Multiphase Flow in Wells
Multiphase Flow in Wells
Multiphase Flow in Wells
1
James P. Brill
Floyd M. Stevenson Endowed Presidential Chair in Petroleum Engineering
Executive Director; Fluid Flow Projects
U.ofTulsa
and
Hemanta Mukherjee
Manager; Production Enhancement, West and South Africa
Schlumberger Oilfield Services
j
'--
--
1
First Printing
Henry L. Doherty Memorial Fund of AIME
Society of Petroleum Engineers Inc.
l Richardson, Texas
1999
-
-
i
I..-
Table of Contents
Chapter I-Introduction ..•••••........•.•••.....••••..•••••....•••••.........••.•••• 1
1.1 Scope 1
1.2 Objectives of Monograph. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 1
1.3 Organization of Monograph ". . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 1
1.4 Historical Background 2
1.5 Nomenclature and Units 4
Chapter 2-Sing1e-Phase-F1ow Concepts ...............•..•....•..••••................• 5
2.1 Introduction 5
2.2 Conservation of Mass , 5
2.3 Conservation of Momentum. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 5
2.4 Pressure-Gradient Equation 5
2.5 Flow in an Annulus 11
2.6 Conservation of Energy 13
i Chapter 3-Multiphase-Flow Concepts .•.•••••.•.•••••.....•••.......••.•.......•.... 19
--
I 3.1 Introduction
3.2 Phase Behavior
3.3 Definition of Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..
19
19
20
3.4 Pressure Gradient . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 23
3.5 Flow Patterns 23
3.6 Liquid Holdup 24
3.7 Pressure-Traverse Computing Algorithm 26
3.8 Dimensional Analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 27
Chapter 4-Multiphase-Flow Pressure-Gradient Prediction ......•...••....••••••.......• 28
4.1 Introduction 28
4.2 Pressure-Gradient Prediction 28
4.3 Evaluation of Wellbore Pressure-Gradient-Prediction Methods 56
4.4 Pressure-Gradient Prediction in Annuli . . . . . . . . . . . . . .. 58
4.5 Evaluation of Annulus Liquid-Holdup and
Pressure-Gradient-Prediction Methods 66
4.6 General Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 67
I
L.
Chapter 5-Flow Through Restrictions and Piping Components .....•.................••. 70
5.1 Introduction 70
5.2 Description of Restrictions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 70
5.3 Flow Through Chokes 70
5.4 Flow Through Piping Components 76
Chapter 6-Well Design Applications •••.•..••......•......•••••..•••••.....••.••.•••• 78
6.1 Introduction 78
6.2 Vertical-Flow Performance 78
6.3 Inflow Performance 80
6.4 Production-Systems Analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 86
6.5 Artificial Lift 91
--
\
6.6 Gas-Well Loading
6.7 Erosional Velocity
92
94
6.8 Special Problems 95
Appendix A-Nomenclature and SI Metric Conversion Factors .••.......•••..•....••..•.. 99
Appendix B-F1uid and Rock Properties •.•••••....••••...••••••........•............ 102
Appendix C-Vapor-fLiquid-Phase Equilibrium •••..•.•••••••.••••...•...•.•••••••.••. 135
Appendix D-Tubing and Casing Properties ..•...••........•........•••..••...•.•.•.. 145
Author Index ••..•••...•••••.••••••••..•...•••••••••••••••••..••••••.•..••..•.••. 149
Subject Index .......••......•••....•.•...•••..•.....•....•••..............•...... 155
--
Chapter 1
Introduction
-
1 predict multiphase-flow behavior has improved dramatically in the
past decade. It is now possible to select tubing sizes, predict pressure
drops, and calculate flow rates in wells with acceptable engineering
essarily work for oil wells. Assumptions that are valid for some
wells are totally invalid for others.
accuracy. This chapter sets the stage for the monograph by describ- 1.2 Objectives of Monograph
ing the nature and occurrence of multiphase flow, and by presenting Multiphase flow in pipes is not unique to the petroleum industry.
important historical events that have impacted on the development The simultaneous flow of two or three phases in a single conduit is
of modem multiphase-flow concepts. encountered in petrochemical plants, refineries, stearn generators,
The common occurrence of multiphase flow in wells can be dis- automobile engines, storm sewers, nuclear reactors, and many other
cussed with the simplified production system shown in Fig. 1.1. areas. Consequently, the technical literature in chemical, civil, me-
Fluids entering the wellbore from the reservoir can range from an chanical, petroleum, and nuclear engineering all contain valuable
undersaturated oil to a single-phase gas. Free water can accompany contributions to multiphase flow in pipes. Unfortunately, each disci-
the fluids as a result of water coning, water flooding, or production pline uses different terminology, units, and nomenclature and is in-
of interstitial water. Alternatively, a free gas saturation in an oil res- terested in different ranges of variables. An important objective of
ervoir can result in a gas/liquid mixture entering the well. Retro- this monograph is to bring together the pertinent technology of all
grade condensation can result in hydrocarbon liquids condensing in relevant disciplines into a single book that uses the nomenclature
a gas condensate reservoir so that a gaslliquid mixture again enters and units standardized by the Society of Petroleum Engineers.
the wellbore. Even when single-phase gas or liquid flow exists near Numerous methods have been proposed to predict multiphase
the bottom of a well, multiphase flow can occur throughout most of flow in wells. Some are very general and others are applicable over
the wellbore. This is a result of evolution of gas from oil or con- a very narrow range of flow variables. Some are very empirical in
densation of gas with reduction of pressure and temperature as the nature and others make an attempt to model the basic phenomena
fluids flow up the well. encountered. It is vitally important that persons engaged in design-
Although many of the wells drilled on land tend to be nearly verti- ing wells be aware of the limitations of all methods. Thus, a second
cal, wells drilled offshore and in other hostile environments, such as objective of this monograph is to present the important methods to
the Arctic, are normally directional or deviated. Inclination angles predict flow behavicr in wells and to discuss their limitations and
can vary from vertical near the surface to horizontal near the produc- ranges of applicability.
tion zone. Flow rates of gas, oil, and water vary widely. Pipe diame- Finally, engineers engaged in the design of wells should have a
ters can be as small as I in. and as large as 9 in. Flow also can occur thorough understanding of the calculation procedures necessary to
in a casing/tubing annulus. Depths can range from a few hundred optimize their design calculations. This includes coupling the wells
feet to more than 20,000 ft. Pressures can be as low as a few atmo- to both surface facilities and reservoirs. It also includes understand-
spheres or as high as 20,000 psia. Temperatures can be above 400°F ing computer algorithms, fluid physical properties, and mass trans-
or approach the freezing point of water. Oil viscosities in wellbores fer for multicomponent mixtures. A significant part of this mono-
can be less than I cp or greater than 10,000 cp. graph is devoted to these concepts.
Fluids entering the wellbore often flow through a complicated
well completion region consisting of perforations, fractures, gravel 1.3 Organization of Monograph
pack, and other such items. The effect of this region must be in- Predicting flow behavior when multiphase flow occurs in wells re-
cluded when coupling the well to the reservoir through inflow per- quires an understanding of concepts that are not a part of the curricu-
formance relationship (IPR) procedures. Most wells contain some lum in most engineering disciplines. Before multiphase-flow
type of well-control device that requires produced fluids to flow technology can be mastered, one must have an adequate knowledge
through a restriction. This can vary from a bottornhole choke to a of not only single-phase-flow fluid mechanics, thermodynamics,
INTRODUCTION
r;:==== @
MQQE LQCAItON
1 SEPARATOR
2 SURFACE CHOKE
3 WEUHEAD
I 4 SAFETY VAlVE
L-
5 RESTRICTION
6 PM
7 PM.
8 Pr
1A GAS SAlES
16 STOCK TANK
INTRODUCTION 3
computer programs that are not only easy to use, but also incorpo- 6. Taitel, YM. and Dukler, A.E.: "A Model for Predicting Flow Regime
rate the very latest technology in a friendly, unintimidating manner. Transitions in Horizontal and Near Horizontal Gas-Liquid Flow,"
The faster computing speeds on PC's and engineering workstations AlChE J. (1976) 22, 47.
7. Brown, K.E.: The Technology of Artificial Lift Methods, Petroleum
also make possible a truly interactive, compositional-fluid-modeI
Publishing Co., Tulsa, Oklahoma (1980) 2a, 3a, 3b, and 4.
approach for more precise calculations of mass transfer and fluid 8. Taitel, Y, Bamea, D., and DukJer,A.E.: "Modeling Flow Pattern Tran-
properties. This approach gives better pressure and temperature pre- sitions for Steady Upward Gas-Liquid Flow in Vertical Tubes," AIChE
dictions and is more robust than the standard black-oil approach. J. (1980) 26, 345.
Also emerging is a trend for large standardized databases contain- 9. Bendiksen, K.H. et al.: "The Dynamic Two-Fluid Model OLGA:
ing information ranging from drilling and completion records to Theory and Application," SPEPE (May 1991) 171; Trans., AIME. 291.
well-test and production data. Soon it will be possible to access and 10. Black, P.S. et al.: "Studying Transient Multi-Phase Flow Using the
cross-reference these databases with a well identification and to in- Pipeline Analysis Code (PLAC)," J. Energy Res. Tech. (March 1990)
put data automatically into a muItiphase-flow design program, thus 112,25.
II. Pauchon, C. et al.: "A Comprehensive Mechanistic Model for Two-
dramatically reducing manual input requirements.
Phase Flow," Proc., Sixth International Conference on Multi-Phase
Production, Cannes, France (June 1993) 29.
1.5 Nomenclature and Units 12. Bamea, D., Shoham, 0., and Taitel, Y: "Flow Pattern Transition for
Much of the technology for multiphase flow in pipes was developed Vertical Downward Two-Phase Flow," Chern. Eng. Sci. (1982) 37, 741.
outside the petroleum industry. Consequently, there will be some 13. Bamea, D., Shoham, 0., and Taitel, Y.: "Flow Pattern Transition for
Downward Inclined Two-Phase Flow; Horizontal to Vertical." Chern.
confusion in nomenclature because many of the important publica-
Eng. Sci. (1982) 37, 735.
tions have different nomenclature and terminology, Occasionally, it 14. Bamea, D. et al.: "Gas Liquid Flow in Inclined Tubes: Flow Pattern
will be necessary to define new symbols or deviate from those rec- Transition for Upward Flow," Chern. Eng. Sci. (1985) 40,131.
ommended by the Society of Petroleum Engineers. As much as pos- 15. Bamea, D.: "Transition From Annular Flow and From Dispersed
sible, this monograph uses the standard symbols adopted by the So- Bubble Flow-Unified Models for the Whole Range of Pipe Inclina-
ciety of Petroleum Engineers.V tions," Inti. J. Multiphase Flow (1986) 12, No.5, 733.
SI is the official abbreviation, in all languages, for the Interna- 16. Bamea, D.: "A Unified Model for Predicting Flow Pattern Transitions
tional System of Units (Ies Systeme International d'Unites). How- for the Whole Range of Pipe Inclinations," Intl. J. Multiphase Flow
ever, customary units 23 still are used frequently in many parts of the (1987) 13, No. I, I.
17. Ozon, P.M., Ferschneider, G., and Chwetzoff, A.: "A New Multiphase
, , world, as well as throughout this monograph (except in Examples
Flow Model Predicts Pressure and Temperature Profiles in Wells," pa-
I 4.9 and 4.10, which deal with mechanistic models that were devel- per SPE 16535presented at the 1987SPE Offshore Europe Conference,
oped on the basis of SI units). Appendix A presents customary units, Aberdeen, 8-11 September.
along with factors to convert to the SI metric system of units. 18. Hasan, A.R. and Kabir, C.S.: "A Study of Multiphase Flow Behavior
in Vertical Wells." SPEPE (May 1988) 263; Trans .. AIME, 285.
References 19. Ansari, A.M. et al.: "A Comprehensive Mechanistic Model for Up-
ward Two-Phase Flow in Wellbores," SPEPF (May 1994) 143;
I. Mach, J.M., Proano, E.A., and Brown, K.E.: "Application of Produc- Trans., AIME, 297,
tion System Analysis to Determine Completion Sensitivity in Gas Well 20. Xiao, 1.1.,Shoham, 0., and Brill, J.P.: "A Comprehensive Mechanistic
Production," paper presented at the 1981 ASME Energy Sources Tech- Model for Two-Phase Flow in Pipelines," paper SPE 20631 presented
nical Conference and Exhibition, Houston, 18-22 January. at the 1990SPE Annual Technical Conference and Exhibition, New Or-
2. Gilbert, W.E.: "Flowing and Gas-Lift Well Performance," Drill. & leans, 23-26 September.
Prod. Prac. (1954) 126. 21. Chokshi, R.N.: "Prediction of Pressure Drop and Liquid Holdup in Ver-
3. Nind, T.E.W.: Principles of Oil Well Production, McGraw-Hill Book tical Two-Phase Flow Through Large Diameter Tubing," PhD disserta-
j Co. Inc., New York City (1964). tion, U. of Tulsa, Tulsa, Oklahoma (1994).
'-- 4. Brill, J.P. and Arirachakaran, S.1.: "State of the Art in Multiphase 22. SPE Letter and Computer Symbols Standardfor Economics, Well Log-
Flow," JPT(May 1992) 538; Trans., AIME, 293. ging and Formation Evaluation, Natural Gas Engineering, and Petro-
5. Dukler, A.E. and Hubbard, M.G.: "A Model for Gas-Liquid Slug Flow leum reservoir Engineering, SPE, Richardson, Texas (1986).
in Horizontal and Near Horizontal Tubes," Ind. Eng. Chern. Fund. 23. The SI Metric System of Units and SPE Metric Standard, SPE. Richard-
(1975) 14, 337. son, Texas (1984).
:
L-
\
j
,......
....
-,
Chapter 2
"I
'--
Single-Phase-Flow Concepts
v = 3t(~). (2.11)
v Because the pressure gradient in Eq. 2.11 results only from walI
shear stresses or friction, the pressure gradient is identical to Eq. 2.9.
Fig. 2.1-Control volume. Combining these equations gives
64
Newtonian fluids and flow through annuli, Fanning friction factors 14 = .M.
f = pvd N ; ••••••••••••••••••••••••••••• (2 . 12)
R
are retained to preserve original equations. Eq. 2.7 can be solved for
;
I shear stress
2.4.2 Turbulent Flow. Our ability to predict flow behavior under
fit
I-
turbulent-flow conditions is a direct result of extensive experimen-
r = (2.8)
tal studies of velocity profiles and pressure gradients. These studies
i have shown that both velocity profile and pressure gradient are very
Substituting Eq. 2.8 into Eq. 2.5, the frictional component of the
...... pressure-gradient equation becomes
sensitive to characteristics of the pipe wall. A logical approach to
defining friction factors for turbulent flow is to begin with the sim-
oro
0.09
,
0.08 LtIIIIittIH" CrifI"ctIl rrtlttS;I;1H/
flo • ... 10-1- l()Iti C_pltll hlrb"l,fIC, flp""
I O.05
0.07
",
I
a04
J
0.06
aOJ
-.~'" E
e-
0.0 5 ."
a.02
~~ _.
r-\ a.015
-
f-- -
r-r-
0.04 -
'- , 0.010
aoos
..
.: 0.03 ?
<,
r-,
c-, "
0006 :li!
II
-u- - _ . - -
I..- ~
.~ 0.025
u
~
.-
-.
0.002 I
.t
0.02
"
~
-- ....... 0001 ~
- c-; 0.0008
,~ 0.0006
~
--
.r. . ~ 0.0004
0015 - ""'" -.,.;
--0--".....
.......
....
........... 0.0001
~
j'lIi~t:- 0000,05
0.0 I c
--
E UG
0.009 fie ·0.000.001
aooo,or
0.001
I 2 ~ 6 8 I 2 } 4 6 8 I 2 3 4 6 8 I 2 3 4 6 8 I """ 2 3 4 6 8 I
I· xlO ·1· x 104 ·1· xl0& ·1· X 10' .1· xl0 7 .1
Reynolds number, HRe = pVd/1l
.......................... ; .. (2.14) I i I , , .,
I ! I'"~ I : I I
NI I: I I I
where 0.0 1 ,
0.008
, I ,
3,000 < N Re < 105 • 0.006 If',
Although the Blasius equation is considered less accurate, it re- 0.004 Concref~
\
0.003
,
1 ceives greater use because of its similarity to the laminar friction- I I
~ !\. III
factor equation. Both can be expressed in the form 0.002
,
I'\.
"'-
., '"
f = CN~e' (2.15)
~~"";q, I I , ' , , I
I'\. I , I ! Ii
Normally, the inside wall of a pipe is not smooth. In turbulent
0.001 '0"" 0",,- ,
~ 0.0008
flow, the roughness can have a significant effect on the friction fac- '" c ~'0: ..
~~ ~ @~11o..."
I •
ui 0.0006 i , ,I
tor and, thus, the pressure gradient. Wall roughness is a function of '"
Q)
-~
; the pipe material, the method of manufacture, the age of the pipe,
c
s: 0.0004
..
~. o-=" C>
~...; '
I 01 o......~ ~c:.. f\.. i
::>
and the environment to which it is exposed. e 0.0003 e-('Il...... ~~
'--
From a microscopic sense, wall roughness is not uniform. Indi-
vidual protrusions and indentations vary in height, width, length,
~
.~
a;
0.0002 ~ -1'{'-
...- K'f'\.
,
~~o '
shape, and distribution. The absolute roughness of a pipe, £, is the a: I
not result from its absolute dimensions, but rather from its dimen- 0.000,03
Cliq.. I I I
, °0'"
Q '
,
,
I 0.... I ~i'.
sions relative to the inside diameter of the pipe, eld. In turbulent I -:>~d.111 I 1'\.1'>. I
flow, the effect of wall roughness has been found to depend on both 0.000,02
j '<,.
"'o~ , I
I the relative roughness and the Reynolds number. If the laminar sub- I r
\,..,..
layer that exists within the boundary layer is sufficiently thick, the 0.000,01
behavior is similar to a smooth pipe. The sublayer thickness is di- 0.000,008 ~
rectly related to the Reynolds number.
, , r
0.000,005 I r-. I , , I
Nikuradse's? famous sand-grain experiments formed the basis 1 2 3 4 5 6 8 10 20 30 40 60 80 100
for friction-factor data from rough pipes. His correlation for fully Pipe diameter, d. in.
rough wall pipe is given as Eq. 2.16 and is still the best one available.
Fig. 2.3-Pipe roughness. 2
~= 1.74 - 2l0g(~) (2.16)
d
The region in whichfvaries both with Reynolds number and rela- _jjl = - 2 log [2£3/7 - 5N·0210g(23£/7d + N13 )]. ... (2.19)
tive roughness is called the transition region or partially rough wall. 'f . Re . Re
I..... Note that Eq. 2.17 degenerates to Eq. 2.16 for large Reynolds num-
in the same friction factor. The only way this can be done is to
compare the behavior of a normal pipe with one that is sand-rough-
bers corresponding to fully developed turbulent or rough pipe flow. ened. Moody/ has done this and his results, given in Fig. 2.3, are
Solving Eq. 2.17 forfrequires a trial-and-error process. Eq. 2.17 can still accepted values. However, these values should not be consid-
i
t be expressed as? ered inviolate and could change significantly as a result of paraffin
\.... deposition, hydrates, erosion, or corrosion. If measured pressure
-2
(d ~)]
gradients are available, a friction factor vs. Reynolds number rela-
2£ 18.7 tionship can be established and an effective relative roughness ob-
fe =
[
1.74 - 210g + . ........ (2.18)
tained from Fig. 2.2. Until it is again updated, this value of eld
NReJfes,
should be used for future predictions.
Values offare estimated.j.s, and then calculated.j-, until they agree Initial values of roughness often are needed for design calcula-
to within an acceptable tolerance. A direct substitution procedure tions. The recommended value for new tubing is e = 0.00005 ft. A
that uses the calculated value as the next assumed value results in common value used to generate pressure-gradient curves is
convergence in only two or three iterations. The initial assumption 0.00015 ft. For tubing exposed to an environment that causes sig-
can be obtained from one of the explicit smooth-pipe equations, or nificant changes in roughness, "very dirty" pipe can have rough-
from explicit approximations to the Colebrook equation. ness values of 0.00075 ft. II For most wells, the friction component
Numerous explicit approximations to the Colebrook equation of the pressure gradient is small compared with the potential ener-
have been proposed. Zigrang and Sylvesterlvhave given one of the gy component. Consequently, approximate values for absolute
most accurate and simple to use. roughness normally are sufficient.
- 3.274 + 62.400 = 59.126 psf/ft Eq. 2.25 can be separated into two expressions, one for each half of
the well.
- 0.0227 + 0.4333 = 0.4106 psi/ft. Upper half:
The pressure change is then
(Imf + I if)
dp = (- 0.0227 + 0.4333)(8,000) 18.75ygL/2 = (Pmf- Pif) 2 (2.26)
Note that the pressure change consists of a loss owing to friction of (I + ImA
- 181.9 psi and a gain from an elevation change of + 3,466.4 psi. 18.75ygL/2 = Pwf - Pm[ ( ) Wf 2 " (2.27)
While this method can be used with any number of steps, Cullender
2.4.3 Single-Phase Gas Flow. When a compressible fluid, such as and Smith l2 demonstrated that the equivalent of four-step accuracy
gas, flows in a well, the density, velocity, and. consequently, the can be obtained with a two-step calculation if the Simpson's rule 13
pressure gradient all vary with pressure. Cullender and Smith 12 de- numerical-integration approach is used. The resulting equation is
veloped the most widely used method to calculate flowing bottom-
hole pressure in gas wells. 18.75ygL = ( PW
f 6- Pif)( I
wf + 4/ mf + lif·) (2.28)
Neglecting kinetic energy, Eq. 2.5 can be written as
Use of the Cullender and Smith method to calculate flowing OOt-
dp fpv? . tornhole pressures in gas wells can best be explained with an exam-
dL = 2d + pgsmO. (2.20)
ple problem.
For gases,
p = pM/ZRT, v = q/A, q =qscBg, and Bg = Psc1Z/Tscp. Example 2.2-Single-Phase Gas Pressure Drop. With the follow-
Combining these expressions with Eq. 2.20 and separating vari- ing data, calculate the flowing bottomhole pressure in a gas well us-
ables gives ing the Cullender and Smith method with two increments.
Yg = 0.75
1: I I
L Pwf P L= 10,000 ft
dL = p 2 ~ dp, (2.21)
1j= 245°F
Prj= 2,000 psia
o Pif (ZT) g sin 0 +C Ts = HO°F
e = 0.ססOO7 ft
·For examples in this monograph, the gravitational conversion constant. g, = 32.2 Ibm flJIbf
d= 2.441 in.
sec2, appears when problems are solved With customary units. qsc = 4.915 MMscflD
\
I
I.-
Fig 2.4-Rheological models.'
(-~~L =en~n~l)(~),
Knudsen and Katz, I describes the types of rheological behavior that
can be encountered. (2.31)
Two methods commonly are used to design piping systems for the
where n' = the slope of a logarithmic plot of f w vs. 8v/d, often
transport of oil/water mixtures. The first method treats the mixture
called the flow-behavior index. From the same plot, the relationship
as a Newtonian fluid with an apparent viscosity that can vary with
of the tangential line to the laminar-flow curve can be written as
water fraction. This method is covered in Chap. 3 because it can in-
volve the combining of the viscosities from each phase to obtain a
mixture viscosity. The second method treats the mixture as a non- (2.32)
Newtonian fluid and is based on the following assumptions 14:
I. The mixture is homogeneous. Combining Eqs. 2.30 and 2.32 gives
2. Slippage between phases is neglected. Thus, in-situ holdups are pv 2 - n 'd n '
the same as their respective input volume fractions. N ReM.R 8n ' - 1K' (2.33)
3. The rheological behavior of the oil/water dispersion system is
suitably described by the Ostwald-de Waele power-law model. 15 and
T = K'yn·. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. (2.29) where N ReM.R = the generalized Reynolds number, and" = apparent
viscosity. Eqs. 2.33 and 2.34 clearly reduce to the normal Reynolds
If n' is unity, Eq. 2.29 will describe Newtonian behavior, and K' will number and the Newtonian viscosity for a Newtonian fluid, when
: be equal to the constant viscosity.zz. For a typical oil/water mixture, n' = I and K' =,u.
\-<
n' usually is less than unity, and Eq. 2.29 will describe pseudoplastic
(shear thinning) behavior. It is also possible for an oil/water mixture Non-Newtonian Friction Factor. In general, the procedure for
calculating friction factors for non-Newtonian fluids is similar to
to have n' greater than unity, resulting in dilatant (shear thickening)
that for Newtonian fluids.
behavior. The type of behavior that a fluid system will follow nor-
mally is unknown but can be determined from laboratory experi- Laminar Flow. From Metzner and Reed,16 the Fanning friction
ments with an appropriate viscometer. factor for non-Newtonian laminar flow can be written as
These tests must be conducted for a specified set of operating
conditions: input water fraction, temperature, and droplet size dis- j' = N 16 . . (2.35)
ReM_R
tribution of the dispersed phase (or, indirectly, mixing speed). Once
'I
the fluid system is correctly characterized, the frictional pressure
Turbulent Flow-Smooth Pipes. Dodge and Metzner'? proposed
gradients for pipe flow in this particular oil/water dispersed system
this implicit friction-factor equation,
readily can be determined.
Generalized Reynolds Number. Metzner and Reed l6 introduced
V(if' = n'·40·07510g[NR eM _
/'II-n/2 1] - ~.412'
n .
........ (2.36)
the concept of a "generalized Reynolds number" for non-Newto-
nian flow. The usual Reynolds number definition is given in Eq.
2.10, which also can be written as Turbulent Flow-Rough Pipes. Govier and Aziz 18 suggested
this friction factor for power-law, pseudoplastic fluids flowing in
N Re = P:wd (~).
rough pipes:
. (2.30)
where T w is the laminar wall shear stress, while the quantity, 8v/d, r = fM.R &), (2.37)
is also the true shear rate at the pipe wall for a Newtonian fluid. Thus,
the ratio, [r w / (8v/ d> ], corresponds to the Newtonian viscosity.«. where f M-R = the friction factor calculated from Eq. 2.36; j: =
For non-Newtonian flow, the relationship between the true shear Newtonian flow Fanning friction factor for rough pipe, calculated
rate and the apparent shear rate can be expressed as at the same generalized Reynolds number; and j; = Newtonian
10-11/2 E ]
- 4.0 log N I/ n' (41')12-n')2n'
[ Re
+ 3.71d ' ... (2.38)
M. R
where
Concentric Partially Eccentric Fully Eccentric
(3 = 1.5IJ/n'(0.~?7 + 2.12) - 4,~,15 - 1.057 . . . . . (2.39) DBC=O Dsc = (de - dr,4 Dsc = (de - drJ2
8=0 8=0.5 8=1
Note that Eq. 2.38 is analogous to the Colebrook equation for New- Fig. 2.5-Annuli configurations. 28
tonian fluids given by Eq. 2.17, and also requires a trial- and-error-
solution procedure, Calculate the pipe inlet pressure, Pr-
Fully Rough Wall Turbulence. Govier and Aziz \8 recommended,
PI = P2 + 21'Pm~ v~ = 205.9 psia.
Iff
-; = 4.0610gd/2e + 6.0 2.65
- - ,.
n
. . . . . . . . . . . .. (2.40)
For Newtonian fluids. Eq. 2.40 essentially reduces to the von Kar-
2.5 Flow in an Annulus
In the petroleum industry, flow in wells normally occurs in a tubing
man equation.lf
string. However, many oil wells with high production rates produce
11
\
\
I
'--
Goldstein.U analytical solutions also can be found for these flow
parameters for an eccentric annulus. These solutions are presented 28r------------------....,
in a later section. ECCENTRICITY
In turbulent flow, even for circular pipes, the mechanisms of tur- 0.0
0.1
bulence are by no means fully understood. However, various semi-
empirical and analytical models have been used successfully to pre- o
.. 24
0.2
dict the velocity distribution and pressure gradient. In noncircular ti 0.3
configurations, where the transport phenomena are intrinsically ~
~20 0.4
i more complex than for circular pipes, the formulation of an analyti- Ii
I.- cal model is even more complicated. Three ways have been used to E 0.5
predict the flow behavior ofa turbulent-flow field in an annulus: em- 2
CJ
pirical correlations, semi-empirical correlations, and application of c:: 0.6
o
universal velocity distributions.
Empirical correlations involve the application of curve-fitting
s
't:
0.7
u..
techniques to experimental data to predict an overall flow quantity, 0.8
12
such as friction factor. The resulting friction-factor correlations nor- 0.9
mally take the form of the Blasius-type expression, as given in Eq. 1.0
2.15 where C and n are determined empirically. Examples include K aO.553
the correlations of Knudsen and Katz.! and Dodge. 22 Winkler23
0.2 0.4 0.6 0.8 1.0
successfully used the Dodge correlation for flow in an annulus;
however, he noted that this procedure did not take into account the Annulus Pipe Diameter Ratio, K = dr/de
annulus pipe-diameter ratio. Fig. 2.6-Friction-geometry parameter in annuli and circular
Semiempirical approaches involve the use ofexperimental data for pipe (laminar single-phase flow at same Reynolds number).28
turbulent flow in combination with characteristics of laminar flow in
the same noncircular configuration. The Gunn and Darling-" proce- Thus,
dure is an example of this prediction category. Gunn and Darling con-
cluded that the similarity existing between friction factors for circular 16(1 - ld ........... (2.50)
and noncircular configurations in the laminar region is also accompa-
nied by a similarity in the turbulent region. Using dimensional analy-
sis, they showed that for turbulent flow in noncircular sections, the
i following functional dependency for friction factor exists.
I
where I CA is the Fanning friction factor and FCA is the friction-ge-
::J
\
L- ometry parameter for a concentric annulus.
INC = INC( N Re, (2.45) Through use of a bipolar coordinate system, Snyder and Gold-
stein 21 presented an analytical solution for laminar flow in an eccen-
where Fe and FNC are the so-called friction geometry parameters for tric annulus. Their solution was based on previous developments by
the circular and noncircular configurations, respectively. At low Heyda 20 and El-Saden. 25 Redberger and Charles.e? who presented
Reynolds numbers, the friction factor is inversely proportional to a numerical solution for the problem, used the same technique. Ap-
the ratio Fe/FNC' However, at high Reynolds numbers, the friction plying the Fanning equation, it can be shown that
factor becomes independent of the Fe/FNC ratio. At intermediate
values of Reynolds numbers, the function given in Eq. 2.45 is estab-
~£A'
JJ
= NF £A = NI 4(1 - K)\I - 1(2)
h4 ••...••••••.. (2.51)
Re Re ¢ sin 1]0
lished from suitable experimental data involving many noncircular
types of configurations. and
Newtonian LAminar Flow. The friction factor in laminar flow is 4(1 - K)2(1 - 1(2)
.......... (2.52)
determined from solution of the continuity equation, the equation of ¢ sinh 41]0
motion, and the Fanning equation. The solution is obtained for New-
tonian, fully developed, steady-state, axial flow. where
The Fanning equation can be written as
(2.46)
¢ = (cosh11; - coshl]ol 2( 11' -I
I
I] 0 - 2 I" e 2 2n 2 )
'"Ij - e nT/o
n=1
pvd;
+iCi~4110 - Sin~41])' (2.53)
N Re = ~. (2.47)
_ K(l + e 2) + (I - e 2)
The Fanning friction factor in circular pipes, I; , is given by cos h 11; - 2Ke ' (2.54)
r,
I; = N = N'
16
(2.48)
and
Re Re
K(l - e 2) + (l + e 2)
cosh 110 = 2Ke . (2.55)
where Fp is a friction geometry parameter, having a constant value
of 16 for pipe flow. Although an infinite-series term is involved in Eq. 2.53, the num-
For a concentric annulusl-' ber of terms can be truncated after several terms. As before, IFA is
the Fanning friction factor and FEA is the friction-geometry parame-
. . . . . . . .. (2.49) ter for an eccentric annulus. Fig. 2.6 presents a plot ofEq. 2.52. Note
that K = 0 for pipe flow, resulting in FEA = 16, and that for fully con-
centric pipes (e = 0), FEA approaches 24 as K approaches 1.0. This
also corresponds to the analytical value for single-phase laminar
Newtonian Turbulent Flow. Caetano et al. recommended the Therefore, with Eq. 2.62, the frictional pressure gradient for a con-
centric annulus can be calculated as
Gunn and Darling-" procedure for turbulent flow because of its sim-
plicity and good performance. The approach combines the behavior 2f'pv 2
for friction factor in noncircular configurations, as given by Eq. ............................ (2.62)
de - dr'
2.45, with a large amount of turbulent-flow data for many non-cir-
cular configurations. Combining the Gunn and Darling develop-
ments into a Nikuradse-type expression, the friction factor for con- Eccentric Annulus. Haciislamoglu and Langlinais-f developed a
centric and eccentric annuli are predicted, respectively, from correlation to predict frictional pressure losses of power-law fluids
in an eccentric annulus. A correlating parameter, R, was defined as
I the ratio of frictional pressure losses in an eccentric annulus to those
I
in a concentric annulus. The empirical correlation, which is valid for
I...-
eccentricities from 0 to 0.95, pipe diameter ratios of 0.3 to 0.9, and
flow-behavior indices, n', of 0.4 to 1.0, is given by
0.8454
R = I - o.on.!',
n a,
(~)
_- 410g[{'
N
(F
ICA F
p )0.45eXP[-INRe.
3.OOO lOb
l/ ]}'h] -0.4 0.1852 0.2527
Re
CA
15e2w(i) + 0.96e 3W(i)
..... (2.63)
( - dV)
dr w
= 4n' + 2_8_v_
4n' de - d, '
(2.58) J!. [pv(e +
dL
--.L-)]
pgcl
= - Qnd.
A
.. . . . . . . . . . . . .. (2.66)
where, n' = flow-behavior index. The parameter J is the mechanical equivalent of heat and is neces-
The generalized Reynolds number concept also can be applied for sary when dealing with customary units where mechanical energy
an annular-flow geometry. The generalized Reynolds number is and thermal energy have different units.
Because specific enthalpy is defined as
h=u+-L. (2.70)
pgJ
Eq. 2.69 can be expressed as
dh - Q7Cd v dv gsinO
dL = --W- - gJ dL - g;J .............. (2.72)
I
I
The heat flux, Q, is defined in terms of overall heat-transfer coef-
ficient and temperature difference between the fluids and the sur-
I
roundings. Thus,
Tr Q = U(Tf - Te ) • . . . . . . . . . . . . . . . . . . . . . . . . . .. . . (2.73)
I
I
• Because of the strong dependence of enthalpy and heat transfer
on temperature, Eq. 2.72 is used to determine temperature change
L... I ~
when fluids flow through pipes. Normally, the kinetic energy term
is negligible. Therefore, for a horizontal pipe, an increase in fluid
~ enthalpy equals the heat transferred to the fluid from the surround-
ings. Also, if no heat transfer occurs, an increase in elevation causes
T2 - T 1 = -q- - - (2.77)
(2.69) 27CM k t - 2 •
q In(;,:) where the Grashof number, NGr> reflects the extent of motion of the
T,j - T w = 2nM -k-,- . (2.80) annulus fluid resulting from natural convection.
( ah )
~ [-;]
_ C (dT) - CJ!l, (2.98)
ap T p dp h
• ..•...••...•..... (2.100)
where YJ = the Joule-Thompson coefficient and represents isenthal-
pic cooling (or heating) by expansion. Combining Eqs. 2.97 and
2.98 gives and Eq. 2.106 degenerates to
(2.99)
T, = (T.; - gGLsinO) + (T; - T.i)e- L/A
Combining Eqs, 2.99 and 2.72 gives
+ g G sin OA ( I - e - L/A) - gsinOA(1 _ e-L/A,),
dTf dp JgcCp
Cp dL - CJ!l dL =
.................... (2.110)
_ gsinO _ 2.-dv _ U;;d(T - T.). which is equivalent to the Ramey 32 equation for an ideal gas.
f (2,100) For the case of an incompressible liquid,
gel gel dL
Eq. 2.100 can be simplified to this differential equation. YJ = - J~{iJ (2.111)
dTf T T. I dp
dL + Af = A + JPCpdLifJ, .. , , (2.101) and
where
CpW ...... (2.112)
A = Und , .. , , , , . . .. (2.102)
16
I
i
L
-
,
Chapter 3
Multiphase-Flow Concepts
5500
Reservoirs
.
~
Gas-Condenaate
Reservoirs r
~
~ fB 8/ 1
l- Critical I ff /1
SODO
.!!
Point I §J I where Bg is derived from the engineering equation of state to be
•
Q. B1 It, I
?it I Bg = PscZT/pZscT" . . (3.4)
; 2500 ~t I
J!,
\
'-- ••I! 0./
I if,
Appendix B gives methods to predict gas compressibility, Z.
Do. For the compositional model, volumetric flow rates are calculated
~
0 2000
/ -al
from
••e ~I
/
I
I
• :,
L a: a:, wr(1 - x g )
1500 '0 qL = PL (3.5)
£'
:,
, and
1000 A1
I qg = wrxg/Pg, (3.6)
I
where xg is the no-slip quality or gas mass fraction and is obtained
L
250 300 350 from the results of a VLE calculation as follows,
VMg
Fig. 3.1-Typical phase diagram.1 x g = VM + LM . ......................•..... (3.7)
g L
phenomena, it should not be used for temperatures approaching the If free water exists when the compositional model is used, the water
critical-point temperature. flow rate must be added to qi: to account for all the liquid.
A second parameter, called the oil formation volume factor, Bo ,
also has been defined to describe the shrinkage or expansion of the
Example 3.t-Compositional-Model Flow Rates. A gas-conden-
oil phase. Oil volume changes occur as a result of changes in dis-
sate well is flowing at a rate of 500,000 IbmID. At a given location
solved gas and because of the compressibility and thermal expan- in the pipe, a VLE calculation is performed on the gas composi-
sion of the oil. Dissolved gas is by far the most important factor that tion, yielding
causes volume change. Oil formation volume factor can be mea-
\ sured in the laboratory or predicted with empirical correlations giv-
L en in Appendix B. Once the black-oil-model parameters are known,
L = O.05(mole liqUid) V = 0.95(mole vapor),
mole feed mole feed
oil density and other physical properties of the two phases can be
calculated. Appendix B gives methods to predict these properties.
When water also is present, solution gas/water ratio, Rs w , and wa-
ter formation volume factor, Bw, can be defined. Appendix B also
ML = IOO(mol~b~uid) Mgv = 20(mol~b~apor)'
gives correlations for these parameters and physical properties of and
the water. The amount of gas that can be dissolved in water and the
corresponding possible changes in water volume are much smaller PL = 50 Ibm/ft 3 Pg = 5.0 lbm/ft'.
than for gas/oil systems.
From Eq. 3.7,
3.2.2 Compositional Model. For volatile oils and condensate (0.95)(20.0)
i'-- fluids, vapor/liquid equilibrium (VLE) or "flash" calculations are Xg = (0.95)(20.0) + (0.05)( 100.0)
more accurate to describe mass transfer than black-oil-model pa-
rameters. Appendix C provides a description of VLE calculations. = 0.792 Ibm vapor/Ibm mixture.
Given the composition of a fluid mixture or "feed," a VLE cal-
culation will determine the amount of the feed that exists in the va- With Eq. 3.5,
por and liquid phases and the composition ofeach phase. From these _ (500,000)(1 - 0.792) _ 3
results, it is possible to determine the quality or mass fraction of gas qL - (86.400)(50) - 0.024 ft /sec.
Il-
in the mixture. Once the composition of each phase is known, it also
is possible to calculate the interfacial tension and densities, enthal- With Eq. 3.6,
pies, and viscosities of each phase. Appendix C also gives methods
_ (500,000)(0.792) _ 3
to predict these properties. qg - (86,400)(5.0) - 0.917 ft [sec.
VLE calculations are considered more rigorous than black-oil-
model parameters to describe mass transfer. However, they also
are much more difficult to perform. If a detailed composition is 3.3 Definition of Variables
available for a gas/oil system, it is possible to generate black-oil When performing multiphase-flow calculations, single-phase-flow
parameters from VLE calculations. However, the nearly constant equations often are modified to account for the presence of a second
composition that results for the liquid phase and the increased phase. This involves defining mixture expressions for velocities and
computation requirements make the black-oil model more attrac- fluid properties that use weighting factors based on either volume
tive for nonvolatile oils. or mass fractions. The choice of variables and weighting factors
often depends on the predicted flow pattern. The following sections
3.2.3 Volumetric Flow Rates. After mass transfer calculations are discuss each parameter.
completed, it is possible to calculate the in-situ volumetric flow
rates of each phase. For the black-oil model, volumetric flow rates 3.3.1 Weighting Factors. When gas and liquid flow simultaneously
are determined from up a well, the higher mobility of the gas phase tends to make the gas
travel faster than the liquid. This is a result of the lower density and
qo = qo"B o, (3.1) viscosity of the gas. Under steady-state conditions, this results in a re-
The water cut,fw' based on in-situ rather than stock-tank flow rates, = 0.778 ft 3/sec.
is simply 1 -/0.
