ABSTRACT
GRAY, PARKER BAHNSON. Ignition of a Liquid Fuel Jet by an Electrical Discharge Propagating along the Jet. (Under the direction of Dr. Alexei Saveliev.)
The desire for more efficient engines motivates the investigation of new ignition
sys-tems. Breakdown through a fluid spray occurs at lower voltages than does breakdown
through air. The reduced voltage necessary for breakdown through a fluid jet or spray
presents the possibility of using such a discharge to ignite the spray. This reduction in
voltage was first seen for a water jet but applies to other fluids as well. Increased
con-ductivity makes such a breakdown easier to achieve. Using such a breakdown mechanism
would ignite the fuel inside the spray instead of at the edge of the spray, giving it the
potential to increase the burning rate.
Exposing certain fluids to an electric field can generates spray. This spray may be used
as an atomization mechanism. The structure of this spray depends on the conductivity
and surface tension of the fluid. Lower surface tension and higher conductivity cause
the fluid to break into droplets more readily. Water and ethanol were found to produce
sprays while kerosene simply dripped. This makes ethanol a good choice for a system
using an electrospray as the atomizer. By using the electric field to produce a spray, the
same electrodes may be used both to produce the spray and to ignite it.
The proposed mechanism ignited the ethanol spray by means of a discharge through
the spray similar to that seen in water. Kerosene ignition was achieved as well. The
preliminary work shows that the ignition system presents a viable alternative to current
c
Copyright 2010 by Parker Bahnson Gray
Ignition of a Liquid Fuel Jet by an Electrical Discharge Propagating along the Jet
by
Parker Bahnson Gray
A thesis submitted to the Graduate Faculty of North Carolina State University
in partial fulfillment of the requirements for the Degree of
Master of Science
Mechanical Engineering
Raleigh, North Carolina
2010
APPROVED BY:
Dr. Tiegang Fang Dr. Gregory Buckner
DEDICATION
BIOGRAPHY
The author spent his childhood in Winston-Salem, North Carolina. He attended high
school at Woodberry Forest School in Virginia. He went to college at Vanderbilt
Univer-sity in Nashville, TN, where he graduated Summa cum Laude with a degree in physics
and mathematics. While at Vanderbilt. he worked with two different groups in the
physics department. One group focused on solid state physics and was growing V O2
nanocrystals. The other group was performing preliminary work and building detectors
for BTeV, a high energy particle physics experiment that was to be built at Fermilab.
The author has always been drawn to scientific disciplines as a way of better
under-standing the world. His work at North Carolina State University has helped to further
this understanding while incorporating practical applications to the knowledge he has
gained. He has presented this work already both at the 2009 Eastern States Section
Meeting of the Combustion Institute in College Park, MD and at the 2010 IEEE
ACKNOWLEDGEMENTS
The author would like to extend his thanks to his advisor, Dr. Saveliev, for his guidance
and help with the research. Without his ideas, none of this would have been possible.
TABLE OF CONTENTS
List of Tables . . . vii
List of Figures . . . viii
Chapter 1 Introduction . . . 1
1.1 Thesis Objectives . . . 1
1.2 Background . . . 2
1.2.1 Classical Spark Ignition and Motivations for New Ignition Systems 2 1.2.2 Capillary Instability of a Fluid Jet . . . 4
1.2.3 Electrosprays and Taylor Cones . . . 5
1.2.4 Jet Surface Charge: A Simple Model . . . 10
1.2.5 Electrospraying as an Atomizer . . . 12
1.2.6 Overview of Relevant Discharge Types . . . 14
1.2.7 Electrical Discharges in Water . . . 18
1.2.8 Discharge through a Fluid Jet . . . 20
Chapter 2 Flow Characterization . . . 22
2.1 Introduction . . . 22
2.2 Calibration of the Pump . . . 24
2.2.1 Procedures . . . 24
2.2.2 Results . . . 25
2.3 Spray Characterization Under No Voltage . . . 27
2.3.1 Procedures . . . 27
2.3.2 Results . . . 28
2.4 Surface Charge on the Fluid Jet . . . 32
2.4.1 Procedures . . . 32
2.4.2 Results: No Fluid Flow . . . 34
2.4.3 Results: Fluid Surface Charge . . . 37
2.5 Charged Jet Breakup Characteristics . . . 55
2.5.1 Procedures . . . 55
2.5.2 Results . . . 57
2.6 Conclusions . . . 70
Chapter 3 Discharge Through a Water Jet . . . 72
3.1 Introduction . . . 72
3.2 Voltage, Current, and Energy for Discharges in Water . . . 73
3.3 Discharge Frequency in Water . . . 85
3.3.1 Procedures . . . 85
3.3.2 Results . . . 86
3.4 Conclusions . . . 89
Chapter 4 Ignition Studies . . . 91
4.1 Introduction . . . 91
4.2 Ethanol: Power Supply and Resistor Only . . . 92
4.2.1 Procedures . . . 92
4.2.2 Results . . . 99
4.3 Ethanol: Pulsed Discharge Ignition . . . 115
4.3.1 Procedures . . . 115
4.3.2 Results . . . 116
4.4 Kerosene . . . 123
4.4.1 Procedures . . . 123
4.4.2 Results . . . 123
4.5 Conclusions . . . 127
Chapter 5 Conclusions . . . 129
Chapter 6 Future Work . . . 131
References . . . 132
Appendix . . . 136
LIST OF TABLES
Table 2.1 Fluid Properties, Physical . . . 23
Table 2.2 Fluid Properties, Electrical . . . 23
Table 2.3 Fluid Properties, Conductivity . . . 23
Table 2.4 Slope of linear fit for current as a function of voltage, in µAkV. . . . 45 Table 2.5 Slope of linear fit for current as a function of velocity, in m/sµA . . . . 45 Table 2.6 Correlation coefficients for tap water with respect to voltage at jet
velocities above 3ms. . . 52
Table 4.1 CMOS Truth Table . . . 97
Table 4.2 Resistance and Capacitance Values for Ethanol Ignition by
LIST OF FIGURES
Figure 2.1 Calibration of the pump: jet velocity of the fluid leaving the needle vs. power setting. . . 26
Figure 2.2 Example image of the jet flow under no applied electric field used
to measure the capillary breakup time. 0.91 m
s jet velocity, or 15.4minmL flow rate. . . 30
Figure 2.3 Length of a tap water jet as it begins to break up as a function of
the velocity of the jet. . . 31
Figure 2.4 Corona discharge currents with no fluid flow. . . 36
Figure 2.5 Current due to surface charge carried by a tap water jet as a
function of increasing voltage. . . 39
Figure 2.6 Current due to surface charge carried by a tap water jet as a
function of increasing jet velocity . . . 40
Figure 2.7 Current due to surface charge carried by a distilled water jet as a
function of increasing voltage. . . 41
Figure 2.8 Current due to surface charge carried by distilled water jet as a
function of increasing velocity. . . 42
Figure 2.9 Current due to surface charge carried by ethanol jet as a function
of increasing voltage. . . 43 Figure 2.10 Current due to surface charge carried by ethanol jet as a function
of increasing velocity. . . 44 Figure 2.11 Measured current for a kerosene jet as a function of voltage. Corona
current under no fluid flow is included to demonstrate that less
cur-rent was measured with a kerosene jet than without. . . 47
Figure 2.12 Subtraction of the corona discharge current for a pan filled with distilled water from the measured current for a distilled water jet. Demonstrates that the corona current is greater than the current carried by the jet. . . 48 Figure 2.13 Measured current from a tap water jet as a function of velocity.
