Applied Thermal Engineering 119 (2017) 438–448

Contents lists available at ScienceDirect

Applied Thermal Engineering

journal homepage: www.elsevier.com/locate/apthermeng

Research Paper

Regulation-controlling of boundary layer by multi-wire-to-cylinder

negative corona discharge
Wei Wang a, Lanjun Yang a,⇑, Kai Wu a, Cen Lin a, Peng Huo b, Shuai Liu a, Dong Huang a, Mei Lin c
a
Xi’an Jiaotong University, State Key Laboratory of Electrical Insulation and Power Equipment, Xi’an, China
b
Xi’an High Voltage Apparatus Research Institution, Xi’an, China
c
Xi’an Jiaotong University, National Specialty Laboratory of Fluid Machinery, Xi’an, China

h i g h l i g h t s

Heat transfer has been enhanced by multi-wire-to-cylinder corona discharge.

The unipolar model for boundary layer investigation is used.

The velocity of boundary layer can be increased with the voltage.

The distribution of corona wire can regulate velocity field and temperature field.

Field synergy for heat transfer enhancement is introduced.

a r t i c l e i n f o a b s t r a c t

Article history: Multi-wire-to-cylinder corona discharge was studied for better understanding of the electrohydrody-
Received 28 November 2016 namic phenomena which directly relate to the performance of heat transfer enhancement. A unipolar
Revised 20 February 2017 approximation model was established and numerical simulations were conducted to determinate the
Accepted 19 March 2017
heat transfer and velocity distribution of ionic wind. The numerical and experiments results show good
Available online 21 March 2017
agreement. It indicates that higher applied voltage and more corona wires can help decreasing thermal
boundary layer thickness and increasing velocity gradient of boundary layer, resulting in the enhance-
Keywords:
ment of heat transfer. When the applied voltage increases from 7 kV to 11 kV and the number of wire
Electrohydrodynamic
Boundary layer
increases from corona wire number 1–3, there will be a 39.8% decrease of the thickness of thermal
Field synergy boundary layer. In the condition of U = 11 kV, the maximum local forced convection heat transfer coef-
Heat transfer enhancement ficient for corona wire number N = 3 is 12 times and the average value is 8 times higher than that of nat-
ural convection.
Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction neutral species). There exist two regions, ionization region and
drift region. The ionization region round the active electrode is
The basic corona discharge physics in atmospheric air has filled with electrons and ions. Positive or negative ions are sub-
already been well-known for over one hundred years, and it can jected to Coulomb force and drifting towards the grounded collect-
be depicted as a self-sustaining electrical gas discharge occurring ing electrode in the drift region. These ions impact on the neutral
around the active electrode with the smallest curvature radius air molecules, resulting in a momentum transfer that produces a
and on which the high voltage is applied [1]. A significant amount gas flow from the active electrode to the grounded collecting
of researches have been conducted into the unipolar/bipolar and one. This phenomenon is well-known as Electrohydrodynamic
AC/DC corona discharge [2,3]. According to the polarity of the (EHD), and the induced flow is usually called ionic wind [4–9]. It
active electrode, there exist two DC corona discharge forms named has been experimentally and theoretically studied in recent years
positive corona and negative corona. And corona discharges are due to its potential applications such as enhancement of heat
non-equilibrium plasmas with an about 108% degree of ionization transfer [10], electrostatic precipitation [11], aerodynamic flow
(the ratio of the number of charged species to the number of control [12,13] and electric propulsion [14–16].
Till now, most of the experiments about ionic wind enhancing
⇑ Corresponding author. heat transfer focused mainly on finding the average heat transfer
E-mail address: [email protected] (L. Yang).
coefficient and there has been some progress. By changing the

http://dx.doi.org/10.1016/j.applthermaleng.2017.03.092
1359-4311/Ó 2017 Elsevier Ltd. All rights reserved.
 W. Wang et al. / Applied Thermal Engineering 119 (2017) 438–448 439