With Eq. 3.10,
I
'--
3.3.2 Velocities. As described previously, the individual phase velo-
cities normally are quite different. Only for the cases of the high- vSL = AqLp 0.778
= 0.196 =.397f/
t sec.
ly turbulent, dispersed-bubble-flow pattern and high-velocity, an-
With Eq. 3.3,
nular-flow pattern, in which the fluids exist as a homogeneous
L mixture, are the phase velocities essentially equal. For all other
cases, significant slippage can occur between the gas and liquid. Un-
der steady-state-t1ow conditions, slippage will result in a dispropor-
qg
= [10 X 10 6 - (l 0, (00)(281 »)(0.0091) = 0 757 f 3/
86,400 . t sec.
, qL 0.778 0507
AL = qL + qg = 0.778 + 0.757 =. .
VSg = qg/A p • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • (3.11)
:: Charles4
~
a:
TIGHT
a-
u
> MEDIUM ~
t::
en 20 iii ~ ] Present StUdy (SN-250)3
0
LOOSE o
8
sa
III
s
> is
10
Guzhov et al.6e
_Russell et al.7
L Studies have shown that Eq. 3.18 often is not valid for the viscos-
ity of two immiscible liquids, such as oil and water. Chap. 2 pres-
ented a treatment of oil/water mixtures as a non-Newtonian fluid.
Input Water Fraction Required To Invert Mixture
L
pend mainly on the determination of which phase is continuous. The
apparent liquid viscosity then will be governed primarily by the vis-
(3.19)
cosity of the continuous phase, because this is the phase that pre-
dominates at the pipe wall where most of the friction losses occur. or
Other factors, such as the dispersed-phase viscosity and the droplet-
size distribution of the dispersed phase, also are important.I u, = (Jl~L) X ~11-HLl] (3.20)
For some oil/water systems, the viscosity of the liquid mixture can
be several times greater than the oil viscosity when the continuous and
phase is oil but the water fraction is approaching the point where an
inversion of the dis persion or ernul sion will occur.l This increase has Iln = IlLA L + Illi - AL)· (3.21)
been attributed to two different causes: a deviation from Newtonian
behavior near the inversion point, and an energy loss requirement to Although Eq. 3.19 has never been used in multiphase-flow design
invert the mixture that is not accounted for in the conservation equa- correlations, Eq. 3.20 is used by the Hagedorn and Brown? correlation
tions used. Thus, when a Newtonian-fluid flow model is used, the en- described in Chap. 4. All other empirical correlations use Eq. 3.21.
ergy loss appears as an additional pressure loss that is seen as an in- The following expressions have been used to calculate multi-
crease in apparent viscosity. The inversion point of an oil/water phase-flow mixture densities.f
mixture occurs at water fractions ranging from 0.2 to 0.5, with inver-
Ps = PLHL + pi l - HJ. . (3.22)
sion taking place at lower water fractions when oil viscosities are
high. 3 Fig. 3.3 gives a plot that demonstrates this for experimental Pn = pLA L + pi l - AJ, ..........•........... (3.23)
data from a variety of sources. 3- 7 Furthermore, by the use of surfac-
j and
tants, inversion can be accomplished at even lower water fractions.
L Although Eq. 3.18 is the most common way to treat the apparent A2 (I _A)2
viscosity of an oil/water mixture, a more accurate method is to use Pk = PL H~ + P g (I _ /iL)' (3.24)
the oil viscosity when oil is the continuous phase and the water vis-
cosity when water is the continuous phase. Fig. 3.3 then could be Eq. 3.24 contains the subscript k because it appears in the kinetic en-
used to estimate the water fraction at which inversion takes place. ergy term for the specific case of a homogeneous-mixture, momen-
An even better alternative is to conduct flow tests on actual crudes tum-conservation equation.
t t f
o
... 0
0 ,,0 .. ", ' . .. . • I_
..... ~ ~. I- •
:r=: ".~.
it • •
0
I".
.
c" Co l'ho
"00 __ 0' II:'
10 •t •• . ~
Oc • 0
"Do. 0
~o·
0,0 '1 0
.· t ...
•
, II
L-
"Ie
e.:l 0' 0 b ..
". +.
.1... .-ft.O:,-,. •
~ .,:
.000
·0 • • -.
0
.,,, G ·r·:
Soo·af
·1. ".;.t'
'" 0 10
I- 01
o00__a·
.. 0
... 10
.
t!e ~
-I
I
(I
,f'
L o
0°
e;.~ 0 D
00"
. " 00
o~
o D" 0 0
t:1
.. .
ri .-
.: 0 10
I
I
•
d .'
~.
I
D1•
.1 • • 1°
• ~
,. •
0 °0 0 0
0 00° .
: .....
..-....
0
o " 00
0°0 00
8 ·
t
Bubble Flow
1
Slug Flow Churn Flows
Bubble
Flow
Front
Slug Flow
Back Churn
Flow
Annular
Flow
When performing temperature-change calculations for multi- phases and with a single liquid phase. Consequently, these predic-
phase flow in wells, it is necessary to predict the enthalpy of the mul- tions may be inadequate for high-pressure, high-temperature wells,
tiphase mixture. Most VLE calculation methods include a provision or for wells producing oil and water or crude oils with foaming ten-
to predict the enthalpies of the gas and liquid phases. If enthalpies dencies. A consensus exists on how to classify flow patterns.
are expressed per unit mass, the enthalpy of a multiphase mixture
can be calculated from 3.5.1 Flow-Pattern Classification in Wells. For upward multi-
phase flow of gas and liquid, most investigators10now recognize the
h = hd I - x g) + hgx g. (3.25) existence of four flow patterns: bubble flow, slug flow, chum flow,
and annular flow. These flow patterns, shown schematically in Fig.
3.4, are described next. Slug and chum flow are sometimes com-
3.4 Pressure Gradient bined into a flow pattern called intermittent flow. It is common to
The pressure-gradient equation derived in Sec. 2.3 for single-phase introduce a transition between slug flow and annular flow that incor-
flow can be modified for multi phase flow by considering the fluids porates chum flow. Some investigatorsl l-J? have named annular
to be a homogeneous mixture. Thus, flow as mist or annular-mist flow.
Bubble Flow. Bubble flow is characterized by a uniformly dis-
tributed gas phase and discrete bubbles in a continuous liquid phase.
Based on the presence or absence of slippage between the two
where the definitions for pfand vfcan vary with different investiga- phases, bubble flow is further classified into bubbly and dispersed-
tors. For vertical flow, () = 90°, sin () = I, dL = dZ, and the equation bubble flows. In bubbly flow, relatively fewer and larger bubbles
can be written as move faster than the liquid phase because of slippage. In dispersed-
bubble flow, numerous tiny bubbles are transported by the liquid
phase, causing no relative motion between the two phases.
Slug Flow. Slug flow is characterized by a series of slug units.
The pressure-drop component caused by friction losses requires Each unit is composed of a gas pocket called a Taylor bubble,13.14
evaluation of a two-phase friction factor. The pressure drop caused a plug of liquid called a slug, and a film of liquid around the Taylor
by elevation change depends on the density of the two-phase mix- bubble flowing downward relative to the Taylor bubble. The Taylor
ture which is usually calculated with Eq. 3.22. Except for conditions bubble is an axially symmetrical, bullet-shaped gas pocket that oc-
of high velocity, most of the pressure drop in vertical flow is caused cupies almost the entire cross-sectional area of the pipe. The liquid
by this component. The pressure-drop component caused by accel- slug, carrying distributed gas bubbles, bridges the pipe and sepa-
eration is normally negligible and is considered only for cases of rates two consecutive Taylor bubbles.
high flow velocities.
Many methods have been developed to predict two-phase, flow- Churn Flow. Chum flow is a chaotic flow of gas and liquid in
ing-pressure gradients. They differ in the manner used to calculate which the shape of both the Taylor bubbles and the liquid slugs are
the three components of the total pressure gradient. Chap. 4 de- distorted. Neither phase appears to be continuous. The continuity of
scribes these methods. the liquid in the slug is repeatedly destroyed by a high local gas con-
centration. An oscillatory or alternating direction of motion in the
3.5 Flow Patterns liquid phase is typical of chum flow.
Predicting the flow pattern that occurs at a given location in a well Annular Flow. Annular flow is characterized by the axial conti-
is extremely important. The empirical correlation or mechanistic nuity of the gas phase in a central core with the liquid flowing up-
model used to predict flow behavior varies with flow pattern. Brill ward. both as a thin film along the pipe wall and as dispersed drop-
and Beggsf summarized numerous investigations that have de- lets in the core. At high gas flow rates more liquid becomes
scribed flow patterns in wells and that made attempts to predict dispersed in the core, leaving a very thin liquid film flowing along
when they occur. the wall. The interfacial shear stress acting at the core/film interface
Essentially all flow-pattern predictions are based on data from and the amount ofentrained liquid in the core are important parame-
low-pressure systems, with negligible mass transfer between the ters in annular flow.
23
Annular Flow. The gas is a continuous phase flowing in the core
··. of the annulus cross-sectional area. The liquid flows upward, par-
tially as wavy films around the tubing and casing walls and partially
· .,'.:-.,
• I .. 0 • • I •
,-,
#I ": 0 in the form of tiny spherical droplets entrained in the gas core. The
0" :go • '. I •• outer film that wets the casing wall is always thicker than the inner
0.1
1 "" . -'" film flowing on the tubing wall.
"
o "I: "
10
·: .;. '
• ',I'.'
Liquid accumulation near the pipe-wall contact point is an addi-
•a 1'G
0
··....," ..
•
•
• I
• I ••
'
.,
tional characteristic of annular flow in a fully eccentric annulus .
This accumulation results from the merging of the casing and tubing
" • " 1\" 0 '. I"
I" 'I •• liquid films, which probably happens as a result of the low local
: 0
· : : I,. gas velocities.
.. "'.'
to
II " ; 0 •
' Comparison among flow patterns occurring in upward vertical
• I "
•..t: ...
L . .: a
Q : " II
"
.' • I'
• • 0'1 • •
.. • flow in a pipe and in an annulus reveals that the existence of an inner
pipe in the annulus changes the characteristics of slug flow and an-
nular flow. The Taylor bubbles in an annulus are not symmetric,
having a preferential liquid flow channel through which most of the
liquid phase is shed backward. Two films exist in the annular-flow
pattern, one flowing around the tubing wall and one around the cas-
ing wall. These flow modifications seem to be a function of the pipe-
Bubble
diameter ratio and the eccentricity of the annulus.
Annular
Flow Flow
3.5.3 Flow-Pattern Occurrence. The following is a description of
Fig 3.6-Flow patterns In upward vertical flow through a fully ec- a typical sequence of flow patterns in an oil wellbore.
centric annulus. 15 Starting with a flowing bottomhole pressure above the bubble-
point pressure, only a liquid phase exists at the bottom. As the liquid
flows upward, it experiences pressure decline resulting in the libera-
3.5.2 Flow-Pattern Classification in Annuli. The experimental tion of some of the gas dissolved in the liquid phase. The liberated
data collected by Caetano et al. 15 reveal that, although the same gas appears as small bubbles in the continuous liquid phase, which
flow patterns described for wellbores occur in annuli, their charac- characterizes the bubble-flow pattern. As the flow continues up-
teristics can be substantially different. Thus, it is essential to define ward, further decrease in pressure and temperature occurs, resulting
the flow patterns in these configurations. Figs. 3.5 and 3.6 show in gas expansion and the liberation of more solution gas from the oil
flow patterns in concentric and fully eccentric annuli, respectively. phase. This creates more and larger bubbles that start coalescing
with each other. The coalescence creates larger Taylor bubbles sepa-
Bubble Flow. The gas phase is dispersed into small discrete rated by the continuous liquid phase. The slug-flow pattern thus oc-
bubbles in a continuous liquid phase, forming an approximately ho- curs. Further continuation of the upward motion of the flow into the
mogeneous flow through the annulus cross-sectional area. The dis- region of lower pressure causes the expansion of Taylor bubbles
crete bubbles occur in two different shapes, namely spherical
along with the liberation of more gas from liquid slugs. This creates
bubbles and cap bubbles. The spherical bubbles are very small, on
a chaotic two-phase flow, defined earlier as chum flow. The churn
the order of 3 to 5 mm diameter, compared with the annulus-cap
flow continues to exist until the gas flow rate is sufficiently high to
bubbles, which are relatively larger but still always smallerthan half
push the liquid against the pipe wall. This characterizes the exis-
of the configuration hydraulic diameter. The upward movement of
tence of annular flow.
the small spherical bubbles follows a zig-zag path, whereas the cap
Because of continuous changes in pressure, temperature. and
bubbles follow a straight path with a faster rise velocity. In a fully
mass transfer between the two phases, the flow pattern having a con-
eccentric annulus, there is a tendency for the small bubbles and the
tinuous liquid phase at the bottomhole can be completely trans-
cap bubbles to migrate into the widest gap of the annul us cross-sec-
formed into a flow pattern with a continuous gas phase at the well-
i...... tional area. This causes a higher local void fraction relative to the
head. Fig. 3.4 shows this progressive change of flow patterns .
cross-sectional average void fraction. At high liquid velocities, the
After identifying the different flow patterns, tools are needed to
mixture appears to flow at the same velocity with no slippage be-
predict their occurrence and flow behavior. This requires a thorough
tween the phases, regardless of the annulus geometry.
understanding of the mechanisms of each flow pattern. Ansari et
Slug Flow. This flow is characterized by large cap bubbles of gas al. 16 used the best models for flow-pattern prediction and for flow-
moving upward, followed by liquid slugs that bridge the entire pattern behavior. They are discussed in Chap. 4.
cross-sectional area and contain small spherically distributed gas
L bubbles. The large gas bubbles, which occupy almost the entire
cross-sectional area of the annulus, are similar to the ones occurring
in pipe flow and also are termed Taylor bubbles. The Taylor bubbles
3.5.4 Flow-Pattern Prediction. Methods to predict the occurrence
of the various flow patterns in wells have fallen into two categories.
Most early attempts were based on conducting experimental tests in
do not occupy the total cross-sectional area because they have a small-diameter pipes at low pressures with air and water. Flow pat-
preferential channel through which most of the liquid ahead of the terns were observed, and the values of various flow parameters at
bubble flows backward (see Fig. 3.5). This preferential channel ex- the transition between flow patterns were determined. Empirical
ists from the top to the bottom of the bubble and from the tubing wall flow-pattern maps were drawn that could be used to predict the tran-
to the casing wall. Because of the presence of this channel, no sym- sitions. Chap. 4 presents some of the more successful maps devel-
metry is observed for the Taylor bubble with respect to either verti- oped with this approach. The second method to predict flow patterns
cal or horizontal planes. The liquid phase flows backward in the considers the basic mechanisms that are important in causing a
form of films, around the Taylor bubble, and through the preferen- flow-pattern change. This approach is not restricted to a narrow
tial channel, wetting both the tubing and the casing walls. This tends range of flow parameters and has proved to be highly successful.
to create a high turbulent region behind the Taylor bubble. Contrary Chap. 4 presents mechanistic models for flow-pattern prediction.
to the concentric annulus case, for a fully eccentric annulus (see Fig.
3.6) the preferential liquid channel always is located where the pipe 3.6 Liquid Holdup
walls are in contact.
Sec. 3.3 discussed the interrelationships between gas and liquid
Churn Flow. The characteristics of chum flow are similar to phases traveling at different velocities (slippage). Eqs. 3.13 and 3.14
those of pipe flow. There is no change in flow characteristics ob- show the importance of liquid holdup in predicting velocities of each
served with the annulus configuration. phase. Eqs. 3.19, 3.20, 3.22, and 3.24 show the importance of liquid
II
L.
Fig 3.B-Nuclear densitometer.
j=2
n = number of segments
m = number of calculation t-.p J(~f )dL. (3.28)
increments in a segment
o
For compressible and slightly compressible fluids, the pressure
gradient from Eq. 3.26 varies throughout the pipe length. This is true
I
\.-
for multiphase mixtures and even for crude oils above the bubble-
point, in part because the temperature also varies throughout the
j=3 well bore. For oil and gas wells, the pressure gradient throughout the
piping system will vary with pressure, temperature, and inclination
angle. Pressure-drop or pressure-traverse calculations require that
;=4 both the pressure-gradient equation and the enthalpy-gradient equa-
tion (Eq, 2.72) be coupled by use of a numerical algorithm.
Eqs. 2.5 and 2.105 constitute an initial-value problem with two
coupled, ordinary differential equations in the variables P and T, re-
spectively. All computer programs for performing steady-state,
multi phase-flow, pressure-traverse calculations use what is often
called a "marching algorithm." To solve this initial-value problem,
\ Fig. 3.9-Segmenting typical well bore.
! start at either the top or bottom of a well and "march" to the other
L- end, numerically integrating the pressure-gradient and enthalpy-
very small. As a result, such things as changes in humidity or minute gradient equations over short increments of pipe.
concentrations of water in the liquid phase can cause significant Fig. 3.9 shows a typical deviated wellbore that has been divided
measurement errors. In addition, Kouba et al. showed that dynamic into five pipe segments, with a new segment resulting whenever a sig-
calibration of the sensors was necessary because the capacitance nificant change occurs in inclination angle. New segments, or nodes,
measurement also was dependent on the location of the liquid phase also would be required where there is a change in mass flow rate,
within the sensor. where a pump is located, or where the pipe diameter or configuration
changes. A change in mass flow rate would occur when a well is gas-
3.6.2 Nuclear Densitometers. Brill et al. 21 used nuclear densitom- lifted or when production from different zones is commingled.
eters or gamma-ray absorption sensors to determine rnultiphase- Segments in Fig. 3.9 could be sufficiently long that there is a sig-
I flow behavior in large-diameter pipelines in the Prudhoe Bay field nificant change in pressure gradient within a segment because of
i of Alaska. They used a pair of densitometers to measure the liquid changes in such variables as densities. superficial and in-situ veloci-
L holdup and the lengths of slugs and bubbles. Since then, these de- ties, or even flow pattern. Thus, even for a pipe segment, it is neces-
vices have been used to acquire data in many different field and lab- sary to divide the pipe into calculation increments within which the
oratory tests. Fig. 3.8 shows a nuclear densitometer. pressure gradient can be considered constant. The pressure drop
Nuclear densitometers use a radioactive source that emits gamma then is determined by
rays. The source, often Cs 137, is contained in a lead-shielded cylinder
for safety reasons to prevent the escape of gamma rays. When the top
..................... (3.29)
of the cylinder is removed, a cone of gamma rays is emitted. They
pass through the pipe wall, through the fluids in the pipe, and through
any insulation or other coatings around the pipe. Each of these media Fig. 3.10 gives a flow chart of the "marching algorithm" for a
will absorb some of the gamma rays, depending in part on the density single pipe calculation increment for which Pi and 1; are known
of the material. The number of gamma rays that successfully pass andpi+1 and 1;+1 are to be calculated. A complete traverse is calcu-
through all media can be counted and converted into an electrical sig- lated by sequentially marching through all increments in all seg-
nal. Because the major variable involved is the amount ofliquid in the ments of the well.
cone of gamma rays at a given time and the liquid phase absorbs more The approximate analytical solutions to the enthalpy-gradient
gamma rays than the less-dense gas phase, the electrical response is equation shown in Fig. 3.10 were covered in Sec. 2.6. Note that the
directly proportional to the liquid holdup. numerical integration of the enthalpy-gradient equation involves a
Nuclear densitometers suffer from several disadvantages. Per- trial-and-error solution for 1;+1 for a guessed value of Pi+l. It also
haps their greatest drawbacks are safety and governmental regula- involves calculation of an average enthalpy gradient, and deter-
tion. Gamma rays are potentially dangerous, although adequate mination of the enthalpy of each phase and the mixture enthalpy at
measurements can be obtained with relatively small sources of ra- the end of the increment. Approximate analytical solutions for inte-
dioactive material. The response time of the electronics is fairly grating the enthalpy-gradient equation normally are adequate and
long, typically on the order of 10Hz, depending on the size and type simplify the calculations significantly. The PVT-calculation box in-
of radioactive source and the method used to count the gamma rays. volves the determination of mass transfer between the phases and
Thus, the measured lengths of rapidly moving slugs and bubbles the densities, viscosities, and interfacial-surface tensions at the
may not always be as accurate as desired. Also, the measurement of guessed average pressure and temperature in the increment. Appen-
liquid holdup can be sensitive to the location of the liquid in the pipe. dices Band C provide procedures to perform these calculations.
Chapter 4
Multiphase-Flow
Pressure-Gradient Prediction
4.1 Introduction Category "b." Slip considered, no flow pattern considered. A cor-
The pressure-gradient equation for single-phase flow in pipes was relation is required for both liquid holdup and friction factor. Be-
developed in Chap. 2, by use of the principles of conservation of cause the liquid and gas can travel at different velocities, a method
mass and linear momentum. The same principles are used to calcu- must be provided to predict the portion of the pipe occupied by liq-
late pressure gradient for multiphase flow in pipes. However, the uid at any location. The same correlations used for liquid holdup and
presence of an additional phase makes the development much more friction factor are used for all flow patterns.
Category "c." Slip considered, flow pattern considered. Not only
complicated. Sec. 1.4 gives the historical background of methods
are correlations required to predict liquid holdup and friction factor,
used to predict multi phase-flow phenomena.
but methods to predict which flow pattern exists are necessary. Once
Early investigators treated multiphase flow as a homogeneous
the flow pattern is established, the appropriate holdup and friction-
mixture of gas and liquid. This approach did not recognize that gas
factor correlations are determined. The method used to calculate the
normally flows faster than liquid. The nonslip approach tended to
acceleration pressure gradient also depends on flow pattern.
underpredict pressure drop because the volume of liquid predicted The following list gives the published empirical correlations for
to exist in the well was too small. vertical upward flow and the categories in which they belong.
Improvements to the no-slip methods used empirical liquid-hold-
up correlations to account for slippage between the phases. Al- Method Category
L ~
c
0 . . ,,
,
' ,
0.2
u: .... ,, ,
'"
....'<::::-:<.:::
0.01
.... 0 10 N
... (N L~/ ~575)
gv
L
( Psc )
---.!:f
N
--- d
I. Determine no-slip mixture density: Liquid-Holdup Prediction. A liquid-holdup value must be deter-
mined to calculate the pressure-gradient component that results
p ; = PLA L + PgO - AL)
from a change in elevation. To correlate the pseudo liquid-holdup
values, Hagedorn and Brown used four dimensionless groups pro-
= (47.61)(0.507) + (5.88)(0.493) posed by Duns and Ros.?
Liquid velocity number:
= 27.04 lbm/ft' .
2. Determine friction factor: N Lv = v5L m· (4.3)
1.8
0.0 1.6
(J
I/J
~-J
1.4
1.2
I
L
Fig. 4.3-Hagedorn and Brown 4 correlation for NLC'
dp
(~t+ (~t
L dZ= 1 - Ek
... , (4.10)
E =
k
dZ(d
dp dZ
P) Ps~(v~)
=~'
(4.9)
1. Determine Duns and Ros dimensionless groups:
ace
= 0.13 cp.
N - (1,488)(27.04)(7.83)(0.5)
Re - 0.13
L Nd = 120.872d Yt:
OL
= 1.21 X 106
and
62)
= (120.872)( 1 ( 47.6
8.41
1) = 1438
. ,
~ = (0.00~6) = 0.00012.
and
From Fig. 2.2 or Eq. 2. 17,f= 0.0135.
6. Determine pressure gradient, neglecting kinetic energy effects:
From Eq. 4.2,
(11.87)(0.0024) (1, 700)°·1 Gray Method. Gray developed a method to determine the pres-
sure gradient in a vertical gas well that also produces condensate
= (11.54)°·575(143.8) l4.7 fluids or water.> A total of 108 well-test-data sets were used to de-
velop the empirical correlations. Of these data sets, 88 were ob-
= 7.81 X 10- 5 tained on wells reportedly producing free liquids. The authors cau-
tioned use of the method for velocities higher than 50 ft/sec, nominal
H diameters greater than 3.5 in., condensatelliquid loadings above 50
:'tpL = 0.3.
bbllMMscf, and waterlliquid loadings above 5 bbllMMscf.
From Fig. 4.4, the abscissa, G, is Gray proposed this equation to predict the pressure gradient for
two-phase flow in vertical gas wells.
. ....••........ (4.11)
Gray suggested this expression foroL in Eqs. 4.12 and 4.13 when Using Eqs. 4.22 and 4.25 to eliminate the gas- and liquid-phase
both condensate and water are present. velocities in Eq. 4.24 gives
Iflo + 0.617/.,.0.,.
°L = 10 + 0.617/.,. (4.17)
Gray stated that the liquid holdup in condensate wells often tends
to be smaller than in oil wells producing at comparable gas/liquid
ratios. This probably arises from natural time lags in the flashing and V Sg + QIVSL - a2
condensing processes, and also from the lower average interfacial . . . . . . . . . . . . . . . . . . . . . .. (4.26)
2a 2
surface tension in condensing systems compared with flashing
black-oil systems. The formulation of Eq, 4.15 is such that the liquid Two limiting cases must be defined to make the holdup expres-
holdup is ordinarily very near to the no-slip holdup. sion complete. When the constant-slip term, a2, (buoyancy) is zero,
the holdup becomes
Friction-Factor Prediction. The effect of liquid holdup on fric-
tion loss can be interpreted as a variation in wall roughness from that
I ordinarily experienced in single-phase. dry-gas flow. Gray pro-
L posed that friction factors for wet-gas wells were wholly dependent
on a pseudo-wall-roughness factor because flow is normally in the
When the liquid superficial velocity approaches zero, the flow situa-
tion may correspond to gas bubbling through stagnant liquid. The
fully-developed turbulent region. Friction factors can be obtained holdup for this case is found by rearranging Eq. 4.26 to obtain
from the Moody diagram (Fig. 2.2) or from Eq. 2.16 for completely
V Sg
turbulent flow. The pseudo-wall-roughness, e, was correlated with H Ll v SL - = I - a2' ............•...••...••.• (4.28)
O+
a modified Weber number, similar to that used in the Duns and Ros 7
method for mist flow. Defining a roughness variable,
Friction-Factor Determination. This two-phase friction-factor
r28.50 correlation was used in Eq. 4.21.
e = - - 0 -L ' (4.18)
Pnv ;"
1 = aJn~:, (4.29)
then, for R ;::= 0.007,
L e = e' (4.19) where Pk is defined in Eq. 3.24. The wall friction factor,f", is deter-
mined from Fig 2.2 for a no-slip Reynolds number defined by
and, for R<0.007,
Pnvmd
N Ren =~ (4.30)
R(e' - e g )
e = eg + 0.0007 ' (4.20)
Parameter Values. For slug flow, the normal values of the three
~ith the restriction that e ;::= 2.77 x 10- 5. eg is the absolute wall parameters are a\ = 1.2, a2 = 0.35 lid,
and a3 = 1.0. For homoge-
L roughness for single-phase gas flow. It is noteworthy that the cor-
relation finally adopted includes both the effects of unusually high
apparent roughness observed in certain laboratory test conditions
neous flow the normal values of the three parameters are at = 1.0,
a2 = 0, and Q3 = 1.0. Asheim showed that, by selecting either slug
flow or homogeneous flow, the three parameters can be adjusted for
and the effects of low apparent roughness in certain field systems. an optimized fit of field data.
~ = (~) (~)
and minimize computation errors.
+ psg +
L The pressure gradient is determined from
dp IPnv~
dL = U- + psgsmu.
. II
(4.21)
f aa
Duns and Ros Method. The Duns and Ros 7 method is a result of
(4.31)
2
_.- ---~--I---I
10
21---1_4
10 - 1 '----:_":-_-':_-LJ'--.....L.~-.l--.l--'--.....:__-"---'---.l----'
10-1 2 10
Ng v
Fig. 4.5-Duns and ROS7 flow-pattern map.
groups were identified as being important and were used to select The following procedure is used to calculate the elevation com-
the range of variables in the experimental program. Eqs. 4.3 through ponent of the pressure gradient.
4.6 gave these four groups. 1. Calculate the dimensionless slip velocity, S, using the ap-
propriate correlation. The correlations for S are different for each
Flow-Pattern Prediction. Fig. 4.5 shows the flow-pattern map flow pattern and are given later. .
developed by Duns and Ros. They identified four separate regions 2. Solve Eq. 4.33 or 4.34 for the slip velocity.
for computation purposes, Regions I through III and a transition re- 3. Calculate the liquid holdup from Eq. 4.35 or 4.36.
gion. Fig. 4.5 illustrates the actual flow patterns existing in each region. 4. Calculate the slip density from Eq: 3.22,
Duns and Ros also identified the heading region as a fifth region, but
Ps = PLH L + pi l - HJ. . (3.22)
this is now considered part of Region II. In this monograph, we will re-
fer to Regions I through III as bubble, slug, and mist flow, respectively. 5. Calculate the elevation component of the pressure gradient, as
The flow-pattern transition boundaries are defined as functions of given in Eq. 4.31.
the dimensionless groups Ngv and NLv. For these transition bound- Bubble Flow. Bubble flow exists if Ng v < N gvB/ S ' For bubble flow.
aries, Duns and Ros proposed these equations.
the dimensionless slip-velocity number is given by
Bubble/slug boundary:
s= Vs ~ ••••••.••••.•••.••••••••••••••• (4.33)
_
S - 1.938v s
4
Vaz·
r;;;. . . . . . . . . . . . . . . . . . . . . . . . . . .. (4.34)
The slip velocity was defined by Eq. 3.15 and, when combined with
Eqs. 3.13 and 3.14, yields a quadratic equation for obtaining HL.
V Sg VSL
Vs = vg - VL = I _ H - H ' •.•••••.••••••••• (4.35)
L L
or
90 l~ o.=.::.:::.=o_...",_ _
80 9
70
'2
9
5
&0
0.5
50 0.4
40 0.3
;,. 1
lA. 0.9 30 ......
:... 0.8 o.2'L-.;~.L.L.u.uL.;:-L..1.LJ.llJ~~.J...l.JJ.Ll.U-.....I.....J.J...l..U.w........L~LJ:I:_'"
~ 0.7 20 w~ 1~
0.6 . 10
0.5
o
0.4
-10
0.3 -20
-30
0.2'-:-~-:!-~~,...-..................._ .....~_~--:..,..~_-+--J Fig. 4.8-Duns and Ros7 bubble-flow, friction-factor parameter.
10-3 2 3 4 5&78~0-2 5 1.0 2
N _ PgVS~
Fig. 4.7-Duns and RoS7 bubble-flow, slip-velocIty parameters. Re, - /-lg . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. (4.46)
,
Duns and Ros noted that the wall roughness for mist flow is the
L From experimental data, Duns and Ros developed this equation fori:
I = II~' , (4.40)
thickness of the liquid film that covers the pipe wall. Waves on the
film cause an increased shear stress between the gas and the film
that, in turn, can cause the greatest part of the pressure gradient.
The friction factor is governed mainly by II, which is obtained from These waves result from the drag of the gas deforming the film in
a Moody diagram (Fig. 2.2) as a function of a Reynolds number for opposition to the surface tension. This process is affected by liquid
the liquid phase. viscosity and also is governed by a form of the Weber number
N - PLVSLd
ReL -~. . ..••.•.•....••.•••.•.•••....• (4.41) PgV~l
N we = -----aL' (4.47)
The factor/2 in Eq. 4.40 is a correction for the in-situ gas/liquid ratio
and is given in Fig. 4.8. The factor/3 is considered by Duns and Ros This influence was accounted for by making NWe a function of a di-
as a second-order correction factor for both liquid viscosity and in- mensionless number containing liquid viscosity,
situ gas/liquid ratio. It becomes important for kinematic viscosities
greater than approximately 50 cSt and is given by
................................ (4.48)
........................ , (4.42) Fig. 4.10 shows the functional relationship, where the coordinates
are NWe vs. Nwe Nil.
Duns and Ros considered the acceleration component of the pres- The value of roughness may be very small, but the relative rough-
sure gradient to be negligible for bubble flow. ness never becomes smaller than the value for the pipe itself. At the
Slug Flow. Slug flow exists if Ngv s/s < Ngv < N gvS/T, For slug transition to slug flow, the waviness of the film may become large,
flow, the dimensionless slip-velocity number is with the crests of opposite waves touching and forming liquid
0.982 bridges. Then eid approaches 0.5. Between these limits, eld can be
gv
S = (I + F s)(N ) + ~~, (4.43) obtained from equations developed from Fig. 4.10.
(I + F 7NL vl
................ (4.49)
where Fs, F6, and F7 are given in Fig. 4.9 as functions of the liquid
viscosity number, NL, and
and
F~ = 0.029N d + F6 • .. . . . . .. . .. .. . . .. .. .. .. .. . (4.44)
The friction pressure-gradient component for slug flow is calcu- ...... (4.50)
lated exactly the same way as for bubble flow. Also, the acceleration
component for slug flow is considered negligible.
where d is in feet, vSg is in feet per second, P g is in pounds per
Mist Flow. Mist flow exists if Ngv > N gv . Duns and Ros as-
sumed that, at high gas flow rates, the liquid'i; transported mainly cubic foot, and 0L is in dynes per centimeter.
as small droplets. The result is nearly a no-slip condition between Values ofI for the mist-flow pattern can be found for e/d> 0.05
the phases. Thus, S = 0, V s = 0, and H L = A. L . The mixture from this extrapolation of the Moody diagram.
density for use in the elevation component of the pressure gradient
then is calculated from Eq. 3.23.
Friction in the mist-flow pattern originates from the shear stress
between the gas and the pipe wall. Thus, the friction component of I = 4
I
2 + 0.067 ( ~ )1.73} . ....... (4.51)
the pressure gradient is determined from
{
[410g lO ( 0.27 a)]
(
cdtz
P )f = IP2gdV~g . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. (4.45)
As the wave height of the film on the pipe wall increases, the actu-
Because there is no slip, the friction factor is obtained from a al area available for gas flow decreases because the diameter open
Moody diagram (Fig. 2.2) as a function of a Reynolds number for to flow of gas is now d - e. Duns and Ros suggested that the friction
the gas phase. component of the pressure gradient could be refined by replacing d
...
1.6
1.4
~ :4f- --
~
~ 0.10
1.2 .....10
..... 0.09
0.08
0.07
0.06 2 4 68
O.Os 10'
0.04
0.03
Fig. 4.1G-Duns and ROS7 mist-flow, film-thickness correlatIon.
where
51 = 1.2 + (0.24)(11.87) + (1.116)(11\51:87)
Ngvr'/M - N gv = 4.946.
A = N _ N (4.56)
gV rr/M gV Sj Tr From Eg. 4.35,
Increased accuracy was claimed in the transition region if the gas (4.946)
1.655 ft/sec.
L density used in the mist-flow pressure-gradient calculation was mo-
dified to be (1.938) ( iJ.4
61
1)
0.25
35
3. Determine friction factor: 0.4 ...__----r---,.-----r---,----,
From Eq. 4.41,
~ = 0'(~06 = 0.00012.
L f I Vsr:
. N2/3
d
4VSL
(0.0175)(3.86)( 143.8)2/3
(4)(3.97)
0.1
o 10 30 40 50
:.12 = 1.0.
From Eq. 4.42,
Fig. 4.11-Griffith and Wallis 13 C1 correlation.
L - I (0.0175) 3.86
f3 - + 4 (50)(3.97)
1.0006.
which is equivalent to Eq. 4.36 for the Duns and Ros correlation. Or-
From Eq. 4.40, kiszewski adopted the Griffith 14suggestion that 0.8 ftlsec is a good
approximation of an average VS' In Sec. 4.2.2 we will show that Vs
(1.0)
1= (0.0175)(1.0006) = 0.0175. is a function of the gas and liquid densities and surface tension. The
liquid holdup determined from Eq. 4.60 then is used to calculate slip
4. Determine pressure gradient, neglecting kinetic energy effects: density with Eq. 3.22, which in tum is used to calculate the elevation
From Eqs. 4.31. 3.22. and 4.39. component of the pressure gradient.
dp (0.0175)(47.61)(3.97)(7.83) The friction pressure-gradient component for bubble flow is
dZ = (2)(32.174)(0.5) given by
L (32.174)
+ [(47.61)(0.559) + (5.88)(1 - 0.559)] (32.174) ........................ (4.61)
= 0.80 + 29.21 = 30.01 psf/ft The friction factor is obtained from a Moody diagram (Fig. 2.2) as
= 0.208 psflft. a function of relative roughness and Reynolds number for the liq-
uid phase.
PL(vsL!HL)d
N Re = #L . . .•..................... " (4.62)
Orkiszewski Method. Orkiszewskif tested several published cor-
relations with field data and concluded that none was sufficiently
The acceleration pressure-gradient component for bubble flow
accurate for all flow patterns. He then selected what he considered
was considered negligible.
L to be the most accurate correlations for bubble and mist flow and
proposed a new correlation for slug flow. The slug-flow correlation
Slug Flow. Slug flow exists if )..g > )..g8/S' and N g\' < Ngvs/T,' The
was developed with the Hagedorn and Brown" data. Orkiszewski
selected the Griffith and Wallis 13.14 method for bubble flow and the slip density is calculated from
Duns and Ros 7 method for mist flow.
Flow-Pattern Prediction. Orkiszewski used the Duns and Ros
flow-pattern transitions for the boundaries between slug flow and
mist flow, including the transition region between them. Eqs. 4.32b Orkiszewski developed Eq. 4.63 by performing mass and volume
and 4.32c defined these. For the boundary between bubble flow and balances on a typical slug unit consisting of a Taylor bubble P and
slug flow, he chose these criteria established by Griffith and Wallis. a liquid slug. A similar Griffith and Wallis development neglected
Bubble/slug transition: the presence of a liquid film around the Taylor bubble and the possi-
)..g8/S = LB. . . . . . . . . . . . . • . . • . . • • . . . . . . . . . . . .. (4.58) bility of liquid droplets being entrained in the Taylor bubble. Conse-
quently. Orkiszewski proposed the last term in Eq. 4.63 to account
where for the distribution of liquid in these regions. This modification was
meant to extend the Griffith and Wallis work to include the high-ve-
v2
L B = 1.071 - 0.2218 ;;. . (4.59) locity-flow range.
Griffith and Wallis correlated the bubble-rise velocity, Vb, by
and Vm = ft/sec, vSg = ft/sec, d = ft, and Ln is constrained algebra- the relationship
I
ically to be ;::0.13.
L Bubble Flow. Bubble flow exists if )..g = I -)..L $ )..88/S' The
Vb = C 1C2 lid. . (4.64)
liquid holdup for bubble flow is determined from where CI and Cz are expressed in Figs. 4.11 and 4.12 as functions
of NRe• and NRc"
V Sg ]
( I + v:;
Vm)2
- 4V; • . .. (4.60) PLV~
NRc. = -yiL , (4.65)
1.4
1.3
1.2
1.1
L NReL
TABLE 4.1-QRKISZEWSKIB LIQUID DISTRIBUTION Orkiszewski used the data of Hagedorn and Brown to calculate and
COEFFICIENT EQUATIONS correlate the liquid distribution coefficient, F. Orkiszewski did not de-
fine criteria to establish which liquid phase is the continuous phase
Continuous Value Equation
Liquid of vm for when both oil and water are present. Fig. 3.3 can be used for this pur-
Phase (IVsec) r pose. Table 4.1 identifies liquid distribution coefficient equations. De-
pending on the continuous liquid phase and the value of the mixture ve-
Water <10 4.72
locity, the value of r is calculated from one of these expressions.
Water >10 4.73
Oil <10 4.74 r = O.OI~/~gJ1.L - 0.681 + 0.23210gv m
Oil >10 4.75
- 0.428 log d, (4.72)
and
0.04510g,uL
r
L
N = PLvmd = 0.709 - 0.16210gv m
ReL J1.L" (4.66) JO.799 -
Fig. 4.12 was extrapolated so that Vb could be evaluated at higher - 0.888 log d, (4.73)
liquid Reynolds numbers. When C2 cannot be read from Fig. 4.12,
Vb can be calculated from this set of equations.
0.012710gCuL + 1)
When N«; :5 3,000, r = d l 415 - 0.284 + 0.167 log V m
and
0.027410gCuL + I)
r = d l 371 + 0.161 + 0.56910gd + X,
Vb = (0.35 + 8.74 6
X 1O- N Re Jlid. . (4.68)
. . . . . . . . . . . . . . . . . . .. (4.75)
O.OI IOgCu L + I) ]
. =!(.. ,+
2
Vbs+~
13.59J1.L) ,
PLvd
............. (4.69)
X = - log V m
[
d1.571 + 0.397 + 0.63 log d
.................... (4.76)
where and ,uL is in centipoise, d is in feet, and Vm is in feet per second.
The value of r is constrained by these limits.
