You are on page 1of 9

International Journal of Heat and Fluid Flow 32 (2011) 1199–1207

Contents lists available at SciVerse ScienceDirect

International Journal of Heat and Fluid Flow


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

Flow pattern based correlations of two-phase pressure drop in rectangular


microchannels
Chiwoong Choi ⇑, Moohwan Kim
Department of Mechanical Engineering, Pohang University of Science and Technology, Pohang, San 31, Hyoja Dong 790-784, Republic of Korea

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

Article history: Numerous pressure drop correlations for microchannels have been proposed; most of them can be clas-
Received 10 January 2011 sified as either a homogeneous flow model (HFM) or a separated flow model (SFM). However, the predic-
Received in revised form 1 August 2011 tions of these correlations have not been compared directly because they were developed in experiments
Accepted 30 August 2011
conducted under a range of conditions, including channel shape, the number of channels, channel mate-
Available online 6 October 2011
rial and the working fluid. In this study, single rectangular microchannels with different aspect ratios and
hydraulic diameters were fabricated in a photosensitive glass. Adiabatic water-liquid and Nitrogen-gas
Keywords:
two-phase flow experiments were conducted using liquid superficial velocities of 0.06–1.0 m/s, gas
Rectangular microchannel
Pressure drop
superficial velocities of 0.06–72 m/s and hydraulic diameters of 141, 143, 304, 322 and 490 lm. A pres-
Flow pattern sure drop in microchannels was directly measured through embedded ports. The flow pattern was visu-
Correlation alized using a high-speed camera and a long-distance microscope. A two-phase pressure drop in the
microchannel was highly related to the flow pattern. Data were used to assess seven different HFM vis-
cosity models and ten SFM correlations, and new correlations based on flow patterns were proposed for
both HFMs and SFMs.
Ó 2011 Elsevier Inc. All rights reserved.

1. Introduction patterns were visualized using a high-speed camera and a long-


distance microscope. Data were used to assess seven different
Two-phase flow pressure drop correlations can be classified as HFM viscosity models and ten SFM correlations, In addition, new
either homogeneous flow models (HFMs) or separated flow models correlations based on flow patterns were proposed for both HFMs
(SFMs). HFMs treat a two-phase flow as a single phase flow with and SFMs. There are modeling works for two-phase pressure drop
mixed properties. In HFMs, two-phase density and viscosity should in microchannel (Garimella et al., 2003; Shiferaw et al., 2011). For
be defined. SFMs are based on a two-phase frictional multiplier example, Garimella et al. (2003) developed pressure drop model,
(u), which is defined as the ratio of the pressure gradient of the which based on a unit cell concept, for intermittent flow regime
liquid phase to that of the gas phase. for non-circular microchannels. In development of model, the
A pressure drop of two-phase flow in a microchannel cannot be number of unit cell was regressed with a hydraulic diameter and
predicted using conventional correlations. Hence, many studies of slug Reynolds number. However, their model shows highly under-
the two-phase frictional pressure drop in microchannels have been estimated the number of unit cell comparing with present data
conducted; most used circular tubes and rectangular multi- (Choi et al., 2010b). Shiferaw et al. (2011) also developed semi-
channels, because of difficulties in fabricating circular micro- mechanistic pressure drop model for specific flow pattern, which
channels and controlling small amounts of fluid and heat. For a is based on a three zone model (Thome et al., 2004). Both models
heat sink application, the quantification of a pressure drops in are considered the specific flow pattern, the periodic elongated
rectangular microchannels is needed. bubble regime, thus two models are exempted from our assess-
In the present study, rectangular glass microchannels were fab- ment correlations.
ricated to allow a visualization of two-phase flow patterns. Exper-
iments were conducted in these microchannels using water liquid
and Nitrogen gas flow for liquid superficial velocities of 0.06– 2. Experimental setup
1.0 m/s, gas superficial velocities of 0.06–72 m/s, and microchannel
hydraulic diameters of 141, 143, 304, 322 and 490 lm. Flow 2.1. Test section

The test section (Fig. 1; the unit is mm) was fabricated in a sheet
⇑ Corresponding author. of a photosensitive glass using UV rays. Details can be found in
E-mail address: chiwoongchoi@gmail.com (C.W. Choi). Choi and Kim (2008). The test section has two inlets (i.e., one for

0142-727X/$ - see front matter Ó 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.ijheatfluidflow.2011.08.002
1200 C.W. Choi, M.H. Kim / International Journal of Heat and Fluid Flow 32 (2011) 1199–1207

Fig. 1. A microchannel test section (top view).

