Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

Zhu 1

Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

International Journal of Heat and Mass Transfer 42 (1999) 34053418

Analysis of cylindrical heat pipes incorporating the eects


of liquidvapor coupling and non-Darcian transporta
closed form solution
N. Zhu a, K. Vafai b,*
a

Lucent Technologies, Mount Olive, NJ 07828, USA


Department of Mechanical Engineering, The Ohio State University, Columbus, OH 43210, USA

Received 10 June 1998; received in revised form 4 January 1999

Abstract
This paper presents a two-dimensional analytical model for low-temperature cylindrical heat pipes. A closed-form
solution which incorporates liquidvapor interfacial hydrodynamic coupling and non-Darcian transport through the
porous wick for the rst time, is obtained for predicting the vapor and liquid velocity and pressure distributions. In
addition, the steady-state vapor and wall temperatures for a given input heat load in the evaporator region and a
convective boundary condition in the condenser region, are obtained. The eects of liquidvapor interfacial
hydrodynamic coupling and non-Darcian transport through the porous wick on the vapor and liquid velocity and
pressure distributions as well as the heat pipe capillary limit are discussed and assessed. The analytical solutions of
the axial vapor and wall temperature distributions, the vapor and liquid pressure distributions, and the centerline
vapor velocities compare very well with both experimental and numerical results. This work constitutes for the rst
time a comprehensive analytical solution which provides closed form solutions for the vapor and liquid ow as well
as the operating temperature and the maximum heat removal capability of the heat pipe. # 1999 Elsevier Science
Ltd. All rights reserved.

1. Introduction
Heat pipes are currently used in a wide variety of
heat transfer related applications. Analyses of heat
pipe operations, both analytical and numerical, have
been performed extensively by many investigators.
Almost all of the analytical studies have been concentrated on the dynamics of vapor ow. Liquid ow and
the liquidvapor coupling were mostly neglected in

* Corresponding author. Tel.: +1-614-292-6560; fax: +1614-292-3163.


E-mail address: vafai.1@osu.edu (K. Vafai)

analytical studies because of their complexity. A comprehensive analytical model for the overall simulation
of steady-state heat pipe operation is not available in
open literature. Cao and Faghri [1] investigated the
eects of heat pipe wall and the porous wick on the
heat pipe operation. They concluded that it is important to include the porous wick and the wall in heat
pipe analysis and to treat the entire heat pipe as a
single system rather than to analyze the vapor ow
alone. Rosenfeld [2] also reported the importance of
heat transfer within the wall and the porous wick in
the case of an asymmetric heat input.
Due to the diculty of obtaining an analytical solution for overall heat pipe operation, more and more

0017-9310/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 1 7 - 9 3 1 0 ( 9 9 ) 0 0 0 1 7 - 4

3406

N. Zhu, K. Vafai / Int. J. Heat Mass Transfer 42 (1999) 34053418

Nomenclature
A1
A2
B
C
D
G1
G2
h
hfg
K
ke
kwall
L
La
Lc
Le
M1
M2
p
pc
Q
r
rc
Ro
Rv
Rw
Re
T
Tb
u
ui
Uv
Ul
v
v1
v2
x

constant dened in Eq. (44)


constant dened in Eq. (45)
constant dened in Eq. (22)
constant dened in Eq. (26)
constant dened in Eq. (30)
constant dened in Eq. (28)
constant dened in Eq. (34)
convective heat transfer coecient, [W/m2 K]
latent heat of working uid [kJ/kg]
permeability of the wick [m2]
eective thermal conductivity of the liquid-saturated wick [W/m K]
thermal conductivity of the heat pipe wall [W/m K]
length of heat pipe [m]
length of the adiabatic section [m]
length of the condenser section [m]
length of the evaporator section [m]
constant dened in Eq. (29)
constant dened in Eq. (35)
pressure [Pa]
capillary pressure [Pa]
input heat [W]
radial coordinate [m]
eective pore radius of the wick [m]
heat pipe wall outer radius [m]
vapor core radius [m]
heat pipe wall inner radius [m]
injection Reynolds number (rvv1Rv/mv)
temperature [K]
bulk temperature of the coolant in cooling jack [K]
axial velocity [m/s]
axial interfacial velocity [m/s]
mean axial vapor velocity [m/s]
maximum axial liquid velocity [m/s]
radial velocity [m/s]
vapor injection velocity [m/s]
vapor suction velocity [m/s]
axial coordinate [m].

