Please do not remove this page
Miniature loop heat pipe with flat evaporator for
cooling computer CPU
Singh, Randeep; Akbarzadeh, Aliakbar; Dixon, Christopher; Mochizuki, Mastaka; Riehl, Roger
https://researchrepository.rmit.edu.au/discovery/delivery/61RMIT_INST:ResearchRepository/12247267810001341?l#13248377940001341
Singh, Akbarzadeh, A., Dixon, C., Mochizuki, M., & Riehl, R. (2007). Miniature loop heat pipe with flat
evaporator for cooling computer CPU. IEEE Transactions on Components and Packaging Technologies, 30,
42–49. https://doi.org/10.1109/TCAPT.2007.892066
Published Version: https://doi.org/10.1109/TCAPT.2007.892066
Downloaded On 2022/08/10 12:46:06 +1000
© 2007 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this
material for advertising or promotional purposes or for creating new collective works for resale or
redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must
be obtained from the IEEE.
Repository homepage: https://researchrepository.rmit.edu.au
Please do not remove this page
42 IEEE TRANSACTIONS ON COMPONENTS AND PACKAGING TECHNOLOGIES, VOL. 30, NO. 1, MARCH 2007
Miniature Loop Heat Pipe With Flat Evaporator
for Cooling Computer CPU
Randeep Singh, Aliakbar Akbarzadeh, Chris Dixon, Mastaka Mochizuki, and Roger R. Riehl
Abstract—This paper presents an experimental investigation
on a copper miniature loop heat pipe (mLHP) with a flat disk
shaped evaporator, 30 mm in diameter and 10-mm thick, designed
for thermal control of computer microprocessors. Tests were
conducted with water as the heat transfer fluid. The device was
capable of transferring a heat load of 70 W through a distance up
to 150 mm using 2-mm diameter transport lines. For a range of
power applied to the evaporator, the system demonstrated very
reliable startup and was able to achieve steady state without any
symptoms of wick dry-out. Unlike cylindrical evaporators, flat
evaporators are easy to attach to the heat source without need of
any cylinder-to-plane reducer material at the interface and thus
offer very low thermal resistance to the heat acquisition process.
In the horizontal configuration, under air cooling, the minimum
value for the mLHP thermal resistance is 0.17
C/W with the
corresponding evaporator thermal resistance of 0.06
C/W. It is
concluded from the outcomes of the current study that a mLHP
with flat evaporator geometry can be effectively used for the
thermal control of electronic equipment including notebooks with
limited space and high heat flux chipsets. The results also confirm
the superior heat transfer characteristics of the copper-water
configuration in mLHPs.
Index Terms—CPU cooling, flat evaporator, miniature loop heat
pipe (mLHP), thermal control.
NOMENCLATURE
A
Area, m
.
R Thermal resistance,
C/W.
T Temperature,
C, K
Heat load, W.
h
Heat transfer coefficient, W/m
C.
M
Merit number, W/m
.
L Latent heat of vaporization, J/Kg.
Greek Symbols
Surface tension N/m.
Density, Kg/m .
Viscosity, Pa.s.
Manuscript received November 4, 2005; revised April 4, 2006. This work was
recommended for publication by Associate Editor C. Patel upon evaluation of
the reviewers’ comments.
R. Singh, A. Akbarzadeh, and C. Dixon are with the Energy CARE Group,
School of Aerospace, Mechanical and Manufacturing Engineering, Royal Mel-
bourne Institute of Technology (RMIT) University, Victoria 3083, Australia
(e-mail: randeep.singh@rmit.edu.au).
M. Mochizuki is with Fujikura, Ltd., Tokyo 135, Japan.
R. R. Riehl is with the Space Mechanics and Control Division, National In-
stitute for Space Research—INPE, São José dos Campos 12227-010, Brazil.
Digital Object Identifier 10.1109/TCAPT.2007.892066
Subscripts
Ambient.
Compensation chamber.
Condenser.
Evaporator.
Junction.
Heat pipe.
Total.
Vapor.
Liquid.