each phase flow), one outlet and three pressure ports. Total chan- rates of the liquid were measured using flow-meters. The pressure
nel length was 60 mm and the distance between pressure ports drop was measured directly through embedded pressure ports in
was 15 mm. The test section was held horizontally with a support the microchannel using four different pressure transducers (Druck
frame to connect inlets and outlet tubing. The interconnections LPM9000, Setra 230-5, 10, and 25 PSID). Data were collected using
were achieved using fitting and ferrule (Upchurch Scientific), and a data acquisition system (Agilent 34970) and were saved on a per-
a fitting holder. sonal computer. The details of the experimental setup were de-
Pressure was measured directly through the three embedded scribed in Choi et al. (2010a).
pressure ports; this overcame the uncertainties in a measurement Before the main experiment, water was pumped into the inlet of
that occurred in most previous studies, in which pressure was the test section to extract non-degassed water from the tubes, and
measured outside of the channel. Moreover, the fully developed tubes connected to pressure ports were fully filled with water to
condition was checked with comparing pressure differences offset the effect of a static head, which was significant under low
among three pressure ports (Fig. 1). Flow patterns were visualized pressure conditions. Then the gas flow was applied to the other in-
directly through the transparent glass using a high-speed camera let to create a two-phase flow. When the temperature and the
and a long-distance microscope. pressure upstream of each phase flow were steady, the pressure
Cross sections of five different fabricated rectangular micro- and the flow rate were measured every 2 s. Then, the gas flow rate
channels were measured using a 3D-profiler (Veeco-Wyko DMEMS was changed while the liquid flow rate was held a constant
NT1100). To verify this measuring method, the cross-section of a (Table 1). All experimental uncertainties are summarized in Table 2.
test sample was measured using a microscope (Table 1).
3. Results and discussion
2.2. Experimental facilities and procedure
3.1. Single phase pressure drop
In the experimental setup (Fig. 2), the liquid and gas flows were
controlled using a pneumatic pumping system, which consisted of Before conducting the two-phase flow experiment, experiments
electric regulators and pressurized gases. Nitrogen gas was used to for a single phase pressure drop in the rectangular microchannels
pressurize the gas phase loop and Helium gas was used to pressur- were conducted to verify the experimental apparatus and to eval-
ize the liquid phase loop. The two-phase flow was mixed in a uate the friction factor of the rectangular microchannel. The value
T-junction. The effect of the mixing method is not well understood of fRe is 16 for circular tubes, and Hartnett and Kostic (1989) pro-
and still hot issues in this research area (Kawaji, 2008). Therefore, posed a correlation of fRe for ducts of different ARs (Eqs. (1)–(3))
in this study, we used T-junction configuration, which is well pro-
ven mixing method in various fields. The volume and mass flow f Re ¼ 24ð1  1:3553AR þ 1:9467AR2  1:7012AR3
þ 0:9564AR4  0:2537AR5 Þ; ð1Þ
Table 1
where
Experimental conditions.

Variables Ranges quch Dh


Re ¼ ; ð2Þ
Diameters of microchannels, Dh (WCh  HCh) (lm) 490(510  470)
l
322(501  237)
143 (503  83) and
304 (332  280)  
141 (201  109) Dh DP
f ¼ : ð3Þ
Liquid mass flux, GL (kg/m2s) 66–1000 2qu2ch Dz SP
Vapor mass flux, Gg (kg/m2s) 0.075–80
Liquid Reynolds number, ReL 32–477 The predicted values from Hartnett and Kostic (1989)’s correla-
Vapor Reynolds number, Reg 2–2134 tion were 19.71, 15.74 and 14.25 for 143, 322 and 490 lm diame-
Liquid superficial velocity, jL (m/s) 0.06–1
ters, respectively; measured fRe values were agreed well with
Vapor superficial velocity, Jg (m/s) 0.06–72
predicted values. Results of the single phase frictional pressure
C.W. Choi, M.H. Kim / International Journal of Heat and Fluid Flow 32 (2011) 1199–1207 1201

Fig. 2. Experimental facilities for the adiabatic two-phase flow in microchannnel.

Table 2
Experimental uncertainties.

Variables Uncertainty
Diameter (lm) ±17 (4.5%)
Area (lm2) ±6311 (3.8%)
Pressure (kPa) ±6% (±0.01, ±0.005, ±0.034, ±0.085)
Temperature (°C) ±0.1–0.5
Mass flux, G (kg/m2s) ±0.1–2%
Friction factor ±9%
Superficial velocities (m/s) ±10%

drop in our test section for water liquid indicate that the conven-
tional theory of pressure drops in a single phase-laminar flow is
acceptable and that our experimental setup is appropriate.

3.2. Flow pattern and two-phase pressure drop

In our experimental region, the observed flow patterns were


bubbly, slug bubble, elongated bubble, liquid ring, and transition
flow (Fig. 3). Definitions are the same as in Choi et al. (2010a). As
gas superficial velocity (jG) increased, the bubble length increased.
The bubble nose and the tail collapsed, and formed a liquid ring;
this flow pattern has been observed and given various names by
various researchers (Kawahara et al., 2002; Chung and Kawaji,
2004 and Saisorn and Wongwises, 2008). A transition region oc-
curred between the elongated bubble flow and the liquid ring flow Fig. 3. A classification of the flow patterns in the rectangular microchannel for
Dh = 490 lm.
regimes; in the transition regime, the elongated bubble flow and
the liquid ring flow occurred periodically. Thus, we defined this
flow regime as ‘‘transition flow patterns’’. Before this transition
point, the bubble flows were stable and formed a perfect bubble Pressure drop increased as jL and jG increased. At the constant jL,
train. As liquid superficial velocity (jL) increased, the bubble length the pressure drop increased as jG increased. However, in the tran-
decreased. Then, the elongated bubble flow changed to the single sition region, the pressure drop decreased as jG increased. The
bubbly flow at higher jL. In this study, main flow regimes were merged bubbles caused the bubble to elongate. The dominant com-
grouped into (1) a bubble flow including bubbly, slug bubble and ponent of pressure drops in a bubble occurs in the nose and tail,
elongated bubble, (2) liquid ring flow, and (3) a transition between rather than in the bubble body. However, Choi et al. (2010b)
elongated bubble flow and ring flow. reported the collapsed bubble reduces the number of parts that
1202 C.W. Choi, M.H. Kim / International Journal of Heat and Fluid Flow 32 (2011) 1199–1207

