Analysis of flow and heat transfer characteristics

of an asymmetrical flat plate heat pipe

K. VAFAI
Department

of Mechanical
Columbus,

(Received

and W. WANG

Engineering,
The Ohio State University,
OH 43210-I 107, U.S.A.

3 .Juiy 199 1 and in ,finul

.fom I 1 September

199 I)

Abstract-An
in-depth integral analysis revealing various physical aspects of an asymmetrical
flat plate
heat pipe is presented in this work. The resulting psuedo-three-dimensional
vapor phase Row field is
bifurcated on the X-J plane due to the asymmetrical
nature of the heat source and sinks. The analytical
results for the shifted vapor velocity profiles on the x-y plane, the overall axial pressure distributions
in
both vapor and liquid phases and the axial vapor temperature
distribution
in the heat pipe are obtained.
There is good qualitative agreement between the present results and those based on the solution of the field
equations for the conventional
symmetrical case. The analysis provides accurate analytical expressions for
velocity and pressure distributions
for this type of non-conventional
heat pipe.

7. INTRODUCTION

HEAT PIPEShave been widely used in heat transfer

related applications
for nearly 30 years. Theoretical
and practical research aspects concerning the transport mechanism, the structure of the container, wick
and materials used in the heat pipe have been well
established for many applications,
such as spacecraft
thermal control, electronic systems cooling and many
commercial thermal devices. The fluid flow within a
heat pipe dealing with the vapor flow in the core region
and condensate Row in the wick is an important topic
in most of these research studies. Most of the investigators have concentrated
on the study of conventional heat pipes, which have a symmetrical
heat
source and sink. The finite difference method has been
extensively employed to solve the governing differential equations in the vapor phase for such cases
[l-4]. But, the asymmetrical
characteristics
of a heat
pipe have been mostly neglected.
A comprehensive
review of the physical phenomena
and applications of heat pipes is given by Tien [5] and
Winter and Bars& [6]. Tien and Rohani [I] have
presented the numerical results displaying vapor pressure distributions
and heat pipe performance
for
axisymmetrical heat pipes, and general trends for pressure drop in the evaporator
section and pressure
recovery in the condenser section. Further numerical
results, including
the pressure and velocity distributions, were reported by Narayana [2], pointing out
that an increase in suction rate in the condenser section produces a vapor flow reversal. Faghri [3] has
presented his results for a double-walled
concentric
heat pipe which indicate that the pressure distribution
and velocity profile for a double-walled
concentric
heat pipe are almost the same as a conventional
heat
pipe. There have been several analytical and experimental studies of the heat pipe phenomena.
For

example, Udell [7] reported

analytical and experimental studies for a one-dimensional,
steady state
problem on the heat pipe effect in porous media. In
the model developed, gravity forces, capillarity and
change of phase were considered. Leverett correlation
relating capillary pressure to pernleability,
porosity
and surface tension was also used in this study. Relative permeability
correlations
found by Fatt and
Klikoff [S] were also employed. Udeli and Fitch [9]
reported a further study considering the effects of noncondensible
gases. In their work, the effect of noncondensible gases on Kelvins equation is explained
and utilized.
In this work, a detailed analysis of the in&a-wick
interactions
and integral solutions for the velocity
and pressure fields within the flat plate heat pipe are
presented for the first time. The integral analysis is
employed to investigate the vapor and liquid flow
in a flat plate heat pipe heated asymmetrically.
The
resulting psuedo-three-dimensional
vapor phase flow
field and the bifurcation
phenomena on the x--;y plane
due to the asymmetrical nature of the heat source and
sinks have been vividly illustrated and the analytical
results for the shifted vapor velocity profiles on the
X-;v plane, the overall axial pressure distributions
in
both vapor and liquid phases and the axial vapor
temperature distribution in the heat pipe are obtained.
A specific application will be considered for the flat
plate heat pipe used in our analysis, namely Boron
Neutron Capture Therapy (BNCT). For this application, the protons which are accelerated by an accelerator bombard a lithium target to produce the neutrons which are used for treatment of brain tumors.
The flat plate heat pipe configuration will then be used
for removing a high heat flux generated as a result of
proton bombardments
of the lithium target. The flat
plate heat pipe will be used as an effective heat sink
as well as an additional moderator
for the neutron
2087

2088

K. VAFAI and W.

WANG

NOMENCLATURE
4

evaporator area [m]

half width of any of the vapor channels

rw

[ml

TLi

position of maximum value of vapor

velocity in y direction [m]
,f +(.r+) dimensionless position of the
maximum vapor velocity in y+
direction
height of vapor space for the heat pipe
h

