You are on page 1of 15

SPE

Society of Petroleum Engineers

SPE 20630

A Comprehensive Mechanistic Model for Upward Two-Phase Flow


in Wellbores
A.M. Ansari, Pakistan Petroleum Ltd.; N.D. Sylvester, U. of Akron; and O. Shoham and
J.P. Brill, U. of Tulsa
SPE Members

Copyright 1990, Society of Petroleum Engineers, Inc.


II
This paper was prepared for presentation at the 65th Annual Technical Conference and Exhibition of the Society of Petroleum Engineers held in New Orleans, LA, September 23-26, 1990.

This paper was selected for presentation by an SPE Program Committee following review of information contained in an abstract submitted by the author(s). Contents of the paper,
as presented, have not been reviewed by the Society of Petroleum Engineers and are sUbject to correction by the author(s). The material, as presented. does not necessarily reflect
any position of the Society of Petroleum Engineers, its officers, or members. Papers presented at SPE meetings are subject to publication review by Editorial Committees of the Society
of Petroleum Engineers. Permission to copy is restricted to an abstract of not more than 300 words. Illustrations may not be copied. The abstract should contain conspicuous acknowledg-
ment of where and by whom the paper is presented. Write Publications Manager, SPE, P.O. Box 833836, Richardson, TX 75083-3836 U.S.A. Telex, 730989 SPEDAL.

ABSTRACT Recently the trend has shifted towards the modeling


approach. The fundamental postulate of the modeling
A comprehensive model is formulated to predict the approach is the existence of flow patterns or flow
flow behavior for upward two-phase flow. The configurations. Various theories have been developed for the
comprehensive model is composed of a model for flow prediction of flow patterns. Separate models were
pattern prediction and a set of independent models for developed for each flow pattern to predict the flow
predicting the flow characteristics such as holdup and characteristics such as holdup and pressure drop. By
pressure drop in bubble, slug and annular flows. considering flow mechanics, the resulting models can be
applied to flow conditions other than used for their
The comprehensive model is evaluated by using a development with more confidence.
well databank that is composed of 1775 well cases covering
a wide variety of field data. The performance of the model The only studies pUblished on comprehensive
is also compared with the six commonly used empirical mechanistic modeling of two-phase flow in vertical pipes
correlations. are by Ozon et al. l and Hasan and Kabir2 • Nevertheless,
more work is needed in order to develop models which
The overall performance of the model is in good describe the physical phenomena more rigorously.
agreement with the data. In comparison with the empirical
correlations, the comprehensive model performs the best, The purpose of this study is to formulate a detailed
with the least average error and the smallest scattering of comprehensive mechanistic model for upward two-phase
the results. flow. The comprehensive model first predicts the existing
flow pattern and then calculates the flow variables by taking
INTRODUCTION into account the actual mechanisms of the predicted flow
.·pattern. The model is evaluated against a wide range of
Two-phase flow is commonly encountered in experimental and field data available in the updated TUFFP
petroleum, chemical and nuclear industries. The frequent well databank. The performance of the model is also
occurrence of two-phase flow presents engineers with the compared with six empirical correlations used in the field.
challenge of understanding, analyzing and designing two-
phase systems. FLOW PATTERN PREDICTION

Due to the complex nature of two-phase flow, the The basic work on mechanistic modeling flow of
problem was first approached through empirical methods. pattern transitions for upward two-phase flow was
presented by Taitel et al. 3 They identified four distinct flow
References and illustrations at end of paper. patterns, and formulated and evaluated the transition
151
A COMPREHENSIVE MECHANISTIC MODEL FOR UPWARD
2 TWO- PHASE FLOW IN WELLBORES SPE20630

boundaries among them. The four flow patterns are bubble Transition to Annular Flow: The transition
flow, slug flow, churn flow and annular flow, as shown in criterion for annular flow is based on the gas phase velocity
Fig.1. Later, modifications of transitions were made by required to prevent fall back of the entrained liquid droplets
Barnea et al. 4 to extend the applicability of the model to in the gas stream. This gives the transition as,
inclined flows as well. In a relatively recent work, Barnea s
combined flow pattern prediction models applicable to
different inclination angle ranges into one unified model.
VSG =3.1 [gO(~-PG)r/4 (6)
Based on these different works, flow pattern can be
predicted by defining transition boundaries among bubble,
slug and annular flows. and is shown as transition D in Fig. 2.

Bubble-Slug Transition: The minimum diameter at The same transition was modified by Barnea s by
which bubble flow occurs is given by Taitel et al. 3 as, considering the effects of film thickness on the transition.
One effect is bridging of the gas core by a thick liquid film at
high liquid rates. The other effect is instability of the liqUid
film causing downward flow of the film to occur at low
liqUid rates. The mechanism of bridging is governed by the
minimum liquid holdup required to form a liquid slug,
For pipe sizes larger than this, the basic transition
mechanism for bubble to slug flow is coalescence of small HLF> 0.24.. · ·.. ·· · · · (7)
gas bubbles into large Taylor bubbles. Experimentally this
was found to occur at a void fraction of approximately 0.25. where HLF is the fraction of pipe cross-section occupied by
Using this value of void fraction, the transition can be the liquid film, assuming no entrainment in the core.
expressed in terms of superficial and slip velocities as,
The mechanism of film instability can be expressed
VSG = 0.25 Vs + 0.333 VSL · · (2) in terms of the Lockhart and Martinelli parameters, X and Y,

where Vs is the slip or bubble rise velocity given by y _


-
2 - 1.5 HLF X- (8)
Harmath y 6 as,
K'LF(1 - 1.5 HLF)

Vs = 1.53 [gO(~-PG)r4 (3) where HLF can be expressed in terms of minimum film
thickness, ~n as,

This is shown as transition A in Fig. 2. HLF = 4 Qrnn (1 - Q.mn) (9)

