Applied Thermal Engineering 113 (2017) 1219–1227

Contents lists available at ScienceDirect

Applied Thermal Engineering

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

Research Paper

Study on thermal cycle in oscillating heat pipes by numerical analysis

Takurou Daimaru ⇑, Shuhei Yoshida, Hiroki Nagai
Department of Aerospace Engineering, Tohoku University, Sendai 980-8579, Japan

h i g h l i g h t s

We developed a one-dimensional model to reproduce thermohydrodynamic in OHPs.

The numerical results revealed pressure propagation within the OHP.

The propagation of energy was identified as a reason for pressure propagation.

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

Article history: This paper discusses the thermal cycle found within oscillating heat pipes (OHPs). An OHP is a two-phase
Received 28 July 2016 heat transfer device using self-exited oscillation. Over the past few decades, a considerable number of
Revised 28 October 2016 studies have been conducted to understand the physics of OHP phenomena. However, little is known
Accepted 16 November 2016
about the thermal cycle in OHPs. In this study, we developed a one-dimensional slug flow model to repro-
Available online 17 November 2016
duce thermal and hydrodynamic phenomena in OHPs. Fast Fourier transform (FFT) and cross-correlation
analysis were used to process oscillation waveform data. A multi-branch OHP consisting of a stainless
Keywords:
steel pipe wall and R134a working fluid was simulated. The numerical results revealed pressure propa-
Numerical analysis
Oscillating heat pipes
gation within the OHP. Moreover, the results indicated that the vapor volume oscillated with the same
Propagation phenomena frequency as the pressure. Additionally, the vapor plug obtained energy or performed work depending
Thermal cycle on the direction of pressure propagation. As a result, the propagation of energy was identified as a reason
for pressure propagation within the OHP.
Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction tional principles of OHPs are not well understood because the
internal thermal and hydrodynamic phenomena are complex [9].
Oscillating Heat Pipes (OHP) or pulsating heat pipes (PHP) were Self-excited oscillation as a primary operational principle of OHPs
invented in 1990 by Akachi in Japan [1]. An OHP consists of bent is a phenomenon in which characteristic oscillation is generated
metal mini channels and saturated working fluid [2]. An OHP has by the system even if the inputs to the system are linear. One rea-
heating, adiabatic, and cooling sections along the direction of heat son that self-exited oscillation in an OHP is more complex than a
transport. Under nominal operating conditions, the working fluid general case is because there multiple turns within the heating
separates into liquid slugs and vapor plugs, ultimately forming a section as opposed to a single turn in the general case. Thus, the
slug flow state because of the strong capillary force in the mini problem is not one of simple expansion or compression of a single
channel [3]. Upon application of heat to the heating section, liquid vapor plug. Oscillations in each channel affect each other and thus
slugs commence self-excited oscillation from the pressure differ- are interrelated (see Fig. 1).
ence along the flow path [4]. OHPs transport heat by forced con- In previous studies, visualization experiments and numerical
vection of liquid slugs and the phase change phenomena in analyses were conducted using a straight tube [10–13]. Nagasaki
liquid films surrounding vapor plugs [5,6]. The self-exited oscilla- et al. studied the oscillation mechanism of a liquid column in the
tion is triggered by initial disturbances including vapor-liquid dis- closed ends single-branch OHP [14]. Fig. 2 shows Nagasaki’s exper-
tribution, boiling, merging of liquid slugs, etc. [7,8]. OHPs are much imental apparatus. They investigated the relationship between the
thinner and lighter than conventional heat pipes because they do position of liquid column x and the pressure of vapor P. As shown
not have internal wicking structures. However, the basic opera- in Fig. 3(a), the vapor pressure showed a periodic oscillation wave-
form and its amplitude was larger than the case assuming adia-
batic conditions Padia. In addition, the position of the liquid
⇑ Corresponding author.
column oscillated in the coordinate phase with the vapor pressure.
E-mail address: [email protected] (T. Daimaru).

