Exp Fluids (2009) 47:439–449 DOI 10.1007/s00348-009-0673-y

RESEARCH ARTICLE

Visualization of the evaporation of a diesel spray using combined Mie and Rayleigh scattering techniques Anne Adam Æ Philippe Leick Æ Gerd Bittlinger Æ Christof Schulz

Received: 29 September 2008 / Revised: 28 April 2009 / Accepted: 29 April 2009 / Published online: 20 May 2009 Ó Springer-Verlag 2009

Abstract Evaporating Diesel sprays are studied by laser Rayleigh scattering measurements in an optically accessible high-pressure/high-temperature cell that reproduces the thermodynamic conditions which exist in the combustion chamber of a Diesel engine during injection. n-Decane is injected into the vessel using a state-of-the-art near-production three-hole nozzle. Global images of the distributions of the liquid and vapor phases of the injected fuel are obtained using a combined Schlieren and Mie scattering setup. More details about the evaporation are revealed when the spray is illuminated by a laser light sheet: laser light can be scattered by molecules in the gas phase (Rayleigh scattering) or comparably large fuel droplets (Mie scattering). The former is seen in regions where the fuel has completely evaporated, and the latter is dominant in regions with high droplet concentrations. Studying the polarization of the signal light allows the distinction of three different regions in the spray that are characterized by a moderate, low or negligible concentration of liquid fuel droplets. The characteristics of fuel evaporation are investigated for different observation times after the start of injection, chamber conditions and injection pressures. For the quantification of This paper was originally presented at the 14th International Symposium on Applications of Laser Techniques to Fluid Mechanics, Lisbon, 2008. A. Adam
P. Leick (&)
G. Bittlinger Department CR/AEE3-Sh, Robert Bosch GmbH, Robert-Bosch-Platz 1, 70839 Gerlingen, Germany e-mail: [email protected] URL: http://research.bosch.com C. Schulz Institute for Combustion and Gasdynamics, University of Duisburg-Essen, Lotharstraße 1, 47057 Duisburg, Germany

the fuel concentration measurements based on Rayleigh scattering, a calibration method that uses propane as a reference gas is presented and tested. At high ambient temperatures, the accuracy of the concentration measurements is limited by pyrolysis of the fuel molecules.

1 Introduction The fuel injection system and the injection strategy strongly affect the fuel/air mixture preparation, and thus the power, fuel consumption, and emissions of internal combustion (IC) engines (Bittlinger et al. 2003). The determination of the local fuel concentration in the combustion chamber is particularly important since it strongly influences the pollutant formation (Enderle et al. 2004). To obtain quantitative information about fuel concentrations, the use of non-intrusive laser-based diagnostic techniques such as Raman scattering, laser-induced (exciplex) fluorescence (LIF/LIEF) and Rayleigh scattering has been reported in the literature. A systematic review of optical diagnostic techniques for measurements of mixing processes in engines is presented by Zhao and Ladommatos (1998). Each one of these techniques has its characteristic advantages and drawbacks. LIF and LIEF visualize the vapor phase of the spray. If an appropriate exciplex tracer pair is added to the fuel, the liquid and vapor phases can be simultaneously detected at different emission wavelengths (Bruneaux 2005; Kim and Ghandhi 2001; Payri et al. 2006). Because the LIF signal is typically of higher wavelength than the incident light, the signal can be spectrally separated from background scattering. Strong signals provide good signal-to-noise ratio. These techniques, however, are influenced by collisional

123

440

Exp Fluids (2009) 47:439–449

quenching, which complicates quantification, especially in the presence of oxygen, and—depending on the composition of the fuel—a fluorescent tracer may be needed to visualize the vapor phase (Schulz and Sick 2005). Since the fuel and the tracer evaporation rates can be different, this doping may cause mismatches in tracer and fuel vapor concentrations if inadequate tracers are selected. Raman scattering is a species-selective technique, but is limited by the weak signal intensity, which makes spatially and temporally resolved measurements very difficult (Arndt et al. 2000). Rayleigh scattering is a promising optical measurement technique for fuel vapor visualization, since the signal is proportional only to the incident laser intensity, the number density of the scattering molecules and their associated Rayleigh scattering cross-sections (Espey et al. 1997; Idicheria and Pickett 2007; Schulz et al. 2004; Zhao and Ladommatos 1998). In addition, the signal amplitude is independent of ambient conditions and orders of magnitude stronger than Raman signal amplitudes. However, it is difficult to separate the Rayleigh and Mie (caused by fuel droplets or the chamber walls) components in the detected signal, since both processes are elastic.

