Professional Documents
Culture Documents
2018 - Hybrid CFD Model
2018 - Hybrid CFD Model
PII: S0140-7007(18)30172-5
DOI: 10.1016/j.ijrefrig.2018.05.014
Reference: JIJR 3982
Please cite this article as: Yonghua You , Zhongda Wu , Hanyu Liu , Anqi Zhang , Xican Zeng ,
Xin Shen , A flexible hybrid CFD model for refrigerant mal-distribution among minichannels in parallel
flow condensers, International Journal of Refrigeration (2018), doi: 10.1016/j.ijrefrig.2018.05.014
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service
to our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and
all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Highlights
A flexible hybrid CFD model with DEFINE Macros is built for parallel flow condenser.
The proposed model could simulate fluid flow in condenser efficiently and accurately.
Notable fluid mal-distributions are found among different tubes and minichannels.
Physical mechanism of fluid mal-distribution is discussed with contours and vectors.
T
IP
CR
US
AN
M
ED
PT
CE
AC
1/28
ACCEPTED MANUSCRIPT
Yonghua You1,2,3 *, Zhongda Wu3, Hanyu Liu3, Anqi Zhang3, Xican Zeng3, Xin Shen3
1, State Key Lab. of Refractories and Metallurgy, Wuhan University of Science and
T
2, National-provincial Joint Engineering Research Center of High Temperature
IP
Materials and Lining Technology, Wuhan University of Science and Technology,
CR
Wuhan 430081, China
US
3, School of Material and Metallurgy, Wuhan University of Science and Technology,
Address: State Key Lab. of Refractories and Metallurgy, Wuhan University of Science
AC
2/28
ACCEPTED MANUSCRIPT
Abstract: In the present work, a flexible hybrid CFD model is constructed for parallel
flow condensers to compute their refrigerant mal-distributions, where the DEFINED
Macros of ANSYS Fluent are utilized to embed the one-dimensional finite element
analysis of the heat transfer and flow resistance of refrigerant flowing through flat
tubes. The parallel flow condenser in the literature is simulated with current hybrid
model, and the comparison between the numerical and experimental performances is
conducted for validation. Besides, the refrigerant flowrates in different flat tubes and
T
minichannels are predicted, and the physical mechanism of fluid mal-distribution is
IP
discussed based on contours and vectors. The current investigation demonstrates that
CR
the proposed hybrid model could simulate the refrigerant flow in parallel flow
condensers efficiently and its prediction has a reasonable precision. Besides, notable
US
mal-distributions among different flat tubes and different minichannels of same flat
tube are observed in the 1st tube pass. The tube flowrates in the rest passes are
AN
relatively uniform, while the mal-distribution among different minichannels of same
tube could be more notable, with the maximal relative deviation about 15.6%. The
M
current numerical model could be referred for the optimal design of heat exchangers.
ED
3/28
ACCEPTED MANUSCRIPT
Nomenclature
A heat transfer area(m2) Greek symbols
cp specific heat capacity (J·kg-1·K-1) ε turbulent energy dissipation rate
d differentiation (m2·s-3)
Dh hydraulic diameter (m) η fin efficiency
f volume force (N·m-3); friction factor μ dynamic viscosity (Pa·s)
h convection heat transfer coefficient ρ density (kg·m-3)
T
(W·m-2 ·K-1) λ thermal conductivity (W·m-1·K-1)
IP
i tensor χ refrigerant quality
CR
j tensor; Colburn factor ζ surface tension (N·m-1);
k turbulent kinetic energy (m2·s-2); overall ζk effective Prandtl number for k
heat transfer coefficient(W·m-2·K-1)
p pressure (Pa)
US ζε effective Prandtl number for ε
δ thickness (m)
AN
△p pressure drop (Pa)
Subscripts
Pr Prandtl number
M
air air
qm mass flowrate of minichannel(kg·s-1)
eff effective
Q heat transfer rate (W)
ED
fin fin
Re Reynolds number
l laminar; liquid saturated
St Stanton number
PT
ref refrigerant
Su Suratman number
t turbulent
t temperature (K)
CE
v vapor saturated
u velocity (m·s-1)
x, y, z coordinate axes (m)
AC
1. Introduction
A condenser is among the indispensable components of vapor compression
refrigerators. Proper design of condenser is greatly highlighted by the designers and
engineers of refrigerators. The augmentation of transport processes with mini- or
micro-scale dimensions has been hotly studied for the past decades (Kandlikar et al.
