Sunteți pe pagina 1din 11

Experimental Study of High-Viscosity

Oil/Water/Gas Three-Phase Flow in


Horizontal and Upward Vertical Pipes
S. Wang, H.-Q. Zhang, C. Sarica, and E. Pereyra, University of Tulsa
Summary
In this experimental study, measurements and observations have
been carried out for high-viscosity oil/water/gas three-phase ows
in horizontal and upward vertical pipes. Oil with viscosities be-
tween 0.15 and 0.57 Pas corresponding to temperatures from
37.8 to 15.6

C, ltered tap water, and natural gas at 2.59 MPa


pressure are used as the three phases. Supercial oil and water
velocities range from 0.1 to 1.0 m/s, and supercial gas velocity
varies from 1.0 to 5.0 m/s. The internal diameter of the pipe is
5.25 cm. The experimental measurements include pressure gradi-
ent and liquid holdup. The ow-pattern and slug characteristics
are observed and the images are recorded with a high-speed video
camera system through a high-pressure sapphire window. The ex-
perimental results are compared with the predictions of the Zhang
and Sarica (2006) unied model (UM), and the discrepancies are
identied.
Introduction
Heavy oil constitutes a signicant portion of the worlds total oil
reserve. It is discovered and produced all around the world and has
become one of the most important future hydrocarbon resources,
with ever-increasing world energy demand and depletion of con-
ventional oils. However, heavy oil possesses high viscosity, which
poses many challenges for its production and transportation.
Three-phase ow of oil, water, and gas is of particular impor-
tance for the oil industry. This frequently occurs in wells, risers
and owlines before reaching the downstream processing facili-
ties. Most of the oil and gas reservoirs naturally contain water
which generally has high salinity. Water could also be injected
into the reservoir to maintain pressure at the later stage of produc-
tion. Understanding of the three-phase ow phenomena is neces-
sary in order to better design the production and transportation
systems. Most of the previous experimental research, correlation
and model developments were conducted using low-viscosity
conventional oils or other liquids. Acikgoz et al. (1992) carried
out the rst experimental investigation of three-phase ow pat-
terns in horizontal pipes. Pan et al. (1995) performed similar hori-
zontal experiments. The tests were conducted at 0.5 MPa pressure
and in a 38.0-m-long horizontal, 7.62-cm-ID pipe. Woods et al.
(1998) reported oil/water/air upward vertical ow in a 2.52-cm-
ID Perspex pipe with a 1.8-m test section. Nine ow patterns were
identied based on visual/video observations and pressure techni-
ques. Langsholt and Holm (2001) studied oil/water/gas ows in
steeply inclined pipes. Hewitt (2005) studied oil/water/air three-
phase ows in a 38.0-m-long, 7.62-cm-ID stainless-steel pipe.
Keskin et al. (2007) proposed a two-step classication method for
oil/water/gas three-phase ow patterns. Twelve ow patterns
were identied for horizontal ows.
High-viscosity oil multiphase ow behaves very differently
than low-viscosity oil multiphase ow. Signicant discrepancies
were also observed in model comparisons. Very few studies have
been conducted on high-viscosity oil/water/gas three-phase ow
until very recently. Bannwart et al. (2009) investigated heavy oil/
water/air ows in horizontal, upward vertical, and inclined pipes.
This oil has a viscosity of 34.95 Pas. Flow patterns were identi-
ed from analogies with gas/liquid ow. In horizontal ow, the
presence of gas would considerably increase the pressure loss
compared with oil/water two-phase ow. On the other hand, the
pressure loss would be reduced with the injection of water caused
by lubrication and with the process of keeping oil from touching
the pipe wall. In upward vertical pipes, where the gravitational
term plays an important role, the three-phase pressure drop can be
reduced to as low as 5% of the single-phase oil pressure drop.
Poesio et al. (2009) examined the effect of air on horizontal oil/
water intermittent ow. Two oils with viscosity of 0.9 and 1.2
Pas were used. It was found that with the increase of supercial
air velocity, the total pressure drop would increase accordingly. A
hybrid model for pressure-drop prediction based on the Lockhart-
Martinelli method was proposed and compared with experimental
data, and it showed fairly good agreement.
So far, very few experimental studies have been conducted on
high-viscosity oil/water/gas three-phase pipe ows, none at ele-
vated pressure and with natural gas. Many aspects of the hydrody-
namic behavior are not well understood. The performance of
the current mechanistic models against high-viscosity oil ex-
perimental results also needs to be assessed. This experimental
study is part of the Tulsa University High-Viscosity Oil Projects
(TUHOP), which aims at a comprehensive understanding of mul-
tiphase ow of high-viscosity oil with water and gas. The horizon-
tal and upward vertical ow conditions are chosen to simulate
multiphase ows in horizontal and vertical wells and in surface
transportation lines.
Experimental Facility
Multiphase Flow Loop. The high-viscosity oil/water/gas ow
tests were conducted on the TUPDP high-pressure multiphase
ow loop. This system was previously used by Vuong et al.
(2009), Akhiyarov et al. (2010), and Sridhar et al. (2011) to inves-
tigate oil/water and oil/gas two-phase ow behaviors. This facility
consists of three main systemsoil, water, and gas systems, and
three auxiliary systems: instrumentation air, glycol temperature
control, and data-acquisition systems.
The current overall process schematic of the multiphase ow
loop is shown in Fig. 1. The facility is capable of conducting
single-phase and multiphase experiments at pressures up to 6.9
MPa, temperatures from 4.4 to 71.1