http://dx.doi.org/10.1016/j.applthermaleng.2016.11.114
1359-4311/Ó 2016 Elsevier Ltd. All rights reserved.
1220 T. Daimaru et al. / Applied Thermal Engineering 113 (2017) 1219–1227

Nomenclature

A area (m2) u velocity (m/s)

h heat transfer coefficient (W/m2/K) v vapor plug (–)
k thermal conductivity (W/m/K) V volume (m3)
l liquid slug (–) z position (m)
L length (m) w pipe wall (–)
m mass (kg)
P pressure (Pa)
T temperature (K)

8
Thus, the volume of the vapor was anti-phased relative to the >
< qh : heating section
vapor pressure. The thermal cycle, which is a focus in our study, qex;k ¼ 0 : adiabatic section ð2Þ
was not mentioned in theirs. However, the cycle should form a >
:
hc ðT c  T w;k Þ : cooling section
straight shape on a DP-DV plot, as shown in Fig. 3(b). This straight
cycle shape means that the energy attained in the vapor via com-
pression was equal to the work performed by the vapor during 2.2. Liquid slug temperature and pressure
expansion. Previous studies helped advance the understanding of
this operational principle. However, this theory has not yet been The energy equations for liquid slugs are also one-dimensional
extended to multi-branch OHPs. heat equations that include heat transfer with the pipe wall. The
This paper presents a numerical slug flow model of a multi- equations for both ends of a mesh are solved using the saturation
branch OHP and describes data processing methods for oscillation temperature of adjacent vapor plugs as boundary conditions. The
waveforms of physical parameters. Additionally, OHP pressure number of sub-volumes in a mesh is determined according to the
propagation and thermal cycles are examined. length of the liquid slugs at the first-time step and is maintained
during calculation. The temperature change of each sub-volume
2. Numerical modeling numbered i is calculated using Eq. (3):
" #
@T l @2T l
A schematic diagram of the numerical model is shown in Fig. 4. cp;l ql Acr ¼ qwf p þ kl Acr 2 ð3Þ
@t @z
It is a one-dimensional model along the flow path and has domains n;i

of the pipe wall and working fluid. As shown in Fig. 5, the working
fluid domain consists of liquid slugs, vapor plugs, and liquid films. qwf ;k ¼ hl;n ðT w;k  T l;n;i Þ ð4Þ
The pipe wall and each liquid slug have meshes for temperature
The value of the heat flux between the pipe wall and the liquid
calculation. The mesh of the pipe wall is fixed spatially, whereas
slug, which is expressed in Eq. (4), is calculated based on the forced
the meshes of the liquids can move in the working fluid domain
convection heat transfer in the circular pipe. The heat transfer coef-
[15]. The primary assumptions are as shown below:
ficient calculated in different equations depends on the flow
regime of the laminar, transient, and turbulent flows. A thermally
(a) The inner flow regime is slug flow, which is uniform along
developing flow is assumed for the laminar flow, but developed
the radial direction. The liquid phase is treated as an incom-
steady flow correlations are employed for transient and turbulent
pressible flow.
flows. The inner flow regime is determined by the Reynolds num-
(b) Vapor plugs are treated as ideal gases. Temperature and
ber. The averaged thermal properties along the length of the liquid
pressure are uniform within each vapor plug.
slug are used to calculate these dimensionless numbers and heat
(c) Liquid films maintain constant thickness. Friction between
transfer coefficient.
vapor plugs and film, and between the pipe wall and liquid  
film, are not considered. q ul Din
Rel ¼ l ð5Þ
(d) Mass and volume changes in the liquid slugs as a result of ll n
the generation and absorption of liquid films and via phase
change in interfaces with vapors are not considered.
(e) The front and rear meniscus of liquid slugs are treated as flat
cross sections in terms of heat transfer.
(f) Heat flux in the heating section is considered to be spatially
uniform and temporally constant.