shape of electrode, the efficiency of heat transfer is enhanced as wind are focused on the increase of average and maximum flowing
high as 100–400% in most of the experimental studies. For velocity. However, this paper is devoted in changing the electrode
instance, Go et al. [17] studied the heat transfer enhanced by cor- structure to control the velocity field in boundary layer and regu-
ona discharge with wire-plate electrodes. The result showed that late thermal boundary layer. As a result, heat transfer will be
ionic wind increased local forced convection heat transfer coeffi- enhanced.
cient to more than twice, compared with natural convection. The numeric simulation calculation of ionic wind is a comple-
Owsenek et al. [18] applied needle-plate to generate corona mentary method for experimental study on local velocity field,
discharge and ionic wind. They proved that local forced convection such as the velocity of boundary layer. In the numeric simulation,
heat transfer coefficient exceeds 65 W/m2 K, which was 25 times single/multiple wires-double parallel plates is widely used, which
higher than natural convective heat transfer coefficient prevails in ESP. Traditionally, unipolar approximation model is
(2.6 W/m2 K). Besides, scholars also studied the mechanism of applied in ionic wind simulation, which ignores the thickness of
ionic wind heat transfer at hot spot with MEMS technology to corona sheath and only takes the single polar particles in drift
microminiaturize corona discharge electrodes. Ong et al. [19] region into consideration. Zhao and Adamiak et al. [26–28] made
adopted mesh-needle-mesh electrodes (the mesh sizes are a simulation of ionic wind produced by single stage needle-mesh
2.1  0.9 mm, the distance of needle and mesh is 1 mm) to inves- and needle-plate corona discharge with finite element method
tigate the influence of the sizes of hole in mesh electrode and the (FEM), boundary element method(BEM) and method of character-
gap of need-mesh electrode on ionic wind. The heat flow rate by istic (MOC). Martins [29] and Colas et al. [30] simulated ionic wind
microminiaturized ionic wind actuator reached to 40 mW and velocity field by utilizing the electrodes type, a positive HV wire,
the temperature of hot spot decreased 4 K, when the forced two grounded cylinders and two negative HV plate electrodes
convection heat transfer coefficient reached the maximum and their results are coordinate with each other. Kim et al. [24]
3200 W/m2 K. simulated the multiple stage ionic wind actuator and drew the
Large amount of breakthroughs in promoting ionic wind veloc- conclusion that the flowing velocity was in direct proportion to
ity have been achieved. For instance, when optimizing EHD gas the square root of stage number, which also agreed with the exper-
pump with needle-mesh electrodes in a hollow tube, Moreau and iment. Jewell-Larsen et al. [31] and Ong et al. [19] investigated the
Touchard [5] measured maximum velocity up to 10 m/s. The aver- influence of ionic wind on local enhancement heat transfer
age ionic wind velocity was about 2 m/s and the efficiency of elec- through microminiaturizing the electrodes. They also optimized
tric energy converting to air kinetic energy was 1.73%. Besides, the electrodes structure by simulation and acquired satisfying
Colas et al. [21] employed wire-cylinder-plate electrode and also results.
recorded maximum velocity up to 10 m/s. The average velocity Today, many applications in which wire-to-cylinder electrode
was nearly 4.5 m/s and the efficiency of electric energy converting configuration are being used, such as aerodynamics [32] and ozone
to air kinetic energy was 1.33%. Other academics also acquired generator [33]. In the industrial field of heat transfer, both the in-
1 m/s–10 m/s velocity with several types of electrodes [1,6,7]. line and cross-line pipes are used to dissipate heat through natural
The results are coordinated with the theoretical predictions of Sig- convection or forced convection. At present, there are few reports
mond [22] and Goldman et al. [23]. The above researches have pro- on the heat dissipation enhancement of ion wind around the pipe-
ven that the single stage ionic wind actuator has reached the line. Therefore, this paper studied ionic wind velocity field of mul-
plateau in promoting ionic wind velocity and energy converting tiple wires-to-cylinder corona discharge and heat transfer
efficiency. Rickard et al. [6], Qiu et al. [9] and Kim et al. [24] sought enhancement of cylinder surface by experiments and simulations.
to improve ionic wind velocity with multistage configuration. The To enhance the heat transfer on the cylinder surface, the thickness
research shows that ionic wind velocity is in direct proportion to of thermal boundary layer was decreased by changing the wire
the square root of stage number. Specifically, Qiu et al. used six amount and the applied voltage on it. Through the study of this
needles-mesh electrodes in tube as single stage ionic wind actuator paper, it can provide positive significance and reference value for
and 25 stages were in serial. The maximum ionic wind velocity the application of the ion wind technology produced by corona dis-
reached around 16.1 m/s. However, the energy converting effi- charge to the pipeline heat dissipation.
ciency was 2.2%.
Leger et al. [25] studied the influence of corona discharge pro- 2. Experimental setup
duced ionic wind on the velocity boundary layer with wire-plate
electrodes configuration. The research illustrates that the velocity Fig. 1 shows the schematic diagram of experimental setup used
of boundary layer can be promoted while ionic wind is applied to measure the velocity field of ionic wind and the performance of
and the maximum flowing velocity in this boundary layer is heat transfer intensified by corona discharge. Four types of elec-
3 m/s, compared with 1.8 m/s without ionic wind. And similar trodes configuration are used in experiments and simulations, as
result has been given by Moreau [12]. The massive experimental shown in Fig. 2. The wire with radius of 25 lm was utilized and
researches indicate that with the flowing velocity of ionic wind connected to a negative DC high voltage supply (0  50 kV) in
in boundary layer improved, the thickness of boundary layer is series with a resistance (R = 10 MX). It was placed 10 mm away
reduced to lower the thermal resistance and enhance the heat from the cylinder which is made of stainless steel and has a diam-
transfer. eter of D = 16 mm and length of 100 mm. The cylinder was
In thermodynamics, non-dimensional Prandtl number Pr is gen- grounded via a micro ammeter. The wall thickness of cylinder
erally adopted to represent the relationship of the thickness of was 0.8 mm, inside of which was nichrome resistance wire twined
thermal boundary layer d2 and velocity boundary thickness d1: if on the magnesium oxide bar and magnesia powder filled up the
Pr > 1, it means the thermal boundary layer is relatively thin and gap. The wall is heated through DC power source (0–200 V). The
the heat transfer is effective; if Pr < 1, it means the thermal bound- calibrated infrared thermometer (IS-CFL500AD series, resolution
ary layer is thick and the heat transfer is poor. In order to achieve 0.1 K) was fixed on a displacement platform and 100 mm away
the heat transfer enhancement target, the thermal boundary thick- from the surface of cylinder. The thermometer was manipulated
ness should be decreased to some extent so as to increase the fluid to scan the temperature of the cylinder surface in the length
average velocity or decrease the velocity boundary layer thickness. direction (10 data points) and in the circumferential direction
In present, most of the works on heat transfer enhanced by ionic (2 data points), and 20 temperature data were processed to obtain
440 W. Wang et al. / Applied Thermal Engineering 119 (2017) 438–448

Light arm
Dual- pulse
laser
-HV Electrodes configuration A Nd: YAG
355/266 nm
Wire electrode Laser CCD
sheet

Heating cylinder
R D.C A
Voltage meter synchronizer
displacement
stage
H.V Current meter Infrared
D.C thermometer

Fig. 1. Schematic diagram of velocity field measurement setup of heat transfer enhancement by ionic wind experiment.

