International Journal of Heat and Mass Transfer 102 (2016) 1267–1281

Contents lists available at ScienceDirect

International Journal of Heat and Mass Transfer

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

Experimental and numerical investigation of a cross flow air-to-water

heat pipe-based heat exchanger used in waste heat recovery
Joao Ramos a, Alex Chong a, Hussam Jouhara b,⇑
a
Faculty of Computing, Engineering and Science, University of South Wales, Pontypridd CF37 1DL, United Kingdom
b
Institute of Energy Futures, College of Engineering, Design and Physical Sciences, RCUK Centre for Sustainable Energy Use in Food Chains, Brunel University London,
Uxbridge UB8 3PH, United Kingdom

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

Article history: This paper applies CFD modelling and numerical calculations to predict the thermal performance of a
Received 10 February 2016 cross flow heat pipe based heat exchanger. The heat exchanger under study transfers heat from air to
Received in revised form 14 June 2016 water and it is equipped with six water-charged wickless heat pipes, with a single-pass flow pattern
Accepted 29 June 2016
on the air side (evaporator) and two flow passes on the water side (condenser). For the purpose of CFD
Available online 25 July 2016
modelling, the heat pipes were considered as solid devices of a known thermal conductivity which
was estimated by experiments conducted on the exact same heat pipe configuration under an entire test-
Keywords:
ing range. The CFD results were compared with the experimental and the numerical results and it was
Heat pipe
Heat exchanger
found that the modelling predictions are within 10% of the experimental results.
CFD modelling Ó 2016 Elsevier Ltd. All rights reserved.
Heat recovery

1. Introduction heat to one end of the heat pipe will cause the working fluid inside
the pipe to boil and, due to the lower density, to travel in vapour
Heat exchangers are commonly employed as heat recovery form towards the cooler end of the pipe, where it condenses and
devices to reuse the wasted heat energy from exhaust outlets so gives away the absorbed latent heat that was collected in the evap-
it may be furtherly reused or stored for a later use. According to orator section; thus completing the thermal cycle [7]. A represen-
the research of Haddad et al. [1] 90% of the wasted heat energy tation of the heat pipe working cycle can be seen in Fig. 2. Due to
is found at low to medium-grade heat applications (temperatures the high effective thermal conductivity of these devices at essen-
from 100 to 400 °C), as can be seen in Fig. 1. It is in this environ- tially constant temperature throughout its length [8,9], heat pipes
ment that heat pipe-equipped heat exchangers are finding wide have been referred as superconductors. Their effective conductivity
use due to an array of advantages ranging from a complete flow can easily be several orders of magnitude greater than pure con-
separation, great redundancy and ease of maintenance. All of the duction through a solid metal [5,10,11].
advantages are a direct result of the mechanism of phase change The small heat pipes used in electronic applications are
happening within the heat pipe [2]. equipped with a porous wick structure, which allows them to func-
tion in any orientation, provided there is a difference in tempera-
ture between both sides of the pipe [7,9]. However, the heat pipe
1.1. Literature review does not require a wick in order to function properly; as long as
the evaporator section is located below the condenser section,
Heat pipes were initially developed by NASA as effective heat the condensate working fluid is pushed back to the evaporator
sinks to cool down small-scale electronic equipment in space [3], through the force of gravity. For that reason, wickless heat pipes
while nowadays they are commonly used for cooling purposes of are also known as gravity-assisted heat pipes or two-phase closed
electronic equipment from mobile phones to CPUs [4–6]. A heat thermosyphons [9]. The term ‘‘thermosyphon” is used throughout
pipe consists of a hermetically-sealed tube filled with a small mass the paper and refer to the devices employed in this study.
of saturated working fluid that exists in liquid and vapour form and Thermosyphon-equipped heat exchangers (TSHE) offer many
occupies the whole of the internal volume of the tube. Applying advantages when used as waste heat recovery devices, such as
an increased redundancy and reliability, ease of cleaning, no
⇑ Corresponding author. additional power input to the system, reduced risk of
E-mail address: [email protected] (H. Jouhara). cross-contamination and no moving parts; all advantages widely

http://dx.doi.org/10.1016/j.ijheatmasstransfer.2016.06.100
0017-9310/Ó 2016 Elsevier Ltd. All rights reserved.
1268 J. Ramos et al. / International Journal of Heat and Mass Transfer 102 (2016) 1267–1281

Nomenclature
U overall heat transfer coefficient
Abbreviations D difference
HPHX heat pipe-equipped heat exchanger DTLM LMTD – logarithmic mean temperature difference
HVAC heating, Ventilation and Air Conditioning e effectiveness
NTU number of Transfer Units l static viscosity of the liquid phase
TSHX thermosyphon-equipped heat exchanger p pi
VOF volume of Fraction (CFD method) q density
Pr Prandtl number r surface tension
Nu Nusselt number
Re Reynolds number Subscripts
act actual
Symbols b boiling
A area c condenser
C heat Capacity Rate (m _
cp ) cd/cond condensation
cp specific heat at constant pressure e evaporator
Cr heat Capacity Ratio ðC e =C c Þ h heat transfer coefficient
csf constant/coefficient dependent on surface-liquid i inside
combination in inner
d characteristic dimension k conduction
g acceleration of gravity l liquid phase
h heat transfer coefficient min minimum
hfg latent heat n number of pipes
k thermal Conductivity o outside
Q heat Transfer Rate s Surface
q’’ heat flux sat saturation
r radius t total
T temperature ts thermosyphon

highlighted in the literature surveyed [12–17]. There is much liter- computing which allow the simulation of the phase change within
ature available in heat exchangers equipped with heat pipes the thermosyphon. However, a divide seems to exist in the
(HPHX) used in Heating, Ventilation and Air-conditioning applica- reviewed literature; the authors either focus solely into the CFD
tions (HVAC) [5,10,18–20], as well as heat recovery [21–23]. For simulation of the phase change process in the thermosyphon or
low to medium grade heat, consisting of temperatures above on the behaviour of the fluid on the shell side of a HPHX.
150 °C, Noie [24] presents an analytical method of characterising The simulation of heat pipes through CFD is a fairly novel field
the HPHX using the Effectiveness (e-NTU method) to predict the of research, made possible due to the increase in computational
performance of the heat exchanger. The same approach is taken power of modern computers. Alizadehdakhel et al. [27] and Fadhl
by Danielewicz et al. [25], Jouhara and Merchant [26] and Han &
Zou [8] with the aid of a computer coding that predicts the effec-
tiveness depending on different inlet conditions. All the surveyed
papers refer to air-to-air HPHX even if it is mentioned by Noie that
the e-NTU method theory applies even if different fluids are used
on the shell side.
The development of CFD codes for simulation of heat pipes and
heat pipe heat exchangers is a relatively new field of research. It
has been receiving renewed interest due to recent advances in

