You are on page 1of 13

i An update to this article is included at the end

International Journal of Heat and Mass Transfer 54 (2011) 564–574

Contents lists available at ScienceDirect

International Journal of Heat and Mass Transfer


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

3D heat transfer analysis in a loop heat pipe evaporator with a fully saturated wick
Ji Li a,⇑, G.P. Peterson b
a
Laboratory of Electronics Thermal Management, College of Physics, Graduate University of the Chinese Academy of Sciences, 19A Yu-quan-lu Road,
Shijingshan District, Beijing 100049, PR China
b
The G.W. Woodruff School of Mechanical Engineering, Georgia Institute of Technology, Atlanta, GA 30332, United States

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

Article history: A practical quasi three-dimensional numerical model is developed to investigate the heat and mass trans-
Received 4 May 2010 fer in a square flat evaporator of a loop heat pipe with a fully saturated wicking structure. The conjugate
Received in revised form 30 July 2010 heat transfer problem is coupled with a detailed mass transfer in the wick structure, and incorporated
Accepted 9 September 2010
with the phase change occurring at the liquid–vapor interface. The three-dimensional governing equa-
Available online 15 October 2010
tions for the heat and mass transfer (continuity, Darcy and energy) are developed, with specific attention
given to the wick region. By comparing the results of the numerical simulations and the experimental
Keywords:
tests, the local heat transfer mechanisms are revealed, through the obtained temperature distribution
Loop heat pipe
Numerical simulation
and the further derived evaporation rates along the liquid–vapor interface. The results indicate that
Three dimensional the model developed herein can provide an insight in understanding the thermal characteristics of loop
Heat transfer heat pipes during steady-state operation, especially at low heat loads.
Mass transfer Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction of the operations of LHPs. A performance comparison for miniatur-


ized or compact LHPs can be found in Li et al. [5,6].
Two-phase capillary pumped heat transfer devices are becom- Historically, theoretical analyses or numerical simulations in-
ing much more prevalent in the thermal control of electronic de- volved in the operation of LHPs has lagged far behind the experi-
vices, particularly in the increasingly demanding thermal control mental investigations, due to the complexity of the operational
problems of high-end electronics. Among these devices, loop heat modalities and the complexity of the problems involved. On the
pipes (LHP) are particularly interesting, due to the advantages, basis of the pioneering theoretical works on traditional heat pipes
which include: (i) greatly increased capacity over conventional (and capillary pumped loops), Ku [3] first developed an analytical
heat pipes at comparable dimensions; (ii) more efficient operation model to describe the thermal-hydraulic behaviors of the LHP
at any orientation in the gravitational field; (iii) lower overall ther- and described the operating characteristics and performance limi-
mal resistance; (iv) increased flexibility in packaging; and (v) high tations as a function of the various physical and operational
heat load capacities over longer distance [1]. A loop heat pipe (LHP) parameters, including the heat load, sink temperature, ambient
is a two-phase heat transfer device that removes heat from a temperature, and elevation change. Then, Chernysheva et al. [7]
source (e.g., an electronic chip) through an evaporator and pas- proposed a semi-empirical theoretical method to calculate the
sively moves it to a condenser region or radiator using capillary LHP operating temperature for both fully and partially saturated
forces to pump the fluid, ultimately releasing the heat to the envi- compensation chambers and discussed the different features and
ronment from the condenser by natural or forced convection. Loop operational advantages and disadvantages for these two operating
heat pipes were first investigated and patented in the USSR in 1979 modes. Most recently, Bai et al. [8] established a mathematical
by Kiseev et al. [1,2]. The patent for LHPs was filed in the USA in model considering the influence of the compound wick over the
1982 [1,2]. There were many different configurations of LHP evap- performance of LHPs coupled with the annular flow model for con-
orators and different types (homogeneous or heterogeneous) of denser. Only a limited number of numerical investigations have
wicks, which were investigated to improve the performance of been presented for LHP evaporators [9–11]. Cao and Faghri [9] pre-
LHPs and determine the compatibility of the materials. A detailed sented a numerical analysis for a completely liquid saturated wick
description of the operating characteristics, working principles in a capillary pumped loop with a flat-plate type evaporator, in
and diversity of LHP configurations can be found in Maidannik which a two-dimensional liquid flow in the wick and a three-
[2], Ku [3], and Launay et al. [4], along with a parametric analysis dimensional vapor flow in the grooves separated by the wick were
considered. By adopting a numerical model for heat pipes with an
⇑ Corresponding author. Tel.: +86 13522278866. inverted meniscus developed from Demidov and Ystsenko [12] and
E-mail address: jili@gucas.ac.cn (J. Li). later Figus et al. [13], as depicted in Fig. 1, Kaya and Goldak [10]

0017-9310/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijheatmasstransfer.2010.09.014
J. Li, G.P. Peterson / International Journal of Heat and Mass Transfer 54 (2011) 564–574 565