C, and inclination angles


from 2 to 90

from horizontal. The test section consists of a U-


shaped, 5.25-cm-ID, Schedule 40, 340 stainless-steel pipe. The
total length of the pipe is 48.8 m. It is mounted on a boom so that
any inclination angle within the experimental range can be
selected using a hydraulic hoist attached to an 18.3-m-high tower.
The test section is partially jacketed with a 10.16-cm nominal di-
ameter CPVC pipe over a length of approximately 16.2 m for heat
exchange of temperature control.
Fig. 2 is the schematic of the test section. A chilled mixture of
glycol and water (50/50%) can be circulated inside the annulus of
the jacket and/or through a heat exchanger countercurrently to the
multiphase mixture owing in the inner pipes. The glycol ow
helps to create a simulated cold ambient environment or maintain
test uid at a constant temperature. The test section consists of a
Copyright VC 2013 Society of Petroleum Engineers
This paper (SPE 163148) was revised for publication from paper OTC 23129, rst presented
at the Offshore Technology Conference, Houston, Texas, USA, 30 April5 May 2012.
Original manuscript received for review 6 January 2012. Revised manuscript received for
review 10 January 2013. Paper peer approved 14 January 2013.
306 August 2013 SPE Production & Operations
U-shaped, 5.25-cm-ID stainless-steel pipe with a length of 24 m
for each leg. The rst segment is a 3.35-m-long hydraulic devel-
oping section. The second segment is a 7.43-m-long thermal
developing section. The pressure drop in this section is measured
by a differential pressure transducer and is used for monitoring
the ow development. The third segment is also a 7.43-m-long
longitudinal measurement section. This section is used to deter-
mine the pressure drop by a differential pressure transducer. There
are ten resistance temperature detectors (RTD) located equidistant
along this section to obtain test uids and glycol-mixture tempera-
tures. Two quick closing valves (QCV) at the inlet and outlet are
installed for liquid holdup measurement. The last segment is a
1.52-m-long removable spool piece. It has a high-pressure sap-
phire window through which ow pattern and ow behavior can
be observed and recorded with a high-speed video camera. The
total internal volume of the system is approximately 15 m
3
. The
volumes of oil and water storage tanks of the facility are 4 and 2.4
m
3
, respectively. The oil viscosity is monitored constantly with
the pipe viscometer. Low-shear Moyno progressing cavity pumps
are used to circulate oil and water. The 12- to 18-hour time inter-
vals between daily experiments help to separate the oil, water,
and gas. There are no signicant emulsion problems.
Measurements and Instrumentation. Flow Pattern. Flow pat-
tern is identied by observing the ow behavior through the high-
pressure sapphire window using a high-speed video recording
system. To better observe the ow characteristics, a Photron
TM
high-speed video camera is used to take video images with a pixel
resolution of 10241024 up to 1,000 frames per second (fps). The
maximum shutter speed is 100,000 fps with a reduced resolution.
Side views and cross-sectional phase distributions corresponding
to different owpatterns will be illustrated in the next section.
Gas
System
Oil Return Line
Oil/Water Separator
Oil Tank
Water Tank
Water Heat Exchanger
Oil Heat Exchanger
Test Section
Gas/Liquid
Separator
Fig. 1Process schematic of multiphase ow loop.
#6 #5
PT 14
T44
T45
T49
T50
T25
T20
T39
T40
Glycol
Test Fluid
33 23 ft 23 ft
22 23 ft 23 ft
11 ft 11 ft 11
T27 T27
PT 7 PT 7
T9 T9
T33W T33W
T32W T32W
T8 T8
PT 8 PT 8
TRIMMERS TRIMMERS TRIMMERS TRIMMERS TRIMMERS TRIMMERS TRIMMERS TRIMMERS TRIMMERS TRIMMERS TRIMMERS TRIMMERS
GLYCOL GLYCOL GLYCOL GLYCOL GLYCOL GLYCOL GLYCOL GLYCOL GLYCOL GLYCOL GLYCOL GLYCOL
44 5 ft 5 ft
11 Hydraulic Developing Segment Hydraulic Developing Segment
Thermal Developing Segment Thermal Developing Segment
Longitudinal Measurement Segment Longitudinal Measurement Segment
Removable Spool Piece Removable Spool Piece
22
33
44
T34
T35
T29
T30
#4 #3 #2 #1
G-L SEP
Fig. 2Schematic of test section.
August 2013 SPE Production & Operations 307
Pressure Gradient. Differential pressures are measured by
two differential pressure transducers. One differential pressure
transducer is used in the single-phase oil pipe viscometer section
to determine live oil viscosity, and the other is installed to mea-
sure the pressure drop over the longitudinal measurement segment.
Liquid Holdup. Liquid holdups are measured by using the
7.4-m longitudinal measurement segment. Fluids owing in the
test section are trapped inside this long pipe by using two QCVs
located at both ends of this segment. The QCVs can be com-
pletely closed from an open position within 1.5 seconds. The
trapped oil/water/gas mixtures are allowed to separate for 5
minutes and then the pipe is raised to upward vertical position.
The water is drained into graduated cylinders from a drainage
port immediately above the bottom QCV. The oil is left in the
pipe and its volume can be back-calculated from the differential
pressure transducer readings.
Test Fluid. The uids used in the experiments consist of a
rened mineral oil (Lubsoil ND-50), ltered Tulsa city tap water
(no salinity), and natural gas supplied by Oklahoma Natural Gas
Company. The Lubsoil ND-50 oil was selected because of its rela-
tively high viscosity. Tap water was ltered before being fed to
the facility. The oil has the following physical properties:
API gravity: 28.5