GDh
ReTP ¼ ; ð8Þ
lTP
where lTP is a two-phase viscosity. For a laminar flow in a circular
tube, n = 1 and N = 16; in a rectangular channel N is a function of AR
(Eq. (1)).
Numerous two-phase viscosity models have been proposed.
Recently, Costa-Patry et al. (2011) conducted flow boiling experi-
ments in 85 lm wide multi-microchannels. They showed that
Cioncolini et al. (2009)’s correlation is well agree with their pres-
sure drop results. The Cioncolini et al.’s correlation is developed
using a dimensionless analysis for an annular flow pattern.
Costa-Patry et al. also mentioned the Cioncolini et al.’s correlation
had a good accuracy for the annular flow pattern. Therefore, the
Cioncolini et al.’s correlation is exempted from our assessment cor-
Fig. 4. Two-phase pressure drop in rectangular microchannel with Dh = 322 lm: relations. In this study, our experimental results were used to as-
Typical trend of the pressure drop in a rectangular microchannel ((I) bubbly or sess seven viscosity models: Owens (1961), MacAdams et al.
liquid slug flow, (II) elongated bubble flow, (III) transition flow regime (transition),
(1942), Cicchitti et al. (1960), Dukler et al. (1964), Beattie and
(IV) liquid ring flow).
Whalley (1982), Lin et al. (1991) and Awad and Muzychka (2008).

contribute to the pressure drop. The typical trend of pressure drops Owen ð1961Þ : lTP ¼ lL ; ð9Þ
in the microchannel (Fig. 4) is composed of three regions.
 1
x 1x
Region I: A bubble flow regime including bubbly, slug bubble McAdams etal: ð1942Þ : lTP ¼ þ ; ð10Þ
lG lL
and elongated bubble flow patterns. Pressure drops increased
as gas superficial velocity increased.
Region II: Transition flow patterns. Pressure drop decreased as Cicchitti etal: ð1960Þ : lTP ¼ xlG þ ð1  xÞlL ; ð11Þ
gas superficial velocity decreased.
Region III: Liquid ring flow. Pressure drop increased as gas Dukler etal: ð1964Þ : lTP ¼ blG þ ð1  bÞlL ; ð12Þ
superficial velocity increased.

This typical trend indicates that the pressure drop in the rectan- Beattie and Whalley ð1982Þ : lTP
gular microchannel is highly related to the flow pattern. ¼ blG þ ð1  bÞð1 þ 2:5bÞlL ; ð13Þ

4. Assessment of two-phase pressure drop models lG lL


Lin etal: ð1991Þ : lTP ¼ ; ð14Þ
lG þ x1:4 ðlL  lG Þ
4.1. Homogeneous flow model (HFM)

An HFM is the simplest two-phase flow model; it treats two- 2lG þ lL  2ðlG  lL Þð1  xÞ
Awad and Myuztchka ð2008Þ : lTP ¼ lG :
phase flow as a single phase flow with mixture properties. A 2lG þ lL þ ðlG  lL Þð1  xÞ
two-phase frictional pressure drop consists of frictional, accelera- ð15Þ
tional, and gravitational terms (Eq. (4)) (see Carey, 1992).
        Pressure drop based on the HFM and viscosity models were com-
dP dP dP dP pared to experimental data. Accuracy of model predictions was as-
¼ þ þ : ð4Þ
dz TP dz Friction dz Acceleration dz Gravitation sessed using the mean absolute error (MAE), expressed as a
percentage:
In this study, the gravitational term was neglected, because a
flow was horizontal, and the accelerational term was neglected be-  
1 X jDPTP;pred  DPTP;exp j
cause the flow was adiabatic. Therefore, the total measured pres- MAE ¼  100 : ð16Þ
M DPTP;exp
sure drop is the frictional term in Eq. (4), which can be defined
using the HFM (Eq. (5)). HFMs were assessed for microchannels having different diame-
    ters (Table 3). The most accurate viscosity model is the Beattie and
dP dP 2f G2 Whalley (1982)’s model, which is based on the volumetric quality.
¼ ¼ TP ; ð5Þ
dz TP dz Friction qTP Dh Experimental pressure drop data and the viscosity models, were
used to calculate f and Re (Fig. 5), and an original HFM for
where fTP is a two-phase friction factor, G is mass flux and qTP is Dh = 322 lm with n = 1 and N = 15.75 (Eq. (7)). The two-phase Re
two-phase density (Eq. (6)) was over-predicted by Dukler et al. (1964)’s model and under-pre-
 1 dicted by other models. The two-phase viscosity predicted by the
x 1x Beattie and Whalley (1982)’s model agreed marginally well with
qTP ¼ þ : ð6Þ
qG qL the original HFM. However, the other models except that of Dukler
et al. (1964)’s model overestimated two-phase viscosity (Fig. 6).
The two-phase friction factor can be expressed as an exponen-
Beattie and Whalley (1982)’s two-phase viscosity model was
tial function of the two-phase Reynolds number (Eq. (7))
developed for the bubble flow and annular flow patterns; this con-
fTP ¼ NRen ð7Þ cept agreed well with our experimental result (Fig. 7), but it was
TP ;
deviated from measurements by approximately ±50%. With other
with microchannels, a similar trend was observed.
C.W. Choi, M.H. Kim / International Journal of Heat and Fluid Flow 32 (2011) 1199–1207 1203

Table 3
Comparison of the two-phase pressure drop based on the homogeneous flow model for two-phase viscosity models with the experimental results.