I. I
NTRODUCTION
T
HE development of high-end and compact computers has
resulted in a considerable rise in the power dissipation
tendency of their microprocessors. At present, the waste heat
released by the central processing Unit (CPU) of a desktop and
server computer is 80 to 130 W and of notebook computer is 25
to 50 W [1]. In the latter case, the heating area of the chipset has
become as small as 1–4 cm
. This problem is further complicated
by both the limited available space and the restriction to maintain
the chip surface temperature below 100
C. It is expected that
conventional two-phase technologies [2] like heat pipes and
vapor chambers will not be able to meet the futuristic thermal
needs of the next generation notebooks. Other technologies like
liquid cooling and thermoelectric coolers have good potential but
still create major integration, reliability and cost issues. With the
development in the two-phase heat transfer systems and porous
media technology, loop heat pipes (LHPs) have come up as a
potential candidate to meet these challenging needs [3].
Originally known as an antigravitational heat pipe (AGHP),
the LHP is a highly efficient two-phase heat transfer device
that operates passively on the basis of a capillary driven loop
scheme. It utilizes capillary pressure developed by the fine pore
wick to circulate the working fluid, and the latent heat of va-
porization and condensation of the working fluid to acquire and
transport heat loads. LHPs possess all the advantages of the
conventional heat pipe and additionally provide reliable oper-
ation over long distance at any orientation in the gravity field.
At present, LHPs with different architectures and performance
capabilities are widely used in space applications [4].
Various aspects of the theory and principle of LHPs have been
examined in detail [5] in the past. Most of this research is con-
centrated on the LHPs with large size cylindrical evaporators,
12 to 30 mm in diameter, for space applications. In order to em-
ploy LHPs for cooling of compact electronic equipments like
1521-3331/$25.00 © 2007 IEEE
Authorized licensed use limited to: RMIT University. Downloaded on November 23, 2008 at 21:47 from IEEE Xplore. Restrictions apply.
SINGH et al.: MINIATURE LHP 43
notebooks, their potential in the direction of miniaturization has
to be evaluated. By now different prototypes of the miniature
loop heat pipe (mLHP) have been created and tested for cooling
of power saturated electronics [6]. Usually, the design configu-
ration of these mLHPs consisted of cylindrical evaporator with
ammonia as working fluid. Pastukhov
et al. [7] developed dif-
ferent designs of cylindrical evaporator mLHP with a nominal
capacity of 25–30 W and a heat transfer distance up to 250 mm
for cooling the CPU of a mobile PC. Under air cooling the total
thermal resistance (interface to ambient) of such a system lies
in the range between 1.7 and 4.0
C/W with the mLHP own
thermal resistance of 0.3 to 1.2
C/W. An ammonia charged
mLHP was also tested [8] for low power management and tem-
perature control for electronic components especially for space-
craft applications.
Apart from the cylindrical design, the flat shape is another
design option for the LHP evaporator. Flat evaporators can be
considered as the optimum design from the point of view that
they do not need any special thermal interface (saddle), i.e.,
cylinder-plane reducer to provide a thermal contact with the heat
load source. In case of cylindrical evaporators, such a saddle
creates an additional thermal resistance and increase the LHP
total mass [6]. Also, flat evaporators can be easily integrated
into the compact space inside the object to be cooled (e.g., note-
book). By now, various investigative prototype of mLHP with
flat evaporators and ammonia-stainless steel configuration have
also been developed for space as well as ground based appli-
cations. Boo and Chung [9] demonstrated a successful working
of a mLHP with a flat evaporator, 40
50 mm active area and
30-mm thickness, using a polypropylene wick. With ammonia
as the working fluid the designed mLHP showed a maximum
heat load of 87 W and thermal resistance of 0.65
C/W in the
horizontal position. Another prototype of ammonia mLHP [10]
was developed with flat rectangular evaporator of 5.5-mm thick-
ness and heat transfer length of 75 mm. The maximum heat
load with air blowing was 30 W and the total thermal resis-
tance “evaporator-air” was 2
C/W. Delil et al. [11] report de-
velopment of a mLHP having a flat disk shaped evaporator with
44-mm diameter and 22-mm thickness. Using nickel wick and
ethanol as the working fluid, the device was able to transfer a
maximum thermal load of 120 W with a thermal resistance in
the range of 0.62 to 1.32
C/W at different orientations in the
gravity field.