2 Rayleigh scattering theory Rayleigh scattering arises from the oscillating dipole moments of individual molecules (or other scattering centers whose size is much smaller than the laser wavelength k) induced by the oscillating electric field of the incident light wave (Eckbreth 1988). Consequently, the frequencies of the incident and the scattered light waves can be considered to be identical, if the Doppler effect is neglected. The Rayleigh signal intensity IR is proportional to the incident laser intensity IL, the gas density (qG) and composition and the Rayleigh scattering cross-sections of the molecules in the interrogation region. It can be written as: X dri IR ¼ g IL X V ; ð1Þ Ni dX i where V is the volume of the interrogation region and X is the collection angle of the detection system. Ni is the number density and dri =dX is the differential scattering cross-section for the species i given by:

dri 3 ¼ ri ðcos2 u cos2 h þ sin2 uÞ: dX 8p

Here, u is the polarization angle of the incident light relative to the scattering plane, which is defined by the incident laser beam, and the angle under which the scattered light is observed, and h is the scattering angle, as illustrated in Fig. 1. When the laser polarization is oriented normally to the scattering plane (u = 90°) and the detector is positioned at h = 90° relative to the direction of the incident laser beam, Eq. 1 can be simplified to: X 3 IR;? ¼ g IL X V Ni ri : ð3Þ 8p i Because the Rayleigh scattering cross-sections ri of the ambient gases (see Table 1) are negligible when compared to the cross-sections of the fuel molecules (rf), Eq. 3 can be rewritten as: 3 g IL X V N f r f : ð4Þ 8p For gases with a refractive index n close to 1, the Rayleigh scattering cross-section r (for the wavelength k) can be calculated according to Miles et al. (2001):
 32p3 n  1 2 rffi : ð5Þ N 3k4 IR;? ¼

The refractive indices n of the individual species can be determined using the empirical Lorenz–Lorentz equation. The optical refractivity RL needed in the Lorenz–Lorentz equation is described by Partington (1953). With the optical refractivities or the refractive indices tabulated by Gardiner et al. (1981), Partington (1953), and Vogel (1946), it is possible to calculate the Rayleigh scattering

Table 1 Rayleigh scattering cross-section of selected molecules Molecule

r 9 10-24 cm2

C10H22 (n-decane)

0.61

N2

0.0051

He

0.000070

C3H8 (propane)

0.067

O2

0.0042

H2O

0.0037

scattering plane to the camera (angle of observation) E⊥

ð2Þ

ϕ θ

E⊥

incident laser light E

E

~ denotes the electric field of the light waves) Fig. 1 Definition of the scattering plane, the scattering angle and the polarization angle (E

123

Exp Fluids (2009) 47:439–449

441

cross-sections of some relevant gases. An overview is given in Table 1.

3 Experimental setup 3.1 High-p/high-T combustion chamber The results presented in this paper have been obtained in an optically accessible high-pressure/high-temperature cell equipped with two side windows and a large front window. Pressures and temperatures are similar to those found in an IC engine cylinder at the moment of injection. The gas temperature (TG), density (qG) and O2 volume fractions can be independently adjusted as the chamber atmosphere is prepared by a pre-combustion (Pauer et al. 1999). The chamber is filled with a mixture of H2, O2 and N2 whose exact composition is chosen according to the desired reaction conditions. The mixture is ignited by a spark plug and the pre-combustion leads to a strong rise in pressure and temperature. After the pre-combustion, the chamber atmosphere slowly cools down at constant volume and the fuel is injected when the desired pressure and temperature conditions are reached. For the measurements presented in this study, the precombustion gas mixtures were chosen in such a way that the oxygen is almost completely consumed during the precombustion, thus enabling the visualization of the spray evaporation without ignition and combustion.

