Professional Documents
Culture Documents
a r t i c l e i n f o a b s t r a c t
Article history: Supercritical carbon dioxide (sCO2) is a promising working fluid for heat transfer applications. But the
Received 23 August 2016 complex nature of heat transfer and fluid flow, especially in near critical region hinders the development
Received in revised form of heat transfer equipment. In this study, the two-layer model for heat transfer to supercritical fluids used
20 March 2017
in Ref. [29] is improved by inclusion of the buoyancy and acceleration effects. Results were then cali-
Accepted 21 March 2017
brated with direct numerical simulation (DNS) data for refinement and were validated with the available
experimental data for sCO2. The model is improved qualitatively in terms of capability to predict heat
transfer deterioration and quantitatively in terms of reduction in mean relative error. A parametric study
Keywords:
Supercritical pressure
is also performed to identify the effects of different variables such as mass flux, pipe diameter and inlet
sCO2 temperature on heat transfer. In the end, the need of further DNS and experiments pertaining to heating
Turbulent flow and cooling of sCO2 is briefly discussed.
Two layer model © 2017 Elsevier Masson SAS. All rights reserved.
Heat transfer deterioration
http://dx.doi.org/10.1016/j.ijthermalsci.2017.03.021
1290-0729/© 2017 Elsevier Masson SAS. All rights reserved.
228 S. Pandey et al. / International Journal of Thermal Sciences 117 (2017) 227e238
affect the heat transfer between the PCHE's wall and bulk fluid desirable which uses limited computer resource and provide re-
flowing inside it. There are many experimental evidence accessible sults within an acceptable time.
for heat transfer enhancement and deterioration depending upon In the quest for a simpler and universal model to predict the
mass and heat flux in the supercritical range for different working heat transfer, two layer model is a foremost alternative [18]. used
fluids in the pipe flow (e.g. Refs. [26,38]). two layer model to develop implicit equations for water, based
In the past, several researchers have developed heat transfer upon numerical wall function. In his work, Laurien employed the
correlations for sCO2 by modifying existing correlations to fit logarithmic law of the wall and variable properties to model the
experimental data [4,5,9,13,32]; but the lack of generality and ac- heat transfer to supercritical water. But the agreement was poor
curacy limits their use in PCHE. Also, correlations developed for with the experimental data, therefore he introduced property ratio
heating conditions don't fit well in cooling conditions because of to empirically fit the data. Later, when [30] tested it for sCO2, they
deceleration and buoyancy effects. Similarly, the upward and found poor agreement especially at higher heat flux where dete-
downward direction of flow creates a distinction between each rioration took place. Therefore, they again provided a new value for
other due to the buoyancy. In addition to it, abnormal heat transfer correction factor in terms of density ratio. Thus generality was not
phenomena like deterioration and enhancement cannot be achieved with two layer model using numerical wall function only.
accounted by most of the correlations. Therefore, there is a strong Simultaneously McEligot and Laurien [24] investigated two
need for a correlation or model which is supported by an adequate layer model based on the thermal resistance analogy for sCO2. They
physical theory and increase the range of applicability. have derived an explicit equation to calculate the convective heat
To analyze the heat transfer at supercritical pressure, numerous transfer coefficient while approximating the bulk temperature to
computational fluid dynamics (CFD) studies based upon turbulent be the same as the center temperature. Later, Pandey and Laurien
model are available but their reliability is limited for a specific case [29] tested and validated it with experimental data for sCO2 and
[25,33,35]. Also, CFD studies including DNS generally require high- supercritical R-22. It was able to account heat transfer deterioration
performance computing resource which increases the overall cost for both the fluids. But the relative error was in between ± 25% for
considerably. Additionally, an advanced level programming skill is sCO2 and supercritical R-22, which might not be acceptable for the
required. To mitigate these limitations a relatively simpler model is actual design of PCHE.
S. Pandey et al. / International Journal of Thermal Sciences 117 (2017) 227e238 229
ZR
2
um ¼ uðrÞrdr (5)
R2
0
ZR r 7
1
2
um ¼ ucl 1 rdr (6)
R2 R
0
um
ucl ¼ (7)
0:8167
Then uvs is given as:
1
Duturb ¼ lnRþb lnyþb
cs utb (10)
k
Fig. 3. Iterative approach for solving equations.
