Transfe

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

International Journal of Heat and Mass Transfer xxx (xxxx) xxx

Contents lists available at ScienceDirect

International Journal of Heat and Mass Transfer


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

Numerical investigation of different heat transfer behaviors of


supercritical CO2 in a large vertical tube
Qian Zhang a, Huixiong Li a,⇑, Jialun Liu b, Xianliang Lei a, Chao Wu a
a
State Key Laboratory of Multiphase Flow in Power Engineering, Xi’an Jiaotong University, Xi’an, Shaanxi 710049, PR China
b
School of Mechanical Engineering, Xi’an Shiyou University, Xi’an, Shaanxi 710065, PR China

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

Article history: The design and optimization of key heat exchange components in supercritical CO2 (sCO2) Brayton-cycle
Received 16 May 2019 need a thorough understanding to heat transfer of sCO2. As a continuation of our experiments, numerical
Received in revised form 28 September studies are performed to explore the mechanisms behind different heat transfer behaviors of sCO2
2019
occurred at different mass fluxes. Seven turbulence models are assessed against the test data, and the
Accepted 22 October 2019
Available online xxxx
v2f model and SST k-x model are recommended for low and normal mass flux cases, respectively. A novel
analysis approach is proposed by treating heat transfer of SCFs as a coupling of heat conduction of bound-
ary layer, pseudo-phase-change heat transfer of large specific heat (cp) fluid and convective heat transfer
Keywords:
Supercritical carbon dioxide
of turbulence core. For low mass flux cases, the special heat transfer enhancement (HTE) in low fluid
Heat transfer enhancement enthalpy (hb) region is mainly caused by strong buoyancy effect, which thins the thickness of viscous
Heat transfer deterioration sub-layer and promotes turbulent kinetic energy (k). But for normal mass flux case, heat transfer deteri-
Low mass flux oration (HTD) occurs due to decreasing fluid thermal conductivity (k) of viscous sub-layer and suppress-
Buoyance effect ing turbulence via buoyancy. The buffer layer plays a bridge for heat transfer from viscous sub-layer to
external turbulence region. Meanwhile, a noteworthy phenomenon is that, the heat conduction process
of boundary layer shows a strong relevance with the evolution of heat transfer behaviors, and has a great
effect on overall heat transfer of sCO2, but this is seldom concerned in former research.
Ó 2019 Elsevier Ltd. All rights reserved.

1. Introduction the effect of operating parameters on heat transfer of sCO2, it is still


hard to make the mechanisms of sCO2 unique heat transfer
The Brayton-cycle using supercritical CO2 (sCO2) as working (including the heat transfer deterioration (HTD) and enhancement
medium has attracted more and more attention in wide applica- (HTE) [16]) clear, which are seriously influenced by the sharply
tions [1], such as concentrating solar system [2], enhanced variable thermophysical properties and some additional effects
geothermal system [3], advanced nuclear system [4], and coal- (buoyancy and thermal acceleration induced by radial and axial
fired power plant [5], due to its higher efficiency and energy flux density gradients) as sCO2 crosses the pseudo-critical point.
density than the conventional Rankine-cycle with water. The Recently, both Wang et al. [17] and our former survey [18] show
sCO2 turbine can achieve higher power with fewer stages and that existing experimental studies on heat transfer of sCO2 were
much smaller size compared with the steam turbines, bringing mostly focused on small tubes with inner diameter (d) less than
lower cost, higher flexibility, and excellent peak-shaving capacity 10 mm, which catered to the application of refrigeration systems
[6]. Until now, studies on sCO2 Brayton-cycle applying to these in the past but might not be suitable for the large tubes in future
advanced systems are mostly at the beginning. The design and coal-fired and nuclear power systems with sCO2 Brayton-cycle.
optimization of key devices involving heat transfer of sCO2 are still Additionally, due to the miniaturization of sCO2 Brayton-cycle, it
a fundamental issue, which is important to the overall system effi- is inevitable to undertake the peak-shaving task during the future
ciency and should be deeply researched [7]. multi-energy complement layout in China. When power load low-
A comprehensive understanding to heat transfer of sCO2 within ers, heat transfer characteristics of sCO2 with low mass flux thus
a wide parameter range is necessary. Over the past several decades, should be considered. Our former experimental study have found
though many experiments [8–15] have been performed to study that [18], heat transfer of sCO2 with lower mass flux were saliently
different from a larger mass flux. As Fig. 1 shows, with increasing
⇑ Corresponding author. heat flux (q), wall temperature (Tw) curve at G = 400 kg/m2s turns
E-mail address: huixiong@mail.xjtu.edu.cn (H. Li). from smooth to peak and HTD occurs. But when G lowers to

https://doi.org/10.1016/j.ijheatmasstransfer.2019.118944
0017-9310/Ó 2019 Elsevier Ltd. All rights reserved.

Please cite this article as: Q. Zhang, H. Li, J. Liu et al., Numerical investigation of different heat transfer behaviors of supercritical CO2 in a large vertical tube,
International Journal of Heat and Mass Transfer, https://doi.org/10.1016/j.ijheatmasstransfer.2019.118944
2 Q. Zhang et al. / International Journal of Heat and Mass Transfer xxx (xxxx) xxx

Nomenclature

cp specific heat (kJ/kgK) Greek symbols


C heat transfer state number k thermal conductivity (kW/mK)
D inner diameter of tubes (mm) l dynamic viscosity (Pas)
g gravity (m/s2) q density (kg/m3)
G mass flux (kg/m2s) d thickness of boundary layer (m)
h fluid enthalpy (kJ/kg) s turbulence shear stress (kg/ms2)
htc heat transfer coefficient (kW/m2K)
k turbulent kinetic energy (m2/s2) Subscripts
Nu Nusselt number, ¼ htc  d=k b at bulk fluid temperature
P pressure (MPa) buf at buffer layer
Pr Prandtl number, ¼ cp  l=k db calculated by D-B correlation
Q heat flux (kW/m2) in at the inlet of the tube
R thermal-conduction resistance (m2K/W) pc at pseudo-critical temperature
Re Reynolds number, ¼ G  d=l sub at viscous sub-layer
T temperature (°C) t turbulent
u axial velocity of fluid (m/s) v viscous
us friction velocity (m/s) w at wall temperature
y distance to the inner wall of tubes (m)
y+ the nondimensional distance to the wall

(a) G = 400 kg/m2·s (normal mass flux) (b) G = 100 kg/m2·s (low mass flux)
Fig. 1. Comparison of heat transfer behaviors of sCO2 under different mass fluxes [18].

