SANDIA REPORTSAND2014-15329Unlimited ReleasePrinted June 2014

Advanced Diagnostics for High Pressure Spray Combustion

Scott A. Skeen, Julien Manin, and Lyle M. Pickett

Prepared bySandia National LaboratoriesAlbuquerque, New Mexico 87185 and Livermore, California 94550

Sandia National Laboratories is a multi-program laboratory managed and operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Corporation, for the U.S. Department of Energy's National Nuclear Security Administration under contract DE-AC04-94AL85000.

Approved for public release; further dissemination unlimited.

2

Issued by Sandia National Laboratories, operated for the United States Department of Energy by Sandia Corporation.

NOTICE: This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government, nor any agency thereof, nor any of their employees, nor any of their contractors, subcontractors, or their employees, make any warranty, express or implied, or assume any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represent that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise, does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government, any agency thereof, or any of their contractors or subcontractors. The views and opinions expressed herein do not necessarily state or reflect those of the United States Government, any agency thereof, or any of their contractors.

Printed in the United States of America. This report has been reproduced directly from the best available copy.

Available to DOE and DOE contractors fromU.S. Department of EnergyOffice of Scientific and Technical InformationP.O. Box 62Oak Ridge, TN 37831

Telephone: (865) 576-8401Facsimile: (865) 576-5728E-Mail: reports@adonis.osti.govOnline ordering: http://www.osti.gov/bridge

Available to the public fromU.S. Department of CommerceNational Technical Information Service5285 Port Royal Rd.Springfield, VA 22161

Telephone: (800) 553-6847Facsimile: (703) 605-6900E-Mail: orders@ntis.fedworld.govOnline order: http://www.ntis.gov/help/ordermethods.asp?loc=7-4-0#online

3

SAND2014-15329Unlimited ReleasePrinted June 2014

Advanced Diagnostics for High Pressure Spray Combustion

Scott A. SkeenJulien Manin

Lyle M. PickettEngine Combustion

Sandia National LaboratoriesP.O. Box 969, MS 9053Livermore, CA 94550

Abstract

The development of accurate predictive engine simulations requires experimental data to both inform and validate the models, but very limited information is presently available about the chemical structure of high pressure spray flames under engine-relevant conditions. Probing such flames for chemical information using non-intrusive optical methods or intrusive sampling techniques, however, is challenging because of the physical and optical harshness of the environment. This work details two new diagnostics that have been developed and deployed to obtain quantitative species concentrations and soot volume fractions from a high-pressure combusting spray. A high-speed, high-pressure sampling system was developed to extract gaseous species (including soot precursor species) from within the flame for offline analysis by time-of-flight mass spectrometry. A high-speed multi-wavelength optical extinction diagnostic was also developed to quantify transient and quasi-steady soot processes. High-pressure sampling and offline characterization of gas-phase species formed following the pre-burn event was accomplished as well as characterization of gas-phase species present in the lift-off region of a high-pressure n-dodecane spray flame. For the initial samples discussed in this work several species were identified, including polycyclic aromatic hydrocarbons (PAH); however, quantitative mole fractions were not determined. Nevertheless, the diagnostic developed here does have this capability. Quantitative, time-resolved measurements of soot extinction were also accomplished and the novel use of multiple incident wavelengths proved valuable toward characterizing changes in soot optical properties within different regions of the spray flame.

4

5

ACKNOWLEDGMENTS

Technical support for this work was provided by Chris Carlen, Dave Cicone, Keith Penney, Paul Schrader, and Paul Fugazzi. We also thank Dr. Nils Hansen, Dr. Arnas Lucassen, and Dr. Yongjin Jung for additional support. The experiments were conducted at the Combustion Research Facility, Sandia National Laboratories, Livermore, CA. Support for this work (SAS, LMP) was provided by Sandia National Laboratories under the Early Career Laboratory-Directed Research and Development (EC-LDRD) program and by the U.S. Department of Energy, Office of Vehicle Technologies. Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the United States Department of Energy’s National Nuclear Security Administration under contract DE-AC04-94AL85000.

6

7

CONTENTS

1. Introduction...............................................................................................................................11

2. Experimental.............................................................................................................................132.1 High-Pressure Pre-Burn Vessel .....................................................................................132.2. High-Speed Sampling at High-Pressure ........................................................................13

2.2.1 Valve selection and performance.....................................................................132.2.2 High-Pressure Valve Mount and Seal..............................................................142.2.3 Sampling Probe and Collection Vessel............................................................142.2.4 Sampling Conditions .......................................................................................16

2.3 High-Speed Soot Extinction Imaging ............................................................................16

3. RESULTS .................................................................................................................................213.1 Gas Sampling.................................................................................................................21

3.1.1 Premixed Reactant Sampling...........................................................................213.1.2 Pre-burn Product Sampling..............................................................................223.1.3 High-Pressure Spray Flame Sampling.............................................................24

3.2 High-Speed Multi-Wavelength Extinction Imaging......................................................263.2.1 Optical Thickness and Soot Volume Fraction......................................................263.2.2 Unique Results from Multi-Wavelength Extinction ............................................283.2.3 Ambient Temperature, Density, and Oxygen Effects on Total Soot Mass ..........30

4. Conclusions and Future Work ...................................................................................................32

5. References.................................................................................................................................34

Distribution ....................................................................................................................................36

FIGURES

Figure 1. Response time of high-speed sampling valve. ..............................................................14Figure 2. High-pressure sampling valve window without valve (left) and with valve (right) .....14Figure 3. High-pressure sampling window with custom probe. ...................................................15Figure 4. Still image of sampling probe drawing gases from lift-off region of n-dodecane spray flame. .............................................................................................................................................16Figure 5. Schematic of experiental setup for high-speed extinction imaging of high-pressure sprays. ............................................................................................................................................17Figure 6. Mass spectrum of premixed reactant gases from pre-burn combustion vessel. ............21Figure 7. Mass spectrum of premixed product gases after pre-burn combustion event sampled near vessel wall where quenching of the pre-burn reactions has occurred....................................23Figure 8. Mass spectrum of premixed product gases after pre-burn combustion event. ..............24Figure 9. Mass spectrum of gas species sampled from the lift-off region of a n-dodecane spray flame at an ambient temperature of 900 K and an ambient pressure of 60 bar. ............................25Figure 10. Time-sequenced images of the soot optical thickness (KL) at Spray A conditions (i.e., T = 900 K, ρ = 22.8 kg/m3 and 15% O2) as measured by the blue LED. ......................................26

