Int. J Multiphase F/o.. Vol 4, pp. 33-39 Pergamon/Elsevier. 1978. Printed in Greal Brilain.

CORRELATION OF THE LIQUID VOLUME

F R A C T I O N IN T H E S L U G F O R
HORIZONTAL GAS-LIQUID SLUG FLOW

G. A. GREGORY,M. K. NICHOLSONand K. Azlz

Department of Chemical Engineering,The Universityof Calgary,Calgary, Alberta, Canada

(Received 16 March 1977; receivedfor publication 1 September 1977)

Abstract--Experimentaldata for gas holdup in liquid slogs are reported for two differentpipe sizes (2.58cm
and 5.12 cm I.D.). A simpleempiricalcorrelationis developedand is shown to be a significantimprovement
over the only other publishedcorrelationproposed by Hubbard (1965). The results of this investigationare
important for the developmentof a mechanistic model for the prediction of pressure drop and holdup for
slog flow in pipes.

INTRODUCTION
Two principal attempts to provide mechanistic models for horizontal gas-liquid slug flow have
appeared in the literature. The first of these due to Kordyban (1961) and Kordyban & Ranov
(1963) was not very successful due to several simplifying assumptions inherent in these authors'
analysis.
Primarily, it was assumed that liquid slugs moved at the gas velocity and slipped over the
top of a slower moving liquid film. Agreement between their model and experimental data was
generally poor.
In the second study, Hubbard (1965) and Dukler & Hubbard (1975) presented a considerably
more detailed analysis, in which it was noted that:
(a) the liquid slug scooped up the liquid film ahead of it and redeposited a new film in its
wake; and
(b) in the process of the scooping action, some gas entrainment occurred within the slug.
In the Kordyban (1961) model, any such entrainment was neglected, and the density of the
fluid in the slug was taken simply as that of the liquid. However, in the Dukler & Hubbard
(1975) model, the pressure drop calculation is based on the actual density of the mixture in the
high velocity slug section, and the effect of the entrained gas is accounted for by the use of a
mixture density defined as,

pM = PLELs+ pc(l - ELs)

= pLELs [1]

where pM is the density of the two phase mixture, and ELS is the liquid volume fraction in the
liquid slug.
Unfortunately, no theoretical method is available for the prediction of ELs and an empirical
correlation was sought.
Hubbard (1965) attempted to measure ELs through the use of an impact probe system. This
proved to be very dit~cult and the results he obtained showed little consistency and reproduc-
tibility was poor. Consequently, only a very crude correlation was proposed which has
represented a basic weakness in the overall Dukler & Hubbard (1975) model.
It should be noted however, that a modelling study for horizontal slug flow currently in
progress at The University of Calgary, has confirmed the need for a reliable method of
predicting ELs.

33
MF VoL 4. No. I--C
34 G. A. GREGORY. M. K. NICHOLSON and K AZIZ

THIS STUDY
Experimental values of ELs for this study were obtained using the capacitance-type liquid
volume fraction sensors that have been described in an earlier paper by Gregory & Mattar
(1972). These sensors used in conjunction with an IKOR Model 545 Vapor-Liquid meter,t
provided a continuous record of the in situ liquid volume fraction. They have a sufficiently fast
response that the average liquid volume fraction may be recorded for each slug as it passes
through the approx. 15 cm long sensor. The average lengths of the liquid slugs have been
observed by the authors to vary between a factor of 10 and 125 times the length of the sensor
and thus, the measurements may be considered as representative averages rather than as single
point determinations. The sensing elements are located on the outside wall of the pipe and thus
do not cause any disruption to, or interference with, the flow.
Several series of experiments were performed Using a light refined oil, with air as the gas
phase. The density and viscosity of the oil were 858 kg/m 3, and 6.75 mP sec respectively, both
measured at the average system temperature of 23°C. Two horizontal test sections were used,
having inside diameters of 2.58 cm and 5.12 cm respectively. In the 2.58 cm test section, a length
of approx 575 pipe diameters was provided upstream of the liquid volume fraction sensor to
minimize entrance effects from the flow. In the 5.12 cm test section, the entrance length was
approx. 340dia. The total length of the flow system is approx. 30 m, and thus the measuring
points were also located well upstream of the point where the flow leaves the test section and
enters the separator. The average mid-point pressure, which was maintained relatively constant
for all flow rates investigated, was 345 kPa for the 2.58 cm pipe, and 255 kPa for the 5.12 cm
pipe. The range of flow rates investigated, expressed in terms of the liquid and gas superficial
velocities, VsL and VsG respectively, were:

0.030 ~ Vst <- 2.316 m/sec;

0.088 -< Vs~ <- 15.376 m/sec.

These ranges cover virtually the entire region of slug flow that can be observed in the flow loop.
A detailed description of the overall flow loop is reported elsewhere (Aziz & Gregory, 1976).

RESULTS AND DISCUSSION

The measured values of ELs and the corresponding phase velocities obtained with the
2.58 cm I.D. pipe are given in table 1, while table 2 contains the data for the 5.12 cm I.D. test
section. It should be noted that each value of ELs reported in these tables actually represents an
average for approximately 30 slugs observed at each set of flow rates. However variations in the
values of ELs for individual slugs were small, and the standard deviations for the samples were
generally about 0.01.
When the data were plotted as ELs vs log VM, where V~, the mixture velocity is defined as,

VM = Vst. + VsG [2]

a definite correlation was apparent, the general form of which could be represented by

1
ELs = ~ [3]

where a and/3 are constants.

~'IKORIncorporated,217 MiddlesexTurnpike, Burlington,MA 01803, U.S.A.

 CORRELATIONOF LIQUIDVOLUMEFRACTIONFOR HORIZONTALSLUGFLOW 35

Table 1. Detailed data for runs performed in 2.58 cm I.D. horizontal pipe

Pipe diameter = 2.58 cm Pipe diameter = 2.58 cm

VsL Vsc VsL Vs~
(h/sec) (h/sec) ELs (h/sec) (h/sec) ELs

0,030 0.088 1.000 0,610 0.088 0,958

0.030 0.140 1.000 0.610 0.140 0,961
0,030 0.244 1,000 0.610 0.247 0.942
0.030 0.460 0.983 0,610 0.460 0.948
0,030 0.930 0.963 0.610 0.911 0.942
0.030 1,838 0,897 0,610 1,807 0.889
0.061 0.088 0.998 0,610 2.914 0,788
0,061 0.140 1.000 0.610 5,437 0.699
0,061 0.244 0.998 0.914 0.088 0,968
0.061 0.460 0.986 0.914 0.140 0.962
0,061 0.927 0,985 0,914 0.247 0.942
0.061 1.835 0.922 0.914 0.460 0.930
0.122 0.088 1.000 0.914 0,911 0.899
0.122 0,140 0.999 0,914 1.807 0.876
0.122 0.241 0.998 0.914 3.042 0.746
0.122 0,460 0.938 0.914 5.337 0.674
0,122 0,923 0.924 0.914 10.293 0.578
0.122 1.820 0.913 0,914 15.626 0.519
0,122 2.917 0.869 1,372 0.244 0.934
0,213 0.088 1.000 1.372 0.457 0.907
0.213 0,140 0.997 1.372 0.917 0,834
0,213 0.241 0.963 1,372 1.643 0.790
0.213 0.460 0.953 1.372 3.011 0.723
0.213 0,927 0.935 1.372 5.605 0.640
0.213 1.838 0.878 1.372 I 1.195 0.532
0,213 2.825 0.849 2.133 1.667 0.671
0.366 0.088 0.989 2,133 2.987 0.621
0,366 0.140 0.953 2.133 5.922 0,557
0.366 0.241 0,935
0.366 0.460 0,937
0.366 0,923 0.934
0.366 1.813 0.866
0.366 3.151 0.807
0.366 5,111 0.680

1.0 --%..~o ~ s...; ~. • • •

• "";~ ..a,.,3 ''b ° .%1°°.