100 kg/m2s, Tw keeps slowly rising at much larger q/G and heat (YS) model [24], and Myong-Kasagi (MK) model [25], were mostly
transfer is yet enhanced instead of deterioration. This monotoni- used and recommended for sCO2 flowing in small tubes (d <
cally rising Tw at low G but high q/G has also been displayed in ~10 mm). But in recent studies with large tubes, the Shear-Stress
few other studies [19,20]. However, the mechanisms to this special Transport k-x (SST) model [26] was popularly employed and pre-
HTE and different heat transfer behaviors under different mass flux sented superior accuracy in simulating heat transfer of sCO2.
conditions have still no explanation in published papers. Table 1 also shows that the model validation and numerical anal-
In another respect, owing to the difficulties in experimentally ysis for heat transfer of sCO2 are mostly focused on relative large
measuring flow fields and distributions of supercritical fluids mass flux conditions (G > 300 kg/m2s) but are deficiently studied
(SCFs) with high pressure and temperature, numerical methods with low mass flux (G < 300 kg/m2s), whose heat transfer behav-
provide an approach for exploring the complicated heat transfer iors are apparently different from a G > 300 kg/m2s (as shown in
mechanisms of SCFs. One of the most important issues for accu- Fig. 1). The evaluation of turbulence models for predicting heat
rately modeling heat transfer of SCFs is the chosen of turbulence transfer of sCO2 with low mass flux is needed.
model. Table 1 summarizes the turbulence models employed by To understand the heat transfer characteristics of SCFs, Xu et al.
different researchers for simulating heat transfer of sCO2. It sug- [38] discussed the relationship between the location of Tpc in radial
gests that, the v2f model [21] and various Low-Reynolds number direction and heat transfer of sCO2, and considered that HTD was
k-e models (LRM), including Launder and Sharma (LS) model mainly induced by buoyancy effect reducing the turbulence inten-
[22], Abe-Kondoh and Nagano (AKN) model [23], Yang and Shih sity when Tpc was in buffer layer. Guo et al. [37] numerically

Please cite this article as: Q. Zhang, H. Li, J. Liu et al., Numerical investigation of different heat transfer behaviors of supercritical CO2 in a large vertical tube,
International Journal of Heat and Mass Transfer, https://doi.org/10.1016/j.ijheatmasstransfer.2019.118944
Q. Zhang et al. / International Journal of Heat and Mass Transfer xxx (xxxx) xxx 3

Table 1
Turbulence models employed by researchers for simulating heat transfer of sCO2.

Authors d (mm) G (kg/m2s) Turbulence models Tubes


He et al. [27] 0.948 582 LS model Circular tube
He et al. [28] 5 646 AKN model and v2f model Circular tube
Sharabi et al. [29] 7.8 314 YS model Circular tube
Jiang et al. [30] 0.0992 1823 AKN model Circular tube
Forooghi et al. [30] 4 73, 272, 545 v2f model Inclined tube
Bae et al. [31] 4.57 400 MK model Circular tube
Cheng et al. [32] 22.8 311 SST k-x model Circular tube
Liu et al. [33] 10 303.3, 538.4 SST k-x model Circular tube
Zhang et al. [34] 19 564.3, 1000 SST k-x model Circular tube
Fan et al. [35] 38 1000–2000 SST k-x model Circular tube
Cui et al. [36] 1.8 157.2 SST k-x model Serpentine tube
Guo et al. [37] 2, 6 200, 500, 1500 SST k-x model Circular tube

analyzed heat transfer of sCO2 in 6-mm vertical tube, and also con- was set as no slip boundary. The interface between wall and fluid
cluded that buoyancy effect reducing turbulence with flatten was set as fluid-solid coupling to consider the effects of both heat
velocity distribution could impair the heat transfer. Generally, for conduction and convection. Since the pressure drop along the tube
turbulent convection in tubes, the main thermal resistance during was much smaller (<0.13%) than operation pressure, physical prop-
heat transfer from wall to main flow is usually concentrated on the erties of sCO2 were treated as independent of pressure and
heat conduction process of boundary layer. To the authors’ knowl- obtained from NIST Standard Reference Database.
edge, however, most existing studies [28,32,33,38–40] paid more
attention on the relation between Tw trend and turbulence effect, 2.2. Governing equations and solution procedure
and the HTD of SCFs was frequently ascribed to the degeneration
of local turbulence strength caused by buoyancy or thermal accel- The coupled heat transfer of sCO2 from heated wall to fluid was
eration effects. The effect of heat conduction on heat transfer of simulated by ANSYS Fluent software. For steady-state flow and
SCFs may also be influenced by buoyancy and acceleration effects heat transfer, the transport equations of mass, momentum, and
via modifying the boundary layer structure and thermal property energy can be described as follows:
distributions, but was extremely seldom discussed in past
research. Therefore, a comprehensive understanding to heat trans- @ ðqui Þ=@xi ¼ 0 ð1Þ
fer mechanisms of sCO2 is still lacked and deep analysis to variable
heat transfer behaviors of sCO2 by combining the turbulence, heat    
@ qui uj =@xj ¼ @P=@xi þ @ sij  qu0 i u0 j =@xj þ qg i ð2Þ
conduction, and pseudo-phase-change effects are necessary.
From the above, this paper performs a continuous work of our    
  l @h
previous experiments and aims to numerically reveal the mecha- @ quj h =@xj ¼ @ =@xj  @ qu0 j h =@xj ð3Þ
Pr @xj
nisms of different heat transfer behaviors of sCO2 occurred at dif-
ferent mass fluxes. Based on the experimental data, seven where q, l and u are the density, viscosity and velocity, respec-
turbulence models are assessed against the heat transfer with tively. sij defined as Eq. (4) is the viscous stress tensor.
low mass flux. The flow field details and distributions are obtained
 
with varying operating conditions. To better reveal the heat trans- sij ¼ l @ui =@xj þ @uj =@xi  2dij =3  @uk =@xk ð4Þ
fer mechanisms of SCFs, a novel analysis approach is proposed and
adopted, and the underlying mechanisms of both HTE and HTD of According to the Boussinesq approximation and turbulent vis-
sCO2 with different mass fluxes are comprehensively discussed by cosity lt, the Reynolds-stress tensor and turbulent heat flux can
considering all three effects of heat conduction of boundary layer, be written as Eq. (5) and Eq. (6).
pseudo-phase-change heat transfer and turbulent convective heat  
transfer. Finally, the relations between heat conduction process
qu0 i u0 j ¼ lt @ui =@xj þ @uj =@xi  2kdij =3 ð5Þ
of boundary layer and local heat transfer characteristic of sCO2
are focused. qu0 j h ¼ lt =Prt  @h=@xj ð6Þ
The finite-volume method was employed to solve the governing
2. Numerical methods equations and the SIMEPEC algorithm was used to treat the pres-
sure and velocity coupling. The second order upwind scheme
2.1. Physical model and boundary conditions was applied to discretize the transport equations for a high accu-
racy. The under-relaxation factors were set as proper values during
The simulating geometry employed in this study is a vertically the simulation to keep the convergence of the iteration. During the
upward circular tube with 16 mm inner-diameter and 2 mm wall simulations, variations of outlet temperature and several local wall
thickness. For a complete modeling against the experiments, a temperatures were monitored until no changing. Meanwhile, the
three-dimensional domain was built, which is exactly the same solution was considered to be convergent when the residual is less
as our experimental section [18], as presented in Fig. 2. The heated than 104 for mass equation and less than 107 for other variables.
length of the tube was 4 m (250d) and two adiabatic sections
(0.35 m) were reserved at both two ends to weak the entrance 2.3. Mesh generation
and exit effects and obtain a full-developed turbulent flow. The
working fluid (sCO2) was assumed as incompressible. Constant To achieve a better convergency, structured meshes were gen-
mass flux with preset temperature was adopted at the inlet, and erated for both fluid domain and solid domain by the ANSYS ICEM
pressure outlet condition was set at the outlet. Uniform heat flux software. It should be noted that, due to flow boundary layer, large
condition was exerted on the outer tube wall, and the inner wall temperature gradient exists near the wall. The sharply varied

Please cite this article as: Q. Zhang, H. Li, J. Liu et al., Numerical investigation of different heat transfer behaviors of supercritical CO2 in a large vertical tube,
International Journal of Heat and Mass Transfer, https://doi.org/10.1016/j.ijheatmasstransfer.2019.118944
4 Q. Zhang et al. / International Journal of Heat and Mass Transfer xxx (xxxx) xxx

Fig. 2. Details and boundary conditions of the present simulating geometry.