V bs = (0.251 + 8.74 x 10 -6N Re Jlid . ... ... .... (4.70)
If V m < 10,
Because Vb and N Reb are interrelated, determination of Vb requires
an iterative procedure when using Figs. 4.11 and 4.12 or Eqs. 4.67 r ~ - 0.065vm (4.77)
through 4.70. The procedure follows.
1. Estimate a value of Vb. A good first guess is and,
(~t==fP;;~[(:~:~:)+fl , (4.79)
:. flow pattern is not bubble flow.
From Eq. 4.32.
i
L wherefis obtained from the Moody diagram (Fig. 2.2) by use of the N g VS/ T, == 50 + (36)(11.84) == 476.2.
Reynolds number given by Eq. 4.66. The acceleration pressure-gra-
dient component for slug flow was considered negligible.
Orkiszewski recommended the Duns and Ros method if either Because N gv == 11.54 < 476.2,
mist flow or the transition region between mist flow and slug flow
is predicted. Flow pattern is slug flow.
Modifications. The Orkiszewski method can cause a convergence 2. Determine slip density:
problem in the computing algorithm, described in Sec. 3.7, for cal- From Eq. 4.66.
culating pressure vs. depth traverses in wells. The problem results
== (1,488)(47.61)(7.83)(0.5) == 286 105
from discontinuities between Eqs. 4.72 and 4.73 for water as the N ReL 0.97 . x
continuous liquid phase, and between Eqs. 4.74 and 4.75 for oil as
IL the continuous phase. Brill 16 showed that the constraints given by
Eqs. 4.77 and4.78 are ineffective in eliminating the pressure discon-
:.Fig. 4.12 cannot be used.
tinuities that result from the use of either pair of equations. Triggia"
suggested that coefficients in Eqs. 4.73,4.75, and 4.76 be modified Guess Vb == 0.5 lid == (0.5) j(32.174)(0.5)
so that the slopes of these curves are retained but the discontinuities
are eliminated. This solves the convergence problem but also may == 2.0 ftlsec.
affect the accuracy of results. The resulting equations for water and
oil, respectively, were From Eq. 4.65,
'Personal communcation with A. Triggia, Petrooras SA, Rio de Janeiro (1984). Yes. Therefore, I' == - 0.159.
ADz et aL Method. The Aziz et al.9 method uses many of the funda-
mental mechanisms that form the basis of modem mechanistic models. .................. , (4.92)
1.0
(4.95)
Bubble Slug
0.1
C = 0.345 1 -
[ e( -0.029N ][ 1 - e(-3.3m
y )
7-N
E)]
- ...... (4.96)
and
0.01.~ _ _""'--::~ -L!-::- ~~ --'
0.1 100
(4.97)
- - - This Work'o
:5 18 25
2.50
The friction pressure-gradient component for slug flow is deter-
mined from
I
! . . . . . . . . . . . . . . . . . . . . . . . . . .. (4.99)
I
'--
The friction factor is obtained from a Moody diagram (Fig. 2.2) UN 2.00
as a function of relative roughness and a Reynolds number given by
PLvmd
N Re =~. . (4.100)
Mist Flow. Mist flow exists when Nx > N3 for Ny < 4 or Nx > 26.5
for Ny > 4. Aziz et al. recommended the Duns and Ros mist-flow
method be used to calculate pressure gradient for this flow pattern.
---
Transition Region. The transition region exists when
N2 < Nx < N3 for Ny < 4. In Fig. 4.14. note that the transition region lOO
does not exist for Ny> 4. When the transition region is predicted. the o 5,000 10,000 15,O(!0
pressure gradients must be calculated with both the slug-flow and
mist-flow equations. To obtain the pressure gradient, linear inter-
polation is performed. similar to the procedure described in the
Duns and Ros method. Thus. Fig. 4.15-Chierici et al. 10 ~ correlation.
dp(dP)
dZ = A dZ + (dP)
(I - A) dZ .' (4.101) 2. Determine liquid holdup:
slug mist
From Eg. 4.98,
where
)(0.5)3(32.174)(47.61)(47.61 - 5.88)
N 3- Nx N; = -'------~(;-:-0.;-:-97::-:/~1.-4-88-:-)- - -
A = N _ N ' •••••••••••••••••••••••••••••• (4.102)
3 2
= 1.371 X 105
Modifications. AI-Najjar and AI-Soof19 showed that improved
results could be obtained with the Aziz et al. method if the flow-pat- :.m = 10.
tern map in Fig. 4.14 was replaced with the Duns and Ros map in
L Fig. 4.5. Their conclusion was based on a comparison of the pre-
dicted and measured pressure drops for 80 tests on 15 flowing wells
From Eg. 4.97,
--
I
ing/tubing annulus.
= 1.812 X 104 •
(32.174) and
+ [(47.61)(0.639) + (5.88)(1 - 0.639)] (32.174)
f = 0.00006 = 000012
d Q5 . .
= 0.92 + 32.55 = 33.47 psf/ft
L From Fig 2.2 or Eq. 2.17,J=0.OI58.
= 0.232 psi/ft.
4. Determine pressure gradient:
dp _ (0.0158)(47.61)(0.642)(7.83)2
Chierici et al, Method. The Chierici et al. IO method to predict dZ - (2)(32.174)(0.5)
flow pattern and to calculate liquid holdup and pressure gradient in
the bubble-, transition-, and mist-flow regions is identical to that of (32.174)
Orkiszewski. The only difference is in the slug-flow region. Thus, + [(47.61)(0.642) + (5.88)(1 - 0.642)] (32.174)
only slug-flow treatment is presented.
= 0.92 + 32.67 = 33.59 psf/ft
Slug Flow. The Chierici et al. slug-flow treatment is almost identi-
cal to that of Aziz et al. Liquid holdup is determined from
= 0.233 psi/ft.
(4.103)
where Beggs and Brill Method. The Beggs and Brill)) method was the
first one to predict flow behavior at all inclination angles, including
Vb = C,C 2 lid (4.104) directional wells. Their test facility was 1- and l.5-in. sections of
acrylic pipe, 90 ft long. The pipe could be inclined at any angle from
C J is obtained from Fig. 4.11 as a function of the bubble Reynolds the horizontal. The fluids were air and water. For each pipe size, liq-
number and is essentially 0.35 for all cases. Values of C2 can be ob- uid and gas rates were varied so that, when the pipe was horizontal,
tained from Fig. 4.15 for liquid Reynolds numbers less than 6,000. all flow patterns were observed. After a particular set of flow rates
For this range, Fig. 4.15 is identical to Fig. 4.12 used by Orkiszews- was established, the inclination of the pipe was varied through the
ki. Chierici et al. show in Fig. 4.15 that the extrapolation equations range of angles so that the effect of angle on holdup and pressure
proposed by Orkiszewski in Eqs. 4.67 through 4.70 can result in dis- gradient could be observed. Liquid holdup and pressure gradient
continuities. From earlier work by Nicklin et al.,20 Chierici et al. were measured at angles from the horizontal of 0°, ± 5°, ± 10°,
proposed that for liquid Reynolds numbers greater than 6,000, ± 15°, ±20°, ±35°, ±55°, ±75°, and ±90°. The correlations
were developed from 584 measured tests.
_ I Beggs and Brill proposed the following pressure-gradient equa-
C2 - (4.105)
I - 0.2 v:
V • •
tion for inclined pipe.
fPnv~
Curves based on Eq. 4.105 for extrapolation also are plotted in Fig. dp ----u- + Psg Sin. () (4.107)
4.15 and show no discontinuities. dL = I - Ek
Combining Eqs. 4.104 and 4.105 gives
where Ek is given by Eq. 4.53 and
Vb = o.z-, + 0.35 lid. (4.106)
The resulting prediction of liquid holdup essentially is identical to
Ps = PLHL(Ol + pAl - HL(Ol]' (4.108)
that of Aziz et al.
Flow-Pattern Prediction. Fig. 4.16 illustrates the horizontal-flow
The pressure-gradient friction component is identical to that of
patterns considered by Beggs and Brill. On the basis of observed
Aziz et al., and the acceleration component is neglected for slug flow. flow patterns for horizontal flow only, they prepared an empirical
map to predict flow pattern. Their original flow-pattern map has
been modified slightly to include a transition zone between the seg-
Example 4.6-Using the Chierici et al. Method, Calculate the regated- and intermittent-flow patterns.U Fig. 4.17 shows both the
Vertical, Multiphase-Flow Pressure Gradient for Example 3.2. original and the modified (dashed lines) flow-pattern maps.
Given: /10 = 0.97 cp, /1g = 0.0 16 cp, ao = 8.41 dynes/em, and Beggs and Brill chose to correlate flow-pattern transition bound-
E = 0.00006 ft. aries with no-slip liquid holdup and mixture Froude number, given by
1. Determine flow pattern:
(4.109)
Orkiszewski predicted slug flow.
--
i
i
Segregated
_------'c~
.. - '" , ,
II
a: _ _ Original map
•• -._._. Revised map
", ,
'Z,u-
,, , ,
"
Q; 10 I-Segregated flow regime
Stratified .c II-jntermiltent flow regime ,, , ,
§
Z
III-Distributed flow regime
IV-Transition flow regime
,' ,
~~
L2 "
Q)
'C
, ',,
:::l , IV'
e
u. '-, ,
' ,
,,
,,
, ,,
Wavy 01 ~----...J....._- __ ....L - - l_ _"--..L.-:.,l
6),L=0.164
c: 0.4 . +),L =0.156
Intermittent o o),L =0.010
~ 0.30
ci
:J
"0
(5 0.20
J:
"0
'5
cr
::J
-30 30 50 70 90
Angle of Pipe From Horizontal, degrees
Fig. 4.18-Effect of inclination angle on liquid hotdup."
or
Distributed
AL ~ 0.01 andNFr < t.,
Transition.
A. L ~ 0.01 and L, ~ N Fr ~ L 3
Intermittent.
or
AL ~ 0.4andL 3 < N Fr ~ L4
Distributed.
A. L < O.4andNFr ~ LI
Rg. 4.16-8eggs and 8rill 11 horizontal-flow patterns.
or
The equations for the modified flow-pattern transition boundaries are
HUll) = HL(O)'P· (4.115) to single-phase liquid flow. For s that is used when I < Y < 1.2,
Beggs and Brill introduced this correlation
The factor to correct liquid holdup for the effect of pipe inclination
is given by s = In(2.2y - 1.2). .. (4.125)
'P = 1.0 + c[sin(1.88) - 0.333sin 3(1.88)], (4.116) Modifications. Two modifications to the Beggs and Brill correla-
tion frequently are used. Payne et al. 22 experimented with a 2-in.-di-
where 8 is the actual angle of the pipe from horizontal and C is de- ameter flow loop constructed with Schedule 40 line pipe that was
fined by more than 500 ft long and contained three hills, each with inclination
angles of about ± 10° from horizontal. Carefully metered mixtures
L C = (1.0 - ).L)ln(eA{NLN~r)' (4.117)
where
+ (I - A)HL(IIJ bu ' (4.118)
ness. This change improved the pressure-drop predictions for the
Beggs and Brill method for rough pipes.
Payne et al. also found that the Beggs and Brill method overpre-
L 3 - N Fr dieted liquid holdup in both uphill and downhill flow. On the basis
A = L _ L (4.119)
3 2 of their limited data, Payne et al. recommended these constant
correction factors to improve liquid-holdup values.
Friction-Factor Prediction. The two-phase friction factor is cal-
If 8 > 0,
culated from
(4.121) However, the resulting liquid holdup for 8> 0° should not be less
than AL. The original Beggs and Brill method has been found to
and,un is obtained from Eq. 3.21. overpredict pressure drops in producing wells. Consequently, im-
The ratio of the two-phase friction factor to the normalizing fric- proved results should be obtained if the Payne et al. liquid-holdup
tion factor was correlated with the Beggs and Brill experimental correction factor is applied for wells.
L data, resulting in
III. = e', (4.122) Example 4.7-Using the Beggs and Brill Method, Calculate the
Vertical, Multiphase-Flow Pressure Gradient for Example 3.2.
L where
Iny
Given: ,uo=0.97 cp, ,ug=0.016 cp, ao=8.41 dynes/em, and
E: = 0.00006 ft.
L Y = ---L A
2'
[HUll)]
(4.124)
From Fig 4.17 for h = 0.507, the horizontal-flow pattern is inter-
mittent.
2. Determine liquid holdup:
Eq. 4.123 contains discontinuities for y values of about From Eq. 4.114,
2.63 x 10- 4 and 1.016. It is unlikely that the discontinuity at the 0.5351
smaller value of y would ever be encountered. However, it is neces- H = (0.845)(0.507) = 0574
sary that s = 0 for y = 1.0 to ensure that the correlation degenerates L(O) (3.81)00173 ..
s = (0.591)/[ -0.0523 + 3.182(0.591) The slug/(annular/mist) transition was identical for horizontal and
all upflow and downflow angles. However, liquid viscosity had a
significant effect on this transition. Increased liquid viscosity accel-
- 0.8725(0.591)2 + 0.01853(0.591)4] erates the transition from slug to annular/mist flow. Eq. 4.130 de-
scribes this transition.
= 0.3873.
N gv = 1Q( 1.401-2.694NL +O.521Nt129 ) . •••••••••••••• (4.130)
SIM
From Eq. 4.122,
!
where
'--
z = 0.321 - 0.017N gv - 4.267 sinO - 2.972N L
Mukherjee and Brill Method. The Mukherjee and Brill 12 method
was initiated in an attempt to overcome some of the limitations of
the Beggs and Brill method and to take advantage of new instrumen-
- 0.033(logNg l - 3.925 sin 20. . : (4.134)
tation developed to measure liquid holdup. Their test facility con- In downflow at higher liquid rates, bubble-/slug-flow transition oc-
sisted of an inverted U-shaped, 1.5-in. nominal ID steel pipe. The curs as gas velocities increase at a fixed liquid velocity. At lower liq-
uid rates, bubble-/stratified-flow transition occurs below inclination
closed end of the U-shaped pipe could be raised or lowered to permit
angles of - 30°. From 0° to - 30° at lower rates, however, bubble
flow at any angle from 0° to ± 90° from the horizontal. Each leg of
flow did not occur for the range of flow parameters considered.
the U was 56-ft long with 22-ft entrance lengths and 32-ft test sec- Fig. 4.19 gives a flow chart to predict flow patterns by use of
tions on both uphill and downhill sides. A transparent section in these flow-pattern transition equations. The subscripts BS. SM.
each leg allowed for flow-pattern observations and permitted the and STin Eqs. 4.128, 4.130, 4.131, and 4.133 and in Fig. 4.19 rep-
use of capacitance sensors to measure liquid holdup. Fluids were air resent the bubble/slug, slug/( annular/mist), and the stratified tran-
and kerosene or lube oil. Approximately 1.000 pressure-drop rneas- sitions, respectively.
I1
'--
Fig. 4.19-Flow chart to predict Mukherjee and Brill 12 flow-pat-
tern transitions.
Liquid-Holdup Prediction. The liquid-holdup data were corre- Fig. 4.2O-Control volume for stratified flow. 12
lated with an equation of the form
L where
PsVmVSg
or Eq. 2.17 for an appropriate relative roughness and a Reynolds
number given by Eq. 4.138. Then,
E, = - p - ' (4.137)
f=f,,(fR) (4.141)
The friction factor is obtained from a Moody diagram (Fig. 2.2)
or Eq. 2.17 for an appropriate relative roughness and a Reynolds Stratified Flow. It is likely that stratified flow will occur only in
number given by highly deviated or horizontal wells. Mukherjee and Brill chose a
separated-flow or two-fluid approach to develop a pressure-gradi-
Pnv".d
I'-- N Re =~. (4.138) ent-prediction method. Fig. 4.20 gives a control volume that defines
all variables for this approach.
where «, is obtained from Eq. 3.21. A steady-state momentum balance on the gas and liquid phases
yields Eqs. 4.142 and 4.143. respectively.
Annular Flow. The pressure gradient for annular flow is deter-
mined from Eq. 4.137 and
fPnv;, + P gsin8
dp 2d s
...................... (4.139)
dL dp
A LdL = - (TwLP L + TiW,) - P LALKsin8 (4.143)
Mukherjee and Brill developed an empirical expression for the fric-
I tion factor that depends on liquid holdup. Liquid holdup first is cal- Either equation can be used to calculate pressure gradient. Muk-
L culated from Eq, 4.135 and appropriate coefficients from Table 4.4. herjee and Brill chose Eq. 4.142. To eliminate the effect of interfa-
A ratio of holdup values then is calculated as cial forces in large-diameter pipes, they suggested the addition of
Eqs. 4.142 and 4.143 to obtain Eq. 4.144.
(4.140)
A~ = - (TwLP L + TwgPg) - (.oLA L + pgAg)gsin8.
and a friction-factor ratio,fR. is interpolated from Table 4.5. A no-
slip friction factor.f". is obtained from a Moody diagram (Fig. 2.2) (4.144)
dp Pg •
dL = - T WgA - pggsm(). . (4.145)
g Example 4.8-Using the Mukherjee and Brill Method, Calcu-
From simple geometrical considerations based on Fig. 4.20, it can late the Vertical, MuItiphase-Flow Pressure Gradient for Exam-
be shown that 0, AL, Ag , PL, and Pg are all related to hi. and d. The ple 3.2.
relations are Given: Ito =0.97 cp, Itg =0.016 cp. ao = 8.41 dynes/em, and
e = 0.ססOO6 ft.
o= 2COS- I ( I - 2j} .< (4.146) I. Determine flow pattern:
From Eg. 4.130,
N, = 10(1.401-2,694(0,01181+0.521(11.87)°3'9). (4130)
HL=(~L)=~(O-SinO)' (4.147)
g"/M •• ••••• • •
= 350.8.
P = PL + p~, (4.148)
Ng v = 11.54 < 350.8. Therefore. flow pattern is not annular/mist.
I and Angle> 0°. Therefore, from Eqs. 4.128 and 4.129,
I
'--
x = logt l 1.54) + 0.940 + 0.074 sin(900)
Pg = (I - t)p, (4.149)
- 0.855 sin 2(900) + 3.695(0.0118)
where 0 is in radians. Hydraulic diameters for the gas and liquid
phases also can be defined by
1.2648.
d hg = d[2Jr - (0 - sinOl] (4.150)
2Jr - 0 + 2sin N. = 10 12 648
Z L'B/S
and = 18.40.
j
I - d(o - sino)
d hL -
NLv = 11.87 < 18.4. Therefore, flow pattern is slug flow.
I.- . ......................... (4.151)
o + 2sin~ 2. Determine liquid holdup:
From Eq. 4.135 for uphill coefficients,
Govier and Aziz 23
suggested that the wall shear stresses can be
evaluated approximately by assuming single-phase flow to occur in
the cross section occupied by a given phase. With these assump-
- 0.3801 + (0,.1299) sin(900)
( - 0.1198 sin 2(900») +
+] }
0.4757
tions, these relations result. H L -- ex p (l1.54)
0.2887
{[ (2.343)(0.0118)2 (l1.87)
fLPLvI
T wL = ----:zg- (4.152)
= 0.560.
and
3. Determine friction factor:
fgpgVi
T wg = ----:zg-. .. (4.153) From Eq. 4.138,
where ji, and /g = Moody friction factors based on Reynolds num- N Re = 1,488(27.10)(7.83)(0.5)
(0.5)
= 315
"
X lOS
bers defined by
From Fig. 2.2 for f/d=0.00012,f= 0.0155.
PLvLd hL
N ReL = 7L (4.154) 4. Determine pressure gradient:
I
I From Eq. 4.136,
'--
pgvgd hg
N Reg = ~ (4.155) dp (0.0155)[(47.6)(0.56) + (5.88)(0.44)](7.83)2
dL = 2(32.174)(0.5)
L . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. (4.156)
+ [(47.6)(0.56) + (5.88)(0.44)](sin 90°) g~: ~~:~
and
V Sg
= 0.864 + 29.249 = 30.113 psf/ft
vg = I _ H ' •••.•.•••••••.•••••••.•..•..•• (4.157)
L
= 0.209 psi/ft.
L
Use the following steps to obtain the pressure gradient for strati-
fied flow.
I. Calculate HL on the basis of Eq. 4.135. 4.2.2 Mechanistic Models. Most mechanistic models that predict
2. Solve Eq. 4.147 iteratively foro. A value of 0.001 is a good ini- two-phase-flow behavior in pipes are for an isolated mechanism,
tial guess for O. AL also is calculated from Eq. 4.147. such as flow pattern, film thickness, or rise velocity of gas bubbles
3. Solve Eq. 4.146 for hL/d. Calculate dhg and dhL from Eqs. 4.150 in liquid columns. Although several methods predict flow behavior
and 4.15 I. for a single flow pattern, in this monograph we discuss only those
4. Knowing 0 and P, Pg and PL are calculated from Eqs. 4.148 that predict flow behavior for all flow patterns. Specifically, we de-
and 4.149. scribe the methods of Ansari et al. 24 and Hasan and Kabir. 25 •27
. ....
'u '3
e
000 00 tQl c
0 00 0 0 Do 0.1 Slug or Churn <
°OOO~O ~
...' ...
I
00 00
III
° 000 I 0
0 0
0°0°0
ORo0°0 0.01
00
0.01 0.1 1 10 100
BUBBLE
t t
SLUG
t
CHURN
t
ANNULAR
Superficial Gas Velocity, mls
L FLOW FLOW FLOW FLOW Fig. 4.22-Typical flow-pattern map for wellbores. 24
v
Vs
Because of their proprietary nature, details of the models are not in-
cluded in this monograph. At high liquid rates, turbulent forces break down large gas
bubbles into small ones, even at void fractions greater than 0.25.
Ansari et al. Method. Ansari et al. 24formulated a comprehensive This yields the Barnea32 transition to dispersed-bubble flow as
mechanistic model for upward, vertical two-phase flow. This study
was conducted as part of the Tulsa U. Fluid Flow Projects (TUFFP). 0.40 L ] Y'(P L)3/5( f ) 2/5 1.2
Although the pressure-gradient equations imply that the model can 2[ (.oL _ pg)g 0L 2d (VSL + VS g) =
be used for deviated wells, no attempt was made to account for the
effect of inclination angle on the mechanisms considered. Because
of the model's complexity and because it was developed with SI 005
v Sg
units, the development and examples also are presented in SI units. 0.725 + 4.15 ( + , (4.161)
vSg vSL )
Flow-Pattern Prediction. Taitel et al 3 D presented the basic work
on mechanistic modeling of flow-pattern transitions for upward wherefis obtained from the Moody diagram (Fig. 2.2) for a no-slip
two-phase flow. They identified four distinct flow patterns and then Reynolds number. This is shown as Transition A in Fig. 4.22.
formulated and evaluated the transition boundaries among them. At high gas velocities this transition is governed by the maximum
Fig. 4.21 shows the four flow patterns: bubble flow, slug flow, churn packing of bubbles to give coalescence. Scott and Kouba-" con-
flow, and annular flow. cluded that this occurs at a void fraction of0.76, giving the transition
Barnea et at. 31 modified these transitions to extend the applicabil- for no-slip dispersed-bubble flow as
ity of the Taitel et at. model to inclined flows. Barnea32 combined (4.162)
flow-pattern-prediction models applicable to different inclination-
angle ranges into one unified model. On the basis of these studies, This is shown as Transition C in Fig. 4.22.
flow patterns can be predicted by defining transition boundaries
among bubble, slug, and annular flows. Transition to Annular Flow. The transition criterion for annular
flow is based in part on the gas-phase velocity required to prevent
Bubble/Slug Transition. Taitel et al. gave the minimum diameter fall back of the entrained liquid droplets in the gas stream. This gives
I for which bubble flow occurs as the transition as
L
d m,n = 19.01[1/lL -t )aL]Yl
PLg
g
(4.158) (4.163)
For larger pipe sizes, the basic transition mechanism for bubbly and is shown as the left Transition D in Fig. 4.22.
to slug flow is coalescence of small gas bubbles into large Taylor Barnea 32 proposed additional transition criteria that considered
bubbles. IS Experimentally, this was found to occur at a void fraction the effects of film thickness. One effect is the bridging of the gas
eratively from
The friction component is
dP ) fTPPTPV}p
.................. (4.172) ( dL + 2d ' (4.183)
f
wherefTP is obtained from a Moody diagram (Fig. 2.2) for a Rey-
nolds number defined by
A good initial guess is Q. min = 0.25. In Fig. 4.22, the right Transition
D is a result of applying the two Barnea film thickness criteria. N - PTPvTpd
ReTP - f.lTP . (4.184)
Flow-Behavior Prediction. Ansari et at. 24 presented physical
models to predict flow behavior in each flow pattern. This resulted Because bubble flow is dominated by a relatively incompressible
in separate models for bubble flow, slug flow, and annular flow. liquid phase, there is no significant change in the density of the flow-
Similarly. the velocity of the gas bubbles in the liquid slug is
If..
vgLS -
_
r.z-, + 1.53
[gaL(PL -
PZ pg)] n..o.s
LLS'······ (4.191)
where the second term on the right side represents the bubble-rise
velocity, as defined in Eq. 4.177.
DEVELOPED
The velocity VLTB of the falling film around the Taylor bubble
TAYLOR can be correlated with the film thickness, ch by use of the Bretz"!
BUBBLE ---H~ expression,
FALLING where dL is the constant film thickness for fully developed slug flow.
FILM From geometrical considerations, VLTB can be expressed in terms of
the Taylor-bubble void fraction to give
o 0° 0 0
LIQUID ° 0 0 VLTB = 9.916[gd(1 - jHgTB)r (4.193)
SLUG ---II~ H LLS Ls
°4·o
v. 0+0 V
0
The liquid-slug void fraction can be obtained from the correlation
developed by Sylvester.f" who used data from Fernandes et al. 39
and Schmidt,42
\ o a
LLS 0 LS
0 .....- - . . . . 1 . . - - . . . . 1 . . -
I H LS =
VSg
."....,=-=---=~-:-=-- ..................... (4.194)
L g 0.425 + 2.65v m
Eqs. 4.185,4.186,4.188 through 4.191, 4.193, and 4.194 can be
solved iteratively to obtain the following eight unknowns that define
the fully developed slug-flow model: {3, HLTB' HgLS' VgTB' VLTB,
VgLS, vus. and VTB. Yo and Shoharrr-' showed that these eight equa-
tions can be combined algebraically to give
Fig. 4.23-Slug-flow sChematic.24
I........
I
MULTIPHASE-FLOW PRESSURE-GRADIENT PREDICfrON 49
L For fully developed slug flow, Ansari et al. 24 assumed that the ac-
celeration component of the pressure gradient can be neglected.
McQuillan and Whalley44 introduced the concept of developing
•I •
slug flow during their study of flow-pattem transitions. In develop-
• • • ing slug flow, the length of the cap at the top of the Taylor bubble
GAS CORE
.1·
I • •
is a significant portion of the total length of the Taylor bubble. The
film thickness varies throughout the Taylor-bubble region. Ansari
•
LIQUID FILM
· ··1
.• ·1-I.·. • et al. presented a detailed description of the complex equations to
predict flow behavior for developing slug flow. Bamea45 contends
that failure to consider the developing slug region can result in sig-
• • nificant pressure-drop-prediction errors. However, pipe in produc-
L ENTRAINED
LIQUID DROPLET •
••
• I ••
Vc
I •
• •
ing wells is so long that only minor errors should result if the entire
slug-flow region is treated as fully developed flow. Consequently,
I • • this monograph does not address the developing slug region .
vF • Annular-Flow Model. Wallis I 8 discussed the hydrodynamics of
annular flow and presented the classic correlations for entrainment
and interfacial friction as functions of film thickness. Hewitt and
Hall-Taylor4 6 gave a detailed analysis of the mechanisms involved
L 'tF
in annular flow. All subsequent models were based on this approach.
Fig. 4.24 shows a fully developed annular flow. Conservation of
momentum applied separately to the core and the film yields, for the
core and the film, respectively,
B and
i
I The core is considered a homogeneous mixture of gas and en-
'--
trained liquid droplets flowing at the same velocity. Thus,
where where iF is obtained from a Moody diagram (Fig. 2.2) for a Rey-
PLvr<!HF
N Re = -IJ--' (4.211)
Eq. 4.200 assumes that the liquid film around the Taylor bubble F /""L
does not contribute to the elevation component. Friction losses were where
assumed to occur only across the liquid slug and are neglected along
the Taylor bubble. Therefore, the friction component of the pressure vsL(l - FE)
gradient is ............... (4.212)
4Q(l - Q)
I
L dP ) =
( dL
/LSPLSv~(1 _ {3) (4.202)
and
2d '
f d HF = 4Q(I - Q)d. .. (4.213)
where/LS is obtained from a Moody diagram (Fig. 2.2) for a Rey- This gives
noIds number defined by
N = PLSvmd (4.203)
Re LS f.-lLS·
where the superficialliquid-friction-pressure gradient is given by The basic unknown in Eqs. 4.228 and 4.229 is the dimensionless
film thickness, O. The pressure gradients in the core and film must
(:tL = IS~~V~L, (4.216)
be the same. Animplicit equation for 0 can be obtained by equating
Eqs. 4.228 and 4.229, giving -
N = PLVSLd (4.217) _ (l - F
E) 2 IF (dP)
= 0
Re SL I·J.[· . . . . . . . . . . . .. (4.230)
64Q3(l - Q)3/sL dL SL .
For the shear stress at the interface,
To simplify this equation, Ansari et al. used the dimensionless ap-
I;pcv~ proach developed by Alves et al. 36 This approach defines the fol-
r, = - 8 - ............................... (4.218) lowing dimensionless groups in addition to using the previously de-
fined modified Lockhart and Martinelli parameters, XM and YM.
where
Vsc 2 _ (dp/dL)c - gPc sin B
¢c - (dp/dL)sc (4.231)
and
Ii = Isc Z • . (4.220) 2 (dp/dL)F - gpLsinlJ
¢F = (dp/dL)sL (4.232)
where Z = a correlating factor relating interfacial friction to the film
thickness. On the basis of the performance of the model, Ansari et
al. found that the Wallis expression for Z works well for thin films Eq. 4.230 then reduces to
or high entrainments, whereas the Whalley and Hewitr'? expression
works well for thick films or low entrainments. Thus,
L Z'
The superficial friction-pressure gradient in the core is given by
4Q(l - Q)[I - 4Q(I - Q)]2.5
fscPcv~c
2d ' (4.224)
2.52[4(1 - 2Q)]
4Q(I - Q)[I - 4Q(l - Q)]3.5
where fse is obtained from a Moody diagram (Fig. 2.2) for a Rey-
nolds number defined by
3X1[4(1 - 2Q)]
N Resc = P~:;, ...... ..... (4.225)
[4Q(I _ Q)]4 . (4.234)
L (4.226)
Then,
F(QJ
and . . . . . . . . . . . . . . . . . . . . . . . .. (4.235)
Qj+1 = Qj - F'(QJ
fisc = fiLA.LC + fi8(1 - A.Le)· (4.227)
The pressure gradient for annular flow can be calculated by sub- Once 0 is known, the dimensionless groups ¢F and ¢e can be obtained
stituting the above equations into Eqs. 4.204 and 4.205. Thus, from the following forms of Eqs. 4.228 and 4.229, respectively.
¢~ =
( ~) 5(~)
(l _Z2Q)5 • . .. .. . .. • . . .. . .. .. • • .. . . . •. (4.236)
= Z +pcgsinB (4.228)
C (l-2Q) sc
and
and
X
(I _~Q)5 - YM }
. . ..... (4.237)
4.227, respectively, as
Pc = (761.7)(0.36) + (94.1)(1 - 0.36) = 334.4 kg/m3
{ .Z 5 - YM[ I - (1 _ 2Q)2]2
(1-2Q) and
Because the pressure gradient in the film and core must be the same, (0.97)(0.36) + (0.016)(1 - 0.36)
the total pressure gradient can be obtained from either Eq. 4.231 or Psc = 1,000
4.232. Thus,
= 0.36 x 1O-3 Pa · s.
(~L = (~fL = t/>~(~ftc + gpc sin 9 (4.238) The superficial core velocity from Eq. 4.226 is
N = (334.4)(1.834)(0.1524) = 260000
Rese (0.36 x 10 -3) ,
Ansari et al. neglected the acceleration pressure gradient for an-
nular flow. Lopes and Dukler48 found that, except for a limited and relative roughness is
range of high liquid flow rates, the acceleration component resulting
from the exchange of liquid droplets between the core and the film
is negligible.
Then, the flow is turbulent.
i
\.-. From Fig. 2.2,
Example 4.9-Using the Ansari et al. Mechanistic Model, Cal-
fsc = 0.0152.
culate the Vertical, Multiphase-Flow Pressure Gradient for Ex-
ample 3.2. Similarly, from Eq. 4.217 the film Reynolds number is
Given: £=0.00006 ft=1.83xlO m, d=6.0 in.=0.1524 m,
VSL = 3.97 ft/sec = 1.208 m/s, VSg = 3.86 ft/sec= 1.173 mis, PL=47.61 N = (761.7)( 1.208)(0.1524) = 144 565
Rel.S (0.97 x 10 -3) ,
I Ibmlft3=761.7 kg/m3, Pg =5.88 Ibmlft3=94.1 kg/m3, 0L =8.41
i dyneslcm=8.4IXIO- 3 kg/s2, PL=0.97 cp=0.97XIO- 3 Pa-s, and the friction factor from Fig 2.2 for a relative roughness of
'--
Pg =0.016 cp=0.016 x 10- 3 Pa- s. 1.2008 x 10- 4 is
Flow-Pattern Determination. fSL = 0.01708.
1. Check dispersed-bubble transition boundary (Eq. 4.161): The superficial gas-core and liquid-film pressure gradients are de-
termined from Eqs. 4.224 and 4.216 as
(0.4)8.41 X 10- 3 '/' 761.7 3/5[ 1.49 x 10- 2 ]2/5
2 [ (761.7 - 94.1)9.81 ] (8.41 x 10 3) (2)(0.1524) dPI = (0.01708)(334.4)(1.8344) = 6313 Palm
2
dL sc 2(0.1524) .
+ 1.173)
1.2 _ (1.173
+ 4.15 1.208 + 1.173
)0.5 and
I x (1.208 - 0.725
2
L dPI = (0.0152)(761.7)(1.208) = 55.43 Palm.
x 1.144 ;I! 3.636. dL SL 2(0.1524)
Then, flow pattern is not dispersed bubble. Assuming thatfFIfsL = I, the modified Lockhart and Martinelli pa-
rameters can be estimated from Eqs. 4.167 and 4.168 as
2. Check annular-flow transition boundary:
From Eq. 4.163, 2 (55.43)
(1 - 0.547) (63.13) = 0.784,
L Vs
gAM
= 3.1[(9.8)(8.41 x 10- 3)(761.7 - 94.1)]
(94.1)2
,/,
xit = 0.615,
and
= 0.87 mls.
Although vSg = 1.173 > 0.87 gives the annular-flow pattern, also YM = (9.81)(76\:3~1~~4.4) = 66.34.
check the Barnea transition criteria.
From Eq. 4.209, Because FE=0.547, using Eq. 4.222 yields
]
~
nun [
d . = 19.01 (761.7 - 94.1~(8.4 x 10-
(76 I.7n9.8 I)
3
) 0.1658
H LTB = 0.15 - 8.4383 = 0.13.
]
O~
II = 0.151 mls.
I...-
From Eq. 4.159, the superficial liquid velocity at the transition is
VSL.t = 3[1.173 - (0.25)(0.151)(sin900)] = 3.4 mls
and
VSL.l = 3.4 m/s > VSL = \.208 mls.
+ 0.13 ]
The flow pattern is then slug flow. 4jO - 0.13)(1 - h - 0.13)
Slug-Flow Modeling.
The step-by-step procedure described in the discussion of the = 8.031,
! slug-flow model is followed to determine all slug-flow variables.
and
L \. Calculate VfB, HgLS, and VgLS from Eqs. 4.190, 4.191, and
0.00132
4.194 as H LTB = 0.13 - 8:03T = 0.1298.
0.5
The difference between the guessed and the calculated values of
VTB = (\.2)(2.381) + (0.35) (9.81)(0.15g~i~~)\.7 - 94.1)
[ ] HLTB is very small. Therefore, HLTB :5 0.13.
3. Solve Eq. 4.193 for VLTB:
= 3.258 mls. Note that HgTB = I - HLTB.
H gTB = I - 0.13 = 0.87
1.173
H gLS = 0.425 + (2.65)(2.881) = 0.1742, and
I
]
~
[
I
L-. and VLTB = 9.916 (9.81)(0.1524)( I - JO.87) = 3.144 mls.
VgLS = \.2(2.381) + (0.151)(0.826)°.5 = 2.994 mls. 4. Solve Eq. 4.188 for vus:
Note that HLLS = I - HgLS'
2. Using Eqs. 4.196 through 4.199, determine HLTB.
A good initial guess is HLTB = 0.15. H LLS = 1 - 0.174 = 0.826,
~LS = (0.97 x 10- 3)(0.826) + (0.016 x 10- 3 )(1 - 0.826) VTB = 0.35 gd
(PL - pg) f.li(1
PL vsm e
+ COSu,
ll)1.2
L ReLS
From Fig. 2.2, Dispersed-Bubble Transition. When large bubbles are dispersed
into small bubbles at high liquid rates, transition to slug flow is in-
hibited, even though the gas void fraction exceeds 0.25. For the tran-
eld = 1.2 x 1O- 4 }
N = 291 284 ...... fLS == 0.0166. sition to dispersed-bubble flow, Taitel et al. 30 proposed
ReL.S '
I 05
I
'-- From Eq. 4.202, v~12 = 4.68JO·48 [ g(PLa~ pg)] ~~f\~~t08
2
(0.0166)(645.4)(2.381) (I - 0.287)
L 2(0.1524)
. . . . . . . . . . . . . . . . . .. (4.243)
When the gas void fraction exceeds 0.52, bubble coalescence can-
not be prevented and transition to either slug, churn, or annular flow
142.0 Palm. must occur.
10. Total pressure gradient: Slug/Churn Transition. The Barnea and Brauner50,51 model pre-
dicts the slug/churn transition. They argued that the transition from
(:'L = 4,779.2 + 142.0
slug to churn flow occurs when the liquid slug trailing a Taylor
bubble attains the maximum possible void fraction of 0.52, as given
by Eq. 4.243. In other words, this transition boundary is a locus of
= 4,921.2 Palm constant mixture velocity values, Vm ' where the turbulent intensity
is maintained at the same level as that in dispersed-bubble flow.
= 4.921 kPalm
Annular-Flow Transition. The transition criterion given by Eq.
= 0.218 psilft. 4.163 also is used in the Hasan and Kabir mechanistic model to pre-
dict the transition to annular flow. Eq. 4.163 suggests that this transi-
tion criterion is independent of the liquid flow rate.
+ fv~p,
. 0 where Ycnt is defined by Eq. 4.209.
= p,gsm 2d + p,v m dv.,
dL '
(4244
. ) In annular flow, fine liquid droplets flow in the gas core with a ve-
I
'--
where
locity the same as the gas phase while a thin liquid film creeps up
the pipe wall. Thus, the friction pressure drop pertains to the gas in-
p, = pil - H L) + PLHL' ..•.•....•.•.••..... (4.245) teracting with the wavy liquid film. The friction pressure gradient,
which is a large component of the total pressure gradient for annular
In general, the acceleration component can be neglected during all but flow, can be written as
the annular-flow pattern. Eq. 4.245 suggests that an accurate estima-
tion of the liquid holdup is essential when computing the elevation
component. This component accounts for most of the pressure drop
P
( ddL ) f = fC P
2dC(1 -
VS~ ALC
)2, (4.251)
occurring in the bubble- and slug-flow patterns. Because of the differ-
ent hydrodynamics in each flow pattern, estimations of holdup, HL' where
in-situ mixture density, Pm' and friction factor.j, are made separately.
where Co and Vs are given by Eqs. 4.241 and 4.160, respectively. To (4.253)
estimate total pressure gradient, Eq. 4.244 can be used with the mix-
ture density calculated from the liquid holdup estimated with Eq. The liquid holdup in the core can be given by this simple equation
4.246. The friction component can be computed by treating the mul- from Wallis 18
tiphase mixture as a homogeneous fluid. Friction factor.j, can be de-
termined from the Moody diagram (Fig. 2.2) for a Reynolds number _ (
ALC - I + AvII.8)
-
- 0.378
, ....................... (4.254)
defined as
where X. the Lockhart and Martinelli 35 parameter, is defined in
N R Cm = P~;d. . (4.247) terms of the gas mass fraction, xg , and fluid properties as
In the case of annular flow, the gas void fraction is for the channel
Slug and Chum Flow. The drift-flux model of Eq. 4.246 also was
volume not occupied by the liquid film. Hence, the gas mass fraction
applied in slug flow, but with different values for Co and Vs given by
1 calculation should include the entrained liquid droplets.