8

Figure 11. Averaged images of the soot optical thickness (KL), as measured by the blue LED, for the 850 K – 1200 K ambient temperature sweep at an ambient density of 22.8 kg/m3 and ambient O2 concentration of 15% (by volume). The lift-off length is indicated by the dashed line........................................................................................................................................................27Figure 12. Soot volume fraction (fv), as measured by the blue LED, in the plane corresponding to center of the flame axis for the 850 K – 1200 K ambient temperature sweep at an ambient density of 22.8 kg/m3 and ambient O2 concentration of 15% (by volume). The lift-off length is indicated by the dashed line...........................................................................................................28Figure 13. Difference between fv corresponding to green LED extinction and fv corresponding to blue LED extinction. The top and bottom panels correspond to the 900 K (Spray A) and 1200 K conditions, respectively, at an ambient density of 22.8 kg/m3 and ambient O2 concentration of 15% by volume. .............................................................................................................................29Figure 14. Soot mass in thin (0.1 mm) radial cross-sections of the jet as a function of axial distance from the injector orifice. Ambient density and O2 concentration were 22.8 kg/m3 and 15% by volume, respectively.........................................................................................................31Figure 15. Inverse lift-off length, 1/H (curves) and total soot mass (bar charts) as a function of ambient temperature. .....................................................................................................................32

9

NOMENCLATURE

αsa Ratio of scattering to absorptionDf Fractal dimensiondp Primary particle diameterDI Direct-InjectionDOE Department of EnergyE(m) Imaginary part of (m2-1)/(m2+2)EGR Exhaust gas recirculationfps Frames per secondfv Soot volume fractionH Lift-off lengthHeNe Helium-NeonI Intensity of light after passing through a medium of interestI0 Intensity of incident light prior to passing through a medium of interestke Non-dimensional extinction coefficientK Dimensional extinction coefficientKL Optical thicknessL Path lengthλ WavelengthLE Laser extinctionLED Light-emitting diodeLIF Laser-Induced FluorescenceLII Laser-Induced Incandescencem Complex refractive indexND Neutral densityNp Number of primary particlesPAH Polycyclic Aromatic HydrocarbonsPLIF Planar laser-induced fluorescencePM Particulate Matterppm Parts per millionRDG Rayleigh-Debye-GansTOF-MS Time-of-flight mass spectrometer

10

11

1. INTRODUCTION

The development of predictive high-pressure spray combustion simulations will enable more rapid and cost-effective engine design that leverages fundamental physical and chemical information to reduce pollutant emissions and maximize efficiency. Anticipated advancements are crucial to the transportation sector as more stringent emissions regulations are imposed and new synthetic and bio-derived fuels with varying properties find greater usage. Presently, limited quantitative experimental information is available regarding the chemical structure of high pressure spray flames under engine-relevant conditions. Such data are needed to further develop existing simulation capabilities; however, probing such flames for chemical information using non-intrusive optical methods or intrusive sampling techniques is difficult because of the physical and optical harshness of the environment. For example, such flames are characterized by high velocities, temperatures, and pressures that make intrusive probing difficult. Further, beam steering effects caused by refractive index gradients are exacerbated at high pressure, complicating optical diagnostics. And finally, engine-relevant injection times are only a few milliseconds resulting in very small time-windows for measurements under quasi-steady conditions.

Particulate matter (PM) (mainly soot) emissions are of particular concern in direct-injection (DI) diesel engines. More stringent PM emissions standards for diesel engines have led to the use of additional exhaust after-treatment such as diesel particulate filters, which increase the cost of diesel engine production while negatively impacting fuel economy. A more economical approach to reducing soot emissions from DI diesel engines lies in the development of combustion strategies to reduce or eliminate soot in-cylinder. The successful development of such strategies, by experimental or computational means, can benefit from quantitative measurements of gas-phase soot precursor species and time-resolved, quantified measurements of soot volume fraction (fv) in high-pressure spray flames as a function of the available operational parameters.

The present work leverages the physical and optical access available in a pre-burn spray vessel [1], which simulates the thermodynamic conditions of modern diesel engines, to develop and apply sampling and optical diagnostics for the characterization of pre-burn reactant and product gases, gas-phase species within the lift-off region, and the temporal evolution of soot formation in n-dodecane spray flames. Prior to the present work, measurements of the minor gas-phase species following the pre-burn event have not been made. Nesbitt et al. [2] simulated the formation of minor species to investigate the effects of nitric oxides (NO, NO2) and OH on n-heptane ignition delay times. In that work, NO was the most abundant of the three minor species discussed, but remained below equilibrium values due to the relatively low peak temperatures achieved during the pre-burn event. Higher concentrations of these species also resulted in shorter ignition delay times, with NO showing the most influence; however, these effects are small relative to the effect of changing the oxygen concentration by exhaust gas recirculation (EGR). In the present work, the pre-burn products are sampled and investigated in an off-line time-of-flight mass spectrometer (ToF-MS) to determine which minor species can be identified and potentially quantified. Further, gas-phase species are also extracted from the lift-off region of an n-dodecane spray flame for identification and quantification where possible. Planar laser-induced fluorescence (PLIF) measurements using an excitation wavelength of 355 nm have demonstrated that polycyclic aromatic hydrocarbons (PAH), which are believed to be precursors

12

to soot particles, are formed downstream of the spray flame lift-off length (H) [3, 4]. In general, excitation with 355-nm radiation is only effective for producing fluorescence in condensed aromatic species having three or more rings. Consequently, higher energy (shorter wavelength) lasers are needed to investigate the formation of the first aromatic ring (e.g., benzene or toluene) [5]. LIF measurements can typically only provide a qualitative assessment of the PAH present in the spray flame. This is because PAH have broad fluorescence emission spectra and many species have spectra that overlap. Consequently, even species identification is formidable if not impossible. The present work addresses the need to identify soot precursor species within the lift-off region of a high-pressure n-dodecane spray flame. Such measurements are the first of this kind.The extensive optical access of the pre-burn spray vessel also permits the development and application of a novel high-speed, quantitative soot diagnostic. In optical engines and high-pressure spray rigs, laser extinction (LE) and laser-induced incandescence (LII) represent the most common in-situ optical diagnostics used to measure fv in flames. In some cases, laser scattering measurements have been coupled to LE measurements to obtain additional information about primary particles and aggregates [6]. While single-color LE measurements alone can only provide information about fv, LII measurements in conjunction with an accurate physical model can yield information about fv and primary particle size. In both cases, knowledge of the particles’ optical properties (i.e., refractive index) is required. The uncertainty in the refractive index of soot is well-documented and is generally accepted to be large [7]. The vast majority of LE and LII measurements have been made at atmospheric pressure and under steady, laminar flame conditions.Prior studies in the Sandia pre-burn spray vessel have used 2-D planar LII in conjunction with path-averaged LE to obtain quantitative fv measurements during the quasi-steady period of spray combustion at pressures and temperatures relevant to diesel engines [8]. Because the duration of the injection and combustion event for these studies is less than 10 milliseconds, the 10 Hz Nd:YAG lasers commonly used for LII provide only a single image per injection, but no information about the temporal evolution of soot during a single injection event. This drawback may be avoided by using high-power burst lasers for high repetition-rate LII [9]; however, this technology is still under development and we anticipate that cost will prohibit the widespread use of such systems in the near-term. Another weakness of 2-D planar LII is exposed when soot concentrations are high enough to cause significant laser attenuation. As an example, for an optical thickness (KL) of 1, less than 40% of the original laser energy is transmitted to soot particles later in the laser path. At KL = 2, almost 90% of the laser energy has been absorbed. In atmospheric pressure flames, laser fluences (laser energy per unit area) greater than ~ 0.2 J/cm2

(for 532 nm light) heat soot particles to their sublimation temperature and the LII signal is relatively insensitive to laser fluence variations [10]. For this reason, LII experiments seeking quantitative measurements of soot volume fraction typically operate in this so-called “plateau” regime [10]. As flames become optically thick, however, LII signal from particles found later in the laser path can be significantly reduced. Further, dense soot clouds outside the laser sheet can block incandescence signal from reaching the detector—a phenomenon referred to as signal trapping. Compensating for these effects can be difficult and in general yield a negative impact on soot level quantification [11]. A key objective in the present work was the development of a quantitative, high-speed diagnostic that could avoid the aforementioned complexities of LII. As such, this work presents the development of a high-speed extinction imaging technique for the investigation of the temporal evolution of fv in high-pressure, high-temperature spray flames.

