Paper No 7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
Droplet Entrainment in Churn Flow
MasroorAhmad*, Deng J. Pengt, Colin P. Halet, Simon P. Walker* and Geoffrey F. Hewittt
Department of Mechanical Engineering, Imperial College London, SW7 2AZ, UK
t Department of Chemical Engineering, Imperial College London, SW7 2AZ, UK
m.ahmad06@imperial.ac.uk and g.hewitt@imperial.ac.uk
Keywords: entrainment, churn flow, axial view photography
Abstract
Chum flow is an important intermediate flow regime between slug and annular flow. In the Taylor bubbles which are
characteristic of slug flow, there is a falling film at the tube wall with an upwards gas flow in the core. The transition from slug
flow to chum flow occurs when the conditions in the Taylor bubbles occurring in slug flow are such as to promote flooding
(Jayanti and Hewitt, 1992). Chum flow is a region in which there are large interfacial waves travelling upwards with falling
film regions between the waves; the transition to annular flow occurs when the (film) flow becomes continuously upwards. In
both chum flow and annular flow, a (sometimes large) proportion of the liquid flow is in the form of entrained droplets in the
gas core. In annular flow systems with evaporation, the entrainment of droplets from the film surface plus the evaporation of
the film may not be sufficiently offset by droplet deposition and film dryout occurs. Dryout can be predicted provided
expressions can be invoked for local entrainment, deposition and evaporation rates and these expressions integrated to
establish the conditions under which the film flow rate becomes zero (Hewitt and Govan, 1990). This integration procedure
needs a boundary value for entrained droplet flow rate at the onset of annular flow (i.e. at the transition between churn flow
and annular flow) and the predictions obtained for dryout may be sensitive to this boundary value for short tubes. Some data
and a correlation for the entrained fraction at the onset of annular flow were obtained by Barbosa et al. (2002) but this
correlation applies only to adiabatic and equilibrium conditions. The work described in this paper is focused on the question of
droplet entrainment in churn flow and the development of a methodology for predicting the entrained fraction at the onset of
annular flow for diabetic (i.e. nonequilibrium) systems. New expressions are given which allow the integration procedure to
be extended to cover both chum flow and annular flow. New experiments were also conducted on the churnannular region
using the axial view photography technique. This allowed the entrainment processes to be visualized starting at the onset of
chum flow and passing into the annular flow regime.
Introduction
In gasliquid mixture flows in vertical pipes, the phases
distribute themselves in a variety of spatial and temporal
distributions, referred to as flow regimes or flow patterns. A
wide variety of names have been given to these phase
distributions; a reasonably well accepted set of descriptions
for vertical upflow is as follows: bubbly flow, slug flow,
chur flow and annular flow. These flow regimes and
corresponding transitions between them are of immense
importance in predicting the transition of a regime of high
heat transfer coefficient to one of greatly reduced heat
transfer coefficient; this transition is referred to by a
number of names, including dryout, critical heat flux (CHF)
and boiling crisis. The accurate prediction of this transition
is of great importance not only in design and safe operation
of nuclear power plants but also many other types of
industrial heat transfer equipment.
Annular film dryout is arguably the most important
mechanism for the onset of reduced heat transfer coefficient.
In annular flows, liquid is lost from the film as a result of
droplet entrainment and evaporation, and it is gained by the
film by droplet deposition rate, which leads to liquid film
dryout. In the annular flow model, the equations for
entrainment, deposition and evaporation are integrated from
the onset of annular flow (i.e. from the churannular
transition); when the film flow rate is predicted to be zero,
then dryout is predicted to occur. However, this integration
process requires an initial value for entrained fraction (IEF)
at the chur flowannular flow transition; the IEF could, in
principle vary between zero and unity. It was observed that
dryout prediction may be a strong function of IEF at the
onset of annular flow especially at high liquid mass fluxes.
In order to solve the problem of IEF at the churannular
transition, an understanding of droplet entrainment in chur
flow is vital.