Re,

,f(.u)

hr.1
L,,
12%
&+
ir:
lrri:
L
k,
k,,,,
K
K
I
I+
42
1,
PI

Pb
PI+
+
P
&&I
44
&A,,,,
cIl1,
A/4+
AP:
6/e

Iml
thickness of the lithium disk [m]
thickness of the heat pipe wall [m]
thickness of the wick [m]
dimensionless half width of any of the
vapor channels, b/h
dimensionless thickness of the wick,
/1,//I
latent heat of working fluid [kJ kg- 1
thermal conductivity of the lithium
[Wm-
C 1
thermal conductivity of the wick
[Wm-
C]
thermal conductivity of the heat pipe
wall [W rn- C- 1
permeability [m]
dimensionless permeabiiity, K/h;
length of the heat pipe [m]
dimensionless length of the heat pipe
width of the heat pipe [m]
length of the evaporator [m]
liquid pressure [Pa]
vapor pressure [Pa]
dimensionless liquid pressure, p,/p, r/i,
dimensionless
vapor pressure,
P,/P UL
capillary pressure head generated by the
wick [Pa]
overall liquid pressure drop along the
heat pipe [Pa]
total pressure head in the heat pipe [Pa]
overall vapor pressure drop along the
heat pipe [Pa]
overall dimensionless liquid pressure
drop along the heat pipe, p: -&
overall dimensionless vapor pressure
drop along the heat pipe, p: -pot
imposed heat flux on the evaporator
[Wmm2]
rate of heat transfer generated by
neutron bombardment
[W]

production. Expressions for the pressure and velocity

distributions
are obtained and discussed in relation
to the BNCT application.
Qualitatively,
our results
compare quite favorably with the results reported by
other researchers [IO, I I].

porous radius of wick [m]

injection Reynolds number, D,h/rV
free surface temperature of the lithium

1Cl
T,
T,v
ALi
ATw
AT,,,,
AT:
w
11,
u:
11:

U,I
UOY
CJ:
L1
F2
0:
r,+
2

vapor temperature
[CJ
dimensionless saturate vapor
temperature,
T,,,h,/R
tem~rature
drop across the lithium
disk [Cl
temperature drop across the wick [Cl
temperature drop across the heat pipe
wall [Cl
dimensionless vapor temperature drop
along the heat pipe, T+ - T,t
liquid velocity [m s- 1
vapor velocity [m s- 1
dimensionless liquid velocity, u,/UO,
dimensionless vapor velocity, u,,/Uoy
maximum liquid velocity [m s 1
maximum vapor velocity [m s- 1
dimensionless vapor velocity
component in X+ direction
vapor injection velocity [m s- 1
vapor suction velocity [m s- 1
dimensionless vapor injection velocity,
c,h/v,,
dimensionless vapor suction velocity,
zl,h/v,.

Greek letters
liquid viscosity [Ns m 1
I4
vapor viscosity [Ns m- 1
11
dimensionless viscosity, P~/I(,
/1+
kinematic viscosity of vapor [ms_ 1

liquid density [kg m3]

PI
vapor density [kg m- 1
pV
surface tension of the heat pipes
61
working liquid [N rn-- 1
ratio of the evaporator length to the
yl
heat pipe length.
Subscripts
C

1
V
W

condenser
evaporator
liquid phase
vapor phase
wick.

Superscript
dimensionless
+

quantities.

The selection of the material which will be used in

manufactu~ng
the heat pipe also needs some careful
consideration.
This is because the interference of the
reflected (or outwardly diffused) neutrons (which are
produced as a result of the interaction
between the

Flow heat transfer characteristics

protons and the lithium) with certain materials produces gamma rays which is unacceptable and must be
avoided. For this reason heavy water (D,O) is chosen
as the operating fluid for the heat pipe and the heat
pipe material itself is chosen to be aluminum. In the
present work the heat pipe will be designed such that
it would be able to comfortably remove the heat generated as a result of the proton bombardments of the
lithium target. Since a large amount of energy falls
upon a relatively small area of the lithium surface,
we need to ensure that there will not be any flash
evaporation. The present analysis clearly shows that
it is possible to attain temperatures well below the
melting point of the lithium for the high heat flux
situation in the BNCT application. This in turn prevents any adverse pressure build up within the system.
2. ANALYSIS

AND

FORMULATION

The schematic diagram of the heat pipe for the

present investigation is shown in Fig. l(a). It can be
seen that the heat pipe is mounted symmetrically
under the lithium target area. A cooling jacket is used
around the outer periphery of the heat pipe for removing the dissipated heat. Based on our analysis the
surface temperature of the heat pipe turns out to be
low enough that the fluid inside the cooling jacket can
pass through an appropriate cooling loop. This is
quite advantageous since it will result in a closed loop
system thus further reducing any possibility of an
accidental contamination
of the surroundings. As
shown in Fig. 1(a), the heat generated from the proton
bombardment of the lithium is imposed on the upper
part, which is then transferred equally to both sides
of the heat pipe. The multiple independent vapor flow
channels and the coordinate system for the flat plate
heat pipe are shown in Figs. l(b) and (c). The generated heat causes the vaporization and subsequent
pressurization of the liquid in the wick. Part of the
vapor is condensed on the opposite surface of the
evaporator, however, most of it is condensed downstream of the evaporator on both sides of the condenser region.
Obviously, the vapor flow in the heat pipe does not
have a symmetrical velocity profile. The location of
the maximum vapor velocity will be shifted towards
the tower plate due to the vapor injection from the
heating side of the upper plate. As the vapor flows
downstream, the location of the maximum velocity
will gradually shift towards the center line due to
the presence of symmetrical cooling conditions. The
vapor space is divided into several channels by the
vertical wicks which transport liquid from the lower
wick to the upper wick. Therefore, the vapor velocity
along the longitudinal direction will be a function of
the x, JJ and z directions. The condensate in the upper
wick (section 2 of Fig. 2) flows back directly to the
evaporator
along the wick by capillary force.
However, in section 4 of the lower wick, the condensate will move directly along the horizontal wick