Fig. 1. Waste heat energy by temperature range (adapted from [1]). Fig. 2. Schematic of a working heat pipe.
 J. Ramos et al. / International Journal of Heat and Mass Transfer 102 (2016) 1267–1281 1269

et al. [28] have both successfully modelled the thermosyphon simulations, by assuming that the heat pipes are solid devices of
using a custom volume of fraction (VOF) code in Fluent, a popular a constant conductivity. The conductivity is predicted using
CFD release. Both works depicted a 2 dimensional study and adapted versions of equations found in literature. The results shall
proved that the software is capable of simulating the phase change prove that the thermal resistance analogy within the heat pipe can
process within a single heat pipe during the evaporation and con- be extended to 3-dimensional CFD simulations.
densation processes albeit after a long processing time. More At first the theoretical modelling of thermosyphons is presented
recently, the same method has been successfully applied to a 3D in detail. It is followed by the experimental set up and the design-
model to simulate Geyser boiling in the heat pipe [29]. Geyser boil- ing conditions of the CFD model. The results are then presented
ing usually takes place at low heat input, when a large amount of and compared.
evaporated fluid bubbles starts to form below the liquid bulk.
When the pressure difference between the bubble and the liquid
2. Theoretical analysis
bulk becomes too great, the liquid is projected into the top of the
thermosyphon [30,31].
In terms of predicting the performance of a heat exchanger
CFD has also been used to calculate the optimum filling ratio for
equipped with thermosyphons, there are quite a few examples in
a thermosyphon by calculating the quantity that will allow the
the literature; Azad & Geoola [38] and Kays and London [39] were
shortest response time and lowest thermal resistance [32]. Accord-
some of the first to report the use of the effectiveness-Number of
ing to Shabgard et al. [32], it is recommended that an extra 5–10%
Transfer Units (e-NTU) method to predict the performance of a
of fluid is inserted in the pipe to prevent breakdown of the liquid
heat exchanger equipped with thermosyphons to great effect. Even
film from the thermosyphon wall. A three-dimensional numerical
to this day authors continue using the same approach as it has pro-
study simulating multi-phase flow inside horizontally oriented
vided satisfactory results; Lukitobudi et al. [40] used it in an
heat pipes was conducted by Hughes et al. [33] for steady-state
approach to recovering waste heat in bakeries, Noie [24] used it
conditions. In this study, a multiphase flow with coupled heat
in an investigation of an air-to-air heat exchanger used in heat
and mass transfer was used. In order to predict the performance
recovery, and Jouhara & Merchant [26] reported the same in their
of the heat pipes, the effectiveness of the heat exchanger was
multi-use apparatus.
determined through an experimental study. A good correlation
was found between the results from the CFD model and the exper-
imental results for the same operating conditions. 2.1. Predicting the performance of a single thermosyphon
As mentioned previously, a progression is being observed in the
application of CFD to the simulation of thermosyphons, however A thermosyphon is, in many ways, a miniature heat exchanger;
there is not much literature on the study of the entire HPHX or so it is only natural to approach it the same way a heat exchanger
TSHX. Selma et al. [34] have designed a working model of a heat is approached. The most reported method of predicting the perfor-
exchanger equipped with heat pipes using OpenFOAM, an open- mance of a thermosyphon is through the thermal network analogy
access CFD release, in order to improve the energy efficiency of [7,9,13,37], also approached in this study. In this analogy, the ther-
an existing model. A 3-dimensional simulation of the external flow mosyphon is broken down into its inner thermal resistances, con-
surrounding the pipes was created and used the outer wall of the duction, boiling, condensation, etc.
pipes as a constant temperature boundary condition gathered from Treating the circuit displayed in Fig. 3 as an electrical circuit;
industrial practice. The results proved very satisfactory and corre- and neglecting the axial thermal conductivity, the total thermal
lated very well with both experimental data and a commercial CFD resistance for the thermosyphon is found through Eq. (1).
release. Peng et al. [35] conducted a CFD study on the effect of fin
RT ¼ Rh;e þ Rk;e þ Rb þ Rin þ Rcd þ Rk;c þ Rh;c ð1Þ
shape on the air-side heat transfer performance of a fin-plate ther-
mosyphon used in electronics cooling. The simulation focused The axial thermal conductivity along thin-walled ther-
solely on the air side and results from the CFD model were within mosyphons with long adiabatic sections may be considered negli-
15% error of the experimental results. CFD has also been employed gible [7,13] therefore simplifying the expression to a simple
to simulate the feasibility of installing heat pipes within a wind addition. In the case of a heat pipe, which is equipped with a wick
tower. In a study by Calautita et al. [36], the heat pipes were mod- structure, an additional parallel network of thermal resistances is
elled as having a constant surface temperature, a reasonable added which includes the convection to enter the wick and the
assumption taking into account there is little difference in the tem- axial conductivity along the length of the device.
perature of the working fluid inside the pipe. The results showed
that the incorporation of heat pipes in this application is capable
2.2. Conduction through the thermosyphon walls
of improving the reduction in inlet air temperature.
It appears that there is a gap in the form of o attempt made at
The flow of heat through the thermosyphon starts with heat
simulating the thermosyphons and the heat exchanger in the same
entering the heat pipe through the evaporator wall. The thermal
simulation. Other than the VOF method, there were no other rec-
resistance at the evaporator wall is deducted from conduction
ommendations in terms of alternative methods of simulating the
through a hollow pipe wall:
thermosyphons using, for example, the thermal network analogy.
The author therefore recommends the application of the thermal 2pklDT lnðr out  r in Þ
Q_ k ¼ ) Rk ¼ ð2Þ
network analogy in order to predict the thermal conductivity of lnðr out  r in Þ 2pkl
the thermosyphons and feeding that value as a boundary condition
into the CFD model of the TSHX. Mroué et al. [37] conducted a where k is the thermal conductivity of the encasing material
study similar to the one present in this work, in which the flow (W/mK), l is the length of the pipe in contact with the hot air flow
on the hot side of the TSHX was allowed to return in order to make (m), r out  r in represent the tube thickness (m) and DT is the differ-
contact with the tubes where film boiling takes place. The thermal ence in temperature between the inside and the outside of the pipe
network analysis is also used in conjunction with the e-NTU (°C). This applies to the evaporator and to the condenser section
method in order to accurately predict the behaviour of the TSHX. equally but with different values for the inner and outer tempera-
The objective of this paper is to implement the analytical and tures. It is mainly affected by the area of exposure and the conduc-
theoretical background analysis of thermosyphons with CFD tivity of the material.
1270 J. Ramos et al. / International Journal of Heat and Mass Transfer 102 (2016) 1267–1281

Thermal resistance from convectioon on

outside of evaporator section
Thermal resistance from conductio on
across wall of evaporator section
Thermal resistance from boiling oon TS
wall
Thermal resistance from change in n
pressure between top and bottom
Thermal resistance from condensaation
on TS wall
Thermal resistance from conductio on
across wall of condenser section
Thermal resistance from convectioon on
outside of condenser section
Thermal resistance from Axial
conduction across thermosyphon

Fig. 3. Schematic of the thermal resistances within the thermosyphon.