 
properties in this thin layer have a significant effect on local heat @ ðqui kÞ=@xi ¼ @=@xj l þ lt =rk  @k=@xj þ Gk  Y k ð7Þ
transfer of SCFs. Therefore, finer meshes were built within the
 
near-wall fluid region to capture the abrupt properties, as dis- @ ðqui xÞ=@xi ¼ @=@xj l þ lt =rx  @ x=@xj þ Gx  Y x þ Dx ð8Þ
played in Fig. 3. The distance from wall to the first node was set
as 0.0005 mm to keep the nondimensional wall distance (y+) is where Gk, Yk, Gx and Yx are the generations and dissipations of k
always less than 1.0, which is recommended by many researchers and x, respectively, and detailed expressions and constants can
[34,40,41] to capture the details in boundary layer. In the solid be referred in [43] and [26].
domain, coarser grid was adopted due to only heat conduction But for lower mass flux cases (G < 300 kg/m2s), which was little
effect. The interface between wall and fluid was covered with gra- involved in former literatures, models was not recommended and
dient meshes to keep a smooth grid transition. should be carefully examined. The v2f model, SST k-x model and
Due to little effect of axial mesh distribution on simulated five k-e models (RNG k-e, Standard k-e, Realizable k-e, two LRMs
results [42], the grid dependence was examined with four grid sys- of AKN and YS) were therefore evaluated against experimental
tems from 3,154,200 to 8,586,400 only by changing radial and cir- data. The k-e models can be generally defined as follows:
 
cumferential nodes numbers. As shown in Table 2, when the total @ ðqui kÞ=@xi ¼ @=@xj l þ lt =rk  @k=@xj þ Pk þ Gk  qðe  DÞ
cell number reaches to 6435500, the relative deviation of average
ð9Þ
Tw between Mesh 3 and Mesh 4 is only about 0.14%. Then consid-
ering less computing cost and an acceptable accuracy, the Mesh 3  
@ ðqui eÞ=@xi ¼ @=@xj l þ lt =re  @ e=@xj
grid system was employed in the following calculations.
þ ðC e1 f 1 Pk þ C e1 f 1 Gk  C e2 f 2 qeÞ þ qE ð10Þ
2
2.4. Turbulence models selection where turbulent viscosity lt ¼ qk C l f l =e, shear production
 
@ui @u
Pk ¼ lt þ @xj @u
@xj @x
i lt @T
and buoyancy term on turbulence Gk ¼ bg i Pr t t @x
.
As mentioned above, the selection of turbulence model should i j i

be seriously considered for precisely simulating complex heat For v2f model, two other equations for v2 and f are needed:
transfer of sCO2. Table 1 suggests that the SST k-x model have  

  
been validated and frequently used for sCO2 flowing with relative @ qui v 2 =@xi ¼ @=@xj l þ lt =rk  @ v 2 =@xj þ qkf  6qv 2 e=k
high mass flux (G > 300 kg/m2s) in large tubes (d > 10 mm), hence
the SST k-x model was adopted for simulations of cases with ð11Þ
G = 400 kg/m2s in this study. The transport equations of k and x
are given as:

Fig. 3. Computational domains and mesh system with near-wall refinement.

Table 2
Results of grid dependence work.

Mesh Axial nodes number Circumferential nodes number Total cell number Relative deviation of Tw against Mesh 4
1 33 76 3,154,200 3.32%
2 43 80 4,266,400 1.24%
3 55 96 6,435,500 0.14%
4 67 108 8,586,400 0%

Please cite this article as: Q. Zhang, H. Li, J. Liu et al., Numerical investigation of different heat transfer behaviors of supercritical CO2 in a large vertical tube,
International Journal of Heat and Mass Transfer, https://doi.org/10.1016/j.ijheatmasstransfer.2019.118944
Q. Zhang et al. / International Journal of Heat and Mass Transfer xxx (xxxx) xxx 5

   
L2  @=@xi ð@f =@xi Þ  f ¼ ðC 1  1Þ v 2 =k  2=3 = f l k =ev 2
2 model in reproducing the more drastic buoyancy effect at much
higher q.
 C 2 Pk =qk  5ev 2 2 =f l k
3
ð12Þ In summary, the v2f model and SST k-x model can better pre-
dict the different heat transfer behaviors of sCO2 with different
The constants and damping terms of turbulence models can be
mass fluxes in most cases and will be used for following analysis.
found in [17,27,43,44].
Fig. 4(a) displays the comparison of predicted Tw by different
3. Results and discussion
turbulence models against the test data with low mass flux
(G = 100 kg/m2s, q = 30 kW/m2). It displays that most models over-
3.1. Analysis approach to heat transfer of sCO2
rate the Tw before hpc but under-predict it when after hpc. The YS
model predicts Tw well in high hb region but has a large deviation
As is known, the particular heat transfer of SCFs are mainly
in low hb region, where special HTE occurs with lower Tw. Com-
induced by sharply varying fluid properties. Buoyancy and thermal
pared to other models, the prediction of v2f model agrees best with
acceleration effects are generated from density gradients and fur-
the experimental data over the whole hb region. Fig. 4(b) compares
ther affect heat transfer via distorting flow and turbulence fields.
the predicted Tw with experimental results at other heat fluxes. It
In fact, for turbulent flow, the boundary layer can be divided
shows that the v2f model can also accurately capture the Tw trends
into three regions from the wall to core successively: viscous
with various q from 10 to 30 kW/m2.
sub-layer, buffer layer and logarithmic-law layer. In viscous sub-
Fig. 4(c) validates the SST k-x model applied for cases with rel-
layer, flow is mostly laminar due to large viscous effect, and heat
ative high mass flux (G = 400 kg/m2s). The predicted Tw curves
transfer is dominated by heat conduction. In logarithmic-law layer,
show a good agreement with the measured ones with q from 20
flow and heat transfer are usually governed by turbulence, while
to 50 kW/m2, but a much overpredicted Tw peak is obtained when
the heat conduction and turbulence convection in buffer layer
q = 75 kW/m2. This may be induced by the failure of conventional
may have comparable effects on heat transfer. However, most

(a) Comparisons of predicted Tw via different turbulence models for low mass flux case.

(b) Validations of v2f model for G = 100 kg/m2·s (c) Validations of SST k- for G = 400 kg/m2·s
Fig. 4. Validations of turbulence models against different experimental cases.

Please cite this article as: Q. Zhang, H. Li, J. Liu et al., Numerical investigation of different heat transfer behaviors of supercritical CO2 in a large vertical tube,
International Journal of Heat and Mass Transfer, https://doi.org/10.1016/j.ijheatmasstransfer.2019.118944
6 Q. Zhang et al. / International Journal of Heat and Mass Transfer xxx (xxxx) xxx

studies only concerned the turbulence effect but neglected the heat Table 3
conduction process, which also had a considerable effect on heat Operating parameters and settings of simulated cases.