'--
Co = 1.2 and
V, = 0.35 ( gd P L ;.pg )
0.5
/sinOO + cos 0)1.2. . . .. (4.248) Example 4.10-Using the Hasan and Kabir Mechanistic Model,
Calculate the Vertical, Multiphase-Flow Pressure Gradient for
It was recommended that the equations developed for slug flow also Example 3.2.
be used for the churn-flow pattern. However, because of the chaotic Given: E = 0.00006 ft= 1.83 x 10- 5 m, d=6.0 in. =0.1524 m,
flow nature that tends to make the gas concentration profile flat, it VSL = 3.97 ft/sec = 1.208 mis, vS g = 3.86 ft/sec = 1.173 mis,
is suggested that a value of 1.15 be used for Co rather than a value PL =47.61 Ibmlft 3 =761.7 kg/m 3. Pg =5.88 lbm/ft' =94.1 kg/m 3.
of 1.2 used for the slug-flow pattern. As in bubble flow, the total aL =8.41 dynes/em = 8.41 x 10- 3 kg/S2./-lL =0.97 cp=0.97 x 10- 3
pressure gradient can be obtained by Eq. 4.244 by use of Eqs. 4.246 Pa- S,/-lg =0.016 cp=0.016 x 10- 3 Pa- s.
I and 4.248. The estimation of the friction component presents some
I difficulty because some of the liquid flows downward in a film Flow-Pattern Determination.
'-- around the Taylor bubble while most of the liquid flows upward in I. Check bubble/slug transition (Eq, 4.240):
the liquid slug. Wallis I 8 suggested that the wall shear stress around
]
~
the vapor bubble be ignored. With this assumption, the friction pres-
sure gradient becomes
dP )
[
v, = 1.53 (9.81)(8.41 x 10- 3)(76 1.7 - 94.1)
(761.7)2
( dL f . . . . . . . . . . . . . . . . . . . . . . . . .. (4.249)
I = 0.151 m/s,
L The product P LHL is very nearly equal to Pm for low-pressure sys-
tems, indicating the similarity in evaluating the friction terms in slug Co = 1.2,
and bubble flow. The contribution of the friction component is no and
longer negligible but is still small (typically 10% of the total gradi-
ent). Acceleration, however, is small and can be neglected.
VSgs/
s
= (4 _11.2)[(1.2)(1.208) + 0.151) = 0.572.
Annular Flow. In annular flow, it is appropriate to estimate the liq-
I uid holdup in the central core rather than for the entire pipe cross Because vS g > vSgs/
s'
bubble flow does not exist.
L to., Of """,nr, _ r r r l r\~" r>nr("C", 'nr' ,-.n , T'T ..... ·..,... nnrr-,f"""'"T"lr-..,.
TABLE 4.6-RANGE OF DATA IN TUFFP WELL DATA BANK24
Nominal Diameter Oil Rate Gas Rate Oil Gravity
Source (in.) (STB/D) (Mscf/D) (OAPI)
Old TUFFP Databank" 1 to 8 o t010,150 1.5t010,567 9.5 to 70.5
Govier and Fogarasi 54 2 t04 8to 1,600 114 to 27,400 17 to 112
Asheim 6 27 / 8 to 6 720 to 27,000 740 to 55,700 35 to 86
Chierici et al.1O 2 7/ 8 to 5 0.3 to 69 6t027,914 8.3 to 46
(~~1 =
05
vL12 = (4.68)(0.1524)0.48 (9.81)(76\.7 - 94.1) 5,115 + 150.5
m [ 8.4 X 10- 3 ]
= 5, 265.? Palm
x (8.41 10_ 3
X )°."(0.9776\.7 )
, 76 \.7 x 10- 3
= 5.266 kPaim
or VmDB = 4.40 I > 2.381. Therefore, neither dispersed-bubble nor
L chum flow exist. The flow pattern is slug. = 0.233 psilft.
Slug-Flow Modeling.
For slug flow, Co = 1.2.
4.3 Evaluation of Wellbore
1. Calculate slip velocity (Eq. 4.248):
Pressure-Gradient-Prediction Methods
I The pressure drop in a well can be calculated with the computing al-
J
........ gorithm described in Sec. 3.7. This requires that the profile of the
well is known, that a pressure-gradient-prediction method is speci-
fied, that sufficient information is available to predict pressure/vol-
ume/temperature (PVT) relationships, and that adequate informa-
tion is available to calculate fluid temperatures as a function of
= 0.401 mis, depth. Each requirement is a potential source of error when calculat-
ing pressure drop.
2. Calculate holdup (Eq, 4.246) and slip density (Eq. 4.245): A common error in deviated wells is to not use accurate well pro-
files in the marching algorithm. Many times the profile is unknown
1.173 and vertical flow is incorrectly assumed.
HL =1- (1.2)(2.38) + 0.401 = 0.64 Black-oil-model PVT calculations, which include the prediction
I
I and of mass transfer and fluid physical properties of the gas and liquid
'-- phases, can be highly inaccurate under many real applications. For
Ps = (76\.7)(0.64) + (94.1)(0.36) = 52\.4 kg/m3• example, bubblepoint pressures calculated from different black-oil
models can have errors in excess of 50%. Depending on the depths
3. Elevation component of pressure gradient: of wells, this can lead to serious errors in pressure-drop calculations.
From Eq. 4.244, PVT property correlations often incorrectly predict the part of a well
in which single-phase flow occurs. This in tum can lead to very con-
(~L = (521.4)(9.81)sin90" = 5,115 Palm, tradictory results when evaluating pressure-gradient-prediction
methods with different PVT correlations.
When using empirical correlations or mechanistic models to pre-
4. Friction component of pressure gradient: dict pressure gradient, another common source of errors is the fre-
quent absence of reality checks, such as ensuring that the calculated
L From Eq. 4.247,
N
ReLS
= (761.7)(2.381)(0,1524) = 284
(0.97 x 10 3)
942
,.
liquid holdup always exceeds the no-slip holdup for upward flow.
Correlations that have been modified normally include reality
checks. Correlations that have not been modified, such as Duns and
Turbulent flow exists. Ros,? Mukherjee and Brill,12 and Aziz et al.9 may be vulnerable to
From Fig. 2.2, such programming inadequacies.
With this introduction, an evaluation of six commonly used cor-
eld = 1.2 x
N = 284 942
1O-4} ... f == 0.0166. relations and two mechanistic models was performed by Ansari et
al.24 ,53 The pressure-drop predictions from applying each correla-
I
'--
ReLS '
tion and model by use of the computing algorithm described in Sec.
3.7 were compared with measured data in a well databank devel-
From Eq. 4.249, oped at TUFFP. The databank contained 1,712 well cases with a
wide range of data, as shown in Table 4.6.
d P ) = (0.0166)(2.381)2(761.7)(0.64)
( dL f 2(0.1524) 4.3.1 Criteria for Comparison With Data. Evaluation of the cor-
relations and models with the databank was based on these statisti-
150.5 Palm. cal parameters.
where
E2 = (4.258)
+E
5.... -
E""n + E 6
5""n 6.... -
;m..
6....
............ (4.264)
The minimum and maximum possible values for Frp are 0 and 6, in-
E2 indicates an average of how large the errors are. dicating the best and worst performances, respectively. Table 4.7
gives an evaluation of the methods in terms of Frp with the best value
n for each column being boldfaced.
E) = I . . . . . . . . . . . . . . . . . . . . .. (4.259)
i=l 4.3.3 Overall Evaluation. An overall evaluation first was per-
E3 indicates the degree to which the errors are scattered about their formed by use of the entire databank. resulting in Col. I of Table 4.7.
average percent error. The performance also was checked for vertical weUcases only (Col.
2) and for deviated well cases only (Col. 3). To make the evaluation
I
'-- E4 = (k Ie;), (4.260)
unbiased with respect to the methods, a second database was created
that excluded 331 sets of data from the Hagedorn and Brown study.
t= I For this reduced databank, Col. 4 shows the results for all vertical
well cases. Col. 5 shows the results for combined vertical and de-
where viated weU cases.
e, = LlPicalc - LlP;meas' •.••••••••••••••••••••• (4.261) 4.3.4 Evaluation of Individual Flow-Pattern Models. Perfor-
E4 indicates the overall trend independent of the measured pressure mance of individual flow-pattern models is based on sets of data that
j are dominant in one particular flow pattern, as predicted by the tran-
\ drop.
'-- sitions described by Ansari et al. 24 To have an adequate number of
cases for the bubble-flow-pattern database, cases with bubble flow
E = (k ~le;I).
5 (4.262) predicted over 75% of the well length were considered. These re-
sults are shown in Col. 6. Cols. 7 through 10 give results for cases
predicted to have slug flow over 100% of the well length. The cases
E5 also is independent of the measured pressure drop and indicates used for Cols. 7 and 8 were selected from the entire databank,
the magnitude of the average error. whereas the cases used for Cols. 9 and 10 were selected from the re-
duced databank. Col. II gives results for those cases in the total da-
n tabank that were predicted to be in annular flow for 100% of the well
E6 = I
;=1
. . . . . . . . . . . . . . . . . . . . . .. (4.263) length. The performance of the Ansari et al. model for annular flow
is significantly better than all other methods.
I
'--
E6 indicates the scattering of the errors about their average error. Ansari et al. concluded that the overall performance of their mod-
el was better than all other methods. However, the overall perfor-
mances of the Hagedorn and Brown, Aziz et al., Duns and Ros, and
4.3.2 Comparison Method. The correlations and models used for
Hasan and Kabir methods are comparable. For the latter three meth-
the comparison were a modified Hagedorn and Brown," Duns and
ods, this can be attributed to the use of flow mechanisms. The excel-
Ros,7 Orkiszewski" with the Triggia* correction, Beggs and Brill 11 lent performance of the Hagedorn and Brown correlation can be ex-
with the Payne et al. 22 correction, Mukherjee and Brill,12 Aziz et plained only by the extensive data that were used in its development
al.,9 Ansari et al.,24 and Hasan and Kabir. 25.26 The evaluation was
accomplished by a comparison of the statistical parameters. The 'Pecsonal communcation wilh A. Tnggia. Petrob,,;s SA. Rio de.Janeiro (1984).
i
1
\--
and modifications that have been made to the correlation. In fact, multaneously flow upward in an annulus. Each is described in fol-
when the data without Hagedorn and Brown are considered. the An- lowing sections.
sari et al. model performed best (Cols. 4 and 5). Although the Hage-
dorn and Brown correlation performed better than the others for de- 4.4.1 Caetano et al. Method. Caetano et al. 66 .67 conducted a com-
viated wells. none gave satisfactory results (Col. 3). bined experimental and theoretical study of upward, gas/liquid,
Only 29 cases satisfied the bubble-flow criteria. The Hasan and two-phase flow in vertical concentric and fully eccentric annuli. The
Kabir model gave the best predictions for these cases. When the experimental part of the study involved flowing air with either water
Hagedorn and Brown data are included (Cols. 7 and 8), their meth- or kerosene up a vertical annulus. The experimental study included
od exceeds the performance of the Ansari et al. slug-flow model. the establishment of flow-pattern definitions, determination of
The Ansari et al. model performed best when Hagedorn and flow-pattern-transition boundaries, development of flow-pattern
I Brown data are excluded for both all well cases and all vertical well maps, and measurements of average volumetric liquid holdup and
! cases (Cols. 9 and 10). average pressure gradient for several tests in each flow pattern. The
L....
For annular flow, the Ansari et al. model is significantly better theoretical part included development of flow-pattern-prediction
than all other methods (Col. II). models and the formulation of models for each flow pattern to pre-
Pucknell et al. 59 compared predicted pressure drops with mea- dict average volumetric liquid holdup and pressure gradient. Caeta-
sured pressure drops for 246 data sets collected from eight fields, in- no et al. used Fanning friction factors throughout the study. They are
cluding a gas and gas-condensate field. None of these data was used preserved here to retain consistency with the original work.
in the development of the pressure-gradient-prediction methods.
They reached four specific conclusions. Flow-Pattern-Transition Prediction. The models proposed to
I. Despite the development of new mechanistic models, no single predict flow-pattern transitions in an annulus are similar to those
method gives accurate predictions of bottomhole flowing pressures proposed by Ansariet al. 24 for circular pipes.
in all fields.
Bubble-/Slug-Flow Transition. The minimum equiperiphery di-
2. Traditional methods to predict pressure drops, such as that of
ameter for which bubble flow occurs is
Duns and Ros, give good results in oil wells, but can give very poor
results in gas wells. The new mechanistic models give reasonable
results in both oil and gas wells. d EP = 19.7 . (4.265)
3. Overall, the Ansari et al. mechanistic model gives the best re-
sults of all the methods evaluated. In oil wells, 62% of the pressure
drops were predicted with errors of less than ± 6%. In gas wells, where dEP is the equiperiphery diameter defined by
68% of the pressure drops were calculated within ± 15%. These
conclusions were derived from deviated wells with larger tubing
d EP = de + d, (4.266)
sizes, typical of North Sea wells. and de and d, are the inside casing diameter and the outside tubing
4. The Ansari et al. and Hasan and Kabir models behave no better diameter, respectively.
than many of the traditional methods in predicting erroneous disconti- For equiperiphery diameters larger than this, transition to slug
nuities. Reliable convergence is desirable in any multiphase-flow flow was found experimentally to occur at average gas void-fraction
model. The Ansari et al. model performed only as well as existing values of 0.20 for flow through a concentric annulus and 0.15 for
methods, but the Hasan and Kabir model performed somewhat better. flow through a fully eccentric annulus. Thus, the Taitel et al. 30mod-
Salim and Stanislavs" compared methods that describe the flow el was modified for the bubble/slug transition in an annulus by use
of gas/liquid mixture in wells with more than 189 data sets collected of values of the gas void fraction measured at this transition. The re-
from five different sources. Most of their field data were taken from sulting equations for the transition then can be expressed by
wells that exhibited the annular/mist-flow pattern. They concluded
I
L
drag forces acting on the largest stable droplet. The transition is giv-
en by Eq. 4.163.
NReTP=~
The mixture viscosity, JiTP, is
.......................... (4.280)
Flow-Behavior Prediction. Mechanistic models were developed JiTP = JiLAL + Jig(I - AJ, (4.281)
to predict flow behavior for all flow patterns in annuli configura-
tions. This includes bubble, dispersed bubble, slug, and annular where the no-slip liquid holdup, AL, is given by
flow. Each model is presented. A - VSL
L - VSL + vsg ' (4.282)
Bubble Flow. The basic concept for modeling the bubble-flow
pattern is the slippage that takes place between the gas and the liquid Combining Eqs. 4.274, 4.275, 4.277, and 4.278, the total pressure
phases, given by gradient for bubble flow is
! dP) v~
= VSL + 41'
vSG
Vs = vg - VL H -
L
I _ H ' ••.••••••••••••• (4.272) ~ dL = PTPg d;PTPz· •••••••.•.•..•••..• (4.283)
L
Eq. 4.177 gave the rise velocity of a single gas bubble in a bubble-
Dispersed-Bubble Flow. The model for dispersed-bubble flow is
swarm medium. The bubble-swarm index, n' = 0.5, determined ex-
based on the no-slip, homogeneous nature of this flow pattern.
perimentally and used in the model, agrees with the value used by
Therefore, Eq. 4.283 also can be used for dispersed-bubble flow by
Ansari et al. 24
replacing liquid holdup in Eq. 4.276 with no-slip liquid holdup, AL,
Combining Eqs. 4.272 and 4.177 and rearranging yields
as given by Eq. 4.282.
Slug Flow. Caetano et al. developed a hydrodynamic model for
slug flow in annuli that considers two possible configurations. The
first configuration is for fully developed Taylor-bubble flow and oc-
curs when the bubble cap length is negligible compared with the to-
tal liquid film length. Under this condition, the film thickness reach-
es a constant terminal value, which can be used as the average
VS
L
'I, = 0, (4.273) thickness for the entire film zone. The second configuration is for
A1 0
oo
0 0 Ta,o
00
'A (VSL + vsg)A = vusHusA + vgdl - «,»: ... (4.287)
o -- where the flow area for an annulus is
I
I
=4
v.
Ls u
B' 00.
Film
Vus = (VSL + VSg) - 1.53
(P L-Pi.pg)gaL] (H LL)
'I>( )
1 - H us .
[
Liquid
00
o 0
° HLLS
0
° D 000
0
0 0
0 ................... (4.289)
Slug ~ 00 .~O LLS
- ~ 0 VLLS f Overall Mass Balances. The flow of gas and liquid within a slug
unit was assumed previously to be incompressible. Thus, mass and
volume balances are equivalent. An overall volume balance on the
liquid phase yields
LLS LLF
VSL = vusHus L - vLTBH LTBL (4.290)
su su
where the slug-unit length, !.;su, is
+g
VS L su = LLF + LLS' .......................•... (4.291)
YSL =superficial liquid velocity Combining Eqs. 4.290 and 4.291 yields
vs g = superficial gas velocity LLS vSL + vLTBHLTB
YLLS = in-situ liquid velocity in liquid slug L = vusH + vLTBH (4.292)
VgLS = in-situ gas velocity in liquid slug
su us LTB'
VLTB = in-situ liquid velocity in liquid film A similar procedure can be applied for the gas phase. Because gas
VgTB = in-situ gas velocity in Taylor bubble in the liquid slug and in the Taylor bubble are both flowing upward,
vTB = Taylor-bubble transitional velocity this yields
HLLS = liquid holdup in liquid slug
LLS = length of liquid slug ........ (4.293)
LLF = length of liquid film
Lsu = length of slug unit
Fig. 4.25-Fully developed slug flow. 67
Liquid-Film Zone. The liquid film flowing downward and sur-
rounding the Taylor bubble can be considered as a free-falling film.
The free-falling film is assumed to have a thickness and flow-rate
Mass Balance in Slug and Film Zone. A mass balance on the liq- relationship at the bottom of the Taylor bubble identical to that for
uid phase between Planes A-A and B-B of Fig. 4.25 yields a falling film on the surface of a vertical plane or cylinder. This situa-
tion is expected to be valid when the entry region used to develop
(VTB - vus)Hus = (VTB + vLTB)H LTB, (4.285) the velocity profile is less than the length of the film. The relation-
ship between the film thickness and the film velocity for this case
where VLTB, the in-situ liquid velocity in the film surrounding the is given by
Taylor bubble, is considered positive for downward flow. HLTB is
the liquid holdup at a cross section that contains the terminal film l!..!:...
thickness. 4PL' (4.294)
Similarly, a mass balance for the gas phase yields
I
(~L = PLSg(~~), (4.298)
L where the slip density for the gas/liquid mixture in the liquid slug is Wall Welled Perimeter
Casing Film, s,
PLS = PLHLLS + pg(1 - H LLS)· (4.299)
Interlace G8SlLlquld
The acceleration pressure-gradient component is related to the Perimeter Ca.lng Film, Sci
amount of energy required to accelerate the liquid film, which is ini-
tially flowing downward, to the existing upward in-situ liquid ve-
locity in the liquid slug.
Section A-A'
Interlace GasILlquld
dP) HLTB
PLy-(VLTB + VTB) (VLTB + vLLS) . . . . .. (4.300) Perimeter Tubing Film, Sri
( dL =
Wall Welled Perimete
ace SU
L For a fully developed Taylor bubble, HLTO and VLTO are the average
liquid holdup and film velocity in the entire film zone, respectively.
Tubing Film, Sr
. . . . . . . . . . . . . . . . . .. (4.302) where (dpldL)c is the total pressure gradient for the mixture in the
core, Pc is the density of the mixture in the core, andAc is the area
Annular Flow. Fig. 4.26 gives a schematic description of annular of the core occupied by the mixture.
flow in a concentric annulus. The model is based on equilibrium ful- The gas/liquid interfaces are considered stable; hence, this equi-
ly developed flow. The phases are assumed to be incompressible. librium condition exists.
The two liquid films are assumed to have uniform thickness, but
with different values. The gas and liquid droplets flowing in the an-
nulus core are assumed to flow as a homogeneous mixture with the
dP ) CJ.' = (dLdP)tj
(dL = (ddPL)c . , (4.306)
same velocity.
With this equilibrium condition and the linear momentum equations
Linear Momentum Equations. The conservation of linear mo- given by Eqs. 4.303 through 4.305, two combined-momentum
I~
mentum for the outer (casing) liquid film yields equations can be written.
dP)
( dL cf + T CAs, - TO
SCi
A cf + PIJ5 = O. ........... (4.303) - TcA-
s, + T SCI + T SCi
OAIf CiAC
cf cf
In this equation, (dpldLkf is the total pressure gradient for the casing
liquid film. Tc and TCi are the shear stresses at the casing waIl and liq- Sri t: ) 0 (4.307)
uid film core-mixture interface, respectively. Sc and SCi are the wetted + TriAc - If'L - Pcg =
, V~c
t i > I PC2 '
j
............................... (4.321)
s, (p L-PCg=O.
+T'iA- ) (4.308) where Ti = interfacial shear stress, Ve = in-situ mixture velocity in
c the core, Pc = mixture density in the core, and f' = Fanning friction
factor associated with the interface. The mixture density and veloc-
On the basis of the annulus geometry given in Fig. 4.26 and because
ity were defined as
the films are considered uniform in thickness, geometrical relation-
I ships for the relevant dimensions can be written. The casing-liquid- Pc = PLHLC + pAl - HLcl (4.322)
L film area, the tubing-liquid-film area, and the core area for the mix-
ture are, respectively,
and
A . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. (4.323)
(4.309) Vc = vscA
c'
where HLC and Vse are given by Eq. 4.332 and Eq. 4.338, respective-
Aif = m5,(d, + 15,), (4.310) ly; Ac is given in Eq. 4.311; and A = annulus cross section. For flow
in an annulus, Caetano et al. modified the Wallis interfacial-friction-
I and
factor correlation for thin films or high entrainment fractions in an-
L Ac = %[d~ - d; - 4<5M, - dJ - 4<5,(d, + 15,)], nular vertical flow, given by Eqs. 4.220 and 4.221. The resulting
casing-film-interface and tubing-film-interface friction factors are,
.................... (4.311) respectively,
where de and dt = casing-liquid-film thickness and tubing-liquid- 1;' = fs~ (I + 300 dd":) (4.324)
film thickness, respectively.
The perimeters associated with the casing-liquid film and tubing- and
liquid film, for both the wall and interface sides, are
S, = xd., (4.312) I, j
= Jsc
1" (I + 300Q;)
d, ' . . . . . . . . . . . . . . . . . . . . . .. (4.325)
\
where f;c = friction factor obtained with Eq. 4.320 for a Reynolds
L SCi = nid; - 2dc>, (4.313)
number based on the superficial mixture velocity, mixture density,
and hydraulic diameter in the core, given in Eq. 4.318.
S, = xd, (4.314)
With the shear stress and geometrical relationships defined in
and Eqs. 4.309 through 4.325, the combined linear momentum equation
for the casing-liquid film and the mixture in the core, given by Eq.
S,j = ntd, + 215,). . (4.315) 4.307, can be rewritten as
L
(4.318)
T = fp V ••••••••••••••••••..••••••••••••. (4.319)
22,
where T is the respective wall shear stress, P is the density for the - (PL - pclg = O. .. (4.326)
phase which wets the wall, v is the in-situ average phase velocity, Similarly, the combined linear momentum equation for the tubing-
andf is the associated Fanning friction factor, which is evaluated liquid film and the mixture in the core, as given by Eq. 4.308, can
f = C~e' (4.320)
be rewritten as
62
Eqs. 4.208 and 4.209 gave the Wallis 18 correlation, which predicted
the liquid fraction entrained in the gas core of long pipes under annu-
lar two-phase flow, FE.
Tubing/Casing-Liquid-Film-Thickness Ratio. A prediction for
the ratio between the thicknesses of the tubing film and the casing
film was developed by assuming equilibrium between the entrain-
ment and deposition rates of liquid droplets. At equilibrium condi-
L x 4(d,. - 2~c)
tions, the liquid droplets deposition and entrainment rates are equal
for each liquid film, resulting in this relationship between the tub-
ing- and casing-liquid-film thicknesses.
[d~ - d; - ~c(d,. - ~c> - ~r(d, + ~,)]
i T = Or = <w;>
I (4.334)
~e (27l - <W; »K'
--- - (PL - pclg = o. . (4.327)
where <W;> is the cross-sectional average of the angle of view
Phase Continuity Equations. The continuity for the liquid phase, associated with the tubing deposition area as seen by the entrained
which flows in the form of two films wetting the bounding walls and droplets. <W; > can be expressed in terms of geometrical variables by
as droplets entrained in the gas core, yields Eqs. 4.335 and 4.336 for concentric and eccentric annuli, respectively.
dd2
I = (4.328)
f sin-I(~r)rdr
V SL vrfHLJf+ vefHf.<f+ vsLF E, •.•••••...•.•••
L where Huj = in-situ liquid holdup in the tubing film, HLcj = in-situ
<W;> = ( ,16
de - d;
)
zr
d,12
liquid holdup in the casing film, and FE = fraction of liquid en-
trained in the gas core.
The continuity for the gas phase, which flows in the core bounded = 0 ~ [(2) [ 2 sin -I(K) + 2K h - K2 - K2n]
by the two liquid films, yields
· , ' (4.335)
V Sg = Vg~ I - H L . , ) , (4.329)
and
where vgc is the in-situ gas velocity in the core and HLJ is the total
W I
in-situ liquid holdup in the annulus. < t > = nO - [(2)
I
L The liquid holdup for the casing-liquid film is defined as the ratio
between the area for this film and the total annulus area. Using Eqs. :r
L ~ (I0 _+~)
and e is the degree of annulus eccentricity. Note that when e = 0
(concentric annulus), a reduces to 0.5. Consequently, the integrand
HLrf = decK [(2)' (4.331) expression in Eq. 4,336 becomes independent of the integration
variable, 0, and the integral result is the one given by Eq. 4.335.
The in-situ liquid holdup in the core, where the gaslliquid mixture However, if e;" 0, no analytical solution is found for Eq. 4.336. The
integration can be carried out numerically by applying a technique
is assumed to be homogeneous, is given by
like Simpson's rule.
I VSLFE
Fig. 4.27 shows the results found for the tubing/casing-liquid-
L H LC = F + (4.332) film-thickness ratio, as given by Eq. 4.334, in terms of the annulus
V SL E v Sg
pipe-diameter ratio and degree of eccentricity. During the numerical
The total liquid holdup in the annulus is the sum of the liquid hold- determination of the average angle of view for an eccentric annulus,
singularities were found when the pipe-diameter ratio equaled zero.
ups in the core and in the two films. Using Eqs. 4.288, 4.309 through
For a fully eccentric annulus, these singularities also appeared for
4.311, and 4.330 through 4.332 yields extremely low values of pipe-diameter ratio. Consequently. the
curves in Fig. 4.27 should not be used for liquid-film ratio deter-
mination for an eccentric annulus in these singularity regions.
Summary of Governing Equations: Dimensionless Form. The
governing equations developed previously for the annular flow
model can be presented in dimensionless form. The normalizing
variable selected for length is the hydraulic diameter for the annulus
configuration, given by Eq. 4.270.
The normalizing variables selected for velocity are the superficial
+ deO ~ [(2)[~e( 1- ~:) +drK( I +~) l liquid velocity for the in-situ velocities in the liquid films and the
core-mixture velocity for the in-situ velocity in this region. This
core-mixture velocity is given by
. . . . . . . . . . . . . . . . .. (4.333) (4.338)
i 63
'-- MULTIPHASE-FLOW PRESSURE-GRADIENT PREDICfION
I
-- 1.0
Also, the dimensionless groups, G!. G2, G3, and G4 are given by
(4.343)
<IOu
--
<lO-
II
0.9
GZ = I +
4 K - -( I - 0
-e
o[
(l - K)]
~
0 0.8
:;:;
III
a:
- !:LTK( I + QJ I 7< K), (4.344)
III
III
CII 0.7 Eccentricity
c:
G =I-+-K- - lQe[
.ll:
U 0.0 -o(l-K)]
'--
:2 3 4K K -e
~
u:::
0.6
................. (4.345)
,;
'5
C'" 0.l5 and
:::i
(4.346)
0.4
0 1.00
where
0,
T = 0' . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. (4.349)
c
Yet
Y f
-('
= -Y ,
SL
••••••••••••••••••••••••••••••••• (4.350)
Yif
'!!.if = YSL' ••.••....•....••••••••..•.•..••... (4.351)
(4.352)
x (1 + 2QJl ~K) J 3
- 4YM } (4.339)
y(Z-m) =
TQe[1 + QJ(l - K)/K] {(
1
- K)
+ 3000-eT1 - - d.if = 4TQe[ I (l - K)] .
+ TQe--K- ................ (4.354)
-if )(l d m K
M-if
In Eqs. 4.341 and 4.342. (dp/dL)sL and (dp/dL)sc = superficial pres-
sure gradients of the liquid phase and the core mixture. respectively.
Caetano et ai.67 provide additional details on the nondimensionaliz-
ing process.
For a given set of phase flow rates, fluid physical property values,
and configuration geometry, a solution is obtained for Eqs. 4.339
L
be used in bubbly and dispersed-bubble flow. They concluded that
the terminal rise velocity, vs, was not affected significantly by either
the inner pipe or the pipe-deviation angle from the vertical. The in-
fluence of pipe diameter becomes significant only when the diame-
ter of the bubble exceeds 20% of the channel diameter. Hasan and
Kabir 26 demonstrated that a pipe deviation of up to 32° from the ver-
tical did not seem to affect the rise velocity of small bubbles. This Fig. 4.28-Schematlc of slug unit for slug flow In an annulus.•
also was assumed to be the case for flow in annuli. (Reprinted with permission from Elsevier Science.)
Slug Flow. Application of the drift-flux model for slug flow is Fig. 4.28 shows a slug unit of length Lsu, consisting of a Taylor
more complicated than for bubble flow because of the different drift bubble of length LTB and a liquid slug of length LLS. The average
velocity of the small bubbles in the liquid slug compared with that holdup for the slug unit can be expressed as
of the Taylor bubbles. Assuming that the liquid slugs do not contain
any gas bubbles, Hasan and Kabir 25 .26 arrived at this expression for L LLS
H L = H LTB - TB + H LLS -L . .. (4.357)
liquid holdup using V'JB for the rise velocity of a Taylor bubble in an L su su
annulus. Hasan and Kabir68 used a simple approach to estimate the Iiquid-
VS g slug void fraction. The circular channel data of Akagawa and Saka-
H LTB = 1 - C + , (4.355) guchi 70 showed that the average volume fraction of gas in a liquid
IVm VTB
slug. based on the total slug-unit volume,-i.e., (I-HLLS)
where C\ is the flow parameter analogous to Co in bubble flow. (LLS/4;u )-is approximately equal to 0.1 when vSg > 0.4 mls and is
The approach used for bubble flow to determine the effect of an- equal to 0.25 vSg for lower superficial gas velocities. Assuming that
nulus dimension and pipe-inclination angle on the flow parameter this approximation also applies to flow in an annulus, Eq, 4.357 can
can be used for slug flow only ifEq. 4.355 alone is used to estimate be rewritten as
the gas void fraction. In Eq. 4.355, vsgl(l- HLTB) varies linearly L TB
!L- with Vm , with a slope of C I. Although imprecise, this approach indi- (I - H L ) = (l - HLTB)L + 0.1 (4.358)
su
cates that the flow parameter is not influenced significantly by either
inner pipe diameter or pipe-inclination angle. Thus, a constant value if vSg > 0.4 mls and as
of 1.2 was proposed for Cj.
L TB
The presence of an inner pipe tends to make the nose of the Taylor (l - HJ = (l - HLTB)L + 0.25vsg (4.359)
bubble sharper. causing an increase in the rise velocity, V'JB. Hasan su
and Kabir6 8 stated that their Taylor-bubble-rise data showed a linear if vSg < 0.4 mls.
relationship with the diameter ratio, dIllie, suggesting this expres- Chum Flow. Hasan and Kabir did not investigate the churn-flow
L sion for the Taylor-bubble-rise velocity for vertical annuli pattern because of its chaotic nature. However. they suggested that the
analyses presented for bubble and slug flow can be applicable for the
churn-flow pattern. Although the bubble shape is quite different from
VTB = (0.345 + 0.1 ::) the classic Taylor bubble. the bubble-rise velocity during churn flow
is probably not much different from that for slug flow. In addition, be-
cause the mixture velocity is much higher than the bubble-rise veloc-
ity during churn flow, a slight error in estimating bubble-rise velocity
x /sin8(1 + COS8)12 . . . . .. (4.356)
does not affect void-fraction estimation significantly.
I
I
I
1
'-- TABLE 4.8-EVALUATION OF CAETANO et al. LIQUID-HOLDUP AND
PRESSURE-GRADIENT PREDICTIONS FOR TWO-PHASE FLOW IN
CONCENTRIC AND FULLY ECCENTRIC ANNULl37
E1 E3
AirlWater Air/Kerosene AirlWater Air/Kerosene
Bubble
Concentric HL -0.46 -2.17 4.75 5.21
Eccentric HL -2.77 4.32
Concentric cp/dL 7.33 -0.61 8.69 3.44
Eccentric cp/dL 12.03 11.69
Dispersed bubble
Concentric HL 1.67 -3.84 2.84 2.93
Eccentric HL 1.12 1.11
Concentric cp/dz. -2.46 -2.89 5.46 1.39
Eccentric q,/dL -3.74 2.43
Slug
Concentric HL 3.94 5.80 7.72 9.97
Eccentric HL 1.75 4.03
Concentric q,/dL 3.51 8.00 3.32 4.07
'- Eccentric cp/dz. -1.06 6.86
Annular
Concentric HL 10.60 -2.31 8.86 23.17
Eccentric HL 15.00 9.71
Concentric cp/dL 17.11 66.00 15.94 25.22
Eccentric cp/dL -14.03 9.33
However, an accurate estimation of C, is very important in the flow was sometimes unstable because of a heading phenomenon.
prediction of void fraction. The bubble concentration profile in which can cause a lower degree of measurement accuracy.
chum flow may be dissimilar to that for slug flow because of the For dispersed-bubble flow, Table 4.8 shows that the model's pre-
characteristic churning motion of this flow pattern. Using Eq,4.355, diction of liquid holdup is very good, with a slight overprediction
with V'rB given by Eq. 4.356 for circular pipes, Hasan 71,72 analyzed tendency for air/water flow and a slight underprediction tendency
the void-fraction data gathered by Ney 73 and Fuentes 74 for circular for air/kerosene flow. The standard deviation about the mean is be-
pipes in the chum-flow pattern. Hasan concluded that a value of low 3.0%. The model performs equally well for either fluid pair or
1.15 was appropriate for the parameter C r- The same value was pro- annulus configuration. Although lower values of pressure gradients
posed for use in annuli. were encountered in the fully eccentric annulus because of the lower
friction losses, the agreement between the model and experimental
4.5 Evaluation of Annulus Liquid-Holdup and results is again very good, showing low values of both average per-
cent error and standard deviation.
Pressure-Gradient-Prediction Methods
For slug flow, Table 4.8 shows that the model's prediction of liq-
Published liquid-holdup and pressure-drop data for two-phase flow uid holdup is good, with a slight overprediction for both air/kero-
through annular geometries are scarce. Caetano-? gathered liquid- sene and air/water mixtures. The standard deviations ranged from
holdup, pressure-drop, and flow-pattern data for air/water and air/ 4 to 10%, confirming reasonable scatter about the mean. The model
kerosene two-phase flow through an annulus with 3-in.-ID outer performed slightly better for the air/water mixture. The agreement
pipe and 1.66-in.-OD inner pipe at a pressure of about 45 psi. Caeta- between the model and experimental results for pressure gradient is
no et al. 67 and Hasan and Kabir6 8 used these data to evaluate their also good. As in the liquid-holdup prediction, the model performed
respective models. slightly better for the air/water mixture. No significant differences
were found in model performance for the two annuli, suggesting
4.5.1 CaetanoetaL Method. The model's performance was obtained that the model adequately accounts for eccentricity. The slight over-
by comparing the predicted results with measured data. Table 4.8 re- prediction of both the average liquid holdup and pressure gradient
ports this performance in terms of the average percent error, £1, and might result from treating the liquid holdup in the liquid slug as a
the percent standard deviation, £3, as given by Eqs. 4.256 and 4.259. constant. However, this value might decrease with an increase in the
Results are provided for different flow patterns and for both concen- gas flow rate. Taking into consideration this variation might im-
tric and fully eccentric annuli. prove the performance of the model.
For bubble flow, Table 4.8 shows that the model's prediction of For annular flow,Table 4.8 shows that the model's prediction of liq-
liquid holdup is very good, with a slight underprediction tendency uid holdup and pressure gradient is adequate for air/water flow in either
of less than 3.0%. The degree of scatter about the mean is also good, annulus type. The model performance is only fair with air/kerosene.
with a standard deviation of about 5.0%. In general, the model per- The agreement between experimental results and the annular-
forms very well, independent of either physical properties or annu- flow model's predictions is less satisfactory than for models devel-
lus configuration. The agreement between the model and exper- oped for other flow patterns. However, the accuracy of experimental
imental results for pressure gradient also is good. The agreement is measurements obtained for the annular-flow pattern must be con-
weaker for the fully eccentric annulus case, with average percent er- sidered. The quick-closing ball valves used by Caetano-? to mea-
ror and standard deviation values slightly higher than 10%. This was sure liquid holdup have a lower accuracy in annular flow than for
caused by the very low superficial phase velocities used when ac- all other flow patterns. This is because of the low overall holdup val-
quiring the data in this configuration. Under these conditions, the ues characteristic of annular flow. The actual reported holdup values
L were sometimes less than the accuracy possible with the ball-valve
approach. Also, the pressure measurement system used by Caetano
8.1 and 7.2%, respectively. These higher percentage-error values re-
flect the generally lower values of liquid holdup rather than any in-
may not be adequate for the cross-sectional-dependent and axial-de- accuracy in predicting the absolute values.
pendent annular-flow pattern. Finally, because of gas compressor li- Hasan and Kabir also proposed a simplified approach for slug
mitations, it was not possible for Caetano to acquire data covering flow in which all the gas is assumed to have the same drift flux as
a broad region in this flow pattern. the Taylor bubble. Considerations of different drift fluxes in the
A second major aspect in the annular-flow-model performance is Taylor bubble and the liquid slug reduced the percent standard devi-
its strong dependence on the entrainment liquid fraction, F£. If a ation by only 0.01 for the air/water data and by 0.004 for the air/ker-
large entrainment fraction is predicted because of a high superficial osene data. However, the rigorous Caetano et al. slug-flow model
gas velocity, then a small overall liquid-holdup value will be pre- predicted the liquid-holdup data for both systems better, with
dicted. Wallis J8 reported that viscous liquids normally exhibit a E, =3.94% and £3=7.72% for the air/water data and £, =5.8%
L the existing flow field and are determined by liquid droplet size.
Because no measurements of liquid droplet size were obtained in
this investigation, Caetano et al. could not use these insights. As a
Hasan and Kabir also compared their model predictions with the
air/water data of Sadatomi et al.. 77 who used a 30-mm-OD,
l5-mm-ID annulus. The model predicted an average error of0.023
result, a deposition-rate expression was adopted. It is a mass-trans- and a standard deviation of 0.0214 for the gas void fraction in all
fer-process type that does not take into account the particle size and flow patterns in this data set. Good agreement also existed between
the inherent type of deposition mechanism. Thus, to refine the mod- the Sadatomi et al. data and Hasan and Kabir predictions for slug-
el, data with more detailed characteristics for this flow pattern and chum-flow patterns.
would be required.
4.6 General Observations
4.5.2 Hasan and Kabir Model. Hasan and Kabir6 8 reported statis- Engineers who perform multiphase-flow design calculations for
tical comparisons between predicted liquid-holdup values and wellbores are clearly faced with an immediate dilemma. Which cor-
measured values from Caetano-? for flow in a concentric annulus. relation or model should be used? Many companies have their fa-
L No comparisons were made on pressure drop. Table 4.9 presents
the statistical results.
vorites, often based on experience not documented in the literature.
Unfortunately, decisions often are made without awareness of the
For bubble flow, general overestimation of the liquid holdup is serious limitations or the availability of more accurate methods. The
evident for both fluid systems, although agreement is better for the evaluations presented in Sees, 4.3 and 4.5 are inconclusive and
airlkerosene system. Caetano's air/water bubble-flow data were based on a very limited set of data.
predicted with an average percentage error of 2.4%. The percentage Nevertheless, several conclusions can be drawn from published
standard deviation was 4.07%. The Caetano et al. 67 model predicted evaluations and experience. Among the empirical correlations pres-
the same data with an average percentage error and percentage stan-
L dard deviation of - 0.46% and 4.75%, respectively. The Hasan and
Kabir model predicted an average percentage error of 1.90% and a
percentage standard deviation of 3.98% for the Caetano airlkero-
ented, the modified Hagedorn and Brown" and Beggs and Brill!!
methods appear to be best. The Beggs and Brill method attempts to
account for the effect of inclination angle on liquid holdup and,
therefore, should perform better than the Hagedorn and Brown
sene bubble-flow data, compared with values of - 2.17% and
L 5.21'70, respectively, for the Caetano et al. model. The general over-
estimation by Hasan and Kabir suggests that a lower value of Co or
a lower value of Vs is needed in Eq. 4.246.
method in deviated wells. Although several attempts have been
made to create hybrid correlations-i.e., by use of one investiga-
tor's flow-pattern map with another's pressure-gradient equation-
none has any technical merit and all should be avoided. Any im-
For dispersed-bubble flow, the air/water liquid-holdup data are provements in accuracy are purely accidental.
overestimated on the average by 5.0%, compared with an underes- Mechanistic models offer the most accurate methods to predict
timation of 1.67% with the homogeneous model proposed by Caeta- pressure gradients in wellbores. Unfortunately, they also are much
no et al. However, treating the dispersed-bubble flow as homoge- more complicated and difficult to understand. Among these, the An-
neous may not be appropriate because it assumes a value of 1.0 for sari et al. 24 comprehensive model has enjoyed great success and is
L Co, when most of the bubbles still flow through the channel core. Cae-
tano's airlkerosene data appear to support this contention. The Hasan
accepted in the petroleum industry. Future models that improve sig-
nificantly on Ansari et al. no doubt will be published. Two mod-
and Kabir model predicts these data with an average percent error of els 28.29 are undergoing peer review. Others, such as the steady-state
0.60% and a standard deviation of 2.16%, compared with values of commercial codes OLGA and TACITE, are proprietary.
