Applied T/wnwl Dlginrering Vol. 17. Nos. 8-10. pp. 705-715.

1997
CTEuropean Communities 1997. Published by Elsevier Science Ltd
Pergamon Printed in Great Bntain
PII: s13s9-4311(%)00071-3 1359-4311197 $17.00 + 0.00

CORRELATIONS FOR THE PREDICTION OF BOILING

HEAT TRANSFER IN SMALL-DIAMETER CHANNELS

Peter A. Kew and Keith Cornwell

Department of Mechanical and Chemical Engineering, Heriot-Watt University,
Edinburgh, EH 14 4AS. UK

Abstract-This paper describes aspects of the work relating to boiling in single, small-diameter tubes as
part of a study of compact two-phase heat exchangers. In order to realise the energy-saving potential of
compact heat exchangers for evaporating duties it is necessary to establish design procedures. A test facility
was commissioned which was used to measure pressure drop and boiling heat transfer coefficients for
Rl4lb flowing through tubes 500 mm long with diameters of 1.39-3.69 mm. Established correlations
predicted the heat transfer coefficients reasonably well for the largest tube but performed badly when
applied to the smaller tubes. It would appear that simple nucleate pool boiling correlations, such as that
of Cooper, best predict the data. While under some conditions increasing quality leads to an increasing
heat transfer coefficient, it is suggested that intermittent dry-out occurs at very low quality in single narrow
channels, thus reducing the average heat transfer coefficient below that expected from the pool boiling
correlations. lc~ European Communities 1997. Published by Elsevier Science Ltd.

Keywords-Compact heat exchangers, heat transfer, boiling, evaporation, process intensification.

NOTATION
BO boiling number (q//r,,G)
CO confinement Number [Equation (I)]
d hydraulic diameter, m
F enhancement factor [Equation (2)]
I: acceleration due to gravity, m/s’
G mass flux, kg/m’s
k liquid thermal conductivity. W/mK
A4 molecular mass
N number of data points
NU Nusselt number (a&k)
Nu, film Nusselt number (rb/k)
Pr reduced pressure
q heat flux, W/m’
Re Reynolds number (GdJp)
s suppression factor [Equation (2)]
We, Weber number (G’d,/p,g)
.\- dryness fraction

s( heat transfer coefficient, W/m’K

6 film thickness, m
P density, kg/m’
(r surface tension, N/m

Subscripts

npb nucleate pool boiling

L-1 liquid
g vapour
pred predicted
cxp experimental

INTRODUCTION

Traditionally, evaporation or boiling processes have occurred in tubular heat exchangers, either
within the tubes or on the outside of the tubes. Recently there has been a growing awareness of

705
706 P. A. Kew and K. Cornwell

the benefits of process intensification-the reduction in plant size, for a given capacity, by an order
of magnitude or more-and this has led to a requirement for smaller evaporators.
Utilisation of a physically small heat exchanger for a given duty is advantageous for several
reasons: often (although not always) small physical size is associated with relatively low capital cost;
installation costs are reduced if the size of a component is reduced; a lower fluid inventory can
be beneficial on both cost and safety grounds; and the nature of many compact heat exchanger
designs is such that close approach temperatures can be achieved. The feasibility of heat recovery
schemes, associated with significant energy savings, is enhanced by the small size and low costs
associated with compact heat exchangers. Close approach temperatures can also result in significant
energy savings in well-integrated plant.
Recognising the importance of this area of research and development the Commission of the
European Communities (CEC) has financed collaborative projects in the area of enhanced
evaporation and compact heat exchangers under the JOULE I and JOULE II programmes relating
to the rational use of energy. The work described in this paper was undertaken as part of a project
entitled “Compact Two-Phase Heat Exchangers” [ 11,a collaborative project involving 18 partners
from 5 EU Countries. The laboratories involved in the work package from which this paper is
derived were Heriot-Watt University (UK), the University of Oxford (UK) and British Gas plc
(UK). Other laboratories involved in the project were IKE (University of Stuttgart), National
Technical University of Athens (Greece) and BEHR (Germany) investigating heat transfer and flow
phenomena in narrow spaces; AEA Technology (UK), the University of Nottingham (UK) and
a group of heat exchanger manufacturers and users studying the design implications for compact
heat exchangers; and GRETh (France), together with three heat exchanger manufacturers
investigating the performance of enhanced plate evaporators.
The aim of the study was to develop technical expertise to enable new types of compact heat
exchangers to be used for evaporation duties and to provide design information required to produce
advanced compact heat exchangers. A fundamental requirement of the heat exchanger designer is
the ability to predict heat transfer coefficients under the conditions of interest with confidence. In
this paper the selection of appropriate design correlations is discussed. It is shown that conventional
in-tube flow boiling correlations may be used for tubes having a confinement number [2] of the
order of 0.3 and above. Consideration of the flow regimes and heat transfer mechanisms in narrow
channels has led to a framework being proposed based on a film flow model [3], which permits
the evaluation of heat transfer coefficients in narrow channels in the absence of dry-out. However,
at relatively low heat fluxes and qualities, intermittent dry-out of the tube wall can occur during
boiling, thus reducing the effective average film coefficient. Until the onset of dry-out can be
predicted reliably it is suggested that established nucleate boiling correlations are appropriate for
the determination of heat transfer coefficients in small diameter tubes.