Table 2 Relevant nozzle properties (explained in detail in Winter et al. 2004) Nozzle type

Orifice geometry

Outlet diameter

Height angle

VCO

ks

145 lm

80°

The fuel is injected into the vessel using a Bosch Common Rail (CR) injector with a state-of-the-art, nearproduction, three-hole nozzle (the relevant nozzle properties are shown in Table 2). Compared to standard nozzles, which typically have between five and eight holes, the reduced number of spray holes allows unlimited optical access to one fuel jet but does not significantly alter the spray properties. The injector is fixed on the back wall of the cell in such a way that one spray cone is injected parallel to the chamber wall. For quantitative measurements, a single-component fuel with a known Rayleigh scattering cross-section and similar properties to Diesel is preferred to standard Diesel, whose exact composition is not generally known, and whose components—with different scattering cross-sections—do not evaporate at the same rate. For evaporation measurements, n-decane is a decent substitute for Diesel fuel. 3.2 Optical setup A frequency-doubled Nd:YAG laser (k = 532 nm, pulse duration 7 ns, laser energy up to 440 mJ) is used for the Rayleigh setup (Fig. 2). The pulse energy can be varied by adjusting a k/2-plate that is followed by a polarization cube. A second k/2-plate is used to adjust the polarization of the laser pulse. The beam is formed into a sheet by a set of four cylindrical lenses and directed through the cell side windows by a mirror system that allows a precise and reproducible adjustment of the light-sheet position. A slit improves the light-sheet quality and controls the sheet height and width (0.8 9 8.3 mm2). The laser energy is measured by a detector that can either receive laser pulses deflected by a beam splitter located between the cylindrical lenses, as shown in Fig. 2, or that can be positioned immediately after the exit window of the chamber. In the latter case, typical measured values of the pulse energy are of the order of 5 mJ.

Fig. 2 Experimental setup for Rayleigh measurements

123

442

The signal is observed through the large front window at a right angle by an intensified CCD camera. The need to avoid signal saturation in order not to damage the image intensifier sets an upper limit to the laser pulse energy that can be used for the Rayleigh/Mie image recordings. Frequency-shifted signals are not expected, nevertheless a narrow bandpass filter (532 ± 3 nm) is inserted in front of the camera as a precaution, as well as a polarizer that filters the scattered light. During the pre-combustion followed by the cooling of the chamber atmosphere, the pressure inside the combustion vessel is constantly monitored. When the desired conditions are reached, the trigger sequence that controls injection, laser illumination and image detection is initiated.

Exp Fluids (2009) 47:439–449

the ambient atmosphere is negligible due to the small scattering cross sections of the corresponding molecules (Table 1). For a quantitative interpretation of the signal intensities, the Rayleigh-scattering images must be calibrated. For this, the chamber is filled homogeneously with a gas that has a large and well-known Rayleigh scattering cross-section. The ratio of signal intensities (Eq. 4) for fuel (index f) and calibration gas (index c) provides a way to determine the fuel concentration:

3.3 Image processing To compare spray measurements carried out under different conditions, the images are corrected [independently for each pixel, denoted by the coordinates (x, y)] for variations in laser pulse energy: the mean dark current of the camera K is subtracted from each spray image; the image is then divided by the integral laser pulse energy EL:  yÞ I0 ðx; yÞ  Kðx; : ð6Þ In ðx; yÞ ¼ EL K is read out when the camera is not exposed to any radiation. I0 is the detected signal and In is the corrected one. Images without spray are acquired and are corrected in order to subtract the scattered laser light from the background from the spray images at a later stage. To improve statistics, a set of at least five images is recorded under the same experimental conditions and the corrected images are averaged. The attenuation of the laser light sheet within the combustion chamber, due to absorption and scattering, is negligible in both the calibration (Sect. 4) and the actual spray experiments. This is easily verified by order of magnitude calculations based on the Lambert–Beer law and the scattering cross sections given in Table 1. Additionally, for the calibration measurements, a decrease of the average Rayleigh signal along the laser light sheet path (from right to left in Fig. 4) could not be observed. A correction of the measured signal due to an attenuation of the laser beam is therefore not necessary.