2.3. Convection outside the thermosyphon cylinder in cross-flow has been extensively studied in literature
and it applies to the case at hand. The correlation used for external
Looking at either section, heat transfer through convection from flow over a single cylinder is that of Zhukauskas [41]:
the surrounding medium to the pipe is given by the following

1=4
equations for the evaporator section and the condenser section, Pr
Nu ¼ C Rem Prn
respectfully: Prs
ð5Þ
1 1 ½0:7 < Pr < 500
Q_ h ¼ hADT ) Rh;e ¼ and Rh;c ¼ ð3Þ
he Ae hc Ac ½1 < Re < 1
106 
h refers to the heat transfer coefficient (W/m2K) between the The constants C and m are functions of the turbulence in the
fluid and the solid surface, which in this case is air-carbon steel vicinity of the cylinder and are available in Table 1. All fluid prop-
for he and water-carbon steel for hc . A is the exposed surface erties are evaluated at the arithmetic mean of the fluid inlet and
area (m2) and DT the difference in temperature between the flow outlet temperatures except for the properties marked with an s,
and the surface area. which are evaluated at the boundary between the solid and the
The heat transfer coefficient is derived from the Nusselt number fluid.
through the expression:
hd 2.3.2. Convection outside the thermosyphon’s evaporator section
Nu ¼ ð4Þ
k In the evaporator section the thermosyphons are organised into
where h represents the heat transfer coefficient (W/m2 K), d the two lines of three thermosyphons each as seen in Fig. 4. Eq. (5) is
characteristic dimension (m) and k the thermal conductivity (W/ not applicable to the evaporator section due to the higher volume
mK) of the surrounding fluid. The Nusselt number is a function of of pipes, and Zhukauskas’ [41] correlation for a range of vertical
the flow conditions, in particular the turbulence and will therefore tubes in a staggered arrangement is preferred. The expression
be different in the evaporator section and in the condenser section. has the form:

1=4
Pr
2.3.1. Convection outside the thermosyphon’s condenser section Nu ¼ C 1 C 2 Rem
max Pr
0:36

A cut-section of the condenser section of the heat exchanger

Prs
under study is schematically represented in Fig. 4. ½NL P 20 ð6Þ
Water flows through each thermosyphon one by one following ½2000 < Remax < 40; 000
a u-shaped path. The heat transfer by convection over a vertical ½Pr P 0:7
All properties of the fluids used in Eq. (6) are evaluated at the
mean film temperature except the properties marked with an s,
which are evaluated at the boundary temperature. C1 and m
depend on the geometry of the tube bundle and are taken from
Table 2, C2 is the correction factor used in case fewer than 20 rows
of tubes (N L < 20) are present and is available in Table 3.
This expression also takes into account the maximum turbu-
lence and therefore Remax is used, a variable based on the maxi-
mum fluid velocity. The maximum velocity occurs at the smallest

Table 1
Constants of Eq. (5) for a circular cylinder in cross flow [41] – excerpt.

Re C m
1–40 0.75 0.4
40–1000 0.51 0.5
Fig. 4. Cross-section of the condenser section of the heat exchanger under study (all 103–2
105 0.26 0.6
dimensions in mm).
 J. Ramos et al. / International Journal of Heat and Mass Transfer 102 (2016) 1267–1281 1271

Table 2 comprehensive correlation as it holds remarkably well and has

Constants of Eq. (6) for airflow over a tube bank of 20 or more rows [41] – excerpt. been reported by many researchers in the literature; Reay and
Configuration Remax C1 m Kew [7], Hagens et al. [13], Mroué et al. [37] and Ramos et al.
Staggered 10–10 2
0.90 0.40 [45] reported the use of Rohsenow & Hartnett’s expression [46]
Staggered 102–103 Approximate as a single to predict the heat transfer from nucleate pool boiling in ther-
(isolated) cylinder mosyphons. Rohsenow & Hartnett’s expression for nucleate pool
Staggered ST =SL < 2 103–2
105 0.35 ðST =SL Þ1=5 0.60 boiling has the form shown in Eq. (8).
Staggered ST =SL > 2 103–2
105 0.40 0.60 The subscript l refers to the liquid phase and v to the gas phase
Staggered 2
105–2
106 0.022 0.84
as during boiling there is a mix of both. The coefficient csf and the
exponent n depend on the surface-liquid combination. The ther-
mosyphons under investigation are composed of carbon steel and
Table 3
Correction factor C2 of Eq. (6) for NL < 20 [41].
filled with water having the following vales of csf = 0.006 and
n = 1.0 [46].
NL 1 2 3 4 5 6 7 8 9  1=2
3
gðql  qv Þ cp;l ðT sat  T s Þ
Aligned 0.64 0.80 0.87 0.9 0.92 0.94 0.96 0.98 0.99 q00 ¼ ll hfg ðW=m2 Þ ð8Þ
Staggered 0.68 0.75 0.83 0.89 0.92 0.95 0.97 0.98 0.99 r csf hfg Prnl
The thermal resistance offered by the boiling process may be
area; transversally or diagonally between the tubes, according to found by first converting the heat flux to heat transfer rate by mul-
Fig. 5. tiplying it with the heat transfer area and then divide the differ-
ence in temperature by this value, as shown in Eq. (9):
2.4. Thermal resistance from vapour pressure drop T s  T sat
Rb;pool ¼ ðK=WÞ ð9Þ
q00 Ae;inside
The thermal resistance from vapour pressure drop (Rin in Fig. 3)
changes as the vapour pressure decreases as it flows from the It is important to note that the temperature of saturation of the
evaporator to the condenser section. The expression used for the fluid (T sat ) and the Temperature of the boundary (T s ) are required
vapour pressure drop is that of Faghri [43] and takes the form: in order to solve the expressions related to the boiling and conden-
  sation of the working fluid. This is often resolved by employing
8Rg lv T 2v ðLe þ Lc Þ=2 þ La thermocouples inside of the thermosyphon and on its surface.
Rv ¼ ð7Þ
ph2fg Pv qv r4i
2.6. Condensation heat transfer
where Rg , hfg , La , T v , Pv , lv and qv are the specific gas constant,
latent heat of vaporisation of the working fluid, adiabatic section Condensation is found to take place in the condenser section
length and temperature, pressure, dynamic viscosity and density where the working fluid, upon coming into contact with the cooler
of the vapour phase, respectfully. The vapour temperature is the walls, condenses and gives up its latent heat energy. For this heat
average temperature between the evaporator and condenser sec- transfer mechanism, the laminar condensation equation from Nus-
tion temperatures and the vapour pressure is the saturation pres- selt is used [47]:
sure correspondent to the vapour temperature. !1=4
kl ql ðql  qV Þghfg
3
hcond ¼ 0:943
ð10Þ
2.5. Boiling heat transfer ll hl
Assuming that ql  qv further simplifies the equation. Based on
Boiling regimes are dependent on the temperature difference
experiments, McAdams [48] also suggests a 20% increase to theo-
between the bulk temperature of the fluid and the heating wall.
retical expressions due to the fact that experimental values are
In addition, evaporation on a pool of liquid is different from evap-
often larger, changing (10) into:
oration of a liquid film. The thermosyphons under study were all
!1=4
engineered to work in the nucleate pool boiling regime. The kl q2l ghfg
3

expression chosen to predict the heat transfer in nucleate pool hcond ¼ 1:13
ð11Þ
ll hl
boiling is that of Rohsenow & Hartnett [44], found to be the most

Fig. 5. Tube arrangements in a bank (a) aligned and (b) staggered [42].
1272 J. Ramos et al. / International Journal of Heat and Mass Transfer 102 (2016) 1267–1281

2.7. Predicting the performance of a thermosyphon-based heat thermosyphon (in W/mK) and A the cross sectional area (in m2).
exchanger Re-arranging the equation for k and considering the cross-
sectional area of a circle, Eq. (17) is obtained:
The heat exchanger under study is equipped with 6 ther-
L
mosyphons arranged in parallel as shown in the schematic repre- k¼ ðW=mKÞ ð17Þ
R1ts pr2
sented by ‘‘Parallel” refers to their arrangement within the
thermal network analogy. From a heat transfer perspective, the Eq. (17) represents the thermal conductivity for a ther-
total thermal resistance for the heat exchanger as shown in Fig. 6 mosyphon if it is assumed to be a solid super conductor. This value
would assume the six thermosyphons are in parallel with each is used as a boundary condition in the CFD simulation.
other. This means that the overall thermal resistance would be
smaller the more thermosyphons are included in the assembly. 2.8. The effectiveness-NTU (e-NTU) prediction method applied to a
This is represented analytically in the following way: TSHX
1 1 1 1 1 1 1
¼ þ þ þ þ þ ð12Þ e-NTU stands for effectiveness-Number of Transfer Units and it is
R6 thermosyphonsðTSÞ Ri Rii Riii Riv Rv Rv i
a method of predicting the performance of a heat exchanger. The
The thermosyphons (TS) are assumed to have the same average number of transfer units is a dimensionless parameter widely used
internal thermal resistance thus simplifying the equation to: in heat exchanger analysis, generally defined as:

