University of Southern California Dissertations and Theses

The physics of pulsed streamer discharge in high pressure air and applications to engine technologies

(USC Thesis Other)

The physics of pulsed streamer discharge in high pressure air and applications to engine technologies

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content THE PHYSICS OF PULSED STREAMER DISCHARGE IN HIGH PRESSURE AIR AND
APPLICATIONS TO ENGINE TECHONOLOGIES

by

Yung-Hsu Lin

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(PHYSICS)

ii
Acknowledgements

During my graduate study, I have received support and encouragement from a great
number of individuals.
First of all, I want to thank my research advisors and mentors, Professor Martin
Gundersen and Dr. Andy Kuthi. With their guidance, support and encouragement, I learn to think
like a physicist and become a better researcher.
I would also like to thank my dissertation committee of Professor Aiichiro Nakano,
Professor Werner Dappen, Professor Chongwu Zhou and Professor William Steier for their time,
ideals and patience.
To my friends and colleagues, their company makes my PhD life colorful and joyful.
Last but not least, I want to thank my family. Thank you for listening to me and
encourage me when I felt frustrated. Without their unconditional support, I would not be able to
finish this journey.

iii
Table of Contents

Acknowledgements......................................................................................................................... ii
List of Tables ................................................................................................................................. vi
List of Figures................................................................................................................................ xi
Abstract......................................................................................................................................... xii
Chapter 1: Introduction to streamer discharges .............................................................................. 1
1.1 What is a streamer discharge?......................................................................................... 1
1.2 Applications of streamer discharges ............................................................................... 2
1.3 Fundamental streamer discharge studies ........................................................................ 3
1.4 Research related to plasma enhanced combustion.......................................................... 5
1.5 Organization of this dissertation ..................................................................................... 6
Chapter 2: Investigation of streamer propagation at high pressures............................................... 8
2.1 Introduction..................................................................................................................... 8
2.2 Basic concepts of streamer discharge ............................................................................. 9
2.2.1 Corona inception voltage ........................................................................................ 9
2.2.2 Mechanism of streamer propagation..................................................................... 10
2.2.3 The similarity law ................................................................................................. 12
2.3 Experimental setup and procedures .............................................................................. 13
2.4 Experimental results and discussion ............................................................................. 16
2.4.1 Streamer inception voltage ................................................................................... 16
2.4.2 Streamer morphology............................................................................................17
2.4.3 Streamer propagation............................................................................................19
2.4.4 Scaling factors at higher pressures........................................................................ 21
2.5 Summary....................................................................................................................... 23
Chapter 3: Streamer physics at high pressures ............................................................................. 25
3.1 Introduction................................................................................................................... 25
3.2 Experimental methods..................................................................................................27
3.2.1 Correlated streamer discharge phases with current waveform ............................. 27
3.2.2 Electrical analysis.................................................................................................29
3.3 Streamer propagation at high pressures (3-18 bar)....................................................... 31
3.3.1 Connection between optical diagnostics and electrical measurement .................. 31
iv
3.3.2 Average velocity estimated by electrical measurement........................................ 32
3.3.3 Average velocity as a function of E/P (the similarity law)................................... 36
3.4 Discussion of the invalidity of similarity law at high pressures ................................... 36
3.5 Dimensional analysis as applied to the similarity law.................................................. 42
3.6 Summary....................................................................................................................... 44
Chapter 4: Application of streamer discharge to diesel engine technology.................................. 46
4.1 Introduction................................................................................................................... 46
4.1.1 Benefits of streamer discharge for ICE combustion............................................. 46
4.1.2 Mechanism of the streamer discharge in combustion application........................ 47
4.2 Electrode design............................................................................................................ 48
4.2.1 Basic concept of electrode design......................................................................... 48
4.2.2 Electrode design for medium size 3-cylinder diesel engine ................................. 49
4.2.3 Impact of electrode configuration on temperature................................................ 51
4.3 Streamer discharge technique applied to small diesel engine....................................... 54
4.3.1 Specification of Kubota diesel engine for research at USC.................................. 54
4.3.2 Engine instrumentation.........................................................................................54
4.3.3 Baseline measurement..........................................................................................58
4.3.4 Experimental setup................................................................................................60
4.3.5 Results................................................................................................................... 63
4.3.6 Further experiments..............................................................................................68
4.4 Conclusion.................................................................................................................... 72
Chapter 5: Application of streamer discharge on diesel exhausts treatment ................................ 73
5.1 Introduction................................................................................................................... 73
5.2 Basic mechanism of NO removal and operation principle of gas analyzer.................. 75
5.2.1 Mechanism of NO removal by pulsed corona discharge...................................... 75
5.2.2 Operation principle of gas analyzer ...................................................................... 76
5.3 Streamer discharge application on diesel exhaust treatment ........................................ 79
5.3.1 Exhaust treatment outside the engine ................................................................... 79
5.3.2 Exhaust treatment in the engine cylinder.............................................................. 81
5.4 Discussion..................................................................................................................... 83
5.5 Calculation of NO removal efficiency and the effect of pulse width & repetition rate 84
Chapter 6: Conclusion and future work........................................................................................ 90
6.1 Conclusion.................................................................................................................... 90
6.1.1 Nanosecond pulsed streamer propagation at high pressures................................. 90
6.1.2 Application of transient plasma technique to diesel engines................................ 91
6.2 Future work................................................................................................................... 92
v
6.2.1 Measurement of radicals generated by streamer discharge at high pressures ...... 92
6.2.2 Other characteristics of pulsed streamer discharge at high pressures................... 92
6.2.3 Transient plasma effects on the stability of low-temperature combustion ........... 93
6.2.4 Pulse generator and electrode design for scaling up to marine diesel .................. 93

References......................................................................................................................................97

vi
List of Figures

Figure 2.1 Illustration of streamer propagation mechanism. (a) Cathode-directed streamer
propagation with secondary avalanches moving towards the positive head of the streamer. (b)
Anode-directed streamer propagation with secondary avalanches in front of the streamer head
[36]................................................................................................................................................ 11
Figure 2.2 (a) Optical diagnostic setup for observing streamer propagation. (b) Voltage and
current waveform of 12 ns pulse generator into 200 ohm load. The original current waveform is
shifted due to measurement setup................................................................................................. 15
Figure 2.3 Time dependence of the position of streamer heads at 2 bar. The data points
represented the position of streamer heads measured from images at different time frames. ...... 16
Figure 2.4 Corona inception voltages at various pressures. Two data sets were taken in different
days which showed slightly deviation due to background environment. ..................................... 18
Figure 2.5 Streamer morphology. The gauge reading showed on the left upper corner of each
image and the gate width was in the lower left corner. The gate width at 1.01 bar is different than
others because the intensity is too high......................................................................................... 19
Figure 2.6 Sequential images of streamer propagation at 2 bar. The bright areas are the trail of
streamer head during the exposure time. The time on the left lower corner shows the delay from
peak voltage and the jitter of time and voltage is 0.1 ns and 0.7 kV. ........................................... 20
Figure 2.7 (a) Velocity versus applied voltage at 1.01 & 2.04 bar (b) Streamer velocity versus
pressure at 40 kV applied voltage................................................................................................. 20
Figure 2.8 (a) Streamer velocity plotted as a function of reduced electric field E/P. The pressures
of navy dots from right to left are 1.01, 1.36, 1.70, 2.05, 2.39, 2.74 and 3.08 bar. (b) Streamer
velocity analyzed as a function of E/ √P at P=1-3 bar................................................................... 22
Figure 3.1 Experimental setup. Camera is synchronized with pulse generator and the delay time
is controlled by software. All signals are monitored by oscilloscope........................................... 28
vii
Figure 3.2 Voltage and current waveform of pulse generator into 200 ohm load. The rise time is
about 5 ns and pulse width is 12 ns. The maximum output voltage is around 46 kV. ................. 29
Figure 3.3 (a) Electrical analysis setup. (b) Voltage and current waveforms of the 65 ns
pseudospark pulse generator into the coaxial electrode with 8 mm gap in the high pressure
chamber. (c) A waveform example to explain the average velocity estimation........................... 30
Figure 3.4 Current and gate waveforms were recorded at 6 bar with 5.75 mm gap and 30 kV. . 31
Figure 3.5 Different streamer discharge phases at 6 bar: (a) streamer initiation (b) streamer
propagation (c) formation of conducting channel (d) arcing formation. The image quality was not
good due to low light intensity at high pressures.......................................................................... 32
Figure 3.6 Average velocity as a function of applied voltage and pressure (2-8 bar). The velocity
increases with voltage but decreases with pressure. ..................................................................... 34
Figure 3.7 The relationship of streamer velocity (6-18 bar) with external voltage (a) and pressure
(b). Circled data points in (a) are off due to the mismatch of gap size and pulse duration. ......... 35
Figure 3.8 The average streamer velocity as defined by the arc formation time deviated from
linear scaling with the reduced electric field E/P for pressures (a) P = 2~8 bar. (b) P = 6~18 .... 39
Figure 3.9 Average velocity as a function of E/ √P for pressures (a) P = 2~8 bar. (b) P = 6~18 bar.
There was a deviation between P=2 bar and 3-8 bar. It showed that the average streamer velocity
fitted better with E/ √P................................................................................................................... 40
Figure 3.10 A schematic view of density profile across shock wave. ......................................... 41
Figure 3.11 Re-analyzed data from literature [24-25] which are consistent with presented work.
....................................................................................................................................................... 44
Figure 4.1 It showed the rate constants of autoignition reactions and electron impact dissociation
versus gas temperature [27]. The great difference indicated that ignition with TPI is more
efficient than traditional ignition. ................................................................................................. 48
Figure 4.2 An example drawing of electrode installed in an engine............................................ 49
Figure 4.3 Electrode design for NPS diesel engine. .................................................................... 50
viii
Figure 4.4 The electrode before (top) and after (bottom) experiment. After experiments, soot is
deposited on the insulator which might cause arcing along the insulator..................................... 51
Figure 4.5 (1) Low heat range (hot) spark plug. (2) High heat range (cold) spark plug.[].......... 51
Figure 4.6 The second version of electrode which is a modified spark plug. The inserted
insulator makes the heat transfer slower....................................................................................... 52
Figure 4.7 Electrodes after experiments. The soot problem is successfully solved..................... 53
Figure 4.8 The latest version of electrode design. ....................................................................... 53
Figure 4.9 Kubota engine instrumentation................................................................................... 56
Figure 4.10 (a) Pressure sensor is mounted on the intake port bolt. (b) The channel is drilled
through the bolt and connected to another channel drilled from the combustion chamber.......... 57
Figure 4.11 Instrumented Kubota engine..................................................................................... 57
Figure 4.12 Pressure signal in combustion chamber.................................................................... 59
Figure 4.13 Partial pressure signals in a run. The peak pressure changes a lot even without TPI
applied........................................................................................................................................... 59
Figure 4.14 The relationship between engine speed (rpm) and peak pressure. ........................... 60
Figure 4.15 Schematic diagram of how high voltage sources are triggered and recorded (a); how
pressure signals are marked with TTL signals (b)........................................................................ 61
Figure 4.16 Optimization of sample rate, buffer size and number of samples to read. ............... 62
Figure 4.17 Peak pressures versus normalization. The peak pressure with TPI is 1040.846 +/-
25.173 psi; without TPI1 is 1025.28 +/- 21.896 psi; without TPI2 is 1027.717 +/- 25.842 psi. .. 64
Figure 4.18 Results of Student’s T test........................................................................................ 64
Figure 4.19 Peak pressure change versus TPI angle. The x-axis error bar (trigger angle jitter)
isn’t shown here but it is less than 0.5 degree (calculated by hand)............................................. 65
Figure 4.20 Peak pressure with time and PV diagram in both control and TPI conditions. The
relationship between the crank angle and the volume:
ix
221/2
11
[1 cos ( sin ) ]
12
d
V
RR
Vr
θθ =+ +− − −
−
, where V
d
(325 cm
3
) is the displacement volume, r
(23) is the compression value, R is the connecting length over crank radius. After obtaining the
volume, PV diagram can be plotted.............................................................................................. 67
Figure 4.21 (a) View of modified engine head; (b) electrodes. ................................................... 69
Figure 4.22 New baseline of Kubota engine when no load is applied......................................... 69
Figure 4.23 New experimental setup. The trigger source is changed to rpm signal to avoid noise
issue............................................................................................................................................... 70
Figure 4.24 (a) 1 kHz burst mode (b) 2 kHz burst mode............................................................. 71
Figure 4.25 (a) Extended cathode (b) Extended cathode and electrode are installed on the engine
head............................................................................................................................................... 72
Figure 5.1 Typical configuration of electrochemical sensor used for gas analyzer []. ................ 77
Figure 5.2 Schematic view of a non-dispersive infrared sensor. ................................................. 78
Figure 5.3 The exhaust treatment procedures. ............................................................................. 79
Figure 5.4 The electrode used for exhaust treatment. .................................................................. 80
Figure 5.5 Plasma treatment reduces NO concentration in the exhaust gas stream by a factor 2.
....................................................................................................................................................... 80
Figure 5.6 Exhaust treatment procedure. ..................................................................................... 81
Figure 5.7 NO concentration varied with time in (a) rich condition (1500 rpm and full load) and
(b) lean condition (780 rpm and no load). C means control, P means TP treatment and the
numbers are the average of NO concentration in a stable range................................................... 82
Figure 5.8 Exhaust treatments at 1600 rpm with full load. (a) With 85 ns pseudospark pulse
generator, there is a 50% NO reduction at 0.1 Hz, 90% at 0.2 Hz and 100% removal at repetition
rate higher than 0.5 Hz; (b) with 12 ns solid state pulse generator, there is a 50% NO decrease at
0.5 Hz and 80% at 1 Hz. ............................................................................................................... 86
x
Figure 5.9 Comparing the energy cost at the condition that the similar percentage of NO is
removed. It showed that the shorter pulse width has better energy efficiency............................. 87
Figure 5.10 Emission treatments with different repetition rate. The higher repetition rate
generates more radicals per unit time and results in more NO reduction..................................... 88
Figure 5.11 Emission treatments with different repetition rate. The higher repetition rate
generates more radicals per unit time and results in more NO reduction..................................... 89

xi
List of Tables

Table 3.1 Experimental parameters in streamer propagation measurements............................... 27
Table 4.1 Engine specifications. .................................................................................................. 54
Table 4.2 Peak pressure measurements with different parameters. From left to right are: number
of pulsed cycle, average peak pressure of pulsed cycles, standard deviation; number of non-
pulsed cycle, average peak pressure of non-pulsed cycles and standard deviation. ..................... 71
Table 5.1 Sensors used to detect different gases in the exhaust gas analyzer and their resolutions.
....................................................................................................................................................... 76
Table 5.2 Calculated results of percentage of NO reduction, number of removed NO molecules
and energy cost per molecule. Exhaust treated by transient plasma produced with (a) the solid
state (12 ns, 40 kV) (b) the pseudospark (85 ns, 60 kV) pulse generator..................................... 88
Table 6.1 Engine specifics and scaling factors of Kubota and Sulzer engine.............................. 95
Table 6.2 Parameters of pulse generator used on Kubota and needed for Sulzer........................ 95
Table 6.3 Estimation of energy needed to remove 50% of NO in the Sulzer engine. ................. 96
xii
Abstract

The goal of this dissertation is to study high pressure streamers in air and apply it to
diesel engine technologies. Nanosecond scale pulsed high voltage discharges in air/fuel mixtures
can generate radicals which in turn have been shown to improve combustion efficiency in
gasoline fueled internal combustion engines. We are exploring the possibility to extend such
transient plasma generation and expected radical species generation to the range of pressures
encountered in compression-ignition (diesel) engines having compression ratios of ~20:1,
thereby improving lean burning efficiency and extending the range of lean combustion.
At the beginning of this dissertation, research into streamer discharges is reviewed. Then,
we conducted experiments of streamer propagation at high pressures, calculated the streamer
velocity based on both optical and electrical measurements, and the similarity law was checked
by analyzing the streamer velocity as a function of the reduced electric field, E/P. Our results
showed that the similarity law is invalid, and an empirical scaling factor, E/ √P, is obtained and
verified by dimensional analysis. The equation derived from the dimensional analysis will be
beneficial to proper electrode and pulse generator design for transient plasma assisted internal
engine experiments.
xiii
Along with the high pressure study, we applied such technique on diesel engine to
improve the fuel efficiency and exhaust treatment. We observed a small effect of transient
plasma on peak pressure, which implied that transient plasma has the capability to improve the
fuel consumption. In addition, the NO can be reduced effectively by the same technique and the
energy cost is 30 eV per NO molecule.

