The Pennsylvania State University

The Graduate School

College of Engineering

GAS JET STUDIES FOR THE CHARACTERIZATION OF ADVANCED

INJECTION SCHEDULES AND BOWL DESIGN IN DIESEL ENGINES

A Thesis in

Mechanical Engineering

by

Meghan J. Borz

© 2016 Meghan J. Borz

Submitted in Partial Fulfillment

of the Requirements

for the Degree of

Master of Science

August 2016

ii

The thesis of Meghan J. Borz was reviewed and approved* by the following:

Jacqueline A. O’Connor

Assistant Professor of Mechanical Engineering

Thesis Adviser

Daniel C. Haworth

Professor of Mechanical Engineering

Mary Frecker

Professor of Mechanical Engineering

Associate Department Head for Graduate Studies

*Signatures are on file in the Graduate School.

iii

Abstract

In-cylinder strategies such as advanced fuel injection schedules and optimal piston bowl designs are often

utilized in diesel combustion for to improve efficiency and/or reduce emissions. The large design space

for injection schedules makes choosing the optimal schedule for an engine particularly challenging.

Additionally, experimental in-engine studies of advanced injection schedules are time-consuming and

costly. Gas jet experiments can provide a good approximation for the behavior of diesel jets and more

tests can be conducted in a shorter period of time without incurring the costs of an engine research

facility. The goal of this work is to gain a further understanding of some of the fundamental fluid

mechanics of multiple injections, jet-jet interactions, and jets impinging on surfaces.

Gas jet experiments are conducted using z-schlieren and acetone tracer planar laser-induced fluorescence

(PLIF). Three studies are conducted focusing on free jets, multiple jet interactions, jet-jet interaction, and

bowl geometry effects. The first study is a comparison of penetration results for helium gas jets with

penetration results for vaporizing and non-vaporizing sprays, which shows that by non-dimensionalizing

the results of gas jet experiments, the penetration curve follows a similar trend to the non-dimensionalized

penetration curve for vaporizing and non-vaporizing liquid sprays.

The second study explores the fluid mechanic interactions between multiple injections and the effects of

injection duration and dwell. The schlieren results of the multiple-injection studies showed that before the

end of injection (EOI) the non-dimensional jet-tip penetration was not significantly different for the first

and second injection, however, the average dispersion half angle during the quasi-steady portion of

injection was higher for the first injection than for the second injection. There are two multiple-injection

cases where the average dispersion half angle of the second injection is higher than that of the first

injection by a statistically significant amount. These differences in jet dispersion angle are indicative of

differences in mixing and entrainment during the first and second injections. Future studies with acetone-

PLIF will allow the concentration to be quantified and differences in the jet composition for the first and

second injections in multiple-injection schemes to be compared.

The last study focuses on the effects of piston bowl geometry and the angle between interacting jets. The

effects of the geometry on fluid recirculation and mixing are studied using schlieren and PLIF with jets of

acetone vapor and air and two different piston bowl designs. The results showed that a deeper bowl and

wider angle between the jets allows for improved mixing and air utilization.

iv

Table of Contents

List of Tables ............................................................................................................................................... vi

List of Figures ............................................................................................................................................. vii

List of definitions and abbreviations ............................................................................................................. x

Acknowledgements ...................................................................................................................................... xi

Chapter 1. INTRODUCTION AND BACKGROUND ................................................................................ 1

Motivation ................................................................................................................................................. 1

Multiple-injection interactions .................................................................................................................. 2

Jet-jet interactions ..................................................................................................................................... 4

Bowl geometry effects .............................................................................................................................. 7

Steady and unsteady jets ........................................................................................................................... 8

Gas jets .................................................................................................................................................... 10

Chapter 2. EXPERIMENTAL OVERVIEW .............................................................................................. 11

Experimental Setup ................................................................................................................................. 11

Diagnostics and Controls ........................................................................................................................ 12

Schlieren ............................................................................................................................................. 12

Acetone Tracer-LIF............................................................................................................................. 13

Image Processing Techniques ................................................................................................................. 15

Injection Studies...................................................................................................................................... 18

Unsteady Multiple Injections .............................................................................................................. 18

Unsteady Simultaneous Injections ...................................................................................................... 18

Wall-impingement .............................................................................................................................. 19

Chapter 3. SCALED INJECTION COMPARISONS ................................................................................. 20

Theory ..................................................................................................................................................... 20

Single Injection ....................................................................................................................................... 21

Scaled Injection Studies .......................................................................................................................... 23

Chapter 4. EFFECTS OF MULTIPLE INJECTIONS ................................................................................ 25

Injection Schedules ................................................................................................................................. 25

Single Injection ....................................................................................................................................... 26

Pressure Traces ....................................................................................................................................... 27

Jet-tip Penetration ................................................................................................................................... 28

Dispersion Angle .................................................................................................................................... 31

Chapter 5. BOWL GEOMETRY AND NOZZLE CONFIGURATION EFFECTS .................................. 34

Experiment Configurations and Scaling ................................................................................................. 34

v

Pressure Curves ....................................................................................................................................... 37

Penetration and Dispersion Angle........................................................................................................... 38

Effects of Wall Impingement .................................................................................................................. 40

Effects of Jet-jet interaction .................................................................................................................... 43

Chapter 6. CONCLUSIONS/FUTURE WORK ......................................................................................... 45

APPENDIX A Dimensioned Injector Drawings ......................................................................................... 47

APPENDIX B Acetone Bubblers and Nozzle Design ................................................................................ 51

APPENDIX C Pressure Transducer Calibration and Hardware Setup ....................................................... 53

APPENDIX D Determination of the Discharge Coefficient ....................................................................... 55

APPENDIX E Convergence Study ............................................................................................................. 59

APPENDIX F Bowl designs for 55% BTE effort ....................................................................................... 62

APPENDIX G Calculations for Piston Bowl Location............................................................................... 64

REFERENCES ........................................................................................................................................... 68

vi

List of Tables Table 1. Dimensions for ASCO 8262H212E two-way valve [71] .............................................................. 12 Table 2. Experiment parameters necessary to calculate characteristic values of gas jet injections ............ 23 Table 3. Experiment parameters for ECN Spray A injector studies [24] .................................................... 23 Table 4. The injection strategies include pilot, split, and post injections with three different dwell

durations given in non-dimensional time (𝑡): short 211.1 (5 ms) dwell (cases 1 – 7), intermediate 422.1

(10 ms) dwell (cases 8 – 14), and long 633.2 (15 ms) dwell (cases 15 – 21). The multiple injections were

compared with a 2532.6 (60 ms) single injection (case 22). ....................................................................... 26 Table 5. Engine parameters and characteristic time (t+

eng) and distance (x+eng) values ............................... 34

Table 6. The ratios of the characteristic time of the engine to the characteristic time of the experiment,

t+ratio (t+

eng/t+exp), are given for test cases with different injection times. ...................................................... 34

Table 7. Test matrix for studying bowl geometry effects and jet-jet interactions ...................................... 35 Table 8. Mean dispersion half angle averaged over the quasi-steady portion of injection for the extruded

bowl test cases............................................................................................................................................. 39

vii

List of Figures Figure 1. Ensemble-averaged quantitative PLIF of formaldehyde/PAH (red) and OH (green) at three times

(12, 14, and 40 CAD ATDC) for three umbrella angles (a) 160°, (b) 152°, and (c) 124°. The injector

location is marked by the white dot to the left in the images and the bowl-wall location is marked by the

white curve at the right of the images. [41] .................................................................................................. 6 Figure 2. Illustration of a pancake piston bowl geometry. ............................................................................ 7 Figure 3. Computational mesh for a 3D CFD study on 1/6 of the Caterpillar 3401 engine [54] .................. 8 Figure 4. Distribution of TKE in the central plane of the cylinder for an engine with a highly re-entrant

piston bowl [59] ............................................................................................................................................ 8 Figure 5. “Hypothetical jet mixture distributions shortly after an end-of-injection transient.” [66] ............ 9 Figure 6. Experimental setup for gas jet experiments showing two gas tanks with regulators (R), three

pressure transducers, three valves, and the injector. ................................................................................... 11 Figure 7. Valve drawings for ASCO 8262H212E two-way valve [71] ...................................................... 11 Figure 8. The cross-sectional cut of the injector body shows the 1/8” tapped holes for the two-way fast-

acting valves that feed air into the injector cavity and the relief valve as well as the 1/4” NPT hole for the

pressure transducer. ..................................................................................................................................... 12 Figure 9. The diagram of the z-schlieren setup shows that the light is emitted from an LED light source,

then passes through a condensing lens, then passes through a pinhole. The light is reflected off of a

concave mirror which collimates the light before it passes through the experiment. Then the light is

reflected off of a second mirror. Some of the light is blocked by the knife edge before it travels through a

lens and into the camera. ............................................................................................................................. 13 Figure 10: Experimental setup of pulsed dye laser for planar laser induced fluorescence. The blue dashed

line represents the laser location. ................................................................................................................ 14 Figure 11: Acetone absorption spectrum [73] ............................................................................................. 14 Figure 12. Acetone fluorescence spectrum [73] ......................................................................................... 15 Because the acetone fluorescence signal is relatively weak, a LaVision high-speed intensifier is used with

a gain of 65, delay of 1700 ns, and gate of 1600 ns. To filter out any ambient green light from the high

power laser, a Schott BG3 filter was placed in front of the lens on the intensifier. The spectral response of

the BG3 filter is shown in Figure 13. .......................................................................................................... 15 Figure 14. Transmittance for BG3 filter (x-axis is wavelength in nm) [74] ............................................... 15 Figure 15. Binary images at three instances in time after SOI2 are shown for case 15. The excess fluid

following the first injection was subtracted during the background subtraction step of image processing.

The intensity of the excess fluid was the same as the intensity of the second injection gas jet, resulting in

the tip of the second injection gas jet being subtracted out in the near-nozzle region. The blue line

represents the location of the jet tip as determined by the original jet-tip tracking algorithm. ................... 16 Figure 16. Images for case 15 at 3.067 ms after SOI2 are shown after the following processing steps: (a)

subtracting ith background subtracted image from the average of i+1, i+2, and i+3 images, (b) binarization

with a 0.1 threshold, (c) binarization with noise reduction, and (d) determination of the location of highest

intensity gradient represented by the blue line. ........................................................................................... 17 Figure 17. The search bounds for case 15 were found by fitting a rational curve to data that followed the

jet tip later in the injection. The concept for finding the search bounds was tested on the first injection (a)

and repeated for the second injection (b). ................................................................................................... 18 Figure 18. Scaled liquid jet penetration measurements compared to penetration correlation derived from a

control volume analysis using conservation of mass and momentum [1]................................................... 21

Figure 19. Comparisons of the dimensionless penetration (𝑆) vs. dimensionless time (𝑡) for the Abani and

Ghandhi B5 test case [64] and a similar injection conducted using the RFDL gas jet setup. ..................... 22

viii

Figure 20. Injection pressure curves for helium gas jet experiments by Abani and Ghandhi [64] and RFDL

researchers................................................................................................................................................... 22 Figure 21. The pressure profile for the 2583.3 (61.2 ms) single injection case +/- 1 standard deviation for

20 ensemble averaged trials shows that the gas jet experiments were highly repeatable. .......................... 27 Figure 22. The pressure profiles for pilot injections with similar injection durations but different dwell

times are compared with the baseline case and shown for (top) cases 1, 8, and 15, (center) cases 2, 9, and

16, and cases (bottom) 3, 10, and 17. .......................................................................................................... 27 Figure 23. The pressure profiles for split injections with similar injection durations but different dwell

times are compared with the baseline case. ................................................................................................ 28 Figure 24. The pressure curves for post injections with similar injection durations but different dwell

times are compared with case 22 and shown for (top) cases 5, 12, and 19, (center) cases 6, 13, and 20, and

(bottom) cases 7, 14, and 21. ...................................................................................................................... 28 Figure 25. The jet-tip penetration of the first injection (S1) and the jet-tip penetration of the second

injection (S2) are shown for pilot injections with different dwell periods. There is not a significant

difference in S1 and S2. ................................................................................................................................ 29 Figure 26. S1 and S2 are shown for split injections with different dwell times. .......................................... 30 Figure 27. S1 and S2 are shown for post injections with different dwell times. .......................................... 30 Figure 28. The half angle, θ1/2 and θ2/2, is given at each instant in time for both the first and second

injection for the 𝑡 = 422.1 dwell split-injection case. The black dashed lines indicate SOI1 and SOI2. ..... 31 Figure 29. The half angle, θ1/2 and θ2/2, are given in relative non-dimensional time from SOI1 and SOI2

for both the first and second injection for the 𝑡 = 422.1 dwell split-injection case. ................................... 31

Figure 30. The averaged dispersion half angle, θ/2, is shown as a function of first injection time for

various dwell periods. There does not appear to be a trend with dwell time nor injection duration. The

error bars show the confidence interval. The average confidence interval for θ1 and θ2 are 0.10 and 0.07,

respectively. Note: the limits of the y-axis are 14 to 18 degrees. ............................................................... 32 Figure 31. The sequence of schlieren images shows the relative jet-tip penetration of the first and second

injection in the split injection cases (4, 11, and 18). The inverted schlieren image is the second injection

fluid penetrating through the remnants of the first injection. Time is given in non-dimensional time from

SOI relative to each injection, where 𝑡 = 0 is SOI1 and SOI2. The red lines overlaid on the image represent

the boundary of the first injection and the blue line represents the location of the jet tip of the second

injection found using the jet tip tracking algorithm explained in the Image Processing Techniques section.

.................................................................................................................................................................... 33 Figure 32. The extruded bowl, single-jet experiments were conducted using Bowl 1 (left) and Bowl 3

(right). The distance between the bowl and the nozzle was the scaled distance for the piston location at

TDC. The bowl was angled at 15° from the vertical centerline of the jet based on a 150° umbrella angle.

The orifice location is marked by the blue dot. The measured angles were 15.1° and 14.9° for Bowl 1 and

Bowl 3, respectively. ................................................................................................................................... 36 Figure 33. The semi-circular bowl, three-jet experiments were setup so that the center of the bowl was

aligned with the centerline of the injector nozzles with 51° jet spacing (left) and 45° jet spacing (right). 36 Figure 34. Injector nozzle plates with three 1 mm orifices with 51° spacing for a 7-hole injector

configuration (left) 45° spacing for an 8-hole injector configuration (right) are used in the schlieren and

LIF experiments with the semi-circular bowl and revolved bowl designs. ................................................ 37 Figure 35. The ensemble-averaged pressure curves are shown for three schlieren experiments extruded

bowl 1 (left) and extruded bowl 3 (right). The subscripts in the legend indicate the injector pressures in

the engine (1200, 1800, and 2400 bar) which correspond to experimental injection pressures of 20, 30,

and 40 kPa, respectively. The vertical dashed lines correspond to the scaled EOI timing because the actual

injection time was longer. ........................................................................................................................... 37

ix

Figure 36. The ensemble-averaged pressure curves are shown for three schlieren experiments with a semi-

circular bowl and 51° nozzle spacing (left) and 45° nozzle spacing (right). The subscripts in the legend

indicate the injector pressures in the engine (1200, 1800, and 2400 bar) which correspond to experimental

injection pressures of 20, 30, and 40 kPa, respectively. The vertical dashed lines correspond to the scaled

EOI timing because the actual injection time was longer. .......................................................................... 38 Figure 37. The penetration curves for schlieren experiments with extruded Bowl 1 and extruded Bowl 3

show very similar trends. The error bars on the figure represent the 95% confidence interval. ................. 38 Figure 38. The ensemble averaged dispersion half angles are similar near SOI and after fluctuations in

light source intensity have stopped (after approximately 2.5 ms aSOI) across test cases with different

extruded bowl designs and injection pressures. The region of the injection that is used to calculate the

mean dispersion half angle is indicated using vertical black lines. ............................................................. 39 Figure 39. The ensemble averaged penetration for schlieren experiments with the semi-circular bowl

shape. .......................................................................................................................................................... 40 Figure 40. The schlieren images for tests with extruded Bowl 1 (left) and Bowl 3 (right) are shown for

increasing time after SOI. The blue dot indicates the location of the injector orifice................................. 41 Figure 41. In the images recorded with a 51° spacing between the jets, Bowl 3 appears to be better for

allowing the acetone-seeded air to spread further in the bowl. This indicates that in an engine, this bowl

design may allow more air to be utilized during combustion processes. .................................................... 42 Figure 42. In the tests with a 45° spacing between the jets, Bowl 3 appears to be better for allowing the

acetone-seeded air to spread further in the bowl. This indicates that in an engine, this bowl design may

allow more air to be utilized during combustion processes. ....................................................................... 43 Figure 43. After the acetone-seeded jets of air interact with the bowl wall, more acetone-seeded fluid

appears to move further back towards the nozzle in the images shown for the 51° spacing (7-hole injector

configuration). The blue arrows show the direction the fluid is moving back (normal to the bowl surface).

