Nuclear Engineering and Design 236 (2006) 19021908

Experimental study on convective heat transfer of water ow

in a heated tube under natural circulation
Ruichang Yang

, Ruolei Liu, Yong Zhong, Tao Liu

Department of Thermal Engineering, Key Laboratory for Thermal Science and Power Engineering of Ministry of Education,
Tsinghua University, Beijing 100084, China
Received 26 September 2005; received in revised form 23 January 2006; accepted 23 January 2006
Abstract
This paper reports on an experimental study on transitional heat transfer of water owin a heated vertical tube under natural circulation conditions.
In the experiments the local and average heat transfer coefcients were obtained. The experimental data were compared with the predictions by
a forced ow correlation available in the literature. The comparisons show that the Nusselt number value in the fully developed region is about
30% lower than the predictions by the forced ow correlation due to ow laminarization in the layer induced by co-current bulk natural circulation
and free convection. By using the Rayleigh number Ra to represent the inuence of free convection on heat transfer, the empirical correlations for
the calculation of local and average heat transfer behavior in the tube at natural circulation have been developed. The empirical correlations are
in good agreement with the experimental data. Based on the experimental results, the effect of the thermal entry-length behavior on heat transfer
design in the tube under natural circulation was evaluated.
2006 Elsevier B.V. All rights reserved.
1. Introduction
Natural circulation has been widely used in thermal equip-
ment of many industries. It is of great use for the safety of light
water reactors. Natural circulation can be used to release the
residual heat within the reactor during reactor accidents and
events. It can also be considered as a major method of circula-
tion cooling during the normal operation of a pressurized water
reactor (PWR). For a PWR operating under natural circulation
condition, the ow rate in the primary circuit is relatively low
and it is mainly related to the heat power of the reactor and the
ow resistance in the primary loop. In principle, to increase the
heat power of the reactor can increase the owrate in the primary
circuit. However, the safety of the reactor is not only related to
the ow rate in the core, but also affected by the heat transfer
behavior of single-phase water ow in the core of the reactor.
If the heat transfer capacity of the uid in the core decreases
during natural circulation, in order to ensure the power output
of the reactor, the temperature difference between the wall of the
fuel rods and the coolant must be increased, which will affect the

Corresponding author. Tel.: +86 10 62781005; fax: +86 10 62770209.

E-mail address: yangrc@mail.tsinghua.edu.cn (R. Yang).
safe operation of the reactor. Therefore, it is very important to
determine correctly the heat transfer performance of the coolant
in the reactor core during design and safety analysis of the PWR
with natural circulation.
So far a tremendous amount of research has been done on
the subject of convective heat transfer of single-phase liquid
ow in tubes. This research has been reviewed, among others,
by Cengel (2003), Incropera and DeWitt (2002), Ebadian and
Dong (1998) and Kays and Perkins (1985). Many good correla-
tions have been proposed for the calculation of convective heat
transfer in tubes. Among these, the correlations developed by
Dittus and Boelter (1930), Petukhov and Kirillov (1958) and
Gnielinski (1976) can be used to estimate the heat transfer coef-
cients in fully developed turbulent ow regions and even in
transition ow regions of tubes. Techniques dealing with con-
vective heat transfer in the entrance region of tubes have also
been proposed, e.g. by Sleicher and Tribus (1957), Kays and
Perkins (1985) and Gnielinski (1976). When the wall and uid
temperature difference is large, methods to take into account
the effect of temperature-varying properties on convective heat
transfer have been suggested, for example, by Sieder and Tate
(1936), Allen and Eckert (1964) and Petukhov (1970). A high
temperature gradient in forced uid ow in a gravity eld can
also increase the natural convection, and thus forced convection
0029-5493/$ see front matter 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.nucengdes.2006.01.013

http://www.paper.edu.cn
R. Yang et al. / Nuclear Engineering and Design 236 (2006) 19021908 1903
Nomenclature
C
p
specic heat (J/kg K)
d tube diameter (m)
F thermal entrance effect factor
f friction factor
G mass ux (kg/m
2
s)
g gravitational acceleration (m/s
2
)
Gr Grashof number
k thermal conductivity (W/mK)
L length (m)
Nu Nusselt number
Pr Prandtl number
Q heat input power (kW)
q heat ux (W/m
2
)
Ra Rayleigh number
Re Reynolds number
T temperature (