1
Chapter 1: Introduction to streamer
discharges

1.1 What is a streamer discharge?
The streamer discharge is a gas ionization process which usually starts with electron
avalanches. Electron avalanches occur as following. While free electrons are in an electric field,
they are accelerated and gain kinetic energy. If the electrons gain enough energy for ionization
before collisions, they can produce more electrons and excited species by colliding with atoms or
molecules in gas and ionizing them. This was first discovered by John Townsend [1]. While the
electrons move forward, the positive ions are left behind and a thin conductive channel is formed,
which is a streamer. When a streamer bridges the gap between the anode and the cathode, the
channel becomes highly conductive and transforms into an arc.
Streamer discharge, also known as transient plasma and non-thermal (non-equilibrium)
plasma, is characterized by its non-thermal equilibrium property. The majority of the electrical
energy goes into the generation of energetic electrons rather than transferring into heat energy.
Thus, the surrounding gas temperature stays close to room temperature while the electron
temperature is much higher.
2
1.2 Applications of streamer discharges
Transient plasma technology has drawn much attention recently because of low energy
cost and its ability to generate active species. It has been successfully demonstrated in many
applications. Below are some of them [2,3,4,5,6,7]:
a) Assisted combustion and ignition: Excited species generated by streamer discharge create a
path with a lower temperature threshold and accelerate chain reactions. It has been
demonstrated that this technology is capable of reducing the ignition delay, lowering the
ignition temperature of combustion, improving energy-efficiency and increasing flame speed
and providing homogeneous and stable combustion.
b) Biomedical: Transient plasma technology is also attractive to biomedical engineers because
of its non-equilibrium and reactive properties. The energetic electrons produce reactive
radicals while the gas temperature is close to room temperature. Thus, transient plasma does
not cause any thermal damage to heat sensitive biological systems. Due to these advantages,
transient plasma technology has been extensively used for sterilization, cancer cell treatments,
blood coagulation, wound healing and dental treatments.
c) Agriculture: Studies have shown that plasma pre-treatment stimulates seed germination and
suppresses pathogenic microorganisms due to the active species of oxygen atoms and
molecules.
3
d) Exhaust treatment: Active species generated by transient plasma interact with nitrogen oxides
(NO
x
), particulate matter and volatile organic compounds and decompose harmful gases. In
particular, pulsed corona discharges and dielectric barrier discharges combined with selective
catalytic reduction are the most popular techniques for the removal of pollutant molecules.
e) Ozone generation: Streamer discharge generates atomic oxygen which recombines with
oxygen molecule then forms ozone. The ozone can be used for water cleaning, equipment
disinfecting and food processing.

1.3 Fundamental streamer discharge studies
Due to its great potential impact on a wide range of applications, many numerical
modeling and experimental studies on streamer discharge at atmospheric pressure have been
reported. Most of the simulation works were based on the hydrodynamic diffusion-drift model.
By using this model, streamer dynamics and structures at atmospheric pressure have been well
established. Along with this theoretical research, many experiments have been carried out on
positive streamers in air as well. Following are some essential topics.
The mechanism of positive streamer propagation in air, which is commonly accepted to
be photo-ionization, has been studied. Since the emitted photons and the absorption length of
4
photo-ionization were related to the concentration of oxygen and nitrogen, streamers in different
nitrogen-oxygen mixtures have been investigated. Yi and William [8] conducted their
experiments with both pure N
2
and N
2
/O
2
mixtures in a plane to plane electrode configuration.
Their results suggested that photoionization played an important role because streamers
propagate faster at higher oxygen concentrations. Nijdam [9] conducted experiments in different
gas mixtures while Wormeester [10] did simulations. They saw some difference with different
gas composition but the velocity is about the same. They studied background ionization as well
by changing the repetition rate and found that the background ionization had an influence on
streamer morphology and initiation [11]. However, it had little impact on the velocity and
minimum diameter of a streamer.
Briels et al [12] experimentally investigated both positive and negative streamers in a
point to plane electrode configuration at atmospheric pressure. They found that the positive
streamers were easier to form than the negative ones. The positive streamers were observed at 5
kV while no negative ones were emerged at voltages below 20 kV. With the same voltage
amplitude, the positive streamers propagate faster and its diameter was about 10% thicker than
the negative streamers. Also, Briels derived an empirical fit, v = 0.5d
2
mm
-1
ns
-1
, for the velocity
and diameter of both positive and negative streamers. At the same time, Luque A et al [13]
modeled streamers of both polarities in in two geometries, plane to plane and needle to plane.
5
Their results were compatible with Briels’ work. The velocity of positive streamers was fitted
well by the same empirical equation, v = 0.5d
2
, without any fit parameter. The positive streamers
were thinner than the negative ones and its minimal diameter was about 0.2 mm, which was the
same as those reported in experiments [12,14,15]. They proposed that the slower velocity of the
negative streamers was due to the broaden streamer head by the electron drift motion and the
resultant decrease of field enhancement leads to a slower propagation than the positive streamers.

1.4 Research related to plasma enhanced combustion
In recent decades, plasma assisted ignition and combustion have received much attention
due to its potential to improve the performance of combustion in engines. Following are several
examples. In pulse detonation engines (PDE), ignition delay is reduced significantly by using
plasma assisted ignition, and this allows the engines to operate at higher repetition rate to achieve
enough thrust [16,17]. In internal combustion engines, a factor of 2 to 3 shorter ignition delay
[18] has been found. Besides, faster flame propagation and 20% peak pressure increase [19] has
been verified as well. Moreover, the lean burn limit is extended by using plasma assisted ignition
as well as knocking reduction [20,21]. In turbine engines, increasing efficiency and emission
reduction have been found as well as better flame stabilization [22,23].
6
Along with developing plasma technology to practical applications, scientists broadly
study the underlying mechanisms. Kinetic models have been built as well as experimental works
have been done. Here are some important contributions. Bozhenkov et al use a shock tube to
investigate plasma combustion kinetics. They found the ignition delay was reduced while the
temperature wasn’t increased [24]. Ombrello and his colleagues developed a lifted-flame
experiment to investigate the effects of ozone and excited oxygen on flame propagation [25,26].
Advanced diagnostic techniques have developed to investigate the active species generated
during streamer discharges such as O, H and OH [27,28,29].

1.5 Organization of this dissertation
The next two chapters are studies of streamer propagation at high pressure conditions.
We investigate streamer propagation at high pressures (1-3 bar) by optical diagnostics and
characterize streamers by inception voltage, morphology and velocity. In addition, we calculate
the average velocity of primary and secondary streamer (3-18 bar) by electrical measurements
and analyze it with the reduced electric field, E/P. An empirical scaling factor, E/ √P, is obtained
and verified by dimensional analysis. The equation derived from dimensional analysis is
beneficial to proper electrode and pulse generator design for transient plasma assisted internal
7
engine experiments. In the following two chapters, works of streamer discharge application on
diesel engine are presented. We observe a small effect of transient plasma on peak pressure, and
exhaust treatments in different places have been done. Last chapter concludes our work and
some suggestions for future extending works.

8
Chapter 2: Investigation of streamer
propagation at high pressures

2.1 Introduction
Understanding the dynamics of nanosecond scale pulse streamer discharges in air at
multi-atmospheric pressures is essential for the development of transient plasma enhanced
combustion in internal combustion engines. The main reason is that the electrode used to
generate transient plasma is usually installed inside the engine where the pressure can raise to 20
bar or higher. Although there are many studies on streamer discharge at atmospheric pressure,
very few such studies have been done at high pressures (above atmospheric pressure).
The first experimental study which had been found is S Achat et al’s work. They
compared the discharge properties in the pressure range of 1-8 bar [30]. After that, N. Yu.
Babaeva and G.V. Naidis developed a model for dense media and found the importance of
electron losses due to three-body recombination [31] while P Tardiveau et al believed that the
increase of heat caused the invalidity of similarity law at high pressures by investigating
discharge development and analyzing the pressure effect on streamer discharge [32-33]. Limited
9
research has been done so far due to the challenge of plasma generation at high pressures. For
example, in order to generate plasma at high pressures, it requires a higher voltage.
This chapter begins with the basic concepts of streamer discharge. Then, it follows by the
descriptions of the experimental setup, results and discussions. Finally, it ends with the summary
of the pressure effects on streamer.

2.2 Basic concepts of streamer discharge
2.2.1 Corona inception voltage
Electron avalanches are needed for streamer initiation. In order to have electron
avalanches, electrons need to gain enough energy for ionization impact. Equation (3) showed
that the energy gained by an electron in a uniform electric field and it is proportional to the mean
free path and external field.
ε = e l E (3)
, where e is electric charge, l is mean free path and E is external electric field.
The inception voltage is depended on gas composition, density, electrode material and
electrode configuration. Peek presented an empirical formula for corona inception as a function
of pressure and temperature:
10

, where r
0
is the inner radius and δ is the relative air density which is given as:

, where P is pressure in torr and T is temperature in ℃ [34].

2.2.2 Mechanism of streamer propagation
In an external electric field, free electrons are accelerated and collided with other
molecules. If the electric field is high enough, then electrons gain enough kinetic energy to
ionize gas molecules by direct collision and produce more free electrons and positive ions. This
process will go on and the number of free electrons increases. The increase of electrons and ions
is called electron avalanche. As more and more electrons are produced and the internal electric
field of avalanche is comparable with the external field, then a streamer initiates.
Equation (1) described that the electric field of the avalanche needs to be close to the
order of the external field. From Raether’s observation, he found that when the electron density
Ne is around 3x10
8
, the internal electric field of avalanche reached the same magnitude of the
external field and leaded to a streamer initiation.
0
0 2
0
exp[ ( )* ]
4
a
A
E e
ExE
rp
α
πε
=≈ (1)
11
8
exp( ) 3 10
e
Nd α =≈⋅ , 20 d α ≈ (2)
Equation (2), the criterion of streamer formation is known as the Raether-Meek criterion [35].

Figure 2.1 Illustration of streamer propagation mechanism. (a) Cathode-directed streamer
propagation with secondary avalanches moving towards the positive head of the streamer. (b)
Anode-directed streamer propagation with secondary avalanches in front of the streamer head
[36].

Figure 2.1 illustrated the mechanism of streamer propagation. After streamer initiation,
photons emitted from primary avalanche caused photoionization in the vicinity and then initiate
secondary avalanches. The positive charges at the streamer head pulled the electrons of
secondary avalanches, formed a quasi-neutral channel and new positive space charges are left in
front of the previous ones. This process will go on and leads to streamer propagation.
(a) (b)
12
There are three main physical mechanisms for streamer propagation: direct ionization by
electron impact, photo-ionization process and background ionization.
Negative streamers propagate in the direction of electron drift but positive streamers propagate
against the direction of electron drift. In most literatures, the electron source for positive
streamers is assumed to be photo-ionization. It was proposed by Flegler and Raether, Bradley
and Snoddy, and Cravath [37,38,39]. In air, the excited nitrogen molecules N
2
(C
3
Π
u
) emitted a
UV photon and in a wavelength of 980-1025A the emission of N
2
(C
3
Π
u
) is mainly absorbed by
oxygen and ionized an oxygen molecule in the gas. Because of difficulty, photo-ionization has
been directly measured only in few works [40-41]. These results shown that photo-ionization had
greater effects in air than in pure gas. Since photo-ionization can still be seen in pure gases,
background ionization has been proposed to be an alternative electron source [41]. It could be
natural background (ex. Radioactivity and cosmic radiation) or charges left from previous
discharge. However, there is no direct measurement of background ionization.

2.2.3 The similarity law
Similarity laws for streamer discharges with the same gas composition but with different
densities were first formulated by Townsend for the so-called Townsend scaling [42,43,44]. The
basic length scale in streamer discharges is mean free path (l), which is inversely proportional to
13
the gas density (n) because in streamer discharges, most of the important processes such as
impact ionization are two-body processes, where an electrode collides with a neutral molecule.
Streamers possess the same characteristics while electrons gain the same energy between
collisions. Therefore, the electric field (E) needs to scale with the gas density to keep similar
discharge.
Based on above, it implied that the electron energy and the velocity should be
independent to gas density while the time should be inversely proportional to the gas density.
That is to say streamer discharges in different densities but with the same gas are similar if
lengths, times, electric fields are properly scaled. Therefore, the similarity law is a useful tool
which allows people to estimate properties of streamer discharge at different pressures, which are
not possible in experimental studies, by known values at low pressure [45].

2.3 Experimental setup and procedures
In this chapter, streamer propagation in the pressure range of 1 to 3 bar was investigated by
optical diagnostics (Figure 2.2(a)). Streamers were generated inside a stainless steel cylindrical
high pressure chamber which is designed for pressure up to 20 bar. In order to make
14
experimental conditions even more consistent, the chamber was evacuated before experiments
and backfilled with synthetic dry air. Electrodes were installed inside the chamber and
cylindrical electrode configuration was specially chosen over other configurations such as point
to plane because it is mostly used on practical systems. A 0.5 mm tungsten wire was used as an
anode and stainless steel tubes with various diameters (5.75-11.75 mm) were served as cathode
and form a radial gap. What connects to the electrode is a pulse generator which delivers a pulse
width of 12 ns FWHM, and its maximum amplitude is 43 kV. Figure 2.2(b) shows the voltage
and current waveforms of the pulse generator into 200 ohm load. A high speed ICCD camera, PI-
MAX 3, was used to take the pictures of light emitted from streamer discharges. The camera gate
width was fixed at 3ns, and it was synchronized with the pulse generator. By changing the delay
time between the gate and the peak of voltage, a series of streamer discharge images (every
image is a different discharge) were taken beginning with streamer initiation until it reached the
cathode in steps of 1 ns. Using these images, we measured the position of streamer head and
plotted as a function of time. Figure 2.3 is an example of streamer head position to time at 2 bar
and the velocity of streamer head was calculated from the slopes of position to time figures.
Same procedures were repeated at different pressures.
15

Figure 2.2 (a) Optical diagnostic setup for observing streamer propagation. (b) Voltage and
current waveform of 12 ns pulse generator into 200 ohm load. The original current waveform is
shifted due to measurement setup.
High pressure chamber
Oscilloscope
Gate signal
V & I
ICCD Camera
Pulse generator
Image storage
Trigger signal
-5
0
5
10
15
20
25
30
35
40
45
100 150 200 250 300
Time (ns)
Voltage (kV)
-25
0
25
50
75
100
125
150
Current (A)
Voltage
Current
16
y = 0.9747x - 4.6936
R
2
= 0.9956
0
2
4
6
8
10
5 7 9 11 13 15
time (ns)
Position (mm)
2 bar

Figure 2.3 Time dependence of the position of streamer heads at 2 bar. The data points
represented the position of streamer heads measured from images at different time frames.

