Вы находитесь на странице: 1из 13

Adiabatic two-phase frictional pressure drops in microchannels

Remi Revellin, John R. Thome


*
EPFL, STI ISE LTCM, ME Gl 464, Station 9, CH-1015 Lausanne, Switzerland
Received 21 March 2006; received in revised form 11 July 2006; accepted 11 July 2006
Abstract
Two-phase pressure drops were measured over a wide range of experimental test conditions in two sizes of microchannels (sight glass
tubes 0.509 and 0.790 mm) for two refrigerants (R-134a and R-245fa). Similar to the classic Moody diagram in single-phase ow, three
zones were distinguishable when plotting the variation of the two-phase friction factor versus the two-phase Reynolds number: a laminar
regime for Re
TP
< 2000, a transition regime for 2000 6 Re
TP
< 8000 and a turbulent regime for Re
TP
P8000. The laminar zone yields a
much sharper gradient than in single-phase ow. The transition regime is not predicted well by any of the prediction methods for two-
phase frictional pressure drops available in the literature. This is not unexpected since only a few data are available for this region in the
literature and most methods ignore this regime, jumping directly from laminar to turbulent ow at Re
TP
= 2000. The turbulent zone is
best predicted by the Mu ller-Steinhagen and Heck correlation. Also, a new homogeneous two-phase frictional pressure drop has been
proposed here with a limited range of application.
2006 Elsevier Inc. All rights reserved.
Keywords: Microchannels; Frictional; Pressure drop; Two-phase ow
1. Introduction
Micro- or mini-heat spreaders are used in the interest of
providing higher cooling capability for microtechnologies.
They are characterized by a high heat ux dissipation
and a better heat transfer coecient compared to conven-
tional processes. Higher eectiveness means, for an identi-
cal power, a reduction of size and cost. Compactness also
reduces the amount of charge of the uid, which has also
a direct positive impact on safety and environment. How-
ever, the negative point is possibly a higher pressure drop
related to the micro- or mini-ow channels.
The total pressure drop of a uid is due to the variation
of kinetic and potential energy and that due to friction, so
that the pressure drop is the sum of the static pressure drop
(elevation head), the momentum pressure drop (accelera-
tion) and frictional pressure drop:
dP
dz
_ _
t

dP
dz
_ _
s

dP
dz
_ _
m

dP
dz
_ _
f
1
The static pressure drop in a horizontal microchannel is 0
dP
dz
_ _
s
0 2
The momentum pressure drop takes into account the accel-
eration of the ow due to the ashing or diabatic eect and
is dened as follows:
dP
dz
_ _
m
G
2
Dxq
L
q
V

Lq
V
q
L

3
The frictional pressure drop
dP
dz
_ _
f
can be determined by
dierent models or correlations for macrochannels as those
reported in Table 1. The simplest one is the homogeneous
model that makes analysis of two-phase ows easier: this
ideal-uid obeys the usual equation of a single-phase uid
and is characterized by suitably averaged properties. Three
possible forms of the two-phase viscosity models are
reported in Table 1.
0894-1777/$ - see front matter 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.expthermusci.2006.07.001
*
Corresponding author.
E-mail addresses: remi.revellin@ep.ch (R. Revellin), john.thome@
ep.ch (J.R. Thome).
www.elsevier.com/locate/etfs
Experimental Thermal and Fluid Science 31 (2007) 673685
The separated ow model considers the two-phases to
be articially separated into two streams, each owing in
its own pipe. The cross-sectional ow areas of the two
pipes are proportional to the void fraction. The basic
equations for the separated ow model are not dependent
on the particular ow conguration adopted. It is assumed
that the velocities of each phase are uniform, in any given
cross-section, within the zone occupied by the phase. The
rst of these analyses was performed by Lockhart and
Martinelli [1].
The Friedel [2] correlation for the two-phase frictional
pressure drop was specially developed for conventional
channels as well as that of Chisholm [3].
A new correlation by Muller-Steinhagen and Heck [4]
for the prediction of frictional pressure drop for two-phase
ow in pipes was suggested which is simple and more con-
venient to use than other prior methods. The correlation
was developed using a data bank containing 9300 measure-
ments of frictional pressure drop for a variety of uids and
conditions, including channel diameters from 4 to 392 mm.
In Table 2, are reported the modied correlations or
models for microchannels.
Mishima and Hibiki [5] measured the frictional pressure
loss for airwater ows in vertical capillary tubes with
inner diameters in the range from 1 to 4 mm. The results
were compared with the Lockhart and Martinelli model.
The frictional pressure loss was reproduced well by
Chisholms equation with a new equation for Chisholms
parameter C as a function of inner diameter. Lee and Lee
[6] proposed new correlations for the two-phase pressure
drop through horizontal rectangular channels with small
gaps (heights) based on 305 data points. The gap between
the upper and the lower plates of each channel ranges from
0.4 to 4 mm while the channel width was xed to 20 mm.
Water and air were used as the test uids. The authors
expressed the two-phase frictional multiplier using the
LockhartMartinelli type correlation but with the modi-
cation on parameter C (see Table 3).
Lee and Mudawar [7] measured the two-phase pressure
drop across a microchannel heat sink that served as an
Nomenclature
B Chisholm parameter,
C constant of Lockhart and Martinelli or Chis-
holm parameter, m
C
0
connement number,
D diameter, m
e surface roughness, lm
E Friedel parameter,
F Friedel parameter,
F Mu ller-Steinhagen parameter, bar/m
Fr Froude number,
f friction factor,
G mass ux, kg/m
2
s
g acceleration of gravity, m/s
2
H Friedel parameter,
L length, m
P pressure, bar
q heat ux, W/m
2
Re Reynolds number,
T temperature, C
We Weber number,
x vapor quality,
X LockhartMartinelli parameter,
Y Chisholm parameter,
z longitudinal abscissa, m
Greek letters
a mean absolute error, a
1
N