Fig. 3 Variation of Rayleigh signal strength as a function of the propane density calculated using either the ideal gas law or the Soave Redlich Kwong law

4 Quantification of the fuel concentration In the fully evaporated portions of the Diesel spray, the overall elastic scattering in the absence of droplets and soot (remnants of previous measurements, during which fuel combustion was not inhibited) is dominated by Rayleigh scattering of fuel molecules, as the Rayleigh scattering of

123

Fig. 4 Rayleigh scattering signal from 0.8 MPa propane in the combustion vessel. The plot shows the average laser light intensity profile of the slope obtained by linear regression

Exp Fluids (2009) 47:439–449

443

Fig. 7 n-Decane pyrolysis for a constant ambient density (qG = 9.6 kg/m3) at different temperatures (continuous lines). Estimation of the number of carbon atoms per fuel molecule and the associated effective Rayleigh cross-sections as a function of time (dotted lines)

Fig. 5 Fuel concentration in the vapor phase of the spray. Averaged image from 20 independent spray measurements at pG = 3 MPa, TG = 1,000 K (qG = 9.6 kg/m3), 1.3 ms after start of energization of the injection system (SoE), injection pressure pinj = 90 MPa, total fuel quantity = 18 mm3. Spray direction from bottom to top

Liquid phase

Nozzle

Vapor phase

Fig. 6 Rayleigh scattering cross-section of selected alkane, alkene and alkyne molecules

Nf ¼

IR;?;f EL;c rc Nc : IR;?;c EL;f rf

ð7Þ

Multiplying Eq. 7 by the molar mass of the fuel vapor (Mf) leads to a similar expression, but written in terms of gas densities: qf ¼

IR;?;f EL;c rc Mf q: IR;?;c EL;f rf Mc c

ð8Þ

Fig. 8 Schlieren/Mie image. pinj = 50 MPa, duration of energization tE = 1 ms, qG = 9.6 kg/m3 (pG = 3 MPa, TG = 1,000 K) recorded at t = 1 ms after SoE of the injector. The diameter of the Schlieren mirror, which corresponds to the bright background illuminated by a HeNe-laser (k = 632.8 nm), is 98 mm

Using only one calibration gas density in Eq. 8 for determining the fuel concentration may lead to high measurement uncertainties due to a weak signal-to-noise ratio (SNR) and an inaccurate assessment of the background intensity. An improved strategy uses Rayleigh measurements at different calibration gas densities. The relationship between IR,\,c and qc is determined for each individual pixel by a linear regression analysis (slope ? calibration

123

444

Exp Fluids (2009) 47:439–449

Distance from nozzle [mm]

Scatte red light intensity [a.u.]

max

min

Fig. 9 Mie image of the liquid phase. Experimental conditions as in Fig. 8; the spray is now illuminated by the (highly attenuated) Nd:YAG laser

constant, offset ? background). The local fuel density is then determined from the slope of the linear regression and a measured background intensity (i.e. images with no injection). Propane is chosen as a calibration gas since it is readily available, gaseous at room temperature and has a high Rayleigh scattering cross section. The density is determined from pressure and temperature measurements. Measurements were carried out between 0.1 and 0.8 MPa propane. Because the maximum pressure is near the liquefaction limit of propane (0.83 MPa at room temperature), the precision of the ideal gas law is not sufficient. Instead, the density needs to be calculated using a real-gas law such as the one proposed by Redlich and Kwong (1949) and Soave (1972). Figure 3 shows the variation of the Rayleigh scattering intensity of propane with the measured pressure and underlines the need to use a real-gas law, since calculations based on the ideal gas law cannot accurately reproduce the experimental results. For this representation, for each pressure, the scattered light was averaged over 20 individual images and within the region of the laser light sheet.

Laser polarization ⊥ to the scattering plane

// to the scattering plane

max

⊥ to the scattering plane

direction of spray propagation

light sheet

high droplet concentration

direction of spray propagation

light sheet

low droplet concentration

direction of spray propagation

light sheet

distance to nozzle tip: 27,5 mm

36,1 mm

Fig. 10 Observation of the tip of the liquid spray using different orientations of the laser polarization and the polarizing filter. The liquid fuel on the left, outside of the light sheet, is illuminated by stray

123

scattered light intensity [a.u.]

high droplet concentration

direction of spray propagation

// to the scattering plane

Camera polarization filter orientation

fuel vapor

light sheet 27,5 mm

min 36,1 mm

light. pinj = 50 MPa, qG = 9.6 kg/m3 (pG = 2.7 MPa, TG = 900 K). The images, which are each from six averaged independent measurements, are recorded at t = 1.0 ms after SoE

Exp Fluids (2009) 47:439–449

445

Light sheet

Light sheet

Distance from nozzle [mm] min

Distance from nozzle [mm] max

Scattered light intensity [a.u.]