Greek symbols
E
porosity of the wick
p
g
porous wick shape parameter ( E=K)
m
dynamic viscosity [N s/m2]
r
density [kg/m3]
sl
surface tension of the working liquid [N/m].
Subscripts
i
liquidvapor interface
l
liquid phase
v
vapor phase.
Superscript
+
dimensionless quantity.

N. Zhu, K. Vafai / Int. J. Heat Mass Transfer 42 (1999) 34053418

numerical models have been developed. Some comprehensive numerical models [35] cover both the vapor
ow and the liquid ow. The boundary and inertial
eects were included in these numerical models by
applying the generalized momentum equation in porous medium to describe the liquid ow in heat pipes.
The coupling of the liquid and vapor momentum
equations was also incorporated in these models by
applying either the LaplaceYoung equation [3,4] or
the momentum jump condition [5] at the liquidvapor
interface. In these models, the matching conditions of
velocity and shear stress at the liquidvapor interface
were neglected by assuming a non-slip condition and
neglecting the interfacial drag. The eects of these
matching conditions as well as the boundary and inertial eects on heat pipe operation have not been investigated in any of the previous studies.
Vafai et al. [612] have developed comprehensive
pseudo-three-dimensional analytical models for asymmetrical at-shaped, including both disk-shaped and
at-plate, heat pipes. They incorporated liquid ow,
secondary vapor ow and the eects of liquidvapor
hydrodynamic coupling and non-Darcian transport in
their models [811]. Their results show that, for the
at-shaped heat pipes, while the eects of liquidvapor
interfacial hydrodynamic coupling are negligible,
neglecting the boundary and inertial eects can lead to
signicant error in predicting liquid ow and the maximum heat transfer capability of the heat pipe.

3407

In the present work, a two-dimensional analytical


model is developed for the overall simulation of the
steady-state cylindrical heat pipe operation. This analytical model employs matched asymptotic expansions
for the liquid ow to incorporate liquidvapor interfacial hydrodynamic coupling and the boundary and
inertial eects. A closed-form solution is then obtained
based on an in-depth integral method. The eects of
liquidvapor hydrodynamic coupling and the boundary and inertial eects on cylindrical heat pipe operation and operating limit are also investigated. The
analytical prediction of vapor and liquid velocity and
pressure distribution as well as vapor and wall temperature distributions compare very well with both the
experimental data reported by Huang et al. [13] and
the numerical results reported by Tournier and
El-Genk [5].
2. Mathematical modeling
The schematic of the cylindrical heat pipe and the
coordinate system used in the present analysis is
shown in Fig. 1. Heat applied at the evaporator section
causes vaporization and subsequent pressurization of
the working liquid. The vapor ows to the condenser
section and releases latent heat as it condenses. The
heat is then removed from the condenser wall surface
by convection. In the present analysis, vapor and

Fig. 1. Schematic of the heat pipe and the coordinate system used in the analysis.

3408

N. Zhu, K. Vafai / Int. J. Heat Mass Transfer 42 (1999) 34053418

liquid ows are assumed to be steady, laminar


and incompressible. The wick is assumed isotropic and
saturated with the working liquid. The liquid and
vapor phases are coupled at the liquidvapor interface.
The vapor injection and suction at the liquidvapor
interface are assumed to be uniform.
2.1. Governing equations for vapor ow
Based on the above assumptions, the continuity and
the momentum equations governing the vapor ow are
given by
@ uv @ vv vv

0
@x
@r
r





@ uv
@ uv
@ pv
@ 2 uv 1 @ uv
rv uv

vv

mv
r @r
@x
@r
@x
@ r2

@ pv
0
@r

rv vv x, r Rv rl vl x, r Rv
8
0RxRLe
< rv v1 ,

0,
Le RxRLe La
:
rv v2 ,
Le La RxRL

The continuity of axial velocity and shear stresses at


the liquidvapor interface yields

The liquid ow within the porous wick is modeled


using the generalized momentum equation [15] which
accounts for the boundary and inertial eects. The
convective term in the generalized momentum equation
is dropped in accordance with the analysis given in
Vafai and Tien [15]. The governing equations for the
liquid ow are
4



ml @ 2 ul 1 @ ul
m
r FE
@ pl

l ul l 1=2 j ul j ul
r @r
K
E @ r2
K
@x

and
mv

2.2. Governing equations for liquid ow

The vapor and liquid phases are coupled at the


vaporliquid interface. The continuity of mass uxes
in the r-direction at the liquidvapor interface yields

uv x, r Rv ul x, r Rv ui x

The x-direction shear stress in the momentum


equations is neglected based on the work of Busse and
Prenger [14] which shows the validity of the boundary
layer approximation for the vapor ow in long cylindrical heat pipes.