2.1. Pipe wall temperature

The energy equation for the wall is a one-dimensional heat

equation including heat transfer with the external heat, and work-
ing fluid. The entire tube wall is subdivided into Nw sub-volumes,
and the temperature of each sub-volume, numbered k, is calcu-
lated using Eqs. (1) and (2):
" #
@T w @2Tw
cp;w qw Aw ¼ qex pout  qwf pin þ kw Aw ð1Þ
@t @z2
k

where Fig. 1. An oscillating heat pipe.

 T. Daimaru et al. / Applied Thermal Engineering 113 (2017) 1219–1227 1221

Fig. 2. Nagasaki’s experimental apparatus [14].

Fig. 3. (a) Displacement of liquid column and pressure oscillation [14] and (b) an expected thermal cycle in a single-branch OHP.
 
ll cp;l
Prl ¼ ð6Þ
kl n

(a) Laminar flow (Rel,n < 2000) [16]

    1=3 
kl
hl ¼  1:953  Rel Prl DLin
Din l
h i n ð7Þ
Din
when Rel Prl L P 33:3
l n

h    i
hl ¼ Dkinl  4:364 þ 0:0722Rel Pr l DLin
l
h i n
ð8Þ
when Rel Prl DLin < 33:3
l n

(b) Transient flow (2000 5 Rel,n 5 10,000)

Gnielinski correlation
"   !# Fig. 4. Schematic diagram of the numerical model.
kl ðf s =8ÞðRel  103 ÞPrl
hl ¼  1=2
ð9Þ
Din 1 þ 12:7ðf s =8Þ ðPrl2=3  1Þ n
2.3. Vapor plug temperature and pressure
where f s is the friction factor of the liquid slug.
(c) Turbulent flow (Rel > 10,000) In this model, the temperature is uniform within each vapor
Dittus-Boelter correlation plug because the speed of the heat transfer is very fast in the vapor
    phase. Therefore, each vapor plug is calculated by only one node.
kl Fig. 6 shows the heat transfer path for the j-th vapor plug. Eq.
hl ¼  0:023Re0:8
l Pr 0:4
l ð10Þ
Din n (11) considers the heat transfer between the tube wall and vapor
through the liquid film. The saturation temperature of the vapor
Assuming the pressure of liquid slugs to be linear along its
pressure is used to calculate the heat flux because the vapor-
length, the pressures of adjacent vapor plugs are used as boundary
liquid film interface should be in the saturation condition. Simulta-
conditions.
1222 T. Daimaru et al. / Applied Thermal Engineering 113 (2017) 1219–1227

h i
dT v ¼ Pv dVcvv;vþdm
mv
v RT v
j ð13Þ
where ½dV v ¼ Acr ðul;n  ul;n1 Þj

½P v ¼ ðmv RT v Þ=V v j ð14Þ

½P v ¼ Psat ðT v Þj when½Pv > Psat ðT v Þj ð15Þ

2.4. Liquid slug momentum and velocity

The momentum equation for n-th liquid slug is written by fol-

lowing an ordinary differential equation. The liquid slug is treated
with a combined parameter approach, as shown Fig. 7. The left
Fig. 5. Thermal network. hand side of Eq. (16) is the change in the momentum at each time
step.
 
dpl
¼ Acr ðdPv  ð1  gturn ÞdPs  gturn dP turn Þ ð16Þ
dt n

The first term on the right hand side is the pressure difference
between adjacent vapor plugs. This term is caused by the pressure
change as a result of the work and phase change of the vapor plugs.
dPv ;n ¼ fPv ðT v Þgj  fPv ðT v Þgjþ1 ð17Þ

The pressure loss in the circular pipes is given by Eq. (18), and
the friction factor, f, is calculated for the cases of laminar flow
and turbulent flow regimes. The correlation for a steady developed
flow is used.
 