At high liquid rates, turbulent forces break down


To account for the effect of the liquid entrainment
large gas bubbles into small ones, even at void fractions
in the gas core, Eq. 7 is modified in this study as,
greater than 0.25. This yields the transition to dispersed
bubble flow given by Barnea et al. 2 as,
(HLF + ALCAc/A) > 0.24 (10)

2 [ 0.4 0 ]1/2 (PL)3/5 [£CL(Q)·n]2/S In Eq. 8, X and Y must be redefined in terms of the
(PL-PG)g 0 0 VL core parameters instead of the gas parameters to account
for the entrainment.
(VSL + VSGf(3.nYs = 0.725 + 4.15 ( VSG )o.s ...... (4)
VSG + VSL FLOW BEHAVIOR PREDICTION

This is shown as transition B in Fig. 2. Following the prediction of flow patterns, the next
step is to develop physical models for the flow behavior in
At high gas velocities this transition is governed by each of the flow patterns. This resulted in separate models
the maximum packing of bubbles to give coalescence. This for bubble flow, slug flow and annular flow. Churn flow has
occurs at a void fraction of 0.52, giving the transition for not yet been modeled due to its complexity, and is treated
no-slip dispersed bubble flow as, as a part of slug flow. The models developed for other flow
patterns are discussed in the following sections.
VSG = 1.08 VSL · · ·· · ·· ·.. (5)

This is shown as transition C in Fig. 2.


Bubble Flow Model:
152
SPE 20630 A M, Ansari N D, Sylyester 0 Shoham and J P Brill 3

The bubble flow model is based on the work by The two-phase pressure gradient is comprised of
Caetano? for flow in an annulus, The two bubble flow three components, Thus"
regimes, viz, bubbly flow and dispersed bubble flow, are

= (dP) + (dP) + (d P)'''''''''''''''''''''''''''' (20)


considered separately in developing the model for the bubble
flow pattern, (dP)
dL dL e dL dL a
t f

Due to the uniform distribution of gas bubbles in the


The elevation pressure gradient is given by,
liquid, and no slippage between the two phases, dispersed
bubble flow can be approximated as a pseudo-single phase,
Due to this simplification, the two-phase parameters can be
expressed as,
(:~t = PTP g sine " .. ", " .. , """"" " .. " (21)

PTP = PL At. + pG (1 - AL).. """" ..... ", .. ,,,,, ........ ,, (11) The friction component is given by,

IlTP = ilL AL + IlG (1 - Ad... """" ... " .. ""." ..... ",,, (12) (:~)f = fTP ~T~ Vfp "",.""""" """""", .. , .. " (22)

VTP = VM = VSL + VSG """" ...... """ ....... """""" (13)


The explicit expression given by Zigrang and Sylvester9 can
where, be used to define fTP as,

f.L = VSL """""" .. """"" ..... "."",." ...... ' (14) _1_ =-2 log
(VSL + VSG)
Yhp
For bubbly flow, the slippage is considered by
taking into account the bubble rise velocity relative to the [(£10)
3.7
_5,02
R9Tp
log ((£/0)
3,7
+ 13,0.)t"""...... ", (23)
R9Tp IJ
mixture velocity. By assuming a turbulent velocity profile
for the mixture with the rising bubble concentrated more at
the center than along the wall of the pipe, the slip velocity where,
can be expressed as,
R _ PTP VTP 0
Vs = VG - 1 ,2VM""" .... """, .. " .. ""· .. " .... ",,·,, .... (15) 9Tp - "'"'''''''''''''''''''''''''''''''''''''''' (24)
IlTP

An expression for the bubble rise velocity was given by The acceleration pressure gradient is negligible
Harmath y6, To account for the effect of bubble swarm, this
compared to the other pressure gradients.
expression was modified by Zuber and Hench 8 as follows
Slug Flow Model:

The first thorough physical model for slug flow was


developed by Fernandes et al. lO A simplified version of this
where the value of n' varies from one study to another. In model was presented by Sylvester 11 . The basic
the present study, a value of 0.1 for n' was found to give simplification made was the use of a correlation for slug
the best results. Thus, Eq. 15 yields, void fraction, An important assumption of fUlly developed
slug flow was used by these models. The concept of
developing flow was introduced by McQuillan and Whalley 12
during their study of flow pattern transitions, Due to the
basic difference in the geometry of the flow, fUlly developed
and developing flow are treated separately in the model.

This gives an implicit equation for the actual holdup for For a fUlly developed slug unit, as shown in Fig.
bubbly flow, The two-phase flow parameters can now be 3(a), the overall gas and liquid mass balances, respectively,
calculated from, give,

PTP = PL H L + pG (1 - HL)""".".""""."""".""" (18)


IlTP = ilL HL + flG (1 - Hd"""""" .... "·""" ...... ,, (19) (1 - ~) VGLS (1 - HLLS) "." •. """""" .. """"""", (25)

153
A COMPREHENSIVE MECHANISTIC MODEL FOR UPWARD
4 TWO-PHASE FLOW IN WELLBORES SPE20630

VSL = (1 - ~) VLLSHLLS - ~ VLTBHLTB (26) To model developing slug flow, as shown in Fig.
3(b) it is necessary to determine the existence of such flow.
where, This requires calculating and comparing the cap length with
the total length of a developed Taylor bubble. The
expression for the cap length, as developed by McQuillan and
~=LTB (27) Whall ey 1o, is given as,
lsu
l: = -1-[VTB + VNGTB (1 - HNLTB) _ ~]2 (35)
Mass balances for liquid and gas from liquid slug to Taylor
2g H!'l.TB HNLTB
bubble, respectively, give,

(VTB - VLLS)HLLS = [VTB - (- VL TB)]HLTB •.........•.. (28)


where vNGTB and HNLTB are calculated at the terminal film
thickness ON (called Nusselt film thickness) given by,
(1 - HLLS) =
(VTB - VGLS)