- 3.84% and 2.93%, respectively, for the homogeneous model. No existing comprehensive model properly accounts for all the ef-
For slug flow, the Hasan and Kabir model overestimates the air/ fects of inclination angle on flow behavior. Although excellent mod-
water slug-flow data by 7.04% and the airlkerosene data by 9.93%, els are available to predict flow patterns at all inclination angles, many
while percentage standard deviations for these two sets of data are of the variables used to predict flow behavior in mechanistic models
L modeling must begin with two-phase oil/water flow before the more
complicated multi phase-flow behavior can be addressed. No prov-
en mechanistic models have been published for oil/water flow in
Vertical Tubing," paper SPE 35676 presented at the 1996 Western Re-
gional Meeting, Anchorage, 22-24 May.
29. Petalas, N. and Aziz, K.: "Development and Testing of a New Mecha-
vertical or deviated wellbores. Analogous to gas/liquid flow. the nistic Model for Multiphase Flow in Pipes," paper presented at the 1996
modeling of oil/water flow-pattern transitions must first be accom- AS ME Summer Meeting, Anaheim, California, II July.
plished. Only then can improved flow-behavior models be devel- 30. Taitel, YM., Barnea, D.. and Dukler, A.E.: "Modeling Flow Pattern
oped to predict pressure gradient. Eventually, mechanistic models Transitions for Steady Upward Gas-Liquid Flow in Vertical Tubes,"
will be developed that are sufficiently accurate for a broad range of AlChE 1. (1980) 26, 345.
gas, oil. and water flow rates, inclination angles, oil viscosities, and 31. Barnea, D., Shoham, 0., and Taitel, Y: "Flow Pattern Transition for
Vertical Downward Two-Phase Flow," Chem. Eng. Sci. (1982) 37, 741.
both diameters and annulus configurations.
32. Barnea, D.: "A Unified Model for Predicting Flow-Pattern Transition
for the Whole Range of Pipe Inclinations," Inti. 1. Multiphase Flow
References (1987) 13, I.
I. Poettmann, EH. and Carpenter. P.G.: "The Multiphase Flow of Gas, Oil 33. Hannathy, T.Z.: "Velocity of Large Drops and Bubbles in Media ofInfi-
and Water Through Vertical Flow Strings with Application to the De- nite or Restricted Extent," AlChE 1. (1960) 6, 281.
sign and Gas-Lift Installations," Drill. & Prod. Prac. (1952) 257. 34. Scott, S.L. and Kouba. G.E.: "Advances in Slug Flow Characterization
2. Baxendell, P.B. and Thomas, R.: "The Calculation of Pressure Gradi- for Horizontal and Slightly Inclined Pipelines," paper SPE 20628 pres-
ents in High-Rate Flowing Wells," lPT (October 1961) 1023; Trans .• ented at the 1990 SPE Annual Technical Conference and Exhibition.
AIME,222. New Orleans, 23-26 September.
3. Fancher, G.H. Jr. and Brown, K.E.: "Prediction of Pressure Gradients for 35. Lockhart, R. W. and Martinelli, R.C.: "Proposed Correlation of Data for
Multiphase Flow in Tubing," SPEJ (March 1963) 59; Trans., AIME, 228, Isothermal Two-Phase, Two-Component Flow in Pipes," Chem. Eng.
4. Hagedorn, A.R. and Brown, K.E.: "Experimental Study of Pressure Prog. (1949) 45, 39.
L Gradients Occurring During Continuous Two-Phase Flow in Small-Di- 36. Alves, LN. et al.: "Modeling Annular Flow Behavior for Gas Wells,"
ameter Vertical Conduits," JPT(April 1965) 475; Trans., AIME, 234, paper presented at the 1988 Annual Winter Meeting of ASME, Chica-
5. User's Manualfor APlI4B. SSCSV Sizing Computer Program. second go, 27 November-2 December.
edition, API (1978) Appendix B, 38-41. 37. Caetano, E.F.: "Upward Vertical Two-Phase Flow Through an Annu-
6. Asheim, H.: "MONA, An Accurate Two-Phase Well Flow Model lus," PhD dissertation, U. of Tulsa, Tulsa, Oklahoma (1985).
Based on Phase Slippage," SPEPE (May 1986) 221. 38. Zuber, N. and Hench, J.: "Steady State and Transient Void Fraction of
7. Duns, H. Jr. and Ros, N.C.1.: "Vertical Flow of Gas and Liquid Mixtures Bubbling Systems and Their Operating Limits. Part I: Steady-State Op-
in Wells," Proc., Sixth World Pet. Cong., Tokyo (1963) 451. eration," Report #62GLloo, General Electric Co., Schenectady, New
8. Orkiszewski, J.: "Predicting Two-Phase Pressure Drops in Vertical York (1962).
Pipes," lPT(June 1967) 829; Trans., AIME, 240. 39. Fernandes, R.C., Sernait, T., and Dukler, A.E.: "Hydrodynamic Model
9. Aziz, K., Govier, G.W., and Fogarasi, M.: "Pressure Drop in Wells Pro- for Gas-Liquid Slug Flow in Vertical Tubes," AlChE 1. (1986) 32, 981.
ducing Oil and Gas," 1. Cdn. Pet. Tech. (July-September 1972) 11,38. 40. Sylvester, N.D.: "A Mechanistic Model for Two-Phase Vertical Slug
10. Chierici, G.L., Ciucci, G.M., and Sclocchi, G.: "Two-Phase Vertical Flow in Pipes," 1. Energy Res. Tech. (December 1987) 109,206.
Flow in Oil Wells-Prediction of Pressure Drop," lPT (August 1974) 41. Brotz, W.: "Uber die Vorausberechnung der Absorptionsgesch-windig-
927; Trans., AIME, 257. keit von Gasen in Stromenden Flussigkeitsschichten,' Chem.lng. Tech.
II. Beggs. H.D. and Brill, J.P.: "A Study of Two-Phase Flow in Inclined (1954) 26, 470.
Pipes," lPT(May 1973) 607; Trans., AIME, 255. 42. Schmidt, Z.: "Experimental Study of Two-Phase Slug Flow in a Pipe-
12. Mukherjee, H. and Brill, J.P.: "Pressure Drop Correlations for Inclined line-Riser Pipe System," PhD dissertation, U. of Tulsa, Tulsa, Oklaho-
Two-Phase Flow," 1. Energy Res. Tech. (December 1985) 107,549. ma (1977).
13. Griffith, P. and Wallis. G.B.: "Two-Phase Slug Flow," 1. Heat Transfer 43. Vo, D.T. and Shoham, 0.: "A Note on the Existence of a Solution for
(August 1961) 83, 307. Two-Phase Slug Flow in Vertical Pipes," 1. Energy Res. Tech. (June
14. Griffith, P.: "Two-Phase Flow in Pipes," special summer program, 1989) 111, 64.
Masschusetts Inst. of Technology, Cambridge, Massachusetts (1962). 44. McQuillan, K.w. and Whalley, P.B.: "F1ow Patterns in Vertical Two-
15. Davies, R.M. and Taylor, G.: "The Mechanics of Large Bubbles Rising Phase Flow," Inti. J. Multiphase Flow (1985) 11, 161.
Through Extended Liquids and Through Liquids in Tubes," Proc .• Roy- 45. Bamea, D.: "Effect of Bubble Shape on Pressure Drop Calculations in
al Soc., London (1949) 200A, 375. Vertical Slug Flow," Inti. 1. Multiphase Flow (1990) 16,79.
16. Brill, J.P.: "Discontinuities in the Orkiszewski Correlation for Predicting 46. Hewitt, G.F. and Hall-Taylor, N.S.: Annular Two-Phase Flow, Perga-
Pressure Gradients in Wells," 1. Energy Res. Tech. (March 1989) 111,34. mon Press, Ltd .. Oxford, U.K. (1970).
17. Govier, G.w., Radford. B.A., and Dunn, J.S.C.: "The Upward Vertical 47. Whalley, P.B. and Hewitt, G.F.: "The Correlation of Liquid Entrain-
Flow of Air-Water Mixtures, Part I," Cdn. 1. Chem. Eng. (1957) 35, 58. ment Fraction and Entrainment Rate in Annular Two-Phase Flow,"
18. Wallis, G.B.: One Dimensional Two-Phase Flow, McGraw-Hili Book UKAEA Report, AERE-R9187, Harwell, U.K. (1978).
Co. Inc., New York City (1969). 48. Lopes, J.C.B. and Dukler, A.E.: "Droplet Entrainment in Vertical An-
19. Al-Najjar, H.S.H. and Al-Soof, N.B.A.: "Alternate Flow-Pattern Maps nular Flow and its Contribution to Momentum Transfer," AIChE 1.
Can Improve Pressure-Drop Calculations of the Aziz et al. Multiphase- (1986) 32, 1500.
Flow Correlation," SPEPE (August 1989) 327. 49. Hasan, A.R., Kabir, C.S., and Rahman. R.: "Predicting Liquid Gradient
20. Nicklin, D.1., Wilkes, J.O .• and Davidson, J.F.: "Two-Phase Flow in in a Pumping-Well Annulus," SPEPE (February 1988) 113; Trans.,
Vertical Tubes," Trans .• AIChE (1962) 40, 61. AIME.285.
21. Brill, J.P. and Beggs, H.D.: Two-Phase Flow in Pipes. U. of Tulsa. Tul- 50. Barnea, D. and Brauner, N.: "Holdup of the Liquid Slug in Two-Phase
sa. Oklahoma (l99\). Intermittent Flow," Inti. 1. Multiphase Flow (1985) 11,43.
22. Payne. G.A. et al.: "Evaluation of Inclined-Pipe Two-Phase Liquid 51. Brauner, N. and Bamea, D.: "Slug/Churn Transition in Upward Gas-
Holdup and Pressure-Loss Correlations Using Experimental Data," Liquid Flow," Chem. Eng. Sci. (1986) 41,159.
JPT(September 1979) 1198; Trans., AIME, 267. 52. Steen, D.A and Wallis, G.B.: AEC report No. NYO-31142-2 (1964).
68
L S3. Ansari. A.M. et al.: "Supplement to paper SPE 20630, A Comprehen-
sive Mechanistic Model for Upward Two-Phase Flow in Wellbores.'
65. Ros, N.C.J.: "Simultaneous Flow of Gas and Liquid as Encountered in
Well Tubing," 1 PT (October 1961) 1037; Trans.• AIME, 222.
paper SPE 28671 available from SPE. Richardson. Texas (May 1994). 66. Caetano. E.F.. Shoham, 0 .. and Brill, J.P.: "Upward Vertical Two-Phase
54. Govier. G.W. and Fogarasi, M.: "Pressure Drop in Wells Producing Gas Flow Through an Annulus, Part I: Single-Phase Friction Factor, Taylor
and Condensate," 1. Cdn. Pet. Tech. (October-December 1975) 14,28.
Bubble-Rise Velocity and Flow-Pattern Prediction," 1. Energy Res.
55. Hagedorn. A.R.: "Experimental Study of Pressure Gradients Occurring
Tech. (March 1992) 114, I.
during Continuous Two-Phase Flow in Small Diameter Vertical Con-
67. Caetano, E.F.. Shoharn, 0., and Brill. J.P.: "Upward Vertical Two-Phase
duits," PhD dissertation, U. of Texas. Austin. Texas (1964)
56. Espanol, H.J.H.: "Comparison of Three Methods for Calculating a Flow Through an Annulus. Pan II: Modeling Bubble, Slug and Annular
Pressure Traverse in Vertical Multi-Phase Flow." MS thesis, U. ofTul- Flow," 1. Energy Res. Tech. (March 1992) 114, 14.
sa. Tulsa, Oklahoma (1968). 68. Hasan, A.R. and Kabir, C.S.: "Two-Phase Flow in Vertical and Inclined
57. Messulam, S.A.G.: "Comparison of Correlations for Predicting Multi- Annuli," Intl. J. Multiphase Flow (1992) 18,279.
I
l
I
L
I
i
'--
I
I
I.-
Chapter 5
Flow Through Restrictions and
Piping Components
q = CA j2g~6.P . (5.1 )
I
i
_
q - 22, 800Cd ch
2
V{iP
P , ........................ (5.2)
:.... Fig. S.3--Ceramlc choke disk operation (after Willis Oil Tool Co.2).
where q is in BID and dch = the choke diameter in inches. Choke di-
and for an upstream pressure and temperature of 1,000 psia and
ameter is frequently called "bean" size and is measured in 64ths of
150°F. respectively. Eq. 5.5 predicts a critical-pressure ratio of 0.56
an inch.
for this gas. For all pressure ratios below Ye, the flow rate is constant
The flow coefficient, C, in Eqs. 5.1 and 5.2 accounts for all irre-
for a given choke size.
versibilities, such as friction. C can be determined experimentally
and depends primarily on the type of restriction (i.e., venturi, The correct ratio of specific heats for hydrocarbon gases varies
nozzle, orifice, or choke), the ratio of the restriction diameter to the with pressure and temperature.P Fig. 5.8 shows that values as high as
pipe diameter, and the Reynolds number. Fig. 5,6 shows the flow- k = 2.0 are possible for methane at pressures between 2,000 and 4,000
coefficient behavior for a nozzle. psia and a temperature of 50°F. This can have a significant effect on
the predicted critical-pressure ratio and resulting critical-flow rate.
(k +2 l)k~l.
I p psia kPa
."
i
1
.......
-si
l-
2,000
l.aoo
r---------------------,
t
LOCKING
MANDREL.
1.995
0
::
1.995 e 1,400
U
1,600
., 1,200
+-
III ::::e
0.687 ;:)
Iii
l-
.! 1,000 Critical Flow
t + lIS
a:
EQUALIZING
L
., 600
SUB lIS ..-- Subcritical
0.720 e Flow
t-
600
0.750 0.890
..... 0.942 ..... 400
> >
~ 0.750
li.> 0.558 2IlO
>
., OTIS
l'l
I
a:
CA/'lCO
BEAN 0l:-__
o ~--~:---~~--~---L----J
0.2 0.8 1':
CERRMIC
-
CIl
l-
e
BEAN 0
u
iu
7::.<:"'----__ 1.50
+-s
0.750 0.800 Fig. 50S-Dependence of choke flow rate on y.
1
0
z
1.995
+:: 0
1.995
where
I
~ l-
!lI-
(5.8)
and
L Yc = ----------;0----------
X g• = w g • + WLI • . 0. 0.' .... 00.. 0... 00. 00' ... 00. (5.11)
(5.6)
where
• • • • • • • • • • 0 ••••••••••
where
Wgi = 0.0764A. gQLj Rp - faRs')'
b = k - I and
k
and
e = -k-'
k + I Eq. 5.7 is dimensionless and any set of consistent units can be
7"')
L 1.20
~ 6,000
1.18
/--- 0.725
5,000
32/ 64 in. k=1.25
----
1.16
-: ~ PI =1.000
T1 = 150°F
1.14
,./
0.70
g 4.000 CD=0.62
~1.12 ,. ~ 0.675
o
en 24/ 64 in.
~--- :::E 3.000
"21.10
:e81.08 /'
-: /---
.......1---
---- 0.65
0.625
~
ai
en
'"
2,000
16/64 in.
--
°106 0.60 (!) 1,000
I :to
/ / ....... 0.575 8/ 64 in.
L ~ 1.04 0.55
~1.02
- ~ ~ ....... 0.50
0.45
a
a 02 04 06 0.8 1.0 1.2
o /.:::: / ....... :.- 0.40 Pressure Ratio, y
c3 1.00 :/;::
0.30
0.98
0.20 Fig. 5.7-Gas flow performance for different choke sizes.
~
0.96
0.94 "':. to define the boundary between critical and subcritical flow. The
curves in Fig. 5.10 were based on a downstream pressure of 19.9
I
'--
0.92
2 4 6 8 104 4 6 810S 4 6 81()6 psia. Eq. 5.14 should be used to calculate the Fortunati mixture ve-
locity from the actual mixture velocity and downstream pressure.
Reynolds Number based on ~. NRe
I1
I-
Fig. 5.6-Flow coefficients for liquid flow through a nozzle (after
Crane Co.S).
vm 2 = V m2F( ltif. . (5.14)
where
I
i- TJ = ( I - A.~2 )
0.38
. ............................ (5.15)
Fig. 5.9 compares the Fortunati critical-flow boundary with bound-
i aries predicted by Ashford and Pierce.
i
'-- Wallis Sonic Velocity. In Eq. 5.16, Wallis presented an expression
to calculate the sonic or compressibility-wave velocity of a homoge-
neous mixture.l-'
..... (5.16)
As for the case of Ashford and Pierce and Sachdeva et al., an itera-
tive procedure is required to determine Yc-
The parallel expression for the effective sonic velocity in the liq-
Fortunati. Fortunati II presented an empirical method that can be uid phase is
used to calculate both critical and subcritical multiphase flow
through chokes. He assumed a homogeneous mixture and conten-
ded that the assumption was valid as long as Vm is greater than 32.8 .................. (5.18)
ftlsec and the mixture Froude number is greater than 600. Using ex-
perimental data, Fortunati developed Fig. 5.10, which can be used
L 2.4-r----------------------., 0.7
0.6
k~l.04
k~1.20
Subcrltlcai Flow • k=l . •
.
N
2.0 0.5
~
It
~ 0
Ue,
..." ~ 0.4
f
1.6
.,.,
:::I
0.3
Critical Flow
l! ec..
iii
1.2~~~~~
I.-. u 0.2
~
u
10 100 1,000 10,000 i.
>c 0.1
Pressure, psia
Fig. 5.B-Ratio of specific heats for methane (after Perkins 6).
0.0 L - _....... ........' - - _......_ _- " - _ - - ' ...... _ ~
*_ vi v; and
v - H' (I ) •. . (5.19)
LVg + - H L VL
~I
i BaxendeU1B 1.93 3.12x 10- 3 0.546 A ForfJ=0.70
'--
,£ ForfJ=O.80
Achong 19 1.88 1.54x 10- 3 0.650 O.
c; ForfJ= 0.50
lI
TABLE 5.3-SUMMARY OF RESULTS C/J
~/'
~
E ForfJ=0.40
"-' I 3=
• o
P2 iI
(psia) .z.. Status
200 0.118 Critical
400 0.235 Critical
600 0.353 Critical
800 0.470 Subcritical
1,000 0.588 Subcritical
1,200 0.706 Subcritical
R
I
= VV Sg l = 3.82
3.97
= 0962
.
SLl
and
1.25
k = 1.3. = (1.84)(10,000) ( ~46/ )
The Eq. 5.6 trial-and-error solution yields Ye = 0.446.
Determine the flow status of the given pressure conditions for 160,674,
PI = 1,700 psia. Table 5.3 summarizes the results.
263N_~~~19}..0657)118
2. For cases of critical flow, determine dch using the Gilbert and
the Omaiia et at. critical-flow correlations. ND = (0 18.21,
Gilbert: . P PI L
i 1/1.89
and
!
'-- d = [(3.86 X 10- 3)(10, 000 )(1, 000 )0.546]
~
120.872dch(~~) 'h = 18.21,
1.700
ND =
I
1 ' = 63.6 .
L =.
O 993 m. 64 m. then
dch = 0.0633 ft
Omaiia et al. :
R 1 = 0.962, = 0.76 in.
i I
~
9.0
8.0
"'"\.."-
8
8 "I' -o-
6
6
.s
7.0
6.0
5.0
rt "
\
\" \
\..~ ....
1\
1.
"
\
V~ --
\"
4
C>
c: 4.0
\ ~\ '\. j 1\ ~
.£ 4 '5.
11'-
.~
en
ii:
1ii
II> 3.0 '\ \\ \. r\ .~
\ '\~\ I~
I~
Ol I-
'- .S 2.5 '0
~ 2
l;;
OJ 2.0
)., f\ I~
0 2
~\ \ ~ ~
i
....
\ ~
>
1.5
1.5
E
as
i5 1.5 ~ i:\ \ 1\
\ ~ l\, ~) \ \; \
1.25-8
1.2 'iii
'-{
.s
1.0
.... l--r .... '\
I'" U\
0.9
~'r- ... ~
0.75 0.8
" '"
I', \ ": 1\
0.5
0.7
0.6
"\. l; ,\ \ \" 1-
0.5
0.5 ~
0
co
e
....
0 0.1 0.15 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.91.0 ,1.5 2.0 2.5 3 4 5 6 7 8 9 10
"s:.o "s:.c
"3 "3
Resistance Coefficient, K
CIl CIl
o = Schedule 40 pipe, 30 diameters long (K = 301) ¢ =600-lbm steel wedge gate valves, seat reduced
cr = 125-lbm iron-body wedge gate valves -<>- = 300-lbm steel venturi ball-cage gate valves
<;> = 600-lbm steel wedge gate valves cf = 125-lbm iron-body V-pattern globe valves
" = 90° pipe bends, rfd= 2 'Q. = 125-lbm brass angle valves, composition disk
6 =90° pipe bends, rfd=3 ):( = 125-lbm brass globe valves, composition disk
-0- = 90° pipe bends, rfd= 1
Fig. 5.11-Varlatlon of resistance coefficient Kwith size and type of selected components (after Crane Co.s).
L
TABLE 5.4-SUMMARY OF RESULTS . 64.4 .
and d ch =.
0 0838 f t =.
I 005 Ill. = 64 in.:
P2 ~h
(psia) (64ths in.) Status . d ch -- 68.6·
at P2 = I, 000 psia, 64 in.;.
200 63.6 Critical
.~
400 63.6 Critical and at P2 = I, 200 psia, d ch = 7~6 in.
600 63.6 Critical
Table 5.4 shows a summary of the results.
800 64.4 Subcritical Clearly, the Gilbert and Ornafia et al. correlations give signifi-
1,000 68.6 Subcritical cantly different results for critical flow.
......
1
D..
P
= (.f!:: K) Pnv~ = [(5)(0. 15) + I .8] (27.04)(7.83)2
L ' 2g 2(32.2)
16. Ros, N.C.J.: "An Analysis of Critical Simultaneous Gas-Liquid Flow
Through a Restriction and Its Application to F1owmetering," Appl. Sci.
i= I C Res. (1960) 9,
17. Gilben, W.E.: "Flowing and Gas-Lift Well Performance," Drill. &
Prod. Prac. (1954) 126.
= 65.64 Ibf/ft 2 = 0.46 psi.
18. Baxendell, P.B.: "Bean Performance-Lake Maracaibo Wells," Inter-
nal Company Report. Shell Oil Co., Houston (October 1967).
19. Achong, I.: "Revised Bean Performance Formula for Lake Maracaibo
References Wells," internal company report, Shell Oil Co., Houston (October 1961).
I. Cook, H.L. Jr. and Dotterweich, EH.: Report on the Calibration ofPosi- 20. Sookprasong, P., Brill, J.P., and Schmidt, Z.: "Two-Phase Flow in Pip-
tive Flow Beans as Manufactured by Thornhill-Craver Company, ing Components," 1. Energy Res. Tech. (September 1986) 108, 197.
I
L
6.1 Introduction jects, there is a continuous pressure gradient from the reservoir to
The mechanics of fluid flow in every component in a well's plumb- the separator. In well design calculations, it is not uncommon to use
ing system affects the flow rate. Accurate well design is the key to wellhead pressure for the separator pressure, assuming that the sep-
achieving optimum flow rate. Consequently, understanding the me- arator is at or very near the wellhead. Such assumptions imply negli-
chanics of fluid flow through each component, from the reservoir to gible pressure loss in the surface flowline. For long flowlines, espe-
the first stage of separation, is imperative for accurate well design. cially in hilly terrain, to ignore pressure losses may lead to
The overall objecti ve of this chapter is to present example applica- substantial error in the production-rate calculation.
tions of multiphase-flow theories to well design and optimization. A node is any point in the production system where the pressure
It also provides a brief discussion of the reservoir flow into the well- can be calculated as a function of the flow rates. As shown in Fig.
bore to enable well design calculations. 6.2, the two extreme nodes are the reservoir drainage boundary
The well design methods presented in this chapter also are used (Node 8) and the separator (Node I). The pressures at these nodes
to troubleshoot well problems. In this context, some of the key are called the average reservoir pressure. Pr- and the separator pres-
constraining phenomena in well design, such as gas-well loading, sure, Psep» respectively. The other two important nodes are the bot-
erosion, and formation of natural gas hydrates, are also discussed. tomhole (Node 6), where the bottomhole flowing pressure, Pwf, is
The plumbing system is an interfacing conduit between the reser-
L voir and the surface handling facilities. Without it, the hydrocarbons
cannot become a tangible asset. For optimal production, a well de-
sign requires complex engineering considerations that depend on
measured by a downhole gauge, and the wellhead (Node 3), where
the wellhead pressure, Pwh, is measured by a gauge attached to the
Christmas tree or the flow arm. If the pressures are measured at each
node, the pressure loss between the nodes can be calculated as a
well components. Optimal production yields a maximum return on
function of the flow rate. Some nodes (Nodes 2, 4, and 5) are called
investment. not a maximum production rate. Fig. 6.1 shows the ma-
functional nodes. These exist where a pressure drop occurs because
jor components with substantial pressure losses in a typical well. A
of a choke, safety valve, or other piping component. For each com-
production system consists of the following major components. 1-9
• Porous medium. ponent, the flow rate, q. is related functionally to the pressure differ-
q = f(t1p)· (6.1)
• Flowline with choke and other piping components (valves, el- Previous chapters established these mathematical relationships
bows, and other such elements) from the wellhead to the first-stage for different component segments of the production system, except
separator. for the reservoir system. The following sections discuss parameters
In an oil or gas production system, the fluids flow from the drain- that are important for the optimization of production through these
age boundary in the reservoir to the separator at the surface. The av- components. Production-system analysis combines all component-
erage pressure within the drainage boundary is called the average system design procedures to help in the design and optimization of
reservoir pressure. This pressure controls the flow through a pro- the total production system.
duction system and is assumed to remain constant over a fixed time
interval during depletion. When this pressure changes, the well's 6.2 Vertical-Flow Performance
performance changes and the well may need to be re-evaluated. The
average reservoir pressure changes because of normal reservoir Vertical-flow performance- is the well's ability to produce under a
depletion or artificial pressure maintenance with water, gas, or constant surface-pressure constraint. In a producing well, this is
chemical injection. called tubing intakev" or outflow performance. The rate vs. pres-
The separator pressure at the surface is designed to optimize pro- sure loss in Nodes 1 through 6 are considered to calculate the tubing
duction and to retain lighter hydrocarbon components in the liquid intake. For deliverability calculations that use any iterative scheme
phase. This pressure is maintained constant by the use of such me- or graphic method, it is convenient to represent the bottomhole
chanical devices as pressure regulators. As the well produces or in- flowing pressure in terms of pressure losses through each of the
Fig. 6.1-Possible pressure losses in the producing system of The algebraic sign convention to determine pressure losses in Eq.
a flowing wel1.6 6.2 is important for production or injection cases. Pressure losses re-
sulting from friction and acceleration always occur in the direction
nodes, starting ata 'fixed wellhead or separator pressure. This rela- of flow, whereas hydrostatic pressure loss occurs in the direction of
tionship can be expressed mathematically as elevation gain. In Eq. 6.2, the friction and acceleration terms should
be positive for production. For injection, the separator pressure,
Pwf = Psep + t1Ph + (t1Pfl + t1PI + t1P"h)f + t1pacc, Psep- in Eq. 6.2 should be replaced by pump discharge pressure; and
the friction and acceleration terms should be negative. Consequent-
(6.2) ly, as the injection rate increases, the bottomhole injection pressure
decreases. Fig. 6.4 shows a set of typical tubing performance curves
where Pwf= bottomhole flowing pressure at Node 6, Psep = separator for an injection well.
pressure,!¥Jh = total hydrostatic pressure loss, (!¥Jfl.!¥Jt,!¥Jch">!= fric- Note that in the case of injection, tubing performance could be
tion pressure losses through the flowline, tubing, choke, or restric- called tubing outflow because the fluid is pumped from the tubing
tions, and !¥Jacc = pressure loss because of acceleration. into the reservoir. However, during production the tubing intake
Note that, in the absence of flowlines, Psep can be set equal to the represents tubing performance. In the rest of this chapter, the tubing
wellhead pressure, Pwh, and friction loss in the flowline, t1Pjl, can performance curve (TPC) is used for both injection and production.
be set to zero in Eq. 6.2. The separator or the wellhead pressure nor- As Fig. 6.5 shows, Eq. 6.2 also can be used to generate gradient
mally is known and the methods to determine the pressure losses in curves, or flowing pressure vs. depth curves. Gradient curves are
each segment were discussed in previous chapters. In this chapter,
the plot of Pwf vs. q (assuming a fixed wellhead/separator pressure)
is called the tubing-intake curve. Fig. 6.3 presents a typical tubing-
intake curve for producing wells. It is important to note that the flow 7,000 r--------------,
condition represented by this curve will be stable only if dp/dq 2: 0
at any given point. Flow rates calculated in the unstable flow region
indicate intermittent production, heading, or loading up of wells. Injectivity
I'-- 6,000
This should be avoided through proper design and will be discussed
in the example problems.
5,000
3,OOO-y----------------,
0)
2,500
Unstable Flow
.~ 4,000
...
,...
1
~~
~
Q.
2,000 3,000 ?'""~......
L r}1,500
2,000
1,000
1,000 400PSi9~
~-- Stable Flow
500
L 0-J--J-+---t__----f----+----1
0+----+----..+---+----1
o 2,000 4,000 6,000 8,000
o 500 1,000 1,500 2,000 2,500
Rate, q Injection Rate, BWPD
Fig. 6.3-Typical tubing-intake curve for producing wells. Fig. 6.4-Typical systems graph for injection wells.
1.000
til
'iii
2.000 a.
~GlR=50
-.-GLR=100
3,000
-<>- GLR = 200
-e-GLR=400
4,000 __ GlR=800
;:: --tr- GLR = 1.500
t
CD
5,000
Q
6.000
11,000 0
Average reservoir pressure. Pro psig 1.000
500 1,000 1.500 2,000 2,500 3,000 3.500 4,000
Water cut o
Pressure. psig Unknown
Oil viscosity. Ito. cp
Formation volume factor, Bo• bbl/STBO Unknown
Fig. 6.5-Typical set of gradient curves for different GlR's. Skin, S o
Pumped off (negligible fluid column above pump section)
very useful in the design of artificial-lift systems, such as gas lift or
bottomhole pumps, as illustrated in Sec. 6.5.
6.3.1 Single-Phase Liquid Flow. For single-phase oil or liquids, the Solution.
IPR shown in Fig. 6.6 is stated by Darcy's law 10.II for radial flow as I. Estimate production rate assuming a 90% drawdown.
\
\.- ,uo = 1 cp,
........... (6.3)
qo = ,uoBo[ln(;:) - 0.75 + Sf + DqoJ' s, = I bbllSTBO,
where % = oil flow rate into the well, STBOID; Bo = formation vol- and
ume factor of oil, bbIlSTBO, (see Appendix B);,uo = viscosity of = (6)(15)(0.9)(1,000) =: 81 STBOID
oil, cp, (see Appendix B); ko = effective permeability of the forma- qo - 1,000 - .
tion to oil, md; h = net thickness of the formation, ft; Pr = average 2. For a well producing 20 STBOID, diagnose possible problems.
reservoir pressure, psia; Pw = bottomhole flowing pressure, psia; a. Check actual viscosity under reservoir conditions.
J
re = radius of drainage, ft = Afx, where A is area of circular drain-
s,
age, ft2; rw = well bore radius, ft; = total skin; and Dq, = pseudo-
b. If viscosity is not the problem, check fluid level in annulus.
L
c. If fluid level has risen, increasing the Pwf or reducing draw-
skin caused by turbulence. In oil wells, this term is insignificant, es- down, the pump is the problem.
pecially for low-permeability reservoirs. d. If fluid level has not risen, the problem may be caused by skin.
I
I
phase liquid reservoir. Thus,
J = qo
P.f)
. ctJ,
r.
o.~112
•
m .':
172iA'12
r.
'-- (Pr -
I
4
. • J.3J.!" 112
~
::.:: 1 194A"
r.
2
5.
On the basis of Darcy's law, 1I2
I It 2.066A
J = 7.08 x 1O-
,ufiu[ln(;:) - 0.75 + SI]
3k
oh qu
(Pr - PW!)'
STBO!D
psi
EB....:..
r...
Q8f!A1I2 EIffi 4W2A' 12
r.
I . . . . . . . . . . . . . . . .. . .. (6.5) Ef}
-.- 1.48 5A1I2 H·1+f Q.52aA'1J2
r•
-- The productivity-index concept is not applicable for gas wells be-
cause the IPR for a gas well is nonlinear.
r.
&. 10.135A1J2
r.
Il -
Example 6,2-0ilwell IPR Problem.
Using Fig. 6.7, plot the IPR curve.
Input Data. Table 6.2 provides the input data.
2. Determine productivity index,
Solution. •
I. Calculate drainage radius.
J = qu = 7.08 x 1O- 3k,,h
(Pr - P.f) ,ufiu[ In(;:) - 0.75 ]
r, = jA x ;:,560 = 1,490 ft.
Assume rw = 0.51 ft and apply Darcy's law for radial flow. 1.22 STBO/D .
pSI
I
' -- 6.3.3 Single-Phase Gas Flow. Darcy's law for single-phase gas is
qOmax = (0.8)(1.2)[lnCo~n _ 0.75] where qg = gas flow rate, Mscf/D; kg = effective permeability to
L gas, md; Z= gas compressibility factor determined at average tem-
perature and average pressure, fraction; T; = average reservoir tem-
= 26,550 = 3 672 STBOID perature, OR; and II g = gas viscosity calculated at average pressure
I 7.23' . and average temperature. cpo
i
l-
6.3.4 Sources of Information. When performing IPR calculations,
3,000 Darcy's Law (liquid) several information sources are available, including
1. Transient well-test interpretation.l-d ' For example, buildup
..
Ui
Q,
and drawdown tests are used to determine khh: and St. Successful
buildup tests also help determine average pressure Pro In injection
wells, the buildup test is called a falloff test and the draw down
Skin,' s -1 216
- 5, -1, 0, 1, 5,
10,50 o 186
Oil viscosity, ~o' cp 1.1 1 163
Formation volume factor, Bo, bbllSTBO 1.2 5 110
Spacing, acres 80 10 78
L Wellbore radius,
'Sensitivity parameter
'w, ft 0.365 50 23
vestigation, whereas logs and cores determine the value of k at dis- TABLE 6.6-INPUT DATA FOR EXAMPLE 6.3, STEP 2
crete points around the wellbore. Reservoir temperature, T" OF 200
6.3.5 Boundary Effects. Most reservoir engineering calculations Gas permeability, kg, md 1
assume radial-flow geometry. Radial geometry implies that the Pay thickness, n, ft 200
drainage area is circular and the well is located at the center of the Average gas viscosity, ,Ug, cp 0.019
drainage circle. In many cases, the drainage area is rectangular or of
some other noncircular shape. Applications of equations based on
Gas compressibility factor, Z 1.1
radial geometry to a noncircular drainage area could lead to substan- Average reservoir pressure, Pr, psig 3,500
tial error. Darcy's law can be modified for a bounded drainage area Spacing, acres 80
of different shapes as
Wellbore radius, 'w, ft 0.365
7.08 X 1O- 3kjz(Pr - PW!) Skin, S 1
qo = , (6.7)
Bo,uo[ln(X) - 0.75 + 5,]
2. For a gas well, tabulate and draw the IPR curve. Calculate the
L for oil and
7.03 x 1O- 6
•
kgh(p;- p~!)
AOFP.
Input Data. Table 6.6 gives the input data.
(6.8)
Solution.
From Part I, for an oil well, In (r.lrw) - 0.75 = 7.22.
for gas. The productivity index for oil then is From Darcy's law (neglecting turbulence),
7.0S x 1O- 3kjz
[
Bu,uu In(X) - 0.75 + 5/
I' ......... (6.9)
where Xis given in Table 6.3 for various drainage shapes and well
locations. Note that A is the drainage area and fA Irw is dimen-
sionless. 7.03 X 10- 4 (1)(200)(3, 500 2 - P:f )
(0.019)(1.l)(660)(7.22 + I)
Example 6.3-IPR Problems for Oil and Gas Wells.
1. For an oil well, draw IPR's and tabulate skin vs. AOFP, qOmax'
Input Data. Table 6.4 gives the input data.
1.24 x 10- 3(3,500 2 - p:X
Fig. 6.9 gives this IPR curve. Table 6.7 gives the IPR data used
Solution. The drainage radius, re , is
in this figure.
re = (SO)(~,560) = 1,053 ft. 3. Solve Part I for a square drainage area and a skin of zero.
Solution.
AOFP is
1,341 1,341
qOmax = InX - 0.75 + 5 = InX - 0.75'
qOmax = ,uuBo[In( r. I r w ) - 0.75 + 5] From Table 6.3,
82
L 3,000 4,000 . , . . - - - - - - - - - - - - - - - - - - - ,
2,500
3,000
2,000
III
Ui
~ 1,500 III
l .~ 2,000
L
Q.
j
1,000 Q.
500 1,000
Fig. 6.8-IPR lines for an oil well in Example. 6.3 (Part 1). Gas Rate, C7g, Mscf/D
j Fig. 6.~PR curve for a gas well in Example 6.3 (Part 2).
I
'-- TABLE 6.7-IPR DATA USED IN FIG. 6.9
2
PM
(psig) (MS'S/D)
~
qo."..
= I - O.2(P_P,f)- O.8(P,:f)
W
r, ............ : (6.10)
3,500 o Flow rate normalization with q 0.".. seemed to make this IPR work
3,000 4,030 in a multiphase system. This IPR can be modified easily for a com-
2,500 7,440 posite reservoir system, as shown in Eqs. 6.15 through 6.17. A Vogel
I
I.....
2,000 10,230 IPR curve can be generated if either the AOFP, qo.".., and the reser-
1,500 12,400 voir pressure, 15" are known or the reservoir pressure, 15" and a flow
1,000 13,950 rate and the corresponding bottornhole flowing pressure are known.
For either case, a buildup test for 15, and a flow test with a bottom-
500 14,880
hole gauge are needed. Though quite simple to apply, the Vogel IPR
o 15,190 is not commonly supported by field data from multirate oilwell tests.
The Fetkovich 17 IPR, on the other hand, quite often is supported by
6.3.6 Two-Phase Flow. Darcy's law is applicable only for single- field data and is more universally used for both oil and gas wells.
phase flow. For the case of an oil reservoir, single-phase flow occurs
Fetkovich IPR. Multipoint or backpressure testing of gas wells is
when the bottomhole flowing pressure is above the bubblepoint
a common procedure to establish gas-well performance curves or
pressure of the reservoir fluid at the reservoir temperature. During
deliverability. Fetkovich applied these tests to oil wells with reser-
I
"-
reservoir depletion, the pressure continues to drop unless main-
tained by fluid injection or flooding. Consequently, during deple-
voir pressures above and below the bubble point pressure. The gen-
eral conclusion from these backpressure tests is that, as in gas wells,
tion, the bottomhole flowing pressure falls below the bubblepoint the IPR in oil wells is of the form
pressure, resulting in a combination of single- and two-phase flow
in the reservoir. The pressure profile in the drainage flow path deter- A=BlackOil
mines the areas of single- and two-phase flow on the basis of the B = Composite
C=WetGas
black-oil phase diagram shown in Fig. 6.10. The IPR of such sys- D=Dry Gas
tems is called a composite IPR. Before discussing composite IPR, A B E = Solution Gas
X=p,
hydrocarbon phase behavior is reviewed briefly. O=Pw[
~=Wellhead
Phase Behavior of Hydrocarbon Fluids. On the basis of fluid
I samples taken at reservoir conditions, a pressure/volumeltempera-
IL- ture (PVT) analysis generates a phase envelope in a pressure-tem-
perature diagram. 15 Fig. 6.10 presents a typical black-oil pressure-
temperature diagram showing the physical state of the fluids. When
the average reservoir pressure, the bottornhole and wellhead flow-
ing pressures, and the corresponding temperatures are plotted, one
can identify the type of reservoir fluid, such as single-phase, two-
phase, or a combination. Such information helps determine the type
of IPR equation to be used.