° . . ° . .. :-L'; "~V.:
4 . "'°'.

o .,~ ~
0.9-- o
•. ~ x . , . " o
°° o~
08--

\o
Q7-- o 1.0 = 2 5 8 cm "~"
• I.D = 5 . 1 2 cm °"
LU \° o
°° ~ .\
06--

° k . •

I
\
0.5-- ELS /V M \139
- .k
\.-.
\
0.4. - -

0.1
l t
1,0 I00
VM ( m / s )

Figure 1. Measured values of E~,s vs mixture velocity for air-oil slug flow in horizontal 2.58 and 5.12cm
I.D. pipes.
36 G A GREGORY.M K NICHOLSONand K. AZlZ

Table 2. Detailed data for runs performed in 5.12 cm I.D. horizontal pipe

Pipe diameter = 5.12 cm Pipe diameter = 5.12 cm Pipe diameter = 5.12 cm

vsL Vs~ VsL Vs. VsL Vs6
(h/sec) (h/sec) ELs (h/sec) (h/sec) ELs (h/sec) (h/sec) ELs

0.061 0.283 0.980 0.610 0.046 0.989 1.372 0.216 0.926

0.061 0.823 0.963 0.610 0.046 0.975 1.372 0.311 0.894
0.061 1.490 0.956 0.610 0.091 0.985 1.372 0.436 0.896
0.061 1.908 0.941 0.610 0.094 0.983 1.372 0.564 0.863
0.061 1.929 0.931 0.610 0.146 0.978 1.372 0.680 0.855
0.122 0.046 1.000 0.610 0.216 0.987 1.372 0.683 0.862
0.122 0.094 0.986 0.610 0308 0.992 1.372 0.802 0.844
0.122 0.094 0.994 0.610 0.436 0.971 1.372 0.933 0.825
0.122 0.152 0.975 0.610 0.558 0.979 1.372 1.137 0.822
0.122 0.155 0.978 0.610 0.680 0.959 1.372 1.457 0.800
0.122 0.216 0.997 0.610 0.808 0.939 1.372 2.874 0.702
0.122 0.216 0.974 0.610 0.933 0.944 1.372 4.179 0.616
0.122 0.311 0.969 0.610 1.198 0.930 1.372 5.751 0.550
0.122 0.436 0.994 0.610 1.448 0.886 1.372 8.%1 0.465
0.122 0.558 1.000 0.610 1.929 0.866 1.372 12.057 0.444
0.122 0.680 1.000 0.610 3.002 0.783 1.372 15.480 0.456
0.122 0.683 1.000 0.610 4.368 0.716 1.823 15.303 0.442
0.122 0.805 1.000 0.610 5.407 0.658 1.856 2.024 0.670
0.122 0.927 0.976 0.610 8.537 0.563 1.856 4.596 0.541
0.122 1.192 0.949 0.610 10.948 0.469 1.862 0.832 0.789
0.122 1.481 0.950 0.914 0.091 0.983 1.865 1.201 0.727
0.122 1.951 0.924 0.914 0.155 0.964 1.868 9.683 0.450
0.305 0.046 1.000 0.914 0.219 0.958 1.871 12.774 0.417
0.305 0.094 0.998 0.914 0.314 0.958 2.316 2.009 0.623
0.305 0.094 0.996 0.914 0.439 0.916 2.316 4.246 0.485
0.305 0.146 1.000 0.914 0.561 0.906
0.305 0.216 0.985 0.914 0.683 0.883
0.305 0.219 0.988 0.914 0.689 0.872
0.305 0.308 0.973 0.914 0.811 0.896
0.305 0.433 0.965 0.914 0.927 0.867
0.305 0.564 0.973 0.914 1.158 0.868
0.305 0.686 0.971 0.914 1.439 0.858
0.305 0.808 0.985 0.914 1.941 0.803
0.305 0.939 0.986 0.914 2.911 0.757
0.305 1.204 0.941 0.914 4.325 0.686
0.305 1.442 0.935 0.914 6.053 0.590
0.305 1.911 0.911 0.914 9.390 0.479
0.305 2.941 0.754 0.914 10.628 0.450
0.305 4.383 0.626 0.914 15.376 0.342
0.305 6.388 0.479
0.305 8.689 0.487