N
1
predicted valueexperimental value
experimental value

, %
b fraction of data predicted to within 20%, %
e void fraction,
k Lee and Lee parameter,
/ two-phase multiplier,
w Lee and Lee parameter,
q density, kg/m
3
l dynamic viscosity, Pa s
r surface tension, N/m
Subscripts
crit critical
f frictional
G gas
h heated
h hydraulic
Hom homogeneous model
int internal
in inlet
L liquid
m momentum
MEV microevaporator
MPH micropreheater
O only
out outlet
s static
sat saturation
sub subcooling
t total
t turbulent
TP two-phase
TS test section
v viscous
V vapor
674 R. Revellin, J.R. Thome / Experimental Thermal and Fluid Science 31 (2007) 673685
evaporator in a refrigeration cycle. The microchannels were
formed by machining 231 lm wide 713 lm deep grooves
into the surface of a copper block. Experiments were per-
formed with refrigerant R-134a.
Zhang and Webb [8] measured adiabatic two-phase ow
pressure drops for R-134a, R-22 and R-404a owing in a
multi-port extruded aluminum tube with a hydraulic diam-
eter of 2.13 mm, and in two copper tubes having inside
diameters of 6.25 and 3.25 mm, respectively. They found
that the Friedel correlation did not predict the two-phase
data accurately, especially for high reduced pressure. Using
the data taken in their present and in a previous study, a
new correlation for two-phase friction pressure drop in
small tubes was developed by modifying the Friedel corre-
lation. The new correlation predicts 119 data points with a
mean deviation of 11.5%.
Two-phase ow pressure drop measurements were made
by Tran et al. [9] during a phase-change heat transfer pro-
cess with three refrigerants (R-134a, R-12 and R-113) at six
dierent pressures ranging from 138 to 856 kPa, and in two
sizes of round tubes (2.46 and 2.92 mm inside diameters)
and one rectangular channel (4.06 1.7 mm). The data
were compared with those from large tubes under similar
conditions, and state-of- the-art correlations were evalu-
ated using the R-134a data. The state-of-the-art large-tube
correlations failed to satisfactorily predict the experimental
data.
Garimella et al. [10] proposed the rst mechanistic
model for two-phase pressure drop during intermittent
ow of refrigerant in circular microchannel. The model
was developed for a unit cells in the channel based on the
Table 1
Pressure drop models and correlations for macrochannels
Authors Equation
Homogeneous
model
dP
dz
_ _
f

2f
TP
G
2
Dq
TP
with
q
TP

x
q
V

1x
q
L
_ _
1
f
TP

16
ReTP
for Re
TP
< 2000 or
f
TP
0:079Re
0:25
TP
for Re
TP
> 2000
and Re
TP

GD
l
TP
Three possible forms of the two-phase
viscosity models are:
McAdams et al. [13]: l
TP

x
l
V

1x
l
L
_ _
1
Cicchitti et al. [16]: l
TP
= (xl
V
+ (1 x)l
L
)
Dukler et al. [17]: l
TP
q
TP
x
l
V
q
V
1 x
l
L
q
L
_ _
Lockhart and
Martinelli [1]
dP
dz
_ _
f

dP
dz
_ _
L
U
2
L
with
U
2
L
1
C
X

1
X
2
with C given by:
C
tt
= 20 (liquid turbulent, gas turbulent),
C
vt
= 12, C
tv
= 10, C
vv
= 5
X
dP
dz

L
dP
dz

V
_ _
0:5
with
dP
dz
_ _
L
f
L
2G
2
Dq
L
1 x
2
and
dP
dz
_ _
V
f
V
2G
2
Dq
V
x
2
f
L

16
ReL
for Re
L
< 2000 or f
L
0:079Re
0:25
L
for Re
L
> 2000
f
V

16
ReV
for Re
V
< 2000 or f
V
0:079Re
0:25
V
for Re
V
> 2000
Re
L

G1xD
l
L
and Re
V

Gx D
l
V
Friedel [2]
dP
dz
_ _
f

dP
dz
_ _
LO
U
2
LO
with
U
2
LO
E
3:24FH
Fr
0:045
We
0:035
Fr
G
2
gDq
2
H
, F = x
0.78
(1 x)
0.224
,
H
q
L
q
V
_ _
0:91
l
V
l
L
_ _
0:19
1
l
V
l
L
_ _
0:7
We
G
2
D
rq
H
with q
H

x
q
V

1x
q
L
_ _
1
E 1 x
2
x
2 q
L
fVO
q
V
fLO
f
LO

16
ReLO
for Re
LO
< 2000 or
f
LO
0:079Re
0:25
LO
for Re
LO
> 2000
f
VO

16
ReVO
for Re
VO
< 2000 or
f
VO
0:079Re
0:25
VO
for Re
VO
> 2000
Re
LO

GD
l
L
and Re
VO

GD
l
V
and
dP
dz
_ _
LO
f
LO
2G
2
Dq
L
Chisholm [3]
dP
dz
_ _
f

dP
dz
_ _
LO
U
2
LO
with
U
2
LO
1 Y
2
1 Bx
2n
2
1 x
2n
2
x
2n
_ _
where the exponent n = 0.25 and
Y
2

dP=dz
VO
dP=dz
LO
and
dP
dz
_ _
VO
f
VO
2G
2
Dq
V
;
if 0 < Y < 9.5:
B
55
G
0:5
for G P1900 kg/m
2
s
B
2400
G
for 500 < G < 1900 kg/m
2
s
B = 4.8 for G < 500 kg/m
2
s
If 9.5 < Y < 28:
B
520
YG
0:5
for G6 600 kg/m
2
s
B
21
Y
for G > 600 kg/m
2
s
For Y > 28:
B
15000
Y
2
G
0:5
Mu ller-Steinhagen
and Heck [4]
dP
dz
_ _
f
F 1 x
1=3

dP
dz
_ _
VO
x
3
with F
dP
dz
_ _
LO
2
dP
dz
_ _
VO

dP
dz
_ _
LO
_ _
x
Table 2
Pressure drop models and correlations for microchannels
Authors Equation
Mishima and Hibiki [5] C = 21(1 e
319D
)
using the LockhartMartinelli model
Lee and Lee [6] C Ak
q
w
r
Re
s
LO
using the LockhartMartinelli model
k
l
2
L
q
L
rD
and w
l
L
j
r
The exponents A, q, r and s are
given in Table 3
Lee and Mudawar [7] For laminar liquid and laminar vapor:
C
vv
2:16Re
0:047
LO
We
0:6
LO
For laminar liquid and turbulent vapor:
C
vt
1:45Re
0:25
LO
We
0:23
LO
with We
LO