Density: 884.4 kg/m


3
at standard conditions
Dead-oil viscosity: 1.1 Pas at 15.6

C
Surface tension: 35.75 dynes/cm at 19.8

C
Interfacial tension with water: 30.4 dynes/cm at 19.8

C
Pour and ashpoint temperatures: 15 and 266

C, respectively.
Test Matrix. On the basis of the research objectives and facility
limitations, high-viscosity oil/water/gas three-phase tests have
been carried out for both horizontal and upward vertical ows.
The experimental ranges are as following:
Supercial oil velocity: 0.1 to 1 m/s
Supercial water velocity: 0.1 to 1 m/s
Supercial gas velocity: 1 to 5 m/s
Water cut: 17 to 77%
On the basis of a rheological study, Li (2009) determined that
the oil/water-phase inversion point for the Lubsoil ND-50 occurs
between 15 to 20% water cut. When the water-ow rate is low
and oil is continuous in the ow loop, it is hard to maintain a sta-
ble water ow rate because of the high discharge pressure. Thus,
most tests were conducted within the water-continuous region.
Live-oil viscosities with dissolved natural gas are 0.15, 0.28, and
0.57 Pas, corresponding to temperatures of 37.8, 26.7, and
15.6

C, respectively. The pressure in the test section is maintained


at approximately 2.59 MPa.
Experimental Results
Flow Pattern. Oil/water/gas three-phase ow patterns have been
viewed as a combination of gas/liquid and oil/water ow patterns
by authors including Bannwart et al. (2005), Trevisan and Bann-
wart (2006), Keskin et al. (2006), and Bannwart et al. (2009).
Gas/liquid ow patterns observed during this study are intermit-
tent (INT) and stratied (STR). For gas/liquid intermittent ow,
the oil/water ow patterns need to be specied in both the slug
body and the lm region. For gas/liquid stratied ow, oil/water
distribution in the liquid lm needs to be classied. Four ow pat-
terns are identied by analyzing the images obtained with the
high-speed video camera system for the three-phase ows in hori-
zontal and upward vertical pipes. Schematic drawings of the cross
sectional views and side views from the high-speed camera at
slug body and lm region for each ow pattern are shown in Figs.
3 through 12. Descriptions of the four ow patterns are presented
next.
1. INT(O/W-S&SOW-F). As shown in Figs. 3 and 4, gas and
liquid are in intermittent ow (or slug ow). Oil is dispersed in
water within the slug body. A continuous water layer may be pres-
ent at the bottom of the slug body depending on the turbulent in-
tensity. In the lm region, oil and water are stratied. Very few
gas bubbles are entrained by liquid in the lm region. This ow

Gas phase with
thick oil film on
the wall
Oil layer
Water layer
Oil and gas
dispersed
in water
(a) (b)
Fig. 3Horizontal intermittent ow with oil-in-water dispersion slug and stratied oil and water lm, denoted as INT(O/W-S&SOW-
F). (a) Slug body region; (b) liquid lm region.
(a) (b)
Fig. 4Snapshots of horizontal INT(O/W-S&SOW-F) (l
O
50.28 Pas, v
SO
50.1, v
SW
50.1, v
SG
51 m/s). (a) Slug body region; (b) liq-
uid lm region.
308 August 2013 SPE Production & Operations
pattern occurs at low supercial gas velocities such as v
SG
1 and
2 m/s and relatively low supercial oil and water velocities.
2. INT(O/W-S&O/W-F). As shown in Figs. 5 and 6, gas and
liquid are in intermittent ow (or slug ow). Oil is dispersed in
water in both the slug body and lm region. This is the dominant
ow pattern for horizontal ows in this study when supercial gas
velocity is at or higher than 2 m/s. In upward vertical ows, this is
the only ow pattern observed (as shown in Figs. 7 and 8). With
increasing either supercial oil or supercial water velocities, liq-
uid holdups or velocities would increase in the three-phase ow,
causing more intensive turbulence in both the slug body and lm
region. Therefore, the free water layer at the bottom of the slug
body disappears because of turbulent mixing. Stratied oil water
layers in the lm region changed to oil in water dispersion.
Depending on the ow rate, oil droplets could be distributed across
almost the whole liquid-lm section.
3. INT(W/O-S&W/O-F). As shown in Figs. 9 and 10, gas and
liquid ow pattern is intermittent (or slug). Water is dispersed in
oil in both the slug body and lm region. Oil is the continuous
phase. This is conrmed with the high frictional-pressure gradient
measurement in the test section. Because of water-pump

Gas phase with
thick oil film on
the wall
Oil dispersed in
water layer
Oil and gas
dispersed
in water
(a) (b)
Fig. 5Horizontal intermittent ow with oil-in-water dispersion slug and oil in water dispersion lm, denoted as INT(O/W-S&O/W-
F). (a) Slug body region; (b) liquid lm region.
(a) (b)
Fig. 6Snapshots of horizontal INT(O/W-S&O/W-F) (l
O
50.28 Pas, v
SO
50.3, v
SW
50.3, v
SG
52 m/s). (a) Slug body region; (b) liq-
uid lm region.

