You are on page 1of 14

Applied Thermal Engineering 188 (2021) 116664

Contents lists available at ScienceDirect

Applied Thermal Engineering


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

Experimental and numerical investigation on heat transfer in the vertical


upward flow water wall of a 660 MW ultra-supercritical CFB boiler
Xihong Zhou, Tiantian Niu, Yafei Xin, Yinlong Li, Dong Yang *
State Key Laboratory of Multiphase Flow in Power Engineering, Xi’an Jiaotong University, Xi’an 710049, Shaanxi, China

A R T I C L E I N F O A B S T R A C T

Keywords: Heat transfer in the vertical upward water wall of a 660 MW ultra-supercritical circulating fluidized bed (CFB)
Supercritical water boiler was experimentally and numerically studied. A new variable turbulent Prandtl (Prt) model was used in the
Buoyancy effect numerical simulation. The effects of specific heat ratio, buoyancy parameter, and acceleration parameter on the
Acceleration effect
heat transfer of supercritical water were investigated through experimental data. Results show that the inde­
Turbulent Prandtl number
Heat transfer deterioration
pendence between the heat transfer coefficient and these dimensionless parameters was weak with the majority
of them, only buoyancy parameter proposed by Jackson showed a strong independence. Additional parameters
are required if specific heat ratio and acceleration parameter are used to predict the supercritical water heat
transfer. A new variable Prt model was proposed as a function of the turbulent viscosity ratio and the molecular
Prandtl number on the basis of the analyses of the effect of Prt on the supercritical water heat transfer. Exper­
imental data of supercritical water were used to validate the accuracy of this Prt model. The results indicated that
this Prt model provided accurate heat transfer predictions, especially for the heat transfer deterioration (HTD)
cases. The variable Prt model demonstrated a significant effect on the radial velocity distribution, thereby leading
to an ‘‘M” shape velocity profile. The zero velocity gradients limit the turbulent mixing, thereby leading to a
significant reduction in the turbulent kinetic energy and HTD. The safe heat transfer characteristics of a 660 MW
ultra-supercritical CFB boiler water wall were analyzed on the basis of the experimental research and simulation
with the new Prt model. The results show that the water wall adopts a smooth tube can ensure the nonexistence
of film boiling and overtemperature, and the water wall heat transfer is safe.

deterioration (HTD), due to the drastic changes in the physical proper­


1. Introduction ties of water near the pseudo-critical point (Fig. 1). Compared with HTE,
HTD has received great attention. If HTD occurs, then the local wall
Supercritical water has been widely used in various engineering temperature will rapidly increase, thereby causing safety problems, such
applications, such as large-capacity supercritical circulating fluidized as boiler water wall explosion. Therefore, many studies have focused on
bed (CFB) boiler and supercritical water-cooled reactor (SCWR). The the prediction of supercritical water flow and heat transfer.
CFB not only possesses several advantages (e.g., low nitrogen oxide Over the past few decades, researchers have conducted numerous
emissions, wide fuel adaptability, high combustion efficiency, and low experimental studies to understand the mechanism of HTD of super­
pollution control costs) but also realizes a wide range of load regulation critical fluids. Ackerman [7] proposed the “pseudo-film boiling” phe­
and direct desulfurization during combustion, which has great devel­ nomenon to explain the occurrence of HTD in supercritical water. Hall
opmental prospects [1,2]. Harbin Boiler Company Limited has recently [8] compared the experimental phenomenon of vertical upward flow
begun to develop a 660 MW ultra-supercritical CFB boiler with the with vertical downward flow. The result showed that HTD is prone to
support of the National 13th Five years Science and Technology Plan occur in an upward flow; thus, buoyancy is the root cause of HTD.
Project of China [2]. Knowledge of the heat transfer characteristics of Jackson et al. [9,10] further confirmed Hall’s idea and made an expla­
supercritical water is the basis for the development of SCWR and CFB nation. They believed that the buoyancy effect caused the forced
boilers; many researchers have carried out substantial basic research convective heat transfer to convert to mixed convective heat transfer
[3–12]. Supercritical water has special heat transfer characteristics, and triggered HTD. They proposed the theory of “a two-layer model” to
including heat transfer enhancement (HTE) and heat transfer explain the main reason. Li et al. [11] found that turbulent mixing at the

* Corresponding author.
E-mail address: dyang@mail.xjtu.edu.cn (D. Yang).

https://doi.org/10.1016/j.applthermaleng.2021.116664
Received 2 July 2020; Received in revised form 24 October 2020; Accepted 28 January 2021
Available online 5 February 2021
1359-4311/© 2021 Elsevier Ltd. All rights reserved.
X. Zhou et al. Applied Thermal Engineering 188 (2021) 116664

Nomenclature Greek letters


β thermal expansion coefficient, K− 1
Cp specific heat, J⋅kg− 1⋅K− 1 η heat efficiency of the experimental section
T temperature, ◦ C λ thermal conductivity, W⋅m− 1⋅K− 1
d diameter, m ω specific turbulent dissipation rate, 1/s
E input voltage, V τ viscous stress tensor
I input current, A
QE input electric power, W Subscripts
g gravitational acceleration, m⋅s− 2 a average
h enthalpy, J⋅kg− 1 b bulk
Δh added enthalpy, J⋅s− 1 t turbulent
k turbulent kinetic energy, m2⋅s− 2 in inner
L length, m max maximum
P pressure, Pa out outer
q heat flux, W⋅m− 2 w wall
G mass flux, kg⋅m− 2⋅s− 1 iw inner wall
u axial velocity, m/s ow outer wall
h heated
Dimensionless numbers adi adiabatic
πc specific heat ratio
KV acceleration number proposed by McEligot et al. Abbreviations/Acronyms
Bo buoyancy number proposed by Jackson. CFB circulating fluidized bed
Nu Nusselt number. SCWR supercritical water-cooled reactor
Re Reynolds number. HTD heat transfer deterioration
Pr molecular Prandtl number. HTE heat transfer enhancement
Prt turbulent Prandtl number. SST shear stress transport
y+ dimensionless distance from wall. HTC heat transfer coefficient
Exp experimental
TKE turbulent kinetic energy

front of the log-law layer (y+ = 30–50) played an important role when thermal acceleration, and buoyancy effects have significant effects on
HTD occurred. At the severe cross-sections of HTD, the flow in the near- the heat transfer of supercritical fluid. Therefore, various dimensionless
wall area is faster than that in the bulk flow. Shiralkar [12] discovered parameters that represent these effects should be studied.
another type of HTD that occurs in high heat and mass fluxes and may Numerical simulations are also used to obtain insight into the flow
appear at upward and downward flows. The author proposed that the field to efficiently understand the mechanism of heat transfer in addition
main reason for this deterioration might be the thermally induced flow to experimental research. Various attempts have been carried out to
acceleration. As previously mentioned, the physical property changes, create different numerical models and propose modifications to

Fig. 1. Physical properties of supercritical water.

