Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

Void Fraction in Bubbly and Slug Flow in

Downward Vertical and Inclined Systems


A.R. Hasan, SPE, U. of North Dakota

Summary
Much work has been done in the area of two-phase upflow, while the The gas void fraction depends on the in-situ velocity of the gas
flow of gas and liquid in the downward direction has not received phase, VG. For upflow, the buoyancy effect and the tendency of the
~uc~ attention. This paper presents a model for estimating void frac- gas b~bbles to ~ow through the channel center cause the in-situ gas
tion m downward flow for vertical and inclined systems for the two velOCity to be higher than the mixture velocity. For downflow, buoy-
dominant flow regimes, bubbly flow and slug flow. We use the drift- ancy opposes the flow of the gas phase. The cross-sectional distribu-
flux approach to determine slip between the phases and the transition tion ofthe gas phase in the channel may also be different from that
between the flow regimes. The flowing mixture densities calculated in upflow. The effect of buoyancy and bubble distribution across the
from the model are useful for pressure gradient estimation. channel depends on the existing flow pattern. In this paper, we de-
Extensive data are used to validate the model. The data show that velop relationships for void fraction in terms of phase velocities and
for both .bubbly and s.lug flow the effect of buoyancy, expressed by system properties for bubbly and slug flow regimes. The upper limit
the termmal bubble-nse velocity, has the same magnitude as that for for the flow regime is then established. We use data from diverse
.the case of upflow. In addition, the flow distribution parameter in sources to validate the model.
bubbly flow appears to have the same value of 1.2 as in upflow. For
slug flow, however, the flow parameter is represented by a some- Bubbly Flow
what lower value (1.12) than for upflow. For upflow in vertical and inclined channels, VG has been expressed
as the sum of the terminal bubble-rise velocity, v 00, and the mixture
Introduction velocity at the channel center, Covm .10-13 Because v 00 acts in op-
Downward simultaneous flow of gas and liquid, although rarer than position to flow in the downward direction, we express VG by
two-phase upflow, is important in chemical process industries and VG = Covm - vox>' (I)
petroleum production. Wet steam injection into high-viscosity-oil-
With VG = vGs/fg, Eq. 1 may be used to arrive at the following ex-
bearing formations is an example oftwo-phase downflow. Another
pression for the gas void fraction:
example is the design of gas anchors, which separate gas from oil
by reversing the direction of flow. fg = VGS ......•.•••.....••........•.•.• (2)
The static head is most often the major contributor to the total Covm - v",
pressure gradient during two-phase flow. The static head is directly
For upflow, V oo is well represented by the Harmathy l4 correlation
pr?portio?al to the in-situ volume fraction of the phases. Therefore,
for both vertical and inclined channels. 12,13
It IS very Important to estimate the gas void fraction,/g, accurately.
However, two-phase flow in the downward direction is not yet suffi-
ciently understood to allow proper estimation of void fraction under .................. (3)
various circumstances. The available correlations for predicting
The flow parameter Co reflects the higher mixture velocity at the
flow. ~attern transition, void fraction, and pressure drop are largely
channel center that the gas phase experiences. For turbulent flow, the
empmcal. Some of these correlations are discussed here before ratio ofthe maximum velocity at the channel center to the channel av-
presenting the method proposed in this work.