C)
Greek symbols
heat transfer coefcient (W/m
2
K)
thermal expansion coefcient (1/K)
dynamic viscosity (kg/ms)
density (kg/m
3
)
Subscripts
f uid
w tube wall
is accompanied by natural convection. Metais and Eckert (1964)
have given the ow pattern maps to determine the ow regimes
of natural, forced and mixed convection for ow through tubes.
A lot of works on mixed convection heat transfer of water ow
in tubes have been reported in the literature. Among these are the
papers of Herbert and Sterns (1972), Rouai (1987), Joye (1996)
and Celata et al. (1998). Until now, almost all the correlations
for the calculation of convective heat transfer of single-phase
liquid ow in tubes were obtained from experiments carried out
under forced owconditions. For most safety analyses of PWRs
with natural circulation the heat transfer calculation for coolant
ow in the reactor cores are performed by using the correlations
taken from forced ow conditions. However, whether the heat
transfer correlations obtained under forced ow conditions are
applicable for natural circulation ow in these analyses remains
a problem. So far only few studies on the convective heat trans-
fer of single-phase liquid ow in natural circulation systems
were published. Zvirin et al. (1981) performed an experiment
in a natural circulation system to investigate the performance
of the system. Based on the experimental data taken at nat-
ural circulation condition, they estimated the convective heat
transfer coefcient of laminar water ow in the primary side
of the heat exchanger connected in the system. It was found
that the convective heat transfer coefcients obtained from the
experiments were 21%belowthat predictedbythe empirical cor-
relation available in the literature. Zvirin et al. considered that
this deviation was due to the multidimensional ow patterns
in the natural circulation loop. Vijayan et al. (1991) reported
an experimental study on convective heat transfer of laminar
water ow in a natural circulation loop. They found that the
natural circulation data of convective heat transfer were not cor-
related by the forced ow correlations available in literature due
to the presence of secondary ows in the tubes under natural
circulation.
This paper reports on an experimental study on local transi-
tional heat transfer of single-phase water ow in a heated tube
under natural circulation condition. The experimental data are
compared with the predictions of a forced owcorrelation found
in the literature. Based on the experimental results, the empir-
ical correlations for the calculations of the local heat transfer
coefcients as well as the average heat transfer coefcients in
the heated tube under natural circulation conditions have been
developed.
2. Experimental loop
A schematic diagram of the test loop is shown in Fig. 1. For
the experiments, demineralized and degassed water was used
Fig. 1. Schematic of test loop.
http://www.paper.edu.cn
1904 R. Yang et al. / Nuclear Engineering and Design 236 (2006) 19021908
as working medium. The test loop essentially consists of an
upow heated test tube, an unheated riser, an upow cooler, a
downowcooler, a downcomer, a return leg and a tank. The loop
can be operated under natural and forced circulation conditions,
respectively. For natural circulation tests, the valves 2 and 3 and
the pump were closed but the valve 1 was opened. The test loop
is 9.25 m high, made of a stainless steel with 21 mm ID and
25 mm OD, except the heated test section and the two coolers.
The heated test section is a tube with 18 mm ID and 22 mm OD.
The heated test tube is made of stainless steel and heated directly
by ACcurrent. The heated length is 2.5 m. Uniformheat ux was
attained by uniform tube wall thickness. In front of the heated
section there is an adiabatic starting length in which the velocity
prole can be developed. At the inlet of the heated section where
heat transfer begins the velocity prole is fully developed. A
venturi ow meter is located in the downcomer. At the top of
the loop, a water tank opened to the atmosphere is connected to
the loop. There is a surge line from the downcomer to the tank.
The surge line is opened to atmosphere by tee union pipe. The
measuring points are shown in Fig. 1. The pressure at the outlet
of the test tube was measured by a Bourdon-type manometer.
The pressure drop over the venturi ow meter was measured by
a U-tube differential manometer. The temperatures of the uid
at different locations of the loop were measured by sheathed
thermocouples. Along with the heated test tube, thermocouples
were welded on the tube surfaces at the locations of 2, 5, 10,
20, 30, 70 and 137 times of the tube diameter from the inlet
of the tube to measure the local outer tube wall temperatures.
The voltage and thermo-emf signals were recorded by a multi-
channel digital recorder and a computer data acquisition system.
The test loop was operated under atmospheric pressure. The
ranges of other experimental variables are as follows: heat input
power to the heated test section Q=1.010.5 kW, uid tem-
perature at the inlet of the heated tube T
in
=2759