TEST FACILITY

The test facility at Heriot-Watt University comprised a rig incorporating two flow loops. One
consisted of a loop for the measurement of local heat transfer and pressure drop in single, narrow
channels. The other permitted flow visualisation together with the measurement of heat transfer
and pressure drop data in geometries with large aspect ratios (76 mm wide by up to 4.5 mm deep)
or multi-channel geometries. The rig was operated with water, R141b and Flutec PPl. The results
reported here were obtained from the single-tube test section using R141b, which has a boiling
point of 32°C at atmospheric pressure.
The rig is shown schematically in Fig. 1. The data logging equipment, fluid reservoir, circulating
pump and condenser were common, however, flow meters and instrumentation were specific to one
of two loops of the rig. These loops are referred to as the single-tube test loop and the flow
visualisation test loop. Liquid was pumped from the reservoir through the appropriate test loop
via manual flow control valves and variable orifice flow meters connected in parallel. In each loop
an electric liquid preheater was incorporated in the pipe between the flow meters and the test
section.
The single-tube test section is illustrated schematically in Fig. 2. All test sections were thin-walled
stainless steel and were heated by passage of a DC electric current along the tube. The test section
 Predicting boiling heat transfer 707

‘uid
F’:
Reservoir

Fig. I. Schematic diagram of test rig.

was enclosed in a 22 mm diameter copper tube which was heated with three independently
controlled guard heaters. The environment immediately surrounding the test section was
maintained within 2 K of the average temperature measured by the surface thermocouples. The
outer wall temperature along the single-tube test section was measured using 10 equally spaced
thermocouples.
Calibrated, Type K thermocouples with an ice reference junction were used for temperature
measurement. The thermocouples used on the single-tube test section were made from 0.2 mm
diameter wires with spot-welded junctions, while other thermocouples were 1.5 mm diameter and
stainless-steel sheathed. Pressures at the inlet and outlet of the test sections were measured using
calibrated WIKA type 891.14.525 pressure transducers giving an output signal of 4-20 mA over
a O-2.5 bar range. Flow rates were measured using variable orifice-type flow meters. As supplied
these were calibrated for water at 20°C. These were individually calibrated for the working fluids
R141b and Flutec PPl. Flow meters having ranges permitting measurement of flows from
lo-350 ml/min of water were used on the single-tube circuit and 50-2000 ml/min on the flow
visualisation circuit.
Two data acquisition systems were used. A Hewlett Packard 3421A was used for monitoring
and recording steady-state and time-averaged values and a Keithley DAS8/EXP16 was used for
recording transient values. The Hewlett Packard was a stable instrument with a low susceptibility
to noise but a relatively low logging speed. The Keithley was more susceptible to electrical noise
when measuring the low voltages associated with thermocouples but was capable of much higher
data acquisition rates. The Keithley data acquisition system was connected to a PC running
‘EASYLX’ data acquisition and analysis software. This arrangement permitted the recording and
analysis of rapidly varying signals.
The multi-channel test section incorporated a window which permitted visualisation of the flow
within the channels either directly or by video or photographic means. Further details of this test
section and the heat transfer results obtained from it are included in ref. [l].