Moderate droplet concentration Isolated droplets (low concentration) No droplets (only vapor)

Fig. 11 The three spray zones: vapor phase, isolated droplets and moderate droplet concentration zones. Ambient conditions as in Fig. 10. Left Original image, laser and camera polarization filter \ to the scattering plane. Right In order to separate the different spray zones, the images (from Fig. 10) are binarized. The region where signal is seen when the polarization of the laser beam and the camera filter are differently oriented (depolarized signal) shows only droplets

in high or moderate concentrations. The area where isolated droplets are present is obtained by binarizing the image where both laser and camera polarization filter are // to the scattering plane, followed by a subtraction of the depolarization area. When the laser polarization direction and the filter orientation are both \ to the scattering plane, and both previous areas are subtracted, the region containing only vapor is found

The calibration images are processed in the same way as the spray images: dark current, laser pulse energy and background intensity are considered. The signal distribution that is determined from the linear regression allows isolation and thus elimination of the background signal from the calibration measurements (Fig. 4). An application of the calibration technique is shown in Fig. 5: the laser light sheet is positioned in a fully vaporized region of the spray and images from 20 independent experiments are averaged. The signal is calibrated according to Eq. 8. The result demonstrates that the fuel concentration is symmetric to the vertical spray axis and slowly decreases with increasing distance from the nozzle outlet.

of carbon atoms per hydrocarbon molecule (see Fig. 6). As a consequence, the mole fraction and the Rayleigh scattering cross-section of each species resulting from the fuel pyrolysis need to be known for a rigorously correct analysis of the data. An estimation of the n-decane pyrolysis has been done for a homogeneous mixture of fuel vapor based on Chemkin calculations [based on the model of Zhao et al. (2005)] and an ambient atmosphere at qG = 9.6 mg/cm3 (see Fig. 7). Up to temperatures of about 1,000 K, the cracking of n-decane is negligible within the time of interest. For higher temperatures, the fuel molecules rapidly break down, so that the average number of carbon atoms per hydrocarbon molecule strongly decreases. The time needed for these reactions is comparable to the duration of the injection (roughly 1 ms). Using the curve of Fig. 6, an approximate ‘‘effective’’ Rayleigh scattering cross-section can be calculated. While it needs to be emphasized that the Chemkin simulations are not an accurate simulation of the processes occurring after fuel injection, they clearly show that—for the ambient conditions considered in this paper—fuel pyrolysis cannot be neglected. The breakdown of the hydrocarbon molecules is mainly due to the high temperatures.

5 Fuel pyrolysis In principle, it should be straightforward to calculate local fuel concentrations from Rayleigh scattering measurements. However, fuel pyrolysis—the breakdown of large hydrocarbon molecules into smaller ones—may be a limiting factor for quantitative measurements, as Rayleigh scattering cross-sections are not proportional to the number

123

Laser polarization

Exp Fluids (2009) 47:439–449 Camera polarization filter orientation

446

pG = 2.55 MPa

pG = 2.7 MPa

pG = 3.0 MPa

pG = 3.3 MPa

pG = 3.6 MPa

pG = 3.75 MPa

TG = 850 K

TG = 900 K

TG = 1000 K

TG = 1100 K

TG = 1200 K

TG = 1250 K 33,4 mm 24,2 mm

⊥ ⊥ 33,4 mm 24,2 mm

//

// 33,4 mm 24,2 mm

⊥ //

33,4 mm 24,2 mm

// ⊥ min

max log. scattered light intensity [a.u.]