transfer of SCFs [45]. Therefore, a novel analysis approach is pro- Case P (MPa) G (kg/m2s) q (kW/m2) g (m/s2)
posed in this study by treating overall heat transfer of SCFs as a 1 7.5 100 10 9.8
coupling of three processes: heat conduction of boundary layer, 2 7.5 100 20 9.8
pseudo-phase-change heat transfer of large cp fluid, and convective 3 7.5 100 30 9.8
heat transfer of turbulence core. 4 7.5 400 20 9.8
5 7.5 400 30 9.8
The processes of sharply varying properties impacting heat 6 7.5 400 50 9.8
transfer of SCFs through above three ways are displayed in Fig. 5. 7 7.5 400 75 9.8
It presents that, the pseudo-phase-change effect is determined by
the distribution of fluid cp. The heat conduction process is mostly
influenced by the thermal resistance (R = d/k) of boundary layer
(especially the sub and buffer layers). The effect of turbulence con-
vection is usually reflected by the turbulent kinetic energy (k) of
main flow. The d and k are mainly influenced by fluid viscosity
(l) and velocity gradient (du/dy) through viscous shear stress
and turbulent shear stress, respectively, and du/dy is mostly con-
trolled by the buoyancy and thermal acceleration effects generated
from sharp density gradient.
Thus, the mechanisms of different heat transfer phenomena of
sCO2 were discussed by considering all above three heat transfer
processes. The cases simulated in this study are listed in Table 3.
The boundaries among three layers depend on local parameters
and fluid properties, and are usually defined as an approximate y+
[46]. According to Bazargan et al. [47] and Hiroaki et al. [48], the
ranges of viscous sub-layer (y+ < 12.26), buffer layer (12.26  y+
h6 0) and logarithmic-law layer (60  y+ < 300) are adopted for
sCO2 in this study. Then, the thickness of viscous sub-layer (dsub)
and buffer layer (dbuf) are determined by y+ = 12.26 and 60 and
can be calculated as follows.

yþ ¼ qyus =l; dsub ¼ 12:26l=qus ; dbuf ¼ 60l=qus  dsub ð13Þ Fig. 6. Variations of calculated C versus hb at different q with low mass flux.

Due to d  d/2, the thermal resistance of cylindrical wall can be determined as enhancement and deterioration if C is larger or
approximately equivalent to a plate. Then, the thermal resistances smaller than 1.0, respectively.
of viscous sub-layer (Rsub) and buffer layer (Rbuf) per unit area in
C ¼ htcreal =htcdb ð15Þ
tube flow can be defined as:
Z Z where htcreal is the real heat transfer coefficient of sCO2 and htcdb is
dsub 60l=qus
calculated by Dittus-Boelter correlation (Nudb = 0.023Re0.8Pr0.4)
Rsub ¼ ð1=kÞ dy; Rbuf ¼ ð1=kÞ dy  Rsub ð14Þ
0 0 using properties of sCO2 at hb = 120 kJ/kg.

where y is the vertical distance to the wall, us ¼ ðsw =qÞ0:5 is the 3.2. Analysis to heat transfer behavior of sCO2 at low mass flux
friction velocity. (G = 100 kg/m2s)
The local buoyancy force of unit mass in the radial is character-
ized by ðqb  qÞ  g=q. Fig. 6 displays the variations of calculated C versus hb at differ-
A heat transfer state number C [18] defined as Eq. (15) is ent q with low mass flux (G = 100 kg/m2s). It displays that heat
introduced to reflect the heat transfer state of sCO2, which is transfer of sCO2 is abnormally enhanced by increasing q in low

Fig. 5. Schematic of the processes of variable properties influencing heat transfer of SCFs.

Please cite this article as: Q. Zhang, H. Li, J. Liu et al., Numerical investigation of different heat transfer behaviors of supercritical CO2 in a large vertical tube,
International Journal of Heat and Mass Transfer, https://doi.org/10.1016/j.ijheatmasstransfer.2019.118944
Q. Zhang et al. / International Journal of Heat and Mass Transfer xxx (xxxx) xxx 7

hb region (hb < ~270 kJ/kg), and reaches to a peak at little less than 3.2.1. Special heat transfer enhancement in low hb region (hb <
hpc followed by a HTD in high hb region. To explain the varying heat ~270 kJ/kg)
transfer behaviors, the case with q = 30 kW/m2 is analyzed as an Fig. 7 displays the radial distributions of fluid temperature,
example. velocity, properties, and turbulence parameters at various hb with

(a) The ratio of T to Tpc (b) Density

(c) Buoyancy effect (d) The ratio of u to u0

(e) Radial gradient of axial velocity (f) Total shear stress


Fig. 7. Radial distributions of various parameters at different hb in case of P = 7.5 MPa, G = 100 kg/m2s, q = 30 kW/m2.

Please cite this article as: Q. Zhang, H. Li, J. Liu et al., Numerical investigation of different heat transfer behaviors of supercritical CO2 in a large vertical tube,
International Journal of Heat and Mass Transfer, https://doi.org/10.1016/j.ijheatmasstransfer.2019.118944
8 Q. Zhang et al. / International Journal of Heat and Mass Transfer xxx (xxxx) xxx

P = 7.5 MPa, G = 100 kg/m2s, q = 30 kW/m2. It indicates that, with Fig. 7(f), in low hb region, due to the increasing du/dy, shear stres-
the increase of heated fluid hb, the near-wall fluid temperature first ses in both near-wall region (sv) and the turbulence core (st) are
exceeds Tpc. The near-wall fluid density decreases rapidly and a enhanced.
large density gradient generated in viscous sub-layer (Fig. 7(b)). Fig. 8(a) and (b) further display the radial distributions of fluid
As shown in Fig. 7(c), increasing buoyancy force is induced by viscosity (l) and thermal conductivity (k), and the variations of
the abrupt radial density gradient in the near-wall region (y+ < thermal resistance and thickness of viscous sub-layer and buffer
~60) and reaches to maximum at hb  237 kJ/kg, where unique layer are also drawn in Fig. 9(a) and (b). It can be seen that, with
HTE occurs. This means that the buoyancy effect may has a strong near-wall fluid temperature exceeding Tpc, fluid viscosities of both
relevance with the special HTE phenomenon of sCO2 occurred in sub-layer and buffer layer drop rapidly, and the decreasing viscos-
low hb region with low mass flux. ity coupling with increasing viscous shear stress cause viscous sub-
Fig. 7(d) and (e) show the radial distributions of axial velocity layer and buffer layer thinner significantly. Meanwhile, the large k
and its gradients. It is seen that, as increasing buoyancy drives in fluid begins moving away from the wall but still within the viscous
low hb region, near-wall fluid apparently accelerates and forms a sub-layer. The the reducing degree of fluid k is less than that of the
distorted ‘‘M-shape” distribution, which causes a forward and thickness dsub and dbuf. Then Rsub and Rbuf decrease in low hb region,
backward increasing velocity gradient (du/dy) within the near- and the heat conduction process of boundary layer is enhanced.
wall region and turbulence core, respectively. Fig. 7(f) gives the To discuss the effects of pseudo-phase-change and turbulence,
radial distribution of total shear stress. In near-wall region, viscous the distributions of turbulent kinetic energy (k) and specific heat
shear stress (sv) is ascendant due to laminar effect, but in the exter- (cp) are displayed in Fig. 10(a) and (b). It shows that, due to the
nal flow, turbulent shear stress (st) is dominant. It can be seen from increasing turbulent shear stress, the k in turbulence region is

(a) Viscosity (b) Thermal conductivity


Fig. 8. Radial distributions of fluid viscosity and thermal conductivity at different hb in case of P = 7.5 MPa, G = 100 kg/m2s, q = 30 kW/m2.

(a) Viscous sub-layer (b) Buffer layer


Fig. 9. Variations of thickness and thermal resistance of viscous sub-layer and buffer layer versus hb at low mass flux.

Please cite this article as: Q. Zhang, H. Li, J. Liu et al., Numerical investigation of different heat transfer behaviors of supercritical CO2 in a large vertical tube,
International Journal of Heat and Mass Transfer, https://doi.org/10.1016/j.ijheatmasstransfer.2019.118944
Q. Zhang et al. / International Journal of Heat and Mass Transfer xxx (xxxx) xxx 9

(a) Turbulent kinetic energy (b) Specific heat


Fig. 10. Radial distributions of turbulent kinetic energy and specific heat at different hb in case of P = 7.5 MPa, G = 100 kg/m2s, q = 30 kW/m2.