C, uid
temperature at the outlet of the heated tube T
ex
=3978

C. The
mass ow rate in the heated test tube is power-dependent. The
range of the mass velocity in the test tube is G=75405 kg/m
2
s.
During experiments, the temperature of the tube wall at the exit
of the heated test section was monitored to ensure the ow in
the heated tube was single-phase liquid ow.
3. Experimental results and discussion
From the measured outer wall temperatures of the heated
tube, the inner wall temperatures can be obtained by solving the
heat conduction equation of the tube wall. The local convective
heat transfer coefcient can be calculated as follows
=
q
w
T
w
T
f
(1)
where q
w
is the heat ux at the inner tube wall, T
w
and T
f
are the
local inner tube wall and bulk uid temperatures, respectively.
The experimental data of the local convective heat transfer under
natural circulation conditions are illustrated in Figs. 2 and 3,
respectively. In Figs. 2 and 3, the note 2d, for example, indicates
the location of two times the tube diameter from the inlet of
the tube. For comparison, the predictions from the correlation
Fig. 2. Experimental data of local convective heat transfer.
of Gnielinski (1976) are also shown in Figs. 2 and 3, because
the Gnielinski correlation is considered to be suitable for the
calculation of forced convective heat transfer in both transition
and turbulent pipe ows (2300 Re 5 10
6
). The Gnielinski
correlation is given by
Nu =
(Re 1000)Prf/2
1.0 +12.7(Pr
2/3
1)