Fig. 12 Observation of the influence of the ambient temperature at constant density on the tip of the liquid spray using different orientations of the laser polarization and the polarizing filter.

pinj = 90 MPa, qG = 9.6 kg m-3. The images, which are each from 6 averaged independent measurements, are recorded at t = 1.0 ms after SoE

6 Visualization of the distribution of liquid and evaporated fuel

to distinguish between light scattered by fuel vapor (molecule size  k), soot [particle size \ k (Khatchikian et al. 2004)] and large fuel droplets (size  k, Mie scattering). According to the Schlieren/Mie images, at sufficiently high temperatures, there seems to be a maximum liquid penetration length. The laser light sheet is then positioned in the vicinity of the leading edge of the liquid part of the spray, where the droplet concentration decreases rapidly with increasing distance from the nozzle outlet. Within the area illuminated by the light sheet, however, both vapor and (minor amounts of) liquid fuel can be present. Depending on the local droplet concentration, different zones can be defined depending on the polarization of the scattered light. In areas without liquid, only Rayleigh scattering can be seen, which gives the strongest signal when both the laser polarization direction and the orientation of a polarization

The penetration length of the liquid and vapor phases of the spray can easily be determined on combined Schlieren/Mie images (Pauer et al. 1999) which are recorded by a highspeed camera under the same p, T-conditions as the Rayleigh measurements. Figure 8 shows an example: flash lamps illuminate the liquid phase of the spray (Mie scattering) and the Schlieren technique, which is based on variations of the refractive index, indicates the vapor phase of spray. For a more precise determination of the liquid penetration length, additional Mie images are recorded with an intensified CCD camera (Fig. 9). Rayleigh scattering is used for a more detailed visualization of the fuel vapor distribution. However, it is difficult

123

Exp Fluids (2009) 47:439–449

447

scattering and should be measured with identical directions of the laser polarization and filter orientation (// or \ to the scattering plane; single scattering, no depolarization). Finally, multiple scattering associated with higher droplet concentrations depolarizes the scattered light and is thus detectable independently of the orientations of the laser polarization and the filter. Some possible configurations of the laser polarization and filter orientation are shown in Fig. 10. With a combination of these images, it is possible to delimitate three spray zones as shown in Fig. 11: regions containing fuel vapor and droplets at moderate, low or negligible concentrations. 6.1 Application of the polarization variation Fig. 13 Injection rates for pinj = 50, 90 MPa. For both pressures, the same amount of fuel (18 mm3) is injected. The measurements were done in a chamber pressurized with nitrogen at room temperature according to a technique described by Harndorf et al. (2002); since the flow rate of the nozzle used for these experiments is not obstructed by cavitation, the effect of the different ambient conditions can be neglected, and the results can be compared to those obtained in the high-p/high-T cell

filter in front of the detector are perpendicular (\) to the scattering plane. The Rayleigh signal intensity is negligible if either the polarization of the incident laser beam or the orientation of the polarizer are parallel (//) to the scattering plane. Isolated fuel droplets lead to Mie-scattering, which is several orders of magnitude stronger than Rayleigh t = 0.6 ms

t = 0.8 ms

t = 1.0 ms

The polarization technique is used to study the influence of the ambient gas temperature at constant density on the liquid and vapor phase distributions. Figure 12 shows such a ‘‘polarization matrix’’, where each image is an average from six independent measurements. Due to the high intensity difference between Mie scattering of the droplets and Rayleigh scattering, the color-scale representation is logarithmic. With decreasing temperature, the maximum liquid penetration length rises noticeably. At moderate and low temperatures (T B 1,000 K), the measurements show that there is no abrupt end of the liquid jet, but rather a fast, but still gradual decrease of the droplet concentration. The t = 1.2 ms

t = 1.4 ms

t = 1.6 ms

pinj = 50 MPa

33.4 mm

24.2 mm

pinj = 90 MPa

33.4 mm

24.2 mm

min

max scattered light intensity [a.u.]

Fig. 14 Time evolution of the Diesel spray for different injection pressures. For each time t (after SoE), six independent measurements have been acquired and then averaged. The orientation of the laser

light polarization and the polarization filter are both normal to the scattering plane. qG = 9.6 kg/m3, TG = 1,100 K

123

448

technique presented here, where part of the spray is illuminated using a bright laser light sheet and observed with an ICCD camera, is much more sensitive to small droplet quantities than more traditional methods such as the Schlieren/Mie technique. Thus, it is not surprising to find that some liquid droplets are still found in regions where, according to corresponding Schlieren/Mie images, the liquid fuel should already have evaporated completely. Nevertheless, for temperatures above 1,000 K, only vapor is present in the light sheet area. However, the scattering intensity decreases with increasing temperature, which is almost certainly due to the n-Decane pyrolysis that lowers the average Rayleigh scattering cross-section of the hydrocarbon molecules (see Figs. 6, 7).