(b) N=2 (c) N=3 (d) N=4

(a) N=1
wire

-HV a=120° a=90°

Heating a=180°
cylinder

Fig. 2. Schematic of wire-to-cylinder electrode configuration (A-A cross-section), N represents the corona wire number.

an average temperature. The velocity field of ionic wind was the conduction term is negligible because of the extremely low
acquired by particle image velocimetry (PIV), which was composed electrical conductivity of air (r  1.6  1023 X1 m1 [35]). The
of dual-pulse laser (Nd: YAG 355/266 nm), CCD digital camera (TSI value of convection and diffusion terms of ions is lower than the
Model 630046, 1000  1016 pixels), laser pulse synchronizer and drift term, but those were retained in this paper. The ion mobility
incense smoke particles tracer [34] with its controlling devices, in air and the mass diffusion coefficient vary slightly in literatures
as showing in Fig. 1. And the frames were analyzed with Insight [33,35], so they are taken to be 2.1  104 m2/V s and
3.2 software. 1.589  105 m2/s [20], respectively.
The thermo-hydrodynamic part of the problem is described by
3. Model description mass (4), momentum (5), and energy (6) of the conservation equa-
tions for steady state incompressible air flow.
3.1. Governing equations r ~
u¼0 ð4Þ

The electric field and the space charge distribution are qair ð~
u  rÞ~ u þ~
u ¼ rP þ lr2~ f ð5Þ
described by the Poisson Eq. (1) and the conservation equation of
charges (2) in the unipolar approximation model of ionic wind pro- @T
duced by corona discharge. qair C p þ qair C p~
u  rT ¼ r  ðkrTÞ ð6Þ
@t
q
r2 U ¼  v ð1Þ where qair is the air density, P is the air pressure, l is the air
e dynamic viscosity, Cp is the specific heat capacity of air, T is the tem-
perature. The body force ~f acting on the air was defined by Stratton
r  ~J ¼ 0 ð2Þ
[36], as the Eq. (7). The three terms on the right side of Eq. (7) rep-
where U, qv and e are the electric potential, the space charge den- resent the Columbic force, the force owing to permittivity gradient,
sity, and the air dielectric permittivity, respectively. The electric and the electrostriction force, respectively. Since the variation for
current density ~
J could be described as follow: air is less than 0.1% over a range of 1000 K [37], the second term
is neglected. Moreover, only in a two-phase interface does elec-
~
J ¼ r~
E þ li~
Eqv þ ~
uqv  Drqv ð3Þ trostriction affect the flow [36]. Therefore, the body force was sim-
ply modeled as a function of the electric field and ion charge
where r is the electrical conductivity, ~ E is the electric field inten- density. Additionally, compared to the body force, the buoyancy
sity, li is the ion mobility, ~
u is the air flow velocity induced by cor- induced by air temperature gradient isn’t considered.
ona discharge, and D is the diffusion coefficient. The four terms on

~ 1 1 @e
the right side of Eq. (3) represent conduction, drift, convection of f ¼ qv ~
E  E2 re þ r E2 qair ð7Þ
ion due to air flow, and the diffusion of ions through air. Typically, 2 2 @ qair
 W. Wang et al. / Applied Thermal Engineering 119 (2017) 438–448 441

Around the corona initiation voltage, a few tufts irregularly

spaces along the wire at negative corona. The number of tufts
100
Inlet (B5)
increases with the voltage and the distribution of tufts becomes
more uniform [38]. This paper assumes that corona discharge
along the wire distributes uniformly under high voltage. Thus,
Axial sysmmetry
the simulation model is simplified from 3D to 2D and calculation line (B1)
time is reduced. The electrostatics mode was utilized to solve Pois-
Wire electrode 25μm
son’s equation (1), and the conservation equation of charge (2) was
(B2)
constructed by the PDE mode in COMSOL. In addition, the laminar
flow mode was employed to solve the conservation equation of
mass (4) and momentum (5). When solving convection and diffu-
sion problems, COMSOL adopts various forms of artificial diffusion,
e.g. isotropic diffusion, streamline diffusion and crosswind diffu-
Open boundary (B4)
sion, which improves the accuracy, stability, and efficiency of
computation. 0
During the natural convection simulation, only Eqs. (4)–(6) are AIR
considered. Body force is buoyancy ~ f ¼ gðq  q Þ; g is gravitation
ref
constant; q is air density; qref is air density at Tamb and 1 atm. Ground (B3)
When simulating ionic wind intensified heat transfer, Eqs. (1)–(7)
are solved together. The whole simulation process is showed in
Fig. 3.

3.2. Calculation domain and boundary conditions

Outlet (B6)
As shown in Fig. 4, the simulation domain of the single wire-to- -100
cylinder negative corona discharge is presented and the mesh is
generated around the electrodes. The boundary conditions are 0 100
summarized in Table 1. For the rest three types of electrodes con-
figuration the calculation regions are similar to single wire-to- Fig. 4. Simulation domain of the single wire-to-cylinder negative corona discharge
(not in scale). And the another three types of electrodes configuration are similar to
cylinder, except that the boundary of B5 and B6 are changed to
this one, the differences are calculation regions, which are 150  200 mm2,
open boundary condition. Furthermore, in order to ensure that 250  250 mm2, 300  300 mm2, respectively, bigger than the original one.
the numerical results are independent of the computational grid,
a grid sensitivity analysis was performed. Morrow suggested that
the corona sheath thickness can be neglected and it could be onset field is generally accepted and justified by Morrow [39].
described by a boundary condition [39]. Generally, three methods Then, the charge concentration q of active electrode under differ-
are available to describe the boundary condition of active elec- ent applied voltages can be acquired through auxiliary sweep func-
trode: ions flux j, oj/ot and the electric field intensity Es [38–40]. tion of CMOSOL. Through this function, the range of charge
The electric field intensity Es is adopted in simulation on account concentration can be preset, and it can be determined when the
of its convenience and simplicity. electric field intensity obtained by the boundary probe satisfies
The corona critical electric field intensity Es on the wire surface the value calculated by the Peek’ formula.
is calculated by Peek’s formula [41], as shown in Eq. (8). The pffiffiffiffi
Es ¼ E0 ð1 þ 0:03= rs Þ ð8Þ
assumption that the electric field strength at the wire equals Peek’s
6
where E0 = 3  10 V/m is the dielectric breakdown electric field
intensity of air; rs is the radius of wire in meter.