13

2. EXPERIMENTAL

2.1 High-Pressure Pre-Burn Vessel

The high-pressure pre-burn spray vessel used in this work has been described in detail previously [1]. Briefly, the vessel is nearly cubical with a volume of approximately 1 L and was designed to simulate the thermodynamic conditions of modern direct-injection diesel engines. The vessel is rated to safely achieve pressures as high as 350 bar at temperatures up to 1400 K. Prior to and during operation, the temperature of the entire vessel is maintained at 188 ºC by electric heaters. Spark-ignition of a combustible gas mixture rapidly elevates the pressure and temperature within the vessel. When imaging of the combustion event is desired, large (~100 mm) sapphire windows provide optical access from four sides. For the sampling studies described in the present work, a new metal window was designed to hold the customized sampling valve. Its design will be discussed in a subsequent section of this report. Following the pre-burn combustion event, the spray is injected when the desired thermodynamic conditions are achieved after a short cool-down period. The spray combustion investigated in this study used n-dodecane (C12H26) as fuel and a common-rail diesel fuel injector equipped with a single 90-μm diameter orifice (# 370) belonging to the family of Spray A ECN injectors [12]. The injection pressure was 1500 bar and the electronic injection duration was set to 6.5 ms resulting in an 8 ms actual injection period.

2.2. High-Speed Sampling at High-Pressure

2.2.1 Valve selection and performance

The high-speed valve system selected for this work was manufactured by Parker Hannafin Corporation’s Precision Fluidics Division (Part Number 009-0181-900). This solenoid valve has a 0.79-mm orifice and is driven by an Iota One® pulse valve controller. This system was selected because the shortest manufacturer specified valve response time was 160 µs, which met the need for a fast opening/closing valve that would complete sampling within 1.5 to 7 ms. To be certain of the sampling valve’s performance, its response time was tested using a high-speed camera equipped with a long-distance microscope. A 635-nm diode laser was used to illuminate the valve orifice and poppet. Both the imaging system and the poppet motion were slightly off-camber such that the axial poppet motion could be easily tracked based on the vertical and horizontal location of the poppet tip. In Figure 1, we present a scatter plot of the poppet position for a number of repeat experiments showing the time delay between the trigger signal and the first valve movement as well as the time between trigger to fully open.

Based on the results presented in Figure 1, the sampling valve begins to open approximately 300 µs after the electronic trigger signal and is fully open between 400 and 500 µs after trigger. These timings are used to ensure that valve opening occurs after the quasi-steady spray flame has been established as will be discussed in more detail below. A similar analysis during valve closing showed an equivalent time period for the transition from valve fully-open to fully-closed

14

Figure 1. Response time of high-speed sampling valve.

2.2.2 High-Pressure Valve Mount and Seal

The valve described above was also selected because its radial dimension on the potential sealing surface nearly matched the dimension of existing high-temperature, high-pressure seals used on low-speed intake and exhaust valves associated with the pre-burn vessel. Because of this, the high-pressure valve mount window and sealing surfaces could be designed for use with existing parts. Photographs of the valve mount window with and without the sampling valve installed are shown in Figure 2.

Figure 2. High-pressure sampling valve window without valve (left) and with valve (right)

2.2.3 Sampling Probe and Collection Vessel

A custom made cover and sampling probe was mounted to the face of the high-speed sampling valve. The manufacturer specified AS-568a-901 o-ring formed the seal between the valve face

15

and sampling probe base. The sampling probe was constructed from 1.5 mm OD, 0.56 mm ID, Type 316 stainless steel tubing, which was welded to a 3 mm thick stainless steel baseplate as shown in Figure 3. The probe length was specified to sample at the center axis of the spray flame.

Figure 3. High-pressure sampling window with custom probe.

The exit of the sampling valve was connected to a ¼” stainless steel Swagelok® “cross” junction. One port of the “cross” junction was connected to a Convectron vacuum gauge. A second port was connected to a 500 mL stainless steel container equipped with a ¼-turn shut-off valve. A third port “cross” junction was connected to a “T” junction. One port of the “T” junction was connected to a bottle of high-purity argon, while the second port was connected to a vacuum pump. Prior to any sampling, the collection bottle and associated lines were pumped down below 25 mTorr. The sampling bottle was then filled with 50-100 Torr of high-purity argon. While the argon acts as a diluent to reduce molecular collisions, its most important role is to provide a known concentration of a species not found in the flame gases to serve as a reference in the mass spectrometry analysis. Signal intensities from the mass spectrometer are made quantitative by determining a calibration constant for the species of interest relative to a species of known concentration. The signal intensity for a given species in the ToF-MS is related to the individual species’ mole fraction, its electron ionization cross section, its mass discrimination factor, and the electron flux from the ion source. In addition, there is an arbitrary instrument constant related to the design and voltage settings of the ion optics. If the ratio of the signal intensity from two species is considered, only the species specific cross sections and mass discrimination factors remain. By using a calibration gas, the ratio of these values for two species can be determined. To determine the oxygen and nitrogen concentrations in the pre-burn reactants and products, a calibration gas mixture of 20% oxygen, 20% nitrogen, and 60% argon was used. The ratio of the product of the oxygen cross section and mass discrimination factor to that of argon was determined to be 5.2 using an electron energy of 12 eV. Under these same conditions, the ratio for nitrogen to argon was determined to be 0.43. Upon increasing the electron energy to 15 eV to obtain more ion signal from nitrogen, these ratios were determined to be 35 and 0.61 for oxygen to argon and nitrogen to argon, respectively. In addition to using argon as a reference for quantitative mole fractions, calibration mixtures having different concentrations of oxygen were sampled to determine if the mass spectrometer response was linear with molar concentration.