2.4 Experimental results and discussion
Pressure effects on streamer discharge are discussed in following aspects: inception
voltage, morphology, velocity and the scaling factor.

2.4.1 Streamer inception voltage
Inception voltage is measured in the pressure range of 9 to 22 bar with cylindrical
configuration, a 0.5 mm tungsten wire as anode and 5.75 mm gap size. Our measurements show
that the inception voltage is linearly increased with pressure (Figure 2.4). It is due to decrease of
17
the mean free path which is inversely proportional to gas density. The electron energy is
proportional to the external electric field and the mean free path, thus higher voltage is needed
for inception when pressure increases.

2.4.2 Streamer morphology
Figure 2.5 shows different structures of streamer at different pressures. Though in
different electrode configuration, our results are similar to [46]. As the pressure increases,
streamers get thinner and more branches. The branching mechanism hasn’t yet been
understood completely but currently there are three models to explain branching phenomena
[47,48,49].
1. Stochastic behavior of secondary avalanches [50]
If the electric field around the streamer head is high enough then there must be a chance for
two avalanches to be developed simultaneously from the streamer head corresponding to
two individual streamer branches.
2. Laplacian instability [51,52,53]
Laplacian growth, a mathematical model, is used to explain why branching occurs at the
boundary between two substances such as mineral structures and cracks. Researchers in
18
Netherlands have shown that the streamer branching is the same kind of process of those
branching pattern of nature.
3. Photoionization [54]
Experimental results from Briels et al. [15] have shown that in nitrogen-oxygen mixtures,
more branching occurs at lower oxygen concentration. They attributed this to the
differences in photo-ionization.

y = 1.1882x + 11.123
R
2
= 0.9804
y = 1.055x + 10.451
R
2
= 0.9723
0
10
20
30
40
0 5 10 15 20 25
Pres s ure(bar)
Ele c tric fie ld (kV/m m )

Figure 2.4 Corona inception voltages at various pressures. Two data sets were taken in different
days which showed slightly deviation due to background environment.

19

Figure 2.5 Streamer morphology. The gauge reading showed on the left upper corner of each
image and the gate width was in the lower left corner. The gate width at 1.01 bar is different than
others because the intensity is too high.

2.4.3 Streamer propagation

Figure 2.6 is a set of images that shows streamer propagation at 2 bar. All images are single shot
per exposure with 3 ns gate width. There is a one second waiting time between shot to shot to
reduce the possibility of background ionization. At each time frame, at least 10 images are
selected so that the accuracy is within 0.1 ns and voltage jitter is about 0.7 kV. The position of
streamer head is measured and plotted as a function of time. Then, the velocity is determined as
slopes of position traces (in the section of steady state propagation). In Figure 2.7(a), the
streamer velocity at atmospheric pressure is 0.95-1.5 mm/ns, which is linearly dependant on
external electric field and is close to [46]; Figure 2.7(b), the streamer velocity decreases with the
increase of pressure because the mean free path decreases thus electrons obtain less energy.

3 ns 10 ns 10 ns
10 ns
2.39 bar 3.08 bar 1.01 bar 1.70 bar
20

Figure 2.6 Sequential images of streamer propagation at 2 bar. The bright areas are the trail of
streamer head during the exposure time. The time on the left lower corner shows the delay from
peak voltage and the jitter of time and voltage is 0.1 ns and 0.7 kV.

Figure 2.7 (a) Velocity versus applied voltage at 1.01 & 2.04 bar (b) Streamer velocity versus
pressure at 40 kV applied voltage.
3mm
16 ns
15 ns 14 ns
13 ns
12 ns 11 ns 10 ns
9 ns
8 ns 7ns 6 ns 5 ns
y = 0.0266x - 0.0825
R
2
= 0.9865
y = 0.04x - 0.161
R
2
= 0.9932
0
0.5
1
1.5
2
20 25 30 35 40 45
Voltage (kV)
velocity (mm/ns)
1.01 bar
2.04 bar
y = 1.5321x
-0.5824
R
2
= 0.9853
0
0.5
1
1.5
2
01 2 3 4
Pressure (bar)
velocity (m m /ns)
21
2.4.4 Scaling factors at higher pressures
It is well known that streamers have similar properties while having the same scaling factor. In
other words, they should have same velocity (or related by linear transformation) at the same
value of E/P if the similarity law is valid. However, from our results, it is shown that the
similarity law is violated, even at low pressure ~ 3 bar. In Figure 2.8(a), streamer velocity is
plotted as a function of E/P. The navy dots are measured at a fixed electric field (fixed voltage
and fixed gap size), about 41 kV/mm, with various pressures while the pink dots and blue dots
are measured at fixed pressures, 1.01 bar and 2.05 bar, with different electric field (varied
voltage but fixed gap size). If the similarity law held, these data sets should have same slopes but
they didn’t. In addition, the slope is larger at higher pressures, which indicates that the deviation
from similarity law is larger with the increase of pressure. Furthermore, comparing data at the
same E/P value, the velocity at higher pressure is faster than lower pressure, which means that
the velocity at high pressure is faster than the prediction of similarity law.
From our empirical fitting, the streamer velocity fits well with (E/P)· √P as shown in
Figure 2.8(b). So the streamer velocity increases with pressure faster than simply the E/P scaling.
22
y = 0.0272x + 0.4462
R
2
= 0.9926
y = 0.0391x - 0.161
R
2
= 0.9932
y = 0.0528x - 0.0825
R
2
= 0.9865
0
0.5
1
1.5
2
010 20 30 40 50
E/P (kV*mm
-1
* bar
-1
)
Velocity (mm/ns)
E fixed
P fixed (1.01 bar)
P fixed (2.05 bar)

y = 0.0394x - 0.1428
R
2
= 0.9807
0.5
1
1.5
2
0 10 203040 50
E/P
0.5
(kV/mm*bar
-0.5
)
Velocity (mm/ns)
E fixed
P fixed (1.01 bar)
P fixed (2.05 bar)

Figure 2.8 (a) Streamer velocity plotted as a function of reduced electric field E/P. The pressures
of navy dots from right to left are 1.01, 1.36, 1.70, 2.05, 2.39, 2.74 and 3.08 bar. (b) Streamer
velocity analyzed as a function of E/ √P at P=1-3 bar.
23
2.5 Summary
The increase of pressure means higher gas density and which results in shorter mean free
path and higher collision frequency. Our investigation of streamer discharge at elevating
pressures is summarized below:
Higher inception voltage
Since the mean free path becomes shorter, the electric field needs to increase. Therefore, higher
streamer inception voltage is expected at higher pressures.
Slower streamer velocity
Primary streamer velocity is a function of external electric field and pressure. The velocity
increases when the external electric field increases and decreases with increasing pressure. Again,
because of shorter mean free path, the kinetic energy gained by the electron is smaller.
More branching structure
A clear mechanism for this hasn’t yet been shown in literatures. However, it possibly due to
more collisions occurred and higher chance to have secondary avalanches around streamer head.
Violation of similarity law
From our result, it indicates that the similarity law is violated even at atmospheric pressure and
the deviation from similarity law increases with pressures.
24
To further understand the behavior of streamer at high pressure, we conduct the measurement of
average velocity of primary and secondary streamer in an extending pressure range from 3 to 18
bar in the next chapter.

25
Chapter 3: Streamer physics at high
pressures

3.1 Introduction
Although the similarity law (similar discharges can be found at different pressures while
the scaling factors are constant) can be used to predict streamer properties at high pressures
based on known properties, recent works [13,14,55,56] showed that the similarity law is not
always true at high pressures. There are three hypotheses for the invalidity of the similarity law
and was indirectly proved by either theoretical model or experimental data. The first hypothesis
is electron-ion recombination. N. Yu. Babaeva et al [31] built an analytical model which
included electron-ion recombination processes. In their numerical data, the loss of electron was
negligible at 1 bar but a significant decrease of electron density was observed at 10 bar. They
attributed it to three-body recombination, which dominated at high densities. Secondly, the
leader-like hot channel is formed at high pressures. P Tardiveau et al [32] observed two
discharges (at 1 bar and 6 bar) under similar conditions and found the discharge current was
much smaller than what the similarity law predicted. They explained it with thermal effect that,
at high pressures, heat was restricted in streamer channels and accordingly increased the
26
temperature. As a result, the density decreased, and E/N increased. Hence, the similarity law is
violated at high pressures. The last is photoionization. The modeling results in V P Pasko et al’s
work [57] demonstrated that the quenching of singlet excited states of molecular nitrogen that
emits photoionizing radiation is the main reason for the non-similar behavior at pressures above
30 Torr. Although their results were consistent with recent experimental works [60,58] at
different pressures, the pressure ranges of both experiments and model were not even higher than
1 bar.
In previous chapter, our experimental results showed that the similarity law was violated
at 1-3 bar. To further understand streamer physics at high pressures, it is necessary to extend the
experimental study up to 20 bar. However, investigating streamer propagation above 3 bar with
optical method is difficult because of weak light emission at high pressures. Therefore, we
complemented optical data obtained from PI-MAX 3 ICCD camera with electrical measurements
to deduce average streamer velocities, and combining with previous results, an empirical scaling
factor was derived in a pressure range of 1-18 bar.
27
3.2 Experimental methods
Table 3.1 summarized important experimental parameters used in both optical (Chapter 2)
and electrical (Chapter 3) measurements. Although different gap sizes were used in different
pressure ranges, the pressure range overlapped one another.

Table 3.1 Experimental parameters in streamer propagation measurements.

3.2.1 Correlated streamer discharge phases with current waveform
In order to measure the average velocity of primary and secondary streamers, the
correlation between streamer discharge phases with discharge current is necessary. Figure 3.1
shows a schematic experimental setup. A coaxial electrode system with tungsten wire (0.5 mm
diameter) as an anode and various gap sizes, 1.75 – 3 mm, were placed in a stainless steel
cylindrical chamber. The chamber was filled with synthetic dry air. What connects to the
electrode is a pulse generator which delivers a pulse width of 12 ns FWHM and its maximum
amplitude is 43 kV (Figure 3.2).
A high speed ICCD camera (PI-MAX 3) having a gate width of 3 ns was used to follow
the evolution of streamer discharge in air. The gate was synchronized with discharge current
Method OpticalElectricalElectrical
Pressure (bar) 1-3 1-8 6-18
Pulse width (ns) 12 85 12
Gap size (mm) 11.75 7.93 1.75
28
measured by a Pearson current probe and the delay was controlled by camera software. The
velocity of primary streamers above 3 bar was difficult to measure using optical diagnostics
alone due to the weak and decreasing light emission with increasing pressure. Hence, our
investigation of streamer propagation in air at high pressures was augmented by electrical
analysis.

Figure 3.1 Experimental setup. Camera is synchronized with pulse generator and the delay time
is controlled by software. All signals are monitored by oscilloscope.

High pressure chamber
Oscilloscope
Gate signal
V & I
ICCD Camera
Pulse generator
Image storage
Trigger signal
29
-100
-50
0
50
100
150
200
250
0 100 200 300 400 500
time (ns)
Current (I)
-10
0
10
20
30
40
50
Voltage (kV)
Current
Voltage

Figure 3.2 Voltage and current waveform of pulse generator into 200 ohm load. The rise time is
about 5 ns and pulse width is 12 ns. The maximum output voltage is around 46 kV.

3.2.2 Electrical analysis
Due to low light levels at high pressures, measuring the velocity of primary streamers
using optical diagnostics is difficult. An alternative method was developed to estimate the
average velocity of primary and secondary streamers. Assuming arcing is formed when the
secondary streamer reached anode after a conductive channel is formed while primary streamer
reached cathode, the average velocity of primary and secondary streamers can be estimated by
the interval time between the rise and collapse of voltage. In this part, two pulse generators were
30
used to generate streamers: (1) Pseudospark pulse generator with 65 ns pulse width and (2) Solid
state pulse generator with 12ns pulse width. Electrode configuration was cylindrical and to match
with different pulse generator, different anode and gap size were used: (1) 8/32 rod with 8 mm
gap size and (2) 0.5 mm tungsten wire with 1.75 mm gap size. Voltage and current waveforms
were shown in Figure 3.3. Voltage collapse in the right hand figure marked the end of the
streamer propagation phase. The average velocity was calculated by streamer propagation
distance (twice gap size) over the delay time to arc current rise, assuming that the secondary
streamer propagated at the same speed as the primary streamer.

Figure 3.3 (a) Voltage and current waveforms of the 65 ns pseudospark pulse generator into the
coaxial electrode with 8 mm gap in the high pressure chamber. (b) A waveform example to
explain the average velocity estimation.
-20
-15
-10
-5
0
5
10
15
20
25
-3.0E-07 -1.0E-07 1.0E-07 3.0E-07 5.0E-07 7.0E-07
-4000
-3000
-2000
-1000
0
1000
2000
3000
4000
5000
Current
Voltage
∆ t
-300 -100 100 300 500 700
Voltage (kV)
Current (10 A)
Time (ns)
-20
-15
-10
-5
0
5
10
15
20
25
-3.0E-07 -1.0E-07 1.0E-07 3.0E-07 5.0E-07 7.0E-07
-4000
-3000
-2000
-1000
0
1000
2000
3000
4000
5000
Current
Voltage
∆ t
-300 -100 100 300 500 700
-20
-15
-10
-5
0
5
10
15
20
25
-3.0E-07 -1.0E-07 1.0E-07 3.0E-07 5.0E-07 7.0E-07
-4000
-3000
-2000
-1000
0
1000
2000
3000
4000
5000
Current
Voltage
∆ t
-20
-15
-10
-5
0
5
10
15
20
25
-3.0E-07 -1.0E-07 1.0E-07 3.0E-07 5.0E-07 7.0E-07
-4000
-3000
-2000
-1000
0
1000
2000
3000
4000
5000
Current
Voltage
∆ t
-20
-15
-10
-5
0
5
10
15
20
25
-3.0E-07 -1.0E-07 1.0E-07 3.0E-07 5.0E-07 7.0E-07
-4000
-3000
-2000
-1000
0
1000
2000
3000
4000
5000
Current
Voltage
∆ t
-20
-15
-10
-5
0
5
10
15
20
25
-3.0E-07 -1.0E-07 1.0E-07 3.0E-07 5.0E-07 7.0E-07
-4000
-3000
-2000
-1000
0
1000
2000
3000
4000
5000
Current
Voltage
∆ t ∆ t
-300 -100 100 300 500 700
Voltage (kV)
Current (10 A)
Time (ns)
-6
-4
-2
0
2
4
6
8
10
12
-5.E-07 0.E+00 5.E-07 1.E-06
-3000
-2000
-1000
0
1000
2000
3000
4000
Current
Voltage
Current (10 A)
Voltage (kV)
-6
-4
-2
0
2
4
6
8
10
12
-5.E-07 0.E+00 5.E-07 1.E-06
-3000
-2000
-1000
0
1000
2000
3000
4000
Current
Voltage
Current (10 A)
Voltage (kV)
(a) (b)
31
3.3 Streamer propagation at high pressures (3-18 bar)
3.3.1 Connection between optical diagnostics and electrical measurement
The correlation between discharge phases and discharge current were shown in Figure
3.4 and 3.5. The waveform shown in Figure 3.4 was a discharge at 6 bar with 5.75 mm gap size
and the applied voltage was 30 kV. The images in Figure 3.5 were taken with 3 ns gate width
and each image was from different discharge events. The discharge phases from (a) to (d) were
streamer initiation, streamer propagation, formation of conducting channel and arcing formation.

