Está en la página 1de 12

Unified Mechanistic Model

for Steady-State Two-Phase Flow:


Horizontal to Vertical Upward Flow
L.E. Gomez, SPE, Ovadia Shoham, SPE, and Zelimir Schmidt,* SPE, U. of Tulsa;
R.N. Chokshi,** SPE, Zenith ETX Co.; and Tor Northug, Statoil

Summary the mechanisms and the important parameters of the flow. All
A unified steady-state two-phase flow mechanistic model for the current research is conducted through the modeling approach. Ap-
prediction of flow pattern, liquid holdup and pressure drop is pre- plication of models in the field is now underway, showing the
sented that is applicable to the range of inclination angles from potential of this method.
horizontal (0) to upward vertical flow (90). The model is based The mechanistic models developed over the past two decades
on two-phase flow physical phenomena, incorporating recent de- have been formulated separately for pipelines and wellbores. Fol-
velopments in this area. It consists of a unified flow pattern pre-
lowing is a brief review of the literature for these two cases.
diction model and unified individual models for stratified, slug,
bubble, annular and dispersed bubble flow. The model can be
Pipeline Models. These models are applicable for horizontal and
applied to vertical, directional and horizontal wells, and
near horizontal flow conditions, namely, 10. The pioneering
horizontal-near horizontal pipelines. The proposed model imple-
ments new criteria for eliminating discontinuity problems, provid- and most durable model for flow pattern prediction in pipelines
ing smooth transitions between the different flow patterns. was presented by Taitel and Dukler.6 Other studies have been
The new model has been initially validated against existing, carried out for the prediction of specific transitions, such as the
various, elaborated, laboratory and field databases. Following the onset of slug flow,7 or different flow conditions, such as high
validation, the model is tested against a new set of field data, from pressure.8 Separate models have been developed for stratified
the North Sea and Prudhoe Bay, Alaska, which includes 86 cases. flow,6,9-11 slug flow,12-14 annular flow15,16 and dispersed bubble
The proposed model is also compared with six commonly used flow the homogeneous no-slip model17. A comprehensive
models and correlations. The model showed outstanding perfor- mechanistic model, incorporating a flow pattern prediction model
mance for the pressure drop prediction, with a 1.3% average and separate models for the different flow patterns, was presented
error, a 5.5% absolute average error and 6.2 standard deviation. by Xiao et al.18 for pipeline design.
The proposed model provides an accurate two-phase flow mecha-
nistic model for research and design for the industry. Wellbore Models. These models are applicable mainly for verti-
cal flow but can be applied as an approximation for off-vertical
sharply inclined flow (60 90) also. A flow pattern predic-
Introduction tion model was proposed by Taitel et al.19 for vertical flow, which
Early predictive means for two-phase flow were based on the was later extended to sharply inclined flow by Barnea et al.20
empirical approach. This was due to both the complex nature of Specific models for the prediction of the flow behavior have been
two-phase flow and the need for design methods for industry. The developed for bubble flow21,22 slug flow23-25 and annular flow.26,27
most commonly used correlations have been the Dukler et al.1 Comprehensive mechanistic models for vertical flow have been
and Beggs and Brill2 correlations for flow in pipelines, and the presented by Ozon et al.,28 by Hasan and Kabir,21 by Ansari
Hagedorn and Brown3 and Ros4/Duns and Ros5 correlations for et al.29 and by Chokshi et al.30
flow in wellbores. This approach was successful for solving two-
phase flow problems for more than 40 years, with an updated Unified Models. Attempts have been made in recent years to de-
performance of 30% error. However, the empirical approach velop unified models that are applicable for the range of inclina-
has never addressed the why and how problems for two- tion angles between horizontal (0) and upward vertical (90)
phase flow phenomena. Also, it is believed that no further or flow. These models are practical since they incorporate the incli-
better accuracy can be achieved through this approach. nation angle. Thus, there is no need to apply different models for
A new approach emerged in the early 1980s, namely, the the different inclination angles encountered in horizontal, inclined
mechanistic modeling approach. This approach attempts to shed and vertical pipes. A unified flow pattern prediction model was
more light on the physical phenomena. The flow mechanisms
presented by Barnea31 that is valid for the entire range of inclina-
causing two-phase flow to occur are determined and modeled
mathematically. A fundamental postulate in this method is the tion angles (90 90). Felizola and Shoham32 presented a
existence of various flow configurations or flow patterns, includ- unified slug flow model applicable to the inclination angle range
ing stratified flow, slug flow, annular flow, bubble flow, churn from horizontal to upward vertical flow. A unified mechanistic
flow and dispersed bubble flow. These flow patterns are shown model applicable to horizontal, upward and downward flow con-
schematically in Fig. 1. The first objective of this approach is, ditions was presented by Petalas and Aziz,33 which was tested
thus, to predict the existing flow pattern for a given system. Then against a large number of laboratory and field data. Recently,
a separate model is developed for each flow pattern to predict the Gomez et al.34 presented a unified correlation for the prediction of
corresponding hydrodynamics and heat transfer. These models are the liquid holdup in the slug body.
expected to be more reliable and general because they incorporate The above literature review reveals that separate comprehen-
sive mechanistic models are available for pipeline flow and well-
*Deceased. bore flow. Only very few studies have been published on unified
**Now with TanData Corp. modeling. The objective of this paper is to present a systematic,
Copyright 2000 Society of Petroleum Engineers comprehensive, unified model applicable for the range of inclina-
This paper (SPE 65705) was revised for publication from paper SPE 56520, presented at tion angles between horizontal (0) and vertical (90). This will
the 1999 SPE Annual Technical Conference and Exhibition held in Houston, 58 October.
Original manuscript received for review 20 October 1999. Revised manuscript received 20
provide more efficient computing algorithms, because the model
May 2000. Manuscript peer approved 9 June 2000. can be applied conveniently for both pipelines and wellbores,

SPE Journal 5 3, September 2000 1086-055X/2000/53/339/12/$5.000.50 339


d CD 2 0.4
L G g
1/2
. 4

The other critical diameter is applicable to shallow inclinations


(10) where, due to buoyancy, bubbles larger than this diameter
migrate to the upper part of the pipe causing creaming and
transition to slug flow as follows:

3 L f M v 2M
d CB . 5
8 L G g cos

Transition to dispersed bubble flow will occur when the maximum


possible bubble diameter, given by Eq. 3, is less than both critical
diameters given by Eqs. 4 or 5, namely,

d maxd CD and d CB . 6

The transition boundary given by Eq. 6 is valid for 0.52,


which represents the maximum possible packing of bubbles for a
cubic lattice configuration. For larger values of void fraction, ag-
glomeration of bubbles occurs, independent of the turbulence
forces, resulting in a transition to slug flow. This criterion is given
Fig. 1Flow patterns in pipelines and wellbores for horizontal by
to vertical flow patterns.
v SG
NS 0.52. 7
v SG v SL
without the need to switch among different models. The proposed
model will be evaluated against new field data, along with other Annular to Slug Transition. Two mechanisms are respon-
published models and correlations. sible for this transition from annular flow to slug flow, causing
blockage of the gas core by the liquid phase. The two mechanisms
Unified Model Formulation are based on the characteristic film structure of annular flow:
The unified model consists of a unified flow pattern prediction 1. Instability of the liquid film due to downward flow near the
model and separate unified models for the different existing flow pipe wall. The criterion for the instability of the film is obtained
patterns. These are briefly described below. from the simultaneous solution of the following two dimension-
less equations:
Unified Flow Pattern Prediction Model. The Barnea31 model is
applicable for the entire range of inclination angles, namely, from 175H L 1
Y X 2, 8
upward vertical flow to downward vertical flow (90 1H L 2.5H L H L3
90). Below is a summary of the applicable transition criteria
for this study, including the stratified to nonstratified, slug to dis- 2 3/2 H L
persed bubble, annular to slug and bubble to slug flow. Y X2 9
Stratified to Nonstratified Transition. The criterion for this
H L3 1 3/2 H L
transition is the same as the original one proposed by Taitel and where X is the Lockhart and Martinelli parameter and Y is a
Dukler,6 based on a simplified KelvinHelmholtz stability analy- dimensionless gravity group defined respectively by
sis given by