As noted by Barbosa et al, (2001b), the word 'CHURN' is
used by different research groups to describe different flow
types. Thus, Zuber & Findley (1965) describes
churnturbulent flow as a type of bubble flow whereas Taitel
et.al (1980) considered it to be a developing slug flow.
However, the most widely accepted definition of churn flow
is that of Hewitt & Hall Taylor (1970) who considered it to
be an intermediate regime between slug and annular flow.
Paper No
Chum flow occurs due to break down of slug flow due to
flooding of the liquid film in the Taylor bubble (Jayanti and
Hewitt, 1992; McQuillan, 1985). It is characterized by large
interfacial waves with flow reversal between the waves. The
highly oscillatory liquid film in chum flow is accompanied
a continuous gas core containing considerable amount of
entrained liquid. Ultimately as the gas flow rises the
periodic downward flow of liquid film ceases and gives rise
to a unidirectional annular flow. This flow reversal point is
described in terms of dimensionless superficial gas velocity
by (Hewitt and Wallis, 1963; Wallis, 1969) as:
UGS UGS PG
gd(pL PG)
Although extensive data for pressure gradient and liquid
holdup are available for chum flow and the corresponding
transition regions (McQuillan, 1985; Govan, 1991; Barbosa,
2001a), experimental data regarding film thickness and
droplet entrainment behaviour is quite scarce. This might be
due to the reason that conventional measurement techniques
for film thickness and entrained liquid flow in annular flow
could not be applied to chum flow due to its chaotic nature.
Wallis et.al (1962) carried out entrained fraction
measurements in the chum and churnannular transition
region by using a 12.7 mm internal diameter vertical tube
fitted with single axially located sampling probe. The results,
shown in Figure 1, indicated a considerable amount of
liquid entrained as droplets in churn flow; the entrained
fraction passes through a minimum around the
churannular transition.
0
30

20
1: 0
z
I
z
_o
_a
DIMENSIONLESS SUPERFICIAL GAS VELOCITY
0.0 2.0 4.0 6.0
I I
0
O
* OE
%, a
** **
0 "
So
0
Superficial liquid
velocity [mIs]
0 0.042 (upflow)
o 0.042 (downflow)
0.084 (upflow)
O 0.084 (downflow)
0.147 (upflow)
[] 0.168 (downflow)
0.0 20.0 40.0 60.0
SUPERFICIAL GAS VELOCITY [m/sl
Figure 1: Variation of entrained fraction with gas velocity
(Wallis, 1962).
Recently, similar sorts of experiments encompassing a wider
range of liquid flow rates and pressure, were carried out by
Barbosa et.al (2002) using an isokinetic probe technique in
a 31.8mm internal diameter and 10.8 meter long vertical
tube facility. The results are presented in Figure 2 & 3 and
show that the liquid entrained fraction decreases with
increasing gas velocity in chur flow, passes through a
minimum around the transition to annular flow and
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
increases again in the annular flow region. It is also worth
noting that entrained fraction in these regions is a function
of liquid mass flux and extrapolation of these results reveals
a very high proportion of liquid entrained as droplets at the
onset of churn flow. Barbosa et al (2002) proposed the,
following correlation for IEF at onset of annular flow:
E.0 (in %)= 0.95+342.55 PL d
PGGG (97
e 40
z
2
I
LaJ
Z
t io
I
0.40 0.80 1.20
DIMENSIONLESS GAS VELOCITY
Figure 2: Liquid entrained fraction as a function of the total
liquid mass flux: p = 2 bara (Barbosa, 2002).
0.40 0.80 1.20 1.60
DIMENSIONLESS GAS VELOCITY
Figure 3: Liquid entrained fraction as a function of the total
liquid mass flux: p = 3.6 bara (Barbosa, 2002).
This paper addresses the problem of IEF at onset of annular
flow in the context of dryout predictions using an annular
flow dryout model. Firstly, pictures of churn flow and the
corresponding transitions, obtained by an axial view
photography method, were analyzed. On the basis of visual
observation and the available (Barbosa et al, 2002) data, a
new methodology for entrainment rate calculation in chur
flow is proposed. Also, the annular flow dryout model is
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
extended to chur flow i.e. the integration process is made
to start from onset of chur flow; this leads to better
prediction of dryout data.