erage velocity is 1.2.'Although not all the gas bubbles flow exactly
The classic study of Beggs and Brill l probably gives the most
throug~ the center, very few flow close to the wall and the velocity
comprehensive method currently available for predicting void frac-
profile IS flat enough for Co= 1.2 to be applicable. Hasan and Kabir l2
tion and pressure gradient in inclined systems. The correlation is
found a value of 1.2 for Co to agree with their void fraction data and
based on a predictive method for horizontal systems and modifica- data from other sources for upflow at all angles of inclination.
tions to account for the system inclination in upward and downward The vertical downflow data of Clark and Flemmer l5 and Hasan 9
directions. Mukherjee and Bri1l2 also present extensive data and a support the applicability ofEq. 2 to downflow. However, the values
correlation for downward two-phase flow. Both the correlations are of Co and v 00 may be different from those for vertical upflow and
quite robust and generally predict void fraction accurately. Howev- may depend on the extent of wellbore deviation. To establish the
er, they are less reliable at low liquid flow rates because oftheir para- values of these parameters, we analyzed data from three sources,
metric dependence on input liquid fraction. including air/water data gathered by Beggs l6 with 25- and 38-mm
Barnea3 presents a unified model for flow pattern transitions in pipes at various inclinations; air/kerosene and air/lube oil data gath-
both upward and downward flow. Crawford et al. 4 present data and ered by Mukherjee l ? in a 38-mm pipe at various inclinations; and
take an empirical approach in predicting flow pattern transition in air/oil data gathered by Kokal 18 in 25.4-, 50.8-, and 76.2-mm pipes
downflow by modifying the Weisman and Kang 5 method in inclined inclined at 5 and 9° from horizontal.
two-phase upflow. Crawford et al. 6 also propose a simplified drift- Beggs' downflow air/water data in terms of VG (= vGs/fg) vs. Vm
flux approach to predicting void fraction in bubbly and slug flow in are shown in Fig. 1 for pipe inclinations (from horizontal) of 90 and
downflow. Spedding et al.? present frictional gradient data and a 85° (shown as squares, D), 75 and 55° ( +), 35 and 20° (<», and 15,
correlation for downward bubbly and slug flow, among other flow 10, and 5° (.6.). The linear relationship observed between the in-situ
regimes. Yamazaki and Yamaguchi 8 find that a modified Bankoff gas velocity and the mixture velocity in this figure supports the ap-
variable-density approach is suitable for estimating void fraction plicability of Eqs. 1 and 2 for two-phase downflow.
~nd p~ess~re drop in all flow regimes in downflow. A preliminary The solid line in Fig. I, with an intercept of - 0.24 m/s and a slope
Investigation by Hasan 9 indicates that the drift-flux approach that of 1.2, appears to represent data for all inclination angles quite well.
recognizes the different flow regimes is likely to be superior to these The Harmathy correlation gives a value of 0.24 m/s for the rise ve-
empirical methods for estimating void fraction in downflow. locity of air bubbles in water at normal temperatures and pressures.
Thus, Fig. 1 suggests that Co and v 00 are unaffected by the flow
Copyright 1995 Society of Petroleum Engineers
direction. In addition, as with upflow (Hasan and Kabir l2 and Ha-
Original SPE manuscript received for review Del. 25, 1993. Revised manuscript received san 13), pipe inclination does not appear to influence the flow param-
Dec. 5, 1994. Paper peer approved Jan. 10, 1995. Paper (SPE 26522) first presented at the
1993 SPE Annual Technical Conference and Exhibition, Houston, Del. :Hl. eter or the terminal rise velocity in downward bubbly flow.