AC Guard heater coils

Fig. 2. Schematic diagram of single-tube test section
708 P. A. Kew and K. Cornwell

Table I. Correlations used for comparison with measured data

Reference Reference No. Equation No. Correlation
Liu and Winterton (1988) F-51 2 a’ = (See.,,)’+ (I%,)’
Cooper (1984) [I81 3 I = 5Q$ I?( _ ,og,,,P,) - 0“M 0ryw
Cooper (1989) 4 * = 35P: I?(_ log,,)J - 0‘SM- 0y 07
Lazarek and Black (1982) ;I;{ 5 Nu = 30Re”X”‘&,““’
Modified Lazarek and Black (1982) 6 No = 3ORe”X’7&,“711(, _ ‘1)-1I131
Tran, Wambsganns and France (1995) [I31 7 1 = 840(Bo?We,)“‘(p,/p, )-I” kW/m’K

BACKGROUND

Widely used flow boiling correlations, for example, those of Chen [4], Shah [5] and Liu and
Winterton [6], have been derived using databanks containing few data relating to small diameter
passages in which the effects of confinement are significant. Examination of studies of capillary
flow [7] and flooding in vertical up-flow [S] and of heat transfer in confined spaces under a variety
of conditions [9-l l] suggests that the effects of confinement will be significant for channels having
hydraulic diameters such that the confinement number, introduced by the present authors [2] and
defined by equation (l), is in excess of 0.5.

co = bMPl - P,W
4 ’
(1)

There are few studies concerned with boiling heat transfer in channels of hydraulic diameter of
the order of 3 mm and below. Lazarek and Black [12] worked with R113 in a 3.15 mm diameter
tube. Wambsganss and co-workers have studied boiling in tubes and rectangular sections having
hydraulic diameters of 2.4 to 2.92 mm with various fluids, this work is summarised in ref. [ 131.The
geometries of Lazarek and Black and Wambsganss and co-workers correspond to confinement
numbers of approximately 0.35. It was concluded that, except at very low heat flux, the boiling
mechanism was predominantly nucleate, since it was observed that the heat transfer coefficient
showed a strong dependence on heat flux, only a weak dependence on mass flux and was not a
function of quality. Studies at greater confinement [14, 151, examining heat transfer in very much
smaller gaps with confinement numbers of the order of 10 and above, concluded that evaporative
heat transfer was predominantly by conduction through a thin layer of liquid on the heated surface
and evaporation from that layer.
Film models, involving determination of the annular liquid film thickness and the heat transfer
through the film have not been widely used in the determination of boiling heat transfer coefficients
in large tubes. Since such an approach involves the determination of film thickness and the film
Nusselt Number using empirical methods, it is preferable to use an empirical correlation to
determine the heat transfer coefficient directly. However, based on tests and analysis of a typical
plate-fin section at low heat and mass fluxes, Robertson [16] demonstrated that the film flow model
could be applied in compact heat exchangers. It was shown that the film Nusselt Number, Nu,,
as defined below, was a function of a dimensionless mass flow rate and the relationship was close
to that given by Hewitt and Hall-Taylor [17].

The Nusselt Number, Nu,, is essentially the ratio of the actual rate of heat transfer through the
film, thickness 6, to that which would occur due to conduction through a film of the same thickness.
For laminar films, Nu, is unity.
In the study reported in this paper experimental results from a range of single tubes are compared
with predicted values of heat transfer coefficient from a number of published correlations. The
correlations applied, together with the appropriate equation numbers, are listed in Table 1.
The Liu-Winterton correlation, the form of which is given in equation (2), is widely used in the
prediction of flow boiling heat transfer coefficients and is appropriate for use in conventionally sized
tubes, It is based upon the assumption that there are two active mechanisms of heat
transfer-nucleate boiling and convection. The nucleate boiling contribution is calculated using the
 Predicting boiling heat transfer 709