G
2
D
q
L
r
using the LockhartMartinelli model
Zhang and Webb [8]
dP
dz
_ _
f

dP
dz
_ _
LO
U
2
LO
with U
2
LO
1 x
2
2:87x
2 P
Pcrit
_ _
1

1:68x
0:8
1 x
0:25 P
Pcrit
_ _
1:64
Tran et al. [9]
dP
dz
_ _
f

dP
dz
_ _
LO
U
2
LO
with U
2
LO
1 4:3Y
2
1
Cox
0:875
1 x
0:875
x
1:75
_ _
Co
r
D
2
gq
L
q
V

_ _
0:5
and Y
2

dP=dz
VO
dP=dz
LO
R. Revellin, J.R. Thome / Experimental Thermal and Fluid Science 31 (2007) 673685 675
observed slug/bubble ow pattern for these conditions. The
unit cell comprises a liquid slug followed by a vapor bubble
that is surrounded by a thin, annular liquid lm. Authors
take into account dierent phenomena. They rst of all
accounted for the continual uptake of liquid from lm into
the front of the slug (DP
lmslug transitions
) as the bubble
travels faster than the liquid slug. They also accounted
for the pressure drop in bubble/lm interface (DP
f/b
) and
the pressure drop in the liquid slug (DP
slug
). The total pres-
sure drop is given by the following equation:
DP DP
slug
DP
f=b
DP
filmslug transitions
4
DP
lmslug transitions
is directly related to the number of unit
cell per meter. An experimental correlation is proposed for
predicting the slug frequency. It was assumed in the model
that the length/frequency/speed of bubble/slugs is con-
stant, with no bubble coalescence and a smooth bubble/
lm interface. The model has been experimentally validated
for condensing R-134a.
2. Description of the test facility
The microchannel test facility is described in detail in
Revellin [11] and available at the university website. The
test facility was designed to operate using either a speed
controlled micropump or the pressure dierence between
its two temperature-controlled reservoirs (the latter mode
was used for all the present tests and is presented in
Fig. 1). A valve installed between the upstream reservoir
and the test section is used to avoid ow oscillations in
the loop and a wide range of stable operating conditions
is thus achieved.
2.1. Test section
The test section consisted of four subsections: (i) an
80 mm long, thin wall stainless steel tube used as a pre-
heater, (ii) a 20 mm long glass tube for electrical insulation,
(iii) a 110 mm long stainless steel tube microevaporator and
(iv) a 100 mm long glass tube for ow pattern visualization
and pressure drop measurements, as shown in Fig. 2. Two
copper clamps were attached to both the preheater and
the evaporator and heating was provided by two Sorensen
DCpower supplies. Two pressure transducers were installed
at the inlet and outlet of the test section and two 0.25 mm
thermocouples were placed in the uid at the same loca-
tions. Agood agreement between the two dierent measure-
ments (pressure and temperature) was observed at the outlet
of the test section (saturation conditions). Four 0.25 mm
thermocouples were also attached on the external surface
of the sight glass tubes (before the inlet and after the outlet
of both the preheater and the evaporator) to measure local
uid saturation temperatures. After calculation it has been
found that the heat conduction through the thermocouples
was negligible (23%). Furthermore, the internal thermal
resistance of the ow is much lower than the wall thermal
resistance thus, the longitudinal heat conduction can also
be neglected. Two more 0.25 mm thermocouples were
installed on the two heated tubes to avoid dryout and over-
heating. All the test section was thermally insulated. Careful
attention was made to match up the ends of the glass and
stainless steel tubes.
3. Measurements and accuracy
A Coriolis mass ow meter was used to measure the
ow rate of the subcooled refrigerant. Joule heating was Fig. 1. Schematic of reservoir loop.
Fig. 2. Schematic of the test section.
Table 3
Constant and exponents for parameter C of Lee and Lee [6]
Flow regime liquid Gas A q R S Range of X Range of Re
LO
Number of data
Laminar Laminar 6.833 10
8
1.317 0.719 0.557 0.77614.176 1751480 106
Laminar Turbulent 6.185 10
2
0 0 0.726 0.3031.426 2931506 2
Turbulent Laminar 3.627 0 0 0.174 3.27679.415 260617,642 85
Turbulent Turbulent 0.408 0 0 0.451 1.30914.781 267517,757 62
Laminar: Re
L
, Re
G
< 2000; turbulent: Re
L
, Re
G
> 2000.
676 R. Revellin, J.R. Thome / Experimental Thermal and Fluid Science 31 (2007) 673685
determined by measuring the DC voltage (0.02%) and the
current by a DC current transformer (3.5% for low cur-
rents and 1% for high ones). The absolute pressure trans-
ducers for monitoring the local pressures were accurate to
5 mbar and the thermocouples to 0.1 C, according to
their calibrations. The vapor quality entering the ow visu-
alization tube was estimated to be accurate to 2% for
most test conditions.
The pressure drops in the visualization tube reached up
to 0.174 MPa, which were obtained indirectly fromthe mea-
sured dierences in saturation temperature between the
inlet (0.25 mm thermocouple attached on the adiabatic sur-
face of the microtube after the microevaporator) and outlet
(0.25 mm thermocouple installed in the uid after the glass
tube) of the visualization tube. This temperature measure-
ment is preferred to pressure measurement as no opening
is made in the microtube to install a pressure tap. The ow
is thus not deformed or disturbed. Moreover, this technique
avoids also the capillary pressure due to the meniscus
liquid/vapor before the pressure transducer that could inu-
ence the measurement. However, for low pressure drops,
the inaccuracy is higher with this method than with the pres-
sure measurement. But as the pressure drop is high in
microtubes, the accuracy of the measurement remains good.
The total database from this study and our prior publi-
cations covers two dierent refrigerants (R-134a and
R-245fa) and two diameter tubes (0.509 and 0.790 mm).
The microevaporator heated length varied from 20 to
70 mm and the inlet subcooling from 2 to 15 C. The mass
uxes ranged from 200 to 2000 kg/m
2
s and the maximum
heat ux was 415 kW/m
2
. Three dierent saturation tem-
peratures were tested: 26, 30 and 35 C. Experimental con-
ditions and uncertainties are summarized in Table 4.
Notably, the present setup involves actual two-phase
ows exiting a microevaporator channel. The bubbles
and subsequent ow regimes observed here originated from
nucleation in the evaporator, just like in a microchannel
cooling element attached to a computer chip, for instance.
Thus, here the resulting ow pattern and bubble character-
istics are determined by the boiling process itself, not
the hydrodynamics of an injector, mixer or header used
in adiabatic tests.
4. Experimental results
There were 2210 experimental two-phase pressure drops
data points measured in this study. Thirty data points are
not taken into account as their values are below zero due
to small instabilities and error measurements. The total
pressure drop is calculated using pressure dierence.
According to Eqs. (1) and (3) the momentum pressure drop
is subtracted from the total pressure drop to obtain the
two-phase frictional pressure drop
dP
dz
_ _
f
. The momentum
pressure drop is caused by the slight ashing eect due to
the pressure drop but it is negligible (less than 0.1%) as
shown in Fig. 3, but is taken into account anyway.
The general trend for the two-phase frictional pressure
drop is illustrated in Fig. 4. The data are plotted according
to the vapor quality calculated at 20 mm from outlet of the
microevaporator. In general, the higher the mass ux, the
higher the vapor quality, the higher the two-phase
frictional pressure drop is (the expected trend). However,
a change in the trend can be observed for G = 1000 kg/
m
2
s and G = 1200 kg/m
2
s. Three arrows show this
change. It corresponds to a change in ow pattern. The
ow is annular for case (a) and (b) according to the ow
pattern map of Revellin et al. [12]. But the annular ow
is wavy annular for case (a) and tends to be separated into
smooth annular and wavy annular ow for case (b) as
shown in Fig. 5(a) and (b). (This can happen also for
semi-annular ow which exhibits both smooth annular
and churn ow zones.) The friction factor is denitely less
for smooth than for wavy annular ow so the two-phase
pressure drop is less for case (b) than for case (a).
For very low vapor quality one can surprisingly observe
the two-phase frictional pressure drops decreasing with
increasing vapor quality. Afterwards the two-phase fric-
tional pressure drop increases with the vapor quality with
Table 4
Experimental conditions and uncertainties
Parameter Range Uncertainties Units
Fluid R-134a, R-245fa
D 0.509, 0.790 1% mm
e
D
<0.002%
L
MEV
3070 <2.5% mm
G 2102094 2% kg/m
2
s
q 3.1415 <5.7% kW/m
2
T
sat
26, 30, 35 0.1 C
P
sat
6.9, 7.7, 8.9 (for R-134a) <0.07% bar
2.1 (for R-245fa) <0.23%
DT
sub
26 0.1 C
x
MEV,out
00.95 >1.3%
dP
dz
_ _
t
014.5 >1.1% bar/m
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
10
3
10
2
10
1
10
0
10
1
Vapor quality []
P
r
e
s
s
u
r
e