Displays the limited charging rate as velocity increases. . . 50
Figure 2.14 Surface charge on a tap water jet vs velocity vs. jet velocity.
Displays the velocity at which the maximum charging rate occurs. 51
Figure 2.15 Model of the current from a fluid jet, assuming a finite charging time. . . 54
Figure 2.16 Setup for flow characterization experiments. . . 57
Figure 2.19 Images of spray produced by a distilled water jet, 45.96 mm gap width, 0.91 msjet velocity, or 15.4minmL. . . 62 Figure 2.20 Breakup time of a jet of distilled water as a function of applied
voltage, 45.95 mm gap width. . . 63
Figure 2.21 Images of spray produced by an ethanol jet, 39.49 mm gap width, 0.91 msjet velocity, or 15.4minmL. . . 65 Figure 2.22 Breakup time of a jet of ethanol as a function of applied voltage,
39.49 mm gap width. . . 66
Figure 2.23 Images of spray produced by a kerosene jet, 44.94 mm gap width, 0.91 msjet velocity, or 15.4minmL. . . 68 Figure 2.24 Breakup time of a jet of kerosene as a function of applied voltage,
44.94 mm gap width. . . 69
Figure 3.1 Experimental setup used to generate a discharge through a water
jet. . . 74
Figure 3.2 Voltage vs time for breakdown in distilled water at a distance of
24.3 mm. . . 76
Figure 3.3 Current vs time for breakdown in distilled water at a distance of
24.3 mm. . . 77
Figure 3.4 Voltage measurement at which electrical breakdown occurred vs
interelectrode distance for distilled water. . . 79
Figure 3.5 Voltage measurement at which electrical breakdown occurred vs
interelectrode distance for tap water. . . 80 Figure 3.6 Discharge along the surface of a tap water jet, 1.82ms jet velocity,
22.5 kV, 34 mm interelectrode distance . . . 81
Figure 3.7 Average discharge energy vs interelectrode distance for distilled
water. . . 83
Figure 3.8 Average discharge energy vs interelectrode distance for tap water. 84
Figure 3.9 Frequency of electrical breakdown through a distilled water jet as
a function of applied voltage. . . 87 Figure 3.10 Frequency of electrical breakdown through a tap water jet as a
function of applied voltage. . . 88
Figure 4.1 Experimental setup for ignition of ethanol and kerosene. . . 93
Figure 4.2 Diagram of CMOS CD54HCT132E logic circuit inputs and
out-puts used to synchronize the camera and intensifier with the spark. 96
Figure 4.3 Photograph of black box used to encase experiment. The exposed
Figure 4.4 Pulse energies for ignition of an ethanol spray vs resistance be-tween the high voltage supply and needle electrode. The pulse energy continues to decrease as the resistance increases, but may be thought of as relatively constant once the resistor is in place. . 101
Figure 4.5 Breakdown voltage vs. distance for an ethanol spray. . . 103
Figure 4.6 Effective capacitance vs. distance for the initial breakdown pulse
in ethanol ignition. Capacitances are the average value across all velocities measured for each gap width. . . 104
Figure 4.7 Effective capacitance vs. velocity for initial breakdown pulse in
ethanol ignition. Capacitances are the average value across all gap widths for each velocity. . . 105
Figure 4.8 Voltage characteristics for ethanol ignition images seen in Figure
4.8. 1.13ms jet velocity, or 19minmL. 21 mm gap width. . . 107
Figure 4.9 Ethanol ignition displaying the continuous discharge column
re-sponsible for ignition. Times are elapsed time after initial spark. Ignition can be seen in (e) and (f). 0.91ms jet velocity, or 15.5minmL. 21 mm gap width. Voltage characteristics may be seen in Figure 4.8.108 Figure 4.10 Ethanol ignition with a secondary ignition in the center of the gap.
The secondary ignition can be seen beginning in (a) and falls past the lower electrode in (c) before leaving the field of view. Times are elapsed time after trigger point. 0.91ms jet velocity, or 15.5minmL. 25.4 mm gap width. . . 110 Figure 4.11 Voltage characteristics for ignition of ethanol at the lowest
avail-able flow rate, 0.2minmL, and seen in Figure 4.12. 21 mm gap width. 112 Figure 4.12 Ethanol ignition at the lowest available flow rate. Ignition begins
in (c) and (d). The flame replaces the discharge column and is all that is visible in (g) and after. 21 mm gap width. 0.05ms jet velocity, or 0.2minmL. . . 113 Figure 4.13 Circuit between the high voltage supply and the needle electrode
for pulse experiments. . . 116 Figure 4.14 Voltage characteristics for ignition of ethanol by a repetitive pulse
discharge, seen in Figure 4.15. Ignition was seen after the eighth pulse. 20 mm gap width. 0.68ms jet velocity, or 11.6minmL flow rate. 118 Figure 4.15 Ethanol ignition by a repetitive pulsed discharge. The eighth pulse
can be seen in (c), after which ignition below the bottom electrode becomes visible. 20 mm gap width, 0.68ms jet velocity, or 11.6minmL flow rate. . . 119 Figure 4.16 Ethanol ignition with as close to a single spark as was possible.
Figure 4.17 Example of voltage characteristics for ignition of a kerosene jet. 0.68 ms jet velocity, or 11.6 minmL flow rate. 12.5 mm gap width. . . 124 Figure 4.18 Ignition of a kerosene jet. Discharge can be seen in (a) and (b),
followed by beginnings of flame in (c). The flame then falls with the jet over (d) through (h). 0.46 ms jet velocity, or 7.8 minmL. 19 mm gap width. . . 125
Figure A.1 Breakup time of a jet of tap water as a function of applied voltage, 31.84 mm gap width, 0.91ms jet velocity. . . 138 Figure A.2 Breakup time of a jet of tap water as a function of applied voltage,
45.83 mm gap width, 0.91m
s jet velocity. . . 139 Figure A.3 Breakup time of a jet of tap water as a function of applied voltage,
45.83 mm gap width, 1.14ms jet velocity. . . 140 Figure A.4 Breakup time of a jet of tap water as a function of applied voltage,
45.83 mm gap width, 1.37ms jet velocity. . . 141
Figure A.5 Breakup time of a jet of distilled water as a function of applied
voltage, 45.96 mm gap width, 0.91m
s jet velocity. . . 142
Figure A.6 Breakup time of a jet of distilled water as a function of applied
voltage, 45.96 mm gap width, 1.14ms jet velocity. . . 143
Figure A.7 Breakup time of a jet of distilled water as a function of applied
voltage, 45.96 mm gap width, 1.37ms jet velocity. . . 144
Figure A.8 Breakup time of a jet of ethanol as a function of applied voltage,
39.49 mm gap width, 0.91m
s jet velocity. . . 145
Figure A.9 Breakup time of a jet of ethanol as a function of applied voltage,
39.49 mm gap width, 1.14ms jet velocity. . . 146 Figure A.10 Breakup time of a jet of ethanol as a function of applied voltage,
39.49 mm gap width, 1.37ms jet velocity. . . 147 Figure A.11 Breakup time of a jet of kerosene as a function of applied voltage,
44.94 mm gap width, 0.91m
s jet velocity. . . 148 Figure A.12 Breakup time of a jet of kerosene as a function of applied voltage,
44.94 mm gap width, 1.14ms jet velocity. . . 149 Figure A.13 Breakup time of a jet of kerosene as a function of applied voltage,
Chapter 1
Introduction
1.1
Thesis Objectives
The first research objective was to characterize the flow of a fluid jet exposed to high
voltage. Further research relied on an understanding of each fluid’s properties: the
reaction of each fluid to an electric field, the structure of the fluid jet, and the conductive
properties of that jet. The characteristics of the fluid spray needed to be known. The
length of the fluid jet under varying voltage as well as the charge carried on the surface
of the fluid were measured.