(a) Buoyancy effect (b) Total shear stress

(c) sub and Rsub (d) Turbulent kinetic energy


Fig. 11. Variations of thermal and flow parameters versus q at P = 7.5 MPa, G = 100 kg/m2s.

Please cite this article as: Q. Zhang, H. Li, J. Liu et al., Numerical investigation of different heat transfer behaviors of supercritical CO2 in a large vertical tube,
International Journal of Heat and Mass Transfer, https://doi.org/10.1016/j.ijheatmasstransfer.2019.118944
10 Q. Zhang et al. / International Journal of Heat and Mass Transfer xxx (xxxx) xxx

increasing and has a positive effect on heat transfer of sCO2. The


large cp fluid always lies in viscous sub-layer and is only a small
part of all fluid, meaning that the enhanced effect of large cp fluid
on heat transfer is limited when fluid is in low hb region.
Fig. 11 further displays the variations of thermal and flow
parameters versus q in low hb region (hb  220 kJ/kg), where
HTE occurs with rising heat flux. It clearly shows that, with the
increase of q, buoyancy effect of near-wall fluid is greatly
strengthened because of larger density gradient, and the shear
stresses in both viscous sub-layer (representing sv) and turbu-
lence core (representing st) are enhanced. From Fig. 11(c) and
(d), it can be seen that, under the action of increasing sv, the
thickness of viscous sub-layer dsub is thinned and Rsub decreases
with the increase in q. The k is also increased with rising q due
to larger st. Therefore, when mass flux is lower, with the increase
of q, the processes of both heat conduction and turbulent convec-
tive heat transfer are enhanced by strong buoyancy force, then
heat transfer of sCO2 is improved and the special HTE occurs in
low hb region.
Fig. 12. Variations of calculated C versus hb at different q with normal mass flux.
3.2.2. htc peak around the hpc
As hb continuously increases to about 295.35 kJ/kg, where heat
transfer is further enhanced to peak, the sub-layer fluid tempera- 3.3. Analysis to heat transfer behavior of sCO2 at normal mass flux
ture has completely exceeded Tpc, and the location of most dra- (G = 400 kg/m2s)
matic property variation moves from viscous sub-layer to buffer
layer (Fig. 7(a)). The variation of density becomes smooth in vis- Fig. 12 displays the variations of C versus hb at different q with
cous sub-layer but abrupt in buffer layer (Fig. 7(b)), then buoyance normal mass flux (G = 400 kg/m2s). Unlike cases in Fig. 6 at
force and du/dy begins to weak in sub-layer while keeps rising in G = 100 kg/m2s, heat transfer with G = 400 kg/m2s is deteriorated
buffer layer, causing a decreasing sv and an increasing st, as shown with q increasing in low hb region, followed by a heat transfer
in Fig. 7(c), (e) and (f). recovery (C > 1.0) around the hpc. To explore the difference of heat
Figs. 8(a) and 9 display that, as hb increasing from 237 to transfer from low mass flux, the case (G = 400 kg/m2s, q = 50 kW/
295.35 kJ/kg, fluid l in viscous sub-layer decreases little, so the m2) is discussed.
decreasing sv makes a slight increase in dsub. The dbuf continu-
ously decreases due to the decrease of fluid l in buffer layer.
However, as shown in Figs. 8(b) and 10(b), with fluid 3.3.1. HTD phenomenon in low hb region
temperature increasing, large k and cp fluid shifts from viscous Fig. 13 displays the radial distributions of flow parameters and
sub-layer to buffer layer. Fluid k and cp dramatically reduce in fluid properties at various hb with P = 7.5 MPa, G = 400 kg/m2s,
viscous sub-layer but largely increase in buffer layer, then the q = 50 kW/m2. It indicates that, with hb increasing to about
Rsub rises from about 7.0 to 10.0 m2K/W but Rbuf declines from 247.8 kJ/kg (corresponding to the Tw peak of HTD), the near-wall
17.2 to the minimum about 8.0 m2K/W. The total thermal resis- fluid temperature increases to larger than Tpc. The near-wall fluid
tance of boundary layer (mainly contains Rsub and Rbuf) declines. density decreases, and large density gradient induces strong buoy-
The capacities of both heat conduction and heat absorption ancy effect, which reaches to the maximum at hb = 247.8 kJ/kg, as
(pseudo-phase-change effect of larger cp fluid) are significantly shown in Fig. 13(c). Fig. 13(d) displays that the velocity distribu-
enhanced in buffer layer, which plays a bridge for heat and tion is distorted as ‘‘M-shape” from the acceleration of near-wall
momentum transferring from near-wall region to main flow fluid driven by buoyancy force, but the velocity deformation at
between viscous sub-layer and turbulence core [40,47]. Mean- G = 400 kg/m2s is more milder and flatter than that at
while, Fig. 10(a) displays that the enhanced effect of turbulent G = 100 kg/m2s, due to a relative weaker buoyancy effect against
convection on heat transfer is also improved by the increasing the forced convection. The vertex of ‘‘M-shape” velocity, where
st and k. As a result, heat transfer of sCO2 is greatly enhanced du/dy is near to zero, is approximately located at buffer layer or
to the most (htc peak). logarithmic layer (y+  100), leading to an increasing du/dy in vis-
cous sub-layer but a diminishing du/dy in buffer layer and logarith-
3.2.3. HTD in high hb region mic layer (Fig. 13(e)). This is different from the variations of du/dy
When hb keeps increasing to high hb region (exceed hpc), all fluid at low mass flux cases, where du/dy in both viscous sub-layer and
changes to low-density fluid, and the buoyancy effect is dimin- turbulent region are enhanced by much stronger buoyancy effect
ished due to smaller density gradient (Fig. 7(c)). As shown in Figs. 7 and severely deformed velocity distribution.
(f) and 8–10, the sv in viscous sub-layer decreases, leading to an Further, as shown in Fig. 16 (a) and (b), due to the diminishing
increase in dsub. More importantly, due to Tw and Tb larger than du/dy in external turbulent region, st within buffer layer and loga-
Tpc, both k and cp of near-wall fluid drop to much lower value, rithmic layer decline and lead to k remaining at much low level,
and the thermal resistance of boundary layer (Rsub and Rbuf) which impairs the turbulent convective heat transfer process and
increases obviously. Though the k of main flow keeps increasing is bad for local heat transfer. For another, as shown in Figs. 14
due to the thermal acceleration effect, the near-wall region is com- and 15(a), the increasing du/dy and decreasing l in viscous sub-
pletely occupied by the fluid with poor heat conduction (high R) layer cause a little changing sv but a significant decrease in dsub.
and heat-carrying (low cp) capacities. The bridge of heat and When hb > ~200 kJ/kg, the sub-layer fluid mostly converts to
momentum transport is damaged. Hence, heat is hard to transfer pseudo-gas fluid with much lower l, then the dsub nearly keeps
from near-wall region to main flow, and then HTD occurs with invariable after reducing to low level. However, due to the poor
rapidly rising Tw. turbulent heat transfer, Tw rises, and the large k and cp fluid begin

Please cite this article as: Q. Zhang, H. Li, J. Liu et al., Numerical investigation of different heat transfer behaviors of supercritical CO2 in a large vertical tube,
International Journal of Heat and Mass Transfer, https://doi.org/10.1016/j.ijheatmasstransfer.2019.118944
Q. Zhang et al. / International Journal of Heat and Mass Transfer xxx (xxxx) xxx 11