4. Numerical results and discuss

4.1. Effect of the voltage

Fig. 5 is the velocity field of ionic wind with single wire-to-

cylinder. When the applied voltage is 0 kV, the velocity field of
cylinder surface under natural convection is shown in Fig. 5(a).
The airflow near the cylinder flows along the surface and doesn’t
separate, and the maximum air flowing velocity excited by buoy-
ancy is only about 0.1 m/s. When the applied voltage is larger than
5 kV, the air molecules are collided by space charges with momen-
tum transfer and accelerated. As a result, a high air flowing velocity
region between the wire and cylinder is formed. When the high
velocity airflow approaches the upper surface of cylinder, the flow
pattern is similar to the free stream fluid across tube: the air fluid
Fig. 3. Simulation process schematic, I without ionic wind, the heat flux of cylinder is brought to rest at the stagnation point, with an accompanying
surface is q W/m2, during the natural convection process the temperature increases rise in pressure. From this point, the pressure decreases downward
from Tamb to Twall2; II with ionic wind, the heat flux of cylinder surface stays the
same, the surface temperature decreases to Twall1 due to forced convection; III with
along the surface of cylinder, according to Bernoulli equation, and a
ionic wind, the heat flux of cylinder surface increases Dq W/m2, the surface higher velocity flow layer with about 5 mm thickness is excited on
temperature increases to Twall2 again. both side of the cylinder surface. In the condition of U = 11 kV,
442 W. Wang et al. / Applied Thermal Engineering 119 (2017) 438–448

Table 1
Boundary conditions used in the model.

Application mode Electrostatics Charge transport Fluid dynamics Heat transfer

B2 Potential U = 7  11 kV Concentration q = const. No-slip –
B3 Ground U = 0 q=0 No-slip Heat source
B1 nD = 0 Zero flux nJ = 0 un = 0 n(k5T) = 0
B5 nD = 0 q=0 Normal flow u = 0 m/s Open boundary: T = const.
B6 nD = 0 q=0 Pressure P = 0 Pa Open boundary: T = const.
B4 nD = 0 q=0 Open boundary: zero viscous stress Open boundary: T = const.

(a) N =0 (b) N =1 (c) N =1 (d) N =1

Fig. 5. Velocity field of ionic wind with single wire-to-cylinder; colors represent the velocity values and the arrows are the flowing direction; the lengths of arrows are
normalized. (a) N = 0 represents natural convection, around the cylinder umax = 0.1 m/s, Gr = qbgd3DTl2  4.17  104, where q is the density of air, b is the gas expansion
coefficient, g is the gravity acceleration, d is the electrode gap, DT is the temperature difference, l is the gas dynamic viscosity. (b) Ionic wind forced convection, U = 7 kV,
umax = 1.39 m/s, Reehd  1400; (c) ionic wind forced convection, U = 11 kV, umax = 3.16 m/s, Reehd  3200; (d) U = 11 kV, PIV measurement result, umax = 2.52 m/s.

the maximum ionic wind velocity is almost 32 times of that in nat-

12 wire U=0kV
ural convection condition, as is shown in Fig. 5(c). And at the same
applied voltage, the velocity field measured by PIV generally agrees r=20 U=-7kV
30° U=-8kV
with the numerical result, but the velocity of boundary layer differs 10
r=8
U=-9kV
Distance of the wall (mm)

from that in simulation for the insufficient resolution of CCD cam-

era at boundary layer, as is shown in Fig. 5(c) and (d). 8
cylinder U=-10kV
The flow patterns of ionic wind across the cylinder remain con- U=-11kV
1.00
stant, as the applied voltage increase: ionic wind flows around the
6 0.75
cylinder surface and a pair of Foppl whirlpool appears in the down-
stream region. The maximum ionic wind velocity increases from 0.50

1.39 m/s to 3.16 m/s, shown in Fig. 5(b) and (c). Apparently com- 4 0.25

paring the natural convection and ionic wind forced convection, 0.00
0.0 0.5 1.0 1.5 2.0 2.5 3.0
the effect of electric field force is much higher than the buoyancy 2
on the velocity of airflow, which also proves the rationality of
ignoring the natural convection in the simulation. Ionic wind
0
velocity distribution along radial direction r = 8 mm to r = 20 mm
at 30 degrees, as is shown in Fig. 6. It’s obvious that the velocity 0.0 0.5 1.0 1.5 2.0 2.5 3.0
gradient of boundary layer increases with the applied voltage on Velocity of ionic wind (m/s)
the wire, while the thickness of boundary layer decreases, which
are given in Fig. 9(a). Likewise, another three types of electrode Fig. 6. Velocity profiles induced by the ionic wind with single wire-to-cylinder
electrode configuration for different voltage values; U = 0 kV represents natural
configuration lead to the same results. convection.

4.2. Effects of the number and position of wire wind near the cylinder surface. And there are four parts divided:
EHD flow part S1, excited by electric field force; laminar flow part
Unlike the single wire-to-cylinder electrode configuration, the S2, along cylinder surface; flow confluence part S3, joined from two
position and number of wire affect more on ionic wind flow pat- adjacent laminar flows; vortex part S4 between S2 and cylinder sur-
terns than the applied voltage. Fig. 7 illustrates ionic wind velocity face, as is shown in Fig. 7. Among the flow parts, S2 and S4 directly
field of multiple wires-to-cylinder corona discharge. Wires in dif- conduct heat transfer from the cylinder surface. Although the max-
ferent positions will effectively control the flow pattern of ionic imum velocity of ionic wind in laminar flow part S2 decreases from
 W. Wang et al. / Applied Thermal Engineering 119 (2017) 438–448 443

Numeric PIV
(a) N=2

S1
S2
S3

S4

(b) N=3

(c) N=4

Fig. 7. Velocity field induced by corona discharge at U = 11 kV for different corona wire number and position.

