International Journal of Heat and Mass Transfer 128 (2019) 208–216

Contents lists available at ScienceDirect

International Journal of Heat and Mass Transfer

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

Theoretical and experimental studies of heat transfer characteristics of a

single-phase natural circulation mini-loop with end heat exchangers
Haojie Cheng, Haiyan Lei, Long Zeng, Chuanshan Dai ⇑
Key Laboratory of Efficient Utilization of Low and Medium Grade Energy, MOE, School of Mechanical Engineering, Tianjin University, 92 Weijin Road, Nankai District,
Tianjin 300072, China

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

Article history: In the present work, the heat transfer and fluid flow characteristics of a single-phase natural circulation
Received 6 March 2018 loop (SPNCL) with both the heating and the cooling ends are theoretically and experimentally studied.
Received in revised form 28 August 2018 Distilled water is chosen as the working fluid circulating in a rectangular loop that the top and bottom
Accepted 31 August 2018
sides are respectively cooled and heated by 245 mm tube-in-tube heat exchangers. The height and diam-
eter of the mini-loop are 250 mm and 4 mm, respectively. Both analytical and experimental results are
obtained by varying the heating fluid temperatures from 30 °C to 60 °C but fixing cooling fluid temper-
Keywords:
ature of 10 °C. Based on the experimental data, a Nu
Re correlation is obtained and applied into the
Natural circulation
Mini-loop
one-dimensional mathematical model. A good agreement between the experimental and theoretical
Heat exchangers results with a new correlation can be observed. Experimental results show that stable flow can be
Optimal ratio reached for the cases with different Th. The start-up time of natural circulation from quiescent state short-
ens with the increase of Th. The Reynolds number and heat transfer rate at steady state are proportional to
the heating fluid temperature Th. Based on the proposed mathematical model, an optimal ratio of the hea-
ter length to the loop height can be reached when the total length and diameter of the mini-loop keep
constants.
Ó 2018 Elsevier Ltd. All rights reserved.

1. Introduction convective heat transfer boundary used in the cooler. A generalized

Natural circulation loop (NCL) is commonly used as an energy
transfer instrument in nuclear reactor [1], solar heater [2],
computer cooling [3], etc., due to its simple construction, none
power input from pump. Working fluid circulation in a single-
phase natural circulation loop (SPNCL) is accomplished by the
interaction between buoyancy force and frictional force. In a
SPNCL, the buoyancy force is generated by the density difference
due to temperature gradient. For the two-phase natural circulation
in a loop (TPNCL), the circulation flow is based on the evaporation
and condensation of working fluid. Normally, the NCL consists of a
heating section (evaporator), a cooling section (condenser) and two
adiabatic sections.
A great deal of experimental studies, theoretical analyses and
numerical simulations on steady state, transient state and stability
behavior of the SPNCL and TPNCL have been carried out. Different
working fluids in the loops and thermal boundary conditions have
been considered in these studies. The commonly used thermal
boundary conditions are constant heat flux in the heater and
⇑ Corresponding author.
E-mail address: csdai@tju.edu.cn (C. Dai).
mathematical model of a SPNCL was presented by Thomas [4], and
an agreement between their mathematical model and those exper-
iment data had been confirmed. The steady state flow equation in a
SPNCL with uniform or non-uniform diameter was derived by Vija-
yan [5], which could be expressed as Ress = C[Grm/NG]r. Meanwhile,
a good agreement between dimensionless correlation and experi-
mental data was verified. Misale et al. [6] reported an experimental
investigation on a natural circulation mini-loop with different
heating power and loop inclination angle. The loop inclination
angle was considered in the modified Grashof number. Further-
more, Garibaldi and Misale [7] performed an experimental study
with water and FC43 as the working fluids, respectively, in two
kinds of mini-loops. It was found that the thermal-hydraulic
behaviors for both the mini-loops are always stable. Meanwhile,
an experimental investigation on thermal-hydraulic performance
in a mini-loop with distilled water and Al2O3 nanofluid was
presented by Miasle et al. [8]. It was observed that the thermal
performances for the mini-loops between distilled water and
Al2O3 nanofluid were similar. A three-dimensional CFD transient
simulation on the natural circulation in a mini-loop was performed
by Wang et al. [9]. The experimental data conducted by Misale
et al. [6] was used to validate the proposed CFD model and an

