04815174
04815174
04815174
ABSTRACT
A detailed review of progress-to-date in the field of electrohydrodynamic conduction
pumping is presented. Analytical, numerical, and experimental research studies that
have been published in this field are summarized and a detailed scientific overview of
the conduction pumping model is provided. Conduction pumping has already been
investigated successfully to resolve several commonplace issues relating to pumping,
flow-distribution, and heat transfer and these applications are also presented.
Index Terms Electrohydrodynamics, conduction phenomenon, dielectric fluids,
non-mechanical pumping.
NOMENCLATURE
b
C0
C
c
cp
D
d
electric permittivity
efficiency
dynamic viscosity
mass density
net charge density
e
electric conductivity
e
Manuscript received on 31 May 2008, in final form 15 September 2008.
potential field
space charge relaxation time (/e)
dimensionless dissociation rate parameter, defined by equation (19)
SUBSCRIPTS
+
positively-charged ions
negatively-charged ions
eq
equilibrium
ohm ohmic
SCL space charge limited
SUPERSCRIPTS
*
dimensionless
INTRODUCTION
(1)
The first term represents the Coulomb force that acts on free
charges in an electric field. The second and third terms, titled
dielectrophoretic and electrostriction forces respectively,
represent the polarization force acting on polarized charges.
The dielectrophoretic term requires a gradient in permittivity
that does not exist in an isothermal liquid, while the
electrostriction term is relevant only for compressible fluids.
EHD pumping has shown extensive potential due to its
simple, lightweight, non-mechanical design, low power
consumption, low acoustical noise, and the ease with which
pumping can be controlled by adjusting the applied voltage.
A variety of EHD pumping mechanisms are based on the
Coulomb force: conduction pumping, induction pumping, and
ion-drag pumping. These methods differ in the manner of
generating the non-zero net charge density that is required for a
Coulomb force to exist. Ion-drag pumping relies on the injection
of ions into the liquid from sharp liquid/solid interfaces, while
induction pumping uses an ac travelling wave to attract and repel
AB A+ + B
(2)
THEORETICAL MODEL
(3)
(4)
p* =
Figure 1. Schematic of heterocharge layers, the attractive force toward the
electrode, and the use of suitable electrode geometries to generate pumping
using this force [4].
425
E
n
, n* =
, E* =
, * = ,
V d
V
neq
neq
P* =
d
2
, t* =
t
tT
(5)
426
(7)
* E* = C0 ( p* n* )
(8)
E* = * *
(9)
(10)
(11)
D k BT
=
bV
eV
neq d 2V
G=
2
(12)
(13)
(u
) u =
*
1
Re EHD
(14)
P + Re1EHD u + M C0 ( p n ) E*
*
*2
2
0
(15)
where the mobility ratio M0 and the EHD Reynolds number,
ReEHD, were defined as
M0 =
Re EHD
b2
bV
=
u* =
(6)
C0 =
(16)
(17)
u
bV d
(18)
I1 ( 2 )
,=
e3 E
4 k B2T 2
(19)
* = 1 , p* = 0 , n * n* = 0
* = 0 , n * p* = 0 , n* = 0
(22)
(23)
When the liquid conductivity is low and the length scales are
sufficiently large (on the order of millimeters), the effect of
electric double layer and the induced zeta potential are
negligible and the conditions simplify to
n * * = 0 , n * p* = 0 , n * n* = 0
(25)
n * p* =
SINGLE-PHASE STUDIES
3.1 THEORETICAL/NUMERICAL
Atten and Seyed-Yagoobi [3] applied their steady-state
model to three electrode configurations: parallel plane
electrodes, point-plane electrodes, and blade-plane electrodes.
Of particular interest was the point-plane configuration, since
it presented significant field asymmetry that could be
harnessed for pumping head generation. The thickness of the
heterocharge layer was approximated in two asymptotic limits
and the net force exerted on the liquid bulk was estimated, by
a series of approximations and simplifications, to vary
linearly with and V2.
