Applied Thermal Engineering 29 (2009) 35113517

Contents lists available at ScienceDirect

Applied Thermal Engineering

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

Compact cooler for electronics on the basis of a pulsating heat pipe

Yury F. Maydanik *, Valery I. Dmitrin, Vladimir G. Pastukhov
Institute of Thermal Physics, Ural Branch of the Russian Academy of Sciences, Amundsen St. 106, Ekaterinburg 620016, Russia

a r t i c l e

i n f o

Article history:
Received 14 April 2009
Accepted 3 June 2009
Available online 7 June 2009
Keywords:
Pulsating heat pipe
Electronics cooling
Heat load
Heat ux
Thermal resistance

a b s t r a c t
The paper presents the results of developing and investigating a compact cooler for electronics made on
the basis of a closed loop pulsating heat pipe (CLPHP). The cooler is made of a copper tube 5.6 m long with
OD of 2 mm and ID of 1.2 mm in the form a 3D spiral containing 17 turns. The device is equipped with a
light copper radiator with a nning area of 1670 cm2, which was blown by an axial fan located inside the
spiral. The thermal interface of the cooler situated in the heating zone is made of a copper plate with a
thermocontact surface measuring 40  35 mm, which was in thermal contact with all the turns of the
device. The cooler overall dimensions are 105  100  60 mm, its mass is 350 g.
The operation of the cooler has been investigated with water, methanol and R141b as working uids at
a uniform and concentrated supply of a heat load in different heating modes. A reliable operation of the
device has been demonstrated in the range of heat loads from 5 to 250 W. A minimum thermal resistance
heat sourceambient air equal to 0.32 C/W was attained with water and methanol as working uids at
a uniform heat load of 250 W. With a heat load concentrated on a section of the thermal interface limited
by an area of 1 cm2, a minimum value of thermal resistance equal to 0.62 C/W was attained at a heat
load of 125 W when methanol was used as a working uid.
2009 Elsevier Ltd. All rights reserved.

1. Introduction
Pulsating heat pipes (PHPs) [13] have taken up a prominent
place among highly efcient heat-transfer devices operating on a
closed evaporationcondensation cycle. In such characteristics as
heat-transfer capacity, thermal resistance, sensitivity to orientation in the gravity eld PHPs in some cases can approach conventional heat pipes, and the extreme simplicity of their design makes
them comparable to two-phase thermosyphons. However, PHPs
will evidently never be able to attain as high thermal efciency,
with comparable dimensions, as that of heat-transfer devices
whose operation is fully based on the use of the latent heat of
vaporization. In PHPs, where heat-transfer is realized at the expense of the chaotic motion of alternating vapor slugs and liquid
plugs, the latent heat of vaporization is used only partially. In this
sense such a brand as heat pipe refers to these devices rather
conventionally.
By their design PHPs may be divided into two main types: open
loop pulsating heat pipes (OLPHPs) and closed loop pulsating heat
pipes (CLPHPs) [4]. On the basis of well-known publications it is
difcult to draw an unambiguous conclusion as to which of these
two types of PHP is preferable. For instance, in Refs. [4,5] it is noted
that there is no considerable difference between these devices. At
the same time in Ref. [6] it is pointed out that OLPHPs with such
working uids as water, ethanol, FC-72 do not operate at all. Evi* Corresponding author. Tel.: +7 343 267 87 91; fax: +7 343 267 87 99.
E-mail address: maidanik@etel.ru (Y.F. Maydanik).
1359-4311/$ - see front matter 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.applthermaleng.2009.06.005

dently the serviceability, as well as the efciency of one or another

type of PHP, depends on many factors, which for a correct comparison should be sufciently comparable. The experience of the
authors themselves shows that the difference between the functional potentialities of an OLPHP and a CLPHP is mainly connected
with the minimum value of the heat load required for a start-up.
For the latter it may be considerably lower. In any case, the structure of any kinds of PHP should correspond to the following principles formulated in the above-mentioned Ref. [4]:
(a) the devices must be made of a tube of capillary diameter,
have the shape of a serpentine with a sufciently large number of turns and be partially lled with a working uid;
(b) they may not contain any capillary structure inside;
(c) they must have zones of heating and cooling, which, if necessary, can be separated by an adiabatic zone.
Usually for evaluating the maximum internal (critical) diameter
of a capillary tube one uses formula [7]:

dcr < 2

q
r=ql g ;

or its specied variant [4]:

dcr  2

q
r=ql  qv g :