f/2
(2)
where Nu (Nu =d/k) is the Nusselt number, Re (Re =Gd/)
the Reynolds number, Pr (Pr =C
p
/k) the Prandtl number, f
the friction factor, f =(1.58ln Re3.28)
2
. It can be found from
Figs. 2 and 3 that the values of the Nusselt number taken from
the experiments at the location of 2d are greater than that cal-
culated by the Gnielinski correlation. The experimental data of
the Nusselt number at the locations of 5d and 10d are approx-
imately close to that predicted by Gnielinski correlation. The
data of the Nusselt number at the locations of 30d and after
30d under natural circulation are lower than that calculated by
Gnielinski correlation. Figs. 2 and 3 show that the values of the
Nusselt number obtained from the experiments decrease along
the ow direction in the entrance region and nally the Nusselt
numbers in the fully developed region almost do not change with
Fig. 3. Experimental data of local convective heat transfer.
http://www.paper.edu.cn
R. Yang et al. / Nuclear Engineering and Design 236 (2006) 19021908 1905
the distance from the tube inlet. It can be well understood that
in the thermal entrance region starting at the inlet of the heated
tube the thermal boundary layer is thin and the convective heat
transfer coefcients become larger than that in the fully devel-
oped region. It can be observed from Figs. 2 and 3 that for the
present experimental situation the thermal entry length is about
30 times the diameter of the tube. Figs. 2 and 3 also show that
the natural circulation data in the fully developed region cannot
be correlated by the Gnielinski correlation. The values of the
Nusselt number in the fully developed region taken at natural
circulation conditions are about 30% lower than that predicted
by the Gnielinski correlation. This behavior can be explained by
the inuence of free convection on heat transfer in the heated
tube. When water ows in the vertical heated tube with a low
ow rate under natural circulation condition, the uid temper-
ature near the heated tube wall is much higher than that in the
bulk of the tube. Hence, the uid density near the wall is less
than that in the bulk of the tube. Due to the buoyancy caused by
the lower density of the uid near the wall, free convection of
the uid occurs in the layer near the tube wall. Since the free
convection ows in the same direction as the bulk ow, this co-
current mixed convection can reduce the turbulence and causes
ow laminarization in the layer, which also reduces the convec-
tive heat transfer. Therefore, the natural circulation data of the
Nusselt number in the fully developed region are lower than that
predicted by the Gnielinski correlation.
For engineering application ones more concern the determi-
nations of average convective heat transfer coefcients in tubes.
The experimental data for the average convective heat transfer
coefcients under natural circulation conditions and the com-
parisons of the data with the Gnielinski correlation (Eq. (2)) are
illustrated in Fig. 4. During the data reduction the mean wall
temperature and the mean bulk uid temperature, which are the
arithmetic averages of the wall and uid temperatures at the inlet
and the exit of the tube, respectively, are used for the evaluation
of the average convective heat transfer coefcients in the tube.
By comparing Fig. 4 with Fig. 3, it can be found that though the
data of the Nusselt number in Figs. 3 and 4 are all evaluated with
the same experimental parameters, the average Nusselt numbers
are higher than the exact Nusselt numbers in the fully developed
Fig. 4. Experimental data for average convective heat transfer coefcients in the
tube.
region. It indicates that whether the average heat transfer coef-
cients obtained fromthis experiment can be used to calculate the
exact heat transfer behavior in the full length of the tube under
natural circulation remains a problem.
To present a method for the prediction of the convective heat
transfer in a heated tube under natural circulation conditions is of
great signicance for engineering applications. In some studies
available in the literature, such as in the works of Morcos and
Bergles (1975) and Petukhov and Polyakov (1988), the Grashof
number was used to describe the effect of free convection on
forced convective heat transfer, while in the works of Vicente et
al. (2002) and Aicher and Martin (1997), the Rayleigh number,
which is the multiplier of the Grashof and Prandtl numbers, was
introduced to correlate the experimental data of combined forced
and free convection heat transfer. In this paper it is considered
that the Rayleigh number can be used to represent the effect of
free convection caused by buoyancy on convective heat transfer
in the heated tube under natural circulation. By adopting the
Rayleighnumber tomodifythe conventional correlationused for
the calculation of heat transfer in forced circulation, we propose
the following correlation for the determination of convective
heat transfer in natural circulation
Nu
Nu
0
= Ra
n
(3)
where Nu is the Nusselt number for the calculation of
convective heat transfer in natural circulation, Nu
0
is
determined by the Gnielinski correlation (Eq. (2)) and
Ra =GrPr =
2
gd
3
(T
w
T
f
)C
p
/k. In this study Eq. (3) is at rst
used to correlate the natural circulation experimental data of heat
transfer at the location of 137 times the tube diameter from the
inlet of the tube, that is, at the exit of the tube, which belongs to
the fully developed region. The exponent in Eq. (3) n =0.03
was found for the data at the exit of the tube. The comparison
of the experimental data at the exit of the tube with the predic-
tion of Eq. (3) for n =0.03 is shown in Fig. 5. The calculated
results agree with the experimental data within 7.6%. It indi-
cates that the correlation proposed in this paper correlates the
experimental data very well.
Based on the formulation for the prediction of convective heat
transfer in the fully developed region (Eq. (3)), we assume that
the local convective heat transfer in the thermal entrance region
of the heated tube under natural circulation can be evaluated by
Fig. 5. Comparison of experimental data for the exit of the tube with Eq. (3).
http://www.paper.edu.cn
1906 R. Yang et al. / Nuclear Engineering and Design 236 (2006) 19021908
Fig. 6. Comparison of experimental data for the location of L =2d with Eq. (4).
introducing an entrance effect factor to modify the expression
of Eq. (3) as follows
Nu = Nu
0
Ra
0.03