.................................................................................................................................................................... 44

x

List of definitions and abbreviations a 0.66, experimentally

determined coefficient from [1]

that relates the angle of the

model equations to the

measured angle for non-

vaporizing sprays

ACE Advanced Combustion Engine

aEOI After the end of injection

ATDC After top dead center

BTDC Before top dead center

Ca Area contraction coefficient

CNL Combustion noise level

df Effective orifice diameter

do Actual orifice diameter

DOI Duration of injection

DOI1 Duration of first injection

DOI2 Duration of second injection

DPF Diesel particulate filters

ECN Engine Combustion Network

EGR Exhaust gas recirculation

EOI End of injection

fps Frames per second

HCCI Homogeneous charge

compression ignition

HRR Heat release rate

IMEP Indicated mean effective

pressure

LIEF Laser induced exciplex

fluorescence

LTC Low temperature combustion

NOx Mono-nitrogen oxides

PAH Polycyclic aromatic

hydrocarbon

PIV Particle Image Velocimetry

PLIF Planar laser-induced

fluorescence

PLII Planar laser-induced

incandescence

PM Particulate matter

RCCI Reactivity controlled

compression ignition

S Jet-tip penetration

SCR Selective catalytic reduction

(exhaust aftertreatment)

SOI Start of injection

SOI1 Start of the first injection

SOI2 Start of the second injection

TDC Top dead center

TKE Turbulent kinetic energy

UHC Unburned hydrocarbons

Uf Nozzle exit velocity

�̃�, 𝝆𝒂, 𝝆𝒇 Density ratio, ambient density,

and fuel density

θ, θ1, θ2 Dispersion angle, dispersion

angle for first injection,

dispersion angle for second

injection

xi

Acknowledgements

I am grateful to have had a supportive and ambitious advisor who continually inspired, motivated, and

encouraged me during these past two years and I express my sincere gratitude to Dr. Jacqueline

O’Connor. In addition to my research advisor, I would like to thank Dr. Haworth for his insight

throughout my time with Volvo Supertruck Program and for being the reader for my thesis. Additionally,

I acknowledge Volvo Technology of America and the U.S. Department of Energy for funding this

research under DOE award number DEEE0004232.

Without the efforts of two hard-working undergraduate students, Dan Ruth and Yoontak Kim, the gas jet

project would not be what it is today and I greatly appreciate everything that they have done for the

project. I am proud to have shared this work with them. I also thank Brandon Smitsky for designing the

laskin nozzles that were used in the acetone bubblers and Alicia Grabiec for her help in the lab.

I thank all of the current graduate students in the Reacting Flow Dynamics Lab: Wyatt Culler, Anand

Makwana, and Ankit Tyagi for help with random lab tasks and for all the laughs in the office when we

needed a break from research. I thank Samuel Hansford for his help in the lab with configuring the

cameras, helping with LabVIEW, and for reviewing presentations and reports for me when I wanted

feedback. I thank Wyatt Culler for his help in the lab as well and for keeping me sane and less stressed by

convincing me to join the ultimate frisbee summer league, running and hiking with me, and watching

countless episodes of Lost.

I thank Larry Horner for his patience and for doing the machining to fabricate almost every piece of my

experiment. Thanks to Mary Newby for all of your support with day-to-day operations. Thanks to Steve

Peluso for helping me setup my experiments to measure my discharge coefficient, giving me your old

shock tube when I broke my experiment, for providing assistance with setting up the photodiode for my

energy measurements, and for explaining various pieces of hardware to me throughout the past two years.

Finally, I thank my friends and family who gave me respite from graduate school when I needed it and

encouraged me throughout the process. Special thanks to Tiffany Szeles for sharing in the new

experiences of graduate school, moving far from home, and going to a much larger university with me. I

have learned so much from my conversations with you and you taught me how to find value in even the

worst situations. Most importantly, I thank my Dad, Theodore Borz, who has encouraged me in every

academic and athletic endeavor I have pursued, convinced me to never give up, and always supported me.

1

Chapter 1. INTRODUCTION AND BACKGROUND

The U.S. Department of Energy’s SuperTruck program aims to develop Class 8 vehicles, often referred to

as “tractor-trailers,” that are more efficient than baseline models and supports the Energy Department’s

Clean Energy Manufacturing Initiative. The primary objectives of the SuperTruck program are to:

achieve at least a 50% brake thermal efficiency (BTE) under highway cruise conditions

determine approaches to improve BTE to 55%

increase vehicle freight efficiency (in ton-miles per gallon) by 50% from a manufacturer’s

predetermined 2009 baseline

The research discussed in this thesis primarily contributed to the effort to determine and evaluate

approaches for achieving 55% BTE in an engine as part of the Volvo SuperTruck program under DOE

award number DEEE0004232.

This chapter details the motivation for studying advanced injection schedules and provides a brief review

of the literature on the interaction of multiple consecutive injections, jet-jet interactions, piston geometry

effects, steady and unsteady jets, and gas jets. An experimental overview of the z-schlieren and acetone

tracer-LIF setups is discussed in Chapter 2. The scaling equations used to relate the fluid mechanic

behavior of gas jets to diesel sprays and comparisons of our experimental gas jet results to results for

vaporizing and non-vaporizing sprays are explained in Chapter 3. An effort to identify the effects of

injection duration and dwell time in multiple-injection schedules is discussed in Chapter 4, the effects of

separation angle between jets and piston bowl geometry are discussed in Chapter 5, and the

conclusions/implications of this work are discussed in Chapter 6.

Motivation

Exhaust emissions from diesel engines contain numerous pollutants including particulate matter (PM),

nitrogen oxides (NOx), and unburned hydrocarbons (UHC); the released quantities of the pollutants are

limited by emissions regulations [2, 3]. Aftertreatment techniques, such as diesel particulate filters (DPF)

and selective catalytic reduction (SCR), have proven to be effective at reducing PM and/or NOx

emissions. However, aftertreatment systems are often costly to install, reduce efficiency by adding weight

to the vehicle and additional thermodynamic constraints to the engine cycle, and have other drawbacks

specific to the each system. For instance, a DPF requires that the collected soot be periodically removed

from the filter and the buildup leads to decreased exhaust flow, which can lead to increases in fuel

consumption [4] and possible engine-out ash emissions. A more practical and cost-effective way to

reduce emissions is to utilize in-cylinder strategies.

Aside from applicability to emissions reduction, in-cylinder combustion strategies can be used to improve

engine efficiency and reduce combustion noise level (CNL). Improvements in engine efficiency are

sought for both environmental reasons like reduction of fossil fuel consumption and greenhouse gas

emissions, such as CO2, and for economic reasons like reduced costs for fuel per mile. In recent years,

there have been a number of U.S. DOE funded programs to increase engine efficiency, such as the

SuperTruck program and the Advanced Combustion Engine (ACE) programs, both of which included

research on advanced combustion strategies and multiple-injection schedules [5, 6]. Unfortunately many

strategies that improve diesel engine efficiency also increase CNL, which is regulated in many European

countries [7]. Lowering engine noise can reduce stress on vehicle components due to vibrations.

2

Additionally, noise reduction is a requirement in the modern age for social acceptance and

competitiveness of vehicles in the market [8].

There are a number of advanced in-cylinder strategies that have the potential to address all of these issues,

including homogeneous charge compression ignition (HCCI), reactivity controlled compression ignition

(RCCI), and advanced injection schedules. HCCI engines utilize autoignition of a fuel-lean homogenous

mixture of fuel and air, which results in lower emissions than typical diesel and gasoline combustion.

However, this combustion method has only been used in research laboratories and is not practical for use

in real engines because a high temperature must be achieved for the autoignition reactions to proceed

making cold start capability infeasible. A second reason HCCI is currently impractical for use in real

engines is its sensitivity to charge gas temperature, which varies with speed, load, external conditions, and

operating condition [9]. RCCI utilizes two fuels of different reactivity, multiple injections, and EGR to

obtain enhanced mixing conditions in the cylinder [9]. The use of low reactivity fuel (such as gasoline)

with high reactivity fuel (such as diesel) necessitates a different piston bowl geometry than those utilized

for compression ignition of just high reactivity fuels. This is a major drawback to RCCI because it

requires a major overhaul of the engine design and operation. It also requires that two fuels be carried on

board at all times. If the diesel fuel is diminished before the gasoline tank empties, the engine will not

operate because most RCCI engines do not include a spark plug [10], although spark-assisted

compression ignition (SACI) engines are being investigated to overcome this issue [11]. All modern

common rail diesel injectors have the capability for advanced injection schedules, which can be

implemented in new engine designs or as an aftermarket solution without significant changes to hardware.

The injection schedule is altered by changing the fuel delivery programmed into the vehicle’s engine

control unit (ECU).

There are a number of ways in which the extent of mixing, and therefore the local equivalence ratio and

combustion process, in an engine can be changed: altering the injection schedule, the piston bowl

geometry, or the injector design. There is a large design space for advanced injection schedules wherein

the following parameters can be altered: duration of main, pilot, and post injections; dwell time between

injections; the number of injections in the schedule; and timing of injections with respect to crank angle

degrees (CAD) before top dead center (BTDC) or after top dead center (ATDC).

When designing engine components these additional parameters can be changed as well: the number of

injector orifices (alters number of simultaneously interacting jets and the angle between those jets),

umbrella angle, piston bowl geometry, and injector rail pressure.

With such a large number of variables, optimizing the engine design rapidly becomes a complex process.

Currently, effective injection schedules and engine component designs are primarily predicted using

simulations and tested experimentally using large parameter sweeps. To more efficiently design engines

and produce better models for determining the most effective injection schedule, a targeted physics-based

approach needs to be developed. The objective of this research is to understand the fundamental fluid

mechanics of unsteady jet mixing, multiple-jet interactions, and wall impingement to develop that

physics-based understanding.

Multiple-injection interactions

Multiple injection schemes have been utilized in practice to alter heat release rate (HRR), peak cylinder

pressure, and combustion emissions. The terminology used for subsets of multiple injections is based on

the quantity of fuel that is used during different portions of the schedule. The injection that uses the most

fuel and the injection of the longest duration is considered the main injection. When injections with a

shorter injection duration precede the main injection, those short injections are referred to as pilot

injections. When short injections are subsequent to the main injection, those shorter injections are called

3

post injections. An injection schedule with two injections of equal injection duration (and approximately

equal fuel quantity) will be referred to as a split injections in this thesis.

Pilot injections have been effective in reducing the peak cylinder pressure and combustion noise level

[12-14]. A mechanism for reducing CNL in diesel engines is phasing HRR. Combustion phasing through

use of a pilot injection reduces HRR and peak cylinder pressure while increasing the temperature and

radical concentrations prior to the main injection, thereby decreasing the ignition delay of the main

injection fuel and effectually reducing engine noise [14]. Busch et al. [14] showed that close-coupled

pilot injections (dwell of 140 μs) reduced the CNL under part-load operating conditions by roughly 9 dB

compared to a single injection without increasing soot, NOx, CO, or UHC emissions. In addition to CNL

reduction, Tanaka et al. [12] showed that pilot injections can simultaneously decrease PM and NOx

emissions. The timing of pilot injections in relation to the main injection (changing dwell) has been used

to change the combustion duration. Zeng et al. [15] used pilot injections with the same injected mass

quantity and same duration of injection (DOI) but different dwell times to show that advanced pilot

injection timing (30 degrees BTDC) decreased the HRR, decreased combustion duration, and produced

the lowest in-cylinder soot levels within their test matrix.

Split injections have been used primarily in injection studies, rather than practical application, to

understand the effects of altering dwell time between injections alone [16-22]. A split injection schedule

is currently the focus of one of the Engine Combustion Network (ECN) spray cases (Spray A) [23, 24].

The results of the ECN studies showed that the fluid mechanic behavior of the second injection in that

split injection scheme with a 0.5 ms injection duration and 0.5 ms dwell time was different from the

behavior of the first injection after the end of injection (aEOI), which has implications on fluid mixing

[23]. Herfatmanesh et al. [22] compared the quantities of soot, NOx, and UHCs produced by single

injections and split injections; the results showed that the split injections did not decrease emissions and

rather significantly increased soot production. However, other authors [16, 18-21] showed that split

injections did decrease NOx emissions but only with certain dwell times. Goldwine and Sher [25]

compared schedules with two different pilot injections with a split injection schedule across a range of

dwell times. Their results showed that the split injection schedule produced the highest NOx emissions but

the lowest CO emissions across all dwell times and the lowest UHC emissions at long dwell times [25].

There is a tradeoff in decreasing the CO and UHC emissions using split injections, whereby NOx

emissions increase (and vice versa).

Close-coupled post injections have been utilized to decrease UHC and soot emissions. Herfatmanesh and

Zhao [26] showed that a post-injection strategy used in place of a single injection with the same total

injected fuel quantity can result in significantly lower UHC emissions, improved mixing, and lower CNL.

O’Connor and Musculus studied close-coupled post-injection efficacy for UHC emissions reduction in a

heavy-duty optical diesel engine at high-EGR, low temperature combustion (LTC) conditions using PLIF

and luminescence images [27]. Their results showed that the post injection enriches the lean regions near

the injector nozzle that reside after the main injection and increases the local equivalence ratio resulting in

a transition to second stage combustion in that region and a reduction in UHC emissions by up to 34% as

compared to a single injection. Once again, there is an emissions tradeoff and reducing UHC results in an

increase in NOx emissions [26]. Numerous studies have shown that post injections can be used to reduce

engine-out soot [28-35]. In a study by O’Connor and Musculus the filter smoke number was measured for

a number of injection schedules and in the best case soot emissions were reduced by 55% (3.0 mg/m3)

using a post-injection schedule as compared to a single injection at the same load [28].

The interaction of multiple consecutive jets can affect the behavior and mixing. The behavior of an

injection close-coupled with a previous injection can change the penetration rate of the second injection

through slipstreaming. Slipstreaming occurs when the penetration rate of a subsequent injection in a

multiple-injection scheme is higher after EOI than it would be if that injection were penetrating into a

4

quiescent media. As part of the ECN’s split-injection study on the spray A injector, Skeen et al. [23]

observed this effect using schlieren. Because the first jet was injected into stagnant fluid, there was a

stagnation plane at the head of the jet. However, the second injection had entered a “slipstream” produced

by the first injection wherein the fluid ahead of the second injection was already moving downstream

after SOI2. A fundamental study of liquid and vapor penetration of injections under engine-like conditions

by Parrish et al. showed similar findings [36]. The split-injection case showed that the liquid penetration

of the second injection was higher than the penetration of the first injection for the two shortest dwell

times. Those results indicate that jet-tip penetration is a function of both DOI1 and the residual flow field

from the first injection [36].

Mixing affects combustion processes and changes with injection duration and timing. Bruneaux and

Maligne [37] compared tracer laser-induced exiplex fluorescence (LIEF) measurements for single

injections and main plus post injections to determine the effects of injection timing and dwell on mixing.

Their results showed that entrainment after the end of injection (EOI) resulted in low fuel concentrations

near the tail of the jet. High concentrations at the jet tip were caused by limited mixing near the head of

the jet. In the post-injection cases, the second injection followed a similar trend to the first injection.

However, the OH-concentration gradient at the tip of the jet was less steep than it was in the first

injection, indicating increased mixing in the jet-tip region. Comparisons of the post injections for two

different dwell times showed that the gradient at the tip of the jet was steeper when the dwell time was

longer, which implies that more mixing occurs at the head of the post injection jet when the dwell time is

shorter [37].

This review of the literature highlights the numerous benefits of properly timed multiple-injection

schemes with regards to CNL and emissions, and provides insight into some of the offsetting weaknesses.

Multiple-injection strategies in diesel engines are already used for engine performance optimization.

However, these methods are typically optimized on a case-by-case basis through an extensive series of

tests. An understanding of the mechanisms of injection interactions would enable optimal combustion

conditions to be determined more efficiently. One objective of this work is to explore the fluid mechanic

interactions between multiple consecutive injections in free jets.

Jet-jet interactions

In most modern common rail diesel injectors there are 6-10 orifices. The number of orifices and umbrella

angle of the orifices can alter mixing processes in the cylinder. The distribution and mixing of fuel and air

in a diesel engine can influence the HRR and the formation and oxidation of pollutants. To maximize fuel

efficiency and reduce emissions, it is necessary to understand jet-jet interactions and how they affect the

flow field and combustion reactions.

The effects of changing the umbrella angle and angle between injections has been studied numerically

[38, 39] and experimentally in diesel engines [40-43]. In part of the numerical study by Abraham and

Khan [38] the azimuthal angle between jets was held constant at 45° (an 8-hole injector was used) and the

results showed that as the umbrella angle was increased (over a range of 128° to 140°), the mixing rate

increased. When the umbrella angle was held constant and the angle between jets was changed, there is an

optimal angle that maximizes mixing. Jet-jet interaction interferes with entrainment of surrounding fluid

because adjacent jets are competing to entrain the same fluid, therefore increased jet-jet interaction can

lead to “entrainment suppression” [44] and a decrease in the characteristic mixing time [38]. Too great of

an angle between adjacent jets reduces jet-jet interaction and can lead to decreased air utilization.

5

In the experimental study by Pierpont et al. [40], two umbrella angles were tested (125° and 140°) for

single injections with 0% and 6% exhaust gas recirculation (EGR) and the 125° umbrella angle showed a

significant reduction in PM emissions at advanced injection timings. A more extensive experimental

study by Genzale et al. [41] investigated the mixing processes for three different umbrella angles (124°,

152°, and 160°) at LTC conditions with EGR using formaldehyde/polycyclic aromatic hydrocarbon

(PAH) PLIF and OH-PLIF. The PLIF results are shown in Figure 1. Interesting regions of the flow field

are labeled in Figure 1 as follows: P represents the very fuel-rich combustion zones that don’t produce

OH, Q indicates a PAH signal (stronger signal than the formaldehyde fluorescence), R indicates

formaldehyde persisting near the center of the chamber, S shows the fuel being redirected upward along

the bowl wall after jet impingement, T indicates dark fuel-rich regions near the bowl wall, U corresponds

to small pockets of soot and PAH, and V indicates fuel-lean regions where only first-stage ignition

occurs, W indicates fuel rich contours (ϕ > 1.4), X indicates dark pockets near bowl wall that are devoid

of formaldehyde or OH, Y indicates pockets of soot and PAH near the bowl wall, and Z indicates a weak

formaldehyde signal near the center of the combustion chamber.