The second objective was to determine the properties of electrical breakdown through
a jet of water under the conditions that produced the fluid spray. This would provide a
basis for comparison with any later work on the discharge through a fuel jet to help to
prove that such a discharge was affected by the presence of the jet.
The main objective, then, was to discover whether a discharge through such a fluid
jet at lower electric field strengths than a discharge through air alone could ignite the
cause of ignition. This involved finding the type and energy of the igniting discharge, as
well as whether other discharge types could be responsible for ignition. Also important
were investigating where the ignition occurred in the spray, the ignition delay time, and
what conditions were necessary to sustain a flame after an ignition.
1.2
Background
Ignition of a fluid jet by means of an electrical discharge through the jet delves into
many different fields of physics and engineering. The flow produced by the charged
needle is a type of electrospray with a higher flow rate than a classical Taylor cone-jet.
It has both characteristics of such an electrospray as well as similarities to a the simple
capillary breakup of a fluid jet. The amount of charge that the fluid holds will prove
to be important as well. The discharge along the jet or through the spray has different
characteristics than breakdown through air alone and so discharges both underwater and
through water jets must be compared to what was seen. In addition, a comparison to
spark ignition systems will help to determine what uses this new ignition system may
have.
1.2.1
Classical Spark Ignition and Motivations for New Ignition
Systems
Classical spark ignition has been widely studied and remains one of the most popular
ignition systems. It relies on the use of a spark plug, generally placed in the side of
the combustion chamber, to generate a spark of sufficient energy to thermally ignite the
that energy is not wasted. The minimum ignition energy also depends on the size of the
fuel droplets and is lower at higher equivalence ratios. The energy required is higher for a
longer interelectrode distance because the heating energy is spread over a larger volume.
In the case of kerosene-air mixtures and as a useful order of magnitude estimate for other
fuels, the minimum ignition energy is on the order of 5 to 10 mJ [1, 2].
The goal with a new ignition system, then, is the ability to ignite the mixture in the
center of the combustion chamber in order to increase the probability of ignition and
reduce the burning time. Unfortunately, the turbulent conditions present in real engines
may spread the budding flame kernel away from the areas in which the fuel/air ratio is
close enough to stoichiometric to allow for a successful ignition. By igniting at the center
of the chamber, there is a greater chance that the flame kernel will begin in an area
where the air/fuel ratio is favorable to its continued development, increasing the chance
that the flame will survive. Once ignition is successfully achieved, the flame will spread
outward from the kernel at a finite rate. Ignition from the center of the chamber will
maximize the surface area of this expanding flame such that the total burn time is at
a minimum, increasing the efficiency of the engine by reducing losses due to the second
law of thermodynamics [3].
Ignition by a discharge through an electrospray has the potential to cause ignition in
the center of the chamber from a discharge that spreads through the fuel spray [4, 5]. It
also has the advantage that it does not require a separate atomization system. It may
also allow for decreased total burn times by igniting in the center of the spray, maximizing
1.2.2
Capillary Instability of a Fluid Jet
The work done by Lord Rayleigh remains a central source on the instability of a fluid
jet flowing out of a capillary. Rayleigh’s work indicates that the time before a fluid jet
falling under gravity breaks up is determined by instability modes in the jet that alter
the surface shape of the jet and eventually pinch it apart. If these instability modes
have a wavelength less than the circumference of the jet, then the jet will remain stable;
wavelengths greater than the circumference of the jet will decrease the surface area of
the jet. This happens during the thin parts of the jet surface’s sinusoidal shape, where
the curved shape is caused by the instabilities. This will result in the jet’s breaking into
droplets once the surface area has been reduced enough that surface tension no longer
holds the jet together. The greatest instability occurs when the wavelength is equal to 1.5
times the circumference, and the instability grows in time at a rate that can be described
byexp(qt) [6, 7]. Ashgriz simplified his equations to find that:
r = 1 +δ0exp(ωt−ikz) (1.1)
ω =
I1(k)
I0(k)
(1−k2)k
1/2
(1.2)
k = 2πRinnerλ−1 is the wavenumber, δ0 is the magnitude of the initial disturbance due
to the instability, r is the ratio of the radius of the jet to its undisturbed radius, and I0
and I1 are Bessel functions [8]. From here he found that the breakup time, tb, is equal
to:
R =
1−1
2δ
2 0
1/2
(1.4)
Applying these to the 0.6 mm inner diameter and 0.9 mm outer diameter needle used,
assuming the maximum instability mode ofλ= 1.5∗circumf erence and that the initial
magnitude of the instability is equal to:
δ0 =
Router Rinner
−1 (1.5)
the predicted breakup time is 5.64 s. This will be compared to the measured breakup
time in Section 2.3. Note that increasing the initial disturbance from a 0.9 mm radius to
a 1.0 mm radius decreases the breakup time by a factor of 2 to 2.47 s, showing that even
a slight deviation from this estimated initial disturbance can have a very large effect on
the predicted breakup time.
1.2.3
Electrosprays and Taylor Cones
When a fluid droplet is sufficiently electrically charged, the repulsive internal electrical
forces between the charges on the fluid will exert an outward pressure that will overwhelm
the inward pressure of the surface tension, breaking apart the droplet. This will occur
when:
q >
q
16πa3
0σ (1.6)
whereq is the total charge contained in the droplet,a0 is the droplet radius, andσ is the
surface tension [9]. When the droplet is at the exit of a capillary and electrical forces pull
to induce at least charge q above, this breakup will take the form of a jet of fluid off of
the droplet. In this case, the fluid droplet outside the capillary’s aperture will take on a
conic shape, a so-called Taylor cone, with the jet of fluid exiting from the tip of the cone
[10]. Taylor discovered that this jet mechanism was true for conducting fluids, whereas
nonconducting fluids were drawn out of the capillary as single droplets. In subsequent
years, de la Mora further examined the effect of conductivity on the structure of such
Taylor cones. He discovered that the charge carried out of the jet in a particular time,
which can be thought of as a current, becomes independent of voltage at higher voltages
in fluids with conductivities greater than 100cmS . This is due to the face that while
the charge per volume continues to increase with voltage, the expelled volume seems
to decrease. This occurs as the emitted jet becomes progressively narrower at higher
voltages [11]. De la Mora determined that the current can be described as follows:
I =f(x)pσKQ/ (1.7)
where I is the current, f(x) is an empirical function determined by their experiments,
K is the conductivity, Q is the flow rate, and is the dielectric constant. Cherney
expanded on this work and found that the density and viscosity have little effect on the
structure of the cone-jet so long as they are large compared to the capillary and meniscus
surface tension forces. Density affects velocity and surface charge but not jet radius, and
viscosity only affects velocity inside the capillary [12]. In other words, the properties that
determine the shape of the cone-jet are only the electrical properties and surface tension.