1 +

d
L

(4)
where Nu is the Nusselt number for the calculation of local con-
vective heat transfer at the location of L from the inlet of the
tube in natural circulation, Nu
0
is determined by the Gnielinski
correlation (Eq. (2)), while the Re, Pr and f involved in Eq. (2)
and Ra in Eq. (3) are all evaluated with the parameters taken
at the location L; here L is the length determined from the inlet
of the heated tube. By using Eq. (4) to reduce the experimen-
tal data, it is found that the exponent in Eq. (4) is n =0.2 for
L/d <30 and (d/L)
n
=0 for L/d 30. It means that the heat trans-
fer in the region of L/d 30 belongs to a fully developed region.
The predictions with Eq. (4) have been compared with the nat-
ural circulation data taken form the present experiments. The
average deviations of the predictions from the corresponding
experimental data are 7.0% for the location of L =2d, 9.1%
for the location of L =5d, 7.6% for the location of L =10d,
6.1% for the location of L =20d, and 5.6% for the location
of L =30d, respectively. The typical comparisons are shown in
Fig. 6 for the location of L =2d, in Fig. 7 for the location of
L =10d and in Fig. 8 for the location of L =30d, respectively. It
is found from the comparisons that the predictions with the cor-
relation (4) presented in this study are in good agreement with
the experimental data.
As mentioned above, for engineering applications most cor-
relations available in the literature are used for the calculation
Fig. 7. Comparison of experimental data for the location of L =10d with Eq.
(4).
Fig. 8. Comparison of experimental data for the location of L =30d with Eq.
(4).
of average convective heat transfer coefcients in tubes. Based
on the experimental data of the average convective heat transfer
coefcients under natural circulation as shown in Fig. 4, we also
use Eq. (3) to correlate the data of the average convective heat
transfer behavior in the tube at natural circulation conditions. It
is given by
Nu
Nu
0
= Ra
0.011
(5)
where Nu is the Nusselt number for the calculation of average
convective heat transfer coefcients in the heated tube under
natural circulation; Nu
0
is determined by the Gnielinski corre-
lation (Eq. (2)). The comparison of the experimental data with
the prediction of Eq. (5) is shown in Fig. 9. Eq. (5) agrees with
the experimental data within 8.7%.
It can be found from the present experiments that the values
of the Nusselt number in the thermal entrance region are higher
than that in the fully developed region. If the Nusselt numbers
in the fully developed region are used directly in heat transfer
design without reference to the entry-length behavior, the rate of
heat transfer in the full length of the tube will be underpredicted.
However, for transitional and turbulent ow, the thermal entry
length is characteristically short, while the Nusselt numbers in
the fully developed region are lower than that reduced by using
the average heat transfer coefcients in the tube. If the Nusselt
numbers reduced by the average heat transfer coefcients are
used for heat transfer calculation in practice, the rate of heat
Fig. 9. Comparison of experimental data for average convective heat transfer
coefcients with Eq. (5).
http://www.paper.edu.cn
R. Yang et al. / Nuclear Engineering and Design 236 (2006) 19021908 1907
transfer in the tube perhaps will be overpredicted. In view of
this situation, it is valuable to evaluate the effect of the thermal
entry-length behavior on heat transfer design in the tube under
natural circulation. To perform this evaluation, we have divided
the full length of the heated tube into seven sections. The length
of each section was determined as follows. For the second to
the sixth sections, the locations at which thermocouples were
welded to measure the local outer tube wall temperatures, were
approximately taken as the center of the sections, while for the
rst and the last sections, the locations at which thermocouples
were welded for the measurement of outer tube wall tempera-
tures, were at the exit of the sections. For each section, we take
the local Nusselt number, which is directly obtained from the
measured local tube wall temperature within the section as the
average Nusselt number in the section, then we weight the local
Nusselt number with the length of the section and sum these
weighted local Nusselt numbers. We also weight the Nusselt
number obtained in the fully developed region and the average
Nusselt number reduced from the average heat transfer coef-
cients in the tube with the full length of the tube, respectively, and
nally we can get the thermal entrance effect factor F as follows
F =