Figure 3.4 Current and gate waveforms were recorded at 6 bar with 5.75 mm gap and 30 kV.
-100
-50
0
50
100
150
200
0 25 50 75 100 125 150 175 200 225 250 275
time (ns)
Current (amp)
0
2
4
6
8
10
Voltage (V)
Current
gate signal
(a)
(b)
(c)
(d)
32

Figure 3.5 Different streamer discharge phases at 6 bar: (a) streamer initiation (b) streamer
propagation (c) formation of conducting channel (d) arcing formation. The image quality was not
good due to low light intensity at high pressures.

Each phase had been observed for 3-5 times. Streamer initiation and arc formation, (a)
and (d), had precise timing and were easy to identify but streamer propagation and the
conducting channel formation had more jitter because of the weak light emission and the fast
drop of discharge current which made position (c) sometimes difficult to identify.

3.3.2 Average velocity estimated by electrical measurement
A. Streamer propagation in the pressure range of 2-8 bar
Figure 3.6 showed the average velocity of primary and secondary streamers as a function
of voltage (fixed gap) and pressure. The average velocity was about 1/3 of the optically
measured streamer velocity because the velocity of secondary streamer was slower than the
(a)
(b)
(c) (d)
1mm
33
primary streamer velocity and the electric field in two measurements were different due to
different anodes (0.5 mm tungsten wire in optical diagnostic but 8/32 threaded rod here).
B. Streamer propagation in the pressure range of 6-18 bar
Figure 3.7 showed the relation connecting average streamer velocity to applied voltage
and pressure. In Figure 3.7 (a), the velocity at higher pressures with low voltages (circled data
points) was insensitive to the parameters due to the voltage pulse duration being shorter than the
streamer propagation time across the gap, extinguishing the electric field and removing the
energy source for the streamer. In Figure 3.7 (b), there was a break at 14.27 bar in the 44 kV
dataset. It may have been due to generation of an alternate discharge path along the insulator.
34
0.05
0.10
0.15
0.20
0.25
0.30
01 2345 678
Pressure (bar)
velocity (mm/ns)
21kV
19kV
17kV
15kV
13kV
0.05
0.10
0.15
0.20
0.25
0.30
0 1020 30 40 5060
Voltage (kV)
velocity (mm/ns)
6.18 bar
5.15 bar
4.12 bar
3.08 bar
2.05 bar

Figure 3.6 Average velocity as a function of applied voltage and pressure (2-8 bar). The velocity
increases with voltage but decreases with pressure.

35

Figure 3.7 The relationship of streamer velocity (6-18 bar) with external voltage (a) and pressure
(b). Circled data points in (a) are off due to the mismatch of gap size and pulse duration.
0.1
0.15
0.2
0.25
0 5 10 15 20
Pressure (bar)
velocity (mm/ns)
30 kV
34 kV
38 kV
42 kV
36 kV
40 kV
44 kV
0.1
0.15
0.2
0.25
25 30 35 40 45 50 55
Voltage (kV)
velocity (mm/ns)
6.10 bar
7.12 bar
8.14 bar
9.16 bar
10.18 bar
11.20 bar
13.25 bar
14.27 bar
15.29 bar
16.31 bar
17.80 bar
12.22 bar

36
3.3.3 Average velocity as a function of E/P (the similarity law)
The average streamer velocity was analyzed as a function of the reduced electric field,
E/P (Figure 3.8) and as a function of a modified reduced field, E/ √P (Figure 3.9). Our data
showed that the similarity law based on the presumed E/P scaling was violated (Figure 3.8a) and
the deviation from the standard similarity law increased with pressure. From our empirical fitting,
both the streamer velocity and the average velocity fit well with (E/P)· √P as shown in Figure 3.9.
Thereby, the streamer velocity increases with pressure faster than simply the E/P scaling.
Although data at the lower end of pressures were slightly off from the best fit line (2 bar in
Figure 3.8a; 6.1 and 7.12 bar in Figure 3.8b), this may be due to the limitation of experimental
setup (gap size and the sensitivity of current probe).

3.4 Discussion of the invalidity of similarity law at high pressures
As mentioned in the beginning of this chapter, there are three hypotheses for the
invalidity of similarity law: electron loss, thermal effect and photoionization. Let’s first consider
how these hypotheses can affect streamer velocity. If electron loss through three body
recombination was solid, then the streamer velocity at high pressures should be lower than the
prediction of regular similarity law. With regard to thermal effect, since the temperature
37
increased due to bad thermal conductivity, the density decreased and resulted in a higher velocity
than regular simply E/N. Same as photoionization, it provided extra electrons ahead of streamer
tip and thereby increased the velocity. Our data showed streamers propagate faster than the
prediction of similarity law. Thus, our results consist with thermal effect and photoionization,
and details were discussed below.
Photo-ionization is usually used to explain the propagation of positive streamers in air.
Indeed, photoionization may provide extra electrons ahead of streamer head and cause streamers
propagate faster than the similarity law predicted. However, from the photoionization model Eq.
(1) used in most literature, the quenching pressure, P
q
, is about 30 ~ 80 mbar, which indicates
that the quenching effect makes the photoionization negligible at high pressures.
3 1
11 2
1
() 1
()
4
q
ion
photo
q V
p
Sr
Sdr rrp
pp
rr
π
=Ψ−
+
−
∫
(1)
Hence, thermal effect might dominate. Tardiveau et al. [32] argued that the thermal conduction
was less effective as pressure increased and the increase of temperature inside the streamer
channel thereby resulted in the decrease of local density. However, the increase of equilibrium
temperature inside the channel was not sufficient for the observed phenomena. The local density
in front of the streamer head must decrease as well if the propagation is to be affected. Pressure
38
equilibration leading to the decrease in density propagates with approximately the sound speed,
which is much smaller than the streamer propagation speed.
As an estimate, we start from the heat equation Eq. (2).

2
2
TT
D
tx
∂ ∂
=
∂ ∂

where D is the thermal diffusivity and T is temperature as a function of time and position.
Approximating Eq. (2) in one dimension by assuming T is an exponential function, gives Eq. (3)
and (4). This shows that temperature varies by ∆T over a distance δ in a time scale of τ.
1 TT
D
τ δδ
∆∆ ⎛⎞
≈
⎜⎟
⎝⎠

D δ τ ∼

Using the thermodynamic properties of air at 10 bar, thermal diffusivity, D = λ/( ρ·C
p
), is
about 4.85 x 10
-3
mm
2
/ns, where λ is thermal conductivity (100 kW/mK), ρ is air density (17.3
kg/m
3
) and C
p
is specific heat capacity (1.189kJ/kg K) [59]. In nanosecond time scale, δ is about
0.07 mm, which is three times smaller than the propagation distance of streamer by comparing to
the measured velocity at 10 bar. This indicated that the measured streamer velocity is faster than
both thermal diffusion and sound speed in the unperturbed gas in front of the streamer. Thus
simple gas heating and resultant density decrease is unlikely to be the full explanation.
(2)
(3)
(4)
39
0.00
0.05
0.10
0.15
0.20
0.25
0.30
01 2345 67
E/P (kV*mm
-1
*bar
-1
)
velocity (mm/ns)
2.05 bar
3.08 bar
4.12 bar
5.15 bar
6.18 bar
7.22 bar
8.25 bar

0
0.05
0.1
0.15
0.2
0.25
0 0.2 0.4 0.6 0.8 1
E/P (kV*mm
-1
*bar
-1
)
velocity (mm/ns)
6.10 bar
7.12 bar
8.14 bar
9.16 bar
10.18 bar
11.20 bar
12.22 bar
13.25 bar
14.27 bar
15.29 bar
16.31 bar
17.80 bar

Figure 3.8 The average streamer velocity as defined by the arc formation time deviated from
linear scaling with the reduced electric field E/P for pressures (a) P = 2~8 bar. (b) P = 6~18
40
0.10
0.15
0.20
0.25
0.30
02 46 8 10
E/P
0.5
(kV*mm
-1
*bar
-0.5
)
velocity (mm/ns)
4.12 ~ 8.25 bar
3.08 bar
2.05 bar

y = 0.1617x - 0.0766
R
2
= 0.9447
0.05
0.1
0.15
0.2
0.25
0.5 1 1.5 2 2.5
E/P
0.5
(kV*mm
-1
*bar
-0.5
)
velocity (mm/ns)
6.10 bar
7.12 bar
8.14 bar
9.16 bar
10.18 bar
11.20 bar
12.22 bar
13.25 bar
14.27 bar
15.29 bar
16.31 bar
17.80 bar

Figure 3.9 Average velocity as a function of E/ √P for pressures (a) P = 2~8 bar. (b) P = 6~18 bar.
There was a deviation between P=2 bar and 3-8 bar. It showed that the average streamer velocity
fitted better with E/ √P.
41
The formation of a supersonic shock wave is another possibility we are exploring in order
to explain the observed phenomena. A shock wave might be formed ahead of the streamer
(Figure 3.10). If the initial shock outruns the streamer head the density depression behind the
shock front but ahead of the streamer head could cause faster streamer propagation compatible
with an increase in E/P seen by the streamer head.

Figure 3.10 A schematic view of density profile across shock wave.
ρ
x
Shock front
Density
hole
42
3.5 Dimensional analysis as applied to the similarity law
Dimensional analysis is a method that enables us to derive relationships between the
physical quantities as well as to understand or characterize a phenomenon. Here, we introduced
dimensional analysis to explain our experimental data.
In the experiment, time for streamers travel from anode to cathode is measured and
output voltage, gap size and gas density (pressure) are known parameters. Thus, time can be
written as a function of voltage, permittivity, length and density.
12 3 4
[] [ ] [ ] [] [ ]
kk k k
tV l ε ρ =
Substituting units for these parameters, a dimensional matrix is shown in the following:

Variable time VoltagePermittivityLength density
Dimension (t) (V) ( ε) (l) ( ρ)
L 0 2 -3 1 -3
M 0 1 -1 0 1
T 1 -3 4 0 0
A 0 -1 2 0 0

Then, four equations are derived from the matrix and k1 ~ k4 can be solved:
21 3 2 3 3 4 0
12 4 0
31 4 2 1
12 2 0
kk k k
kk k
kk
kk
− +− =
−+ =
−+ =
−+ =

k1 = -1, k2 = -1/2, k3 = 2, k4 = 1/2.
The travel time and the average velocity cross the gap are therefore:
43
2
l
t
V
lV
v
t l
ρ
ε
ε
ρ
=
==

This is consistent with our data since V/l is electric field and pressure is proportional to ρ.
Moreover, our empirical scaling factor is consistent with early studies by Briels [60] and
Pancheshnyi [61]. We re-interpolated their data as a function of E/ √P (Figure 3.11). The blue
dots, which was taken from Pancheshnyi’s work [61, Figure 7a], were streamer velocity
measured at atmospheric pressure with various voltages; the pink data, which originated from
Briels [60, Figure 8], were numerical results with fixed electric field and were in the pressure
range of 300-700 torr. The slopes of these data sets are similar to presented work which shows
that (E/P)· √P might be a better scaling factor for streamer velocity. In addition, these results
imply that something is missing in the well-known similarity law since the data include not only
high pressures but also low pressures.
44
y = 0.039x - 1.7338
y = 0.0353x - 0.9733
0
1
2
3
4
0 40 80 120 160
E/P
0.5
(kV/mm*bar
-0.5
)
Velocity (mm/ns)
Ebert 2008 ( 1 bar, PM) Experiment, Positive
Pancheshnyi 2005 (E fixed) Simulation, Negative
Presented work

Figure 3.11 Re-analyzed data from literature [24-25] which are consistent with presented work.
These relations are used to design proper high voltage pulse amplitudes and pulse widths for the
high pressure full scale marine diesel experiments.

3.6 Summary
In this chapter, the average streamer velocity in the pressure range of 2 -18 bar has been
investigated. The velocity increased with the external electric field and decreased with elevating
pressure. The similarity law was checked as well by analyzing streamer velocity with the scaling
45
factor, E/P. The presented data demonstrated that the similarity law was violated at high
pressures and the deviation increased with higher pressure.
Combining with optical measurement, an empirical scaling factor, E/ √P, was derived in
the pressure range of 1-18 bar. The result will be beneficial for electrode and pulse generator
design in Transient Plasma Assisted Internal Combustion experiments.
Thermal diffusion and the resultant density decrease is not a likely explanation of the
observed scaling. A shock wave formed in the initial stage and outrunning the streamer head may
lead to the decrease of local density in front of the streamer head and enable the streamer to
propagate with higher velocity than the prediction of similarity law. Also, our empirical scaling
factor agreed with the dimensional analysis, which showed that the average velocity across the
gap is a function of 1/ √ρ.
46
Chapter 4: Application of streamer discharge
to diesel engine technology

4.1 Introduction
4.1.1 Benefits of streamer discharge for ICE combustion
Streamer discharge in combustion applications has been studied extensively because
comparing to traditional spark ignition, plasma assisted ignition and combustion provide many
advantages. First, streamer discharges generate many discharge paths, which results in multi-site
ignition and volumetric effect while spark discharge only have a single path [62]. In addition,
due to the thermal ignition, more energy is wasted in heat loss. Hence, streamer discharge has
lower energy consumption than spark discharge. Meanwhile, since the excited species are
generated by the high electron energy, streamer discharge has been shown to reduce the ignition
delay, lower the ignition temperature, decrease the pressure rise time and increase the peak
pressure [63,64]. Last but not least, it extends the stable operating regimes of engines by
improving the lean burn capability, flame stability and flame velocity [65].
47
4.1.2 Mechanism of the streamer discharge in combustion application
The main mechanism of transient plasma technology is generating active species such as
H, O and OH to speed up chemical reactions and results in a shorter ignition delay and lower
ignition temperature. Following is an example (hydrogen-oxygen stoichiometric mixture) [66].
(R1) is a reaction of chain initiation of autoignition. (R2) - (R4) are reactions of chain
propagation. These reactions are fast and don’t lead to recombination of radicals.
H2 + O2 -> 2OH + 78kJ/mole (R1) Chain initiation
H + O2 -> OH + O + 70kJ/mole (R2) Chain branching
O + H2 -> OH + O + 8kJ/mole (R3) Chain branching
OH + H2 -> H2O + H -62kJ/mole (R4) Chain development
Take TPI effect into consideration. At a reduced electric field of E/N=100-300Td, the
dissociation rate will be within the range 10
-10
– 10
-8
cm
3
/s, which is two orders larger than the
reaction rate constant of (R1).
48

Figure 4.1 It showed the rate constants of autoignition reactions and electron impact dissociation
versus gas temperature [66]. The great difference indicated that ignition with TPI is more
efficient than traditional ignition.

