
2(lnN 1,2 ) 
lnN 1,2 ⋅ 2Œ

1
(2)
Where:
PDF – the overall probability density function
PDF1,2 - probability density function of large and small droplets mode respectively
N1,2 – modes standard deviations
D1,2 – modes mean diameters
C1,2 – modes weight factors
14
12
14
Malvern data
NLF
P = 3 [MPa]
12
10
8
8
6
6
4
4
PDF
10
2
0
2
1
10
100
14
12
0
Malvern data
NLF
P = 4.36 [MPa]
12
10
8
8
6
6
4
4
PDF
1
10
100
14
10
2
0
Malvern data
NLF
P = 3.64 [MPa]
Malvern data
NLF
P = 4.63 [MPa]
2
1
10
D [µm]
100
0
1
10
100
D [µm]
Figure 3. NLF (Non-Linear Fit) results for pure fuel sprays under various injection pressures (Malvern data
represent experimental observations)
The fit procedure was performed using MATHEMATICA’s Statistics ‘NonlinearFit’ by Chi square method.
As it is seen from figs. 2 and 3 the fitted curve almost coincide with the corresponding experimental one for
both bimodal (17.3 % CO2 mixture) and monomodal (pure gasoline) distributions.
14
12
14
Malvern data
NLF
P = 3.09 [MPa]
12
10
8
8
6
6
4
4
PDF
10
2
0
2
1
10
0
100
14
PDF
12
1
10
100
14
Malvern data
NLF
P = 4.05 [MPa]
12
10
10
8
8
6
6
4
4
2
0
Malvern data
NLF
P = 3.48 [MPa]
P = 4.6 [MPa]
Malvern data
NLF
2
1
10
0
100
D [µm]
1
10
D [µm]
100
Figure 3. NLF (Non-Linear Fit) results for 17.3 % CO2 mixture under various injection pressures (Malvern
data represent experimental observations)
Fig. 4 presents the dependence of parameters N1,2 on the injection pressure for pure gasoline and for 17.3 %
CO2 mixture. For pure gasoline spray there is a slight increase in both N1 and N2 with increasing the injection
pressure, which indicates a more dispersed droplets diameter distribution. For 17.3% CO2 mixture, N1
remains nearly constant and N2 decreases with increasing the injection pressure. This indicates more uniform
droplets diameter distribution.
2.6
2.6
pure gasoline
2.2
2.0
2.0
1.8
1.8
1.6
1.4
1.6
N1
N2
1.2
17.3% CO2 mixture
2.4
2.2
N1,2
N1,2
2.4
1.4
N1
N2
1.2
1.0
2.8 3.0 3.2 3.4 3.6 3.8 4.0 4.2 4.4 4.6 4.8 5.0
1.0
2.8 3.0 3.2 3.4 3.6 3.8 4.0 4.2 4.4 4.6 4.8 5.0
P [MPa]
P [MPa]
Figure 4. Standard deviations for pure gasoline and 17.3% CO2 mixture droplet diameter distributions
Another parameter of the droplet diameter distribution is the weight mean diameter, as defined by eq. (3).
Fig. 5 presents the behavior of this parameter for different experimental conditions.
C1 D1 + C 2 D2
C1 + C 2
Dmean [µm]
Dmean =
34
32
30
28
26
24
22
20
18
16
14
12
(3)
3 [MPa]
3.5 [MPa]
4 [MPa]
4.5 [MPa]
0
2
4
6
8
10
12
14
16
CO2 content [%]
Figure 5. Weight mean diameter
18
It follows from Fig. 5, that mixtures with more than 9.8 % CO2, generate smaller mean diameters for the
entire range of injection pressures. Smaller amounts of dissolved CO2 (less than 10% in this study) result in a
fairly deterioration of the Dmean. When the injection pressure increases, this effect becomes less and less
significant. We suggest that for low CO2 mixtures, the upstream pressure is relatively low, and the pressure
drop across the inlet orifice is insufficient to produce the required amount of nuclei to result in a rich-bubble
mixture at the discharge orifice. Instead, the main nucleation process occurs at the discharge orifice where the
pressure falls to a lower value. At this location, the nuclei that are generated at the jet surface, do not allow to
mix with the liquid bulk, as occurs (inside the mixing chamber) when the nuclei are generated at the inlet
orifice. This results in an annular flow (liquid core with gas envelope), which neutralizes the flashing effect,
and reduces to an extent also the mechanical breakup mechanism of the swirl injector.
Conclusions
Volumetric analysis shows a double peak appearance, with an increasing portion of small droplets as the CO2
content increases. The higher the CO2 concentration in the mixture, the more uniform spray is obtained. The
impact of the carbon dioxide on the large droplets is a major effect. It suggests that not only the SMD is
smaller, but also the droplet volume fraction distribution is better.
References
[1] Zhao, F., Lai, M.-C., and Harrington, D.L., “Automotive Spark-Ignited Direct-Injection Gasoline
Engines”, Progress in Energy and Combustion Science 25:437-562 (1999).
[2] Harada, J., Tomita, T., Mizuno, H., and Mashiki, I.Y., ”Development of a Direct Injection Gasoline
Engine,” SAE-960600 (1996).
[3] Park, B.S., and Lee, S.Y., “An Experimental Investigation of the Flash Atomization Mechanism”,
Atomization and Sprays 4:159-179 (1994).
[4] Senda, J., Asai, T., Kawaguchi, B., and Fujimoto, H., “Characteristics of Gas-Dissolved Diesel Fuel
Spray (Spray Characteristics and Simulating Flash Boiling Process)”, JSME International Journal Series
B 43:503-510 (2000).
[5] Zhen, H., Yiming, S., Shiga, S., Nakamura, H., and Karasawa, T., “The Orifice Flow Pattern, Pressure
Characteristics, and Their Effects on the Atomization of Fuel Containing Dissolved Gas”, Atomization
and Sprays 4:123-133 (1994).
[6] Zhen, H., Yiming, S., Shiga, S., Nakamura, H., Karasawa, T., and Nagasaka, T., “Atomization Behavior
of Fuel Containing Dissolved Gas”, Atomization and Sprays 4:253-262 (1994).
[7] Zeigerson-Katz, M., and Sher, E., “Spray Formation By Flashing of a Binary Mixture: a Parametric
Study”, Atomization and Sprays 8:255-266 (1998).
[8] Sher, E., and Zeigerson-Katz, M., “Spray Formation By Flashing of a Binary Mixture: An Energy
Approach”, Atomization and Sprays 6:447-459 (1996).
[9] Zeigerson-Katz, M. and Sher, E., “Fuel Atomization by Flashing of a Volatile Liquid in a Liquid Jet,”
SAE, Vol. SP-1151, pp. 27-34 (also as SAE-960111), (1996).
[10] Solomon, A.S.P., Rupprecht, S.D., Chen, L.D., Faeth, G.M., “Flow and Atomization in Flashing
Injectors”, Atomization and Spray Technology, 1:53-76 (1985)
[11] Solomon, A.S.P., Chen, L.D., and Faeth, G.M., “Investigation of Spray Characteristics for Flashing
Injection of Fuels Containing Dissolved Air and Superheated Fuels”, NASA Contractor Report 3563.
[12] Sher, E., Rashkovan, A., deBotton, G., and Kholmer, V., “Gas-Dissolved Gasoline Spray – An
Experimental Study”, SAE paper no. 2002-01-0841, 2002.