172 SPE Production & Facilities, August 1995


10,-------------------------,
12

11

.!!!
E
10
o .0
":?
'"
";».0> ..
".
o.
.·0 o •• ;3-
'0

.
0
Q; o

. ....
>
l/) o~
o ttl
<!l
[J 90 & 8:· degree
T 75 & 55 degree .2 c 90 <'legree
o <) 35 & 20 degree 'f ... 80 & 70 oeq'ee
t:. 15, 10 & = degree E o
to
5C & 30 deg~ee
20 & 5 degreo:>

6 10

Mixture Velocity, Vm, m/s Mixture Velocity, Vm, m/s

Fig. 1-ln-situ gas velocity vs. mixture velocity, Beggs bubbly Fig. 2-Predlcted vs. experimental void fraction, Beggs bUbbly
flow data. flow data.

The airlkerosene and air/lube-oil data of Mukherjee 17 and the air/ vertical downflow may be expressed in terms of the superficial
oil data of Kokal 18 also support the conclusion that the flow parame- phase velocities by using 18 = 0.25.
ter and the terminal rise velocity in downflow are the same as in up-
flow. Note that for these data, the magnitude of the terminal rise COVLS - v",
vas = (/.J = 0.43 v LS - 0.36v",. . (4)
velocity varies from 0.18 to 0.24 mls. Therefore, we propose Eq. 2 1 0.25 - Co
for estimating void fraction in downward bubbly flow with a value
of 1.2 for the flow parameter Co, and the terminal rise velocity esti- Although the rise velocity appears to be unaffected by pipe in-
mated from the Harmathy correlation, Eq. 3. clination angle, it is necessary to modify Eg. 4 before it can be ap-
Fig. 2 compares our predictions with the downward bubbly flow plied to bubbly/slug flow transition in deviated wells. In inclined
data of Beggs. 16 Eq. 2 underestimates these data with an average er- systems, bubbles flow preferentially at the upper wall of the pipe,
e:
ror, of - 0.005 and a standard deviation of 0.022. We did not ob- causing a higher local void fraction there compared with the cross-
sectional average value. Consequently, the local void fraction at the
serve a definite pattern in prediction error with the inclination angle;
the errors appeared to be random. Our attempts at optimizing the val- upper pipe wall exceeds the transition value of 0.25 even though the
ue of the flow parameter resulted in a slight improvement in the aver- cross-sectional average void fraction is well below 0.25.
age error and standard deviation of predictions for a slightly lower Assuming that the gas velocity at the upper portion of the pipe is
value of Co = 1.18 (e = 0.0, standard deviation equals 0.0217). e
higher by a factor of IIsin (ratio of actual to projected flow area)
We predict the airlkerosene bubbly-flow data of Mukherjee with compared with the cross-sectional average value, Hasan 13 and Ha-
an average error of - 0.011 and a standard deviation of 0.0497. The san and Kabir l2 derive an expression that is similar to Eq. 4 for tran-
higher standard deviation for the Mukherjee data reflects larger sition from bubbly to slug flow for upwardly inclined pipes. The
scatter in the data. For this data set the optimum value of Co is 1.16, same transition equation is proposed for downflow as well, using
giving an average error of 0.0003 and a standard deviation of the negative sign for the terminal rise velocity.
0.0472. The air/lube-oil void fraction data of Mukherjee and the air/
oil data of Kokal 18 are also estimated with similar accuracy by Eq. .vas = (0.43VLS - 0.36v",) sine, (5)
.
2. Table 1 shows the performance of Eg. 2 in predicting these data. where eis the positive angle of inclination from horizontal. Barnea3
The marginal improvement in predicting void fraction with op- proposes a criterion for bubbly/slug flow transition that is quite sim-
timized values of the flow parameters for these data sets lead us to ilar to Eq. 5. The data of Crawford et al. 4 also show that, except at
recommend that Co = 1.2, as used in upflow, is appropriate for low mass flow rates, the transition to slug flow occurs at a void frac-
downflow. tion of about 0.25.
Fig. 3 compares the predictions ofEq. 5 with the bubbly-to-slug-
Bubbly/Slug Flow Regime Transition. The transition from bubbly flow transition data for downflow systems of Crawford et al. 4 as
to slug flow occurs because of bubble agglomeration resulting from well as those of others. 4 The agreement appears acceptable. Note
increased collision between bubbles at higher void fractions. For that the log-log axes used in Fig. 3 make comparisons between the
vertical upflow, the transition from bubbly to slug flow occurs at a predictions and the data somewhat difficult. However, like many
void fraction of about 0.25. 3,12,19,20 We propose that this criterion other researchers, we used log-log axes to accomodate the wide
would apply to downward flow as well. This transition criterion for range of data and to avoid de-emphasizing low-flow-rate data. 4

TABLE 1-eOMPARISON OF THE PROPOSED MODEL WITH DATA FROM VARIOUS SOURCES
Inclination
Data Angles Flow Average Standard
Source (degrees) Fluids Regime Error Deviation
Beggs 5,10,15, Air/water Bubbly -0.005 0.022
20,35,55, Air/water
75,85,90 Slug 0.0048 0.044
Mukherjee 5, 20, 30, 50, Air/water Bubbly -0.011 0.0472
70,80,90 Air/water Slug 0.021 0.069
30,90 Air/lube oil Bubbly 0.060 0.109
Slug -0.0056 0.108
Kokal 5,9 Air/oil Bubbly 0.011 0.043
Slug 0.086 0.101