(VTB - VGTB) (1 - HLTB) (29)


ON = [a
4
D VNLTB ILd
1 - HNLTB)]1/3
g(PL-PG)
(36)

The Taylor bubble rise velocity is equal to the centerline The geometry of the film flow gives HNLTB in terms of ON
velocity plus the Taylor bubble rise velocity in a stagnant as,
liquid column, Le.,

VTB = 1.2 VM + 0.35


9 D (PL
PL
- pG)
1/2

.••...•..••. (30)
HNLTB = 1 -(1 -2~Nr •.•••...•.••.....•....•...••...•.•.. (37)
[ ]

Similarly, the velocity of the gas bubbles in the liquid slug is To determine vNGTB, the net flow rate at ON can be
used to obtain,

v - v
NGTB - TB
- (v
TB
_v )
GLS (
(1 - HLLS)) (38)
1 - H NLTB
where the second term on the right hand side represents the
bubble rise velocity as defined earlier in Eq (16). The length of the liquid slug can be calculated
empirically from,
The velocity vL B T of the falling film can be
correlated with the film thickness OL using the Brotz 1 3 LLS = C'D (39)
expression,
where C' was found by Dukler et al. 15 to vary from 16 to
VLTB = -J 196.7 g CL (32) 45. It is taken as 30 in this study. This gives Taylor bubble
length as,

where OL is the constant film thickness for developing flow,


and can be expressed in terms of Taylor bubble void fraction bB = (1 ~~) ~ (40)
to give,

From the comparison of Lc and LTB, if Lc > LTB, the


VLTB = 9.916 [9 D(1 - y~)r'2 (33)
flow is developing slug flow. This require new values for
LTBo, HLTBo and vL TBo calculated earlier for developed
The liquid slug void fraction can be obtained by the
flow.
correlation developed by Sylvester 11 from Fernandes et
al. 10 and Schmidt 14 data,
For LTBo, Taylor bubble volume can be used,

Has = VSG (34) .


0.425 + 2.65 VM
t A~B (L) dL
VGTB = Jo
TB
(41)
Equations 25-26, 28-31, and 33-34 can be solved
iteratively to obtain all eight unknowns that define the
developed slug flow model.
154
SPE 20630 A. M. Ansari- N. p. Sylvester. 0 ShQham. and J. P. Brill 5

where ATB*(L) can be expressed in terms Qf local holdup h~TB (L) = (VTB - VLLS) HLLS
h L T B (L), which in turn can be expressed in terms of V2gL ""."" (50)
velQcities by using Eq. 28. This gives,

In calculating pressure gradients, the effect Qf


A~B (L) = [1 - (VTB ~)HLLS] A."".".".""." (42) varying film thickness is considered and the effect Qf
frictiQn alQng the TaylQr bubble is neglected.

The vQlume VGTB*(L) can be expressed in term of FQr develQped flQw, the elevatiQn compQnent
f1QW geQmetry as, occurring across a slug unit is given by,

, ,
VGTB = Vsu - V LS .. " .. "" ... " ... " ....... " ....... " ... ". (43a) (.dP) = [(1 - 13) PLS + 13 PG]g sin9....... " .. " .... "" (51)
dL 9

or
where,

V~TB = VSG JL~B + LLS) -


PLS = PL HLLS + pG (1 - HLLS) ...."" ... "." ... " .. " ... (52)
'\ VTB

VGLS A (1 - HLLS) ks "" ".."" (43b) The elevation component fQr develQping slug flQW is
VTB given by,

SubstitutiQn of Eqs. 42 and 43 into Eq. 41 gives, P


( d ) = [( 1 - 13 *) PLS + 13* PTBA]g sin9"" .. " .. ". (53)
dL 9
VSG (LTB + LLS) - VGLS (
1H ) -LLS =
- LLS
VTB VTB where PTBA is based Qn average VQid fractiQn in the TaylQr
bubble sectiQn with varying film thickness. It is given by,

PTBA = PL HLTBA + pG (1 - HLTBA)"" .. " .... """"". (54)


where HLTBA is obtained by integrating Eq. 50 and dividing
by LTB* giving,
EquatiQn 44 can be integrated and then simplified to
give, HLTBA = 2 (VTB - VLLS)HLLS""." ... " .. "" .. " ... "" .. (55)

[-2ab _ 4c 2 ] 2 ~2gL~B
'2
LTB + ,
LTB + -b -_ 0 .. " .. "" ...... " ... (45)
a2 a2
The frictiQn cQmpQnent is the same fQr bQth the
developed and developing slug flQWS as it occurs Qnly across
where,
the liquid slug. This is given as,

a = 1 - VSG """""... " .... """"".. """"".. """"" (46)


VTB
(:~t = fLS ~L~ V~ (1 - 13 L""""".""..."....."... (56)

b = VSG - VGLS (1 - HLLS) LLS"""'''''''''''''''''''''' (47)


VTB where fLS can be calculated by using,

c -- VTB - VLLS H
V2G LLS " "".. " ". (48) R 0 PLS VM
eLS = .. ""." .. """ .. "" .. " .... " ..""..".". (57)
ilLS
After calculating LTB·, the other local parameters
For stable slug flow, the acceleratiQn cQmpQnent Qf
can be calculated from,
pressure gradient can be neglected.

VLTB (L) = V2gL - VT B ......... """.... " .. " .. "." ..... (49) Annular FIQW Model:

A discussion Qn the hydrQdynamics Qf annular flow


was presented by Wallis 16 . Along with this, Wallis also

155
A COMPREHENSIVE MECHANISTIC MODEL FOR UPWARD
6 TWO- PHASE FLOW IN WELLBORES SPE 20630

)J2
presented the classical correlations for entrainment and 'tF-- -1 fSL (1 - FE)2-n
J PL[ VSL
( (68)
interfacial friction as a function of film thickness. Later, 2 4~1-~
Hewitt and Hall-Taylor 17 gave a detailed analysis of the
mechanisms involved in an annular flow. All the models that
where,
followed later are based on this approach.