Models MAE (%)


Dh = 490 Dh = 322 Dh = 143 Dh = 304 Dh = 14 l Average
Owens (1961) 352.10 647.37 1247.28 801.18 709.07 761.98
MacAdams et al. (1942) 40.81 101.35 236.46 123.84 169.87 134.46
Cicchitti et al. (1960) 300.76 554.13 1076.53 715.23 650.12 659.35
Dukler et al. (1964) 66.84 51.76 23.25 63.29 42.10 49.44
Beattie and Whalley (1982) 30.69 22.72 75.38 24.41 43.50 39.34
Lin et al. (1991) 111.99 225.10 464.32 309.52 346.56 291.49
Awad and Muzychka (2008) 107.87 204.17 426.47 145.01 298.29 236.36

Fig. 7. Two-phase friction factor and Reynolds number for Dh = 322 lm using
Fig. 5. Relation of friction factor and Reynolds number based on different two- Beattie and Whalley’s two-phase viscosity model.
phase viscosity models for Dh = 322 lm and jL = 0.4 m/s.

    0:5  0:5  0:5  0:5


Dp Dp 1x qG lL
X vv ¼ ¼ : ð19Þ
Dz L Dz G x qL lG
In the right-hand side of Eq. (18), one represents liquid-only
pressure drop, C/Xvv represents mixed pressure drop, and 1=X 2vv
represents gas phase-only pressure drop. Therefore, C represents
the interactional effect of the two-phase flow.
Previously proposed correlations for frictional pressure drop in
a microchannel have been developed by modifying the C-value in
the SFM. In this study, our experimental data were used to assess
ten correlations for the SFM: Lockhart and Martinelli (1949),
Chisholm (1967) and Muller-Steinhagen and Heck (1986) for
macro-scale correlations; Zhang et al. (2007), Tran et al. (2000)
and Lee and Lee (2001) for mini-scale correlations, and Moriyama
and Inoue (1992), Qu and Mudawar (2003), Lee and Mudawar
(2005) and Li and Wu (2010) for micro-scale correlations.
Fig. 6. Two-phase viscosity for water liquid and Nitrogen gas for different two-
Lockhart and Martinelli (1949) proposed the C value for differ-
phase viscosity models. ent flow structures. For example, C = 5 for laminar flow of both
liquid and gas phases.
Chisholm (1967) proposed a modified two-phase multiplier (Eq.
4.2. Separated flow model (SFM) (20)):
h 2s 2s
i
SFMs are based on a two-phase multiplier (u), which is defined /2LO ¼ 1 þ ðY 2  1Þ Bx 2 ð1  xÞ 2 þ x2s ; ð20Þ
as
        where s = 0.25, x is a quality and Y is a sort of Martinelli parameter
Dp Dp Dp Dp
/2L ¼ or /2LO ¼ : ð17Þ (Eq. (21))
Dz TP Dz L Dz TP Dz LO Dp
1
Lockhart and Martinelli (1949) suggested that /2L is a function of Y 2 ¼ DDzpGO ¼ ; ð21Þ
a Martinelli parameter (Eq. (18)) Dz LO X 2vv

C 1 and B is a function of mass flux and Y.


/2L ¼ 1 þ þ ; ð18Þ
X vv X 2vv Muller-Steinhagen and Heck (1986) proposed a simple and con-
venient correlation for a two-phase frictional pressure drop using
where Xvv is a parameter of a laminar liquid–laminar gas flow (Eq. data points of 9300 various fluids and flow conditions, including
(19)). channel diameters from 4 mm to 392 mm (Eqs. (22) and (23))
1204 C.W. Choi, M.H. Kim / International Journal of Heat and Fluid Flow 32 (2011) 1199–1207

   
Dp Dp where k is a fluid property, Caj and Rej are capillary number and
¼ Fð1  xÞ1=3 þ x3 ; ð22Þ
Dz TP Dz GO Reynolds number based on total superficial velocity (j), respectively,
and A, p, q and r are empirically-determined coefficients. For lami-
      