2
X. Zhou et al. Applied Thermal Engineering 188 (2021) 116664

efficiently predict supercritical fluid flows [13–18]. Bae et al. [13] used considering all parameters affecting heat transfer, including the pipe
direct numerical simulation (DNS) for the first time to simulate the heat diameter, mass flux, pressure, heat flux, and working fluid. However,
transfer to supercritical pressure fluid in a vertical smooth tube. How­ these parameters have limited experimental data to verify the proposed
ever, the DNS method can only be used in the case of low Reynolds Prt model; thus, additional data are needed to verify the validity of the
number due to the high computational cost. Kim et al. [14] evaluated 11 Prt model, especially for the HTD cases. Bae [27] considered the prop­
low Reynolds models on the basis of the data simulated by Bae et al. erty variations and theoretically derived a new formula for Prt. How­
They found that the simulation results of these models were not accurate ever, the model has a complicated form, which is unsuitable for
enough. Therefore, they recommended using the k–ε–ν2–f model for engineering applications. The Prt model proposed by Tang et al. [28] is a
simulation. In HTD cases, the model could not predict the wall tem­ function of molecular Prandtl number of molecules that was only
perature profile in good agreement with the experimental data. Never­ applicable to supercritical CO2. Thus, further investigation is needed on
theless, the shear stress transport (SST) k–ω model provides a good the turbulent Prandtl model of the supercritical water heat transfer.
prediction for some HTD cases [15–18]. Zhang et al. [18] modified the This study first analyzed the effect of various dimensionless param­
effect of buoyancy on the basis of the original SST k–ω model to ensure eters on the heat transfer of supercritical water through experimental
that the calculated results are consistent with the experimental data. data. A new variable Prt model was proposed on the basis of the analyses
Therefore, the SST k–ω model is used for the numerical analysis of su­ of the effect of Prt on the supercritical water heat transfer. The experi­
percritical fluid flow and heat transfer a in circular tube. mental data of supercritical water were used to validate the accuracy of
Many researchers concentrated on the turbulent Prandtl number in this Prt model. The results indicated that this Prt model provided accu­
addition to improving the turbulent models. In turbulent flow, the Prt is rate heat transfer predictions, especially for the HTD cases. The safe heat
defined as the ratio of the eddy diffusivity of momentum to the eddy transfer characteristics of a 660 MW ultra-supercritical CFB boiler water
diffusivity of heat transfer. This ratio is used to explain the contribution wall were analyzed on the basis of the experiment and simulation with
of turbulence to energy transfer. In most fluid turbulent flow and heat this Prt model.
transfer simulations, the Prt is assumed to a constant near unity [19],
which is usually fit for the heat transfer in fluids with constant proper­ 2. Experimental method and results
ties. However, the physical properties of supercritical fluids have un­
dergone strong variations, thereby resulting in the questionable 2.1. Experimental system and test section
applicability of the constant Prt. Kang et al. [20,21] experimentally
showed that the value of Prt varied between 0.4 and 1.4. Numerical Fig. 2 schematically depicts the experimental system diagram of the
simulations by using LES revealed that the range of Prt varied between supercritical water flow and heat transfer. The diagram consists of a
0.1 and 20. Kawamura et al. [22] studied the effect of the Prandtl deionized water tank, a Ram-type pump with a maximum operating
number from 0.025 to 5 on the turbulent heat transfer through DNS. pressure of 40 MPa, regulating valves and the bypass, an experimental
They found that the Prt is independent of the molecular Prandtl number section, an independent cooling system, and a data collecting system.
except for fluids with a low Prandtl number. Mohseni and Bazargan [23] Deionized water from the water tank is driven by the high pressure ram-
tested the results of four different constant Prt ranging from 0.72 to 0.9, type pump. Some water returns to the water tank through the bypass,
thereby indicating that the buoyancy effect of upward flow leads to a while others are pumped into the heat exchanger designed to absorb
decrease in Prt. These studies show that a constant Prt near unit is un­ heat from the hot water at the outlet of the test section and for heat
reasonable for flows with significant property variations under super­ recovery. Afterward, the water reaches the preheating and vertical
critical pressures. Thus, a new Prt model should be developed to show experimental sections, which are directly heated by alternating current
the effect of property variations. power supplies of the low voltage and high current. The pressure and
Due to the important role of Prt in predicting heat transfer, there mass flux are precisely controlled by regulating the valves, and the
exists extensive literatures [19,24–28] (as shown in Table 1) describing temperature of the experimental section is simultaneously measured by
various attempts to identify its magnitude. However, research on the Prt the arranged thermocouples. Subsequently, the water from the experi­
model for supercritical water heat transfer is limited. Kays [19] dis­ mental section is cooled in the condenser with cold water from the
cussed the previous experimental results of air and water and concluded cooling tower and flows back to the tank for recycling. The working state
that Prt is primarily a function of the turbulent Peclet number in the is terminated until an abrupt increase of the wall temperature occurs, or
‘‘log” region. Prt is approximately equal to 0.85 at a large value of the when the vapor in the test section overheats.
turbulent Peclet number. The Prt is close to 1.00 in the region of y+ from Fig. 3 shows a schematic of the vertical experimental section and
0 to 5. However, these studies were based on the experimental data arrangement of the thermocouple of the smooth tube. The outer diam­
obtained from fluids at normal pressure, which do not undergo large eter and wall thickness of the smooth tube are 30 and 5.5 mm, respec­
variations in the physical properties. Mohseni et al. [26] developed a tively, and its material is 12CrMoVG. The heated length between the two
new Prt model for upward supercritical flow in circular tubes heating plates is 2000 mm. The thermal insulator is coated outside the
experimental tube to minimize heat loss. The outer wall temperature is
measured by 0.2 mm NiCr–NiSi k-type thermocouples at different cross-
Table 1
sections. The fluid temperature at the inlet and outlet of the tube is
Existing variable Prt models.
measured by 3 mm NiCr–NiSi armored thermocouples. The pressure and
Author Year Models pressure drop over the test section are measured by a pressure trans­
Myong et al. [24] 1989 1.63 ducer and a differential pressure transducer, respectively. The heat flux
Prt = 0.75 +
Ln(1 + Pr/0.0015) is calculated by the heat loss input and electric power.
Hollingsworth et al. 1989 Prt = 1.855 − tanh[0.2(y+ − 7.5) ]
[25]
Kays [19] 1994

⎨ 1.07 2.2. Data reduction and uncertainty analysis
μt /μ < 0.2
Prt = 2
⎩ + 0.85 μt /μ⩾0.2
Pet ( )( The thermal efficiency of the experimental is estimated through the
Mohseni et al. [26] 2016 pcr pref M
)( )0.3
d q
Prt,cr = − 1.067
pcr,co2 p Mref dref G2
+ enthalpy increase and input electric power, which is given by the
0.9
following:

Tang et al. [28] 2016 1.0 μt /μ < 0.2 Δh
⎨ Pr η= (1)
Prt = 0.85 + 0.2⩽μt /μ⩽10
⎩ A QE
0.85 μt /μ > 10

3
X. Zhou et al. Applied Thermal Engineering 188 (2021) 116664

Fig. 2. Components of the experimental system.