Nomenclature
C concentration of droplets (kg m 3)
D deposition rate (kg m2 sec 1)
d diameter (m)
E entrainment rate (kg m2 sec 1)
Ef entrained fraction ()
G mass flux (kg m2 sec'1)
g gravitational constant (m sec 2)
kD deposition coefficient (m sec ')
p pressure (bar)
q heat flux (W m 2)
UGS superficial gas velocity (m sec ')
ULS superficial liquid velocity (m sec ')
UG dimensionless superficial gas velocity
Greek letters
P density (kg m 3)
C7 surface tension (kg sec 2)
Yr viscosity (kg m1 sec'1)
Figure 4: Schematic diagram of LOTUS air water facility
Subscripts
L
LE
LF
LFC
G
liquid
entrained liquid
liquid Film
critical liquid film
Reflect mirow
Experimental Facility
In view of importance of flow regimes and corresponding
transitions between these flow regimes in prediction of the
dryout position and postdryout behaviour, axial view
photography experiments have been carried out on the
LOTUS (LOng TUbe System) facility at Imperial College
London.
The LOTUS system is a vertical airwater two phase facility
with a test section consisting of a 10.41m long vertical
acrylic (PerspexTM) tube with an internal diameter of
0.032m. Water is supplied from a separation vessel by two
pumps with 1.5kw and 8.0kw power. Air is taken from the
Imperial College site mains supply and the airwater
mixture leaving the test section is separated using a cyclone,
with the air being discharged to atmosphere and the water
being returned to the separation vessel. An axial viewer was
fitted to the top of the test section; essentially, this consists
of a high speed video camera (Olympus iSPEED 3) which
is focused at an illuminated plane at a distance of 470 mm
from the end of the tube. The water flow is diverted away
from the camera window using an arrangement of air jets.
The camera can then capture events occurring at the tube
cross section in the illuminated plane. The principle of this
method is described, for instance, by Hewitt and Whalley
(1980). The schematic diagram of LOTUS facility and axial
view photography setup is shown in Figure 4 & 5. The test
conditions were adjusted to cover the churn and annular
flow regions and the transition between them.
Visualize position
3rd floor
///////////
Figure 5: Axial view photography setup at LOTUS
Results and Discussion
IEF Problem:
The problem of initial entrained fraction (IEF) input for the
annular flow dryout model at high mass fluxes is a long
standing one. Information on the amount of entrained liquid
at the transition from chur to annular flow boundary is
essential as it is the starting point of integration of
entrainment/deposition processes in a heated channel. In
order to demonstrate the problem, the annular film dryout
model (as described by Hewitt and Govan, 1990 and
Paper No
Paper No
embodied in the Imperial College computer code GRAMP)
is applied to the uniform and nonuniform heated tube
dryout data of Bennett et al. (1966) and Keeys et al. (1971).
The results, as indicated in Figure 6&7, confirmed that at
medium to high liquid mass fluxes, dryout location is a
strong function of IEF at onset of annular flow. The analysis
also revealed that high IEF values, normally in range of
0.650.95, are required to predict the dryout data at these
high liquid flow rates.
48
46 A
44
42 *
S40
38
36
34
U % (Run 5358_G=380 kg/m2 /sec)
32 * % (Run 5273_G=1020 kg/m2/sec)
30 % (Run 5374_G=3850 kg/m2/sec)
00 02 04 06 08 10
Initial Entrained Fraction ()
Figure 6: IEF effect on the dryout predictions in uniformly
heated tubes dryout data
E
0
C3
J
28
26
24.
22
20
Keeys141kw_G=720 kg/m2/sec
18 Keeys191kw_G=2000 kg/m2/sec
00 02 04 06
Initial Entrained Fraction
08 10
Figure 7: IEF effect on the dryout predictions in
nonuniformly heated tubes dryout data
The same phenomena could be demonstrated by
investigating the effect of IEF on dryout qualities, as
illustrated in Figure 8&9. At low liquid flow rates, the IEF
effect is minimal but at high flow rates dryout quality
increases considerably with increasing IEF at onset of
annular flow.