SPE Production & Facilities, August 1995 173


... -15 0 d=2.5cm
~ -30· d=2.5cm
f:,. -45· d=2.5cm
0-60·
o -90·
• -90·
---
d=2.5cm
d=2.5cm
d=3.8cm
-90· Prediction
- . - -45· Prediction
0. •
(J:
'"
. 0 '"
~,/
~,,'
"
" r -1I VooT
LT
____ -15· Prediction ~ """"" I
1.0
L
,,"
INTERMITTENT
f ;,,/
<:Y o

tj /,' ~

a Voo
D I 0
O.

o D

0.1 fl: BUBBLE


• 0 •
c:l 0

'iI ?'"V'V'~ 0'"


I
I
I " .
0 1 0

Fig. 4-Cellular model for slug flow.
0.1 10
Slug Flow
{ Vgs + VLsl/-J;j
Slug flow is characterized by a large bullet-shaped bubble, which
almost fills the entire channel cross section, followed by a liquid
Fig. 3-Transition from bubbly to intermittent (slug) flow. Com-
parison of predictions of the model with data.
slug that contains small gas bubbles. The large bubble, called the
Taylor bubble, is governed by dynamics that are different from the
small bubbles in the liquid slug. However, if we assume that the drift
Dispersed Bubbly Flow. The bubbly/slug transition criteria dis-
flux for the bubbles in the liquid slug is approximately the same as
cussed so far apply to low or moderate flow rates only. At higher
that for the Taylor bubble, then we arrive at an expression for the gas
flow rates, the shear stress caused by turbulence tends to break up
void fraction in a manner very similar to that for bubbly flow. Using
the larger bubbles, inhibiting the transition to slug flow even when v ooT for the rise velocity of a Taylor bubble in downflow, we express
void fraction exceeds the transition value of 0.25. Bubbly flow re- void fraction during ideal slug flow by
sulting from the dispersion of larger bubbles is known as dispersed
bubbly flow. Taitel et al. 20 develop a model, later modified by Sho- ig = Clv - V ' ••.•.....•.••••••••••••.•.••••• (7)
ham 21 , for the onset of dispersed bubbly flow for vertical upflow. m ooT
Hasan and Kabir22 find the model to be applicable for flow through where the value of the flow parameter C1, in analogy to bubbly flow,
annuli in vertical and inclined channels. Based on the maximum is expected to be 1.2.
bubble diameter possible under highly turbulent conditions, the Assuming that the entire gas phase has the same drift velocity as
model outlines the following relationship between phase velocities, that of the Taylor bubble, Hasan 13 and Hasan and Kabir 12 are able to
pipe diameters, and fluid properties: correlate inclined upflow void fraction data in slug flow from several
sources with satisfactory accuracy. The applicability of the simple ap-

2'~2f'·u,n'd)"'[g(o~:"'P,J " 0.725 + 415(:~t


proach derives from the fact that the gas phase in the liquid slug is usu-
ally a small fraction of the total gas phase and the difference in drift
velocities in the slug and the Taylor bubble usually is small.
To account for the differing drift velocities in the liquid slug and the
(6) Taylor bubble, a few rigorous models have been advanced for vertical
Because of the high velocities associated with dispersed bubbly upflow (Sylvester23 and Fernandes et al. 24). Caetano et al. 25 success-
flow, pipe inclination and flow direction are unlikely to have a sig- fully apply this approach to a vertical annular channel. The cellular
approach pioneered by Fernandes et al. 24 for vertical upflow is sim-
nificant effect and Eq. 6 may be used to determine transition to this
plified by Hasan and Kabir2 2 for slug flow in vertical and inclined an-
flow regime even for downflow. Bamea3 finds a similar equation in
nuli. In this work, we adapt the Hasan-Kabir22 approach for down-
satisfactory agreement with the data of Shoham,21 although he ob-
flow. Fig. 4 shows a model cell of length L, consisting of a Taylor
serves a slight dependency of this transition on the inclination angle. bubble of length LT and a liquid slug of length 4. Denoting the gas
Thus, if the mixture velocity is greater than that given by Eq. 6, fraction in the Taylor bubble portion by iT and that in the liquid slug
bubbly flow will persist even whenig is greater than 0.25. However, portion by fs, we express the average void fraction for the cell by
Taitel et al. 20 show that even for small gas bubbles, the gas void frac-
tion cannot exceed 0.52. For estimating void fraction, Eq. 2 with v 00
given by the Harmathy14 equation, which treats dispersed bubbly
ig = i iT + i is. (8)

flow as ordinary bubbly flow, is proposed. The high fluid velocities We use Eq. 7 for the Taylor bubble portion of the cell for calculat-
involved imply that pipe inclination would have little influence on ingiT . For estimatingfs, we note that the circular channel data pres-
Co. Therefore, a value of 1.2 is recommended for Co in Eq. 2 for dis- ented by Akagawa and Sakaguchi 26 show that the average volume
persed bubbly flow. fraction of gas in the liquid slug (based on the total system volume,