Once the initial jet has formed, various instabilities in addition to the capillary
insta-bility assert themselves. One possiinsta-bility is that the breakup takes the form of a simple
instability. In the case of high surface charge, however, this is a different mechanism
than the capillary instability. The droplets’ charge is sufficiently high, as described in
equation 1.6, for them to then break up further into smaller droplets. This breakup mode
reaches a limit with fluids of very low conductivities, such as kerosene. Such
nonconduc-tive fluids simply drip rather than forming a jet. The droplets they form are smaller than
the capillary from which the fluid flows. The electrical forces have difficulty drawing the
fluid out of the capillary, resulting in very low flow rates [13]. The lack of charge mobility
due to effectively zero conductivity is what prevents forming and sustaining a jet driven
by the flow of charge out of the meniscus. A second possibility is that the jet will begin
to develop ”kinks” or potentially even swirl around its axis with a progressively larger
radius. In a third breakup mode, the jet emits smaller jets off of it in a Christmas-tree
like pattern. These smaller jets then break into tiny droplets [14]. The third pattern was
confirmed for ethanol by Dunn, who also discovered that increasing the surface charge
will increase the number of tiny jets off of the main jet. Further down the spray, the
charge present on the droplets formed by these jets resembles a Fermi-Dirac distribution
[15]. Such a distribution implies that each droplet can be thought of as having energy
lev-els that are open to ions. The addition of each ion, then, occupies one of the energy levlev-els
and makes the addition of another ion to that droplet less favorable than to a different
droplet that does not yet have that energy level filled. It also implies that the droplets
should all be roughly the same size in order to minimize the energy and maximize the
entropy of the system. This will prove to be important in the use of electrosprays as an
atomization method for combustion.
Taylor cone-jets have been applied to textiles as a means of spinning highly viscous
polymers into threads. In modeling such spinning, Koombhongse and Yarin [16] derived
resulting jet. For fluids where viscosity is a factor, the bending (or kink) perturbations
will grow much faster than the capillary instabilities if:
πµ2
ρλ2ln(L/a
0)
>>1 (1.8)
where λ is the charge per unit length, L is the length of the jet before breakup, µis the
viscosity, and a0 is the radius of the jet. In all fluids used in the experiments herein,
surface tension will prove to be much more important than viscosity. For such a case,
the electrical repulsive forces will overwhelm the surface tension if:
λ2ln(L/a0)> πa0σ (1.9)
Here again one sees the same sort of relationship as in Equation 1.6 for the Rayleigh
instability of a charged droplet, this time derived for a cylinder of fluid instead of a
droplet. As before, the maximum charge is merely a function of the jet size and the
surface tension.
The length before breakup of a charged jet depends on both the stress due to the
mutual repulsion of charges on its surface and also a second stress from the presence of
the electric field produced by the charging electrode. According to Panofsky, this field
exerts stresses on the liquid both parallel and transverse to the field that are equal to:
0E2/2 (1.10)
An external electric field exerts a volume force on a dielectric such that:
F =ρeE−
0
2∇+
0
2∇
E2d
dρρ
(1.11)
where ρe is the charge density in the volume. The Clausius-Mossotti relation provides a
relationship between and ρ:
−1
+ 2 =
N0ρα
3M (1.12)
whereN0 is Avogadro’s number, M is the molecular weight, and α is the polarizeability.
This can be used to find ddρ, such that:
d
dρ =
N0α
9M (+ 2)
2
(1.13)
Panofsky used this relation to derive the pressure difference between two regions of
a dielectric liquid, but the same principle applies for the total volume force [17]. This
leads equation 1.11 to become:
F =ρeE−
0
2∇+
0
2∇
E2(+ 2)(−1)
3
(1.14)
The first term describes the force on the charge carried by the jet. The second can
be thought of as acting outward from the center of the jet along the surface of the jet.
It exists where changes from the value of the fluid to the value of air and so acts in
the opposite direction of the surface tension. The third term depends on the change in
the electric field. Some simplification of this term will aid in understanding the jet. If
we look at the component of the force purely down the axis of the jet and assume thatρ
equation becomes:
F =ρeE+
0
6(+ 2)(−1)∇(E
2) (1.15)
The first term will generally be a repulsive force downward from the charging electrode
and will simply increase the velocity of the jet. This increase in velocity, of course, will
result in a decrease in the jet’s radius as it falls and thus will facilitate breakup. Keeping
in mind that this ignores the boundary layer between the fluid and air, the second term
describes the stress on the jet due to the external electric field. The more pronounced
the inhomogeneities in the electric field, the greater this force becomes.
According to calculations by Thong and Weinberg [18], the enhanced electric field
strength about a small tip charged to a particular voltage is equal to:
E =
√
2V0
rln(4z0/r)
(1.16)
where z0 is the interelectrode distance and V0 is the applied voltage. The smaller the
needle and larger the ratio between the interelectrode distance and needle radius, the
greater the electric field enhancement around the needle tip. A strong localized electric
field about the needle tip will increase the field gradient as the jet flows away from the
needle tip, increasing the stress on the jet and facilitating breakup into droplets.
1.2.4
Jet Surface Charge: A Simple Model
The jet used in experiments differs somewhat from the electrospray jets described in
Section 1.2.3 in that the flow rate is much higher. For such a jet, one simple model that
null at the center of the jet. The jet is modeled as an infinitely long cylinder, allowing
edge effects to be ignored. Starting with Gauss’s law, as well as the definition of electric
potential:
Z Z
E·dA= q
0
(1.17)
−∇V =E (1.18)
where V is the electric potential and A is the area. From here, it follows that:
E = q
0A
= q
2π0rl
(1.19)
wherer is the radius of the cylinder and l is its length. The linear charge densityλ may
be defined here as q/l. The potentialV may be found by integrating with respect to the
radius r, the only direction in which the field will vary:
V =−
Z
E·dr =−
Z λ
2π0r
=− λ
2π0
[ln(r)−ln(r0)] (1.20)
If we set the potential V = 0 at r0 → ∞,, then the linear charge density is found to be:
λ = −2π0V0
ln(rs)
(1.21)
where V0 is the potential at the surface, v is the jet velocity, and rs is the radius of the
surface of the cylinder. The surface charge will thus approach a singularity as r → 0.
Because this cylinder is moving as the surface of the fluid jet, a current may be defined
I =λv= −2π0V0v
ln(rs)
(1.22)
This current is what is largely looked at experimentally. Plugging in values forπ,0, and
rs = 0.0003m, this equation reduces to:
I =
6.85∗10−12 A∗s
V ∗m
∗V0∗v =
0.00685 µA∗s
kV ∗m
∗V0 ∗v (1.23)
This model will be compared to measured values in Section 2.4.