S. Pandey et al. / International Journal of Thermal Sciences 117 (2017) 227e238 231
pipe flow of sCO2 is given in Ref. [6]. From the complete DNS
4qw mb bb database, 3 cases of vertical flow in pipe viz. forced (i.e. without
Kv ¼ ; cv ¼ Au2m
GDum rb Cpb buoyancy), upward and downward are employed in the current
work. Table 1 depicts the simulation conditions for DNS.
Due to density variations in the radial direction of flow, its ef- An additional parameter cv was introduced earlier in equation
fects cannot be neglected especially at high heat flux. The flow will (11) to model the acceleration effect and it consists of an arbitrary
change from forced to mixed type, where natural convection plays constant A and u2m . The value of u2m depends upon the flow while
an important role. In the previous attempt of authors, buoyancy the magnitude of A was adjusted to have an optimized value of
was included by empirically fitting of the model with DNS data, goodness of fit. The mean square error (MSE) was chosen as the
which considers the internal and external effects of buoyancy [19]. goodness of fit (see equation (14) for definition) and value of cv is
But, generality wasn't achieved, therefore a different analytical found to be 7:3 106 while optimizing MSE for wall temperature
approach is introduced in this study. To take buoyancy into account, and modified wall shear stress as shown in Fig. 4. Fig. 5a and b
a buoyant shear stress is inserted in addition to wall shear stress. shows the variation for the wall temperature and shear stress
This approach to include buoyancy effect by an extra shear stress respectively along the heated length. Both calibrated value and DNS
has already been used for drag reducing fluids by Ref. [36]. Equation values are shown. In this case, gravity is set to zero in DNS to have a
(12) shows the modified form of wall shear stress. reference case only for acceleration effects without buoyancy. From
Fig. 5a, it can be observed that due to uniform heat flux, wall
tw;m ¼ tw þ ðflow tb Þ (12) temperature keeps rising, which is the result of increased thermal
The resultant magnitude of the wall shear stress depends on the resistance across the layers. On the other hand, wall shear stress
’flow’ direction and intensity of buoyancy in a flow. The value of flow decrease with heated length, which will result in lesser pressure
is calibrated with DNS data for upward and downward flow. drop so lesser the heat transfer coefficient.
Buoyant shear stress is given by equation (13) and it is calibrated
with DNS in the next section 3. 1X N
MSE ¼ ðmodel DNSÞ2 (14)
N n¼1
tb ¼ yb gðrb rm Þ (13)
The second step in the process of calibration is to calibrate the
Where, data for buoyancy for equation (13). For that, parameter 0 flow0 was
varied and MSE was optimized similarly to relaminarization. Fig. 6a
ZTw rffiffiffiffiffiffiffiffiffiffi and b shows the variation of the wall temperature and wall shear
1 yvs tw;m
rm ¼ rdT; yb ¼ ; utw ¼ stress for upward flow, which has combined effect of acceleration
ðTw Tb Þ 1
Prcs
3 rw and buoyancy. DNS results show an oscillating pattern which is
Tb
assumed to be damped out after a certain heated length. During
heating in upward flow, buoyancy effects will stabilize the flow as
the direction of buoyancy force and flow are same. As a result of this
3. Calibration and validation stabilizing effect, the tendency of development of ‘M'shaped ve-
locity profile will increase. Therefore, the magnitude of wall shear
3.1. Introduction to DNS for presented case stress will increase and turbulence is decayed. This stabilization
effect is responsible for the increase in the wall temperature and
For calibration purpose, DNS data have been used. DNS is poor heat transfer between the wall and fluid.
considered as a reliable numerical method for fundamental tur- Fig. 7a and b illustrate the variations for downward flow. The
bulence research, although its usage is limited to the low Reynolds wall temperature increment is very less and this can be explained
number as a result of high demand on computational resources. by stabilizing effect. Due to constant heat flux, fluid is heated and
The inconvenience of obtaining turbulence statistics in the exper- buoyancy force acts upward while the flow direction is downward.