2089

towards section 3 of the lower wick, then laterahy

towards the vertical wicks and upwards through them
towards the evaporator section. Therefore, within
0 < x < tpl region (where rp will be established later),
the liquid in the upper and lower wicks possesses two
distinct velocity components, U,in the x direction, and
W, in the z direction, since, in this region, the upper
wick (evaporator) gets part of the liquid supply from
the lower wick. However, the liquid will have only one
velocity component, u,, along the x direction in the
cpl < x 6 1 section, since both upper and lower wicks
act as condensers. That is over the $6 x < 1 region
due to the symmetrical boundary conditions, there is
no liquid exchange through the vertical wicks. In the
heat pipe research performed to date, researchers have
made some assumptions in modeling and analysis. An
important assumption made by various researchers is
that the capillary porous wick is always saturated with
liquid phase working fluid and vapor flows only in the
core region during the operation of the heat pipe.
Hence, evaluation of the effective thermal conductivity and formulation of the capillary pressure
were simplified. This assumption will be employed in
the present investigation. In order to obtain the integral solutions, we will make some common assumptions which are usually made in analyzing the heat
pipes. The assumptions made in this work are :
(1) Vapor and liquid flow are steady, laminar and
subsonic.
(2) Transport properties for the vapor and liquid
are taken as constants.
(3) The vapor injection and suction rate are uniform in the evaporator and condenser sections.
(4) The vapor velocity component in the z direction
is negligible since there is no injection or suction on
vertical wicks.
2. I. Vapor phase analysis
Any one of the internal channels can be considered
as a building block for the flat heat pipe. Therefore,
we will concentrate on fluid flow considerations within
one of these channels (Fig. l(c)). Once the fluid
characteristics within one of the channels is determined, the fluid flow characteristics of the entire heat
pipe can be easily established. In our analysis the
position coordinates, velocities and pressure are nondimensionalized by h2U~~l~~, h, b, U,,, v,/h and
pyUib in the vapor phase, and by ~~~,~~~,,Uor, p,Uz,
in the liquid phase and the temperature is non-dimensionalized by h,/R factor, where v, and v, are the
kinematic viscosities, U,, and UO,are the maximum
vapor and liquid velocities, p and p, are the vapor
and liquid densities, h and b are the height and half
width of the vapor space of one vapor flow channel,
h, is the thickness of the wick and R is the ideal gas
constant. Based on the numerical results given by
Narayana {Z], Faghri [3] and Sorour et al. [12], a
parabolic velocity profile will be used for vapor flow
within the heat pipe. The dimensionless velocity distri-

K. VAFAI and W. WANG

2090

A
I \

,
I

Proton Beam

Lithium Target

(a)

(b)

the BNCT system. (a) Schematic of the proposed BNCT system

(b) Parametric dimensions of the heat pipe. (c) The coordinate system used in the analysis.

FIG. 1. Flat plate heat pipe within

bution u,+(x+, y+, z+) will be represented by a functional product in the x+, y+ and z+ directions. That
is
U~(X,~,Z)

= v,(n+)[a,+a,y+

+~*w)21k*+c,z+ +&+)*I

(1)

where Uyf(~+) denotes the maximum velocity for

u:(x+, y+, z+) on every transverse surface along the
x+ axis. The location of the maximum velocity will
be shifted towards the lower plate due to the vapor
injection from the heating side of the upper plate.
Because of the symmetrical conditions in the Z+ direc-

209 1

Flow heat transfer characteristics

FIG. 2. IXfferent regimes which are analyzed

for the vapor and liquid phases in the .X-J- plane of the heat
pipe.

tion (i.e. non-slip boundary conditions

at Z+ = -1
and Z+ = I), the constants cO. c, and cZ for the velocity
component in the Z+ direction are easily specified as
[C,,+c,Z+ +cz(Z+)]

= [l -(=)I

(2)

where Z+ = z/b. Since the velocity component in the

I+ direction given by equation (2) is uncoupled from
X+ and ,v+, the two velocity components in the X+ and
_v+directions can be determined by concentrating
only
on the .Y+-.Y+ plane. Therefore, the velocity profile in
the .Y-y+ plane is

U:(.~+)[U,,$.a,y++a,(y+)*].

ut(x+,y+)=

= 0,

u,+(X+,,f+(X+))

au: (x+, _V)

iV
_

i.+_ {+r, )

< p

(x)).

_p)

= u:(x+),
= 0.

>+= I+ (X )

The use of the above boundary conditions in equation

(3) will result in the following profile for the upper
part

(.f+(x+) < y < 1).

(5)

To obtain the complete velocity distribution,

the two
unknowns U,i-(x) and ,f(x-) in equations (4) and
(5) must be determined. First, Uz (.x+) is determined
by integrating the continuity equation with respect to
y+ from 0 to 1, i.e.