4/28
ACCEPTED MANUSCRIPT
2006), and mini-channel condensers were proposed and widely studied for automobile
air conditioning system (Lee and Yoo, 2000; Shao and Wang, 2009; Tian et al., 2014),
where small volume and light weight are greatly concerned. The mini-channel
condensers could also improve energy efficiency and decrease refrigerant charge, and
could be applied for the residential and commercial vapor refrigerators. Illán-Gómez
(2017) compared the mini-channel condensers and fin-and-tube counterparts, and the
former were found to have notable advantages of energy saving and environmental
T
protection. Qi (2016) performed experiments to evaluate the performances of heat
IP
transfer and flow resistance of mini-channel condensers for residential/commercial
CR
refrigeration system with the refrigerant of R22, R410A or R407C.
Besides experiments, numerical method is adopted for the research of mini-channel
US
condensers. Jabardo and Mamani (2003) developed a numerical model for parallel
flow condensers with flat tubes and louvered fins based on three thermodynamic
AN
states of refrigerant, and the predicted behaviors had an acceptable deviation against
experimental data. With the assumption that the condenser could be divided into three
M
performance was obtained by the method of analogy. Huang et al. (2014a) built a
flexible model of mini-channel condensers based on arithmetic mean temperature
CE
difference, where the condenser had variable tube and fin geometries and the flow and
heat transfer in each port-segment were modeled. It is noted that the above numerical
AC
superheated vapor flowing through the hole of flow distributor, the refrigerant could
enter the next pass with more uniform quality, which resulted in both the condensation
heat transfer enhancement and pressure drop reduction (Ye et al., 2009). Wang et al.
(2011) investigated the single-phase flow in the parallel flow heat exchangers with
rectangular inlet and outlet headers. They found that the first several tubes took half
the flowrate of last tube and the mal-distribution could be improved via smaller tubes
or larger inlet distance. With the experimental refrigerant quality in vertical header as
T
the input of a heat exchanger model, Zou et al. (2014) found that the performance
IP
reduction due to fluid mal-distribution could reach 30% and 5% for the refrigerants of
CR
R410a and R134a, respectively.
Nowadays, the commercial CFD software of ANSYS Fluent is widely applied for
US
the optimal design of heat transfer devices (Bhutta et el., 2012; Guo et al., 2015; Luo
et al., 2016; Yang et al., 2014; You et el., 2015). With flexible DEFINE Macros
AN
(ANSYS, 2013), the ANSYS Fluent could be qualified to simulate various complex
thermo-fluid processes. Recently, Shojaeefard et al.(2016) developed a hybrid model
M
with Fluent for the parallel flow condenser, and how the refrigerant mal-distribution
among flat tubes was influenced by the location and diameter of inlet tube was studied
ED
(Shojaeefard et al., 2017). It is noted that in their hybrid model, the traditional
effectiveness-NTU method was adopted to evaluate the heat transfer rate of flat tubes,
PT
and the effect of uneven refrigerant distribution within the same flat tube was
neglected. Besides, the boundary condition of porous jump used in Shojaeefard et al.