iments due to extreme conditions of supercritical fluids prevents a This makes flow unstable and creates a chaotic condition. It will
further investigation of complex heat transfer phenomena, which enhance the heat transfer, therefore, wall temperature increases
happens to be the strength of DNS [39]. In the current DNS, the slowly in the flow direction. The presented model also predicts a
dimensioned low-Mach-number governing equations in Cartesian peculiar behavior in wall shear stress, it shows a negative value
coordinates, as described in Ref. [6]; were discretized with the open during entire flow. This indicates the limitation of presented model
source finite volume code OpenFOAM V2.4 [28]. The Pressure- in the prediction of hydraulic behavior in the downward flow. To
Implicit with Splitting of Operators (PISO) algorithm was applied mitigate it, a wise selection of goodness-of-fit parameter is required
for pressure-velocity coupling. The temporal term was discretized along with weights assign to objective functions in the
with the second order implicit Euler scheme. The spatial dis- optimization.
cretization was handled with the central differencing scheme and a The presented model over-predicts or under-predicts the value
third order upwind scheme QUICK was adopted for the convective
term in the energy equation. For heated domain mesh size was
Table 1
168 172 2400 (radial, circumferential and axial direction), Simulation conditions for the DNS.
which is significantly higher than the well-known DNS work from
Parameter Value Parameter Value
Bae et al. [2] and [3] under the condition of the same second-order
accuracy in both codes. The authenticity of the current numerical D 2 mm Lh 60 mm
method has been tested for air flowing in a vertical pipe with Rei 5400 ½ Pi 8 MPa
qw 30.87 W/m2 Ti 301.15 K
measured data from experiments [7]. Another validation of DNS in
232 S. Pandey et al. / International Journal of Thermal Sciences 117 (2017) 227e238
of wall temperature at the inlet section. The main reasons for this 4.2. Validation with available experiments
disparency are thermal inlet effects in the DNS. These effects were
not within the scope of this study; therefore they are not modeled For validation purpose, there is a scarcity of experimental data
in this study. Also, they are not of importance as for longer heated for sCO2 with 2 mm pipe diameter comparable to PCHE's channels,
length, one would be more interested in overall flow rather than with both upward and downward flow. Experiments conducted by
Ref. [16] are well accepted by the scientific community and
Table 2 experiments, pressure drop (head loss) occurs due to major and
Summary of calibration parameter. minor losses in the flow and this drop will certainly enhance the
Case A (in cv ¼ Au2m ) Flow m in eq. (4) MSE for Tw MSE for tw heat transfer. A similar trend can also be seen in high flux case after
Forced 0 mcs 18.5090 0.0053
the heat transfer deterioration where the model predicts higher
7:3 106
Upward 7:3 106 1 mw 65.5777 0.0028 wall temperature than the experiments. In lower heat flux case
Downward 7:3 106 10 mw 7.0906 0.0776 (with qw ¼ 10 kW/m2), one can observe a deviation from old model
which predicts heat transfer very well. This is because of at lower
heat flux, property variations are not significant, therefore insig-
nificant effect of the density variation. Nevertheless, presented
model excels most of the correlation in term of prediction of heat
transfer along with hydraulic characteristics.
Fig. 10 shows the variation of wall shear stress for all three cases
of heat flux. The wall shear stress is nearly constant for lower heat
flux case due to the insignificant variation of properties. For me-
dium and high heat flux, a distortion can be seen. These oscillations
are a result of sharp radial properties variation sourced by the heat
transfer deterioration. For e.g. wall temperature suddenly increases
after bulk hb z230 kJ/kg due to deterioration at a high heat flux of
50 kW/m2. At the same time oscillation can be observed in wall
shear stress. These oscillations damp out shortly after this region.
Yet, cooling of sCO2 was not aimed during the development of
this model, but in the future, it is one of the aim of research.
Therefore, comparing this model with available cooling experiment
will provide an insight. Fig. 11 illustrates the comparison with the
experiments of [9]. Horizontal layout with comparatively larger
Fig. 8. Variation of bulk Nusselt number with bulk Reynolds number.
mass flux to heat flux ratio was considered during the experiments.
Therefore, smaller property variation and it resembles near-
therefore serve as a reliable database for the validation. The constant property case. An excellent agreement can be observed,
including in the prediction of heat transfer enhancement at pseu-
experimental parameters were: internal diameter of pipe was
4.4 mm, heated length was 2.1 m, inlet pressure and temperature docritical temperature.