= i&!-(.x).
= 0,

Applying the above boundary conditions to equation

(3) will result in the following velocity profile for the
lower part

(0 < x

&,t(Xf,
++

u:(x+,f+(x+))

(3)

Due to the asymmetry

of velocity distribution
in
the X-Y+ plane, the velocity profile will be divided
into two parts in the y+ direction (Fig. 2), the lower
part
(0 < _v+ < ,f(x))
and
the
upper
part
(,f+(x+) dy+ < I), by the curve JJ+ =,f+(x+),
where ,f+(-u) = ,f(s)/l~. The curve r+ = f(x+)
is
the location corresponding
to zero shear stress for
the velocity distribution
on the x+-y+ plane. For the
lower part (0 < y+ <f(x)),
the boundary
conditions arc
u:(x+,o)

u,f(x-, I) = 0,

(4)

For the upper part (.f+(~+) < y+ < I), the boundary
conditions are

Substituting equation (4) into the first part of the right

hand side of equation
(6) (lower part (0 6 !a+ <
,f(.r))
and equaLion (5) into the second part of the
right hand side ofequation
(6) (upper part (,f+~(.u+) <
I+ < 1). Also noting that in the 0 < x+ < (ol+ scction, u; (.u+, 0) = 0, u, (.x+. 1) = 0, VT(~1 = 0) =
-z>: and E: (_Y+= I) = --CT, and in the cpP <
x+ d /+ section, u:(.x+,O) = 0 and u:(x,
I) = 0,
z$(y+ = 0) = -2::
and ~T(Y+ = 1) = -+-I:: will
result the final expression for U;(.u+)

2092

K. VAFAI and W. WANG

U:(x+)=

/+I.\+)

3(c:-c+) +
2.x

(0<x+ 6 cpl+)

(-3$(x+--I+)

(ql+<.x+ <If).

(UT (x+, y))

(7)

To obtain the above equation, equation (6) is used

once for the 0 < x+ < cpl+ section and once for the
cpl+ < x+ < I+ section subject to proper boundary
conditions. As expected, U:(x)
is related to vapor
injection and suction velocities. Recognizing the fact
that at x+ = cpl+, C/$(x) must be continuous will
result

(8)

dY+

/+(x+1 ~u~(x+,y+)

II

p-Tm-dy+
a(Y )

(cpl+ < x+ d I+).

(10)

Using the vapor velocity profile (equation

average vapor pressure in each transverse
equation (10) will result

(4)) and the

direction in

Clf+(x)
This relation can also be obtained through a mass
balance on the whole vapor space and therefore is
independent of the velocity profile. Using the relation
(8) and the definition of the dimensionless transverse
velocity c: = r, h/v, as the injection Reynolds number
Re,,where u, is related to the input power through
r , = Q/p,A,h,, equation (7) can be written as
+

3(1 -cp)Re,
--727x

u:

dx+

(0 < x+ < rpl+)

(x+) =
- 2:;

(0

(cpl+ < x+ < I+).

(.*--If)

_!!!F$+(x+)+jT;~ g5(I+-x+)

(9)
The f(x ) will be determined by integrating the x+momentum equation with respect to Y+ for either the
lower or upper part of the heat pipe. This is because
the integration
of the momentum
equation for the
upper or the lower part in the x+-Y+ plane produces
identical results. It should be noted that in the
0 d x+ < cpl+ region, there is mass flow crossing the
interface between the two control volumes I and III
in the vapor phase (Fig. 2) due to the vapor injector
from the upper wick and suction from the lower wick.
However, in the cp/+ < x+ d I+ region, due to the
symmetrical boundary conditions on both upper and
lower wicks, there is no mass exchange between the
two control volumes II and IV. Since the longitudinal
length is much larger than the transverse length in a
heat pipe, the axial sheer stress will be neglected [2,
31. The above mentioned considerations
will result in
the following set of equations for the lower part of
the vapor space

Since U,?(O)
= U,'(l+)
= 0, and the velocity distributions at x+ = 0 and X+ = I+ are uniform, the
boundary
conditions
for the function f (x)
are
taken as ,f(O) = ,f(/+) = l. The use of equations
(2), (4), (5), (9) and (11) in equation (I) will provide
us with the complete veiocity profile within the flat
heat pipe assembly.
To obtain the pressure distribution
in the vapor
phase, the integrated x+-momentum
equation will be
used. Integrating the x+-momentum
equation within
a channel bounded by porous wicks results in the
following equation in which i = 1 corresponds
to the
evaporator section and i = 2 corresponds
to the condenser section

I ss
$1

/+(\I

-=z
x+< qd+)

~~.
qu:)

-1

ax+

dx+

dy+

dz+

IL,(x, Y) dY+
s0
I+(\+) a?:
-u:

(x+)c;

= s0

/+(.r+) zu,t(x+,
-amm
s0

ax+dYi

y+)
dy+

(0 <x + d cpf+)

I
a*u+
+ (h,+jT Y&$

dx+ dy+ dz+.