CE
(2016) and Shojaeefard et al. (2017) is incompatible with the user-defined functions
of Fluent, which could be against updating the pressure drop of refrigerant through
AC
T
that investigated by Hu et al. (2012). It consists of 40 flat tubes and three pieces of
IP
baffles are properly arranged in vertical headers so that four refrigerant passes,
CR
respectively with tube numbers of 18, 10, 7 and 5, are created. To enhance the heat
transfer rate on the air side, louvered Aluminum fins are soldered between
US
neighboring flat tubes. In the present hybrid model, the flow and heat transfer in each
minichannel of flat tubes are treated by the method of one-dimensional finite element
AN
analysis, so that the fluid mal-distribution among different minichannels along with
the temperature variation of air flowing across flat tubes could be identified. The
M
ANSYS Fluent is employed to conduct the three-dimensional CFD simulation for the
refrigerant flow in vertical headers. To obtain the continuous refrigerant flow, 40 fake
ED
flat tubes, which are much shorter than real ones and the pressure drops of refrigerant
flowing through real tubes are imposed on, are introduced to connect the left and right
PT
vertical headers, as depicted in Fig. 1. It is noted that only 13 of the total 40 rows of
flat tube channels along with one of the total three baffles are presented for good
CE
distinction. The CFD model and one-dimensional analysis are coupled during the
hybrid simulation of the refrigerant flow in condenser, as presented in Fig. 2.
AC
7/28
ACCEPTED MANUSCRIPT
liquid and vapor of vertical headers, the assumption of homogeneous flow is adopted
in the current model to decrease the computation load (Shojaeefard et al., 2016).
Besides, the refrigerant flow through condenser is assumed to be steady and turbulent.
As the heat transfer occurring at each minichannel of flat tubes will be treated with
the method of one-dimensional analysis, no energy conservation is included in the
current CFD model, and the governing equations, consisting of the continuity and
momentum conservation equations, are expressed as below (Shojaeefard et al., 2016;
T
You et el., 2015).
IP
Continuity equation:
CR
ρui
=0 (1)
xi
Momentum equation:
US (2)
AN
where the p is fluid pressure, while u and f, with the tensor subscripts expressed by i
M
and j, stand for fluid velocity and volume force, respectively. The ρ and μeff represent
refrigerant density and effective dynamic viscosity, and the latter is calculated by μeff
ED
=μl+μt. It is noted that with the vapor condensation in flat tubes, the average physical
properties of mixture could vary notably, and the quality-weighted average method
PT
are adopted. In more details, for the wet vapor with the quality of χ, its density
ρ=(1-χ) ·ρl+χ·ρv and dynamic viscosity μ=(1-χ) ·μl+χ·μv.
CE
the standard wall function. The conservation equations of turbulent kinetic energy and
its dissipation rate are given below:
For turbulence kinetic energy k:
(ρku j ) μ k
= [(μ l + t ) ]+G k -ρε (3)
x j x j σ k x j
8/28
ACCEPTED MANUSCRIPT
ρεu j μ ε C1ε ε ε2
= μ l + t + G k -C2ε ρ (4)
x j x j σε x j k k
k2 1 u u
where μ t =ρCμ ; Gk=2μtEij.Eij; Eij = [( i )+( i )] .
ε 2 x j x j
The constants for the current turbulent model are set as below:
Cμ=0.09; C1ε=1.44; C2ε=1.92; ζk=1.0; ζε=1.3.
2.3 One-dimensional finite element analysis of flat tubes
T
The flow and heat transfer of refrigerant through each minichannel of flat tubes are
IP
treated by the one-dimensional finite element approach. Three distinct zones of
CR
superheated and wet vapors and subcooled liquid, which are identified by comparing
the refrigerant temperature and the saturated one of local pressure, are used in the
modeling.
2.3.1 Heat transfer analysis
US
AN
The energy conservations of refrigerant flowing through minichannels during the
single-phase and condensation heat transfer processes are respectively expressed as
M
dt ref
cp,ref qm,ref kA(t air t ref ) (5)
dx
dref kA(t air t ref )
qm,ref (6)
ED
dx r
Here qm stands for the mass flowrate of a minichannel, and x is the flow direction in
minichannel (see Fig. 1); r and χ refer to the vaporization potential heat and quality of
PT
refrigerant, respectively; k and A stand for the overall heat transfer coefficient and
CE
heat transfer area, respectively. As the flat tubes take a large thermal conductivity and
their wall thickness is small, the product of k and A could be modeled by
AC
1 1 1
(7)
kA hair (Aair Afin fin Afin ) href Aref
Here η and h refer to the fin efficiency and convection heat transfer coefficient. As the
heat transfer on the air side is enhanced by multi-louvered fins, the hair is modeled
with the following empirical correlation of Colburn factor (j)(Kim and Bullard, 2002).