1 1 R1TS UA
¼
6 ) R6TS ¼ ð13Þ NTU  ð18Þ
R6TSs R1TS 6 C min
When looking at the larger picture as displayed in Fig. 6, the where U represents the overall heat transfer coefficient, A the total
overall thermal resistance for the thermosyphon-equipped heat heat transfer area and Cmin the minimum heat capacity rate
exchanger (Rth,TSHX) may be found through the following between the hot and cold flows. The heat capacity rate is a measure
expression: of the mass flow rate (m _ in kg/s) multiplied by the specific heat
capacity (cp in J/kgK).
Rth;TSHX ¼ Rh;e þ R6TSs þ Rh;c ð14Þ
Effectiveness (e) is the ratio of the actual heat transfer rate to
The subscript TS stands for thermosyphon, e for evaporator, c for the maximum possible heat transfer rate; which would be
condenser and o for outer. achieved if the temperature of the outlet of the cold flow would
Placing Eq. (13) into Eq. (14): equal the inlet temperature of the hot flow [26]:

R1TS qact C c ðT h;i  T h;o Þ C c ðT c;o  T c;i Þ

Rth;TSHX ¼ Rh;e þ þ Rh;c ð15Þ e  cand e¼ ð19Þ
6 qmax C min ðT h;i  T c;i Þ C min ðT h;i  T c;i Þ

where Ch and Cc represent the heat capacity of the hot and cold
2.7.1. Determination of the thermal conductivity of a single flows, respectively, and Cmin the smallest between the two. By def-
thermosyphon inition, effectiveness is dimensionless and must be valued between
If the thermosyphon is assumed to be a solid super-conductor, 0 and 1; theoretically only a heat exchanger of infinite length would
this means that the total axial conductivity of a single ther- be able to achieve an effectiveness of 1.
mosyphon may be taken as the axial conduction through a solid The e-NTU analysis of a heat exchanger equipped with ther-
pipe: mosyphons may be done by separating it into two separate heat
exchangers, the condenser and the evaporator, and consider them
L
R1TS ¼ ðK=WÞ ð16Þ coupled by the thermosyphon working fluid [6,24,38,43]. The
kA
effectiveness of the evaporator and condenser sections of the heat
where R1 TS is the overall thermal resistance of a single exchanger is determined separately and is given by:
thermosyphon, L is correlated to the length of the thermosyphon
(in m), k is the effective thermal conductivity for a single ee ¼ 1  eðNTUe Þ and ec ¼ 1  eðNTUc Þ ð20Þ
The number of transfer units for the evaporator and for the con-
denser is found through:
U e Ae U c Ac
NTUe ¼ and NTUc ¼ ð21Þ
Ce Cc
where A refers to the total heat transfer area in the respective row
or stage, and Ce and Cc represent the average heat capacity of the
shell-side fluid. The overall heat transfer coefficient U must be
found for each section using the thermal network analogy explained
in a later chapter.
Faghri [43] defined the effectiveness of an individual ther-
mosyphon to be related to the minimum and maximum values of
effectiveness between the evaporator and the condenser sections:


1 Cr C min
ets ¼ þ ; Cr ¼ ð22Þ
emin emax C max
where emin and emax take the minimum and maximum values of ee
and ec. Cr is the heat capacity ratio and Cmin and Cmax follow the
Fig. 6. Schematic of the thermal resistances within the thermosyphons equipped in same logic of the effectiveness, taking the minimum and maximum
the heat exchanger. values of Ce and Cc, respectively.
 J. Ramos et al. / International Journal of Heat and Mass Transfer 102 (2016) 1267–1281 1273

Fig. 7. Experimental apparatus of the heat exchanger in cross flow. From left to right: the heat exchanger before installation; the heat exchanger after being thermally
insulated; representative schematic of the thermosyphon heat exchanger and the size of its respective sections.

 n
1C r;c ec
1ec
1
ec;n ¼  n ð24Þ
1C r;c ec
1ec
 C r;c

Cr is the heat capacity ratio of the fluid streams on each side of the
thermosyphon; the ratio between the heat capacity rate of the
shell-side fluid to the heat capacity rate of the thermosyphon’s
working fluid. However, since the working fluid is at constant tem-
perature, its specific heat and capacity rate is effectively infinite,
making the variables Cr,e and Cr,c equal to zero [39]. Eqs. (23) and
(24) are then simplified into the forms seen in Eqs. (25) and (26),
Fig. 8. Schematic of the single pass test setup. Description: TCI/TCO – Thermocou- respectively:
ples at inlet/outlet of condenser; HE-C/E/A – Thermosyphon heat exchanger
Condenser/Evaporator/Adiabatic section; FM-1 – Turbine Flow Meter; WP – Water ee;n ¼ 1  ð1  ee Þn ð25Þ
pump; WT – Water tank; TEI/TEO – Thermocouple at inlet/outlet of evaporator; AP
– Air Pump; H-1 – Air Heater.
ec;n ¼ 1  ð1  ec Þn ð26Þ
The overall effectiveness depends on which fluid side has the
largest heat capacity; if the heat capacitance of the evaporator side
fluid is the largest; Ce > Cc:

1
1 C c =C e
et ¼ þ ð27Þ
ec;n ee;n
On the other hand, if Cc > Ce:

1
1 C e =C c
et ¼ þ ð28Þ
ee;n ec;n
Using the overall effectiveness of the heat exchanger, the outlet
temperatures for the evaporator and the condenser can be pre-
dicted from:
C min
T h;out ¼ T h;in  et ðT h;in  T c;in Þ ð29Þ
Ce

Fig. 9. The brazed thermocouples on the surface of the thermosyphon. C min

T c;out ¼ T c;in þ et ðT h;in  T c;in Þ ð30Þ
Cc
The effectiveness of a multistage heat exchanger in counter flow
for an n number of rows has been adapted from Incropera and
DeWitt [47] in order to apply to thermosyphon-equipped heat 3. Experimental set up
exchangers.
 n This chapter describes the physical design of the experimental
1C r;e ee
1ee
1 rig, the working conditions of the heat exchanger, including the
ee;n ¼  n ð23Þ
inlet temperatures and flow rates of both shell sides, as well as
1C r;e ee
1ee
 C r;e
the control apparatus used.
1274 J. Ramos et al. / International Journal of Heat and Mass Transfer 102 (2016) 1267–1281