https://doi.org/10.1016/j.ijheatmasstransfer.2018.08.136
0017-9310/Ó 2018 Elsevier Ltd. All rights reserved.
 H. Cheng et al. / International Journal of Heat and Mass Transfer 128 (2019) 208–216
Nomenclature
A cross sectional area (m2)
Ar ratio
Cp specific heat capacity (J/kgK)
D diameter (m)
f friction factor
g gravitational acceleration (m/s2)
Gr Grashof number
h heat transfer coefficient (W/m2k)
H the loop height (m)
K local resistance coefficient
L length (m)
NG non-dimensional geometry parameter
Nu Nusselt number
P loop perimeter (m)
Pe Pelect number
Q heat transfer rate (W)
Re Reynolds number
s coordinate along the flow direction (m)
St Stanton number
T temperature (K)
t time (s)
U overall heat transfer coefficient (W/m2K)
agreement between those experiment and simulation could be
concluded. Compared with the mini-loop. the stability behavior
for a SPNCL with a relatively large geometry scale was experimen-
tally investigated by Nayak et al. by using water and Al2O3 nanoflu-
ids with various concentrations [10]. It was found that the use of
nanofluids can suppress the flow instability and enhance the natu-
ral circulation flow rate. A unified model for a natural circulation
loop with different geometries and boundary condition was pro-
posed by Basu et al. [11]. An analytical study related to the entropy
generation in a SPNCL was performed by Goudarzi and Talebi [12]
where entropy generation formulas in each section were obtained.
Saha et al. [13] experimentally and numerically investigated the
thermal-hydraulic behavior of a SPNCL under low and moderate
heating power with a variable ambient temperature. The periodic
oscillations were observed at a moderate heating power. The sta-
bility behavior for the SPNCLs with four kinds of heater and cooler
orientations was experimentally investigated by Vijayan et al. [14].
The maximum flow rate and the instability were observed for the
orientation that both heater and cooler were horizontal. A three-
dimensional CFD simulation for the instability in four different
configurations of SPNCLs was performed by Kudariyawar et al.
[15], which showed that for the configuration of horizontal heater
and horizontal cooler, the steady state could not be achieved and
uni-directional and bi-directional oscillation were observed at dif-
ferent heating powers. The validity for applying Boussinesq
approximation to simulate a natural circulation loop was pre-
sented by Krishnani and Basu [16]. Compared with the complete
variation of all relevant thermos-physical properties, Boussinesq
model was only valid in transient simulation at low heating power.
An experimental investigation of a SPNCL with molten salt as the
working fluid in a wide range of input powers was presented by
Srivastava et al. [17]. The start-up of natural circulation and heat
loss in heat sink had been studied. The enhanced heat transfer per-
formance of water-based nanocapsules in a SPNCL was experimen-
tally studied by Ho et al. [18]. It was found that the heat transfer
effectiveness increases with the increase of the heating power.
In addition, a single-phase natural circulation loop (SPNCL) can
be found in many practical applications where a convective bound-
ary condition was used in the heater of a SPNCL. Employing the
number of heat transfer unit, Rao et al. [19–22] investigated the
209
V velocity (m/s)
W mass flow rate (kg/s)
Z coordinate in the direction of gravity (m)
Greek symbols
h dimensional temperature
q density (kg/m3)
k heat conductivity coefficient (W/mK)
l viscosity (Pas)
b volumetric expansion coefficient (K1)
Subscripts
c cooler
cl cold leg
h heater
hl hot leg
hy hydraulic diameter
i inner
o outer
r reference
ss steady state
effects of heat exchangers performance parameters, different exci-
tations and core capacitance on the steady state, stability behavior
and dynamics performance of a SPNCL. A mathematic model for
calculating the heat transfer performance of the natural circulation
loop at steady state with CO2 as the working fluid was developed
by Kumar and Gopal [23]. By using the CFD code, FLUENT, Yadav
et al. [24] numerically investigated the steady state performance
of the CO2 natural circulation loop with end heat exchangers at
subcritical and supercritical states. It was found that the heat
transfer rate in the supercritical region was higher than that in
the subcritical region. A transient simulation of subcritical/super-
critical CO2 in a SPNCL with end heat exchangers was carried out
by Yadav [25]. The effects of different operating pressures and tem-
peratures, tilt angle and water mass flow rate on the thermal-
hydraulic characteristics of a SPNCL were investigated. Cheng
et al. [26] carried out a three-dimensional numerical simulation
of a SPNCL with end heat exchangers and analyzed the entropy
generation and stability of natural circulation.
Different from SPNCLs, TPNCLs have complex structure and
stronger instability [27] due to the boiling and condensation, but
higher heat transfer capacity. Pandey and Singh [28] used linear
stability method to analyze the chaos behaviors of two-phase flow
in a loop. A novel stability boundary, named Type A, was found in
this study, which was different from the Ledinegg stability bound-
ary. Chen et al. [29] presented an experimental investigation of a
TPNCL and flow characteristics were observed through the moni-
toring the pressure variation. Flow reversal was found due to the