3 m/s to 1.7 m/s with the increase of corona wire number, the the heat transfer enhancement on the cylinder surface is saturated
vortices of vortex S4 decreasesand the area of cylinder surface when corona wire number N  3, which also fits the experiment
covering by laminar flow part S2 expands. This explains why results of Section 5.
444 W. Wang et al. / Applied Thermal Engineering 119 (2017) 438–448

(a) N=0 (b) N=1 (c) N=2 (d) N=3 (e) N=4

r(T, θ)

θ r(Twall, θ)
(0, 0)

red: 340K red: 310K red: 308K red: 307K red: 307K

bule: 300K bule: 300K bule: 300K bule: 300K bule: 300K
Fig. 8. Air temperature field around the cylinder at U = 11 kV for different corona wire number and position, and white outlines stand for isotherms, calculated from Eq. (9).
The N = 0 represents natural convection.

(a) N=1 (b) N=2

2.0 2.0
-7kV -7kV
1.8 1.8
-8kV -8kV
1.6 -9kV 1.6 -9kV
-10kV -10kV
1.4 1.4
-11kV -11kV
1.2 1.2
TBLT /mm
TBLT /mm

1.0 1.0 80
80
60
0.8 0.8
60
TBLT /mm
TBLT /mm

40
0.6 40 0.6
20
0.4 20 0.4
0
0.2 0
0.2 0 20 40 60 80 100 120 140 160 180
angle
0 20 40 60 80 100 120 140 160 180
angle
0.0 0.0
0 20 40 60 80 100 0 20 40 60 80 100 120 140 160 180
angle θ angle θ

(c) N=3 (d) N=4

2.0 2.0
-7kV 80 80
-7kV
1.8 -8kV 1.8 60 -8kV
TBLT /mm

60
-9kV -9kV
TBLT /mm

40
1.6 -10kV 40 1.6 -10kV
-11kV 20 -11kV
1.4 20
1.4 0
0 20 40 60 80 100 120 140 160 180
0
angle
1.2 0 20 40 60 80 100 120 140 160 180
1.2
TBLT /mm

angle
TBLT /mm

1.0 1.0
0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.2
0.0
0 20 40 60 80 100 120 140 160 180 0.0
0 20 40 60 80 100 120 140 160 180
angle θ
angle θ

Fig. 9. TBLT versus angle h for different applied voltage values.

 W. Wang et al. / Applied Thermal Engineering 119 (2017) 438–448 445

Table 2
TBLT for different applied voltage and wire amount. TBLT ¼ j~
rðT; hÞ  ~
rðT wall ; hÞj ð10Þ
Voltage/kV TBLT/mm
where T stands for the upper limit temperature in thermal bound-
N=1 N=2 N=3 N=4
ary layer, Twall is the average temperature of cylinder surface, Tamb
7 1.13 1.01 0.97 0.98 is the environment temperature, r is radial vector, the origin (0, 0)
8 1.04 0.89 0.85 0.87
of which is the center of cylinder, and it is a function of temperature
9 0.96 0.83 0.78 0.78
10 0.91 0.77 0.72 0.73 T and the angle h between r and the axis symmetry line, as shown in
11 0.86 0.73 0.68 0.68 Fig. 8(a). Compared with the velocity field distribution (in Figs. 5
and 7) and the temperature distribution (in Fig. 8), .the distribution
4.3. Effect of velocity field on temperature gradient situation of temperature on and near the surface of cylinder is
determined by the ionic wind. In Fig. 8(a), airflow is heated in the
Fig. 8 shows that the thickness of thermal boundary layer downstream region and the heat flux of cylinder surface is sup-
decreases apparently in ionic wind forced convection, compared pressed, as the vortex stirs, but it is still larger than that of natural
with that in natural convection. Moreover, with the increase of cor- convection. When corona wire number N  2, the vortex’ position
ona wire number, so it decreases gradually since the heat flux near transfers and the vortices decrease, so that the heat flux of cylinder
the cylinder surface is dissipated by ionic wind rapidly. In this surface increases apparently. As a result, the temperature of its hot
paper, the thermal boundary layer thickness (TBLT) of ionic wind point decreases 2 K.
is defined as: The TBLT of the cylinder, which is attained from Eq. (10), is
reduced with the applied voltage, and the minimum values are
T  T wall obtained at the stagnation point, as is displayed in Fig. 9. The aver-
¼ 0:99 ð9Þ
T amb  T wall age TBLT are acquired by integral mean calculation of Eq. (10) and

(a) N=1 (b) N=2

14 14
U=-7kV
U=-7kV U=-8kV
wire
12 wire U=-8kV 12 U=-9kV
0
U=-9kV 0 U=-10kV
U=-10kV cylinder
U=-11kV
10 U=-11kV 10 90
90

cylinder 180
8 180 8
h1/h0

h1/h0

wire

6 6

4 4

2 2

0 0
0 20 40 60 80 100 120 140 160 180
0 20 40 60 80 100 120 140 160 180
Angle θ
Angle θ

(c) N=3 (d) N=4

14
wire U=-7kV
wire U=-7kV
U=-8kV 12 U=-8kV
0 U=-9kV
12 0 U=-9kV
wire U=-10kV
cylinder U=-10kV 90 wire U=-11kV
90 U=-11kV 10
10 cylinder
180
180 wire
wire
8
8 wire
h1/h0

h1/h0

6
6

4 4

2 2

0 0
0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 140 160 180
Angle θ Angle θ