F 2
1 2
v G dA L /dh L
1, 1 d
4C L L v SL d
L 2
n
L v SL
2

dp
dL


SL


1h L 2 X 2 2 , 10
A G 4C G G v SG d m G v SG dp
where the superscript tilde symbol represents a dimension- d G 2 dL SG
less parameter length and area are normalized with d and d 2 ,
respectively, and the phase velocity is normalized with the corre- L G g sin


sponding superficial velocity. F is a dimensionless group given Y . 11
dp
by
dL SG

F
L
G

v SG
G dg cos
. 2 Note that Eq. 8 yields the steady-state solution for the liquid
holdup H L , while Eq. 9 yields the value of the liquid holdup that
satisfies the condition of the film instability.
Slug to Dispersed Bubble Transition. The slug to dispersed 2. Wave growth on the interface due to large liquid supply from
bubble transition occurs at high liquid flow rates, where the tur- the film. If sufficient liquid is provided, the wave will grow and
bulent forces overcome the interfacial tension forces, dispersing bridge the pipe, resulting in slug flow. The condition for occur-
the gas phase into small bubbles. The resulting maximum bubble rence of this mechanism is
size can be determined from


H L 0.24. 12

0.4
v SG 0.5
0.6
2 f M v 3M
d max 4.15 0.725 . 3 Transition from annular to slug flow will occur whenever one
vM L d
of the two criteria is satisfied. A smooth change between the two
Two critical bubble diameters are considered. The first is the mechanisms is obtained when the inclination angle varies over the
critical diameter below which bubbles do not deform, avoiding entire range of inclinations, or when a change occurs in the op-
agglomeration or coalescence, given by erationing conditions.

340 Gomez et al.: Unified Mechanistic Model for Steady-State Two-Phase Flow SPE Journal, Vol. 5, No. 3, September 2000
Bubble to Slug Transition. The transition from bubble to slug
flow occurs at relatively lower liquid flow rates compared to the
transition from slug to dispersed bubble flow. Under these condi-
tions the turbulent forces are negligible, and the transition is
caused by coalescence of bubbles at a critical gas void fraction of
0.25, as follows:

v SL
1

v SG 1.53 1 0.5g L G
L2 1/4
sin . 13

The bubble regime can exist at low liquid flow rates as given
by Eq. 13, provided that the pipe diameter is larger than
d19 ( L G ) / L2 g 0.5 and only for sharply inclined pipes Fig. 2Physical model for stratified flow.
with inclination angles between approximately 60 and 90.

Elimination of Transition Discontinuities. Mechanistic models


for the prediction of pressure traverses in multiphase flow are liquid velocity and the superficial gas velocity on the transition
notorious for creating discontinuities. This is the result of switch- boundary to annular flow, predicted by the Barnea model.31 This
ing from one flow pattern model to another as the transition averaging eliminates numerical problems and ensures a smooth
boundary is crossed. Different models are used for different flow pressure gradient across the slug to annular boundary.
patterns to predict the liquid holdup and pressure drop, which
might result in a discontinuity. In order to avoid this problem in Unified Stratified Flow Model. The physical model for stratified
the proposed model, the following criteria were implemented to flow is given in Fig. 2. A modified form of the Taitel and Dukler6
smooth the transitions between the different flow patterns. model is used here. Two modifications are introduced: the liquid
Bubble to Slug and Slug to Dispersed Bubble Transitions. wall friction factor is determined by Ouyang and Aziz35 and the
Near the transition boundaries from slug to bubble or dispersed interfacial friction factor is given by Baker et al.36
bubble flow, the liquid film/gas pocket region behind the slug Momentum Balances. The momentum force balances for the
body, namely, L F , becomes small. The short film/gas length can liquid and gas phases are given, respectively, by
prevent the slug flow model from converging. Thus, to solve this
problem, when slug flow is predicted near these transition bound- dp
aries, the following constraints were developed: A L WL S L I S I L A L g sin 0, 16
dL
if L F 1.2d and v SL 0.6 m/s bubble flow, dp
A G WG S G I S I G A G g sin 0. 17
if L F 1.2d and v SL 0.6 m/s dispersed bubble flow. dL
14
Eliminating the pressure gradient from Eqs. 16 and 17, the com-
The value L F /d1.2 is based on the mechanism that once the bined momentum equation for the two phases is obtained as fol-
Taylor bubble length approaches the pipe diameter, it becomes lows:


unstable and might break into small bubbles. Under these condi-
tions, for high superficial liquid velocities, due to turbulence in- SL SG 1 1
tensity and bubble breakup and dispersion, the resulting flow pat- WL WG IS I L G g sin 0.
AL AG AL AG
tern will be dispersed bubble flow. However, for low superficial 18
liquid velocities, due to low turbulence intensity and coalescence
of the small bubble to larger ones, the resulting flow pattern will The combined momentum equation is an implicit equation for h L
be bubble flow. or h L /d, the liquid level in the pipe. Solution of the equation,
Slug to Annular Transition. A two-fold problem is associated carried out by a trial and error procedure, requires the determina-
with this transition boundary. First, a discontinuity in the pressure tion of the different geometrical, velocity and shear stress vari-
gradient between slug flow and annular flow occurs. Also, if slug ables. Under high gas and liquid flow rates, multiple solutions can
flow is predicted near this transition boundary, due to the high gas occur. It can be shown that, in this case, the smallest of the three
rates, the film/gas zone becomes long, resulting in a very thin film solutions is the physical and stable solution.
thickness, one approaching zero. This can prevent the slug flow Once the liquid level h L /d is determined, the liquid holdup,
model from converging. To alleviate the two problems, a transi- H L , can be calculated in a straightforward manner from geometri-
tion zone is created between slug flow and annular flow based on cal relationships as follows:
the superficial gas velocity. The transition zone is predicted by the
critical velocity corresponding to the droplet model used by Taitel
et al.19 as follows: cos1 2 hL
d
hL
1 2 1
d 1 2
hL
d
1 2