onset of critical heat flux and the mechanism of flow reversal
was compared between macro-channel and mirco-channel. Bod-
jona et al. [30] built a mathematical model of a TPNCL and used
finite volume method to carry out a numerical simulation. Result
showed that the density, temperature, pressure of loop fluid chan-
ged considerably with heating power. An experimental investiga-
tion of steady state performance in a TPNCL with multiple
channels was performed by Bhusare [31] and it was found that
two-phase pressure drop decreased with the increasing superficial
gas velocity.
As a summary about the previous studies on the single-phase
natural circulation loops, it can be concluded that the experimental
studies are still limited and most of which had a heating section by
210 H. Cheng et al. / International Journal of Heat and Mass Transfer 128 (2019) 208–216
Fig. 1. The schematic diagram of a SPNCL.
electricity rather than by hot fluids. An optimal geometry design at
some confinements is seldom discussed so far. Therefore, the the-
oretical and experimental investigations for a SPNCL with both the
heating and cooling ends are reported in the present paper. The
experimental data is compared with the mathematical model
and a Nusselt number correlation is proposed. Besides, an optimal
ratio Ar for higher Reynolds number and heat transfer rate can be
obtained based on the model with the new correlation.
2. The configuration of a SPNCL
A single-phase natural circulation loop (SPNCL) with two tube-
in-tube heat exchangers at the heating and cooling sections is
shown in Fig. 1, which is fixed at a vertical plate. Distilled water
is chosen as the working fluid in both the SPNCL and heat exchang-
ers and the thermophysical properties are shown in Table 1 with
thermal expansion coefficient and dynamic viscosity chosen based
on the reference temperature Tr [32], which equals to the average
value of the heating and cooling fluids temperature, (Th + Tc)/2. The
rectangular SPNCL is composed of two tube-in-tube heat exchang-
ers, two vertical adiabatic pipes and four elbow pipes. The bottom
heat exchanger is used as a heater and the upper one is a cooler.
The uniform inner diameter d of the mini-loop is 4 mm with a wall
thickness d of 2 mm and the inner diameter D of the annular tubes
is 16 mm. The length of heating and cooling section, Lh and Lc is
245 mm with L1 of 40 mm and the height of the mini-loop, H is
250 mm. This rectangular loop is made of organic glass, which
has a thermal conductivity of 0.19 W/(mK). The water inside the
loop once circulated is to be continuously heated in the heater
and cooled in the cooler.
Table 1
Thermo-physical properties of distilled water.
q (kg/m3) Cp (J/kgk) Average b(1/K)
temperature
Water 997.1 4182 20 °C 2.0  104
25 °C 2.5  104
30 °C 3.0  104
35 °C 4.0  104
2.1. Experimental system
As shown in Fig. 2, the experimental system consisted of a ver-
tically placed mini-loop, two thermostatic tanks, data acquisition
system, an electronic scale, several valves and pipes. Four cali-
brated T-type thermocouples (±0.1 °C) with a diameter of 0.5 mm
were used to measure the local temperature, which were placed
at the cross-sections of the middle of two vertical pipes. Two ther-
mocouples, marked as T1 and T2 from left to right, were radially
placed at a distance of d/4 from the wall surface of the left leg.
The same arrangement was applied for other two thermocouples
at the right leg, which were also marked as T3 and T4, respectively.
Meanwhile, two thermostatic water tanks were used as the supply
of heat and cold source, which could achieve the precise control of
the water temperature with a temperature fluctuation of ±0.01 °C.
The mass flow rate of the secondary fluids was adjusted through
altering the pump power of water bath and valves, which was fixed
at about 14 g/s. The experimental data for loop fluid temperature
was acquired and stored over a period of 1500 s by the high-
speed data acquisition system with time interval of 0.5 s. The mass
flow rate of secondary fluids was obtained by mass weighting
method. A precise electronic scale was used to measure the mass
flow rate of heating or cooling water with an uncertainty of 0.5 g.
In our experiments, both the bottom and top end heat exchangers
and vertical pipes were wrapped with about 30 mm thick good
thermal insulation material to reduce the heat loss and the heat
loss was ignored in the following heat balance calculation. Fig. 3
shows the photograph of experimental system, where a SPNCL
made of organic glass was fixed at a vertical plate fully filled with
distilled water. The mini-loop was equipped with thermal insula-
tion material.
Each experimental run was started from a quiescent state of
room temperature. The initial temperature of the loop fluid and
the secondary fluids was set to about 19 °C. Before the start-up
l(Pas) of each experiment, the mass flow rate of secondary fluids was
adjusted by changing valve openness and measured repeatedly
0.001 and then the temperature of each thermostatic water tanks was
0.00089 set to the fixed value. In this experiment, the cooling water tem-
0.000797
perature was fixed at 10 °C while the temperature of heating water
0.000719
was altered from 30 °C to 60 °C with a temperature step of 10 °C.
 H. Cheng et al. / International Journal of Heat and Mass Transfer 128 (2019) 208–216 211