3.1. Design of test rig section, and within the thermosyphons. The thermocouple place-
ments are marked in red in the simplified schematic of Fig. 7.
The experimental rig consisted of a heat exchanger equipped Four thermocouples were placed on the surface of each ‘‘corner”
with six thermosyphons in a cross-flow arrangement. The design thermosyphon: one in the evaporator section, two in the adiabatic
of the rig was based on a real heat exchanger used by the partner section and one in the condenser section. These thermocouples
company, built with a modular design in mind in order to allow were brazed into the surface of the thermosyphon at 1 mm depth.
further investigation of different flow configurations and boiling Two brazed thermocouples are shown in Fig. 9.
regimes. The experimental procedure was the same in all the tests and it
The test rig was equipped with six thermosyphons vertically is described as follows:
arranged in two staggered rows. The thermosyphon tubes were
made of carbon steel measuring 2 m in length and with a diameter 1. The cold water was allowed to run in the cold water circuit.
of 28 mm. The surrounding wall had an average thickness of 2. The air was then bled from the condenser section through the
2.5 mm. The working fluid was distilled water and the filling ratio bleed valve located at the top.
was 100% (of the evaporator section), which roughly translates into 3. After the condenser section was filled with water, the heater
0.7 m in height from the bottom of the thermosyphon. All tubes was turned on and the hot air flow was flowing in the evapora-
were chemically treated before insertion of water to avoid tor section.
corrosion. 4. The temperature was set to 300 °C. Data for this temperature
As can be seen in Fig. 7, the condenser section occupied the top setting would be recorded for a fan rate of 10, 20, 30, 40 and
0.2 m of the thermosyphons and the evaporator section the lower 50 Hz. The same flow rates were then tested for 250, 200,
0.6 m. The remaining 1.2 m were kept fully insulated as they 150, 100 and 50 °C.
served the adiabatic section of the heat pipe. Both the evaporator 5. Data were recorded and collected for each setting every10 min
and the condenser were separated from the adiabatic section by at steady state conditions.
a 10 mm-thick division plate in order to prevent leaks.
A total of 30 tests were conducted, one for each different inlet
condition.
3.2. Experiment design

The experimental rig was divided between two circuits; a 4. The CFD model
closed air circuit – the heat source – and an open water circuit –
the heat sink –, both included in the schematic represented in A 3 dimensional computational model was run in parallel with
Fig. 8. The hot air circuit consisted of a closed air loop equipped the experimental tests conducted on the heat exchanger. In this
by a fan and a heater. The flow was directed to pass through the simplified model the thermosyphons were modelled as super-
fan and the heater and then to enter the heat exchanger. After leav- conductors whose thermal conductivity had been deduced accord-
ing the heat exchanger, it was sucked into the fan once again, ing to the inlet conditions using the analytical method explained in
repeating the cycle. The fan frequency ranged between 10 and chapter 2.
50 Hz in 10 Hz increments. The mass flow rate was controlled by
an analogue pitot tube installed at the inlet of the evaporator sec- 4.1. Assumptions
tion and ranged between an average 0.05and 0.14 kg/s in 0.03
increments. The following assumptions were made prior to running the
The heater power could be regulated according to a desired simulation:
temperature thanks to a feedback loop. The feedback loop con-
trolled the heater power through a thermocouple located at the (a) Constant mass flow rate across the heat exchanger in both
outlet of the heat exchanger. The temperature of the air varied flow sides.
between 100 and 300 °C in 50 °C increments. The higher tempera- (b) Neglectable axial heat transfer from conduction across the
tures correspond to the normal working conditions of the heat thermosyphon wall.
exchangers encountered in waste heat recovery [12]. The lower (c) No heat transfer across the walls of heat exchanger.
temperatures were employed to test the lower operating limits (d) No heat transfer at the adiabatic section of the
of the thermosyphons. thermosyphon.
The cold water circuit consisted of an open loop and included a (e) Constant inlet mass flow rate across inlet area.
water tank to help regulate the inlet flow rate into the pump as can (f) Same thermal conductivity for all the thermosyphons.
be seen in Fig. 8. The mass flow rate of water was kept constant (g) The thermosyphons were assumed to be solid
throughout the test at 0.08 kg/s and at an average temperature of superconductors.
10 °C. After leaving the water tank, the water was pumped through
the heat exchanger. After flowing through the heat exchanger, the
4.2. Methodology
warmed-up water would freely flow into another process. The flow
inside the condenser section followed a U-shaped path as depicted
ANSYS Fluent was the computational fluid dynamics (CFD) soft-
in Fig. 4.
ware used to simulate the heat flow within the heat exchanger. The
model was created with the purpose of assessing the potential of
3.3. Gathering and processing of data simulating heat pipes as solids rods of constant conductivity for
the modelling of future heat exchangers. The realizable k-epsilon
20 k-type thermocouples were placed at specific locations in turbulence model (k–e) was used in each of the simulations as it
the heat exchanger to measure the temperature of the flows and is found to be more accurate at higher Reynolds number and smal-
the working temperatures of the thermosyphons. The thermocou- ler pressure gradients [49,50], which is the case in this particular
ples were placed in key sections, such as the inlet and outlet of experimental test range. A coupled pressure-based solver is also
both the evaporator and condenser sections, on the surface of each recommended as it is more efficient in steady-state simulations
‘‘corner” thermosyphon (numbered 1–4 in Fig. 4), on the adiabatic and offers better results for single-phase fluid flow [45,51].
 J. Ramos et al. / International Journal of Heat and Mass Transfer 102 (2016) 1267–1281 1275

Table 4 Table 5
Mesh dependency. Boundary Conditions.

Level No of cells Type of cells Skewness Time per iter. Type Mass flow rate (kg/s) Temperature (°C)
Coarse 1,408,658 Hex + Tetra Avg: 26%, stdev: 16% 2–10 s Evaporator Mass flow 0.05–0.17 at 0.03 50–300 at 50
Medium 2,291,364 Hex + Tetra Avg: 21%, stdev: 13% 7–15 s inlet inlet intervals intervals
Fine 3,099,230 Hex + Tetra Avg: 21%, stdev: 13% 25–50 s Evaporator Mass flow – Desired output
outlet inlet
Condenser Mass flow Constant 0.0715 Constant 10.0 ± 0.3
inlet inlet
The thermosyphons were modelled as solid objects using the Condenser Mass flow – Desired output
value for thermal conductivity achieved from the method outlet inlet
described in chapter 3. For the fluid properties, Fluent’s own stan-
dard tables of substance properties were used to determine the
characteristics of the fluids simulated (water in the condenser All the walls of the heat exchanger were insulated during the
and air in the evaporator). experimental test and were thus assumed to be adiabatic in the
CFD simulation. The only contact between the hot and cold flows
4.3. Mesh selection is made through the thermosyphons.
The thermosyphons were modelled not as two-phase devices,
The mesh selection was done by running the same simulation but as solid bars. The thermal conductivity of the tubes was pre-
with different mesh sizes and comparing the accuracy of the dicted using the thermal network analogy described in chapter 2.
results. A mesh was deemed ‘‘good” if the maximum skewness is
lower than 0.7 for hexahedron and tetrahedrons and 0.8 for trian- 5. Results and discussion
gular elements [20]. A comparison of the results is available in
Table 4 and can be observed in Fig. 10. This chapter presents the main results and includes a compar-
It was found that a fine mesh would take 3 times longer to con- ison between the experimental, theoretical and CFD results. First
verge (on average) and the results would not be significantly more the results from the experimental tests are presented and then a
accurate (±0.8%); therefore the medium mesh was used in all the comparison of the results obtained from the CFD and the numerical
tests run in Fluent. predictions is included.
The relaxation factors were set at 1e-6 and the test allowed to
run until no change was observed in the scaled residuals. 5.1. Experimental Results