A fully developed annular flow is shown in Fig. 4. fSL = CF[VS~LDJn (69)


The conservation of momentum applied individually to the
core and the film yields,
Using the definition of superficial frictional pressure

(ddLP) -'tl SI - pc Ac 9sine = 0


gradient for liquid, Eq. 68 reduces to,
Ac (58)
c

AF (dP) + 'tl S, - 'tF SF - PL AF 9sine = 0 (59)


For the shear stress at the interface, exactly the
dL F
same approach can be adopted to give,

Pc
't1=~(1 ~~t (:~L·····································
The core density is given by,

(71)
pc = PL ~c + pG (1 - A.Lc) (60)

where Z is a correlating factor for interfacial friction and


where,
the film thickness. Based on the performance of the model,
the Wallis expression for Z works well for thin films or high
ALc = 1 - VSG (61) entrainments, whereas the Whalley and Hewitt 18 expression
VSG + FE VSL is good for thick films or low entrainments. Thus,

Z = 1 + 300 Qfor FE > 0.9································· (72)


FE is the fraction of the total liquid entrained in the core,
given by Wallis as,
PL)1/3
Z = 1 + 24 (PG ~ for FE < 0.9 (73)
FE = 1 - exp [-0.125 (v crit - 1.5)]. (62)

where, The pressure gradient for annular flow can be


calculated by substituting the above equations into Eqs. 58
IlG (PG)1/2 and 59. Thus,
VCrit. -_ 10000 VSG
-cr-- ~ (63)

Z P)
(ddL + pc 9 sine
\5 (74)
The shear stress in the film can be expressed as, (1 - 2~JSC

'tF = fF PL V~ (64) (1 - FErn (dP)


2
64 ~3 (1 _ ~t dL SL
where,

fF = CF [DH:LVF r (65) 4 ~ (1 -
Z
~)( 1 - 2~)
3
P)
(ddL
sc
+ PL 9 sine (75)

The basic unknown in the above equations is the


VF _ QL (1 - FE) _ VSL (1 - FE) dimensionless film thickness, ~. An implicit equation for ~
- AF - 4 ~ (1 _ ~) (66)
can be obtained by equating Eqs. 74 and 75. This gives,

DHF=4.Q.(1 - li)D (67)


Z (dP ) + pc 9 sine _
(1 - 2~)5 dL sc
This gives,

156
SPE 20630 A M Ansari N D Sylvester 0 Shoham. and J p Brill 7

(1 - FErn
3
P) (d Z P) (d the exchange of liquid droplets between the core and the film
is negligible.
64 li (1 - lif dL SL + 4 li (1 - li)( 1 - 2lif dL sc
EVALUATION
PL 9 sine = 0 .. ".".""."""""""""."."."."""".""."" (76)
The evaluation of the comprehensive model is
To simplify this equation, the dimensionless carried out by comparing the pressure drop from the model
approach developed by Alves et al. 19 is used. This approach with the measured data in the updated TUFFP well databank
defines the following dimensionless groups, that comprises 1775 well cases with a wide range of data
as given in Table 1. The performance of the model is also
~ _ (dp/dLht- compared with that of the six commonly used correlations in
- (dp/dL)sc "".""."""""".""""""".""""" (77) the petroleum industry.

Criteria for Comparison with pata


Y _ 9 sine (PL - pc)
M
- (dp/dL)sc .. "" " .. " (78) The evaluation of the model using the databank is
based on the following statistical parameters,

2 (dp/dLk - 9 pc sine Average percent Error


cl>c = (dp/dLhc " ""."" (79)

2 (dp/dL~ - 9 PL sine
E1 = [l-
N 1-1
f eri] X 100..... """" ......." ......" .." .. (84)
cl>F = (dp/dLh .... "" .... """"" ......."" (80)
where,
By using these dimensionless groups, Eq. 76 reduces to,
e . _ L\pi calc - L\pI meas
n - "." """ (85)
~ (1 - FEf - n L\pi meas

[1 - (1 - 2lif]3 E1 indicates the overall trend of the performance.

Z + YM - 0 Absolute Average percentage Error


[1-(1-2~f](1-2~t - ....."".."..... (81)

The above equations can be solved iteratively to obtain li.


E2 = [l-fieri
N 1-1
I] X 100"...... """" ...... " ......... (86)
Once li is known, the dimensionless groups "F and "c can be
obtained from the dimensionless form of Eqs. 74 and 75. By E2 indicates how large the errors are on the average.
using the definitions of "F and "c' the total pressure
percent Standard Deviation
gradient can be obtained as,

(:~t = cI>~ (:~tc + 9 pc sine """..""." .."""" .. (82) E3 = f


i _ 1
V (eri - E1
N - 1
f """ " (87)

or E3 indicates the degree of scattering of the error about its


average value.
P
(d ) = cI>~ (dP ) + 9 PL sine "" (83) Average Error
dL F dLsL

The two pressure gradients calculated from the above


equations should be equal.
E4 = [l- f
N i-1
el] "" " (88)

It is important to note that the above calculated where,


total pressure gradient does not include accelerational
pressure gradient. This is based on results found by Lopes ei = L\pi calc - L\PI meas " .. """"." .. " (89)
and Dukler20 indicating that, except for a limited range of E4 indicates the overall trend independent of the measured
high liqUid flow rates, the accelerational component due to pressure drop.

157
A COMPREHENSIVE MECHANISTIC MODEL FOR UPWARD
8 TWO-PHASE FLOW IN WELLBORES SPE 20630

used before for the evaluation of any correlations. The


Absolute Average Error results are shown in Tables 2 to 5.