Also, in the present study, the only available IEF correlation
of Barbosa et.al (2002) was employed initially in the
annular flow dryout model to predict the dryout data. The
correlation yielded IEF values of 0.2 to 0.3 at the onset of
annular flow and these values were much lower than the
entrained fractions required to fit the dryout data for
medium and high mass fluxes using the annular flow dryout
model. A major problem with the Barbosa et al (2002)
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
model is that it is essentially for nearequilibrium flows.
However, in heated channels, the flows will not be in
equilibrium. A "memory" of the high entrained fractions at
the onset of chum flow will persist since the entrained
fraction does not immediately relax to the equilibrium
values as the churn flow regime is traversed along the
channel as a result of the increase in quality due to
evaporation. What are needed, therefore, are local values of
entrainment rate and deposition rate in chum flow so that
the same type of integration can be carried out in chum flow
as is done in the annular flow model for annular flow.
) Dryout Point .: 
S   IEF =0.9
" . IEF = 0.7
S ..... IEF = 0.5
.IEF =0.3
.. ...... IEF = 0.1
S Onset of annular Flow
005 000 005 010 015 020 025 030 035 040
Quality
Figure 8: IEF effect on the dryout quality predictions at
high liquid mass fluxes (G = 3850 kg m2 sec1)
10
08
*5 06
u
0
04
02
00
00 02 04 06 08 10
Quality
Figure 9: IEF effect on the dryout quality predictions at low
liquid mass fluxes (G = 380 kg m2 sec1)
Experimental Results:
In view of role of droplet entrainment behaviour in churn
and chumannular transition region, axial view photography
experiments were carried out at LOTUS facility. The visual
evidence, as shown in Figure 10, also supported the
experimental data of Barbosa et.al (2002) i.e. the amount of
liquid entrained is high as the (chaotic) chum flow regime is
entered and it passes through a minimum around the
churnannular transition, leading to an increase in entrained
fraction with increasing gas velocity in annular flow.
Paper No 7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
predictions. In the absence of detailed data, it was assumed
that the droplet deposition rate in churn flow could be
predicted from a model identical to that used for annular
flow. Specifically, the model suggested by Hewitt and
Govan (1990) (given in AnnexA) is used. Thus, in
equilibrium chum flow, it is assumed that:
Echurn = Dchurn = (kD )annular C (3)
For the equilibrium chum flow data, (such as that of
Barbosa et al, 2002), the value of C (the concentration of
drops in the gas core in kg/m3, calculated on the assumption
of a homogeneous core flow) is known and (kD) .annular can be
calculated from the equations given in AnnexA, Thus,
Echurn can be calculated. Since C is higher in chum flow
a) than would be expected in annular flow (and increases with
decreasing gas flow rate) it follows that Echurn follows a
similar trend (Figure 11 a,b,c). A simple correlation which
gives an approximate representation of the data calculated
churn = 8.73U + 9.73
Eannular,local (4)
where Eannuar,loca1 is the value of entraimnent rate for annular
flow calculated for annular flow from the relationships
given in AnnexA and for the local values of gas and liquid
flow rates. Thus, Equation 4 is a representation of the
enhancement of entrainment rate above that for annular
flow in the chum flow region. The enhancement factor
becomes unity at UG=1.
036
0 34  GL=22 kg/m2/sec (a)
032. GL=48 kg/m2/sec
S030. GL=126kg/m2/se
S028. GL=218 kg/m2/se
024 GL=331 kg/m /se
E 022
0207
018
016'
014.