d
r
o
p

[
b
a
r
/
m
]
Total pressure drop
Frictional pressure drop
Momentum pressure drop
Fig. 3. Comparison between the total, the frictional and the momentum
pressure drop as a function of vapor quality from laser 1 for R-134a,
D = 0.509 mm, L
MEV
= 70.70 mm, T
sat
= 30 C, G = 1800 kg/m
2
s and
DT
sub
= 3C.
R. Revellin, J.R. Thome / Experimental Thermal and Fluid Science 31 (2007) 673685 677
two dierent gradients, where the higher gradient is for
lower vapor qualities. The change in the gradient of the
two-phase frictional pressure drop will be explained later.
4.1. Eect of the inlet subcooling
As expected, the inlet subcooling at the entrance to the
microevaporator had no eect on the saturated two-phase
frictional pressure drops as shown in Fig. 6. The small inlet
subcoolings tested here had no inuence on the ow pat-
terns and thus the two-phase mechanisms are not aected.
4.2. Eect of the saturation temperature
As presented in Fig. 7, the saturation temperature inu-
ences the two-phase frictional pressure drop. The higher
the saturation temperature, the lower the two-phase fric-
tional pressure drop is. When increasing the saturatizon
temperature, the vapor density increases, the vapor velocity
decreases and by consequence the two-phase frictional
pressure drop decreases. The results here show the expected
tendency from macroscale theory.
0 0.2 0.4 0.6 0.8 1
0
5
10
15
Vapor quality []
T
w
o

P
h
a
s
e

f
r
i
c
t
i
o
n
a
l

p
r
e
s
s
u
r
e

d
r
o
p

[
b
a
r
/
m
]
G=1000 kg/m
2
s
G=700 kg/m
2
s
G=500 kg/m
2
s
G=350 kg/m
2
s
0 0.2 0.4 0.6 0.8 1
0
5
10
15
Vapor quality []
T
w
o

P
h
a
s
e

f
r
i
c
t
i
o
n
a
l

p
r
e
s
s
u
r
e

d
r
o
p

[
b
a
r
/
m
]
G=2000 kg/m
2
s
G=1800 kg/m
2
s
G=1500 kg/m
2
s
G=1200 kg/m
2
s
a
b
Fig. 4. Two-phase frictional pressure drop as a function of the vapor
quality from laser 1 for R-134a, D = 0.509 mm, L
MEV
= 70.70 mm,
T
sat
= 30 C and DT
sub
= 3 C.
Fig. 5. Flow patterns for R-134a and D = 0.509 mm. (a) Wavy annular
ow and (b) smooth annular ow.
0 0.2 0.4 0.6 0.8 1
0
5
10
15
Vapor quality []
T
w
o

P
h
a
s
e

f
r
i
c
t
i
o
n
a
l

p
r
e
s
s
u
r
e

d
r
o
p

[
b
a
r
/
m
]
T
sub
=2C
T
sub
=3C
T
sub
=5C
Fig. 6. Inuence of the inlet subcooling at entrance to the microevapora-
tor on the two-phase frictional pressure drop for R-134a, D = 0.509 mm,
L
MEV
= 70.70 mm, T
sat
= 35 C and G = 1500 kg/m
2
s.
0 0.2 0.4 0.6 0.8 1
0
5
10
15
Vapor quality []
T
w
o

P
h
a
s
e

f
r
i
c
t
i
o
n
a
l

p
r
e
s
s
u
r
e

d
r
o
p

[
b
a
r
/
m
]
T
sat
=26C
T
sat
=30C
T
sat
=35C
Fig. 7. Inuence of the saturation temperature on the two-phase frictional
pressure drop for R-134a, D = 0.509 mm, L
MEV
= 70.70 mm, DT
sub
= 3 C
and G = 1800 kg/m
2
s.
678 R. Revellin, J.R. Thome / Experimental Thermal and Fluid Science 31 (2007) 673685
4.3. Eect of the microevaporator length
As the microevaporator length has no eect on the ow
patterns, it was expected that the two-phase frictional pres-
sure drops measured in the sight glass tube would not be
aected by this variation. Fig. 8 shows this to be the case.
4.4. Eect of the microevaporator diameter
As can be seen in Fig. 9, the diameter of the sight glass
tube has a strong inuence on the two-phase frictional
pressure drop. As expected, the larger the diameter, the
lower the two-phase pressure drop is. The same trends
are observed for both diameters.
4.5. Eect of the uid
The uid has an important inuence on the two-phase
frictional pressure drop. The values for R-245fa are
strongly higher than those for R-134a. Actually
q
V
(R-134a) 3q
V
(R-245fa) so the vapor velocity for
R-245fa is higher than for R-134a and the result is the
increase of the pressure drop for R-245fa (see Fig. 10).
4.6. The two-phase friction factor
As observed in the previous gures, the variation of the
two-phase frictional pressure drop can be divided into
three dierent zones according to the two-phase Reynolds
number. Firstly, the two-phase frictional pressure drop
decreases, secondly the two-phase frictional pressure drop
increases with a certain gradient and thirdly the two-phase
frictional pressure drop increases with a lower gradient as
can be seen in Fig. 4.
It is possible to plot the pressure drop data as the two-
phase friction factor versus the two-phase Reynolds num-
ber as shown in Fig. 11. l
TP
[13] and q
TP
are calculated
using the denitions of Table 1 for the homogeneous model.
The two-phase friction factors are determined from the
measured pressure gradients using Eq. (5). The values of
Re
TP
are obtained using the denition of Table 1 for the
homogeneous model. Three dierent zones are detected.
An analogy with the behavior of the single-phase friction
factor is made as follows:
dP
dz
_ _
f

dP
dz
_ _
TP

2f
TP
G
2
Dq
TP
5
Re
TP
< 2000: The rst zone is called the laminar zone.
The two-phase friction factor diminishes when increas-
ing the two-phase Reynolds number.
0 0.2 0.4 0.6 0.8 1
0
5
10
15
Vapor quality []
T
w
o

P
h
a
s
e

f
r
i
c
t
i
o
n
a
l

p
r
e
s
s
u
r
e

d
r
o
p

[
b
a
r
/
m
]
L
MEV
=30 mm
L
MEV
=50 mm
L
MEV
=70 mm
Fig. 8. Inuence of the microevaporator length on the two-phase frictional
pressure drop for R-134a, D = 0.509 mm, T
sat
= 30 C, DT
sub
= 3 C and
G = 1800 kg/m
2
s.
0 0.2 0.4 0.6 0.8 1
0
5
10
15
Vapor quality []
T
w
o

P
h
a
s
e

f
r
i
c
t
i
o
n
a
l

p
r
e
s
s
u
r
e

d
r
o
p

[
b
a
r
/
m
]
D=0.509 mm
D=0.790 mm
Fig. 9. Inuence of the microevaporator diameter on the two-phase
frictional pressure drop for R-134a, T
sat
= 30 C, L
MEV
= 70.70 mm,
DT
sub
= 3 C and G = 1200 kg/m
2
s.
0 0.2 0.4 0.6 0.8 1
0
5
10
15
Vapor quality []
T
w
o

P
h
a
s
e

f
r
i
c
t
i
o
n
a
l

p
r
e
s
s
u
r
e

d
r
o
p

[
b
a
r
/
m
]
R134a
R245fa
Fig. 10. Inuence of the uid on the two-phase frictional pressure drop
for D = 0.509 mm, T
sat
= 35 C, L
MEV
= 70.70 mm, DT
sub
= 56 C and
G = 700 kg/m
2
s.
R. Revellin, J.R. Thome / Experimental Thermal and Fluid Science 31 (2007) 673685 679
2000 6 Re
TP
< 8000: The second zone is the transition
zone. The two-phase friction factor increases with the
Reynolds number or decreases, depending to the condi-
tions. A case could also be made here to set the upper
boundary at 4500 rather than 8000 in viewing Fig. 11
but in (b) and (c) some transition eects still appear to
fall in this range.
Re
TP
P8000: The third zone corresponds to the turbu-
lent zone. The data are grouped and decrease with
increasing two-phase Reynolds number.
The two-phase friction factor behavior is the same as for
the single phase friction factor. It is interesting to empha-
size here that each zone corresponds to each trend observed
earlier when plotting the two-phase frictional pressure drop
versus the vapor quality.
The laminar zone corresponds most often to the bubbly
or bubbly/slug regime [12]. Usually, the pressure drop
should not decrease and then increase afterwards as there
should be a continuous increase of the pressure drop with
the two-phase Reynolds number. However, macroscale
two-phase pressure drop data nearly always fall only in
the turbulent zone and this trend is not thus observed. Fur-
thermore, it is interesting to emphasize that the transition
zone occurs in the same range of Reynolds number as for
single-phase ow. The laminar zone will be ignored for
this study as it is dicult to make a detailed analysis
since it corresponds to only 74 data points, or 3.3% of
the database.
The rest of the database will be divided into two sets of
data (936 data points for the transition zone and 1200 data
points for the turbulent region) and compared to available
correlations or models. As the transition zone cannot be
predicted by any of the existing methods, only the turbu-
lent zone will be presented here.
5. Comparison with existing methods
The comparison between the present experimental data
(for R-134a, R-245fa and both sight glass tube diameters)
and twelve models and correlations available in the litera-
ture (described in Section 1) is shown in Table 5 for the
10
3
10
4
10
3
10
2
10
1
10
0
Twophase Reynolds number []
T
w
o

p
h
a
s
e

f
r
i
c
t
i
o
n

f
a
c
t
o
r

[

]
10
3
10
4
10
3
10
2
10
1
10
0
Twophase Reynolds number []
T
w
o

p
h
a
s
e

f
r
i
c
t
i
o
n

f
a
c
t
o
r

[

]
G=1200 kg/m
2
s
G=1000 kg/m
2
s
G=700 kg/m
2
s
G=500 kg/m
2
s
Laminar Transition Turbulent
G=1000 kg/m
2
s
G=700 kg/m
2
s
G=500 kg/m
2
s
G=350 kg/m
2
s
Laminar Transition Turbulent
10
3
10
4
10
3
10
2
10
1
10
0
Two phase Reynolds number []
T
w
o
p
h
a
s
e

f
r
i
c
t
i
o
n

f
a
c
t
o
r

[

]
G=1000 kg/m
2
s
G=200 kg/m
2
s
G=500 kg/m
2
s
G=350 kg/m
2
s
G=200 kg/m
2
s
Laminar Transition Turbulent
Fig. 11. Two-phase friction factor versus two-phase Reynolds number. (a) R-134a, D = 0.509 mm, T
sat
= 30 C, L
MEV
= 70.70 mm, DT
sub
= 3 C.
(b) R-134a, D = 0.790 mm, T
sat
= 30 C, L
MEV
= 70.70 mm, DT
sub
= 3 C. (c) R-245fa, D = 0.509 mm, T
sat
= 35 C, L
MEV
= 70.70 mm, DT
sub
= 6 C.
680 R. Revellin, J.R. Thome / Experimental Thermal and Fluid Science 31 (2007) 673685
transition and turbulent region. No prediction method
works very well to predict the present experimental data,
except for the Mu ller-Steinhagen and Heck [4] correlation
that predicts more than 62% of the data within a 20%
error band for the turbulent data with a mean absolute
error of 18.3%. Furthermore, it predicts 83% of the data
for R-134a in the 0.509 mm tube within this 20% error
band. For R-134a in the 0.790 mm tube, the results are
not well predicted, as only 29% of the data fall within a
20% error band while 62% of the data are within 20%
for R-245fa. Ribatski et al. [14] found also that Mu ller-
Steinhagen and Heck [4] was the best correlation to predict
two-phase frictional pressure drop in microchannels. They
used a huge database from the literature covering eight u-
ids and compared with twelve existing predictions methods.
The homogeneous models under predict the data. In
fact, some methods over predict the data while others
under predict them. A new prediction method has been
developed and is proposed below.
6. Comparison to a phenomenological model
Fig. 12(a) shows a comparison between experimental
data of two-phase frictional pressure drop and the Garim-
ella et al. [10] model for intermittent ows of R-134a and
R-245fa in a 0.509 and 0.790 mm tube (874 data points).
As can be seen, only 20.4% of the data fall within a
20% error band. The mean deviation is 51.7%. The repar-
tition of the data as a function of the two-phase Reynolds
number is represented in Fig. 12(b). Most of the data are in
the transition zone and only a few of them in the turbulent
zone. The comparison has also been done inputting our
original experimental bubble frequencies directly into the
model. The results are then in even poorer agreement than
the model.
Some explanations may explain the dierence between
the experimental data and the Garimella et al. model. First
of all, the model has been experimentally validated for con-
densing R-134a whereas the data in the present study are
for evaporating R-134a and R-245fa. The break down of
annular to intermittent ow in condensation has dierent
mechanisms than coalescence in evaporation. Secondly,
Table 5
Statistical analysis of pressure drop (the fraction of data predicted to within 20%, b, and the mean absolute error, a)
Frictional pressure drop prediction method Transition regime Turbulent regime
2000 6 Re
TP
< 8000 Re
TP
P8000
b (%) a (%) b (%) a (%)
McAdams et al. [13] 28.1 45.9 5.75 39.8
Dukler et al. [18] 23.8 46.2 3.58 43.5
Cicchitti et al. [17] 34.7 45.2 52.3 25.9
Lockhart and Martinelli [1] 26.6 69.3 22.8 51.4
Friedel [2] 33.3 55.9 44.7 28.1
Chisholm [3] 22.8 162 28.7 65.2
Mishima and Hibiki [5] 10.5 56 1.58 51.7
Lee and Lee [6] 23.7 69 18.9 51.4
Lee and Mudawar [7] 16.9 109 18.1 70
Zhang and Webb [8] 30.2 81.9 46.8 34.7
Tran et al. [9] 11.2 474 1.75 355
Mu ller-Steinhagen and Heck [4] 30.1 68 62.5 18.3
0 1 2 3 4 5
0
1
2
3
4
5
Experimental pressure drop [bar/m]
P
r
e
s
s
u
r
e

d
r
o
p

f
r
o
m

t
h
e

m
o
d
e
l

[
b
a
r
/
m
]
20.4 % of the values within 20 %
Garimella model
Experimental data
20 % error band
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
x 10
4
0
10
20
30
40
50
60
70
80
90
100
Twophase Reynolds number []
N
u
m
b
e
r

o
f

d
a
t
a

p
o
i
n
t
s
Fig. 12. (a) Comparison between the experimental data of two-phase
frictional pressure drop in the slug ow regime and the Garimella et al.
[10] model and (b) repartition of the experimental data of two-phase
frictional pressure drop as a function of the two-phase Reynolds number.
R. Revellin, J.R. Thome / Experimental Thermal and Fluid Science 31 (2007) 673685 681
in the model, a constant frequency has been assumed but
actually the bubble frequency is not constant due to the
coalescence or break down phenomena. Furthermore, the
dierence between the model and the experimental data
could be explained by the wide range of Re
TP
. AS shown
in Fig. 12(b), most of the data in this work are in the tran-
sition zone. As expected, this transition is dicult to pre-
dict accurately with any of the existing methods. The
model of Garimella et al. has been validated for parallel
circular multichannels whereas here a single circular chan-
nel is used. This particularity could also explain the dier-
ence between the model and the present study.
7. New prediction method
Two possibilities are oered to develop a new prediction
method: In rst, modication of the homogeneous model
using the two-phase friction factor. This assumption is
not far away from reality as it has been shown in Revellin
et al. [12] that the homogeneous model predicted the void
fraction rather well. Moreover, Agostini and Bontemps
[15] found that homogeneous model predicted their R-
134a two-phase pressure drop data very well. Secondly,
the classic LockhartMartinelli method can be modied
with a new C parameter. Both approaches are widely used
in the literature to predict two-phase pressure drops. Only
the results for the homogeneous model will be shown here
as this approach gives the best prediction.
Based on Section 4.6, it is possible to determine a new
two-phase friction factor for the turbulent data where
8000 6 Re
TP
. Fig. 13 shows the new predicted two-phase
friction factor determined with a least square method and
given by the following equations:
f
TP
0:08Re
1=5
TP
for D 0:509 mm 6
f
TP
6Re
3=5
TP
for D 0:790 mm 7
The friction factor for smooth tubes used in the homoge-
neous model (see Table 1) is also plotted for comparison.
The experimental two-phase friction factor is higher than
that for the single-phase smooth tubes. Actually, the two-
phase friction takes into account the friction between the
liquid and the wall as well as the friction between the liquid
and the vapor, the latter being much higher in case of
wavy-annular regime (see Fig. 5(a)). This dierence is thus
expected as the annular regime is the ow pattern present
in the turbulent zone.
There are two dierent relations for f
TP
, one for each
tube. This could be explained by the fact that even if the
sight glass tubes come from the same manufacturer, there
could be a dierence between them; however, they are very
smooth (see Table 4) as can be expected with good optical
quality glass and hence surface roughness can be excluded
as inuencing the results. More likely, it is the buoyancy
eect that is responsible, which is greater for the 0.790 mm
tube than for the 0.509 mm tube as shown in Fig. 14(a)
(c) and this eect explains the dierence in the results. Buoy-
ancy eects for R-245fa are the same as for R-134a in the
0.509 mm glass tube as shown in Fig. 14(d). Supporting this
conclusion, the data for R-134a and R-245fa in the
0.509 mm glass tube are well grouped together and are well
predicted by Eq. (6) whereas those for the 0.790 mmtube are
well predicted by Eq. (7).
Fig. 15 presents the comparison between the 1200 exper-
imental data points for the turbulent zone with the homo-
geneous model using a new two-phase friction factor
expressions. It can be seen that the data are very well pre-
dicted. The trends are captured and the data are well
grouped with 85.7% of the data falling in a 20% error
band and more than 96% within 30%. The transition zone
data cannot however currently be correctly predicted by
any of the existing methods and no method is proposed
here for this regime. In fact none of the literature methods
were specically developed to handle the transition regime,
so the lack of accuracy is not surprising. On the other
10
4
10
3
10
2
10
1
10
0
Twophase Reynolds number []
T
w
o

p
h
a
s
e

f
r
i
c
t
i
o
n

f
a
c
t
o
r

[

]
10
4
10
3
10
2
10
1
10
0
Twophase Reynolds number []
T
w
o

p
h
a
s
e

f
r
i
c
t
i
o
n

f
a
c
t
o
r

[

]
Experimental data
Fit curve
0.079Re
1/4
TP
Experimental data
Fit curve
0.079Re
1/4
TP
Fig. 13. Two-phase friction factor versus two-phase Reynolds number
for Re
TP
P8000. (a) 768 data points for R-134a, R-245fa, D = 0.509 mm.
(b) 432 data points for R-134a, D = 0.790 mm.
682 R. Revellin, J.R. Thome / Experimental Thermal and Fluid Science 31 (2007) 673685
hand, most literature methods eliminate the transition
regime in the two-phase analysis, jumping directly from
laminar to turbulent (see for example Friedel [2]). Fig. 11
provides a very good example of why this is not at all
appropriate.
The new prediction method has also been compared
with data from the literature. Cavallini et al. [16] have
performed two-phase frictional pressure gradient tests for
R-134a inside multi-port minichannels. The test section
consisted of eleven parallel rectangular cross-section
Fig. 14. Elongated bubble in 2.0, 0.8 and 0.5 mm tubes. (a) 2.0 mm tube for R-134a. (b) 0.8 mm tube for R-134a. (c) 0.5 mm tube for R-134a. (d) 0.5 mm
tube for R-245fa.
0 5 10 15
0
5
10
15
Experimental pressure drop [bar/m]
P
r
e
s
s
u
r
e

d
r
o
p

f
r
o
m

t
h
e

m
o
d
e
l

[
b
a
r
/
m
]
89.1 % of the values within 20 %
Experimental data
20 % error band
0 5 10 15
0
5
10
15
Experimental pressure drop [bar/m]
P
r
e
s
s
u
r
e

d
r
o
p

f
r
o
m

t
h
e

m
o
d
e
l

[
b
a
r
/
m
]
81.7 % of the values within 20 %
Experimental data
20 % error band
0 5 10 15
0
5
10
15
Experimental pressure drop [bar/m]
P
r
e
s
s
u
r
e

d
r
o
p

f
r
o
m

t
h
e

m
o
d
e
l

[
b
a
r
/
m
]
75 % of the values within 20 %
Experimental data
20 % error band
0 5 10 15
0
5
10
15
Experimental pressure drop [bar/m]
P
r
e
s
s
u
r
e

d
r
o
p

f
r
o
m

t
h
e

m
o
d
e
l

[
b
a
r
/
m
]85.7 % of the values within 20 %
Experimental data
20 % error band
Fig. 15. Comparison between the experimental data of two-phase frictional pressure drop and the homogeneous model using a new two-phase friction
factors for Re
TP
P8000. (a) R-134a and D = 0.509 mm. (b) R-134a and D = 0.790 mm. (c) R-245fa and D = 0.509 mm. (d) R-134a, R-245fa,
D = 0.509 mm and D = 0.790 mm.
R. Revellin, J.R. Thome / Experimental Thermal and Fluid Science 31 (2007) 673685 683
channels with a hydraulic diameter of 1.4 mm. Zhang and
Webb [8] measured adiabatic two-phase pressure drop for
R-134a owing in a multi-port extruded aluminum tube
with hydraulic diameter of 2.13 mm. Tran et al. [9] carried
out two-phase ow pressure drop measurements for
R-134a in a 2.46 mm diameter tube. The test sections used
in these three studies have a hydraulic diameter in a so-
called transition zone, i.e. minichannels and not microchan-
nels. As a result, these data have been compared with the
prediction method detailed in Eq. (7) as it has been devel-
oped for a transition diameter. Eq. (6) cannot be tested
due to the lack of data in the literature for refrigerants ow-
ing in microchannels. Fig. 16 shows the comparison
between the prediction model and the data for Re
TP
P8000
(l
TP
calculated with the McAdams relation). The data are
well grouped with 70.3% of the data falling in a 30% error
band and an absolute error of 26%. The same comparison
with the conventional homogeneous model would give
16.5%of the data falling in a 30%error band and an abso-
lute error of 47%. As a result, the new prediction methods
are validated.
8. Conclusions
In this study, 2210 experimental two-phase frictional
pressure drop data points were taken in 0.790 and
0.509 mm adiabatic glass tubes for R-134a and R-245fa
for a wide range of test conditions. The following conclu-
sions can be made here:
Similar to the classic Moody diagram in single-phase
ow, three zones were distinguishable when plotting
the variation of the two-phase friction factor versus
the two-phase Reynolds number: a laminar zone for
Re
TP
< 2000 (74 data points), a transition zone for
2000 6 Re
TP
< 8000 (936 data points) and a turbulent
region for Re
TP
P8000 (1200 data points).
The laminar zone is actually not taken into account as
the gradient for the pressure drop versus vapor quality
is negative. This zone corresponds to small instabilities.
The transition region is not predicted well by any of the
prediction methods. In the literature, only a few data are
available for this region and most methods ignore this
region, jumping directly from laminar to turbulent ow
at Re
TP
= 2000.
The turbulent zone can be reasonably well predicted by
the Mu ller-Steinhagen correlation.
New prediction methods have been proposed here for
both size of tubes using the homogeneous model and
determining a new two-phase friction factor f
TP
for the
two dierent tubes. More than 85.7% of the data fall
in a 20% error band and more than 96% within
30%. The trend is captured for both diameters and
both uids. The new prediction methods have also been
successfully validated with data from literature.
In summary, none of the methods are able to capture
the laminar, transition and turbulent trends in the present
two-phase friction factors shown in Fig. 11, not even the
LockhartMartinelli method with its dierent C values
for laminarlaminar, turbulentlaminar, laminarturbu-
lent and turbulentturbulent combinations in the liquid
and vapor phases. Since most methods either ignore the
transition regime and laminar regime all together, or jump
directly from a laminar to turbulent formulation, it is not
surprising that databases containing transition regime data
are poorly predicted (as these data are not usually segre-
gated from the database as was done here). The Garimella
et al. model is compared with experimental pressure drop
for intermittent ow. Most of the data fall in the trouble-
some transition region and are poorly predicted. Proposing
a new prediction method that captures all these three
regimes requires more laminar data and more accurate
measurements at low pressure drops to do this.
Acknowledgement
R. Revellin is supported by the European Communitys
Human Potential Programme under contract HPRN-
CT-2002-00204 [HMTMIC] funded by OFES, Bern,
Switzerland.
References
[1] R.C. Lockhart, R.W. Martinelli, Proposed correlation of data for
isothermal two-phase, two component ow in pipes, Chem. Eng.
Progr. 45 (1949) 3948.
[2] L. Friedel, Improved friction pressure drop correlations for horizon-
tal and vertical two-phase pipe ow, in: European Two-Phase Flow
Group Meeting, Ispra, Italy, 1979, paper E2.
[3] D. Chisholm, Pressure gradients due to friction during the ow of
evaporating two-phase mixtures in smooth tubes and channels, Int. J.
Heat Mass Transfer 16 (1973) 347348.
[4] H. Mu ller-Steinhagen, K. Heck, A simple friction pressure drop
correlation for two-phase ow pipes, Cem. Eng. Process. 20 (1986)
297308.
10
2
10
1
10
0
10
2
10
1
10
0
Experimental pressure drop [bar/m]
P
r
e
s
s
u
r
e

d
r
o
p

f
r
o
m

t
h
e

m
o
d
e
l

[
b
a
r
/
m
]
70.3 % of the values within 30 %
Cavallini et al.
Zhang & Webb
Tran et al.
30 % error band
Fig. 16. Comparison between the new prediction method (Eq. (7)) and
data from the literature for Re
TP
P8000.
684 R. Revellin, J.R. Thome / Experimental Thermal and Fluid Science 31 (2007) 673685
[5] K. Mishima, T. Hibiki, Some characteristics of airwater ow in
small diameter vertical tubes, Int. J. Multiphase Flow 22 (1996) 703
712.
[6] H.J. Lee, S.Y. Lee, Pressure drop correlations for two-phase ow
within horizontal rectangular channels with small heights, Int. J.
Multiphase Flow 27 (2001) 783796.
[7] J. Lee, I. Mudawar, Two-phase ow in high heat ux microchannel
heat sink for refrigeration cooling applications: Part Ipressure drop
characteristics, Int. J. Heat Mass Transfer 48 (2005) 928940.
[8] M. Zhang, R.L. Webb, Correlation of two-phase friction for refriger-
ants in small-diameter tubes, Exp. Therm. Fluid Sci. 25 (2001) 131139.
[9] T.N. Tran, M.C. Chyu, M.W. Wambsganss, D.M. France, Two-
phase pressure drop of refrigerants during ow boiling in small
channels: an experimental investigation and correlation development,
Int. J. Multiphase Flow 26 (2000) 17391754.
[10] S. Garimella, J.D. Killion, J.W. Coleman, An experimental validated
model for two-phase pressure drop in the intermittent ow regime for
circular channel, J. Fluid Eng. 124 (2002) 205214.
[11] R. Revellin, Experimental two-phase uid ow in microchannels,
Ph.D. thesis, Ecole polytechnique Federale de Lausanne, 2005.
Available from: <http://library.epn.ch/en/theses/?nr=3437>.
[12] R. Revellin, V. Dupont, T. Ursenbacher, J.R. Thome, I. Zun,
Characterization of diabatic two-phase ows in microchannels: Flow
parameter results for R-134a in a 0.5 mm channel, Int. J. Multiphase
Flow, 32, 755774, in press.
[13] W.H. McAdams, W.K. Woods, R.L. Bryan, Vaporization inside
horizontal tubesIIbenzeneoil mixtures, Trans. ASME 64 (1942)
193.
[14] G. Ribatski, L. Wojtan, J. Thome, An analysis of experimental data
and prediction methods for two-phase frictional pressure drop and
ow boiling heat transfer in microscale channels, Exp. Therm. Fluid
Sci., in press, doi:10.1016/j.expthermusci.2006.01.006.
[15] B. Agostini, A. Bontemps, Vertical ow boiling of refrigerant R-134a
in small channels, Int. J. Heat Fluid Flow 26 (2005) 296306.
[16] A. Cavallini, D.D. Col, L. Doretti, M. Matkovic, L. Rossetto, C.
Zilio, Two-phase frictional pressure gradient of R-236ea, R-134a and
R-410a inside multi-port minichannels, Exp. Therm. Fluid Sci. 29
(2005) 861870.
[17] A. Chicchitti, C. Lombardi, M. Silvestri, G. Soldaini, R. Zavattarelli,
Two-phase cooling experiments-pressure drop, heat transfer and
burnout measurements, Energ. Nucl. 7 (6) (1960) 407425.
[18] A.E. Dukler, M. Wicks, R.G. Cleveland, Pressure drop and hold-up
in two-phase ow part Aa comparison of existing correlations and
part Ban approach through similarity analysis, AIChE J. 10 (1)
(1964) 3851.
R. Revellin, J.R. Thome / Experimental Thermal and Fluid Science 31 (2007) 673685 685

Вам также может понравиться