With the particular piston bowl geometry that was used in [41] fuel-rich zones would form near the floor

of the piston bowl in the region of jet-jet interaction, but the equivalence ratio in those zones is reduced

when the umbrella angle is increased (to 160°) or decreased (to 124°) leading to a reduction in soot

formation. The wide umbrella angle (160°) leads to a reduction in soot production as compared to the

baseline case, but there is less bulk-fluid motion near the bowl wall which reduces late-cycle mixing,

causes quenching in the center of the chamber, and increases UHC and CO emissions. The narrow

umbrella angle (124°) improves bulk-fluid motion by redirecting the jet momentum along the bowl wall

thereby preventing the formation of fuel-rich zones between jets, reducing soot formation near the bowl

wall, and improving late-cycle oxidation of fuel-lean regions through enhanced mixing in the center of the

combustion chamber.

6

Figure 1. Ensemble-averaged quantitative PLIF of formaldehyde/PAH (red) and OH (green) at three times

(12, 14, and 40 CAD ATDC) for three umbrella angles (a) 160°, (b) 152°, and (c) 124°. The injector

location is marked by the white dot to the left in the images and the bowl-wall location is marked by the

white curve at the right of the images. [41]

This review of the literature highlights the benefits of properly spaced injections with regards to mixing

and emissions. Multiple simultaneous injections are already used in diesel engines with modern injector

systems. An understanding of the physics involved in jet-jet interactions would enable optimal injector

designs to be determined efficiently. One objective of this work is to explore the fluid mechanic

interactions between multiple adjacent jets at various orifice spacings.

7

Bowl geometry effects

Changes in spray geometry (umbrella angle, number of injector orifices, etc.) are not the only means by

which mixing and combustion processes in the cylinder can be altered. Changing the shape of the piston

bowl can change the late-cycle flow field after the jets of fuel have impinged on the bowl wall. Simple

geometries, such as flat plates and “pancake” bowls, are often used for experimental research, particularly

in optical engines, while more complex geometries are used for industrial and commercial applications.

Fundamental studies on jet-wall interactions have been conducted for jets impinging normally onto flat

walls [45-49] and travelling tangentially along plane walls [50]. Pickett and Lopez [45] conducted

experiments in a CVCC with jets of T70 fuel impinging on a flat plane wall located at multiple positions

ranging from 25-50 mm downstream of the injector nozzle. They compared soot formation (measured

using planar laser-induced incandescence, PLII), wall temperature (measured with thermocouples), and

heat flux for free jets, jets impinging on walls without confinement (no recirculation and minimal jet-jet

interaction), and jets impinging on walls with confinement (increased jet-jet interaction and secondary jet-

wall interaction). Jets impinging on a flat wall without confinement produced less than half of the soot

that a free jet produced, which likely resulted from heat loss to the wall. Sooting also decreased as the

distance between the injector and wall decreased. The wall-jet and jet-jet interactions in the cases with

confinement resulted in higher measured soot quantities. It was found that soot formation could be

prevented in the confined cases by ending the injection prior to the time of significant interaction of

combustion gases moving back from the bowl wall between jets after impingement.

Taking simplified impingement studies on stationary flat plates a step further, a “pancake” piston bowl is

often used in research optical engines [51]. A “pancake” piston bowl geometry has a flat bowl without a

pip, as shown in Figure 2.

Figure 2. Illustration of a pancake piston bowl geometry.

Genzale et al. [52] studied the effects of piston bowl geometry on mixing, combustion, and pollutant

formation with pancake bowls of various bore diameters. The results of that study showed the complexity

of jet-wall and jet-jet interactions as they relate to bowl geometry. The piston bowls with a cylinder bore

that was 60% and 70% of the original bore size (140 mm) had significant jet-wall and jet-jet interactions

that resulted in the formation of fuel-rich zones near the bowl wall and increased soot formation, but bulk-

fluid motion was enhanced which promoted late-cycle oxidation further back in the cylinder. The bowl

with a bore that was 80% of the original bore size resulted in less jet-wall and jet-jet interaction, lower

soot formation, but increased UHC emissions due to fuel-lean mixtures stagnating near the injector and

center of the combustion chamber.

“Mexican hat” bowls or “Hesselman” chambers [53] have a raised pip in the center of the bowl

surrounded by a toroidal recess, giving it the appearance of a sombrero. To illustrate this geometry, an

image of a computational mesh for 1/6 of a Mexican hat bowl design that is used in the Caterpillar 3401

engine is shown in Figure 3 [54]. Using a piston bowl with this surface geometry where the fuel jets

interact can significantly alter the flow field in the cylinder. The pip height and recess depth are optimized

for the particular engine of interest to generate the desired amount of swirl and mixing in the cylinder.

Mexican hat piston bowl geometries are used commonly in research and in commercial applications.

8

Oftentimes a Mexican hat bowl geometry is included in numerical [55-57] studies as a baseline geometry

for validation or comparison with complex geometries.

Figure 3. Computational mesh for a 3D CFD study on 1/6 of the Caterpillar 3401 engine [54]

More advanced geometries have been developed, such as the re-entrant bowl, and have been found to

improve swirl and intensify turbulence which can improve mixing [58, 59]. Results from a CFD

simulation of turbulent kinetic energy (TKE) in a re-entrant bowl are shown in Figure 4 to provide an

illustration of the bowl geometry [59]. Typically modified re-entrant bowls are proposed as new piston

bowl designs for modern engines and simulations are conducted using CFD and combustion models.

Many numerical studies have investigated the effect of varying parameters such as pip shape, toroidal

radius, and bowl lip shape in re-entrant piston bowls [58-60].

Figure 4. Distribution of TKE in the central plane of the cylinder for an engine with a highly re-entrant

piston bowl [59]

Piston bowl geometry can significantly alter late-cycle mixing, which impacts the extent of combustion

and emissions. Advanced piston bowl geometries are used in modern diesel engines but there is not

extensive literature on experimental studies of piston geometry effects. One objective of this work is to

explore the impact of varying piston bowl shape and depth on jet-wall interactions and mixing using gas

jet experiments.

Steady and unsteady jets

Steady jet behavior is a well-studied area of research. Many researchers have performed steady jet

experiments and developed analytical models to gain an understanding of entrainment and turbulence in

steady jets [1, 61-63]. Naber and Siebers [1] developed a one-dimensional model (constant axial velocity

across jet) for steady jet penetration using a control volume analysis assuming constant dispersion angle,

an instantaneously started jet, constant injection velocity, and no velocity slip between the injected fuel

9

and entrained gases. The model was confirmed to capture the penetration rate of starting diesel jets

through their experiments on vaporizing and non-vaporizing sprays [1].

In unsteady jets there are several other factors and phenomena that can have a significant impact on jet

behavior such as the ramp rate [64] and the “entrainment wave” that follows a transient end of injection

(EOI) [65, 66]. Abani and Ghandhi [64] studied the effects of ramp-up rate in unsteady turbulent gas jets

and found that the jet-tip penetration curve deviated from the jet-tip penetration of a steady jet. For cases

where there was a step increase in pressure (positive ramp-up rate), the jet-tip deviation occurred earlier

and increased in magnitude with the difference in injection velocity. For cases where there was a step

decrease in pressure (negative ramp rate), the time where the jet-tip penetration deviated was independent

of the injection velocity difference. The rate at which the jet penetrates downstream affects the

entrainment rate of the jet and the timing at which impingement on the piston bowl wall will occur, which

then affects the extent of mixing. Therefore, varying the ramp rate can affect the combustion process.

Experimental measurements [66], a one-dimensional model [65], and large eddy simulations [67] have

shown that transient ramp-down at EOI results in a region of increased entrainment, an “entrainment

wave,” that travels downstream at a faster rate than the initial jet propagation rate and improves mixing.

This could be beneficial to air utilization near the piston bowl wall, combustion efficiency, and the

distribution of heat and fuel in the cylinder. However, increased entrainment in a decelerating jet after

EOI has also been shown to create a fuel-lean region near the injector tip that cannot achieve complete

combustion [65, 67]. Naber and Siebers [1] have shown that steady gas jets behave similarly to vaporizing

fuels during the quasi-steady portion of injection and that the equivalence ratio, ϕ, of vaporizing and non-

vaporizing jets decreases inversely as 1/x, with x being the downstream distance from the injector nozzle.

But, after a transient event (e.g. fast EOI) a jet is no longer steady. A gradual injection ramp-down would

be required for the jet to remain relatively steady and for the local equivalence ratio near the injector to be

fuel-rich after EOI. An illustration of gradual ramp-down and fast ramp-down at EOI are given in Figure

5.

Figure 5. “Hypothetical jet mixture distributions shortly after an end-of-injection transient.” [66]

Real diesel injections have ramp-down rates somewhere in between what is shown in case i and case ii in

Figure 5. If a fuel-air mixture in the near-injector region is too lean, such as in case ii, most of the fuel is

carried away from the injector and the lack of fuel in that region may prevent second-stage ignition from

occurring, increasing UHC emissions. This is especially prevalent in low temperature combustion (LTC)

diesel combustion. In low temperature ranges (<1000 K) aliphatic fuels having three or more carbon

atoms can exhibit two-stage ignition behavior where only about 5-10% of the hydrocarbons are consumed

and peroxide and formaldehyde are produced during first-stage ignition [68]. However, at the operating

10

conditions for most typical diesel engines combustion recession occurs and second stage ignition is

reached near the nozzle after combustion recession, reducing UHC emissions [69].

Injection ramp-up rate was considered in this work and quantified by taking the derivative of the rising

edge of the pressure curve. The effects of entrainment and EOI transients are considered throughout this

work, especially in the multiple-injection studies, but entrainment rates are not quantified.

Gas jets

Gas jets have been shown to capture entrainment rates and penetration rates of vaporizing and non-

vaporizing sprays [1, 64, 65]. Fundamental gas jet experiments by Abani and Ghandhi [64] have also

demonstrated the important physics related to injection ramp rates and showed strong parallels with diesel

fuel injection. The goal of this work is to explain the fluid mechanics of multiple injections, jet-jet

interaction, and wall impingement using unsteady gas jets. Gas jet experiments are utilized as an

alternative means by which key physical phenomena can be observed and measured. There are a number

of benefits and a few tradeoffs to using gas jet experiments to understand the fundamental physics of jets

as opposed to in-engine testing. The tradeoffs include the inability to capture the effects of

compressibility due to high rail pressures and liquid droplet breakdown in vaporizing liquid sprays. Near

the injector tip liquid droplet breakdown is important because the mixing processes and dispersion of the

jet are heavily driven by atomization, however, further downstream mixing is dominated by shear layer

mixing effects [1]. Because much of the mixing in diesel combustion, outside of the near nozzle region,

happens in the gaseous phase, gas jets should still be able to capture most of the key physics.

From an experimental perspective, there are several benefits including reduced time per test and easy

application of numerous diagnostic techniques. A large number of tests can be conducted in a short period

of time compared to in-engine experiments, which allows results to be studied on an ensemble-averaged

basis so statistical significance can be understood. Additionally, a wide range of high fidelity, high-speed

diagnostic techniques that are difficult to perform in optical engines, such as PIV, can be easily applied

with the gas jet experimental setup.

The measurements that are recorded in this study are jet composition (or equivalence ratio, ϕ), penetration

rate (momentum), and jet dispersion (entrainment, mixing). The jet composition is studied using both

qualitative and quantitative planar laser induced fluorescence (PLIF). Qualitative LIF results can be used

to understand the relative distribution of fuel and observe bulk fluid motion, while quantitative LIF

provides scalar concentrations allowing local equivalence ratios to be determined and compared across

test cases. The jet penetration and dispersion angle are determined from flow visualizations obtained

using a z-schlieren technique. The theory behind scaling gas jet results to infer diesel spray behavior is

explained in Chapter 3.

11

Chapter 2. EXPERIMENTAL OVERVIEW

Experimental Setup The experimental setup for the gas jet experiments is shown in Figure 6 and follows that of Abani and

Ghandhi [64, 70].

Figure 6. Experimental setup for gas jet experiments showing two gas tanks with regulators (R), three

pressure transducers, three valves, and the injector.

The gas from the supply canisters is fed into a single injector through two fast-acting two-way ASCO

8262H212E solenoid valves. The valves are normally closed and open when supplied 24 V DC. Valve

timing is set in LabView and triggering is achieved with an NI 9482 4-channel electromechanical relay

module. The flow factor of the valve (Cv) is 0.76 and the response time is 5-10 ms (appears to be about 7

ms from experimental testing in the RFDL lab facility). The valve pipe size is 1/8” and the other

dimensions of the valve are given in Figure 7 and Table 1.

Figure 7. Valve drawings for ASCO 8262H212E two-way valve [71]

12

Table 1. Dimensions for ASCO 8262H212E two-way valve [71]

Units H K L P W

inches 3.12 1.79 1.56 2.76 1.95

mm 79 45 40 70 50

In the setup for PLIF experiments with acetone-seeded air there are two acetone bubblers located between

the regulators for the gas tanks and the valves attached to the injector body. For details on the design for

the acetone bubblers, see Appendix B. An Omega PX329 pressure transducer is located on the side of the

injector to monitor the pressure in the injector, and a relief valve allows for fast ramp down times. For

some tests, the line pressures before the inlet valves are additionally measured using pressure transducers.

The pressure transducers require a 5 V excitation, they can measure pressures 0 – 50 psig, and the

connection fitting is a ¼ -18 male NPT. An NI 9237 bridge analog input module is used to send the

excitation voltage to the pressure transducer and read in the voltage at a set sampling rate (typically 45

kHz). The voltage is then converted to a pressure reading in LabView using the calculated calibration

factors and the pressure at each instant in time is saved to a textfile. For information on wiring the

pressure transducers and calculating the calibration values, see Appendix C. The pressure recorded by the

pressure transducer on the injector cavity is used to generate the injection pressure profiles presented in

the Results section.

The gas jet is produced from an orifice in the center of the nozzle plate on the injector into quiescent air at

ambient temperature and pressure. Detailed drawings of the injector are given in Appendix A. In all of the

studies discussed in this document, the orifice diameter is 1 mm. A top-view cross-section of the injector

cavity is shown in Figure 8.

Figure 8. The cross-sectional cut of the injector body shows the 1/8” tapped holes for the two-way fast-

acting valves that feed air into the injector cavity and the relief valve as well as the 1/4” NPT hole for the

pressure transducer.

Diagnostics and Controls

Schlieren

The z-type schlieren configuration [72] shown in Figure 8 is used to visualize the flow. The jet-tip

penetration and dispersion angle are determined from the schlieren images. Light is provided by a

continuous LED. 6” parabolic mirrors are used on either side of the setup. The collimated light that

reflects off of the first mirror is bent in proportion to the density of the gas through which it travels as it

passes over the experiment. After the second mirror, the undisturbed light is partially blocked by a knife

edge and the light that is deflected produces an image of the flow due to the refraction of that light

13

through regions of the flow with density gradients. High-speed images are captured with a Photron SA1.1

Fastcam. The camera has a viewing region that is 384 by 864 pixels for the multiple-injection studies and

a viewing region that is 576 by 504 pixels for the bowl-wall impingement studies. The maximum height

and width of the images from the multiple injection studies are 12.5 cm and 5.8 cm, respectively,

providing a resolution of 65.5 pixels/cm. The resolution of the images in the bowl-wall impingement

studies is approximately 47.5 pixels/cm. The images are recorded at 15 kHz.

Figure 9. The diagram of the z-schlieren setup shows that the light is emitted from an LED light source,

then passes through a condensing lens, then passes through a pinhole. The light is reflected off of a

concave mirror which collimates the light before it passes through the experiment. Then the light is

reflected off of a second mirror. Some of the light is blocked by the knife edge before it travels through a

lens and into the camera.

Acetone Tracer-LIF

An Edgewave pump laser and Sirah dye laser with Rhodamine 6G dye are used for acetone PLIF

measurements of the scalar concentrations in the jet. The pulse duration is 10 ns and the repetition rate for

all PLIF tests in this study is 10 kHz. The beam from the high power, 532 nm Edgewave laser is first sent

to a high power optical mirror to bend the light into the dye laser. A beam splitter is then used to send the

laser into the resonator and amplifier sections of the dye laser, after which the wavelength of the light is

566 nm. A second harmonic generator (SHG) inside of the amplifier is used to double the frequency and

produce a 283 nm UV beam. A series of four Pellin-Broca prisms are used to separate the visible beam

(566 nm) from the UV beam (283 nm) before the UV beam exists the dye laser. The power output of the

UV laser beam typically ranges from 1 – 2.6 W and is tested daily using a Gentec UP25-H power meter.

A small percentage of the laser is reflected (approximately 0.16%) towards an energy meter using quartz

glass. The reflected beam is sent through a piece of quartz glass, then a diverging lens to decrease the

energy density, and then through a second piece of quartz glass before travelling into a Thorlabs PDA25K

GaP amplified photodetector. The output signal from the photodetector is sent into a Stanford Research

Systems gated boxcar average (model SR250) and the output from the boxcar average is sent to the NI

DAQ. An NI 9237 module is used to determine the voltage and the values are converted to energy values

in LabView and written to a textfile. The transmitted portion of the laser beam (approximately 96%) from

the first piece of quartz then travels through several 90° prisms and a sheet optic. For qualitative

experiments where an energy meter is not necessary, the beam travelled from the dye laser, through the

14

90° prisms, and through a sheet optic that expanded the beam into a sheet with a vertical Gaussian

distribution, as shown in Figure 10. .