Water film
with oil
dispersion
Oil and gas
dispersed
in water
Gas core
(a) (b)
Fig. 7Vertical intermittent ow with oil-in-water dispersion slug and oil in water dispersion lm, denoted as INT(O/W-S&O/W-F).
(a) Slug body region; (b) liquid lm region.
(a) (b)
Fig. 8Snapshots of upward vertical INT(O/W-S&O/W-F)
(l
O
50.28 Pas, v
SO
50.3, v
SW
50.3, v
SG
52 m/s). (a) Slug body
region; (b) liquid lm region.
August 2013 SPE Production & Operations 309
limitations, only a few low water-cut tests close to the inversion
point were performed with this ow pattern.
4. STR(O/W-F). At supercial gas velocity v
SG
5 m/s, the
gas/liquid ow pattern changes from intermittent (INT) to strati-
ed (STR) ow. As shown in Figs. 11 and 12, oil and gas are
entrained and dispersed in the water-continuous phase, which
traveled on the pipe wall as a lm. Because of the high shear
stress between gas and liquid, liquid droplets are also entrained in
the gas phase. In the lm oil, water and gas are well mixed and
can be viewed as a homogenous mixture. The lm spreads upward
around the pipe wall, causing a concave interfacial shape.
It is observed that in horizontal ows, by keeping supercial
oil and water velocities constant, increasing the supercial gas ve-
locity will cause the ow pattern to shift from INT to STR ow.
On the other hand, keeping supercial gas velocities constant and
increasing supercial oil or water velocity will cause the ow pat-
tern in the lm region to change from stratied oil water (SOW)
to oil in water (O/W) or water in oil (W/O) dispersion, depending
on water fraction and inversion point.
In upward vertical ows within the current experimental
ranges, the only ow pattern observed is INT(O/W-S&O/W-F).
The oil and water are well mixed in the slug body with signicant
entrained gas. In the lm, the oil and water are also mixed. Gas
stays in the liquid lm until being picked up by the next upcoming
slug. At low gas-ow rate, the liquid lm falls downward and is
picked up by the upcoming slug.
In Fig. 13, current high-viscosity oil/water/gas three-phase hor-
izontal ow patterns are compared with Keskin et al. (2007) low-
viscosity (l
O
0.0135 Pas) horizontal three-phase ow patterns
at 50% water cut. The brown and blue solid curves are the oil/gas
and water/gas ow pattern transition boundaries predicted by the
Zhang and Sarica (2006) unied model assuming two-phase ow
conditions. Because of the narrower test range and high oil viscos-
ity of this study, only three out of twelve ow patterns identied

Gas phase with
thick oil film on
the wall
Water
dispersed in oil
layer
Water and
gas dispersed
in oil
(a) (b)
Fig. 9Horizontal intermittent ow with water-in-oil dispersion slug and water in oil dispersion lm, denoted as INT(W/O-S&W/O-
F). (a) Slug body region; (b) liquid lm region.
(a) (b)
Fig. 10Snapshots of horizontal INT(W/O-S&W/O-F) (l
O
50.28 Pas, v
SO
50.5, v
SW
50.1, v
SG
51 m/s). (a) Slug body region; (b) liq-
uid lm region.

Liquid Film Region
Gas phase with
oil film on the
pipe wall
Oil and gas
dispersed
in water layer
Fig. 11Horizontal stratied gas-liquid ow with oil in water
dispersion lm, denoted as STR(O/W-F).
Liquid Film Region
Fig. 12Snapshot of Horizontal STR(O/W-F) (l
O
50.28 Pas,
v
SO
50.3, v
SW
50.3, v
SG
55 m/s).
310 August 2013 SPE Production & Operations
by Keskin can be used for comparison. They are intermittent-water
continuous (IN-WC), stratied-stratied (ST-ST), and stratied-
water continuous (ST-WC). Unlike current ow pattern classica-
tions, Keskin et al. didnt consider the differences between oil and
water distributions in the liquid slug and lmregion. At high viscos-
ity, oil moves slower because of high friction. Compared with low-
viscosity oil multiphase ow, the lm height is much greater
because of its slower velocity (as shown in Figs. 4, 6, 10, and 12).
The low oil velocity and high liquid holdup are the main reasons
that the slug-ow region on a ow pattern map expands signi-
cantly compared with low-viscosity oil multiphase owas observed
in experiments and predicted by the Zhang and Sarica UM.
Pressure Gradient. Three-Phase Pressure Gradient. In hori-
zontal ow, pressure gradient is only caused by friction between
the uids and the pipe wall, assuming the acceleration pressure
gradient is negligible. In upward vertical ow, frictional and grav-
itational terms are presented separately. The inversion point of the
oil and water mixture is at approximately 20% water cut, based
on laboratory experiments.
The measured frictional three-phase pressure gradients are
plotted with respect to water cut at 0.5-m/s supercial oil veloc-
ities in Figs. 14 and 15a. Both horizontal and upward vertical
cases show that with the increase of supercial gas velocities, fric-
tional pressure gradient increases. There is also a clear trend of
the pressure gradient increasing within the oil-continuous region,
peaking around the inversion point, and then decreasing again
with the increase of water cut. At around 4050% water cut, it
reaches the minimum value and then increases again. This means
there is an optimum water cut that gives the minimum pressure
gradient. The water cut increase (by increasing v
SW
) within the
oil-continuous region does not reduce the pressure gradient. The
minimum pressure gradient seems to move to lower water cut
with the increase of supercial oil velocities. Probably with more
oil in the pipe, core annular ow can be formed, and the water-
lubrication effect is more signicant.
In upward vertical ows, gravitational pressure drop is meas-
ured from the hydrostatic head of the uids trapped by the QCVs
at both ends of the test section. Frictional pressure can be calcu-
lated by subtracting the gravitational pressure drop from the total
0.01
0.1
1
10
0.1 1 10 100

S
L

(
m
/
s
)

SG
(m/s)
0.01
0.1
1
10
0.1 1 10 100

S
L

(
m
/
s
)