Nomenclature

cp specific heat, J/kg K Subscripts


hfg latent heat of evaporation, J/kg a ambient
Kp permeability, m2 c condenser
k thermal conductivity, W/m K cap capillarity
ke effective thermal conductivity, W/m K cc compensation chamber
m_ mass flow rate, kg/s cu copper
n number of vapor removal channels e evaporator
P pressure, Pa g gravity
Qload total heat load l liquid
Rg gas constant, J/kg K ll liquid line
T temperature, °C n normal direction
q heat flux
Greek symbols s substrate; solid phase material in wick
a heat transfer coefficient, W/m2 K v vapor
C border of wick and vapor removal channel vl vapor line
m kinetic viscosity, m2/s vc vapor removal channel
q density, kg/m3 w wick
e porosity

conducted a two-dimensional numerical analysis of heat and mass formation of the meniscus are desirable. Inspired by the concept
transfer in the wick of a loop heat pipe with a cylindrical evapora- of mini-/micro-channel heat transfer and by the fundamental find-
tor based on the results of previous works [9,12,13], e.g., the valid- ings reported by Li et al. [14], Li and his co-workers first proposed a
ity of the Darcy law for a wick with a varying pore-size unique square, flat LHP evaporator with a wicked fin directly sin-
distribution, and the accuracy of a two-dimensional model for heat tered onto the substrate (or casing) of the LHP evaporator [5,6],
and mass transfer in the porous structure. They reported that the see Fig. 2. With this novel design of the porous structure, an extre-
overall temperature distribution is not strongly affected by the mely low thermal resistance in the LHP evaporator was obtained
choice of the correlation used to predict the effective thermal con- [5,6]. The mechanism behind this phenomenon was also carefully
ductivity of the wick. In this model, only the heat and mass transfer examined [6] and it is believed that Demidov and Ystsenko’s argu-
in the wick and the evaporator cover were given consideration. An ment is correct.
empirical correlation was used to define the boundaries of the sur- Based upon the analysis above, through a thorough literature
rounding wick and the evaporator casing. They also recommended investigation, it is apparent that a detailed numerical simulation
that a three-dimensional model was necessary to solve the heat of a loop heat pipe evaporator coupled with a detailed determina-
and vapor flow in the vapor removal channels. Recently, Chernysh- tion of the local heat transfer mechanism is rare, but prerequisite
eva and Maydanik [11] numerically investigated the transient for the further optimization of the performance of loop heat pipes
startup process for a loop heat pipe with a cylindrical evaporator in the future. In order to fulfill this purpose, in the current investi-
using a transient heat conduction model with energy and mass gation, a practical quasi three-dimensional numerical model is
conservative boundary conditions. The mechanism for the temper- developed to investigate the heat and mass transfer in the LHP
ature overshoot during the LHP startup was discussed. evaporator as explored by Li et al. [5,6]. This model incorporates
As pointed out by Demidov and Ystsenko [12], the evaporation the semi-empirical analytical solution proposed by Chernysheva
from the meniscus formed in the corner of the wick in intimate
contact with the fin (refer to Fig. 1) could be much higher than that
occurring from the surface of the wick. Designs facilitating the

Fig. 1. Schematic of evaporation and inverted meniscus inside the evaporator Fig. 2. Schematic structure of a unique square flat LHP evaporator with wicked fin
(adopted from [10]). directly sintered on the substrate [5,6].
566 J. Li, G.P. Peterson / International Journal of Heat and Mass Transfer 54 (2011) 564–574

et al. [7] into the determination of the saturation pressure of the (1) the process is steady-state,
working fluid in the compensation chamber as a predetermined (2) the capillary structure is homogeneous and isotropic,
boundary condition under a heat load. The other governing equa- (3) the radiative and gravitational effects are negligible, and
tions and the boundary conditions are all constructed from their (4) the fluid is Newtonian and has constant properties at each
own physical aspects. In addition, the thermal and dynamic proper- phase.
ties of the materials are all taken directly from the handbook values
except for those which are addressed individually in the context. In addition, there are other assumptions in the present model
involving the boundary conditions, the properties of the porous
structure and the liquid–vapor interface, which are listed below:
2. Mathematical formulation
(1) the wick structure is perfectly saturated,
A schematic of the computational model for the LHP evaporator (2) the liquid-vapor interface has zero thickness,
is shown in Fig. 3 with the detailed geometric parameters as given (3) the sharp discontinuities of the properties are maintained
in Fig. 4 (here is the actual domain due to the symmetry of the across the interface,
structure). In this approach, a uniform heat flux is added to the bot- (4) the temperature at the liquid-vapor interface (here the wick
tom surface of the substrate. Consistent with previous investiga- interface) is the saturation temperature corresponding to
tions [10,12,13], the mathematical model developed herein is the local static pressure,
based on the following assumptions: (5) with the exception of the bottom of the evaporator substrate
and the top surface of the liquid in the compensation cham-
ber, all other thermal boundary conditions are chosen as adi-
abatic condition.