Fig. 2. Schematic diagram of the experimental set-up.

Each run was started through simultaneous start-up of data acqui- [33] through a root-sum-square method, which is expressed as
sition system, two ball valves. The run time was continued for follows
1200 s until switching off the circulating pumps and data acquisi-
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
tion system. u  2
uX @R
In this experiment, the only measurable parameter was the @R ¼ t @xi ð1Þ
local temperature of loop fluid. Other experimental results such n
@xi
as Reynolds number, Nusselt number and heat transfer rate cannot
be obtained directly, which have to be calculated through several The uncertainty of Reynolds number, Nusselt number and heat
equations. The uncertainty of these parameters was calculated transfer rate were 5.6%, 8.9% and 10.6%, respectively.

2.2. Mathematical model

In order to simplify the current mathematical model, several

assumptions have been made as follows.

(1) The model is one-dimensional. In other words, there is no

variation of temperature and velocity in the circular section
area of the SPNCL.
(2) Boussinesq approximation is employed to estimate density
variation with the temperature gradient accounting for the
buoyancy force effect in the momentum equation, while
density maintains constant in other terms. This approxima-
tion is given as q = q0[1  b(T  T0)] where b represents the
volumetric thermal expansion coefficient of the working
fluid.
(3) The flow regime in the SPNCL is supposed to be laminar.
(4) In the energy equation, the effects of viscous dissipation and
axial heat conduction in the working fluid and in the solid
wall are not considered.
(5) The temperature difference of the heating and cooling fluids,
between the inlet and the outlet, respectively, at the heater
and the cooler, is tiny so that the temperature of heating and
cooling fluid can be considered to be constant.
(6) In this model, the length of horizontal adiabatic section L1 is
supposed to be a constant.
Fig. 3. The photographs of experimental system.
212 H. Cheng et al. / International Journal of Heat and Mass Transfer 128 (2019) 208–216

2.2.1. Governing equations at steady state @h @h 4Lt

þw ¼ St m h cooler ð12Þ
The one-dimensional form of continuity, integral momentum @s @S d
and energy conservation equations are, respectively, shown as A modified Stanton number is defined as follows, which is a
flows: function of Reynolds number
@ q 1 @ðWÞ 1
þ ¼0 ð2Þ Stm ¼ ð13Þ
@t A @s RlC p Ress
d
DSto
þ 1
St i
þ d
X I 4  
4
Li dW W2 X fLi  
¼ q0 bg Tdz  þ K ð3Þ d D
i¼1
A dt 2q0 A2 i¼1 d R¼ ln ð14Þ
2k d
@T W @T UðT h  TÞp By eliminating the time-dependent term in the governing equa-
þ ¼ heater ð4Þ
@t q0 A @s qC p A tions, a solution at steady state can be obtained. At the same time,
the dimensionless mass flow rate, w, will become unity due to def-
@T W @T inition. In this case, the governing equations reduce to
þ ¼ 0 pipes ð5Þ
@t q0 A @s 4  
Grm 1X Li
¼ f þK ð15Þ
@T W @T UðT c  TÞp Re2ss 2 i¼1 d
þ ¼ cooler ð6Þ
@t q0 A @s qC p A
@h 4Lt
In the above equations, W is the mass flow rate, which is ¼ Stm ð1  hÞ heater ð16Þ
@S d
defined as W = q0Au and A is the cross-section area of the loop.
In addition, f and K denote the frictional coefficient and local resis- @h
¼ 0 pipes ð17Þ
tance coefficient, respectively. d represents the diameter of the @S
loop while Li is the length in ith section. In the energy conservation
equations, U represents the overall heat transfer coefficient, which @h 4Lt
¼ St m h cooler ð18Þ
is defined as @S D
1 The dimensionless temperature distributions along the loop are
U¼   ð7Þ  
d
ho D
þ d
2k
ln Dd þ h1 4Lt Lh
i
hh ðSÞ ¼ ðhcl  1Þ exp  Stm S þ 1 0 6 S 6 ð19Þ
d Lt
where ho and hi are, respectively, the local convective heat transfer
coefficient between the external fluid and the outer wall surface,   
4Lt 1 1 1 Lh
the internal fluid and the inner wall surface. Furthermore, k is heat hc ðSÞ ¼ hhl exp  St m S  6S6 þ ð20Þ
d 2 2 2 Lt
conductivity of the wall. To make the issue concise, some dimen-
sionless parameters are defined below so that those governing On account of the closed relationship of temperature along the
equations can be rewritten in a dimensionless form: loop, the minimum dimensionless temperature, hcl and the maxi-
W T  Tc qV z s mum, hhl, in the cold and hot adiabatic vertical sections, respec-
w¼ ; h¼ ; tr ¼ 0 ; Z ¼ ; S ¼ ; DT r ¼ T hl  T cl tively, can be obtained.
W ss Th  Tc W ss H Lt
   
exp  4Ldc St m  exp  4ðLcdþLh Þ St m
q20 bg DT r d2 H W ss d Nu hA hcl ¼   ð21Þ
Grm ¼ ; Ress ¼ ; St ¼ ¼ ð8Þ 1  exp  4ðLcdþLh Þ St m
l2 lA Ress Pri W ss C p
 
Here, Wss represents the steady-state mass flow rate. Th and Tc
1  exp  4LDh Stm
denote the temperatures of hot fluid in the heater and cold fluid hhl ¼   ð22Þ
in the cooler, respectively. A modified Grashof number, Grm is pro- 1  exp  4ðLcDþLh Þ St m
posed in the paper. In this definition of Grm, the reference temper-
ature difference DTr and the characteristic length are different where Lc and Lh represent the length of heater and cooler in the
from those used in the previous studies. It is considered that the loop. Since there are four parts of horizontal pipe under the thermal
defined Grashof number should reflect inherently the intensity of isolation with a fixed length L1, the effective length L should be cor-
natural convection happened inside the SPNCL. Therefore, the max- rected as Lt-4L1.
imum and minimum mean temperatures, Thl and Tcl along the Eqs. (19) and (20) can be rewritten as
SPNCL are chosen for accounting the reference temperature differ-    
G Ar G Ar
ence. In addition, since the buoyancy force exists on in the vertical exp 2 N1þAr St m  exp 4 N1þAr St m
hcl ¼   ð23Þ
sections in the gravity field, the loop height, H should be included in 1  exp 4 N1þAr G Ar
St m
the Grashof number. As a consequence, the dimensionless momen-
tum and energy conservation equations can be rewritten as  
G Ar
4   1  exp 2 N1þAr Stm
dw Gr m w2 X Li hhl ¼   ð24Þ
¼ 2  f þK ð9Þ G Ar
1  exp 4 N1þAr Stm
ds Ress 2 i¼1 d

where NG denotes the ratio between the length L and diameter of

@h @h 4Lt
þw ¼ Stm ð1  hÞ heater ð10Þ the loop and Ar is the ratio between the length of heating section
@s @S d and height, which are given by
@h @h L Lh
þw ¼ 0 pipes ð11Þ NG ¼ Ar ¼ ð25Þ
@s @S D H
 H. Cheng et al. / International Journal of Heat and Mass Transfer 128 (2019) 208–216 213