Figure 10: Experimental setup of pulsed dye laser for planar laser induced fluorescence. The blue dashed

line represents the laser location.

Acetone has a wide range of excitation wavelengths, most in the UV spectrum as shown in Figure 11, and

283 nm is a wavelength that falls within that absorption spectrum. Acetone exhibits broadband

fluorescence at wavelengths in the visible spectrum, primarily blue, as shown in Figure 12.

Figure 11: Acetone absorption spectrum [73]

15

Figure 12. Acetone fluorescence spectrum [73]

Because the acetone fluorescence signal is relatively weak, a LaVision high-speed intensifier is used with

a gain of 65, delay of 1700 ns, and gate of 1600 ns. To filter out any ambient green light from the high

power laser, a Schott BG3 filter was placed in front of the lens on the intensifier. The spectral response of

the BG3 filter is shown in Figure 13.

Figure 14. Transmittance for BG3 filter (x-axis is wavelength in nm) [74]

Image Processing Techniques

Schlieren images of gas jet experiments are processed in MATLAB to determine the jet-tip penetration

and spreading angle. In multiple-injection cases, the second injection images require a more rigorous

image processing method than single injections for finding the jet-tip penetration due to increased

background noise from the first injection and excess fluid. Initially, these images are processed in the

same manner as the gas jet images for the first injection, but the binary images revealed that the actual jet

tip was not tracked due to noise from excess fluid between injections that results due to buoyancy.

Because the excess fluid was subtracted from the images in the background subtraction step and was

16

nearly the same intensity as the second injection gas jet, the background subtraction left a gap in the

center of the second jet, as shown in Figure 15.

Figure 15. Binary images at three instances in time after SOI2 are shown for case 15. The excess fluid

following the first injection was subtracted during the background subtraction step of image processing.

The intensity of the excess fluid was the same as the intensity of the second injection gas jet, resulting in

the tip of the second injection gas jet being subtracted out in the near-nozzle region. The blue line

represents the location of the jet tip as determined by the original jet-tip tracking algorithm.

To mitigate these issues, a sliding background subtraction technique is utilized and the region with the

highest intensity gradient is used to determine the location of the jet tip. First, the start of the injection is

determined using the pressure measurements taken inside of the injector cavity. This method was

validated using visual inspection of the first injection images, which showed that the start of the gas jet

could be seen within 5 frames of the time where the pressure first rose. Thirty images prior to the frame

for SOI1 are averaged and subtracted from the injection images in the pre-background subtraction step.

The absolute value of the image is taken to account for the negative intensity on one side of the jet and the

positive intensity on the opposite side of the jet (resulting from using a vertical knife edge) following the

pre-background subtraction step. A sliding background technique has also been reported by Skeen et al.

[23] for use in multiple diesel injections.

The sliding background subtracted images are obtained by averaging three pre-background subtracted

images following the image of interest (i+1, i+2, and i+3) and subtracting the image of interest (ith image)

from the averaged image as shown in Figure 15a. Next, the image is binarized using a threshold of 0.08

for the first 15 frames and a threshold of 0.1 for all following frames. The Matlab function ‘bwareaopen’

is used to reduce noise by eliminating all open areas that are smaller than 350 pixels as shown in Figure

15b. To evaluate the intensity at each downstream location, pixel intensity is averaged over 30 columns

(around the centerline) and 5 rows (the row of interest and two rows above and below it). Changes to

these parameters are noted where appropriate in the results section. The gradient of the intensity averaged

matrix is found using the ‘diff’ function, which calculates the difference between adjacent elements. The

maximum gradient is found at each instant in time, which provides the downstream location where the

image transitions from the dark region where the jet is located to the white region where the jet will be in

the next three frames, as shown in figure 15d.

17

Figure 16. Images for case 15 at 3.067 ms after SOI2 are shown after the following processing steps: (a)

subtracting ith background subtracted image from the average of i+1, i+2, and i+3 images, (b) binarization

with a 0.1 threshold, (c) binarization with noise reduction, and (d) determination of the location of highest

intensity gradient represented by the blue line.

However, near SOI2, there is a significant amount of noise in the region of the images where the jet tip is

located. Oftentimes, a region that does not correspond to the location of the actual jet tip is the region

with the highest intensity gradient. The search region was is limited in the algorithm to prevent erroneous

points from being chosen as the jet tip location. This is achieved by fitting a rational or polynomial curve

(determined on a case-by-case basis using the ‘cftool’) to the points later in time on the penetration plot,

which accurately track the jet tip, and placing upper and lower bounds +/- 30 pixels above and below the

curve as shown in Figure 17. The jet tip location near SOI2 is the location of the highest intensity gradient

that does not vary more than 5 pixels above or below the previously found point and is within the search

bounds. If the data point with the highest intensity gradient is not within 5 pixels of the jet tip location

found before it, the algorithm finds the next highest intensity gradient. The search process for one point

iterates until the search criteria is met or 5 iterations have occurred. If the jet tip location has not been

found after 5 iterations, the point is thrown out because there is too much noise in the image to accurately

find the jet tip.

(a) (b) (c) (d)

18

Figure 17. The search bounds for case 15 were found by fitting a rational curve to data that followed the

jet tip later in the injection. The concept for finding the search bounds was tested on the first injection (a)

and repeated for the second injection (b).

Injection Studies

The experiments conducted in this study included unsteady multiple-injections (discussed in Chapter 4),

unsteady simultaneous injections (discussed in Chapter 5), and wall-impingement (discussed in Chapter

5).

Unsteady Multiple Injections

Multiple-injection studies were conducted using both schlieren and PLIF. A single 1 mm injector orifice

is used. For schlieren experiments the working fluid is helium, the peak injector cavity pressure is 100

kPa, and the total injection duration is 60 ms. For PLIF experiments, the working fluid is acetone-seeded

air and the pressure and injection durations are scaled accordingly to account for differences in density

and specific gas constant. The test matrix for these experiments is provided in Chapter 4.

Unsteady Simultaneous Injections

The effects of jet-jet interaction are studied using both schlieren and PLIF with acetone-seeded air jets.

Two different injector nozzle configurations with three orifices are used to represent a 7-hole injector (51°

between jets) and an 8-hole injector (45° between jets). More details on the injector nozzle designs and

the test matrix for these experiments are given in Chapter 5.

19

Wall-impingement

Piston-bowl impingement studies are conducted using schlieren with acetone-seeded air jets. The bowl

designs for the piston-bowl impingement studies are described in Appendix C and the test matrix is

discussed in Chapter 5.

20

Chapter 3. SCALED INJECTION COMPARISONS

Theory

Naber and Siebers [1] developed an analytical, 1-D model for non-vaporizing spray penetration by using a

control volume analysis of conservation of mass and momentum. Their assumptions included a uniform

axial velocity profile, constant injection velocity with an instantaneous start, no velocity slip between the

fuel and entrained air, and a quasi-steady flow. The characteristics of diesel sprays can be inferred from

gas jet results by non-dimensionalizing experimental gas jet conditions and scaling to diesel conditions

using the characteristic equations given in [1]. Naber and Siebers [1] showed that ambient gas density has

a large effect on jet-tip penetration, and their scaling equations account for that effect, as well as the

gas/fuel density and injector geometry. The equations for characteristic length (x+) and time (t+) are given

in Equation 1 and Equation 2, respectively [1].

𝑥+ = 𝑑𝑓 ∙ √�̃�

𝑎∙tan(𝜃 2⁄ ) (1)

𝑡+ = 𝑑𝑓 ∙ √�̃�

𝑎∙tan(𝜃 2⁄ )∙𝑈𝑓 (2)

The angle, θ, is the dispersion angle of the spray, the term a is equivalent to 0.66 and is used to relate the

tangent of the measured spray angle to the theoretical spray dispersion angle. The effective diameter (𝒅𝒇)

of the gas exiting the injector orifice is approximately 0.8 for this experiment and was calculated using

Equation 3 [1], while the actual diameter of the orifice (𝒅𝒐) is 1 mm. Ca is the area contraction

coefficient.

𝑑𝑓 = √𝐶𝑎 ∙ 𝑑𝑜 (3)

The fuel density ratio (�̃�), given in Equation 4 [1], is the ratio of the fuel density (𝜌𝑓) to the ambient

density (𝜌𝑎).

ρ ̃ = ρf ρa⁄ (4)

The velocity (Uf) of the gas jet as it exits the orifice is calculated using Equation 5 [1], where the flow

coefficient is Cv, the injection pressure of the fuel is Pf and the ambient pressure in the laboratory is Pa.

𝑈𝑓 = 𝐶𝑣 ∙ √2 (𝑃𝑓− 𝑃𝑎)

𝑃𝑓 (5)

The dimensionless penetration time and dimensionless penetration distance are given in Equation 6 and

Equation 7, respectively, where t is the actual time and S is the measured penetration distance [1].

�̃� = 𝑡/𝑡+ (6)

21

�̃� = 𝑆/𝑥+ (7)

The engine’s characteristic values (𝒙𝒆𝒏𝒈+ and 𝒕𝒆𝒏𝒈

+ ) can be used to determine the experiment parameters.

The ratios of the characteristic values of the engine and experiment (𝒙𝒆𝒏𝒈

+

𝒙𝒆𝒙𝒑+ and

𝒕𝒆𝒏𝒈+

𝒙𝒆𝒙𝒑+ ) along with the engine

parameters can be used to determine the injection timing and scaling for the bowl shape in the gas jet

experimental setup. It is necessary to determine the proper scaling for the experimental setup prior to

testing to ensure that the data obtained will correspond to relevant conditions for the engine timing and

geometry of interest. Dimensionless penetration and time, calculated using Equation 6 and Equation 7,

will be used to discuss the results to enable easy comparison with vaporizing spray results from other

studies. Schlieren will be used to obtain the jet-tip penetration. The concentration of injected fluid will be

determined using acetone tracer-PLIF, which provides an indication of the mixture fraction in the jet

during fuel injections.

Single Injection

The results of single injection experiments for vaporizing and non-vaporizing sprays by Naber and

Siebers showed good agreement with their penetration correlation, as shown in Figure 18 [1].

Figure 18. Scaled liquid jet penetration measurements compared to penetration correlation derived from a

control volume analysis using conservation of mass and momentum [1].

Helium gas jet experiments with a fast ramp-up rates were reported by Abani and Ghandhi [64] and used

as a baseline to test the current experimental setup. The results, shown in Figure 19, are normalized using

the characteristic x+ and t+ values of those experiments. The results are compared with the theoretical

penetration curve that was obtained using the Naber and Siebers correlations [1].

22

Figure 19. Comparisons of the dimensionless penetration (�̃�) vs. dimensionless time (�̃�) for the Abani and

Ghandhi B5 test case [64] and a similar injection conducted using the RFDL gas jet setup.

The primary differences between the current experiments and those from Abani and Ghandhi were the

nozzle orifice sizes and ramp-up rate of the injection pressure. The injector orifice for the current injector

has a 1 mm diameter and the injector orifice in Abani and Ghandhi’s experiments had a 0.5 mm diameter

[64]. The difference in ramp rate is evident in the plot of the pressure curves for these experiments given

in Figure 20. The slight differences in the penetration rate of the Abani and Ghandhi and current helium

jets after EOI may be due to random error and uncertainty in the measurement of jet-tip penetration or

differences in the ramp up rate, which was shown to affect the time at which jet-tip deviation (from the

steady case) starts and the magnitude of the difference in penetration rate in [64]. The penetration rate

curve from the Naber and Siebers correlation, which is for a steady jet, is similar to the experimental

penetration rates at the beginning of the injection but not after the end of the ramp up period; this was

expected based on the results of [64].

Figure 20. Injection pressure curves for helium gas jet experiments by Abani and Ghandhi [64] and RFDL

researchers.

23

The results in this section are proof-of-concept for the gas jet experimental design and the ability to

capture similar jet tip penetration as predicted for a steady jet using the Naber and Siebers correlations

[1].

Scaled Injection Studies

To obtain data from gas jet experiments that correspond to relevant conditions for a given engine and

injection timing, the parameters of the experiment need to be defined prior to testing. The gas jet

experiments are designed with characteristic length and time scales that allow achievable pressures and

injection timings, enable sufficient spatial and temporal resolution, and provide results that will scale to

the conditions of the injection of interest.

The engine conditions and scaling values for the study of injector and bowl geometry are described in

detail in Chapter 5. The injection timings and pressures chosen for the multiple-injection study, however,

are not scaled to be compared with a particular engine or injection test case but rather are chosen based on

the ability to produce repeatable pressure curves with both a fast ramp rate and top hat injection profiles.

The results of the multiple injection studies are normalized to allow for easy comparison with results from

experiments at engine conditions. Injection timings for other recent studies in the diesel literature are

scaled for comparison with the injection timings in the multiple-injection cases to check that the results

are relevant to diesel applications. The x+ and t+ values for the multiple-injection tests are calculated using

Equation 1 and Equation 2 with the parameters given in Table 2 and were equal to 0.0024 and 6.71x10-6,

respectively. The Cd was approximated close to the measured discharge coefficient of the single hole

nozzle plate, which was 0.84, for an injection pressure that is twice the ambient pressure (see Appendix

D). The discharge coefficient of the nozzle was not measured until after the following scaling analysis

was complete.

Table 2. Experiment parameters necessary to calculate characteristic values of gas jet injections

df (mm) ρf (kg/m3) ρa(kg/m3) Pf (kPa) Pa (kPa) γ Cd θ/2

0.8 0.3247 1.1839 200 101.3 1.66 0.8 15°

The non-dimensionalized injection times, �̃�, were calculated for various injection durations that are used

in the ECN’s Spray A studies [24]. The parameters of the experiments along with the calculated t+ and �̃�

values are given in Table 3. In all of the test cases in Table 3, x+ is 0.0032.

Table 3. Experiment parameters for ECN Spray A injector studies [24]

Reference df (mm) ρf

(kg/m3) ρa(kg/m3) Pf (kPa) Pa (kPa) Cd θ/2

Injection

duration (ms) t+ �̃�

Skeen et al. [23] 0.08 698 22.8 150000 35000 0.888 12° 0.5 6.24E-6 80.1

Skeen et al. [75] 0.08 698 7.6, 15.2,

22.8 150000

26586,

53024,

78873

0.91 12° 4

9.40E-6,

6.70E-6,

5.52E-6

425.6,

596.4,

723.9

Pickett et al. [76] 0.08 698 22.8 150000 35000 0.888 12° 1.5 6.18E-6 242.6

Benajes et al. [77] 0.08 698 22.7 150000 60500 0.91 12° 5.3 5.51E-6 962.6

The shortest injection duration for a multiple-injection test case in this work was �̃� = 633.2 and the

longest injection duration was �̃� = 2532.6 for the single injection case. The shortest dwell time was �̃� =

211.1 and the longest dwell time was �̃� = 633.2. While all of the multiple-injection durations and dwell

24

times were longer than the injection duration and dwell time in the ECN split injection test case [23], the

shorter injection durations do fall within the range for single injections that have been studied by the

ECN. Unfortunately, the dwell time between injections is system-limited and shorter dwell times between

injections are not achievable at the injection pressure that was chosen.

25

Chapter 4. EFFECTS OF MULTIPLE INJECTIONS

The objective of this portion of the study is to explore the fluid mechanic interactions between multiple

consecutive injections in free jets. There are several physical effects of multiple injections that have

previously been studied including slipstreaming [23] and mixing [37]. The possibility of slipstreaming

was initially investigated as a function of DOI1 and dwell using penetration measurements, and mixing

was studied using the dispersion angle of the jet. A follow-up study using acetone tracer-PLIF will look at

the effects of ambient density and turbulence, penetration rates after the end of injection (aEOI), and

quantitative scalar concentration measurements.

Injection Schedules

The test matrix for this study, shown in Table 1, includes injection schemes with two injections of varying

duration and three different dwell times. A single injection with an injection duration that is the same as

the total injection time for the multiple-injection cases is used for comparison. The non-dimensional

injection durations are of the same order of magnitude of a typical diesel engine injection, and are chosen

to optimize the clarity of the experimental results.

This test matrix allows effects of the relative duration and dwell between the first and second injection to

be investigated. Variations in dwell time were expected to affect the extent of mixing of the fluid from the

first injection prior to SOI2. The dispersion of the first injection subsequently affects the amount of

entrainment of the fluid from the first injection into the second injection. In an engine, the velocity of the

second injection relative to the first injection and the entrainment of first injection fluid into the second

injection would affect the local mixture composition.

The relative duration of the injections also influences the physics of the interaction. The flow field and

local mixtures that result from a main injection interacting with a short pilot injection are different than

for a short post injection interacting with a main injection. It was expected that pilot and post injections of

varying durations would behave differently. The test schedule is devised to include pilot, split and post

injections. For each of the three dwell times (�̃� = 211.1, 422.1, and 633.2) there are seven first- and

second-injection timings, which sum to a total �̃� = 2532.6 (60 ms) injection duration.

26

Table 4. The injection strategies include pilot, split, and post injections with three different dwell

durations given in non-dimensional time (�̃�): short 211.1 (5 ms) dwell (cases 1 – 7), intermediate 422.1

(10 ms) dwell (cases 8 – 14), and long 633.2 (15 ms) dwell (cases 15 – 21). The multiple injections were

compared with a 2532.6 (60 ms) single injection (case 22).