Flow heat transfer characteristics

i=

1,2

(o<x+

(2)

<I+)

Introducing equations (2), (4), (5), (9) and (11) into

equation (12) and considering the average vapor pressure at each cross-section,
and carrying out the integration will result

2093

From Fig. 2, it can be seen that the liquid velocity

u,+ is related to both vapor injection and suction velocities. Adding the lower and upper wick mass conservation equations for the section 0 < X+ < (~1 to
cancel out w,+ in z+ direction will result in the following one-dimensional
velocity distribution
in the x+
direction

Appy(x+) =
= hW+p+(u: -0:)

Zd$($(x+))
r

4(1-V)&

16(lMRe
25(2--q)

(2-V)
Y+
+

s0

+
h

(16)

X+

.f+(-~+)u

where pi = L/K and h,f = h,/h.

cpl+ ,< x+ < 1 section, we obtain

-.f.+(x+))dY+

= -h,fp+u:

&(z$(x+))
4(?
< AP(@+) - (2-0

Rei,

(0 < x+ < qd+)

1
~
(x)
2(h,+)2 1
While

(ql+

for

the

< x+ < I+).

16
cp
{[ 25 (2 - cp) Reh

(17)

1
x+--l+
_s0Y+
.fT(')(l-J+(X+))dx+
Subject to the boundary

- 2(1;+ji

[(x+ -r+)-(cpl+

conditions

-I+)21

UT(O) = Ui+(l) = 0

the above equations

distribution as

($I/+

<x+ < If).

(13)

The numerical solutions for the coupled equations,

pressure Ap,?(x)
and J+(x+), are then obtained
through an iterating procedure involving equations
(11) and (13). An approximate
solution for the
,f+(x+) can be found if Bernoullis equation is used
sty+ = ,f(x).
This will result

provide the liquid phase velocity

h,+P+(v: -Gx+

u: (xf

)=

(0 <x+ < (pl+)

2
(h,f pu:

(I+ -x+)

(cpl+ <x+

< I+).
(18)

At x+ = cpl+, u,+(x) given by both sides of the above

equation should be equal. This gives the value for cp as

df+(x+) _
dx+

q=-

2v:
u: +tQ

(0 <

x+ < cpl+)

_~f+(_l+)+()P2!?

Re,

As expected, this is the same as the result obtained from

the vapor phase analysis (equation (8)). The pressure
distribution in the wick in the x+ direction is obtained
by substituting equation (18) into Darcys law, equation
(15). This will result

(qd+ <

f +(x+)

x+ < I+).

7p(l+

Ap,+(x+) =

1
-x+)

(14)

The above approximate

solution for f+(x+) is used
as the initial guess in our iterative numerical scheme.
2.2. Liquid phase analysis
Within the thin porous
applied. The dimensionless

wick, Darcys law will be

Darcys law is

1
aP,+
__=-n+
ax+
K+

(15)

where K+ = K/hi. It should be noted that the boundary and inertia effects are neglected in this work [13].

r
~~

I+ h,+p+

px+

dx+

----&(I+-x+)dx+

s y?/+ K+

(0 < x+ < cpl+)

(cpl+ < x+ < I+).

(19)

The boundary condition for equation (19) is obtained

by using the following physical assumption [5,6, 11, 141
Ap,+ (I+) = Ap;(Z+).
Integrating (19) subject to the above boundary condition
results in the pressure distribution in the liquid phase

K. VAFAI and

2094

h,+pL+(I -cp)Re,
App: (I+) _ __~~~~

{CPU -cp)U)

2(2-qn)K+

[(qd+)- (,K+)qj

APT = P,(O)

(0 ,< x+ < qd+)

Therefore,

h: p+cpRe,,

(If -.x+)2

2(2-q)K+

(q/t

(20)

The total capillary

is given by

distribution

Once the vapor pressure distribution

is found, the
vapor temperature
distribution
can be specified by
using the ClausiussCIapeyron
equation. In this work,
the vapor is assumed to behave as an idcal gas and
the voIumc of the liquid within the vapor phase is
neglected [6, 121. The dimensionless temperature protile based on the ClausiussClapeyron
equation can be
found from

is thedimensionless

satu-

where
Apto,;,, = AP\ + A/J, + A~pp.

head ApP can be neglected

0.0

0.1

0.2

pressure

for the

(24)

head along the heat pipe

where G, denotes the surface tension of the liquid in

the wick, and rw is the porous radius of the wick.
Using the results given by equations (13) and (20), we
will show in the next section that the operational
criterion for the heat pipe is indeed established.
In
what follows it will be shown that the boiling limit
criterion is also satisfied. For a water heat pipe near
atmospheric pressure, the critical heat flux for boiling
limit is of the order of IO Wrn~- . However, for our
case. the critical heat flux is [5]

which is almost an order of magnitude higher than

the maximum heat flux imposed on the heat pipe.
Therefore, both the capillary and boiling limit criteria
are satisfied for high heat flux situations such as the
BNCT application.

0.3

0.4

0.5

0.6

0.7

0.6

Dimensionless

= P,(O)-p,(O).