f
dPs ¼ s Ll ql u2l ð18Þ
2Din n
Fig. 6. Schematic of the vapor plug sub-section.
(a) Laminar flow (Rel < 2000)
½f s ¼ 64=Rel n ð19Þ
neously, this means that a steep temperature gradient exists
(b) Turbulent flow (Rel = 2000)
around the vapor and liquid film [11]. The total mass change
within a vapor plug is calculated by Eq. (12). The temperature of ½f s ¼ 0:3164Re0:25
l n ð20Þ
the j-th vapor plug is calculated by Eq. (13), where the work of
the vapor plug is considered. Finally, the ideal gas law, Eq. (13), cal- The third term on the right hand side of Eq. (16) is the pressure
culates the pressure of the vapor. It is important to note that the loss in the turn sections. The pressure loss in the turn sections is
pressure cannot exceed the saturation pressure of the vapor tem- defined by Eq. (21). The friction factor, fK, is calculated by using
perature calculated from Eq. (14). If the pressure is higher than the friction factor in the straight sections [17,18]. The transient
the saturation pressure, then the ideal gas law is no longer valid, Reynolds number (Return) is determined by Eq. (24).
and the pressure is set to the saturation pressure calculated by
 
f
Eq. (15) [5]. dPturn ¼ turn Ll ql u2l ð21Þ
2Din n
qwf ;k ¼ kl =dfilm ðT w;k  T sat ðPv ;j ÞÞ ð11Þ
(a) Laminar flow (Rel < Return) [17]
" Z #
zf ;n1  f turn =f s ¼ 0:1008De1=2 ð1 þ 3:945De1=2 þ 7:782De1
dmv ¼ qwf ;k pin =HLG dz ð12Þ ð22Þ
zb;n þ9:097De3=2 þ 5:608De2 Þ
j
(b) Turbulent flow (Rel = Return) [17]

Fig. 7. Schematic of the force balance in the liquid slug. Fig. 8. Schematic of the turn section.
 T. Daimaru et al. / Applied Thermal Engineering 113 (2017) 1219–1227 1223

(  2 )0:05 Table 1
Din
f trun =f s ¼ Rel ð23Þ Design parameters.
2r trun
Pipe wall material Stainless steel (–)
Working fluid R134a (–)
Here, the Dean number (De) is calculated in Eq. (23).
Charging ratio of W.F. 50 (%)
sffiffiffiffiffiffiffiffiffiffiffiffi Heating section length Lh 100 (mm)
Din Adiabatic section length La 80 (mm)
De ¼ Rel ð24Þ
2rturn Cooling section length Lc 100 (mm)
Pipe diameter (In/Out) Din/Dout 0.8 / 1.6 (mm)
 0:32 Number of turns 16 (–)
Din Radius of turns rturn 3 (mm)
Return ¼ 2  104 ð25Þ
2r turn
When the liquid slug exists in turn sections, the spatial ratio is Table 2
taken into account, as shown Fig. 8. The spatial ratio for turn sec- Calculation conditions.
tions are described as:
Heat input Qh 20 (W)
½gturn ¼ Lturn =Ll n ð26Þ Heat flux qh 2488 (W/m2)
Initial temperature Tint 10 (°C)
Finally, the velocity of the liquid slug is estimated from the Cooling temperature Tc 10 (°C)
momentum of the liquid slug: H.T. coeff. in cooling plate hc 1000 (W/m2/K)
Liquid film thickness dfilm 30 (lm)
  Time step dt 0.0010 (s)
pl
ul ¼ ð27Þ Number of time steps 100,000 (–)
Acr Ll ql n Mesh size dz 1 (mm)
Thermal conductivity of the pipe wall kw 16 (W/m/K)
Displacements of liquid slugs are described as: Specific heat of the pipe wall cp,w 499 (J/K/kg)
Density of the pipe wall qw 7920 (kg/m3)
½dz ¼ ul dtn ð28Þ

2.5. Numerical procedure

As a result of volume and mass changes in the vapor plugs, the

vapor plugs may collapse and the liquid slugs merge into each

Fig. 10. Computational domain and initial distribution.