1.2.5
Electrospraying as an Atomizer
Based on the work of Dunn and Snarski [15] described in Section 1.2.3, one may conclude
that electrospraying would be a convenient way to produce an atomized spray. This is
indeed the case, and electrospraying has been studied for applications in combustion
systems. It has been found that such a spray produces somewhat evenly sized droplets
far downstream, once the droplets reach a form of electrostatic equilibrium. The spacing
between the droplets is due to the field produced by the charge on the droplets. The
rate of acceleration of the expansion of the area containing the droplets will decrease as
the spray travels farther downstream [19]. Increasing the applied electric field strength
will thus increase the spreading distance of the charged spray [20]. Gomez and Tang
found that larger charged droplets break up more readily than smaller droplets, even
when the charge-to-volume ratio of the larger droplets is lower than that of the smaller
droplets [21]. Droplet size should thus asymptotically approach a minimum such that,
given sufficient time, the droplets should be about the same size. Together, these findings
The conductivity of the fluid is important in producing a useful atomized spray for
much the same reason as it matters in producing a Taylor cone-jet. The ability to produce
a jet seems essential in that it helps to initiate further breakup into smaller droplets.
Without a jet, the fluid will simply drip as single droplets and will not spread outward.
This presents a problem considering that the majority of widely used hydrocarbon fuels
are nonconductive, minimizing the effect that electric forces have on them. One solution
to this problem is to add an anti-static additive to the fuel [18]. This will increase its
conductivity such that normal electrospraying becomes possible, but it unfortunately can
also limit the choice of fuel and increase the cost of that fuel. A second popular solution
is to take advantage of the enhancement to electric field strength about a sharp tip
described in equation 1.16. The placement of a sharp nonconductive tip, such as a glass
filament, results in an enhanced charging rate of the nonconductive fluids and dispenses
with the need for any additives [20, 22, 23]. For such a setup, using a smaller aperture
and higher flow rate results in more charge being added to each volume of liquid, leading
to better atomization [20].
For more conductive fuels, such as ethanol and butanol, no further enhancement of the
spraying needle is necessary. Ethanol has already been seen in Section 1.2.3 to produce
an electrospray. Butanol can be used in much the same way, producing a spray similar
to that of ethanol in a Taylor cone-jet setup [24]. The spray produced is unfortunately
not entirely evenly sized, but effective atomization can still be achieved in this way. In
much the same way, hydrocarbon electrosprays were found to be not perfectly evenly
sized either. At the center of the spray are larger droplets with a smaller charge to mass
ratio, while smaller droplets with a higher charge to mass ratio spread out to the edges
[20]. The variation in droplet size should not be overstated, however.
version of the electrosprays described in the literature. Many of the instabilities that
exist are very similar to those in Taylor cone-jets, but simply on a larger scale. This
facilitated the investigation of a surface discharge along or through these sprays of the
type described in Section 1.2.8.
The presence of charge in the fuel has an effect on the flame shape and burning speed.
Positive charge on a fuel droplet serves to hold the flame together into more of a spherical
shape, counteracting the buoyancy of the heated products [25]. The downward force on
the charged particles from the field produced by the anode is the dominant effect in
this case. The addition of negative charge, on the other had, causes the flame to be
more turbulent than an uncharged flame [26]. This may be due to the positive charge
affecting the soot primarily, effectively only adding a charge to the products [25]. The
excess negative charge, on the other hand, is not as easily cast aside and so affects the
flame shape itself. One beneficial effect of excess negative charge, however, is that it
seems to result in reduced NOx production as the excess electrons are released when the
droplets evaporate [23]. The electrons here may simply act as a competitive oxidizing
agent, reducing the probability that a given nitrogen molecule will react with an oxygen
radical by taking its place.
1.2.6
Overview of Relevant Discharge Types
The Townsend discharge occurs silently and invisibly in the majority of situations. It is
a low power, low current discharge that occurs at electric field strengths that are too low
to cause a self-sustained discharge in the gas. According to Fridman and Kennedy [27],
i= i0exp(αd)
1−γ[exp(αd)−1] (1.24)
where d is the gap distance, dnedx =αne where ne is the free electron number density and
α describes the rate of electron formation, and γ is a function of α such that:
αd =ln
1
γ + 1
(1.25)
The Townsend discharge thus occurs when the denominator in Equation 1.24 is
nonzero, meaning when the electron formation rate is sufficiently low. The value ofα as
a function of the reduced electric fieldE/phas been determined for many substances. It
can be calculated using the equation:
α
p =Aexp
− B
E/p
(1.26)
and experimentally determined values ofA and B. Values of A and B are 15cm·1torr and 365cmV·torr for air, and 13cm·1torr and 290cmV·torr for water vapor. These equations describe the beginnings of an electron avalanche, but below the field strength at which a streamer
could form. It should be noted that they ignore electron recombination with gases, but a
recombination rateβmay be added to reduceαsuch that dnedx = (α−β)ne[27]. Streamers
will begin to form once the space charge left behind by such an avalanche is sufficiently
high (generating a potential equal to that of the anode) to sustain itself to the cathode
[28]. A spark is formed when a streamer reaches the cathode. The cathode, acting as an
electron reservoir, provides electrons that begin the propagation of a negatively charged
spark in the opposite direction, along the now positively charged path left behind by the
own space charge and by the attractive field left behind by the original streamer, causing
it to move at much higher speeds than the streamer did [27].
Electrical breakdown in an aerosol operates under similar principles to one in the gas
phase, except that charge attachment to and ionization of the aerosol particles must be
taken into account as well. Following the work of Karachevtsev and Fridman [27, 29, 30],
the electron formation rate in such a plasma depends not only on the simple electron
formation and recombination rates with molecules in the gap, but also on collisions
between electron or ions and the aerosol particles that may produce or absorb an electron.
Collisions between aerosol particles and photons emitted by electron-molecule collisions
(or emitted by absorbed electrons) may result in additional electrons as well. In the case
of a plasma in an aerosol, the Townsend limit becomes:
γef f
exp[α−β]d−1 + β
α−β[exp(α−β)d−1]
= 1 (1.27)
where α and β are the Townsend coefficients of detachment and attachment discussed
above, d is the gap distance, and γef f takes into account the production of electrons
from secondary collisions, primarily with the aerosol. Because α > β always, increasing
γef f will result in a decrease in the breakdown voltage. The only difference from the
Townsend discharge in a gas, then, is in the γef f term:
γef f =γikexp
−naσiad
νi
2νi d
+γia
1−exp
−naσiad
νi
2νi d
+γνk
µ(1−ν)
ki
+γνa µν
ki
(1.28)
where γik is the probability that a collision between an ion and aerosol will produce an
and aerosol,νi is the ion RMS velocity,νi
dis the ion drift velocity,γikis the probability of
a collision between an ion and the cathode producing an electron, γνk is the probability
that a collision between a photon and the cathode will produce an electron, µ is the
probability of emitting a photon in an electron-molecule collision, ν is the probability
that that photon collides with an aerosol, ki is the probability that an ion will form in a
collision between an electron and a gas molecule, andγνais the probability that a collision
between a photon and aerosol will produce an electron. The third term assumes that all
photons that do not collide with an aerosol reach the cathode, whereas the cross-sectional
area of the cathode is very small compared to the total area toward which the photons
can travel. Thus, the term may be ignored. The fourth term will increase as the ratio
of the aerosol’s area to the total photon flux’s travel area increases. In the setup used,
the total volume of all aerosols remains constant, so this term will increase as the size
of each particle decreases. The second term will increase in the same manner, as more
ions collide with the aerosols, but this will in turn decrease the first term. Considering
that the first term, concerning electron formation at the cathode, is likely already small
due to ions being lost to drift away from the cathode (and this drift is increased by the
presence of positive charge on the aerosols), the first term may be considered to be very
small.