(a) The ratio of T to Tpc (b) Density

(c) Buoyancy effect (d) Axial velocity

(e) Radial gradient of axial velocity (f) Viscosity


Fig. 13. Radial distributions of various parameters at different hb in case of P = 7.5 MPa, G = 400 kg/m2s, q = 50 kW/m2.

shifting from viscous sub-layer to buffer layer. The k of viscous Fig. 16(c), though large cp fluid shifts to buffer layer, due to the lit-
sub-layer falling greatly, makes Rsub strikingly increase and reach tle amount, the enhanced effect of large cp fluid on heat transfer is
to the maximum at hb = 247.8 kJ/kg, where processes of both heat limited compared to the negative effect of impaired heat conduc-
conduction and turbulent heat transfer are impaired most. From tion and turbulent convective heat transfer. Therefore, the overall

Please cite this article as: Q. Zhang, H. Li, J. Liu et al., Numerical investigation of different heat transfer behaviors of supercritical CO2 in a large vertical tube,
International Journal of Heat and Mass Transfer, https://doi.org/10.1016/j.ijheatmasstransfer.2019.118944
12 Q. Zhang et al. / International Journal of Heat and Mass Transfer xxx (xxxx) xxx

(a) Viscous shear stress (b) Thermal conductivity


Fig. 14. Radial distributions of sv and k at different hb in case of P = 7.5 MPa, G = 400 kg/m2s, q = 50 kW/m2.

(a) Viscous sub-layer (b) Buffer layer


Fig. 15. Variations of thickness and thermal resistance of viscous sub-layer and buffer layer versus hb at normal mass flux.

heat transfer state is impaired and HTD occurs with Tw rising to the this interacts with the decrease of l, leading to dsub and dbuf thin-
peak at hb = 247.8 kJ/kg. ner. Moreover, due to Tw decreasing and Tb increasing, fluid k in
viscous sub-layer has a little recovery, and more fluid is close to
3.3.2. Heat transfer recovery around the hpc the pseudo-critical region. The buffer layer are almost occupied
As hb continually increasing to about 312.8 kJ/kg, it can be seen by fluid with large k and cp. Thereby, the Rsub and Rbuf significantly
form Fig. 13(a)–(c), fluid temperature of viscous sub-layer exceeds decline and the heat conduction process of sCO2 near the wall is
Tpc, and fluid in buffer layer and external flow is close to pseudo- greatly improved. Meanwhile, the heat-carrying capacity of the
critical point. The fluid density in buffer layer decreases rapidly, bridge (buffer layer) between viscous sub-layer and main flow is
leading to buoyancy diminishing and the acceleration of main flow. also promoted by large cp fluid. As a result, all three processes of
As shown in Fig. 13(d) and (e), the acceleration effect turns weak in heat conduction, turbulent convection and pseudo-phase-change
near-wall fluid but strong in main flow, causing the abrupt ‘‘M- effect are improved, and heat transfer is recovered from HTD to
shape” velocity distribution smooth and an upswing of du/dy in HTE. Combined with the discussion in Section 3.2, it presents that
buffer layer and logarithmic layer. And then, as Fig. 16 displays, the reason of heat transfer recovery in HTD case with normal mass
the st and k in external layer are improved in some extent. The tur- flux is exactly similar to that of the htc peak in low mass case.
bulent convective heat transfer is recovered and Tw begins to
decline. 3.3.3. The second HTD in high hb region
For another, Figs. 13–16 also displays that, due to the constant When hb increases to larger than hpc (~337 kJ/kg), all fluids at
mass flux and decreasing density of heated fluid, flow is overall cross-section have changed to pseudo-gas state. As shown in
accelerated. The du/dy and sv in viscous sub-layer increase, and Figs. 13–16, fluid thermal properties (contain q, k, l and cp)

Please cite this article as: Q. Zhang, H. Li, J. Liu et al., Numerical investigation of different heat transfer behaviors of supercritical CO2 in a large vertical tube,
International Journal of Heat and Mass Transfer, https://doi.org/10.1016/j.ijheatmasstransfer.2019.118944
Q. Zhang et al. / International Journal of Heat and Mass Transfer xxx (xxxx) xxx 13

(a) Turbulent shear stress (b) Turbulent kinetic energy

(c) Specific heat


Fig. 16. Radial distributions of turbulent parameters and specific heat at different hb in case of P = 7.5 MPa, G = 400 kg/m2s, q = 50 kW/m2.

decrease to low value. The dsub and dbuf decrease and tend to con- the enhancement and impairment of heat transfer. However, in
stant. But the Rsub and Rbuf turn to increase rapidly because near- the present study, a noteworthy phenomenon from Figs. 4, 9 and
wall region (especially the viscous sub-layer and buffer layer) are 15 shows that, for G = 400 kg/m2s case, with Rsub increasing and
completely filled by low k and cp fluid, and then the capacities of dropping, Tw rises to the peak followed by a decline, indicating that
heat conduction and heat-absorption in near-wall region are sig- the thermal-conduction resistance of boundary layer may be clo-
nificantly aggravated. sely related with the Tw curve, but this was mostly neglected in
At this point, though the st and k in turbulence region drasti- the past.
cally increases due to the pronounced acceleration effect of main To explore the relationship between heat conduction process
flow, the bridge of heat exchange between near-wall region and and local heat transfer characteristic of sCO2, the total thermal-
core flow is insulated by low k and cp fluid. The heat transfer from conduction resistance of boundary layer (Rtotal) is introduced.
the wall to turbulence core is difficult. As a result, as shown in Due to that in buffer layer, the effects of conduction and convection
Figs. 4(c) and 12, Tw goes up and heat transfer of sCO2 is deterio- on heat transfer is comparable, so the heat conduction effect of
rated in high hb region once again. From the above analysis, it is buffer layer on overall heat transfer is approximatively expressed
concluded that, the mechanism of the second HTD at normal mass by the half of Rbuf. Then, the Rtotal can be defined as Rtotal = Rsub +
flux cases is the same as that of the HTD occurred in high hb region (Rbuf/2), and the inverse of Rtotal (1/Rtotal) represents the heat trans-
with low mass flux. fer capacity via conduction among overall heat transfer process.
Fig. 17 displays the relevance between overall heat transfer
3.4. Relationship between heat conduction process of boundary layer coefficient and 1/Rtotal at different cases. It clearly shows that, for
and local heat transfer characteristic both low mass flux cases (G = 100 kg/m2s) and normal mass flux
cases (G = 400 kg/m2s), when 1/Rtotal increases to the maximum
As mentioned above, during discussion on heat transfer of SCFs, (Rtotal drops to minimum), the overall htc of sCO2 almost simulta-
most previous studies only focused on the turbulence factor and neously rises to the peak, while when 1/Rtotal decreases, the overall
agreed that the upturn and downturn of k were responsible for htc is declining. The htc peak and valley, where heat transfer is best

Please cite this article as: Q. Zhang, H. Li, J. Liu et al., Numerical investigation of different heat transfer behaviors of supercritical CO2 in a large vertical tube,
International Journal of Heat and Mass Transfer, https://doi.org/10.1016/j.ijheatmasstransfer.2019.118944
14 Q. Zhang et al. / International Journal of Heat and Mass Transfer xxx (xxxx) xxx

(a) Low mass flux, Case 1 (b) Low mass flux, Case 2

(c) Low mass flux, Case 3 (d) Normal mass flux, Case 4

(e) Normal mass flux, Case 5 (f) Normal mass flux, Case 6
Fig. 17. The variations of htc and 1/Rtotal versus hb at different mass flux conditions.