@ ul 1 @
rul 0

r @r
@x

2.3. Vaporliquid hydrodynamic coupling



@ uv
@ ul

m
l
@ r rRv
@ r rRv

where ui(x ) is the interfacial velocity at the liquid


vapor interface. The vapor injection velocity, v1 is related to the input power Q by the following relation:
v1

Q
2rv pRv Le hfg

where hfg is the latent heat of the working uid. The


vapor suction velocity v2 is determined by the mass
balance which requires that the uid entering the
vapor region in the evaporator section to ow out
through the condenser section.
2.4. Boundary conditions
The boundary conditions are as follows:
x 0:

uv vv ul vl 0

x L:

uv vv ul vl 0,

r 0:

vv 0,

@ uv
@ pv

0
@r
@r

10
pv pl

11

12

In the above governing equations, rv is the vapor density and rl the liquid density, mv the vapor dynamic
viscosity, ml the liquid dynamic viscosity, K and E are
the permeability and porosity of the wick and F is a
geometric function based on the porous wick structure
and is calculated using the expression outlined in Vafai
[16], i.e., F=1.75/Z150E 3/2.

r Rw :

ul vl 0

13

The above governing equations, coupling equations


and boundary conditions are used to obtain a closedform analytical solution for the vapor and liquid velocity and pressure distributions which accounts for the
eects of liquidvapor coupling and non-Darcian
transport through the porous wick.

N. Zhu, K. Vafai / Int. J. Heat Mass Transfer 42 (1999) 34053418

3. Analytical solution
An in-depth integral analysis along with the method
of matched asymptotic expansions is employed to
obtain the closed-form analytical solution for the
vapor and liquid velocity and pressure distributions.
The temperature values are then obtained through the
use of a simple conduction model.
3.1. Vapor velocity prole
The following velocity prole is utilized for the
vapor ow within the heat pipe:
uv x, r Uv xa0 a1 r a2 r2

14

where
1
Uv x
pR2v

Rv
0

uv x, r2pr dr

15

8


r Rv
>
>
>
Ul x ui x Ul x exp p
>
>
K=E
>
<
ul x,r
Ul x
>


>
>
>
r Rw
>
>
Ul x exp p
:
K=E

3409

three parts: an inner solution for the interface zone


between the liquid-wick and the vapor phase, an outer
solution for the main wick region and an inner solution for the interface zone between the liquid-wick
and the heat pipe wall. This results in the following
velocity prole:
ul x, r
8


r Rv
>
>
>
c1 c2 exp p
>
>
K=E
>
<
Ul x
>


>
>
>
r Rw
>
>
: c3 c4 exp p
K=E

r Rv
0R p R1
K=E
outer solution
r Rw
1R p R0
K=E
17

where Ul(x ) is the maximum liquid velocity. Applying


the boundary conditions given by Eqs. (7) and (13)
and matching the inner and outer solutions yields

r Rv
0R p R1
K=E
outer solution

18

r Rw
1R p R0
K=E

Eq.
(18) can be written in a more compact form, by utilizing the fact that the thickness of the interface regions,
p
K=E, is much smaller than (RwRv)/2. This results in
8


>
r Rv
>
>
p

u
x

U
x
exp

U
>
l
i
l
<
K=E
ul x,r


>
r

R
w
>
>
Ul x exp p
>
:
K=E

Rv RrR

uv x, r ui x 2Uv x ui x 1

r
Rv

2 #

19

Rw Rv
RrRRw
2

is the mean vapor velocity. Applying the boundary


conditions given by Eqs. (7) and (12) to Eq. (14) yields
the following vapor velocity prole:
"

Rw Rv
2

16

3.3. The mean vapor velocity


The mean vapor velocity Uv(x ) is determined by
integrating the vapor continuity equation. Utilizing the
vapor velocity prole and the boundary conditions
given by Eqs. (6)(8), (10) and (11), the integration of
the vapor continuity Eq. (1) with respect to r from 0
to Rv yields

3.2. Liquid velocity prole


Vafai and Thiyagaraja [17] have shown that the
momentum boundary layer thickness at the interface
between a porous medium and a uid or an impermeable medium is of the order of (K/E )1/2. Utilizing the
method of matched asymptotic expansions the liquid
velocity prole within the wick region is derived in