The above assumptions are mostly identical to those used by


Cao and Faghri [9], and Kaya and Goldak [10]. The problem studied
here is a conjugate heat transfer problem coupled with mass trans-
fer in the compensation chamber (liquid phase), the vapor removal
channel (vapor phase), and in the wick structure (liquid phase),
and also coupled with phase change taking place at the liquid–
vapor interface (evaporation at the wick border C, referring to
Fig. 3). The governing equations for the heat and mass transfer
(continuity, Darcy and energy) can be summarized as follows:
Governing equations:
Liquid flow and heat transfer in the compensation chamber
_l
m
u ¼ 0; v¼ ; w ¼ 0; ð1Þ
ql Acc
!
@T k @2T @2T @2T
v ¼ l þ þ ð2Þ
@y ql  cp;l @x2 @y2 @z2

_ l is the liquid mass flow rate per pitch.


here Acc ¼ Lpitch  Le , and m
Vapor flow and heat transfer in the vapor removal channel
Z     !
@uðxÞ 1 1 @T @T
¼  kw  kv dC;
@x W c  Hc C qv  hfg @n C @n Cþ
v ¼ 0; w ¼ 0; ð3Þ
!
@T kv @2T @2T @2T
u ¼ þ þ : ð4Þ
@x qv  cp;v @x2 @y2 @z2

Liquid flow and heat transfer in the wick


@u @ v @w
þ þ ¼ 0; ð5Þ
@x @y @z

K p @P w K p @Pw K p @Pw
u¼ ; v ¼ ; w¼ ; ð6Þ
ll @x ll @y ll @z
!
@T @T @T kw @2T @2T @2T
u þv þw ¼ þ þ : ð7Þ
@x @y @z ql  cp;l @x2 @y2 @z2

The adequacy of the Darcy law to describe the flow inside the
wick has been discussed extensively by Kaya and Goldak [10],
especially for low heat flux rates. With the boundary conditions
(hydrodynamic and thermal) shown in Table 1, the problem is
Fig. 3. Schematic of the computational model for the LHP evaporator. mathematically closed and the pressure, velocity and the
J. Li, G.P. Peterson / International Journal of Heat and Mass Transfer 54 (2011) 564–574 567

Fig. 4. Geometric parameters of the numerical domain.

temperature can be solved numerically. A temperature boundary


condition at the liquid-vapor interface is adopted here due to the
At y = H (liquid phase),
assumption of a fully saturated wick. Kaya and Goldak [10] pro-
posed the following interfacial conditions to identify the vapor–
liquid interface in the wick, T ¼ T l;cc ¼ T s ðPl;cc Þ ð15Þ
At z = 0
Table 1
Boundary conditions.
@P @u @v @T
At x = 0, ¼ 0; ¼ 0; ¼ 0; w ¼ 0; ¼0 ð16Þ
for wick region and compensation chamber region @z @z @z @y
At z = Lpitch/2
@P @u @v @w @T
¼ 0; ¼ 0; ¼ 0; ¼ 0; ¼0 ð8Þ
@x @x @x @x @x @P @u @v @T
¼ 0; ¼ 0; ¼ 0; w ¼ 0; ¼0 ð17Þ
for vapor region @z @z @z @y
At the inner wick boundaries,
  At y = d,
25:511  1:065  hfg =Rg
Pv ;e ðx ¼ 0Þ ¼ exp ð9Þ
T e;v þ 273:15
@P
¼ 0; u ¼ 0; v ¼ 0; w¼0 ð18Þ
@y
@T At y = Hw,
u ¼ 0; v ¼ 0; w ¼ 0; ¼0 ð10Þ
@x
At x = L, _l
m
for wick region and compensation chamber region P ¼ Pl;cc ; u ¼ 0; v ¼_  ; w¼0 ð19Þ
ql Aw
here, Aw  Lpitch  Le  e
@P @u @v @w @T
¼ 0; ¼ 0; ¼ 0; ¼ 0; ¼0 ð11Þ At C,
@x @x @x @x @x
for vapor region     ! 
1 @T  @T  @P 
un ¼  kw   kv  ; 
ql hfg @n C @n Cþ @n C
 
25:511  1:065  hfg =Rg ll
Pv ;e ðx ¼ LÞ ¼ exp  DP v c ð12Þ ¼ un ; ð20Þ
T e;v þ 273:15 Kp
Z x¼Le Z  
1 1 @T 
uðx ¼ Le Þ ¼  kw
W c  Hc x¼0 C q v  hfg @n C x
  ! T n ¼ T s ½Pv ðxÞ; Pv ðxÞ ¼ Pv ;e  DPv c ð21Þ
Le
@T 
kv dC dx;
@n Cþ

@T The mass balance at C,


v ¼ 0; w ¼ 0; ¼0 ð13Þ
@x ðun Þv  qv ¼ ðun Þl  ql : ð22Þ
At y = 0 (solid boundary conditions)
And the energy balance at C,
     
@T  @T 
kcu
@T
¼q ð14Þ kw   kv  ¼ ðun Þl  ql  hfg ¼ ðun Þv  qv  hfg : ð23Þ
@y @n C @n C
568 J. Li, G.P. Peterson / International Journal of Heat and Mass Transfer 54 (2011) 564–574

The above considerations have been incorporated in the present Table 2


model along with the boundary conditions described in Eqs. (13) Wick properties.