and worst, are well corresponding to the decrease and increase of helpful for more thoroughly understanding the mechanisms of
Rtotal. Moreover, it also can be seen that, with high q/G, the rele- sCO2 complex heat transfer behaviors, and provides a feasible path
vance between htc and 1/Rtotal at low mass flux are more consistent for improving the accuracy of heat transfer predictive models by
than normal mass flux cases. It can be concluded that, the variation correlating the heat conduction of boundary layer of SCFs.
of Rtotal has a strong relevance with the overall heat transfer and
can significantly dominate it. The evolution of heat transfer behav- 4. Conclusions
iors of sCO2 is highly dependent on the heat conduction process of
boundary layer, especially under low mass flux conditions, how- As a continuity of our experiments, numerical works are per-
ever, which is seldom noticed in former studies. This would be formed to explore the mechanisms behind different heat transfer

Please cite this article as: Q. Zhang, H. Li, J. Liu et al., Numerical investigation of different heat transfer behaviors of supercritical CO2 in a large vertical tube,
International Journal of Heat and Mass Transfer, https://doi.org/10.1016/j.ijheatmasstransfer.2019.118944
Q. Zhang et al. / International Journal of Heat and Mass Transfer xxx (xxxx) xxx 15

behaviors of sCO2 occurred at varying mass flux conditions. Follow- [3] S.-M. Lu, A global review of enhanced geothermal system (EGS), Renew.
Sustain. Energy Rev. 81 (2018) 2902–2921.
ing conclusions are obtained:
[4] V. Dostal, M.J. Driscoll, P. Hejzlar, A supercritical carbon dioxide cycle for next
generation nuclear reactors, MIT-ANP-TR-100, 2004.
(1) Compared with the turbulence models of SST k-x, RNG k-e, [5] J. Xu, E. Sun, M. Li, H. Liu, B. Zhu, Key issues and solution strategies for
Standard k-e, Realizable k-e, AKN and YS, the v2f model is supercritical carbon dioxide coal fired power plant, Energy 157 (2018) 227–246.
[6] Y. Zhang, H. Li, W. Han, W. Bai, Y. Yang, M. Yao, Y. Wang, Improved design of
more better to predict heat transfer behaviors of sCO2 at supercritical CO2 Brayton cycle for coal-fired power plant, Energy 155 (2018)
low mass flux (G = 100 kg/m2s). For normal mass flux cases 1–14.
with large tubes (G = 400 kg/m2s), SST k-x model is recom- [7] J. Guo, X. Huai, Performance analysis of printed circuit heat exchanger for
supercritical carbon dioxide, J. Heat Transfer 139 (6) (2017) 061801–061809.
mended and can well capture the heat transfer deterioration [8] P.J. Bourke, D.J. Pulling, L.E. Gill, W.H. Denton, Forced convective heat transfer
and recovery but seriously over-predicts the Tw peak when q to turbulent CO2 in the supercritical region, Int. J. Heat Mass 13 (8) (1970)
is much higher. 1339–1348.
[9] J. Fewster, Mixed Forced and Free Convective Heat Transfer to Supercritical
(2) By treating the overall heat transfer of SCFs as a coupling of Pressure Fluids Flowing in Vertical Pipes, University of Manchester,
three processes: heat conduction of boundary layer, pseudo- Manchester, 1976.
phase-change heat transfer of large cp fluid, and convective [10] P.X. Jiang, Y. Zhang, R.F. Shi, Experimental and numerical investigation of
convection heat transfer Of CO2 at supercritical pressures in a vertical mini-
heat transfer of turbulence core, a new analysis approach tube, Int. J. Heat Mass 51 (11–12) (2008) 3052–3056.
is proposed to thoroughly understand the heat transfer [11] Y.Y. Bae, H.Y. Kim, D.J. Kang, Forced and mixed convection heat transfer to
behaviors of sCO2. supercritical CO2 vertically flowing in a uniformly-heated circular tube, Exp.
Therm. Fluid Sci. 34 (8) (2010) 1295–1308.
(3) In low hb region, the special HTE at low mass flux cases is
[12] M.M. Ehsan, Z. Guan, A.Y. Klimenko, A comprehensive review on heat transfer
mainly induced by the strong buoyancy effect, which and pressure drop characteristics and correlations with supercritical CO2
enhances the processes of both heat conduction and turbu- under heating and cooling applications, Renew. Sustain. Energy Rev. 92 (2018)
lence convection by reducing dsub and promoting k. But rel- 658–675.
[13] N. Kline, F. Feuerstein, S. Tavoularis, Onset of heat transfer deterioration in
ative weak buoyancy effect suppressing k and the shift of vertical pipe flows of CO2 at supercritical pressures, Int. J. Heat Mass 118
large k fluid from viscous sub-layer to buffer layer causing (Suppl. C) (2018) 1056–1068.
increasing thermal-conduction resistance are the main rea- [14] B. Zhu, J. Xu, X. Wu, J. Xie, M. Li, Supercritical ‘‘boiling” number, a new
parameter to distinguish two regimes of carbon dioxide heat transfer in tubes,
son of HTD occurring at normal mass flux. Int. J. Therm. Sci. 136 (2019) 254–266.
(4) The buffer layer plays a bridge for heat transferring from [15] X. Lei, J. Zhang, L. Gou, Q. Zhang, H. Li, Experimental study on convection heat
near-wall region to main flow between viscous sub-layer transfer of supercritical CO2 in small upward channels, Energy 176 (2019)
119–130.
and turbulence core. When hb is close to hpc, buffer layer is [16] I.L. Pioro, R.B. Duffey, Experimental heat transfer in supercritical water flowing
mostly occupied by large cp and k fluid. The capacities of inside channels (survey), Nucl. Eng. Des. 235 (22) (2005) 2407–2430.
both heat conduction and pseudo-phase-change heat [17] J. Wang, Z. Guan, H. Gurgenci, K. Hooman, A. Veeraragavan, X. Kang,
Computational investigations of heat transfer to supercritical CO2 in a large
absorption are greatly enhanced, and this is the reason for horizontal tube, Energy Convers. Manage. 157 (2018) 536–548.
HTE occurring in near hpc region. But in high hb region, near [18] Q. Zhang, H. Li, X. Kong, J. Liu, X. Lei, Special heat transfer characteristics of
wall region is completely occupied by low cp and k fluid. The supercritical CO2 flowing in a vertically-upward tube with low mass flux, Int. J.
Heat Mass 122 (2018) 469–482.
processes of both heat conduction and pseudo-phase-
[19] M.E. Shitsman, Paper 6: natural convection effect on heat transfer to a
change heat transfer are sharply weakened. The heat turbulent water flow in intensively heated tubes at supercritical pressures,
exchange bridge between the wall and main flow is broken, Proc. Inst. Mech. Eng. 182 (9) (1967) 36–41.
and then heat is difficult to transfer from near-wall region to [20] D.E. Kim, M.H. Kim, Experimental investigation of heat transfer in vertical
upward and downward supercritical CO2 flow in a circular tube, Int. J. Heat
the strong turbulence core. This causes the HTD of sCO2 in Fluid Flow 32 (1) (2011) 176–191.
high hb region. [21] M. Behnia, S. Parneix, P.A. Durbin, Prediction of heat transfer in an
(5) The variation of 1/Rtotal versus hb is highly consistent with axisymmetric turbulent jet impinging on a flat plate, Int. J. Heat Mass 41
(12) (1998) 1845–1855.
the overall htc trends of sCO2. The heat conduction process [22] B.E. Launder, B.I. Sharma, Application of the energy-dissipation model of
of boundary layer has a strong relevance with the evolution turbulence to the calculation of flow near a spinning disc, Lett. Heat Mass
of heat transfer behaviors of sCO2. This would be helpful for Transfer 1 (2) (1974) 131–137.
[23] K. Abe, T. Kondoh, Y. Nagano, A new turbulence model for predicting fluid flow
deep-understanding the mechanisms of sCO2 complex heat and heat transfer in separating and reattaching flows—I. Flow field
transfer behaviors and developing more precise heat trans- calculations, Int. J. Heat Mass 37 (1) (1994) 139–151.
fer predictive models of SCFs by considering the heat con- [24] Z. Yang, T.H. Shih, New time scale based k-epsilon model for near-wall
turbulence, AIAA J. 31 (7) (1993) 1191–1198.
duction effect of boundary layer in further research. [25] H.K. Myong, N. Kasagi, A new approach to the improvement of k-&epsilon;
turbulence model for wall-bounded shear flows, JSME Int. J.. Ser. 2, Fluids Eng.,
Heat Transfer, Power, Combust., Thermophys. Properties 33 (1) (1990) 63–72.
[26] F.R. Menter, Two-equation eddy-viscosity turbulence models for engineering
Declaration of Competing Interest applications, AIAA J. 32 (8) (1994) 1598–1605.
[27] S. He, P.X. Jiang, Y.J. Xu, R.F. Shi, W.S. Kim, J.D. Jackson, A computational study
The authors declared that there is no conflict of interest. of convection heat transfer to CO2 at supercritical pressures in a vertical mini
tube, Int. J. Therm. Sci. 44 (6) (2005) 521–530.
[28] S. He, W.S. Kim, J.D. Jackson, A computational study of convective heat transfer
Acknowledgement to carbon dioxide at a pressure just above the critical value, Appl. Therm. Eng.
28 (13) (2008) 1662–1675.
[29] M. Sharabi, W. Ambrosini, Discussion of heat transfer phenomena in fluids at
This study was supported by the National Basic Research Pro- supercritical pressure with the aid of CFD models, Ann. Nucl. Energy 36 (1)
gram of China (973 Program, Grant No. 2015CB251502) and the (2009) 60–71.
National Natural Science Foundation for Young Scientists of China [30] P. Forooghi, K. Hooman, Numerical study of turbulent convection in inclined
pipes with significant buoyancy influence, Int. J. Heat Mass 61 (2013) 310–322.
(Grant No. 51506170).
[31] Y.-Y. Bae, E.-S. Kim, M. Kim, Assessment of low-Reynolds number k-e
turbulence models against highly buoyant flows, Int. J. Heat Mass 108
References (2017) 529–536.
[32] H. Cheng, J. Zhao, M.K. Rowinski, Study on two wall temperature peaks of
supercritical fluid mixed convective heat transfer in circular tubes, Int. J. Heat
[1] J. Guo, X. Huai, K. Cheng, X. Cui, H. Zhang, The effects of nonuniform inlet fluid
Mass 113 (2017) 257–267.
conditions on crossflow heat exchanger, Int. J. Heat Mass 120 (2018) 807–817.
[33] S.-H. Liu, Y.-P. Huang, J.-F. Wang, L.K.H. Leung, Numerical investigation of
[2] M. Zeyghami, F. Khalili, Performance improvement of dry cooled advanced
buoyancy effect on heat transfer to carbon dioxide flow in a tube at
concentrating solar power plants using daytime radiative cooling, Energy
supercritical pressures, Int. J. Heat Mass 117 (Suppl. C) (2018) 595–606.
Convers. Manage. 106 (2015) 10–20.