Case # Peak Injector Cavity

Pressure (kPa)

1st Injection

Duration Dwell 2nd Injection

Duration

1 100 633.2 211.1 1899.5 2 100 844.2 211.1 1688.4

3 100 1055.3 211.1 1477.4

4 100 1266.3 211.1 1266.3

5 100 1477.4 211.1 1055.3

6 100 1688.4 211.1 844.2

7 100 1899.5 211.1 633.2

8 100 633.2 422.1 1899.5 9 100 844.2 422.1 1688.4

10 100 1055.3 422.1 1477.4

11 100 1266.3 422.1 1266.3

12 100 1477.4 422.1 1055.3

13 100 1688.4 422.1 844.2

14 100 1899.5 422.1 633.2

15 100 633.2 633.2 1899.5 16 100 844.2 633.2 1688.4

17 100 1055.3 633.2 1477.4

18 100 1266.3 633.2 1266.3

19 100 1477.4 633.2 1055.3

20 100 1688.4 633.2 844.2

21 100 1899.5 633.2 633.2

22 100 2532.6 N/A N/A

There are nine pilot injection cases (1, 2, 3, 8, 9, 10, 15, 16, and 17), three split injection cases (4, 11, and

18), nine post injection cases (5, 6, 7, 12, 13, 14, 19, 20 and 21), and a single injection case (22) in total.

The split injection cases are of particular interest because a multiple-injection case consisting of equal

split injections (Spray A) is currently a focus of the Engine Combustion Network (ECN) [24].

Single Injection

A single 2532.6 non-dimensional injection time, �̃� , (60 ms in real time) is used for baseline comparisons

for the 21 multiple-injection cases in Table 4. The pressure profile for the single injection case, given in

Figure 21, shows that the pressure peaked at 107.5 kPa and the actual non-dimensional DOI is 2583.3.

The experimental setup provides highly repeatable injections, as shown in Figure 21, which shows +/- one

standard deviation for the ensemble averaged pressure profile.

27

Figure 21. The pressure profile for the 2583.3 (61.2 ms) single injection case +/- 1 standard deviation for

20 ensemble averaged trials shows that the gas jet experiments were highly repeatable.

Pressure Traces

The injection pressure profiles presented in this section are ensemble averaged for 20 ensembles of the

same injection timing based on the results of the convergence study described in Appendix D. The

pressure profiles for pilot, split, and post injection cases with varying dwell times but the same injection

durations are given in Figure 22, Figure 23, and Figure 24, respectively. Injections with the same dwell

time but different injection durations are also compared. A lower peak injection pressure during the

second injection is an overall trend in all of the test cases.

Figure 22. The pressure profiles for pilot injections with similar injection durations but different dwell

times are compared with the baseline case and shown for (top) cases 1, 8, and 15, (center) cases 2, 9, and

16, and cases (bottom) 3, 10, and 17.

28

Figure 23. The pressure profiles for split injections with similar injection durations but different dwell

times are compared with the baseline case.

Figure 24. The pressure curves for post injections with similar injection durations but different dwell

times are compared with case 22 and shown for (top) cases 5, 12, and 19, (center) cases 6, 13, and 20, and

(bottom) cases 7, 14, and 21.

Jet-tip Penetration

The jet-tip penetration, S, is determined by tracking the tip of the helium gas jets in the schlieren images.

The camera viewing window has a height of 12.5 cm, limiting the distance over which the jet-tip

penetration is measured. The penetration rate of the first injection is determined for each trial and

averaged over 20 ensembles. To enhance image processing, ensemble averaged images are used to track

the jet-tip penetration of the second injection. Therefore, the standard deviation of S2 is not provided but

is expected to be similar to that of S1.

Comparisons of S1 and S2 across all multiple-injection cases show that there is not a significant difference

in the jet-tip penetration up to 12.5 cm downstream of the injector. Figure 25 and Figure 27 show the jet-

tip penetration for pilot injections and post injections, respectively, with the same injection durations but

varying dwell time. Figure 26 shows the jet-tip penetration for the split injections.

29

The slipstreaming phenomenon that was previously noted in other experimental work on multiple

injections is not observed in the results of this study. While slipstreaming is not observed under these test

conditions, previous pilot injection tests with this injector did result in a higher jet-tip penetration for the

second injection near EOI.

Figure 25. The jet-tip penetration of the first injection (S1) and the jet-tip penetration of the second

injection (S2) are shown for pilot injections with different dwell periods. There is not a significant

difference in S1 and S2.

30

Figure 26. S1 and S2 are shown for split injections with different dwell times.

Figure 27. S1 and S2 are shown for post injections with different dwell times.

31

Dispersion Angle

The jet dispersion angle, θ, is a metric that correlates with the amount of fluid entrained into a jet [61].

Previous studies of non-vaporizing sprays have shown that the dispersion angle depends on the ambient

density [78]. Because the second gas jet is injected into air mixed with helium from the first injection, the

ambient density is lower during the second injection than during the first injection. Additionally, the

ambient turbulence level increases after the end of the first injection. Therefore, it is expected that the

dispersion angle of the first and second injections will not be the same. A difference in dispersion angle

implies that the entrainment of surrounding fluid into the gas jet and the extent of mixing are different

during the first and second injections.

At the beginning of an injection, the jet dispersion angle is high and decreases with time until it becomes

nearly constant [1, 79]; this phenomenon is observed in the instantaneous dispersion angle results as

shown in Figure 28 for a split injection (case 11).

Figure 28. The half angle, θ1/2 and θ2/2, is given at each instant in time for both the first and second

injection for the �̃� = 422.1 dwell split-injection case. The black dashed lines indicate SOI1 and SOI2.

Generally, the first injection has a higher half angle than the second injection, as shown for the split

injection cases in Figure 29. With the exception of cases 17 and 19, any cases where θ2/2 is less than θ1/2

the half angle for the second injection is within +/- one standard deviation of the dispersion angle of the

first injection.

Figure 29. The half angle, θ1/2 and θ2/2, are given in relative non-dimensional time from SOI1 and SOI2

for both the first and second injection for the �̃� = 422.1 dwell split-injection case.

32

To compare the average dispersion angle during the quasi-steady portion of the injection across test cases,

the half angle i averaged over 30 frames that represent the same time from SOI for each injection, as

shown in Figure 29. The average half angle for the first and second injections are 15.4 degrees and 15

degrees, respectively. The error bars on the data indicate the 95% confidence interval for each averaged

angle. Comparisons across different dwell times and injection durations show that injection angle does

not vary with the duration of injection nor dwell time. It is apparent, however, that θ1 is generally higher

than θ2, as shown in Figure 30.

Figure 30. The averaged dispersion half angle, θ/2, is shown as a function of first injection time for

various dwell periods. There does not appear to be a trend with dwell time nor injection duration. The

error bars show the confidence interval. The average confidence interval for θ1 and θ2 are 0.10 and 0.07,

respectively. Note: the limits of the y-axis are 14 to 18 degrees.

Differences in the dispersion angle are also observed by looking at the ensemble-averaged schlieren

images, as shown in Figure 31. Following Skeen et al. [23], the first injection is shown by the red outline,

while the ensemble-averaged inverted schlieren image shows the second injection. Figure 31 shows the

three split-injection cases at various instances in time. At �̃� = 168.84 after SOI, the second injection

dispersion angle is visibly less than the dispersion angle of the first injection.

33

Figure 31. The sequence of schlieren images shows the relative jet-tip penetration of the first and second

injection in the split injection cases (4, 11, and 18). The inverted schlieren image is the second injection

fluid penetrating through the remnants of the first injection. Time is given in non-dimensional time from

SOI relative to each injection, where 𝑡 ̃= 0 is SOI1 and SOI2. The red lines overlaid on the image represent

the boundary of the first injection and the blue line represents the location of the jet tip of the second

injection found using the jet tip tracking algorithm explained in the Image Processing Techniques section.

34

Chapter 5. BOWL GEOMETRY AND NOZZLE CONFIGURATION

EFFECTS

Experiment Configurations and Scaling

A series of schlieren and acetone tracer-PLIF experiments were conducted to determine the optimal bowl

geometry and nozzle configuration for effective in-cylinder mixing as part of Volvo’s 55% BTE effort.

Given the engine parameters in Table 5 along with an ambient density (ρa) of 75 kg/m3, a fuel density (ρf)

of 835 kg/m3, a discharge coefficient (Cd) of 0.721, and a dispersion angle (θ) of 26° the characteristic

values of the engine (x+eng and t+

eng) were calculated using the scaling equations given in Equation 1 and

Equation 2 in Chapter 3 [1].

Table 5. Engine parameters and characteristic time (t+eng) and distance (x+

eng) values

Number of

Orifices

Orifice

Diameter, df

(mm)

Rail Pressure,

Pinj (bar)

Injection

Duration (ms)

Distance from

injector to bowl wall

(cm)

x+eng t+

eng

7 0.257 1200 2.96 4.24 5.18E-03 1.14E-05

7 0.257 1800 2.42 4.24 5.18E-03 9.29E-06

7 0.257 2400 2.1 4.24 5.18E-03 8.04E-06

8 0.238 1200 2.96 4.24 4.80E-03 1.06E-05

8 0.238 1800 2.42 4.24 4.80E-03 8.62E-06

8 0.238 2400 2.1 4.24 4.80E-03 7.46E-06

For an actual orifice diameter of 1 mm, the effective orifice diameter (df) of the experiment was 0.8 mm.

The discharge coefficient used in the scaling calculations was 0.8 based on the measurements described in

Appendix C.

The jet dispersion angle was assumed to be approximately 30° based on the results of the study described

in Chapter 4. The specific gas constant of the acetone and air jet was calculated using the partial pressures

of acetone and air and the ideal gas law, assuming that the temperature is 300 K and the seeder tank

pressure is 10 kPa higher than the peak pressure in the injector cavity during the injection. The pressure

difference between the peak injector cavity pressure and the pressure in the acetone seeders was gathered

from observation by reading the pressure gages on the acetone seeders. The velocity of the jet at the

orifice exit was assumed to be close to sonic conditions (Uf = 300 m/s ). The x+ratio (x+

eng/x+exp) is 1.28 and

the t+ratio (t+

eng/t+exp) values for the various injection times are given in Table 6.

Table 6. The ratios of the characteristic time of the engine to the characteristic time of the experiment,

t+ratio (t+

eng/t+exp), are given for test cases with different injection times.

Angle between

orifices (degrees)

Injection

Pressure (kPa)

Engine Injection

Duration (ms)

Experiment Injection

Duration (ms) t+

ratio

51° 120 2.96 18 6.21

51° 130 2.42 16 6.61

51° 140 2.1 14 6.67

45° 120 2.96 18 6.21

45° 130 2.42 16 6.61

45° 140 2.1 14 6.67

35

The test matrix given in Table 7 included two bowl shapes and two nozzle configurations for schlieren

and LIF experiments with acetone-seeded air jets. The 2-D bowl designs that were supplied by Volvo are

given in Appendix F. The gas jet tests are conducted at three different injector cavity pressures

(corresponding to injector rail pressures, Pinj, of 1200, 1800, and 2400 bar) and an umbrella angle of 150°.

Table 7. Test matrix for studying bowl geometry effects and jet-jet interactions

Diagnostic Technique Injector Design Bowl Design Injection Pressure

(kPa)

Injection Duration

(ms)

schlieren

Single Extruded Bowl 1 120 18

Single Extruded Bowl 1 130 16

Single Extruded Bowl 1 140 14

Single Extruded Bowl 3 120 18

Single Extruded Bowl 3 130 16

Single Extruded Bowl 3 140 14

3-hole, 51° Semi-circular 120 18

3-hole, 51° Semi-circular 130 16

3-hole, 51° Semi-circular 140 14

3-hole, 45° Semi-circular 120 18

3-hole, 45° Semi-circular 130 16

3-hole, 45° Semi-circular 140 14

PLIF

3-hole, 51° Revolved Bowl 1 130 16

3-hole, 51° Revolved Bowl 3 130 16

3-hole, 45° Revolved Bowl 1 130 16

3-hole, 45° Revolved Bowl 3 130 16

Schlieren experiments are utilized to qualitatively understand how bowl shape and nozzle number affect

the motion of the fluid near the bowl wall and the extent of air utilization, which is indicative of increased

mixing in the cylinder. The extent of air utilization is qualitatively observed by tracking the fluid moving

towards the nozzle after interacting with a wall. Two experimental configurations are used: single

injections with the extruded bowl shape allow bowl effects to be observed and three simultaneous

injections with nozzle hole spacing of 51° (corresponding to an 7-hole injector configuration) and 45°

(corresponding to a 8-hole injector configuration) with a semi-circular bowl shape are used to explore the

effects of jet-jet interaction. The experimental setups for the extruded bowl, single-jet test cases and the

semi-circular bowl, three-jet test cases are shown in Figure 32 and Figure 33, respectively. The injector

orifice diameter is 1 mm for all of the tests conducted in this study. The designs for the injector nozzle

plates are given in Figure 34.

36

Figure 32. The extruded bowl, single-jet experiments were conducted using Bowl 1 (left) and Bowl 3

(right). The distance between the bowl and the nozzle was the scaled distance for the piston location at

TDC. The bowl was angled at 15° from the vertical centerline of the jet based on a 150° umbrella angle.

The orifice location is marked by the blue dot. The measured angles were 15.1° and 14.9° for Bowl 1 and

Bowl 3, respectively.

Figure 33. The semi-circular bowl, three-jet experiments were setup so that the center of the bowl was

aligned with the centerline of the injector nozzles with 51° jet spacing (left) and 45° jet spacing (right).

37

Figure 34. Injector nozzle plates with three 1 mm orifices with 51° spacing for a 7-hole injector

configuration (left) 45° spacing for an 8-hole injector configuration (right) are used in the schlieren and

LIF experiments with the semi-circular bowl and revolved bowl designs.

A smaller number of experiments using acetone tracer-PLIF are utilized to understand fluid motion and

mixing with the realistic bowl geometry and a 3-jet nozzle configuration so jet-jet interaction can be

observed in a plane through the centerline of the jets rather than from an integrated line-of-sight

perspective. Eight experimental configurations are used. As shown in Table 7, two bowl shapes and two

injector nozzle plates with nozzle hole spacings of 51° (corresponding to an 7-hole injector configuration)

and 45° (corresponding to a 8-hole injector configuration) were used. Two different laser optic alignments

and injector positions were used in order to look at the flow field in the near nozzle region and near the

bowl wall. The alignment of the extruded bowl setups for schlieren and the revolved bowl setups for PLIF

are discussed in detail in Appendix D.

Pressure Curves

The ensemble-averaged pressure curves for schlieren experiments are shown in Figure 35 and Figure 36

for the test cases with the extruded bowl and the semi-circular bowl, respectively. The actual injection

times for the experiments were longer than the calculated scaled injection times due to software and

experimental response time limitations, so the calculated EOI times (14, 16, and 18 ms) are given by

vertical dashed lines. The penetration and spreading angle after the scaled EOI time would be irrelevant,

however, in all cases the jets hit the bowl wall well before EOI. It is still necessary to only consider the

motion of fluid moving back until the scaled EOI and not after EOI.

Figure 35. The ensemble-averaged pressure curves are shown for three schlieren experiments extruded

bowl 1 (left) and extruded bowl 3 (right). The subscripts in the legend indicate the injector pressures in

38

the engine (1200, 1800, and 2400 bar) which correspond to experimental injection pressures of 20, 30,

and 40 kPa, respectively. The vertical dashed lines correspond to the scaled EOI timing because the actual

injection time was longer.

Figure 36. The ensemble-averaged pressure curves are shown for three schlieren experiments with a semi-

circular bowl and 51° nozzle spacing (left) and 45° nozzle spacing (right). The subscripts in the legend

indicate the injector pressures in the engine (1200, 1800, and 2400 bar) which correspond to experimental

injection pressures of 20, 30, and 40 kPa, respectively. The vertical dashed lines correspond to the scaled

EOI timing because the actual injection time was longer.

Penetration and Dispersion Angle

In the extruded bowl schlieren test results the penetration curves showed similar trends at the three

different pressures that were tested, as shown in Figure 37. The penetration curves are expected to be very

similar because the difference in the injection pressure between the minimum (120 kPa) and maximum

(140 kPa) pressures tested was small (less than 17% of the minimum pressure) compared to the injection

pressures. The schlieren images for the extruded bowl test cases with a 140 kPa injection pressure were

not clear enough and the brightness of the light source was not consistent enough over the test period to

obtain measurements of penetration and dispersion angle using the MATLAB algorithm that is described

in Chapter 2.

Figure 37. The penetration curves for schlieren experiments with extruded Bowl 1 and extruded Bowl 3

show very similar trends. The error bars on the figure represent the 95% confidence interval.

39

The ensemble-averaged dispersion half angle did not change significantly as a function of pressure or

bowl design either. The dispersion half angle is high near SOI in each case, however, the large

fluctuations in dispersion angle between 1 ms and 3 ms are not real and are caused by fluctuations in the

intensity of the light source. The mean dispersion half angle is averaged from 2.9 ms to 3.5 ms (10

frames) are shown in Table 8 and is between 14° and 15° for each extruded bowl test case.

Figure 38. The ensemble averaged dispersion half angles are similar near SOI and after fluctuations in

light source intensity have stopped (after approximately 2.5 ms aSOI) across test cases with different

extruded bowl designs and injection pressures. The region of the injection that is used to calculate the

mean dispersion half angle is indicated using vertical black lines.