H L .
g sin L G 0.25
v SG,crit 3.1 . 15 19
G2
Once the liquid holdup is determined, the pressure gradient can be
Thus, for a given superficial liquid velocity, the transition region
determined from either Eq. 16 or 17. Either equation provides the
is defined when the superficial gas velocity is greater than the
frictional and the gravitational pressure losses, and neglects the
critical gas velocity given in Eq. 15 and less than the superficial
gas velocity on the transition boundary to annular flow predicted accelerational pressure losses.
by the Barnea model31. Hence, when slug flow is predicted in the Closure Relationships. The wall shear stresses corresponding
transition zone, the pressure gradient is averaged between the to each phase are determined based on single-phase analysis using
pressure gradient under slug flow and annular flow conditions. the hydraulic diameter concept, as follows Fanning friction factor
The corresponding slug flow pressure gradient is calculated at the formulation:
given superficial liquid velocity and the critical superficial gas
velocity, given by Eq. 15. Similarly, the corresponding pressure L v L2 G v G2
WL f L and WG f G . 20
gradient under annular flow is calculated at the given superficial 2 2

Gomez et al.: Unified Mechanistic Model for Steady-State Two-Phase Flow SPE Journal, Vol. 5, No. 3, September 2000 341
The respective hydraulic diameters of the liquid and gas phases
are given by

4A L 4A G
d L and d G . 21
SL S G S I
The Reynolds numbers of each of the phases are

dL vL L dG vG G
N ReL and N ReG . 22
L G
Taitel and Dukler6 proposed that both the liquid and gas wall
friction factors, f L and f G , can be calculated using a standard
friction factor chart. However, Ouyang and Aziz35 found this pro-
cedure to be appropriate for the gas phase only. This is due to the
fact that the liquid wall friction factor can be affected significantly
by the interfacial shear stress, especially for low liquid holdup
conditions. Thus, f G is determined from a standard chart, while f L
is determined by a new correlation developed by Ouyang and
Aziz35 that incorporates the gas and liquid flow rates, given as

16
f G for N ReG 2,300, 23
N ReG Fig. 3Physical model for slug flow.



f G 0.001 375 1 2104
10
d N ReG
6


1/3

unit. Applying this balance on cross sections in the liquid slug


for N ReG 2,300, body and in the liquid film region gives, respectively,

0.0926 v M v SL v SG v LLS H LLS v GLS 1H LLS , 28


1.6291 v SG
f L 0.5161 . 24
N Re v SL v M v LTB H LTB v GTB 1H LTB . 29
L

The interfacial shear stress is given, by definition, as


Eq. 28 can be used to determine v LLS , the liquid velocity in the
G v G v L v G v L slug body, since the other variables are given in the form of clo-
I f I . 25 sure relationships. Then, the liquid film velocity, v LTB , can be
2 determined from Eq. 27 for a given liquid holdup in this region,
The interfacial friction factor for stratified smooth flow is taken as H LTB . Also, from Eq. 29 it is possible to determine v GTB , the gas
the friction factor between the gas phase and the wall. However, velocity in the gas pocket.
for stratified wavy flow, as suggested by Xiao et al.,18 the inter- The average liquid holdup in a slug unit is defined as
facial friction factor is that given by Baker et al.36
H LLS L S H LTB L F
Unified Slug Flow Model. The unified and comprehensive analy- H LU . 30
LU
sis of slug flow, presented by Taitel and Barnea,37 is used in the
present study with the following features: a uniform film along the Using Eqs. 2628, the expression for the liquid holdup becomes
liquid film/gas pocket zone; a global momentum balance on a slug
unit for pressure drop calculations, and a new correlation Gomez
et al.34 for the liquid holdup in the slug body. The original Taitel v TB H LLS v GLS 1H LLS v SG
H LU . 31
and Barnea37 model was extended to vertical flow by assuming a v TB
symmetric film around the Taylor bubble for inclination angles
between 86 and 90. Eq. 31 shows an interesting result, namely, that the average liquid
With the above characteristics, the original model is simplified holdup in a slug unit is independent of the lengths of the different
considerably, as given below, avoiding the need for numerical slug zones.
integration along the liquid film region. The proposed simplified Hydrodynamics of the Liquid Film. Considering a uniform
model is considered to be sufficiently accurate for practical appli- liquid film thickness, a combined momentum equation, similar to
cations. Refer to Fig. 3 for the physical model for slug flow. that in the case of stratified flow, can be obtained for the film/gas
Mass Balances. An overall liquid mass balance over a slug pocket zone as follows:


unit results in
WF S F WG S G 1 1
LS LF IS I L G g sin 0.
v SL v LLS H LLS v LTB H LTB . 26 AL AG AF AG
LU LU 32
A mass balance can also be applied between two cross- Solution of Eq. 32 yields the uniform equilibrium film thickness
sectional areas, namely, in the slug body and in the film region, in or the liquid holdup in this region, H LTB . This value can be used,
a coordinate system moving with the translational velocity, v TB , in a trial and error procedure, to determine the gas and liquid
yielding velocities in the slug and film/gas pocket regions, as discussed
below Eq. 29.
v TB v LLS H LLS v TB v LTB H LTB . 27
The liquid film length can be determined from
A continuity balance on both liquid and gas phases results in a
constant volumetric flow rate through any cross section of the slug L F L U L S . 33

342 Gomez et al.: Unified Mechanistic Model for Steady-State Two-Phase Flow SPE Journal, Vol. 5, No. 3, September 2000
The slug length, L S , is given as a closure relationship while the
slug unit length, L U , can be determined from Eq. 26, as follows:

v LLS H LLS v LTB H LTB


L U L S . 34
v SL v LTB H LTB

Pressure Drop Calculations. The pressure drop for a slug


unit can be calculated using a global force balance along a slug
unit. Since the momentum fluxes in and out of the slug unit con-
trol volume are identical, the pressure drop across this control
volume for a uniform liquid film is

dp S d L S WF S F WG S G L F
U g sin , 35
dL A LU A LU
where U is the average density of the slug unit given by

U H LU L 1H LU G . 36
The first term on the right-hand side of Eq. 35 is the gravitational
pressure gradient, whereas the second and third terms represent
the frictional pressure gradient that results from the frictional
losses in the slug and in the film/gas pocket regions. No accelera-
tional pressure drop occurs in the slug unit control volume formu-
lation.
Closure Relationships. The proposed model requires four clo-
sure relationships, namely, the liquid slug length, L S , the liquid
holdup in the slug body, H LLS , the slug translational velocity, Fig. 4Physical model for annular flow.
v TB , and the gas velocity of the small bubbles entrained in the
liquid slug, v GLS . The closure relationships are given below.
A constant length of L S 30d and L S 20d is used for fully


developed and stable slugs in horizontal and vertical pipes, re-
spectively. For inclined flow, an average slug length is used based SF SI dp
WF I L g sin 0, 41
on inclination angle. However, for horizontal and near horizontal AF AF dL F
( 1) large diameter pipes (d2 in.), the Scott et al.38 cor-


relation is used, as given below
SI dp
I C g sin 0. 42
ln L S 25.428.5 ln d 0.1, 37 AC dL C

where d is expressed in inches and L S is in feet.