Exp Fluids (2009) 47:439–449

of the corresponding vapor plume is higher at pinj = 90 MPa than at 50 MPa. It is therefore not too surprising that similar local fuel concentrations are found in the illuminated area at both injection pressures. The situation is different at t = 1.6 ms after SoE, which is just after the end of injection at pinj = 90 MPa. Here, the local fuel concentrations are much lower than for pinj = 50 MPa, where the injection has not yet stopped. They are also lower than during the injection event. This gradual decrease of the local fuel concentrations shows that the mixing process slows down after the end of the injection, but does not end abruptly.

7 Conclusion 6.2 Temporal evolution of the diesel spray The temporal evolution of the structure of the vapor plume of an evaporating spray at two Diesel injection pressures of pinj = 50 and 90 MPa is investigated by Rayleigh imaging. Since the injection rate increases with the injection pressure, the duration of energization is shortened at pinj = 90 MPa, so that the same fuel quantity (18 mm3 total, 6 mm3 per spray hole) is injected for both conditions. Figure 13 shows the corresponding injection rates, which are helpful for the interpretation of the following results. Figure 14 shows the spray plume’s temporal evolution. For each time t after SoE, the images are obtained by averaging six individual images. The time of data acquisition spans from t = 0.6 ms to t = 1.6 ms after SoE. At pinj = 50 MPa, vapor is visible in the observation area about 0.8 ms after SoE. For the higher injection pressure of pinj = 90 MPa, this delay is reduced to about 0.6 ms. The difference is mainly due to higher injection velocity at higher injection pressure, which leads to shorter times-offlight between the nozzle outlet and the observation area. The lower delay between SoE and start of injection (the difference is about 50 ls, Fig. 13) at higher injection pressure also contributes to this effect. For most of the injection event, the local fuel concentrations have a similar order of magnitude for both values of pinj, and the spray structure in the light sheet area seems to be nearly steady until the end of injection (around t = 1.4 ms for pinj = 90 MPa). This observation can easily be explained using the assumption that higher injection pressures accelerate the fuel/air mixing process. To illustrate this, the images at t = 1.2 ms after SoE are analyzed in more detail: for both injection pressures, the injection is still ongoing, but a higher fuel quantity has already been injected at pinj = 90 MPa than at 50 MPa. However, the propagation of the spray tip (located downstream of the area illuminated by the light sheet for t [ 0.8 ms) is faster at higher pinj—and thus, the volume

123

Rayleigh and Mie scattering were used to study the local fuel distribution and to quantify the vapor concentration in a high-pressure/high-temperature cell under Diesel engine conditions. A calibration method to quantify the local fuel concentration in the vapor phase of the spray with Rayleigh scattering was presented, which is based on reference measurements where the chamber is filled with a pure gas (propane) at various pressures. For the reference gas, the ideal gas law is not sufficiently precise; instead, the Soave Redlich Kwong law should be used to derive gas density from pressure. At high temperatures (TG [ 1,000 K), fuel pyrolysis considerably affects the effective Rayleigh-scattering cross-section of the fuel/gas mixture and thus limits the possibility of determining fuel concentrations from Rayleigh scattering in the vapor phase. The polarization properties of the Mie/Rayleigh imaging technique were used to determine different zones in the spray according to the local droplet concentrations. In further work, a more detailed study of the polarization properties will be carried out: Instead of using a single camera to acquire the scattered light from the combustion cell, two cameras and a polarizing cube will be used to acquire two polarization components simultaneously. Acknowledgments The pyrolysis of n-decane was simulated by Mustapha Fikri from IVG, University of Duisburg-Essen. His support is greatly appreciated.