Table 8. Mean dispersion half angle averaged over the quasi-steady portion of injection for the extruded

bowl test cases.

Bowl Design Engine Injection

Pressure (bar)

Experiment Injection

Pressure (kPa)

Mean Dispersion Half

Angle, θ/2 (degrees)

95% Confidence

Interval (degrees)

1 1200 20 14.07 0.34

1 1800 30 14.16 0.24

1 2400 40 14.52 0.29

3 1200 20 14.73 0.15

3 1800 30 14.34 0.31

There are slight differences observed in the penetration rate of the three adjacent jets in the test cases with

the semi-circular bowl shape. However, these differences are attributed to random error and error from

the jet tip tracking algorithm inaccurately tracking the jet tip location after the LED light source flashed

(at about 3 ms aSOI). Looking at the jet tip location plotted on the schlieren images, it is evident that the

right and center jets were tracked more accurately than the left jet. A difference in the penetration rate of

the center jet compared to the side jets is expected if jet-jet interactions are significant enough to impact

the jet development. Comparing the penetration rates of the right jet and the center jet in each case shows

that there is not a significant difference in penetration and that changing the angle between adjacent jets

by 6° does not have a noticeable effect on jet-tip penetration before the jet interacts with the bowl wall.

40

Figure 39. The ensemble averaged penetration for schlieren experiments with the semi-circular bowl

shape.

Effects of Wall Impingement

From qualitative observations of the fluid motion near the bowl wall in the schlieren images for the

extruded bowl cases, the deeper bowl appears to allow the fuel to spread more, resulting in better air

utilization. However, it is important to keep in mind that for the schlieren test cases the bowls were both

positioned much further from the injector in the x’ position than would be at the time of impingement (see

Appendix D). It is also important to note that while Figure 40 shows that the fluid that interacts with the

wall of Bowl 3 moves back towards the nozzle alongside the jet, the fluid that interacts with the wall of

Bowl 1 is being pushed into the head of the jet.

41

Figure 40. The schlieren images for tests with extruded Bowl 1 (left) and Bowl 3 (right) are shown for

increasing time after SOI. The blue dot indicates the location of the injector orifice.

This effect is also observed in the LIF results, as shown in Figure 41 and Figure 42. The LIF images were

recorded for a single plane through the centerline of the jets. After the jets impinge on the bowl wall (at

about 4 ms aSOI), it appears as though the jets are spreading wider, but this is likely due to fluid coming

back from the bowl wall into the centerline plane of the jet as the two regions are somewhat disconnected.

The increased quantity of acetone-seeded fluid around the edges of the jets appears to be more prevalent

with the Bowl 1 geometry. It is possible that this occurs because after the jets interacts with the wall of

Bowl 1, they move back towards the jet in this plane. However, after the jets interact with the wall of

Bowl 3, they move back towards the nozzle along the bowl wall and the fluid is not pushed towards the

jet as much, and it would not appear in the images taken in the plane along the centerline of the jets.

42

Figure 41. In the images recorded with a 51° spacing between the jets, Bowl 3 appears to be better for

allowing the acetone-seeded air to spread further in the bowl. This indicates that in an engine, this bowl

design may allow more air to be utilized during combustion processes.

43

Figure 42. In the tests with a 45° spacing between the jets, Bowl 3 appears to be better for allowing the

acetone-seeded air to spread further in the bowl. This indicates that in an engine, this bowl design may

allow more air to be utilized during combustion processes.

Effects of Jet-jet interaction

One metric that is used to understand the effects of jet-jet interaction on mixing is the rate at which the

acetone-seeded air moves back towards the nozzle after interacting with a wall. In the near nozzle region

there were no obvious differences in the PLIF images when comparing images from tests with the same

bowl shape and different injector orifice angles other than the spacing between the jets. However, in the

region near the bowl wall there appeared to be more acetone-seeded fluid moving back towards the nozzle

in the space between the jets in the configuration with a 51° angle between injector orifices (7-hole

injector design) compared to a 45° angle between injector orifices (8-hole injector design), which is

pointed out using a blue arrow in Figure 43. Therefore, the 7-hole injector design also allows for better

mixing.

44

Figure 43. After the acetone-seeded jets of air interact with the bowl wall, more acetone-seeded fluid

appears to move further back towards the nozzle in the images shown for the 51° spacing (7-hole injector

configuration). The blue arrows show the direction the fluid is moving back (normal to the bowl surface).

45

Chapter 6. CONCLUSIONS/FUTURE WORK

This thesis considers the effects of altering injection duration, dwell time between multiple injections, jet

spacing, and bowl shape on jet penetration and mixing for applications to diesel combustion. The results

of the multiple-injections studies show that the penetration rate of the second injection in the near-nozzle

region is unaffected by changes in DOI1 and the dwell between injections before EOI. Comparisons of

the dispersion angles between the first and second injection show that the first injection has a higher

dispersion angle than the second injection. The significance of the differences in dispersion and mixing

during 1st and 2nd injections will be further investigated using acetone tracer-PLIF. The comparisons of

bowl geometry showed that for the particular designs used in these studies, a deeper bowl allowed for

better mixing of the acetone-seeded air and improved air utilization. Qualitative analysis of two injector

configurations (45° and 51° jet spacing) showed that the injected fluid tended to spread more rapidly and

further in the regions between jets when the spacing is wider and there is less jet-jet interaction.

The results of this work can be applied to diesel engine combustion in a number of ways. The

experimental results shown here are proof that gas jet experiments can be used to guide the design of

advanced engines. Differences in the jet spreading angles for multiple injections may indicate a difference

in the mixing processes in the near-nozzle region between the first and second injections. These near-

nozzle mixing processes determine important quantities in a reacting jet, including ignition delay time,

lift-off length, and soot volume fraction in the spray. Differences in the flow field using PLIF images for

injections with multi-hole nozzles and wall-impingement indicate that bowl and injector geometry can

alter jet entrainment and especially affect late-cycle mixing. Late-cycle mixing can affect the flow field

and local combustion chemistry having an impact on the extent of combustion and emissions. Developing

a better understanding of multiple-injection mixing, adjacent jet interactions, and jet-wall interactions can

help improve the way that injection schedules are designed. Additionally, the image processing tools that

were developed in this work for tracking the penetration and dispersion of a jet moving through the

remnants of a prior close-coupled injection can be used by future researchers in post-processing and

understanding their gas jet experimental results.

For future work, the multiple-injection studies could be greatly expanded. The sensitivity of the jet tip

tracking algorithm to changing the search parameters for finding the location of the jet tip based on the

maximum intensity gradient (i.e. how many pixels to look ahead of the previous jet tip location, how

many pixels to average the intensity over in the vertical and horizontal directions, how tight to make the

search bounds, etc.) should be determined. Near-nozzle mixing processes during multiple injections

should be further investigated using high-speed acetone-tracer PLIF. Obtaining quantitative

measurements of the mixture fraction in the jet using acetone-tracer PLIF will allow the significance of

the differences in dispersion of the 1st and 2nd injections to be determined. The cause of the differences in

dispersion angle between the two jets should also be determined if the differences in mixture fraction

throughout the flow field are found to be significant. The differences in jet dispersion could be due to

increased ambient turbulence level or lower ambient density during the second injection. To see if the

results show a difference in dispersion when the ambient density is the same, two experiments with the

same injection timings and pressures should be tested: an injection of air followed by an injection of

acetone-seeded air, and in a second experiment, an injection of acetone-seeded air followed by an

injection of acetone-seeded air. In the first experiment, the ambient density will be the same for the first

and second injections, but the turbulence field will be different. In the second experiment, both the

ambient density and turbulence field will be different for the second injection as compared to the first. If

the results are the same for both experiments, then the difference in dispersion that was previously

46

observed is the result of differences in ambient turbulence. If the results are significantly different for

both experiments, the difference in dispersion is the result of differences in the ambient density into

which the gas jets are injected.

While slipstreaming was not observed at the locations measured in this multiple-injection study, the

distance and time over which the images were recorded may not have been long enough to see that effect.

In the study by Skeen et al. [23] the penetration rate of the 2nd injection only deviated from the

penetration curve for the 1st injection after EOI. In the gas jet multiple-injection studies the penetration

rate was only recorded for the first 8 ms (�̃� ≈ 338), but the shortest injection time was 15 ms (�̃� ≈

4220). In future multiple-injection studies the penetration rate should always be recorded after EOI. The

experimental setup will need to be adjusted so that a second set of schlieren images can be recorded at

locations further downstream if the same test parameters are used in the future. A second option is to

record both near-nozzle and far field images of the acetone-seeded gas jet while using acetone-tracer

PLIF and track the jet tip penetration in the PLIF images.

In addition to expanding on the multiple-injection studies, it would be useful to determine the uncertainty

in our penetration and mixture fraction measurements. Uncertainty quantification is necessary to

determine if differences in results from different test cases or results recorded in different facilities are

significant and due to physical phenomena or due to bias and random error. It is also necessary for

accurate model validation. Sensitivity analyses performed previously by Eagle et al. [80] on liquid length,

vapor penetration, ignition delay, and flame lift-off length quantified bias error due to systematic

uncertainties, propagated uncertainty due to fluctuations in boundary conditions, and the standard

deviation. Vapor penetration was only briefly discussed and standard deviation was quantified, but

sensitivity to bias error was not discussed in depth due to a lack of measurements with suitable parametric

variations and was left to future work. However, the methodology and uncertainty/sensitivity equations

explained in [80] could be applied to find the uncertainty of penetration and mixture faction from gas jet

experiments conducted in the RFDL.

Ramp-up rates are quantified for gas jet experiments by taking the derivative of the ramp-up portion of

the pressure curve. As discussed in the Steady and Unsteady Jets section of Chapter 1, the ramp-up rate

can affect the penetration rate of the jet and the ramp-down rate can affect entrainment after EOI [65, 66].

In future studies, the ramp-down rate could be calculated similarly using the pressure measurements in the

injector cavity. In developing the test matrix, ramp-down rates could be put into the reduced order model

by Knox et al. [81] to determine conditions under which combustion recession would be likely to occur in

the engine of interest. The jet-tip penetration, velocity field, and scalar concentrations should be measured

after EOI to observe the effects of the “entrainment wave” on the flow field with both transient ramp-

down and slow ramp-down conditions.

47

APPENDIX A

Dimensioned Injector Drawings

The following drawings are given for the stainless steel injector described in Chapter 2. The injector was

machined by Larry Horner.

Figure A. 1. Injector body side view

Figure A. 2. Injector body tilted view with dimensions for nozzle plate and mounting plate screw holes

48

Figure A. 3. Injector body top view

Figure A. 4. Injector cross-sectional view of valve and pressure transducer connections

The injector was designed to be able to mount on a stand allowing it to be rotated about its center and

raised/lowered on a post. To mount the injector body on its stand, rather than on the optics bench, the

rotation plate is attached to the 3 ¼-20 holes on the side of the injector body. The rotation plate is attached

to the mounting plate using 10-24 screws. The mounting plate can then be slid onto a ½” diameter post

and held in place by tightening the set screw. The bottom of the ½” diameter post is held in a collar that is

attached to a based with ¼”-20 holes for attaching the stand to either a turntable for the optics bench.

49

Figure A. 5. Rotation plate

Figure A. 6. Mounting plate

50

Figure A. 7. Base for injector stand

Figure A. 8. Final assembly of injector body mounted on the stand

51

APPENDIX B

Acetone Bubblers and Nozzle Design

The acetone bubbler design was based on a similar bubbler that was used at Georgia Tech to study

synthetic jets with acetone-tracer PLIF [82] and is used in the ExCCL at Penn State University. A

pressure tank with several ports for inlet flow, outlet flow, a pressure gage, and a pressure relief valve was

partially filled with liquid acetone. A nozzle was attached to the end of tube for the inlet in the center of

the tank and submerged below the liquid acetone. The design of the submerged nozzle is based on the

Laskin nozzle originally described by Laskin in 1948 [83]. The nozzle is made from stainless steel and a

dimensioned drawing of the nozzle is shown in Figure B. 1. The nozzle has an inlet tube with a 0.25”

inner diameter attached to a plate at the bottom with a hollow ring. The hollow ring has a diameter of

0.875” and a depth of 0.0625”. After the hollow ring was machined, a solid circular flat plate that was

0.0625” thick was welded to the bottom.

Figure B. 1. A dimensioned side view drawing (left) and top view drawing (right) are given with units in

inches for the Laskin nozzles used in the acetone bubblers.

Pressurized air is sent through the inlet tube to the hollow ring with four radial 0.02” holes at the top. As

the compressed air rapidly flows out of the holes, acetone is pulled into the airflow and atomized into the

air bubbles. When the bubbles reach the surface of the liquid acetone, they burst and release the acetone

particles into the air above the liquid, creating a mixture of air and acetone vapor in the top of the bubbler

that travels through the outlet tube and to the injector as shown in Figure B. 2. A needle valve is used to

relieve the pressure in the tanks when necessary to prevent backflow of acetone to the compressed air

supply tanks. A photo of the acetone bubbler setup is given in Figure B. 3.

52

Figure B. 2. The inlet air travels through a Laskin nozzle near the bottom of the pressure tank and air

bubbles carry atomized acetone to the surface of the liquid acetone in the bubbler producing acetone-

seeded air that travels through the outlet.

Figure B. 3. The photo of the acetone bubbler setup shows the location of the inlet airline, the outlet

airline, the pressure gage, and the needle valve that is used for pressure relief.

53

APPENDIX C

Pressure Transducer Calibration and Hardware Setup

Calibration calculations were conducted to obtain the offsets for the pressure transducers (PT) using the

calibration sheets from Omegadyne, Inc (see PT folder in Research East 122 for details). Equation C. 1,

can be used to calculate calibration factor (CF) for the output of the transducer based on its sensitivity.

The pressure is found using Equation C. 2, where the calibration factor that is input into LabView to

calculate the pressure based on the output voltage is given by the full scale capacity of the transducer

(Prange) divded by the product of the CF and the excitation voltage. The offset for the transducer is found

by using the voltage output of the transducer given on the Omegadyne Inc. calibration sheet for a pressure

of zero psig and solving for x.

Calibration factor (CF): 𝐶𝐹 = 𝑠𝑒𝑛𝑠𝑖𝑡𝑖𝑣𝑖𝑡𝑦 (𝑚𝑉)

𝑒𝑥𝑐𝑖𝑡𝑎𝑡𝑖𝑜𝑛 𝑣𝑜𝑙𝑡𝑎𝑔𝑒 (𝑉) (C. 1)

𝑃 =𝑉𝑜𝑢𝑡𝑝𝑢𝑡

𝐶𝐹×

𝑃𝑟𝑎𝑛𝑔𝑒

𝑉𝑒𝑥𝑐𝑖𝑡𝑎𝑡𝑖𝑜𝑛+ 𝑥 (C. 2)

Alternatively, the pressure in psig and measured voltage in mVdc from the calibration sheets can be

plotted in Excel. The calibration factor by which the output voltage should be multiplied to convert it to

pressure can be determined by fitting a linear trend-line to the data and finding the slope. The voltage

offset is the intercept. An example of these plots for three pressure transducers currently used in the

RFDL is provided in Figure C. 1.

Figure C. 1. The plot of pressure vs. output voltage from the calibration sheets for Omega PX329 can be

used to determine the calibration factor and offset.

Both Omega PX329 and Twist Lock PX329 pressure transducers have been purchased for use in the

RFDL. The Twist Lock Type requires the user to solder the wires for output and excitation to the device

before use. A white wire is typically used for the positive output voltage, but for the pressure transducer

with serial number 020307I058, purple wire was used in place of the white wire. The pressure transducers

are wired into an NI 9923 screw-terminal connector block. The pins of the NI 9923 connect to the NI

9237 bridge analog input module, which has four channels and can sample at 50kS/s/ch. Each terminal in

the NI 9923 corresponds to a pin in the NI 9237 as shown in Figure C. 2 and Table C. 1. The analog

signal names are defined in Table C. 1. The analog input (AI) specified in the pin assignment for each PT

is the AI corresponds with that device in the LabVIEW code.

y = 6899.8x - 0.0468

R² = 1

0

100

200

300

400

0 0.02 0.04 0.06

Pre

ssure

(kP

a)

μVdc

serial # 020307I057

y = 6902.9x - 0.2858

R² = 1

0

100

200

300

400

0 0.02 0.04 0.06

Pre

ssure

(kP

a)

μVdc

serial # 020307I058

y = 6912.6x - 0.442

R² = 1

0

100

200

300

400

0 0.02 0.04 0.06

Pre

ssure

(kP

a)

μVdc

serial # 020307I059

54

Figure C. 2. Pin assignments for NI 9237 [84]

Table C. 1. Pin assignment descriptions

Signal Names Signal Descriptions

EX- Excitation current return

EX+ Positive excitation current output

AI- Negative analog input

AI+ Positive analog input

To ensure that the values that are written to the textfile with the pressure reading information are written

to the correct columns the pressure transducers for the chamber, line 1, and line 2 should be wired to

channel 0, channel 1, and channel 2, respectively. The proper terminals for this wiring configuration are

given in Table C. 2.