The liquid holdup in the slug body, H LLS , is predicted using Eliminating the pressure gradients from the equations results in
the Gomez et al.34 unified correlation, given by the combined momentum equation for annular flow, namely,


3 2.48106 N
H LLS 1.0e (7.8510 ReSL ) , 0 900 , 38 SF 1 1
WF IS I L C g sin 0. 43
AF AF AC
where the slug superficial Reynolds number is calculated as

Lv M d Eq. 43 is an implicit equation for the film thickness or /d that


N ReSL . 39 can be solved by trial and error, provided the proper geometrical,
L velocity and closure relationships are provided. These are de-
The slug translational velocity is determined from the scribed below.
Bendiksen39 correlation, given by Mass Balances. The velocities of the liquid film and the gas
core can be determined from simple mass balance calculations
v TB 1.2v M 0.542gd cos 0.351gd sin . 40 yielding, respectively,

The gas velocity of the small bubbles entrained in the liquid 1E d 2


slug, v GLS , can be determined in the manner suggested by Hasan v F v SL , 44
4 d
and Kabir,21 given in a later section by Eqs. 57 and 58. Note that
for this case the liquid holdup in the slug body, H LLS , should be
used. v SG v SL E d 2
v C . 45
27 d2 2
Unified Annular Flow Model. The model of Alves et al. de-
veloped originally for vertical and sharply inclined flow has been
The gas void fraction in the core and the core average density
extended in the present study to the entire range of inclination
and viscosity are given, respectively, by
angles from 0 to 90, as given below. The physical model for
annular flow is given in Fig. 4.
The annular flow model equations are similar to the stratified v SG
C , 46
flow model ones, since both patterns are separated flow. The dif- v SG v SL E
ferences between the two models are the different geometrical and
closure relationships, and the fact that the gas core in annular flow
includes liquid entrainment. C G C L 1 C , 47
Momentum Balances. The linear momentum force balances
for the liquid and gas core phases are given, respectively, by C G C L 1 C . 48

Gomez et al.: Unified Mechanistic Model for Steady-State Two-Phase Flow SPE Journal, Vol. 5, No. 3, September 2000 343
Closure Relationships. The liquid wall shear stress is deter-
mined from single-phase flow calculations based on the hydraulic
diameter concept.
The most difficult task in modeling annular flow is the deter-
mination of the interfacial shear stress, I , and the entrainment
fraction, E. By all means this is an unresolved problem even for
vertical or horizontal flow conditions.
The definition of the interfacial shear stress for annular flow is

v C v F v C v F
I f I C . 49
2
As suggested by Alves et al.,27 the interfacial friction factor can
be expressed by

f I f SC I, 50
where f SC is the friction factor that would be obtained if only the
core gas phase and entrainment flows in the pipe. Calculation of
f SC should be based on the core superficial velocity ( v SC v SG
E v SL ) and the core average density and viscosity given, respec-
tively, by Eqs. 47 and 48. The interfacial correction parameter I is
used to take into account the roughness of the interface. Different
expressions for I are given by Alves et al.27 for vertical flow only.
In the present study, the parameter I is an average between a
horizontal factor and a vertical factor, based on the inclination
angle, , as follows:

I I H cos2 I V sin2 . 51
The horizontal correction parameter is given by Henstock and
Hanratty41 as
Fig. 5Physical model for bubble flow.
I H 1850F A , 52
where


0.0379N Re
0.5 2.5
0.707N Re
0.9 2.5 0.4 angles was carried out by taking the component of the bubble rise
SL SL vL L 0.5
velocity in the direction of the flow, as given below see Fig. 5 for
F A 53
0.9
N Re vG G the bubble flow physical model.
SG
The gas velocity is given by
and N ReSL and N ReSG are the liquid and gas superficial Reynolds
numbers, respectively. The vertical correction parameter is given v G C 0 v M v 0 sin H L0.5 , 57
by Wallis17 as where v M is the mixture velocity, C 0 is a velocity distribution
coefficient, v 0 is the bubble rise velocity and H L0.5 is a correction
I V 1300 . 54 for bubble swarm. In the present study, the velocity distribution
d coefficient C 0 1.15, as suggested by Chokshi et al.,30 and the
The entrainment fraction, E, is calculated by the Wallis17 corre- bubble rise velocity is given by Harmathy41 in SI units as fol-
lation, given by lows:

E1e [0.125( 1.5)] ,


where
55
v 0 1.53 g L G
L2 0.25
. 58


Substituting for the gas velocity in terms of the superficial veloc-
v SG G G 1/2
ity yields
104 . 56
L
v SG
Unified Bubble Flow Model. Extension of the Hasan and Kabir21 C 0 v M v 0 sin H L0.5 . 59
bubble flow model for the entire range of wellbore inclination 1H L

TABLE 1 DATABASE FOR INDIVIDUAL FLOW PATTERN MODELS VALIDATION

Pipe Diameter Liquid Density Pressure


Data Source Flow Pattern Inclination (in.) Fluids (lbm/ft3) (psia) Data Points

Minami (Ref. 44) Stratified 0 3 Air-kerosene/water 50/62.4 50 100


Nuland et al. (Ref. 42) Slug 10 60 4 Dense gas (SF6)-oil 51 145 52
Felizola and Shoham (Ref. 32) Slug 0 90 2 Air-kerosene 50 250 72
Schmidt (Ref. 43) Slug 90 2 Air-kerosene 50 225 15
Caetano et al. (Ref. 22) Bubble 90 Annulus Air-kerosene/water 50/62.4 45 19
Alves et al. (Ref. 27) Annular 90 2.5 Natural gas-Crude 27 1750 2 (75)
Total260

344 Gomez et al.: Unified Mechanistic Model for Steady-State Two-Phase Flow SPE Journal, Vol. 5, No. 3, September 2000
TABLE 2 INDIVIDUAL FLOW PATTERN MODELS VALIDATION RESULTS

Pressure Gradient Liquid Holdup


Average Abs. Average Average Abs. Average
Data Source Flow Pattern Inclination Error (%) Error (%) Error (%) Error (%)

Minami (Ref. 44) Stratified 0 20.8 33.5


Nuland et al. (Ref. 42) Slug 10 60 7.5 10.2 6.7 9.6
Felizola and Shoham (Ref. 32) Slug 0 90 20.6 25.0 0.6 13.2
Schmidt (Ref. 43) Slug 90 9.3 15.0
Caetano et al. (Ref. 22) Bubble 90 2.3 2.7
Alves et al. (Ref. 27) Annular (2 points) 90 1.5 1.5
Annular (75 points) 90 0.9 9.8