H2 + e -> H + H + e (R5)
O2 + e -> O + O + e (R6)

4.2 Electrode design
4.2.1 Basic concept of electrode design
Figure 4.2 is a drawing of an electrode installed in an engine. L1 is the discharge gap; L2
- L4 are the possible paths for arcing from anode to cathode; L3 is the thickness of insulator. The
49
basic concept for electrode design is to have electric discharge at the expecting position without
any electrical breakdown along/across the insulator. Thus, L2 and L4 should be larger than L1,
and L3, is determined by the dielectric strength of insulator.

Figure 4.2 An example drawing of electrode installed in an engine.

4.2.2 Electrode design for medium size 3-cylinder diesel engine
Figure 4.3 is the first version of electrode. In high pressure conditions, a good insulator is
important to prevent electrical breakdown across the insulator. The main idea here is to use
compressed air as the insulator. The design of the metal shell (cathode) is based on the glow plug,
and the front part is porous so that the radicals can mix with fuels easily. A straightened tungsten
wire is used as anode and it slides into a thin alumina tubing except the first 5 mm from tip. Then,
L1
L2
L3
L4
Engine head (Ground cathode)
Anode
Cathode
Insulator
50
the tubing is assembled with a macor insulator by applying silicone paste in between to keep the
alumina from ejecting during combustion.

Figure 4.3 Electrode design for NPS diesel engine.

Figure 4.4 shows the electrode before and after experiments. Electrode sooting might
result in electrical loading of the pulse generator. Therefore, we try to improve the design so that
the electrodes can self-clean inside the engine.

Alumina tubing
Tungsten wire glued with alumina
Porous cathode to
enhance air fuels mixing.
Macor insulator
Silicone paste applied
between macor insulator
O-ring added to reduce strains.
51

Figure 4.4 The electrode before (top) and after (bottom) experiment. After experiments, soot is
deposited on the insulator which might cause arcing along the insulator.

4.2.3 Impact of electrode configuration on temperature
In the engine, an igniter absorbs heat produced from combustion, the heat is transferred
through the centre electrode and insulator nose to the metal shell, which then transfers the heat
into the engine casing and coolant. Igniters have long insulator noses will heat up easily since the
heat release path from the insulator to the housing is long. Thus, heat dispersal is low, and the
temperature of the center electrode rises easily. On the contrary, igniters with short insulator
noses have less surface area exposed to the hot combustion gases. It thereby will dissipate heat
faster results in a lower temperature of the center electrode (Figure 4.5).

Figure 4.5 (1) Low heat range (hot) spark plug. (2) High heat range (cold) spark plug[67].
52
Thus, low power engines which do not produce a large amount of heat use a low heat
range (or hot) spark plug. Then the spark plug will heat up easily and reach its optimal operating
temperature, vice versa. If the igniter is not designed properly, it will cause carbon deposits (a
cold spark plug) or electrode melting (a hot spark plug). In our case, there is soot deposit, which
means that the heat transfers too fast. Thus, a modified design with longer insulator might be
needed.
However, the length of insulator is limited to the machining technique and it is difficult to
drill a hole with such a small diameter through a long ceramic rod. Therefore, we modified a
non-resistance spark plug. The discharge arm is taken off, and the anode is extended by welding
a stainless steel rod. Figure 4.6 is the drawing and Figure 4.7 showed the electrode after the
experiments.

Figure 4.6 The second version of electrode which is a modified spark plug. The inserted
insulator makes the heat transfer slower.
Spark plug
Ceramic
Insulator
Adapter
1 mm Tungsten
wire
53

Figure 4.7 Electrodes after experiments. The soot problem is successfully solved.

An optimal ratio of anode radius to cathode radius is calculated as well. With ratio, 1/3,
the electric field between channel and anode is minimized and less transient plasma is generated.
Thus, arcing and electrical breakdown might be eliminated. A piece of ceramic is used to keep
electrode centered. A sharp disc is added to the tip to enhance the electric field (Figure 4.8).

Figure 4.8 The latest version of electrode design.
Spark plug
Ceramic Insulator
Adapter
2 mm Tungsten wire
with small disc at the tip
54
4.3 Streamer discharge technique applied to small diesel engine
4.3.1 Specification of Kubota diesel engine for research at USC
One cylinder Kubota diesel engine is used for our engine study and the reasons for
choosing this engine are because of the possibility and ease of head modification to instrument
and install TPI electrode system.

Parameter Setting
Engine type 4-Stroke, single cylinder
Bore x stroke 77 x 70.0 mm
Compression ratio 23
Fuel injecting timing 24 deg before top dead center(BTDC)
Table 4.1 Engine specifications.

4.3.2 Engine instrumentation
Combustion chamber pressure is monitored by a fiber optic isolated pressure sensor
(Optrand D32294-Q) mounted on the intake port bolt closest to the combustion chamber and
connected to the space between the piston and the head through a 1.5mm diameter channel as
shown in Figure 4.9 & 4.10. The signal has been calibrated by comparison with a second
identical sensor, temporarily mounted in the glow-plug thread, without the narrow channel.
55
Piston position is sensed by an optical sensor which is mounted on the engine block
above the flywheel. A reflecting tape marked top dead center on the flywheel. When the sensor
observed the marking, it converted the resulting position signal into a 5V negative going pulse.
The signal is also used for instantaneous rpm.
Angular position of shaft is detected by an incremental encoder (NorthStar SL56) which
is mounted on the shaft. The z-index signal, which is aligned with top dead center (TDC), is used
to reset the encoder counter value at TDC to zero, simplifying data analysis. A low-pass filter is
placed at the Z signal input to the data acquisition system to eliminate the noise caused by pulse
generator. The encoder signals are converted to angles and recorded by NI-DAQ. It will be used
to calculate the PV diagram and engine heat release.
Other measurements such as engine exhaust analysis includes measurement of the levels
of CO
2
, CO, O
2
, HC, and NO
x
from the exhaust and monitor the temperature of air intake, engine
head and exhaust with thermocouples.
Data acquisition and control subsystem based on the NI USB-6216 data recorder, a TC-
08 8-channel Thermocouple USB recorder, and a computer under LabVIEW software collects all
instrument signals into a single control and measurement system with synchronized recording of
piston position, cylinder pressure, instantaneous rpm and all relevant temperatures.
56
The engine load is a belt driven 120Vac generator, with a maximum of 1.8 kW
continuous output. The output power is monitored by recording the generator output voltage and
current with the Labview data acquisition system. The generator is then connected to the final
engine load, an electric water heater. Heated water is recycled into a sink, replaced continuously
with fresh tap water. This arrangement avoids heating the ambient air and will keep the
temperature drift of the intake air to a minimum.

Figure 4.9 Kubota engine instrumentation.

57

Figure 4.10 (a) Pressure sensor is mounted on the intake port bolt. (b) The channel is drilled
through the bolt and connected to another channel drilled from the combustion chamber.

Figure 4.11 Instrumented Kubota engine.
Belt-driven generator
Optical sensor Exhaust analysis
Thermocouple
Thermocouple
TPI electrode
Magnetic encoder
58
4.3.3 Baseline measurement
Figure 4.12 showed a typical pressure signal. Fuel injecting timing is 24 degree before
top dead center (BTDC) and the pressure at that moment is around 280 psi. The first peak of
pressure signal, ~755 psi, means that it’s top dead center while the second peak of pressure
signal, ~855 psi, indicates a complete burn at around 11.4 degree after top dead center (ATDC).
By triggering pressure signal on oscilloscope, an output signal is sent out to trigger the
pulse generator and by adjusting triggering level, TPI can be applied at different angles (Figure
14). TPI is applied at different timing and the effect of TPI should be determined by the peak
pressure difference although a small change is expected.
The relationship between peak pressure and rpm is shown in Figure 4.14. During
experiments, engine speed (rpm) fluctuates (+/- 50) because it is a single cylinder engine.
Variation of peak pressure with engine speed is measured and the minimum at 1800 rpm. This
speed is used for the following experiments.

59

Figure 4.12 Pressure signal in combustion chamber.

Figure 4.13 Partial pressure signals in a run. The peak pressure changes a lot even without TPI
applied.

0
200
400
600
800
1000
1200
5000 5100 5200 5300 5400 5500 5600 5700 5800 5900 6000
time(ms)
Pressure(psi)
60

Figure 4.14 The relationship between engine speed (rpm) and peak pressure.

Figure 4.13 shows part of the pressure signals in one run. The peak pressure variation is
quite large from cycle to cycle. Therefore, we apply TPI on every other cycle and sort data points
into two groups, TPI and non-TPI (Figure 4.15). Then, statistical analysis is used.

4.3.4 Experimental setup
Pressure and rpm signals are monitored on oscilloscope and the oscilloscope is triggered
by pressure signal. Whenever the oscilloscope is triggered, a TTL signal is sent out from
oscilloscope to NI-DAQ and triggers NI-DAQ for sending a TTL signal, TTL1 which is
800
850
900
950
1000
1050
1100
1500 1600 1700 1800 1900 2000 2100
rpm
Pressure(psi)
NO LOAD
W/ LOAD
61
modified to cover two cycles for statistical analysis. Meanwhile, TTL1 is fed to NI-DAQ to
trigger another TTL signal, TTL2, to trigger HV pulse generator.

Figure 4.15 Schematic diagram of how high voltage sources are triggered and recorded (a); how
pressure signals are marked with TTL signals (b).

Pulsed Non-Pulsed
62

Figure 4.16 Optimization of sample rate, buffer size and number of samples to read.

1
2
3
63
Figure 4.16 is the labview program used for collecting data and generating TTL signal
to trigger pulse generator. There are three parts in this program: analog input (pressure, rpm and
pulse timing), digital input (encoder signal) and digital output (TTL signal). To record all the
data simultaneously, analog and digital inputs are shared the same sample clock which is based
on encoder signal (A). The parameter of sample rate, buffer size and number of samples read
need to be optimized as well to record data correctly.

4.3.5 Results
Peak pressures are plotted as a function of normalization shown in Figure 4.17. Straight
lines mean that the distributions of data samples are Gaussian (normal) distribution although the
values are fluctuated with time. The mean peak pressure is 1.5% higher in TPI than non-TPI.
Since this difference is very small, we applied student’s t-test.
Statistical Student’s t-test (Figure 4.18) implies that 1.5% is significant, with 99.9%
confidence.
64
960
980
1000
1020
1040
1060
1080
1100
1120
-3 -2 -1 0 1 2 3
Normalize
Pressure (psi)
TPI
No-TPI1
No TPI2

Figure 4.17 Peak pressures versus normalization. The peak pressure with TPI is 1040.846 +/-
25.173 psi; without TPI1 is 1025.28 +/- 21.896 psi; without TPI2 is 1027.717 +/- 25.842 psi.

Figure 4.18 Results of Student’s T test.
No-TPI1 No-TPI2 TPI No-TPI
TPI pulse at 22deg BTDC
Fuel inject at 24deg BTDC
65
The data table and Figure 4.19, summarizes a new data set after the engine re-adjustment
in order to replicate and extend previous results. This data set includes measurements of the
engine with load and without load. The engine speed was 1750 rpm, as determined from the
speed sensitivity of the peak pressure. A single pulse with ~43 kV amplitude was applied to the
electrode at varying crankshaft angles between 35
o
and 22
o
BTDC. A small difference of peak
pressure between TPI and non-TPI conditions was seen at a TPI pulse position of 23.9
o
BTDC.
However, a further check of the P-V area, the work done by engine, indicated no significant
difference between two conditions (Figure 4.20). The difference in work delivered in the two
cases is not statistically significant. Therefore, to resolve this inconclusive issue multiple TPI
pulses with higher voltage are needed.

Figure 4.19 Peak pressure change versus TPI angle. The x-axis error bar (trigger angle jitter)
isn’t shown here but it is less than 0.5 degree (calculated by hand).
66
Kubota engine at 1750 rpm with single pulse (~43 kV)

Load
NP/P
events
BTDC
(deg)
*

%
**
T-test
+
80/80 -35.490 0.336 0.320
73/73 -31.587 -0.028 0.940
72/72 -30.829 -0.287 0.490
73/73 -28.522 -0.421 0.290
72/72 -26.782 -0.011 0.980
72/72 -25.179 0.475 0.260
65/65 -23.882 0.983 0.006
66/66 -22.791 -0.534 0.190
73/73 -21.924 0.205 0.590

*
BTDC: Before Top Dead Center
**
%: Peak pressure difference between TPI and No TPI.
+
T-test: The higher number indicates a higher probability to be null, the difference of peak
pressure is by chance.

No Load
NP/P
events
BTDC
(deg)
*

%
**

T-
test
+
73/73 -35.490 0.100 0.78
65/65 -31.587 -0.144 0.68
73/73 -30.829 -0.332 0.4
65/65 -28.522 -0.036 0.92
73/73 -26.782 0.382 0.27
65/65 -25.179 -0.204 0.58
73/73 -23.882 -0.345 0.33
65/65 -22.791 0.453 0.23
72/72 -21.924 -0.015 0.96
67

Figure 4.20 Peak pressure with time and PV diagram in both control and TPI conditions. The
relationship between the crank angle and the volume:
221/2
11
[1 cos ( sin ) ]
12
d
V
RR
Vr
θθ =+ +− − −
−
, where V
d
(325 cm
3
) is the displacement volume, r
(23) is the compression value, R is the connecting length over crank radius. After obtaining the
volume, PV diagram can be plotted.
68
4.3.6 Further experiments
Diesel engine modification
In order to operate the small diesel engine closer to the operating method of large, open
combustion chamber which one finds in marine diesel engines, we modified the engine head by
drilled out the bottom of pre-combustion chamber up the pre-combustion chamber within the
head gasket area in order not to interfere with the seals (Figure 4.21(a)). Without prep-
combustion chamber, radicals generated by plasma will have better interactions with fuel and air.
The modified head had swapped on the engine and tested with the new electrode (Figure 4.21(b)),
which has a thin disc at the end of the tip. The sharp edge of the disc will increase the local
electric field and is benefit to plasma generation. New baseline is in Figure 4.22. The
compression pressure is dropped from 360 psi to 340 psi and the maximum rpm is 1700, which
was 2800.

69

Figure 4.21 (a) View of modified engine head; (b) electrodes.

600
650
700
750
800
850
900
950
1000
600 1000 1400 1800 2200
Rpm
Pressure (psi)
Before
After

Figure 4.22 New baseline of Kubota engine when no load is applied.

New Old
70
Diesel engine experiments
Instead of trigger from pressure signal, the trigger source is changed to rpm signal for
better consistency and less noise issue (Figure 4.23). The rpm signal has been re-verified by cold
cranking and is 50 degree BTDC.
The engine was run at 780 rpm with no load on it. The pulse generator is at full voltage,
which is about 40kV. Different parameters (single pulse at 25 degree BTDC, which is close to
fuel injection timing, and multiple pulses with various repetition rates) have been tested and data
is listed in Table 4.1. The standard deviation at 1 kHz burst mode is big due to the noise interfere
with peak pressure (Figure 4.24) and can be eliminated by reducing the pulse numbers.