174 SPE Production & Facilities, August 1995


i.e., J~L,/L) is approximately equal to 0.1 when vcs is greater than Note that Eq. 14 predicts zero rise velocity for Taylor bubbles in hor-
0.4 mls and is equal to 0.25 vcs for lower superficial gas velocities. izontal systems, while some researchers have noted nonzero veloci-
ties in horizontal pipes (Alves 30 and Weber et al. 31). Other ap-
fg (i)fT +
= 0.1 for vcs > 0.4 mls proaches to estimate v ocT(} are available. 27 .30 However, because of
the reduced importance of void fraction for calculating pressure
drop in near-horizontal systems, refining estimates for VocT(} for
andfg = (i)fr+ 0.25vcsfor vcs S; 0.4 m/s (9) such systems is perhaps unwarranted.
The applicability of the proposed method for void fraction es-
timation during slug flow in downward two-phase flow is verified
For superficial gas velocities higher than 0.4 mis, Hasan and Ka- in Fig. 5, where the void fraction data of Beggs l6 for pipe inclina-
bir22 derive the following expression for the fraction LT/L tions of90 and 85° (shown as squares, D), 75 and 55° (+),35 and
[= 1 - (L,/L)] using Eq. 2 for gas void fraction in bubbly flow to the 20° (0), and 15, 10, and 5° (Ll) are plotted againstthe predicted val-
liquid slug and noting that it is equal to 0.1. ues. A value of 1.12 for CI is used in these predictions for all pipe
inclinations, which appears to minimize the standard deviation of
cs
(LLs)fs = JLsILJ...v
oVm v:o
= 0.1 (10) predictions. On average, we underestimate the data by 0.0048 with
a standard deviation of 0.044.
The proposed method predicts the airlkerosene data of Mukher-
L
or L
s = 01
.
(Cov vcs v",) .
m -
(11)
jee 17 with an average error of -0.021 and a standard deviation of
0.069 using CI = 1.12. Mukherjee's airllube-oil data are underesti-
mated with an average error of 0.0056 and a standard deviation of
Similarly, for vcs lower than 0.4 mis, we obtain 0.108. The slug flow data of Kokal l8 also is well represented by the

7= 0.25(Covm - v",j (12)