were varied in between 7.75 and 8.85 MPa and 5e30 C respec-
tively. In the experiments, they used the vertical orientation of pipe 4.3. Parametric study
with only upward flow. Fig. 9 depicts the validation for presented
model with experiments of [16] along with the previous model of The aim of the parametric study is to investigate the sensitivity
[29]. The mean relative error (MRE) comes down to 13.1% from of the solution to the parameters. A parametric study is an impor-
18.3% for high heat flux case (i.e. qw ¼ 50 kW/m2). Also, heat tant tool to optimize the design of a device, trend base analysis, and
transfer deterioration is predicted with higher accuracy than the what-if type of analysis. In the following sub-sections a similar
previous model. For medium heat flux (correspond to qw ¼ 30 kW/ analysis is performed, in which parameters are varied to identify
m2), there is not much improvement in term of MRE. Also, pre- the effects on heat transfer.
sented model predicts a peak in wall temperature, which wasn't
observed in experiments. The reason for this dissimilarity can be 4.3.1. Influence of thermophysical properties
either inability of presented model or increase in heat transfer rate For the sCO2, very little change to the model parameter or
in the experiments. One explanation can be that during the boundary conditions may imply large discrepancies. Especially,
234 S. Pandey et al. / International Journal of Thermal Sciences 117 (2017) 227e238
Fig. 9. Variation of wall temperature with bulk fluid enthalpy (a) qw ¼ 50 kW/m2 (b) qw ¼ 30 kW/m2 (c) qw ¼ 10 kW/m2.
Fig. 10. Variation of shear stress with heated length corresponding to Fig. 9. Fig. 11. Comparison with experiments for cooling at P ¼ 9 MPa, G D ¼ 2.4 kg/m s,
qw ¼ 12 kW/m2.
coefficient for variable properties and for different scenarios where bulk fluid and wall of the pipe, similar to constant properties fluids.
one of the thermophysical property is held constant based on inlet As a result of it, wall temperature will decrease with the increasing
conditions. The inlet Reynolds number is set to 6000, wall heat flux mass flux. A peak can be seen in a at zz150D for G ¼ 219 kg/m2 s.
to 35 kW/m2, inlet pressure and temperature to 8 MPa and 298 K With the decreasing mass flux, this peak is shifting towards the
while the diameter is kept 2 mm and the heated length is 750 times inlet because increased bulk temperature will reach the pseudoc-
of diameter to resemble the channels of PCHE. ritical point earlier. Fig. 13b shows the Nusselt number (Nu ¼ alD)
b
Out of all 5 cases, thermal conductivity and dynamic viscosity variation, Nu increases with the mass flux as expected. For a given
have extreme effect on heat transfer. While holding thermal con- mass flux, it decreases as heat transfer coefficient is decreasing but
ductivity constant, heat transfer coefficient shows its peak at as soon as it reaches near the pseudocritical point, it shows a steep
94 kW/m2 (not shown in the graph). In this case, the thermal rise.
resistance offered by laminar sub-layer is zero and the temperature
difference between bulk and wall is only because of the turbulent
core layer. In another extreme scenario where dynamic viscosity 4.3.3. Influence of inlet temperature
held constant, wall temperature has the highest value compared to Fig. 14a depicts the variation of a with the heated length for
other cases due to the higher resistance offered by turbulent core different inlet temperature of sCO2. The magnitude of a decreases
layer. with the increase in inlet temperature of fluid. The reason for this
behavior is a drop in the value of isobaric specific heat and thermal
conductivity. Besides, it can be observed that the peak in heat
4.3.2. Influence of mass flux transfer coefficient is shifting towards the inlet and it disappears
Fig. 13a illustrates the effect of mass flux on the local heat after 306 K. The peak in a is only observed at the pseudocritical
transfer coefficient. As the mass flux increases, Reynolds number point but as the fluid inlet temperature increases and as it crosses
also increases proportionally thereby intensifies the turbulence. the pseudocritical temperature, this peak starts vanishing. In this
This increased turbulence augments the heat transfer between the case Nu is a function of heat transfer coefficient and thermal
Fig. 12. Influence of thermophysical properties on (a) Wall temperature (b) Heat Fig. 13. (a) Local heat transfer coefficient (b) Bulk Nusselt number variation; for
transfer coefficient. different mass flux of CO2 for upward flow, P ¼ 8 MPa, D ¼ 2 mm, Ti ¼ 301.1 K.