QE η
q= (4)
πdin L
The HTC can be defined as follows:
qiw
HTC = (5)
Tiw − Tb

where Tb is the bulk fluid temperature, and Tiw is the inner wall tem­
perature that corresponds to Tb in the same profile. A 1D heat conduc­
tion model is used to calculate Tiw, which is illustrated as follows:
( )
qdin 1 din2 dout
Tiw = Tow − − 2 ln (6)
2λw 2 dout − din2 din

The uncertainties of the measured and calculated parameters in the


experiment are examined using the method proposed by Moffat [29],
and the results are listed in Table 2.

2.3. Effects of dimensionless parameters

In the previous investigation, effect of operating parameters


including mass flux, heat flux and pressure has been studied extensively.
It is well acknowledged that heat transfer is deteriorated with increasing
pressure, increasing heat flux and decreasing mass flux [30]. However,
little work has been devoted to studying effect of dimensionless pa­
rameters at supercritical pressure. In the present paper, the effects of
various dimensionless parameters on heat transfer mechanism are
investigated.
The theory of single-phase subcritical forced convection heat transfer
based on the constant physical property or weak variable physical
Fig. 3. Structure of the experimental section and layout of the property cannot accurately describe the heat transfer characteristics
measuring points.
Table 2
Uncertainties of variables.
where ΔH and QE are calculated as given below:
Parameter Type Unit Relative uncertainty
1
Δh = πdin2 G(hout − hin ) (2) Pressure Measured MPa 0.95%
4 Wall temperature Measured ◦
C 0.19%
Fluid temperature Calculated ◦
C 0.29%
QE = EI (3) Mass flux Measured kg⋅m− 2⋅s− 1 1.79%
Heat flux Calculated kW⋅m− 2 6.54%
The inner wall heat flux q is calculated as shown below:
HTC Calculated kW⋅m− 2⋅K− 1 8.23%

4
X. Zhou et al. Applied Thermal Engineering 188 (2021) 116664

near the pseudo-critical point of the supercritical fluid. Consequently, no 4


gβb diw qiw
theoretical solution has been obtained for the heat transfer in the su­ Grb = (10)
percritical fluid. Many researchers have proposed a variety of relation­ λb ν2b
ships to predict the supercritical heat transfer, and most of them have McEligot et al. [33] proposed the following formula to represent the
the modified form of the Dittus–Boelter equation: effect of thermal acceleration.
Nu = aReb Prc F (7) 4βb diw qiw
KV = (11)
Reb μb Cp,b
The heat transfer deviation calculated by Dittus–Boelter correlation
is considered by correlation factor F, which is caused by the physical Fig. 4 shows the profile of the heat transfer coefficient with the
property variations, buoyancy effect, and thermal acceleration. A vari­ specific heat ratio at different test conditions to study the effect of spe­
ety of dimensionless parameters have been proposed to study the heat cific heat ratio on the heat transfer of supercritical water. The peak value
transfer characteristics. of the HTC appeared in the position of πc < 1 in each case. This finding
Gu et al. [31] proposed a specific heat ratio to describe the heat indicates the HTE may occur when the inner wall temperature is higher
transfer characteristics of supercritical fluid by investigating the phys­ than the pseudo-critical temperature, and the bulk fluid temperature is
ical properties of supercritical fluid. The specific form is as follows: lower than the pseudo-critical temperature. Such phenomenon is
CP,a because of the specific heat that reaches the maximum value at the
πC = (8) pseudo-critical point, and the inner wall temperature is higher than the
CP,b
pseudo-critical temperature. The specific heat Cp,w calculated by the
Where the average specific heat (Cp,a) is calculated as: CP,a = hw − hb
Tw − Tb
inner wall temperature is smaller than specific heat Cp,b calculated by
Jackson [32] proposed a buoyancy number (Bo) representing the the bulk fluid temperature, and Cp,a is the weighted average value of
effect caused by buoyancy in the following form: both; thus, Cp,a is smaller than Cp,b (i.e., πc < 1). The specific heat ratio
and HTC have poor variation regularity, and they are not a single-value
Bo =
Grb
(9) relationship. Therefore, the heat transfer of supercritical water cannot
Re3.425
b Prb0.8 be predicted through a specific heat ratio alone. The specific heat ratio is
The Grashof number is defined as follows: not the only factor affecting heat transfer. Other physical properties,
such as viscosity and density, still have great influence and should be
comprehensively considered.

Fig. 4. Effect of specific heat ratio on the heat transfer of supercritical water.

5
X. Zhou et al. Applied Thermal Engineering 188 (2021) 116664

Fig. 5 shows the profile of heat transfer coefficient with the buoyancy HTC values. The thermal acceleration effect dominates the heat transfer
parameter at different test conditions to study the effect of buoyancy on of supercritical water before the pseudo-critical temperature and then
the heat transfer of supercritical water. The buoyancy parameter shows gradually loses its dominant role. Therefore, the heat transfer of su­
a good single-valued relationship with HTC, and the relationship be­ percritical water cannot be accurately predicted by using the accelera­
tween them can be divided into two regions. In the region before HTC tion parameters only. The other parameters should be supplemented as
reaches the peak, HTC increases with the increase in the buoyancy the judgment basis.
parameter. Meanwhile, in the region after HTC reaches the peak, HTC
begins to decrease with the increase in the buoyancy parameter. This 3. Numerical simulation method and the variable Prt model
phenomenon indicates that the buoyancy parameter proposed by
Jackson can efficiently predict heat transfer of supercritical water. The 3.1. Numerical simulation method
cooling upward flow correlation proposed by Bruch et al. [34] uses the
Bo proposed by Jackson to characterize the effect of buoyancy and ob­ This study simulated the upward flow of supercritical water in cir­
tains a good prediction result. cular tubes with circumferentially uniform heating, and the computa­
Fig. 6 shows the profile of the heat transfer coefficient with the ac­ tional model is shown in Fig. 7. The finite volume method was used to
celeration parameter at different test conditions to study the effect of solve the governing equations. These governing equations can be
thermal acceleration on the heat transfer of supercritical water. In the described in Favre averaging form as follows:
regions where the acceleration parameters are small, and before HTC ( )
reaches the peak value, HTC basically shows a corresponding relation­ ∂ ρ̃uj
ship with the acceleration parameter. HTC increases with the increase in =0 (12)
∂xi
the acceleration parameter, and it has a strong correlation with the ac­
celeration parameter. This single-value relationship is no longer ( ) ( )
obvious. After HTC reached its peak, it began to decline with the in­ ∂ ρ̃ui ̃uj ∂ τij − ρũ
′ ′
i uj
∂p
crease of bulk fluid temperature. The acceleration parameter also =− + + ρgi (13)
∂xj ∂xi ∂xj
decreased with the increase of bulk fluid temperature. This unique
correlation disappears in the region where the acceleration parameter
value is large, and one acceleration parameter can correspond to two

Fig. 5. Effect of buoyancy on the heat transfer of supercritical water.