7
x=1
(L
x
)Nu
x

7
x=1
(L
x
)

Nu

7
x=1
(L
x
)

Nu
100% (6)
where x is the number of the divided section, Nu
x
the local
Nusselt number taken from section x, Nu the Nusselt number
in fully developed region or the average Nusselt number of the
tube. During the evaluation of the thermal entrance effect factor
F, the arithmetic average of the Nusselt numbers obtained at the
locations of 70d and 137d is taken as the Nusselt number in fully
developed region. In Eq. (6), L is the length of the divided
section. According to the dividing method of the sections
mentioned above, we take L
1
=2d, L
2
=5d, L
3
=7d,
L
4
=10d, L
5
=10d, L
6
=70d, and L
7
=35d. The thermal
entrance effect factor F is calculated from Eq. (6) based on
1 1 1 run data points obtained from the experiments. It is found
that when the Nusselt number in fully developed region Nu is
used for the evaluation of the thermal entrance effect factor F
in Eq. (6), the average value of F for the 1 1 1 run data points
is +8%, while the average Nusselt number Nu is used for the
calculation of the thermal entrance effect factor F in Eq. (6), the
average value of F is 18%. It means that if the Nusselt number
in the fully developed region is used in heat transfer design
without reference to the entry-length behavior, the rate of heat
transfer in the tube will be underpredicted by 8%. On the other
hand, if the average Nusselt number is used for the calculation
of heat transfer, the rate of heat transfer in the tube will be
overpredicted by 18%. It indicates that instead of the average
Nusselt number or Eq. (5), the Nusselt number in the fully
developed region or Eq. (3) with n =0.03 is more suitable for
the heat transfer design in the tube under natural circulation.
4. Conclusion
The experiments of convective heat transfer of single-phase
water ow in a heated vertical tube under natural circulation
condition have been performed. In the experiments, the local
heat transfer coefcients and Nusselt numbers for transitional
ow in both the thermal entrance and fully developed regions
of the tube were obtained. The experiments indicate that the
thermal entry length is about 30 times the diameter of the tube.
Based on the experimental results, the average Nusselt numbers
in the tube were obtained.
The experimental data were compared with the predictions by
the Gnielinski correlation, which is suitable for the calculation of
forced transitional and turbulent heat transfer in tubes. The com-
parisons show that the Nusselt numbers in the fully developed
region taken at natural circulation are about 30%lower than that
predicted by the Gnielinski correlation. This behavior is due to
the ow laminarization in the layer induced by co-current bulk
natural circulation and free convection. By using the Rayleigh
number Ra to represent the inuence of free convection on heat
transfer and to modify the Gnielinski correlation, the empirical
correlations for the calculation of local and average heat transfer
behavior in the tube under natural circulation have been devel-
oped. The empirical correlations are in good agreement with the
experimental data.
Based on the experimental data, the effect of the thermal
entry-length behavior on heat transfer design in the tube was
evaluated. It is shown that the Nusselt number in the fully devel-
oped region is more suitable for the heat transfer calculation in
the tube under natural circulation.
Acknowledgements
This research was supported by the Natural Science Founda-