(25)

(22)

as

Based on Tien [5], Dunn and Reay [IO], Ivanovskii rt

[I41 the wetting angle between
the liquid and wick in the evaporator section is 0, = 0
whereas in the condenser section this wetting angle is
0, = n/2. Therefore the maximum capillary pressure
head of ApPcapcan be written as

2.4. Operational criteria

In general. the operational limits which need to be
satisfied for a heat pipe under steady state and low
temperature
situations [5] are usually based on the
boiling and capillary limit criteria. First, the capillary
operational limit is established. To satisfy the capillary
limit. the maximum capillary pressure head (ApPc.lP)m,lr
must be greater than the total vapor, liquid and gravitational pressure drops [5, IO, I I]. That is

0.0

head is presented

al. [I I] and Chisholm

Inpz2(x+)-lnp:
AT, (_v+) = (To:,)? ~~ ~~_..
~~~~
I - TA(Inp,f,-lnp,f(x+))

The gravitational

the total pressure

(23)

< x+ d I+).
A~m,.i, = AP\ +AP,

where T,,(.u+) = T,,,.(h,,/R)

rated vapor temperature.

-pv(O

AP, = p,(I)-P,(O).

A/C (I+)

2.3. Temperature

WANG

horizontal heat pipe, and the overall vapor and liquid

pressure drops are established as

Ap,+(.u+) =

c+

W.

length of the heat

FIG. 3. Variations

pipe ( x*/l*)

of U: (x+) vs x/i.

0.9

1.0

Flow heal

0.0

0.1

0.2

characteristics

lranskr

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Dimensionless length of the heat pipe ( x+/I+)

FIG. 4. Variations

3. RESULTS

AND

of ,f(.x)

for different

DISCUSSION

The results presented here are based on a flat heat

pipe with heavy water as the working fluid. The pipe

0.0

0.1

0.2

0.3

0.4

values of injection

Reynolds

number,

Re,.

dimensions based on our application

(see Fig. l(b))
are chosen as : I = 0.46 m, lh = 0.25 m, b = 0.0125 m,
and h = 0.0254 m. The evaporator
section, 1, =
0.101 m, is only on the top surface of the heat pipe.

0.5

0.6

0.7

0.8

0.9

1.0

Dimensionless length of the heat pipe ( x*/l+)

FE

5(a). Effects of the injection

Reynolds

number

on the velocity

distributions

(I

= 0).

Dimtnsionfess icngrh of the heat pipe ( x*/l+)

FIG. 5(b). The velocity distributions at different lateral locations for Re, = 1460

The rest of the heat pipes external area acts as the

condenser section. The maximum heat Aux at the
evaporator section of tbe heat pipe can then be calculated as
qe = f

= 1.8293 x lo6 Wm- = 182.93 Wcm-2.

c
(26)

Since the light water will hamper the traverse of neutrons, heavy water (D&I), which has almost the same
thermo~hys~cai properties except having a larger density than light water, is chosen as the working fluid
for the heat pipe. The material for the container and
the wick is chosen to be aluminum. The heat pipe
must be designed such that the lithium target will be
solid at all times. Since the melting temperature for
lithium is 180C. the target will remain solid if the
vapor temperature is contro~ied at Tvap, = 80C. This
can be seen through examination of the corr~spondjng
lithium temperature
r,i = Tvnporc A 7wlck
+ A Tw,,,+ A rr_i < 18oC
where

(27)

and where h,, = I x tow3 m, h,,lr = 2 x IO- m and

h, = 2 x 10 m are &hethicknesses of l~~h~urntarget,
the container watt and the wick, respectively, and
lcL,= 42.8 W m- C- , k,,l = 236 W me- C- and
k wick= 94.65 W mm Xl- are thermal ~oilductivit~es
of the lithium, aluminunl and the wick, respectively.
Based on the above calculations, the vapor temperature should be less than 88.9C to satisfy equation
(27). Therefore, the operating vapor temperature is
chosen to be 80C to ensure that there will be no phase
change in the lithium layer.
The thermophysical properties of the heavy water
used in our calculations (at 7: = 80C) are : h, = 2128
kJkg.,hb,== l.1876x10~Nsm~2,~,=41.6x10~5
Ns mm2,py = 0.3055 kg rn-- 3and p, = 1078.3 kg rne3.

Flow heat transfer characteristics

The saturated
vapor
pressure
corresponding
to
TV = 80C isp,,, = 4.44 x lo4 Pa. The imposed energy
rate based on the BNCT application is Q = 75 kW,
which will result an average injection velocity in the
evaporator section

Q = 2.813mss.
A&p,

u, = ~

This will correspond to an injection Reynolds number

of Reh = 1460 for Q = 75 kW. As seen from equations
(9) (ll), (13), (18), (20), thef+(x+)
and non-dimensional velocity and pressure distributions
for both the
vapor and liquid phases are functions of two pertinent
parameters.
These are: Re,, which represents
the
input power, and cp, which represents the geometric
ratio of the length of evaporator to the entire length
of the heat pipe. Equations (11) and (13) were solved
by using a standard
Runge-Kutta
method.