Dp Dp Dp nar–liquid and laminar–gas flow, A = 6.833  108, p = 1.317,
F¼ þ2  x: ð23Þ
Dz LO Dz GO Dz LO q = 0.719 and r = 0.557.
Moriyama and Inoue (1992) studied a two-phase pressure drop
The correlation of Muller-Steinhagen and Heck has a form that in 35–110 lm gaps using R113 under boiling conditions, and
is similar to Chisholm’s correlation without the B-coefficient. Nitrogen and R113 under adiabatic conditions. They proposed a
Mishima and Hibiki group (1993, 1996) investigated the effect modified L–M correlation (Eq. (30))
of a mini-sized diameter on two-phase frictional pressure in exper-
iments on an air–water adiabatic two-phase flow in minichannels K
/2L ¼ 1 þ : ð30Þ
with diameters of 1–4 mm. They proposed that the C-value of the X 2vv
L–M correlation is a function of the diameter. They defined the C-
This correlation lacks the second term in the original L–M correla-
value as
tion (Eq. (19)), i.e., the C-value that is an interactional effect in the
C ¼ 21ð1  e333D Þ; ð24Þ two-phase pressure drop. They multiplied the third term (the gas
phase portion in the two-phase pressure drop) by a factor K, which
for circular tubes and as
they postulated to be a function of liquid Reynolds number. This
C ¼ 21ð1  e319Dh Þ; ð25Þ means that the influential parameter in the two-phase pressure
drop in a microchannel is the gas phase portion rather than the
for rectangular tubes.
interactional portion. Recently, a similar correlation was proposed
Zhang et al. (2007) modified Mishima and Hibiki’s correlation to
by Saisorn and Wongwises (2008).
extend to the micro-scale and verified it using different experi-
Qu and Mudawar (2003) investigated the frictional two-phase
mental results. The hydraulic diameter was replaced by a Laplace
pressure drop in water flow boiling experiments using 21 parallel
constant:
microchannels of size 231 lm  713 lm. They proposed a modi-
C ¼ 21ð1  e358=Nconf Þ; ð26Þ fied Mishima et al. (1993)’s correlation, which considers the effect
of mass flux (Eq. (31))
where Nconf is a confinement number (Eq. (27))
C ¼ 21ð1  e319Dh Þð0:00418G þ 0:0613Þ: ð31Þ
lc lc ½r=ðqL  qV Þg1=2
Nconf ¼ ¼ ¼ ; ð27Þ Lee and Mudawar (2005) conducted flow boiling experiments
D Dh Dh
with R134a using Qu and Mudawar (2003)’s experimental facili-
where lc is the Laplace constant, and the unit of D or Dh is a meter. ties. They proposed a new correlation composed of dimensionless
The applicable range was limited to 0.014 mm <Dh < 6.25 mm and numbers (Eq. (32))
to laminar flow structures of liquid and vapor phases. They recom-
mended that the constant of 0.358 is replaced by 0.674 for the C ¼ c1 RecLO2 WecLO3 ; ð32Þ
adiabatic two-phase flow. where c1, c2 and c3 are experimentally-determined coefficients; for
Tran et al. (2000) performed a series of experiments with laminar–liquid and laminar–gas flow, c1 = 2.16, c2 = 0.047 and
R134a, R12 and R113 flows using circular minitubes with diame- c3 = 0.60.
ters of 2.46 mm and 2.92 mm, and a rectangular mini-channel with Li and Wu (2010) studied dominant parameters using database
dimensions of 4.06 mm  1.7 mm. They proposed a modified Chis- of 769 points, which include 12 different working fluids, diameter
holm correlation with the confinement number for the turbulent range from 0.148 to 3.25 mm, and a channel configuration of circu-
region (Eq. (28)). lar, rectangular cross section and multi-channel. They found that
h i
2s 2s influential parameters are Bond number (Bo) and liquid Reynolds
/2LO ¼ 1 þ ð4:3Y 2  1Þ N conf x 2 ð1  xÞ 2 þ x2s : ð28Þ
number (ReL). Finally, they proposed a new correlation based on
Lee and Lee (2001) conducted air–water adiabatic two-phase the Bo and ReL as follows:
flow experiments for different rectangular minichannels with low Bo 6 1:5 : C ¼ 11:9Bo0:45 ; ð33Þ
aspect ratios. They proposed a new C-value as a function of three
dimensionless numbers for laminar and turbulent regions:
1:5 < Bo 6 11 : C ¼ 109:4ðBoRe0:5
L Þ
0:56
: ð34Þ
 p  q  
l2
jl L qL jDh r
C ¼ Ak q
Carj Resj ¼A L
; ð29Þ The ten correlations based on the SFM were assessed for chan-
qL rDh r lL nels with different diameters (Table 4). The most accurate correla-
tion is that of Qu and Mudawar (2003), which considered the

Table 4
Comparison of the two-phase pressure drop correlations based on the separated flow model with the experimental results.

Reference MAE (%)


Dh = 490 Dh = 322 Dh = l43 Dh = 3O4 Dh = HI Average
Lockhart and Martinelli (1949) 27.50 29.24 100.84 22.07 52.79 46.49
Chisholm (1967) 71.13 60.18 42.91 74.06 54.14 60.48
Muller-Steinhagen and Heck (1986) 56.96 39.39 24.88 48.92 31.25 40.28
Zhang et al. (2007) 45.32 30.52 12.40 41.91 29.75 31.98
Tran et al. (2000) 73.00 92.99 283.17 89.88 108.41 129.49
Lee and Lee (2001) 51.80 41.07 18.88 51.15 39.31 4044
Moriyama and Inoue (1992) 48.55 31.93 30.20 38.57 26.39 35.13
Qu and Mudawar (2003) 25.65 19.75 23.70 23.53 20.36 22.60
Lee and Mudawar (2005) 43.30 23.70 19.42 38.63 24.82 29.97
Li and Wu (2010) 45.00 29.37 11.12 44.94 30.53 32.19
C.W. Choi, M.H. Kim / International Journal of Heat and Fluid Flow 32 (2011) 1199–1207 1205