SG
(m/s)
IN-WC
ST-ST
ST-WC
Oil/Gas
Water/Gas
DB
SLUG
ANN
Oil/Gas
Water/Gas
INT(O/W-S&SOW-F)
INT(O/W-S&O/W-F)
STR(O/W-F)
DB
SLUG
ANN
Fig. 13Three-phase ow pattern comparison between Keskin (2007) observations and current study for horizontal ows with
50%water cut. (a) Keskin (2007) observations; (b) current study (l
o
50.15 Pas).
0
500
1000
1500
2000
2500
0.0 0.2 0.4 0.6 0.8 1.0

d
p
/
d
L
F
,

E
x
p

(
P
a
/
m
)
Water Cut
V
sg
= 1 m/s
V
sg
= 2 m/s
V
sg
= 5 m/s

SO
= 0.5 m/s
Fig. 14Three-phase frictional pressure-gradient measure-
ments in horizontal ow at l
o
50.15 Pas.
0
1000
2000
3000
0.0 0.2 0.4 0.6 0.8 1.0

d
p
/
d
L
F
,

E
x
p

(
P
a
/
m
)
Water Cut
0.0 0.2 0.4 0.6 0.8 1.0
Water Cut
(a) (b)
V
sg
= 1 m/s
V
sg
= 2 m/s
V
sg
= 5 m/s
0
2000
4000
6000
8000

d
p
/
d
L
G
,

E
x
p

(
P
a
/
m
)
V
sg
= 1 m/s
V
sg
= 2 m/s
V
sg
= 5 m/s

SO
= 0.5 m/s
SO
= 0.5 m/s
Fig. 15Three-phase pressure-gradient measurements in upward vertical ow at l
o
50.15 Pas. (a) Frictional pressure gradient
vs. water cut; (b) gravitational pressure gradient vs. water cut.
August 2013 SPE Production & Operations 311
pressure drop, which is measured by the differential pressure
transducer. At the same supercial oil velocity and water cut, the
higher the supercial gas velocity, the lower the gravitational
pressure gradient, as shown in Fig. 15b. This is caused by the
lower liquid holdups at higher gas velocities. Because the gravita-
tional pressure gradient term is dominant in upward vertical ow,
the total pressure gradient shows the same trend as the gravita-
tional pressure gradient.
Comparison With Two-Phase Flow. Horizontal oil/gas two-
phase data were acquired in the current study. Vertical oil/gas
two-phase ow data were collected by Akhiyarov (2010) for his
MS thesis using the same ow loop, with the same uids and sim-
ilar test conditions. These data are used to study the water-lubrica-
tion effect for oil/gas two-phase ow and the gas-injection effect
for oil/water two-phase ow, respectively, in Figs. 16 and 17.
As shown in Fig. 16, the same supercial gas and total super-
cial liquid velocities are used to compare oil/water/gas three-
phase ow and oil/gas two-phase ow results in horizontal pipes.
At low water cut (e.g., 17% with oil/water either stratied or
water in oil dispersed), the three-phase frictional pressure gra-
dients are comparable or slightly higher than the correspondent
two-phase frictional pressure gradients. This suggests that water
injection does not help to reduce pressure gradient in oil continu-
ous region. By increasing water cut to 23, 38, and then 50%, fric-
tional pressure gradient is signicantly reduced to as low as 30%.
After water cut is higher than 50%, further increasing the water
ow rate will not produce any more benet.
Fig. 17 presents current oil/water/gas three-phase pressure gra-
dients compared against Akhiyarov (2010) oil/gas two-phase
upward vertical ow pressure gradients. The supercial liquid ve-
locity is the same for both current three-phase ow and Akhiyar-
ovs two-phase ow. Due to different test matrices and insufcient
data, only 50% water cut three-phase ow pressure drop experi-
mental results are used to compare with oil/gas two-phase ow. It
can be observed when water is used to replace part of the oil in the
upward vertical oil/gas ow that the frictional pressure gradient is
greatly reduced. However, the gravitational pressure gradients are
close between two-phase and three-phase ows. In upward vertical
pipe with medium ow rates, the gravitational pressure gradient is
usually much higher than the correspondent frictional pressure
gradient. As a result, the reduction on the total pressure gradient is
limited with water injection based on the observations in this
study. In upward vertical ow, water is mixed with oil and some
entrained gas. As long as the water remains continuous, water cut
change may not affect the overall liquid holdup, which determines
the gravitational pressure gradient.
Comparison With Model Predictions. The measured three-
phase pressure gradients are compared with the predictions in the
Zhang and Sarica (2006) UM in Figs. 18 and 19. Zhang and Sarica
developed the UM based on slug dynamics. Both slug characteris-
tics and transitions from slug ow to other ow regimes can be pre-
dicted by solving the continuity and momentum equations of slug
ow. It can be seen that there is signicant scatteredness in the
comparisons. Most of the overpredictions correspond to horizontal
0
1000
2000
3000
0 1000 2000 3000

d
p
/
d
L
F
,

T
h
r
e
e

(
P
a
/
m
)
dp/dL
F, Two
(Pa/m)
WC = 17%
WC = 23%
WC = 38%
WC = 50%
WC = 63%
WC = 77%
+50%
70%
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
0 0.2 0.4 0.6 0.8 1

d
p
/
d
L
F
,

T
h
r
e
e
/

d
p
/
d
L
F
,

T
w
o
Water Cut
(b) (a)
INT(O/W-S&SOW-F)
INT(O/W-S&O/W-F)
STR(O/W-F)
Fig. 16Oil/water/gas frictional pressure gradient compared with oil/gas frictional pressure gradient (current study) in horizontal
ow at l
o
50.15 Pas, various water cut, same v
SL
and v
SG
. (a) Three-phase vs. two-phase pressure gradient; (b) pressure-gradient
ratio vs. water cut.
0
2000
4000
6000
8000
10000
0 2000 4000 6000 8000 10000

d
p
/
d
L
T
h
r
e
e

(
P
a
/
m
)
dp/dL
Two
(Pa/m)
Gravitational
Frictional
Total
+10%
35%
0
0.4
0.8
1.2
1.6
2
8 6 4 2 0 10 12