4.4. Boundary Conditions This section outlines the results for the experiment with six
thermosyphons. Air was used as the evaporator-side fluid and
The boundary conditions used in the CFD to describe the inlet water as the condenser-side fluid. The mass flow rate of air varied
and outlet of the heat exchanger model are displayed in Table 5. between 0.05 and 0.17 kg/s and the inlet air temperature varied
The CFD model was run several times for each different inlet con- between 50 and 300 °C. On the condenser side, the inlet tempera-
dition to reduce the variance of the results. ture and the mass flow rate of water were both kept constant at
The boundary conditions used to simulate the inlets and outlets approximately 7 °C and 0.08 kg/s respectively.
were of type ‘‘mass flow inlet”, with the outlets having the oppo- Fig. 11 displays the temperature distribution within the heat
site direction. This assumption is valid for both circuits: the air cir- exchanger for inlet temperatures of 50, 100, 150, 200, 250 and
cuit consisted of a closed system, as observed in Fig. 8, so it was 300 °C. The temperatures were logged from five different locations;
being ‘‘pulled” out of the evaporator outlet at the same rate it namely at the inlet and outlet of the evaporator (Tein, Teout) and con-
was pushed into the evaporator inlet. denser sections (Tcin, Tcout) and inside the thermosyphons (Tpipe).
At normal atmospheric conditions, water is incompressible. It can be seen that the trend is for the difference in temperature
Since the condenser section had been completely purged of air across the evaporator section to decrease as the mass flow rate of
prior to the start of any test, the assumption is that the mass flow the incoming hot air increases. In the condenser section the differ-
rate of water at the exit of the condenser must be the same as the ence in temperature increases with increasing mass flow rate on
mass flow rate of water flowing into the condenser. the evaporator side. From a thermodynamic perspective this is a

Fig. 10. Comparison between the three different meshes – Coarse, Medium and Fine.
1276 J. Ramos et al. / International Journal of Heat and Mass Transfer 102 (2016) 1267–1281

logical outcome as the increasing mass flow rate of air into the Eq. (31) is a correlation that allows the prediction of the average
evaporator increases the heat transfer coefficient which in turn working temperature of the thermosyphons according to the mass
increases the heat flow into the thermosyphon thus resulting in flow rate of air on the evaporator side and the difference in tem-
more heat being transferred to the water on the condenser section. perature across the entire heat exchanger. The expression only
From the experimental results, a regular pattern emerged applies to the heat exchanger under study with constant tempera-
between the thermosyphon working temperature and the overall ture and mass flow rate on the condenser side. The output of the
difference in temperature across the entire heat exchanger applied correlation to the inlet conditions of the heat exchanger
(Te,avg – Tc,avg). A plot of the average working temperature of the is presented in Fig. 13 which seems to follow the tendency of
thermosyphons against the overall difference in temperature Fig. 12 with reasonable accuracy.
between the evaporator and the condenser section for the range The heat transfer rate was found to be directly proportional to
of mass flow rates tested is displayed in Fig. 12. Data from the the inlet air mass flow rate and to the temperature of the flow at
trend lines, shown in their respective colour, allowed the creation the inlet, as can be seen in Fig. 14. The profile of the lines indicate
of Eq. (31) an expression able to predict the average working that if a higher mass flow rate was provided, the heat exchanger
temperature of the thermosyphons for each different working would be capable of transporting that much more heat, however,
conditions. the lines also start to become more flat as the mass flow rate
increases. The average maximum duty according to the data gath-
_ 0:5146 ðDT ov erall  40Þ þ 32 ð CÞ
T pipe ¼ 0:589m ð31Þ ered was 900 W maximum heat flux per heat pipe.

Fig. 11. Temperature distribution within the heat exchanger for inlet air temperatures ranging from 50 to 300 °C.
 J. Ramos et al. / International Journal of Heat and Mass Transfer 102 (2016) 1267–1281 1277

Fig. 12. Average working temperature of the thermosyphons for different overall Fig. 15. Relation between Qout and the overall thermal resistance.
DT at different mass flow rates.

Fig. 16. Effectiveness of the heat exchanger.

Fig. 13. Predicted average working temperature of the thermosyphons for different
overall DT at different mass flow rates.

0.20
0.18
0.16
0.14
Effectiveness

0.12
0.10
0.08
0.06
0.04
0.02
0.00
0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20
Number of Transfer Units
50 °C 100 °C 150 °C 200 °C 250 °C 300 °C

Fig. 17. Comparison between the effectiveness of the heat exchanger and the
Fig. 14. Total heat transfer rate of the heat exchanger. Number of transfer units.

The thermal resistance from Eq. (18) was also plotted against The plot clearly shows the thermal resistance is inversely propor-
the overall heat transfer rate as shown in Fig. 15. A higher differ- tional to the overall heat transfer coefficient.
ence in temperature produced a lower overall thermal resistance An uncertainty study was conducted in order to find the error
in the heat exchanger due to the more stable boiling regime inside propagation from the measurement instruments used in the exper-
the thermosyphon. Once again, the results obtained at 50 °C are the imental rig. It was observed that for all the tests, the uncertainty
oddball with a thermal resistance higher than 0.2 K/W. The other when determining the Qout is lower than 10%, which is an accept-
results have a lower value ranging between 0.05 and 0.1 K/W. able Fig. for engineering applications.
1278 J. Ramos et al. / International Journal of Heat and Mass Transfer 102 (2016) 1267–1281

Fig. 18. Percentage temperature difference at the outlets between the experimental test and the CFD simulation.

rate, the temperature difference between the inlet and the outlet
of the condenser section is reduced, reducing the overall effective-
ness. The plot is in agreement with Jouhara and Merchant [26] as
higher inlet temperatures result in higher effectiveness.
The effectiveness was also plotted against the Number of Trans-
fer Units (NTU) and it can be concluded that there is a quasi-linear
relation in agreement with the literature for this type of plots
[47,42] as seen in Fig. 17. The heat exchanger under study had a
small area of exposure and was incapable of transferring more than
0.2 transfer units, however, the trend shown in Fig. 17 displays a
linear increase. The effectiveness of the heat exchanger will
increase at a rate of 9:10 to the number of transfer units until an
effectiveness of 1 which at the current rate could be found at 1.2
NTU.
Fig. 19. Comparison between experimental and CFD results for the temperature at
the outlet of the evaporator at different mass flow rates.
5.3. Comparison of results

Regarding the temperature at the outlets, the CFD results com-

pared well to the experimental results, with a maximum difference
of 10% at the evaporator section and 15% at the condenser section
as can be seen in Fig. 18.
Fig. 19 portrays a more detailed comparison between the outlet
temperature of the evaporator section and the condenser section; a
good agreement is found on both as they tend towards the centre
of the graph. According to Fig. 19, it is found that the CFD simula-
tion is slightly over-estimating the temperature at the outlet at low
evaporator mass flow rates and under-estimating them as the mass
flow rates increase.
The heat transfer rate achieved in the CFD was compared with
that found from the experimental results and plotted in Fig. 20.
Following the same behaviour as Fig. 19, the CFD simulation
over-estimates the performance of the heat exchanger. However,
this is to be expected as in reality it is impossible to have an insu-
Fig. 20. Comparison between experimental and CFD results for the temperature at lation that is completely adiabatic; particularly with the geometry
the outlet of the condenser at different mass flow rates. of the heat exchanger under study. The CFD simulation also
appears to be slightly under-estimating the performance of the
5.2. Effectiveness heat exchanger at higher mass flow rates (10%); this may be due
to higher inlet turbulence in the experimental rig as a result of
Effectiveness is a variable that is an integral part of the an increased mass flow rate which is impossible to predict as an
Effectiveness-Number of Transfer Units (e-NTU) method and is a inlet boundary condition for the CFD model.
measure of a heat exchanger’s heat transfer potential. The effec- The difference in temperature is partially reflected in the heat
tiveness is a rate of the actual heat transfer of a heat exchanger transfer rate. Fig. 20 includes a comparison of the overall heat
to the maximum possible heat transfer rate. transfer rate between the experimental tests and CFD simulation.
Fig. 16 represents a plot of the effectiveness of the heat exchan- All the results are included within a 10% fluctuation showing a
ger against the mass flow rate of incoming air. A downward trend good correlation between the experimental data and the data
is observed in all of the results, as with the increased mass flow obtained from the simulations. The theoretical results were plotted
 J. Ramos et al. / International Journal of Heat and Mass Transfer 102 (2016) 1267–1281 1279

in a similar-type graphs in Figs. 21 and 22 and it was observed that

the CFD simulation clearly follows the same trend. This leads to the
conclusion that the inherent error is caused by the theoretical
expressions which may require some fine tuning in order to better
adapt to the problem at hand.