The space charge of the charged aerosols will increase the effective electric field in
the gap, which will increase the effective electron formation rate and thus α. It will
increase the recombination rate β as well, but because β depends on a second collision
with non-unity probability by the newly formed electron, α will increase more than β
and the charge on the aerosols will, again, result in an increased electron production rate.
This, in turn, will decrease the voltage for the Townsend limit. The field produced by
odds of a collision in the second term and also the odds that that collision will produce an
electron due to the higher available energy. The result is that, though uncharged aerosols
often increase the breakdown voltage due to the reduced probability that charges will
traverse the gap, charged aerosols result in a decrease in the breakdown voltage as a
result of increased electron formation.
1.2.7
Electrical Discharges in Water
Electrical discharges in water, in particular underwater discharges, have become the
subject of much research in recent years. Water purification has been the main goal of
many of these studies, utilizing the UV radiation and radical species generated by the
breakdown to decompose contaminants [31, 32, 33, 34]. The field at which breakdown
occurs in water is about 30 times the breakdown field in air, however. This is in large
part due to the low probability of an electron avalanche on account of the much shorter
mean free path available to electrons in a liquid. In addition, the high dielectric constant
of water decreases the intensity of localized electric fields, further limiting electron energy
gain. The mechanism behind the breakdown in water is thus still not entirely understood,
but the dominant theory is that it is actually the end result of discharges in small gas
bubbles within the discharge path [35]. These bubbles may be pre-existing gas pockets
within the liquid or may be formed when the pre-breakdown current in the liquid heats
and vaporizes some of the liquid. The bubbles then grow as the current continues to heat
the liquid until they are large enough for a breakdown avalanche to form in the gas [36].
Even in bubbles within the water, however, the cause of breakdown does not seem
to be a pure gas-phase electron avalanche. In de Baerdemaeker’s study of underwater
a large gas bubble intentionally placed in the discharge path in water, the majority of
the voltage drop occurred in the bubble. This, along with the fact that the presence
of a large bubble reduced the breakdown field strength by up to 2 orders of magnitude,
lends credence to the idea that water discharges are still predominantly gas phase
dis-charges. Gershman also discovered that charge accumulates along the surface of the gas
bubble while voltage is applied to the system. A second discharge caused by this charge
propagates back in the opposite direction across the bubble’s surface after the voltage is
turned off. The reduction in the field strength necessary for breakdown makes the whole
process similar to a dielectric barrier discharge [31]. These findings were later seen by
Bruggeman in studying bubble discharges in capillaries. He observed that the breakdown
electric field decreases as the length of the bubbles in the capillary increases. The
break-down field in the bubbles was lower than that in air, confirming that a dielectric barrier
discharge along the surface is a likely mechanism [38].
For underwater discharges, the effect of conductivity seems to be to increase the
breakdown voltage. This happens in part because the more conductive water provides a
path for current to flow in the absence of breakdown, making the buildup of higher charge
concentrations and localized fields more difficult. According to Sunka, it also is due to
the fact that the ions dissolved in the water act similar to the free charge in a conductor,
counteracting some of the electric field due to the localized space charge at the heads
of streamers [28]. Clements discovered, however, that higher conductivity results in an
increase in the number of long magenta streamers. The difference here may be due to
the fact that Clements discovered that a positive applied voltage results in much more
and longer streamers than a negative applied voltage [36]. It will be shown in Section 3.2
that conductivity does decrease the breakdown voltage in the discharge along the surface
findings. It must be stated, however, that the fact that increased conductivity decreases
breakdown voltage in the fluid jet case does not necessarily indicate that the same is true
for underwater discharges. In fact, the fluid jet does not provide a direct contiguous link
between the needle and ground electrodes in the experiments described herein, meaning
that the reduction in effective voltage due to current through the conductive fluid is not
applicable in the fluid jet case. The increased streamer production still applies, then, but
the detrimental effects on the voltage do not. This makes conductivity a net benefit for
the surface discharge along such a jet.
1.2.8
Discharge through a Fluid Jet
In studying the mechanism behind lightning in 1931, Macky discovered that water
droplets traveling through an electric field are elongated by the field. Tiny filaments,
very much like Taylor cone-jets, propagate from each end of the elongated droplet and
facilitate a discharge through the drop between the two electrodes forming the field. A
high enough electric field will virtually always result in a spark though the falling drop,
and the interelectrode voltage decreases as the drops become larger [39]. Water drops
were once again found to facilitate a discharge along a long path, guided by the stream of
drops, by Takaki in 2008. His experiment used highly conductive tap water that traveled
between two high voltage electrodes, resulting in a surface discharge along the water
droplets that then hopped between the droplets to traverse a 50 cm gap [40]. In both
cases, it is clear that the water provides a means of reducing the breakdown voltage
between the electrodes and helps to provide a path for the breakdown that occurs.
Shmelev [32, 41] studied the properties of a discharge through a contiguous jet of
from the needle through a grounded ring electrode. Once the capacitor charged to a
sufficiently high voltage, a discharge propagated along the surface of the water jet and
the resulting shock blew the jet into small droplets. The discharge then repeated and
blew apart the jet again once the contiguous jet had reformed the connection between
the two electrodes. In this case, the charging time of the capacitor was much less than
the travel time of the water between the electrodes, making the jet formation the cause
of the periodicity. It should be noted that the voltage used Shmelev’s his setup to charge
the needle was negative, unlike the positive voltage used for the experiments in Chapters
3-4.
Shmelev found three different discharge modes for the jet: a glow discharge mode,
a surface discharge, and an arc discharge. The important one to consider will be the
surface discharge, which resembles a dielectric barrier discharge. In such a discharge,
imperfections in the surface of the dielectric, such as the instabilities in the fluid jet
discussed in Section 1.2.2, locally enhance the electric field and cause the breakdown
voltage along the surface of the dielectric to be greatly reduced as compared to the
breakdown voltage in air [27]. The avalanche that results in this discharge still occurs in
the gas next to the dielectric, but it is aided by the localized field enhancements caused
by the presence of the dielectric. This low power surface discharge dissipates little of its
Chapter 2
Flow Characterization
2.1
Introduction
A fluid jet will behave differently when subjected to electrical forces than when simply
falling downward. This behavior resembles that described for Taylor cone-jets in Section
1.2.3. Depending on the fluid’s properties, it may produce a simple jet similar to how it
would behave under no electrical forces, may develop a spiraling jet that eventually breaks
up into droplets, or may simply break into a spray with a conic shape. The behavior of
conductive tap water, much less conductive distilled water, ethanol, and kerosene with
no anti-static additives will be examined. These fluids will pass through a needle that
is positively charged to several kilovolts and then travel downward through a grounded
ring. This will allow a qualitative comparison of these macroscopic, pump-driven jets to
the characteristics of Taylor cone-jets that are driven almost entirely by electrical forces.