Uv x

8
>
>
>
>
>
>
>
>
<

2v1
x,
Rv
2v1
Le ,
Rv

>
>
>
>
>
>
2v2
>
>
L x,
:
Rv

0RxRLe
Le RxRLe La
Le La RxRL

20

3410

N. Zhu, K. Vafai / Int. J. Heat Mass Transfer 42 (1999) 34053418

vapor and liquid velocity proles in Eq. (8). This


yields

3.4. The maximum liquid velocity


The maximum liquid velocity Ul(x ) is determined by
integrating the liquid continuity Eq. (4) with respect to
r from Rv to Rw. Applying the liquid velocity prole
given by Eq. (19), the mean vapor velocity given by
Eq. (20), and the boundary conditions given by Eqs.
(6)(8), (10) and (13) within the integrated liquid continuity equation yields
8
<

Bv1 x,
Ul x
Bv1 Le ,
:
Bv2 L x,

0RxRLe
Le RxRLe La
Le La RxRL

21

where

4
Uv x
24
m gRv
p
The fact that 1/g= K=EW(RwRv)/2 is used in the
derivation of Eq. (24). Substituting Eqs. (20) and (21)
into Eq. (24) results in the following expression for
ui(x):

ui x Ul x

ui x

Cv1 x,
Cv1 Le ,
:
Cv2 L x,

0RxRLe
Le RxRLe La
Le La RxRL

25

where
!

2Rv
8r
22
1 2 2
2
Rv
m g Rv
p
where g= E=K is the shape parameter of the porous
wick, and r +=rl/rv, m +=ml/mv. If the hydrodynamic
coupling is neglected at the liquidvapor interface, the
following expression is obtained for B based on a noslip condition at the liquidvapor interface:
B

8
<

r R2w

2Rv
B 2
r Rw R2v

23

3.5. Interfacial velocity


The interfacial velocity is determined by utilizing the
8
<

G1 v21 M1 v1 x 2 ,
pv x pv 0
M1 v1 L2e 2M1 v1 Le x,
:
G1 v22 M1 v2 x L2 M1 v2 L La Lc ,
G1 v21

8
B
m gR2v

26

Neglecting the hydrodynamic coupling at the liquid


vapor interface leads to C=0.
3.6. Vapor pressure distribution
The pressure distribution in the vapor phase is
obtained by integrating the vapor momentum
equations with respect to r from 0 to Rv. Introducing
vapor continuity equation, vapor velocity prole,
vapor-liquid coupling conditions and the related
boundary conditions into the integrated momentum
equations results in the following derived expression
for vapor pressure distribution:
0RxRLe
Le RxRLe La
Le La RxRL

27

where
G1

rv
D2 7D 16
3R2v

28

4mv
D 2
R3v

29

M1

D CRv

30

If the hydrodynamic coupling at the liquidvapor interface is neglected, Eq. (27) reduces to
8
>
>
>
>
>
>
>
>
>
<


16rv 2 8mv
2
v

v
1 x ,
3R2v 1
R3v


16rv 2 8mv
16m
v

L2e 3 v v1 Le x,
pv x pv 0
1
2 1
3
3R
R
Rv
>
v
v
>
>
>


>
>
16rv 2 8mv
8mv
>
2
>
>
: 3R2 v2 R3 v2 x L R3 v2 L La Lc ,
v
v
v


0RxRLe
Le RxRLe La
Le La RxRL

31

N. Zhu, K. Vafai / Int. J. Heat Mass Transfer 42 (1999) 34053418

Eq. (31) is similar to that obtained by Faghri [18],


which in turn was in agreement with similar relations
developed by Busse [19] earlier. It should be noted that
in Ref. [18], in addition to not accounting for the
liquidvapor coupling, the non-Darcian eects and for
that matter the liquid ow transport was not taken
into account. In fact, Ref. [18] considered only the
vapor ow.
3.7. Liquid pressure distribution
The liquid pressure distribution is determined by
integrating the generalized momentum Eq. (5) with
respect to r from Rv to Rw. Integrating Eq. (5) and utilizing all necessary boundary conditions and the cited
relationships results in the derivation of the following
expression for the liquid pressure distribution:
pl x pl 0

2La L2c

33

and


ml B
1
2
1
2K
gRv 1 Rw =Rv
"