9/28
ACCEPTED MANUSCRIPT
(9)
Here Ll, Lα and Lp refer to the louver height, angle and pitch respectively; Tw and Tp
refer to the width and pitch of flat tubes; the δfin, FH and Fp stand for the fin thickness,
height and pitch, respectively.
Under the process without phase change, the convection heat transfer coefficient
on the refrigerant side is calculated by the empirical Dittus-Boelter correlation, i.e.,
T
λ
IP
h=0.023ReDh 0.8 Pr 0.3 (10)
Dh
CR
where the Dh refers to the hydraulic diameter of minichannel.
As for the condensation heat transfer of refrigerant in flat tube mini-channels, the
h=0.023Relo0.8Prl 0.4
λl
US
convection heat transfer coefficient is calculated by Shojaeefard et al.(2016)
[(1-χ)0.8 +
3.8χ 0.76 (1-χ)0.04
] (11)
AN
Dh (p/pcrit )0.38
where the Relo takes the value where all the mixture is assumed to be liquid, and pcrit
M
modeled with the energy conservation equation of single-phase flow similar to Eq. 5.
2.3.2 Flow resistance analysis
PT
Under the process of single-phase flow, the pressure gradient of the refrigerant
flowing through a minichannel is calculated by
CE
dp 2fu 2
(12)
dx Dh
AC
dp C 1 dp
(1 2 ) ( )l (13)
dx X X dz
dp 2fl 2u 2 (1 ) 2 1 fl v 0.5
Here ( )l and X ( ) , where fl (or fv) is the friction
dx l Dh f v l
10/28
ACCEPTED MANUSCRIPT
C=0.39Re0.03 0.10
lo Su vo ( ρl ρ v )
0.35
for tt flow regime
C=8.7 10 Relo Su vo ( ρl ρ v )
-4 0.17 0.50 0.14
for tl flow regime
(14)
C=1.5 10 Relo Su vo ( ρl ρ v )
-3 0.59 0.19 0.36 for lt flow regime
for ll flow regime
C=3.5 10 Relo Su vo ( ρl ρ v )
-5 0.44 0.50 0.48
T
Here the Suratman number Su vo vDh 2v ; the above flow regime of tt, tl, lt or ll
IP
depends on both the liquid and vapor flow regimes, for an example, the "tl" stands for
CR
the turbulent liquid flow accompanied by laminar vapor flow.
3. Numerical Solution and validation
US
3.1 Computation domain, meshes and boundary conditions
The CFD computation domain consists of the two vertical headers on the left and
AN
right sides of condenser, together with the minichannels of fake short flat tubes, as
depicted in Fig. 1. To minimize the computation load, the whole condenser is divided
M
into many blocks and the hexahedral elements are adopted in the meshes generation.
Besides, the denser meshes are generated at the regions near the entrances and exits of
ED
flat tubes. Three internal planes (i.e., three baffles) in the left and right headers are set
as the wall boundary, and the refrigerant flow takes four passes in the condenser. The
PT
inlet and outlet of condenser take the boundary conditions of velocity inlet and
pressure outlet, respectively.
CE
The uniform meshes are adopted for each minichannels when one-dimensional
analysis is conducted. The fluid pressure and velocity obtained by CFD computation
AC
are taken as the inputs of one-dimensional model, while its predicted pressure drops
together with fluid temperature and quality are exported to the CFD model.