and (20) for the completely saturated wick. However, if the wick Powder Porosity, Permeability, Effective thermal
is partially saturated, not as described by the present model, the diameter (lm) e Kp (m2) conductivity, kw (W/m K)
boundary conditions Eqs. (13) and (20) may still be applicable, 128.0 ± 22.0 0.5 ± 0.1 6.075  1011 10.36
but the position of the interface will no longer be at the wick
boundary, C. This situation will need to be investigated further
in future studies along with the vapor flow inside the wick, which
should be added to the model, in a manner similar to the liquid
flow controlled by the Darcy law.
Through careful examination of the present model, as given by
Eqs. (1)–(21), one may be curious in how to determine the vapor
saturation pressure Pv,e as it appears in Eq. (9), the saturation pres-
sure in the compensation chamber Pl,cc in Eq. (15) and the thermal
conductivity of the liquid saturated wick kw, which are the three
critical parameters that govern the final results of the simulation.
Pv,e is the vapor pressure in the vapor removal channel, and can
be approximated from the vapor temperature, Tv,e, coupled with
the Clausius–Clapeyron equation. The semi-empirical model pro-
posed by Chernysheva et al. [7] can be used to determine the vapor
temperature in the vapor removal channel for a loop heat pipe un-
der a given heat load and a detailed derivation for Pv,e and Pcc is
presented in Appendix A.
The permeability of the porous structure, Kp, used in Eqs. (6) and
(20) can be calculated directly from the well known Blake–Kozeny
equation [15] for the sintered spherical metal powder,
2
dpowder  e3 Fig. 5. A SEM scanning of the copper powder wick.
Kp ¼ : ð24Þ
150ð1  eÞ2
depending on the various powder densities or different sintering
Determination of the effective thermal conductivity of the li-
forces used during the wick preparation process. However, using
quid saturated wick utilized several well-accepted models for sin-
a comparative weight analysis method, which compares the mea-
tered spherical metal powder:
sured weight of the fabricated wick and the calculated weight of
Gorring and Churchill’s model
the wick from its material density and volume, the measured data
kl ½ð2kl þ ks Þ  2ð1  eÞðkl  ks Þ is close to the data reported in Ref. [18], i.e., approximately 40%).
kw ¼ ; ð25Þ
½ð2kl þ ks Þ þ ð1  eÞðkl  ks Þ
where ks is the thermal conductivity of the wick material (i.e., 3. Numerical procedure
copper).
Maxwell’s model The governing partial differential equations with the boundary
  conditions were discretized using a control volume method with
2 þ kl =ks  2eð1  kl =ks Þ grid points placed at the center of each cell. The upwind scheme
kw ¼ ks : ð26Þ
2 þ kl =ks þ eð1  kl =ks Þ was used for the convection term in Eq. (2), Eqs. (4) and (7), and
were solved by marching downstream. In order to expedite the
Chaudary and Bhandari’s model
convergence of the calculations, a line-by-line iteration and a Tri-
kw ¼ ðkmax Þn  ðkmin Þ1n ; ð27Þ diagonal Matrix Algorithm (TDMA) along with a Thomas algorithm
solver, and successive under-relaxation iterative methods were
where used to obtain the three-dimensional temperature distribution,
kmax ¼ e  kl þ ð1  eÞ  ks ðparallel caseÞ; T(x, y, z). The overall numerical procedure is illustrated in Fig. 6.
The convergence criteria were established as:
kmin ¼ ðkl  ks Þ=½e  ks þ ð1  eÞ  kl  ðseries caseÞ: XXX
jP w;new ði; j; kÞ  P w;old ði; j; kÞj 6 103 ð28Þ
From other literatures [16,17], it was found that Gorring and
Churchill’s model underestimated the value of kw, while Maxwell’s for pressure the distribution in the wick and
model over-predicted the value of kw, by several times of the value XXX
jT new ði; j; kÞ  T old ði; j; kÞj 6 102 ð29Þ
as measured by Singh [16]. Chaudary and Bhandari’s model has
been demonstrated to be appropriate for the present type of wick for the temperature distribution in the domain.
structure when n = 0.42 [8,17]. The total grid number is 84,000 (i(x)  j(y)  k(z) is
All the necessary property information associated with the wick 50  56  30) for the domain as shown in Fig. 4. Eq. (29) yields
is shown in Table 2, as determined by Eqs. (24) and (27) with the an approximate criterion for the mean square root error (MSRE)
aid of experimental measurements for the copper powder wick. P P P pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2
ðTði;j;kÞT 0 ði;j;kÞÞ
Based on the data reported in the literature [18], the SEM measure- of jTði; j; kÞ  T 0 ði; j; kÞj  N
6 1:2  107 . This
ments as shown in Fig. 5, and the weight analysis method, for the type of a fine grid mesh for the x, y and z directions was chosen
present wick structure, the porosity was determined to be 50% for in order to properly resolve the boundary conditions, and to better
the sintered powder with mesh numbers between 100 and 140 (it define the conjugate heat transfer at the surface of the channel,
should be noted here that for such a wick, the measured porosity as thereby improving the temperature resolution. At the border of
determined from SEM measurements was between 50% and 60%, the different regions, the interface thermal conductivity is set as
J. Li, G.P. Peterson / International Journal of Heat and Mass Transfer 54 (2011) 564–574 569