M2

rl FE
1
CC 2B
B2
3K 1=2
gRv 1 Rw =Rv 2

ml
1
2C
B
gRv 1 Rw =Rv 2
2K

M2 0

0RxRLe
Le RxRLe La
Le La RxRL

32

uniformly over the evaporator section and convectively


cooled in the condenser section. The vapor temperature is assumed uniform along the heat pipe, and a
one-dimensional heat conduction model is used for the
wall and liquid-wick regions. For steady-state operation, the convective boundary condition at the wall
outer surface (r=Ro) is:

34

Q 2pRo Lc hTwall,c Tb

35

where Q is the heat input, h is the convective heat


transfer coecient, Twall,c is the wall temperature in
the condenser section, and Tb is the bulk temperature
of the coolant. The vapor and wall temperatures are
obtained as follows:

If the boundary and inertial eects are neglected, G2


and M2 are reduced to the following relationships:
G2

For a fully-thawed low-temperature heat pipe, the


vapor phase can be assumed to be saturated and uniform. Therefore, there is no need to solve the energy
equation for the vapor phase. The validity of this approximation was examined by Huang and El-Genk
[20] experimentally. They measured the axial vapor
and wall temperature distributions of a copper heat
pipe with water as the working uid. Their results
show that the vapor temperature is uniform along the
heat pipe. They also found that the wall temperature
in the evaporator and the condenser sections is almost
uniform, except near the interfaces with the adiabatic
section where axial conduction in the wall is most pronounced. In the present study, the heat pipe is heated

G2 v1 x 2 M2 v21 x 3 ,
G2 v1 Le 2x Le M2 v21 L2e 3x 2Le ,
:
G2 v2 x L2 L La Lc M2 v22 x L3 L 2La L2c ,

pl 0 pv L G2 v2 L La Lc M2 v22 L

G2

3.8. Vapor and wall temperatures

8
<

where

3411

!
36

37

Q
Tv Tb
2pLc

1

hRo

38

lnRo =Rw
lnRw =Rv

kwall
keff
39

and

8
"
#


>
Q
lnRo =Rw
lnRw =Rv
Lc
1
>
>
>
1
0RxRLe

Tb
>
>
2pLc
kwall
keff
hRo
Le
>
>
>
<


Q
lnRo =Rw
lnRw =Rv
1
Twall x
Le RxRLe La
Tb

>
>
2pLc
kwall
keff
hRo
>
>
>
>
>
Q
>
>
:
Tb
Le La RxRL
2phRo Lc

40

3412

N. Zhu, K. Vafai / Int. J. Heat Mass Transfer 42 (1999) 34053418

where kwall is the thermal conductivity of the heat pipe


wall, and ke is the eective thermal conductivity of
the liquid-saturated wick.
3.9. The maximum heat transport capillary limit
For a heat pipe under steady-state operation, stable
circulation of the working uid in the heat pipe is
achieved through the capillary pressure established by
the wick structure. In conventional heat pipes, there
exists a maximum capillary pressure that can be developed for a liquidvapor pair. The maximum heat
transport capillary limit for a heat pipe is achieved
when the sum of the pressure losses along the circulation path of the working uid reaches the maximum
capillary pressure; that is
Dpv x max x min Dpl x min x max pc x min

2sl
rc

41

where pc(xmin)=pv(xmin)pl(xmin) is the minimum


capillary pressure, sl is the surface tension of the
liquid, rc is the eective pore radius of the wick, and
xmax/xmin denotes the location where the capillary
pressure is maximum/minimum. The notation
Dp(xmaxxmin) refers to evaluation of Dp over the distance (xmaxxmin). In the present study, as stated in
Eq. (11) the condenser end (x=L ) is assumed to be
the wet point ( pv(L )=pl(L )) during steady-state operation. This reduces Eq. (41) to
pv 0 pl 0

2sl
rc

42

Substituting Eqs. (9), (31) and (33) into Eq. (42) and
solving for the maximum heat transport capillary limit
yields
Qmax

q
1
A22 8A1 sl =rc A2
2A1

43

where Qmax is the maximum heat transport capillary


limit, and
A1

M2 L 2La
12p2 R2v r2v h2fg

44

A2

G2 2M1 L La
4pRv rv hfg

45

When boundary and inertial eects are neglected, Eq.