I
'--
For the case of two-phase flow, such as Case E in Fig. 6.10 where
fi, is below the bubblepoint pressure, either a Vogel 16 or a Fetko-
vich 17 IPR is recommended. For composite reservoirs shown in Fig.
6.10, Darcy's law is used above the bubblepoint pressure and a Vo-
gel or a Fetkovich IPR is used below the bubblepoint. Temperature. OR
VogeiIPR. Through simulation of saturated oil reservoirs, Vogel
developed this IPR. Fig. 6.1O-Typical phase diagram for black 011.
L WELL DESIGN APPLICATIONS 83
I
'--
TABLE 6.9-INPUT DATA FOR EXAMPLE 6.5 TABLE 6.1G-TRANSIENT IPR's FOR EXAMPLE 6.5
Drainage radius, 'e, ft 2,000 t q,max
(hours) (STBO/D)
Permeability, ko, md 1
0.1 79
Average reservoir pressure, Pr, psig 2,000
10 48
Net pay thickness, h, ft 20
100 40
Porosity, <p 0.2 1,000 35
Oil viscosity, .uo, cp 1 7,584 31
Wellbore radius, 'w, ft 0.5
1
162. .".s.{[IOg(••;",) - 3.23 + 0.87'] + IOg(t)} and
I 0.000264kg t
...... .................... (6.13) t Dxf = I.. ) 2 ;
¢\f'oc, i xf
Also known as the transient IPR equation.I? this equation is valid
I for the transient radial-flow period. It is characterized by flowing and dimensionless fracture conductivity, CjD, is
I times less than lpss when pseudosteady-state flow begins, where
'-- CjD = kfw/kxf,
¢/-loc,r; where w = fracture width and xJ= fracture half-length.
I tp " = 948- hours. . (6.14)
k--,
o Fig. 6.13 presents a few of these types of curves. It is important
!...... to note that the early-time pressure behavior depends on CjD. How-
ever, at late times or after depletion starts, the pressure response is
Example 6.5- Transient IPR Problem. Calculate the duration of
influenced by the shape and size of the drainage area.
infinite-acting radial flow and tabulate the AOFP.
I As in the homogeneous reservoir case, transient IPR curves can
i Input Data. Table 6.9 gives the input data. be generated for fractured reservoirs by use of curves from Fig. 6.13
'--
Solution. Duration of infinite-acting radial flow, tpss• 1()2 . . . . - - - - - - - - - - - - - - - - - : - : : - : - - - - - - ,
Fraclure
Penetration
0.2 x 10 -5(2, 000)2 RaIio~
tpss = 948 I = 7,584 hours (316 days). x.
(20)(2,000)
I
L 162.6[{IOg[ I
0.2 x 10- 5(0.5)
2]-3.23+0.87(1.21)}+IOgt]
246
4.12 + log r '
Note that for a single-phase oil reservoir, average reservoir pressure Fig. 6.13-Constant rate type curve for finite-conductivity frac-
ture in a closed-square system. 22
L
and the AOFP at any time uniquely define the IPR for that time.
L a = L[ 0.5 +.
2"
Rate, 'lsc
for li2<0.9 reh and lani' permeability anisotropy = Jkh/k v • Sub-
Fig. 6.11-Different forms of IPR's. scripts h and v refer to horizontal and vertical, respectively. All vari-
ables are in oilfield units and can be converted easily fora gas-well
IPR, as presented in Example 6.4.
. . . . . . . . . . . . . . . . . . . . . . . . . . . (6.11)
This equation is called the oil or gas deliverability equation. The Example 6.4-Horizontal-Well Problem. Calculate the flow rate
exponent n ranges between 0.5 and 1.0 for both oil and gas wells; through a horizontal gas well at a bottomhole flowing pressure of
I n less than 1.0 often is attributed to non-Darcy flow effects. The co- 400 psia.
L efficient C represents the reservoir productivity index and is a strong
Input Data. Table 6.8 gives the input data.
function of phase mobility. Consequently, this coefficient increases
with increasing values of k and decreases with an increasing skin Solution.
factor. For pseudosteady-state production without rate-dependent
skin, C changes as a result of depletion. It also is important to note q=
that for gas wells, this equation is applicable at all stages of deple-
tion. Assuming pseudosteady-state flow, neither C nor n changes.
i
J
Golan and Whitson? presented a detailed discussion of this IPR. The
'-- Fetkovich IPR is customized for a well and is obtained by multipoint
backpressure testing, such as flow after flow or isochronal testing.
,
I ,,,
I
'--
1:}
)J,oUo
{
I
n
[
a+ a2 -
(L/2)
L/2
( )2
+
hlani I (hl ani) }
L n 2r w
,,.
.................... (6.12)
for L > hlani and (li2) < 0.9reh and where a is one-half the major
axis of a drainage ellipse in a horizontal plane. Fig. 6.12-Horizontal-well drainage model (from Joshi 18).
For gas wells, Steps I through 5 are followed in the range of pres- The system sensitivity study is a major engineering application of
sures considered to generate IPR curves. Example 6.8 shows how production-systems analysis. A system sensitivity is defined as the
to use transient IPR to generate production decline curves for homo- functional relationship of the production or injection rate with any
I I.....-
geneous reservoirs, with and without induced hydraulic fractures. system parameter, such as tubing internal diameter, gas/oil ratio
(GOR), wellhead pressure, permeability, and skin among others. This
is accomplished by repeating IPR or intake curves in the system graph
6.4 Production-Systems Analysis
for different values of the sensitivity parameter. Once the parameter
Production-systems analysis is a simple engineering tool to couple is optimized, other parameters are considered. This also is called se-
IPR with the tubing intake, allowing the determination of surface pro- quential optimization.
duction rate through the whole production system. Mathematically, Sequential optimization in single-variant steps is a convenient way
such coupling allows the reservoir to produce fluids into the wellbore to learn the procedure and troubleshoot. However. multivariant opti-
and enables the piping system to lift these fluids to the separator at the mization-" enables one to determine the most profitable well design
surface. Additional pressure losses in the production system (e.g., configuration while considering a number of system parameters si-
through perforations, induced fractures, chokes, and the like) are ac- multaneously. A few example systems analyses with a computer pro-
counted for by combining these losses with the appropriate total-sys- gram called PERFORM25 will show their importance in optimized
\ tem losses, such as those in the tubing or reservoir system. production-systems design and troubleshooting. In the input-data
I
I This technique is used widely in the design, economic evaluation,
l-.
and troubleshooting of oil and gas wells. The graphical presentation
of the coupled IPR with the tubing-intake curves often is called the
systems graph. The intersection of these curves is the solution point
or natural flowing point and determines the producing rate and Pr
the pressure.
If these curves do not intersect, the well probably is loading up
and artificial-lift methods, such as gas lift or subsurface pumps, may
mitigate these problems. Fig. 6.14 elucidates this point graphically
before actual problems and sensitivity studies are discussed. This
figure shows a typical IPR curve with three hypothetical intake
curves. Intake Curve A does not intersect the IPR curve, indicating
this well's inability to produce to the surface. However, any artifi-
cial-lift method will lower this intake curve by either reducing the
total pressure loss in the tubing through aeration of the fluid column
or by adding a net positive pressure change, /)"P, to the fluid column
at the bottom of the tubing string by use of a subsurface pump. The
production rate improves to qpump as a result of the new pumped
well, Intake Curve C. Intake Curve B is the natural flowing curve
with production rate of qo. FIg. 6.14-Typical systems graph.
,
I
'-- 86
1
L 5,000
---5=-4 TABLE 6.12-INPUT DATA FOR EXAMPLE 6.7
4,500 - -5=12 Reservoir pressure, Pr, psia 3,200
-..---...- d= 1.5 in.
4,000
- - - d=O.824 in. Reservoir temperature, T" of 160
3,500 Permeability, k, md 1.5
,
1 Cl
'iii
3,000 Pay thickness, n, ft 20
a.
i 2,500 Skin: s -4, -2,0,10, and 17
\.- j
2,000 Wellbore radius, 'w, in. 3.5
1,500
Drainage area, 'eh, acres 160
\ IPR Darcy
1,000
\
I Perforation interval, ft 20
500
\ Perforation shot density, shots/ft 4
I
20 60 80 160 Tunnel diameter, Dp , in. 0.35
Oil Rate, STBO/O Tunnel length, L, in. 12
Fig. 6.15-Effects of tubing 10 and skin on production perform-
ance. kckt 0.5
i Porosity, tfJ, % 10
I-
80 Water cut, % o
Oil gravity, °API 25
70
Gas specific gravity, Yg 0.65
60
o GOR, scf/STBO 300
-,
0 50 Water specific gravity, Yw '1.03
aI
I-
C/)
40 Wellhead pressure, Pw, psig 50
'-- iIII 30 Perforation top, ft 8,840
a:
Wellhead temperature, Tw, of 70
I is 20 /8BOPD
I'- Tubing correlation Hagedorn and Brown
10
0 -- Casing 10, in.
Tubing 10, in.
5.012
2.75
-5 0 5 10 15 20 'Sensitivity parameter
Skin
Fig. 6.16-0iI rate sensitivity to total skin. combination of I.5-in. tubing ID and a suitable stimulation opti-
mizes the production system design.
tables for the following examples, the sensitivity parameters are indi-
cated by an asterisk. As mentioned earlier, optimum production refers Example 6.7-Production-Systems-Analysis Problem on Effect
to the maximum economic production. of Skin Improvement by Acidizing an Oil Well. This vertical well
produces with a beam pump installed 100 ft above the top perfora-
Example 6.6-Production-Systems-Analysis Problem on Effects tion. The pump suction pressure is 50 psig. Give a critical review of
of Skin and Thbing ID. This problem shows important effects of tub- skin sensitivity analysis and assume the skin caused by damage is
ing ID and skin damage on production performance of an oil well. 6 out of a total skin of 17.
Input Data. Table 6.11 gives the input data. Input Data. Table 6.12 provides the input data
Solution. The system graph presented in Fig. 6.15 shows that the Solution. On the basis of the sensitivity analysis presented in Fig.
damaged well with a high skin of 12 produces about 22 STBO/D 6.16, use of a very efficient matrix acid treatment to remove skin re-
I with both tubing sizes, although the well surges with the larger sulting from damage yields a skin of 11 with a very marginal in-
\.-
1.5-in. tubing ID. This unstable flow condition is very apparent crease in the production rate. Note that hydraulic fractures can re-
from the location of the point of intersection between the IPR and move damage and then impose some negative skin. Thus, if the total
skin is reduced to 5 or less through hydraulic fracturing, the well
the intake curves. For the I.5-in. tubing ID, the unstable flow condi-
production doubles to about 16 STBO/D, resulting in a more cost-
tion continues until production reaches about 60 STBO/D. This
effective treatment.
spread in the flow rate is enough to question the accuracy of the pres-
sure-loss correlation. However, even for the smaller 0.824-in. tub-
ing ID, the well would surge if the calculated rate was 18 STBO/D Example 6.8-Production-Systems-Analysis Problem on Effect
or less, The rate difference is so marginal that it may need further ofFracturing a Gas Well. This problem compares the transient rate
verification with another method of calculation. However, this well declines over a year for a fracture-stimulated well vs. a nonfractured
i offers excellent opportunity for stimulation. If the source of damage gas well.
is properly diagnosed, a matrix acid treatment or a fracture treatment
I
"- can be used to remove skin. A successful fracture stimulation treat-
Input Data. Table 6.13 provides the input data.
ment can reduce the skin to - 4 or less. Note that this stimulation Solution. This is a typical low-permeability gas reservoir with
effect improves the inflow performance to an extent that the poten- very little liquid production. The high reservoir pressure of 6,800
tial for surging with any size tubing is eliminated. Additionally, pro- psia makes it a perfect candidate for fracture stimulation. Fig. 6.17
duction will range from 89 to 107 STBO/D, depending upon the tub- gives rate-decline curves with transient IPR's for a homogeneous
ing size. The larger tubing size not only increases the production by reservoir and a finite-conductivity, vertical-fractured reservoir. To
, a factor of five but also completely stabilizes the production. The study the rate decline, time must be the sensitivity variable. Thus,
,I \
',--
WELL DESIGNAPPLICATIONS
I
L TABLE 6.13-INPUT DATA FOR EXAMPLE 6.8 1,200 r---------------------,
Reservoir pressure, Pr, psia 6,800 <,
Reservoir temperature,
Permeability, k, md
T" of 220
0.05
1,000
--- - .... -- .... .... ---- ..
Q 800
I
Pay thickness, h, ft 50 "":;:
:Ii
- - x,~500ft
_ _ x,~Oft
L Skin, S
Wellbore radius, 'w, in.
2.0
3.5
t
II:
• 600 ~
:I 400 ----------------
Drainage area, 'eh, acres 160 e
L IPR
Perforation interval, ft
Transient
50
200
culated by the Orkiszewski method and the higher value by the Aziz
et al. and Mukherjee and Brill methods. This is a discrepancy of Example 6.10-Production-Systems-Analysis Problem on Ef-
nearly 30%. The Ansari et al. method is based on mechanistic mod- fect of Water Cut. This problem is a typical water-cut-sensitivity
eling of vertical upward flow and comes within 5% of the Beggs and analysis in an oil well with three-phase oil, water, and gas produc-
Brill prediction, which is based on an empirical approach. The Or- tion. Understanding this problem should clarify any other sensitiv-
kiszewski and Hagedorn and Brown pressure-loss-calculation ity analysis where the same sensitivity parameter is repeated.
6,000 . . . - - - - - - - - - - - - - - - - - - - - - - - - ,
3.000·..--------------------,
5,000
4.000
Cl §
~
~ 3.000
~
(J)
Q.
~
2.000
-----IPR
- - - - - - WC~0.1
- - - - - WC=0.3
II:
6 1,00
- - - - - WC=0.5
1.000 WC =0.7
- - - - WC=0.9
----WC=O
oL- --- J,.... ....I 01-1-_....__---->--.........---+----<.----+--+---1
o 1,000 2.000 3.000 4,000 5,000 6,000
o 10 20 30 40 50 60 70 80 90 100
Oil Rate, STeO/D WaterCut. %
• • IG,=0.001 md 1.000 . - - - - - - - - - - - - - - - - - - - - - - ,
- - - - - IG,=0.01 md
kV=0.03 md
-
_-__- - _-
- - kV=0.1
k=0.1 mdmd
- - - - - Intake
~
II: 400
en
o'" 200
4,500
1,400
1,200 -
4,000 '5 1,000
a>
3,500 f0- 800
Ul
013,000 .
.~
~
a:
600
-e 400
12 ,500 's
Q.
2,000 .
:3" 200
0
1,500 . _ _ d=0.75 0.2 0.3 0.4 0.5 0.6 0.7 0.8
_d=0.0.5 Choke Diameter, in.
1,000· -d=0.375
L 500.
O.L.-
--d=IPR
. l - -.......- _ _..J
Fig. 6.21B-Choke-ID sensitivity.
Permeability (radial), k. md 30
A
Pay thickness, h, ft 20 3,000 ,/
, / .:«:
Skin,s 1.0 .;'--
Wellbore radius, 'w, in.
Drainage area, 'eh' acres
3.5
160
----- --~
/
,/
IPR ,/
DarcyNogel
Water cut, % 30
kckr 0.8
Fig. 6.22A-Production-systems graph for GLR sensitivity.
Porosity, r/J, % 10
Wellhead pressure, Pw, psig 100 1,200 , - - - - - - - - - - - - - - - - - - - - - ,
Pertoration top, ft 6,865 983 STBLJD
1.000r---'/..,---- _
Wellhead temperature, Tw, OF 70
Tubing correlation
Casing 10, in.
Tubing 10, in.
Duns and Ros
5.012
1.867
c
:::J
a>
f0-
Ul
~
a:
800
600
/
L 'Sensitivity parameter ~
400
"e-
.:J
6.5 Artificial Lift 6.5.1 Subsurface Pump. Fig. 6.23 presents a typical pressure tra-
The previous section introduced artificial-lift diagnostics based on verse for a downhole-pumped well. Without the pump this well is
systems analysis principles. This section presents the application of dead, with the fluid column at Point A in the tubing. A designed rate,
multiphase-flow principles in the design of artificial-lift methods. Ar- qo. and the corresponding bottomhole flowing pressure, Pw[, are
tificial-lift methods help manipulate the tubing-gradient curves for a identified from the IPR curve. With the same dead-oil gradient, the
predesigned bottornhole oil or liquid rate. q, with a designed flowing oil level in the annulus should be at Point B with the pump suction
A rTubinB Gn4iCIII
A
L
Pr Pr
at Point C. As Fig. 6.23 shows, the pump must be designed to pro- 6.6 Gas-Well Loading
vide a pressure boost of I1Ppump. This pressure difference is between Gas wells often produce a liquid phase, such as oil, condensates, or
the pump intake and the pump discharge. When the pump dis- even interstitial water. Depending on the phase behavior of the gas.
charges the fluid into the tubing at a rate, %' with the discharge pres- it is even conceivable that a well producing dry gas may have liquid
sure at the depth shown, the fluids flow to the surface at the specified in the wellbore. This problem can be particularly severe in conden-
wellhead pressure, Pwh. This tubing-gradient curve is generated by sate or retrograde condensate reservoirs. If the velocity of the pro-
use of one of the two-phase-flow models with the designed rate, %, ducing gas is sufficiently high, the wellbore liquids are produced to
and designed wellhead pressure, Pv.h. as in Example 6.9. the surface and the well does not accumulate any liquids. However,
in low-permeability gas wells, particularly with low reservoir pres-
6.5.2 Gas Lift. Gas-lift design is optimized based on a GLR sensitiv- sures, the gas velocity may not be sufficient to lift the liquid phase
ity study, as presented in Example 6.13 (Fig. 6.22B), where the opti- to the surface unless the tubing diameter is reduced to attain the liq-
mum GLR was 700 scf/STBL and the corresponding production rate uid unloading velocity. Lower velocities result in accumulation of
was 983 STBUD with a wellhead pressure of 100 psig. Theoretically, liquid in the well. This liquid accumulation eventually can create
gas lift aerates a liquid column with injected gas, allowing the liquid enough hydrostatic pressure in the wellbore to curtail gas produc-
to flow to the surface with a predesigned flowing wellhead pressure. tion severely, even completely stopping it with time. This phenome-
The gas is injected through a port in the gas-lift valve with a de- non is called gas-wellloading. 34-4 2 In oil wells with high liquid ve-
locity this may not be a problem. Artificial-lift methods can be used
signed differential pressure, I1p, at the port. 33 Fig. 6.24 presents an
to mitigate the problem. In gas-condensate reservoirs, particularly
example gas-lift pressure traverse in the production tubing. For sim-
in the presence of a large retrograde envelope, the effective perme-
plicity, this well also is assumed dead before the installation of gas
ability to gas can be severely reduced if condensation occurs near
lift. Note that the tubing is packed above the pay zone, which is a the wellbore. This may drastically reduce the gas velocity in the
precondition to a gas-lift installation that ensures injection into the wellbore, causing severe well-loading problems.
I tubing only through the valve port. If the oil is produced through the Gas-well-loading problems often can be diagnosed from produc-
L annulus, the tubing forms the gas-injection string and needs to be
blocked at the bottom.
tion-systems-analysis plots, as shown in Fig. 6.3. The intersection
of the IPR curve with the tubing-intake curve in the unstable zone
From the GLR sensitivity study in Example 6.13, the optimum is a good diagnostic of such problems, particularly if the proper PYT
rate, qo, is calculated for the optimum GLR. At that optimum rate, relationships given in Appendix B are used. To identify any problem
the bottomhole flowing pressure is obtained from the IPR curve. To with retrograde-condensation-induced loading, we highly recom-
calculate the fluid pressure at the point of injection (Point B), a mend the use of compositional models presented in Appendix C.
constant fluid gradient is calculated below the point of gas injection. This gas-well-loading problem can be mitigated by increasing the
L duced gas rate from the optimized gas rate that was calculated from
the sensitivity study for the optimum liquid rate and GLR. With the
available surface casing injection pressure, Pc, and the calculated
to achieve sufficient gas velocity. This permits complete unloading
of liquids from the wellbore, as discussed in Example 6.6. Normally,
intermittent gas lift and plunger lift (which is intermittent in nature)
are a second choice because of higher cost and associated operation-
gas injection rate through the annulus, the injection gradient or cas-
al problems. Plunger-lift efficiency also is very sensitive to the vol-
ing gradient is calculated. Any dynamic gas-gradient method can be
ume and pressure of gas in the annulus. The higher the volume and
used. Normally, all the multiphase pressure-loss methods revert to pressure, the greater the plunger-lift efficiency.
the single-phase-flow case based on superficial gas velocity. The Surface production rates and phase characteristics often can be
I
' --
upstream injection pressure is determined from the calculated cas-
ing gradient. The pressure difference, I1p, between the casing and
the tubing, at the point of injection, is then fixed. The port size of the
quite deceptive in predicting gas-well-loading problems. Histori-
cally, the prediction of the conditions conducive to this problem has
been very unreliable.P Turner et al. 36presented the best known pro-
gas-lift valve then can be determined with the Thornhill-Craver cedure to calculate the minimum stable-gas velocity. Unlike with
[ equation'[ or any type of choke equation for the passage of the de- the continuous film model presented earlier by Dukler.I? Turner et
t
'-- signed injection gas rate with the differential pressure, I1p. This al. showed that to unload a gas well. the largest droplet in the gas
should allow lifting the well at rate % to the surface at a predesigned stream must be carried to the surface. Dukler proposed a method to
wellhead pressure, Pwh. determine the minimum gas velocity required to move the liquid
L 92
!
--- TABLE 6.19-1NPUT DATA FOR EXAMPLE 6.14 To unload a well, the gas velocity has to exceed this terminal veloc-
ity at every point in the tubular until the liquid droplet reaches the sur-
Reservoir pressure, Pr, psia 4,000 face. Gregory35 reported an error in the coefficient of this equation in
Reservoir temperature, t; of 160 the original Turner et al. 36 paper, which was repeated in other refer-
ences to this equation. However, the equation presented here has been
Permeability: k, md 1.0,3.0
corrected and does not include the Tumer et al. arbitrary 16% upward
Pay thickness, n, ft 20 correction to this velocity. Coleman et al. 39 .4O concluded from field
I
'--
Skin, S
Wellbore radius, 'w, in.
o
3.0
tests that this 16% upward correction in terminal velocity is not
needed to calculate the minimum gas flow rate or critical gas rate
required to keep low-pressure gas wells unloaded. Based on the ter-
Drainage area, 'eh, acres 160 minal-velocity equation presented here, the critical gas flow rate,
L IPR
Completion
Darcy
Open hole
qg,' can be calculated as
(6.22)
Water cut, % o
Oil gravity, °API 54 where qg, = critical gas-flow rate, MMscflD; A = area of cross-sec-
tion of flow, ft 2; p = pressure, psia; T = temperature, OR; and Z = gas
Gas specific gravity, Yg 0.65
compressibility factor.
GOR, scf/STBO 150 Turner et al. suggested that wellhead conditions controlled the liq-
Water specific gravity, Yw 1.03 uid load up in most cases. This was substantiated by Coleman et at. 39
It is important to note that the critical unloading rate determined from
Wellhead pressure, Pw, psig 200
the liquid droplet model may not be applicable to slug flow.-?
Perforation top, ft 5,990 With the annular transition criteria described in Sec. 4.2.2 for the
Wellhead temperature, Tw, of 80 Ansari et al. 26mechanistic model, gas-well loading from downward
Tubing correlation flow of a liquid film on the tubing wall also can be considered. Gas-
Gray
well loading does not occur if annular flow is maintained throughout
Casing 10, in. 4.408 the wellbore. The similarity of Eq. 6.21 to Eq. 4.163, which predicts
Tubing ID: in. 0.824, 1.0, 1.6,2.0, 3.0 the transition to annular flow, suggests that gas-well loading will not
'Sensitivity parameter occur if annular flow is maintained throughout the wellbore.
film on the tubing wall continuously upward. The Turner et al. mod- Example 6.14-EtTect ofThbing ID on Gas-Well Unloading and
el is accepted as a more reliable method of liquid unloading. These Erosion. The best way to mitigate any flow-related problem is to
authors suggested that the minimum gas velocity in the well had to identify the potential problem before it manifests and design the sys-
equal the terminal velocity of the largest liquid droplet in the flow- tem around it. This example helps identify a gas-well-loading prob-
stream. For a spherical droplet, the terminal velocity is given by the lem and provides a method of solution. Velocity is the key to liquid
relation developed by Foust et al. 38 unloading. The higher the velocity is, the more efficient the unload-
ing. Velocity depends on permeability and pressure, which are con-
trolled by tubing !D. Consequently, a tubing !D sensitivity study
= [4(PL - pg)gdD]';' must be performed to optimize tubing size.
VI 3CvPg , (6.18)
Input Data. Table 6.19 gives the input data.
where VI = terminal velocity, ft/sec: PL= liquid density, Ibrnlft 3; Solution. In this example, two gas wells are studied: one with low
P g = gas density, lbm/ft': g = gravitational acceleration, 32.174 permeability and a suspected loading problem and one with higher
ft/sec-: do = droplet diameter, ft: and CD = drag coefficient for permeability and a suspected erosion problem. A tubing ID sensitiv-
the droplet. ity analysis is performed. A modified Turner unloading rate is used.
The maximum stable droplet diameter can be determined from The gas-well-unloading and erosion-rate envelopes are superim-
the relation 35 posed in Fig. 6.25 with the tubing-ID sensitivity plots for two differ-
ent permeabilities.
v;pgdD As long as the actual predicted production rates are higher than the
-0- == 30, (6.19)
L corresponding unloading rate, the wells are unloaded. In Fig. 6.25, the
gas rates for all the tubing !D's considered in the I-md permeability
where OL = surface tension, lbm/sec.I
From Eq. 6.19,
6.000 , . - - - - - - - - - - - - - - - - - - - - - - - ,
30a L
dD = -1-'
V"iP g 5,000
__ k=1 md
which can be substituted into Eq. 6.18 to obtain _._. k=0.3 md
o 4.000 ___ Erosion
_ Unloading
13VI
. . . . . . . . . . . . . . . . . . . .. (6.20)
*
::;
a:
gj
3,000
Assuming spherical droplets, Turner et al. showed that the drag co- C) 2,000
efficient for the droplet, CD, is 0.44. Substituting this value of CD
r..........
and converting the units of surface tension, 0, to dynes/em, the Turn-
1.000
er et al. drop-terminal-velocity equation can be written as
where VI is in ftlsec, is in dynes/em, and P Land P g are in Ibrnlft3. Fig. 6.25-Gas-well-loading erosion envelopes.
L
OL
Stag~n:8J
Erosion is the physical removal of pipe material in contact with a
L flowing fluid. Erosion limits the life of flow strings. Continuous ex-
cessive erosion often leads to mechanical failure, leakage, or both.
x /
v
L illustrates the use of Eq. 6.23 to mitigate pipe erosion. This problem
uses a value of C= 100. In Fig. 6.25, as long as the producing gas
rates are lower than the erosion envelope, the problem should not
where h = erosion penetration rate, rnil/yr; qsd = sand production
rate, ft 31D; vp = particle impact velocity, ft/sec: T= elbow metal
hardness, psi; and d = elbow diameter, in.
exist. For multi phase mixtures, the erosional velocity is compared Salama and Venkatesh obtained this expression for erosional ve-
with the mixture velocity in a pipe and the no-slip mixture density locity in ftJsec. Assuming particle impact velocity equals flow
is used in Eq. 6.23. For the higher permeability of I md, erosion stream velocity in Eq. 6.25.
problems are predicted to occur for tubing ID's of less than 2 in.
Salama and Venkateshv' reviewed the API equation for erosional v, = 1.73d/ ~. . (6.26)
velocity and concluded that the form can be justified from a Ber-
This equation assumes hardness T = 1.55 x 105 psi and an allow-
noulli-type relationship. However, assuming a 3,000- to 5,000-psi
able penetration rate of h = 10 mil/year.
maximum pressure-drop range in the flow conduit for clean two- However, Salama and Venkatesh suggested that the applicability
phase fluids, the predicted Ve is very conservative. Allowing for a of Eq. 6.26 be limited to gas flow only, noting that the particle-im-
10 mil/yr erosion rate, they recommended the following relation-
pact velocity in gas flows (with low density and viscosity) nearly
L ship to calculate the erosional velocity caused by liquid impinge-
equals the flow-stream velocity. This assumption is not valid for liq-
ment on a steel surface.
uid flow, where the particle-impact velocity is lower than the flow-
300 stream velocity. In this case, the sand flow rate, qsd, must be adjusted
v, = IP' (6.24) for the number of particles actually impinging on the pipe wall. This
adjustment is necessary because with high fluid density and viscos-
This equation seems to be more reasonable for limiting the de- ity most of the sand grains may pass through the core of the elbow
signed fluid velocity in tubing to control any erosional damage to the without impinging on the wall, thus reducing erosion. This flow-re-
L pipe. In producing wells, this results in curtailment of production and,
thus, affects the economics. It is often more economic to produce
wells at optimum rates, allowing for some acceptable penetration
gime dependence of erosion calls for consideration of fluid proper-
ties, such as the density, viscosity, and composition.
Shirazi et al. 45presented a new, more rigorous mechanistic model
rates resulting from erosion. Because Eqs. 6.23 and 6.24 are indepen- that accounts for a number of variables, including flow geometry,
dent of penetration rate, they cannot be used for such calculations. In pipe size, pipe material, sand density, particle size, sand-particle
L fact, the only parameter considered in these equations is density, sug- sharpness, flow-stream velocity, fluid viscosity, and fluid density.
gesting that the erosional velocity can be increased by reducing the The model to compute the maximum penetration rate in elbows for
density of the flowing fluid or decreased by increasing the density of carbon-steel material is
the flowing fluid. This does not agree with experimental observations
for sand-laden fluids. Sand in gases with lower densities causes more h = AFsFp [ qsdPpv~73 /(B 0.59d2 )] , ...••..•••••••••• (6.27)
erosion than sand in liquids with higher densities.i"
The presence of sand in the flowing fluid results in erosion dam- where A = empirical constant (0.9125 for carbon steel); B = Brinell
I
'--
age caused by abrasive wear mechanism. 46.47 Rabinowicz,48.49 has
shown that the volume of ductile metal eroded from the surface be-
hardness; Fp = penetration factor for steel based on l-in. pipe diam-
eter, in.zlbm: F, = sand sharpness factor (empirical); and PP= par-
cause of solid particle impingement is directly proportional to the ticle density, Ibm Ift 3 .
total weight of impinging solids. Thus, in the case of pipe flow, el- The other variables were defined previously. This equation is
bows will be more susceptible to erosion because more solids will based on extensive empirical information gathered by Shirazi et
impinge on the projected area of the incoming pipe. It is esti- al. 45 and others. 52,53 Unlike the previous equations, this one is more
mated 50,51 that for low-density gas systems, 100% of these solids general because vp is the true characteristic impact velocity on the
will impinge on the pipe. As the density increases, this percentage pipe wall and not the flow-stream velocity. Consequently, this ve-
2,000
.
.!
a.
! 1.soo
L
Temperature
--
I
To avoid or mitigate these problems,
I. Determine the geothermal and hydrothermal gradients before
4. At or near subsurface safety valves or other choke devices
where Joule-Thompson cooling is present.
the pipeline design. 5. During cold startup of producing wells after shut-ins for well
2. Determine the flowing temperature gradient in the well or pipe- testing, workover, or routine surface-facility maintenance.
line before the pipeline design. Gas-hydrate phase behavior and kinetics are well understood.
3. Determine the phase behavior of the flowing fluids in the oper- SIoan 61 presented a very comprehensive review of these processes.
ating minimum and maximum pressure and temperature range ex- Formation of solid gas hydrates in oil and natural gas wells or trans-
perienced between the reservoir and first stage of separator. mission pipelines'v causes physical blockages to flow and is a major
4. After combining the information in the first three steps, genera-
nuisance. When designing gas or multiphase-flow pipelines in gas-
tion of a gas-hydrate envelope in a pressure/temperature phase dia-
hydrate-prone systems, consider the phase behavior of gas hydrates
gram, as shown in Fig. 6.29, can identify pipe segments ideal for hy-
and, ifpossible, avoid the formation of gas hydrates or hydrate plug-
drate formation during cold startup. Similar phase envelopes can be
drawn for wax deposition. ging. If a hydrate-free system cannot be designed, hydrate-related
5. Displace the pipe segments identified in Step 4 with chemical blockage in the pipeline should be considered while troubleshooting
inhibitors, such as diesel, methanol, etc., during any shutdown. flow or production problems, especially after shutdowns or during
6. To mitigate these problems, the oil and gas industry uses prop- cold startups.
erly insulated pipeline segments or hot oil pumping when the prob-
Inhibition/Dissociation. The common method of hydrate inhibi-
lem occurs. Economics permitting, permanent chemical injection
I tion is by inducing thermodynamic instability to the hydrate phase,
lines also are used for mitigation.
I also known as thermodynamic inhibition. This is accomplished by
'- the following method.
6.8.2 Gas Hydrates. The formation of gas hydrates is another very
common problem, particularly in a high-pressure and low-tempera- • Heating the system above the hydrate-formation temperature at
L ture operation. 60 -64 Gas hydrates (also called c1athrates for encag-
ing) are crystalline compounds that result when water forms a cage-
like molecular structure around very specific smaller guest
system pressure.
• Reducing the system pressure below hydrate stability at system
temperature.
• Changing composition or removing one of the components of
molecules. They are commonly composed of water and a combina-
tion of methane, ethane, propane, normal butane, iso-butane, nitro- hydrate, such as the hydrocarbon or the water.
gen, carbon dioxide, and hydrogen sulfide.P! • Adding or injecting an inhibitor, such as methanol or glycol, to
Gas hydrates are crystalline solid compounds resembling dirty ice the hydrate system to reduce its stability at the system pressure and
in appearance. Unlike ice, however, hydrates form anywhere water temperature.
and the guest hydrocarbon molecules come in contact at temperatures The injection of inhibitors moves the hydrate stability conditions
above or below 32°F and at elevated pressures dictated by the com- toward higher pressures and lower temperatures. 65 .65 New inhibi-
position-dependent phase behavior. Being solid, hydrates are a nui-
sance in flowing wells or pipelines, where they can totally or partially
I block fluid passage. Subsurface restrictions, such as chokes or safety TABLE 6.21-EXAMPLE OF GAS COMPOSITION
L valves, are particularly susceptible to hydrate blockage even during
normal operation because of Joule-Thompson cooling.
Component Mole Fraction
Hydrates also concentrate hydrocarbons. One cubic foot of hy- Methane 0.6741
drate may contain as much as 180 scf ofgas. In high-pressure oil and Ethane 0.0842
gas wells, hydrates often form during the startup phase. Although Propane 0.0460
the very low temperature of a permafrost environment provides the lso-butane 0.0063
perfect condition for gas-hydrate formation, they are found any- n-butane 0.0171
where in the world, especially in high-pressure and marine environ-
ments. Ideal situations for gas-hydrate formation include Iso-pentane 0.0051
I. High-pressure gas wells or flowlines, where the gas contains n-pentane 0.0077
moisture or free water or load-water produced during post-fracture Hexane 0.0093
cleanup. Heptane 0.0677
2. In marine pipelines, riser pipes, and high-pressure oil and gas Nitrogen 0.0024
wells.
C02 0.0801
3. In arctic wells, especially near the permafrost zone.
L
envelope with respect to a typical North Sea gas-phase envelope. 21. Cineo-L, H., Samaniego-Y., E, and Domingues-A., N.: "Transient
Table 6.21 gives the gas composition. Pressure Behavior for a Well With a Finite-Conductivity Vertical Frac-
A phase envelope for this gas in a feed stream with 1,000 Ibm- ture," SPEl (August 1978) 253.
mol/hr dry gas and 2lbm-mol/hr water was generated with the Peng- 22. Meng, H.Z. and Brown, K.E.: "Coupling of Production Forecasting,
Robinson equation of state from Appendix C and is presented in Fig. Fracture Geometry Requirements, and Treatment Scheduling in the
6.29. Note that the hydrate equilibrium line (Line A) in the middle Optimum Hydraulic Fracture Design," paper SPE 16435 presented at
of the gas/liquid phase envelope determines the thermodynamic sta- the 1987 SPE Low Permeability Reservoirs Symposium, Denver, Colo-
bility conditions of pressure and temperature. The area to the left of rado, 18-20 May.
this curve provides the ideal stable-gas-hydrate envelope. If the 23. AI-Hussainy, R. and Ramey, H.J. Jr.: "Application of Real Gas Flow
pressure and temperature in this gas well lie on the left side of this Theory to Well Testing and Deliverability Forecasting," lPT (May
envelope during cold startup, hydrates may cause blockage to the 1966) 637; Trans., AIME, 237.
line. Injection of inhibitor moves this curve to the left, requiring 24. Carroll, J.A. III and Horne, R.N.: "Multivariate Optimization of Pro-
higher pressure and/or lower temperature to form hydrates. With the duction Systems," lPT (July 1992) 782; Trans., AIME, 293.
25. "'PERFOR:vt' Well Performance Analysis Program," Dwight's Ener-
same gas/water stream, injection of 3.607 Ibm-mol/hr, 80 wt%
gydata Inc., Dallas (1995).
methanol in water moves the hydrate equilibrium line to the left, as
26. Ansari, A.M. et al.: "A Comprehensive Mechanistic Model for Two-Phase
shown by Line B. This shows a little shrinkage of the hydrate-phase
Row in Wellbores.' SPEPF (May 1994) 143; Trans., AIME, 297.
envelope with the addition of methanol. The extent of control of the 27. Hagedorn, A.R. and Brown, K.E.: "Experimental Study of Pressure
hydrate-phase envelope depends on the composition of the produc- Gradients Occurring During Continuous Two-Phase Flow in Small-Di-
ing hydrocarbons and the type and amount of hydrate inhibitor add- ameter Vertical Conduits," lPT(April 1965) 475; Trans., AIME, 234.
ed. Thus, an understanding of such hydrate-phase behavior forms 28.0rkiszewski, J.: "Predicting Two-Phase Pressure Drops in Vertical
the basis of its thermodynamic inhibition and is highly recom- Pipes," JPT(1une 1967) 829; Trans., AIME, 240.
i
I.....
mended for troubleshooting and solving hydrate-related problems. 29. Beggs, H.D. and Brill, J.P.: "A Study of Two-Phase Flow in Inclined
Pipes," lPT(May 1973) 607; Trans., AIME, 255.
References 30. Mukherjee, H. and Brill, J.P.: "Pressure Drop Correlations for Inclined
Two-Phase Flow," 1. Energy Res. Tech. (December 1985) 107,549.
I. Versluys, J: "Mathematical Development of the Theory of Flowing Oil
31. Aziz, K., Govier, G.W., and Fogarasi, M.: "Pressure Drop in Wells Pro-
Wells," Trans., AI ME (1930) 86, 192.
ducing Oil and Gas," 1. Cdn. Pel. Tech. (July-September 1972) 11,38.
2. Gilbert, W.E.: "Flowing and Gas-Lift Well Performance," Drill. &
32. Ashford, F.E. and Pierce, P.E.: "Determining Multiphase Pressure
Prod. Prac. (1954) 126.
Drops and Flow Capacities in Down-Hole Safety Valves," lPT (Sep-
I 3. Nind, T.E.W.: Principles of Oil Well Production, McGraw-Hill Book
I tember 1975) 1145.
Co. Inc., New York City (1964).
'-- 33. Mach. J.M. et al.: "A New Concept in Continuous-Flow Gas-Lift De-
4. Proano, E.A., Mach, J.M .. and Brown, K.E.: "Systems Analysis as Ap-
sign," SPEl (December 1983) 885; Trans., AIME, 275.
plied to Producing Wells," paper 3.1 presented at the 1979 Congreso
34. Lea. J.E: "Avoid Premature Liquid Loading in Tight Gas Wells Using
Panamericano de Ingenieria del Petroleo, Mexico City, 19-23 March.
j Prefrac and Postfrac Test Data," Oil & Gas 1. (20 September 1982) 123.
.... 5. Crouch, E.e. and Pack, K.J.: "Systems Analysis Use for the Design and
Evaluation of High-Rate Gas Wells," paper SPE 9424 presented at the
35. Gregory, G.A.: "Comments on the Prediction of Minimum Unloading
Velocities for Wet Gas Wells," Technical Note No. 14, Neotechnology
1980 SPE Annual Technical Conference and Exhibition, Dallas, 21-24
September. Consultants Ltd., Calgary (December 1989).