correlation due to Cooper [18] with a suppression factor, S, applied. The convective component
is determined using the Dittus Boelter equation together with an enhancement factor, F.
Based on the assertion of Lazarek and Black [12] and Wambsganss and co-workers [13] that the
boiling in small diameter tubes would be predominantly nucleate, the results were tested against
the Cooper [18] correlation, equation (3), which was derived for nucleate pool boiling and the
correlation with a reduced constant, equation (4) which Cooper [19] tentatively proposed for
predicting the boiling heat transfer coefficient during flow boiling at low quality.
Lazarek and Black [12] proposed a correlation based upon boiling of a single fluid in a 3.19 mm
tube, this is presented as equation (5). A modified form has been suggested by the present authors
to allow for the observed increase in heat transfer coefficient with quality in the larger tubes tested
during the present study. The modified Lazarek and Black equation is given here as equation (6).
Wambsganss and co-workers [13] proposed a correlation, presented here as equation (7) in
which the dependency on the mass flux was eliminated by inclusion of the Weber number. It is
notable that the dependence of the heat transfer coefficient on the Weber number to the power
0.3 implies

which suggests that the heat transfer coefficient decreases with confinement. However, the
correlation was based upon tests spanning a small range of equivalent diameters.

FLOW REGIMES

Experimental observations have shown that, for boiling flow in small channels, the flow regimes
differ slightly from those observed in large channels and described in, for example, ref. 1201.Three
flow regimes are sufficient to describe the patterns observed. These have been defined as follows.

Isolated bubble jlow

Similar to bubble flow in large channels, bubbles detach from the nucleation sites and flow as
discrete units in the liquid, as the flow proceeds further heat addition results in an increase in the
number and size of the bubbles.

Confined bubble ,frow

Bubbles span the gap (in spaces confined in one dimension) or fill the channel (in spaces confined
in two dimensions), they are separated from the wall by a layer of liquid which evaporates and
causes the bubble to grow exponentially. The bubbles may be formed by isolated bubbles growing
or coalescing, alternatively single bubbles may reach a sufficient size to be regarded as confined
before becoming detached from their nucleation sites. This latter phenomenon has been observed
in relatively large tubes having polished surfaces and has been referred to as cavitation slug
flow [21]. A similar phenomenon also occurs in the boiling of liquid metals [22]. The rapid growth
of confined bubbles can lead to significant fluctuations in the pressure at entry to the channel and
apparent instability in multi-channel arrangements [23].

Annular-slug jlow
As the confined bubbles expand, liquid in the slugs between them is deposited on the channel
wall and the flow becomes basically annular with random, irregular slugs of liquid interspersed with
the vapour.
Determination values of the parameters governing the boundaries between the regimes was not
possible using the multi-channel flow visualisation test section since the flow was, as noted above,
frequently unstable with constant variations in the local conditions. It was also observed that, in
both the confined bubble and annular slug flow regimes, temporary local dry-out of the wall
occurred under some conditions. The three flow regimes, together with a representation of the
condition of partial dry-out, are illustrated in Fig. 3.
710 P. A. Kew and K. Cornwell