Eq. 59 must be solved numerically to determine the liquid holdup, holdup (H LU ) and the pressure gradient, averaged over a slug
H L . Once the liquid holdup is computed, the gravitational and unit. The results for the different data sources are given in Table
frictional pressure gradients are determined in a straightforward 2.
manner. Unified Bubble Model. The data of Caetano et al.22 were used
For dispersed bubble flow, the homogeneous no-slip model17 is to test the model for bubble flow. Note that the Caetano et al. data
used. Details of this simple model are omitted here for brevity. were acquired in an annulus configuration with a 3-in. casing
inner diameter ID and 1.66-in. tubing outer diameter OD. For
Results and Discussion this reason the comparison was carried out only for the liquid
This section includes the validation of the developed unified holdup. An equivalent diameter was used that provides the same
model with published laboratory and field data, and the perfor- cross-sectional area and superficial velocities that occur in the
mance of the model with new field data. annulus. The results show excellent agreement with an average
error and an average absolute error of 2.3 and 2.7%, respec-
Unified Model Validation. Initially, the individual flow pattern tively.
models for slug flow, stratified flow, bubble flow and annular flow Unified Stratified Model. The stratified flow model was tested
were validated against several sets of available laboratory and against the liquid holdup data of Minami.44 The data were col-
limited field data. Tables 1 and 2 present the range of data and lected for air-water and air-kerosene. The model systematically
the validation results, respectively. underpredicted the data, with an average error and average abso-
Unified Slug Model. Validation of the proposed slug flow lute error of 20.8 and 33.5%, respectively, as shown in Table 2.
model was carried out using the following sets of data: Note that, as reported by Minami,44 the original Taitel and
1. the Felizola and Shoham32 data provide detailed slug char- Dukler6 model performed poorly against his data. Modification of
acteristics, liquid holdup and pressure drop, for the entire range of both the liquid wall friction factor and the interfacial friction fac-
upward inclination angles between 10 and 90 at 10 increments; tor, implemented in the present study model, improves the predic-
2. the Nuland et al.42 data for 10, 20, 45, 60, and 80 including tions of the stratified model considerably.
liquid holdup and pressure drop; Unified Annular Model. As shown in Table 1, Alves et al.27
3. the Schmidt43 data for vertical flow with liquid holdup only. provided 2 new field data points, in addition to the 75 data points
Fig. 6 presents a typical comparison of the predictions of the taken from the Tulsa U. Fluid Flow Projects TUFFP database, in
Gomez et al.34 slug body liquid holdup correlation with published which the wells are under annular flow. The model of Alves et al.
experimental data including additional data other than the above shows excellent agreement with the data: For the 2 data points the
mentioned three sets. As can be seen, the correlation follows the average error and average absolute errors are 1.5%. For the 75
trend of decreasing slug liquid holdup as the inclination angle database points the average error is 0.9% and the average ab-
increases. solute error is 9.8%.
Comparisons between the predictions of the unified slug model Entire Unified Model Validation. Following validation of the
and the experimental data were carried out for both the liquid individual flow pattern models, the entire unified model was
evaluated against the TUFFP wellbore databank, as reported by
Ansari et al.29 The databank includes a total of 1,723 laboratory
and field data, for both vertical and deviated wells. The data cover
a wide range of flow conditions: pipe diameter of 1 to 8 in.; oil
rate of 0 to 27,000 B/D; gas rate of 0 to 110,000 scf/D and oil
gravity of 8.3 to 112API. Additionally, six commonly used cor-
relations and models have been evaluated against the databank.
They are those of Ansari et al.,29 Chokshi et al.,30 Duns and Ros,5
Beggs and Brill,2 Hasan and Kabir,21 and the modified Hagedorn
and Brown.3 The modifications of the Hagedorn and Brown cor-
relation are the Griffith and Wallis45 correlation for bubble flow
and the use of no-slip liquid holdup if greater than the calculated
liquid holdup. Note that, except for the Beggs and Brill2 correla-
tion, the other five methods were developed for vertical upward
flow only. These methods are adopted in this study for deviated
well conditions by incorporating the inclination angle in the gravi-
tational pressure gradient calculations. The proposed unified
model is the only mechanistic model applicable to all of the incli-
nation angle range, from horizontal to vertical.
The overall performance of the unified model showed an aver-
Fig. 6Comparison between predicted and measured slug liq- age error of 3.8% and an absolute average error of 12.6%. The
uid holdup Ref. 34. Hagedorn and Brown3 correlation showed a minimum average

Gomez et al.: Unified Mechanistic Model for Steady-State Two-Phase Flow SPE Journal, Vol. 5, No. 3, September 2000 345
346
TABLE 3 PERFORMANCE OF UNIFIED MODEL AND OTHER METHODS FOR DATA SET No. 1 Ref. 3

Present Study Chokshi et al. (Ref. 30) Hagerdon and Brown (Ref. 3) Ansari et al. (Ref. 29)
Gas/Oil Ratio Water Cut
Case (scf/stbl) (%) P Measured P Calculated Error (%) P Calculated Error (%) P Calculated Error (%) P Calculated Error (%)

1 336 68.0 1,851 1,743 5.8 1,826 1.4 1,684 9.0 1,819 1.7
2 1,747 20.2 1,282 1,325 3.4 1,333 4.0 1,161 9.4 1,219 4.9
3 537 87.0 1,990 1,842 7.4 1,885 5.3 1,804 9.3 1,837 7.7
4 511 5.0 2,220 2,308 4.0 2,479 11.7 2,261 1.8 2,554 15.0
5 1,044 45.0 1,518 1,650 8.7 1,405 7.4 1,349 11.1 1,260 17.0
6 527 60.0 2,588 2,146 17.1 2,444 5.6 2,265 12.5 2,448 5.4
7 1,841 0.0 1,540 1,094 29.0 1,105 28.2 1,189 22.8 1,012 34.3
8 1,135 36.9 1,371 1,319 3.8 1,383 0.9 1,338 2.4 1,223 10.8
9 1,196 59.2 1,386 1,268 8.5 1,100 20.6 1,120 19.2 948 31.6
10 894 62.3 1,817 1,496 17.7 1,413 22.2 1,289 29.1 1,284 29.3
11 344 80.0 2,998 2,539 15.3 2,783 7.2 2,579 14.0 2,881 3.9
12 1,490 0.2 1,160 908 21.7 791 31.8 815 29.7 703 39.4
13 1,758 55.6 1,638 1,240 24.3 1,301 20.6 1,145 30.1 1,290 21.2

Gomez et al.: Unified Mechanistic Model for Steady-State Two-Phase Flow


14 1,898 0.3 840 722 14.0 743 11.5 787 6.3 676 19.5
15 1,494 27.0 1,016 1,136 11.8 1,041 2.5 876 13.8 966 4.9
16 1,382 70.2 1,199 1,267 5.7 1,076 10.3 1,078 10.1 960 19.9
17 5,716 33.1 802 716 10.7 512 36.2 659 17.8 469 41.5
18 872 66.7 2,006 1,664 17.0 1,883 6.1 1,751 12.7 1,822 9.2
19 2,118 26.7 767 973 26.9 678 11.6 821 7.0 592 22.8
20 1,498 0.3 1,221 1,249 2.3 1,153 5.6 1,036 15.2 1,090 10.7
21 2,847 9.0 680 813 19.6 632 7.1 820 20.6 562 17.4

Average error % 5.2 10.5 11.7 16.1


Std. dev. avg. error 14.7 12.2 12.1 14.0
Average absolute error % 13.1 12.3 14.5 17.5
Std. dev. abs. avg. error 8.1 10.3 8.3 12.0

SPE Journal, Vol. 5, No. 3, September 2000


error and absolute average error of 1.2 and 9.3%, respectively.
However, the databank includes about 400 data points collected
by Hagedorn and Brown3 to develop their correlation. An objec-
tive comparison should exclude these data points from the data-
bank.