2.2.2. Empirical correlations of heat transfer and fluid flow used in the
SPNCL
In order to get a quantitative solution at a given set of parame-
ters, the correlation of heat transfer with the fluid inside the SPNCL
has to be selected. Since that there are no heat transfer correlations
available yet specifically proposed for the case of a SPNCL, a few
empirical correlations calculating the frictional coefficient f in the
circular tube, the local flow resistance coefficient K through a right
angle curved pipe and the averaged Nusselt numbers Nu inside a
tube and annular are selected accordingly.
Due to the small temperature difference between the heat sink
and the heat source in the present analysis, the correlation for cal-
culating the frictional coefficient inside the SPNCL is chosen as that
in the laminar flow regime as follows
64
f ¼ Re < 2300
Re
The local flow resistance coefficient, K through a curved pipe
with a right angle is calculated by,
 3:5
d
K ¼ 0:131 þ 0:163
R
3.1. Experimental results
With regard to the mini-loop in this experiment, the only avail-
able experimental data is the temperature and temperature differ-
ence of the two legs since the lower pressure drop is difficult to be
measured. The temporal variation of local temperature T1, T2, T3
and T4 is shown in Fig. 4 for two cases Th = 30 °C and Th = 50 °C
when the mass flow rate of both the heating and cooling streams
is about 17 g/s (the Reynold number is about 1014). It is easy to
observe that the temperature T1 and T2 is always higher than T3
and T4 for the two cases. In other words, the left leg is the hot
leg while the right leg is cold leg and the stable clockwise circula-
tion flow is generated in the mini-loop. Meanwhile, the similar
temperature curves for T1 and T2 (T3 and T4) can be seen which
means that there is the uniform temperature distribution at the
ð26Þ cross-section of the middle of the two legs. Generally, at the begin-
ning of the supply of heating and cooling fluids, the loop fluid is
under the quiescent state and the main heat transfer mechanism
is heat conduction. The local natural circulation at the radial direc-
tion may exist [9]. After the quiescent state, since the buoyancy
ð27Þ

where R is the curvature radius of the curved pipe.

The heat transfer correlations [13,23] inside a tube and an
annular are successively given as below

Nui ¼ 3:66 ð28Þ

"  0:5 #
D 0:19ðPe  Dhy =Lh Þ0:8
Nuo ¼ Nu1 þ 1 þ 0:14 ð29Þ
d 1 þ 0:117ðPe  Dhy =Lh Þ0:467

where Nu1 denotes the Nusselt number of fully developed flow in

an annular, which is defined as
 0:5
D
Nu1 ¼ 3:66 þ 1:2 ð30Þ
d
In addition, Pe is the Peclet number, the product of Prandtl
number and Reynolds number. D and Dhy represent the outer
diameter and the hydraulic diameter of tube-in-tube heat exchan-
ger, respectively.
It can be seen from Eqs. (21) and (22), that the dimensionless
temperature distribution along the loop is a function of Reynolds
number Ress at steady state. Therefore, in the dimensionless
momentum equation, the modified Grashof number depends on
the steady state Reynolds number. In this case, an iteration method
accompanied by several empirical correlations is utilized to solve
the energy and momentum equations. In the present study, the
heat transfer rate of a SPNCL with convective heating and cooling
boundaries can be defined as

Rei lC p pDDT r
Q¼ ð31Þ
4

3. Results and discussions

In our study, the effect of heating fluid temperature Th from

30 °C to 60 °C on the thermos-hydraulic characteristics of the
SPNCL was experimentally investigated with the same mass flow
rate Wc and Wh of 17 g/s. The steady state experiment data was
analyzed and compared with the one-dimensional mathematical
model. Based on the proposed mathematical model, the non-
dimensional temperature distribution along the mini-loop could
be determined and the further optimal ratio between the length
of heater (cooler) and the height of the mini-loop was obtained Fig. 4. Temporal development of local temperature T1, T2, T3 and T4 at (a) Th = 30 °C
subjected to the fixed total length and diameter. and (b) Th = 50 °C.
214 H. Cheng et al. / International Journal of Heat and Mass Transfer 128 (2019) 208–216

force in the left leg is higher than that in the right, the loop fluid
begins to circulate clockwise along the mini-loop and the local
temperature T1
T4 also starts to rise or fall at the same time. After
a period of small oscillation, the local temperature tends to become
stable. Fig. 5 shows the variation of temperature difference
between the two legs with Th = 30–60 °C. As the heating fluid tem-
perature Th increases, it is shown that the time from static state to
the start-up of natural circulation is shortened. Since there is a lar-
ger thermal resistance of the mini-loop, the maximum tempera-
ture difference for Th = 50 °C is only 25 °C. As for the steady state,
it is obvious that the temperature difference between two legs
increases with the increase of Th.