Fig. 11. Temperature history of the pipe wall.

other. In such a case, the meshes of the liquid slugs are combined
for temperature calculation. Further, the momentum and length of
the new liquid slug are set to the sum of the values of the old
merged liquid slugs:
pl;n;new ¼ pl;n;old þ pl;n1;old ; Ll;n;new ¼ Ll;n;old þ Ll;n1;old
ð29Þ
when V v ;j < 0 or mv ;j < 0
Fig. 9. Calculation flowchart.
1224 T. Daimaru et al. / Applied Thermal Engineering 113 (2017) 1219–1227

correlation [18] for an isothermal case may be used to obtain a

rough estimate. This calculation case assumes a mean velocity of
0.25 m/s, which is obtained from the vapor velocity calculated by
assuming that 50% of the heat input is consumed in evaporation,
and a temperature of 10 °C. However, sensitivity to temperature
is much lower than sensitivity to velocity in the case of R134a.
2=3
0:67Cal
dfilm ¼ Din 2=3
ð30Þ
1 þ 3:35Cal

ll ul
where Cal ¼ ð31Þ
rl
Thermal properties of the working fluid are calculated by using
REFPROP version 9.0 [19], which can calculate properties using
functions of temperature. However, constant thermal properties
are used for the pipe wall.
Fig. 12. Momentum history of the liquid slug No. 5. Fig. 10 shows the computational domain and initial vapor-
liquid distribution. Liquid slugs are equally distributed within the
flow path in the initial distribution. The length of each liquid is
A flowchart of the calculation procedure is shown in Fig. 9. A 92 mm.
calculation is implemented in each sub-section. The fourth order
Runge-Kutta method is used for numerical stability of calculation. 4. Results and discussion
When merging occurs during the loop, modification of temperature
meshes and momentums of liquid slugs is dealt outside of the loop Figs. 11 and 12 show the temperature history of the pipe wall
[6]. and the momentum history of the liquid slug, respectively. The
temperature in the heating section increased linearly in the begin-
ning, then decreased at around 40 s and converged after that. Here,
3. Calculation conditions the momentum of one liquid slug (No. 5) is shown in Fig. 12. The
momentum also started oscillation at the almost same time as
Table 1 shows the design parameters of the calculated OHP, and the decrease in temperature. This change implied that the liquid
Table 2 shows the calculation conditions. Nikolayev [6] has slug initiated oscillation. Thus, it is clear that oscillation of the
reported that the regime of oscillation changes from intermittent working fluid transported heat from the heating to cooling section.
to periodic with increasing heat input. Finally, oscillation collapses This result agrees with the basic operational characteristics of
at high heat input because condensation of vapor plugs occurs OHPs. The average value of the temperature difference between
owing to the large inertia of the liquid slugs. Therefore, we set the heating section T1 and cooling section T5 from 90 to 100 s
the heat input to reproduce periodic and stable OHP operation. was 34.2 K, and the thermal resistance Rth, which was calculated
The thickness of the liquid films is a suitable parameter, but the by Eq. (32), was 1.71 K/W.

Fig. 13. Pressure distribution and pressure field.

 T. Daimaru et al. / Applied Thermal Engineering 113 (2017) 1219–1227 1225

Fig. 14. Pressure and volume histories of vapor No. 5.