Unified Model Performance and Results. The ultimate goal of


any model is to predict the flow behavior under field conditions.
The performance of the proposed unified model under field con-
ditions was evaluated by comparison between its predictions and
directional well field data provided by British Petroleum and Sta-
toil. Two sets of data were provided. The first data set includes 21
data points while the second data set includes 65 cases. The data
include wells with different flow conditions: pipe diameter of
7
2 8 to 7 in.; inclination angles of 0 to 90; oil rate of 79 to 2,658
B/D; gas rate of 42 to 23,045 Mscf/D, and water-cut of 0 to 80%.
Of the total cases, 59 wells were producing naturally and the
remaining 27 were on artificial lift. Each data point included, in
addition to the geometrical and operational variables, the wellhead
pressure, the wellhead and bottomhole temperatures and the total
pressure drop.
Physical Properties. The pressure/volume/temperature PVT
properties used were summarized by Brill and Beggs.46 The Glaso Fig. 7Comparison between unified model predictions and
data set No. 2 65 cases.
correlation was used for the prediction of the solution gas/oil ra-
tio, oil formation volume factor and oil viscosity. The Standing z
factor was used in the calculations of the gas phase properties.
The Lee et al. correlation was used for the gas viscosity. The
gas/oil surface tension was predicted by the Baker and Swerdloff For the combined data sets the unified model shows excellent
correlation. The liquid phase oil and water properties, namely, performance, with an average error of 1.3% s.d. 8.2 and ab-
density, viscosity and surface tension, are calculated based on the solute error of 5.5% s.d. 6.2. These results are also shown
volume fraction of the oil and water in the liquid phase. The graphically in Fig. 8. The Chokshi et al.30 model shows an aver-
volume fractions were calculated based on the in-situ flow rates, age error and absolute error of 0.9% s.d. 9.6 and 7.1% s.d. 6.4,
assuming no-slip between the oil and water. respectively. As can be seen from Table 4, except for the three
For the gas lift wells, the gas properties are calculated as fol- small diameter well cases, the unified model shows better perfor-
lows. Up to the point of gas injection, the calculations are per- mance than the Chokshi et al. model, especially for large diameter
formed using the flow rate and specific gravity of the formation tubing and deviated wells. It is believed that the unified slug flow
gas. At the point of gas injection, the formation gas flow rate is model is the main reason for this behavior, since it is more suit-
combined with the injection gas rate to give the total gas flow rate, able for directional flow. Both models perform equally well for
with a weighted average specific gravity based on the two flow the entire range of water cuts.
rates at standard conditions. From the point of injection to the
surface, the PVT properties, including the solution gas oil ratio
and hence free gas quantity, are determined based on the com- Conclusions
bined total gas specific gravity. No tuning of the PVT data was
done. A unified steady-state two-phase flow mechanistic model for the
Results and Discussion. Table 3 reports the pressure drop pre- prediction of flow pattern, liquid holdup and pressure drop was
diction performance of the unified model, along with that of presented that is applicable to the range of inclination angles from
Chokshi et al.,30 Hagedorn and Brown3 and Ansari et al.,29 vs. the horizontal (0) to upward vertical flow (90). The model consists
first data set 21 data points. Note that Table 3. includes, in of a unified flow pattern prediction model and five individual uni-
addition to the pressure drop, the gas/liquid ratio and the water fied models for the stratified, slug, bubble, annular and dispersed
cut. The comparison shows good agreement, with an average error bubble flow patterns.
of 5.2% and a corresponding standard deviation s.d. of 14.7 The proposed unified model was evaluated and compared to the
and an average absolute error of 13.1% with a s.d. of 8.1 for the other six most commonly used models or correlations. This was
unified model. Corresponding errors for the other methods are as carried out by running the unified model and the other methods
follows: 10.5% s.d. 12.2 and 12.3% s.d. 10.3 for Chokshi against the TUFFP wellbore databank.29 The databank includes a
et al.,30 11.7% s.d. 12.1 and 14.5% s.d. 8.3 for Hagedorn total of 1,723 laboratory and field data for both vertical and devi-
and Brown,3 and 16.1% s.d. 14.0 and 17.5% s.d. 12 for the ated wells. The overall performance of the unified model showed
Ansari et al.29 model. an average error of 3.8% and an absolute average error of
Fig. 7 shows a comparison between the predicted results of the 12.6%.
unified model and measured pressure drops for the 65 cases of the The performance of the unified model and of other models and
second data set. The predictions of the proposed unified model correlations was evaluated against 86 new directional well field
show excellent agreement vs. this data set, with an average error data cases provided by British Petroleum and Statoil. The predic-
of 0% s.d. 3.9, as compared to 4.5% s.d. 4.5 for the Chokshi tions of the unified model show excellent agreement with data,
et al.30 model. The average absolute error for the unified model with an average error of 1.3% and an absolute average error of
and the Chokshi et al.30 model are 3.0% s.d. 2.5 and 5.5% s.d. 5.5%, with respective standard deviations of 8.2 and 6.2. A sen-
3.2, respectively. sitivity analysis of the model performance was conducted with
The overall performance of the model was evaluated vs. the respect to tubing diameter, method of lift and maximum wellbore
combined two data sets, including all 86 well cases. The results inclination angle. The unified model showed superior performance
were compared with the predictions of only the Chokshi et al.30 except for a limited number of small diameter wells.
model. In addition, a sensitivity analysis was carried out based on The predictions of the unified model were carried out without
the maximum deviation angle of the well, production method any tuning of either the model or the PVT data. It provides an
natural or artificial lift and tubing diameter. All the results are accurate two-phase flow mechanistic model for research and de-
summarized in Table 4. sign for the industry.

Gomez et al.: Unified Mechanistic Model for Steady-State Two-Phase Flow SPE Journal, Vol. 5, No. 3, September 2000 347
348
TABLE 4 OVERALL PERFORMANCE OF UNIFIED MODEL AND SENSITIVITY ANALYSIS RESULTS

Present Study Chokshi et al. (Ref. 30)

Average Standard Absolute Average Standard Average Standard Absolute Average Standard
No. of Wells Classification Error (%) Deviation Error (%) Deviation Error (%) Deviation Error (%) Deviation

Inclination
3 Vertical 1.7 2.7 2.5 1.3 7.4 4.0 7.4 4.0
90
19 Horizontal to vertical 2.2 5.1 4.2 3.4 0.4 5.2 4.0 3.3
0 90
64 Deviated wells 1.1 9.1 6.0 6.9 0.7 10.7 8.1 6.9
45 90

Production
59 Naturally flowing 0.1 4.1 3.2 2.5 4.6 4.7 5.7 3.4
27 Gas Lifted 4.3 13.0 10.5 8.7 7.2 12.4 10.5 9.6

Diameter
3 Tubing 3.3 5.8 5.5 2.1 0.9 4.6 3.5 2.0

Gomez et al.: Unified Mechanistic Model for Steady-State Two-Phase Flow


7
d2 8 in.
24 Tubing 0.1 10.8 5.9 9.0 1.7 9.5 7.3 6.1
1
d4 2 in.
28 Tubing 3.3 8.8 6.7 6.4 3.3 12.0 8.7 8.8
1
d5 2 in.
31 Tubing 0.1 4.9 4.0 2.7 4.1 5.7 6.0 3.7
d 7 in.