Such a comparison will demonstrate how experiments performed on the discharge and
on ignition in Chapters 3-4 should apply to smaller scale jets and electrosprays.
Table 2.1: Fluid Properties, Physical
Fluid Density (ccg) Viscosity (cP) Surface Tension (dynecm )
Water 1.027 0.911 73.56
Ethanol 0.787 1.057 23.39
n-Dodecane 0.745 1.390 24.94
n-Tridecane 0.754 1.718 25.5
n-Tetradecane 0.758 2.110 26.68
n-Pentadecane 0.765 2.558 26.68
Table 2.2: Fluid Properties, Electrical
Fluid Dipole Moment (Debye) Dielectric Constant Ionization Energy (eV)
Water 1.85 79.99 10.47
Ethanol 1.69 25.02 12.612
n-Dodecane 0 – 9.75
n-Tridecane 0 – 9.72
n-Tetradecane 0 – 9.72
n-Pentadecane 0 – 9.68
Kerosene – 2.086 –
Table 2.3: Fluid Properties, Conductivity
Fluid Conductivity (cmµS)
Tap Water 213+11
Distilled Water 1.3+0.2
Ethanol 0.09+0.01
n-alkanes listed are included as representative of the major constituents of kerosene. The
majority of these data are from the Chemical Properties Handbook [42]. The n-alkane
ionization energies are from experimental data obtained by Zhou et al. [43]. Water and
ethanol ionization energies are from Lange’s Handbook of Chemistry [44]. Dielectric
constants were experimentally determined for ethanol and water by Mohsen-Nia et al.
[45] and for kerosene by El-Sharkawy et al. [46]. Fluid conductivities were measured
using an Oakton CON400 Series conductivity meter.
2.2
Calibration of the Pump
2.2.1
Procedures
The pump used in all experiments, an Ismatec Reglo-z, gave no indication of the fluid
flow rate but rather simply displayed a power setting as a percentage of its maximum
power. This meant that a calibration had to be performed to correlate the fluid flow
rate with the displayed power setting. A simple function describing this correlation was
found and used for all subsequent analysis.
The procedure for determining the fluid flow rate involved much the same setup that
will be used for all experiments, in particular the flow characterization under no applied
voltage described in Section 2.3. The pump input was connected to a fluid source bottle.
Its output fed into a tube leading to the 0.6 mm inner diameter and 0.9 mm outer
diameter source needle that was used for all experiments. The pump calibration was
performed using tap water. The fluid flowed out of the needle into a large graduated
cylinder, and the time required to feed a particular volume into the graduated cylinder
for the measurements to be taken in a practical amount of time for lower pump settings
while increasing precision at higher settings. The lower amount of water used to calibrate
at the lower end contributed to the greater variance of points at lower settings, as the
exact time elapsed to feed the fluid into the cylinder was harder to determine. This is
because the water adheres to the sides of the cylinder and a few drops on the sides of the
cylinder rather than in the pool of fluid at the bottom can change the measured time by
several seconds at such low flow rates. Also, the water level in the cylinder passes the
line indicating the amount of water measured much more slowly at low flow rates than at
higher flow rates; the slow movement of the water level made precise time measurements
more difficult with the equipment available.
2.2.2
Results
Figure 2.1 shows the jet velocity vs the pump power setting, where the jet velocity is
determined by:
v = Q
A =
4V
πd2t (2.1)
Where v is the jet velocity, Q is the flow rate, A is the area of the needle aperture, V is
the volume collected, d is the inner diameter of the needle, and t is the time over which
the volume V was collected.
The relationship between the pump power setting and jet velocity is assumed to be
linear, with the y-intercept assumed to be zero because no fluid will flow when the pump is
turned off. The slope B, its uncertaintyσB, and the uncertainty in each point (assuming
that each point has the same uncertainty) σy may be found by using the equations:
B =
PN
i=1(xiyi) PN
0 10 20 30 40 50 60 70 80 0
0.5 1 1.5 2 2.5 3 3.5
Pump Power Setting
Jet Velocity, m/s
Figure 2.1: Calibration of the pump: jet velocity of the fluid leaving the needle vs.
σy = v u u t 1
N −1
N
X
i=1
(yi−Bxi)2 (2.3)
σB =
σy
q PN
i=1x2i
(2.4)
where xi and yi in this case are the pump power setting and jet velocity at that power
setting, respectively, and N is the total number of measurements, in this case 24 [47].
The slope of the fit seen in Figure 2.1 of the jet velocity vs power setting is thus
0.0456+0.0006ms, and the uncertaintyσy in each point is 0.13 ms. As can be seen from the
graph, this is likely an overestimate of the actual uncertainty due to the lower precision
of measurements at lower flow rates. The real imprecision in the pump’s flow rate should
be much lower than reported.
2.3
Spray Characterization Under No Voltage
2.3.1
Procedures
To measure the breakup time due to the capillary instability under no electric field, a
similar setup was used as to calibrate the pump. Tap water flowed from the Ismatec
Reglo-z pump to the 0.6 mm diameter stainless steel needle. The water traveled directly
down from the needle into the collection pan.
An Allied Vision Technologies Stingray camera was used to observe the spray. This
camera allowed for 1 megapixel images at exposure times of a few tens of microseconds.
This allowed the viewing of the jet at a high enough shutter speed that individual drops
To find a quantitative description of the flow, the vertical length of the jet was
mea-sured. This vertical length is the distance directly down from the needle tip to the point
at which the jet begins to break up into droplets. These measurements were made by
counting the pixels in the images of the jet. To calibrate the length scale, a ruler was
placed behind the needle in the field of view of the camera. The ruler was only about a
centimeter behind the needle while the lens’s focal distance is 1 meter, so the ruler can
be said to be effectively in the same plane as the needle and jet. The number of pixels
that corresponded to one millimeter could then be counted.
Multiple images, such as the image for a jet speed of 0.91 ms seen in Figure 2.2, were
taken for each flow rate. The sets of images were constructed by setting the camera to
record 12-22 images in sequence. What this involved was the camera being set to run at
15 fps and save every other image. The total saved images, then, numbered 12-22 and
all but two of the 41 flow rates measured had 17 images taken for them.
2.3.2
Results
Beginning at a flow velocity of 0.82 m/s, the jet breakup length depends linearly on the jet
velocity. Each jet length plotted in Figure 2.3 is the average length of all measurements
made at that velocity. The length is assumed to be normally distributed, such that the
uncertainty σx in the average length ¯x across N measurements is equal to:
σx=
v u u t
1
N(N−1)
N
X
i=1
(xi−x)¯ 2 (2.5)
This may be used to find a linear fity =A+Bxthat weights each point by its uncertainty
A =
P
(wx2)P
(wy)−P
(wx)P
(wxy)
P
wP
(wx2)−[P
(wx)]2 (2.6)
σA =
s
P
(wx2) P
wP
(wx2)−[P
(wx)]2 (2.7)
B =
P
wP
(wxy)−P
(wx)P
(wy)
P
wP
(wx2)−[P
(wx)]2 (2.8)
σB =
s
P
w
P
wP
(wx2)−[P
(wx)]2 (2.9)
whereσAandσB are the uncertainties in A and B, respectively [47]. The slope of this
lin-ear relationship is 39.67+0.39mmm/s, and it crosses the y-axis at a length of 9.30+0.56mm.