6. Mach, J.M., Proano, E.A., and Brown, K.E.: "Application of Produc- 36. Turner, R.G., Hubbard, M.G., and Dukler, A.E.: "Analysis and Predic-
tion System Analysis to Determine Completion Sensitivity on Gas Well tion of Minimum Flow Rate for the Continuous Removal of Liquids
Production," paper presented at the 1981 ASME Energy Sources Tech- From Gas Wells," lPT (November 1969) 1475; Trans., AIME, 246.
nical Conference and Exhibition. Houston, 18-22 January. 37. Dukler, A.E.: "Fluid Mechanics and Heat Transfer in Vertical Falling
7. Mach, J.M.: "Apply NODALTM Analysis to Production Systems," Well Film Systems," Chem. Eng. Prog. (1960) 56.
Servicing (JanuarylFebruary 1981) 38. 38. Foust, A.S. et al.: Principles of Unit Operations, second edition, John
8. Brown, K.E. et af.: "Production Optimization of Oil and Gas Wells by Wiley & Sons, New York City (1980).
NODALTM Systems Analysis," Technology ofArtificial Lift Methods, 39. Coleman, S.B. et af.: "A New Look at Predicting Gas-Well Loadup,'
Penn Well Publishing Co., Tulsa, Oklahoma (1984) Chap. 4. lPT (March 1991) 329; Trans., AIME, 291.
9. Golan, M. and Whitson, C.H.: Well Performance, Prentice-Hall Book 40. Coleman, S.B. et al.: "Understanding Gas-Well Load-Up Behavior,"
Co. Inc., Englewood Cliffs, New Jersey (1991). lPT (March 1991) 334; Trans., AIME, 291.
10. Darcy, H.: Les fontaines publiques de la ville de Dijon (1856). 41. Oudeman, P.: "Improved Prediction of Wet-Gas-Well Performance,"
II. Muskat, M.: The Flow ofHomogeneous Fluids Through Porous Media, SPEPE (August 1990) 212; Trans., AIME, 289.
McGraw-Hill Book Co. Inc., New York City (1937); reprinted by SPE, 42. Upchurch, E.R.: "Expanding the Range for Predicting Critical Row
Richardson, Texas (1982); copyrighted by IntI. Human Resources De- Rates of Gas Wells Producing From Normally Pressured Waterdrive
velopment Corp., Boston (1982). Reservoirs," SPEPE (August 1989) 321; Trans., AIME, 287.
1 12. Earlougher, Re. Jr.: Advances in Well Test Analysis, Monograph Se- 43. RP 14E. Recommended Practicefor Design and Installation ofOffshore
Production Platform Piping System, third edition, API, Washington,
i ries, SPE, Richardson, Texas (1977) 5.
DC (1981).
'-- 13. Matthews, e.S. and Russell, D.G.: Pressure Buildup and Flow Tests in
Wells. Monograph Series, SPE. Richardson, Texas (1967) 1. 44. Salama, M.M. and Venkatesh, E.S.: "Evaluation of API RP14E, Ero-
14. Dietz, D.N.: "Determination of Average Reservoir Pressure From sional Velocity Limitations for Offshore Wells," paper SPE 1253 I pres-
Build-Up Surveys," lPT (August 1965) 955; Trans., AIME, 234, ented at the 1983 Offshore Technology Conference, Houston, 2-5 May.
15. Ahmed, T.H.: Hydrocarbon Phase Behavior, Gulf Publishing Co., 45. Shirazi, S.A. et al.: "Generalization of the API RP14E Guideline for
Houston (1989). Erosive Services," lPT(August 1995) 693.
16. Vogel, J.Y.: "Inflow Performance Relationships for Solution-Gas Drive 46. Thompson, T.L. and Aude, T.e.: "Slurry Pipeline Design and Operation
Wells," lPT(January 1968) 83; Trans., AIME, 243. Pitfalls to Avoid," Trans., ASME (1976).
I
I
'--
I
'--
I
I MULTIPHASE FLOW IN WEII.s
'- 98
IL.
I
!
\-
I Appendix A
\-
Nomenclature and
81 Metric Conversion Factors
!I
\-
a= abscissa F= friction geometry parameter for annulus
\
A= Helmholtz energy, mL 2/t2 F= moles of feed mixture, mol
I
I A= pipe cross-sectional area, L2 F= formation resistivity factor
L. A= well drainage area, L2 FE= liquid entrainment fraction, L3/L3
A= parameter defined in Eq. 4.196 Fp= penetration factor, Urn
Bg= formation volume factor of gas, L3/L3 Frp = relative performance factor in Eq. 4.264
Bo= formation volume factor of oil, L3/L3 Fs= sand sharpness factor
B(= total formation volume factor, L3/L3 g= acceleration of gravity, Ut2
Bw= formation volume factor of water, L 3/LJ 8c= gravitational conversion factor
c= compressibility, L 2/m so» geothermal temperature gradient, T/L
i c= wellbore storage coefficient, L4t 2/m h= convective film coefficient, m1t3T
I
\- C= correction factor h= specific enthalpy, L2/t2
C= flow coefficient in Eqs. 5.1 and 5.2 h= net reservoir thickness, L
C= liquid-holdup parameter in Eq. 4.117 h= erosion penetration rate, Ut
C= specific heat, L2/t2T hL= liquid interface height, L
'-- CD= discharge coefficient, drag coefficient H= enthalpy, mL 2/t2
CFD = dimensionless fracture conductivity HL= slip gas/liquid volume fraction (liquid holdup), L 31L3
CK.CM = film indices in Eq. 4.294 HLF= liquid holdup from liquid film, L3/L3
CL = specific heat of liquid, L2/t2T HR= holdup ratio in Eq. 4.140
Co = flow coefficient in Eqs. 4.240 and 4.241 i= injection rate, L 3/t
Cp = specific heat at constant pressure, L 2/t2T lani= permeability anisotropy
I Cv = specific heat at constant volume, L 2/t2T J= mechanical equivalent of heat
i
thermal conductivity, mUt3T
--
\
j
d=
D=
D=
pipe diameter, L
dimensionless parameter in Eqs, 5.7 and 5.9
turbulence coefficient
k=
k=
k=
ratio of specific heats, Cp/Cv
effective permeability, L2
DBC = distance between pipe centers, L K= pipe diameter ratio for annulus, LIL
e= eccentricity of annulus, LIL K= piping component factor in Eq, 5.33
e= intrinsic specific energy, L2/t2 K= equilibrium constant, mol/mol
e= error Kw= Watson characterization factor
er= relative error K'= power-law consistency index in Eq. 2.29
1
£k= dimensionless kinetic-energy pressure gradient in L= pipe length, L
'-- Eqs. 4.9 and 4.53 L= liquid mole fraction, mol/mol
£1= average percent error in Eq. 4.256 L= moles of liquid, mol
£2= absolute average percent error in Eq. 4.258 L= horizontal well length, L
£3= percent standard deviation in Eq. 4.259 LB= parameter in Eq. 4.59
'-
£4= average error in Eq. 4.260 4= latent heat of vaporization, L2/t2
£5= absolute average error in Eq. 4.262 m= mass, m
£6= standard deviation in Eq. 4.263 m= parameter in Eq. 4.96
j= Moody or Darcy-Weisbach friction factor m(p)= real-gas pseudopressure function
j= no-slip volume fraction, L3/L3 M= molecular weight, m
f= Fanning friction factor M= volumetric heat capacity, m1Lt2T
j= fugacity, m1Lt2 n= number of moles
'-
NnMFNri ATI'RF 00
IL- n= parameter in Eqs, 5.9 and 5.10 x.. = drainage boundary length, L
n= polytropic expansion exponent in Eq. 5.13 xf= fracture half length, L
n,..=
number of components XM= Lockhart and Martinelli parameter in Eq. 4.167
n'= power-law flow behavior index in Eq. 2.29 y= mole fraction of component in vapor phase
'-
n'= bubble-swarm parameter in Eq. 4.177 y= parameter in Eq. 4.124
N= dimensionless parameter in Eqs. 5.7 and 5.8 y= parameter in Eq. 4.132
Nd= pipe-diameter number in Eq, 4.5 y= ratio of downstream to upstream pressure
1 ND= dimensionless diameter numbers in Yc= critical pressure ratio
L Eqs. 4.13 and 5.27 Ye= drainage boundary width, L
NE= dimensionless number in Eq. 4.97 YM= Lockhart and Martinelli parameter in Eq. 4.168
NFr= Froude number in Eq. 4.109 7=
~ mole fraction of component in feed mixture, mol/mol
Ngv= gas-velocity number in Eq. 4.4 z= parameter in Eq. 4.134
NGr= Grashof number in Eq. 2.93 z= gas compressibility factor, L 31L3
NL= liquid-viscosity number in Eq. 4.6 z= vertical distance, L
NLC= corrected liquid-vicosity number in Figs. 4.2 and 4.3 z= interfacial friction parameter in Eqs. 4.221 and 4.222
NLv= liquid-velocity number in Eq. 4.3 Q= thermal diffusivity, L2/t
'"- NNu= Nusselt number in Example 2.4 Q= void fraction
Npl= dimensionless pressure number in Eq. 5.26 {3= coefficient of thermal expansion, Iff
I Npr= PrandtI number in Eq. 2.94 {3= ratio of Taylor bubble to slug unit length
tI NqL= dimensionless liquid flow rate number in Eq. 5.28 0= subtended angle by interface chord
""- NRe= Reynolds number ~= dimensionless film thickness
Nv= dimensionless velocity number in Eq. 4.12 ~n= minimum dimensionless film thickness
NWe= Weber number in Eq. 4.47 °L= liquid film thickness, L
Nx= flow-pattern coordinate in Eq. 4.85 and Fig. 4.14, Ut L1= difference
I...,...
Ny= flow-pattern coordinate in Eq. 4.86 and Fig. 4.14, Ut e= absolute pipe roughness, L
Nil = liquid viscosity number in Eq. 4.48 e= convergence tolerance
Nv= dimensionless number in Eq. 4.98 ¢= parameter in Eqs. 2.53 and 2.103
N=p dimensionless density ratio in Eq. 5.25 ¢= parameter in Eqs. 4.231 and 4.232
"'- p= pressure, mlLt 2 ¢= porosity
PD= dimensionless wellbore pressure drop <1>= fugacity coefficient
\ p= perimeter, L y= specific gravity
i
I
Pch = parachor YAPI= oil gravity, 0 API
'- q= volumetric flow rate, L3/t y= shear rate, 1ft
q= heat flow rate, mL2/t3 r= liquid distribution coefficient in Eqs. 4.72 through
qmax= absolute-open-flow potential, L3/t 4.81
qsc = volumetric flow rate at standard conditions, L3/t 1]= louIe-Thompson coefficient, TLt2/m
Q= heat flux, mlt 3 A= no-slip gas/liquid volume fraction, L31L3
r= radius, L A= thermal conductivity, mUt 3
I
rd= drainage radius, L ~= absolute viscosity, mILt
L rw = well radius, L
R= annulus concentricity parameter in Eq. 2.63, LIL
y= kinematic velocity, L2/t
0= inclination angle from horizontal
R= gas constant, mL2/t2T 0= reduced boiling point
R= gas/oil ratio, L31L3 p= density, mlL 3
R= in-situ gas/liquid volume ratio, L3/L3 a= surface tension, mlt 2
Rp= producing gas/oil ratio, L31L3 a= percent standard deviation
s= skin factor r= shear stress, mlLt 2
s= parameter in Eqs. 4.122, 4.123, and 4.125 w= Pitzer acentric factor
s= slip velocity number 1J!= parameter in Fig. 4.4
s= surface area, L2 1J1= liquid-holdup inclination angle correction in Eq.
s= saturation 4.116
s= entropy, mL 2/t2T
t= time, t Subscripts
'-
T= temperature, T I= upstream conditions
T= metal hardness, m/Lt? 2= downstream conditions
u= specific internal energy, L2/t2 a= air, atmospheric, apparent
:..- U= overall heat-transfer coefficient, mlt 3T acc= acceleration
v= velocity, Ut an= annulus
y • = sonic velocity, Ut b= bubble, boiling point, bulk
vI = effective sonic velocity, Ut bf= bubble in flowing liquid
I..- Vs = specific volume, L3/m bs= bubble in static liquid
v= vapor mole fraction, mol/mol B/5= bubble/slug boundary
v= volume, L3 c= casing, circular, critical
v= moles of vapor, mol calc = calculated
""- w= mass flow rate, mit cern = cement
w= fracture width, L ch= choke
w;= angle of view average in Eqs. 4.335 and 4.336, rad ci= inside casing
W;= interface chord length, L co= outside casing
x= mass fraction crit= critical
x= mole fraction of component in liquid phase, mol/mol C= calculated, corrected, gas core
x= parameter in Eq. 4.129 CA= concentric annulus
,
i
'-
I
L NOMENCLATURE
\
Appendix B
Fluid and Rock Properties
8.1 Introduction wide, the use of laboratory data measured on representative samples
Integration of the pressure-gradient equation requires determina- taken from the reservoir is strongly recommend whenever available.
tion of individual phase velocities, densities, viscosities, and, in Most of the fluid and rock properties included in this appendix have
some cases, surface tension at different pressures and temperatures. been reported previously.l> However, we decided to incorporate
In dynamic conditions of multiphase flow in pipes, the pressure and some of these materials, in some cases verbatim or with minor
t temperature of the fluids change continuously. Consequently, very changes, and to provide supplementary information when necessary.
I active mass transfer occurs between the liquid and the gas phases.
L As the pressure decreases in the direction of flow below the bubble- 8.2 Hydrocarbon Physical Properties
point, gas evolves from solution in the oil, increasing the gas veloc-
Crude oils and natural gases contain several pure organic and inor-
ity and the oil density and viscosity. Such flow and fluid property
changes can be predicted with either compositional or black-oil ganic chemical components, each with its own properties. Table
models. The compositional models presented in Appendix C can B-1 presents selected physical properties of several of these individ-
represent the mass-transfer phenomena more accurately and are ap- ual components. Refs. 6 and 7 present more complete data.
plicable universally. Table B-1 shows a very wide range of molecular weights, boiling
This appendix presents methods to determine fluid physical prop- points, phase densities, and critical temperatures and pressures
erties required in the black-oil model. The assumptions in the black- among the components constituting reservoir fluids (crude oils or
oil model are that, at any fixed temperature, pressure, API gravity natural gases). Clearly, the physical properties of reservoir fluids are
of the liquid phase and specific gravity of gas, the liquid phase has pressure- and temperature-dependent. The critical pressure and
a fixed gas solubility and formation volume factor. These assump- temperature of a multicomponent mixture are called pseudocritical
tions imply that the composition of the oil and gas do not change pressure, Ppc- and pseudocritical temperature, 'Fpc, respectively. In
with pressure and temperature. This constant composition assump- this appendix, these pseudocritical properties are used as correlating
tion may be valid for the oil phase, but can cause errors in predicting parameters in many empirical correlations. If the mixture composi-
I
the physical properties of the gas phase. tion is known, these quantities may be estimated from
'-- Gas solubility is defined as the volume of gas dissolved per unit
volume ofliquid at a fixed temperature and pressure. Formation vol-
ume factor determines the change in the volume of a phase at differ- r.; = i>jTcj (B-1)
ent pressures and temperatures. This appendix presents a detailed j=1
Most black-oil-model-related fluid physical properties!·28 can be where TIc = number of components in the mixture; Yj = mole fraction
determined by use of pressure/volume/temperature (PV1) cells.
of Componentj; Tcj = critical temperature of Componentj, OR, and
These are often called PVT properties. This appendix presents the
Pcj = critical pressure of Componentj, psia.
most widely used and reliable fluid and rock physical property cor-
If the system composition is not known, Figs. B-1 through B-3
relations. These correlations are very useful in determining the fluid
and rock physical properties required in the application of the multi- can be used to estimate Tpc andppc. Fig. B-1 provides a way to esti-
phase flow theories presented in this monograph. However, the cor- mate these quantities for undersaturated oil at reservoir pressure; the
relations should be used with proper caution. Most of these correla- oil specific gravity corrected to 60°F (the value normally reported)
tions are empirical in nature and are based on a limited quantity of is used. "The specific gravity of a petroleum oil and of mixtures of
representative samples of data. Some of these PVT properties are de- petroleum products with other substances is the ratio of the weight
veloped from reservoir fluid samples belonging to particular geo- of a given volume of the material at a temperature of 60°F ... to the
graphical areas and may not be applicable in other areas where reser- weight of an equal volume of distilled water at the same tempera-
voir fluid composition is very different. Although PVT property ture, both weights being corrected for the buoyancy of air."9 This
correlations are very widely used in the petroleum industry world- often is referred to as Yo at 60°F/60°F. Sometimes the weight of the
oil is not determined at 60°F, in which case the specific gravity is With gas in solution, oil physical properties also depend on gas sol-
either corrected to 60°F to give Yo at 60°F/60°F or is reported as Yo ubility, in addition to the pressure, temperature, and API gravity of oil.
at the given temperature. The pressure at which the liquids are
weighed is not specified. In Fig. B-1, the specific gravity at reservoir B.3.1 Gas Solubility. Gas solubility is defined as the volume of gas
pressure corrected to 60°F is the ratio of the oil density at reservoir dissolved in one stock tank barrel of oil at a fixed pressure and tem-
pressure and temperature corrected to 60°F to the density of distilled perature. Gas solubility in oil increases as the pressure increases, up
water at 60°F and I atm. Fig. B-2 applies to bubblepoint liquids by to the bubblepoint pressure of the oil. Above the bubblepoint pres-
use of the specific gravity corrected to 60°F. Fig. B-3 applies to con- sure, gas solubility stays constant (Fig. B-4) and crude oil is often
densate well fluids and natural gases. To use Fig. B-3, the gas gravity called undersaturated. In black-oil models, gas solubility deter-
must be known. Tpc and Ppc normally are used to estimate the mines the mass transfer between the liquid and gas phases.
pseudoreduced temperature, Tp r , and pressure, Ppr- This parameter fails to predict any retrograde condensation ef-
fects. For light oils or condensates, where retrograde condensation
T pr = Jpc
(B-3)
and i!?
:>
1.300 ~...,..'t""I"'''I'''''Ir-r..,..I'"T''''''''''''''''''''I'''''I''''''''I'"T'''''I'"T''''''T'''l''
~ 1,200
Ppr -
- Ppc'
P ••••••.•••••••••••••••••••••••••••• (B-4)
a.
E
~a: 1,100
_0
~.~
'E
500
!t 400
r--------__ Pressure
, gQ.
141.5 -g 300
!I..- YAPI = y;;- - 131.5, (B-5)
~ 200 a........................_ ........._ ..............._ ....._ ..... ..J
with Yo at 60°F. In Eq. B-5, Yo = specific gravity of stock tank or 0.62 0.65 0.70 0.74 0.78 0.82 0.86
dead oil at 60°F/60°F. If the oil gravity is reported at other than 60°F, Spedic Gravity of Undersaturated Reservoir Liquid
at Reservoir Pressure Corrected to 60°F
it can be corrected to 60°F by use of the technique described in Ref.
9. (In Ref. 9, Table 5 is used for hydrometer measurements at other
than 60°F; Table 7 allows the correction of volume at a given tem- Fig. B-1-Approxlmate correlation of liquid pseudocrltlcal pres-
\ perature to volume at 60°F.) The API gravity of water is 10. sure and temperature with specific gravity (after Trube 8).
I
L
FLUID AND ROCK PROPERTIES 103
1()4
8 700 .,..,., ...,.., ""'.,..,., ""'.,..,.,~
6
Micellaneous G
ases
Conde
4 nsate We" FiW;;;.__
Specific Gravity of Underaaturated
il "- Reaervolr Liquid at 60°F, y
a.
~, 1\ '\
u:
0
\ \j
I.-
s
OJ '"1'-
1\.r\1I\
'\
1\ 1\ 1\ 'i
,
,i ..e
::> D6l 2?~>..>.
. '(;'.
0 0 0 0 0 0 0 0
.>.>..~.~ 'ct-
0 -i' V 6' ~ .'§.
L- f 1()3
a..
C 8
'0 500
6
a.
CII
:is !
.tl
:::l
m 4
~CII
\ a.
E
CIIlX:
1\ ~ 0. 400
2
~ ~
E .....
:;
o
"C
l " ,..
tablishing equilibrium with the liquid phase. Fig. B-4 compares gas
solubility with these two types of vaporization processes. s, = CIYgIOOPCzexp[ C3(T :A~60) (B-8)
Several empirical correlations to determine gas solubility are
presented. They assume a flash-vaporization process. where the values for the coefficients are presented in Table B-2.
The gas specific gravity, Yg 100, refers to a separator pressure of
Standing Correlation. The Standing correlation-! states
100 psig and can be calculated from
1.2048
R s = Yg [(L
18.2 + I .4)IOo.oI2~API-0.0009IT] ..... (B-6) YglOO = Y«( 1.0 + 5.912 x 10
-5 Psep )
YAPITsepiog 114.7 '
0.989 ) 1.2255
YAPI •
where Yo = specific gravity of oil; Mo = effective molecular weight s, = Yg ( TO.172 v» (B-IO)
of tank oil obtained from Fig. B-6; and Yg = mole fraction of gas, a
correlating parameter obtained from Fig. B-7. where T is in OF and P; = 1O{2.8869-[14.1811-3.3093Iog(p)j0.5I.
11M
voir to surface conditions. Thus, the formation volume factor is al-
ways equal to or greater than 1.0. A mathematical definition is
Differential
(Vo)p.T
B; = -(V) , (B-14)
o sc
I
• above the bubblepoint pressure, the oil stops dissolving more gas
and the formation volume factor decreases because of the compress-
I•
ibility of the liquid. Fig. B-8 describes this behavior. It is important
to note that the oil formation volume factors are determined by use
I of different equations above and below the bubblepoint pressure.
Consequently, it also is imperative to determine the bubblepoint
'Pb pressure before the oil formation volume factor is determined.
Pressure There are different correlations to calculate the oil formation vol-
ume factor and the bubblepoint pressure. A few of the more com-
monly used correlations are presented here. The bubblepoint pres-
Fig. B-4-ldealized comparison of flash and differential gas sol-
ubilities. sures are calculated from the solution gas/oil ratio correlations given
in Sec. B.3.1 by setting R, equal to the producing gas/oil ratio, Rp ,
and solving for the corresponding pressure.
The accuracy of this correlation is believed to decrease for solu-
tion gas/oil ratios exceeding 1,400 scf/STBO.18 Analysis of the ex- Standing Correlation. This is the oldest and most commonly
pression for P; shows that the correlation cannot be used for pres- used empirical correlation.U It was developed primarily from
sures in excess of 19,285 psi. California crude oils. The Standing correlation can be expressed in
r. . .
this mathematical form below the bubblepoint pressure.
Al-Marhoun Correlation. AI-Marhoun 15 presented this correla-
L
Required: Bubblepoint pressure at 200°F of liquid with a
gas/oil ratio (GOR) of 350 ft3/bbl, a gas gravity of 0.75,
and a tank-oil gravity of 30 0API.
Bubblepoint
Fig. B-S-Properties of natural mixtures of hydrocarbon gas and liquids at bubblepoint pressure (after Standing 21 ).
TABLE B-3-COEFFICIENTS FOR THE VAZQUEZ AND TABLE B-4-COEFFICIENT VALUES FROM VAZQUEZ AND
BEGGS CORRELATION13 BEGGS MODIFICATION13
where a= -C3 YAPI/T. Table B-4 tabulates the values of coeffi- Glase also presented a bubblepoint-pressure correlation and sug-
cients C]. C z, and C3 in Eq. B-20. gested a procedure to account for the presence of nonhydrocarbon
content of the crude oil system. The proposed bubblepoint-pressure
Glast' Correlation. The Standing!" PVT correlations are widely correlation 19 can be expressed as
used in the oil industry, although they are based primarily on
California crude oils. These correlations do not correct for other oil log P» = ,
1.7669 + 1.7447 log Pb - 0.30218 log Pb •
(,)2
types or nonhydrocarbon content. Glase!" modified the Standing
................... (B-22a)
IO(i
'\ . , 'f \ T" ••• "r--' • ~
55
J
•. 4
I 50
L 0
I
"0
45
~
1-
I
40 It..
s
'>
...co 35
e"
u
0 30
~
e
4.. oc;.
...co
1
25
4..
20
,, E
15 '"-c
;...
J
''0
100 200 300 500 ~
0 I.,
ti
01
LL
E
'0
• (Rs)a c
Pb = Yg T'/YAPI, ......................... (B-22b)
0.
Q>
:0
2..
.t:J
:>
al
where Rs = gas solubility, scf/STBO; T= temperature, OF; Yg = av- 2.4
erage specific gravity of the total surface gases, and ab,c = coeffi-
cients with respective values of 0.816, 0.172, and 0.989. For volatile
oils, b = 0.130. For nonhydrocarbon content, Glase suggested these 2.0
corrections to the bubblepoint pressure, Pb.
a2 = 5.5 X 10- 3 , Fig. B·7-8ubblepolnt pressure factor vs. gas mole fraction
(after Lasater12).
a3 = 0.0391,
Pressure, psia
~ [O.0315~,',W~;""~"'I>"~.]
below the bubblepoint pressure that can be expressed mathemati-
cally as p, .... (B-29b)
j
B; = 0.98496 + 0.000IFI.50, (8-27)
I""- Note that Eqs. 8-29a and 8-29b are identical to Eqs. B-12a and
where F = R~·755Y~i~y;1.50 + 0.45T. where T=temperature. of; B-12b. but solve for p with R, equal to the produced gas/oil ratio,
Yg 100 = specific gravity of gas at 100 psig, and Yo = specific gravity Rp • These correlations were developed with nonlinear regression
of stock-tank oil. This formation volume factor is recommended for analysis.
the flash-vaporization process normally encountered in pipe flow. For
application to reservoir flow. such as in inflow performance relation- B.3.3 Oil Density. The density for saturated crude oils below the
ship (IPR), a differential-vaporization formation volume factor bubblepoint pressure can be calculated at a given presssure and tem-
Example
ReqUired: Formation volume at 200°F of bubblepointliquid
1
1--.
I
1
Fig. B-9-Chart to determine oil formation volume factor with the Standing correlation. 21
2,000
\
1.3 1,000 \
Co BOO
u
BOO
6 400
CIl
1.2 "0
:::J
U
200 1\\ \
"
~
CIl
CIl
100
80
\\
"
-
II
~
1.1 It
"'
<II
60
40 ~~
....
~
e>
Cl
0 20
'\."\.
"
'\."-
,"-. <; ~
.
~
0
~
III
0
o
III
10
8
6
4
" ~
-::~
~
5 <, /."§,;
<,"O'b:, -........J <,
~ roo
"0
III
.0 1.0
<, <, -<; ' .... ......
-c 0.8
-- --
0,6
0.4
0.2
0.1
10 15 20 25 30 35 40 45 50 55 60 65
62.4yu + 0.0136R,Y gd where produced gas/oil ratio, Rp , is calculated as the gas solubility
p, = B; (B-30)
at or above the bubblepoint pressure. In gas/oil, two-phase-flow cal-
culations, the physical properties of the free gas, such as density and
where Yo = stock-tank oil specific gravity and Ygd = dissolved-gas
viscosity, should be calculated based on the free-gas gravity, Ygf, as
gravity described in Sec. B.3.4.
functions of pressure and temperature, which changes the gas solu-
For saturated oils above the bubblepoint, oil density can be calcu-
bility and the specific gravity of the dissolved gas. Most commercial
lated by first determining the density at the bubblepoint pressure,
computer programs fail to account for the changing composition of
62.4yu + 0.0 136RpYgt the free and dissolved gas as pressure and temperature change .
I
i.- Pob = B . . . . . . . . . . . . . . . . " (B-31)
ob
B.3.5 Oil Viscosity. The viscosity of crude oil with dissolved gas is
Then, given the oil compressibility, Eq. B-32 can be used for density an important parameter in pressure-loss calculations for flow in
whenp>pb' pipes or in porous media. Whenever possible, the oil viscosity
should be determined in the laboratory for the required pressure and
Po = Pobeco(P-Pb) . . (B-32) lemperature ranges. This section presents empirical correlations to
calculate the oil viscosity based on frequently available hydrocar-
bon-system parameters, such as temperature, pressure, oil gravity,
B.3.4 Specific Gravity of Free and Dissolved Gas. In thermody-
gas gravity, and gas solubility.
namically stable mixtures of oil and gas, as the equilibrium is dis-
turbed with change in pressure under isothermal conditions, the
B.3.5.1 Dead-Oil Viscosity Correlations. For empirical correla-
e ~~~.=J
o
----
1.000 2.000 1.000 •. 000 ).000
I
\-
Q.
c: Pressure. psi
0Q,
.,
:is 1.3 Detail
J:J
:l
.,
III 1.2
>
0
J:J
c(
(5
.,
"0
Q,
u
0
2 i-
0;
'0 .
o
u
j
......
~
0;
0
u
..
s., L----- ,~-- --- s
.. 1-_-------- >-- - - -
:; 10
10
0
J:J
c(
0
~_.-.-_------ ) - - - - --
0.00 1
2 3 4 5
0 )00 1.000 1.)00) 2.000 2.)00 1,000 1.)00 •. 0Q0;l •• lO\l
Undersaturated Pressure. p = p - Pb. psia Pressure. 1.000 psia
I
'-
Fig. 8-12-Variation of oil viscosity with pressure. 20
I.... Beggs and Robinson Correlation. Beggs and Robinson 22 pro- B.3.5.2 Saturated Crude Oil Viscosity. The reservoir oil viscosity
posed a different empirical correlation to determine the dead-oil vis- depends on the solution-gas content. Oil viscosity decreases with
cosity, It is based on 460 dead-oil viscosity measurements and can rising pressure as the solution gas increases, up to the bubblepoint
be expressed mathematically as pressure. There are few empirical correlations to determine the vis-
cosity of saturated or undersaturated crude oil systems.
).lod = lOX - I, ., , (B-37)
Beggs and Robinson Correlation. This correlation-? is based on
where 2,073 saturated oil viscosity measurements. The empirical form of
this equation is
10(3.0324 - 0.02023 YAPI)
x = .:;.'------~
TI.163
u; = [IO.715(R J + 100) -0515]).l:d' ...••••••••... (B-40)
and T is in "F,
where
Glast' Correlation. Glasel ! presented an empirical correlation
based on North Sea data. Sutton and Farshadlf found this correla-
b = 5.44 (Rs + 150) -0.338.
tion to be the most accurate dead-oil viscosity correlation among the This correlation was developed from these ranges of data:
three described here. This correlation can be expressed as pressure = 132 to 5,265 psia,
temperature = 70 to 295°F.
and
. . . . . . . . . . . . . . . • . . .. (B-39) y = 10-o.ooo8IR,.
E c:
~
c:
o
U
~ 30 ~
-g 12 8 8
t:> 26 ;!!.
24
22
L 20 0~--:,:",-~:---~-~:--~~-~""':::...... ~--1
10 20 30 40 50 80
Oil Gravity at 60°F, °API
Saturation Pressure, psia
Fig. B-13-Surface tension of crude oils at atmospheric pres-
sure (after Baker and Swerdloff23). Fig. B-14-Correction to dead-oil surface tension. 23
B.3.5.3 Undersaturated Crude Oil Viscosity. Above the bubble- o; = acJx/100), (B-44)
point pressure, rising pressure increases the viscosity of oil because
of its compressibility. Fig. B-12 presents this viscosity vs. pressure where x = percent correction from Fig. B-14.
relationship.
8.4 Water Physical Properties
Vazquez and Beggs Correlation. Vazquez and Beggs 13 proposed
1 Water is often a very important liquid component in the oil and gas
this correction to the saturated crude oil viscosity at the bubblepoint
I
-- pressure for pressures above the bubblepoint pressure.
_ (p)m
production system.I? Consequently, the physical properties of water
play an important role in multiphase flow calculations. Among the
physical properties of water, gas solubility, formation volume factor,
#0 - #ob Pb ' .............................. (B-42)
viscosity, surface tension, and compressibility are of particular inter-
where est. Specific gravity or density of water is also an important property
and is normally readily available from an oilfield water analysis.
m = 2.6 p1.l87 lOa
i B.4.1 Water Density. The density of pure water at standard condi-
and
~ tions is 62.4lbm/ft3. Neglecting gas solubility in water, the density
of water can be calculated from
This correlation is based on data in the following ranges: P.sc 62.4ywsc
pressure = 141 to 9,515 psia, p; = s:: = -s:;-' (B-45)
gas solubility = 90.3 to 2,199 scf/STBO,
viscosity = 0.117 to 148 cp, where p ; = density of water at any pressure and temperature, Ibm/
gas specific gravity = 0.511 to 1.351, and ft 3; Pwsc = density of water at standard conditions, Ibm/ft 3;
oil gravity = 15.3 to 59,SO API. Bw = formation volume factor of water; and Ywsc = specific gravity
of water at standard conditions.
Kartoatmodjo and Schmidt Correlation. This correlationl'' al-
lows correction of the saturated crude oil viscosity at the bubble-
I point, #ob' based on Eq. B-41 for undersaturated pressure, p.
B.4.2 Gas Solubility in Water. The solubility of hydrocarbon gas
I'"- components is inversely proportional to their molecular weights.
#0 = 1.00081#ob + 0.001127(p - Pb) Thus, methane is more soluble than ethane, ethane is more soluble
than propane, and so forth. McCain 27 states that the solubility of
each component is two to three times greater than that of the next
x (- 000651711
.
18148
rob
+ 0038.11
.
1.590)
rob ' ........ , (B-43) heavier paraffinic component. Methane is the most soluble compo-
nent of natural gas in water. The solubility of methane in water can
where #0 = viscosity of undersaturated oil, cp.
be used to estimate the solubility of natural gas in pure water with
B.3.6 Surface Tension. Surface tension 23-26 is a measure of the im- an accuracy of5% or better. Fig. B-15 can be used to estimate meth-
balance in the interfacial molecular forces for two phases in contact. ane solubility in pure water. 28
In multiphase pipe flow, gas/liquid and IiquidJIiquid surface or in- To account for the effect of water salinity, Dodson and Standing-!
terfacial tension values are used to determine flow patterns and liq- suggested the corrections presented in Fig. B-16.
I\ Ahrnedl? recommends the use of this gas-solubility correlation
uid holdup.
l- The empirical work of Baker and Swerdloff,23 presented in Fig. for water.
B-13, commonly is used to estimate the surface tension of crude oil
at atmospheric pressure (dead oil), aod. In this figure, the surface R sw = A + Bp + Cp 2 , (B-46)
tension of dead oil is correlated vs. temperature and the API gravity where A=2.12+(3.45 x 1O- 3)T-(3.59x IO- S)T2, B=0.0107
of the oil. Because the precise effects of temperature on dead-oil sur- -(5.26 XIO- s)T+(1.48XIO- 7)T2, and C= -(8.75 x 10- 7)
face tension is unknown, extrapolation beyond the data presented in + (3.9 x 10 - 9) T - (1.02 x 10 - 11)T2. The gas solubility is cor-
Fig. B-13 is not recommended.
rected for the effect of water salinity.
The surface tension of live oil with dissolved gas, ao , can be ob-
tained with an appropriate correction to aod for the dissolved gas. (B-47)
Fig. B-14 presents this correction to the surface tension of the dead
oil as a percentage reduction of the dead-oil surface tension, as a where T is in OF and (Rsw)b = gas solubility in brine, scf/STBW;
function of pressure. The effect of increasing pressure is to increase Rsw = gas solubility in pure water; Cs = salinity correction factor
gas solubility, thus decreasing surface tension. The effect of pres- = 1.0 - [0.0753 - 0.000 173 nS, where S = salinity of water, weight
sure on surface tension of oil with dissolved gas can be determined percent of NaC!.
from Eq. 8-44. Ref. 19 is the source of this gas-solubility correlation.
111
FLUID AND ROCK PROPERTIES
04 ical polynomial relation to determine the formation volume factor
of water.
--
estimated from Fig. B-17. 31 The top graph yields Cw for pure water
-- -- ~Cf!J.--
10
...........
...... "- i-- .ft
while the bottom graph is used to determine a correction factor for
!--J" Cw to correct for the effects of gas solubility.
-<,
B
on
Meehan 32 presented an empirical correlation to determine the
r- .. ' 1O isothermal compressibility coefficient of gas-free and gas-saturated
1 '- III water. For gas-free water. this empirical relationship is proposed.
I 0 "
'-' 80 100 140 180 220 2eO 340 (CW)f = 1O-6(A + BT + CT 2 ) . • •••.••...•••••.•. (B-50)
Temperature, of
where (cw}c= isothermal compressibility coefficient of the gas-free wa-
Fig. B-15-Solubility of methane in pure water (after Culberson ter,psi- 1; A=3.8546-0.<XXlI34p; B= -0.01052+(4.77 x 1O-7 )p ;
and McKena28 ). C= (3.9267 x 10- 5) - (8.8 x 1O- 10)p; p = pressure, psia; and T=tem-
perature. oF.
j
\ B.4.3 Water Formation Volume Factor. Because gas solubility is
I
i..- very low in water compared to that in oil, it is common practice to
4.0 ~.......,..-r-'T'""'....,-r..,..-r-T"""~r--r-r...,..-r-,....,......
neglect compressibility and gas solubility of water. Assuming negli-
gible thermal expansion, water formation volume factor is assumed
to be 1.0. In actual field situations, water shows gas solubility and.
-I
Ui
Q,
)(
i:
Q
.
~
~
!
::J
Q,
.5
"0
~
'0III
III
C Temperature, OF
III
l'Cl
~C o 1.3r-r"""I""T'T"'l~T"T.,...r-r""M"'''''''''''''T'"T-r-r-T'T'''1
.-: ~~
I1 m - IX Correction for Gas in Solution
=>0
l- .5 :!:!:t::
"0 :l:s
~ 11-
"'1/1
Q,I/I
'0 Ell
III
III OD.
C °E
co 1.1
III 00
l'Cl ;: ...
:::J1l
CJ "01ij
(1)3=
O.t50'------.....-
10C,COO
......-~--'
......-200.000 300.000 5 10 15 20 25
TDS, ppm GaslWater Ratio, ft 3lbbl
Fig. B-16-Effect of salinity on gas solubility: TDS = total dis- Fig. B-17-Compressibility of pure water, Including effects of
solved solids (after Dodson and Standing 31). gas in solution (after Dodson and Standing 31).
I
\.....
112 MULTIPHASE FLOW IN WELl S
I
I
'--
1.6
~
'iii
0
I o
en 1.5
'- s
-
~
~
cu
as
cu
~
1.4
::J I.
---
Q.
>-
'iii
Critical Point
0
o
en
scu 700 BO(
Temperature. OF
e
'i: Fig. B-19-Kinematic viscosity of steam and water as a function
m of temperature (after Prats 5 ).
8 12 16 20 24
uid phase. Thus, actual laboratory viscosity data should be used in
TDS,% this case.
Fig. B-18-Ratio of brine viscosity to pure water viscosity vs. sa- B.4.5 Surface Tension of Water. HocoU36 and Hough et al. 37 in-
linity (after Frick35 ). vestigated the surface tension of water/gas systems. Katz et al. 26
combined their works into a single diagram (Fig. B-21).
Meehan also proposed a correlation to determine the isothermal Although these results,26,36,37 suggest that all experiments have
compressibility of gas-saturated water. been valid, their approach is highly questionable in its prediction of
water surface tension as a function of such variables as water salinity
(c,.J g = (cw}j (1.0 + 8.9 x 1Q-3R,w), (B-5la) and gas composition. Nevertheless, from the work of Hough et al.
one can linearly interpolate between the curves for 74 OF and 280°F
where (cw}g = isothermal compressibility coefficient of the gas-sat-
and obtain acceptable estimates of the surface tension of water.
urated water, psi - I, and Rsw = gas solubility in water, scf/STBW.
To account for the salinity of water, this adjustment is suggested.
B.5 Gas Physical Properties
I c; = (cw)gC S ' •..•••.••.••••..•.••••••.•.... (B-5Ib) A gas is defined as a homogeneous fluid oflow density and viscosity
L where Cw = isothermal compressibility coefficient of the brine,
without a definite volume. Gas occupies the volume of its container
without regard to shape or size. According to the kinetic theory of
psi - I, and Cs = salinity correction factor. gases, gas is composed of a large number of molecules of insignifi-
Numbere et al. 33 proposed this mathematical expression for the cant volume compared to the total volume of the container. These
salinity correction factor. molecules are assumed to have no attractive or repulsive forces be-
tween them, and they collide at random. These intermolecular colli-
Cs = 1.0 + (- 0.052 + 2.7 x 1O-4T - 1.14 x 1O-6T2
sions are perfectly elastic. An equation of state is a mathematical
+ 1.121 X 1Q-9T3)SO.7, (B-52) relationship between pressure, volume, and temperature for a fixed
i mass of gas and is expressed mathematically for n moles of gas as
'--
where S = salinity of the water, weight percent of NaCI, and
pV = nRT. . .. , , ,., " (B-54)
T= temperature, "F,
This equation is called the ideal gas law, where p = absolute pres-
B.4,4 Water Viscosity. Data on the viscosity of oilfield water are sure, psia; V= volume of gas (gas container), ft3; T= absolute tem-
very scarce. Water viscosity increases with pressure and dissolved perature, OR; n = number of moles of gas; m = mass of gas, Ibm;
solids and decreases with increased gas solubility. No data are avail- M = molecular weight of gas; and R = universal gas constant.
able on the effects of gas solubility. However, van Wingen-" re- Hydrocarbons or natural gases are not pure or single-component
ported the effect of temperature on water viscosity. His empirical gases. These are mixtures of multicomponent gases and are called
correlation relating water viscosity with temperature in OF is pres- real gases. For real gases, the basic assumptions of negligible inter-
ented as molecular forces and insignificant volume of molecules compared
Temperature
-LB.......