Table C. 2. Designated pin terminals for wiring pressure transducers to NI 9923 screw-terminal connector

block

Pressure Transducer Wire color 9932 terminal 9237 Pin assignment

Chamber

Red 2 EX0+

Black 21 EX0-

White 3 AI0+

Green 22 AI0-

Line 1

Red 6 EX1+

Black 25 EX1-

White 7 AI1+

Green 26 AI1-

Line 2

Red 12 EX2+

Black 31 EX2-

White 13 AI2+

Green 32 AI2-

55

APPENDIX D

Determination of the Discharge Coefficient

The flow coefficient (Cv) of the flow as it exits the injector orifice accounts for losses in the flow through

the orifice and is necessary to accurately predict the jet velocity (Uf), as shown in Equation 5. The

velocity is a parameter that is used to obtain the characteristic time (t+) of the experiment. The Cv is the

ratio of the discharge coefficient (Cd) and the area contraction coefficient (Ca) as shown in Equation D. 1.

𝐶𝑣 = 𝐶𝑑

𝐶𝑎 (D. 1)

The discharge coefficient of a nozzle or constriction accounts for non-ideal effects such as viscosity and

boundary layer development in the orifice [85]. The empirical discharge coefficient is determined using

the ratio of the measured (actual) flow rate to the theoretical isentropic, adiabatic mass flow rate. The area

contraction coefficient is assumed to be unity because we are using a gaseous fluid and there is no

cavitation or hydraulic flip to account for.

The discharge coefficient in [1] was calculated assuming incompressible flow. The working fluid is

always a gas in the gas jet experiments. The Mach number for all experiments with air and a pressure

drop greater than 5 psi was greater than 0.6, meaning that an incompressible approximation is inaccurate.

Therefore, it is necessary to calculate the velocity of the fluid using the compressible Bernoulli equation

given in Equation D.2. The discharge coefficient is calculated using Equation D.3, where the density of

the fluid exiting the nozzle is the ambient density because the jet exiting the nozzle will quickly approach

that density, ρ2.

𝑈𝑓 = √2𝛾

𝛾−1(

𝑃1

𝜌1−

𝑃2

𝜌2) (D. 2)

𝐶𝑑 =�̇�

𝐴𝑜𝜌2𝑈 (D. 3)

The discharge coefficients of the nozzles used in the experiments in this study were measured using a

standard method described in ISO 5167-1 [86] and several engineering textbooks [85, 87]. The

experimental setup is shown in the photos in D.2.

56

Figure D. 1. The experimental setup shown from a side view (top) and front view (bottom) included a

Coriolis flow meter which measured the flow rate in kg/s, a ball valve to control the pressure in the

apparatus, and pressure transducer at the flange tap to measure the pressure before the nozzle plate.

Air at 150 psig was sent to an Endress + Hauser Proline Promass 80A Coriolis flow meter through a 1/8”

tube. The flow of inlet air into the development length was regulated manually via a ball valve, allowing

the pressure before the nozzle plate to be controlled. Once the pressure was set, it was held at that

constant pressure for a few minutes to ensure that steady flow was exiting the nozzle when the mass flow

rate was recorded. Leakage testing conducted over the course of an hour prior to these measurements,

revealed that the leakage rate of the experiment was as low as 9.203∙10-9 m3/s. The effects of any possible

leakage were assumed to be negligible and the pressure readings did not change within the few minutes

that pressure was held constant. An Omega PX309 gage pressure transducer was located at the flange tap

shown in Figure D. 1.

The mass flow rate through the Coriolis flow meter was recorded along, the ambient temperature and

pressure were measured using a thermometer and barometer, and the pressure was measured using the

pressure transducer. These measured values were used to calculate the discharge coefficient with

incompressible flow assumptions (using Bernoulli equation for velocity) that are followed in [1] as shown

57

in Equation D. 4 and with compressible flow assumptions that used the compressible Bernoulli equation

to find velocity (Equation D. 2) as shown in Equation D. 5.

𝐶𝑑 = �̇�

𝐴𝑜∙𝜌1∙√2∙(𝑃1 − 𝑃𝑎𝑡𝑚

𝜌1)

(D. 4)

𝐶𝑑 = �̇�

𝐴𝑜∙𝜌2∙√2𝛾

𝛾−1∙(

𝑃1

𝜌1 −

𝑃2

𝜌2)

(D. 5)

The mass flow rate is �̇�, Ao is the effective orifice area, ρ2 is the ambient density, ρ1 is the density in the

tube, gamma is the heat capacity ratio for air (1.4), P1 is the ambient pressure, and P2 is the pressure in the

tube. The results with incompressible and compressible assumptions are shown in Figure D. 2 and Figure

D. 3, respectively. From the results with compressible flow assumptions, the CD of the nozzle plate

designs used in this study range from 0.8 – 0.88.

Figure D. 2. Discharge coefficient calculated with incompressible flow assumption

Figure D. 3. Discharge coefficient accounting for compressible flow

58

As stated earlier, the incompressible flow assumption is not accurate because the Mach number for all of

the flows tested was greater than 0.6. Additionally, comparing the results with experimental results in the

literature for the measurement of the discharge coefficient for small straight bore orifices (see Figure D.

4) [88] shows that the discharge coefficient calculated with compressibility considered follows a similar

trend but the discharge coefficient calculated with the incompressible assumption does not.

Figure D. 4. Discharge coefficient for small, straight bore orifices tested with air at 295 – 700 K [88]

59

APPENDIX E

Convergence Study

The number of ensembles of gas jet experiments required to obtain statistical convergence of the results

was calculated using a student’s t-distribution. The student’s t-distribution equation is given in Equation

E. 1.

𝑡 = �̅�−𝜇

𝑆 √𝑛⁄ (E. 1)

where n is the number of data points in the sample, S is the sample standard deviation, �̅� is the sample

mean, and μ is the population mean. Rearranging Equation E.1, the population mean term can be broken

into the sample mean and the confidence interval, as shown in Equation E.2.

𝜇 = �̅� ± 𝑡𝛼/2𝑆

√𝑛 (E. 2)

where for this study n is the number of ensembles, S is the sample standard deviation at each point on the

pressure curve, �̅� is the average pressure at each point, tα/2 is the critical value for a two-sided confidence

interval on the distribution, and α is the level of significance. In this convergence study α = 0.05 for a

95% confidence interval.

Forty ensembles of pressure tests at 20 kPa (gage) for cases with a slow ramp rate (one inlet valve

opening and the relief valve being closed during the entire injection duration) where the time to reach 20

kPa was 6.4 ms and a fast ramp rate (both inlet valves and the relief valve were used) where the time to

reach 20 kPa was 3.4 ms. The pressure curves for the slow ramp and fast ramp cases with outliers

removed are shown in E. 4 and E. 5, respectively.

60

Figure E. 4. The pressure curves for the slow ramp case shown with the outliers removed, leaving 36

pressure profiles. In two cases there were erroneous voltage readings from the pressure transducers

around 20 ms in to the test, which led to the spikes seen in the pressure curve.

Figure E. 5. The pressure curves for the fast ramp case.

0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.0450

5

10

15

20

25

All runs for slow ramp case

Time (s)

Ch

am

be

r P

ressu

re (

kP

a)

0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.0450

5

10

15

20

25

All runs for fast ramp case

Time (s)

Ch

am

be

r P

ressu

re (

kP

a)

61

The mean, confidence interval, and percent uncertainty were then calculated for random sets of 5, 10, 15,

20, 25, and 36 (slow ramp cases) or 40 (fast ramp cases), as shown in Table E. 1 and Table E. 2. The

results were considered to be converged when the difference in percent error between one set of

ensembles and the next highest set of ensembles was less than 0.1% for both the fast and slow ramp cases,

which occurred when 20 ensembles were used.

Table E. 1. The average confidence and uncertainty for calculated for different n (number of ensembles)

chosen at random from the set of 36 slow ramp pressure curves.

n +/- Average Confidence Uncertainty at 20 kPa

5 0.3826 1.9 %

10 0.1586 1.0 %

15 0.1185 0.7 %

20 0.0977 0.6 %

25 0.0896 0.5 %

36 0.0772 0.4 %

Table E. 2. The average confidence and uncertainty for calculated for different n (number of ensembles)

chosen at random from the set of 40 fast ramp pressure curves.

n +/- Average Confidence Uncertainty at 20kPa

5 0.2229 1.1%

10 0.1256 0.7%

15 0.1025 0.5%

20 0.0825 0.4%

25 0.0760 0.4%

40 0.0583 0.3%

62

APPENDIX F

Bowl designs for 55% BTE effort

Two bowl profiles were provided for testing by Volvo Powertrain. Two-dimensional designs of the piston

bowl are shown in Figure F.1. The notable difference between the two designs is the bowl depth; Bowl 3

is deeper than Bowl 1.

Figure F.1. Two-dimensional designs for Volvo Bowl 1 and Bowl 3

The engine piston bowl design was scaled by 1.28 times based on a scaling analysis that used the

characteristic x+ and t+ values [1] of the experiment and the engine. Three-dimensional versions of these

bowl designs were created in SolidWorks and 3-D printed at the Pennsylvania State University from PLA

Makerbot filament. A trapezoidal section that fits into a Promaster 4200 tripod was added to the edge of

each bowl for mounting the bowl to the tripod and positioning it above the injector. By extruding half of

the two-dimensional designs shown in Figure F.1, 3-D versions were created as shown in Figure F.2 for

looking at a single injection from a two-dimensional view with line-of-sight integrated flow visualization

techniques, such as schlieren.

Figure F.2. Extruded 3-D designs for Bowl 1 (left) and Bowl 3 (right)

A revolved bowl design was created in order to visualize the flow with 3-jets and 3-D wall interactions as

shown in Figure F.3.

Bowl - v1

Bowl - v3

63

Figure F.3. Revolved 3-D designs for Bowl 1 (left) and Bowl 3 (right)

64

APPENDIX G

Calculations for Piston Bowl Location

When conducting experiments involving bowl impingement, it is important to determine the location of

the piston bowl with respect to the injector nozzle and ensure that the bowl is positioned accurately.

Figure G.1 is provided for illustrative purposes to show the coordinate frames for the injector nozzle in

the engine and the experiment.

Figure G. 1. The illustration (not to scale) shows the rotation from the engine coordinate frame (x and y)

to the experiment coordinate frame (x’ and y’). The location of the red dot is found in the engine

coordinate frame and then transformed to find the distance of that point on the half-piston bowl from the

injector orifice in the experiment coordinate frame.

In the engine coordinate frame the distance the piston has traveled from TDC and the squish are given in

the y-direction. The location of the injector nozzle is at the origin in this frame of reference. The angle of

the injector orifices is defined by an umbrella angle (see Equation G.1).

𝑈𝑚𝑏𝑟𝑒𝑙𝑙𝑎 𝑎𝑛𝑔𝑙𝑒: 180° − 2𝛼 (G. 1)

The distance between the top of the cylinder and the piston, y, can be calculated using Equation G. 2 [89]

if the timing in crank angle degrees after TDC, umbrella angle, stroke, and length of the connecting rod

are given.

𝑦 = 𝑙 + 𝑎 − 𝑎2 ∗ sin2(𝐶𝐴𝐷𝑎𝑇𝐷𝐶) + 𝑎 ∗ 𝑐𝑜𝑠2(𝐶𝐴𝐷𝑎𝑇𝐷𝐶) (G. 2)

where y is the distance between the top of the cylinder and the piston, a is the length of the crank shaft

(half of the stroke), l is the length of the connecting rod, and CADATDC is the number of crank angle

degrees that the piston has moved from TDC when the jet is expected to hit the bowl wall based on the

results of CFD simulations.

In the experiment coordinate frame the origin is positioned at the center orifice where y’ is the direction in

which the jet moves downstream normal to the injector and x’ is the direction parallel to the injector

nozzle plate. The distance of the tip of the bowl from the injector in the x’- and y’-directions are

calculated using a transformation matrix to rotate the coordinates by 90°-α and multiplying by the scaling

factor (x+ratio) as shown in Equation G.3 and Equation G.4, respectively.

𝑥′ = cos(180° − 𝛼) ∗ 𝑦 ∗ 𝑥𝑟𝑎𝑡𝑖𝑜+ (G. 3)

65

𝑦′ = sin(𝛼) ∗ 𝑦 ∗ 𝑥𝑟𝑎𝑡𝑖𝑜+ (G. 4)

For the schlieren experiments with the extruded bowls and the PLIF experiments with the revolved bowls

that were recorded in the near-nozzle region the angle of the bowl with respect to the centerline of the jet,

θ, and the distance of the tip of the bowl from the center injector orifice in the x’- and y’-directions are

determined by processing side-view images of the experiment in MATLAB. The locations of various

objects in the image such as the injector orifices, the bowl tip, the back edge of the bowl, and the dots on

the calibration plate are determined by the using selecting the points using ‘ginput’.

Figure G. 5. A photograph of the side-view of the experiment is used to check the angle of the bowl and

the (x’,y’) location of the tip of the bowl with respect to the center nozzle of the injector. The yellow

diamond marker indicates the location of the center orifice and the red circle marker represents the tip of

the bowl. A calibration plate is placed in the frame so the resolution of the image can be determined.

PLIF experiments recorded in the near-bowl region required that the injector and bowl be tilted. The

angles of the edges of the bowl with respect to the nozzle plate (β1 and β2) and the distance of the tip of

the bowl from the center injector orifice in the y’-direction are determined by processing front-view

images of the experiment in MATLAB. The angles between the injector plate and edges of the cut bowl,

β1 and β2, are expected to be about 33° on each side. The β1 and β2 angles don’t need to be exact, but need

to be close to these values to ensure that the full width of the two side jets impinge on the bowl wall.

66

Figure G. 6. A photograph of the front-view of the tilted experiment for PLIF experiments for visualizing

the flow in the near-bowl region is used to check the angles of the edges of the bowl with respect to the

nozzle plate and the y’ location of the tip of the bowl with respect to the center nozzle of the injector. The

magenta diamond marker indicates the location of the center orifice. A ruler plate is placed in the frame

so the resolution of the image can be determined.

For the bowl impingement experiments discussed in Chapter 5, the engine parameters necessary to

calculate the x’- and y-locations of the tip of the extruded bowls and revolved bowls are given in Table

G.1. The angle α that is given in Table G. 1 is the angle between the bowl and the jet centerline that is

shown as θ in Figure G. 2 and should have a value close to 15° for all tests to account for the 150°

umbrella angle. The distances from the nozzle to the tip of the bowl should be -0.39 mm and 1.04 mm in

the x’ and y’ directions, respectively. The distances were matched very closely for PLIF experiments,

however, the x’ distances were not matched closely for schlieren experiments. It is important to note that

while there appear to be differences in spray development in the PLIF images due to bowl depth and

results that the results seen in schlieren experiments may be due to both a difference in bowl depth and an

approximately 1 mm difference in the bowl tip location. Because Bowl 3 was located a slightly further

distance away, it is possible that difference in distance is part of the reason so much fluid is observed

moving back towards the nozzle in the schlieren images.

Table G. 1. The engine parameters given are for the engine used in the Volvo 55% BTE effort and are

used to calculate x’ and y’.

Umbrella Angle

(degrees)

Alpha

(degrees)

Stroke

(m)

a (m) l (m) Squish

(m)

CADaTDC

(degrees)

x’ (m) y’ (m)

150 15 0.152 0.076 0.225 0.0009 4.5 -0.00039 0.000104

The measured angles and distances from the injector nozzle to the bowl tip in the x’- and y’-directions for

the various experimental setups are given in Table G. 2.

67

Table G. 2. The values for the angles and distances that describe bowl position with respect to the injector

nozzle in the experiment were measured.

Experiment

Description Bowl Design

Nozzle

Plate

Angle between bowl and

jet CL, θ (degrees) x’ (mm) y’ (mm)

Schlieren Extruded Bowl 1 Single hole 15.1 -0.0080 Not

measured

Schlieren Extruded Bowl 3 Single hole 14.9 -0.0093 Not

measured

PLIF, near-nozzle Revolved Bowl 1 3-hole, 51° 14.9 -0.0008 0.0004

PLIF, near-nozzle Revolved Bowl 1 3-hole, 45° 15.0 -0.0009 0.0005

PLIF, near-nozzle Revolved Bowl 3 3-hole, 51° 15.1 -0.0009 0.0004

PLIF, near-nozzle Revolved Bowl 3 3-hole, 45° 15.0 -0.0005 0.0005

PLIF, near-bowl Revolved Bowl 1 3-hole, 51° 15.0 -0.0011 0.0005

PLIF, near-bowl Revolved Bowl 1 3-hole, 45° 15.0 -0.0011 0.0003

PLIF, near-bowl Revolved Bowl 3 3-hole, 51° 15.0 -0.0013 0.0001

PLIF, near-bowl Revolved Bowl 3 3-hole, 45° 15.2 -0.0010 0.0004

68

REFERENCES

[1] Naber, J., and Siebers, D. L., 1996, "Effects of gas density and vaporization on penetration and

dispersion of diesel sprays," SAE technical paper.

[2] EPA, U. S., 2014, "Control of Air Pollution from Motor Vehicles: Tier 3 Motor Vehicle Emission and

Fuel Standards," Federal Register, pp. 23413 – 23886.

[3] 2009, "Regulation (EC) ", European Parliament.

[4] Singh, N., Rutland, C. J., Foster, D. E., Narayanaswamy, K., and He, Y., 2009, "Investigation into

different DPF regeneration strategies based on fuel economy using integrated system simulation," No.

0148-7191, SAE Technical Paper.

[5] 2013, "FY 2013 PROGRESS REPORT FOR ADVANCED COMBUSTION ENGINE

TECHNOLOGIES."

[6] 2016, "ADVANCED SYSTEMS LEVEL TECHNOLOGY DEVELOPMENT, INTEGRATION AND

DEMONSTRATION FOR EFFICIENT CLASS 8 TRUCKS, SUPERTRUCK II," DOE National Energy

Technology Laboratory.