RESULTS

The results presented here relate to tests carried out with R141b working fluid. Less extensive
tests using water and Flutec PPl yield results which indicate the same general characteristics.
Sample results showing the measured variation in heat transfer coefficient with quality are given
in Fig. 4. It can be seen that the results in the 3.69 and 2.87 mm channels [Fig. 4(a) and (b)] follow
similar trends to those observed in conventionally sized channels. The heat transfer coefficient
increases with heat flux at low quality, while at higher qualities the heat transfer coefficient is a
function of quality and is essentially independent of heat flux. These trends are less apparent in
the smaller tubes. Results obtained at two values of mass flux for the 1.39 mm tube are shown in
Fig. 4(b) and (c). It can be seen that at high mass flux the heat transfer coefficient falls rapidly
with increasing quality.
The percentage deviations between the experimental values and the heat transfer coefficients
predicted by the six correlations listed in Table 1 are given in Table 2. It can be seen that all
correlations, with the exception of equation (4), perform reasonably well for the 3.69 mm diameter
tube. The deviation for all correlations increases with increasing confinement. The Liu and
Winterton correlation performs very poorly when applied to the 2.05 and 1.39 mm diameter tubes.
Comparisons of the predicted and measured heat transfer coefficients are presented for selected
cases as Fig. 5(a)-(f). Clearly the correlations which do take no account of mass flux or quality
result in a single value for each heat flux, while the experimental results span a considerable range
at each heat flux. Since a variation of heat transfer coefficient with quality has been observed,
neither equation (3) nor equation (7) can represent the physical reality under all conditions. The
modified Lazarek and Black correlation [equation (6)] does take into account an increase in heat
transfer coefficient with quality. Examination of Fig. 5(a) and (e) suggests that while there is
considerable scatter of the data around the predicted value for equation (3), equation (7) tends to
underpredict the data. The value predicted by equation (7) appears to form a lower bound to the
measured value. This may imply that the correlation predicts the nucleate boiling component of
the heat transfer but does not take into account an additional convective component.
Considerable scatter is apparent in the measured data when compared with all of the correlations
applied to the 1.39 mm diameter tube. It should be noted that since overheating of the tube
occurred at relatively low heat fluxes and qualities, all tests with this tube were at low quality and
low to moderate heat flux. These conditions are not generally conducive to obtaining high heat
transfer coefficients. The tendency to overheat combined with the observation that the heat transfer
coefficient fell rapidly with quality under some conditions, as illustrated in Fig. 4(c), suggests that
local dry-out occurred. This was due to the expulsion of the liquid in the tube by a rapidly
expanding bubble which ultimately filled the tube, dry-out then occurred and there was a rapid
increase in tube wall temperature until the liquid was replenished.
The relatively good performance of simple nucleate boiling correlations in the prediction of heat
transfer coefficients in single narrow tubes may be explained by consideration of the mechanisms
involved in pool boiling and in boiling in narrow tubes. Flow visualisation has shown that in
narrow channels discrete bubbles form which then grow to fill the width of the channel. These

t
Flow

B n

IB CB ASF PD
Fig. 3. Schematic diagram showing flow regimes: 0, vapour; ??, liquid; IB, isolated bubble; CB, confined
bubble; ASF, annular slug flow; PD, partial dry-out.
 Predicting boiling heat transfer 711

6 -

5 - 0

, 0 QA
4-A 0
A A
3%
PO0 SB
2* 00 O

3
1 -

0’
-0.05 0.15 0.35 0.55 0.75 0.95 -0.05 0.15 0.35 0.55 0.75 0.95

Dryness fraction Dryness fraction

(a) 3.69mmdiameter tube, G=212kg/m% (b) 2.87mm diameter tube, G=188kg/m%

6 -

OoO
3 - 0 0
. 0 CP
2 ,
;s”O
1
J
0
-0.05 0.15 0.35 0.55 0.75 0.95 -0.05 0.15 0.35 0.55 0.75 0.95

Dryness fraction Dryness fraction

(c) 1.39mm diameter tube, G=478kg/m% (d) 1.39mm diameter tube, G=1480kg/m%
Fig. 4. Variation of heat transfer coefficient with dryness fraction. (a) 3.69 mm diameter tube,
G = 212 kg/m%. (b) 2.87 mm diameter tube, G = 188 kg/m%. (c) 1.39mm diameter tube, G = 478 kg/
m’s, (d) 1.39 mm diameter tube, G = 1480 kg/m%.

confined bubbles then grow and move along the tube. The work of Aligoodarz and Kenning [24]
involved measurement of the variation in wall superheat during the passage of a sliding bubble
through a narrow channel. They observed that the temperature variations were similar to those
due to microlayer evaporation under pool boiling bubbles.
It is thus concluded that if the liquid film between the wall does not break down, a substantial
portion of the heat transferred is by conduction through this film and evaporation from the surface
of the film. The proportion of the tube length which is occupied by bubbles at any one time is

Table 2. Comparison of measured single-tube heat transfer data with selected correlations
Percentage Deviation
Tube Number of Equation Equation Equation Equation Equation Equation
description CO points (2) (3) (4) (5) (6) (7)
3.69 mm 4 0.33 146 21 2s 36 22 23 24.9
2.87 mm $ 0.42 141 42 21 46 19 19 32.1
2.05 mm I$ 0.59 182 118 30 43 47 48 37.3
1.39 mm 4 0.87 79 250 36 69 69 71 39.3
2.10 mm 0 0.57 I49 72 33 54 39 39
2.92 mm #* 0.34 15
@Xrcular section tube.
0 Square section channel.
* Dsta of ref. [ 13)
712 P. A. Kew and K. Cornwell