6
X. Zhou et al. Applied Thermal Engineering 188 (2021) 116664

Fig. 6. Effect of acceleration on the heat transfer of supercritical water.

stress tensor ρu’i u’j and turbulent heat flux terms − ρuj h’ as follows:
⎛ ⎞
∂ui ∂uj 2 ⎠
− ρu’’i u’’j = μt ⎝ + − kδij (16)
∂xj ∂xi 3

λt ∂̃
h
uj h′ =
− ρ̃ (17)
cp ∂xj

Eqs. (16) and (17) are substituted into the momentum and energy
Fig. 7. Configuration of the smooth tube. transport equations as follows:
( ) ( )
⎧ ⎡⎛ ⎞
( ) ∂ ρ̃ui ̃uj ( )
∂ ρ̃uj ̃h
∂ λ ∂̃h ∂
( ) ∂p ∂ 2 ∂ ⎨ ⎣ ⎝∂̃ui ∂̃uj ⎠
(14) = − + ρg i − ρkδ + (μ + μt ) +
∂xj
=
∂xj Cp ∂xj
+
∂xj
− ρ̃
uj h′ ∂xj ∂xi ∂xj 3 ij ∂xj ⎩ ∂xj ∂xi
⎤⎫
2 ∂̃uk ⎦ ⎬
where - represents the time average; ~ represents the variable that used − δij
the Favre average method; and ρ, u, τij, P, λ, and h indicate the fluid 3 ∂xk ⎭
density, axial velocity, fluid viscous stress tensor, pressure, thermal (18)
conductivity, and enthalpy, respectively. The fluid viscous stress tensor
( )
is defined as follows: [(
∂ ρ̃uj ̃h ) ]
[( / / ) / / ] ∂ λ λt ∂̃
h
= + (19)
τij = μ ∂̃ui ∂xj + ∂̃uj ∂xj − 2δij 3⋅∂̃uk ∂̃xk (15) ∂xj ∂xj Cp Cp ∂xj

The turbulent viscosity (μt) and turbulent thermal conductivity (λt) The turbulent Prandtl number can be defined by:
are introduced by Boussinesq hypothesis theory to simplify the Reynolds

7
X. Zhou et al. Applied Thermal Engineering 188 (2021) 116664

μ /ρ μ Cp F1 and F2 can be referred to [35], where the constants in the above


Prt = /t = t (20)
λ t ρC p λt equation can be also found.
ANSYS FLUENT was used to simulate the heat transfer process of
Eq. (20) is substituted to Eq. (19), and energy equation (19) can be supercritical water flow in the vertical upward tube. The fluid flows in
slightly rearranged as follows: the tube at constant pressure because the pressure drop in the tube is
( ) small that it can be ignored. The thermo-physical properties of the water
∂ ρ̃uj ̃h [ ( ) ]
∂ 1 μt /μ ∂̃ h were derived from NIST REFPROP database and placed into the Fluent
= μ + (21) by using the piecewise-linear function of temperature. An adiabatic
∂xj ∂xj Pr Prt ∂xj
section was set before the heating section for a fully developed flow, and
The equation set is not closed due to the existence of turbulent vis­ structured meshes were generated using ANSYS ICEM (Fig. 8). The near
cosity μt and turbulent kinetic energy k, and the SST k–ω turbulence wall region was carefully meshed to ensure that the dimensionless wall
model is introduced to obtain numerical solutions. The turbulent vis­ distance of the first layer (y+) was always less than 1, which was sug­
cosity and kinetic energy can be determined as follows: gested according to [36]. The mesh had a growth rate of 1.1 along the
( ) radial direction. The radial mesh size was adjusted for each case to
∂ ρ̃ui k [( ) ] ensure that the near wall flow features were efficiently reflected. The
∂ μ ∂k
= μ+ t + Gk − Yk (22) grid independence analysis was conducted with the cells ranging from
∂xi ∂xj σ k ∂xj
1.6 × 106 to 7.2 × 106. When the number of cells exceeds 4.9 × 106, the
( ) results are close to each other. A constant heat source term was set in the
∂ ρ̃ui ω [( ) ] solid wall to simulate the circumferentially uniform heat flux. The inlet
∂ μt ∂ω
= μ+ ⋅ + Gω − Yω + Dω (23) and outlet boundary conditions were set as the mass flux inlet and
∂xi ∂xj σω ∂xj
pressure outlet, respectively. The SIMPLEC algorithm based on the
where ω is the specific dissipation rate; and μt, Gk, Yk, Gω, Yω, and Dω pressure–velocity coupling equation was used to solve the discrete
in Eqs. (22) and (23) can be calculated as follows: equations. The momentum, turbulent kinetic energy, specific dissipation
rate, and energy equations were solved using the second-order upwind
ρa1 k
μt = (24) method to improve the accuracy. When the residuals during calculation
max(a1 ω, SF2 )
were sufficiently small, and several area-weight surface monitors
( ) remained constant, convergence was considered.
Gk = μt 2Sij Sij (25)

Yk = ρβ* kω (26) 3.2. Theoretical analysis of Prt effect on heat transfer


α
Gω = Gk (27) Eq. (21) exhibits that the effect of Prt on the heat transfer prediction
vt
can be expressed as follows:
Y ω = ρβ ω 2 (28) 1 μt /μ
+ (32)
Pr Prt
1 ∂k ∂ω
Dω = 2(1 − F1 )ρ (29)
ωσ ω,2 ∂xj ∂xj where the first (1/Pr) and second (μt/μ)/Prt terms can be regarded as the
where σ k and σω indicate the turbulent Prandtl numbers for k and ω, heat transfer contribution of the molecular conduction and the turbulent
respectively. conduction contribution, respectively.
In the viscous sublayer, the turbulence intensity is weak, and μt tends
σk = /
1
/ (30) to 0, thereby making (μt/μ) close to 0. The turbulent term is much
F1 σk,1 + (1 − F1 ) σ k,2 smaller than the molecular term, and Prt has a less effect on the heat
transfer. Meanwhile, 1/Pr has a dominant effect on the heat transfer. In
σω = /
1
/ (31) the fully developed core region, μt is large, thereby resulting in a large
F1 σ ω,1 + (1 − F1 ) σω,2 μt/μ. Accordingly, the turbulent term is much larger than molecular
term, and the effect of the latter can be ignored. Relevant experiments

Fig. 8. Computational meshes adopted in the simulation.