proposed method. The average error of prediction for Kokal's data
is 0.086, and the standard deviation is 0.010.
As can be expected, the standard deviation of prediction for each in-
Thus, Eq. 9 in combination with Eqs. 7 and 11 allows one to calcu- clination angle for these data sets may be improved for slightly different
late gas void fraction in slug flow provided that values of C1and v ocT values of CI. However, our attempts at optimizing the parameter values
are available. for each data set show that the improvement in predictions are minimal
Using arguments similar to those used in bubbly flow, we can and CI values optimized in this manner do not exhibit any particular
show that the value of the parameter C 1would be 1.2 for vertical up- pattern with pipe inclination angles. Therefore, a constant value of 1.12
flow, especially because turbulent flow is very likely in slug flow. for C1 is recommended for downward slug flow.
During flow in inclined channels, bubbles tend to flow closer to the We argued that slight deviation of the well from vertical makes
upper wall of the channel where the mixture velocity is expected to the bubble nose sharper during upflow. A sharper bubble nose
be less than 1.2 vm . In spite of this, Bendiksen,27 Hasan,13 and Hasan causes a greater decrease in drag force compared with the decrease
and Kabir22 find that the value of C1 remains constant at 1.2 for slug in buoyancy, resulting in a higher rise velocity. Similar arguments
flow in upwardly inclined circular channels and annuli. The work probably apply for downward slug flow, although the opposing mo-
of Martin28 with a single Taylor bubble indicates a lower value than tion of the liquid might deform the shape of the Taylor bubble. In-
1.2 for CI in downflow. In this work, the void fraction data of deed, Martin28 shows that even for vertical downward flow of a
Beggs,16 Mukherjee,17 and Kokal 18 are used to establish an ap- single Taylor bubble, the bubble is often distorted and unstable.
propriate value for CI. Martin's data for single Taylor-bubble-rise velocity also indicate a
Besides the value of C I, an expression for the rise velocity of Tay- lower value for C1(about 0.9) and a larger value for the terminal rise
lor bubbles in downwardly inclined systems is needed for estimat- velocity. However, Martin does not report any void fraction data.
ing void fraction in Eq. 9. For vertical upflow, Eq. 13, proposed by
Nicklin et al. 29 for Taylor-bubble-rise velocity, has become well es- Discussion and Conclusions
tablished: In this paper, we present a modification of the drift-flux approach for
predicting void fraction in bubbly and slug flow.in downwardly in-
V"'T = 0.345 gd(pe ;/g) (13) clined systems. We show data analysis to support the proposed model.
For downward bubbly flow, no modification from upflow is needed
except to note that the terminal bubble-rise velocity acts in the direc-
For upwardly inclined pipes, significant variation in the rise ve- tion opposite to that of the flow. Crawford et al. 4 note a somewhat
locity of a Taylor bubble with pipe deviation from vertical has been lower value of Co of between 0.8 to 0.85 for this flow regime,
reported. 13,22,27,30 Taylor-bubble-rise velocity increases with pipe compared with a value of 1.2 for upflow. However, they use the Nick-
deviation from vertical up to a deviation angle of about 45°. The rise lin et al. 29 equation for Taylor-bubble-rise velocity in bubbly flow
velocity then gradually declines as the channel approaches a hori- instead of the Harmathy l4 equation suggested in this work.
zontal position. We can derive an expression for the rise velocity of
a bubble in an upwardly inclined pipe by balancing the buoyancy
force against the drag force experienced by a rising bubble. 12, 13
0.9 -
Analyses show that the Taylor-bubble-rise velocity in inclined
0.8
channels depends on the drag coefficient and the projected area of
the bubble, as well as the reduction in buoyancy with inclination an- 0.7

gels. For small deviations from vertical, the reduction in drag forces 0.6

on the bubble outweighs the decrease in buoyancy, while the oppo- 0.5
site is true when the pipe is nearly horizontal. When the deviation
0.4 -
is small the net effect is an increase in the terminal rise velocity with
pipe deviation, compared to that for a vertical system. This effect re- 0.3 - o 90 &: 85 degree
+ 75 &: 55 degree
sults in a decreased void fraction in a deviated well compared to that 02 <:> .35 &: 20 degree
[:,. 15, 10 &: 5 degree
for a vertical well. At large deviations from vertical, the buoyancy 0.1
force decreases more quickly than the drag force, resulting in de- . ---,--,--:-.,--------.-------.--.-.....--,--- ------,- - - , - - -
creasing bubble-rise velocity with increasing deviation from verti- 0.2 0.4 0.6 0,8

cal. We propose the following expression, suggested by Hasan 13 Experimental f9

and Hasan and Kabir l2 , for downflow:


Fig. 5-Predicted vs. experimental void fraction, Beggs slug
(14) flow data.