computational results are compared with the experimental data


previously published in the literature [5]. It should be noted that
the working fluid is pure water and the ambient temperature is as-
sumed to be 26.5 °C. The heat loads utilized in the simulation were
31.25 W (q = 5 W/cm2 in Eq. (14)), 93.75 W (q = 15 W/cm2), and
187.5 W (q = 30 W/cm2), respectively.
Referring to Fig. 6, a detailed thermodynamic analysis was con-
ducted to determine the pressure drop at each component in the
loop, for a given heat load under a fixed ambient temperature. In
addition, the saturation temperature corresponding to the pressure
was calculated using the Clausius–Clapeyron Eq. (A.11). An illus-
tration of the thermodynamic cycle for the loop is illustrated in
Fig. 7. The numbers shown indicate the location of the working
fluid.
Fig. 8 represents the numerical results for the temperature dis-
tribution in the LHP evaporator for the case q = 15 W/cm2. Fig. 8a
shows the local cross-sectional temperature distribution in the
x–y plane at z = Lpitch/2. As shown in Fig. 8a, the temperature is lay-
ered, which is characteristic of these types of structures in the LHP
evaporator. Referring to Figs. 4 and 8, from the bottom to the top, is
the substrate, the vapor removal channel, the wick and the liquid,
respectively. Fig. 8b illustrates the cross-sectional temperature dis-
tribution in the LHP evaporator in the y–z plane at x = Le/2. And
Fig. 8c shows the local cross-sectional temperatures inside the
wick at x = Le/2, corresponding to Fig. 8b. From Fig. 8c, it can be
found that most parts of the wick do not have significant difference
in the temperature distribution except for where the wicked fin
foot contacts the substrate which is subjected to the heat load.
The highest temperature occurs at the central point of the wicked
fin foot. This result agrees with the experimental observations ob-
tained through the use of an infrared camera [5,6]. From the sim-
ulations, for the other cases (q = 5 W/cm2, and q = 30 W/cm2), it
is found that the pattern of the temperature distribution inside
the evaporator and the wick is similar in shape to the case
q = 15 W/cm2 as shown in Fig. 8. This similarity is mainly due to
the assumption that there is no superficial liquid-vapor interface

Fig. 6. Overall numerical procedure.

a compound value of the two different materials. This approach is


widely used in these types of situations [9]. Based on previous
investigations [9,10], a grid sensitivity was performed and the cell
number used was the optimal result based upon the computational
efficiency and accuracy. Validation of the numerical program has
been previously discussed elsewhere [19,20].

4. Results and discussion

A series of numerical simulations have been conducted and pre-


sented to determine the effects of the heat loads on the velocity Fig. 7. Illustration of thermodynamic cycle of working fluid in a LHP on temper-
field and temperature distribution in the LHP evaporator. The ature–volume diagram.
570 J. Li, G.P. Peterson / International Journal of Heat and Mass Transfer 54 (2011) 564–574

formed inside the wick as made in the present model, but the mag- perature at the wick border in the x-direction is slightly decreasing
nitudes of the temperature are different for different cases. based on the Clausius–Clapeyron equation and the vapor pressure
As illustrated in Fig. 8a, the temperature does not exhibit any drop in the vapor removal channels (actually, this difference is
obvious differences in the x-direction (or vapor flow direction) in negligible, e.g. 0.0012 °C decrease for q = 15 W/cm2, 0.0025 °C for
the evaporator, primarily due to the fact that the saturation tem- q = 15 W/cm2, 0.0031 °C for q = 30 W/cm2, respectively). This result

Fig. 8. Temperature distribution in the LHP evaporator for q = 15 W/cm2.


J. Li, G.P. Peterson / International Journal of Heat and Mass Transfer 54 (2011) 564–574 571

also indicates that the majority of the heat is absorbed and deliv- corresponding to Fig. 8, which was obtained through numerically
ered through the evaporation of the working fluid, and only a very solving the Laplace equation r2 P ¼ 0, which comes from combin-
small amount of heat is transferred by convection resulting from ing Eqs. (5) and (6), and with the associated boundary conditions
the vapor flowing through the vapor removal channels. Kaya and Eqs. (18)–(21). The pressure variation in the wick is compared with
Goldak [10] have previously stated that the temperature difference the local vapor pressure and is represented by variations in color,
is not significant along the axial direction of the cylindrical evapo- with the pressure scale located at the side of each figure. The vapor
rator in LHPs and thus they adopted a two-dimensional mathemat- pressure at a given position of the vapor removal channel along the
ical model in their simulation work. The present study verifies that x-direction were predetermined using the mathematical model
such an assumption is acceptable if no other mechanism than the presented in the Appendix A prior to initiating the numerical sim-
model developed herein is involved. ulation as discussed before. The sharp pressure difference at the
The absolute pressure distribution inside the wick for the case wick-channel border can be recovered by the capillary force
q = 15 W/cm2 is illustrated in Fig. 9 in the y–z plane at x = Le/2, DPcap ¼ Pv ;e  P l;w ¼ r 2r , which is equal to Eq. (A.2). The rough
meniscus

Fig. 9. Pressure distribution in the wick compared to the local vapor pressure in the vapor removal channel at x = Le/2 for q = 15 W/cm2.

Fig. 10. Velocity vector field in the wick at x = Le/2 for q = 15 W/cm2.