tion of China (approved number 50476012) and National Key
Laboratory of Bubble Physics and Natural Circulation.
References
Aicher, T., Martin, H., 1997. New correlation for mixed turbulent natural
and forced convection heat transfer in vertical tubes. Int. J. Heat Mass
Transfer 40, 36173626.
Allen, R.W., Eckert, E.R.G., 1964. Friction and heat transfer measurements
to turbulent pipe ow of water at uniform wall heat ux. J. Heat Transfer
86, 301310.
Celata, G.P., Francesco, D., Andrea, C., Cumo, M., 1998. Upow turbulent
mixed convection heat transfer in vertical pipes. Int. J. Heat Mass Transfer
41, 40374054.
Cengel, Y.A., 2003. Heat Transfer: A Practical Approach, second ed.
McGraw-Hill.
Dittus, F.W., Boelter, L.M.K., 1930. Heat transfer in automobile radiators of
the tubular type. Univ. California Publ. Eng. 2, 443461.
Ebadian, M.A., Dong, Z.F., 1998. Forced convection, internal ow in ducts.
In: Rohsenow, W.M., Hartnett, J.P., Cho, Y.I. (Eds.), Handbook of Heat
Transfer, third ed. McGraw-Hill.
Gnielinski, V., 1976. Equations for heat and mass transfer in turbulent pipe
and channel ow. Int. Chem. Eng. 16, 359368.
Herbert, L.S., Sterns, U.J., 1972. Heat transfer in vertical tubesinteraction
of forced and free convection. Chem. Eng. J. 4, 4652.
Incropera, F.P., DeWitt, D.P., 2002. Introduction to Heat Transfer, fourth ed.
Wiley.
Joye, D.D., 1996. Comparison of aiding and opposing mixed convection heat
transfer in a vertical tube with Grashof number variation. Int. J. Heat
Fluid Flow 17, 96101.
http://www.paper.edu.cn
1908 R. Yang et al. / Nuclear Engineering and Design 236 (2006) 19021908
Kays, W.M., Perkins, H.C., 1985. Forced convection, internal ow in ducts.
In: Rohsenow, W.M., Hartnett, J.P., Ganic, E.N. (Eds.), Handbook of
Heat Transfer Fundamentals, second ed. McGraw-Hill.
Metais, B., Eckert, E.R.G., 1964. Forced, mixed and free convection regimes.
J. Heat Transfer 86, 295296.
Morcos, S.M., Bergles, A.E., 1975. Experimental investigation of combined
forced and free laminar convection in horizontal tubes. J. Heat Transfer
97, 212219.
Petukhov, B.S., 1970. Heat transfer and friction in turbulent pipe ow with
variable physical properties. In: Irvine, T.F., Hartnett, J.P. (Eds.), Heat
Transfer, vol. 6, pp. 503564.
Petukhov, B.S., Kirillov, V.V., 1958. The problem of heat exchange
in the turbulent ow of liquids in tubes. Teploenergetika 4, 6368
(in Russian).
Petukhov, B.S., Polyakov, A.E., 1988. Heat Transferin Turbulent Mixed Con-
vection. Hemisphere.
Rouai, N.M., 1987. Inuence of buoyancy and imposed ow transients on
turbulent convection heat transfer in a tube. Ph.D. thesis, University of
Manchester.
Sieder, E.N., Tate, G.E., 1936. Heat transfer and pressure drop of liquids in
tubes. Ind. Eng. Chem. 28, 14291435.
Sleicher, C.A., Tribus, M., 1957. Heat transfer in a pipe with turbulent ow
and arbitrary wall-temperature distribution. Trans. ASME 79, 789796.
Vicente, P.G., Garcia, A., Viedma, A., 2002. Experimental study of mixed
convection and pressure drop in helically dimpled tubes for laminar and
transition ow. Int. J. Heat Mass Transfer 45, 50915105.
Vijayan, P.K., Mehta, S.K., Date, A.W., 1991. On the steady-state per-
formance of natural circulation loops. Int. J. Heat Mass Transfer 34,
22192230.
Zvirin, Y., Jeuck III, P.R., Sullivan, C.W., Duffey, R.B., 1981. Experimental
and analytical investigation of a natural circulation system with parallel
loops. J. Heat Transfer 103, 645652.
http://www.paper.edu.cn