236 S. Pandey et al. / International Journal of Thermal Sciences 117 (2017) 227e238
Fig. 14. (a) Local heat transfer coefficient (b) Bulk Nusselt number variation; for
Fig. 15. (a) Local heat transfer coefficient (b) Bulk Nusselt number; for different in-
different inlet temperature of CO2, P ¼ 8 MPa, D ¼ 2 mm, Rei ¼ 5400.
ternal diameter of pipe, p ¼ 8 MPa, Rei ¼ 5400, Ti ¼ 301.1 K.
increasing again. For D ¼ 2:5 mm, bulk temperature crosses the Tpc pressure is a long path and the model presented here is step for-
at zz76D, thereby after it Nu is increasing. A similar observation ward in that direction.
can be noticed from other cases as well.
6. Acknowledgement
4.4. Need of further experiments and DNS database
The first and third author acknowledge the financial support
In the present study, only 3 cases of DNS were used to calibrate provided by Forschungsinstitut für Kerntechnik und Ener-
the model. To have a more comprehensive, reliable and generalize giewandlung (KE) e.V. in terms of doctoral research fellowship. All
model, there is need to calibrate with more DNS cases. In this study, authors extend their sincere thanks to the High Performance
the focus was only on the heating of sCO2 but there are wide ap- Computing Center (HLRS) of the University of Stuttgart for granting
plications of sCO2 in cooling ranges from refrigeration industry to access to high-performance computers for DNS.
power plants. Therefore, DNS cases are needed for cooling as well.
Furthermore, DNS provides an understanding of the physics, which
is still unavailable for cooling of sCO2 as none DNS study was References
conducted in this domain. Moreover, validation is an important
[1] Anderson JD. Ludwig Prandtl's boundary layer. Phys Today 2005;58(12):42e8.
part in the process of reliable model development. But the avail- http://scitation.aip.org/content/aip/magazine/physicstoday/article/58/12/10.
ability of experimental data obstructs the development process. In 1063/1.2169443.
the literature, few experiments are available with the exhaustive [2] Bae JH, Yoo JY, Choi H. Direct numerical simulation of turbulent supercritical
flows with heat transfer. Phys Fluids 2005;17(10). http://scitation.aip.org/
detail of parameters for small diameter pipes which correspond to content/aip/journal/pof2/17/10/10.1063/1.2047588.
PCHE's channels. Therefore, authors urge to experimentalist to [3] Bae JH, Yoo JY, McEligot DM. Direct numerical simulation of heated CO2 flows
conduct the experiments in the range of application. As this model at supercritical pressure in a vertical annulus at Re¼8900. Phys Fluids
2008;20(5). http://scitation.aip.org/content/aip/journal/pof2/20/5/10.1063/1.
is based upon coupled hydraulic and heat transfer in which both 2927488.
wall shear stress and wall temperature were used to optimize the [4] Bae YY, Kim HY. Convective heat transfer to CO2 at a supercritical pressure
model. So the measurements of skin friction coefficient along with flowing vertically upward in tubes and an annular channel. Exp Therm Fluid
Sci 2009;33(2):329e39. http://www.sciencedirect.com/science/article/pii/
wall temperature are required from the experiments. The temper- S0894177708001519.
ature gradient in the radial direction assumed to follow the loga- [5] Bovard S, Abdi M, Nikou MRK, Daryasafar A. Numerical investigation of heat
rithmic law similar as velocity profile in the turbulent flow, in the transfer in supercritical CO2 and water turbulent flow in circular tubes.
J Supercrit Fluids 2017;119:88e103. http://www.sciencedirect.com/science/
theory presented here. But at higher Re with strong heat flux, this
article/pii/S0896844616303059.
has to be verified with experiments. As stated earlier cooling is [6] Chu X, Laurien E. Flow stratification of supercritical CO2 in a heated horizontal
equally important as heating thus experiments are necessary pipe. J Supercrit Fluids 2016;116:172e89. http://www.sciencedirect.com/
especially with near-constant heat flux with both upward and science/article/pii/S0896844616301048.
[7] Chu X, Laurien E, McEligot DM. Direct numerical simulation of strongly heated
downward direction flow. air flow in a vertical pipe. Int J Heat Mass Transf 2016;101:1163e76. http://
www.sciencedirect.com/science/article/pii/S0017931016305142.
5. Conclusion [8] Colburn AP. A method of correlating forced convection heat transfer data and
a comparision with fluid friction. Trans AIChE J 1933;29:174e210.
[9] Dang C, Hihara E. In-tube cooling heat transfer of supercritical carbon dioxide.
This work was initiated with the aim of development of a reli- part 1. experimental measurement. Int J Refrig 2004;27(7):736e47. http://
able analytical model which is simple in implementation like cor- www.sciencedirect.com/science/article/pii/S0140700704000817.