3.2 Numerical solution scheme
The current CFD simulation of parallel flow condenser is integrated with the
one-dimensional finite element analysis of the flow and heat transfer in each
minichannel of flat tubes. The iteration schematic is depicted in Fig. 2. The
11/28
ACCEPTED MANUSCRIPT
T
general-purpose DEFINE-ADJUST Macro (ANSYS, 2013), called at the beginning of
IP
each iteration to adjust or modify FLUENT variables that are not passed as arguments,
CR
is adopted to conduct the one-dimensional analysis of the minichannels of flat tubes,
and the calculated momentum sources are imposed on the minichannels of fake short
US
flat tubes (see Fig. 1(a)) through the Macro of DEFINE-SOURCE. In addition, the
DEFINE-PROPERTY Macro is complied to express the refrigerant density and
AN
viscosity varying notably due to the vapor condensation. The considerable momentum
source along with the greatly varied properties could weaken the robust of CFD
M
hybrid model. Based on the compromise between computation load and prediction
precision, the final CFD model takes about 3.5 million cells and 400 longitudinal
PT
To validate current flexible hybrid model, the automobile parallel flow condenser,
experimented by Hu et al. (2012) under six different refrigerant and air flowrates, is
AC
numerically simulated, and the predicted performances of heat transfer rate and flow
resistance are compared with the experimental counterparts. It is noted that the above
condenser was manufactured by 40 pieces of 605mm-long flat tubes with a constant
pitch of 8.9mm and the refrigerant of R134a was adopted as the working fluid. These
tubes, each taking 19 minichannels of 0.8mm×0.61mm (Hu et al., 2012), have the
width of 18mm and louvered Aluminum fins are soldered between them.
The comparison of refrigerant pressure drops between numerical and experimental
12/28
ACCEPTED MANUSCRIPT
data is depicted in Fig. 3(a). It is clearly seen that the current hybrid model somewhat
underestimates the refrigerant pressure drops, with the averaged relative deviation
between the predicted and experimented values equal to about -9.8%. This prediction
precision, greatly related to the precision of the adopted correlation of two-phase flow
resistance, is generally acceptable and better than that in Hu et al.(2012).
The refrigerant outlet temperatures and heat transfer rates predicted by current
model, along with the numerical and experimental counterparts of Hu et al.(2012), are
T
presented in Fig. 3(b) and (c), respectively. It is clear that the predictions of heat
IP
transfer performances meet experiments well. In more details, for the six experimental
CR
conditions, the refrigerant outlet temperature of current model deviates by about 1.2℃
on average from experimental counterparts(see Fig. 3(b)), and the relative deviations
US
of heat transfer rates against experimental data are within -1.0%~2.9% (see Fig. 3(c)),
both better than the numerical predictions of Hu et al. (2012). The improvement could
AN
be owed to the adoption of advanced hybrid model along with accurate empirical
correlations of convection heat transfer.
M
With the above comparisons, it could be said that the current model has a
reasonable precision, and will be used to study the refrigerant mal-distribution within
ED
under the condition that the air inlet temperature equals 35℃ and the condensing
temperature is 10℃ greater than the experimental counterpart of Hu et al. (2012).
The computation results are added to the Fig. 3 with the columns of sparse horizontal
lines. It is clear from Fig. 3 that the heat transfer rates and pressure drops at the air of
35℃ are slightly lower than the counterparts of the air of 25℃. The reason for this
phenomenon could be that the potential heat and kinetic viscosity of the working fluid
decreases with the rising condensing temperature.
13/28
ACCEPTED MANUSCRIPT
T
among flat tubes of parallel flow condenser is found to be uniform except the 1st pass
IP
of refrigerant, Fig. 4(a) depicts only the refrigerant flowrates in the top 18 flat tubes.
CR
From Fig. 4(a), one could see that the top 8 flat tubes have the refrigerant flowrate
decreasing fom 9.41 to 8.60kg·h-1 and the flowrate rises from 8.59 to 8.63kg·h-1 for
US
the other tubes. In more details, the maximal relative deviation of the refrigerant
flowrates of flat tubes against the average counterpart takes the value of 7.57%, and
AN
the standard deviation is equal to about 2.56%. It is noted that these figures are
comparable to those obtained by Shojaeefard et al. (2016).
M
the 3rd, 9th and 15th flat tubes of 1st pass. It is clear from Fig. 4(b) that the
minichannels at the upstream of air take the larger refrigerant flowrates, and the
PT
3flowrates against their average counterparts are respectively equal to 5.90%, 7.40%
and 8.06%, with the standard deviations of 2.96%, 3.46% and 3.52%, respectively. As
AC
a contrast, the refrigerant flowrate distributions among minichannels are the same for
different tubes of 2nd, 3rd or 4th refrigerant pass approximately, and the minichannels
at the upstream of air have greater refrigerant flowrates than those downstream the air.