3.2. Steady state data analysis

One-dimensional mathematical model is proposed in our study

where two empirical equations to calculate the convective heat
transfer coefficient hi and ho are used. It is known that for the sec-
ondary fluids in the annular tubes, the fluid flow is forced convec-
tion driven by circulating pump. As a result, the empirical equation Fig. 6. Variation of Nusselt number for both heater and cooler with corresponding
Reynolds number.
is suitable for this case. But for the loop fluid, the fluid flow is nat-
ural circulation caused by temperature difference. The constant
Nusselt number 3.66 is applied to the fully developed laminar
forced flow inside a circular tube subjected to a constant wall tem-
perature and may be inappropriate for the natural circulation.
Therefore, the theoretical calculation is carried out at first with
the use of Nu = 3.66. From Fig. 7, it is clearly shown that the use
of Nu = 3.66 can cause a greater computation error and the maxi-
mum errors of Reynolds number and heat transfer rate in the
mini-loop are 34.5% and 83.9%, respectively, when Th = 30 °C.
Therefore, the correlation of Nui
Ress needs to be obtained based
on the experimental data.
Fig. 6 shows the Nusselt numbers, Nui,c and Nui,h, of loop fluid in
the heater and cooler with corresponding Reynolds number at
steady state. The value of Nui,c and Nui,h is approximately 0.62–
1.45 and 0.63–1.13, respectively, which are far lower than 3.66.
When Th = 30 °C, both the Nusselt numbers are approximately
equal. As the Th increases, it is found that the Nusselt number
Nui,c gradually becomes a little lager than Nui,h due to lower loga-
rithmic mean temperature difference. Obviously, both the Nusselt
numbers of natural circulation in the heater and cooler increase
with the increase of Reynolds number. For the sake of theoretical
calculation, the average Nusselt number correlation for both Nui,c
and Nui,h is put forward in Fig. 6, which is expressed as

Fig. 5. Temporal development of temperature difference between two legs for

Fig. 7. Comparison between experimental data and theoretical calculation with
Th = 30, 40, 50, 60 °C.
two models.
 H. Cheng et al. / International Journal of Heat and Mass Transfer 128 (2019) 208–216 215

Nu ¼ 0:2702Re0:4713 0 < Re < 2300 ð32Þ

Based on the proposed Nusselt number correlation, the theoret-
ical solutions of Reynolds number and heat transfer rate at steady
state are obtained. Fig. 7(a) and (b) show the comparison of exper-
imental results and two models with different Nusselt number cor-
relations under different Th from 30 °C to 60 °C. Clearly, the new
Nu
Re correlation is suitable for the proposed model and a good
agreement between theoretical curves and experimental data can
be seen. The maximum errors of Reynolds number and heat trans-
fer rate decline to 9.7% and 18.6% when Th = 30 °C.
The value of Ress is only 10–42.5, indicating that the loop fluid
velocity is of millimeter or centimeter per second. As known, the
value of Reynold number is proportional to the heating fluid tem-
perature Th. Different with the case of constant heating power, the
heat transfer rate transmitted by natural circulation is unknown
and need to be calculated. The value of Q is approximately 1.7–
6.2 W. Similarly, it increases with the increase of Th.
When using the one-dimensional mathematic model with pro-
posed Nu
Re correlation, the dimensionless temperature distri-
bution under steady state along the mini-loop can be obtained,
which is shown in Fig. 8 with Th from 30 °C to 60 °C. It is shown
that as the heating fluid temperature Th increases, the dimension-
less temperature difference of the loop decreases.

3.3. The optimal geometry ratio

The optimal geometry ratio is investigated subjected to a fixing

total length Lt based on the mathematic model. The ratio Ar
between the length of heater Lh and the height H has a great influ-
ence on the flow and heat transfer characteristics. It is well known
that the natural circulation is driven by the static pressure differ-
ence between the two vertical pipes, which is closely related to
the height of the SPNCL. When the height of the loop increases,
the static pressure difference between the two vertical pipes
increases so that the Reynolds number or the mass flow rate will
increase accordingly. However, with the increase of the height,
the length of the heater and the cooler could decrease due to a
fixed total length of the SPNCL that resulting in a reduction of
the mass flow rate. Therefore, the optimal value of the ratio, Ar,
has to be discussed based on the proposed mathematical in the
present work.
Fig. 9(a) and (b) show the effect of the ratio Ar on the steady
state Reynolds number and heat transfer rate for two cases
Fig. 9. The variation of steady state Reynolds number and heat transfer rate with
different Ar at (a) Th = 30 °C and (b) Th = 50 °C.
Th = 30 °C and Th = 50 °C subjected to a fixing length Lt and diame-
ter D. For both cases, the steady state Reynolds number initially
increases with the increase of Ar, reaches a maximum and then
starts decreasing. It is clearly observed from Fig. 9 that when Ar
is approximately equal to 1.0, the maximum Reynolds number is
achieved at Th = 30 °C and 50 °C. The reason of maximum Ress at
Ar = 1.0 is probably that the loop height H and heater length Lh
have the approximate contribution to the steady state Reynolds
number, which is a linearly function of Grashof number. In addi-
tion, as the Ar increases, the heat transfer rate reaches a peak at
Ar = 2.5 when Th = 30 °C and 50 °C, and then keeps almost a con-
stant. In addition, it is clearly shown that the maximum Reynolds
number Ress and heat transfer rate Q do not occur at a same Ar in
general since the heat transfer rate is dependent on the H and Lh
and Lh has a larger contribution to the heat transfer rate Q from
Eq. (31). Therefore, the maximum heat transfer rate could occur
at a larger Ar.