8
X. Zhou et al. Applied Thermal Engineering 188 (2021) 116664

show that Prt is determined to be a constant (approximately equal to underestimates the effect of turbulence reduction on heat transfer.
0.85) in the core region [19]. In the transition layer, μt has the same Accordingly, the predicted wall temperature is lower than the experi­
magnitude with μ. Accordingly, the molecular and turbulent terms mental data. These characteristics indicate that Prt should be related to
contribute to the same degree, and the value of Prt will determine the flow parameters and physical properties.
contributions of the turbulent conduction and the molecular conduction.
Thus, the Prt needs to be carefully determined in this layer.
The right side of Eq. (21) is the diffusion term of the energy gov­ 3.3. Variable Prt model
erning equation. The energy diffusion term will decrease with the in­
crease in Prt, and the predicted wall temperature will increase. Fig. 9 The above analyses reveal that the following assumptions should be
uses the SST k–ω model with different constant Prt to predict the wall adopted to develop the variable Prt model for heat transfer to super­
temperature of the supercritical water upward flow in the vertical critical water:
smooth tube. The predicted wall temperature rapidly increases with the
increase in Prt, which is consistent with the theoretical analysis. When 1. Prt has a strong influence on the predicted wall temperature. The
Prt = 0.85 is used, the predicted wall temperature is much less than the increase in Prt leads to a decrease in the energy diffusion term,
experimental one, and the position of HTD deviation is observed. When thereby leading to the increase in the predicted wall temperature.
Prt = 1.00 is used, the predicted wall temperature greatly exceeds the 2. The variation of Prt in y+ = 5–100 is key to predict the heat transfer
experimental wall temperature, thereby indicating that a constant Prt to supercritical water. Prt should be a function of the flow parameters
near unity is no longer applicable. and physical properties.
We still need to analyze the radial profile of the key parameters (fluid 3. In the viscous sublayer (y+ < 5) and the fully developed core region
density [ρ], molecular Prandtl number [Pr], turbulent viscosity ratio [μt/ (y+ > 100), Prt can be treated as a constant. According to the sug­
μ], and turbulent kinetic energy [k]) to further study the effect of Prt on gestion of Kays [19], Prt is assumed to be 1.00 in the viscous sublayer
heat transfer. Prt = 0.85 was used to simulate the experimental condi­ and is 0.85 in the fully developed core flow region.
tions with HTD in literature [7]. Fig. 9 demonstrates that the severe
cross-section of the HTD in simulation is the M section. The cross-section We propose Prt as a function of Pr to relate Prt to the physical
where the HTD does not occur in the upstream is the L section. Mean­ properties and as a function of the turbulent viscosity ratio to relate Prt
while, the cross-section where the low predicted wall temperature is in to the flow parameter on the basis of the above-mentioned assumptions.
the downstream recovery area is the N section. The radial profile of the After trying a variety of different function forms and comparing a large
key parameters of each cross-section is shown in Fig. 10. number of numerical simulations with experimental data, it is found that
Fig. 10(a) shows a slight change of the density in the sublayer of y+ the function in logarithmic form fits best. We use μt/μ to identify the
< 5 and y+ > 100. In the layer of y+ = 5–100, the density dramatically fully developed core flow region, the transition region, and the viscous
changes from gas to liquid state. Fig. 10(b) exhibits that the molecular sublayer. Thus, the following equation for Prt is proposed (denoted as
Prandtl number of the M and N sections has obvious peaks in the layer of ZY):
y+ = 5–100. This result shows that Pr in this layer is large, and the in­ ⎧
⎨ 1.0 (μ ) μt /μ < 0.2
crease in Pr means a decrease in 1/Pr; thus, the molecular conduction Prt = a − b⋅ln t ⋅Pr , 0.2⩽μt /μ⩽10 (33)
term has a low contribution to heat transfer. However, the property ⎩
0.85
μ μt /μ > 10
variation effect on heat transfer cannot be an effective response because
Prt is constant. Many numerical simulations have shown that the buffer Constants a and b in Eq. (33) are the model parameters that will be
layer and the front of the log-law region play important roles [11], optimized by the experimental data. The new Prt equation was tested in
which is consistent with the result in the layer of y+ = 5–100 in Fig. 10. several cases, including the normal heat transfer and HTD cases. By
Jackson’s “relaminarization” theory [37] indicates that when HTD oc­ contrast, a Prt equation proposed in previous literature was also tested.
curs, the turbulence in the buffer layer will be restrained because of the The model (denoted as K1) was proposed by Kays et al. [19] as follows:
buoyancy force caused by the radial density change. Fig. 10(c)–10(d) ⎧
⎨ 1.07
demonstrate that μt/μ of the M section is smaller than those of the other Prt = 2
μt /μ < 0.2
(34)
two cross-sections, thereby leading to a reduction in the turbulent ki­ ⎩ + 0.85 μt /μ⩾0.2
Pet
netic energy. This finding shows that the simulation can reproduce the
decay and recovery of turbulence. However, the constant Prt of 0.85 Where Pet = (μt /μ)Pr
cannot reproduce the heat transfer characteristics because it
4. Numerical result and discussion

4.1. Prediction of wall temperature with different Prt model

The experimental data of upward flow in the vertical tubes were


selected as reference to evaluate the performance of the different Prt
models. The working fluid was supercritical pressure water. A summary
of the operating conditions of these cases is listed in Table 3, including
the HTD cases. Figs. 11–13 show the comparison between the measured
and the predicted wall temperatures by using different Prt models.
Fig. 11 shows that the three cases (A1–A3) have the same mass flux of
407 kg/(m2/s) and pressure of 24.8 MPa. By contrast, the heat flux is
reduced from 315 kW/m2 to 158 kW/m2. Obvious HTD is observed in
cases A1 and A2, and the measured wall temperature shows a peak
followed by a valley. The predictions with constant Prt = 0.85 show
smooth variations, and nearly no HTD occurs in cases A1 and A2. The
prediction with the K1 model shows a similar feature with the experi­
mental data in case A1. However, the peak value of the wall temperature
Fig. 9. Wall temperature prediction by using different constant Prt. is much higher than the experimental data. The peak position is close to

9
X. Zhou et al. Applied Thermal Engineering 188 (2021) 116664

Fig. 10. Predicted radial profile of the key parameters at different axial locations.