E5 = [i f Iad]
N ja 1
(90)
Evaluation of Individual Elow Pattern Models:

The performance of individual flow pattern models


is based on sets of data that are dominant in one particular
E5 is also independent of the measured pressure drop and
flow pattern. Eor the bubble flow model, well cases with
indicates the magnitUde of the average error.
bubble flow over 75%, or more, of well length are
considered in order to have an adequate number of well
Standard Deviation
cases, whereas for slug and annular flow models, well cases

E6 = f
i.1
y(a i
-
N - 1
E4f (91)
with 100% slug and annular flows, respectively, are
considered. The performance of the slug flow model is also
checked for all vertical well cases as well as for vertical
well cases without Hagedorn and Brown data, which is one-
E6 indicates the scattering of the results, independent of the third of all the vertical well cases. The statistical results
measured pressure drop. are shown in Tables 6 to 9.

Criteria for Comparison with other Correlations: CONCWSIONS

The correlations used for the comparison are From Tables 2 - 9, the performance of the model
modified Hagedorn and Brown 26 , Duns and ROS34, and other empirical correlations indicates that,
Orkiszweski 3o with Triggia correction 35 , Beggs and Brill 36
with Palmer correction 37 , Mukerjee and Brill 36 , and Aziz et -The overall performance of the comprehensive
a1. 39 . The comparison is accomplished by comparing the model is superior to all the correlations. This
statistical parameters. The parameter E1 was found to give superiority is further improved when only
very small values for the well cases within the range of the vertical data without Hagedorn and Brown well
data used in developing empirical correlations. To remove cases are considered. In fact, for the latter two
this biasing effect, E1 is not considered in the comparison of sets of data (Tables 4 and 5) the performance of
the model with the correlations. However, its effect is the model is the best in all respects.
considered through E4. The comparison involves the use of
a relative performance factor (RPE) which is obtained by -The performance of the bubble flow and the
dividing each statistical parameter for each correlation and annular flow models are exceptionally better
the model by the minimum values of the respective than all the correlations for all the variety of
parameter and then adding all the fractions together. data in the databank.
Mathematically,
-The performance of the slug flow model is
RPF = E2/E2 MN + E3/E3 MN + E41 / I I E4MN I + exceeded by the Aziz et al. correlation for non-
vertical well cases. This is due to the fact that
E5/E5MN + E6/E6MN (92) the model is valid only for vertical flow, and
does not include the mechanisms related to
directional wells. Indeed, for the vertical well
The minimum possible value for RPE is 5 indicating the best
cases, the performance of the model is improved.
performance in all respects.
The best performance of the model is obtained
when Hagedorn and Brown data are not included.
Oyerall Evaluation:
RECOMMENDATIONS
The overall evaluation involves the entire
comprehensive model so as to study the combined Based on the above conclusions, the following
performance of all the independent flow pattern behavior recommendations are suggested.
models together. The evaluation is first done by using the
entire databank. The performance of the model is also -The entire comprehensive model should replace all
checked for vertical well cases only. To make the existing empirical correlations used to predict
comparison unbiased with respect to the correlations, two two-phase flow behavior in wells.
different sets of well cases are considered. One such set is
composed of all vertical well cases excluding 331 well -The slug flow model should be modified to include
cases from the Hagedorn and Brown data. The other set is flow mechanisms related to directional wells.
composed of all new vertical well cases that were never

158
SPE 20630 A M Ansari. N D Sylyester 0 ShQham and J P Brill 9

·The degree Qf emplClClsm in the cQmprehensive 11 Sylvester, N. D.: "A Mechanistic MQdel fQr TWQ-Phase
mQdel shQuld be reduced tQ further imprQve the Vertical Slug FIQW in Pipes," ASME J. Energy Resources
model. Tech. (1987), .1Q2., 206-213.

As a final remark, it shQuld be mentiQned that the


12McQuillan, K. W., and Whalley, P. B.: "FIQW Patterns in
present study is Qnly a first step tQwards develQping a
Vertical Two-Phase FIQw," Int. J. Multiphase Flow
cQmprehensive mQdel that shQuld replace existing pressure
(1985), 11., 161-175.
gradient cQrrelations for the entire range of operating and
design parameters.
13 BrQtz, W.: "Uber die VQrausberechnung der
REFERENCES AbsQrptiQnsgesch- windigkeit VQn Gasen in StrQmenden
Flussigkeitsschichten," Chern. Ing. Tech. (1954), .2..2.,
10ZQn, P. M. , Ferschneider, G., and Chwetzoff, A.: "A New 470.
Multiphase FIQW Model Predicts Pressure And
Temperature Profiles," paper SPE 16535 presented at the 14Schmidt, Z.: Experimental Study of Gas-Liqujd FIQW in a
OffshQre EurQpe CQnference, Aberdeen, Sept. 8-11, Pipeline-Riser System. M. S. Thesis, The University
1987. Qf Tulsa (1976).

2Hasan, A. 8., and Kabir, C. S.: "A StUdy of Multiphase FIQW .15Dukler, A. E., MarQn, D. M., and Brauner, N.: "A Physical
Behavior in Vertical Wells," SPE Prod. Eng. J. (May MQdel fQr Predicting the Minimum Stable Slug Length,"
1988). 263-272. Chern. Eng. Sci. (1985), 1379-1385.