0 1 2 3 4 5 6 7 8 9 10 6 0 1 2 3 4 5 6 7 8 9 10
Measured beat transfer coefficient (kW/m%) Measured heat transfer coefficient (kW/m2K)
(a) 3.39mm diameter tube-prediction of equarion 3 (b) 1.39mm diameter tube-prediction of equarion 3

“E 10
5 9
‘g=: 8
._
I
iz
$ 6
,o
$ 5
B 4
e
: 3
2
G 2
.u
$1
._
B 0
I3 - 0 1 2 3 4 5 6 7 8 9 10 & 0 1 2 3 4 5 6 7 8 9 10
Measured heat transfer coefficient (kW/m2K) Measured heat transfer coefficient (kW/mZK)
(c) 3.39mm diameter tube-prediction of equation 6 (d) 1.39mm diameter tube-prediction of equation 6

2 10 5 10

$9 $9
; 8 : 8
iz” 7
.I 5
2.- 7

“0 6 % 6
8 8
E 5 & 5

a 4 a 4
2 3 $ 3
s 2 2 2
Pz 1 3‘; 1
a a *
E
& O 01234567 8 9 10 aP O01234567 8 9 10
Measured heat transfer coefficient (kW/mZK) Measured heat transfer coefficient (kW/m*K)
(e) 3.39mm diameter tube-prediction of equcarion 7 (f) 1.39mm diameter tube-prediction of equation 7

Fig. 5. Comparison of measured data with predictions. (a) 3.69 mm diameter tube-prediction of
equation (3). (b) 1.39 mm diameter tube-prediction of equation (3). (c) 3.69 mm diameter
tube-prediction of equation (6). (d) 1.39 mm diameter tube-prediction of equation (6). (e) 3.69 mm
diameter tube-prediction of equation (7). (f) 1.39 mm diameter tube-prediction of equation (7).
 Predicting boiling heat transfer 713

a function of the heat flux, which determines the propensity of bubbles to nucleate on the wall
and the net rate of vapour generation. The processes are thus similar to those occurring during
nucleate pool boiling, except that the bubbles are constrained during their expansion. It is therefore
logical that nucleate pool boiling correlations give reasonable results when applied to narrow tubes.

A NUCLEATE BOILING AND FILM FLOW MODEL

A model for the prediction of heat transfer coefficients to a fluid evaporating in a single channel
has been formulated. The model is based upon the observation that heat transfer to a fluid
evaporating in a narrow channel may be through one of four mechanisms:
1. Nucleate boiling.
2. Confined bubble boiling.
3. Convective boiling.
4. Partial dry-out
Note that these mechanisms do not correspond exactly with the flow regimes described in Section
4. In the isolated bubble regime it is reasonable to assume that heat transfer occurs entirely through
nucleate boiling, this is similar to the situation suggested by Cooper [18] for larger tubes. At any
point in the channel occupied by the confined bubble regime then either nucleate boiling or confined
bubble evaporation dominates. In the annular slug flow regime nucleate boiling or convective
boiling determines the heat transfer. The contribution of single-phase convection to liquid slugs
is assumed to be small throughout.
In testing the proposed model the heat transfer coefficient attributable to mechanism 1 was
obtained from an appropriate nucleate boiling correlation. Mechanism 2 was considered to be
active in the confined bubble regime and heat transfer is by conduction through the layer deposited
on the wall by the passing bubble and evaporation of the film. The film thickness was determined
using the data of ref. [25] as reported in ref. [7]. The Nusselt number was assumed to be unity and
the proportion of the surface covered by the liquid layer was taken to be equal to the void fraction
calculated for homogeneous flow.
Mechanism 3 was considered when the flow was in the annular slug flow regime with the liquid
phase deposited on the wall of the channel or passing through as liquid slugs; heat transfer was
assumed to be by conduction and convection through and evaporation of the liquid film.
Calculation of the heat transfer by mechanism 4 was similar to that for mechanisms 2 or 3, except
that at any given time a portion of the wall is dry and therefore heat transfer from that portion
is negligible. The film thickness in the annular slug flow regime was estimated using the CISE
correlation for void fraction [26], neglecting the volume of liquid in the slugs. The film Nusselt
number was then determined from the data presented in ref. [17].
In testing the model it was assumed that there was a minimum film thickness which could exist
on the wall. In the annular slug flow regime as the film thickness reduced there was a tendency
for the film to break up and dry areas form, this led to the occurrence of partial dry-out. Partial
dry-out also occurred if the film evaporated fully before being replenished, this has been observed
in the confined bubble regime.
The criterion for transition from confined bubble to annular slug flow was set as the dryness
fraction at which the film thickness calculated for confined bubble flow, which increased with
dryness fraction, was equal to the film thickness calculated for annular slug flow, which decreased
with increasing dryness fraction. Thus, the selected flow regime was that which yielded the thinner
film of liquid on the wall.
The approach used in the annular slug flow regime would lead to a film thickness approaching
zero at as the dryness fraction approached unity. .This in turn would result in the heat transfer
coefficient approaching infinity. In practice, during evaporation at high values of dryness fraction,
the heat transfer coefficient must reach a peak and then sharply declines, due to partial dry-out
of the walls. During the flow visualisation studies [l] intermittent local dry-out was noted at low
dryness fraction and moderate heat flux in the confined bubble regime. This was due to complete
evaporation of areas of the liquid film deposited beneath the confined bubbles. The film flow model
proposed has not yet been modified to account for the phenomenon of partial dry-out in either
714 P. A. Kew and K. Cornwell