16

2.2.4 Sampling Conditions

Three different sampling conditions were considered. The first test involved sampling the pre-burn reactants that contain hydrogen, acetylene, oxygen, and nitrogen. This test allowed us to confirm that oxygen and nitrogen concentrations in the pre-burn reactants meet the specified tolerances when filled manometrically. As specified previously, samples were collected only

after pumping the collection vessel and associated lines down below 25 mTorr and subsequently filling the collection vessel to 50 Torr with high-purity argon. The samples were then collected

by purging the vessel with the premixed reactants five times before filling the combustion vessel a sixth time to a pressure of 200 psi. The sampling valve was then activated repeatedly for 500

ms until the desired pressure was achieved in the collection vessel. The second sampling condition involved acquiring sample following the pre-burn event to quantify the oxygen and

nitrogen concentrations in the pre-burn products and to determine if minor species such as NOx could be detected. As described in the introduction, the presence of NOx in the pre-burn products

could impact ignition delay times. Following the aforementioned pumping and argon filling procedures, the pre-burn products were collected by activating the sampling valve for 500 ms at

the timing associated with fuel injection for the Spray A condition. Because this set of experiments did not involve a spray, only the pre-burn products were collected for offline

analysis. The third sampling condition involved drawing sample from within the lift-off region of the n-dodecane spray flame. A still image from a high-speed movie acquired during sampling

after the flame has reached the quasi-steady period is shown in Figure 4Figure 4. Still image of sampling probe drawing gases from lift-off region of n-dodecane

spray flame.

2.3 High-Speed Soot Extinction Imaging

As discussed in the introduction to this work, the use of LII in optically thick flames can be challenging due to laser attenuation and signal-trapping. Moreover, high-pressure spray flames are extremely turbulent requiring high-speed diagnostics to extract time-resolved information. Presently, the cost and technical limitations of high-speed, high-power lasers and high-speed intensified cameras prohibits their wide-spread use for high-speed LII. In the present work we have developed a high-speed extinction imaging diagnostic to addresses the need for a lower-cost, quantitative, high-speed soot diagnostic that can be applied to optically thick flames.

The main components of the high-speed imaging extinction setup consist of high-output, ultra-fast light-emitting diodes (LEDs), a dichroic beam splitter to permit the use of two different incident wavelengths of light, engineered diffusers, a field lens, and a high-speed camera (running at 82 kfps) equipped with a dual-bandpass filter. The dual-bandpass filter was customized to transmit the wavelengths of incident light used for a specific experiment. A neutral density (ND) filter was also added to reject broadband flame luminosity. The optical density of the ND filter can be changed as needed depending on the wavelengths transmitted by the dual-bandpass filter. The diffusers have a 50o circular pattern and the field lens is a Fresnel type with a focal length of 152 mm. The engineered diffusers in conjunction with the Fresnel lens create a large (60-mm diameter) uniformly illuminated background with minimal light intensity losses (more than 90% transmission) [13]. This light source setup directs a bundle of

17

diffused rays toward the object plane resulting in minimal light extinction by density gradients (i.e., beam steering, schlieren effects). A schematic of the combustion vessel and optical arrangement used in this work is provided in Figure 5.

Figure 5. Schematic of experiental setup for high-speed extinction imaging of high-pressure sprays.

The background corrected extinction imaging measurements were converted to optical thickness, KL, using the well-known Beer-Lambert law.

I/I0 = exp(-KL) (1)

In Eq. 1, I is the transmitted LED intensity, I0 is the incident intensity, K is the dimensional extinction coefficient, and L is the path length through the soot cloud. We corrected the transmitted intensity for flame luminosity by capturing a frame between the blue and green LED pulses when only flame luminosity was present (i.e. with both LEDs turned off). For the highest ambient temperature cases (1100 and 1200 K), we observed infrequent, intermittent localized regions where flame luminosity was larger than the sum of the transmitted LED intensity and the flame luminosity in the surrounding frames. This phenomenon was attributed to the rapid spatial fluctuations in extinction and flame emission. Specifically, local regions of high extinction and high flame emission observed during the LED pulse move both downstream and stochastically during the time between two captured frames. Spatial mismatches between a region of high extinction and high flame emission can thus result in negative values for the background corrected transmitted intensity, I. We minimized these effects via a weighted temporal averaging scheme. Nevertheless, our analysis required the specification of an upper threshold limit for KL. We assessed the uncertainty associated with this phenomenon in the higher ambient temperature cases as ±10%, which, for example, is well below the uncertainty due to variation in soot refractive index.

18

We derived soot volume fraction, fv, from the KL images using the following relationships based on small particle Mie theory.

fv = Kλ/ke (2)

ke = (1+αsa)6πE(m) (3)

In Eqs. 2 and 3, λ is the wavelength of incident light, ke is the non-dimensional extinction coefficient, αsa is the scattering-to-absorption ratio, and E(m) is the imaginary part of (m2-1)/(m2+2), in which m is the complex refractive index of soot.

We calculated the non-dimensional extinction coefficient (ke) using Rayleigh-Debye-Gans (RDG) theory for fractal aggregates following the work of Köylü and Faeth [14, 15]. In our analysis, we specified the number of primary particles per aggregate (Np) as 150, with a primary particle diameter (dp) of 16 nm (Aizawa et al.[16]; Kook and Pickett [17]). The fractal dimension, Df, associated with the above parameters is 1.79. As noted by Shaddix et al. [18], the RDG scattering approximation is considered to be accurate provided that Df < 2. For simplicity, the refractive index was held constant at m = 1.75-1.03i (Williams et al. [19]) for the two wavelengths evaluated (406 nm, 17 nm FWHM and 519 nm, 34 nm FWHM) resulting in ke values of 7.76 and 7.46 for the blue and green light, respectively. The refractive index reported by Williams et al. [19] is considered more appropriate for the present work because they derived m by measuring the ke of soot particles within the flame, as opposed to prior studies that measured the ke of post-flame soot. We understand that the refractive index of soot depends on the soot morphology, primary particle size and structure, and the wavelength of incident light. A full analysis on the effects of assumed dp, Np and m on our evaluation of fv, as well as the uncertainty associated with the use of RDG theory for large m and when multiple internal scattering may be important, is beyond the scope of this work, but has been addressed in a journal publication [20].

At the pressures and temperatures of this study, extinction due to beam steering effects could not be completely avoided. Because our diffused light source is not perfectly Lambertian, beam steering can increase local transmission if diffused ray bundles are preferentially steered toward a common image point. Conversely, extinction occurs at locations where light is steered away from the expected path. Upon time averaging, the measured extinction under non-sooting flame conditions was near zero. In the individual time-resolved images, however, extinction caused by beam steering defined our lower KL detection limit. For non-sooting cases, this lower limit was determined to be approximately 0.1. For path lengths on the order of 10 mm and a non-dimensional extinction coefficient (ke) near 8, this corresponds to a lower detection limit of about 1 ppm for fv. Based on the factors considered above, we assign a conservative uncertainty of ± 15% to the quantitative soot volume fractions presented below.

For validation of the extinction imaging technique, laser-extinction measurements were made using a 10-mW, 1-mm HeNe (632.8 nm) laser beam gated at 100 kHz by an acousto-optic modulator. The full details of this setup were described by Musculus and Pickett [21]. The HeNe laser passed through the flame axis 60 mm downstream of the injector orifice. Collection optics and components for the laser extinction measurement included a 50-mm diameter, 200-mm focal

19

length lens to collect the laser after passing through the testing section, a 100-mm integrating sphere, a 633 nm (1 nm FWHM) bandpass filter sandwiched between a 75-mm focal length lens pair, and a photodiode. Previous work has demonstrated this system’s insensitivity to extinction caused by beam-steering [21].