As a rule, the internal diameter of PHPs varies in the range from 0.5
to 3 mm and the number of turns is usually equal to 10 or more. The

3512

Y.F. Maydanik et al. / Applied Thermal Engineering 29 (2009) 35113517

Nomenclature
CFM
dcr
FH
g
N
Q
qmax

cubic foot per minute

critical diameter of a capillary, m
area of the heated surface, m2
gravity constant, m/c2
number of turns
heat load, W
max. heat ux, W/m2

PHP operation also largely depends on the degree of lling with a

working uid. As mentioned in Ref. [8], with a relative volume of
lling of 20% and less each of the turns of the device functions as
an independent two-phase thermosyphons, and with a lling of
more than 70% the conditions of the device operation in the pulsating regime become difcult. Filling ratios in the range from 50% to
70% are thought to be optimum. In this case an efcient operation of
PHPs is possible at different orientations when the heat load is supplied in the bottom (bottom heating mode) or in the top (top heating mode), and also in the horizontal position (horizontal heating
mode).
The simplicity of the design, in combination with sufciently
high heat-transfer characteristics and the possibility of various design embodiments, makes pulsating heat pipes quite an attractive
object for actual application. One of the rst publications in this
direction was connected with the development and investigations
of the so-called L-KENZAN FINs and VL-KENZAN FINs devices
created on the basis of PHP technology and intended for cooling
powerful electronic devices [9]. This paper, in particular, discusses
a cooler created on the basis of a PHP made of a copper tube with
OD = 1.6 mm, ID = 1.2 mm which has 500 turns. On one side the
cooler is equipped with a thermal interface for supplying a heat
load made in the form of a copper plate measuring 80  80
 2 mm, and on the other with an axial fan of appropriate size.
The overall dimensions of the cooler are approximately
105  80  80 mm. At a heat load of 450 W and a blowing rate of
3 m/s the temperature at the surface of the thermal interface was
about 80 C, and its thermal resistance was equal to 0.089 K/W.
In this case the heat load density at the surface of the thermal
interface reached 7 W/cm2, and the heat load per PHP turn was
equal to 0.9 W.
A similar technology was also used for cooling IGBT modules,
which are capable of dissipating heat ows up to several kilowatts
[10]. For instance, a cooler on the basis of a copper PHP with
OD = 2.0 mm, ID = 1.6 mm and 424 turns was developed for these
purposes. The dimensions of the cooler section to be cooled
were 200  150 mm, and those of the thermal interface 110 
190 mm. At a heat load of 1800 W and a blowing rate of 3 m/s
the thermal resistance of the device was equal to 0.022 K/W. Such
low values of thermal resistance of these coolers are caused by the
large dimensions of thermal interfaces and the well-developed
cooled surface, which is formed by a great number of sufciently
long PHP turns.
Quite interesting solutions based on the use of PHPs have been
suggested for cooling CPUs of notebooks, where the problem of
limitedness of the inner space is critical. Such a system manufactured on the basis of T-shaped and O-shaped at PHPs with
R141b as a working uid made of copper capillaries with an internal diameter of 2 mm is discussed in Ref. [11]. The PHPs had from 3
to 6 turns and were equipped with a copper thermal interface measuring 40  40 mm, which was in contact with the CPU. The length
of the device varied from 170 to 260 mm. The lling ratio was
4080%. Tests were conducted in the horizontal heating mode.

RPM
R
TH
Tamb

r
ql
qv

rotation per minute

thermal resistance, C/W
temperature of the thermocontact surface of the heater, C
ambient air temperature, C
surface tension on a liquidvapor boundary, N/m
liquid density, kg/m3
vapor density, kg/m3