Jeong et al [5] were the first to take the model developed
by Atten and Seyed-Yagoobi [3] and obtain a full numerical
solution to the governing equations (6)(10) to numerically
investigate the prospects of an electrode configuration
consisting of a ground ring and a high-voltage hollow tube
centered in the flow channel (refer to Table 1 for a sketch of
a hollow-tube electrode). The study neglected fluid motion
and focused on static pressure-head generation. The
computation considered the potential field in both the liquid
and the solid insulating material that surrounds the liquid
channel, with appropriate boundary conditions at the solidliquid interface. The numerical solution provided valuable
insight by allowing the easy visualization of contours of
potential field intensity, electric field intensity, and
positive/negative/net charge density. From the results, the
heterocharge layers around the electrodes were clearly
visible. The numerical results suggested that a pressure
generation of 1 kPa was plausible by combining 15 electrode
pairs in series, with sufficient spacing to minimize
interaction between pairs.
Jeong and Seyed-Yagoobi [11] used the addition of the
continuity and momentum equations to the theoretical model
in order to numerically solve the fluid motion inside a twodimensional enclosure when subject to an electric field
imposed by the electrodes shown in Figure 2. Numerical
results showed two counter-rotating vortices to each side (left
and right) of the high voltage electrode. To study the effect
that the charge convection term and the field-enhanced
dissociation term, F(), could have on the charge distribution,
three distinct cases were considered and compared: 1) zero
charge convection (p*u* = n*u* = 0) and field-independent
dissociation (F() = 1), 2) charge convection and fieldindependent dissociation, and 3) both charge convection and
field-enhanced dissociation. Charge convection was shown to
deform the heterocharge layers in certain regions, while fieldenhanced dissociation served to cause heterocharges with
high-density values to exist in larger areas around the
electrodes, implying an increase in the resultant net electric
body force compared to the field-independent dissociation
case. This higher body force resulted in stronger flow
circulation within the enclosure.
Feng and Seyed-Yagoobi [13] investigated equations (6)
(8) with the same simplifications as Atten and Seyed-Yagoobi
[3] (i.e. steady-state, spatial variations limited to x-direction,
diffusion terms neglected). However, the characteristic length
scale used to nondimensionalize was the characteristic
427
neq d
(26)
V
The characteristic heterocharge layer thickness, , was
defined as the point x = at which the non-dimensional
positive ion density, p*, is equal to 0.99, corresponding to the
location where the charge concentration has almost reached
the equilibrium level. Furthermore, the exact analytical
solutions to equations (6)(8) were used, and from these
solutions the parameter C was shown to assume a constant
value of 1.8, meaning that the parameter is independent of the
properties of the dielectric fluid and the applied electric field
provided that << d. The theoretical solution was shown to
have an error of just 0.0001% even for = d/3, suggesting
that the developed model is widely applicable. The analysis
predicted the thickness of the heterocharge layer, as a function
of the fluid properties and the electric field intensity, as
V
= 1.8
, neq e
(27)
2b
neq d
V2
d2
(28)
428
= 3.6
V b
f ( u* )
e d d
(29)
3.2 EXPERIMENTAL
A number of electrode designs have been considered for the
pumping of single-phase liquid dielectrics using conduction
pumping. Not-to-scale sketches that illustrate the most
commonly utilized electrodes are shown in Table 1. In all
single-phase studies, the ground electrodes have been ring
electrodes that are flushed to the circumference of the tube.
For additional details about electrodes and dimensions,
readers should refer to the references. Additional electrode
designs that have been considered for the pumping of singlephase and two-phase fluids can be found in [15, 16].
For the experimental portion of their study, Atten and
Seyed-Yagoobi [3] considered two different electrode
configurationssingle-tube and multi-needle. The minimum
axial distance between the electrode pair (high-voltage and
ground) was 4.5 mm (single-tube) and 4.2 mm (multi-needle).