Fig.10. Normalized local ionic wind forced convection heat transfer coefficient h1/h0, synergizing angle b versus angle h for different applied voltage values, and average
natural heat convection coefficient h0 = 8.7 W/m2 K.
446 W. Wang et al. / Applied Thermal Engineering 119 (2017) 438–448

45
EXP SIM N=1
900 N=1 40 N=2
N=2 N=3
N=3 35 N=4
750
N=4
fit curve I=CU(U-U0) 30

Power (W/m)
600
25
I (uA)

450 20

15
300
10
150
5

0 0
7 8 9 10 11 7 8 9 10 11
U (kV) U (kV)
(a) (b)
Fig.11. Current versus applied voltage on wire (I-U, (a)), power versus applied voltage (P-U,(b)) for 4 types of electrode configuration, the experimental data are represented
by solid symbol, the simulation data are denoted by hollow symbol, the line is fitted using the formula I = CU(U-U0), C is a constant, and U0 is corona onset voltage.

9
SIM EXP
8 N=1
N=2
7 N=3
N=4
6
h1/h0

2
7 8 9 10 11
Voltage (kV)

Fig. 12. The normalized average ionic wind forced convection heat transfer
 1 =h
coefficient h  0 versus the applied voltage for different wire amount configuration
in simulation and experiment.

Fig. 13. The cylinder surface temperature versus the wire amount and the applied
shown in Table 2. Increasing the applied voltage or the number of
voltage.
wire will reduce the average TBLT. If corona wire number N  3,
the average TBLT remain the same. The TBLT decreases 39.8% when
the applied voltage increases from U = 7 kV to U = 11 kV and the of the vortex S4 which will be weaken with the increase of wire
number of wire increases from corona wire number N = 1 to N = 3. amount, as are pictured in dashed red box in Fig. 10. Meanwhile,
the local forced convection heat transfer coefficient increases with
the applied voltage owing to the decrease of TBLT.
4.4. Heat transfer enhancement by ionic wind

For the mechanism of single phase convection intensified heat 5. Experimental results and discuss
transfer, Refs. [42,43] offer many explanations: (a). decrease of
the TBLT; (b). the increase of disturbance in airflow; (c). the The I-U and P-U characteristic curves for wires-to-cylinder
increase of velocity gradient on the wall. In the paper, the numer- corona discharge were shown in Fig. 11, corona current and
ical results indicate that increasing the applied voltage and wire electrical power consumption increased with the applied voltage
number have much effect on the decrease of TBLT and the increase increasing. And the corona current can be fitted using the formula
of the velocity gradient. Fig. 11 presents the normalized local I = CU(U-U0). During the experiment, it was found that incense
forced convection heat transfer coefficient changing with angle h. smoke particles have no influence on the corona current, this result
It begins to decrease at the stagnation point, and reaches the min- is consistent with [34] refs. At the same excitation voltage, the
imum at the separation point where the boundary layer separation corona discharge power decreases with the increase of the wire
occur. However, it ends up with gradual increase due to the effect electrodes number, but the electric power no longer decreased
 W. Wang et al. / Applied Thermal Engineering 119 (2017) 438–448 447

when the wire electrode number increase to three or more than temperature decreases with the wire amount in the same applied
three. voltage condition, and the minimum reaches to 305.5 K in the
The cylinder is heated when the applied voltage on wire is experiment. With the same electrode configuration, the cylinder
U = 0 kV, and its surface temperature is stable at Twall = 340.1 K surface temperature decreases linearly with the applied voltage.
conducted by natural convection, according to the energy In this paper, the maximum surface temperature drop of cylinder
conservation: we got at U = 11 kV for three and four wires-to-cylinder elec-
trodes ionic wind actuator are 34.4 K and 34.6 K, respectively.
Q 0 ¼ Q conv 0 þ Q loss0 ð11Þ
where Q0 is the input heat power of the cylinder, Qloss0 is the heat 6. Conclusion and future work
power loss of the cylinder, including two major parts: first, the heat
dissipation through fixed support made of wood for the cylinder; Ionic wind velocity field and heat transfer enhancement of mul-
second, the thermal radiation on the cylinder surface; Qconv0 is heat tiple wires-to-cylinder corona discharge have been studied
flow rate through natural convection, which can be expressed as through simulation and experiments in this work. The conclusions
following: are as follows:
0 SðT
Q conv 0 ¼ h wall  T amb Þ ð12Þ
(1) The unipolar approximation model of ionic wind produced
When the applied voltage on wire increases to U = 7 kV  by multi-wire-to-cylinder negative corona discharge is
11 kV, ionic wind acts on the cylinder for forced convection. In available on velocity and temperature distribution predic-
order to make the surface temperature of cylinder constant tion. The experiment results are agree with the numerical
(Twall = 340.1 K), the voltage applied on the resistance wire of cylin- ones. This model will be convenient in multi-wire-to-
der increases from U0 to U1, and its current form I0 to I1. So, the Eq. multi-cylinder electrode configuration.
(13) can be derived under the new thermal equilibrium condition. (2) The heat transfer performance of cylinder will be enhanced
Q 1 ¼ Q conv 1 þ Q loss1 ð13Þ by increasing the wire number and applied voltage. But,
there exist certain limits in specific situation, such as the
Consequently, the thermal radiation flux of cylinder is equal in applied voltage on wire can’t exceed the breakdown poten-
both conditions of natural and forced convection. Meanwhile, tial (11.5 kV), and the wire number shouldn’t beyond 3.
assuming the heat conduction is basically the same (Qloss0 = Qloss1), (3) The 3 wires-to-cylinder corona discharge investigated in this
so the heat dissipation of the heating cylinder surface forced by paper is optimized electrode configuration, considered both
ionic wind is derived from Eqs. (11)–(13): of power consumption and cooling effect of cylinder. In the
DQ ¼ Q 1  Q 0 ¼ U 1 I1  U 0 I0 future work, the optimization of multi-wire-to-multi-
cylinder configuration will be conducted. And the plasma
¼ ðQ conv 1  Q conv 0 Þ þ ðQ loss1  Q loss0 Þ ð14Þ
of the corona sheath in numerical model will be considered.
1  h
¼ ðh 0 ÞSðT
wall  T amb Þ