d
p
/
d
L
T
h
r
e
e
/

d
p
/
d
L
T
w
o

M
(m/s)
Gravitational
Frictional
Total
(a) (b)
Fig. 17Three-phase pressure gradient compared with Akhiyarov (2010) oil/gas pressure gradient in upward vertical ow at
l
o
50.15 Pas, 50% water cut, same v
SL
and v
SG
, all corresponding to INT(O/W-S&O/W-F). (a) Three-phase vs. two-phase pressure
gradient; (b) pressure-gradient ratio vs. mixture velocity.
312 August 2013 SPE Production & Operations
slug ows with oil in water slug and stratied oil water lm INT(O/
W-S&SOW-F). For vertical ow, the UM can predict the positive
frictional pressure gradient because the liquid lm falling back is
allowed in the momentum equations. The gravitational pressure
gradient is underpredicted. This is caused by the underprediction of
the liquid holdup by the model. The UMalso underpredicts the ver-
tical frictional pressure gradient as a general trend.
Holdups. Three-Phase Holdup. Fig. 20 shows the oil and water
holdups in horizontal ows. It is shown that with the increase of
water cut, oil holdup decreases while the water holdup increases.
With the combined effect, the total liquid holdup in the pipe is rel-
atively stable, with a small dip at around 4050% water cut corre-
sponding to all supercial gas velocities. Holdups in upward
vertical pipes exhibit similar trends.
0
500
1000
1500
2000
2500
3000
0 500 1000 1500 2000 2500 3000

d
p
/
d
L
F
,

U
M

(
P
a
/
m
)
dP/dL
F, Exp
(Pa/m)
0.28 Pas Vertical
0.15 Pas Vertical
+40%
90%
0
2000
4000
6000
8000
0 2000 4000 6000 8000

d
P
/
d
L
G
,

U
M

(
P
a
/
m
)
dP/dL
G, Exp
(Pa/m)
0.28 Pas Vertical
0.15 Pas Vertical
50%
(a) (b)
Fig. 19Measurements of pressure gradient compared with Zhang and Sarica (2006) UM predictions in upward vertical three-
phase ows. (a) UM predictions vs. frictional pressure gradient; (b) UM predictions vs. gravitational pressure gradient.
0
1000
2000
3000
4000
0 1000 2000 3000 4000

d
p
/
d
L
F
,

U
M

(
P
a
/
m
)
dp/dL
F, Exp
(Pa/m)
0.57 Pas Horizontal
0.28 Pas Horizontal
0.15 Pas Horizontal
+300%
40%
0
1
2
3
4
5
6
0 2 4 6 8 10

d
p
/
d
L
M
o
d
e
l
/

d
p
/
d
L
E
x
p

M
(m/s)
0.57 Pas Horizontal
0.28 Pas Horizontal
0.15 Pas Horizontal
(a) (b)
Fig. 18Measurements of frictional pressure gradient compared with Zhang and Sarica (2006) unied model (UM) predictions in
horizontal three-phase ows. (a) UM predictions vs. frictional pressure gradient; (b) pressure-gradient ratio vs. mixture velocity.
0.0
0.2
0.4
0.6
0.8
1.0
0.0 0.2 0.4 0.6 0.8 1.0
H
O
Water Cut
(a)
0.0
0.2
0.4
0.6
0.8
1.0
0.0 0.2 0.4 0.6 0.8 1.0
H
W
Water Cut
(a)
V
sg
= 1 m/s
V
sg
= 2 m/s
V
sg
= 5 m/s

SO
= 0.5 m/s
V
sg
= 1 m/s
V
sg
= 2 m/s
V
sg
= 5 m/s

SO
= 0.5 m/s
Fig. 20Measured holdups in oil/water/gas horizontal ows for l
o
50.15 Pas. (a) Oil holdup vs. water cut; (b) water holdup vs.
water cut.
August 2013 SPE Production & Operations 313
Comparison With Two-Phase Flow. Approximately 50 hori-
zontal oil/gas ow experiments were performed using the same
uids and facilities along with the three-phase ow study. At the
same total supercial liquid velocities, experimental results show
liquid holdups in horizontal oil/water/gas ow are generally lower
than horizontal oil/gas ow (Fig. 21). This suggests that in hori-
zontal three-phase ow, the liquid phase ows faster because of
water lubrication. In upward vertical ow, liquid holdups are gen-
erally the same between the two-phase and three-phase ows, as
shown in Fig. 22.
Comparison With Model Predictions. The three-phase hold-
ups are compared with Zhang and Sarica (2006) UM predictions
in Fig. 23. As a general trend, the UM underpredicts most of the
oil and water holdups. The discrepancy between the predicted val-
ues and experimental results is likely caused by the closure rela-
tionships used within the UM. These correlations were developed
on the basis of low-viscosity oil multiphase ow experimental
results. Correlations suitable for both low- and high-viscosity oil
multiphase ows are needed to improve the model performance.
Conclusions
Experiments on high-viscosity oil/water/gas ows in horizontal
and upward vertical pipes have been carried out. Four three-phase
ow patterns are identied by combining gas/liquid ow pattern
and oil/water mixing status within experimental range. Compared
with low-viscosity multiphase ow, the slug ow region expands
signicantly due to high oil viscosity and low liquid lm velocity.
Water injection can lubricate the ow and signicantly reduce the
frictional pressure gradient. This is most benecial for horizontal
ows. The maximum reduction of horizontal oil/gas pressure gra-
dient occurs at approximately 4050% water cut. For upward ver-
tical ow, the water injection effect on total pressure gradient is
limited due to the dominance of gravitational pressure gradient.
The Zhang and Sarica (2006) UM gives reasonable predictions
for water continuous ow. However, the pressure drop and hold-
ups are underpredicted as a general trend. The closure relation-
ships (such as slug liquid holdup, slug translational velocity,
interfacial friction factor, and entrainment fraction) used in the
unied model were developed on the basis of low-viscosity multi-
phase ow experimental results. They need to be modied, or
new closure relationships need to be developed, to fully incorpo-
rate the viscosity effect.
Nomenclature
dp/dL
F
frictional pressure gradient, Pa/m
dp/dL
G
gravitational pressure gradient, Pa/m
dp/dL
T
total pressure gradient, Pa/m
Exp experimental results
G-L SEP gas/liquid separator
H
L
total liquid holdup
H
O
oil holdup
H
W
water holdup
UM Zhang and Sarica (2006) unied model
v
M
mixture velocity, m/s
v
SG
supercial gas velocity, m/s
v
SL
supercial liquid velocity, m/s
0
0.2
0.4
0.6
0.8
1
0 0.2 0.4 0.6 0.8 1
H
L
,