Accordingly, b is larger for A2 than for A1, and no HTD occurs in case
Table 3
A3. Almost the same result is obtained when b ranging from 0.25 to 0.3
Operating conditions of the supercritical water heat transfer cases for validation.
is used to predict the wall temperatures. Constants coefficients a and b
Cases References P, G, kg/ q, Hydraulic Tin, need to be optimized by experimental data at the current stage.
MPa (m2s) kW/ diameter, mm C
Nevertheless, these analyses of the effects on a and b show that a and b in

m2
the ZY model can be modeled as functions related to the heat transfer
A1 Ackerman 24.8 407 315 18.54 311 characteristics of the supercritical water, such as representing the
[7]
A2 Ackerman 24.8 407 284 18.54 311
buoyancy effect dimensionless number Bo or representing the acceler­
[7] ation effect dimensionless number Ac. Therefore, a and b have physical
A3 Ackerman 24.8 407 158 18.54 311 significance and can be determined with additional data in the future.
[7] Fig. 12 shows the comparison between the measured and the pre­
O1 Our 32 650 474 19 312
dicted wall temperatures with different Prt models in case O1. A slight
experiment
Z1 Zhang et al. 23 597 773 10 301 increase in the measured wall temperature is observed near the pseudo-
[18] critical temperature, and the speed accelerated, thereby indicating that
HTE occurs in case O1. All Prt models show similar features with the
measured wall temperature. However, the K1 model overpredicts the
the inlet, and the K1 model failed to reproduce the measured wall wall temperatures along all the bulk fluid temperature. The prediction
temperature profile in case A2. The ZY model successfully captured the with constant value of Prt = 0.85 underpredicts the wall temperature
heat transfer features in A1 and A2. The values of constant coefficients a before HTE occurs. The predicted temperature with the ZY model has a
and b in the ZY model significantly affect the Prt calculation and the better agreement with the measured wall temperature compared with
simulation results; thus, they must be carefully determined. Fig. 11(a) those of the other Prt models.
and 11(b) demonstrate that the predicted wall temperature is in agree­ Another comparison of the measured and predicted wall temperature
ment with the experimental data with a = 1.25. However, the wall profiles for case Z1 is shown in Fig. 13. The heat fluxes for case Z1 is
temperature is predicted to be high along all the bulk fluid temperature slightly high with severe HTD. The constant Prt = 0.85 and K1 models
when a large constant (a = 1.35) is used. Nevertheless, the feature of the overpredict the wall temperatures up to 1200 ◦ C, which is slightly
predicted wall temperature is similar to when a = 1.25. Constant coef­ different from the measured wall temperature. The ZY model greatly
ficient a influences the magnitude of the wall temperature but not the improves the simulation results, and the predicted wall temperature is in
feature of the predicted wall temperature profile. Meanwhile, constant good agreement with the measured data.
coefficient b represents the degree of the contributions of the physical Overall, this study compared several measured data and the pre­
property variations and the flow conditions to heat transfer. The results dicted results obtained with different Prt models. The comparison re­
of Fig. 11(a) and 11(b) show that serious HTD will lead to a small b. veals that the ZY model is in best agreements with the measured data

10
X. Zhou et al. Applied Thermal Engineering 188 (2021) 116664

Fig. 11. Comparison of the measured and predicted wall temperature profiles as a function of bulk fluid temperature for cases A1–A3.

Fig. 12. Comparison of measured and predicted wall temperature profiles as a Fig. 13. Comparison of the measured and predicted wall temperature profiles
function of the bulk fluid temperature for case O1. as a function of the bulk fluid temperature for case Z1.

than other Prt models, thereby improving the heat transfer predictions, effects of the variable Prt model on the prediction of HTD. This typical
especially for the HTD cases. HTD case is selected from case A1, and the radial profile of each
parameter is taken at the axial location where the maximum tempera­
4.2. Analysis of the HTD mechanisms ture appears in each simulation. Accordingly, the axial position of the
cross-section for each Prt model is different. The comparison of the
The radial profiles of the key parameters in the near wall region with predicted Prt radial profile with variable Prt models is shown in Fig. 14
different Prt models are compared in Fig. 14 to deeply understand the (a). All variable Prt models have a sharp peak within y+ = 5–30.

11
X. Zhou et al. Applied Thermal Engineering 188 (2021) 116664

Fig. 14. Radial profile of Prt, axial velocity and TKE as a function of y+ at the position where the highest predicted wall temperature appears in case A1 with different
Prt models.

Accordingly, the energy diffusion in the buffer layer will reduce ac­ 50.2, and 59.4 m, respectively.
cording to Eq. (21), and the peak point will move close to the wall with Point A: Pressure (P) is 31.76 MPa, the mass flow rate (G) is 1077.6
the decrease in the predicted wall temperature. Fig. 14(b) demonstrates kg/(m2/s), the horizontal average heat flux is 107.5 kW/m2, the
that the axial velocity profile predicted through all the Prt models ap­ maximum peak heat flux is 300 kW/m2, the working fluid enthalpy is
pears in an ‘‘M” shape, which has been confirmed by many researchers 1608.2 kJ/kg, the working fluid temperature is 345.3 ◦ C, and the
[13,16]. Zero velocity gradients occur in the layer of y+ = 20–100 with working fluid state is ultra-supercritical pressure sub-cooled water (the
the variable Prt model. Consequently, the low velocity gradients corresponding pseudo-critical temperature under 32 MPa pressure is
restrained the turbulent mixing in this region, thereby resulting in a 408.02 ◦ C).
great decrease in the turbulent kinetic energy (Fig. 14[c]). The turbulent Point B: Pressure (P) is 31.66 MPa, the mass flow rate (G) is 1077.6
kinetic energy decreases to a minimum with the K1 model. Accordingly, kg/(m2/s), the horizontal average heat flux is 82.1 kW/m2, the
the wall temperature predicted by the K1 model reaches to the maximum peak heat flux is 300 kW/m2, the working fluid temperature is
maximum. 385.5 ◦ C, and the working fluid state is ultra-supercritical pressure sub-
cooled water.
5. Safety analysis of water wall heat transfer based on Point C: Pressure (P) is 31.54 MPa, the mass flow rate (G) is 1077.6
experiment and simulation kg/(m2/s), the horizontal average heat flux is 64.5 kW/m2, the
maximum peak heat flux is 300 kW/m2, the working fluid temperature is
This study aims at a design scheme of a 660 MW high-efficiency 403.5 ◦ C, and the working fluid state is ultra-supercritical pressure sub-
ultra-supercritical CFB boiler and the various possible operating condi­ cooled water.
tions, the safe heat transfer characteristics of water wall are analyzed Point D: Pressure (P) is 31.48 MPa, the mass flow rate (G) is 1077.6
according to our experimental results and numerical simulation results kg/(m2/s), the horizontal average heat flux is 56.4 kW/m2, the
with the ZY model. maximum peak heat flux is 300 kW/m2, the working fluid temperature is
According to the design scheme of the CFB boiler, the feed water 409.6 ◦ C, and the working fluid state is ultra-supercritical pressure
pump water supply pressure P = 31.91 MPa, the water wall tube size is steam.
Φ31.8 × 7 mm, and the average mass flow rate of water wall tube G = In the experiment and numerical investigation, the circumferentially
1042.631 kg/(m2/s) at the boiler maximum continue rate load. The uniform electric heating method is adopted, while the actual 660 MW
schematic of the furnace structure of the CFB boiler and the heat flux CFB boiler is a half circumferentially heating surface that receives the
distribution in the furnace along the height direction are shown in radiant heat of flue gas, and the heat transfer method is different. In the
Fig. 15. In the figure, A, B, C, and D points are four typical positions previous study, the heat transfer of supercritical fluid in tube under half
located in the 8th loop of the front wall, with heights of 17.06, 32.01, circumferentially heating condition has been investigated extensively