where h 1 is ionic wind forced convection heat transfer coefficient,

 0 is natural convective heat transfer coefficient, S is the area of Acknowledgment
h
cylinder surface, the environment temperature is Tamb = 300 K.
This work was supported by the National Natural Science Foun-
Considering the thermal radiation, the natural convection heat
dation of China (51377132).
transfer coefficient is about 8.9 W/m2 K obtained through
Eqs. (11) and (12), which is similar to the numerical result
(8.7 W/m2 K) in this paper. But, this result is different from other References
references, such as 2.5 W/m2 K by Owsenek et al. [18], 5 W/m2 K [1] M. Robinson, Movement of air in the electric wind of the corona discharge, Am.
by Owensek and Seyed-Yagoobi [44], 9–12 W/m2 K by Kalman Inst. Electr. Eng. (AIEE) Trans. 80 (1961) 143–150.
and Sher [45], arising from the diversity of electrode shape and [2] V.T. Dau, T.X. Dinh, T. Terebessy, T.T. Bui, Ion wind generator utilizing bipolar
discharge in parallel pin geometry, IEEE Trans. Plasma Sci. 44 (2016) 1–9.
environment condition. That’s the reason why the different values
[3] V.T. Dau, T.X. Dinh, T.T. Bui, C.D. Tran, H.T. Phan, T. Terebessy, Corona based air-
of normalized heat transfer coefficient had been reported in many flow using parallel discharge electrodes, Exp. Therm. Fluid Sci. 79 (2016) 52–
papers. Generally, the natural convection heat transfer coefficient 56.
is set in the range of 5–25 W/m2 K is appropriate [20]. [4] M. Goldman, R.S. Sigmond, Corona and insulation, IEEE Trans. Electr. Insul. 17
(1982) 90–105.
Fig. 12 shows that the ionic wind forced convection normalized [5] E. Moreau, G. Touchard, Enhancing the mechanical efficiency of electric wind
heat transfer coefficient is enhanced with the applied voltage on in corona discharges, J. Electrostatics 66 (2008) 39–44.
wire, mainly arising from increase of ionic wind velocity and drop [6] M. Rickard, D. Dunn-Rankin, F. Weinberg, F. Carleton, Maximizing ion-
drivengas flows, J. Electrostatics 64 (2006) 368–376.
of the velocity boundary layer thickness showing from the numer- [7] N. Takeuchi, K. Yasuoka, Efficiency of a wire-rod type electrohydrodynamic gas
ical result, which is the same with Jewell-Larsen et al. [46]. Besides, pump under negative corona operation, IEEE Trans. Plasma Sci. 37 (2009)
it increases with corona wire number at same applied voltage, e.g. 1021–1026.
[8] W. Qiu, L.Z. Xia, L.J. Yang, Q.G. Zhang, L. Xiao, L. Chen, Experimental study on
the value is 1.78 times larger than N = 1 compared with N = 4 at the velocity and efficiency characteristics of a serial staged needle array-mesh
U = 11 kV. As is presented in numerical result, this is primarily type EHD gas pump, Plasma Sci. Technol. 13 (2011) 693–698.
because the TBLT decreases and the velocity gradient increases [9] W. Qiu, L.Z. Xia, L.J. Yang, The velocity characteristics of a serial staged EHD gas
pump in air, IEEE Trans. Plasma Sci. 38 (2010) 2848–2853.
with the corona wire number, respectively. Meanwhile, compared [10] D.B. Go, S.V. Garimella, T.S. Fisher, Ionic winds for locally enhanced cooling, J.
with the simulation, the similar result from the experiment is Appl. Phys. 102 (2007) 1–8.
achieved, when corona wire number N  3, there is little or no pro- [11] J. Podlinski, J. Dekowski, J. Mizeraczyk, D. Brocilo, J.S. Chang,
Electrohydrodynamic gas flow in a positive polarity wire-plate electrostatic
gress on ionic wind forced convection heat transfer performance.
precipitator and the related dust particle collection efficiency, J. Electrostatics
The reasons have been given in Section 4.2 of this paper. 64 (2006) 259–262.
Fig.13 shows the relation of cylinder surface temperature and [12] E. Moreau, Airflow control by non-thermal plasma actuators, J. Phys. D-Appl.
the wire amount and the applied voltage on wire. The numerical Phys. 40 (2007) 605–636.
[13] N. Benard, E. Moreau, Electrical and mechanical characteristics of AC surface
result agrees with the experiment study, the error is less than dielectric barrier discharge plasma actuators applied to airflow control, Exp.
2.5 K. Under ionic wind forced convection, the cylinder surface Fluids 55 (2014) 1–43.
448 W. Wang et al. / Applied Thermal Engineering 119 (2017) 438–448