The heat-transfer part of the PHP was cooled by means of blowing

at a rate of 2 m/s. The best results were demonstrated by an
O-shaped PHP with six turns and 60% of a working uid. At a
nominal heat load of 50 W the thermal resistance of the system
CPUambient air was equal to 1.1 C/W. The CPU temperature
in this case was maintained at a level of 80 C.
Ref. [12] presents the results of testing at exible straighttype and wing-type PHPs developed for the same purpose. The
PHPs were also made of a copper capillary, but the working uid
was R134a with a lling ratio of 70%. The dimensions of the heating
and the cooling zone were, respectively, 110  50 and 110 
150 mm. Devices with capillary internal diameters of 1.0, 1.2 and
1.7 mm with 8 and 12 turns were put to a test. On trials the heating
section was set horizontally and the cooling section was tilted 60
towards the outer side. The temperature of the cooling section was
maintained at about 50 C with a fan. The highest results were
shown by a PHP of the straight type with a capillary diameter
of 1 mm and 12 turns. At a maximum heat load of 100 W it had
a thermal resistance of about 0.28 K/W, maintaining the temperature of the CPU thermal simulator at a level of about 75 C.
The aim of the present work consisted in creating a simple, light
and compact cooler for electronics capable of operating efciently
in the temperature range from 50 to 100 C at different heating
modes and investigating such a device based on a copper closed
loop pulsating heat pipe with different working uids, which was
tested both at a uniform and at a concentrated heat load.
2. Description of the experimental device
Two identical coolers based on OLPHP and CLPHP were made at
the stage of development. Preliminary tests showed that the devices had closely analogous thermal characteristics, but the

Fig. 1. Scheme of 3D PHP.

Y.F. Maydanik et al. / Applied Thermal Engineering 29 (2009) 35113517

3513

PHP

thermal
interface
heat sink

fan
Fig. 2. External view of a cooler on the basis of CLPHP.

CLPHP-based cooler started up at lower heat loads. Therefore, all

further tests were conducted only with this device.
The PHP was made of a copper tube 5.6 m long with OD = 2 mm
and ID = 1.2 mm in the form of a 3D spiral asymmetrical with respect to the longitudinal axis, which has 17 turns. The scheme of
the PHP is given in Fig. 1.
On one of the spiral lateral sides the turns were located close to
one another and soldered to a copper thermal interface a copper
plate measuring 40  35  0.5 mm. On the opposite side of the
spiral the turns spreading in a semicircle from the thermal interface were soldered to a light copper radiator, whose nning area
was 1670 cm2. Inside the spiral between the radiator and the thermal interface there was an axial fan with an impeller diameter of
92 mm, which ensured an air-ow rate of 31.8 CFM at 1800 RPM.
The cooler overall dimensions were 105  100  60 mm, and the
mass, including the fan, was equal to 350 g. An external view of
the cooler is shown in Fig. 2.
Water, methanol and R141b, which have a good compatibility
with copper, were chosen as working uids. Besides, such a choice
made it possible to study the cooler operation with working uids
whose thermal properties were widely different. By means of
experiments it has been found that the optimum lling ratio for
methanol and R141b is 65%, and for water 60%. Such an amount
of the working uid for the cooler under discussion ensured the
minimum thermal resistance at the minimum dependence on the
heating mode. The internal diameter of the capillary tube of the
PHP for the chosen working uids corresponded to the criteria
(1) and (2) in the whole range of operating temperatures at which
the cooler was tested.
3. Testing procedure
The cooler was tested in the top heating mode, the bottom heating mode and the side heating mode. The last may be regarded as a
certain equivalent to the horizontal heating mode, which in this
case was impossible to realize because of the 3D shape of the cooler. The positions of the cooler at different heating modes are presented schematically in Fig. 3.
Two different heat sources were used for tests. The rst of
them was at in shape. The dimensions of its thermocontact surface ensured a uniform supply of a heat load to the whole surface of the cooler thermal interface, the area of which was
14 cm2. The other heater made it possible to concentrate a heat
load on a limited section of the thermal interface by the method
of a thermal wedge. It was made of copper in the form of a

Fig. 3. Cooler positions at different heating modes: a top heating mode, b side
heating mode, c bottom heating mode.