Fig. 11. Heat flux distribution in the wick at x = Le/2 for q = 15 W/cm2.
572 J. Li, G.P. Peterson / International Journal of Heat and Mass Transfer 54 (2011) 564–574

averaged magnitude of the pressure drop (flow resistance) in the


wick are 2.75 Pa, 9.1 Pa, and 18.9 Pa for q = 5 W/cm2, 15 W/cm2,
30 W/cm2, respectively from the computations.
Correspondingly, the velocity vector field in the wick is illus-
trated in Fig. 10 in the y–z plane at x = Le/2 for q = 15 W/cm2, which
was obtained from Eq. (6) with the boundary conditions described
in Eqs. (18)–(21) after the pressure distribution in the wick was
obtained.
In order to help in understanding the heat transfer mechanism
along the liquid–vapor interface inside the evaporator, an illustra-
tion of the heat flux distribution is given in Fig. 11. The highest heat
flux occurs at the wick border where it contacts intimately with
the substrate and the heat flux decreases along the border of the
wicked fin quickly. This numerical observation verifies Demidov
and Ystsenko’s argument as being pointed out in the section of
introduction.
In order to verify the accuracy of the numerical model and the Fig. 13. Case temperature comparison between the numerical simulations and the
ability to accurately predict the performance of LHPs, a comparison experimental measurements for different heat loads.

of the simulation results and the experimental data was per-


formed. A detailed description of the test facility and the various
the compensation chamber. However, at the present time, the
parameters utilized during the LHP operation have been reported
influence from the above two matters are very difficult to
previously [5].
evaluate.
It should be noted that for a given heat load, with the aid of the
Fig. 13 illustrates a comparison between the numerical results
semi-empirical theoretical model given by Chernysheva et al. [7]
for the bottom surface temperature of the LHP evaporator and the
(as partially summarized in Appendix A), the saturation status at
experimental results under the same operation parameters
the top surface of the liquid in the compensation chamber and
(referring to [5,6]). In Fig. 13, the dotted line shows the experimental
at the vapor removal channel was predetermined quantitatively
results of the average temperature at the bottom surface of the LHP
at the start of the numerical simulation. Fig. 12 shows the theoret-
evaporator for Qload = 30 W, 100 W, and 200 W, respectively, and the
ical calculations of saturation status of the working fluid inside the
starred line shows the numerical results at the bottom surface of
loop and the comparison with the experimental measurements at
the LHP evaporator for Qload = 31.25 W (q = 5 W/cm2), 93.75 W
different places on the external surface of the tested loop heat pipe
(q = 15 W/cm2), and 187.5 W (q = 30 W/cm2), respectively.
for q = 15 W/cm2 with an illustration of a schematic of the loop
If we take into account the numerical solution error, the exper-
heat pipe in test (here five microscale T-type Omega thermocou-
imental uncertainties, and the limitations of the present model, the
ples were firmly attached onto the loop heat pipe by film adhesive
present numerical approach correlates quite well, particularly in
as shown, to measure: the temperature at the bottom surface of
the region surrounding the point where q = 15 W/cm2. It can be in-
the evaporator (102); the temperature at the exit of the evaporator
ferred from these results that in this region, the wick satisfies the
(103); the temperature at the entrance of the condenser (104); the
assumption of a perfectly saturated wick, which was also con-
temperature at the exit of the condenser (105); the temperature at
firmed by a previous investigation [6] for Qload = 100–150 W after
the entrance of the evaporator (106); and the temperature at the
the LHP reached steady-state operation. Besides the reasons as dis-
top surface of the compensation chamber (107), respectively).
cussed above, the disparity between the numerical results and the
From Fig. 12, it can be seen that the theoretically predicted
experimental measurements at heat loads other than q = 15 W/cm2
temperature is somewhat different from the measurements. The
can be divided into two different situations: First, if the heat flux is
reasons which cause this difference occur primarily from the fol-
smaller than the value required to force all of the liquid inside the
lowing unpredictable conditions: (1) the subcooling status at the
vapor removal channels out, large oscillations in the LHP tempera-
exit of the condenser; (2) the heat leak from the evaporator to
ture during operation will occur [5,6]. The blockage of the vapor
removal channels can significantly degrade the performance of
the loop heat pipe; second, if the heat flux is higher than the value
required to maintain the superficial liquid–vapor interface at the
wick-channel border, a liquid–vapor interface will be formed
inside the wick and the evaporation interface area will be enlarged.
This will result in an enhanced heat transfer mechanism in the
evaporator. A new computational methodology is currently being
developed that will be able to identify the evolution process of
the liquid–vapor interface in the wick automatically for different
heat loads.

5. Conclusions

In this work, a practical quasi-3D numerical model for fluid flow


and heat transfer in a flat square LHP evaporator was developed
and the local heat transfer mechanism inside the LHP evaporator
was determined by analyzing the simulation results. The results
Fig. 12. Theoretical calculations of fluid temperature inside the loop and their
of the numerical model compare quite well with the available
comparison with the experimental measurements at different places on the experimental results for the fully saturated operational mode in
external surface of the tested loop heat pipe under q = 15 W/cm2. the wick at low heat loads. Based on the present model, further
J. Li, G.P. Peterson / International Journal of Heat and Mass Transfer 54 (2011) 564–574 573