. . . . . . . . . . . . . . Cooper (Eqn. 3)
9
-------- Film model
2, - Predicted htc.
“f t
Experimental

Mass flux 1EOg/m*s

Heat flux 36 kWlm*
I , I . I. I, I * , , I. 1 1 I,

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Dryness fraction
Fig. 6. Application of the proposed model to sample set of data.

regime. A simple approach involving specification of a minimum film thickness, below which the
film would be expected to break up, leaving dry areas through which there would be negligible heat
transfer, has not yielded satisfactory results.
This model correctly explained the trends observed for many of the measured data sets, an
example of which is shown as Fig. 6. However, in many cases, particularly in the smaller tubes,
the model significantly overpredicted the heat transfer coefficient. Examination of the experimental
results, examples of which are presented Fig. 5, shows that in the smaller diameter tubes which
were tested, the measured heat transfer coefficient frequently fell below that predicted for nucleate
boiling alone, even at low dryness fraction, implying that intermittent dry-out occurred. It is
suggested that the model and the correlations applied provide a basis for the prediction of boiling
heat transfer coefficients in narrow tubes, however, intermittent dry-out occurs under a wider range
of conditions than expected and the model must be refined to account for this.

CONCLUSIONS

Existing flow boiling heat transfer correlations do not perform well when applied to narrow
channels having a confinement number of the order of 0.5 and above. Nucleate boiling type
correlations yield better results in spite of not taking account of the convective component of heat
transfer or the tendency for intermittent local dry-out to occur in confined tubes. A model based
upon film evaporation and nucleate boiling has yielded promising results but does not yet account
satisfactorily for intermittent dry-out.

Acknowledgements-The authors gratefully acknowledge the financial support received from DC&XII of the European
Commission through Contract No. JOU2-CT92-0045, in the framework of the JOULE II Programme.

REFERENCES

1. K. Cornwell et nl., Compact two-phase heat exchangers, Final Report, CEC Contract No. JOU2-CT92-0045 (1995).
2. K. Cornwell and P. A. Kew, Boiling in small parallel channels. In Energy &ficiency in Process Technology, Edited by
P. A. Pilavachi. Elsevier Applied Science, London, pp. 624638 (1993).
3. K. Cornwell and P. A. Kew, Evaporation in micro channel heat exchangers. Proc. 4th UK National Cot~f. on Heat
Tran$fer, Manchester, UK, pp. 289-294. 1.Mech.E (1995).
4. J. C. Chen, A correlation for boiling heat transfer to saturated fluids in convective flow, ASME Paper 63-HT-34. 6th
International Heat Transfer Conf., Boston, MA (1963).
5. M. M. Shah. Chart correlation for saturated boiling heat transfer: equations and further study. Trans. ASHRAE 88,
185-196 (1982).
 Predicting boiling heat transfer 715