cylinder which consisted of two parts with different diameters

and lengths. An ohmic heater of nichrome wire was located on
the lateral surface of the wider part of the cylinder. The other,
shorter part, with a relatively small diameter ensured the concentration of a heat load on a thermocontact surface with an
area of 1 cm2. This part also served for measuring the temperature at the thermocontact surface. For this purpose use was
made of three radial channels situated at different distances
along its length, in which thermocouples were set. The heat load
was measured by a wattmeter. Maximum systematic uncertainties in the course of heat load measurements did not exceed
2.5%. On trials its value was changed stepwise with the help
of a laboratory autotransformer, beginning with 5 W and up
to the maximum value, at which the temperature at the
thermocontact surface of the heat source reached 100 5 C.
Temperature measurements were conducted with conventional
copperconstantan thermocouples OMEGA of TT-T-30 type
with an electrode diameter of 0.254 mm, which were used as a
unit with a personal computer and a data acquisition unit Agilent 34970A. The temperature of the at heater used for a uniform supply of a heat load was measured by a thermocouple
positioned just at its thermocontact surface. The temperature
of the thermocontact surface of the thermal wedge used for
the concentration of a heat load was determined by linear
extrapolation of indications of the three thermocouples located
in its measuring part. Uncertainties in determining temperature
average values amounted to 1 C. All tests were conducted at
an ambient temperature of 23 1 C. The value of the total thermal resistance of the system heat sourcesurrounding air was
calculated by the formula:

T H  T air
:
Q

Calculations were also made of the maximum value of the heat ux

corresponding to the maximum heat load:

qmax

Q max
:
FH

The maximum value of heat load per PHP turn was determined as:

qmax

Q max
:
N

3514

Y.F. Maydanik et al. / Applied Thermal Engineering 29 (2009) 35113517

minimum to the maximum one for the side heating mode. For
other heating modes these proles had a similar form.
In analyzing these results one, rst of all, pays attention to the
fact that the characteristics of temperature pulsations of the cooler
differ considerably depending on the working uid used. In particular, for water the amplitude of pulsations is much higher, and the
frequency lower than for methanol and R141b. It can also be seen
that with increasing heat load, when the temperature and the corresponding pressure in the PHP increase, the amplitude of pulsations decreases and the frequency increases. This is connected
with the fact that during the boiling of liquids at low pressures

4. Results of tests at a uniform heat load

The most important characteristic of the cooler efciency is the
temperature of the cooled object (heater) which it can ensure at a
given heat load and under certain external conditions. In the present case, as such conditions one may consider the ambient air temperature and the rate of blowing of the radiator, which should be
constant, and also the cooler orientation, which changed 90 in
accordance with the heating modes.
Fig. 4 presents typical temperature proles of the heater for different working uids obtained under changes of heat load from the

250 W

110

225 W

100

200 W
175 W
150 W

Temperature, oC

90
80

125 W
100 W

70

50 W

75 W

60
50

25 W

10 W

40
30
20
0

1000

2000

3000

4000

5000

120

250 W

110

225 W

100

200 W

90

Temperature, oC

6000

175 W

80

150 W
125 W

70

100 W

60

75 W
50 W
25 W

50
40

5W

10 W

30
20
0

1000

2000

3000

4000

5000
190 W

110
180 W

100

160 W

Temperature, oC

90

150 W
125 W

80
70

100 W
25 W

60

75 W

20 W
15 W

50

50 W

10 W

40

5W

30
20
0

1000

2000

3000

4000

5000

6000

Time, sec
Fig. 4. Temperature proles of the heater in the side heating mode: a water, b methanol, c R141b.

Y.F. Maydanik et al. / Applied Thermal Engineering 29 (2009) 35113517

the minimum size of vapor bubbles will grow considerably, and

the frequency of their origination will decrease [13]. In comparing
water and R141b in this respect one can see that in approximately the same temperature range the corresponding pressures
of saturated vapor vary from 6 to 140 kPa for water and from
80 to 670 kPa for R141b. For methanol the range of pressure variation contains intermediate values. The amplitude of temperature pulsations changes in accordance with the sizes of bubbles
that form in the heating and collapse in the cooling zone. As for
the pulsation frequency, it is evidently related to the latent heat
of vaporization of the liquids used as working uids. Despite
the fact that it participates in the heat-transfer process only partially, at the same heat load the rate of formation of vapor bubbles and the circulation rate must be higher in the working
uid with a lower latent heat of vaporization. Another signicant
difference in the operation of PHPs with these working uids is
that the minimum value of heat load at which stable pulsations
begin is 25 W for water, and 5 W for R141b. For methanol it varies from 10 to 15 W depending on the heating mode. This may be