20

21

3. RESULTS

3.1 Gas Sampling

3.1.1 Premixed Reactant Sampling

As described above, the premixed reactants contain hydrogen, acetylene, nitrogen, and oxygen. A mass spectrum showing the observed peaks from a representative sample of the premixed reactants is shown in Figure 6. The electron ionization source voltage was 12 eV for this spectrum and it consists of 221 sweeps. A sweep refers to a single trigger event on the mass spectrometer’s ion optics. Thus, the spectrum shown below represents the summation of ion signal for more than two million sweeps. At an electron energy of 12 eV, signal from hydrogen was not observed. As such, the ordinate is limited to a mass range from 24 amu to 60 amu. Hydrogen is not observed for two reasons. First, hydrogen has an ionization threshold of ~15.4 eV which is significantly higher than the setpoint of 12 eV; and, although the ion source produces electrons over a wide range of energies above and below the setpoint, the concentration of hydrogen in the premixed reactants is relatively low (0.5%) resulting in a low probability for hydrogen ionization. Second, the detection efficiency is mass dependent and is approximately a factor of 10 lower for m/z = 2 relative to m/z = 26, for example. It is also important to note that water (m/z = 18) is not observed in the spectrum. Water has an ionization threshold of 12.62 eV, thus it would be easily ionized at the current ion source energy given its broad energy distribution. The lack of water in this spectrum is an indication that the vacuum system is free from leaks (since water is present in the room air) and that the vacuum system has been pumped down sufficiently to remove residual water from internal surfaces. Thus, very little background signal for oxygen and nitrogen are anticipated.

Figure 6. Mass spectrum of premixed reactant gases from pre-burn combustion vessel.

22

The acetylene (m/z=26), nitrogen (m/z = 28), and oxygen (m/z = 32) peaks are easily identified, while the peak at m/z = 58 is attributed to the acetone in which the acetylene supply is dissolved. In the inset axis of Figure 6 we show three additional peaks. The peak at m/z = 25 is C2H, which begins to fragment from acetylene at electron energies around 16.7 eV. The peak at m/z = 27 represents the carbon-13 isotope of acetylene. The peak at m/z = 43 is assigned to C2H3O, which fragments from acetone at electron energies greater than about 10.3 eV. The presence of this fragment provides additional confirmation that m/z = 58 is indeed acetone.

3.1.2 Pre-burn Product Sampling

As described above, the pre-burn combustion event raises the pressure and temperature within the spray vessel to meet the thermodynamic conditions of modern direct-injection diesel engines. By changing the composition of the premixed reactants, different levels of exhaust gas recirculation (EGR) can be simulated. If a stoichiometric reactant premix is used, the resulting products should contain 0% oxygen. Such cases are used to examine non-reacting spray conditions. For example, non-reacting sprays experiments using high-speed schlieren imaging are used to measure the rate of fuel vapor penetration [22]. For the Spray A condition investigated in this work, the desired oxygen concentration remaining after the pre-burn is 15%. The other major products expected include water and CO2. Near the spray vessel walls and within seal crevices, the pre-burn combustion event may be quenched leaving hydrocarbon residuals that could potentially mix into the core of the vessel and influence the spray combustion event. To investigate this, gas samples were extracted through an exhaust valve near the vessel wall as well as from the core of the vessel using the sampling probe described above. A representative mass spectrum from the sampled pre-burn product gases near the vessel wall is shown in Figure 7. The major products of combustion described above (water and CO2) are easily identified at m/z = 18 and m/z = 44, respectively. Nevertheless, the spectrum clearly indicates that flame quenching near the walls results in residual acetylene along with other hydrocarbon species. A large number of unidentified peaks are observed because at the relatively high electron energy used (12 eV) for ionization many hydrocarbon species are easily fragmented. We note also that both CO and N2 are expected at m/z = 28, since quenching of the reactions would inhibit complete oxidation of CO to CO2.

23

Figure 7. Mass spectrum of premixed product gases after pre-burn combustion event sampled near vessel wall where quenching of the pre-burn reactions has occurred.

A representative mass spectrum of the pre-burn product gases sampled from the core of the vessel, which have been diluted with argon, is shown in Figure 8. As before, water and CO2 are easily identified. The inert nitrogen is most likely the only species present at m/z = 28 since the stoichiometry of this mixture is fuel lean (i.e., CO is expected to completely oxidize to CO2). Nevertheless, at a single electron energy it is not possible to differentiate between ion counts due to nitrogen and those potentially due to CO. Future experiments using an energy sweep or a different ion source could confirm that CO is not present in the pre-burn products. The argon-36 and argon-38 isotopes are also apparent in this spectrum. Argon-36 constitutes around 0.3% of the naturally occurring argon while argon-38 constitutes only about 0.06%. The mass peak observed at m/z = 16 is most likely attributed to an oxygen atom fragmenting from water, while the mass peak at m/z = 14 can be attributed to fragmentation of N2. One of the key observations from this spectrum is the lack of signal from acetylene (m/z = 26) or other hydrocarbon species. The lack of acetylene indicates that even with the small disturbance caused by the sampling probe, the combustion event in the core of the vessel proceeds to completion. Moreover, residual hydrocarbons observed in samples drawn near the walls due to quenching of the pre-burn combustion reactions are not transported into the core of the vessel and therefore are not expected to influence the spray combustion event. Lastly, nitric oxide (NO) is not observed in this spectrum; however, future work will investigate higher temperature pre-burn events in which significant thermal NOx may be formed. These future studies will investigate the effect of NO on ignition delay times.

24

Figure 8. Mass spectrum of premixed product gases after pre-burn combustion event.

3.1.3 High-Pressure Spray Flame Sampling

As described above and demonstrated in Figure 4, gas samples were extracted from the lift-off region of a high-pressure (60 bar) n-dodecane spray flame. A representative mass spectrum of the species sampled from the lift-off region is shown in Figure 9. With the electron ionization source at an energy of 12 eV, it is not feasible to identify all of the peaks observed in this spectrum since some may be the result of larger species fragmenting; however, the hydrocarbon species are easily grouped based on the number of carbon atoms. Further, some of the strongest peaks in the spectrum likely correspond to the most common combustion intermediates and the most stable aromatic species.

The species that can be identified with the greatest confidence are as follows: at m/z = 42, propene (C3H6) and ketene (C2H2O), at m/z = 54 1,3-butadiene (C4H6), at m/z = 66 cyclopentadiene (C5H6), at m/z = 78 benzene (C6H6), at m/z = 92 toluene (C7H8), and at m/z = 104 styrene (C8H8). At m/z = 116, indene is the most commonly anticipated species; however, several other isomers may also be present. Recent aerosol mass spectrometry experiments indicate that phenylallene as well as other propargyl-substituted phenyl-rings likely contribute to the ion signal at this mass [23]. Finally, naphthalene is typically recognized as the most thermodynamically stable species at m/z = 128.