geometric optimization in term of the performance of loop heat T v ;e ¼ T a


pipes is possible for a specific heat load.  !
1 1 X dT  1
þ þ Rwall þ þ DPi   þ
ac;ext Sc;ext ac;int Sc;int i
dP T ae Se;activ e
Acknowledgements
 Q;
This work was partially supported by the President Fund of ðA:10Þ
Graduate University of the Chinese Academy of Sciences
where ac,ext is the convective heat transfer coefficient at the external
(085101AM03) and partially supported by National Science and
surface of the condenser and Sc,ext is the total heat transfer surface
Technology Ministry (2009GB104001).
area of the fins in the condenser; ac,int the heat transfer coefficient
during the vapor condensation in the condenser serpentine pipe
Appendix A. Determination of saturated pressures and and Sc,int is the inner surface area of the condenser pipe; Rwall is
temperatures inside the loop the thermal resistance of the condenser pipe wall; ae is the
equivalent evaporation heat transfer coefficient in the evaporator
Referring to Fig. 7, given a certain heat load, the mass flow rate and Se,active is the evaporator surface area to which heat Q is sup-
P
of liquid in a loop heat pipe is, plied (active area); i DP i is the total pressure drop of vapor during
the motion of vapor from the evaporator into the condenser and
Q load given by Eq. (A.3); Rwall ¼ dwall =ðkwall  Swall Þ, dwall is the thickness of
_ ¼
m : ðA:1Þ
hfg the condenser tube and Swall is  the outer surface area of the con-
denser tube; The derivative dT  is a thermophysical characteristic
It is well known that the pressure losses in the loop should be dP T

balanced by the capillary force in order to maintain continuous of the working fluid which is taken along the liquid–vapor satura-
operation of the loop heat pipe, tion line and can be calculated from the Clausius–Clapeyron equa-
tion approximately at a reference temperature. It should be noted
DPcap ¼ DPv þ DPl þ DPw  DP g : ðA:2Þ that ac,ext, ac,int and ae are very difficult to calculate accurately from
any theoretical correlations (of course, in principle, the theoretical
The pressure loss for the vapor phase in the loop can be calcu-
solutions for these coefficient may exist) and more realistically
lated from
these coefficients should be identified by careful experiments. For
X
DP v ¼ DPi ¼ DP v c þ DPv l þ DPv ;condenser : ðA:3Þ the prototype of the loop heat pipe in this study, all of these coeffi-
i cient and parameters were presented in Refs. [5,6].
With the aid of Clausius–Clapeyron equation for water–vapor
The terms in the right side of the equation above takes into ac- saturation lines, the fitting equation for water based on the pub-
count the pressure drop in the vapor removal channels (rectangu- lished data is
lar shape) in the evaporator DPvc, in the vapor line (circular pipe)  
DPvl, and in the vapor section of the condenser (circular pipe) 25:511  1:065  hfg =Rg
Pv ;e ¼ exp : ðA:11Þ
DPv,condenser respectively and can be evaluated from the Fanning T e;v þ 273:15
friction equation by substituting different friction factors in the
And the saturation pressure of vapor in the vapor removal chan-
equation for the different geometries respectively,
nels can be determined by substituting Eq. (A.10) into Eq. (A.11).
In the vapor removal channel, 2r
From Young–Laplace equation DPcap ¼ Pv ;e  P l;w ¼ rmeniscus , the li-
 v c  c  lv  lv c
u quid pressure at the liquid–vapor interface in the wick can be ob-
DP v c ¼ 2 2
; ðA:4Þ
dv c tained as,
_
m 1þa H 2 Pl;w ¼ Pv ;e  DP cap : ðA:12Þ
qv , c ¼ 4:7 þ 19:64 ð1þaÞ2 , and a ¼ W .
 v c ¼ n HW
where u vc
In the vapor line, By adding the pressure loss during the liquid flowing inside the
128  lv l  tv  m
_ wick, the saturated pressure inside the compensation chamber can
DP v l ¼ : ðA:5Þ be identified as,
p  d4v l
Pl;cc ¼ Pl;w þ DPl;w : ðA:13Þ
And in the vapor section of the condenser,
Once more, by applying the equation of (A.11), the temperature
128  lv ;c  tv  m
_
DPv ;condenser ¼ : ðA:6Þ of two phase mixture of water in the compensation chamber can
p  d4c be fixed!
For the pressure loss of the liquid phase in the loop,
References
128  lll  tl  m
_
DPl ¼ Pl;c  Pl;cc ¼ : ðA:7Þ
p  d4ll [1] Y.F. Maydanik, S.V. Vershinin, M.A. Korukov, J.M. Ochterbeck, Miniature loop
heat pipes – a promising means for cooling electronics, IEEE Trans. Compon.
Pack. Technol. 28 (2) (2005) 290–296.
For the pressure loss in the wick during liquid flow,
[2] Y.F. Maydanik, Loop heat pipes (review), Appl. Thermal Eng. 25 (2005) 635–
ll  ul;w  dw 657.
DPw ¼ Pl;cc  P l;w ¼ Darcy Equation; ðA:8Þ [3] J. Ku, Operating Characteristics of Loop Heat Pipes, in: 29th International
Kp Conference on Environmental System, Paper No. 1999-01-2007, Denver, USA,
1999.
 l;w ¼ q mA_ .
where u [4] S. Launay, V. Sartre, J. Bonjour, Parametric analysis of loop heat pipes
w l
operation: a literature review, Int. J. Thermal Sci. 46 (2007) 621–636.
And finally, from the gravity,
[5] J. Li, D. Wang, G.P. Peterson, Development of a Robust Miniature Loop Heat
DPg ¼ ðql  qv Þgh: ðA:9Þ Pipe for High Power Chip Cooling, Paper No. MNHMT2009-18011, ASME 2nd
Micro/Nanoscale Heat & Mass Transfer International Conference, Shanghai,
Thus, all of the terms in Eq. (A.2) can be fixed theoretically. China, 2009.
[6] J. Li, D. Wang, G.P. Peterson, Experimental studies on a high performance
From Chernysheva et al. [7], for the evaporator temperature un- compact loop heat pipe with a square flat evaporator, Appl. Thermal Eng. 30
der a certain heat load Q, (6–7) (2010) 741–752.
574 J. Li, G.P. Peterson / International Journal of Heat and Mass Transfer 54 (2011) 564–574