4. Conclusion

Fig. 8. Dimensionless temperature distribution along the SPNCL with different Th.
In the present work, the heat transfer and fluid flow character-
istics of a SPNCL with tube-in-tube heat exchangers at both the
heating and the cooling ends are theoretically and experimentally
studied. Distilled water is used as the working fluid in the mini-
loop and secondary fluids. Experimental runs are carried out with
216 H. Cheng et al. / International Journal of Heat and Mass Transfer 128 (2019) 208–216
a fixing cooling fluid temperature of 10 °C and varying heating fluid
temperature from 30 °C to 60 °C when the mass flow rate of the
both secondary fluids is about 17 g/s. A one-dimensional analytical
model of a SPNCL at steady state is developed. The analytical and
the experimental results at steady state show a good agreement
when using the new proposed correlation. Several major conclu-
sions can be drawn as follows.
(1) From the temporal development of the local temperature at
the vertical leg, it can be seen that the stable flow in a mini-
loop can be reached for all the cases in this experiment with
clockwise natural circulation. Clearly, the start-up time from
static state decreases with the increase of Th.
(2) As for the steady state results, heat transfer rate and Rey-
nolds number are proportional to the Th. The maximum Ress
and Q are 42.5 and 6.2 W, respectively.
(3) When using the correlation of Nu = 3.66 in the mathematic
model, a greater error between theoretical results and
experimental data can be observed. Therefore, a novel Nus-
selt number correlation of natural circulation has been
acquired, which is Nu = 0.2702Re0.4713, far lower than 3.66
in this experiment. A good agreement can be seen with the
use of new correlation.
(4) Based on the proposed mathematical model, an optimal
ratio between the length of heater (cooler) Lh and the height
H is presented with a purpose of higher Reynolds number
and heat transfer rate.
Conflict of interest statement
We declare that we have no financial and personal relationships
with other people or organizations that can inappropriately influ-
ence our work, there is no professional or other personal interest
of any nature or kind in any product, service and/or company that
could be construed as influencing the position presented in, or the
review of, the manuscript entitled, ‘‘Theoretical and experimental
studies of heat transfer characteristics of a single-phase natural cir-
culation loop with end heat exchangers”
Acknowledgements
This work is financially supported by the National Science
Foundation of China (Grant No. 41574176 and 41672234).
References
[1] M. Wang, H. Zhao, Y. Zhang, G. Su, W. Tian, S. Qiu, Research on the designed
emergency passive residual heat removal system during the station blackout
scenario for CPR1000, Ann. Nucl. Energy 45 (2012) 86–93.
[2] M.A. Siddiqui, Heat transfer and fluid flow studies in the collector tubes of a
closed-loop natural circulation solar water heater, Energy Convers. Manage. 38
(1997) 799–812.
[3] S.I. Haider, Y.K. Joshi, W. Nakayama, A natural circulation model of the closed
loop, two-phase thermosyphon for electronics cooling, J. Heat Transf. 124
(2002) 881.
[4] T.A. Nyce, F. Rosenberger, A general method for calculating natural convection
flows in closed loops, Chem. Eng. Commun. 134 (1995) 147–155.
[5] P.K. Vijayan, Experimental observations on the general trends of the steady
state and stability behaviour of single-phase natural circulation loops, Nucl.
Eng. Des. 215 (2002) 139–152.
[6] M. Misale, P. Garibaldi, J.C. Passos, G.G. de Bitencourt, Experiments in a single-
phase natural circulation mini-loop, Exp. Therm. Fluid Sci. 31 (2007) 1111–
1120.
[7] P. Garibaldi, M. Misale, Experiments in single-phase natural circulation
miniloops with different working fluids and geometries, J. Heat Transf. 130
(2008) 104506.
[8] M. Misale, F. Devia, P. Garibaldi, Experiments with Al2O3 nanofluid in a single-
phase natural circulation mini-loop: preliminary results, Appl. Therm. Eng. 40
(2012) 64–70.
[9] J.Y. Wang, T.J. Chuang, Y.M. Ferng, CFD investigating flow and heat transfer
characteristics in a natural circulation loop, Ann. Nucl. Energy 58 (2013) 65–
71.
[10] A.K. Nayak, M.R. Gartia, P.K. Vijayan, An experimental investigation of single-
phase natural circulation behavior in a rectangular loop with Al2O3