3Taitel, Y., Barnea, D., and Dukler, A. E.: "MQdelling FIQW 16Wallis, G. B.: One-DimensiQnal TWQ-phase FIQw,
Pattern TransitiQns for Steady Upward Gas-Liquid FIQW in McGraw-Hili (1969).
Vertical Tubes," AIChE J. (1980), .aa.,
345-354.
17Hewitt, G. F., and Hall-TaylQr, N. S.: Annular TWQ-phase
4 Barnea, D., ShQham, 0., and Taitel, Y.: "Flow Pattern ~, PergamQn Press (1970).
TransitiQn fQr Vertical DQwnward TWQ·Phase FIQw,"
Chern. Eng. Sci. (1982),li, 741-746.
18Whalley, P. B., and Hewitt, G. F.: "The Correlation of
Liquid Entrainment FractiQn and Entrainment Rate in
5Barnea, D.: "A Unified MQdel fQr Predicting FIQw-Pattern Annular TWQ-Phase FIQw," UKAEA RepQrt, AERE-
Transition for the Whole Range of Pipe InclinatiQns," Int. R9187, Harwell (1978).
J. Multiphase Flow (1987), ll, 1-12.
19A1ves, I. N., CaetanQ, E. F., Minami, K., and Shoham, 0.:
6Harmathy, T. Z.: "VelQcity Qf Large DrQPs and Bubbles in "MQdeling Annular FIQW BehaviQr for Gas Wells,"
Media of Infinite or Restricted Extent," AIChE J. (1960), presented at the Winter Annual Meeting Qf ASME,
2., 281. ChicagQ, Nov. 27 - Dec. 2, 1988.

7Caetano, E. F.: Upward Vertjcal TWQ-phase FIQW ThrQugh 20LQpes, J. C. B., and Dukler, A. E.: "Droplet Entrainment in
an Annulus, Ph.D. DissertatiQn, The University Qf Tulsa Vertical Annular FIQW and its CQntributiQn tQ MQmentum
(1985). Transfer," AIChE J. (1986), 1500-1515.

8Zuber, N. and Hench, J.: "Steady State and Transient VQid 21 Brill, J. P., and Beggs, H. D.: TWQ-phase FIQW in Pipes,
Fraction Qf Bubbling Systems and Their Operating Limits. The University Qf Tulsa, 1988.
Part 1: Steady State OperatiQn," General Electric RepQrt,
62GL100 (1962). 22GQvier, G. W., and Fogarasi, M.: "Pressure Drop in Wells
Producing Gas and Condensate," J. Can. Pet. Tech.
9Zigrang, D., and Sylvester, N. D.: "Explicit ApprQximatiQn (Oct.-Dec. 1975), 28-41.
tQ the SQlutiQn Qf ColebrQok's FrictiQn factQr Equation,"
AIChE J. (1982), .2..a, 514. 23Asheim, H.: "MONA, An Accurate TWQ-Phase Well FIQW
Model Based Qn Phase Slippage," SPE Prod. Eng. J. (May
10Fernandes, R. C., Semait, T., and Dukler, A. E.: 1986), 221·230.
"HydrQdynamic MQdel for Gas-Liquid Slug FIQW in
Vertical Tubes," AIChE J. (1986), ZQ, 981-989. 24Reinicke, K. M., Remer, R. J., and Hueni, G.: "CQmparison
Qf Measured and Predicted Pressure DrQps in TUbing fQr

159
A COMPREHENSIVE MECHANISTIC MODEL FOR UPWARD
10 TWO- PHASE FLOW IN WELLBORES SPE 20630

High-Water-Cut Gas Wells; SPE Prod. Eng. J. (Aug. Pressure Gradients in Wells," ASME JERT,
1987), 165-177. 111 , 34 - 3 6, (M ar ch, 1 9 89) .

25Chierici, G. L., Cuicei, G. M. ,and Sclocci, G.: "Two-Phase 36Beggs, H. D. and Brill, J. P.: "A Study of
Vertical Flow in Oil Wells -- Prediction of Pressure Two-Phase Flow in Inclined Pipes," J. Pet.
Drop," SPE J. Pet. Tech. (Aug. 1974), 927-938. ilQ..b., 607 - 617, (May, 1973).

26Poettmann, F. H., and Carpenter, P. G.: "The 37palmer, C. M.: "Evaluation of Inclined Pipe
Multiphase Flow of Gas,Oil and Water Two- Phase Liquid Holdup Correlations Usi ng
Thr ough Vert i cal Flow Str ings wit h Experimental Data," M. S. Thesis, The
Application to the Design of Gas-Lift University of Tulsa (1975).
Installations," API Drilling and Production
Practices, 257 - 317 (1952).
38Mukherjee, H. and Brill, J. P.: "Pressure
Drop Corr elations for Inclined Two- Phase
27Fancher, G. H., and Brown, K. E.: "Prediction Row," Trans. ASME, JERT (Dec., 1985).
of Pressure Gradients for Multiphase Flow in
Tubing," Trans. AIME(1963), 2...2..a, 59-69.
39Aziz, K., Gov ier, G. W. and Fogar asi, M.:
"Pressure Drop in Wells Producing Oil and
28Hagedorn, A. R.: Experimental Study of Gas," J. Cdn. Pet. Tech .. 38 - 48, (July -
Pressure Gradients Occurring during September, 1972).
Continuous Two-Phase flow in Small
Diameter Vertical Conduits, Ph.D. NOMENCLATURE
Dissertation, The University of Texas at
Austin (1964). Descrjption