So far, most of the mLHP prototypes have been focused on
the thermal control of electronics in space. Under such circum-
stances, a low temperature working fluid like ammonia and a
durable material like stainless steel for the loop container are
the best options. In order to exploit the potential of the mLHP in
ground based electronic cooling including personal and mobile
computers, certain safety measures have to be observed that
restrict the use of high pressure, toxic or inflammable working
fluids like ammonia, acetone or different grades of alcohol. In
this regard, water can be considered as the ideal working fluid.
Water is the best low temperature working fluid with efficient
heat transfer characteristics, presents no hazard to people, and
is fully compatible with high thermal conductive material like
copper. For ground based electronic cooling, the copper and
water combination is considered very competitive and is widely
used in conventional heat pipes for this purpose. Kim et al.
[12] developed a high performance remote heat exchanger
(RHE) from a copper-water heat pipe that was able to transfer
75-W heat load dissipated by a desktop CPU of 35
35 mm
area. Moon et al. [13] built and performed test on a copper
based miniature heat pipe, 2-mm thick, with woven wire wick
and water as the working fluid. The miniature heat pipe de-
vice was able to effectively manage the heat load of 11.5 W
from a 35
35 mm laptop CPU. Water due to its excellent
heat transfer characteristics like high latent heat, high surface
tension and easy of availability in pure state is a widely used
heat pipe working fluid in the computer cooling applications.
Various copper-water based thermal designs using heat pipe or
vapor chamber has made it possible to extend and maximize the
cooling capability of the heat sinks for electronic applications
[14]. However, it is expected that the convectional heat pipe
design will not be able to provide reliable thermal control
for the next generation high-end computers with extensive
processing capabilities. This has instigated the need to focus
on the alternative and efficient thermal designs using mLHPs.
Miniature LHPs with copper-water arrangement can serve as
potential replacements of conventional heat pipes in electronic
cooling applications.
Although the copper-water combination is not yet fully ex-
plored in mLHPs, some of the investigative prototypes which
have been developed with this combination have shown supe-
rior performance and high heat manageable potential. In their
work on mLHPs, Maydanik et al. [15] developed and tested dif-
ferent prototypes of ammonia-stainless steel and copper-water
mLHPs with cylindrical evaporators, 5 and 6 mm in diameter,
and heat transfer distances up to 300 mm. Tests have shown that
a nominal heat load of 70 W for ammonia mLHP and 130 W for
water mLHP can be obtained. The work clearly pinpoints the
potential of the copper-water mLHP for electronic cooling ap-
plications. Kiseev et al. [16] presented his results on the mLHP
with flat evaporator, 31.8 mm diameter and 15-mm thick, using
different configuration of capillary structure (i.e., nickel and ti-
tanium) and working fluids (i.e., acetone and water). With water
as the heat transfer agent, the device showed higher heat po-
tential of 160 W while the thermal resistance was within 0.5
to 0.8
C/W. It is evident from the current body of knowledge
that very limited research has been done on the mLHPs with
water as the working fluid. In order to utilize the unique thermal
characteristics of the mLHPs for cooling compact electronics
like laptop, development of these devices in this respect is very
crucial.
In this study, the design, development and results of tests on a
mLHP with flat evaporator geometry are presented. A flat evap-
orator of copper with an active diameter of 30 mm and character-
istic thickness of 10 mm was developed to transport 70 W over
a distance of up to 150 mm. Here, a flat evaporator is preferable
due to lower interface resistance and easy of integration inside
the limited space of notebook PC. The device uses water as the
working fluid to promote its viability for cooling portable and
personal computers. Results of the experiment demonstrated
satisfactory operation and thermal behavior of LHPs on a minia-
ture scale.
Authorized licensed use limited to: RMIT University. Downloaded on November 23, 2008 at 21:47 from IEEE Xplore. Restrictions apply.
44 IEEE TRANSACTIONS ON COMPONENTS AND PACKAGING TECHNOLOGIES, VOL. 30, NO. 1, MARCH 2007
Fig. 1. Schematic of the experimental prototype and test layout for the mLHP:
(a) cross-sectional details of the mLHP evaporator; (b) cross section view of the
heater block showing cartridge heaters; (c) sectional view of the heater face
showing position of the thermocouple groove and temperature measurement
point; (d) bottom view of the mLHP showing thermocouple locations and heater
position; (e) condenser cross section; (f) top view of the mLHP showing dif-
ferent parts; and (g) side view of the mLHP clearly showing the heater, evapo-
rator and compensation chamber positions.