SPE Production & Facilities, August 1995 175


For downward slug flow, the Taylor-bubble-rise velocity estimated 9. Hasan, A.R.: "Void Fraction in Gas-Liquid Two-Phase Flow in Down-
from the expression for upflow represents the void fraction data of ward Vertical Systems," paper presented at the 1989 AIChE Annual
Beggs t6 , Mukherjee 17, and Kokal t8 quite well. However, a lower Meeting, San Francisco, Nov. 5-10.
value of 1.12 for Ct, compared with 1.2 for upflow, is indicated by 10. Zuber, N. and Findlay, J.: "Average Volumetric Concentration in Two-
Phase Flow Systems," Trans.• ASME J. Heat Transfer (1965) 87,453.
the data for all inclination angles. Crawford et at. 4 note even lower
11. Govier, G.W. and Aziz, K.: The Flow ofComplex Mixtures in Pipes. Van
values of between 0.9 and 0.95 for this parameter for their two-phase Nostrand Reinhold Co., New York City (1972) 381.
flow (refrigerant 113) data gathered in a 2.5-cm vertical tube. 12. Hasan, A.R. and Kabir, C.S.: "Predicting Multiphase Flow Behavior in
A criterion for transition from bubbly flow to slug flow is pres- a Deviated Well," SPEDE (Nov. 1988) 474; Trans.• AIME, 285.
ented in this work. The transition criterion appears to be well sup- 13. Hasan, A.R.: "Inclined Two-Phase Flow: Flow Pattern, Void Fraction
ported by data from several sources. and Pressure Drop in Bubbly, Slug and Chum Flow," Particulate Phe-
nomena and Multiphase Transport, Hemisphere Publishing Corp., New
Nomenclature York City (1988) 229.
14. Harmathy, T.Z.: "Velocity of Large Drops and Bubbles in Media oflnfi-
Ap = projected area of bubble, L2, m2 nite or Restricted Extent," AlChE J. (June 1960) 6, 281.
CD = drag coefficient for bubble, dimensionless 15. Clark, N.N. and Flemmer, R.L.: "Predicting the Hold-Up in Two-Phase
Co = flow parameters in bubbly flow, dimensionless Upflow and Downflow Using the Zuber and Findlay Drift-Flux Model,"
Ct = flow parameters in slug flow, dimensionless AlChE J. (March 1985) 31, 500.
d= pipe diameter, L, m 16. Beggs, RD.: "An Experimental Study of Two-Phase Flow in Inclined
de = equivalent diameter of annuli = do - dj, L, m Pipes," PhD dissertation, U. of Tulsa, Tulsa, OK (1972).
e= average error = sum of errors/no. of data 17. MukheIjee, H.: "An Experimental Study of Inclined Two-Phase Flow,"
PhD dissertation, U. of Tulsa, Tulsa, OK (1979).
f= Fanning friction factor, dimensionless
18. Kokal, S.: "Study of Two-Phase Flow in Inclined Pipes," PhD disserta-
/g = in-situ volume fraction of gas phase, dimensionless
tion, U. of Calgary, Calgary (1987).
Is = in-situ gas fraction in liquid slug portion of cell, 19. Griffith, P. and Snyder, G.A.: "Bubbly-Slug Transition in a High Veloc-
dimensionless ity Flow," MIT Report No. 5003-29 (TID-20947) Massachusetts Inst. of
fT = in-situ gas fraction in the Taylor bubble portion of Technology, Boston (1964).
cell, dimensionless 20. Taitel, Y., Barnea, D., and Dukler, A.E.: "Modelling Flow Pattern Tran-
g= acceleration caused by gravity, Llt2, mls 2 sition for Steady Upward Gas-Liquid Flow in Vertical Tubes," AlChE
L= cell length in slug flow, L, m J. (May 1980) 26,345.
Ls = length of liquid slug in slug flow, L, m 21. Shoham, 0.: "Flow Pattern Transition and Characterization in Gas-Liq-
LT = length (\f Taylor bubble in slug flow, L, m uid Two-Phase Flow in Inclined Pipes," PhD dissertation, Tel Aviv U.,
VG = in-situ gas velocities, LIt, mls Tel Aviv (1982).
22. Hasan, A.R. and Kabir, C.S.: "Two-Phase Flow in Vertical and Inclined
VI = in-situ liquid velocities, LIt, mls
Annuli," Inti. J. Multiphase Flow (1992) 18, No.2, 279.
Vm = superficial velocity of mixture = total volumetric 23. Sylvester, N.D.: "A Mechanistic Model for Two-Phase Slug Flow in