29Baxendell, P. B.: "The Calculation of Pressure a coefficient defined in Eq. 46


Gradients in High Rate Flowing Wells," SPE A cross-sectional area of pipe, m2
J. Pet. Tech. (Oct. 1961), 1023. b coefficient defined in Eq. 47
c coefficient defined in Eq. 48
300rki szew ski, J.: "Pr edict ing Two- Phase
C constant factor relating friction factor to Reynolds
Pressure Drops in Vertical Pipes," SPE J. number for smooth pipes
C' coefficient defined in Eq. 39
Pet. Tech. (J une 1967), 829- 838.
d differential change in a variable
D pipe diameter, m
31Espanol, H. J. H.: Comparison of Three
e error function
Methods for Calculating a Pressure Traverse El average percentage error, %
in Vertical Multi-Phase Flow, M. S. Thesis E2 absolute average percentage error, %
The University of Tulsa (1968). E3 standard deviation, %
E4 average error, psi
32Messulam, S. A. G.: Comparison of E5 absolute average error, psi
Correlations for Predicting Multiphase E6 standard deviation, psi
Flowing Pressure Losses in Vertical Pipes, f friction factor
M.S. Thesis, The University of Tulsa (1970). FE fraction of liquid entrained in gas core
g gravity acceleration, m/s 2
h local holdup fraction
33Camacho, C. A.: Comparison of Correlations
H average holdup fraction
for Predicting Pressure Losses in High Gas-
L length along the pipe, m
LiQujd Ratio Vertical wells. M.S. Thesis, The n exponent relating friction factor to Reynolds
University of Tulsa (1970). number for smooth pipes
n' exponent to account for the swarm effect on bubble
34Duns, H., Jr. and Ros, N. C. J.: "Vertical rise velocity
Flow of Gas and Liquid Mixtures in Wells," N number of well cases successfully traversed
Proc. 6th World Pet. Congress, 451, p pressure, psi [ N/m2 ]
(1963) . Q flow rate, m3 /s
Re Reynolds number
35Brill, J. P.: "Discont inui ties in the RPF Relative Performance Factor, defined in Eq. 92
Orkiszewski Correlation for Predicting S wetted perimeter, m

160
SPE 20630 A M Ansar i N P Sy Iy est er 0 Shoham. and J P Br ill 11

v velocity, m/s
V volume, m3
X Lockhart and Martinelli parameter
Y Lockhart and Martinelli parameter
Z empirical factor defining interfacial friction

Greek letters

p length ratio, defined in Eq. 27


1) film thickness, m
~ ratio of film thickness to diameter
,1. difference
E absolute pipe roughness, m
cjl dimensionless groups, defined in Eqs. 79 and 80
" no-slip holdup fraction
Il dynamic viscosity, kg/m-s
v kinematic viscosity, m2 /s
8 angle from hortizontal, rad or deg
p density, kg/m 3
cr surface tension, dyne/cm
't shear stress, N/m 3

Subscripts
a acceleration
A average
c Taylor bubble cap, core
crit critical
e elevation
f friction
F film
G gas
H hydraulic
ith element
I interfacial
L liquid
LS liquid slug
M mixture
mn minimum
N Nusselt
r relative
s slip
S superficial
SO slug unit
t total
1B Taylor bubble
TP two-phase

Superscript

* developing slug flow

161
TABLE 1 TABLE 3

RANGE OF WELL DATA STATISTICAL RESULTS USING ALL VERTICAL WELL CASES

E2 E3 E4 E5 E6 RFP
Nom. Dia Oil Rate Gas Rate Oil Gravity ("Al) ("Al) (psi) (psi) (psi) (- )
(in.) (STBO/D) (MSCF/D) (0 API)
HAGBR 10.8 15.1 -7.5 95.9 173.9 5.380

Old TUFFP' MODEL 14.5 19.2 -17.7 81.3 144.9 6.987


Databank 1-8 0-10150 1.5-10567 9.5-70.5
AZIZ 14.0 19.3 -18.6 98.4 182.5 7.542
Govier &
Fogarasi 22 2-4 8-1600 114-27400 17-112 DUNROS 14.7 21.9 23.2 102.0 176.3 8.392

BEGBR 16.7 23.0 52.0 121.7 199.5 12.913


Asheim 23 2~-6 720-27000 740-55700 35-86
ORKIS 21.1 39.5 50.9 154.9 298.8 15.374
Reinicke &
Remer 24 2~-7 0.3-5847" 448-44980 MUKBR 20.9 22.0 78.0 147.2 211.0 17.122

Chierici et al 25 2~-5 0.3-69 6-27914 8.3-46


TABLE 4
Prudhoe Bay 5~-7 600-23000 200-110000 24-86
STATISTICAL RESULTS USING ALL VERTICAL
WELL CASES WITHOUT HAGEDORN AND BROWN28 DATA

E2 E3 E4 E5 E6 RPF
("Al) (%) (psi) (psi) (psi) (- )

MODEL 10.1 14.8 -6.4 97.8 172.2 5.000


'Includes data from Poettmann and Carpenter26. Fancher and Brown27,
Hagedorn and Brown 2 8, Baxendell and Thomas29, Orkiszewski30, HAGBR 12.2 17.0 -12.2 130.6 207.2 6.801
Espanol 31 , Messulam 32 , and Camach033 field data from several oil
companies. AZIZ 12.8 18.0 -21.9 126.3 216.1 8.459

"Water flow rate DUNROS 15.0 22.8 33.1 135.2 209.2 10.814

BEGBR 18.2 23.3 81.2 167.2 235.6 19.168

MUKBR 20.6 22.6 92.5 181.9 239.5 21.301

ORKIS 27.4 46.8 77.5 223.4 362.4 22.400

TABLE 2 TABLE 5
STATISTICAL RESULTS USING ENTIRE DATABANK STATISTICAL RESULTS USING
ALL NEW VERTICAL WELL CASES
E2 E3 E4 E5 E6 RFP
("Al) (%) (psi) (psi) (psi) (- ) E2 E3 E4 E5 E6 RPF
(%) ("Al) (psi) (pSi) (pSi) (- )
MODEL 12.1 17.1 9.3 101.3 163.9 5.573
MODEL 8.6 12.3 -3.0 109.0 164.4 5.000
AZIZ 12.2 16.8 -20.8 116.6 190.4 7.049
HAGBR 10.6 14.8 13.1 122.1 166.2 8.934
HAGBR 9.2 13.6 -28.5 102.8 178.4 7.101
DUNROS 18.1 27.1 -6.4 165.8 216.7 9.281
DUNROS 12.2 18.5 33.4 110.9 177.7 8.470
AZIZ 10.2 14.7 -90.9 154.6 280.5 35.685
ORKIS 16.1 32.2 12.2 151.3 273.3 8.653
MUKBR 18.2 19.8 110.3 176.5 191.3 43.140
BEGBR 14.4 20.2 41.3 134.9 207.9 10.102
BEGBR 24.5 25.7 152.6 215.9 198.0 58.808
MUKBR 17.6 20.2 78.7 159.8 217.2 14.751
ORKIS 60.7 71.9 295.6 453.5 539.1 118.515
TABLE 6 TABLE 7