[14] C.O. Porter, J.W. Baughn, T.E. McLaughlin, C.L. Enloe, G.I. Font, Plasma actuator Annual Gaseous Electronics Conference and 7th International Conference on
force measurements, Am. Inst. Aeronautics Astronautics (AIAA) 45 (2007) Reactive Plasmas American Physical Society, 2010.
1562–1570. [31] N.E. Jewell-Larsen, S.V. Karpov, I.A. Krichtafovitch, V. Jayanty, C.P. Hsu, A.V.
[15] J. Wilson, H.D. Perkins, W.K. Thompson, An investigation of ionic wind Mamishev, Modeling of corona-induced electrohydrodynamic flow with
propulsion, NASA Report NASA/TM 2009–2158222009. COMSOL multiphysics, in: Proc. ESA Annual Meeting on Electrostatics, 2008.
[16] E. Moreau, N. Benard, J.D. LanSunLuk, J.P. Chabriat, Electrohydrodynamic force [32] T. Miyano, T. Moriya, H. Nagaike, N. Ikeuchi, T. Matsumoto, Modelling wire-to-
produced by a wire-to-cylinder DC corona discharge in air at atmospheric wire corona discharge action on aerodynamics and comparison with
pressure, J. Phys. D: Appl. Phys. 46 (2013) 475204. experiment, J. Phys. D: Appl. Phys. 41 (2008) 035205.
[17] D.B. Go, R.A. Maturana, T.S. Fisher, S.V. Garimella, Enhancement of external [33] K. Yanallah, F. Pontiga, A. Fernandez-Rueda, A. Castellanos, A. Belasri, Ozone
forced convection by ionic wind, Int. J. Heat Mass Transfer 51 (2008) 6047– generation using negative wire-to-cylinder corona discharge: the influence of
6053. anode composition and radius, Electr. Insul. Dielectric Phenomena (2008)
[18] B.L. Owsenek, J. Seyed-Yagoobi, R.H. Page, Experimental investigation of 607–610.
corona wind heat transfer enhancement with a heated horizontal flat plate, J. [34] M. Hamdi, M. Havet, O. Rouaud, D. Tarlet, Comparison of different tracers for
Heat Transfer 117 (1995) 309–315. PIV measurements in EHD airflow, Exp. Fluids 55 (2014) 1–12.
[19] A.O. Ong, A.R. Abramson, N.C. Tien, Electrohydrodynamic microfabricated ionic [35] M. Boulos, P. Fauchais, E. Pfender, Thermal Plasmas: Fundamentals and
wind pumps for thermal management applications, J. Heat Transfer 136 Applications, Plenum Press, New York, 1994.
(2014) 061703. [36] J.A. Stratton, Electromagnetic theory, McGraw-Hill, New York, 1941.
[20] F.P. Incropera, D.P. Dewitt, Fundamentals of Heat and Mass Transfer, Wiley & [37] E.W. Washburn, International critical tables of numerical data, physics,
Sons, New York, 2011. chemistry and technology, Norwich NYK novel Corporation (2003) 1926–
[21] D.F. Colas, A. Ferret, D.Z. Pai, C.O. Laux, Ionic wind generation by a wire- 1930.
cylinder-plate corona discharge in air at atmospheric pressure, J. Appl. Phys. [38] J. Chen, J.H. Davidson, Model of the negative DC corona plasma: comparison to
108 (2010) 103306. the positive DC corona plasma, Plasma Chem. Plasma Process. 23 (2003) 83–
[22] R.S. Sigmond, Simple approximate treatment of unipolar space charge 102.
dominated coronas: the Warburg law and the saturation current, J. Appl. [39] R. Morrow, The theory of positive glow corona, J. Phys. D Appl. Phys. 30 (1997)
Phys. 53 (1982) 891–898. 3099–3114.
[23] M. Goldman, A. Goldman, R.S. Sigmond, The corona discharge, its properties [40] A. Samusenko, Y. Stishkov, P. Zhidkova, Computer simulation of ionic wind in
and specific uses, Pure Appl. Chem. 57 (1985) 1353–1362. the point-torus electrodes system, J. Plasma Environ. Sci. Technol. 9 (2015) 24–
[24] C. Kim, D. Park, K.C. Noh, J. Hwang, Velocity and energy conversion efficiency 28.
characteristics of ionic wind generator in a multistage configuration, J. [41] F.W. Peek, Dielectric phenomena in high-voltage engineering, McGraw-Hill,
Electrostatics 68 (2010) 36–41. New York, 1929.
[25] L. Leger, E. Moreau, G.G. Touchard, Effect of a DC corona electrical discharge on [42] F.P. Incroper, D.P. Dewitt, Introduction to heat transfer, Wiley & Sons, New
the airflow along a flat plate, IEEE Trans. Ind. Appl. 38 (2002) 1478–1485. York, 1996.
[26] L. Zhao, K. Adamiak, Numerical simulation of the electrohydrodynamic flow in [43] Y.A. Cengel, Heat transfer: a practical approach, McGraw-Hill, Boston, 1998.
a single wire-plate electrostatic precipitator, IEEE Trans. Ind. Appl. 44 (2008) [44] B.L. Owsenek, J. Seyed-Yagoobi, Theoretical and experimental study of
683–691. electrohydrodynamic heat transfer enhancement through wire-plate corona
[27] L. Zhao, K. Adamiak, EHD flow in air produced by electric corona discharge in discharge, J. Heat Transfer 119 (1997) 604–610.
pin–plate configuration, J. Electrostatics 63 (2005) 337–350. [45] H. Kalman, E. Sher, Enhancement of heat transfer by means of a corona wind
[28] K. Adamiak, Simulation of corona in wire-duct electrostatic precipitator by created by a wire electrode and confined wings assembly, Appl. Therm. Eng. 21
means of the boundary element method, IEEE Trans. Ind. Appl. 30 (1994) 381– (2001) 265–282.
386. [46] N.E. Jewell-Larsen, C.P. Hsu, I.A. Krichtafovitch, S. Montgomery, CFD analysis of
[29] A.A. Martins, Modelling of an improved positive corona thruster and actuator, electrostatic fluid accelerators for forced convection cooling, IEEE Trans.
J. Electrostatics 71 (2013) 61–67. Dielectrics Electr. Insul. 15 (2008) 1745–1753.
[30] D. Colas, A. Ferret, I. Sciacca, D.Z. Pai, D.A. Lacoste, C.O. Laux, Novel electrode
configuration for ionic wind generation in air at atmospheric pressure, in: 63rd