[7] M.A. Chernnysheva, S.V. Vershinin, Y.F. Maydanik, Operating temperature and [14] C. Li, G.P. Peterson, J. Li, Visualization of boiling and evaporation of water on
distribution of a working fluid in LHP, Int. J. Heat Mass Transfer 50 (2007) micro porous media, ASME Summer Heat Transfer Conference, Paper No.
2704–2713. HT2008-56352, Florida, USA, 2008.
[8] L. Bai, G. Lin, H. Zhang, D. Wen, Mathematical modeling of steady-state [15] S.W. Chi, Heat Pipe Theory and Practice, McGraw-Hill, New York, 1976.
operation of a loop heat pipe, Appl. Thermal Eng. 29 (13) (2009) 2643–2654. [16] R. Singh, A. Akbarzadeh, M. Mochizuki, Effect of wick characteristics on the
[9] Y. Cao, A. Faghri, Conjugate analysis of a flat-plate type evaporator for capillary thermal performance of the miniature loop heat pipe, ASME J. Heat Transfer
pumped loops with three-dimensional vapor flow in the groove, Int. J. Heat 131 (2009) (paper #082601).
Mass Transfer 37 (3) (1994) 401–409. [17] M.L. Parker, Modeling of Loop Heat Pipe with Applications to
[10] T. Kaya, J. Goldak, Numerical analysis of heat and mass transfer in the capillary Spacecraft Thermal Control, Faculty of Mechanical Engineering and
structure of a loop heat pipe, Int. J. Heat Mass Transfer 49 (2006) 3211–3220. Applied Mechanics, University of Pennsylvania, Pennsylvania, United
[11] M.A. Chernysheva, Y.F. Maydanik, Numerical simulation of transient heat and States, 2000.
mass transfer in a cylindrical evaporator of a loop heat pipe, Int. J. Heat Mass [18] D. Reay, P. Kew, Heat Pipes, fifth ed., Elsevier, Burlington, 2006.
Transfer 51 (2008) 4202–4215. [19] J. Li, G.P. Peterson, 3-Dimentional numerical optimization of silicon-based high
[12] A.S. Demidov, E.S. Yatsenka, Investigation of heat and mass transfer in the performance parallel-channel micro-heat sink with liquid flow, Int. J. Heat
evaporation zone of a heat pipe operating by the ‘inverted meniscus’ principle, Mass Transfer 50 (15–16) (2007) 2895–2904.
Int. J. Heat Mass Transfer 37 (14) (1994) 2155–2163. [20] J. Li, G.P. Peterson, P. Cheng, Three-dimensional analysis of heat transfer in a
[13] C. Figus, Y. Le Bray, S. Bories, M. Prat, Heat and mass transfer with phase micro-heat sink with single phase flow, Int. J. Heat Mass Transfer 47 (19–20)
change in a porous structure partially heated: continuum model and pore (2004) 4215–4231.
network simulations, Int. J. Heat Mass Transfer 42 (1999) 2557–2569.
Update
International Journal of Heat and Mass Transfer
Volume 54, Issue 17–18, August 2011, Page 4152

DOI: https://doi.org/10.1016/j.ijheatmasstransfer.2011.04.036
International Journal of Heat and Mass Transfer 54 (2011) 4152

Contents lists available at ScienceDirect

International Journal of Heat and Mass Transfer


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

Corrigendum

Corrigendum to ‘‘3D heat transfer analysis in a loop heat pipe evaporator


with a fully saturated wick’’ [Int J Heat Mass Transfer 54 (2011) 564–574]
Ji Li a,⇑, G.P. Peterson b
a
Laboratory of Electronics Thermal Management, College of Physics, Graduate University of the Chinese Academy of Sciences, 19A Yu-quan-lu Road,
Shijingshan District, Beijing 100049, PR China
b
The G.W. Woodruff School of Mechanical Engineering, Georgia Institute of Technology, Atlanta, GA 30332, United States

The authors regret that an error occurred in the Introduction section of the above paper. In the first paragraph of the Introduction, the
third sentence should read as follows:
‘‘A loop heat pipe (LHP) is a two-phase heat transfer device that removes heat from a source (e.g., an electronic chip) through an evap-
orator and passively moves it to a condenser region or radiator using capillary forces to pump the fluid, ultimately releasing the heat to the
environment from the condenser by natural or forced convection. Loop heat pipes were first investigated and patented in the USSR in 1974
by Y. Maydanik and his collaborators [1,2].’’

DOI of original article: 10.1016/j.ijheatmasstransfer.2010.09.014


⇑ Corresponding author. Tel.: +86 13522278866.
E-mail address: jili@gucas.ac.cn (J. Li).

0017-9310/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijheatmasstransfer.2011.04.036

You might also like