Values of a and /3 were determined for various subsets of the data using a method of
non-linear least squares, and the results of these calculations are given in table 3.
All of the 167 data points are shown in figure 1 as a semi-log plot of ELs vs VM. The dashed
line in figure 1 is given by the relation,

1
ELS= 1 + ~.-.-.-~(VM)
1"39 [4]

where the mixture velocity, VM has units of (m/sec).

Equation[4] is actually based on a culled data set in which all observations where
Vs6 > 10 m/sec have been discarded. These points are included in figure l, however. The basis
for neglecting these points in the evaluation of a and/3 lies in the observations from a detailed
investigation of flow pattern transitions for the same oil/air system and flow loop used in this
study. Gregory et al. (1977) noted that the transition from intermittent (i.e. slug) flow to annular
flow generally lies within the region of 10 ~ VsG <- 15 m/sec. A similar observation can be made
 CORRELATIONOF LIQUIDVOLUMEFRACTIONFOR HORIZONTALSLUGFLOW 37

Table 3. Comparison of the values of a and/3 in[3l calculated using various

data subsets

I
Model: ELS = 1 + ( _ ~ ) a

di = (ELs)i, pred-- (ELs)i. obs

d~ i=l
n
n
~, d;2
S=,~-~
n-I

Description Number of
of data data points a 13 d S

All data for

2.58 and 5.12
cm I.D. pipes 167 10.2 1.21 -0.006 0.042

All data for

2.58 cm I.D.
pipe 62 12.8 1.06 -0.004 0.035

All data for

5.12 cm I.D.
pipe 105 9.48 1.26 -0.008 0.042

Vsc < 10 m/sec

for 2.58 and
5.12 cm I.D.
pipes 157 8.66 1.39 -0.002 0.034

Vsc < 10 m/sec

for 2.58 cm
I.D. pipe 59 9.95 1.27 -0.004 0.029

Vsc < 10 m/sec

for 5.12 cm
I.D. pipe 98 8.30 1.42 -0.005 0.035

with the generalized flow pattern map that has been proposed by Mandhane et al. (1974). The
relatively wider data scatter that is evident in figure 1 for VM > 10 m/sec is thus thought to be
due to the somewhat unstable nature of slug flow as the transition to annular flow is
approached. It is significant that ELS values in this region are of the order of 0.45, indicating
that the slug is in fact a very frothy mixture under these conditions. It is clear from table 3 that
the 10 observations which lie in or very close to the transition region exert a substantial effect
on the values of a and/3.
Examination of figure I and table 3 suggests the existence of a modest diameter effect, with
values of ELs tending to be slightly higher for the 2.58 cm pipe at values of VM > 1.0 m/sec, and
slightly lower when VM < 1.0 m/sec. If, in fact, [3] is rewritten as

o~~
E L S - °t ~ + V M ~

it can be noted that there is only a small difference between the values of a ~ for the culled 2.58
and 5.12 cm I.D. data sets (i.e. 18.5 and 20.2 respectively). Thus it is possible that a ~ represents
a single constant for the two diameters and any true diameter effect is reflected solely in the
value of/3. Detailed examination of the data in tables I and 2 also suggests that there might be
a modest effect on ELS due to VsG. However, in view of the general degree of data scatter, the
38 G A. GREGORY. M K NICHOLSON and K AZIZ

I.O - -

.'x," ",,. "N"

09 -- \\ ,'..\ .\ \
-\; \ .vsL=o,65
• ~ "\ \o oN/

0.8 - - • ",, X .x,"

\ \ Fvs,: o.,.~2
0.7--

u,.I

0.6--

05-- 0
•
•
VSL = 0 . 4 4 2
VSL = 0 . , 6 5
VSL = 0.971
rn/sec
m/sec
m/sec
'2 o ° \\
k