flow rate/cross-sectional area, LIt, mls Pipes," J. Energy Resource Tech. (Dec. 1987) 109, 206.
VGS = superficial gas velocities, LIt, mls 24. Fernandes, R.C., Semiat, R., and Dukler, A.E.: "Hydro-Dynamic Model
VLS = superficial liquid velocities, LIt, mls for Gas-Liquid Slug Flow in Vertical Tubes," AlChE J. (Nov. 1983) 29,
v 00 = rise velocity of small bubbles, LIt, mls 981.
v ooT = rise velocity of Taylor bubbles, LIt, mls 25. Caetano, E.F., Shoham, 0., and Brill, J.P.: "Upward Vertical Two-Phase
e= channel inclination angle from horizontal, degrees Flow Through an Annulus, Part II: Modeling Bubble, Slug, and Annular
fl = viscosity, mILt Flow," J. Energy Resource Tech. (Sept. 1991).
P= density, mlL3 26. Akagawa, K. and Sakaguchi, T.: "Fluctuation of Void Ratio in Two-
Phase Flow," Bull. JSME (1966) 9,104.
p (= liquid density, mlL3
27. Bendiksen, K.H.: "An Experimental Investigation of the Motion of
pg = gas density, mlL3 Long Bubbles in Inclined Tubes," IntI. J. Multiphase Flow (1984) 10,
a= surface tension, mlt2 No. 4,467.
28. Martin, S.C.: "Vertically Downward Two-Phase Slug Flow," J. Fluid
Acknowledgment Eng. (Dec. 1976) 98,715.
I am grateful to a number of graduate (M.M. Hossain, R. Zielin- 29. Nicklin, DJ., Wilkes, J.O., and Davidson,J.F.: "Two-Phase Flow in Ver-
tical Tubes," Trans., Inst. Chern. Engrs. (1962) 40, 61.
ski, and R. Holmes) and undergraduate (G. Anderson, J. Dunklee,
30. Alves, LN.: "Slug Flow Phenomena in Inclined Pipes," Report, Tulsa U.
and P. Kaspari) students who helped with data analysis. Thanks also Fluid Flow Projects (1991).
to J.P. Brill and S. Kokal for sharing data with me. 31. Weber, M.E., Aleri, A., and Ryan, M.E.: "Velocities of Extended
Bubbles in Inclined Tubes," Chern. Eng. Sci. (1986) 41, 2235.
References
1. Beggs, H.D. and Brill, J.P.: "A Study of Two-Phase Flow in Inclined 51 Metric Conversion Factors
Pipes," JPT(May 1973) 607; Trans.• AIME,255.
2. MukheIjee, H. and Brill, J.P.: "Liquid Holdup Correlations for Inclined ftx 3.048* E-Ol =m
Two-Phase Flow," JPT(May 1983) 1003; Trans.. AIME,275. in. x 2.54* E+OO =cm
3. Barnea, D.: "A Unified Model for Predicting Flow Pattern Transition for
the Whole Range of Pipe Inclination," IntI. J. Multiphase Flow (1986)
*Conversion factor is exact. SPEPF
13, No.1, 1-12.
4. Crawford, T.J., Weinberger, C.B., and Weisman, J.: "Two-Phase Flow
Patterns and Void Fractions in Downward Flow. Part I: Steady-State A. Rashid Hasan is a professor of chemical engineering at the U.
Flow Patterns," Inti. J. Multiphase Flow (1985) 11, No.6, 761. of North Dakota in Grand Forks. His research interests include
5. Weisman, J. and Kang, S.Y.: "Flow Pattern Transitions in Vertical and heat transfer and multiphase flow in wellbores. rheology of non-
Upwardly Inclined Lines," Inti. J. Multiphase Flow (1981) 7, 271. Newtonian fluids, and pressure-transient analysis. He holds MS
6. Crawford, TJ., Weinberger, C.B., and Weisman, J.: "Two-Phase Flow and PhD degrees in chemical engineering from the U. of Water-
Patterns and Void Fractions in Downward Flow. Part II: Void Fractions 100. Canada. Hasan serves on the Editorial Review Committee.
and Transient Flow Patterns," IntI. J. Multiphase Flow (1986) 13, No.
2,219.
7. Spedding, PL., Chen, UJ., and Nguyen, V.T.: "Pressure Drop in Two-
Phase Gas-Liquid Flow in Inclined Pipes," Inti. J. Multiphase Flow
(1982) 8, No.4, 407.
8. Yamazaki, Y. and Yamaguchi, K.: "Characteristic of Cocurrent Two-
Phase Downflow in Tubes; Flow Pattern, Void Fraction and Pressure
Drop," J. Nuclear Sci. & Tech (April 1979) 161, 245.

176 SPE Production & Facilities, August 1995

You might also like