tion of the position axis as time elapsed. The larger peak (a) main-
tained close to the same value, but the smaller peak (b) slightly
transformed during the propagation. The pressure difference
between the higher and lower pressure region was close to
100 kPa. The pressure distribution and propagation could be seen
more clearly in the pressure field. The pressure gradient was smal-
ler within the same channel, but was larger between different
channels. This pressure propagation has been suggested by Miya-
zaki et al. [4]. He observed that the positions of liquid slugs formed
a wave-like shape during oscillations in a visualization experiment.
He then suggested the presence of pressure propagation and devel-
oped a theoretical model by using the wave equation to explain the
operational principle of OHPs. However, pressure propagation has
never been confirmed in either experiments or numerical simula-
tions. Recently, Yoshida et al. [20] has presented results that sup-
port the presence of pressure propagation. They measured the
pressure of the working fluid in the center and at one edge of a
Fig. 15. Oscillation frequencies. transparent OHP.
The two pressure histories showed similar oscillation wave-
forms in different phases. In addition, this phase difference agreed
Tl  T5 with values obtained by assuming that one pressure wave traveled
Rth ¼ ð32Þ
Qh in the OHP.
Therefore, this result of pressure distribution was a remarkable
4.1. Propagation phenomenon result that could not be obtained by experiment using point mea-
surements of a pressure sensor. Furthermore, this serves as evi-
Pressure propagation occurred, when liquid slugs oscillated. dence that oscillations of liquids are connected within multi-
Fig. 13 shows the pressure distribution in the working fluid along branch OHPs because the motions of liquid slugs are eventually
the flow path and the pressure field in the OHP during one period driven by a pressure difference.
of the oscillation. In the pressure distribution, one large (a) and one We investigate the reason for the pressure propagation in the
small (b) pressure peak existed. They then traveled along the direc- following sections.

Fig. 16. Pressure and volume histories of the upper channel.

1226 T. Daimaru et al. / Applied Thermal Engineering 113 (2017) 1219–1227

Fig. 17. Pressure and volume histories of the down channel.

Fig. 18. Thermal cycles: (a) single vapor plug, (b) upper channel, and (c) down channel.

4.2. Thermal cycles

First, we discuss single vapor, as in the case of single-branch

OHP. Fig. 14 shows the pressure and volume history of vapor plug
No. 5. The pressure and volume were completely anti-phased with
respect to one another. The results of the Fast Fourier Transform
(FFT) are shown in Fig. 15. The power spectrums of both pressure
and volume displayed two peaks in frequency. The frequency of
6.2 Hz corresponded to a larger peak in the pressure distribution
shown in Fig. 13, and the 12.4 Hz peak included the smaller peak.
The thermal cycle was made from pressure and vapor volume by
normalization, as shown in Fig. 18(a). The thermal cycle had a
straight shape, similar to the case of single-branch OHP. However,
this was not a direct reason for pressure propagation because this
type of cycle did not represent the attainment of external energy or
exertion of energy.
Next, the thermal cycles that occurred within each channel are
discussed. The volume of single vapor was divided into two parts. Fig. 19. Energy waveforms of vapors No.1 and 5.
According to the direction of pressure propagation, these were
called the ‘‘upper channel” and the ‘‘down channel”. The pressure
and volume histories are shown in Figs. 16 and 17. In these cases, energy, and is likely the reason for pressure propagation in the
the pressure and vapor histories were not completely anti-phased. OHP.
Similar to Fig. 18(a), the thermal cycles are shown in Fig. 18 To validate this hypothesis, the phase difference between the
(b) and (c). These cycles had elliptical shapes, which was different energy waveforms was calculated using vapor plug No. 1 as a
from the case of single vapor plug. The cycle was in the left direc- Ref. The period of energy propagation should have been the same
tion in the upper channel and in the right direction in the down as that of the pressure oscillation if energy propagation was the
channel. The vapor obtained energy via compression in the right reason for pressure propagation. Fig. 19 shows examples of the
direction of the thermal cycle. Contrarily, the vapor exerted work waveforms. First, a time delay between two waveforms was calcu-
on its surroundings in the left direction. Thus, it could be consid- lated by the cross-correlation analysis as shown in Fig. 20. Next,
ered that the vapor plug obtained energy in the upper channel, the phase difference was calculated using the strongest frequency
and then transmitted that energy to the next channel in the down of pressure oscillation, 6.2 Hz, as shown in Fig. 21. The phase differ-
channel. This phenomenon was responsible for the propagation of ence between each vapor was almost 45°, which is equal to the
 T. Daimaru et al. / Applied Thermal Engineering 113 (2017) 1219–1227 1227

The vapor plug received or exerted energy according to the direc-

tion of propagation. Based on these results, we determined that
pressure propagation resulted from energy exerted by the vapor
in the form of work. Finally, this supposition was validated via
waveform analysis.