(43) reduces to
Q max

2sl
A 2 rc

46

4. Results and discussion


To verify the model predictions, results are compared with the experimental data reported by Huang
et al. [13] and the numerical results reported by
Tournier and El-Genk [5], for a copper heat pipe with
water as the working uid. The physical dimensions of
the heat pipe are chosen as: Ro=9.55 mm,
Rw=9.4 mm, Rv=8.65, Le=60 cm, La=9 cm, and
Lc=20 cm which are based on the physical dimensions
used in Refs. [5] and [13]. The wick eective pore
radius, porosity and permeability are chosen as 54 mm,
0.9 and 1.5  109 m2, respectively. The eective thermal conductivity of the liquid-wick is calculated to be
1.965 W/m K, using the equation given by Chi [21].
The thermophysical properties of the working uid are
obtained at the calculated vapor temperature and are
assumed to be constant along the heat pipe. The input
heat is taken as 455 W. The cooling water enters the
condenser cooling jacket at 294.5 K and 11.33 g/s. The
temperature of the cooling water at the exit of the
cooling jacket is calculated using the equation
Q=m cp(ToutTin). The bulk temperature of the cooling water is taken as the arithmetic mean of Tin and
Tout. The convective heat transfer coecient in the
cooling jacket is taken as 1800 W/m2 K [5]. It should
be noted that a great advantage of the presented closed
form analytical solution is the ease and speed in which
a comprehensive solution can be obtained subject to
changes in various parameter and properties. In here
some representative values were chosen for the simulations. We were able to show that the main conclusions remain unchanged for wide range of variation
of these parameters.
Fig. 2 compares the model predictions with the numerical and experimental results of the vapor and wall
temperature distributions along the heat pipe. The
model predictions of both the vapor and the wall temperatures agree well with the numerical and the experimental values. As can be seen in Fig. 1, the
assumption of uniform vapor temperature along the
heat pipe is good for low-temperature heat pipes.
Although there is a discrepancy between the calculated
wall temperature and the experimental data at the
beginning and the end of the condenser section, the
model predictions agree very well with the numerical
results reported by Tournier and El-Genk [5]. Since
same parameter values for the convective cooling
boundary condition are used in the present analytical
model as those used in the numerical model, the good
agreement between the analytical and the numerical
results of the wall temperatures demonstrates that the
one-dimensional approximation for the heat transfer
within the wall and liquid-wick regions is adequate for
low-temperature heat pipes. This is in agreement with
the observation of Tournier et al. [22].

N. Zhu, K. Vafai / Int. J. Heat Mass Transfer 42 (1999) 34053418

3413

Fig. 2. Comparison of the calculated vapor and wall temperature distributions with the numerical results given by Tournier and ElGenk [5] and the experimental results given by Huang et al. [13].

Fig. 3 compares the analytical vapor and liquid


pressure distributions along the heat pipe with the numerical results of Tournier and El-Genk [5]. The absolute value of the vapor pressure at the evaporator end,
pv(0), is related to the calculated vapor temperature
based on the assumption that the vapor phase is saturated for a low-temperature fully-thawed heat pipe.
The analytical results agree very well with the numerical results. As the results show, the vapor pressure
variation along the heat pipe is small, which means
that a uniform vapor temperature prole is expected.
The variations of the mean vapor velocity, the centerline vapor velocity, the maximum liquid velocity
and the interfacial velocity along the heat pipe are
shown in Figs. 4 and 5. The positive velocity value
denotes that the ow is along the x direction and the
negative velocity value denotes that the ow is along
the negative x direction. The analytical centerline
vapor velocity at the exit of the evaporator is compared to the numerical result by Tournier and El-Genk
[5]. Again, as seen in Fig. 4, the analytical predictions
agree very well with the numerical results. This further
demonstrates that the analytical model provides accurate predictions of the low-temperature heat pipe operations.

The eects of liquidvapor interfacial hydrodynamic


coupling and the boundary and inertial eects are analyzed next. Results were obtained for the radial
Reynolds number (Re=rvv1Rv/mv) values of 2.0, 5.0
and 10.0 and are displayed in Figs. 68. To accommodate vapor and liquid pressures at dierent Re values
in the same gure and to make the comparison more
meaningful, the vapor pressure at the evaporator end,
pv(0), is taken as zero in Figs. 7 and 8 for all the Re
values. The eects of liquidvapor interfacial hydrodynamic coupling on the maximum liquid velocity
along the heat pipe is shown in Fig. 6. It should be
mentioned that liquidvapor interfacial hydrodynamic
coupling has no eect on the mean vapor velocity, as
indicated in Eq. (20). This is because the vapor ow
rate at steady-state operation is purely determined by
the vapor injection/suction velocities, i.e., the evaporation/condensation rates. However, liquidvapor interfacial hydrodynamic coupling do aect the vapor
velocity prole, as illustrated in Eq. (16). For example,
the interfacial eects result in a larger centerline vapor
velocity. Since the interfacial velocity is negligible compared to the mean vapor velocity, as can be seen in
Figs. 4 and 5, the interfacial eects on the vapor velocity prole is negligible. The analytical results in Fig.