[7] 2014, "Curbing the sound level of motor vehicles," E. Parliament, ed.

[8] Schwarz, A., and Janicka, J., 2009, Combustion noise, Springer Science & Business Media.

[9] Kokjohn, S., Hanson, R., Splitter, D., and Reitz, R., 2011, "Fuel reactivity controlled compression

ignition (RCCI): a pathway to controlled high-efficiency clean combustion," International Journal of

Engine Research, 12(3), pp. 209-226.

[10] Curran, S., 2014, "Fuel Economy Estimates of Dual-Fuel Low Temperature Combustion Operation

Using Vehicle System Simulations.," Penn State University, University Park, PA.

[11] Szybist, J. P., Steeper, R. R., Splitter, D., Kalaskar, V. B., Pihl, J., and Daw, C., 2014, "Negative

Valve Overlap Reforming Chemistry in Low-Oxygen Environments."

[12] Tanaka, T., Ando, A., and Ishizaka, K., 2002, "Study on pilot injection of DI diesel engine using

common-rail injection system," JSAE review, 23(3), pp. 297-302.

[13] Zhang, L., 1999, "A study of pilot injection in a DI diesel engine," No. 0148-7191, SAE Technical

Paper.

[14] Busch, S., Zha, K., Miles, P. C., Warey, A., Pesce, F., Peterson, R., and Vassallo, A., 2015,

"Experimental and Numerical Investigations of Close-Coupled Pilot Injections to Reduce Combustion

Noise in a Small-Bore Diesel Engine," SAE International Journal of Engines, 8(2015-01-0796), pp. 660-

678.

[15] Zeng, W., He, X., Jin, S., Liu, H., Li, X., and Liu, F., "Analysis of the influence of pilot injection

timing on diesel combustion in an optical engine by the two-color method," Proc. ASME 2013 Internal

Combustion Engine Division Fall Technical Conference, American Society of Mechanical Engineers, pp.

V001T003A015-V001T003A015.

[16] Bakenhus, M., and Reitz, R. D., 1999, "Two-color combustion visualization of single and split

injections in a single-cylinder heavy-duty DI diesel engine using an endoscope-based imaging system,"

SAE Technical Paper.

[17] Bower, G. R., and Foster, D. E., 1993, "The effect of split injection on fuel distribution in an engine-

fed combustion chamber," No. 0148-7191, SAE Technical Paper.

[18] Han, Z., Uludogan, A., Hampson, G. J., and Reitz, R. D., 1996, "Mechanism of soot and NOx

emission reduction using multiple-injection in a diesel engine," No. 0148-7191, SAE Technical Paper.

[19] Tow, T., Pierpont, D., and Reitz, R. D., 1994, "Reducing particulate and NOx emissions by using

multiple injections in a heavy duty DI diesel engine," SAE Technical Paper.

[20] Nehmer, D. A., and Reitz, R. D., 1994, "Measurement of the effect of injection rate and split

injections on diesel engine soot and NOx emissions," SAE Technical Paper.

69

[21] Ehleskog, R., Golovitchev, V., Denbratt, I., Andersson, S., and Rinaldini, C. A., 2006, "Experimental

and numerical investigation of split injections at low load in an HDDI diesel engine equipped with a piezo

injector," SAE Technical Paper.

[22] Herfatmanesh, M. R., Lu, P., Attar, M. A., and Zhao, H., 2013, "Experimental investigation into the

effects of two-stage injection on fuel injection quantity, combustion and emissions in a high-speed optical

common rail diesel engine," Fuel, 109, pp. 137-147.

[23] Skeen, S., Manin, J., and Pickett, L. M., 2015, "Visualization of Ignition Processes in High-Pressure

Sprays with Multiple Injections of n-Dodecane," SAE International Journal of Engines, 8(2015-01-0799),

pp. 696-715.

[24] 2015, "Engine Combustion Network," http://www.sandia.gov/ecn/.

[25] Goldwine, G., and Sher, E., 2009, "Experimental investigation of the effect of split injection on

diesel engine performance," International Journal of Vehicle Design, 50(1-4), pp. 66-90.

[26] Herfatmanesh, M. R., and Zhao, H., 2012, "Experimental investigation of effects of dwell angle on

fuel injection and diesel combustion in a high-speed optical CR diesel engine," Proceedings of the

Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering, p. 0954407012450656.

[27] O'Connor, J., and Musculus, M., 2013, "Optical Investigation of the Reduction of Unburned

Hydrocarbons Using Close-Coupled Post Injections at LTC Conditions in a Heavy-Duty Diesel Engine,"

SAE International Journal of Engines, 6(2013-01-0910), pp. 379-399.

[28] O'Connor, J., and Musculus, M., 2014, "Effect of Load on Close-Coupled Post-Injection Efficacy for

Soot Reduction in an Optical Heavy-Duty Diesel Research Engine," Journal of Engineering for Gas

Turbines and Power, 136.

[29] O'Connor, J., and Musculus, M., 2014, "In-cylinder mechanisms of soot reduction by close-coupled

post-injections as revealed by imaging of soot luminosity and planar laser-induced soot incandescence in

a heavy-duty diesel engine," SAE International Journal of Engines, 7(2014-01-1255), pp. 673-693.

[30] Hessel, R., Reitz, R. D., Musculus, M., O'Connor, J., and Flowers, D., 2014, "A CFD Study of Post

Injection Influences on Soot Formation and Oxidation under Diesel-Like Operating Conditions," SAE

International Journal of Engines, 7(2014-01-1256), pp. 694-713.

[31] Yun, H., Sun, Y., and Reitz, R. D., "An experimental and numerical investigation on the effect of

post injection strategies on combustion and emissions in the low-temperature diesel combustion regime,"

Proc. ASME 2005 Internal Combustion Engine Division Spring Technical Conference, American Society

of Mechanical Engineers, pp. 25-35.

[32] Bobba, M., Musculus, M., and Neel, W., 2010, "Effect of post injections on in-cylinder and exhaust

soot for low-temperature combustion in a heavy-duty diesel engine," SAE International Journal of

Engines, 3(2010-01-0612), pp. 496-516.

[33] Chen, S. K., 2000, "Simultaneous reduction of NOx and particulate emissions by using multiple

injections in a small diesel engine," No. 0148-7191, SAE Technical Paper.

[34] Hotta, Y., Inayoshi, M., Nakakita, K., Fujiwara, K., and Sakata, I., 2005, "Achieving lower exhaust

emissions and better performance in an HSDI diesel engine with multiple injection," No. 0148-7191, SAE

Technical Paper.

[35] Vanegas, A., Won, H., Felsch, C., Gauding, M., and Peters, N., 2008, "Experimental investigation of

the effect of multiple injections on pollutant formation in a common-rail DI diesel engine," No. 0148-

7191, SAE Technical Paper.

[36] Parrish, S. E., Zhang, G., and Zink, R. J., 2012, "Liquid and vapor envelopes of sprays from a multi-

hole fuel injector operating under closely-spaced double-injection conditions," SAE Technical Paper.

[37] Bruneaux, G., and Maligne, D., 2009, "Study of the mixing and combustion processes of consecutive

short double diesel injections," SAE international journal of engines, 2(2009-01-1352), pp. 1151-1169.

[38] Abraham, J., Khan, A., and Magi, V., 1999, "Jet-jet and jet-wall interactions of transient jets from

multi-hole injectors," No. 0148-7191, SAE Technical Paper.

[39] Lee, B., Song, J., Chang, Y., and Jeon, C., 2010, "Effect of the number of fuel injector holes on

characteristics of combustion and emissions in a diesel engine," International Journal of Automotive

Technology, 11(6), pp. 783-791.

70

[40] Pierpont, D., Montgomery, D., and Reitz, R. D., 1995, "Reducing particulate and NOx using multiple

injections and EGR in a DI diesel," No. 0148-7191, SAE Technical Paper.

[41] Genzale, C. L., Reitz, R. D., and Musculus, M. P., 2009, "Effects of spray targeting on mixture

development and emissions formation in late-injection low-temperature heavy-duty diesel combustion,"

Proceedings of the combustion institute, 32(2), pp. 2767-2774.

[42] Chartier, C., Aronsson, U., Andersson, Ö., Egnell, R., and Johansson, B., 2013, "Influence of jet–jet

interactions on the lift-off length in an optical heavy-duty DI diesel engine," Fuel, 112, pp. 311-318.

[43] Lequien, G., Berrocal, E., Gallo, Y., e Mello, A. T., Andersson, O., and Johansson, B., 2013, "Effect

of Jet-Jet Interactions on the Liquid Fuel Penetration in an Optical Heavy-Duty DI Diesel Engine," No.

0148-7191, SAE Technical Paper.

[44] O’Connor, J., 2016, "Entrainment Suppression," M. Borz, ed.

[45] Pickett, L. M., and Lopez, J. J., 2005, "Jet-wall interaction effects on diesel combustion and soot

formation," No. 0148-7191, SAE Technical Paper.

[46] Gutmark, E., Wolfshtein, M., and Wygnanski, I., 1978, "The plane turbulent impinging jet," Journal

of Fluid Mechanics, 88(04), pp. 737-756.

[47] Landreth, C. C., and Adrian, R. J., 1990, "Impingement of a low Reynolds number turbulent circular

jet onto a flat plate at normal incidence," Experiments in Fluids, 9(1), pp. 74-84.

[48] Parameswaran, V., and Alpay, S., 1974, "Normal impingement of jets," Journal of Aircraft, 11(3),

pp. 189-191.

[49] Bruneaux, G., Causse, M., and Omrane, A., 2011, "Air entrainment in diesel-like gas jet by

simultaneous flow velocity and fuel concentration measurements, comparison of free and wall impinging

jet configurations," SAE International Journal of Engines, 5(2011-01-1828), pp. 76-93.

[50] Gogineni, S., and Shih, C., 1997, "Experimental investigation of the unsteady structure of a

transitional plane wall jet," Experiments in fluids, 23(2), pp. 121-129.

[51] Espey, C., and Dec, J. E., 1993, "Diesel Engine Combustion Studies in a Newly Designed Optical-

Access Engine Using High-Speed Visualization and 2-D Laser Imaging," SAE International.

[52] Genzale, C. L., Reitz, R. D., and Musculus, M. P., 2008, "Effects of piston bowl geometry on

mixture development and late-injection low-temperature combustion in a heavy-duty diesel engine," SAE

International Journal of Engines, 1(2008-01-1330), pp. 913-937.

[53] Hesselman, K. J. E., 1925, "Hesselman Heavy-oil High-compression Engine," NASA.

[54] Hossainpour, S., and Binesh, A. R., 2009, "Investigation of fuel spray atomization in a DI heavy-

duty diesel engine and comparison of various spray breakup models," Fuel, 88(5), pp. 799-805.

[55] Reitz, R. D., and Rutland, C. J., 1995, "Development and testing of diesel engine CFD models,"

Progress in Energy and Combustion Science, 21(2), pp. 173-196.

[56] Wickman, D. D., Senecal, P. K., and Reitz, R. D., 2001, "Diesel engine combustion chamber

geometry optimization using genetic algorithms and multi-dimensional spray and combustion modeling,"

No. 0148-7191, SAE Technical Paper.

[57] De Risi, A., Manieri, D., and Laforgia, D., 1999, "A theoretical investigation on the effects of

combustion chamber geometry and engine speed on soot and NOX emissions," ASME-ICE Paper(99-

ICE), p. 207.

[58] Venkateswaran, S. P., and Nagarajan, G., 2010, "Effects of the re-entrant bowl geometry on a DI

turbocharged diesel engine performance and emissions—a CFD approach," Journal of engineering for gas

turbines and power, 132(12), p. 122803.

[59] Prasad, B. V. V. S. U., Sharma, C. S., Anand, T. N. C., and Ravikrishna, R. V., 2011, "High swirl-

inducing piston bowls in small diesel engines for emission reduction," Applied Energy, 88(7), pp. 2355-

2367.

[60] Zhu, Y., Zhao, H., Melas, D., and Ladommatos, N., 2004, "Computational study of the effects of the

re-entrant lip shape and toroidal radii of piston bowl on a HSDI diesel engine's performance and

emissions," No. 0148-7191, SAE Technical Paper.

[61] Han, D., and Mungal, M., 2001, "Direct measurement of entrainment in reacting/nonreacting

turbulent jets," Combustion and flame, 124(3), pp. 370-386.

71

[62] Tennekes, H., and Lumley, J. L., 1972, A first course in turbulence, MIT press.

[63] Ricou, F. P., and Spalding, D., 1961, "Measurements of entrainment by axisymmetrical turbulent

jets," Journal of fluid mechanics, 11(01), pp. 21-32.

[64] Abani, N., and Ghandhi, J. B., 2012, "Behavior of Unsteady Turbulent Starting Round Jets," Journal

of Fluids Engineering, 134(6), p. 061202.

[65] Musculus, M. P., and Kattke, K., 2009, "Entrainment waves in diesel jets," SAE International Journal

of Engines, 2(2009-01-1355), pp. 1170-1193.

[66] Musculus, M. P., Lachaux, T., Pickett, L. M., and Idicheria, C. A., 2007, "End-of-injection over-

mixing and unburned hydrocarbon emissions in low-temperature-combustion diesel engines," No. 0148-

7191, SAE Technical Paper.

[67] Hu, B., Musculus, M. P., and Oefelein, J. C., 2012, "The influence of large-scale structures on

entrainment in a decelerating transient turbulent jet revealed by large eddy simulation," Physics of Fluids

(1994-present), 24(4), p. 045106.

[68] Law, C. K., 2006, Combustion physics, Cambridge University Press.

[69] Knox, B. W., Genzale, C. L., Pickett, L. M., Garcia-Oliver, J. M., and Vera-Tudela, W., 2015,

"Combustion recession after end of injection in diesel sprays," SAE International Journal of Engines,

8(2015-01-0797), pp. 679-695.

[70] Ghandhi, J., 2013, "Gas jet experimental design," J. O’Connor, ed.

[71] "ASCO Product Catalog," 2/2 Series 8262 8263, ASCO, ed., pp. 37-40.

[72] Settles, G. S., 2012, Schlieren and shadowgraph techniques: visualizing phenomena in transparent

media, Springer Science & Business Media.

[73] Lozano, A., Yip, B., and Hanson, R., 1992, "Acetone: a tracer for concentration measurements in

gaseous flows by planar laser-induced fluorescence," Experiments in Fluids, 13(6), pp. 369-376.

[74] Schott, 2014, "BG3 Filter Data Sheet," Schott Optics.

[75] Skeen, S. A., Manin, J., Dalen, K., and Pickett, L. M., "Extinction-based imaging of soot processes

over a range of diesel operating conditions," Proc. 8th US National Combustion Meeting, Salt Laker City,

Utah.

[76] Pickett, L. M., Genzale, C. L., Bruneaux, G., Malbec, L.-M., Hermant, L., Christiansen, C., and

Schramm, J., 2010, "Comparison of diesel spray combustion in different high-temperature, high-pressure

facilities," SAE International Journal of Engines, 3(2010-01-2106), pp. 156-181.

[77] Benajes, J., Payri, R., Bardi, M., and Martí-Aldaraví, P., 2013, "Experimental characterization of

diesel ignition and lift-off length using a single-hole ECN injector," Applied Thermal Engineering, 58(1),

pp. 554-563.

[78] Wakuri, Y., Fujii, M., Amitani, T., and Tsuneya, R., 1960, "Studies on the penetration of fuel spray

in a diesel engine," Bulletin of JSME, 3(9), pp. 123-130.

[79] Jung, Y., Manin, J., Skeen, S., and Pickett, L. M., 2015, "Measurement of Liquid and Vapor

Penetration of Diesel Sprays with a Variation in Spreading Angle," No. 0148-7191, SAE Technical Paper.

[80] Eagle, W. E., Malbec, L.-M., and Musculus, M. P. B., 2016, "Measurements of Liquid Length,

Vapor Penetration, Ignition Delay, and Flame Lift-Off Length for the Engine Combustion Network

‘Spray B’ in a 2.34 L Heavy-Duty Optical Diesel Engine."

[81] Knox, B. W., and Genzale, C. L., 2016, "Reduced-order numerical model for transient reacting diesel

sprays with detailed kinetics," International Journal of Engine Research, 17(3), pp. 261-279.

[82] Ritchie, B., 2006, "Quantitative Acetone PLIF Measurements of Jet Mixing with Synthetic Jet

Actuators,"PhD, Georgia Institute of Techonology, Atlanta, GA.

[83] Laskin, S., 1948, "Submerged aerosol unit," AEC Project Quaterly Report UR-38, Univ. of

Rochester.

[84] 2015, "Getting Started Guide NI 9237," National Instruments, Austin, TX, pp. 1-18.

[85] Okiishi, M. Y., Munson, B., and Young, D., 2006, "Fundamentals of Fluid Mechanics," John Wiley

& Sons, Inc.

[86] ISO, 2003, "Measurement of fluid flow by means of pressure differential devices inserted in circular

cross-section conduits running full," 5167-1, ISO.

72

[87] Fox, R. W., McDonald, A. T., and Pritchard, P. J., 1985, Introduction to fluid mechanics, John Wiley

& Sons New York.

[88] Kayser, J. C., and Shambaugh, R. L., 1991, "Discharge coefficients for compressible flow through

small-diameter orifices and convergent nozzles," Chemical engineering science, 46(7), pp. 1697-1711.

[89] Ferguson, C. R., and Kirkpatrick, A. T., 2015, Internal combustion engines: applied thermosciences,

John Wiley & Sons.