12
X. Zhou et al. Applied Thermal Engineering 188 (2021) 116664

Fig. 15. Schematic of the furnace structure of the CFB boiler and distribution of heat flux.

[38–40]. In comparison with the circumferentially uniform heating, half circumferentially uniform heating method were selected for experiment
circumferentially heating can effectively delay the occurrence of HTD of and numerical simulation. Table 4 shows the comparison between the
supercritical fluid, that is, its critical heat flux is higher. At the same experimental and numerical simulation results of A, B, C, and D. The
circumferential position, the wall temperature of half circumferentially measured inner wall temperature and the inner wall temperature ob­
heating is lower than that of circumferentially uniform heating. From tained by numerical simulation are similar. The wall temperatures at
the point of view of the operation safety of water wall, the circum­ points A, B, C, and D are all less than 550 ◦ C. The working condition we
ferentially uniform heating condition is more severe, and the wall choose is worse than the operating condition of the actual CFB boiler,
temperature obtained by replacing half circumferentially heating with and the obtained wall temperature is still in the safe range, which ver­
circumferentially uniform heating is more conservative. In the present ifies the water wall of a 660 MW high-efficiency ultra-supercritical CFB
study, the working conditions under the maximum peak heat flux and boiler adopts the smooth tube can ensure the nonexistence of film
boiling and overtemperature, and the water wall heat transfer is safe.

Table 4 6. Conclusions
Comparison between the experimental and the simulation results of A, B, C, and
D.
The heat transfer to supercritical water vertical upward flow in a
Item Unit Experimental Numerical Simulation smooth tube under uniform heating was experimentally and numerically
results results
studied. A new variable Prt model was used in the numerical simulation.
Ain wall temperature ◦
C 388.7 391.1 The conclusions drawn from the analysis were listed as follows:
Aout wall ◦
C 416.1 451.3
temperature
Bin wall temperature ◦
C 420.1 423.6
(a) The effects of specific heat ratio, buoyancy parameters, and ac­
Bout wall ◦
C 447.8 483.8 celeration parameters on the heat transfer of supercritical water
temperature were investigated through experimental data. The independence
Cin wall temperature ◦
C 434.1 436.3 between HTC and these dimensionless parameters was weak with
Cout wall ◦
C 462.0 496.5
the majority of them, only the buoyancy parameters proposed by
temperature
Din wall temperature ◦
C 441.4 442.9 Jackson showed a strong independence. Adding other parameters
Dout wall ◦
C 469.7 503.1 is necessary when specific heat ratio and acceleration parameters
temperature are used to predict the supercritical water heat transfer.