[10] Dittus FW, Boelter LMK. Heat transfer in automobile radiators of the tubular
relations. The advantage of presented model over turbulent model type. Publications in Engineering. Berkeley 2: University of California; 1930.
is, its simplicity and easy applicability, also this model uses very p. 443.
little computer resources as compared to DNS and typical CFD [11] Dostal V, Driscoll M, Hejzlar P. A supercritical carbon dioxide cycle for next
generation nuclear reactors. Tech. rep. MIT; March 2004. http://web.mit.edu/
modelings. In this work, two layer theory for the turbulent flow was
22.33/www/dostal.pdf.
used with a different approach. The presented model has shown [12] EcoCute. Transcritical CO2 heat pumps. the future is now, and it's beautiful.
superiority over the previous model while taking the buoyant and May 2016. http://www.mayekawa.com/products/features/eco_cute/.
[13] Fang X, Xu Y. Modified heat transfer equation for in-tube supercritical CO2
acceleration effects. To improve the model, DNS database served as
cooling. Appl Therm Eng 2011;31(14e15):3036e42. http://www.
a benchmark to calibrate. The validity of presented model is sup- sciencedirect.com/science/article/pii/S1359431111002973.
ported by verification with three cases videlicet (i) constant prop- [14] Gnielinski V. New equations for heat and mass transfer in turbulent pipe and
erty; (ii) experimental database in the pseudocritical range and (iii) channel flow. Int Chem Eng 1976;16(2):359e67.
[15] Kays WM, Crawford ME, Weigand B. Convective heat and mass transfer. Tata
cooling case with high mass flux to heat flux ratio. In all cases, an McGraw-Hill Education; 2012.
acceptable agreement was achieved. A parametric study was also [16] Kim H, Bae Y, Kim H, Soong J, Cho B. Experimental investigation on the heat
presented to identify the effect of different variables on heat transfer characteristics on a vertical upward flow of supercritical CO2. In:
Proc. ICAPP; June 2006 [Reno, NV].
transfer. Following points summarize the parametric study: [17] Kuang G, Ohadi MZY. Experimental study of miscible and immiscible oil ef-
fects on heat transfer coefficients and pressure drop in microchannel gas
Thermal conductivity and viscosity have the limiting effect on cooling of supercritical co2. In: ASME 2003 heat transfer summer conference;
2003 [Las Vegas, Nevada, USA].
the heat transfer to sCO2. [18] Laurien E. Implicit model equation for hydraulic resistance and heat transfer
Heat transfer increases with mass flux for a given diameter, inlet including wall roughness. J Nucl Eng Radiat Sci April 2016;2(2). http://
temperature, and pressure. nuclearengineering.asmedigitalcollection.asme.org/article.aspx?
articleID¼2512927.
The pipe with smaller diameter has the minor effect of buoyancy [19] Laurien E, Pandey S, McEligot DM. Two-layer model for the heat transfer to
for given inlet conditions. supercritical CO2. In: 5th international supercritical CO2 power cycles sym-
posium; March 2016. p. 28e31. San Antonio, Texas, USA, http://www.swri.
org/4org/d18/sco2/papers2016/FluidMechanics/030paper.pdf.
The validity of this model can only be warranted in the validated
[20] Lemmon E, Huber M, McLinden M. NIST standard reference database 23:
range of given study. For other cases, credibility is uncertain reference fluid thermodynamic and transport properties-refprop. version 9.1.
because of unavailability of experimental data within the range of 2013.
PCHE. Therefore, a request is made to the experimentalist for a [21] MATLAB, 2013. version 8.1.0.604 (R2013a), The MathWorks Inc., Natick,
Massachusetts.
comprehensive database. Development of a universal heat transfer [22] McEligot D, Coon C, Perkins H. Relaminarization in tubes. Int J Heat Mass
model for sCO2 that remains valid for all ranges of temperature and Transf 1970;13:431e3. http://www.sciencedirect.com/science/article/pii/
238 S. Pandey et al. / International Journal of Thermal Sciences 117 (2017) 227e238
0017931070901183. S0065271708701539.
[23] McEligot DM, Jackson JD. “Deterioration” criteria for convective heat transfer [32] Pitla SS, Groll EA, Ramadhyani S. New correlation to predict the heat transfer
in gas flow through non-circular ducts. Nucl Eng Des 2004;232(3):327e33. coefficient during in-tube cooling of turbulent supercritical CO2. Int J Refrig
http://www.sciencedirect.com/science/article/pii/S0029549304001621. 2002;25(7):887e95. http://www.sciencedirect.com/science/article/pii/
[24] McEligot DM, Laurien E. Insight from simple heat transfer models. In: The 7th S0140700701000986.