II.
MLHP PROTOTYPE DESCRIPTION
The design schematic of the mLHP is as shown in Fig. 1(f).
It consists of a flat disk shaped evaporator with a characteristic
diameter of 30 mm and thickness of 10 mm. Fig. 1(a) shows
TABLE I
M
AIN
DESIGN PARAMETERS OF THE
mLHP
the cross-sectional view of the evaporator. The thickness of the
mLHP evaporator also incorporates the compensation chamber
which is connected to the evaporator through the wick struc-
ture. The function of the compensation chamber is to accom-
modate the excess liquid inventory displaced from evaporator
grooves, vapor line and condenser during startup. Apart from
this, the compensation chamber also provides the wick struc-
ture with direct access to the liquid and thus promotes its wet-
ting at all the times. The compensation chamber design has to
take into consideration the total volume of the loop so that it can
accommodate most of the liquid inventory present in the loop.
The flat diametric face of the evaporator was used as the active
heating area and provides direct connection with the heat load
simulator without need of any cylinder-to-plane reducer mate-
rial (i.e., saddle) at the interface. An efficient system of vapor
removal channel was formed on the inner face of the heating
zone by machining 15 grooves with rectangular cross-section of
1-mm depth and 0.5-mm width. Nickel wick of 3-mm thickness
with 3–5-
m mean pore radius and 75% porous volume pro-
vided sufficient capillary pressure for continuous circulation of
the working fluid around the loop during operation at different
heat loads. Apart from providing capillary pumping, the wick
structure also functions as a thermal and hydraulic lock to in-
hibit back conduction of heat and prevent any back flow of vapor
from the evaporation zone to the compensation chamber. The
vapor lock is provided by the presence of the liquid in the fine
pores of the capillary structure while heat leaks to the compensa-
tion chamber can be decrease by using low conductive capillary
structure. Within the framework of the proposed design, nickel
serves as an optimum choice to achieve these functionalities due
to its low thermal conductivity and capability to be sintered in
smaller pore sizes with relatively high porosity. The body and
the transport lines of the mLHP were made of copper.
The condenser of the mLHP was provided with external fins
[Fig. 1(e)] measuring 20
10 mm with a thickness of 0.2 mm.
Table I gives the main design parameters of the mLHP. Water
was used as the heat transfer fluid that ensured excellent heat
transfer characteristics in the temperature range between 50 to
100
C. The charging volume of the water was decided on the
basis of the volume of the liquid line, compensation chamber
and porous volume of the wick. In the cold state, at least 50% of
the compensation chamber should be filled with liquid for the
proper wetting of the wick and reliable startup of the loop.
Authorized licensed use limited to: RMIT University. Downloaded on November 23, 2008 at 21:47 from IEEE Xplore. Restrictions apply.
SINGH et al.: MINIATURE LHP 45
III. TESTING METHOD
The thermal performance of the mLHP was tested by using a
heater of 3.75 cm
area. In this case, the active area of the evap-
orator, i.e. the surface where capillary structure makes contact
with the evaporator wall and there are vapor removal channels,
is more than the active thermal footprint of the heater face. As
a result, the current testing provides a condition of nonuniform
heating to the evaporator active face. Here, nonuniform heating
was done on approximately 3/5 of the active zone area. The heat
load simulator [Fig. 1(b))] was in the form of copper block em-
bedded with two cylindrical cartridge heaters. During testing,
the heater block was attached symmetrically to the center of the
circular heat absorbing face of the mLHP evaporator [Fig. 1(d)
and (g)].
The condenser cooling was accomplished by forced convec-
tion provided by an air cooling fan using ambient air with an
inlet temperature of 24
2 C. The temperature was measured
at different points on the mLHP using K-Type thermocouples
with an accuracy of
0.1 C. Fig. 1(c), (d), and (f) shows the
experimental set up for testing the mLHP along with the loca-
tion of the thermocouples. Data from these thermocouples was
acquired every 10 s by a “Keyence Thermo Pro 3000” based
data acquisition system. The thermal performance of the mLHP
was measured on the basis of the evaporator temperature, max-
imum heat capacity, evaporator thermal resistance, mLHP/heat
pipe thermal resistance and total thermal resistance of the de-
vice. The following relations (1–3) were used to calculate the
thermal resistances.