Overall
86 Entire Database 1.3 8.2 5.5 6.2 0.9 9.6 7.1 6.4

Database Diameter Inclination Oil Rate Gas Rate Water Cut


7
86 cases 28 to 7 in. 0 90 79 to 2,658 B/D 42 to 23,045 Mscf/D 0 to 80%

SPE Journal, Vol. 5, No. 3, September 2000


max maximum
NS no-slip
R radians
S slug body
SC superficial core
SL superficial liquid
SG superficial gas
TB Taylor bubble
U total slug unit
V vertical
W wall

Superscripts
dimensionless
m, n Blasius equation exponents

Acknowledgment
This paper is dedicated to the memory of Dr. Zelimir Schmidt.

References
Fig. 8Overall performance of the unified model vs. the entire 1. Dukler, A.E., Wickes, M. III, and Cleveland, R.G.: Frictional Pres-
new database 86 cases. sure Drop in Two-Phase Flow: B. An Approach Through Similarity
Analysis AIChE J. 1964 10, No. 1, 44.
2. Beggs, H.D. and Brill, J.P.: A Study of Two-Phase Flow in Inclined
Nomenclature Pipes, JPT May 1973 607; Trans., AIME, 255.
3. Hagedorn, A.R. and Brown, K.E.: Experimental Study of Pressure
A area, L2, ft2 Gradient Occurring During Continuous Two-Phase Flow in Small
C Blasius equation coefficient Diameter Vertical Conduits, JPT April 1965 475; Trans., AIME,
C0 flow distribution coefficient 234.
d diameter, L, ft 4. Ros, N.C.J.: Simultaneous Flow of Gas and Liquid as Encountered
in Well Tubing, JPT October 1961 1037; Trans., AIME, 222.
E entrainment fraction
5. Duns, H. Jr. and Ros, N.C.J.: Vertical Flow of Gas and Liquid
F dimensionless group Mixtures in Wells, Proc., Sixth World Petroleum Congress,
FA annular flow parameter Frankfurt-am-Main, Germany 1963 451.
f Fanning friction factor 6. Taitel, Y. and Dukler, A.E.: A Model for Predicting Flow Regime
g acceleration due to gravity, L/t2, ft/sec2 Transition in Horizontal and Near Horizontal Gas-Liquid Flow,
h liquid level height, L, ft AIChE J. 1976 22, No. 1, 47.
H liquid holdup 7. Lin, P.Y. and Hanratty, T.J.: Prediction of the Initiation of Slug
I interfacial annular parameter Flow with Linear Stability Theory, Int. J. Multiphase Flow 1986
12, No. 1, 79.
L length, L, ft
8. Wu, H.L. et al.: Flow Pattern Transitions in Two-Phase Gas/
N Re Reynolds number Condensate Flow at High Pressure in an 8-Inch Horizontal Pipe,
p pressure, M/Lt2, lbf/ft2 Proc. 3rd Intl. Conference on Multiphase Flow, The Hague 1987
S perimeter, L, ft 13.
v velocity, L/t, ft/sec 9. Cheremisinoff, N.P. and Davis, E.J.: Stratified Turbulent-Turbulent
v 0 single bubble rise velocity, L/t, ft/sec GasLiquid Flow, AIChE J. 1979 25, No. 1, 48.
X Lockhart and Martinelli parameter 10. Shoham, O. and Taitel, Y.: Stratified Turbulent-Turbulent Gas-
Y dimensionless group Liquid Flow in Horizontal and Inclined Pipes, AIChE J. 1984 30,
377.
11. Issa, R.I.: Prediction of Turbulent Stratified Two-Phase Flow in In-
Greek Letters
clined Pipes and Channels, Int. J. Multiphase Flow 1988 14, No.
void fraction 21, 141.
film thickness 12. Dukler, A.E. and Hubbard, M.G.: A Model For Gas-Liquid Slug
viscosity, M/Lt, lbm/ft-sec Flow In Horizontal And Near Horizontal Tubes, Ind. Eng. Chem.
3.141 5926 Fundam. 1975 14, 337.
13. Nicholson, K., Aziz, K., and Gregory, G.A.: Intermittent Two Phase
annular entrainment parameter
Flow In Horizontal Pipes, Predictive Models, Can. J. Chem. Eng.
inclination angle measured from horizontal 1978 56, 653.
density, M/L3, lbm/ft3 14. Kokal, S.L. and Stanislav, J.F.: An Experimental Study of Two-
shear stress, M/Lt2, lbf/ft2 Phase Flow in Slightly Inclined Pipes II: Liquid Holdup and Pressure
surface tension, M/t2, lbf/ft Drop, Chem. Eng. Sci. 1989 44, No. 3, 681.
15. Laurinat, J.E., Hanratty, T.J., and Jepson, W.P.: Film Thickness
Subscripts Distribution for Gas-Liquid Annular Flow in a Horizontal Pipe, Int.
J. Multiphase Flow 1985 6, Nos. 1/2, 179.
c core 16. James, P.W. et al.: Developments in the Modeling of Horizontal
crit critical Annular Two-Phase Flow, Int. J. Multiphase Flow 1987 13, No. 2,
CB critical buoyancy 173.
CD critical diameter 17. Wallis, G.B.: One Dimensional Two Phase Flow, McGraw-Hill Book
F film Co. Inc., New York City 1969.
18. Xiao, J.J., Shoham, O., and Brill, J.P.: A Comprehensive Mechanis-
G gas
tic Model for Two-Phase Flow in Pipelines, paper SPE 20631 pre-
H horizontal sented at the 1990 SPE Annual Technical Conference and Exhibition,
I interface New Orleans, 2326 September.
LS liquid slug 19. Taitel, Y., Barnea, D., and Dukler, A.E.: Modeling Flow Pattern
L liquid Transition for Steady Upward Gas-Liquid Flow in Vertical Tubes,
M mixture AIChE J. 1980 26, No. 3, 345.