6. Z. Liu and R. H. S. Winterton. A general correlation for saturated and subcooled flow boiling in tubes and annuli
based on a nucleate boiling equation. Int. J. Heat Mass Tram&v 34, II, 2759-2766 (1991).
7. G. B. Wallis, One-dimensional Two-phase Flow, pp. 288-302. McGraw-Hill, New York (1969).
8. G. B. Wallis and S. Makkenchery. The hanging film phenomenon in vertical annular two-phase flow. Trans. ASME
J. Fluids Engng 96, 3, 297-298 (1974).
9. 1. I. Chernobyl’skii and 1. M. Tananaiko. Heat exchange during boiling of liquids in narrow annular tubes. Sorier Phrs.
Tech. Ph,vs. 1, 22442249 (1956).
10. E. Ishibashi and K. Nishikawa. Saturated boiling heat transfer in narrow spaces. li?r. J. Heat Mass Transf>r 12,863-894
(1969).
I I. D. L. Bondurant and J. W. Westwater. Performance of transverse fins for boiling heat transfer. C/rem. Engng Prog.
Synp. Series 67, 113, 30-37 (1971).
12. G. M. Lazarek and S. H. Black. Evaporative heat transfer, pressure drop and critical heat flux in a small vertical tube
with RI 13. Int. J. Hear h4as.y Transftir 25, 7, 945-960 (1982).
13. T. N. Tran, M. W. Wambsganss and D. M. France, Boiling heat transfer with three fluids in small circular and
rectangular channels, Argonne National Laboratory, Report ANL-95-9. NTIS, Springfield. VA (1995).
14. S. G. Bankoff and T. E. Rehm. Convective boiling in narrow concentric annuli. Trans. ASME J. Engng Gas Turbines
Pon,er 112, 607-613 (1980).
15. K. Moriyama and A. Inoue. The thermohydraulic characteristics of two-phase flow in extremely narrow channels (Parts
1 and 2). Heat Tray/& Jap. Res. 21, 8, 823-856 (1992).
16. J. M. Robertson, The correlation of boiling coefficients in plate-fin heat exchanger passages with a film-flow model.
7th Int. Heat Trans Conf:, Munich, Vol. 6, pp. 341-345 (1982).
17. G. F. Hewitt and N. S. Hall Taylor, Annular Tlt,o-phase Flon,. p. 194. Pergamon Press, Oxford (1970).
18. M. G. Cooper, Saturated nucleate pool boiling-a simple correlation. Is/ UK National Hear Trans/& Con/: IChemE
Symp. Series No. 86. Vol. 2, pp. 785-793 (1984).
19. M. G. Cooper. Flow boiling-the apparently nucleate regime. In/. J. Heat Mass Transfer 32, 4591164 (1989).
20. J. G. Collier, Convective Boiling and Condensation, 2nd Edition, pp. 8-14. McGraw-Hill, New York (1981).
21. Y. Y. Hsu and R. W. Graham, A visual study of two-phase flow in a vertical tube with heat addition, NASA-TN-D
1564 (1963).
22. Y. D. Morozov et al.. Mechanism of the transition to two-phase flow and flow regimes in the boiling of liquid metals
in a once through steam boiler. Translated from Teplofizik V.vsokikh Temperature 26(6), 1163-l 170 (1988).
23. P. A. Kew and K. Cornwell, On pressure fluctuations during boiling in narrow channels. l4f/r UIT Na/ional Hear
Transfer Corzj:, Rome, Edizioni ETS, Pisa, pp. 1323-1330 (1996).
24. M. R. Aligoodarz and D. B. R. Kenning. Vapour bubble behaviour in a single narrow channel. Proc. 4tlt UK Narional
Conf: on Heat Transfbr, Manchester, UK, pp. .273-276. 1.Mech.E (1995).
25. G. 1. Taylor. Deposition of a viscous fluid on the wall of a tube. Fluid Mechanics IO, 2, 161-165 (1961).
26. A. Premoli, D. Francesco and A. Prina, An empirical corrlation for evaluating two-phase mixture density under
adiabatic conditions. European Tlt,o-phase Flow Group Meeting, Milan. ENEA, Paper B9 (1970).