Please cite this article as: Q. Zhang, H. Li, J. Liu et al., Numerical investigation of different heat transfer behaviors of supercritical CO2 in a large vertical tube,
International Journal of Heat and Mass Transfer, https://doi.org/10.1016/j.ijheatmasstransfer.2019.118944
16 Q. Zhang et al. / International Journal of Heat and Mass Transfer xxx (xxxx) xxx

[34] Y. Zhang, Y. Yao, Z. Li, G. Tang, Y. Wu, H. Wang, J. Lu, Low-grade heat utilization [41] X. Liu, X. Xu, C. Liu, W. Bai, C. Dang, Heat transfer deterioration in helically
by supercritical carbon dioxide Rankine cycle: analysis on the performance of coiled heat exchangers in trans-critical CO2 Rankine cycles, Energy 147 (2018)
gas heater subjected to heat flux and convective boundary conditions, Energy 1–14.
Convers. Manage. 162 (2018) 39–54. [42] Z. Li, Y. Zhai, K. Li, H. Wang, J. Lu, A quantitative study on the interaction
[35] Y.H. Fan, G.H. Tang, Numerical investigation on heat transfer of supercritical between curvature and buoyancy effects in helically coiled heat exchangers of
carbon dioxide in a vertical tube under circumferentially non-uniform heating, supercritical CO2 Rankine cycles, Energy 116 (Part 1) (2016) 661–676.
Appl. Therm. Eng. 138 (2018) 354–364. [43] ANSYS FLUENT 14.0 Theory Guide, ANSYS Inc., 2011.
[36] X. Cui, J. Guo, X. Huai, H. Zhang, K. Cheng, J. Zhou, Numerical investigations on [44] Q.L. Wen, H.Y. Gu, Numerical simulation of heat transfer deterioration
serpentine channel for supercritical CO2 recuperator, Energy 172 (2019) 517– phenomenon in supercritical water through vertical tube, Ann. Nucl. Energy
530. 37 (10) (2010) 1272–1280.
[37] J. Guo, M. Xiang, H. Zhang, X. Huai, K. Cheng, X. Cui, Thermal-hydraulic [45] D.E. Kim, M.H. Kim, Two layer heat transfer model for supercritical fluid flow
characteristics of supercritical pressure CO2 in vertical tubes under cooling in a vertical tube, J. Supercritical Fluids 58 (1) (2011) 15–25.
and heating conditions, Energy 170 (2019) 1067–1081. [46] H. Schlichting, J. Kestin, Boundary-Layer Theory, seventh ed., McGraw-Hill,
[38] R.N. Xu, F. Luo, P.X. Jiang, Buoyancy effects on turbulent heat transfer of New York, USA, 1979.
supercritical CO2 in a vertical mini-tube based on continuous wall [47] M. Bazargan, M. Mohseni, The significance of the buffer zone of boundary layer
temperature measurements, Int. J. Heat Mass 110 (2017) 576–586. on convective heat transfer to a vertical turbulent flow of a supercritical fluid,
[39] J.H. Bae, J.Y. Yoo, D.M. McEligot, Direct numerical simulation of heated CO2 J. Supercritical Fluids 51 (2) (2009) 221–229.
flows at supercritical pressure in a vertical annulus at Re=8900, Phys. Fluids 20 [48] T. Hiroaki, T. Ayao, H. Masaru, N. Nuchi, Effects of buoyancy and of acceleration
(5) (2008) 055108. owing to thermal expansion on forced turbulent convection in vertical circular
[40] Z. Li, Y. Wu, J. Lu, D. Zhang, H. Zhang, Heat transfer to supercritical water in tubes—criteria of the effects, velocity and temperature profiles, and reverse
circular tubes with circumferentially non-uniform heating, Appl. Therm. Eng. transition from turbulent to laminar flow, Int. J. Heat Mass 16 (6) (1973) 1267–
70 (1) (2014) 190–200. 1288.

Please cite this article as: Q. Zhang, H. Li, J. Liu et al., Numerical investigation of different heat transfer behaviors of supercritical CO2 in a large vertical tube,
International Journal of Heat and Mass Transfer, https://doi.org/10.1016/j.ijheatmasstransfer.2019.118944

You might also like