The
numerical
results for functions
U,(x+),
f+(x+),
vapor velocity distributions
u: (x+, y+, z) at different Z+ locations, average vapor and liquid pressure

2097

distributions
and the average vapor temperature
distribution are shown on Figs. 3-7.
Figure 3 demonstrates
the variation of Ut(x)
along the x+ axis. It can be seen that UT (x+) increases
in the 0 < x+ < cpl+ region corresponding
to the
decreases
in
the
evaporator
section
and
cpl+ < X+ < I+ region corresponding
to the condenser section due to the vapor injection (increasing
mass) and suction
(decreasing
mass) over the
corresponding
regions. This variation of U,(x)
is
linear because of uniform vapor injection and suction
velocities. This behavior
is similar to the results
reported by Bankston and Smith [4].
Figure 4 illustrates the functional distribution
for
the location
for the
J+(X+), which represents
maximum value of u: (x, y), at different injection
Reynolds numbers. It can be seen that right after
x+ > 0, the maximum velocity is profoundly
shifted
towards the cooling side. As expected for larger values
of Re,. the maximum vapor velocity will be shifted
more prominently
towards the cooling side, and it
also takes a longer length before f '(x')approaches

14
28

42

;;

56

a
&.

%
%

70
a4
98

126

0.0

0.1

02

0.3

0.4

0.5

0.6

0.7

0.8

Dimensionless length of the hear pipe ( x+/I+)

FIG. 6. Vapor and liquid pressure distributions

along the heat pipe,

0.9

1.0

K. VAFAI and W. WANG

0.12

0.08 -

?
2

0.00
-0.04 -

x+-no

-0.08 -0.12
0.0

R-P1460

I
0.1

I
02

0.3

0.4

0.5

0.8

0.7

0.8

0.9

1.0

Dimcnsionllss length of&e heat pipe ( x*/l)

Fit.

7. Vapor temperature

profiles for different

the center line. Figure 5(a) clearly shows that for

smaller Reynolds numbers, the symmetrical transition
of the velocity profile will occur earlier. Figure 5(b)
shows the x-_v+ velocity distribution at different lateral locations for Re, = 1460.
To obtain a better feel for the operational criterion,
the solutions for the pressure and temperature
distributions are presented dimensionally.
Figure 6 presents the vapor and liquid pressure distributions along
the heat pipe at different Reynolds numbers. These
profiles show the effects of the pressure head loss in
the evaporator
region due to vapor velocity acceleration in that region, and the pressure head recovery
in the condenser region due to the deceleration of the
vapor velocity in that region. In the liquid phase, there
is only pressure head loss, since the flow is dominated
by the bulk frictional resistance from the porous
matrix.
As expected,
the smaller the Reynolds
number, the smaller the pressure head loss. The qualitative behavior of the overall pressure distribution
along the heat pipe is quite close to results presented
by Tien and Rohani [ 11, Winter and Barsch [6], Dunn
and Reay [lo] and Ivanovskii et al. [ 1I]. Figure 6
demonstrates
the variations for the mean vapor temperature along the longitudinal direction at different
injection Reynolds numbers. It can be seen that the
temperature
is quite uniform (AT<< 1C) even for
the highest input power (Rq, = 1460). This is consistent with the numerical and experimental
results
reported by Chen and Faghri [ 151. This further demonstrates the successful operation of the heat pipe [6].
Finally, the capillary operational
criterion for the
heat pipe can be contrasted from the numerical results
given on Fig. 6. The total pressure drop in the heat
pipe is

A/I,,,,, = 133 Pa

injection

Reynolds

numbers.

corresponding
to Re, = 1460 (Q = 75 kW).
equation (25) results
(Apc;,Jman = tz

Using

= 545 Pa.

Obviously,
(A~PcaJman> App,,,,,. Thus the capillary
operational
criterion for the heat pipe under the
design conditions dictated by the BNCT application
is also satisfied.

4. CONCLUSIONS

An in-depth integral analysis for an asymmetrical

flat plate heat pipe is presented in this work. Analytical
results for the shifted vapor velocity profiles on the
.X-Y plane, the overall axial pressure distributions
in
both vapor and liquid phases and the axial vapor temperature distribution
in the heat pipe are obtained.
The feasibility and the pertinent operational criteria
for using the flat plate heat pipe for high heat flux
situations such as the BNCT application
were also
investigated and it has been shown that the heat pipe
can be used as an effective heat sink for removal of
the heat generated from proton bombardments
of a
lithium target as well as an additional moderator for
the neutron production.
Ackno~~~ledgemmt.v-This work was supported
by the
Department