Besides, a tube of downstream refrigerant pass has a greater mal-distribution among
different minichannels, as depicted in Fig. 4(c). In more details, for the central tubes
of the 2nd, 3rd and 4th passes, the maximal relative deviations of minichannel
flowrates against their average counterparts take the values of 11.81%, 12.68% and
14/28
ACCEPTED MANUSCRIPT
15.63%, respectively, and their standard deviations are equal to 5.66%, 5.79% and
7.76% respectively.
4.2 Discussions based on contours and vectors
The above refrigerant mal-distributions among different flat tubes could be related
to the flow and pressure fields, thus Fig. 5(a) and (b) presents the contours of pressure
and velocity at the longitudinal symmetry plane of parallel flow condenser (i.e., y=0
in Fig. 1(b)) to conduct the discussions. It is noted in Fig. 5(b) that the notable density
T
variation due to the condensation results in the great difference of velocity magnitude
IP
on the two sides of fake short tubes. From Fig. 5, it is seen that different flat tubes of
CR
2nd, 3rd or 4th pass take similar pressure and velocity fields, which indicates that they
could have approximately the same refrigerant flowrate. As a contrast, the pressure
US
and velocity contours of 1st pass are not that uniform and the refrigerant distribution
in the top 18 tubes is uneven. In more details, as the refrigerant flows with a high
AN
speed into the left vertical header from the top of condenser, the jet results in that the
top 9 flat tubes have larger inlet velocities and pressures, which facilitates the 9 tubes
M
having larger flowrates. Besides, as the fluid velocity and pressure drop with the
refrigerant flowing downward in the left header, the flowrate of the tube near the inlet
ED
exceeds that of the tube far from the inlet, consistent with the columns in Fig. 4(a). On
the other hand, for the rest 9 tubes of 1st pass, the impact of the inlet jet is weaker.
PT
The static pressures at the inlets of those tubes depend on the compromise between
the pressure decrement caused by the viscous flow and pressure increment due to the
CE
dynamic pressure being recovered. As for the current study, the inlet pressure of the
above 9 flat tubes vary slightly, while the outlet pressure keeps nearly the same (see
AC
Fig. 5(c)), resulting in a flowrate distribution consistent with the columns in Fig. 4(a).
The mal-distribution among different minichannels of same flat tube could be
greatly related to the uneven heat transfer rate within flat tubes. As is seen from Fig.
5(a), the flow resistance of parallel flow condenser is notable and thus the temperature
of downstream saturated refrigerant is notably lower than that at the upstream, which
indicates that less refrigerant could be condensed at the 4th pass than that at the 2nd or
3rd pass due to a smaller heat transfer rate. On the other hand, as the ambient air
15/28
ACCEPTED MANUSCRIPT
absorbs thermal energy of refrigerant when it flows through the condenser, the
minichannel at the downstream of air has a smaller heat transfer rate because the air
has a higher temperature there. It results in less refrigerant vapor condensed into
liquid and thus facilitates a larger flow resistance. With above two factors, it could be
expected that the minichannels at the upstream of air have larger refrigerant flowrates
and the minichannel flowrates of the tubes at the downstream of refrigerant have
larger relative deviations against the averages, consistent with Fig. 4(c). As for the
T
central minichannels of the tubes of 1st pass having the least flowrate (see Fig. 4(b)),
IP
it could be reasoned by the non-uniform pressure and velocity fields, as depicted in
CR
Fig. 6(a) and (b). Similar to that in Fig. 5(b), the refrigerant velocity in Fig. 6(b) has a
considerable drop at the downstream of flat tube because of the condensation. As the
US
flat tubes are arranged with a constant pitch accompanied by two notable triangular
gaps on their two sides in the vertical headers, the non-uniform refrigerant distribution
AN
between the minichannels on the downstream and upstream of air could result in a
notable refrigerant recirculation, which is clearly observed in Fig. 6(b). Because of the
M
recirculation, the fluid pressure and velocity at the central minichannels could be
relatively small. As a result, despite the minichannel on the downstream of air has a
ED
larger flow resistance, its refrigerant flowrate could be larger than those of central
minichannels.