5.4. Results from CFD simulation

CFD was useful as the visualisation of the problem allowed the

identification of recirculation zones within the unique geometry of
the heat exchanger. Fig. 23 illustrates a vector plot of the velocity
within the evaporator section of the heat exchanger. The higher
velocities were found in the vicinity of the heat pipes. In this vector
plot, the recirculation zones are clearly identifiable in blue at the
top and bottom of the figure.
Fig. 23 presents the average velocity profile within the con-
denser section. The inlet temperature and mass flow rate were kept
constant in the condenser section allowing Fig. 24 to represent the
condenser section for all of the conditions tested. The blue areas Fig. 23. Vector velocity plot of the evaporator section.
represent re-circulation, common after the cylinders in parallel
arrangements.
A temperature contour of the evaporator is shown in Fig. 25; the
air is entering the evaporator at 300 °C from the left and leaves the
evaporator at approximately 255 °C. Temperatures lower than
250 °C are shown within the pipes. The variation of the tempera-
ture on the hot flow is particularly noticeable in this figure. The

Fig. 24. Velocity profile within the condenser section of the heat exchanger.

temperature of the thermosyphons at that particular height is also

Fig. 21. Comparison of the heat transfer rate between the experimental test and the shown in its respective location.
CFD model for different operating temperatures. Fig. 26 illustrates the temperature contour of the condenser sec-
tion for the same inlet conditions as Fig. 25. The water enters the
condenser at 10 °C and leaves the condenser at approximately
20 °C. The numbers in the pipes represent the temperatures of
the pipes at the condenser section. It can be observed that they
diverge from the temperatures shown in Fig. 25 and that is because
the pipes are not isothermal within as a normal thermosyphon
would be; their temperature varies along the y-axis due to the dif-
ferent fluid temperatures surrounding the pipes. The overall aver-
age temperature of the pipes, however, coincided with the
saturation temperature of the pipes.

5.5. Error propagation

An uncertainty study was conducted on the error propagation

from the measurement instruments used in the experimental rig
and the results are presented in Fig. 27.
It was observed that the smaller the difference in temperature
Fig. 22. Direct comparison between the experimental and the predicted heat between the inlet and the outlet temperatures, the higher the
transfer rate. uncertainty. This is particularly striking at 50 °C inlet air tempera-
1280 J. Ramos et al. / International Journal of Heat and Mass Transfer 102 (2016) 1267–1281

6. Conclusions

An experimental and a numerical investigation of a heat pipe-

based heat exchanger was successfully carried out and verified
through comparison with a real-world test experiment. The fol-
lowing conclusions were obtained according to the results:

– Higher temperatures and mass flow rates result in higher heat

transfer rates up to a limit of 900 W/pipe.
– Equations found in the literature seem to over-predict the
results at low evaporator temperatures and under-predict at
higher evaporator temperatures – an update is suggested.
– The modelling of the thermosyphons as solid bodies with a
conductivity extracted from an analytical study involving the
e-NTU method has been tried and tested in a CFD simulation
and the results proved very satisfactory.
– Good agreement between the experimental and numerical
Fig. 25. Average temperature contour of the air in the evaporator section at 300 °C
results for all temperatures was found for a wide range of flow
and 0.08 m/s.
conditions on the evaporator side; an average 5% temperature
difference was observed between the numerical and
experimental results in the evaporator section and 7% in the
condenser section.

Acknowledgements

The authors would like to thank the funding bodies, the

European Development Fund, and Econotherm UK Ltd. for their
in kind contribution.

References

[1] C. Haddad, C. Périlhon, A. Danlos, M.-X. François, G. Descombes, Some efficient

solutions to recover low and medium waste heat: competitiveness of the
thermoacoustic technology, Energy Procedia 50 (2014) 1056–1069.
[2] Y.H. Yau, The use of a double heat pipe heat exchanger system for reducing
energy consumption of treating ventilation air in an operating theatre-A full
Fig. 26. Average temperature contour of the water in the condenser for the year energy consumption model simulation, Energy Build. 40 (5) (2008) 917–
evaporator conditions of 300 °C and 0.08 m/s. 925.
[3] T.D. Swanson, G.C. Birur, NASA thermal control technologies for robotic
spacecraft, Appl. Therm. Eng. 23 (9) (June 2003) 1055–1065.
[4] J. Choi, M. Jeong, J. Yoo, M. Seo, A new CPU cooler design based on an active
cooling heatsink combined with heat pipes, Appl. Therm. Eng. 44 (November
2012) 50–56.
[5] H. Jouhara, R. Meskimmon, Heat pipe based thermal management systems for
energy-efficient data centres, Energy 77 (2014) 265–270.
[6] A. Faghri, Heat pipes: review, opportunities and challenges, Front. Heat Pipes 5
(1) (2014) 1–48.
[7] D. Reay, P. Kew, Heat Pipes: Theory, Design and Applications, Elsevier Limited,
2006.
[8] C. Han, L. Zou, Study on the heat transfer characteristics of a moderate-
temperature heat pipe heat exchanger, Int. J. Heat Mass Transfer 91 (2015)
302–310.
[9] H. Shabgard, M.J. Allen, N. Sharifi, S.P. Benn, A. Faghri, T.L. Bergman, Heat pipe
heat exchangers and heat sinks: Opportunities, challenges, applications,
analysis, and state of the art, Int. J. Heat Mass Transfer 89 (2015) 138–158.
[10] H. Jouhara, R. Meskimmon, Experimental investigatuion of wraparound loop
heat pipe heat exchanger used in energy efficient air handling units, Energy 35
(12) (2010) 4592–4599.
[11] A. Faghri, Review and Advances in Heat Pipe Science and Technology, J. Heat
Transfer 134 (12) (October 2012) 12.
Fig. 27. Error propagation for Qout. [12] Econotherm, ‘‘Econotherm – Waste Heat Recovery & Recycling Technology,”
2012. [Online]. Available: <http://www.econotherm.eu/index.html>.
[Accessed 1 2012].
[13] H. Hagens, F. Ganzevles, C. van der Geld, M. Grooten, Air heat exchangers with
ture where the uncertainty hovers the 300% due to the fact that the
long heat pipes: experiments and predictions, Appl. Therm. Eng. 27 (2007).
temperature variation is less than 1 °C. At 100 °C the DTc already 2646-2434.
fluctuates close to 2 °C and therefore the uncertainty propagation [14] S. Noie, Heat transfer characteristics of a two-phase closed thermosyphon,
Appl. Therm. Eng. 25 (2005) 495–506.
is reduced. For all the other tests the uncertainty when determin-
[15] A. Nuntaphan, J. Tiansuwan and T. Kiatsiriroat, Heat Transfer Coefficients of
ing the Qout is lower than 10% and stays within the 5% range, which Thermosyphon Heat Pipe at Medium Operating Temperature, Thailand,
is a more acceptable range for most engineering applications. 2001.
Overall the trend is for the error propagation to reduce as the [16] L. Vasiliev, Heat Pipes in Modern Heat Exchangers, Elsevier, Belarus Minsk,
Russia, 2005.
mass flow rate and the difference in temperatures increase, a trend [17] J.T. Cieslinski, A. Fiuk, Heat transfer characteristics of a two-phase
seen in all the tests. thermosyphon heat exchanger, Appl. Therm. Eng. 51 (1–2) (2012) 112–118.
 J. Ramos et al. / International Journal of Heat and Mass Transfer 102 (2016) 1267–1281 1281