Figure 4.23 New experimental setup. The trigger source is changed to rpm signal to avoid noise
issue.
Rpm
Signal
Agilent
Quantum
Composer
Pulse
generator
Oscilloscope
trigger trigger
Pulse every 4th
Pulse timing/ mode
trigger
Kubota
Engine
Pulse timing
Current
Pressure RPM
71

P_# P_ave P_std NP_# NP_ave NP_std % diff
Single 133 773.669 15.3556133 774.72618.772 -0.136
1 kHz 405 772.653 60.916 405 774.17415.565 -0.197
2 kHz 406 775.271 13.307 406 773.9 15.102 0.177
8 kHz 212 785.462 17.532 212 779.62319.377 0.749

Table 4.2 Peak pressure measurements with different parameters. From left to right are: number
of pulsed cycle, average peak pressure of pulsed cycles, standard deviation; number of non-
pulsed cycle, average peak pressure of non-pulsed cycles and standard deviation.

Figure 4.24 (a) 1 kHz burst mode (b) 2 kHz burst mode.

72

Figure 4.25 (a) Extended cathode (b) The cathode and electrode installed on the engine head.

4.4 Conclusion
Initial experiments in a single-cylinder diesel engine show a 1.5% increase in peak
pressure using a single transient plasma discharge per cycle. These data imply that transient
plasma is capable of improving fuel consumption in diesel engines, thereby reducing harmful
emissions. However, the result is difficult to reproduce due to many uncontrollable factors. The
possible explanations of why TPI system appeared to have little effect on diesel combustion are
(1) the plasma was not being generated effectively inside the cylinder and (2) the fuel oxidation
rate was limited by the evaporation rate and that there is less room for benefit from the
generation of radicals.
Extended cathode
Gap is about 1.5 mm
73
Chapter 5: Application of streamer discharge
on diesel exhausts treatment

5.1 Introduction
The air pollution from combustion and industrial exhaust becomes worse and causes
environmental problems. Due to the concern of environmental impact, stricter regulations and
laws are enforced. Thus, several techniques, such as electron beam [68-69], dielectric barrier
discharge [70-71] and pulsed streamer discharge [72,73,74], have been studied to remove
pollutants from exhaust. Among these techniques, pulsed streamer discharge shows a great
potential on treating exhaust efficiently and inexpensively. Even though the streamer discharge
technique has the best energy efficiency, only 30% of input energy is transformed into plasma
[75]. Hence, lots of research has been done to improve the energy efficiency.
V. Puchkarev et al. studied the energy cost of NO
x
removal from diesel exhaust by using
pulsed corona discharge with different parameters such as pulse width, pulse polarity, current
density, repetition rate and reactor design [76]. They obtained the best efficiency (<10 eV/mol) at
the optimal condition that current density is less than 0.2 A/cm2 and repetition rate is about 1
74
kHz. They found that the energy cost decreases with shorter pulse, the decrease of current
density as well as the increase of repetition rate.
T Namihira et al conducted similar measurements but with positive voltage, and they
confirmed that the energy efficiency of removal NO increases when the pulse width decreases
[77]. However, there is a disagreement on the effect of pulse repetition rate. Yankelevich et al.
found that the NO removal efficiency decreases as the pulse repetition rate increases and
attributed it to the higher current density at a high repetition rate [78].
Another important study had done by B.S. Rajanikanth et al.. They investigated how NO
x

removal can be affected by carbonaceous soot. By treating filtered and unfiltered diesel exhaust,
they found that the presence of carbonaceous soot enhanced NO
x
removal efficiency due to the
directly reaction with NO
2
and thus the decrease of NO
x
concentration. Besides, they conducted
experiments with different after-treatment techniques, such as catalyst and absorbent processes,
associated with plasma and confirmed that both techniques have better performance with the
assist of plasma. They also found the NO removal efficiency decreases with the increase of
exhaust temperature [79].
In this chapter, pulsed streamer discharge technique is used for exhaust treatment. At the
beginning, the basic mechanism of technique and the principle of gas analyzer are mentioned
first. Then, exhaust is treated in two places- inside the engine cylinder and outside the engine
75
exhaust port. After that, we study the effects of repetition rate and pulse width on energy
efficiency as well as calculate the energy efficiency.

5.2 Basic mechanism of NO removal and operation principle of gas
analyzer
5.2.1 Mechanism of NO removal by pulsed corona discharge
The mechanism of NO removal by using pulsed corona discharge is based on plasma
enhanced chemical reactions. Following are some of the possible reactions:
e + O
2
→ e + O + O (1)
e + N
2
→ e + N + N (2)
e + H
2
O → e + H + OH (3)
O + O
2
→ O
3
(4)
N + NO → N
2
+ O (5)
O + NO → NO
2
(6)
O
3
+ NO → NO
2
+ O
2
(7)
OH + NO → HNO
2
(8)
Reactions 1-4 are initial steps that energetic electrons generated by streamer discharge collide
with neutral molecules and produce radicals such as O, N, OH and O
3
. Reaction 5 is NO removal
by N atom while reactions 6-8 are NO oxidized to NO
2
or HNO
2
. NO oxidation is well believed
to be dominant in NO removal [80].
76
5.2.2 Operation principle of gas analyzer
In the emission measurement, a compact commercial exhaust gas analyzer which has two
types of sensors to detect five different gases (Table 5.1). One is electrochemical sensor which is
used to detect oxygen and oxides of nitrogen, and the other is infrared sensor used to detected
rest of the gases. The readings are insensitive to the flow rate and temperature.

Gas Sensor Resolution
Carbon Monoxide (CO) Infra-red 0.01%
Oxygen (O2) Electrochemical 0.01%
Hydrocarbon (HC) Infra-red 1 ppm
Carbon Dioxide (CO2) Infra-red 0.10%
Oxide of Nitrogen (NO) Electrochemical 1 ppm

Table 5.1 Sensors used to detect different gases in the exhaust gas analyzer and their resolutions.

A. Electrochemical gas sensor
The electrochemical sensors are operated by reacting with sample gas and producing an
electrical signal which is proportional to the gas concentration. Typically, an electrochemical gas
sensor has an anode, cathode and electrolyte (Figure 5.1). The sample gas passes through a gas
selective membrane, which only allows for the interest gas to pass through, and diffuse into an
electrolyte solution. Then, the gas reacts with electrodes (either oxidation or reduction) and
77
produces electrical signals (current). The current can be measured and is used to determine the
gas concentration.

Figure 5.1 Typical configuration of electrochemical sensor used for gas analyzer [].

B. Nondispersive infrared gas sensor
Infrared spectroscopy is commonly used for gas analysis due to the ability of infrared
radiation to interact with a large variety of molecules. When infrared radiation passes through an
infrared active sample, some of the radiation is absorbed and causes the temperature increase in
gas and the decrease of the radiation.
For a molecule to be infrared-active, its dipole moment need to be changed. An electric
dipole occurs when two adjacent atoms within a molecule have different charges and separated
by a distance r. The dipole moment, µ, is defined as the product of the charge q and the vector
78
separation r. When the distance between two atoms changes due to vibration or rotation, the
dipole moment is changed.
Atomic species which contain only one atom such as Ar and diatomic molecules which
have identical charge do not meet the criteria for infrared active molecules since atomic species
have no dipole moment, and the vibration and rotation of diatomic molecule cause no change in
dipole moment.
A nondispersive infrared sensor (or NDIR) is called nondispersive because the
wavelength which passes through the sampling chamber is not pre-filtered instead a filter is used
before the detector. The key components are an infrared source (lamp), a sample chamber or
light tube, a wavelength filter, and an infrared detector. The gas is pumped or diffuses into the
sample chamber, and gas concentration is measured electro-optically by its absorption of a
specific wavelength in the infrared (IR).

Figure 5.2 Schematic view of a non-dispersive infrared sensor.
79
5.3 Streamer discharge application on diesel exhaust treatment
5.3.1 Exhaust treatment outside the engine
The Kubota diesel engine was run at 1800 rpm with full (3 kW) load applied. A
pseudospark pulse generator is used for exhaust treatment. The time for emission gas transport
from engine exhaust port to gas analyzer is about 1.5 minute. Figure 5.3 shows the treatment
procedures.
Control
(5 min)
Analyzer zero
1 min waiting
Control
(5 min)
TPI
(5 min)
TPI
(5 min)
Analyzer zero
1 min waiting
Analyzer zero
1 min waiting

Figure 5.3 The exhaust treatment procedures.

An electrode (Figure 5.4) is connected in series with a diesel engine exhaust. This
electrode is designed to be used with the pseudospark pulse generator (60 kV, 85-100 ns pulse)
since the gap size is greater than 1 cm. The electrode is used to treat a volume of exhaust gas and
NO
x
concentration in an exhaust gas sample is measured. Preliminary results indicate that NO
concentration can be reduced by a factor 2 when transient plasma pulses are applied as shown in
Figure 5.5. The repetition rate of the pulse generator in this experiment is 2 Hz.

80

Figure 5.4 The electrode used for exhaust treatment.

Figure 5.5 Plasma treatment reduces NO concentration in the exhaust gas stream by a factor 2.
81
5.3.2 Exhaust treatment in the engine cylinder
The Kubota engine was run at both rich (high rpm and full load) and lean conditions (low
rpm and no load). Pulses are applied in the exhaust cycle. The treatment procedure is in Figure
5.6.
Control
(5 min)
Analyzer zero
1 min waiting
Control
(5 min)
TPI
(5 min)
TPI
(5 min)
Analyzer zero
1 min waiting
Analyzer zero
1 min waiting

Figure 5.6 Exhaust treatment procedure.

The electrode used in this part is the same as the one used in Chapter 4, a modified spark plug
and it is installed in the glow plug hole. 10 pulses (12 ns pulse width and 40 kV at 9 kHz) were
applied in every exhaust cycle and the pulse timing was before exhaust valve open and close to
valve open (130 degree ATDC). Figure 5.7 shows the results in both rich (1500 rpm with full
load) and lean (780 rpm without load) conditions and there is no big effect on plasma treatment
inside the engine cylinder.

82

Figure 5.7 NO concentration varied with time in (a) rich condition (1500 rpm and full load) and
(b) lean condition (780 rpm and no load). C means control, P means TP treatment and the
numbers are the average of NO concentration in a stable range.

NO (ppm)
Time (s)
83
5.4 Discussion
Comparing two different places for exhaust treatment, our results show that exhaust
treatment inside the engine cylinder is not as efficient as the treatment outside the engine. The
main reason is the amount of exhaust gas being treated. We first estimate the volume of exhaust
gas treated outside the engine per second by calculating the flow rate. The total volume of tubes
and electrode is about 1.86 x 10
3
cc and the travel time from exhaust port to the analyzer is about
90 second. Thus, the flow rate is roughly 2 cc/sec. However, the volume of exhaust gas treated
inside the cylinder is about 325 cc per cycle. Depending on the rpm, the amount of exhaust is
changed. For example, when the engine run at 780 rpm, there are 6.5 exhaust cycles per second.
So the amount of exhaust gas treated inside the engine cylinder is about 1000 times higher than
those treated outside the engine.
In addition, the treatment outside the engine is continuous pulses while inside the engine
cylinder is only 10 pulses per cycle due to the limitation of pulse generator. Furthermore, the
turbulent flow inside the cylinder, which might affect the interaction between radicals and
exhaust gas, as well as the temperature of exhaust gas have impact on the NO removal efficiency
[79].

84
5.5 Calculation of NO removal efficiency and the effect of pulse width &
repetition rate
After head modification, the exhaust treatment outside the engine has been repeated. The
engine ran at 1600 rpm with full load. The temperature at engine head is ~ 800 F, which is higher
than it used to be (~400 F) because the pre-combustion chamber was drilled out. The
pseudospark pulse generator (85 ns) and the solid state pulse generator (12 ns) are applied and
the repetition rate is varied.
Figure 5.8 showed the emission data. The NO reading in control was much lower than
before probably due to the pre-combustion chamber drill out. Since the compression ratio
dropped, the peak temperature decreased and resulted in less NO. With the pseudospark pulse
generator, ~ 50% of NO is removed at 0.1 Hz while with the 12 ns pulse generator, a different
repetition rate, 0.5 Hz, is applied to remove the same percentage of NO. From the data, we also
noticed that the repetition rate is not linearly related to the percentage of removed NO. Thus, to
analyze the impact of pulse width on energy efficiency, we compare the data while ~50% NO is
removed by both pulse generators (Figure 5.9).
The NO removal energy per molecule at is estimated as below:
(a) Pseudospark pulse generator (85 ns, 0.1 Hz)
85
Since the average flow rate is 2 cc/ sec and the plasma removed about 28 ppm of NO, the
number of removed NO molecules:
() ( )
()
6
23
3
15
2/sec 28 10
6.02 10
22.4 10
~1.5 10 /sec
cc
cc
NO molecules
−
××
××
×
∆×

,where 22.4 liter is gas volume at STP condition. The energy output of each pulse is about 800
mJ. So the energy required for decomposing one NO molecule is:
()
3
17
15
800 10 0.1
5.3 10 /
1.5 10 / sec
JHz
J molecule
molecules
−
−
××
=×
×

,which is about 330 eV / molecule.
(b) Solid state pulse generator (12 ns, 0.5 Hz)
The flow rate is 2 cc/sec and about 24 ppm of NO is removed at 0.5 Hz. Thus, the number of
removed NO molecules:
() ( )
()
6
23
3
15
2 / sec 24 10
6.02 10
22.4 10
~ 1.29 10 / sec
cc
cc
NO molecules
−
××
××
×
∆×

The energy deposition to streamer per pulse is about 10.4 mJ. Thus, the energy used to remove
one NO molecule is:
()
3
18
15
10.4 10 0.5
4.03 10 /
1.29 10 / sec
JHz
J molecule
molecules
−
−
××
=×
×

, which is about 25.2 eV / molecule.
86
0
10
20
30
40
50
60
70
0
50
100
150
200
250
300
350
time (s)
NO (ppm)
85 ns_C1
85 ns_0.5 Hz
85 ns_C2
85 ns_0.1 Hz
85 ns_C3
85 ns_0.2 Hz

0
10
20
30
40
50
60
70
030 60 90 120 150
time (sec)
NO (ppm)
12 ns_0.5 Hz
12 ns_1 Hz
Control1
Control2

Figure 5.8 Exhaust treatments at 1600 rpm with full load. (a) With 85 ns pseudospark pulse
generator, there is a 50% NO reduction at 0.1 Hz, 90% at 0.2 Hz and 100% removal at repetition
rate higher than 0.5 Hz; (b) with 12 ns solid state pulse generator, there is a 50% NO decrease at
0.5 Hz and 80% at 1 Hz.
87
From the above estimation, it indicates that the shorter the pulse width, the higher the
energy efficiency.
Furthermore, to understand the effect of repetition rate on energy efficiency, the same
exhaust treatment procedure is repeated several times with different repetition rate. The results
are shown in Table 5.2, Figure 5.10 and Figure 5.11. The energy cost increases with the
repetition rate and saturates after 50% of NO removal.