T
h
r
e
e
H
L, Two
0.15 Pas Horizontal
+15%
25%
0.5
0.7
0.9
1.1
1.3
0 2 4 6 8
H
L
,

T
h
r
e
e
/
H
L
,

T
w
o

M
(m/s)
0.15 Pas Horizontal
(a) (b)
Fig. 21Three-phase total liquid holdups compared with current oil/gas liquid holdup in horizontal ows for l
o
50.15 Pas, same
v
SL
and v
SG
. (a) Three-phase vs. two-phase liquid holdup; (b) holdup ratio vs. mixture velocity.
0
0.2
0.4
0.6
0.8
1
0 0.2 0.4 0.6 0.8 1
H
L
,

T
h
r
e
e
H
L, Two
0.15 Pas Vertical +20%
10%
0.5
0.7
0.9
1.1
1.3
1.5
0 2 4 6 8
H
L
,

T
h
r
e
e
/
H
L
,

T
w
o

M
(m/s)
0.15 Pas Vertical
(a) (b)
Fig. 22Three-phase total liquid holdup compared with Akhiyarov (2010) oil/gas liquid holdup in upward vertical ows for
l
o
50.15 Pas, same v
SL
and v
SG
. (a) Three-phase vs. two-phase liquid holdup; (b) holdup ratio vs. mixture velocity.
314 August 2013 SPE Production & Operations
v
SO
supercial oil velocity, m/s
v
SW
supercial water velocity, m/s
l
o
oil viscosity, Pas
Acknowledgment
The authors would like to thank the TUHOP member companies
for supporting this research project.
References
Acikgoz, M., Franca, F., and Lahey Jr, R.T. 1992. An experimental study
of three-phase ow regimes. Int. J. Multiphase Flow 18 (3): 327336.
http://dx.doi.org/10.1016/0301-9322(92)90020-H.
Akhiyarov, D.T., Zhang, H.-Q., and Sarica, C. 2010. High-Viscosity Oil-
Gas Flow in Vertical Pipe. Presented at the Offshore Technology Con-
ference, Houston, 36 May. OTC-20617-MS. http://dx.doi.org/
10.4043/20617-MS.
Akhiyarov, D.T. 2010. An Experimental Study on High-Viscosity Oil/Gas
Upward Flow in Vertical Pipes. MS thesis, the University of Tulsa,
Tulsa, Oklahoma.
Bannwart, A.C., Rodriguez, O.M.H., Trevisan, F.E. et al. 2009. Experi-
mental investigation on liquidliquidgas ow: Flow patterns and
pressure-gradient. J. Pet. Sci. Eng. 65 (12): 113. http://dx.doi.org/
10.1016/j.petrol.2008.12.014.
Bannwart, A.C., Vieira, F., Carvalho, C.-H.M. et al. 2005. Water-Assisted
Flow of Heavy Oil and Gas in a Vertical Pipe. Presented at the SPE/
PS-CIM/CHOA International Thermal Operations and Heavy Oil
Symposium, Calgary, 13 November. SPE-97875-MS. http://
dx.doi.org/10.2118/97875-MS.
Hewitt, G.F. 2005. Three-phase gasliquidliquid ows in the steady and
transient states. Nucl. Eng. Des. 235 (1012): 13031316. http://
dx.doi.org/10.1016/j.nucengdes.2005.02.023.
Keskin, C., Zhang, H.-Q., and Sarica, C. 2007. Identication and Classi-
cation of New Three-Phase Gas/Oil/Water Flow Patterns. Presented at
the SPE Annual Technical Conference and Exhibition, Anaheim, Cali-
fornia, USA, 1114 November. SPE-110221-MS. http://dx.doi.org/
10.2118/110221-MS.
Langsholt, M. and Holm, H. 2001. Oil-Water-Gas in Steeply Inclined
Pipes. Presented at the 10th International Conference (Multiphase
01), Cannes, France, 1315 June.
Li, M. 2009. High-Viscosity Oil/Water Emulsion Rheology. Presented at
the 4th Advisory Board Meeting of Tulsa University High-Viscosity
Oil Projects (TUHOP), Tulsa, 24 March.
Pan, L., Jayanti, S., and Hewitt, G.F. 1995. Flow Patterns, Phase Inversion
and Pressure Gradient in Air-Oil-Water Flow in a Horizontal Pipe.
Presented at the 2nd International Conference on Multiphase Flow:
95 Kyoto, Kyoto, Japan, 37 April.
Poesio, P., Strazza, D., and Sotgia, G. 2009. Very-viscous-oil/water/air
ow through horizontal pipes: Pressure drop measurement and predic-
tion. Chem. Eng. Sci. 64 (6): 11361142. http://dx.doi.org/10.1016/
j.ces.2008.10.061.
Sridhar, S., Zhang, H.-Q., Sarica, C. et al. 2011. Experiments and Model
Assessment on High-Viscosity Oil/Water Inclined Pipe Flows. Pre-
sented at the SPE Annual Technical Conference and Exhibition, Den-
ver, 30 October2 November. SPE-146448-MS. http://dx.doi.org/
10.2118/146448-MS.
Trevisan, F. and Bannwart, A.C. 2006. Three-Phase Flow Patterns and
Pressure Drop for Heavy Crude Oil-Water-Gas Horizontal Flow. Pre-
sented at the 13th International Heat Transfer Conference, Sydney,
Australia, Paper IHTC 13-1322.
Vuong, D.H., Zhang, H.-Q., Sarica, C. et al. 2009. Experimental Study on
High Viscosity Oil/Water Flow in Horizontal and Vertical Pipes. Pre-
sented at the SPE Annual Technical Conference and Exhibition, New
Orleans, 47 October. SPE-124542-MS. http://dx.doi.org/10.2118/
124542-MS.
0
0.2
0.4
0.6
0.8
1
H
O
,