110

Temperature, oC

90

70

50

30

50

100

150

200

250

b 120

3515

caused by the fact that the operation and, consequently, the energy required for the formation of a viable vapor bubble depend
on the value of surface tension, which is higher for water than
for methanol and R141b [14]. Fig. 5 presents experimental values
of the average heater temperature depending on the heat load at
different heating modes.
These results show that despite the differences in the character
of pulsations the value of heat load at which a maximum temperature close to 100 C is achieved varies only slightly. For water and
methanol it is equal to 250 W, and for R141b to 190 W. The difference does not exceed 24%. Accordingly, differences are also observed in the minimum values of thermal resistance, which are
0.32 0.2 C/W for water and methanol and 0.42 0.2 C/W for
R141b.
The maximum heat load density attained in testing the cooler
under uniform heating of the thermal interface was 17.8 W/cm2
for water and methanol and 13.6 W/cm2 for R141b. As for the
heat load per turn, these values were equal respectively to
14.7 W/cm2 and 11.2 W/cm2. It is necessary to mention another
interesting effect, which was connected with the appearance of
a characteristic sound accompanying the PHP operation when
the working uid was water. That sound appeared at a heat load
of 75 W, and its frequency and intensity increased with increasing heat load. The sound effects were also accompanied by quite
noticeable mechanical vibrations of the cooler. From the laws of
acoustics it is known that any process that causes a local change
of pressure or a mechanical stress in an elastic medium may
serve as a sound source. Different hypotheses of the appearance
of noise during the boiling of a liquid are considered in Ref. [15].
One of them, in particular, points to the fact that the noise
source during boiling is pulsations of the volume of vapor bubbles causing a local compression and decompression of the surrounding liquid. It is not improbable that it is precisely such a
process that is observed in a PHP, which is aggravated by constraint conditions. When the device operated with methanol
and R141b, no sound effects were observed in the audible frequency range.

Temperature, oC

100

5. Results of tests at a concentrated heat load

80
60
40
20

0
0

Temperature, oC

50

100

150

200

250

100

80

60

top heating mode

bottom heating mode
side heating mode

40

20
0

50

100

150

200

Heat load, W
Fig. 5. Heat load dependence of the average heater temperature: a water, b
methanol, c R141b.

For an efcient operation of a PHP in the pulsating mode, which

is normal for it, this PHP should, as has been mentioned above, possess a sufciently large number of turns, in each of which there
should be a generation of vapor bubbles alternating with liquid
plugs. Therefore heat must be supplied to every turn. For this purpose use is usually made of a thermal interface, i.e. a plate of a heat
conducting material, which distributes the heat obtained from the
heater among all the turns. It is evident that the thermal interface
should have corresponding dimensions. The problem that arises
here lies in the fact that these dimensions sometimes considerably
exceed those of the thermocontact surface of the object being
cooled. In this case one can speak about a concentrated supply of
heat load. A heat ow is transferred from the heat source to the remote turns only at the expense of thermal conductivity along the
thermal interface. If this heat ux is not sufcient, there may be
problems with the PHP start-up. It is such a case that took place
in testing a cooler with water in the top heating mode, when all attempts to start it up failed even at a heat load of 75 W despite the
fact that the heater temperature increased up to 115 C. Sporadic
pulsations observed at a heat load of 50 W subsided rapidly. At
the same time in the side and bottom heating modes a start-up
was successfully realized at 25 W, as well as during uniform heating. When the working uids were methanol and R141b, the PHP
began to operate in the pulsating regime at a heat load of 5 W in
all heating modes. Fig. 6 presents temperature proles of the PHP

3516

Y.F. Maydanik et al. / Applied Thermal Engineering 29 (2009) 35113517

110

100 W

90 W
85 W

100
75 W

Temperature, oC

90
80
50 W

70
60
25 W

50
10 W

40
30
20
0

1000

2000

3000

4000

5000

6000

7000

8000

110
125 W

100

Temperature, oC

90

100 W

80
75 W

70
50 W

60

25 W
20 W

50
15 W

10 W

40

5W

30
20
0

1000

2000

3000

4000

5000

6000

110
100

100 W

Temperature, oC

90
80

75 W

70
50 W

60
25 W
20 W

50
10 W

40

15 W

5W

30
20
0

1000

2000

3000

4000

5000

6000

Time, sec
Fig. 6. PHP temperature prole at concentrated heat load: a water, b methanol, c R141b.

with different working uids in the side heating mode, which have
similar form for other heating modes too.
Here it is seen that despite the local heat load supply temperature pulsations are observed in all PHP turns. In this case a temperature close to 100 C is achieved at a heat load of 125 W when the
working uid is methanol, which corresponds to a heat ux of
125 W/cm2. The value of the cooler thermal resistance in this case
is at a level of 0.62 C/W.
The heat load dependence of the heater average temperature at
concentrated heating, which is shown in Fig. 7, has a form similar
to that observed at uniform heating.
This dependence is of a linear character, and the effect of the
heating mode here is also insignicant.