E 012
a 010*
0 08
0 06
002
000
02 04 06 08 10 12 14 16 18
UG ()
040
(b)
035 
030
E 025
2(c) 020
Figure 10: Axial view photography experiments undertaken
at superficial liquid velocity ULS = 0.165 m/s in (a) Chum 01 GL23kg/
E GL=23 kg/m2 /sec
Flow (UG = 0.461), (b) ChumtoAnnular Flow transition I 010 GL=48 kg/m2/sec
(UC = 0.968) and (c) Annular Flow (U = 1.066) 005 GL=128 kg/m /sec
S05 GL=215 kg/m2/sec
GL=330 kg/m /sec
Entrainment Rate in Churn Flow: 000o 2 0304 06 0708091011121314151617
UG ()
In order to predict the IEF at the onset of annular flow, it is
necessary to have a model for the chum flow region. Even
an approximate model would in principle allow the IEF to
be predicted accurately enough to give better dryout
Paper No
(c)
u b ) GL=23 kg/m2/sec
045 GL=48 kg/m2/sec
040 GL= 128 kg/m2/sec
0 35 GL=214kg/m2/sec
.3\ GL=302 kg/m2/sec
0 30 
0 25
020 \v 
015 \
0 10 \
005
0 00 
04 06 08 10 12 14 16
UG ()
u 1. 
036
0 34
S032
E 030
0 28
026
5 024
022
L 020
018
016
06 07 08 09 10 11 12 13 14 15
UG ()
Figure 11: Entrainment rate variation in chum and annular
flow (a) p = 2 bar (b) p = 3.5 bar (c) p = 5 bar (d) Liquid
Mass Flux = 215 kg m2 sec1
The application of this methodology supported the idea,
depending upon different flow conditions, that the variation
of entrained droplet flow with length in heated tubes ceases
to follow the equilibrium curve; rather, the departure from
equilibrium leads to maintenance of high values of
entrained fraction at start of annular flow as indicated in
Figure 12.
Dryout Point
S .... ......
S... .... . ",
S
  q = 5 94E05 W/m'
.* q = 7 2E05 W/m2
  q = 8 8E05 W/m2
..... q = 1 OE06 W/m2
  q = 1 2E06 W/m2
Onset of Churn Flow
1v
00 02 04 06 08 10
Quality
Figure 12: Churn flow methodology (G = 654 kg m2 sec1)
In order to complete the proposed dryout model, an initial
value of entrained fraction at the start of chum flow is still
required. In order to test the proposed methodology,
predictions were made of dryout data for uniformly and
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
nonuniformly heated tube (Bennett, 1966; Keeys, 1971)
with IEF at the start of chum flow chosen to be 0.9. The
predictions, as indicated in Figure 13, of dryout data are
encouraging i.e. within 20%.
Measured Dryout Location
Figure 13: Prediction of dryout data
(IEF for chum flow = 0.9)
Work is continuing on searching for an improved correlation
for the IEF in chum flow. However, the approximate
approach taken has already yielded much more realistic
results, particularly for high mass fluxes.
Conclusions
In the present study, the behaviour of droplet entrainment in
churn flow is analysed with particular reference to the
prediction of the value of entrained fraction at the start of
annular flow. For this purpose, new equation for
entrainment rate in chum flow is proposed, which allows
the dryout model to start integration of entrainment and
deposition processes from the start of chum flow rather than
the start of annular flow. The prediction of dryout location,
for both uniform and nonuniform heated tubes cases, by
employing the new proposed methodology led to improved
results i.e. within 20%.
Acknowledgements
This work was carried out as part of the TSEC programme
KNOO, and we are grateful to the EPSRC for funding under
Grant EP/C549465/1. One author (M. Ahmad) would like to
acknowledge the Higher Education Commission (HEC) of
Pakistan and Pakistan Institute of Engineering & Applied
Sciences (PIEAS) for funding his PhD studies at Imperial
College London.
References
Barbosa J.R., Govan A.H., Hewitt G.F., Visualisation and
modelling studies of chum flow in vertical pipe, Int. J.
Multiphase Flow, Vol. 27, Issue 12, pp. 21052127, (2001a)
Barbosa J.R., Hewitt, G.F., Konig G. and Richardson S.M.,
Liquid entrainment, droplet concentration and pressure
gradient at the onset of annular flow in a vertical pipe, Int. J.