This slope is actually a time value of 39.67 ms, which is the breakup time of the jet.
The fact that the breakup length depends linearly on the jet velocity indicates that the
breakup time is constant, as would be expected if it is due to the instability modes
de-scribed in Section 1.2.2. The length increase is only due to the longer distance that the
jet travels in the breakup time. When the breakup time was calculated using Equations
1.2 through 1.5 and assuming that the initial disturbed amplitude of the jet was equal
to the outer diameter of the needle, the breakup time was calculated to be 5.64 s. This
overestimated the breakup time by a factor of about 15 by underestimating the initial
disturbance amplitude. Solving for the initial disturbed amplitude using the
experimen-tally determined breakup time yields an initial disturbed diameter of 1.34 mm, which is
1.5 times the outer diameter of the needle.
At velocities less than 0.82ms, if we consider the breakup to be because of instabilities
0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 0
10 20 30 40 50 60 70 80 90
Jet velocity, m/s
Jet Length at Breakup, mm
greater. This is the transition region from the flow simply falling out of the needle in
droplets at lower flow rates to the formation of a jet at higher flow rates. The initial
disturbance amplitude, then, must be higher. One may think of this phenomenon as a
string of droplets coming out of the needle that are close enough together that they form
a contiguous jet. At higher velocities, this effect is smoothed out more because the fluid
has less time next to the needle for its surface tension to cause it to begin to form into
a droplet.
In order to yield results that were not heavily affected by the transition region in the
flow, the majority of later measurements were taken at jet velocities greater than 0.82ms.
2.4
Surface Charge on the Fluid Jet
2.4.1
Procedures
Measuring the surface charge carried by the fluid jet required measuring the current
flowing from the collection pan to ground. The basic setup was largely the same as
described in Section 2.3.1, with a few changes. The collection pan had the dimensions
of 30.48 cm long, 17.78 cm wide, and 7.62 cm tall. The pan was placed on top of either
a 0.75 cm thick sheet of nylon, electrically isolating it from the grounded optical table
surface. Given that the measured voltages were all less than 1 V and thus insufficient
to produce any sort of appreciable discharge, the fact that the sheet of nylon did not
cover the entirety of the gap between the pan and grounded table should not reduce the
measured current. A resistor connected the pan to ground, and the voltage across this
resistor was measured using a Extech Multimeter 420 so that the current carried by the
did not survive its use in preliminary ignition trials, so it was later replaced with a 1 MΩ
resistor.
Measurements were made for tap water, distilled water, ethanol, and kerosene. The
pan was half-filled with tap water for the ethanol measurements as a safety precaution
against accidental ignition. For the other fluids, the pan was filled with just a thin layer
of the fluid that ran into it before measurements were taken. As a control condition,
the current was measured with no fluid flow as well. The conditions for no fluid flow
were for the empty pan, the empty pan with a screen on top, the pan half filled with
either tap water or distilled water, the pan bottom coated with a layer of kerosene, and
the pan bottom coated with a layer of kerosene with the screen over the top of the
pan. The screen was present in kerosene ignition experiments in order to prevent a pool
fire. Measurements were taken with the needle mount charged from 1 - 9 kV. At higher
voltages, the discharge became less stable and so made the voltage across the resistor too
erratic to measure. Three sets of such data were taken for each fluid so as to compare
different flow rates. To measure the effect of flow rate more closely, the flow rate was
varied while the voltage was held constant at 3 kV, 5 kV, and 7 kV for each fluid as
well. The flow rates corresponded to jet velocities of between 0.684ms and 1.64ms for
these experiments.
Further investigation was performed to find whether there existed a limit to the
charging mechanism. The measurements made for tap water were further extended at 3
kV, 5 kV, and 7 kV far past the highest velocities used for the other fluids. The highest
velocity used in these trials was 3.88ms, which is near the pump’s upper limit. Another
set of trials were taken with the velocity held constant, as before, but again with much
2.4.2
Results: No Fluid Flow
With no fluid flow, a measurable current out of the pan started at just above 6 kV. This
is the result of a corona discharge, beginning in the enhanced field region surrounding
the needle and ending at the pan. The magnitude of this discharge increases with voltage
and was measured between 6 kV and 10 kV. Many of the voltages for which no current
could be measured are omitted from any plots for the sake of clarity, as seen in Figure
2.4. The uncertainty in all voltmeter measurements was assumed to be either + 0.002 V
or 5%, whichever is greater. This uncertainty in voltage propagates through for current
such that:
I = V
R →δI =
∂ ∂V V R
δV = 1
RδV (2.10)
where r is the radius of the needle, v is the jet velocity, I is the current, V is the measured
voltage on the voltmeter, and R is the resistor across which voltage measurements were
taken.
It is immediately clear that the presence of water in the pan, whether more conductive
tap water or distilled water, greatly reduces the discharge current and increases the
minimum voltage necessary to initiate such a discharge to a measurable level. The
insulating effects of the water required that the voltage increase from 6.3 kV to 7.6 kV,
an increase of 23%, before a measurable discharge began. It is unclear, then, whether
any current flowed through the water at all. The pan was only filled halfway with water,
meaning that the conductive sides remained exposed. This would greatly reduce the
ground electrode’s surface area, decreasing it by 71% from 1276 cm2 to 367 cm2. That
no charge flowed into the water does not seem likely considering that the difference in
the effective electric field toward the parts of the pan that contained the water, reducing
the total measured discharge current while not eliminating charge flow into the pan. One
must also consider that, with much of the pan covered by a dielectric, the other grounds
surrounding the needle may become more significant. These grounds include the stage
and the optical table. Current continues to flow, but less of it goes to the pan.
The pan filled with tap water registered a slightly higher discharge current than the
pan filled with distilled water, indicating that the conductivity of the fluid in the pan
may play a small role. The uncertainties in the measurements for distilled water and
tap water overlap, however, so these results may be misleading. Assuming they are
indicative of a trend, however, indicates that greater conductivity seems to facilitate the
discharge by reducing the impact of covering so much of the pan’s surface area. There
was less current with a thin layer of kerosene in the bottom of the pan (simulating the
small amount of kerosene present during other experiments) than for the empty pan as
well. The exposed surface of the pan was roughly double with the kerosene layer what
it was with the pan half-filled with water, so the simplest explanation here must be that
the decrease in exposed surface area was the dominant cause of the decrease in current.
In this vein, it can be stated that the reduction in surface area depends on both the
thickness of the layer of insulating fluid in the pan and the conductivity of that fluid.
Extrapolating from Figure 2.4, one can see that at higher voltages the conductivity will
begin to have a larger effect, as the corona discharge through the kerosene looks as if
it will soon converge with and become less than that of the tap water. The effect of
the dielectric constant of the fluid on the magnitude of the discharge current cannot be
determined from these data.
As is to be expected in a discharge at constant voltage, the greatest factor is the
6 6.5 7 7.5 8 8.5 9 9.5 10 0
0.1 0.2 0.3 0.4 0.5 0.6 0.7
Applied Voltage, kV
Corona Discharge Current, uA
Empty Pan
Empty Pan with Screen Empty Pan with Kerosene Layer
Empty Pan with Kerosene Layer and Screen Pan Filled with Tap Water
Pan Filled with Distilled Water