[18] Y. Yau, M. Ahmadzadehtalatapeh, A review on the application of horizontal [32] H. Shabgard, B. Xiao, A. Faghri, R. Gupta, W. Weissman, Thermal characteristics
heat pipe heat exchangers in air conditioning systems in the tropics, Appl. of a closed thermosyphon under various filling conditions, Int. J. Heat Mass
Therm. Eng. 30 (2–3) (2010) 77–84. Transfer 70 (2014) 91–102.
[19] M. Ahmadzadehtalatapeh, An air-conditioning system performance [33] B. Hughes, H. Chaudhry, J. Calautit, Passive energy recovery from natural
enhancement by using heat pipe based heat recovery technology, Scientia ventilation air streams, Appl. Energy 113 (2014) 127–140.
Iranica 20 (2) (April 2013) 329–336. [34] B. Selma, M. Désilets, P. Proulx, Optimization of an industrial heat exchanger
[20] T. Jadhav, M. Lele, Theoretical energy saving analysis of air conditioning using an open-source CFD code, Appl. Therm. Eng. 69 (1–2) (2014) 241–250.
system using heat pipe heat exchanger for Indian climatic zones, Eng. Sci. [35] H. Peng, X. Ling, J. Li, Numerical simulation and experimental verification on
Technol. Int. J. 18 (4) (2015) 669–673. thermal performance of a novel fin-plate thermosyphon, Appl. Therm. Eng. 40
[21] F. Yang, X. Yuan, G. Lin, Waste heat recovery using heat pipe heat exchanger (2012) 181–188.
for heating automobile using exhaust gas, Appl. Therm. Eng. 23 (3) (2003) [36] J.K. Calautita, H.N. Chaudhrya, B.R. Hughesa, S.A. Ghanib, Comparison between
367–372. evaporative cooling and a heat pipe assisted thermal loop for a commercial wind
[22] Y. Wang, X. Han, Q. Liang, W. He, Z. Lang, Experimental investigation of the tower in hot and dry climatic conditions, Appl. Energy 101 (2013) 740–755.
thermal performance of a novel concentric condenser heat pipe array, Int. J. [37] H. Mroué, J. Ramos, L. Wrobel, H. Jouhara, Experimental and numerical
Heat Mass Transfer 82 (March 2015) 170–178. investigation of an air-to-water heat pipe-based heat exchanger, Appl. Therm.
[23] W. Srimuang, P. Amatachaya, A review of the applications of heat pipe heat Eng., 1359-4311 78 (2015) 339–350.
exchangers for heat recovery, Renew. Sustainable Energy Rev. 16 (6) (2012) [38] E. Azad, F. Geoola, A design procedure for gravity assisted heat pipe heat
4303–4315. exchanger, Heat Recovery Syst. 4 (2) (1984) 101–111.
[24] S. Noie, Investigation of thermal performance of an air-to-air thermosyphon [39] W. Kays, A. London, Compact Heat Exchanger Design, third ed., McGraw-Hill,
heat exchanger using e-NTU method, Appl. Therm. Eng. 26 (5–6) (2006) 559– New York, 1984.
567. [40] A. Lukitobudi, A. Akbarzadeh, P. Johnson, P. Hendy, Design, construction and
[25] J. Danielewicz, M.A. Sayegh, B. Sniechowska, M. Szulgowska-Zgrzywa, H. testing of a thermosyphon heat exchanger for medium temperature heat
Jouhara, Experimental and analytical performance investigation of air to air recovery in bakeries, Heat Recovery Syst. CHP 15 (5) (1995) 481–491.
two phase closed themrosyphon based heat exchangers, Energy 77 (December [41] A. Zhukauskas, Heat transfer from tubes in cross flow, in: J. Hartnett, T. Irvine
2014) 82–87. (Eds.), Advances in Heat Transfer, Academic Press, New York, 1972.
[26] H. Jouhara, H. Merchant, Experimental investigation of a thermosyphon based [42] Y.A. Çengel, in: Heat Transfer: A Practical Approach (Ed.), McGraw-Hill, 2002,
heat exchanger used in energy efficient air handling units, Energy (2012) 82– p. 7. 515.
89. [43] A. Faghri, Heat Pipe Science and Technology, Taylor & Francis, 1995.
[27] A. Alizadehdakhel, M. Rahimi, A. Alsairafi, CFD modeling of flow and heat [44] W. Rohsenow, J. Hartnett, Y. Cho, Handbook of Heat Transfer, third ed.,
transfer in a thermosyphon, Int. Commun. Heat Mass Transfer 37 (2010) 312– McGraw-Hill, 1998.
318. [45] J. Ramos, A. Chong and H. Jouhara, Numerical investigation of a cross flow air-
[28] B. Fadhl, L.C. Wrobel, H. Jouhara, Numerical modelling of the temperature to-water heat pipe-based heat exchanger used in waste heat recovery, SusTEM
distribution in a two-phase closed thermosyphon, Appl. Therm. Eng. 60 (1–2) 2015, Newcastle, 2015.
(2013) 122–131. [46] W.M. Rohsenow, A method of correlating heat-transfer data for surface boiling
[29] H. Jouhara, B. Fadhl, L.C. Wrobel, Three-dimensional CFD simulation of geyser of liquids, Trans. ASME 74 (1952) 969–976.
boiling in a two-phase closed thermosyphon, Int. J. Hydrogen Energy (2016). In [47] F. Incropera, D. DeWitt, Fundamentals of Heat and Mass Transfer, fourth ed.,
press. John Wiley & Sons, New York, 1996.
[30] K. Negishi, T. Sawada, Heat transfer performance of an inclined two-phase [48] W. McAdams, Heat Transmission, McGraw Hill, New York, 1954.
closed thermosyphon, Int. J. Heat Mass Transfer. 26 (1983) 1207–1213. [49] ANSYS, Ansys Fluent 14.5 Help, 2012–2014. [Online]. [Accessed 2015].
[31] I. Khazaee, R. Hosseini, S. Noie, Experimental investigation of effective [50] K. Ekambara, R. Sanders, K. Nandakumar, J. Masliyah, CFD simulation of bubbly
parameters and correlation of geyser boiling in a two-phase closed two-phase flow in horizontal pipes, Chem. Eng. J. 144 (2) (2008) 277–288.
thermosyphon, Appl. Therm. Eng. 30 (5) (2010) 406–412. [51] Fluent Inc., FLUENT 6.3 User’s Guide, 2014.