effects of confinement and mass flux. The second most accurate C-values do not have a constant value for different flow conditions
correlation is that of Lee and Mudawar (2005), which had been val- and well matched flow patterns (Fig. 9). Therefore, the flow pattern
idated previously using that of Qu and Mudawar (2003). Although should be considered when the pressure drop analyzed in a
Qu and Mudawar (2003)’s experiments were conducted in multi- microchannel.
microchannels under boiling conditions, their correlation is well-
predictable with a deviation of approximately ±50%.
Awad and Muzychka (2006) compared correlations of the two- 5. New flow pattern based correlation
phase frictional pressure drop in microchannels, and reported that
a modified C-value with L–M correlation predicts a pressure drop 5.1. New flow pattern based HFM
well. They proposed bounds on the two-phase frictional pressure
drop in the microchannel: the upper bound was the L–M correla- In the basic HFM, n in Eq. (7) indicates flow structure. In this
tion with C = 5 (i.e., the original correlation for laminar–liquid study, three flow regimes were defined: (1) the bubble regime
and laminar–gas); the lower bound was Ali et al. (1993)’s correla- including bubbly, slug and elongated bubble flows, (2) the liquid
tion (i.e., the original correlation with the C = 0). Awad and ring regime, (3) the transition regime between bubble and liquid
Muzychka (2006)’s upper bound was slightly underestimated, ring flows. The relation between f and Re was clearly distinct for
though the lower bound matched the data well (Fig. 8). As hydrau- the flow regimes (Fig. 7), which means that different flow regimes
lic diameter increased, the two-phase multiplier decreased; this have different values of n. Therefore, a flow regime based on the
trend is similar to that of Zhang et al.’s correlation. The correlation HFM was developed using Beattie and Whalley (1982)’s viscosity
suggested by Zhang et al. uses C values of 1.31, 0.87, 0.39, 0.82 and model, which was assessed as the best two-phase viscosity model.
0.38 for hydraulic diameters of 490, 322, 143, 304 and 141 lm, The exponent n and the coefficient N in Eq. (7) were evaluated for
respectively. However, Zhang et al. (2007)’s correlation under-pre- flow regimes using a regression method. New correlations for dif-
dicted pressure drop for larger hydraulic diameters. Originally, the ferent flow regimes (Fig. 10) show no dependency on AR and Dh in
L–M correlation represented the two-phase multiplier monotoni- the bubble and transition regimes, but an obvious effect of hydrau-
cally increased as the Martinelli parameter decreased. This means lic diameters in the liquid ring regime. As AR or Dh increased, a
that the pressure drop always increased with increasing gas coefficient N in Eq. (7) increased. We proposed a new correlation
superficial velocity. The reason is that the Martinelli parameter is of a function of AR, because the AR shows higher correlation
inversely proportional to this quantity in Eq. (19). Moreover, the (Eqs. (35)–(37)).

Bubble regime : f TP ¼ 6:48Re0:85


TP ; ð35Þ

Transition regime : f TP ¼ 5:43Re0:85


TP ; ð36Þ

Liquid ring regime : f TP ¼ ðAR þ 0:46ÞRe0:6


TP : ð37Þ

The new correlation based on the flow pattern more accurately


predicted the two-phase pressure drop with MAE (%) of 10.48, 7.47,
21.5, 10.26 and 16.34 for Dh = 490, 322, 143, 304 and 141 lm,
respectively. In the most of data, the error is reduced to approxi-
mately ±35% (Fig. 11).

5.2. New flow pattern based SFM

In the SFM, the hydraulic diameter was related to a two-phase


multiplier (Fig. 8). As diameter decreased, C-value decreased in
Fig. 8. Two-phase multiplier and Martinelli parameter for different diameter the same manner as in Zhang et al. (2007)’s correlation. The flow
microchannels. pattern was highly related to the pressure drop (Fig. 9). The C-value
was higher in the bubble regime than in the liquid ring regime,
which means that the interaction between the two phases is higher
in the bubble regime than in the liquid regime. The C-value is lin-
early related to the mass flux in each flow pattern (Fig. 12); this
relationship is the clearest in the liquid ring flow regime. There-
fore, the flow pattern based correlation was developed by modify-
ing Qu and Mudawar (2003)’s correlation, which was considered to
be the best correlation. The correlation of Zhang et al. (2007)
underestimated this experimental result, so the coefficient
0.358 in Eq. (19) was changed to 1.612, which is modified
Zhang’s C-value (CM). Finally, the flow pattern based correlations
were obtained using regression (Eqs. (38)–(40)):

Bubble regime : C ¼ C M ð0:0012G þ 1:473Þ; ð38Þ

Transition regime : C ¼ C M ð0:0008G þ 0:95Þ; ð39Þ

Liquid ring regime : C ¼ C M ðaG þ bÞ;


Fig. 9. A C-value of Lockhart and Martinelli’s correlation in different flow conditions
for Dh = 322 lm.
ða ¼ 0:658 AR þ 0:13; b ¼ 0:0016 AR þ 0:0003Þ: ð40Þ
1206 C.W. Choi, M.H. Kim / International Journal of Heat and Fluid Flow 32 (2011) 1199–1207

Fig. 10. Two-phase friction factor and Reynolds number for different flow regimes: (a) bubble regime, (b) transition regime and (c) liquid ring regime.

Fig. 11. Comparison of new flow pattern based homogeneous flow model with all Fig. 13. Comparison of a new flow pattern based separated flow model with all
experimental data. experimental data.