3414

N. Zhu, K. Vafai / Int. J. Heat Mass Transfer 42 (1999) 34053418

Fig. 3. Comparison of the calculated vapor and liquid pressure distributions with the numerical results given by Tournier and ElGenk [5].

Fig. 4. Variations of the mean vapor velocity and the centerline vapor velocity along the heat pipe.

N. Zhu, K. Vafai / Int. J. Heat Mass Transfer 42 (1999) 34053418

Fig. 5. Variations of maximum liquid velocity and interfacial velocity along the heat pipe.

Fig. 6. The eects of liquidvapor interfacial hydrodynamic coupling on the maximum liquid velocities.

3415

3416

N. Zhu, K. Vafai / Int. J. Heat Mass Transfer 42 (1999) 34053418

Fig. 7. The eects of liquidvapor interfacial hydrodynamic coupling on vapor and liquid pressures.

Fig. 8. Boundary and inertial eects on the vapor and liquid pressure distributions.

N. Zhu, K. Vafai / Int. J. Heat Mass Transfer 42 (1999) 34053418

6 show that the interfacial eects lead to a larger maximum liquid velocity. The larger the injection Reynolds
number, the larger the dierence caused by the interfacial eects. Fig. 7 shows the eects of liquidvapor
interfacial hydrodynamic coupling on the vapor and
liquid pressures. The interfacial eect on the vapor
pressure is negligible. However, neglecting liquid
vapor interfacial hydrodynamic coupling will lead to a
smaller liquid pressure drop along the heat pipe.
The boundary and inertial eects on vapor and
liquid pressure proles are shown in Fig. 8. As indicated in Eq. (27), the pressure distribution in the vapor
phase is not aected by the boundary and inertial
eects. However, neglecting the boundary and inertial
eects leads to an underestimation of the liquid pressure. The larger the Reynolds number, the larger the
error involved in the calculation of liquid pressures by
using a Darcian model. This is in agreement with the
results by Vafai and Tien [15].
Fig. 9 shows the interfacial and the boundary and
inertial eects on the heat transport capillary limit of
the heat pipe. It can be seen in Fig. 9 that either
neglecting the interfacial eects or neglecting the
boundary and inertial eects can lead to an overestimation of the capillary limit. This is because a smaller
overall liquid pressure drop along the heat pipe is

3417

expected when the interfacial eects or the boundary


and inertial eects are neglected. While the interfacial
eects on the maximum heat removal capability is relatively small, the error involved in neglecting the
boundary and inertial eects is substantial.
5. Conclusions
The steady-state operation of a low-temperature
cylindrical heat pipe has been studied analytically. An
analytical model, which incorporates liquidvapor
interfacial hydrodynamic coupling and non-Darcian
transport through the porous wick, was developed for
predicting the vapor and liquid ow and the maximum
heat transfer capability of the heat pipe. A closed-form
solution was obtained for the vapor and wall temperatures as well as the vapor and liquid velocity and
pressure distributions for a convective cooling condition in the condenser region. A closed-form solution
of the heat pipe capillary limit during steady state operation was also obtained. These closed-form analytical
solutions, provide a quick, accurate prediction method
for low-temperature heat pipe operation and were
found to be in very good agreement with both experimental and numerical results. The eects of liquid

Fig. 9. The eects of liquidvapor interfacial hydrodynamic coupling and the boundary and inertial eects on the heat pipe capillary limit.