25

Figure 9. Mass spectrum of gas species sampled from the lift-off region of a n-dodecane spray flame at an ambient temperature of 900 K and an ambient pressure of 60 bar.

Overall, these results show promise toward a better chemical characterization of engine-relevant high-pressure spray flames. Although the present study was unable to quantify the mole fractions of the observed species, with some additional effort quantification for several species should be possible. Nevertheless, even qualitative axial profiles of combustion intermediates will be highly valuable for comparison with modeling results. Thus, efforts should be made to sample at several axial distances along the spray flame axis. Understanding the pressure dependencies that exist in the formation and growth of aromatic rings is also a valuable target for further work. By using an ion source with more tunability and improved resolution, isomers can be identified leading to a better understanding of the aromatic chemistry.

26

3.2 High-Speed Multi-Wavelength Extinction Imaging

3.2.1 Optical Thickness and Soot Volume Fraction

Temporally resolved images showing the progression of the soot optical thickness (KL) during injection at the Spay A condition are presented in Figure 10. In these images, the fuel injector is located to the left and the flame is propagating from left to right. The KL values in this figure correspond to extinction of the blue (406 nm) LED at Spray A conditions. Under these conditions, soot is first detected at an axial location approximately 35 mm from the injector orifice less than 1 millisecond after the start of injection (ASOI). Though KL time-sequences are not shown here for the other ambient conditions studied, the axial distance from the injector orifice to the location where soot is first observed and the time to soot ASOI decrease with increasing temperature, density, and oxygen concentration.

Figure 10. Time-sequenced images of the soot optical thickness (KL) at Spray A conditions (i.e., T = 900 K, ρ = 22.8 kg/m3 and 15% O2) as measured by the blue LED.

The images in Figure 10 show that a large amount of soot forms in the head of the jet as it propagates into the ambient. The soot cloud associated with the head also begins separating from the upstream soot cloud in the region between 50 and 60 mm at about 2 ms ASOI. At 2.75 ms ASOI, the head has completely separated from the upstream soot cloud. This occurs presumably due to rapid entrainment of the ambient oxidizer behind the jet head leading to a brief transient period of enhanced mixing and soot oxidation. Such phenomena were observed here under a wide range of ambient conditions, and may be important in understanding the efficacy of multiple injection strategies [24].

The quasi-steady time period of the spray flame begins after the head of the jet passes through the camera’s field of view and ends at injector closing. We determined K for use in Eq. 2 via tomographic inversion of time- and ensemble-averaged KL images from multiple identical injection events. Time-averaging was performed over the quasi-steady period of the flame

27

(approximately 4 ms, representing more than 100 images per injection and per color). Ensemble averaging included a minimum of 5 injections for each specific experimental condition. The averaged KL images result in a relatively smooth (i.e., free from stochastic variations) and nearly axisymmetric KL field as shown in Figure 11. Note that the results in Figure 11 only correspond to extinction of the blue LED. Results for the green LED will be discussed later. When comparing these five cases, we first observe that the maximum KL under the 1200 K ambient condition is more than one order of magnitude larger than that at an ambient temperature of 850 K. As previously mentioned, Figure 11 also shows that soot formation begins closer to the injector orifice for the higher temperature cases. In fact, at 1200 K soot seems to form outside the leftmost region of the visualization window. This was confirmed by inspection of the luminosity images.

Figure 11. Averaged images of the soot optical thickness (KL), as measured by the blue LED, for the 850 K – 1200 K ambient temperature sweep at an ambient density of 22.8

kg/m3 and ambient O2 concentration of 15% (by volume). The lift-off length is indicated by the dashed line.

We performed the tomographic reconstruction of the KL field using an inverse Radon transform, which then returns fv after Eq. 2 has been applied. Figure 12 shows the 2-D fv field in the plane corresponding to the center of the flame axis for the 850 K – 1200 K conditions at an ambient density of 22.8 kg/m3 and ambient O2 concentration of 15% by volume. The plane cut in the

28

tomographic reconstruction yielding fv more clearly reveals the higher soot concentrations near the periphery of the 1100 and 1200 K cases at the flame’s base. These soot “wings” as they have been called, provide further evidence that spray flames having shorter lift-off lengths have partially-premixed mixtures that are more fuel-rich [11]. An interesting feature is observed in the 1100 K and 1200 K cases at an axial position between about 35 mm and 50 mm from the injector orifice. Near 35 mm, fv appears to decrease in the central region of the flame. The phenomenon is more apparent for the 1200 K case. We suspect that this may be due to the intermittent localized regions in the 1100 K and 1200 K cases where the background luminosity was larger than that observed during the extinction image. Future work will entail increasing the LED intensity, increasing the optical density of the neutral density filter to block more flame emission, and reducing the bandwidth of the dual-bandpass filter. Increasing the number of identical injections for ensemble averaging will also provide better resolution and higher fidelity.

Figure 12. Soot volume fraction (fv), as measured by the blue LED, in the plane corresponding to center of the flame axis for the 850 K – 1200 K ambient temperature sweep at an ambient density of 22.8 kg/m3 and ambient O2 concentration of 15% (by

volume). The lift-off length is indicated by the dashed line.

3.2.2 Unique Results from Multi-Wavelength Extinction

A novel feature of the two-color extinction diagnostic is observed when comparing fv evaluated with blue and green incident light under a given condition. Figure 13 shows the difference between the soot volume fraction based on our analysis of green light extinction (fv,green) and that based on our analysis of the blue light extinction (fv,blue). For the 900 K (Spray A) case shown in the upper panel, fv,blue is generally larger than fv,green. Assuming that molecular absorption of light

29

near 400 nm is insignificant, then fv,blue and fv,green should be equal. This can be achieved by adjusting the non-dimensional extinction coefficient, ke, in Eq. 2. Holding the refractive index, m, and the primary particle diameter, dp, constant while varying the number of primary particles per aggregate, Np, the RDG scattering approximation yields ke,blue/ke,green ratios that increase slightly with increasing Np. Thus, the difference between fv,blue and fv,green in the upper panel of Figure 13 might suggest that we should have used a larger value for Np. However, it is unlikely that Np is constant throughout the flame. Thus, if effects from Np were dominant, one might expect a gradual, but noticeable change in fv,green – fv,blue spatially at the 900 K case as the aggregates vary in size. Similarly, using a larger dp results in higher ke,blue/ke,green ratios at a given Np from the RDG calculations. A similar argument as above can be used to rule out dominant effects from changing dp.

Figure 13. Difference between fv corresponding to green LED extinction and fv corresponding to blue LED extinction. The top and bottom panels correspond to the 900 K (Spray A) and 1200 K conditions, respectively, at an ambient density of 22.8 kg/m3 and

ambient O2 concentration of 15% by volume.