Gomez et al.: Unified Mechanistic Model for Steady-State Two-Phase Flow SPE Journal, Vol. 5, No. 3, September 2000 349
20. Barnea, D. et al.: Gas Liquid Flow in Inclined Tubes: Flow Pattern 46. Brill, J.P. and Beggs, D.H.: Two-Phase Flow in Pipes, sixth edition,
Transition for Upward Flow, Chem. Eng. Sci. 1985 40, 131. third printing January 1991.
21. Hasan, A.R. and Kabir, C.S.: A Study of Multiphase Flow Behavior
in Vertical Wells, SPEPE May 1988 263; Trans., AIME, 285.
22. Caetano, E.F., Shoham, O., and Brill, J.P.: Upward Vertical Two-
SI Metric Conversion Factors
Phase Flow through an Annulus Part II: Modeling Bubble, Slug and
Annular Flow, ASME J. Energy Resour. Technol. 1992 114, 13. bbl 1.589 873 E01 m3
23. Fernandes, R.C., Semiat, R., and Dukler, A.E.: Hydrodynamic ft 3.048* E01 m
Model for Gas-Liquid Slug Flow in Vertical Tubes, AIChE J. ft2 9.290 304* E02 m2
1983 29, 981.
24. Sylvester, N.D.: A Mechanistic Model for Two-Phase Vertical Slug ft3 2.831 684 E02 m3
Flow in Pipes, ASME J. Energy Resour. Technol. 1987 109, 206. in. 2.54* E00 cm
25. Vo, D.T. and Shoham, O.: A Note on the Existence of a Solution for lbf 4.448 222 E00 N
Upward Vertical Two-Phase Slug Flow in Pipes, ASME J. Energy
lbm 4.535 924 E01 kg
Resour. Technol. 1989 111, 64.
26. Oliemans, R.V.A., Pots, B.F.M., and Trompe, N.: Modeling of An- psi 6.894 757 E00 kPa
nular Dispersed Two-Phase Flow in Vertical Pipes, Int. J. Multi- *Conversion factors are exact.
phase Flow 1986 12, No. 5, 711. SPEJ
27. Alves, I.N. et al.: Modeling Annular Flow Behavior for Gas
Wells, SPEPE November 1991 435.
28. Ozon, P.M., Ferschneider, G., and Chwetzof, A.: A New Multiphase Luis E. Gomez a member of Sigma Xi, is a PhD-degree candi-
Flow Model Predicts Pressure and Temperature Profiles, paper SPE date at the U. of Tulsa in Tulsa, Oklahoma. e-mail: luis-gomez-
16535 presented at the 1987 Offshore Europe Conference, Aberdeen, morillo@utulsa.edu. He previously taught in the Mechanical
811 September. Engineering Dept. of the U. de Los Andes. Gomez holds a BS
29. Ansari, A.M. et al.: A Comprehensive Mechanistic Model for Up- degree in mechanical engineering from the U. de Los Andes
ward Two-Phase Flow in Wellbores, SPEPE May 1994 143; and an MS degree in petroleum engineering from the U. of
Trans., AIME, 297. Tulsa. Ovadia Shoham is Professor of Petroleum Engineering
30. Chokshi, R.N, Schmidt, Z., and Doty, D.R.: Experimental Study and and Director of the Separation Technology Projects at Tulsa U.
the Development of a Mechanistic Model for Two-Phase Flow in Tulsa, Oklahoma. e-mail: os@utulsa.edu. He teaches and
Through Vertical Tubing, paper SPE 35676 presented at the 1996 conducts research in modeling of two-phase flow in pipes and
SPE Western Regional Meeting, Anchorage, 2224 May. its applications in oil and gas production, transportation, and
31. Barnea, D.: A Unified Model for Predicting Flow Pattern Transi- separation. Shoham holds BS and MS degrees in chemical en-
tions for the Whole Range of Pipe Inclinations, Int. J. Multiphase gineering from the Technion, Israel, and the U. of Houston, re-
Flow 1987 13, No. 1, 1. spectively, and a PhD degree in mechanical engineering
32. Felizola, H. and Shoham, O.: A Unified Model for Slug Flow in from Tel Aviv U., Israel. He served as a 198992 and 19982000
Upward Inclined Pipes, ASME J. Energy Resour. Technol. 1995
member of the Production Operations Technical Committee
117, 1.
and as a 199192 member of the Forum Series in North
33. Petalas, N. and Aziz, K.: Development and Testing of a New
Mechanistic Model for Multiphase Flow in Pipes, Proc., ASME, America Steering Committee. Zelimir Schmidt, deceased,
Fluid Engineering Div. 1996 236, No. 1, 153. was Professor of petroleum engineering and Director of Artifi-
34. Gomez, L.E., Shoham, O., and Taitel, Y.: Prediction of Slug Liquid cial Lift Projects at the U. of Tulsa, Oklahoma. He spent 10 years
HoldupHorizontal to Upward Vertical Flow, Int. J. Multiphase as a production engineer with INA-Naftaplin in the former Yu-
Flow 2000 26, No. 3, 517. goslavia and served as a consultant to various companies be-
35. Ouyang, L.B. and Aziz, K.: Development of New Wall Friction fore joining the U. of Tulsa faculty. Schmidt held an engineer-
Factor and Interfacial Friction Factor Correlations for Gas/Liquid ing degree from the U. of Zagreb and MS and PhD degrees in
Stratified Flow in Wells and Pipes, paper SPE 35679 presented at petroleum engineering form the U. of Tulsa. He served as a
the 1996 SPE Western Regional Meeting, Anchorage, 2224 May. 198788 Distinguished Lecturer and was a 199495 Forum Se-
36. Baker, A., Nielsen, K., and Gabb, A.: Pressure Loss, Liquid Holdup ries in South America and Caribbean Steering Committee
Calculations Developed, Oil & Gas J. 14 March 1988 55. member, a 199195 Editorial Review Committee member, and
37. Taitel, Y. and Barnea, D.: Two Phase Slug Flow, Academic Press 198182 and 199496 U. of Tulsa Student Chapter Faculty
Inc., New York City 1990. Sponsor. Rajan N. Chokshi is a program project manager
38. Scott, S.L., Shoham, O., and Brill, J.P.: Prediction of Slug Length in with TanData Corp. in Tulsa, Oklahoma. e-mail:
Horizontal Large-Diameter Pipes, SPEPE August 1989 335; chokshir@hotmail.com. His current interests are change man-
Trans., AIME, 287. agement, enterprise software architecture, and emerging
39. Bendiksen, K.H.: An Experimental Investigation of the Motion of technologies in computing. He has more than 15 years expe-
Long Bubbles in Inclined Tubes, Int. J. Multiphase Flow 1984 10, rience in research and design of fluid-flow and artificial-lift
467. problems. He has developed software for and taught profes-
40. Henstock, W.H. and Hanratty, T.J.: The Interfacial Drag and the sional courses in these areas and managed consulting
Height of the Wall Layer in Annular Flow, AIChE J. 1976 22, No.
projects in the U.S., Canada, Venezuela, and India. Chokshi
6, 990.
holds BS and MS degrees in chemical engineering from Gu-
41. Harmathy, T.Z.: Velocity of Large Drops and Bubbles in Media of
jarat U., India, and Indian Ints. of Technology, Kanpur, respec-
Infinite or Restricted Extent, AIChE J. 1960 6, 281.
42. Nuland, S. et al.: Gas Fractions in Slugs in Dense-Gas Two-Phase tively, and a PhD degree in petroleum engineering from the U.
Flow From Horizontal to 60 Degrees of Inclination, Proc., ASME, of Tulsa. Tor Northug is a principal engineer in the R&D Dept.
Fluids Engineering Div. Summer Meeting June 1997. of Statoil in Trondheim, Norway. e-mail: THUG@statoil.com. His
43. Schmidt, Z.: Experimental Study of Two-Phase Slug Flow in a research interests include multiphase flow, fluid mechanics,
PipelineRiser Pipe System, PhD dissertation, U. of Tulsa, Tulsa, leak detection, and gas leakage/subsea blowouts. He previ-
Oklahoma 1977. ously worked for Technical U. of Trondheim, Sintef Hydrody-
44. Minami, K.: Liquid Holdup in Wet Gas Pipelines, MS thesis, U. of namic Laboratory, and Reinertsen Engineering Co. Northug
Tulsa, Tulsa, Oklahoma 1982. holds a BS degree in civil engineering from Technical U. of
45. Griffith, P. and Wallis, G.B.: Two-Phase Slug Flow, J. Heat Trondheim and an MS degree in fluid mechanics from Norwe-
Transfer 1961 83, 307. gian U. of Science and Technology.

350 Gomez et al.: Unified Mechanistic Model for Steady-State Two-Phase Flow SPE Journal, Vol. 5, No. 3, September 2000

También podría gustarte