U
M
0.57 Pas Horizontal
0.28 Pas Horizontal
0.15 Pas Horizontal
0.28 Pas Vertical
0.15 Pas Vertical
+100%
80%
0
0.5
1
1.5
2
2.5
H
O
,

U
M
/
H
O
,

E
x
p
0.57 Pas Horizontal
0.28 Pas Horizontal
0.15 Pas Horizontal
0.28 Pas Vertical
0.15 Pas Vertical
0
0.2
0.4
0.6
0.8
1
0 0.2 0.4 0.6 0.8 1
H
W
,

U
M
H
W
,
Exp
0.57 Pas Horizontal
0.28 Pas Horizontal
0.15 Pas Horizontal
0.28 Pas Vertical
0.15 Pas Vertical
+15%
60%
0
0.5
1
1.5
2
0 2 4 6 8
H
W
,

U
M
/
H
W
,

E
x
p

M
(m/s)
0 0.2 0.4 0.6 0.8 1
H
O
,
Exp
0 2 4 6 8

M
(m/s)
0.57 Pas Horizontal
0.28 Pas Horizontal
0.15 Pas Horizontal
0.28 Pas Vertical
0.15 Pas Vertical
(a) (b)
(c) (d)
Fig. 23Measured holdups compared with Zhang and Sarica (2006) UM predictions in three-phase ow. (a) UM predictions vs. oil
holdup experimental results; (b) oil holdup ratio vs. mixture velocity; (c) UM predictions vs. water holdup experimental results; (d)
water holdup ratio vs. mixture velocity.
August 2013 SPE Production & Operations 315
Woods, G.S., Spedding, P.L., Watterson, J.K. et al. 1998. Three-Phase
Oil/Water/Air Vertical Flow. Chem. Eng. Res. Des. 76 (5): 571584.
http://dx.doi.org/10.1205/026387698525252.
Zhang, H.-Q. and Sarica, C. 2006. Unied Modeling of Gas/Oil/Water
Pipe FlowBasic Approaches and Preliminary Validation. SPE Proj
Fac & Const 1 (2): 17. SPE-95749-PA. http://dx.doi.org/10.2118/
95749-PA.
Shufan Wang is currently a Flow Assurance Consultant with
MSi Kenny. email: shufan.wang@woodgroupkenny.com. His
research interests include oil/water/gas multiphase flow, high-
viscosity oils, and slug-flow modeling. Before joining the Univer-
sity of Tulsa (TU), he focused mainly on CO
2
pipeline corrosion.
He holds a BS degree in materials science and engineering
from the Beijing University of Chemical Technology, an MS
degree in chemical engineering from Ohio University, and a
PhD degree in petroleum engineering from TU.
Hong-Quan Zhang is an Associate Professor of Petroleum Engi-
neering at TU, and serves as the director of the universitys Arti-
ficial Lift Projects (TUALP, www.tualp.utulsa.edu). email: hong-
quan-zhang@utulsa.edu. From 19982003, he was a senior
research associate of the Tulsa University Fluid Flow Projects
(TUFFP). From 20032012, he served as the Associate Director
of TUFFP and principal investigator of the Tulsa University High-
Viscosity Oil Projects (TUHOP). Before joining TU in 1998, he was
an Associate Professor and Professor at Tianjin University. In
1993 and 1994, as an Alexander von Humboldt Research Fel-
low, he conducted research at the Max Planck Institute of
Fluid Mechanics and the German Aerospace Research Estab-
lishment in Go ttingen, Germany. Zhang holds BS and MS
degrees from Xian Jiaotong University and a PhD degree from
Tianjin University, China.
Cem Sarica is currently a Professor of Petroleum Engineering
and the director of three industry-supported consortia at TU:
Fluid Flow, Paraffin Deposition, and Horizontal Well Artificial Lift
Projects. He was an Associate Professor of Petroleum and Nat-
ural Gas Engineering at the Pennsylvania State University and
an Assistant Professor of Petroleum and Natural Gas Engineer-
ing at Istanbul Technical University (ITU) before joining TU. He
has over 100 publications, mostly in SPE journals and proceed-
ings, and his research interests include production engineer-
ing, multiphase flow in pipes, flow assurance, and horizontal
wells. He holds BS and MS degrees in petroleum engineering
from ITU and a PhD degree in petroleum engineering from TU.
He currently serves as a member of SPE Projects, Facilities and
Construction Advisory Committee and a member of the SPE
Production and Operations Award Committee. He has previ-
ously served as a member of the SPE Production Operations
and Books Committees, and was a member of the SPE Journal
Editorial Board between 19992007. He is also the recipient of
the 2010 SPE International Production and Operations Award,
and was recognized as a Distinguished Member of SPE in 2012.
Eduardo Pereyra is a research associate with the Fluid Flow
Project at TU. His research interests include multiphase flow sys-
tems and transport, flow assurance, and separation technolo-
gies. Pereyra has appeared in several refereed journals and
has written conference papers in these areas. Pereyra holds
two BE degrees, one in mechanical engineering and one in
systems engineering, from the University of Los Andes, Vene-
zuela, and MS and PhD degrees in petroleum engineering
from TU.
316 August 2013 SPE Production & Operations

S-ar putea să vă placă și