13
X. Zhou et al. Applied Thermal Engineering 188 (2021) 116664

(b) A new variable Prt model was proposed as a function of the tur­ [13] J.H. Bae, J.Y. Yoo, H. Choi, Direct numerical simulation of turbulent supercritical
flows with heat transfer, Phys. Fluids 17 (2005) 105104.
bulent viscosity ratio and molecular Prandtl number, which
[14] W.S. Kim, S. He, J.D. Jackson, Assessment by comparison with DNS data of
represent the contributions of the flow conditions and the phys­ turbulence models used in simulations of mixed convection, Int. J. Heat Mass
ical property changes, respectively, on the basis of the analyses of Transf. 51 (5–6) (2008) 1293–1312.
the effect of Prt on supercritical water heat transfer. The experi­ [15] D. Palko, H. Anglart, Theoretical and numerical study of heat transfer deterioration
in high performance light water reactor, Sci. Technol. Nucl. Instrum. (2008) 1–5.
mental data of supercritical water were used to validate the ac­ [16] Q.L. Wen, H.Y. Gu, Numerical simulation of heat transfer deterioration
curacy of this Prt model. The results indicated that this Prt model phenomenon in supercritical water through vertical tube, Ann. Nucl. Energy 37
provided accurate heat transfer predictions, especially for the (10) (2010) 1272–1280.
[17] Z. Shen, D. Yang, S. Wang, et al., Experimental and numerical analysis of heat
HTD cases. transfer to water at supercritical pressures, Int. J. Heat Mass Transfer 108 (2017)
(c) The variable Prt model demonstrated a significant effect on the 1676–1688.
radial velocity distribution, thereby leading to an ‘‘M” shape ve­ [18] G. Zhang, H. Zhang, H. Gu, et al., Experimental and numerical investigation of
turbulent convective heat transfer deterioration of supercritical water in vertical
locity profile. The zero velocity gradients limited the turbulent tube, Nucl. Eng. Des. 248 (2012) 226–237.
mixing, thereby leading to a significant reduction in turbulent [19] W.M. Kays, Turbulent Prandtl number – where are we, J. Heat Transfer 116 (2)
kinetic energy and resulting in HTD. (1994) 284–295.
[20] S. Kang, B. Patil, J.A. Zarate, et al., Isothermal and heated turbulent upflow in a
(d) The safe heat transfer characteristics of a 660 MW ultra- vertical annular channel – part I. Experimental measurements, Int. J. Heat Mass
supercritical CFB boiler water wall were analyzed on the basis Transfer 44 (2001) 1171–1184.
of the experiment and simulation with the ZY model. The [21] S. Kang, G. Iaccarino, Computation of turbulent Prandtl number for mixed
convection around a heated cylinder, in: Annual Research Briefs 2010, Center for
experimental research and numerical simulation results showed
Turbulence Research, Stanford University, 2010, pp. 295–304.
that the water wall of the CFB boiler adopted the smooth tube can [22] H. Kawamura, K. Ohsaka, H. Abe, K. Yamamoto, DNS of turbulent heat transfer in
ensure the nonexistence of film boiling and overtemperature, and channel flow with low to medium-high Prandtl number fluid, Int. J. Heat. Fluid
the water wall heat transfer is safe. Flow 19 (5) (1998) 482–491.
[23] M. Mohseni, M. Bazargan, Effect of turbulent Prandtl number on convective heat
transfer to turbulent flow of a supercritical fluid in a vertical round tube, ASME,
Declaration of Competing Interest J. Heat Transfer 133 (7) (2011) 071701.
[24] H.K. Myong, N. Kasagi, M. Hirata, Numerical prediction of turbulent pipe flow heat
transfer for various Prandtl number fluids with the improved k-ε turbulence model,
We declare that we do not have any commercial or associative in­ JSME Int. J., Ser. II 32 (4) (1989) 613–622.
terest that represents a conflict of interest in connection with the work [25] D. Hollingsworth, W. Kays, R. Moffat, Measurement and Prediction of the
submitted. Turbulent Thermal Boundary Layer in Water on Flat and Concave Surfaces,
Department of Mechanical Engineering, Stanford University, 1989, Report No.
HMT-41.
Acknowledgements [26] M. Mohseni, M. Bazargan, A New Correlation for the Turbulent Prandtl Number in
Upward Rounded Tubes in Supercritical Fluid Flows, ASME, J. Heat Transfer 138
(8) (2016) 081701.
This work was supported by the National Natural Science Foundation [27] Y. Bae, A new formulation of variable turbulent Prandtl number for heat transfer to
of China (No. 52076172). supercritical fluids, Int. J. Heat Mass Transfer 92 (2016) 792–806.
[28] G. Tang, H. Shi, Y. Wu, et al., A variable turbulent Prandtl number model for
simulating supercritical pressure CO2 heat transfer, Int. J. Heat Mass Transfer 102
References
(2016) 1082–1092.
[29] R.J. Moffat, Describing the uncertainties in experimental results, Exp. Thermal
[1] J. Pan, G. Wu, D. Yang, Thermal-hydraulic calculation and analysis on water wall Fluid Sci. 1 (1988) 3–17.
system of 600 MW supercritical CFB boiler, Appl. Therm. Eng. 82 (2015) 225–236. [30] W.Y. Wang, Z.Y. Liang, L. Wan, et al., Experimental investigation on heat transfer
[2] G. Tang, M. Zhang, J. Gu, et al., Thermal-hydraulic calculation and analysis on characteristics of smooth water wall tube of an ultra-supercritical CFB boiler, in:
evaporator system of a 660MWe ultra-supercritical CFB boiler, Appl. Therm. Eng. Proc. ASME 2018 Int Mechanical Engineering Congress and Exposition,
151 (2019) 385–393. IMECE2018-86137.
[3] M.E. Shitsman, Impairment of the heat transmission at supercritical pressures, High [31] H.Y. Gu, M. Zhao, X. Cheng, Experimental studies on heat transfer to supercritical
Temp. 1 (2) (1963) 237–244. water in circular tubes at high heat fluxes, Exp. Therm. Fluid Sci. 1 (2015) 22–32.
[4] M.E. Shitsman, Natural convection effect on heat transfer to a turbulent water flow [32] J.D. Jackson, Progress in developing an improved empirical heat transfer equation
in intensively heated tubes at supercritical pressures, in: Proceedings of the for use in connection with advanced nuclear reactors cooled by water at
Institution of Mechanical Engineers, SAGE Publications, 1967, pp. 36–41. supercritical pressure, in: Proc. 17th Int. Conf. Nucl. Eng, ICONE17, 2009.
[5] J. Licht, M. Anderson, M. Corradini, Heat transfer to water at supercritical [33] D.M. McEligot, C.W. Coon, H.C. Perkins, Relaminarization in tubes, Int. J. Heat
pressures in a circular and square annular flow geometry, Int. J. Heat Fluid Flow 29 Mass Transf. 13 (1970) 431–433.
(1) (2008) 156–166. [34] A. Bruch, A. Bontemps, S. Colasson, Experimental investigation of heat transfer of
[6] S. Wang, D. Yang, Y. Zhao, et al., Heat transfer characteristics of spiral water wall supercritical carbon dioxide flowing in a cooled vertical tube, Int. J. Heat Mass
tube in a 1000 MW ultra-supercritical boiler with wide operating load mode, Appl. Transf. 52 (11–12) (2009) 2589–2598.
Therm. Eng. 130 (5) (2018) 501–514. [35] ANSYS, Inc, ANSYS FLUENT Theory Guide, Canonsburg, PA, 2010.
[7] J.W. Ackerman, Pseudo-boiling heat transfer to supercritical pressure water in [36] Z. Li, Y. Wu, J. Lu, et al., Heat transfer to supercritical water in circular tubes with
smooth and ribbed tubes, Trans. ASME, J. Heat Transfer 92 (1970) 490–498. circumferentially non-uniform heating, Appl. Therm. Eng. 70 (1) (2014) 190–200.
[8] W.B. Hall, Heat transfer near the critical point, Adv. Heat. Transf. 7 (1) (1971) 86. [37] J. Jackson, Fluid flow and convective heat transfer to fluids at supercritical
[9] J.D. Jackson, W.B. Hall, Influences of buoyancy on heat transfer to fluids flowing in pressure, Nucl. Eng. Des. 264 (2013) 24–40.
vertical tubes under turbulent conditions, in: Turbulent Forced Convection in [38] J.B. Kitto, M Wiener. Effects of non-uniform circumferential heating and
Channels and Bundles, Hemisphere Publishing Corporation, New York, 1979, inclination on critical heat flux in smooth and ribbed bore tubes, in: 7th
pp. 613–640. International Heat Transfer Conference, Munich, 1982, 297-302.
[10] J.D. Jackson, M.A. Cotton, B.P. Axcell, Studies of mixed convection in vertical [39] T.K. Chen, Y.S. Luo, Z.H. Hu, et al., Investigation on the heat transfer
tubes, Int. J. Heat Fluid Flow 10 (1) (1989) 2–15. characteristics of supercritical pressure boiler spirally water wall tube, J. Eng.
[11] Z. Li, Y. Wu, G. Tang, et al., Comparison between heat transfer to supercritical Thermophys. 25 (2) (2004) 247–250.
water in a smooth tube and in an internally ribbed tube, Int. J. Heat Mass Transfer [40] M.F. Qu, D. Yang, Z.Y. Liang, et al., Experimental and numerical investigation on
84 (2015) 529–541. heat transfer of ultra-supercritical water in vertical upward tube under uniform
[12] B.S. Shiralkar, P. Griffith, The Deterioration in Heat Transfer to Fluids at and non-uniform heating, Int. J. Heat Mass Transf. 127 (2018) 769–783.
Supercritical Pressure and High Heat Fluxes, Department of Mechanical
Engineering, Massachusetts Institute of Technology, 1968, Report No. DSR-70332-
70355.

14

You might also like