3418

N. Zhu, K. Vafai / Int. J. Heat Mass Transfer 42 (1999) 34053418

vapor interfacial hydrodynamic coupling and nonDarcian transport through the porous wick were also
analyzed. The results show that the interfacial eects
are small and can be neglected. However, substantial
errors can occur when using Darcy's law in calculating
the liquid pressure distributions as well as the maximum heat removal capability of the heat pipe.
Acknowledgement
The
grant
(DE-F602-93ER61612)
by
the
Department of Energy is acknowledged and greatly appreciated.
References
[1] Y. Cao, A. Faghri, Transient two-dimensional compressible analysis for high temperature heat pipes with
pulsed heat input, Numerical Heat Transfer 114 (1990)
10281035.
[2] J.H. Rosenfeld, Modeling of heat transfer into a heat
pipe for a localized heat input zone, in: AIChE
Symposium Series, no. 257, vol. 83, 1987, pp. 7176.
[3] V.H. Ransome, H. Chow, ATHENA heat pipe transient
model, in: Trans. 4th Symposium on Space Nuclear
Power Systems, CONF-870102-Summs., Albuquerque,
NM, 1987, pp. 389392.
[4] J.M. Doster, M.L. Hall Numerical modeling of hightemperature liquid-metal heat pipes, 1989 Joint ASME/
A.I.Ch.E National Heat Transfer Conference,
Philadelphia, PA, vol. 89-HT-13, 1989, pp. 19.
[5] J.M. Tournier, M.S. El-Genk, A heat pipe transient
analysis model, Int. J. Heat Mass Transfer 37 (1994)
753762.
[6] K. Vafai, W. Wang, Analysis of ow and heat transfer
characteristics of an asymmetrical at plate heat pipe,
Int. J. Heat Mass Transfer 35 (1992) 20872099.
[7] K. Vafai, N. Zhu, W. Wang, Analysis of assymetrical
disk-shaped and at plate heat pipes, ASME J. Heat
Transfer 117 (1995) 209218.
[8] N. Zhu, K. Vafai, The eects of liquidvapor coupling
and non-Darcian transport on asymmetrical disk-shaped
heat pipes, Int. J. Heat Mass Transfer 39 (1996) 2095
2113.

[9] N. Zhu, K. Vafai, Optimization of asymmetrical diskshaped heat pipes, AIAA J. Thermophys. Heat Transfer
10 (1996) 179182.
[10] N. Zhu, K. Vafai, Numerical and analytical investigation of vapor ow in a disk-shaped heat pipe incorporating secondary ow, Int. J. Heat Mass Transfer 40
(1997) 28872900.
[11] N. Zhu, K. Vafai, Analytical modeling of the startup
characteristics of asymmetrical at-plate and diskshaped heat pipes, Int. J. Heat Mass Transfer 41 (1998)
26192637.
[12] N. Zhu, K. Vafai, Vapor and liquid ow in an asymmetrical at plate heat pipe: a three-dimensional analytical and numerical investigation, Int. J. Heat Mass
Transfer 41 (1998) 159174.
[13] L. Huang, M.S. El-Genk, J-M. Tournier, Transient performance of an inclined water heat pipe with a screen
wick, ASME National Heat Transfer Conference,
Atlanta GA, Heat Pipes and Capillary Pumped Loops,
HTD-vol. 236, 1993, pp. 8792.
[14] C.A. Busse, F.C. Prenger, Numerical analysis of the
vapor ow in cylindrical heat pipes, in: Proc. 5th Int.
Heat Pipe Conf., Part 1, JIATEC, Tokyo, Japan, 1984,
pp. 214219.
[15] K. Vafai, C.L. Tien, Boundary and inertia eects on
ow and heat transfer in porous media, Int. J. Heat
Mass Transfer 24 (1981) 195203.
[16] K. Vafai, Convective ow and heat transfer in variable
porosity media, J. Fluid Mech 147 (1984) 233259.
[17] K. Vafai, R. Thiyagaraja, Analysis of ow and heat
transfer at the interface region of a porous medium, Int.
J. Mass Transfer 30 (1987) 13911405.
[18] A. Faghri, Performance characteristics of a concentric
annular heat pipe: Part IIvapor ow analysis, ASME
J. Heat Transfer 111 (1989) 851857.
[19] C.A. Busse, Pressure drop in the vapor phase of long
heat pipes, in: Proc. Thermionic Conversion Specialist
Conference, Palo Alto, CA, 1967, pp. 391398.
[20] L. Huang, M.S. El-Genk, Experimental investigation of
transient operation of a water heat pipe, in: Proceedings
of 10th Symposium on Space Nuclear Power and
Propulsion, Albuquerque, NM, 1993, pp. 365374.
[21] S.W. Chi, Heat Pipe Theory and Practice, Hemisphere,
Washington, DC, 1976.
[22] J-M. Tournier, M.S. El-Genk, A.J. Juhasz, Heat-pipe
transient model for space applications, in: Proceedings
of 8th Symposium on Space Nuclear Power Systems,
Albuquerque, NM, 1991, pp. 857868.

You might also like