0
10
20
30
40
50
60
70
0 50 100 150 200 250 300
time (s)
NO (ppm)
12 ns pulser_control
12 ns pulser_0.5Hz
85 ns pulser_control
85 ns pulser_0.1Hz

Figure 5.9 Comparing the energy cost at the condition that the similar percentage of NO is
removed. It showed that the shorter pulse width has better energy efficiency.
88
0
10
20
30
40
50
60
70
0 50 100 150 200 250 300 350
time (sec)
NO (ppm)
12ns_C1
12ns_C2
12ns_C3
12ns_C4
12ns_0.3Hz
12ns_0.5Hz
12ns_0.7Hz
12ns_1Hz
12ns_1.2Hz

Figure 5.10 Emission treatments with different repetition rate. The higher repetition rate
generates more radicals per unit time and results in more NO reduction.

Rep.
rate NO ∆ NO %
# of
molecule J/# eV/#
0 55 - - - - -
0.3 38 17 30.909 9.138E+14 3.415E‐18 21.341
0.5 31 24 43.636 1.290E+15 4.031E‐18 25.194
0.7 26 29 52.727 1.559E+15 4.670E‐18 29.190
1 13 42 76.364 2.258E+15 4.607E‐18 28.793
1.2 5 50 90.909 2.688E+15 4.644E‐18 29.023

Rep.
rate NO ∆ NO % # of molecule J/# eV/#
Control 55 ‐ ‐ ‐ ‐ ‐
0.1 28 27 49.09 1.45E+15 5.51E‐17 3.45E+02
0.2 6 49 89.09 2.63E+15 6.07E‐17 3.80E+02
Table 5.2 Calculated results of percentage of NO reduction, number of removed NO molecules
and energy cost per molecule. Exhaust treated by transient plasma produced with (a) the solid
state (12 ns, 40 kV) (b) the pseudospark (85 ns, 60 kV) pulse generator.
89
0
20
40
60
80
100
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Repetition rate (Hz)
Removed NO(%)
0
5
10
15
20
25
30
35
Energy cost (eV/#)
Removed NO%
Energy cost

Figure 5.11 Emission treatments with different repetition rate. The higher repetition rate
generates more radicals per unit time and results in more NO reduction.

90
Chapter 6: Conclusion and future work

6.1 Conclusion
6.1.1 Nanosecond pulsed streamer propagation at high pressures
Positive streamer discharges in synthetic dry air were investigated at high pressures in
this dissertation. By using the optical diagnostics, we took images of streamer propagation at the
pressure range of 1-3 bar. At pressures higher than this, we implemented the electrical analysis to
estimate the streamer velocity since the light intensity was very low. The electrode configuration
was coaxial with various radial gaps from 5.75 to 11.75 mm, and two pulse generators were used
to generate the streamers. One was the pseudospark pulse generator (65 ns with maximum
voltage of 65 kV) and the other was the solid state pulse generator (12 ns with maximum voltage
of 40 kV). We measured the time of streamer propagate across the gap, plotted its head position
as a function of time, and calculated its velocity in the pressure range of 1-18 bar. Our data
shows that the velocity is a function of external electric field and pressure, which is consistent
with literature. Meanwhile, we checked the similarity law by analyzing it as a function of the
scaling factor (E/P). Although the result, the invalidity of the similarity law at high pressures, is
91
consistent with other studies, we provide a new argument to explain it. In addition, we find a new
scaling factor, E/ √P, and it is explained by the dimensional analysis.

6.1.2 Application of transient plasma technique to diesel engines
We developed rules for igniter design. An optimal ratio of anode radius to cathode radius,
which is 1/3, is calculated. By keeping this ratio, the electric field between the channel and anode
is minimized and thereby less transient plasma is generated in between. Thus, arcing and
electrical breakdown might be eliminated. In addition to the igniter design, we applied pulsed
streamer discharge on both of the engine efficiency and exhaust treatment. In the aspect of
engine efficiency, a 1.5% increased peak pressure was observed. Although it was only a small
improvement, it implied that transient plasma can improve the fuel consumption of diesel
engines. In order to improve it more, transient plasma needs to be generated in side the cylinder
more effectively. In the exhaust treatment, our results show that the NO can be efficiently
reduced by using the transient plasma technique. The results agreed with other experimental
work that the energy efficiency is higher by using a shorter pulse width. According to our data,
the lowest energy cost to remove each NO molecule is 30 eV by using the 12 ns pulse generator.
92
6.2 Future work
6.2.1 Measurement of radicals generated by streamer discharge at high pressures
It is well accepted that the active species generated by pulsed streamer discharge, such as
O, H and OH, can significantly improve combustion by lowering the fuel oxidation temperature,
increasing the flame propagation velocity, and improving the flame stability and efficiency
[81,82,83]. These radicals have been studied at low pressures in many aspects, ex. quenching
effect, densities and kinetic reactions. However, none of these studies have extended to a
pressure range of a practical system. In order to fully understand the mechanism of plasma
assisted combustion, it is essential to measure the concentration and lifetime of these species at
high pressures. These radical characteristics at high pressures can be measured using Two-
photon absorption laser induced fluorescence (atomic oxygen) or Laser Induced Fluorescence
(OH). Knowing the behaviors of active species can clarify the underlying mechanism of plasma
assisted combustion as well as develop numerical model.

6.2.2 Other characteristics of pulsed streamer discharge at high pressures
In addition to the radical species and the lifetime of radicals, the background temperature
of streamer discharge at high pressures can be measured as well. Research has been done to
93
measure the temperature of streamer discharge at atmospheric pressure by using optical
spectrometers [84]. The background temperature can be extracted from the rotational profile of
the
33
22
(,'0) ( ,''2)
ug
NC v N B v Π= → Π = transition at 380nm. Via measuring the background
temperature, we can tell if the streamer discharges at high pressures are leader forms.

6.2.3 Transient plasma effects on the stability of low-temperature combustion
Exhaust gas recirculation (EGR) is an effective technique to reduce the NO emissions
from diesel engines because it lowers the flame temperature and dilutes the intake oxygen
concentration. However, the con of diesel EGR is that while decreasing NO, the particulate
matter (PM) increases. In addition, as the EGR ratio increases, the engine destabilizes. The
instability caused by the diesel EGR may be improved since transient plasma has been
demonstrated to improve flame stabilization [85]. Thus, a diesel EGR associating with transient
plasma might be the next step for engine technology.

6.2.4 Pulse generator and electrode design for scaling up to marine diesel
In order to apply transient plasma to marine diesel engines, a new pulse generator and
electrode need to be developed. The electrode must mechanically survive under extremely high
pressures and temperatures (the pressure is above 100 bar and the temperature is about 1600-
94
2000 C). In addition, the electrode design needs to consider the erosion caused by fuel spray
impingement. Another consideration of the electrode design is that the insulator needs to hold up
a higher voltage extrapolated from our inception voltage measurement, which might be higher
than 100 kV.
As to the new design of the pulse generator, we calculate the scaling factors of the marine
diesel engine to the single cylinder diesel engine based on the engine specifications (Table 6.1).
To estimate the power of the pulse generator needed in the marine diesel engine, we scale it up
with the ratio of engine output power. The burst power is calculated as
Pulse energy Pulse number
Burst duration
×
and the power needed in the marine diesel engine is hence 50 kW.
To generate 50 kW in power, we need to increase the pulse energy, repetition rate and pulse
numbers. The specifics of the pulse generator used on the Kubota engine and attainable features
of the new pulse generator are list in Table 6.2.
For the exhaust treatment, we assume that the composition of the exhaust in the Sulzer
engine is the same as that in the Kubota engine (even though we expected it to be higher). Since
the cylinder volume of the Sulzer engine is 20,000 times larger than that of the Kubota engine,
the number of NO should be 20000 times more. The energy needed to remove 50% NO in
Kubota engine is 0.1 J (from our results, the energy cost for removing one NO molecule is 30
95
eV). The energy needed in the Sulzer engine is therefore 20000 times more, which is 2000 J, and
the power is 5 kW (Table 6.3).

Kubota Sulzer Scaling Factors
Max RPM 3000 150 20 x Slower RPM
Bore 75 mm 680 mm 9 x Larger Bore
Stroke 70 mm 1550 mm 22 x Longer Stroke
Cycle Four Stroke Two Stroke Fire 2 x as Often
Output power 6 kW 650 kW 100 x Higher power

Table 6.1 Engine specifics and scaling factors of Kubota and Sulzer engine.

Specification Kubota Sulzer
Pulse Energy 50 mJ 500 mJ
Rep Rate 10 kHz 100 kHz
Pulses/Burst 10 1000
Burst Duration 1 ms 10 ms
Burst Power 0.5 kW 50 kW

Table 6.2 Parameters of pulse generator used on Kubota and needed for Sulzer.
96

Kubota Sulzer
RPM 1600 150
Volume 31 cm
3
5.6E+06 cm
3

Removed # of NO 2.08E+16 4.16E+20
Energy 0.1 J 2000 J
Time/cycle 37.5 ms 400 ms
Power 2.6 W 5000 W
Table 6.3 Estimation of energy needed to remove 50% of NO in the Sulzer engine.
97
References

1 Townsend J S E, Electricity in Gases (Oxford:Clarendon Press 1915)

2 Starikovskii A Yu, “Plasma supported combustion”, Proc. Combustion Inst. 30 2405–
2417 (2005)

3 A Starikovskiy and N Aleksandrov, “Plasma-assisted ignition and combustion”, Progress
in Energy and Combustion Science vol. 39 61-110 (2013)

4 Seiji Samukawa et al, “The 2012 plasma roadmap”, J. Phys. D: Appl. Phys. 45 253001
(2012)

5 Kong M G, Kroesen G, Morfill G, Nosenko Y, Shimizu T, J van Dijk
2
and J L
Zimmermann, "Plasma medicine: an introductory review”, New J. Phys. 11 115012
(2009)

6 G Y Park et al, “Atmospheric-pressure plasma sources for biomedical applications”,
Plasma Sources Sci. Technol. 21 043001 (2012)

7 W. Sun, Y. Ju , “Non-equilibrium plasma assisted combustion- A review of recent
progress”, Journal of Plasma and Fusion Research, 89(4), 209-219 (2013)

8 Won J Yi and P F Williams, "Experimental study of streamers in pure N2 and N2/O2
mixtures and a ~ 13 cm gap", J. Phys. D: Appl. Phys. vol. 35, pp. 205, (2002)

9 Nijdam S, van de Wetering F M J H, Blanc R, van Veldhuizen E M and Ebert U,
“Probing photo-ionization: experiments on positive streamers in pure gases and mixtures”,
J. Phys. D: Appl. Phys. 43 145204 (2010)

10 G Wormeester, S Pancheshnyi, A Luque, S Nijdam and U Ebert, “Probing photo-
ionization- simulations of positive streamers in varying N2 O2 mixtures”, J. Phys. D:
Appl. Phys. 43 505201 (2010)

98

11 Nijdam S, Wormeester G, van Veldhuizen E M and Ebert U, “Probing background
ionization: positive streamers with varying pulse repetition rate and with a radioactive
admixture”, J. Phys. D: Appl. Phys 44 455201 (2011)

12 Briels T M P, Kos J, Winands G J J, Veldhuizen E M and Ebert U, “Positive and negative
streamers in ambient air: measuring diameter, velocity and dissipated energy”, J. Phys. D:
Appl. Phys. 41 234004 (2008)

13 Luque A, Ratushnaya V and Ebert U, “Positive and negative streamers in ambient air:
modeling evolution and velocities”, J. Phys. D: Appl. Phys. 41 234005 (2008)

14 Briels T M P, Kos J, van Veldhuizen E M and Ebert U, “Circuit dependence of the
diameter of pulsed positive streamers in air”, J. Phys. D: Appl. Phys. 39 5201 (2006)

15 Briels T M P, van Veldhuizen E M and Ebert U, “Positive streamers in air and nitrogen
of varying density: experiments on similarity laws”, J. Phys. D: Appl. Phys. 41 234008
(2008)

16 J. Lefkowitz et al., 49th AIAA/ASME/SAE/ASEE Joint Propulsion Conference and
Exhibit (2013)

17 F. Schauer et al., 39th AIAA Aerospace Science Meeting, AIAA-2001-1129 (2001)

18 Wang F, Liu J B, Sinibaldi J, Brophy C, Kuthi A, Jiang C, Ronney P and Gundersen M A,
“Transient plasma ignition of quiescent and flowing air/fuel mixtures”, IEEE Trans.
Plasma Sci. 33 844–9 (2005)

19 Cathey C, Tang T, Shiraishi T, Urushihara T, Kuthi A and Gundersen M A, “Nanosecond
plasma ignition for improved performance of an internal combustion engine”, IEEE
Trans. Plasma Sci. 35 1664–8 (2007)

20 W. Kim, H. Do, M. G. Mungal, and M. A. Cappelli, “A study of plasma-stabilized
diffusion flames at elevated ambient temperatures”, IEEE Trans. Plasma Sci. vol. 34 no.
6, pp.2545 -2551 (2006)

99

21 C. Cathey, J. Cain, H. Wang, M. A. Gundersen, M. Ryan, and C. Carter, “OH production
by transient plasma and mechanism of flame ignition and propagation in quiescent
methane-air mixtures”, Combustion and Flame Vol. 154, Issue 4, Pages 715-727 (2008)

22 S Serbin et al., “Improvement of the gas turbine plasma assisted combustion
characteristics”, 49
th
AIAA Aerospace Science Meeting AIAA-2011-61 (2011)

23 J. P. Moeck et al., “Control of combustion dynamics in a swirl-stabilized combustor with
nanosecond repetitively pulsed discharges”, 51
st
AIAA Aerospace Science Meeting
AIAA-2013-0565 (2013)

24 S. A. Bozhenkov, S. M. Starikovskaia, and A. Y. Starikovskiia, “Nanosecond gas
discharge ignition of H2 and CH4 containing mixtures”, Combustion and Flame 133,
133-146 (2003)

25 Timothy Ombrello, Sang Hee Won, Yiguang Ju, Skip Williams, “Flame propagation
enhancement by plasma excitation of oxygen Part I effects of O3”, Combustion and
Flame 157, 1906-1915 (2010)

26 Timothy Ombrello, Sang Hee Won, Yiguang Ju, Skip Williams, “Flame propagation
enhancement by plasma excitation of oxygen Part II effects of O2 (a1 ∆g)”, Combustion
and Flame 157, 1916-1928 (2010)

27 E. Mintusov et al, “Mechanism of plasma assisted oxidation and ignition of ethylene-air
flows by a repetitively pulsed nanosecond discharge”, Proc. Combustion Institute 32
3181 (2009)

28 A. Bao, Y.G. Utkin, S. Keshav, Lou Guofeng, I.V. Adamovich, “Ignition of Ethylene-air
and Methane-air flows by low-temperature repetitively pulsed nanosecond discharge
plasma”, IEEE Trans. Plasma Sci. 35 1628 (2007)

29 G. Lou et al., “Ignition of premixed hydrocarbon-air flows by repetitively pulsed
nanosecond pulse duration plasma”, Proc. Combustion Institute 31, 3327 (2007)

100

30 Achat S, Teisseyre Y and Marode E, “The scaling of the streamer-to-arc transition in a
positive point-to-plane gap with pressure”, J. Phys. D: Appl. Phys. 25 661-8 (1992)

31 N. Yu. Babaeva and G.V. Naidis, “On streamer dynamics in dense media”, Journal of
Electrostatics 53 2, 123–133 (2001)

32 P Tardiveau, E Marode, A Agneray and M Cheaib, “Pressure effects on the development
of an electric discharge in non-uniform fields”, J. Phys. D: Appl. Phys. 34 1690 (2001)