Both electrodes were designed to create a strong electric field
in the axial direction of the tube in order to produce
appreciable liquid flow rates. The pressure generation of the
conduction pump was measured by using a manometer.
Results were presented for two different working fluids: nHexane and low-pressure refrigerant HCFC-123.
For HCFC-123 with the single-tube electrode configuration,
the pressure head reached approximately 500 Pa at an applied
voltage of 20 kV and a power level of 1.4 W, as shown in
Figure 4. In the absence of viscous pressure losses, this
pumping head was shown to correspond to a flow velocity of
approximately 1 m/s, demonstrating the potential of EHD
conduction pumping for practical use. The needle electrode
configuration was not as effective, producing only 100 Pa of
pressure head at an applied voltage of 15 kV. Further increases
in the applied voltage caused the pressure generation to cease
(at 17.5 kV) and then reverse (at 20 kV). This flow reversal was
attributed to the effect of direct ion-injection by the sharp
needle electrodes overcoming the effect of the heterocharge
layers. For both electrode configurations, the pressure
generation was a quadratic function of the applied voltage;
however, departure from this quadratic behavior began around
10 kV for the needle electrode configuration, due to the onset of
direct ion injection. Pressure head generation using n-Hexane
as the working fluid was shown to be approximately one-sixth
of the pressure head generation using HCFC-123.
429
430
Endurance tests were conducted by Jeong and SeyedYagoobi [4, 17] to investigate the long-term performance of
EHD conduction pumps, with multiple electrode pairs
operating continuously in series for 8 hours. The pressure
generation and current levels were found to slowly decrease
over time to some asymptotic value. These decays were
attributed to a purification process caused by gradual
precipitation of fine particles, which is a phenomenon that is
generally seen in electrical conduction measurements of a
fluid.
Feng and Seyed-Yagoobi [13] used the 1.59 mm perforated
high voltage electrode for the experimental portion of their
study. Experiments were conducted using a single-phase loop,
fitted with a flow-meter, valve, and a pressure-transducer to
measure the pumping head generated by the EHD pump.
HCFC-123 was chosen as the working fluid. Due to friction
losses within the pump itself and the external load placed on
the pump due to the setup of the single-phase loop, the
maximum flow velocity was 8.9 cm/s at an applied voltage of
15 kV and a generated pressure head of 425 Pa. The
corresponding mass flux was found to be 130 kg/m2s. The
pressure head generation and current level were shown to
agree with the theoretically predicted values in the absence of
net flow. To explain the discrepancies caused by the presence
of net flow, a magnitude analysis of the current level was
completed, which showed that charge mobility and charge
convection could significantly affect the current level, while
charge diffusion had a negligible effect at high voltages [13].
TWO-PHASE STUDIES
4.1 NUMERICAL
The application of the model described in Section 2 to
examine the pumping of a liquid-film (stratified two-phase
flow) was first conducted by Yazdani and Seyed-Yagoobi
[12] by considering electrodes on the base of a channel that
were designed to remain submerged in the liquid film
(without penetrating into the vapor phase). The design of the
electrodes was motivated by two simple and counter-acting
concepts: 1) that greater penetration of the electrodes into the
liquid film, away from the surface, would generate a greater
electric body force in the fluid and provide greater pumping
DEMONSTRATED APPLICATIONS
5.1
431
2
The first such demonstration came from Bryan and SeyedYagoobi [8], but at that time the conduction pumping
phenomenon was not understood and the flow generation was
erroneously credited to polarization forces. Once the
conduction pumping phenomenon had been clarified [3, 9],
the EHD conduction driven heat pipe was revisited by Jeong
and Seyed-Yagoobi [22] using electrodes better optimized for
conduction pumping (the same porous electrodes as studied
50.80
pressure port
3.17
17.46
1.59
9.53
ground
electrode
high voltage
electrode
432
3 mm
(Favorable)
520 W
920 W
0 mm
(Horizontal)
400 W
840 W
3 mm
(Adverse)
200 W
800 W
433
CONCLUSIONS
434
REFERENCES
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]