Certainly the refractive index, m, plays an important role in the wavelength dependent differences observed in our evaluated fv. Increasing only the imaginary part of m, which is the absorption component, leads to larger ke values. Thus, the results at the 900 K ambient condition may indicate that absorption of blue light is more significant for the type of soot produced in this flame. Moving on to the 1200 K ambient condition shown in the lower panel of Figure 13, we observe spatial variation in fv,green – fv,blue. The larger soot volume fraction corresponding to the blue incident light data nearer the injector and for the green incident light data farther from the injector is an indication of the soot maturation process. As discussed in Shaddix et al. [18], spectrally resolved extinction measurements in sooting premixed flames performed by Millikan [25] and D’Alessio et al. [26] showed that the dispersion coefficient decreased with increasing height above the burner. The dispersion coefficient provides a measure of how much m varies with wavelength [27]. Soot having a higher dispersion coefficient was found to have a higher molar hydrogen-to-carbon (H/C) ratio. Soot precursor particles have been characterized by H/C ratios as high as 0.6, while the H/C ratio for carbonaceous soot is typically less than 0.2 [18]. In the 900 K conditions of Figure 13, fv,green – fv,blue does not change significantly in space suggesting that the soot particles’ optical properties are relatively constant. In the 1200 K

30

conditions of Figure 13, an abrupt transition occurs in the quantity fv,green – fv,blue near 30 mm. Based on the previous works discussed above, this transition is likely associated with a rapid decrease in the particles’ H/C ratio as they transition from nascent to mature soot. Nevertheless, if multiple internal scattering by the agglomerates in these spray flames is significant, phenomena may be present that are not captured under the RDG approximation. Follow-on work will attempt to address these issues in more detail.

3.2.3 Ambient Temperature, Density, and Oxygen Effects on Total Soot Mass

The mass of soot within 0.1-mm radial cross-sections along the flame axis is plotted in Figure 14 for the 850-1200 K ambient temperature conditions at an ambient density of 22.8 kg/m3 and an ambient O2 concentration of 15% by volume. The solid curves correspond to mass computed from KLblue. The green curves correspond to soot mass computed from KLgreen. The total mass was calculated from KL (total soot mass is proportional to KL) using an assumed density of 1.8 g/cm3 for soot (Choi et al. 1995). Since the soot density was held constant, the computed soot mass is proportional to fv. Thus, any error in our evaluation of fv is replicated in the calculation of soot mass. In Figure 14, we see an increase in the length of the region where the soot mass computed based on blue incident light is greater than that using green incident light at 1100 K when compared to the 1200 K case. This means that the region containing less mature soot is stretched out in physical space as the ambient temperature is reduced. The reverse is obviously true as ambient temperature increases. That is, we see more mature soot earlier in the flame at higher ambient temperatures. At ambient temperatures of 1000 K and below, the data indicate that most of the soot formed in the flame absorbs stronger in the blue than in the green. We also point out that at axial positions near 60 mm and beyond, the mass of soot contained in the radial cross-sections for the 1000 K, 1100 K, and 1200 K cases is nearly the same. As discussed by Pickett and Siebers (2004), higher ambient temperatures lead to shorter lift-off lengths and larger equivalence ratios at H. Consequently, the soot mass peaks at an axial position closer to the injector, soot consumption (oxidation) begins at an earlier axial location, and soot oxidation kinetics are accelerated under higher ambient temperature conditions.

In Figure 15, we plot the inverse of the measured lift-off lengths (1/H) along with the total soot mass in micrograms as a function of ambient temperature. The lift-off length (H) is defined as the distance from the injector orifice to the location where the quasi-steady spray flame stabilizes. The total soot mass is shown by the bar charts, while the inverse lift-off lengths are represented by the lines. Oxygen concentrations and ambient densities in the left and right panels, respectively, are distinguished by the different line styles for the inverse lift-off length and by the color of the bars for the soot mass. We observe that the inverse lift-off length is nearly linear with temperature for all ambient densities and oxygen concentrations. In the left panel, we find that the 13%, 15%, and 21% O2 cases yield nearly the same total amount of soot at 900 K, while the 13% and 15% cases remain similar at 1000 K as well. Although the total soot data are similar at these temperatures, at 13% and 15% O2, significant soot persists beyond the field of view. Thus, our integration does not include all soot formed in these flames. Since the 13% O2 flame is lifted slightly more than the 15% O2 flame, one might expect more soot outside our field of view in the former case leading to a larger total amount of soot as observed by Idicheria and Pickett (2005). Of all cases studied, the 13% O2, 22.8 kg/m3, 1200 K case produces the largest amount of soot within our experimental field of view. In practice, this case represents conditions

31

with high temperature, pressure, and exhaust gas recirculation (EGR). At ambient temperatures above 900 K, the 21% O2 case forms about half the total amount (or less) of soot as the others. Although the higher O2 concentration reduces soot due to the higher rate of oxidation, studies have shown that higher O2 (and the resultant higher diffusion flame temperature) can also lead to significant amounts of nitrogen oxides (Plee et al. [28]; Ahmad and Plee [29]; Dec [30]).

The data in the right panel of Figure 15 show that the total soot mass increases significantly with increasing ambient density. As noted above, our field of view does not include the entire chamber. Consequently, the total soot mass evaluated for the three different density cases may be artificially low; however, this effect gets progressively worse as density is reduced and soot formation begins farther downstream. An interesting comparison can be made between the 13% O2, ρ = 22.8 kg/m3 data in the left panel and the 15% O2, ρ = 15.2 kg/m3 data in the right panel. Both cases are characterized by similar lift-off lengths as a function of ambient temperature; however, the 13% O2, ρ = 22.8 kg/m3 case yields between 40% and 100% more total soot mass depending on the ambient temperature. Though not shown here, a comparison of the axial soot mass profile for these two cases indicates that the onset of soot begins at an earlier axial location for the 13% O2, 22.8 kg/m3 case. Thus, the distance between the lift-off length and the onset of soot is reduced at the higher density. At least two factors drive greater soot production at higher ambient densities. First, higher densities promote more rapid soot formation kinetics since collisions occur more frequently, and second, but of lesser importance, higher ambient densities result in more fuel-rich mixtures at the lift-off length (Pickett and Siebers[8]). These results are consistent with prior work showing that the ambient oxygen concentration corresponding to the maximum soot yield decreases with increasing ambient density (Musculus and Pickett [31]).

Figure 14. Soot mass in thin (0.1 mm) radial cross-sections of the jet as a function of axial distance from the injector orifice. Ambient density and O2 concentration were 22.8

kg/m3 and 15% by volume, respectively.

32

Figure 15. Inverse lift-off length, 1/H (curves) and total soot mass (bar charts) as a function of ambient temperature.