The coefficient a and b had higher correlations with AR than with


Dh. Finally coefficients were correlated (Eq. (40)).
The new correlation based on the flow pattern was more accu-
rately predicted a two-phase pressure drop with MAE (%) of 4.61,
8.08, 11.03, 20 and 11.93 for Dh = 490, 322, 143, 404 and 141 lm,
respectively. For most of the data, an error was reduced to approx-
imately ± 30% (Fig. 13).

6. Conclusions

A two-phase pressure drop in a microchannel has been widely


studied. However, few studies have reported a pressure drop in
single rectangular microchannels. Although the flow pattern is a
critical parameter in two-phase flow phenomena, visualization re-
sults are rare due to the difficulties of fabricating microchannels. In
this study, single glass rectangular microchannels with different
diameters were fabricated.
The adiabatic two-phase flow in the rectangular microchannels
Fig. 12. C-value and mass flux according to flow regimes for Dh = 490 lm. was achieved using water-liquid and Nitrogen-gas. A frictional
C.W. Choi, M.H. Kim / International Journal of Heat and Fluid Flow 32 (2011) 1199–1207 1207

pressure drop was directly measured using embedded pressure Choi, C.W., Kim, M.H. 2008 The fabrication of a single glass microchannel to study
the hydrophobicity effect on two-phase flow boiling of water. J. Micromech.
ports in the microchannels. The transparent channel walls allowed
Microeng., doi: 18.105016.
visualization of the flow pattern using a high-speed camera and a Choi, C.W., Yu, D.I., Kim, M.H., 2010a. Adiabatic two-phase flow in rectangular
long–distance microscope. Experimental data were used to assess microchannels with different aspect ratios: Part I – Flow pattern, pressure drop
seven two-phase HFM viscosity models and ten correlations based and void fraction. Int. J. Heat Mass Transfer. doi:10.1016/
j.ijheatmasstransfer.2010.07.067.
on SFMs. Finally, an HFM and an SFM based on flow patterns were Choi, C.W., Yu, D.I., Kim, M.H., 2010b. Adiabatic two-phase flow in rectangular
proposed. microchannels with different aspect ratios: Part II – Bubble behaviors and
From the present study, the following main conclusions can be pressure drop in single bubble. Int. J. Heat Mass Transfer. doi:10.1016/
j.ijheatmasstransfer.2010.07.035.
drawn. Chung, P.M.-Y., Kawaji, M., 2004. The effect of channel diameter on adiabatic two-
phase flow characteristics in microchannels. Int. J. Multiph. Flow 30, 735–761.
(1) Major flow patterns in the rectangular microchannels are (1) Cicchitti, A., Lombardi, C., Silvestri, M., Solddaini, G., Zavalluilli, R., 1960. Two-phase
cooling experiments-pressure drop, heat transfer, and burnout measurement.
a bubble flow including a bubbly, slug bubble and elongated Energia Nucl. 7, 407–425.
bubble, (2) liquid ring flow, and (3) transition between elon- Cioncolini, A., Thome, J.R., Lombardi, C., 2009. Unified macro-to-micro method to
gated bubble flow and liquid ring flow. Two-phase frictional predict two-phase frictional pressure drops of annular flows. Int. J. Multiph.
Flow 35, 1138–1148.
pressure drops in rectangular microchannels are in accord Costa-Patry, E., Olivier, J., Nichita, B.A., Michel, B., Thome, J.R., 2011. Two-phase flow
with the flow patterns. This means that the flow pattern of refrigerants in 85 lm-wide multi-microchannels: Part I – Pressure drop. Int.
was highly related to the frictional pressure drop J. Heat Fluid Flow 32, 451–463.
Dukler, A.E., Wicks, M., Cleveland, 1964. Pressure drop and hold-up in two-phase
mechanism.
flow: Part A – A comparison of existing correlations and part B-an approach
(2) The experimental data were used to assess seven two-phase through similarity analysis. AIChE J. 10, 38–51.
viscosity models. The most accurate two-phase viscosity Garimella, S., Killion, J.D., Coleman, J.W., 2003. An experimentally validated model
model is that of Beattie and Whalley (1982). This model for two-phase pressure drop in the intermittent flow regime for non-circular
microchannels. J. Fluid. Eng. 125, 887–894.
was developed considering the bubble and annular flow pat- Hartnett, J.P., Kostic, M., 1989. Heat transfer to newtonian and non-newtonian fluids
terns. However, the minimum deviation is approximately in rectangular ducts. Adv. Heat Transfer 19, 247–356.
±50%. Kawahara, A., Chung, P.M.-Y., Kawaji, M., 2002. Investigation of two-phase flow
pattern, void fraction and pressure drop in a microchannel. Int. J. Multiph. Flow
(3) The experimental data were used to assess ten correlations 28, 1411–1435.
based on the SFM. The most accurate correlation is that of Kawaji, M. 2008. Unique aspects of adiabatic two-phase flow in microchannels. In:
Lee and Mudawar (2005), which considers the effects of International Conference on Heat Transfer and Fluid Flow in Microscale,
Whistler, Canada.