nanofluids, Exp. Therm Fluid Sci. 33 (2008) 184–189.
[11] D.N. Basu, S. Bhattacharyya, P.K. Das, Development of a unified model for the
steady-state operation of single-phase natural circulation loops, Int. J. Heat
Mass Transf. 62 (2013) 452–462.
[12] N. Goudarzi, S. Talebi, An approach to stability analysis and entropy generation
minimization in the single-phase natural circulation loops, Energy 80 (2015)
213–226.
[13] R. Saha, S. Sen, S. Mookherjee, K. Ghosh, A. Mukhopadhyay, D. Sanyal,
Experimental and numerical investigation of a single-phase square natural
circulation loop, J. Heat Transf. 137 (2015) 121010.
[14] P.K. Vijayan, M. Sharma, D. Saha, Steady state and stability characteristics of
single-phase natural circulation in a rectangular loop with different heater and
cooler orientations, Exp. Therm. Fluid Sci. 31 (2007) 925–945.
[15] J.Y. Kudariyawar, A.M. Vaidya, N.K. Maheshwari, P. Satyamurthy,
Computational study of instabilities in a rectangular natural circulation loop
using 3D CFD simulation, Int. J. Therm. Sci. 101 (2016) 193–206.
[16] M. Krishnani, D.N. Basu, On the validity of Boussinesq approximation in
transient simulation of single-phase natural circulation loops, Int. J. Therm. Sci.
105 (2016) 224–232.
[17] A.K. Srivastava, J.Y. Kudariyawar, A. Borgohain, S.S. Jana, N.K. Maheshwari, P.K.
Vijayan, Experimental and theoretical studies on the natural circulation
behavior of molten salt loop, Appl. Therm. Eng. 98 (2016) 513–521.
[18] C.J. Ho, Y.Z. Chen, F.J. Tu, C.M. Lai, Thermal performance of water-based
suspensions of phase change nanocapsules in a natural circulation loop with a
mini-channel heat sink and heat source, Appl. Therm. Eng. 64 (2014) 376–384.
[19] N.M. Rao, M. Mishra, B. Maiti, P.K. Das, Effect of end heat exchanger
parameters on the performance of a natural circulation loop, Int. Commun.
Heat Mass Transf. 29 (2002) 509–518.
[20] N.M. Rao, B. Maiti, P.K. Das, Dynamic performance of a natural circulation loop
with end heat exchangers under different excitations, Int. J. Heat Mass Transf.
48 (2005) 3185–3196.
[21] N.M. Rao, B. Maiti, P.K. Das, Stability behavior of a natural circulation loop with
end heat exchangers, J. Heat Transf. 127 (2005) 749.
[22] N.M. Rao, B. Maiti, P.K. Das, Steady State performance of a single phase natural
circulation loop with end heat exchangers, J. Heat Transf. 130 (2008) 084506.
[23] K. Kiran Kumar, M. Ram, Gopal, Steady-state analysis of CO2 based natural
circulation loops with end heat exchangers, Appl. Therm. Eng. 29 (2009) 1893–
1903.
[24] A.K. Yadav, M.R. Gopal, S. Bhattacharyya, CFD analysis of a CO2 based natural
circulation loop with end heat exchangers, Appl. Therm. Eng. 36 (2012) 288–
295.
[25] A.K. Yadav, M. Ram Gopal, S. Bhattacharyya, Transient analysis of
subcritical/supercritical carbon dioxide based natural circulation loops with
end heat exchangers: numerical studies, Int. J. Heat Mass Transf. 79 (2014) 24–
33.
[26] H. Cheng, H. Lei, C. Dai, Thermo-hydraulic characteristics and second-law
analysis of a single-phase natural circulation loop with end heat exchangers,
Int. J. Therm. Sci. 129 (2018) 375–384.
[27] S. Bhattacharyya, D.N. Basu, P.K. Das, Two-phase natural circulation loops: a
review of the recent advances, Heat Transf. Eng. 33 (2012) 461–482.
[28] V. Pandey, S. Singh, Bifurcation analysis of density wave oscillations in natural
circulation loop, Int. J. Therm. Sci. 120 (2017) 446–458.
[29] X. Chen, P. Gao, S. Tan, Z. Yu, C. Chen, An experimental investigation of flow
boiling instability in a natural circulation loop, Int. J. Heat Mass Transf. 117
(2018) 1125–1134.
[30] S. Bodjona, E. Videcoq, R. Saurel, A. Chinnayya, A.M. Benselama, Y. Bertin,
Transient simulation of a two-phase loop thermosyphon with a model out of
thermodynamic equilibrium, Int. J. Heat Mass Transf. 108 (2017) 2321–2332.
[31] V.H. Bhusare, R.K. Bagul, J.B. Joshi, A.K. Nayak, U. Kannan, D.S. Pilkhwal, P.K.
Vijayan, Steady state flow analysis of two-phase natural circulation in multiple
parallel channel loop, Nucl. Eng. Des. 305 (2016) 706–716.
[32] W. Rashmi, A.F. Ismail, M. Khalid, Y. Faridah, CFD studies on natural convection
heat transfer of Al2O3-water nanofluids, Heat Mass Transf. 47 (2011) 1301–
1310.
[33] R.J. Moffat, Describing the uncertainties in experimental results, Exp. Therm
Fluid Sci. 1 (1988) 3–17.