Multiphase Flow, Vol. 28, pp943961, (2002)
Barbosa, J.R., Richardson, S., Hewitt, G.F, Churn flow:
(d) P= 2 bar
P= 3 5 bar
P= 5 bar
Paper No
myth, magic and mystery. In: 39th European TwoPhase
Flow Group Meeting, Aveiro, Portugal, 1820 June, (200 lb)
Bennett A.W., Hewitt G.F, Kearsey H.A., Keeys, R.K.F,
Heat transfer to steam water mixtures flowing in uniformly
heated tubes in which the critical heat flux has been
exceeded. AERER5373, (1966).
Govan, A.H., Hewitt, G.F., Richter, H.J., Scott, A., Flooding
and churn flow in vertical pipes. Int. J. Multiphase Flow 17,
2744 (1991)
Hewitt G.F, Flow regimes: Transitions and flow behaviour.
Multiphase Science and Technology, 15:131143 (2003)
Hewitt G.F and Govan A.H., Phenomena and prediction in
annular twophase flow: Invited Lecture, Symposium on
Advances in GasLiquid Flows, Dallas, November, 1990 (at
Winter Annual Meeting of ASME) ASME Volume
Reference FEDVol. 99, HTD Vol. 155 pp 4156, (1990)
Govan A.H. and Hewitt G.F, Phenomenological modelling
of nonequilibrium flows with phase change. Int. J. Heat
Mass Transfer, 33:22942 (1990)
Hewitt, G.F, HallTaylor, N.S., Annular GasLiquid Flow.
Pergamon Press, Oxford (1970)
Hewitt G. F. and Jayanti S., Prediction of the slugtochurn
flow transition in vertical twophase flow, Int. J. Multiphase
Flow, Vol. 18, pp 847860, (1992)
Hewitt, G.F, Wallis, G.B., Flooding and associated
phenomena in falling film in a vertical tube. In: Proceedings
of MultiPhase Flow Symposium, Philadelphia, PA, 1722
November, pp. 6274 (1963)
Hewitt, G.F, and Whalley, PB., Advanced optical
instrumentation methods. Int. J. Multiphase Flow, Vol. 6,
No. 12 p136156. (1980)
Keeys, R.F.K., Ralph, J.C., & Roberts, D.N., Post burnout
heat transfer in high pressure steamwater mixtures in a tube
with cosine heat flux distribution, AERER6411 (1971).
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
AnnexA
Hewitt and Govan droplet entrainment and
deposition correlation
A method for calculating the entrainment and deposition
rates has been proposed by Hewitt & Govan (1990). These
are as follows.
0 ;if GLF
0 316
E= 5.75x10 5GG L(GLF GLF)2 if GL >GLF,
(Al)
where GLFC is a critical liquid film mass flux given by,
GLC L =exp 5.8504+0.4249 PLG
d )l AL PG U
The droplets deposition rate is expressed as,
D= kC
where C is the concentration of droplets in the core,
GLE
0 __ LE
GG + GLE
+
PG PL
and the droplet deposition transfer coefficient, kd,
by,
(A2)
(A3)
(A4)
is given
0.185 r ;if C < 0.3
PGd PG
0.083 0 ;if 2 0.3
SpGd ) pG
McQuillan, K.W., Whalley, P.B., Hewitt, G.F., Flooding in (A5)
vertical twophase flow. Int. J. Multiphase Flow 11,
741760 (1985)
Taitel, Y, Barnea, D., Dukler, A.E., Modelling flow pattern
transitions for steady upward gasliquid flow in vertical
tubes. A1ChE J. 26, 345354 (1980)
Wallis, G.B., The onset of droplet entrainment in annular
gasliquid flows. General Electric Report No. 62GL127
(1962)
Wallis, G.B., OneDimensional TwoPhase Flow.
McGrawHill, New York (1969)
Zuber, N., Findlay, J.A., Average volumetric concentration
in twophase flow systems. J. Heat Transfer 87, 453468
(1965)