PT
5. Conclusions
In the current work, a flexible hybrid CFD model is developed for minichannel
CE
Besides, the refrigerant distributions among different flat tubes and different
minichannels of same tube are obtained. In addition, the discussion on the physical
mechanism of refrigerant mal-distribution is conducted based on contours and vectors.
With the present work, some conclusions are obtained as below:
1) With the DEFINED Macros of ANSYS Fluent to embed the one-dimensional
finite element analysis of the heat transfer and flow resistance in the minichannels of
flat tubes, the current hybrid model could simulate the refrigerant flow in parallel flow
16/28
ACCEPTED MANUSCRIPT
T
transfer rate, which facilitates a smaller flow resistance and a higher refrigerant
IP
flowrate. Meanwhile, the notable refrigerant recirculation in the vertical header could
CR
result in central mini-channels having smaller flowrates than that downstream the air.
References
US
AN
ANSYS, ANSYS FLUENT Theory Guide, ANSYS Inc., 2013.
Bhutta M.M.A., HayatN., BashirM.H., et el., 2012, CFD applications in various heat
exchangers design: A review, Appl. Therm. Eng., 32, 1-12.
M
Brix W., Kærn M.R., Elmegaard B., 2009, Modelling refrigerant distribution in
microchannel evaporators, Int. J. Refrigeration 32, 1736-1743.
ED
Byun H., Kim N., 2011, Refrigerant distribution in a parallel flow heat exchanger
having vertical headers and heated horizontal tubes, Exp. Therm. Fluid Sci. 35,
920-932.
PT
Cuevas C., Lebrun J., Lemort V., et el., 2009, Development and validation of a
condenser three zones model, Appl. Therm. Eng. 29, 3542-3551.
CE
Guo J., Yan Y.X., Liu W., et al., 2015, Enhancement of laminar convective heat
transfer relying on excitation of transverse secondary swirl flow, Int. J. Therm.
AC
17/28
ACCEPTED MANUSCRIPT
T
conditioning, Energ. Buildings, 144: 104-116.
IP
Jabardo J.M.S., Mamani W.G., 2003,Modeling and experimental evaluation of parallel
flow micro channel condensers, J. Braz. Soc. Mech. Sci. 25(2), 1663–1670.
CR
Kandlikar S.G., Garimella S., Li D., et al. Heat Transfer and Fluid Flow in
Minichannels and Microchannels, 2006.
US
Kim M.H., Bullard C.W., 2002, Air-side thermal hydraulic performance of multi-
louvered fin aluminum heat exchangers, Int. J. Refrigeration 25, 390-400.
Kim N.H., Han S.P., 2008, Distribution of air-water annular flow in a header of a
AN
parallel flow heat exchanger, Int. J. Heat Mass Transfer 51, 977-992.
Kim S.M., Mudawar I., 2012, Universal approach to predicting two-phase frictional
M
Lee G.H., Yoo J.Y., 2000, Performance analysis and simulation of automobile air
conditioning system, Int. J. Refrigeration 23(3), 243-254.
Luo X.B., Hu R., Liu S., et al., 2016, Heat and fluid flow in high-power LED
PT
18/28
ACCEPTED MANUSCRIPT
T
with the effect of refrigerant mal-distribution, Int. J. Refrigeration, 60(6): 234-
IP
246.
Yang J., Ma L., Bock J., et al., 2014, A comparison of four numerical modeling
CR
approaches for enhanced shell-and-tube heat exchangers with experimental
validation, Appl. Therm. Eng. 65, 369-383.
US
Ye L., Tong M.W., Zeng X., 2009, Design and analysis of multiple parallel-pass
condensers, Int. J. Refrigeration 32, 1153-1161.
You Y.H., Chen Y.Q., Xie M.Q., et el., 2015, Numerical simulation and performance
AN
improvement for a small size shell-and-tube heat exchanger with trefoil-hole
baffles, Appl. Therm. Eng. 89, 220-228,.