4. CONCLUSIONS AND FUTURE WORK

In this work, we developed an applied intrusive sampling and non-intrusive optical diagnostics to investigate the chemical composition and soot volume fraction in high-pressure spray flames relevant to direct-injection diesel engines. The gas sampling diagnostic was used to determine the species present in the reactants and products associated with the pre-burn event in the high-pressure vessel. The presence of unburned hydrocarbon species following the pre-burn event near the vessel walls or in the crevices associated with seals was confirmed. Samples extracted from the core of the vessel did not contain unburned hydrocarbons or detectable amounts of hydrocarbon species indicating that the spray combustion event occurs in a well-controlled chemical environment. Gas samples extracted from the lift-off region of the high-pressure n-dodecane spray flame contained a large number of hydrocarbon species including polycyclic aromatic hydrocarbons (PAH). Gas-phase species were not quantified due to the poor performance of a vacuum gauge during sample extraction; however, successful sampling and subsequent offline analysis by mass spectrometry represents a significant step forward in our effort to provide quantitative modeling targets. The presence of single- and multiple-ring aromatic species in the lift-off region has not been observed previously and represents a qualitative modeling target on its own. That is, modelers should consider that in the correct chemical progression to soot under the Spray A conditions, one to three ringed aromatic species are observed in the lift-off region. A novel multi-wavelength imaging extinction setup was developed and applied to investigate soot formation in high-pressure spray flames. The technique uses high intensity, ultra-fast LEDs in conjunction with engineered diffusers and a field lens to produce pulsed, highly efficient diffused back-illumination. The diffused incident light greatly reduces the effects of beam steering, such that over the quasi-steady period of injection, the observed extinction due to beam steering is near zero. A relatively small amount of extinction attributed to beam steering was observed when viewing the individual time-resolved images, which defined the lower sensitivity limit of this technique for detecting soot volume fraction (~ 1

33

ppm). The measurements varying ambient density, temperature and oxygen concentration showed good agreement with previous works using laser extinction and/or LII measurements. We observed distinct differences between the soot volume fractions determined using the two different wavelengths of incident light. The blue light (406 nm) showed sensitivity toward the detection of younger soot particles characterized by larger hydrogen-to-carbon ratios. The green light (519 nm) showed sensitivity toward the detection of more mature soot particles characterized by smaller hydrogen-to-carbon ratios. The largest amount of total soot mass was observed under high-temperature, high-density, and low oxygen concentration conditions. These results indicate that the use of too much EGR at high temperature and pressure in diesel engines can lead to copious amounts of soot.

Future work with regard to the intrusive sampling and offline mass spectrometry will involve further efforts to quantify the major species and some of the minor species observed in the pre-burn reactants and within the lift-off region of the n-dodecane spray flame. In addition, measurements will be acquired at different axial locations to provide spatial profiles of major and minor species as modeling targets. Higher temperature pre-burn scenarios will also be used to increase the presence of NO in the ambient gases for an investigation into the effect of NO on ignition delay times. Future efforts on the imaging extinction diagnostic will involve optimization of the optical arrangement to yield diffused incident light with a more Lambertian profile. The diagnostic will also be applied in conjunction with LII to provide a real-time assessment of the optical thickness throughout the flame such that a more accurate correction for signal trapping can be made.

34

5. REFERENCES

1. D.L. Siebers, SAE Technical Paper 980809 (1998).2. J.E. Nesbitt, S.E. Johnson, L.M. Pickett, D.L. Siebers, S.-Y. Lee, J.D. Naber, Energy & Fuels 25 (3) (2011) 926-936.3. G. Bruneaux, International Journal of Engine Research 9 (3) (2008) 249-265.4. C.A. Idicheria, L.M. Pickett, SAE Technical Paper 2006-01-3434 (2006).5. T. Aizawa, H. Kosaka, International Journal of Engine Research 9 (1) (2008) 79-97.6. R. Puri, T.F. Richardson, R.J. Santoro, R.A. Dobbins, Combustion and Flame 92 (3) (1993) 320-333.7. C.M. Sorensen, Aerosol Science and Technology 35 (2) (2001) 648-687.8. L.M. Pickett, D.L. Siebers, Combustion and Flame 138 (1-2) (2004) 114-135.9. J. Sjöholm, R. Wellander, H. Bladh, M. Richter, P.-E. Bengtsson, M. Alden, U. Aronsson, C. Chartier, O. Andersson, B. Johansson, SAE Int. J. Engines 4 (1) (2011) 1607-1622.10. C. Schulz, B.F. Kock, M. Hofmann, H. Michelsen, S. Will, B. Bougie, R. Suntz, G. Smallwood, Appl. Phys. B 83 (3) (2006) 333-354.11. L.M. Pickett, D.L. Siebers, International Journal of Engine Research 7 (2) (2006) 103-130.12. L.M. Pickett, available at < http://www.sandia.gov/ECN/ >.13. J.B. Ghandhi, D.M. Heim, Review of Scientific Instruments 80 (5) (2009).14. U.O. Koylu, G.M. Faeth, Journal of Heat Transfer-Transactions of the ASME 116 (4) (1994) 971-979.15. U.O. Koylu, G.M. Faeth, Journal of Heat Transfer-Transactions of the ASME 116 (1) (1994) 152-159.16. T. Aizawa, H. Nishigai, K. Kondo, T. Yamaguchi, J.-G. Nerva, C. Genzale, S. Kook, L. Pickett, SAE Int. J. Fuels Lubr. 5 (2) (2012) 665-673.17. S. Kook, L.M. Pickett, SAE Int. J. Fuels Lubr. 5 (2) (2012) 647-664.18. C.R. Shaddix, Á.B. Palotás, C.M. Megaridis, M.Y. Choi, N.Y.C. Yang, Int. J. Heat Mass Transf. 48 (17) (2005) 3604-3614.19. T.C. Williams, C.R. Shaddix, K.A. Jensen, J.M. Suo-Anttila, Int. J. Heat Mass Transf. 50 (7-8) (2007) 1616-1630.20. J. Manin, L.M. Pickett, S.A. Skeen, SAE Int. J. Engines 6 (4) (2013).21. M.P.B. Musculus, L.M. Pickett, Combustion and Flame 141 (4) (2005) 371-391.22. D.L. Siebers, in: SAE International: 1999.23. S.A. Skeen, H.A. Michelsen, K.R. Wilson, D.M. Popolan, A. Violi, N. Hansen, Journal of Aerosol Science 58 (2013) 86-102.24. J. O'Connor, M. Musculus, SAE Technical Paper 2013-01-0917 (2013).25. R.C. Millikan, in: Temperature: Its measurement and control in science and industry, Reinhold Publishing: New York, 1962; Vol. 3, pp 497-507.26. A. D'Alessio, F. Beretta, C. Venitozzi, Combustion Science and Technology 5 (1) (1972) 263-272.27. W. Dalzell, A. Sarofim, J. Heat Transfer 91 (1) (1969) 100-104.28. J.T. Plee, J.P. Myers, T. Ahmad, SAE Technical Paper 811195 (1981).29. T. Ahmad, S.L. Plee, SAE Technical Paper 831734 (1983).30. J.E. Dec, SAE Technical Paper 980147 (1998).

35

31. M.P.B. Musculus, L.M. Pickett, in: Advanced direct injeciton combustion engine technologies development. Volume 2: Diesel engines, H. Zhao, (Ed.) Woodhead Publishing Limited: Cambridge, UK, 2010; pp 644-675.

DISTRIBUTION

1 MS9055 Dennis Siebers 083621 MS9055 Paul Miles 08362

1 MS0899 Technical Library 9536 (electronic copy)

1 MS0359 D. Chavez, LDRD Office 1911