Evaporator thermal resistance
(1)
Heat Pipe thermal resistance
(2)
Total thermal resistance
(3)
Calculation for the overall heat transfer coefficient of the
evaporator was made by using
(4)
In the above equations,
is the external temperature of
the evaporator active zone which was measured by averaging
the temperatures of the thermocouples fixed on the evaporator
heating face (
1, 2, 3, 4). The vapor temperature
was taken to be equal to the temperature at the evaporator
outlet as indicated in the Fig. 1(d). Temperature of the external
condenser surface
was calculated by taking mean of the
readouts from the thermocouples installed at the condenser
inlet
, condenser finned surface and
condenser outlet
. For measuring the temperature at
the junction
of the mLHP evaporator and heat simulator, a
special groove [Fig. 1(c)] was made at the center of the heating
face of simulator block in which the thermocouple point was
fixed using thermally conductive epoxy resin.
A digital wattmeter with an error of
0.1 W was used to mea-
sure and control the input heat load to the heat simulator. During
the experiment the input power to the heat simulator was in-
creased in steps of 5 W. The error in determining the thermal
resistance—
is formed from an estimation of the measurement
errors of the input heat load
and temperature measurements.
It should be noted that the error of estimation will be larger at the
low heat loads due to the higher values of the associated thermal
resistances at such loads. Uncertainties in the reported thermal
resistances were carried out over the entire range of applied heat
load in the experiment and lie between 1.26% to 6.23%. Simi-
larly in the determination of the evaporator heat transfer coef-
ficient—
, (from temperatures and heat load measurements)
uncertainties were estimated to be within 3.50% to 9.04%, re-
spectively. Heat losses due to the natural convection and radi-
ation heat transfer from the outer surface of the heat load sim-
ulator and the mLHP evaporator was found to be the functions
of the temperature difference of the hot surface and the ambient
air. In the present test, the heat losses were estimated [17] to be
about 6% to 8% of the input power.
Testing of the mLHP prototype was done in the horizontal
configuration with the evaporator and condenser at the same
level. Startup of the mLHP was assumed to occur with the rise in
temperature of the entire vapor line and consequently clearing
of liquid from the vapor line. For a successful startup at a given
heat load the temperature difference between the outlet of evap-
orator and inlet of condenser should be less than or equal to
1
C. At a given heat load, steady state was characterized by the
constant evaporator temperature.
IV. T
EST RESULT AND
DISCUSSION
As heat load is applied to the evaporator active zone, the tem-
perature of the evaporator rises and results in the vaporization
of the working fluid. The resulting vapor pushes the liquid from
the grooves and vapor line. Evaporating meniscus is formed
at the wick-wall interface of the capillary structure inside the
evaporation zone which is responsible for developing the cap-
illary pressure to circulate the working fluid around the loop.
This is followed by a rise in temperature of evaporator outlet
and then condenser inlet which registers the startup of the LHP.
The excess liquid cleared from the evaporator grooves and vapor
line by the vapor is displaced to the compensation chamber. In
Fig. 1(d), (f), and (g) the shaded parts of the mLHP shows the
portion occupied by the liquid while the blank portion is filled
with vapor.
Fig. 2(a) shows the startup process with 20-W input power.
The device showed reliable startup under low as well as high
heat loads and achieved steady state under every step (5 W)
increase in input load. Fig. 2(b) shows startup of the device from
the cold state under 50-W heat load. The capillary evaporator
does not show any symptoms of wick depriming like overshoot
of evaporator temperature or back vapor flow from evaporator
to compensation chamber for high heat loads or abrupt power
down conditions.
Fig. 3 gives the heat load dependence of the evaporator sur-
face temperature. The plot clearly shows that the LHP is able to
Authorized licensed use limited to: RMIT University. Downloaded on November 23, 2008 at 21:47 from IEEE Xplore. Restrictions apply.