STATISTICAL RESULTS USING ALL WELL CASES STATISTICAL RESULTS USING ALL
WlTH OVER 75% BUBBLE FLOW WELL CASES WlTH 100% SLUG FLOW

E2 E3 E4 E5 E6 RPF E2 E3 E4 E5 E6 RPF
(%) (%) (psi) (psi) (psi) (- ) (%) (%) (psi) (pSi) (psi) (- )

MODEL 3.2 3.7 -25.3 67.0 76.9 5.000 AZIZ 14.8 19.8 5.6 102.9 173.8 6.016

AZIZ 3.2 3.7 -30.3 68.9 79. I 5.286 MODEL 16.2 20.4 13.0 101.2 160.8 7.413

ORKIS 3.3 4.3 -26.9 69.4 90.6 5.493 HAGBR 10.1 14.8 -19.7 90.4 176.8 7.605

DUNROS 3.6 4.0 -47.9 77.5 8.2 6.374 ORKIS 14.6 26.3 17.4 116.3 212.9 8.920

HAGBR 3.8 4.3 -44.9 78.7 90.1 6.511 BEGBR 15.5 21.3 43.7 114.8 184.9 13.181

BEGBR 3.8 4.8 -46.6 79.2 102.6 6.842 DUNROS 15.1 21.4 56.6 108.2 170.7 15.276

MUKBR 7.3 3.8 -154.0 155.6 83.3 12.852 MUKBR 21.5 21.3 99.1 153.2 197.2 24.146

en
'"

TABLE 8
TABLE 9
STATISTICAL RESULTS USING ALL VERTICAL
WELL CASES WITH 100% SLUG FLOW WITHOUT STATISTICAL RESULTS ALL WELL CASES
HAGEDORN AND BROWN28 DATA WITH 100% ANNULAR FLOW

E2 E3 E4 E5 E6 RPF E2 E3 E4 E5 E6 RPF
(%) (%) (psi) (pSi) (psi) (- ) (%) (%) (psi) (pSi) (psi) (- )

MODEL 16.2 20.3 -7.9 10.7 198.7 5.331 MODEL 9.7 12.4 -21.8 90.7 132.9 5.000

AZIZ 19.1 24.1 5.9 126.7 226.3 5.696 AZIZ 12.4 16.5 22.3 106.1 145.4 5.896

HAGBR 17.0 21.1 14.4 140.5 252.6 7.118 HAGBR 15.1 16.4 70.6 128.7 148.2 8.652

DUNROS 24.3 29.3 100.0 169.4 241.9 22.694 DUNROS 20.0 24.8 -79.0 174.9 223.1 11.293

ORKIS 29.6 43.5 101.3 199.8 321.2 24.619 MUKBR 25.5 19.9 202.1 219.9 196.7 17.409

BEGBR 24.7 26.3 118.9 177.0 251.2 25.873 BEGBR 32.2 18.0 250.7 261.9 180.2 20.515

MUKBR 33.2 24.2 152.3 215.4 253.3 32.319 ORKIS 78.7 68.2 504.0 544.9 407.9 45.810
0
t t t t
a
O
0 00 Oc?°080 0 20
0 0
0
° 10 I

D
0000 Ul I
"-
00 0 ° E I
000 ° 0
.03 0°0 >-
l-
I
I
,
0000 °0
U
o°0 °°00
() 008 0
...J BUBBLY BARNEA I
°O()Oo TRANSITION
0°0 000 LlJ
°0 ° > I
o 0000 0 0 0 0 0.°
0 00
ANNULAR
0 00
°0 0 0 0
0
I

S I
o000
0 ° 0'boooo 0

°°0 0
0 0
0
o °0·0 :::i 0.1 D
...J I
°000~0 ~ A I
a 000 u SLUG OR CHURN
I
0°00°
000 00 .. .. u..
0:: I
Ogo~o
LlJ
ll. 0.01 I
00 ::> I
(/)
I

BUBBLE
t SLUG
t t
CHURN ANNULAR
t 0.002 .
0 02 0.1 1 10
I
100
FLOW FLOW FLOW FLOW SUPERFICIAL GAS VELOCITY (m/s)

Fig. 1-Flow patterns in upward two-phase flow. Fig. 2-Typical flow pattern map for wellbores.

I
I
VTB

DEVELOPED
TAYLOR
BUBBLE--~

DEVELOPING
TAYLOR
BUBBLE

(o) DEVELOPED SLUG UNIT (b) DEVELOPING SLUG UNIT

Fig. 3-Schernatic diagram of slug flow.

164
10.

+ CALCULATED PRESSURE

9.0 T x MEASURED PRESSURE


ANNULAR
GAS CORE
I

8.0
1
'·1 ~
UJ
LIQUID FILM I
UJ

ENTRAINED
.j- lJ..
0
0
7.0

LIQUID DROPLET
.'1
I
....
0

6.0
V .
. C ~
I 0
~ SLUG
VF
l
. H LC .
t vF
~
0
CD
UJ
:c
5.0

. ~ ~
4.0
4"'1' I T I' ~
0

~ c:
TF TF
'.' . '.- i..' lJ..
UJ
u
3.0
"'ic
z "-
'\)
<::
~
(f)
H 2.0
~...
Cl
8 "'"""......
. "-
~~
1.0

0.0
0.0 500.0 1000.0 1500.0 2000.0
PRESSURE ( PSI
Fig. 4-Schematic diagram of annular flow. Fig. 5-Performance of the comprehensive model-typical pressure profile.

165

You might also like