of Energy under contract number DE-FG0289ER60872. The authors would like to thank Dr Thomas
Blue for his help on this work.

REFERENCES
1. C. L. Tien and A. R. Rohani, Analysis of the effect of
vapor pressure drop on heat pipe performance,
Int. .I.
Heat Mass TranJ:fer 17,61--67 (1974).
2. K. B. Narayana,
Vapor flow characteristics
of slender

Flow heat transfer

3.

4.
5.
6.
I.

8.

9.

cylindrical
heat pipe-a
numerical approach,
Numer.
Heat Transfer 10,79-93 (1986).
A. Faghri, Vapor flow analysis in a double-walled
concentric heat pipe, Numer. Heat Transfer 10, 583-595
(1986).
C. A. Bankston
and H. J. Smith, Vapor flow in cylindrical heat pipes, J. Heat Transfer 95,371L376 (1973).
C. L. Tien, Fluid mechanics of heat pipes, Ann. Rev.
Fluid Mech. 7, 167-185 (1975).
E. R. F. Winter and W. 0. Barsch, The heat pipe, Ado.
Heat Transfer 7,219-320 (1971).
K. S. Udell, Heat transfer in porous media considering
phase change and capillarity-the
heat pipe effect, Int.
J. Heat Mass Transfer 28,485495
(1985).
1. Fatt and W. A. Klikoff, Effect of fractional wettability
on multiphase flow through porous media, AIME Trans.
216, 256 (1959).
K. S. Udell and J. S. Fitch, Heat transfer in porous media
and particulate
flows, ASME HTD 46, 103-l IO (1985).

ANALYSE

characteristics

2099

10. P. D. Dunn and D. A. Reay, Heat Pipes, 3rd Edn,

Chauter 2. Pereamon Press. New York (1982).
Il. M. N. Ivanovckii, V. P. Sorokin and 1: V. kagodkin,
The Physical Principles of Heat Pipes, Chapters
1. 2.
Clarendon Press, Oxford (1982).
Devel._.
12. M. M. Sorour. M. A. Hassab and S. Estafanous,
oping laminar flow in a semi-porous
two dimensional
channel with nonuniform
transpiration,
Int. J. Hear
Fluid Flow 8,44-53 (1987).
13. K. Vafai, Convective flow and heat transfer in variable
porosity media, J. Fluid Mech. 147,233-259 (1984).
14. D. Chisholm, The Heat Pipe, Chapter 2. Mill & Boon,
London (1971).
15. M. M. Chen and A. Faghri, An analysis of the vapor
flow and the heat conduction
through the liquid-wick
and pipe wall in a heat pipe with single or multiple
heat sources, Int. J. Heat Mass Transfer 33, 1945-1955
(1990).

DES CARACTERISTIQUES
DECOULEMENT
ET DE TRANSFERT
CHALEUR
DUN CALODUC
ASYMETRIQUE
PLAT

DE

R&rum&Une
analyse integrale en profondeur
revele differents aspects physiques dun caloduc asymetrique
plat. Le champ pseudo-tridimensionnel
dtcoulement
de la phase vapeur est bifurque sur ie plan x-y a
cause de la nature disymttrique
des sources et des puits de chaleur. Les rtsultats pour les profils de vitesse
en .x-y, les distributions
axiales de pression dans les phases vapeur et liquide et la distribution
axial de la
temperature
de la vapeur sont obtenus. 11 y a un bon accord qualitatif entre les resultats present&s et ceux
bases sur la resolution du cas symetrique conventionnel.
Lanalyse fournit des expressions analytiques
precises pour les distributions
de vitesse et de pression dans le type non conventionnel
de caloduc.

UNTERSUCHUNG
VON STROMUNG
UND WARMEtiBERGANG
IN EINEM
ASYMMETRISCHEN
PLATTENFORMIGEN
WARMEROHR
Zusammenfassung-In
der vorliegenden
Arbeit werden die unterschiedlichen
physikalischen
Gesichtspunkte eines asymmetrischen
plattenfiirmigen
Wirmerohres
eingehend untersucht.
Es ergibt sich ein
pseudo-dreidimensionales
Striimungsfeld
fur den Dampf, das wegen der asymmetrischen
Art von Warmequelle und Warmesenke in der x-y Ebene dargestellt wird. Die analytischen Ergebnisse fiir die verschobenen
Geschwindigkeitsprofile
fur den Dampf in der x-y Ebene, die Druckverteilungen
in axialer Richtung fiir
Dampf und Fhissigkeit sowie die axiale Verteilung der Dampftemperatur
im Warmerohr
werden ermittelt.
Die vorgestellten
Ergebnisse und die Ergebnisse aufgrund der Liisung der Feldgleichungen
fiir den herkommlichen
symmetrischen
Fall stimmen qualitativ
gut iiberein. Es ergeben sich genaue analytische
Ausdriicke fur die Verteilungen von Geschwindigkeit
und Druck in dem betrachteten
nichtherkiimmlichen
Warmerohr.

AHAJIA3

XAPAKTEPHCTHK

TEIEHHR
I? TEIIJIOI-IEPEHOCA
HJIOCKOR
TEI-IJIOBOR
TPYEE

B ACHMMETPHsHOm

hmoTauww-~onpo6fio
aHaJrH3HpyroTcn pasnmmbre
$ni3HHecKHe xapaKTepHcrHKH acHMhfeTpHHHofi
n~~ocrofi TennoBofi ~py6br. OBpasyromeecn
nceBnoTpexhiepH none TeHeHHB napa paaserannerca
B mrocKOCTHX-y a CH,ty aCHMMeTpHYHoCTHTeMOBOrO ACTO'IHHKa H CTOKOB.I-IO,tyK%bt tiHiU,HTHWZCKBe pS,',H.TaTbr mll npO@lmeii CKOpoCTeii napa B IIBOCKOCTHx-y, ZlKCll~bHbIX paCLl~J,W+%IHii naBneHB B
napoo6pasHoti
H m~nltoii @sax, a TaKXe aKcHanbHor0 pacnpeneneHnn
TehmepaTyp napa. PesynbTaTbr
KaKXTBeHHOCOrBaCymTCK C AaHHblMH, nOJtyteHHbtMH npH pemeHHH ypaBHeHHfi nOBX B CHMMeTpHHHOM
cnysae.