diameter and mass flux. However, the minimum deviation Lee, H.J., Lee, S.Y., 2001. Pressure drop correlations for two-phase flow within
is approximately ±50%. horizontal rectangular channels with small heights. Int. J. Multiph. Flow 27,
(4) The relationship between two-phase friction and Reynolds 783–796.
Lee, J., Mudawar, I., 2005. Two-phase flow in high-heat-flux microchannel heat sink
number in the HFMs, and the C-value in the SFMs indicates for refrigeration cooling applications: Part I – Pressure drop characterisitics. Int.
that the flow pattern is strongly related to the two-phase J. Heat Mass Transfer 48, 928–940.
frictional pressure drop in the rectangular microchannels. Li, W., Wu, Z., 2010. A general correlation for adiabatic two-phase pressure drop in
micro/mini-channels. Int. J. Heat Mass Transfer 53, 2732–2739.
In addition, the flow patterns in the rectangular microchan-
Lin, S., Kwok, C.C.K., Li, R.Y., Chen, Z.H., Chen, Z.Y., 1991. Local frictional pressure
nel are different from existing patterns. Therefore, correla- drop during vaporization for R-12 through capillary tubes. Int. J. Multiph. Flow
tions based on the flow pattern were proposed for both 17, 95–102.
HFM and SFM. The new HFM based on flow pattern is devel- Lockhart, R.W., Martinelli, R.C., 1949. Proposed correlation of data for isothermal
two-phase, two-component flow in pipes. Chem. Eng. Prog. 45, 39–48.
oped from Beattie and Whalley (1982)’s viscosity model, and MacAdams, W.H., Woods, W.K., Bryan, R.L., 1942. Vaporization inside horizontal
is improved to its MAE of 13.21%. The new SFM based on the tubes in benzene-oil mixture. Trans. ASME 64, 193.
flow pattern was developed by modifying Qu and Mudawar Mishima, K., Hibiki, T., Nishihara, H., 1993. Some characterisitics of gas-liquid flow
in narrow rectangular ducts. Int. J. Multiph. Flow 19, 115–124.
(2003)’s correlation, and decreased its MAE of 11.13%. In Mishima, K., Hibiki, T., 1996. Some characterisitics of air-water two-phase flow in
both new correlations, a dependency on a channel diameter small diameter vertical tubes. Int. J. Multiph. Flow 22, 703–712.
was observed only in the liquid ring flow regime. This could Moriyama, K., Inoue, A., 1992. The thermal-hydraulic characteristics of two-phase
flow in extremely narrow channels (the frictional pressure drop and heat
be due to the different ARs of the rectangular cross-sections. transfer of boiling two-phase flow, analytical model). Heat Transfer-Jap. Res. 21,
To achieve more accurate correlations for rectangular micro- 838–856.
channels, studies of the effects of both diameter and AR on Muller-Steinhagen, H., Heck, K., 1986. A simple friction pressure drop correlation for
two-phase flow pipes. Chem. Eng. Process. 20, 297–308.
the pressure drop in microchannels are necessary. Owens, W.L., 1961. Two-phase pressure gradient. In: International Development in
Heat Transfer, Part II. ASME, New York, United States.
Qu, W., Mudawar, I., 2003. Measurement and prediction of pressure drop in two-
phase micro-channel heat sinks. Int. J. Heat Mass Transfer 46, 2737–2753.
References
Saisorn, S., Wongwises, S., 2008. Flow pattern, void fraction and pressure drop of
two-phase air-water flow in horizontal circular micro-channel. Exp. Thermal
Ali, M., Sadatomi, M., Kawaji, M., 1993. Adiabatic two-phase flow in narrow Fluid Sci. 32, 748–760.
channels between two Flat plates. Can. J. Chem. Eng. 71, 657. Shiferaw, D., Mahmoud, M., Karayiannis, T.G., Kenning, D.B.R., 2011. One-
Awad, M.M., Muzychka, Y.S. 2006. Bounds on two-phase frictional pressure drop dimensional semimechanistic model for flow boiling pressure drop in small
gradient in minichannels and microchannels. In: International Conference on to miro passages. Heat Transfer Eng. 32, 1150–1159.
Nanochannels, Microchannels and Minichannels, Limerick, Ireland. Thome, J.R., Dupont, V., Jacobi, A.M., 2004. Heat transfer model for evaporation in
Awad, M.M., Muzychka, Y.S., 2008. Effective property models for homogeneous microchannels. Int. J. Heat Mass Transfer 47, 3375–3385.
two-phase flows. Exp. Thermal Fluid Sci. 33, 106–113. Tran, T.N., Chyu, M.-C., Wambsganss, M.W., France, D.M., 2000. Two-phase pressure
Beattie, D.R.H., Whalley, P.B., 1982. A simple two-phase flow frictional pressure drop of refrigerants during flow boiling in small channels: an experimental
drop calculation method. Int. J. Multiph. Flow 8, 83–87. investigation and correlation development. Int. J. Multiph. Flow 26, 1739–1754.
Carey, V.C., 1992. Liquid–Vapor Phase-Change Phenomena. Taylor and Francis, Zhang, W., Hibiki, T., Mishima, K. 2007. Two-phase frictional pressure drop in mini-
United States (Chapter 10). channel. In: 6th International Conference on Multiphase Flow. International
Chisholm, D., 1967. A theoretical basis for the Lockhart–Martinelli correlation for Conference Multiphase Flow S7_Wed_C_38.
two-phase flow. Int. J. Heat Mass Transfer 10, 1767–1778.

You might also like