33 P Tardiveau et al, “Diffuse mode and diffuse-to-filamentary transition in a high pressure
nanosecond scale corona discharge under high voltage”, J. Phys. D: Appl. Phys. 42
175202 (2009)

34 Peek F W, Dielectric Phenomena in High-Voltage Engineering (New York: McGraw-Hill)
(1929)

35 Meek, J. M. A theory of spark discharge. Phys. Rev. 57, 722–728 (1940)

36 Raizer Yu P, Gas Discharge Physics (Berlin: Springer) (1991)

37 Flegler E and Raether H, Untersuchung von Gasentladungsvorg¨ angen mit der
Nebelkammer Z. Tech. Phys. 11 435 (1935)

38 Bradley C D and Snoddy L B, “Ion distribution during the initial stages of spark
discharge in nonuniform fields”, Phys. Rev. 47 541 (1935)

39 Cravath A M, “Photoelectric effect and spark mechanism”, Phys. Rev. 47 254 (1935)

40 Nijdam S, van de Wetering F M J H, Blanc R, van Veldhuizen E M and Ebert U,
“Probing photo-ionization: experiments on positive streamers in pure gases and mixtures”,
J. Phys. D: Appl. Phys. 43 145204 (2010)

41 Nijdam S, Wormeester G, van Veldhuizen E M and Ebert U, “Probing background
ionization: positive streamers with varying pulse repetition rate and with a radioactive
admixture”, J. Phys. D: Appl. Phys 44 455201 (2011)
101

42 Meek, J. M. and Craggs, J. D., Electrical breakdown of gases (1978)

43 Raizer, Yu. P., Gas discharge physics (Springer-Verlag Berlin, 1991).

44 Roth, J. R., Industrial plasma engineering: Principles (Institute of Physics Publishing,
London, 1995).

45 Ebert U, Nijdam S, Li C, Luque A, Briels T, van Veldhuizen E, “Review of recent results
on streamer discharges and discussion of their relevance for sprites and lightning”,
Journal of Geophysical Research vol. 115, A00E43 (2010)

46 Takao Namihira, Douyan Wang, Sunao Katsuki, Reuben Hackam, Hidenori Akiyama,
“Propagation velocity of pulsed streamer discharges in atmospheric air”, IEEE TRANS
PLASMA SCI vol. 31, no. 5, pp. 1091-1094 (2003)

47 M. Arrayás, U. Ebert, W. Hundsdorfer, “Spontaneous branching of anode-directed
streamers between planar electrodes”, Phys. Rev. Lett. 88, 174502 (2002).

48 B. Meulenbroek, A. Rocco, and U. Ebert, Streamer branching rationalized by conformal
mapping techniques, Phys.Rev.E 69, 067402 (2004).

49 A. Luque and U. Ebert, “Electron density fluctuations accelerate the branching of
positive streamer discharges in air”, Phys. Rev. E 84, 046411 (2011)

50 S Pancheshnyi, “Role of electronegative gas admixtures in streamer start propagation and
branching phenomena”, Plasma Sources Sci. Technol. 14 645 (2005)

51 U Ebert, C Montijn, T M P Briels, W Hundsdorfer, B Meulenbroek, A Rocco and E M
van Veldhuizen, “The multiscale nature of streamers,” Plasma Sources Sci. Technol., vol.
15, no. 2, pp. S118–S129 (2006)

52 Luque, A., F. Brau, and U. Ebert, “Saffman Taylor streamers: Mutual finger interaction
in spark formation”, Phys. Rev. E, 78, 016206 (2008)

102

53 U Ebert et al., “Multiple scales in streamer discharges with an emphasis on moving
boundary approximations”, Nonlinearity 24 C1 (2011)

54 A. Luque et al., “Photoionization in negative streamers: Fast computations and two
propagation modes,” Appl. Phys. Lett. vol. 90, no. 8, p. 081 501, (2007)

55 Gennadii A Mesyats , “Similarity law for pulsed gas discharges”, Phys.-Usp. 49
1045(2006)

56 O. V. Bolotov, V. I. Golota, B. B. Kadolin, V. I. Karas’, V. N. Ostroushko, L. M. Zavada,
A. Yu. Shulika, “Similarity laws for cathode-directed streamers in gaps with an
inhomogeneous field at elevated air pressures”, Plasma Physics Reports Vol. 36, Issue 11,
pp 1000-1011 (2010)

57 N Liu and V P Pasko, "Effects of photoionization on similarity properties of streamers at
various pressures in air", J. Phys. D: Appl. Phys. vol. 39 pp. 327, (2006)

58 Briels T M P, Veldhuizen E M van and Ebert U, “Branching of positive discharge
streamers in air at varying pressures”, IEEE Trans. Plasma Sci. 33 264-5 (2005)

59 www.thermopedia.com/content/553/#AIR_PROPERTIES_OF_TABLE1

60 T M P Briels, E.M. van Veldhuizen, U Ebert, “Positive streamers in air and nitrogen of
varying density- experiments on similarity laws”, J. Phys. D: Appl. Phys. 41 234008
(2008)

61 Pancheshnyi S, Nudnova M and Starikovskii A, “Development of a cathode-directed
streamer discharge in air at different pressures: experiment and comparison with direct
numerical simulation”, Phys. Rev. E 71 016407 (2005)

62 Singleton D, Pendleton S J and Gundersen M A, “The role of non-thermal transient
plasma for enhanced flame ignition in C2H4-air”, J. Phys.D: Appl. Phys. 44 022001
(2001)

103

63 C. Cathey, F. Wang, T. Tang, A. Kuthi, M. A. Gundersen, J. Sinibaldi, C. Brophy, J.
Hoke, F. Schauer, J. Corrigan , J. Yu, E. Barbour, and R. Hanson, “Transient Plasma
Ignition for Delay Reduction in Pulse Detonation Engines,” 45th AIAA Aerospace
Sciences Meeting and Exhibit, Reno, Nevada, 8-11 (2007)

64 Shiraishi T, Urushihara T and Gundersen M A, “A trial of ignition innovation of
gasoline engine by nanosecond pulsed low temperature plasma ignition”, J. Phys. D:
Appl. Phys. 42 135208 (2009)

65 J.B. Liu, P.D. Ronney, F. Wang, L.C. Lee, and M.A. Gundersen, "Transient Plasma
Ignition For Lean Burn Applications," Proceedings of American Institute of Aeronautics
and Astronautics, 41st Aerospace Sciences Meeting, Reno, Nevada, No. 2003-6208

66 A Starikovskiy, “Plasma assisted ignition and combustion”, Progress in energy and
combustion science vol 39 1 61-110 (2006)

67 www.globaldenso.com/en/products/aftermarket/plug/basic_knowledge/heatrange/index.html

68 Penetrante, B. M.; Hsiao, M. C.; Bardsley, J. N.; Merritt, B. T.; Vogtlin, G. E.; Wallman,
P. H.; Kuthi, A.; Burkhart, C. P. and Bayless, R. J., “Electron beam and pulsed corona
processing of volatile organic compounds and nitrogen oxides”, Proceedings of 10th
IEEE Int. Pulsed Power Conf., 144-149, Albuquerque, NM, USA (1995)

69 Penetrante, B. M.,” Removal of NO from diesel generator exhaust by pulsed electron
beams”, Proceedings of 11th IEEE Int. Pulsed Power Conf., 91-96, Baltimore, MD, USA
(1997)

70 Urashima, K.; Chang, J.-S. & Ito, T. “Reduction of NO from combustion flue gases by
superimposed barrier discharge plasma reactors”, IEEE Trans. Indust. Appl., Vol. 33,
879-886.3 (1997)

71 Rosocha, L.; Anderson, G. K.; Bechtold, L. A.; Coogan, J. J.; Heck, H. G.; Kang, M.; M.
McCulla, M.; M. Tennant, M. & Wantuck, P. J., “Treatment of hazardous organic wastes
using silent discharge plasmas”, In: Non-Thermal Plasma Techniques for Pollution
104

Control, Penetrante, B. M. & Schultheis S. E. (Ed.), pt. B, 281-308, Springer-Verlag,
New York (1993)

72 Akiyama, H.; Kawamura, K.; Takeshita, T.; Katsuki, S.; Maeda, S.; Tsukamoto, S. &
Murata, M., “Removal of NO using discharges by pulsed power”, Proceedings of 10th
IEEE Int. Pulsed Power Conf., pp. 133-137, Albuquerque, NM, USA (1995)

73 Masuda S. & Nakao, H., “Control of NO by positive and negative pulsed corona
discharges”, IEEE Trans. Indust. Appl., Vol. 26, 374-383 (1995)

74 Mizuno, A., Shimizu, K., Chakrabarti, A., Dascalescu, L. and Furuta, S., “NO removal
process using pulsed discharge plasma”, IEEE Trans. Indust. Appl. vol. 31 957-963
(1995).

75 Wang. D., Namihira, T., Fujiya, K., Katsuki S. and Akiyama, H., „The reactor design for
diesel exhaust control using a magnetic pulse compressor”, IEEE Transactions on Plasma
Science, Vol. 32 No. 5, 2038-2044 (2004)

76 V. Puchkarev and M. Gundersen, "Energy efficient plasma processing of gaseous
emission using a short pulse discharge," Applied Physics Letters, vol. 71 pp. 3364-3366,
(1997)

77 Namihira, T., Tsukamoto, S., Wang, D., Katsuki, S., Hackam, R., Akiyama, H., Uchida,
Y. and Koike, M., “Improvement of NO removal efficiency using short width pulsed
power”, IEEE Trans. Plasma Sci., Vol. 28, 434-442 (2000)

78 Yankelevich Y, Wolf M et al., “NOx diesel exhaust treatment using a pulsed corona
discharge: the pulse repetition rate effect”, Plasma Sources Sci. Technol. 16 386 (2007)

79 B S Rajanikanth, A D Srinivasan and V Ravi., “Discharge Plasma Treatment for NOx
Reduction from Diesel Engine Exhaust: A Laboratory Investigation”, IEEE Trans. on
Dielectrics and Electrical Insulation, 12 (1):72-79 (2005)

105

80 Young S M, “Direct and indirect applications of dielectric barrier discharge plasma to
catalytic reduction of nitrogen oxides from exhaust gas”, Plasma Sci. Technol. 8 207
(2006)

81 W Sun, M Uddi, T Ombrello, S H Won, C Carter, and Y Ju, “Effects of non-equilibrium
plasma discharge on counterflow diffusion flame extinction”, Proc. Combustion Institute
33, 3211 (2011)
82 M. Uddie, N Jiang, E Mintusov, I V. Adamovich, W R. Lempert “Atomic oxygen
measurements in air and air-fuel nanosecond pulse discharges by two photon laser
induced fluorescence”, Proc. Combustion Institute 32, 929 (2009)

83 K. Niemi, V Schulz-von der Gathen and H F Döbele “Absolute atomic oxygen density
measurements by two-photon absorption laser-induced fluorescence spectroscopy in an
RF-excited atmospheric pressure plasma jet”, Plasma Sources Sci. Technol. 14 3.75
(2005)

84 D Singleton, S J Pendleton and M A Gundersen “The role of non-thermal transient
plasma for enhanced flame ignition in C2H4-air” (2011)

85 G. Pilla, D Galley, D A Lacoste, F Lacas, D Veynante, and C O. Laux, “Stabilization of a
turbulent premixed flame using a nanosecond repetitively pused plasma”, IEEE Trans.
Plasma Sci. 34, 2471(2006)

Abstract (if available)

Abstract The goal of this dissertation is to study high pressure streamers in air and apply it to diesel engine technologies. Nanosecond scale pulsed high voltage discharges in air/fuel mixtures can generate radicals which in turn have been shown to improve combustion efficiency in gasoline fueled internal combustion engines. We are exploring the possibility to extend such transient plasma generation and expected radical species generation to the range of pressures encountered in compression‐ignition (diesel) engines having compression ratios of ~20:1, thereby improving lean burning efficiency and extending the range of lean combustion. ❧ At the beginning of this dissertation, research into streamer discharges is reviewed. Then, we conducted experiments of streamer propagation at high pressures, calculated the streamer velocity based on both optical and electrical measurements, and the similarity law was checked by analyzing the streamer velocity as a function of the reduced electric field, E/P. Our results showed that the similarity law is invalid, and an empirical scaling factor, E/√P, is obtained and verified by dimensional analysis. The equation derived from the dimensional analysis will be beneficial to proper electrode and pulse generator design for transient plasma assisted internal engine experiments. ❧ Along with the high pressure study, we applied such technique on diesel engine to improve the fuel efficiency and exhaust treatment. We observed a small effect of transient plasma on peak pressure, which implied that transient plasma has the capability to improve the fuel consumption. In addition, the NO can be reduced effectively by the same technique and the energy cost is 30 eV per NO molecule.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

The physics and application of compact pulsed power to transient plasma ignition

PDF

Spatial and time resolved study of transient plasma induced oh production in quiscent CH4-air mixtures

PDF

Modeling the hot-dense plasma of the solar interior in and out of thermal equilibrium

PDF

Physics and applications of energetic electrons resulting from nanosecond-scale transient plasma

PDF

Topics in high voltage pulsed power plasma devices and applications

PDF

Transient pulsed plasma for pollution remediation and energy conversion applications

PDF

Biophysical studies of nanosecond pulsed electric field induced cell membrane permeabilization

PDF

On the Feynman path into the sun

PDF

Design and implementation of full solid state high voltage nanosecond pulse generators

PDF

An experimental investigation by optical methods of the physics and chemistry of transient plasma ignition

PDF

Molecular simulations of water and monovalent ion dynamics in the electroporation of phospholipid bilayers

PDF

Optoelectronic properties and device physics of individual suspended carbon nanotubes

PDF

Growth, characterization of gallium arsenide based nanowires and application in photovoltaic cells

PDF

Laboratory investigations of the near surface plasma field and charging at the lunar terminator

PDF

Nanostructure electronic band modeling for solar cell and lighting applications

PDF

GaAs nanowire optoelectronic and carbon nanotube electronic device applications

PDF

Genome-wide studies reveal the function and evolution of DNA shape

PDF

Reinforcement learning in hybrid electric vehicles (HEVs) / electric vehicles (EVs)

PDF

A signal processing approach to robust jet engine fault detection and diagnosis

PDF

Plasma emission across metal nanoparticle surfaces and semiconductor -liquid interface characterizations

Asset Metadata

Creator Lin, Yung-Hsu (author)

Core Title The physics of pulsed streamer discharge in high pressure air and applications to engine technologies

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Physics

Publication Date 05/01/2014

Defense Date 02/24/2014

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag high pressure streamer discharge,OAI-PMH Harvest

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Gundersen, Martin A. (committee chair), Daeppen, Werner (committee member), Däppen, Werner (committee member), Nakano, Aiichiro (committee member), Steier, William Henry (committee member), Zhou, Chongwu (committee member)

Creator Email yunghsul@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-409881

Unique identifier UC11295318

Identifier etd-LinYungHsu-2481.pdf (filename),usctheses-c3-409881 (legacy record id)

Legacy Identifier etd-LinYungHsu-2481-0.pdf

Dmrecord 409881

Document Type Dissertation

Format application/pdf (imt)

Rights Lin, Yung-Hsu

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA