Heat Transfer Engineering , 23:523, 2002

C 2002 Taylor & Francis

Copyright
01457632/02 $12.00 + .00

Two-Phase Flow Patterns,

Pressure Drop, and Heat
Transfer during Boiling in
Minichannel Flow Passages
of Compact Evaporators
SATISH G. KANDLIKAR
Mechanical Engineering Department, Rochester Institute of Technology, Rochester, New York, USA

The small hydraulic diameters employed during ow boiling in compact evaporator passages are
becoming more important in diverse applications including electronics cooling and fuel cell
evaporators. The high pressure drop characteristics of these passages are particularly important as
they alter the ow and heat transfer, especially in parallel multichannel con gurations. The pressure
drop oscillations often introduce dryout in some passages while their neighboring passages operate
under single-phase mode. This article presents a comprehensive review of literature on evaporation
in small-diameter passages along with some results obtained by the author for water evaporating in
1-mm-hydraulic-diameter multichannel passages. Critical heat ux is not covered in this article due
to space constraints.

Major progress in compact evaporator development

has been made by the automotive, aerospace, and cryogenic industries over the last 50 years. The thermal duty
and the energy ef ciency increased during this period,
while the space constraints became more vital. The
trend was toward greater heat transfer rates per unit
volume. The hot side of the evaporators in these applications was generally air, gas, or a condensing vapor.
The air-side n geometry also saw signi cant improvements resulting from increased heat transfer coef cients
as well as greater surface area densities.
As the air-side heat transfer resistance decreased,
more aggressive n designs were employed on the evaporating side, resulting in narrower ow passages. The

narrow refrigerant channels with large aspect ratio were

brazed in small cross-ribbed sections to provide a better
refrigerant distributio n along the width of the channel.
The major changes in recent evaporator designs involve
individual, small-hydraulic-diamete r ow passages, arranged in multichannel con guration for the evaporating uid.
Figure 1 shows a plate- n evaporator geometry widely used in compact refrigerant evaporators. The
refrigerant-side passages are made from two plates
brazed together, and air-side ns are placed between
two refrigerant ow passages. The plates have cross ribs
that are oriented in opposite directions so that the top
and bottom plates forming a refrigerant passage have
contact points only at the intersections of these ridges.
The two-phase ow of refrigerant is distributed across
the width of the ow passage. This feature is important

Address correspondenc e to Dr. Satish G. Kandlikar, Mechanical

Engineering Department, Rochester Institute of Technology, Rochester, NY
14623, USA. E-mail: [email protected]

Figure 1

Schematic of refrigerant and air-side ow passages in a compact plate- n type of evaporator.

in that the localized ow oscillations , caused by nucleate boiling and expanding vapor bubbles, are dissipate d
across the width and do not affect the upstream ow.
Figure 2 shows two geometries being used more
widely in compact evaporators. The parallel minichannel geometry shown in Figure 2a is used extensively in
condensation applications , whereas the geometry
shown in Figure 2b has received quite a bit of attention
for boiling applications . The channels are fabricated
by a variety of processes depending on the dimensions
and plate material. Conventional machining and electrical discharge machining are two typical options, while
semiconductor fabrication processes are appropriate for
microchannel fabrication in chip cooling applications .

FLOW PATTERNS IN SMALL-DIAMETER

TUBES
Previous Studies
Visualization of boiling phenomena inside ow
channels provides insight into the heat transfer mechanisms. Bubble formation, coalescence, formation of
slugs or plugs, and local dryout conditions are all

Figure 2

Recent developments in multichannel evaporators.

heat transfer engineering

important in understanding the heat transfer phenomena. Although some of the heat transfer and pressure
drop equations employed in the design of commercial
equipment are derived from ow pattern-based models, the major bene t of ow pattern information lies in
understandin g the causes for premature dryout or critical heat ux (CHF) condition in an evaporator. Another
major bene t is in the design of the inlet and the outlet
manifolds in multichannel evaporators.
The ow pattern maps available in the literature
were rst developed for the petrochemical industry
(Baker, [1] ) for ow of oil and gas in large-diameter
pipes. Subsequently, adiabatic ow pattern maps were
developed as general ow pattern maps (for example,
Hewitt and Roberts [2], and Taitel et al. [3] ). In recent
years, a number of ow pattern maps have been developed for speci c conditions such as small-diameter
tubes, evaporation or condensation , and compact heat
exchanger geometries.
Earlier investigator s extensively studied ow patterns for gas liquid ows in channels with small hydraulic diameters. A representative survey of the ow
patterns was presented by Fukano et al. [4]. They identi ed bubbly, plug, slug, and annular ow patterns and
compared the ow pattern transitions with the available ow pattern maps. Subsequently, a detailed study
by Wambsganss et al. [5] provided a more comprehensive summary and representation of gas liquid ow
patterns. The role of surface tension becomes more important in smaller-diameter channels. Triplett et al. [6]
explain that, due to the dominance of surface tension,
strati ed ow is essentially absent, slug (plug ) and churn
ow patterns occur over extensive ranges of parameters,
and the slip velocity under these patterns is small. Strati ed ow can exist at very low ow rates, as observed by
Kasza et al. [7] for a mass ux of 21 kg/m2 s. Hewitt [8]
gives a comprehensive summary of ow pattern studies
available in the literature. For large-diameter tubes, the
generalized ow pattern map for an air water system,
developed by Mandhane et al. [9], was quite representative of other ow conditions as well. However, the
vol. 23 no. 1 2002

theoretically based transition criteria presented by Taitel

and Dukler [10] are among the most widely used ow
pattern maps. Hewitt [8] notes that both evaporation and
condensation processes have a signi cant effect on the
ow patterns.
The effect of evaporation on the ow pattern transitions was considered to be quite small in large-diameter
tubes. This is one of the reasons why the ow pattern
studies from gas liquid systems, such as air and water,
were extended to the case of evaporation. In smallerdiameter tubes, the effect of evaporation could be quite
dramatic.
The evaporation of the liquid phase affects the ow
in two ways. First, it alters the pressure drop characteristics by introducing an acceleration pressure drop
component that can be quite large at higher heat uxes.
Second, the tube dimension is quite small, and the effect of surface tension forces becomes more important
in de ning the two-phase structure.
Table 1 includes the ow pattern studies for circular
tubes and narrow rectangular channels with hydraulic
diameters of the order of 3 mm or smaller. Cornwell and
Kew [12] conducted experiments with R-113 owing in
1.2-mm 0.9-mm rectangular channels. The parallel
channels were machined in aluminum and a 6-mm-thick
glass plate was used to observe the ow pattern. In the
ow ranges investigated , three ow patterns were observed as shown in Figure 3. These authors also observed a strong link between the ow pattern and the
heat transfer coef cient. In the isolated bubble region,
h / q 000:7 , indicating the dominance of nucleate boiling. When the bubbles occupied the entire channel cross
section in the con ned bubble region, h dependence
on q decreased. The convection effects were dominant
in the annular-slug region. In a subsequent study, Lin
et al. [33] compared the ow region transitions with
those predicted by Barnea et al. [34], developed for
an evaporating steam water system in 4-mm-diameter
tubes, and Mishima and Hibiki [35], developed one for
an adiabatic air water system in a 2.05-mm-diameter
tube. The results indicate that the ow pattern maps
developed for air water ow are in general applicable,
but the transition boundaries need to be re ned for ow
boiling in small-diameter tubes and channels.

Figure 3 Flow patterns observed by Cornwell and Kew [12] in

1.2-mm 0.9-mm parrallel rectangular channels.

heat transfer engineering

Moriyama and Inoue [13] conducted experiments in

single narrow rectangular channels with R-113. They
observed (1 ) attened bubbles, con ned by the ow
channel (some of them coalescing ), (2 ) liquid strips
owing along the wall, and (3 ) liquid lm ow.
Mertz et al. [18] investigated single- and multichannel test sections with water and R-141b owing in rectangular multichannels , 1, 2, and 3 mm wide. They observed the presence of nucleate boiling, con ned bubble
ow, and annular ow, and discovered that the bubble
generation in the channels was not a continuous process.
In addition, the vapor seemed to stay in the channels,
blocking the ow and, in some cases, causing a reverse
ow to occur in the channels. In both single-channel
and multichannel cases, large pressure pulsations were
noticed.
Kasza et al. [7] presented a detailed study on ow visualization of nucleation activity in a rectangular ow
channel of 2.5 mm 6 mm cross section. They viewed
the ow using a high-speed video camera with a maximum frame rate of 12,000 frames per second. The mass
ux was 21 kg/m2 s. They observed nucleate boiling on
the wall, as the individual bubbles nucleated and grew
on the wall. The bubble growth rates were similar to
those in the pool boiling case when the bubbles grew
without interacting with the wall or any liquid vapor
interface.
Kasza et al. [7] made interesting observations on individual bubbles and their interaction with other bubbles and vapor slugs. The vapor slugs were separated
from the wall by a thin liquid lm with a 0.67-mm
average thickness. Nucleation was observed in this liquid lm. The bubbles growing in this lm did not easily coalesce with the vapor in the slug. Bubbles growing under the liquid slug were attened and covered
a larger wall area compared to those growing in the
liquid ow. The heat transfer in the thin liquid microlayer underneath the bubbles improved the heat transfer; the bubble frequency and the vapor volume both
increased for such bubbles. Their ndings clearly indicate the occurrence of nucleate boiling in thin liquid lms that exist in both slug ow and annular ow
conditions.
Bonjour and Lallemand [22] report ow patterns of
R-113 boiling in a narrow space between two vertical surfaces. The ow patterns observed are similar to
those observed by other investigators : isolated bubbles,
coalesced bubbles, and partial dryout. Comparing the
bubble dimensions with the channel size is vital in determining whether the small channel size in uences the
bubble growth and leads to the con ned bubble ow
pattern. Following the work of Yao and Chang [36],
the Bond number, D e[r =g (q L q G )]1=2 , along with
the Boiling number, D q =( Gh f g ), provided the basis for
vol. 23 no. 1 2002

R-113,
G D 200 1,000,
q D 4 30

R-113,
G D 50 100,
q D 8.8 90.7
R-113,
q D 1,000 2,000,
0.28 1.1 ml/s

Moriyama
and Inoue
[13], 1992

Wambsganss
et al. [14],
1993
Bowers and
Mudawar
[15], 1994;
[16], 1994;
[17], 1994
Mertz et al.
[18], 1996

Water, G D 21,
q D 110

Water, G D 25,000
45,000, CHF of
50 80 MW/m2

Kasza et al.
[7], 1997

Tong et al.
[21], 1997

Ravigururajan
et al. [19], 1996
Tran et al.
[20], 1996

R-113,
G D 124 627,
q D 3 33

Cornwell
and Kew
[12], 1992

Water and R-141b,

G D 50, 100, 200,
300, q D 3 227
R-124, 0.6 5 ml/s,
20 400 W
R-12,
G D 44 832,
q D 3.6 129

R-113,
G D 125 750,
q D 14 380

Lazarek and
Black [11],
1982

Fluid and
ranges of
parameters
G (kg/m2 s),
q (kW/m2 )

Circular,
D D 1.05 2.44

Rectangular,
2.5 6.0 500

Rectangular channels,
1, 2, and 3 mm wide,
aspect ratio up to 3
270 l m wide, 1 mm
deep, 20.52 mm long
Circular, D D 2.46;
rectangular,
D h D 2.4

Mini- and
microchannels,
D D 2.54 and
0.51

Parallel rectangular;
75 channels,
1.2 0.9 deep;
36 channels,
3.25 1.1 deep
Narrow rectangular
channel, 0.035 0.11
gap, width D 30,
L D 265
Circular,
D D 2.92 mm

Circular,
D D 3.1, L D 123
and 246

Channel size/
D (mm),
horizontal (unless
otherwise stated)

Subcooled
ow boiling

Not reported

Heat transfer
coeff. and
heat ux
Heat transfer
coeff.
Heat transfer
coeff.

h as a
function of
x, G, and q
Heat transfer
rate

Heat transfer
coeff.

Heat transfer
coeff.

Heat transfer
coeff. and
CHF

Heat transfer

Summary of investigations on evaporation in mini- and microchannels

Author/year

Table 1

Focus on
pressure drop

Not reported

Not studied

Bubbly, Slug

Not studied

Not studied

Not studied

Nucleate boiling,
con ned bubble
and annular
Not studied

Not studied

Flattened bubbles,
w/coalescence,
liquid strips,
liquid lm
Not reported

Isolated bubble,
con ned bubble,
annular-slug

Not observed

Flow patterns

Not measured

Pressure drop
components

Pressure drop
measured and
components
calculated
Not reported

Not reported

Pressure drop
measured and
correlated

Pressure drop

Single- and multichannel test sections. Flow boiling

pulsation was observed in multichannels, with
reverse ow in some cases. Nucleate boiling dominant.
Experiments were conducted over 0 0.9 quality and
5 C inlet subcooling. Wall superheat 0 80 C.
Local heat transfer coeff. obtained upto x D 0.94.
Heat transfer in nucleate boiling-dominant and
convective dominant regions obtained. New
correlation for nucleate boiling-dominant region.
Increased bubble activity on wall at nucleation
sites in the thin liquid lm responsible for high
heat transfer in small channels.
Pressure drop measured in highly subcooled ow
boiling, correlations presented for both
single-phase and two-phase ow.

Subcooled and saturated data obtained, h almost

constant in the two-phase region, dependent
on heat ux. Behavior similar to large-dia.
tubescombination of nucleate and convective
boiling. Pressure drop correlations obtained.
The heat transfer coef cient was found to be
dependent on the existing ow pattern. In the
isolated bubble region, hq 0:7 , lower q
effect in con ned bubble region, convection
dominant in annular-slug region.
Two-phase ow boiling data in narrow gaps
obtained and correlated with an annular lm
ow model. Nucleate boiling ignored,
although a dependence of h on q was observed.
Except at the lowest heat and mass uxes, both
nucleate boiling and convective boiling
components were present.
Analytical and experimental studies comparing
the performance of mini- and microchannels.
Minichannels are preferable unless liquid
inventory or weight constraints are severe.

Remarks

Rectangular,
1 1 60 mm

Kandlikar
et al. [32], 2001

Water,
G D 80
560 kg/m2 s

Rectangular,
1 20 357

Lakshminarasimhan R-11,
et al. [31], 2000
G D 60 4,586

Kennedy et al.
[30], 2000

Hestroni et at.
[29], 2000

Subcooled
and saturated
ow boiling

Subcooled and
saturated
ow boiling

Onset of nucleate
boiling starts
instability

Measured,
but data
not reported

Not studied

Periodic annular

Not studied

Not studied

Con ned bubble,

cell, annular

Not studied

Not studied

3 ow patterns
with nucleate
boiling
Not observed

Nucleate boiling with isolated bubbles,

nucleate boiling with coalesced bubbles and
partial dryout, criteria proposed for transitions.
The heat transfer results indicate that both nucleate
and forced-convection boiling are present. No
bubbles were observed, and the authors propose a
ctitious boiling. The authors did not use proper
microscope and high-speed video techniques,
resulting in contradictory conclusions.
Bubble nucleation model uses a vapor bubble
growing on a heater surface with heat diffusion
in the vapor phase. This corresponds to post-CHF
heat transfer and is an inaccurate model of heat
transfer during nucleation and bubble growth.
Extremely high heat transfer coef cients up to
11 kW/m2 C were observed. Fully developed
subcooled boiling and CHF were obtained.
Capillary forces important in ow patterns,
thin lm suppresses nucleation, leads to
convective boiling.
Heat transfer coeff. obtained as a function
of quality and heat ux. Trends are
similar to large-tube data.
As the helical coil radius became smaller,
pressure drop reducedpossibly
due to rearrangement in ow pattern.

Periodic annular ow observed in

microchannels. There is a signi cant
enhancement of heat transfer during
ow boiling in microchannels.
Not reported
Not studied
Heat ux at the onset of ow instability was
0.9 of the heat ux required for saturated
vapour at exit, similarly, G at OFI was 1.1
times G for saturated exit vapor condition.
Not measured
Boiling incipience
Boiling front observed in laminar ow, not
observed
visible in turbulent ow due to comparable
through LCD
h before and after, saturated ow boiling data
correlated by Kandlikar (1990) correlation.
Measured
High-speed photography Flow oscillations and ow reversal linked to the
instantaneous values on single and
severe pressure drop uctuations, often
of pressure drop
multiple channels
leading to ow reversal during boiling.

Not reported

Single- and
Two-phase
ow

Not studied

Heat transfer
coeff.

Circular, D D 1

Not studied

Single- and
Not reported
two-phase, max
h 11 kW/m2 C
h 1 20 kW/m2 C Not studied

Bubble nucleation
model

Not reported

Not studied

Circular,
D D 1.59, 2.78,
3.97, 4.62
Annulus,
0.9 gap 500

Theoretical

Rectangular, 0.5 2 mm Not studied

gap, 60 wide,
120 long, vertical
Rectangular,
Heat transfer
width D 0.2 0.4,
coeff., heat
height D 0.1 0.3,
ux
L D 50; triangular,
Dh D 0.2 0.6, L D 120

Circular multichannels
in helical coils,
Dh D 0.23 1.86,
helix dia. D 2.8 7.9
Water,
Triangular parallel
Re D 20 70,
chann, D 55 ,
q D 80 360
n D 21, 26, Dh D
0.129 0.103, L D 15
Water,
Circular,
G D 800 4,500,
D D 1.17 and
q D 0 4 MW/m2
1.45, L D 160

R-113, power
25 700 W,
Re D 190 1,250
R-318C,
G D 200 900,
q D 2 110
R-141b,
G D 300 2,000,
q D 10 150
R-113, ranges not
clearly stated

Kamidis and
Ravigururajan
[25], 1999
Kuznetsov
and Shamirzaev
[26], 1999
Lin et al.
[27], 1999

Dowing et al.
[28], 2000

Theoretical

Water, ethanol
and mixtures

R-113,
q D 0 20

Peng et al.
[24], 1998

Bonjour and
Lallemand
[22], 1998
Peng and
Wang [23],
1998

the transition from individual to con ned bubble ow.

For a smaller Bond number, the channel dimension is
smaller than the departure bubble diameter, resulting in
con ned bubble ow pattern. For large Bond numbers,
the channel size does not interfere with the bubble ow.
A more detailed treatment of the forces acting on the
bubble is needed to identify this boundary clearly.
Kuznetsov and Shamirzaev [26] studied the ow patterns during ow boiling of R318C in a 0.9-mm annular
gap between two circular tubes. This work was an extension of previous research on air water systems by
Kuznetsov and Vitovsky [37]. The isolated bubble region was called the small bubble region. It was followed
by long Taylors bubbles, similar to con ned bubbles
(de ned by earlier investigator s ), which were elongated
in the ow direction. These bubbles spread along the periphery of the annular gap and formed a cell structure
referred to as the cell ow regime. As the vapor quality
increased, the annular ow pattern was established with
a ripple of waves.
Hestroni et al. [29] studied the evaporation of
water in multichannel evaporators. The evaporators
consisted of 21 26 parallel ow passages with channel hydraulic diameters of 0.103 0.129 mm. Hestroni
et al. observed periodic behavior of the ow patterns
in these channels. The ow changed from single-phase
ow to annular ow with some cases of dryout. The
dryout, however, did not result in a sharp increase in
the wall temperature. This clearly indicates that there is
still some evaporating liquid lm on the channel walls
that could not be observed in the video images. These
authors also reported the presence of vapor phase in the
inlet plenum.
Lakshminarasimhan et al. [31] studied the ow boiling in a narrow rectangular channel, 1 mm high
20 mm wide 357 mm long. They used liquid crystal
display (LCD ) on the heated wall to observe the nucleation front and locate the onset of nucleate boiling
(ONB ). The ONB was clearly identi ed in the laminar ow region with subcooled R-11 entering the ow
channel. As the ow rate increased to the turbulent
region, the ONB could not be identi ed through this
technique due to the high heat transfer coef cient in
the subcooled single-phase region. However, it is possible that the ONB may have occurred at discrete locations rather than appearing as a clear identi able
front.

Discussion on Flow Patterns and Flow Pattern

Transitions in Small-Diameter Channels
The three ow patterns shown earlier in Figure 1 represent the characteristic ow patterns associated with
10

heat transfer engineering

two-phase ow in minichannels. It is clear that discrete

bubbles, resulting from nucleation activity on the wall,
are present in the subcooled boiling and low-quality regions. The presence of nucleation in the small-diameter
tubes is also evident through the available studies. The
observations of nucleating bubbles in the thin liquid
lms by Kasza et al. [7], shown in Figure 4, are also
very revealing.
The experimental conditions employed by Kasza
et al. represent low-Reynolds-numbe r conditions . With
their hydraulic diameter of 3.53 mm and a mass ux of
21 kg/m 2 s, the single-phase Reynolds number is 262
at 1 atm pressure. The shear stress effects under these
conditions are expected to be quite low, and it should
not come as a surprise that the growth rate exponent
for nucleating bubbles re ects that under pool boiling
conditions. In general, most of the visual studies are for
low ow rates, as the bubble activity cannot be easily
traced at higher mass uxes.
Kandlikar and Stumm [38], and Kandlikar and
Spiesman [39], demonstrated the presence of nucleating bubbles under highly sheared ow conditions in
a rectangular channel. It was noted that bubble nucleation occurred when the nucleation criterion for available cavity sizes was satis ed. The effect of the ow
and wall temperature on bubble characteristics was illustrated by Kandlikar [40] and is shown in Figures 5
and 6.
Figure 5 shows the effect of ow Reynolds number on
the bubble growth rate. As the ow velocity increased,
ow changed from laminar to transition region with an
associated increase in the single-phase heat transfer coef cient. This caused the bubble to grow much faster,
reaching a departure condition in about 25 ms at Re D
2,280, as opposed to 300C ms at Re D 1,267. The departure bubble diameter is also in uenced by the ow.
As the ow velocity increases, the drag forces cause the
bubbles to depart at smaller diameters.
Figure 6 further illustrates the sensitivity of the wall
temperature conditions on the bubble growth rates. A
higher wall temperature, with a greater associated heat
ux, causes the bubbles to grow rapidly and reach the
departure conditions much sooner. The departure sizes
appear to be more dependent on the ow velocity for
the conditions depicted in these gures.
From the observations above, it may be concluded
that the nucleating bubbles are present in ow boiling
under high shear conditions . Kasza et al. [7] observed
such bubbles in thin lms, shown in Figure 4b, con rming that nucleation can occur in annular ow as well.
Under these conditions, the nucleating bubble size decreases, and bubble departure frequency increases. This
further con rms the conclusions made by Kandlikar
and Stumm [38] that a high speed camera with suitable
vol. 23 no. 1 2002

Figure 4

Flow patterns and bubble nucleation in the liquid lm observed by Kasza et al. [7].

magni cation is needed to observe the nucleating bubbles (1 ) under high shear stress conditions , and (2 ) at
high wall temperatures. In fact, the use of high-speed
photography is essential in clearly observing the ow
patterns in small-diameter tubes.
The bubbles departing in the ow can exist as individual bubbles unless their size is smaller than the
channel dimension normal to the nucleating surface.
Further growth of these bubbles results in their con-

Figure 5 Effect of ow on bubble growth, subcooled ow of water

at 1 atm pressure in 3-mm 40-mm rectangular channel, Twall D
108 C, Tbulk D 80 C, cavity radius 3.2 l m, Kandlikar [40].

heat transfer engineering

nement by the channel walls under the con ned ow

pattern. In reality, the con ned ow pattern is similar
to the early stages of the plug ow pattern seen in the
conventional two-phase ow patterns for larger diameter tubes (>3 mm ). Annular ow pattern then follows
at higher qualities.
Flow Instabilitie s
The ow instabilit y is another concern in the design
of evaporators that employ small channels. The ow
instabilitie s can be reduced considerably by increasing
the upstream pressure drop prior to the ow entering a
channel. A large-diameter ow section, however, generally precedes the test section in a number of experimental studies conducted on gas liquid ows in smalldiameter tubes. In the study conducted by Lin et al. [33],
air was introduced in a large mixing chamber upstream
of the test section to reduce the disturbances resulting
from gas injection. The presence of such low-pressuredrop regions immediately before the test section leads
to signi cant ow instabilitie s that cause large pressure
drop excursions, and occasionally result in a negative
pressure drop with a corresponding ow reversal in the
channel.
vol. 23 no. 1 2002

11

Figure 6 Effect of wall temperatures on bubble growth, subcooled ow of water at 1 atm pressure in 3-mm 40-mm rectangular channel,
Twall D 108 C, Tbulk D 80 C, cavity radius 3.2 l m, Kandlikar [40].

The instabilit y occurs in the negative-pressure-dro p

region of the demand curve plotted as the pressure drop
versus inlet ow velocity of the subcooled liquid.
Kennedy et al. [30] studied the onset of ow instability and noted that instability sets in at a slightly
lower mass ux than the onset of signi cant void
condition.
Cornwell and Kew [12] conducted experiments
on ow boiling of R-113 in parallel microchannels.
They observed that the ow was unstable at lower
ow rates. The pressure drop uctuations were not
reported.
Kandlikar et al. [32] viewed the ow boiling of water in electrically heated multichannel evaporators consisting of six parallel channels. The ow structure was
visualized using a high-speed video camera up to a
maximum speed of 1,000 frames per second. The typical bubbly ow, slug ow, and annular ow patterns
were observed. Nucleation was also observed in the
bulk liquid as well as in the liquid lm. However, the
most interesting discovery made, in an attempt to understand the severe pressure uctuations (described
later under Pressure Drop ), was a visual con rmation of complete ow reversal in some of the
channels.
Figure 7 shows the schematic of the multichanne l
evaporator investigate d by Kandlikar et al. [32]. The
evaporator was heated electrically from the back wall
of the test section. The front part was covered with a
high-temperature-resistan t glass for ow visualization .
Figure 8 shows the sequence of the events that lead
to ow reversal in the ow channels. Two adjacent cen12

heat transfer engineering

tral channels are shown at 2-ms intervals in frames (a )

through (e ). Both channels (1 ) and (2 ) exhibit slug ow
in the visible section.
Vertical lines (y ) and (z ) indicate the initial boundaries of a vapor slug in channel (2 ) of Figure 8a. Vertical
lines (x ) and (w ) are reference lines to aid visualizatio n
of slug motion throughout the frames.
In Figure 8b, channel (1 ) has ow in the direction
of bulk ow (left to right ). The vapor slug in channel (2 ) has expanded in the direction of bulk ow, yet
the inlet-side uid/vapor interface has not moved. The
uid/vapor interface on the inlet side of the slug in
channel (2 ) is still stationary in Figure 8c, although
the outlet-side interface moves in the bulk ow direction. In Figure 8d the ow in channel (1 ) moves along
the direction of bulk ow, but the inlet-side uid/vapor
interface of the slug in channel (2 ) progresses in the

Figure 7 Multichannel evaporator investigated by Kandlikar

et al. [32].

vol. 23 no. 1 2002

PRESSURE DROP IN SMALL-DIAMETER FLOW

BOILING CHANNELS
Single Channel
Pressure drop in small-diameter tubes has been studied by a number of investigators . Lazarek and Black
[11] conducted systematic experiments to evaluate the
three components of pressure drop. The desired quality was generated in the heated inlet section and the
frictional pressure drop was measured under adiabatic
conditions in the discharge section. The frictional pressure was correlated using a correlation recommended
by Collier [41] with the Martinelli parameter v tt :
pTP
C
1
D1C
C 2
v tt
v tt
pLO

Figure 8 Successive frames (a) through (e) at 2-ms intervals of

two channels interacting. G D 40 kg=m2 s, surface temperature D
110.6 C, entering water temperature D 24.7 C, outlet water temperature D 99.3 C, x > 0.

direction counter to the bulk ow. In Figure 8e, the

inlet uid/vapor interfaces in both channels move in
the direction counter to bulk ow. For this particular
case, it appears that both of the channels experience a
vapor-clogging condition where the differential pressure across the channels increases due to vapor generation, and the mass ow rate is consequently reduced
through these two channels. The reaction to this condition in the other four channels would be an increased
ow rate.
Concluding Remarks on Flow Patterns
The ow patterns observed in small channels indicate that the nucleating bubbles play an important
role in small-diameter channels. The three predominant ow patterns are (1 ) isolated bubbles, (2 ) con ned
bubbles, and (3 ) annular-slug. Note that the ow patterns under high mass ux conditions (G > 500 kg/m 2 s )
have not been studied in the literature due to dif culty in capturing the high-speed movement under
these conditions. Further work in this area is
recommended.
heat transfer engineering

(1 )

The subscript TP corresponds to the two-phase value,

while LO corresponds to the value with the total ow
in the liquid phase. Lazarek and Black found that using a value of C D 30 produces results that are in good
agreement with their experimental data. The value of C
recommended by Collier is 20 for large tubes.
The acceleration pressure drop was accurately predicted using the Martinelli and Nelsons [42] separated
ow model with a multiplier K sa :
"
psa
q
D K sa
2
G =(2q L )
q

2
xex

L
G

ex

2
x in

L
G

in

(1 x )2
1 a ex

(1 x in )2

1 a in

(2 )

Here x is the vapor quality and a is the void fraction.

K sa is an empirical constant. Lazarek and Black found
that a value of K D 2.5 correlated most of their data
within 20%.
Moriyama and Inoue [13] measured pressure drop of
R-113 boiling in narrow annular gaps of 35 110 l m.
Their experimental values for frictional pressure drop
were correlated by slightly modifying Eq. (1 ). From
their study, it is evident that the separated ow model
is applicable for the narrow gaps typically encountered
in microchannel applications .
On the other hand, Bowers and Mudawar [17] employed a homogeneous ow model with f TP D 0.003
as recommended by Collier [41]. Their results for both
minichannels and microchannels were correlated to
within 30% with this model.
vol. 23 no. 1 2002

13

Tong et al. [21] present an exhaustive treatment of

pressure drop during subcooled ow boiling in minichannels. In addition, they presented a correlation to
predict the two-phase pressure drop. Since the void
fraction was very small, a two-phase friction factor was
applied. They observed a roughness effect on the singlephase laminar-to-turbulen t transition in these tubes. The
effect of the tube diameter-to-lengt h ratio was also noted
to be quite signi cant.
From the studies available in the literature, the effect of channel dimension on two-phase pressure drop
is not clearly established. Although several investiga tors provide different correlation schemes to correlate
their data, they do not provide a clear indication of
the effect of small passage sizes on pressure drop. The
added effect of channel wall roughness on pressure
drop, seen in the single-phase data, is also not incorporated while analyzing the two-phase pressure drop
parameters. These effects will become more important
as the channel size decreases from minichannel to microchannel geometries.

HEAT TRANSFER IN SMALL-DIAMETER FLOW

BOILING CHANNELS

Multiple Channels
As the tube diameter decreases, vapor slugs ll the
tubes. Under two-phase ow conditions , ow instabilities occur when the pressure drop in the upstream
section is relatively small. Introducing a large pressure
drop through a throttle valve in the liquid line immediately prior to the test section considerably reduces the
instabilities . These instabilitie s have a signi cant effect
on pressure drop and heat transfer under ow boiling
conditions.

Figure 9 Differential pressure history for a six-channel (1-mm

1-mm) parallel con guration. G D 48 kg=m2 s, Pmax =L D 4,688
Pa/m, Pmin = L D 1,793 Pa/m, average surface temperature D
125 C, q 00 D 74:3 kW=m 2 , water inlet and outlet temperatures 25.0
and 90.2 C, Kandlikar et al. [32].

14

Figure 9 shows the pressure drop uctuations measured in a multichannel evaporator with six parallel
1-mm 1-mm square microchannels. Similar observations were made by Kew and Cornwell [43] during
ow boiling of R-141b in 2-mm square channels and in
2.87-mm-diameter circular tubes. The pressure drop
uctuations observed by Kandlikar et al. [32] are quite
large and result in ow reversals as discussed earlier
in the ow patterns section. The compressibility of the
two-phase mixture, in adjacent channels, acts in a manner similar to the negative slope in the upstream section
of a single evaporator tube. The large pressure drop
uctuations lead to instantaneou s localized ow reversal in some of the parallel channels. There are no models
currently available that predict the pressure drop uctuations and the ow reversals under ow boiling conditions. Knowledge of these conditions is essential for
safe operation of evaporators employing minichannels
and microchannels.

heat transfer engineering

Flow boiling in small-diameter tubes and compact

heat exchanger passages has been a subject of interest
in the automotive , aerospace, air liquefaction, chemical and petroleum industries, and in electronics cooling applications . Nakayama and Yabe [44], and Kew
and Cornwell [45] present a good overview of the recent advances in this area. Table 1 includes some of
the recent works on ow boiling heat transfer in
minichannels.
Flow boiling heat transfer in 1 3-mm-diameter channels has been a subject of investigatio n for a long time.
In one of the earlier articles, Bergles [46] studied the
critical heat ux in 0.584-, 1.194-, and 2.388-mmdiameter electrically heated tubes. He indicated that
when the bubble diameter approaches the tube diameter, considerable nonequilibriu m vapor volume exists
in the evaporator section, and ow oscillation s cause a
premature burnout in the small-diameter tubes.
Bowers and Mudawar [17] studied ow boiling pressure drop and CHF in a minichannel of 2.54 mm diameter, and a microchannel of 510 l m diameter. Boiling
curves for the two channels were obtained at nearly
equal values of liquid Reynolds number. Their results
are reproduced in Figure 10. Despite the large variation in the tube diameter, the two curves overlap in the
boiling region. It is believed that these experiments fall
under the fully developed nucleate boiling regime. The
differences between the two boiling curves are only evident at low heat ux (near-single-phas e region ) and
high heat ux values (approaching CHF condition ).
vol. 23 no. 1 2002

Figure 10 Flow boiling characteristics in minichannel and

microchannel evaporators, Bowers and Mudawar [15].

This indicates that in spite of the differences in the ow

characteristics of the channels, the ow boiling behavior is quite similar in the two geometries.
The detailed ow boiling data by Lazarek and Black
[11] provide a clear comparison between the ow boiling characteristics of minichannels and regular tubes
(>3 mm diameter ). Figure 11 shows a comparison of
Lazarek and Blacks ow boiling data in a 3.1-mmdiameter tube and Kandlikars [47] ow boiling
correlation. The correlation represents the data quite
well, although some differences exist in the high-qualit y
region. Although a detailed study is warranted, as a
rst-order approximation , one may use the correlations

Figure 11 Comparison of ow boiling heat transfer coef cient

data by Lazarek and Black [11] with Kandlikar [50] correlation.

heat transfer engineering

developed for the large-diameter tubes for predicting

the heat transfer coef cients in minichannels.
Cornwell and Kew [12] conducted experiments in
two sets of parallel channel geometries, 1.2 mm
0.9 mm deep, and 3.25 mm 1.1 mm deep. Their results
indicate that the ow boiling in such small channels
exhibits fully developed nucleate boiling characteristics in the isolated bubble region at lower qualities. At
higher qualities (when the bubbles ll the entire cross
section ), and in the annular ow region, convective effects dominate heat transfer. These characteristics are
similar to those observed for the large-diameter tubes.
The isolated and con ned bubble regions exhibit characteristics similar to the nucleate boiling-dominan t region, while the annular-slug region exhibits the convective dominant trend seen in large-diameter tubes
(Kandlikar, [47] ).
Another aspect of ow boiling heat transfer in small
channels is the oscillatory nature of the ow. The timeaveraged value between two regions (i.e., between the
con ned bubble and the annular regions ) would yield
a combination of nucleate boiling-dominan t and convective boiling-dominan t regions. Purely ow patternbased correlations need to include this averaging
effect. Since the large-diameter correlations combine
these effects, rather than using distinct boundaries, they
are expected to provide the basis for accurate
correlation schemes for small-diameter tubes and
channels.
Continuing the work in this area, Lin et al. [27] obtained ow boiling data with R-141b in 1-mm-diameter
tubes. Their results indicate that the heat transfer coef cient exhibits behavior similar to that reported by
Cornwell and Kew [12]. The role played by bubbles is
clearly an important one. Speci cally, they presented
detailed results at G D 500 kg/m 2 s with q from 18
to 72 kW/m2 . At high heat uxes, they observed considerable uctuations in the wall temperature readings,
indicative of ow oscillation s that cause changes in the
instantaneou s values of local saturation temperature and
heat transfer rate.
Wambsganss et al. [14] conducted extensive experiments on ow boiling of R-113 in a 2.92-mm-diameter
tube. Their results indicate that the heat transfer coef cient was sensitive to both heat ux and mass ux
changes, except for the lowest mass ux result. At the
lowest value, G D 50 kg/m 2 s, changing the heat ux
from 8 to 16 kW/m2 did not have any in uence on the
heat transfer coef cient. One possible explanation is
that for this case, the nucleate boiling is in the partial
boiling region at low heat fluxes, and the effect of heat
ux is therefore quite small. For their other test conditions, the mass ux was varied from 100 to 300 kg/m2 s,
and the heat ux was varied from 16 to 63 kW/m2 .
vol. 23 no. 1 2002

15

For these tests, the heat transfer coef cient exhibited a

dependence on both heat ux and mass ux. This indicates the contribution s from both nucleate boiling and
convective boiling mechanisms. They also compared
their data with the available correlations and found that
the correlations by Liu and Winterton [48], Shah [49],
and Kandlikar [50] predicted their results with a mean
deviation of less than 20%. In particular, Wambsganss
et al. found that the speci c correlation developed by
Lazarek and Black [11], who used their own smalldiameter-tube data with R-113 in the correlation development, predicted the data with a mean error of 12.7%.
The Chen [51] correlation predicted the results with a
mean deviation of 36%.
Mertz et al. [18] conducted extensive experiments
with water and R-141b boiling in six different minichannel con gurations. The ow boiling was observed as
oscillatory phenomena in multichannels . It is interesting to note that although the heat transfer coef cient
increased with heat ux in almost all cases for singlechannels, the trend for water owing in the multichanne l
con guration at G D 200 kg/m 2 C was different. For all
channels in multichannel con guration, the heat transfer coef cient decreased with increasing heat fluxes. It
is suspected that the ow oscillation s and reversals observed in multichannels are responsible for the degradation in heat transfer. For R-141b boiling in multichanne l
con guration, the heat ux effects were weak and somewhat mixed.
Mertz et al. [18] observed that the heat transfer coef cient for both uids in the multichannel con guration was considerably higher than the corresponding single channel values under the same operating
conditions. Figure 12 shows the comparison for water
(at 200 kg/m 2 s in 2-mm 4-mm channels ) in singleand multichannel con gurations under the same operating conditions. The oscillatory behavior found in
the multichanne l evaporator is therefore believed to improve the heat transfer, although the improvement decreases at higher heat fluxes. Another fact to note is that
the heat transfer coef cient is lower for the 1-mm
1-mm multichannel con guration than for its 2-mm
4-mm counterpart as shown in Figure 12.
Tran et al. [20] conducted experiments with R-12 in
small circular and rectangular channels, both of
2.46-mm hydraulic diameter. Their results indicated
two distinct regions, a convective boiling-dominan t region at lower wall superheat values, and a nucleate
boiling-dominan t region at higher wall superheat values. They compared their data with the Kandlikar [50]
correlation, and found that it exhibited similar trends,
but underpredicted the results. One reason for this difference is the fact that the single-phase heat transfer
coef cients in small-diameter tubes is generally higher
16

heat transfer engineering

Figure 12 Comparisons of the average heat transfer coef cient

trendlines for water in 2-mm 4-mm single- and multichannel
con gurations, and 1-mm 1-mm multichannel con guration. G D
200 kg=m2 s, Tsat D 120 C, Mertz et al. [18].

than those predicted by the Dittus-Boelter type of correlations. It is recommended that the measured
single-phase heat transfer coef cients be used in the
correlation as recommended by Kandlikar [47].
It is interesting to note that the heat transfer coef cients obtained by Tran et al. at higher qualities
exhibited a dependence on heat ux alone. The mass
ux had virtually no in uence on the heat transfer coef cient. These observations are supported by the visual observations made by Kasza et al. [7], who studied ow boiling of water in rectangular channels of
2.5-mm 6-mm cross section. They concluded that the
increased bubble activity at nucleation sites in the thin
liquid lm is responsible for high heat transfer coef cient in small-hydraulic-diamete r tubes and channels.
Kuznetsov and Shamirzaev [26] conducted experiments
with R-318C in an annular gap of 0.9 mm. They observed that at higher values of quality, nucleation was
seen to be suppressed. However, their experimental results were in agreement with the correlation by Tran
et al. [20], which was developed for the nucleate boiling dominant region.
Kamidis and Ravigururajan [25] conducted ow
boiling experiments with R-113 in circular tubes of
1.59, 2.78, 3.97, and 4.62 mm diameter. Their results
vol. 23 no. 1 2002

Figure 13

Comparison of Kamidis and Ravigururajan (1999) data with Kandlikar correlation (1990) for 1.59-mm-diameter tube, R-113.

indicate that very high heat transfer coef cients of the

order of 10 kW/m2 C are obtained in the ow boiling region. Figure 13 shows a comparison of their data
for 1.59-mm-diameter tube with the Kandlikar [50]
correlation. For a Reynolds number of 5,720, the agreement is excellent. For Re D 2,370, the correlation underpredicts the data. It is suspected that the
Reynolds number is in the transition region where the
turbulent single-phase heat transfer correlation is not
applicable.
Kennedy et al. [30] studied the instabilit y during ow
boiling of water in circular tubes of 1.17 and 1.45 mm
diameter. Based on their results, they proposed that the
instability is initiated in the tube when the impressed
heat ux is 90% of the value required to obtain saturated
vapor condition at the exit section.
Lakshminarasimhan et al. [31] conducted experiments with R-11 in 1-mm 20-mm rectangular channels. They noted that the their saturated ow boiling
data was accurately predicted using the large-tube
correlation by Kandlikar [50]. They also observed a
clear boiling front with liquid ow in the laminar region. In the turbulent region, the boiling front could
not be clearly viewed due to the high heat transfer
coef cient in the single-phase region prior to
nucleation.
The experimental studies available in the literature
provide some preliminary data on ow boiling heat
transfer in small-diameter tubes and channels.
heat transfer engineering

In the case of Lazarek and Blacks [11] and

Lakshminarasimhan et al. [31] data, the Kandlikar [50]
correlation for large-diameter tubes predicted the results satisfactorily. However, Tran et al. [20] indicated
a signi cant enhancement over the large-diameter correlations. Pressure uctuations and multichannel effects
are not clearly understood for small-diameter tubes.
As noted by many investigators , including Lin et al.
[27] and Mertz et al. [18], pressure uctuations have
a signi cant effect on the ow characteristics and associated heat transfer performance. The periodic lling
of the ow channel with large vapor plugs, followed
by all-liquid ow in the channel, make it very dif cult
to apply ow pattern-based models to predict the heat
transfer rates. Further research in this area is highly
recommended.
Another class of ow channels that have received
some attention in literature are those with hydraulic diameters below 500 l m. These channels are referred
to as microchannels, although the precise boundary is
not well de ned. There are very few quantitativ e studies available for the microchannel geometry under ow
boiling conditions . Further efforts are needed in this
area to generate high-qualit y experimental data in microchannels under ow boiling conditions. This geometry has been investigated with its potential application
in microelectronic s cooling. Table 1 clearly indicates
that there is very little quantitative data available for
microchannels.
vol. 23 no. 1 2002

17

Ravigururajan et al. [19] studied ow boiling in a microchannel 270 l m wide and 11 mm deep. The working
uid was R-124, and was tested over the entire quality
range. They found that the heat transfer coef cient decreased from a value of 11 kW/m2 C at x D 0.01 to
about 8 kW/m2 C at x D 0.65. Although no conclusions were drawn, this behavior may be the results of
the two trends: (1 ) the nucleate boiling heat transfer is
dominant, leading to its suppression at higher qualities;
or (2 ) the higher vapor fraction leads to ow oscillation s
in multichannels with a consequent change (increase? )
in the heat transfer coef cient.
A number of investigator s (for example, Peng and
Wang [23], Peng et al. [24] ) have indicated that the
ow boiling heat transfer in microchannels may be quite
different than that in larger-diameter tubes. They also
indicated that the regular nucleate boiling phenomenon
does not exist in microchannels.
Peng and Wang [23] conducted experiments with
water, ethanol, and their mixtures in different shaped
microchannel geometries (listed in Table 1 ). They noted
the presence of both nucleate boiling and convective
boiling in various regimes. They did not observe any
bubble activity in the rectangular and triangular passages with hydraulic diameters between 0.1 to
0.6 mm. In turn, they called this a ctitious boiling
phenomenon.
It is dif cult to accept the notion of the ctitious boiling presented by Peng and Wang [23]. Similar studies
reported by Kandlikar and Stumm [38] and Kandlikar
and Spiesman [39] with a channel height of 3 mm
indicate that bubbles as small as 10 l m are seen to
depart from the nucleating sites. The key to observing
bubble activity in small channels is to employ highspeed photography along with a high-resolutio n
microscope.
The theoretical analysis presented by Peng and Wang
[23] considers a bubble nucleus that completely lls the
tube. The microchannel dimension is of the order of
100 l m, while the cavity sizes for active nucleation are
on the order of a few micrometers or smaller. It is expected that the nucleation criterion for ow boiling, established for large-diameter tubes, will hold true unless
the tube diameter approaches the cavity dimensions.
Such a condition may exist only in submicrometer-size d
tubes.
In conclusion, ow boiling in microchannels is an
area where further research is needed. The dif culty in
observing the bubbles and in the accurate measurement
of heat fluxes at the wall make it very dif cult to understand the mechanism of ow boiling heat transfer
in this geometry. With the availability of more accurate
data, we may be able to nd some of the answers in the
near future.
18

heat transfer engineering

DESIGN CONSIDERATIONS
IN MINICHANNEL EVAPORATORS
Mehendale et al. [52] and [53] present a good
overview of the design consideration s for heat exchangers employing mini- and microchannels.
Flow Instability in Multichannel Evaporators
The small-diameter multichanne l evaporators differ
from the small-hydraulic-diamete r compact heat exchangers in one vital aspect: there are no cross ow connections available for the uid to ow across the width
of the ow channel as it passes through the evaporator. This cross-connectio n helps the nucleating bubbles
grow in the crosswise direction without blocking the
entire ow passage, as in the case of small-diameter
channels. This ow structure is clearly illustrate d by
the ow pattern investigatio n conducted by Kuznetsov
and Shamirzaev [26] in an annular gap between two
concentric tubes. They observed a cell pattern that effectively allows the bubbles and vapor to grow in the
cross ow direction without blocking the ow. In the
case of square channels, as shown by Kandlikar et al.
[32], the vapor bubble growth leads to large pressure
uctuations that are not desirable for stable operation
of the evaporator. The presence of ns or ribs in the gap
is expected to further provide stability to the ow by
increasing ow resistance.
In the case of multichannel evaporators employing
individual small-diameter tubes or channels, the channels running parallel to any given channel (which is
experiencing vapor expansion in a direction opposite
to the ow ) act in a manner similar to reducing the
upstream pressure drop characteristics in a two-phase
system. Severe pressure drop uctuations, coupled with
the back ow of vapor into the inlet manifold, are not
desirable. These could lead to premature CHF in some
of the channels where vapor may ow preferentially,
without being accompanied by the liquid ow. Some
of the systems that employ such evaporators may not
tolerate such severe uctuations in the ow rate.
With these considerations , it is necessary to design
multichannel evaporators that avoid the severe pressure
uctuations found in parallel channels. Further research
in this area is warranted.
Design Guidelines for Sizing Small-Diameter
Multichannel Evaporators
Small channels present a number of advantages, making them attractive for speci c systems. Their compact
size, low weight, low liquid/vapor inventory, and fast
vol. 23 no. 1 2002

response are just some of the desirable features. In this

section, preliminary guideline s are presented for designing multichannel evaporators with small-diameter
channels. The design conditions considered in this analysis are as follows.

Substituting n from Eq. (3 ) into Eq. (4 ), the ratio of

the lengths in the ow direction is given by

Design Conditions

For the same exit quality conditions, the total heat

transferred in the heat exchanger remains the same. The
heat fluxes in the two cases are related by the following
equation:

It is assumed that an existing single - or multichannel evaporator employing large-diameter (D ) tubes is

to be replaced with a multichannel evaporator employing small-diameter (d ) tubes. Capital letter subscripts
refer to the large-diameter tube, while lowercase subscripts refer to smaller-diameter tubes. For the qualitative analysis presented here, it is assumed that both
geometries employ circular tubes, and that one largerdiameter tube is being replaced by n number of smallerdiameter tubes. The total mass ow rate and the total
heat transfer rate are identical in both cases. The objective is to arrive at the number of small-diameter tubes
needed to replace each large-diameter tube, and the new
length of the evaporator. Another consideration is the
need for equal pressure drop in the two cases. The analysis is presented for a single-pass evaporator. It can be
extended to a multipass evaporator con guration, but
the treatment will introduce many additional parameters. The purpose of the following exercise is to provide some simple guidelines to help in designing the
new evaporator. Extensive design efforts will be needed
to arrive at the nal design and all of the associated
details.
Design Comparison
If n parallel channels are employed in the new evaporator (for each tube in the original design ), the mass
ux and heat ux for the new heat exchanger with smalldiameter (d ) tubes would be different than those of the
original heat exchanger with large-diameter (D ) tubes.
Since the total mass ow rate remains constant between
the two designs, we get
nD

G D D2
Gd d2

(3 )

The total heat transfer rates in the two cases are also
identical, since the evaporators are being designed for
the same heat duty. If we apply the heat transfer rate
equations with the respective average heat transfer coef cients, and assume that the operating temperature
difference in the two cases is identical, we can write
Ld
h D 1 D
D
LD
h d n d

(4 )
heat transfer engineering

Ld
h D G d d
D
LD
h d G D D

(5 )

qd
1 D LD
D
qD
n d Ld

(6 )

The ow boiling in the small-diameter tubes exhibits

a nucleate boiling-dominan t region at low qualities.
Here the heat transfer coef cient varies as q 0:7 . In the
higher-quality region, the ow becomes convective
boiling-dominant , with the heat transfer coef cient independent of q, and varying as G 0:7 . In general, the
dependence of h may be expressed as h / q m G p . The
ratio of the two heat transfer coef cients is
given by
h d
D
h D

qd
qD

Gd
GD

(7 )

The exponents m and p depend on the ow boiling

heat transfer characteristics under the prescribed operating conditions . Their values range from m D 0 to 0.7,
and p D 0 to 0.8 (Kandlikar, [40] ). Substituting the heat
ux ratio from Eq. (6 ) into Eq. (7 ), combining it with
Eq. (3 ), and then substitutin g the heat transfer coef cient ratio into Eq. (5 ), the ratio of the two lengths is
obtained as
Ld
D
LD

Gd
GD

(1mp)=(1m)

d
D

(8 )

Equation (8 ) provides a preliminary estimate of the

length of the heat exchanger needed to obtain the same
vapor generation rate using a small-diameter evaporator. The number of small-diameter tubes replacing each
large-diameter tube is given by Eq. (3 ).
A few observations can be made regarding the effect
of boiling characteristics on the length ratio presented
in Eq. (8 ). If the nucleate boiling is the dominant mode
in both cases, then p D 0. This results in a mass ux
ratio dependence given by ( L d=L D ) D (G d=G D )(d=D ).
On the other hand, if the heat transfer is convective
dominant, then m D 0 and p D 0.8. In the latter case,
the mass ux ratio has a weak effect on the length ratio, ( L d=L D ) D (G d=G D )0:2 (d=D ). However, the actual
vol. 23 no. 1 2002

19

variation lies between the two extreme cases discussed

here.
Another important factor to be considered in the design of heat exchangers is the pressure drop. The design
guideline provided by Eq. (8 ) does not address this issue. A similar analysis is now presented for pressure
drop comparison in the two cases.
The pressure drop analysis is a bit more complex. To
simplify the analysis, the gravitational pressure drop
is considered negligible when compared to the friction and acceleration pressure drop terms. Assuming
the respective exit quality and exit void fractions to be
equal in both cases, the pressure drop may be expressed
in terms of the mass ux, tube length, tube diameter,
exit quality, and uid properties. The frictional pressure drop varies as p f / f TP L G 2 = D, while the acceleration pressure drop varies as p f / G 2 . In addition,
the void fraction plays a role in the pressure drop terms.
The two-phase friction factor may be considered to vary
as f TP / G 0:25 . Assuming that the same equations apply for the small- and large-diameter tubes, the ratio of
pressure drops may be expressed as

L d G 1:75
d C 1 F1 (a
pd
d

D
1:75
pD
L D G D D C1 F1 (a

d) C

D) C

G 2d C 2 F2 (a

d)
2
G D C2 F2 (a D )

(9 )

The constants C 1 and C 2 include additiona l variables

in the respective pressure drop terms. F1 and F2 are
functions of the void fractions in the two cases. Although the effects of mass ux, diameter, and length
on pressure drop are not immediately obvious, assuming the frictional pressure drop to be dominant provides
some degree of guidance. In addition, assuming the void
fractions and their effects to be similar, Eq. (9 ) may be
simpli ed as follows:

L d G 1:75
d
pd
d

D
1:75
pD
LDGD D

(10 )

For the case of equal pressure drop between the two

con gurations, the length ratio is obtained as
Ld
D
LD

Gd
GD

1:75

d
D

(11 )

Comparing Eqs. (8 ) and (11 ), it is clear that the effect of mass ux is more severe on the pressure drop
than on the heat transfer. The diameter effect is same
in both cases. If the mass ux is held constant for
the two con gurations, then the length ratio is identical to the diameter ratio, L d=L D D d=D. However, in
20

heat transfer engineering

practical system designs, a higher pressure drop is generally accepted with evaporators employing small channels, and tube lengths larger than that given by Eq. (11 )
are employed.
The negative exponent in the mass ux ratio of
Eq. (11 ) indicates that increasing the mass ux results
in shorter tube lengths for the same pressure drop. In
other words, increasing G d causes the pressure drop to
increase, and shorter tube lengths are needed to meet the
pressure drop requirements. From the heat transfer perspective, a larger tube length may be needed to accommodate higher mass fluxes. Consequently, the design
mass ux is a compromise between these considerations and other system requirements.
The preceding discussion provides a preliminary basis for the selection of a mass ux value for the smallertube-diameter heat exchanger being designed to replace
an existing larger-tube-diameter evaporator. Needless
to say, a number of additional parameters, including
uid properties, the local heat transfer coef cient and
pressure drop relationships, differences in manifolding
and number of passes, and the differences in allowable pressure drop will affect the design of the new
evaporator with smaller-diameter tubes. Another major consideration in the design of the evaporator is the
performance on the hot- uid side. In the analysis presented here, the wall temperatures were considered to be
identical in both evaporators. The comparisons should
therefore be treated as qualitative in determining rstorder effects.
CONCLUSIONS
On the basis of a critical literature review and the
work conducted by the author, the following conclusions are drawn:
1. Three ow patterns are commonly encountered during ow boiling in minichannels: isolated bubble,
con ned bubble or plug/slug, and annular. The visual
studies available in the literature have been generally
conducted for low mass ux values in tubes of 1-mm
or larger hydraulic diameters.
2. Large pressure drop uctuations are noted in
multichannel evaporators. Flow pattern observations revealed a ow reversal in some channels
with expanding bubbles pushing the liquid vapor
interface in both the upstream and downstream
directions.
3. Heat transfer studies in the minichannels indicate
that, as a rst-order estimate, heat transfer may be
predicted using the ow boiling correlations developed for large-diameter tubes.
vol. 23 no. 1 2002

4. The heat transfer rate in multichannel evaporators

is different from that in single-channel evaporators
under the same set of operating conditions. The role
of ow uctuations due to ow rate instabilitie s are
not clearly understood at this stage.
5. Severe pressure drop uctuations are not included in
any pressure drop prediction schemes for minichannel evaporators. Both separated and homogeneous
ow models have been used with some degree of
success by previous investigators .
6. In designing evaporators with small-diameter channels, the length-to-diamete r ratio depends on the
heat transfer and pressure drop characteristics. Larger
pressure drops are generally accepted in evaporators
with small-diameter channels.

latent heat of vaporization, J/kg

average heat transfer coef cient in the
evaporator, W/m2 C
length of the evaporator tube
number of parallel small-diameter tubes
for each large-diameter tube
pressure drop, N/m2
heat ux, W/m2
quality
void fraction
density, kg/m3
surface tension, N/m
Martinelli parameter fD (q V =q L )0:5
(l L =l V )0:1 [(1 x )=x]0:9 g

h ffg
h
L
n
p
q
x
a
q
r
v

tt

Subscripts
FUTURE RESEARCH NEEDS
Future research needs are summarized below.
1. Conduct high-speed video studies to obtain ow pattern information under high-mass- ux conditions in
small-diameter tubes and channels.
2. Compare the performance of single-tube evaporators and multichannel evaporators under the same
operating conditions and identify the reasons for the
differences in their performance.
3. Study the effects of inlet ow conditions and manifold design on the performance of the multichannel
evaporator.
4. Conduct more experiments with minichannel evaporators to obtain accurate ow boiling heat transfer and pressure drop data as a function of quality, heat ux, mass ux, and tube/channel hydraulic
diameter.
5. Critical heat ux is an important factor in the design
of evaporators. Although not covered in this article
due to space constraints, there is a need to obtain
more experimental data for CHF in single and parallel minichannels.
NOMENCLATURE
C
C1 and C2
D
d
e
F1 and F2
g
G
K sa

constant in Eq. (1 )
constants in Eq. (9 )
diameter of large-diameter tube
diameter of small-diameter tube
gap size, m
constants in Eq. (9 )
acceleration due to gravity, m/s2
mass ux, kg/m2 s
pressure drop multiplie r in acceleration
pressure drop, Eq. (2 )
heat transfer engineering

D
d
ex
f
in
L
LO
TP
tt
V

large-diameter tube
small-diameter tube
exit
friction
inlet
liquid
all ow as liquid
two-phase
turbulent-turbulent
vapor

REFERENCES
[1] Baker, O., Simultaneous Flow of Oil and Gas, Oil Gas J.,
vol. 53, pp. 185 190, 1954.
[2] Hewitt, G. F., and Roberts, D. N., Studies of Two-Phase Flow
Patterns by Simultaneous X-Ray and Flash Photography, UK
AEA Rep. ASRE-M2159, 1969.
[3] Taitel, Y., Barnea, D., and Dukler, A. E., Modeling Flow Pattern Transition for Gas-Liquid Transitions for Steady Upward
Gas-Liquid Flow in Vertical Tubes, AIChE J., vol. 26, no. 3,
pp. 345 354, 1980.
[4] Fukano, T., Kariyasaki, A., and Kagawa, M., Flow Patterns and
Pressure Drop in Isothermal Gas-Liquid Flow in a Horizontal
Capillary Tube, ANS Proc. 1989 Natl. Heat Transfer Conf.,
vol. 4, pp. 153 161, 1989.
[5] Wambsganss, M. W., Jendrzejczyk, J. A., and France, D. M.,
Two-Phase Flow Patterns and Transitions in Small, Horizontal,
Rectangular Channels, Int. J. Multiphase Flow, vol. 17, no. 3,
pp. 327 342, 1991.
[6] Triplett, K. A., Ghiaasiaan, S. M., Abdel-Khalik, S. I.,
LeMouel, A., and McCord, B. N., Gas-Liquid Two-Phase Flow
in Microchannels Part II: Void Fraction and Pressure Drop, Int.
J. Multiphase Flow, vol. 25, pp. 395 410, 1999.
[7] Kasza, K. E., Didascalou, T., and Wambsganss, M. W.,
Microscale Flow Visualization of Nucleate Boiling in Small
Channels: Mechanisms In uencing Heat Transfer, in R. K.
Shah (ed.), Proc. Int. Conf. on Compact Heat Exchanges for
the Process Industries, pp. 343 352, Begell House, New York,
1997.

vol. 23 no. 1 2002

21

[8] Hewitt, G. F., Fluid Mechanics Aspects of Two-Phase Flow,

in S. G. Kandlikar, M. Shoji, and V. K. Dhir (eds.), Handbook
of Boiling and Condensation, Chap. 9, Taylor and Francis,
New York, 1999.
[9] Mandhane, J. M., Gregory, G. A., and Aziz, K., 1974, A Flow
Pattern Map for Gas-Liquid Flow in Horizontal Pipes, Chem.
Eng. Prog., vol. 45, pp. 39 48, 1974.
[10] Taitel, Y., and Dukler, A. E., A Model for Predicting Flow
Regime Transitions in Horizontal and Near Horizontal GasLiquid Flow, AIChE J., vol. 22, pp. 47 55, 1976.
[11] Lazarek, G. M., and Black, S. H., Evaporative Heat Transfer,
Pressure Drop and Critical Heat Flux in a Small Diameter
Vertical Tube with R-113, Int. J. Heat Mass Transfer, vol. 25,
no. 7, pp. 945 960, 1982.
[12] Cornwell, K., and Kew, P. A., Boiling in Small Parallel Channels, Proc. CEC Conf. on Energy Ef ciency in Process Technology, Athens, Greece, October 1992, Elsevier Applied Sciences,
pp. 624 638.
[13] Moriyama, K., and Inoue, A., The Thermohydraulic Characteristics of Two-Phase Flow in Extremely Narrow Channels (The Frictional Pressure Drop and Heat Transfer of Boiling Two-Phase Flow, Analytical Model), Heat TransferJpn.
Res., vol. 21, no. 8, pp. 838 856, 1992.
[14] Wambsganss, M. W., France, D. M., Jendrzejczyk, J. A., and
Tran, T. N., Boiling Heat Transfer in a Small Diameter Tube,
ASME J. Heat Transfer, vol. 115, no. 4, pp. 963 972, 1992.
[15] Bowers, M. B., and Mudawar, I., High Flux Boiling in Low
Flow Rate, Low Pressure Drop Mini-channel and Microchannel Heat Sinks, Int. J. Heat Mass Transfer, vol. 37, no. 2,
pp. 321 334, 1994.
[16] Bowers, M. B., and Mudawar, I., Two-Phase Electronic Cooling Using Mini-Channel and Micro-channel Heat Sinks
Part 1. Design Criteria and Heat Diffusion Constraints, ASME
J. Electronic Packaging, vol. 116, pp. 290 297, 1994.
[17] Bowers, M. B., and Mudawar, I., Two-Phase Electronic Cooling Using Mini-Channel and Micro-channel Heat Sinks
Part 2. Flow Rate and Pressure Drop Constraints, ASME J.
Electronic Packaging, vol. 116, pp. 290 297, 1994.
[18] Mertz, R., Wein, A., and Groll, M., Experimental Investigation
of Flow Boiling Heat Transfer in Narrow Channels, Calore e
Technologia, vol. 14, no. 2, pp. 47 54, 1996.
[19] Ravigururajan, T. S., Cuta, J., McDonald, C. E., and Drost,
M. K., Effects of Heat Flux on Two-Phase Flow Characteristics of Refrigerant Flows in a Micro-Channel Heat Exchanger, 1996, Natl. Heat Transfer Conf., vol. 7, HTD-Vol. 329,
pp. 167 178, ASME, New York, 1996.
[20] Tran, T. N., Wambsganss, M. W., and France, D. M., Small
Circular- and Rectangular-Channel Boiling with Two Refrigerants, Int. J. Multiphase Flow, vol. 22, no. 3, pp. 485 498,
1996.
[21] Tong, W., Bergles, A. E., Jensen, M. K., Pressure Drop with
Highly Subcooled Flow Boiling in Small-Diameter Tubes,
Exp. Thermal Fluid Sci., vol. 15, pp. 202 212, 1997.
[22] Bonjour, J., and Lallemand, M., Flow Patterns during Boiling in a Narrow Space between Two Vertical Surfaces, Int. J.
Multiphase Flow, vol. 24, pp. 947 960, 1998.
[23] Peng, X. F., and Wang, B. X., Forced Convection and Boiling Characteristics in Microchannels, Heat Transfer 1998,
Proc. 11th IHTC, August 23 28, 1998, Kyongju, Korea, vol. 1,
pp. 371 390.
[24] Peng, X. F., Hu, H. Y., and Wang, B. X., Boiling Nucleation
During Liquid Flow in Microchannels, Int. J. Heat Mass Transfer, vol. 41, no. 1, pp. 101 106, 1998.

22

heat transfer engineering

[25] Kamidis, D. E., and Ravigururajan, T. S., Single and Twophase Refrigerant Flow in Mini-channels, Proc. NHTC2000:
33rd Natl. Heat Transfer Conf., Albuquerque, NM, August
20 22, 2000, Paper NHTC2000-12100 ; pp. 1 8.
[26] Kuznetsov, V. V., and Shamirzaev, A. S., Two-Phase Flow Pattern and Flow Boiling Heat Transfer in Noncircular Channel
with a Small Gap, Two-Phase Flow Modeling and Experimentation, pp. 249 253, 1999.
[27] Lin, S., Kew, P. A., and Cornwell, K., Two-Phase Evaporation
in a 1 mm Diameter Tube, 6th UK Heat Transfer Conf. in
Edinburgh, September 1999.
[28] Downing, R. S., Meinecke, J., and Kojasoy, G., The Effects of
Curvature on Pressure Drop for Single and Two-Phase Flow in
Miniature Channels, Proc. NHTC 2000: 34th Natl. Heat Transfer Conf., Pittsburgh, PA, August 2000, Paper NHTC200012100.
[29] Hetsroni, G., Segal, Z., and Mosyak, A., Nonunifrom Temperature Distribution in Electronic Devices Cooled by Flow
in Parallel Microchannels, Packaging Electronic Photonic Devices, EEP-vol. 28, pp. 1 9, 2000.
[30] Kennedy, J. E., Roach, G. M., Jr., Dowling, M. F., AbdelKhalik, S. I., Ghiaasiaan, S. M., Jeter, S. M., and Quereshi,
Z. H., The Onset of Flow Instability in Uniformly Heated
Horizontal Microchannels, ASME J. Heat Transfer, vol. 122,
pp. 118 125, 2000.
[31] Lakshminarasimhan, M. S., Hollingsworth, D. K., and Witte,
L. C., Boiling Incipience in Narrow Channels, HTD-Vol. 3664, Proc. ASME Heat Transfer Division 2000, ASME IMECE
2000, vol. 4, pp. 55 63.
[32] Kandlikar, S. G., Steinke, M. E., Tian, S., and Campbell,
L. A., High-Speed Photographic Observation of Flow Boiling
of Water in Parallel Mini-Channels, ASME Natl. Heat Transfer
Conf., June 2001.
[33] Lin, S., Kew, P. A., and Cornwell, K., Two-Phase Flow Regimes
and Heat Transfer in Small Tubes and Channels, Heat Transfer
1998, Proc. 11th Int. Heat Transfer Conf., Kyongju, Korea,
1998, vol. 2, pp. 45 50.
[34] Barnea, D., Luninsky, Y., and Taitel, Y., Flow Pattern in Horizontal and Vertical Two-Phase Flow in Small Diameter Pipes,
Can. J. Chem. Eng., vol. 61, pp. 617 620, 1983.
[35] Mishima, K., and Hibiki, T., Some Characteristics of Air-Water
Two-Phase Flows in Small Diameter Tubes, Int. J. Multiphase
Flow, vol. 22, no. 4, pp. 703 712, 1996.
[36] Yao, S.-C., and Chang, Y., Pool Boiling Heat Transfer in a
Con ned Space, Int. J. Heat Mass Transfer, vol. 26, no. 6,
pp. 841 848, 1983.
[37] Kuznetsov, V. V., and Vitovsky, O. V., Flow Pattern of TwoPhase Flow in Vertical Annuli and Rectangular Channel with
Narrow Gap, in G. P. Celata and R. K. Shah (eds.), TwoPhase Flow Modelling and Experimentation 1995, Pisa, Italy,
Edizioni ETS, 1995.
[38] Kandlikar, S. G., and Stumm, B. S., A Control Volume Approach to Predict Departure Bubble Diameter in Flow Boiling,
ASME J. Heat Transfer, vol. 117, pp. 990 997, 1995.
[39] Kandlikar, S. G., and Spiesman, P. H., Effect of Surface Characteristics on Flow Boiling Heat Transfer, Engineering Foundation Conf. on Convective and Pool Boiling, Irsee, Germany,
May 18 25, 1997.
[40] Kandlikar, S. G., Flow Boiling in Circular Tubes, in S. G.
Kandlikar, M. Shoji, and V. K. Dhir (eds.), Handbook of Boiling
and Condensation, Taylor & Francis, Philadelphia, PA, 1999.
[41] Collier, J. G., Convective Boiling and Condensation, McGrawHill, London, 1981.

vol. 23 no. 1 2002

[42] Martinelli, R. C., and Nelson, D. B., Prediction of Pressure Drop during Forced Convection Boiling of Water, ASME
Trans., vol. 70, pp. 695 702, 1948.
[43] Kew, P. A., and Cornwell, K., On Pressure Drop Fluctuations
during Boiling in Narrow Channels, in G. P. Celata, P. Di
Marco, and A. Mariani, (eds.), 2nd European Thermal Sciences and 14th UIT Natl. Heat Transfer Conf., Pisa, Italy,
pp. 1323 1327, Edizioni ETS, 1996.
[44] Nakayama, W., and Yabe, A., Flow Boiling in Narrow Channels for Thermal Management of Microelectronic Equipment,
in S. G. Kandlikar, M. Shoji, and V. K. Dhir (eds.), Handbook
of Phase Change: Boiling and Condensation, pp. 403 411,
Taylor & Francis, Philadelphia, PA, 1999.
[45] Kew, P. A., and Cornwell, K., Flow Boiling in Compact Heat
Exchangers, in S. G. Kandlikar, M. Shoji, and V. K. Dhir
(eds.), Handbook of Phase Change: Boiling and Condensation,
pp. 412 427, Taylor & Francis, Philadelphia, PA, 1999.
[46] Bergles, A. E., Subcooled Burnout in Tubes of Small Diameter,
ASME Paper 63-WA-182, 1964.
[47] Kandlikar, S. G., Development of a Flow Boiling Map for
Saturated and Subcooled Flow Boiling of Different Fluids
in Circular Tubes, J. Heat Transfer, vol. 113, pp. 190 200,
1991.
[48] Liu, Z., and Winterton, R. H. S., Wet Wall Flow Boiling Correlation with Explicit Nuclear Term, 5th Miami Symp. on Multiphase Transport and Particulate Phenomena, 1988.
[49] Shah, M. M., Chart Correlation for Saturated Boiling Heat
Transfer: Equations and Further Study, ASHRAE Trans.,
vol. 88, pp. 185 196, 1982.
[50] Kandlikar, S. G., A General Correlation for Saturated TwoPhase Flow Boiling Heat Transfer Inside Horizontal and Ver-

heat transfer engineering

tical Tubes, ASME J. of Heat Transfer, vol. 112, pp. 219 228,
1990.
[51] Chen, J. C., A Correlation for Boiling Heat Transfer to Saturated Fluids in Convective Flow, I & EC Process Design Devel.,
vol. 5, no. 3, pp. 322 329, 1966.
[52] Mehendale, S. S., Jacobi, A. M., and Shah, R. K., Heat Exchangers at Micro- and Meso-Scales, in R. K. Shah (ed.), Compact Heat Exchangers and Enhancement Technology for the
Process Industries, pp. 55 74, Begell House, New York, 1999.
[53] Mehendale, S. S., Jacobi, A. M., and Shah, R. K., Fluid
Flow and Heat Transfer at Micro- and Meso-Scales with
Applications to Heat Exchanger Design, Appl. Mech. Rev.,
vol. 53, pp. 175 193, 2000.

Dr. Satish G. Kandlikar is a Professor in

Mechanical Engineering at Rochester Institute of
Technology in Rochester, New York. He has been
with R.I.T. since 1980. Prior to coming to R.I.T.,
he obtained his ph.D. in 1975 from the Indian
Institute of Technology, Bombay, India, where
he continued to become Associate Professor before coming to R.I.T. He received the Eisenhart
outstanding teaching award at R.I.T. in 1997.
Dr. Kandlikar is a fellow of the ASME and heads the Heat Transfer Chapter
of the Rochester Section of the ASME. He is involved in research in the
areas of ow boiling, pool boiling, critical heat ux, microchannels , and
cooling of electronic components . He has published over 60 conference and
journal articles and has delivered several keynote lectures. He is also the
chief editor of the Handboo k of Phase Change: Boiling and Condensatio n,
published by Taylor & Francis. He is the current Heat and History Editor of
Heat Transfer Engineering journal.

vol. 23 no. 1 2002

23