international symposium on supercritical water-cooled reactors(ISSCWR-7); [33] Pucciarelli A, Borroni I, Sharabi M, Ambrosini W. Results of 4-equation tur-
15-18 March, 2015 [Helsinki, Finland]. bulence models in the prediction of heat transfer to supercritical pressure
[25] Mohseni M, Bazargan M. Modification of low reynolds number k-epsilon fluids. Nucl Eng Des 2015;281:5e14. http://www.sciencedirect.com/science/
turbulence models for applications in supercritical fluid flows. Int J Therm Sci article/pii/S0029549314005809.
2012;51:51e62. http://www.sciencedirect.com/science/article/pii/ [34] Shah RK, Sekulic DP. Fundamentals of heat exchanger design. John Wiley &
S129007291100247X. Sons, Inc; 2007.
[26] Mori H, Kaida T, Ohno M, Yoshida S, Hamamoto Y. Heat transfer to a super- [35] Sharabi M, Ambrosini W, He S, Jackson J. Prediction of turbulent convective
critical pressure fluid flowing in sub-bundle channels. J Nucl Sci Technol heat transfer to a fluid at supercritical pressure in square and triangular
March 2012;49(4):373e83. http://www.tandfonline.com/doi/pdf/10.1080/ channels. Ann Nucl Energy 2008;35(6):993e1005. http://www.sciencedirect.
00223131.2012.669240. com/science/article/pii/S030645490700309X.
[27] Narasimha R, Sreenivasan KR. Relaminarization in highly accelerated turbu- [36] Shenoy A. Effects of buoyancy on heat transfer during turbulent flow on drag
lent boundary layers. J Fluid Mech 1973;61(3):417e47. http://journals. reducing fluids in vertical pipes. Heat Mass Transf January 1987;21(1):15e8.
cambridge.org/action/displayAbstract? http://link.springer.com/article/10.1007%2FBF01008213.
fromPage¼online&aid¼372898&fileId¼S0022112073000790. [37] Taylor G. The criterion for turbulence in curved pipes. Proc R Soc Lond Ser A
[28] OpenFOAM. OpenFOAM. The open source CFD toolbox. User guide. Open- June 1929;124(794). Containing Papers of a Mathematical and Physical
FOAM Foundation Ltd; 2015. Character, http://rspa.royalsocietypublishing.org/content/124/794/243.
[29] Pandey S, Laurien E. Heat transfer analysis at supercritical pressure using two [38] Yamagata K, Nishikawa K, Hasegawa S, Fujii T, Yoshida S. Forced convective
layer theory. J Supercrit Fluids March 2016;109:80e6. http://www. heat transfer to supercritical water flowing in tubes. Int J Heat Mass Transf
sciencedirect.com/science/article/pii/S0896844615301789. 1972;15(12):2575e93. http://www.sciencedirect.com/science/article/pii/
[30] Pandey S, Laurien E, Kaushik SC. Hydraulic resistance and heat transfer 0017931072901482.
equations based on two layer theory for supercritical carbon dioxide. In: [39] Yoo JY. The turbulent flows of supercritical fluids with heat transfer. Annu Rev
Proceedings of the 23rd national heat and mass transfer conference and 1st Fluid Mech 2013;45(1):495e525. http://dx.doi.org/10.1146/annurev-fluid-
international ISHMT-ASTFE heat and mass transfer conference; 17e20 120710-101234.
December 2015. Thiruvananthapuram, India. No. IHMTC2015-1457. [40] Zhang X, Yamaguchi H. An experimental study on evacuated tube solar col-
[31] Petukhov B. Heat transfer and friction in turbulent pipe flow with variable lector using supercritical CO2. Appl Therm Eng 2008;28(10):1225e33. http://
physical properties. Advances in heat transfer, vol. 6. Elsevier; 1970. www.sciencedirect.com/science/article/pii/S1359431107002487.
p. 503e64. http://www.sciencedirect.com/science/article/pii/