0.4--

I L
I0 I0.0
VM(m/s)

Figure 2. Comparison of the correlation proposed in this study with the data of Hubbard (1965).

fact that only two diameters have been considered, and the simplicity of [4], no attempt has
been made at this time to quantify these effects.
Finally, it is of interest to compare the ELs data of Hubbard (1965) referred to earlier with
the correlation represented by [4]. This comparison is shown in figure 2 in which [4] is shown by
the dashed line. It must be noted that Hubbard's data were obtained in a 3.81 cm I.D. pipe with
the air-water system and the different fluid system may account for at least some of the data
shift to the right. However, the scatter is so bad that any explanation is tenuous. The solid lines
in figure 2 represent the "smoothed" fit to the data for the various indicated values of VSL
which were used by Hubbard for the purposes of testing his model. The correlation and data
presented in this paper are an obvious improvement over those of Hubbard, which have been
the only data known by the present authors to have been reported in the literature up to this
time.
CONCLUSIONS
1. The capacitance-type in situ liquid volume fraction sensors described by Gregory &
Mattar (1972) are capable of providing reasonably reproducible measurements of ELs when
used~with a non-electrolyte system such as air-oil.
2. A correlation has been proposed which can be used to predict ELs over the entire range
of flow rates in which slug flow occurs.
3. The form of the correlation might eventually have to be modified to include a diameter,
flow rate and possible fluid property effect as appropriate data become available.

Acknowledgements--Financial support for this study has been provided by the National
Research Council of Canada.

NOMENCLATURE
d mean deviation between predicted and observed values of ELS;
ELs liquid volume fraction in the slug under flowing conditions:
 CORRELATION OF LIQUID VOLUME FRACTION FOR HORIZONTALSLUG FLOW 39

S standard deviation between predicted and observed values;

VM mixture velocity, -VsL + Vsc (m/sec);
Vsc superficial velocity of the gas phase (m/sec);
VsL superficial velocity of the liquid phase (m/sec);
a parameter in [3];
/3 parameter in [3];
pc density of the gas phase (kg/m3);
pL density of the liquid phase (kg/m3);
pM density of the gas-liquid mixture, defined by [1], (kg/m3).

REFERENCES
Azlz, K. & GREGORY,G. A. 1976 Prediction of flow pattern, holdup, and pressure drop in
multiphase (Oil-Gas) pipelines. Proceeding of the symposium on submarine gas pipeline
developments sponsored by the Royal Inst. of Engrs. in the Netherlands, Amsterdam, The
Netherlands, April.
DUKLER,A. E. 8~ HUBBARD,M. G. 1975 A model for gas-liquid slug flow in horizontal and near
horizontal tubes. Ind. Engng Chem. Fundam. 14, 337-347.
GREGORY,G. A. • MATTAR,L. 1972 An in-situ volume fraction sensor for two-phase flows of
non-electrolytes. J. Can. Pet. Tech. 12, 48-52.
GREGORY,G. A., NICHOLSON,M. K. t~ AzIZ, K. 1977 Intermittent two phase flow in horizontal
pipes: I. Analysis of flow regimes and transitions. Presented at the 27th Can. Soc. Chem.
Engng Conference, Calgary, Alberta, October.
HOaaARD, M. G. 1965 An analysis of horizontal gas-liquid slug flow. Ph.D. Dissertation. The
University of Houston, Houston, Texas.
KORDYBAN,E. S. 1961 Flow model for two-phase slug flow in horizontal tubes. J. Bas. Engng 83,
613-618.
KORDYaAN,E. S. & RAnov, T. 1963 Experimental study of the mechanism of two-phase slug flow
in horizontal tubes. Multiphase Flow Symp., Winter Annual Mtg. of A.S.M.E., Philadelphia,
PN, November.
MANDHANE,J. M., GREGORY,G. A. & AzIz, K., 1974 A flow pattern map for gas-liquid flow in
horizontal pipes. Int. J. Multiphase Flow 1,537-553.