6. Conclusions
1. A compact cooler made on the basis of a CLPHP has been tested
at uniform and concentrated heat loads with water, methanol
and R141b as working uids.
2. An efcient operation of the cooler has been demonstrated in
the top, side and bottom modes. At the top heating mode at a
concentrated heat load it was impossible to start-up the cooler
with water.
3. A maximum average temperature of the heater equal to
100 5 C was attained at a uniform heat load of 250 W when
water and methanol were used as working uids. For R141b
this value was 190 W. The corresponding values of heat ux

Y.F. Maydanik et al. / Applied Thermal Engineering 29 (2009) 35113517

Temperature, oC

3517

were 17.8 W/cm2 and 13.6 W/cm2, and those of thermal resistance 0.32 C/W and 0.42 C/W. In this case no critical phenomena were observed in the heating zone.
4. With concentration of a heat load on a thermal interface area of
1 cm2 a maximum heat ux value equal to 125 W/cm2 was
achieved with methanol as a working uid.
5. The results of tests have shown that for the cooler under discussion the most suitable working uid is methanol, which ensures
its serviceability in the widest range of heat loads in all heating
modes at both uniform and concentrated heating.

110

90

70

50

30
0

Temperature, oC

20

40

60

80

100

120

References
100
80
60
40
20
0

Temperature, oC

20

40

60

80

100

120

120
100
80
60
40

top heating mode

bottom heating mode
side heating mode

20
0
0

20

40

60

80

100

120

Heat load, W
Fig. 7. Heat load dependence of the cooler average temperature at concentrated
heating: a water, b methanol, c R141b.

[1] H. Akachi, Structure of a heat pipe, US Patent # 4991041, 1990.

[2] H. Akachi, Structure of micro-heat pipe, US Patent # 5219020, 1993.
[3] G. Smirnov, Method of action of the pulsating heat pipe, its construction and
the device on its base, US Patent Pub. No. US 2003/0037910 Al, 2003.
[4] S. Khandekar et al., Closed and open pulsating heat pipes, in: Proceedings of
the 13th Int. Heat Pipe Conf., Shanghai, 2004, pp. 3648.
[5] H. Yang, S. Khandekar, M. Groll, Performance comparison of open loop and
closed loop pulsating heat pipes, in: Proceedings of the 8th Int. Heat Pipe
Symp., Kumamoto, 2006, pp. 119124.
[6] X.M. Zhang, J.L. Xu, Z.Q. Zhou, Experimental study of a pulsating heat pipe
using FC-72, ethanol and water as working uids, Experimental Heat Transfer
17 (2004) 4767.
[7] S. Maezawa et al., Thermal performance of capillary tube thermosyphon, in:
Proceedings of the 9th IHPC, Albuquerque, 1995, pp. 791795.
[8] H. Yang, S. Khandekar, M. Groll, Operational characteristics of at plate closed
loop pulsating heat pipes, in: Proceedings of the 13th IHPC, Shanghai, 2004, pp.
291297.
[9] H. Akachi, F. Polasek, P. Stuls, Pulsating heat pipes, in: Proceedings of the 5th
IHPS, Melbourne, 1996, pp. 208217.
[10] H. Akachi, F. Polasek, Thermal control of IGBT modules in traction
drivers by pulsating heat pipes, in: Prep. of the 10th IHPC, Stuttgart,
1997 (paper E-1).
[11] K. Gi et al., CPU cooling of notebook PC by oscillating heat pipe, in: Proceedings
of the 11th IHPC, Tokyo, 1999, pp. 469472.
[12] Y. Miyazaki, Flexible oscillating heat pipes for cooling of notebook PCs, in:
Proceedings of the 8th IHPS, Kumamoto, 2006, pp. 226230.
[13] S.S. Kutateladze, Fundamentals of Heat Transfer, Academic Press Inc., New
York, 1963.
[14] V.P. Scripov, Metastable Liquids, John Wiley & Sons, New York, Toronto, 1974.
[15] E. Nesis, Liquid Boiling, Nauka, Moscow, 1973 (in Russian).