M
Zou Y., Tuo H., Hrnjak P.S., 2014, Modeling refrigerant maldistribution in
microchannel heat exchangers with vertical headers based on experimentally
ED
19/28
ACCEPTED MANUSCRIPT
Figure Captions:
[1] Computation domain of current parallel flow condenser and meshes generation.
Only 14 of the total 40 rows of flat tube minichannels along with one of the total three
baffles are presented for a distinct description,. (a) Front view; (b) Top view.
[2] Computation flowchart of current hybrid model of parallel flow condenser
integrating CFD simulation with one-dimensional finite element analysis.
T
[3] Validation of current model by comparing the predicted thermal and hydraulic
IP
behaviors against the experimental and numerical counterparts obtained by Hu et al.
(2012) at the air temperature of 25℃. (a) Comparison of refrigerant pressure drop; (b)
CR
Comparison of refrigerant outlet temperature; (c) Comparison of heat transfer rate.
[4] Variations of refrigerant flowrates among different flat tubes and different
US
minichannels of same flat tube. The flat tube of No. 1 is the nearest to the refrigerant
inlet of condenser and the minichannel of No. 1 is at the most upstream of air. (a)
AN
Variations among flat tubes of 1st pass; (b) Variations among different tube
minichannels of 1st pass; (c) Variations among minichannels of central tubes of 2nd,
M
condenser (i.e., the plane y=0 in Fig. 1(b)). (a) Pressure contour; (b) Velocity contour;
(c) Detailed pressure contour of 1st tube pass.
PT
[6] Pressure contour and velocity vector at the transversal section of No. 3 flat tube
(i.e., the I-I plane in Fig. 1(a)). (a) Pressure contour; (b) Velocity vector colored by
CE
z-component velocity.
AC
20/28
ACCEPTED MANUSCRIPT
T
IP
CR
US
AN
M
ED
PT
CE
AC
Fig. 1 Computation domain of current parallel flow condenser and meshes generation.
Only 14 of the total 40 rows of flat tube minichannels along with one of the total three
baffles are presented for a distinct description,. (a) Front view; (b) Top view.
21/28
ACCEPTED MANUSCRIPT
T
IP
CR
US
AN
M
ED
PT
CE
22/28
ACCEPTED MANUSCRIPT
T
IP
CR
US
(a) Comparison of refrigerant pressure drop
AN
M
ED
PT
CE
AC
23/28
ACCEPTED MANUSCRIPT
T
IP
CR
US
AN
Fig. 3 Validation of current model by comparing the predicted thermal and hydraulic
behaviors against the experimental and numerical counterparts obtained by Hu et al.
ED
(2012) at the air temperature of 25℃. (a) Comparison of refrigerant pressure drop; (b)
Comparison of refrigerant outlet temperature; (c) Comparison of heat transfer rate.
PT
CE
AC
24/28
ACCEPTED MANUSCRIPT
T
IP
CR
US
AN
(a) Variations among flat tubes of 1st pass
M
ED
PT
CE
AC
25/28
ACCEPTED MANUSCRIPT
T
IP
CR
US
AN
(c) Variations among minichannels of central tubes of 2nd, 3rd and 4th passes
Fig. 4 Variations of refrigerant flowrates among different flat tubes and different
M
minichannels of same flat tube. The flat tube of No. 1 is the nearest to the refrigerant
inlet of condenser and the minichannel of No. 1 is at the most upstream of air. (a)
ED
Variations among flat tubes of 1st pass; (b) Variations among different tube
minichannels of 1st pass; (c) Variations among minichannels of central tubes of 2nd,
PT
26/28
ACCEPTED MANUSCRIPT
T
IP
CR
US
AN
Fig. 5 Pressure and velocity contours at the longitudinal symmetry plane of parallel
flow condenser (i.e., the plane y=0 in Fig. 1(b)). (a) Pressure contour; (b) Velocity
M
27/28
ACCEPTED MANUSCRIPT
T
IP
(a) Pressure contour
CR
US
AN
M
(i.e., the I-I plane in Fig. 1(a)). (a) Pressure contour; (b) Velocity vector colored by
z-component velocity.
PT
CE
AC
28/28