References Arndt S, Gartung K, Pauer T, Zeh D, Bru¨ggemann D, Egermann J, Ipp W, Wagner V, Leipertz A (2000) Analysis of high pressure sprays for Diesel and SI engines with direct fuel injection: evaluation of evaporation and spray induced turbulence—Part II: spray evaporation and quantification of the local fuel air ratio inside the spray. In: Proceedings of ILASS-Europe 2000

Exp Fluids (2009) 47:439–449 Bittlinger G, Heinold O, Hertlein D, Kunz T, Weberbauer F (2003) Die Einspritzdu¨senkonfiguration als Mittel zur gezielten Beeinflussung der motorischen Gemischbildung und Verbrennung. In: Proceedings of the 6th conference on engine combustion processes and modern techniques, HdT, pp 19–30 Bruneaux G (2005) Mixing process in high pressure Diesel jets by normalized laser induced exciplex fluorescence—Part I: free jet, SAE Technical Paper Series 2005-01-2100 Eckbreth AC (1988) Laser diagnostics for combustion temperature and species, 2nd edn, Combustion science and technology book series, vol 3, Gordon and Breach Publishers Enderle C, Nolte A, Paule M, Duvinage F (2004) Der PKW im Spannungsfeld der Emissionsgesetzgebung, 2. Emission Control, Dresden Espey C, Dec J, Litzinger T, Santavicca D (1997) Planar laser Rayleigh Scattering for quantitative vapor-fuel imaging in a diesel jet. Combust Flame 109:65–86 Gardiner WC, Hidaka Y, Tanzawa T (1981) Refractivity of combustion gases. Combust Flame 40:213–219 Harndorf H, Bittlinger G, Drewes V, Kunzi U (2002) Analysis of nozzle parameters and their influence on fuel mixture preparation in current and future diesel combustion systems. In: Proceedings of 5th international symposium on internal combustion diagnostics. AVL Deutschland GmbH Idicheria CA, Pickett LM (2007) Quantitative mixing measurements in a vaporizing diesel spray by Rayleigh imaging, SAE Technical Paper Series 2007-01-0647 Khatchikian P, Harndorf H, Knopf M (2004) Dependence of the particle size distribution on the injection pressure of a modern diesel engine with a 1600 bar common-rail injection system. In: Proceedings of FISITA World Automotive Congress, Paper V027, 2004 Kim T, Ghandhi JB (2001) Quantitative 2-D fuel vapor concentration measurements in an evaporating diesel spray using the exciplex fluorescence method, SAE Technical Paper Series 2001-01-3495

449 Miles RB, Lempert WR, Forkey JM (2001) Laser Rayleigh scattering. Meas Sci Technol 12:R33–R51 Partington JR (1953) An advanced treatise on physical chemistry, vol IV. Physico-Chemical Optics, Longmans, Green, London Pauer T, Wirth R, Bru¨ggemann D (1999) Zeitaufgelo¨ste Analyse der DI-Dieselgemischbildung mittels kombinierter Schlieren-/Streulichttechnik. In: Proceedings of 4th conference on engine combustion processes and modern techniques, HdT Payri F, Pastor JV, Pastor JM, Julia JE (2006) Diesel spray analysis by means of planar laser-induced exciplex fluorescence. Int J Eng Res 7:77–89 Redlich O, Kwong JNS (1949) On the thermodynamics of solution. V-An equation of state. Chem Rev 44:233–244 Schulz C, Sick V (2005) Tracer-LIF diagnostics: quantitative measurement of fuel concentration, temperature and air/fuel ratio in practical combustion situations. Progr Energy Combust Sci 31:75–121 Schulz C, Gronki J, Andersson S (2004) Multi-species laser-based imaging measurements in a Diesel spray, SAE Technical Paper Series 2004-01-1917 Soave J (1972) Equilibrium constants from a modified Redlich– Kwong equation of state. Chem Eng Sci 27:1197–1203 Vogel AI (1946) Physical properties and chemical constitution of aliphatic hydrocarbons. J Chem Soc 9:133–142 Winter J, Dittus B, Kerst A, Muck O, Schulz R, Vogel A (2004) Nozzle hole geometry—a powerful instrument for advanced spray design. In: Proceedings of THIESEL Conference, Valencia Zhao H, Ladommatos N (1998) Optical diagnostics for in-cylinder mixture formation measurements in IC engines. Progr Energy Combust Sci 24:297–336 Zhao Z, Li J, Kazakov A, Dryer FL, Zeppieri SP (2005) Burning velocities and a high temperature skeletal kinetic model for n-decane. Combust Sci Tech 177:89–106

123