Acknowledgements

This work was partially supported by a Japan Society for the

Promotion of Science (JSPS) Grant-in-Aid for JSPS Fellows (Grant
No. 26-3928).

References

[1] H. Akachi, U.S. Patent, Patent No. 4921041, May 1, 1990.

[2] H. Akachi et al., Pulsating Heat Pipes, Proc. of the 5th Int. Heat Pipe Symp.,
Melbourne, Australia (1996) 208–217.
Fig. 20. Time delay of energy between vapors No. 1 and 5. [3] S. Khandekar et al., An insight into thermos-hydrodynamic coupling in closed
loop PHPs, Int. J. Therm. Sci. 43 (2004) 13–20.
[4] Y. Miyazaki et al., Oscillatory Flow in the OHP, in: Proc. of 11th IHPC, Tokyo,
Japan, 1999, pp. 367–372.
[5] B. Shafii et al., Thermal modeling of unlooped and looped PHPs, J. H. T. 123
(2001) 1159–1172.
[6] V. Nikolayev, A dynamic film model of the PHP, J. Heat Transf. 113 (2011),
081504-1-9.
[7] V. Nikolayev, Oscillatory instability of the gas-liquid meniscus in a capillary
under the imposed temperature difference, Int. J. Heat Mass Transf. 64 (2013)
313–321.
[8] T. Daimaru et al., Mathematical Modeling and Experimental Validation of
OHPs, Proc. of 44th Int. Conf. on Environmental Systems, Tucson, Arizona,
2014, ICES-2014-121.
[9] Y. Zhang et al., Advanced and unsolved issues in PHPs, Heat Transf. Eng. 29 (1)
(2008) 20–44.
[10] S. Kato et al., A single, straight-tube PHP, Int. J. Heat Mass Transf. 64 (2013)
254–262.
[11] P. Gully et al., Evaluation of the vapor thermodynamic state in PHP, in: Proc. of
the 17th International Heat Pipe Conf., Kanpur, India, 2013.
[12] R. Dobson, Theoretical and experimental modeling of an open OHP including
gravity, Int. J. Therm. Sci. 43 (2004) 113–119.
[13] P. Das et al., Thermally induced two-phase oscillating flow inside a capillary
tube, Int. J. Heat Mass Transf. 53 (2010) 3905–3913.
[14] T. Nagasaki et al., Study on the mechanism of liquid column oscillation in PHP,
Fig. 21. Phase difference. in: Proc. of 48th Japanese Symp. on H. T., Okayama, Japan, 2011.
[15] M. Mameli et al., Thermal simulation of a PHP: effects of different liquid
properties on a simple geometry, H. T. Eng. 33 (14) (2012) 1177–1187.
[16] K. Shah, Thermal entry length solutions for the circular tube and parallel
value of dividing one period (360°) by the number of vapor plugs.
plates, in: Proc. of the 3rd National H. Mass T. Conf., India, 1975, HMT-11–75.
In addition, the total amount of difference was 360°. These findings [17] H. Ito, Laminar flow in curved pipes, J. Appl. Math. Mech. 40 (1969) 653–663.
suggest that energy propagates through vapor, and the propaga- [18] P. Aussillous et al., Quick deposition of a fluid on the wall of a tube, Phys. Fluid
tion period is same as pressure propagation. From the above dis- 12 (10) (2000) 2367–2371.
[19] NIST Reference Fluid Thermodynamic and Transport Properties Database
cussion, we can conclude that pressure propagation occurred by (REFPROP) Version 9.1. https://www.nist.gov/srd/refprop, 2016 (accessed
that of vapor energy as work. 21.10.2016).
[20] S. Yoshida et al., A study on circulation flow of oscillating heat pipe with check
valves by visualization, in: Proc. Joint 18th Int. Heat Pipe Conf. and 12th Int.
5. Conclusions Heat Pipe Symp., Jeju, Korea, 2016.

This paper presented a numerical modeling method for a multi-

branch OHP. Propagation of pressure occurred in the simulation.