40 to 120 1 5
120 to 212 5 5
212 to 400 10 5
Q.
U
.....::t
I ~
....
\.- ::J
en
en
~ 100 100 too 400
c,
T, "F
.!,2
CD
s:
c,
Pressure-correction factor. '.
en 1.1 for water vs. T ("F). presumed
0
E applicable to brines but not
«
"0
1.0
confirmed experimentally.
c: Viscosity at elevated pressure
co
e
::J
0..
IJ p,r =lJ r'p.r
~
Q) 0..
Q.
I E
i\.- ~
Q)
0.7 ViSCOSity,1'''
at 1-atm pressure at
less than 212°F. at saturation of
'0
~ 0.. / pressure of water at more than 212°F.
Q)
en
Q)
a: 0.5
a;
~ 0.4
en
0
u
en
s o.~
L 0..
0.1
1
! 0.0 ':40:--::eo!=--:eo~---IOO~~I
':--,~40:--,~eo:--::_~-:aao~-:'UO~...".MO~-Z~eo':--::z~eo~-::.co~-:'~~ZO:-~~~40:-':'MO~-S8Cl~--"!400
...
'- Temperature, of
Fig. B-2D-Water viscosities at various salinities and temperatures (after Matthews and RusseIl 1) .
viously. The specific gravity of gas, Yg, is defined as the ratio of the Thus, the specific gravity of gas is
gas density to that of air, Pa, at standard conditions, expressed math-
ematically as Mg
Ys = 28.96' (B-58)
.......................... (B-57) Gas density in Ibm/ft 3 can be determined easily by combining the
real gas law with the definition of specific gravity of gas.
The molecular weight of air, Ma , is 28.96 and its specific gravity is I.
(B-59)
1,?
!
iI
1.6"
I.-
O.B
14S
1.4
0.7
Us
'"
~O_6
I.J
1.5
'j"
L
a
US
'-- ~
~
...
:0
.~
~O.:i
E
r.
1.4
.a
:i
E
L- a
u a
1.\ u
I.Z
L
1.1
1.0
Jon.or, I, ,g41
ns
11 12 13 14 15
'--
Fig. B-22-Real-gas deviation factor for natural gases as a function of pseudoreduced pres-
sure and temperature.t"
to represent the properties of pure gas mixtures, such as hydrocar- After determining the pseudocritical pressure and temperature,
bon or natural gases. Standing-! presented a set of empirical equa- Eqs. 8-3 and 8-4 can be used to calculate the reduced pressure and
tions to determine the pseudocritical pressure and temperature to temperature. Then, the gas deviation factor can be obtained from
approximate the curves in Fig. B-3. Fig. B-22.
Several empirical correlations are available, representing the
Case 1: Natural Gas Systems.
Standing and Katz u Z-factor nomogram (Fig. B-22). For ease of
Tpc = 168 + 325Yg - l2.5y~ (B-60) calculation with computer programs, these mathematically ex-
pressed correlations are convenient for Z-factor determination. The
and Dranchuk and Abu-Kassem correlation'? is presented for the sake
of completeness. It is an l l-constant empirical equation used to fit
Ppc = 677 + I5.Dyg - 37.5Yi. (B-6I)
Z-factor curves in Fig. B-22, and can be expressed as
Case 2: Gas Condensate Systems.
and
AI = 0.3265, 8.5.5 Gas Viscosity. The viscosity of a fluid is defined as the ratio
of shear stress to shear rate. The commonly used oilfield unit for
A 2 = - 1.0700, viscosity is centipoise, where I poise is defined as a viscosity of
A 3 = - 0.5339, I dyne-s/cm-.
Carr et al.4 1 presented a widely used method to estimate natural
A 4 = 0.01569, gas viscosity. It requires knowledge of the gas composition and the
As = - 0.05165, viscosity of each component at atmospheric pressure and in-situ
temperature. The viscosity of the gas mixture at atmospheric pres-
A 6 = 0.5475, sure is estimated from
A 7 = - 0.7361, n
As = 0.1844,
LYjJ4j~
j=1
J4 ga = - n- - - - .......................... (B-70)
A 9 = 0.1056,
LYj/M;
AID = 0.6134, )=1
and where J4ga = viscosity of the gas mixture at the desired temperature
and atmospheric pressure, cp; n = number of components in the gas;
All = 0.7210. Yj = mole fraction of Cornponentj.rz, = viscosity of Componentj at
Eq. B-64 is implicit in Z and must be solved by some iterative the desired temperature and atmospheric pressure; and Mj = molec-
method, such as the Newton-Raphson iteration technique. This cor- ular weight of Component j from Table B-1.
relation represents the Standing and Katz Z-factor chart within If the gas composition is not known. Fig. 8-23 can be used with
j
0.585% average absolute error and is applicable over the ranges, the gas molecular weight or gas gravity to estimate the gas viscosity
IL.. 0.2 $,Ppr < 30 and 1.0 $, Tp r $, 3.0. Although presented for Ppr up to at reservoir temperature and atmospheric pressure. Molecular
15, the Standing and Katz Z-factor chart can be extrapolated beyond weight is related to gas gravity by
that reduced pressure because of the linear relationship of Z vs. Ppr (B-71)
I Mg = 28.97Yg·
at constant Tpr for Ppr > 15.
L 8.5.3 Correction for Nonhydrocarbons. Standing and Katz devel-
The gas viscosity at reservoir pressure is estimated by determining
the ratio J4g/J4ga at the appropriate temperature and pressure from Fig.
oped their Z-factor correlation based on mixtures of hydrocarbon 8-24. Then, that ratio is applied to J4ga. obtained from either Eq. B-70
gases with molecular weights less than 40. Natural gases often con- or Fig. B-23. The pseudoreduced temperatures and pressures for use
tain nonhydrocarbon components, such as carbon dioxide, nitrogen, in Fig. B-24 are estimated from Eqs. B-3 and B-4 and Fig. B-3.
and hydrogen sulfide. At nonhydrocarbon gas-content values below In the presence of nonhydrogen gases, such as nitrogen, carbon
I 5%, there is negligible effect on the Z factor. Higher concentrations dioxide, and hydrogen sulfide, the gas viscosity at atmospheric
Il..- of nonhydrocarbon gases can cause substantial error when calculat- pressure and desired temperature must be corrected by use of the in-
ing the compressibility factor and, thus, require correction. Wichert serts in Fig. B-23.
and Aziz40 presented a simple gas compressibility correction proce- Lee et al. 42 presented a semiempirical equation to calculate gas
dure to compensate for the presence of hydrogen sulfide and carbon viscosity. This equation, which cannot be used for sour gases, can
dioxide in natural gases. This method suggests the following adjust- be presented as
ments to the pseudocritical properties used to determine the Z factor
X(~~4fl
from the Standing and Katz chart.
4Kex
J4g = 1O- p[ (B-72)
T/x: = Tpc - e (B-66)
where 'Fpc = pseudocritical temperature, OR; Ppc = pseudocritical 8.5,6 Gas Compressibility. Isothermal gas compressibility can be
pressure, psia; fpc = corrected pseudocritical temperature, OR; defined as the change in volume per unit volume of gas for a unit
P;'" = corrected pseudocritical pressure, psia; YH 2S = mole fraction change in pressure, 13 expressed mathematically as
of hydrogen sulfide in the gas mixture; and e = pseudocritical tem-
~(~~t·
perature adjustment factor, defined mathematically by
(B-73)
Cg = -
With the real gas law to replace volume in the previous equations
where the Coefficient A is the sum of the mole fractions of hydrogen and after proper differentiation,
sulfide and carbon dioxide in the gas mixture or
(B-68b)
Cg = ~ - ~(~L, (B-74)
so 60 70
Molecular Weight
where Z= gas deviation factor at absolute pressure, p, in psia and Oil Formation Volume Factor (from Vazquez and Beggs).
absolute temperature, T, in "R. Note that for an ideal gas, Z = I, and
From Eg. B-18,
_ I
cg - p' (B-75) B; = 1.0 + 4.677 x 10-4(281)
Trube 44presented a correlation to estimate gas compressibility.
He defined isothermal gas compressibility, cg , as the ratio of
+ (180 - 60)(0.~~2)[ 1.1 x 10- 5 + 1.337 x 10- 9(28 1)]
pseudoreduced compressibility, Cpr- to the pseudocritical pressure,
1.0 + 0.131 + 0.067
Ppc- as
1.197 bbIlSTBO.
Cpr
Cg = Ppc' •••••••••..•••••••.•.••••••••••••. (B-76) Specific Gravity of Dissolved Gas (from Katz).
To estimate the gas compressibility, Trube presented correlations to From Fig. B-lO, Ygd = 0.88.
estimate the pseudoreduced compressibility as a function of pseudo-
Specific Gravity of Free Gas.
reduced pressure and temperature (Figs. B-25 and B-26). Note that
these two correlations are similar. They present pseudoreduced com- From Eg. B-35,
pressibility at two different ranges of compressibility values.
= (1,000)(0.75) - (281)(0.88) = 070
Ygf 1, 000 - 281 . .
Example B.I-Calculate Black-Oil-Model Mass-Transfer Pa- Oil Density.
rameters and Physical Properties. A crude oil and natural gas mix-
ture is flowing through a wellbore. Determine the black-oil-model From Eg. B-5,
mass-transfer parameters and the physical properties of the gas and 141.5
oil phases at a location in the well where the pressure is 1,700 psia and Yo = 131.5 + 33 = 0.86.
the temperature is 180°F. Also given are Yg = 0.75 at Psep = 14.7 psia
and Tsep = 60°F, YAPI = 33 0, and Rp = 1,000 scflSTBO. From Eg. B-30,
Solution Gas/Oil Ratio (from Vazquez and Beggs). (62.4)(0.86) + (0.0136)(281 )(0.88)
Po = 1.197
From Eg. B-9,
53.66 + 3.36
1.197
YglOO = yg{ 1.0 + 5.912 x 1O-5(33)(60)IOg[I\~:7J} = 47.61 lbm/tt".
= (0.75)(0.896)
Pseudocritical Pressure and Temperature (from Standing).
From Eg. B-61,
= 0.672.
From Eq, B-8 for YAPI> 30°, Ppc = 677 + (15)(0.7) - (37.5)(0.7)2
= 669 psia.
R, = (0.0178)(0.672)( I, 700) 1187
From Eg. B-60,
X ex p[ (23. 931)C 80 ~ 460) ] Tpc = 168 + (325)(0.7) - (12.5)(0.7)2
= 281 scf/STBO. = 389°R.
""
I
'--
LO
a.
ClII
II
el7
" 1/ U Pseudoreduced
_'I./ ..4 Temperature
\ "l\-u
"' ; ,
1I
\ 1.0
\' ~
\ \ \ ~
L 1\ \
\ I'
\ 'V ~
1\ I\~ ~
z
\l'\
"\ \
1/ u
1\ 1\ ~1/ V',.
\\ 1\ \\~~ /'
I.O},I.I,
'"\1\ I\~ ~K J
L o.1
z 4
I\'~
671'10
I
'--
and
b = 5.44(281 + 150)-0338 = 0.7.
From Eg. B-37,
Mg = (28.97)(0.7) = 20.28Ibm/lbm-mol.
+ 460 = I 65
T pr = 180389 "
From Eg. B-4, K = [9.4 + (0.02)(20,28)](640) 1.5
128,7.
209 + (19)(20.28) + 640
L Ppr =
1,700
669 = 2.54.
and
Z=0.853.
Y = 2,4 - (0.2)(5.243) = 1.35 I.
Gas Formation Volume Factor. From Eg. B-72,
From Eg. B-69,
L =
Gas Density.
0.0091 ft 3/scf.
Surface Tension (from Baker and Swerdloff).
From Fig. B-13,
_ (2.7)(0.7)(1,700) _ 3
From Fig. B-14,
pg - (0.853)(180 + 460) - 5.88 Ibmlft . %=29,
I
'--
Q.
E
0
.
U
'0
004
1.0~ L\
\
\L~
\\
L4
\\ \\ -,
\\\\ \
•
• -- r""'- "' "'
~
-
U \
::::I 4
Q03
\~~
'0 1.10
e \ \
0 \! " "x
...
'0
--
::::I
K '\ \
\\~
a_
.
"- "-
'"
II. O.M)
\ ..... '< -,
002
°i~
'
'--
\'
I', "
~ I~\
10·s
• " OJO
•
"
\~
O.Hr-
• O.$tJ
O.
:\\
a • •• '0
• • •• I()ll
QOI
Pseudoreduced Pressure, Ppr
3 4 5 6 7 8 9 10 I~
B.6.1. Total Formation Volume Factor. It is often convenient to B.6.2 Total Compressibility. Compressibility, c. is defined as the
express the formation vol ume factor below the bubblepoint pressure change in volume per unit volume for unit change in pressure at
in terms of total formation volume factor, independent of the num- constant temperature.
ber of phases present. Total formation volume factor is defined as
the ratio of the total volume of hydrocarbons per unit stock-tank bar-
rel of oil at the prevailing pressure and temperature. This can be ex-
C = t (~~) T' ••••.•.•••...•••••••••••••••••• (B-80)
••
4
4 0
I
.s
... • .s
- 0
00 0
B
0
0
\ l
0
~
I- VAN DER KNAAP'$
CORRELATION
I
0 o.
..
.. • l)
• • 0
q 0
~
• 1\ . ..
•
-•
'0
-- -.
0
"' ~ - .-, •
!'It .. I
HALL'S ~
• 4~ CCRRELATION q
• • s-
4-·HALL'S 0
COIfJItELATION
•
.. :
0 .. 0
~ s'
0
.....
• ~ • 0
0 0 0
I
• ~ 0
•
, o 5 ~ ~ ~ e ~
o & ~ ~ ~ " ~ H INITIAL POROSITY AT ZERO NET PRESSURE, "',
INmAL POROSITY AT Z£RO NET PRESSUR£•• 1
Fig. B-29--Pore-volume compressibility at 75% lithostatic pressure
Fig. B-28--Pore-volume compressibility at 75% Iithostatic pres- vs. initial sample porosity for friable sandstones (after Newman45).
I
'--
sure vs. initial sample porosity for limestones (after Newman 45).
pressibility can be determined from laboratory experiments, from
where B, = formation volume factor of Phase i. the correlations presented earlier in this section, or from the Trube
Accordingly, Co, cw , and cg in Eq. B-81 can be replaced by the pre- correlationf shown in Fig. B-27. Trube correlated pseudoreduced
vious expression to yield compressibility, cpr, with pseudoreduced pressure, Ppr> and pseudo-
reduced temperature, Tpr. The oil compressibility then can be esti-
aBo)
- IT
So ( a (- I aBw) mated from
C, =
oPT + s; B w -a-
P T Co = cpr/ppc' ............................••.. (B-86)
+ Sg ( a aB g )
- IT + C/, (B-83) where ppc = pseudocritical pressure from Fig. B-1 and
s P T Tpc = pseudocritical temperature from Fig. B-1.
Taking into account the gas in solution, the apparent compressibili-
where Bw = water formation volume factor, bbVSTBW; Bo = oil ty of oil, COQ , below the bubblepoint pressure can be calculated by
formation volume factor, bbVSTBO; and Bg = gas formation vol-
oa = Co + (0.83p R
+ 21.75) (BBo) . .
ume factor, bbl/scf, s g
Eq. B-83 can be modified slightly to include the effects of solu- C (B-87)
tion gas on the change in liquid phase volumes and presented as
Water Compressibility. For an isothermal condition, the com-
, = Sa[(-=--1
pressibility of oilfield water can be defined as
C
Bap aBo) + Bg(aR
Bap ) ]
s
TOT
J (a~w )T'
o
Cw= -
w
(B-88)
Oil Compressibility. The isothermal compressibility of an under- Formation Compressibility. Rock or formation compressibility
saturated oil (above the bubblepoint pressure) can be defined as under isothermal conditions can be defined as
3 o 0 a 0
0
J
0
0
I
0
00
0 ....
0
I 000
0
0
0 o 9' ~
0
L 0 00
0
. •• 0
..
0
cr
~
---....
~-'C;'L ii HALL'S-'~
II)
0
!......
ILl
II:
:Ii
q.,
oa-
0
,~ '--:--'
0
iii
II)
'"e,
II:
4 - CORRELAnON
• 0
8 0
ILl I 0- ::IE
:If 0
0 o
o
~
0 00
I
'":IE
~
0
ILl
~
I
o 5 ~ ~ ~ U ~ U
INITIAl. POROSITY AT ZERO NET PRESSURE, +1 ~ I
~ 0 5 10 15 20 25 30 3lI
Fig. B-3O-Pore-volume compressibility at 75% Iithostatic pres- INITIAL POROSITY AT ZERO NET PRESSURE, +1
sure vs. initial sample porosity for consolidated sandstones (af-
ter Newman 45 ) . Rg. B-31-Pore-volume compressibility at 75% Iithostatic pres-
sure vs. initial porosity for unconsolidated sandstones (after
M~ {[tR~~)'"' w(NJ'] R~ + }
.................... (B-90) p,
I
'--
................... (B-93)
where (LlhIRTpc)lO) is found in Fig. 8-40, (LlhlRTpc)(I) is found in
Figs. 8-41 and 8-42, and the quantities Tpc' Ppc, w, and RTpclM are
found in Fig. 8-43. The second term of Eq. B-93 is omitted when
the reduced temperature is less than 0.8.
The corrections for pressure given in Figs. B-40 through B-42 de- E
.c
pend on the reduced temperature and pressure, defined by Eqs. B-3 :a5OD
and B-4, and are independent of the Watson characterization factor, III
Kw ' and acentric factor. w. If they are not available from measure-
I ments or other sources. the values of Kw and w may be estimated
I'--
from Fig. B-43, when the API gravity and the pseudocritical temper-
ature or pressure of the crude are known. Kesler and Lee47 provide
additional details. The base enthalpies, h", are referred to a value
hO= 0 at T= - 200°F. Enthalpies calculated by this procedure take
into account changes in the isobaric specific heat, Cpo, with temper-
ature and of the latent heat of condensation and evaporation of the
crude fractions. Average specific heats. Cpo, of the liquid and vapor
phases can be estimated by taking the chord secant of the enthalpies
over the desired temperature range.
C = hTI - h T2
L po T1 - Tz . ............................ (B-94)
When only the density or API gravity of the crude is known, its
heat capacity may be estimated from the Gambi1l50 relationship.
.... •.
•._
:ilki: ..
...;If~
IJ£~
,~~~.
. 350
358.43
134.62
150
321.64
330.53
1,192.6
1,194.4
871.0
863.9
l." T·: :... .... j;:i~1 _ _ I I r~~: 381.82 200 355.40 1,198.7 843.3
I I· :: __ _. ,," ",. ,,,. :~;i;;~ :d,"i' :m ~
.• .".... ....... .. .. ... :~~:~ '.!'f:!!'11'~ ... 400 247.25 374.97 1,201.2 826.2
700 800 900 1,000 1,100 400.97 250 376.04 1,201.4 825.4
1,200
Temperature, OF 417.33 300 393.85 1,203.2 809.3
Fig. B-3S-Enthalpy of petroleum fractions (K'w= 11, T=600 to 420 308.82 396.78 1,203.5 806.7
1,200°F, after Kesler and Lee 47 ) . 431.71 350 409.70 1,204.4 794.7
440 381.59 418.91 1,204.8 785.9
L 460
467.00
466.97
500
441.42
449.40
1,205.0
1,204.9
763.6
755.5
480 566.12 464.37 1,204.2 739.8
486.21 600 471.59 1,203.6 732.0
500 680.80 487.80 1,202.0 714.2
520 812.68 511.9 1,198.4 686.5
540 962.80 536.6 1,193.3 656.7
544.56 1,000 542.4 1,191.9 649.5
560 1,133.4 562.2 1,186.3 624.1
I
1
E
.0
580 1,326.1 588.9 1,177.0 588.1
'-- '§ 596.20 1,500 611.4 1,167.7 556.3
m 600 1,543.2 616.8 1,165.2 548.4
>;
a.
iii 620 1,787.0 646.5 1,150.2 503.7
&;
C 635.78 2,000 671.7 1,135.2 463.5
w
640 2,060.3 678.7 1,130.7 452.0
660 2,366.2 714.4 1,104.9 390.5
680 2,708.4 757.2 1,067.2 310.0
695.37 3,000 802.6 1,019.3 216.7
700 3,094.1 823.9 995.6 171.7
705.34' 3,206.2' 910.3 910.3 0
• Critical properties.
Temperature, OF
Figs. B-47 and B-48 give the enthalpies of water and steam at sat-
Fig, B-36-Enthalpy of petroleum tractions (K'w = 11.8, T= 0 to uration pressures and temperatures.
6000F,after Kesler and lee 47 ). At temperatures higher than its boiling point, steam is said to be
superheated. The additional enthalpy of superheated steam, above
8.8.2 Water and Steam. Table B-5 presents the enthalpy of water its saturation value as given in Table B-5, is found by multiplying
and steam at the saturation pressure and temperature. The differen- the mean specific heat, Cs , given in Fig. 8-49, by the excess temper-
cebetween these two quantities represents the heat required to con- ature above the saturation value, T,. This product is called superheat.
vert water from liquid to its vapor (steam) and is called the latent Sh, = CiT - Ts ) , T ~ Ts • . •••••••.•.•••••••• ,. (B-97)
heat of evaporation.
1
Note that the mean specific heat of steam at atmospheric pressure
I Lv = h, - h.. . (B-96) over a temperature range indicated in Fig. B-49 is somewhat lower
L
123
L 1,000
i 900
i.-
800
E 700
.Q
.:::
::J
iii
>;
Co
iii 600
~
C E
W .Q
""
::J
iii
500 >;
Co
iii
s:
C
W
Temperature, of
i....... Temperature, OF
than the actual specific heat of superheated steam at a given temper-
ature, shown in Fig. 8-44. Fig. 8-46 shows the specific heat of satu-
rated water. Fig. B-38-Enthalpy of petroleum fractions (Kw=12.5, T=O to
600°F, after Kesler and Lee 47 ) .
B.8.3 Reservoir Rocks and Minerals. Sornerton'i'' provided data
in Figs. B-50 and B-51 and Table B-6 on the heat capacity of dry ties of pure quartz and pure calcite. They agree rather well with the
reservoir rocks. Fig. 8-51 also shows the calculated heat capacity experimental values of the two samples. This is an example of the
corresponding to the porosities of Samples 1 and 8 of Table B-7. rule that the heat capacity of a mixture is the mass average of that
Those calculated heat capacities are based on reported heat capaci of its constituents. For fluid-saturated rocks, the heat capacity may
be estimated from
Mf
Cf = PI' (8-98)
124
MULTIPHASE FLOW IN WELLS
,
I
!
I
'--
Reduced Pressure, Pr
Reduced Pressure, Pr
Fig. B-42-Enthalpy departure from zero-pressure-deviation
Fig. B-41-Enthalpy departure from zero-pressure-devlatlon function low-temperature part (after Kesler and Lee 47 ).
function high-temperature part (from Kesler and Lee 47) .
where Po, Po, Pw, and Pg = densities of the solid (rock), oil, water,
37
36
3.5 1ft
34
3.3
u.
0,
1.2
E
.0
e;
:J
co 1.1
.~
u
al 1.0
C.
al
o
16 0.9
I '"
J:
I.- 0.8
II
i 0.7
Q.
k
0.6
10 20 30 40 50 60 70 80
0.2 0 200 400 600 800 1.000 1,200 1.400 1,800 1,800 2,000
011 Gravity, °API 011Gravity, °API
Temperature, OF
Fig. B-43-Tpc. ppc, wand RtpclMas functions of oil gravity and Fig. B-44-Heat capacity of gases at 1 atm (after Touloukian
Watson's characterization factor, Kw (after Kesler and Lee47 ) . et al. S1 ).
L 2
3
Sandstone
Silty sand
Poorly consolidated, medium-fine grain
Poorly consolidated, poorly sorted
0.273
0.207
40
20
IIlite(?)
Kaolinite type
Feldspars
Feldspars
4 Silty sand Medium hard, poorly sorted 0.225 20 Kaolinite type Feldspars
5 Siltstone Medium hard, broken 0.296 20 Kaolinite type Feldspars
6 Siltstone Hard 0.199 25 Illite Feldspars
7 Shale Hard, laminated 0.071 40 Illite/kaolinite
8 Limestone Granular, uniform texture 0.186 Calcium carbonate
9 Sand Unconsolidated, fine-grained 0.38 100
10 Sand Unconsolidated, coarse-grained 0.34 100
B.9.1 Liquids. Thermal conductivities of saturated organic liquids where Yj = mole fraction of the jth gas component.
decrease with increasing temperature. Plotted in Fig. B-52 are val- Lenoir and Comings'v and Eckert and Drake 64 presented thermal
ues for some pure organic liquids. For petroleum fractions and hy- conductivities of superheated steam vs. temperature at several val-
drocarbon mixtures, Cragoevl proposes the relationship ues of pressure and for the saturated vapor.
(B-lOI)
100r-~~--,-"--'--'----.~r-:-:::"II"I~r--r-1LO
to estimate the thermal conductivity in BtuID-ft-°F. For 0.78 RlduCld
601----...--_-..--r-~r_....,..
<Yo <0.95 and 32°F<T<392°F, Eq. B-lOl gave "average and
L maximum errors of 12 and 39%." 401--~1_--+~-+--+-__I~iI'i
L
U
The thermal conductivity of saturated water, shown in Fig. B-52,
.......
I
A = 1{~>jAj + [f(t.)]-I},
J=I J=I J
(B-103) Fig. B-45--Pressure correction to the molar heat capacity of
gases (after Chemical Engineers' HandbooJc52).
L Concrete
Iron
119 to 144 0.21 11 to 19 0.45 to 0.65
B.9.3 Reservoir Rocks and Other Solids. Somerton et ai. 66 and corresponding to those in Kern River oil sands. In these expressions,
Anand et al. 68 have reported extensively on the thermal conductiv- no distinction is made about the fluids other than the brine.
ity of reservoir rocks, including the effects of fluid and mineral con- The effect of temperature is given by
tent, particle size, temperature, and pressure. For unconsolidated
quartzitic sands saturated with water and oil, the thermal conductiv-
ity at 125°F is given by
(B-I06)
A.t = 0.735 - 1.3D.p + 0.390A. m IS:, (B-I04) where A.r is given by Eq, B-103, and the effect of pressure, which is
where the thermal conductivity may be estimated from generally minor, is estimated from
... 1,200 1
~
r--
~
... I / 1,000
u,
L \.,
E
.l:l
Latent Heat )f r't ' cal
Pol~t
---
0
E
JIJ,'! I) ~ 800
L
.0
"'-
., 'l~ f
~~
.2 .: '.5 >:
7/// 7~
CD Q.
€a. I \.'
u,
iii
.II:
c 600
/
...
"e
'"
o'" ... V/;/ I/f/d* L 1.3
.0
"'-
.E
W
/'
/
~ If
/
1ii 400
III
J:
0. •
~ ..-I-~~ /
\.2
;i-
'0
\., o'"'"
a.
200
(
~
.-I~V , .. 1ii
III
J:
... ~
100 2DG _
Temperature, OF
400 sao IDD 'JDO
o
o 1,000 2,000
Saturation Pressure, psla
3,000
Fig. B-46-Heat capacity of saturated pure liquids (after Toulou- Fig. B-47-Enthalples of water and steam as a function of the
kian et al. 51 ). saturation pressure (reference temperature is 32°F).53
-
1
1,200
1,000
800
-- 'fo~--'-
""V
Crlt cal Point
,\
\
sm Latent Heat
Ll.
0
I
L ~
600
E
.0
""'-
I
\\
\\ i\
iii ::>
.a::
C
W V iii
~
./
. ~'\ -,
II
400
~&
J:
~ \.\
:s
ke(\t' e
'u
~
~ <,
Co
200
...... III
....
c I'.... ...........
o ./
V :::E r-, 1'-3,000 psis
11". .....
, ~ i"""'oo.... ........
--
o 100 200 300 400 500 600 700 0 ...... ........ ............ -....
- 2,510
0
.~""" ..... 2,000
Saturation Temperature, of .~
............ ~
...... ~ ....... .1,500
Fig. B-48-Enthalpies of water and steam as a function of the 0
.~
..... ~
~--
1,000
saturation temperatures (reference temperature is 32°F).53
in Fig. 8-56 show the typical effects of fluid content and porosity on
the thermal conductivity of unconsolidated oil sands.
For consolidated sands, Anand et al.68 give the relation
0
0
100...
-
........
-- 75~~
1
0.320 0.320
, .II
0.300
-
0.300
ff
14.~1--9' -... »:
-
u.
~
.0
'§
0.280
V~
./
;~ -;.;'~
¢'""'.:- t
.0
E
I
0.280
~ ~ ~-
~- --
~r)\e
Iii 0.260 V.... ~ 0.260
~?
~
U
~ 0.240
~
til 0.220
A~ ~
h!i
* A Sandstone 0- - -
A Silty Sand t:.
- -
-
Iii
.~
U
[
'"
0.240
0.220
~~~~
«00"
~ ....~,o"
~~
"
~O~~
J:
Ol
~ ~'
L 0.180
Ai Shale 0--
0.'80
0.160
0.160 0 100 200 300 400 500 600 700 600 900 1.000 o 100 200 300 400 500 600 700 800 900 1,000
Temperature, OF Temperature, of
Fig. B·So-Heat capacities of some reservoir rocks (after Fig. B-51-Comparlson of calculated heat capacities of reser-
Somerton 54). voir rocks and principal constituents (after Somerton S4).
Fig. B-57 shows the measured thermal conductivity of Berea where the value ofAj.w may be estimated from Eq. B-I08 or consoli-
sandstone vs. temperature when the sample is saturated with four dated samples, and I, andA.g are the thermal conductivities of the oil
I different fluids. Thermal conductivities of reservoir rocks contain- and gas phases, respectively. A similar volumetrically weighted
I'-- ing several fluids are treated differently. Average values may be esti- geometric mean may be used to approximate any average value
mated from figures, such as Fig. B-57, or from the values obtained when no better procedure is available.
with the equations given previously. Such averages must consider Table B-ll gives thermal conductivities of selected minerals
the relative volumes of the fluids present in the reservoir rock. Ac- found in reservoir rocks. Table B-8 reported the thermal conductivi-
cording to Anand et al.,68 the conductivity of the wetting phase has ties of some metallic and nonmetallic materials. Fluid flow in-
the dominant effect. The liquid conductivity used in Eq. B-107 creases the apparent thermal conductivity of porous rocks.
should be biased accordingly.
Table B-9 presents data on the thermal conductivity of several dry B.9.4 Thermal Diffusivity. Thermal diffusivities are related to the
consolidated rock samples, calculated from thermal diffusivities thermal conductivity and volumetric heat capacity by the relationship
and densities measured at 200°F. Repeat values were generally low-
er because of reactions that occurred in the samples as they were a = A.IM. . (B-112)
heated to temperatures of I,800°F. Thus, the temperature history of
L a sample may have a significant effect on thermal properties. Fig.
B-58 shows the effect of temperature on the thermal conductivities.
Values calculated in this manner are quite adequate for engineer-
ing applications. Because of difficulties in the measurement techni-
Tables B-9 and B-10 give initial sample properties. ques, calculated values actually may be better than directly mea-
In the absence of specific data, it appears that Eq. B-108 could be sured values.
used, together with thermal conductivities of dry saturated rocks Table B-9 presents information on the thermal diffusivities of
given in Table B-9 and Fig. B-58, to estimate thermal conductivities several dry consolidated rock samples at 200°F, measured by use of
of saturated consolidated rocks other than sandstones. Errors in this two different techniques. Repeat measurements were always lower
I
approach should be recognized as being potentially significant. The because of reactions that occurred as the samples were heated to
i'-- following approximate averaging procedure is useful to estimate the temperatures of 1,800°F. Fig. B-59 shows the effect of temperature
value of the thermal conductivity of a porous medium containing on the thermal diffusivity of dry consolidated rocks. Tables B-9 and
water, oil, and gas when the thermal conductivity of the brine-satu- B-IO give the initial sample properties.
rated sample is available.
B.9.5 Coefficients of Heat Transfer. For condensing steam, the film
coefficient of heat transfer, hj, is large and generally adequate.P''
................... (B-111)
hI = 48,000 Btulft 2-D-°F. when NRe> 2,100, (B-1 13)
pq
I
'-- 3
.....-
e::: t::::== I-- ~
irf'\ 0
t::; ~r"
c-- :::::: ~
~~
::- r- -
---- -- ---
~
/ \\ 14
-
~ ::::::: ~ ...-- ~ :::::- f::::"
I--
/ , 13
--
~ ~ t::= ,.....
- - - -
V ~ ~ ~ ~ -:::- :::: ~ ::::: .....- ~ r-
-~
.~
1 i\ ....- -
::::: -
.. 12 r:::: E::::: E:::::: ...-- k-':
~.- ~ ~ ~ l::::= ~ ~
- t-'
~
'+
2
~U ~ ~ ::..::: ~ ~ :.:::: ;:;:.. 10-
.- r::::: ~ r::.::
~~ \,
1
1~
0
u:. ..,c
::J
. / ~ ~~ ~ ~ ~ ~ I--
t-'
I-- ~ t::::
r- -
~ ~ t/:: ~ ~ ~ ~ l-- ~I-- 10-
- -
0, 0 t-'
~
tJ
f§
m o~
r-
0 ..-:::: ~ ,./" ....- ~ ~
.'i'J
~
~
'\
,'\ re -, 1\ a
.~
11
:::J
"C
c:
'/ ~ ~ ~ /
v~ ~
v::
.>
~f\. ~ &0_v
" 0
I'-
\ o
iii
E 1//1VI V r,. = Reduced Preaaure
~
Q)
'/1 1/
\ t= 8
III, I
~ \ 7 '/I
~I
I
\ 2 3 4 5 6 7 8 9 10 11
\
L 7
Reduced Pressure, Pr
\
0 TABLE B-11-THERMAL CONDUCTIVITY OF
o 100 200 300 400 SIlO 600 700 ROCK-FORMING NATURAL MINERALS-s
Temperature, OF
Thermal
Fig. B-52-Thermal conductivity of saturated pure liquids (after Density Conductivity
Touloukian et al. 60 ). Mineral (lbmlft3) (Btu/ft°F-D)
Calcite 169.9 49.8
where the Reynolds number, NRe, is defined as
Dolomite 178.4 76.35
lr ] 2i..{Jw.sc Siderite 237.9 41.7
N Re = [ 2(0.0616) lrfJ1j' (B-114)
Aragonite 176.5 31.0
where iw = steam-equivalent injection rate as condensed water in Anhydrite 185.9 65.97
barrels per day, Pw,sc= density of water at standard conditions (62.4
Gypsum
Ibmlft 3),P,s =viscosity of steam at the injection temperature in cen- 17.4
tipoise, and tt = inner radius of the conduit in feet. Halite 84.7
Because the viscosity of saturated steam over typical temperature Quartz 165.2 106.6
ranges is approximately 0.0 18 cp, turbulence prevails (NRe > 2,100) Chert 159.8 51.4
\
whenever iw is greater than 9.85r;.
I For hot water; the film coefficient of heat transfer is found from 70
Pyrite 306.8 266.2
L Hornblende 198.7 38.9
hI = 0.0115~; N?i:~;4, (B-115) Albite 163.8 29.7
when NRe exceeds 2, 100. NRe also is given by Eq. B-114; A.w is the Microcline 159.7 34.2
thermal conductivity of water in Btu!ft-D-°F. The dimensionless Orthoclase 161.3 32.1
Prandtl number is Kaolinite 38**
N pr = 58.I(Cp,jA t, (B-1I6) Illite 181.3 30.49
I'-- where Cw = specific heat of water at constant pressure in Btu!
Montmorillonite 176.6 112.8
lbm-Ff' andp,w = viscosity of the hot water in centipoise. In the tem- Chlorite 171.8 71.36
perature range of 350 to 600°F, the value of factors representing the Biotite 186.1 28.0
water properties [(A.wjp,~8)Ng;4] does not vary by more than Muscovite 178.1 32.2
± 10%. Accordingly, the approximate film coefficient is derived as
"Except for kaolinite, an data are from Horai 65 and were obtained by use of naturally
hI = I.6i~8ri-L8,
occurring minerals at room temperatures and pressures.
(B-l17) ··Value used by Somerton et al.66
130
MULT/PHASE FLOW IN WELLS
1.5
1.4
0.8
1.3
1.2
L. 1.1 0.7
0
LL 1.0
t 0.9
0
LL
ffi
~ 0.6
~ 0.8
ffi
U 0
;:.
:::I
"C LL 's
c 0
0
U t:::I u:::I
"C
OJ 2.1 iii c
0.5
E
Ql .~
0
U
s: 2.0 .~ OJ
I- U
:::I
"C
E
Ql
c s:
1.9 0 I-
U
OJ
0.4
1.8 E Ql
.r:
I-
1.7
1.8
0.3
Temperature, of
i
.r:
--____
Solvenl~-------
Data
_
I-
o o
0.20
IBerea Sandstone I
i
L
Fig. 9-56-Thermal conductivity of unconsolidated oil sands Fig. 9-57-Thermal conductivity of Berea sandstone-eflects of
(after Somerton et al.66 ). temperature and fluid saturant (after Anand et al.68 ).
when 350°F < T < 600°F, iw.sc > 'i, and the subscript sc refers to
standard conditions. Fig. B-60 shows a plot of the results calculated 2 Wg
N Re = [.7l(0.01096)/2 ]7i II r.: ..............•.. (B-118)
from Eq. B-l17. r:« I
For hot gases, the film coefficient is found by evaluating Eq. Accordingly, the film coefficient is given by
B-115 for gas properties. For gases, it generally is found that
~4 = 0.92 ± 0.07 and the Reynolds number, including the con-
version factor associated with the oilfield units for gas, is given by
hI = 2.86 x 1O_4~(~)o.8
rJ /-lgrl . •••••••••••••••• (B-119)
131
1
\
L. 1.2
........ 4>
"...,., .e
-,' 0/'
.'
1.0
--------- ~ 0.8
~ ,~
.l!
~ ,, a' 0.7 1 _ _ Bandera sandstone
~'.\ j 2 _ _ _ BereaSandstone
lie:
0.8 3 ••••••• Boise Sandstone
0
\\'\' ' .~ 0.6
4 _ . _ Umestone
'\\~\
5 _ •._ CaJci<Jm Oxide
~ ·il 6 ShaJe
II:
~
''':~
1 __ ' _ Rocl< 5aJt
~
" . .. 8 __" _ Tull
u \ \ \
0.5
:::l
"C
c
0
0.6
"
....
~'...:."---- 0.4
0
1li
. tl> "'-
E
CI> " . ..........-G "'- - -(!)
0.4
~ 1,600 1,600 2,000
1- - Bandera Sandstone Temperature, of
2 - - - Berea Sandstone
3 - - - - - Boise Sandstone Fig. B-59-Thermal diffusivity relative to that at 200°F-samples
0.2 4 - . - Limestone SUbjected to Increasing temperatures (after Somerton and
6 - - - Shale Boozer 70) .
7- - - - Rock Salt
8 _____ Tuff
being dislodged, thus possibly plugging the injection well bore or
0 other parts of the flow system.
0 400 800 1,200 1,600 2,000
Temperature, of Values ofht'c. For surface injection lines exposed to winds, McA-
dams 69 gave the coefficient of heat transfer caused by forced con-
Fig. B-58-Thermal conductivity relative to that .. 200°F-sam- vection (air currents) at the outer surface of a conduit, h.jc, as
pies SUbJected to Increasing temperatures (after Somerton and
Boozer7 ).
. , (B-120)
when wg > I 92,ug rj and ~~ = i p, 103 IbmID; i = injection rate, for I ,000 <NRe < 50,000, where.l, = thermal conductivity ofair,
MscflD; P = density, Ibmlft ;,ug = viscosity, cp; Ag = thermal con- Btu/ft-Dv'F, re = external radius of the conduit exposed to air, ft;
ductivity, Btu/ft-Dv''F; and r, = inner radius of the conduit, ft.
NRe = [Y2(4, 364))(2r evwPa .sc1,ua); Vw = wind velocity normal to
Fig. B-61 shows the values of h.j for hot-gas injection obtained
from Eq. B-119. the pipe, mile/hr; Pa,sc = density of air.Ibm/ft-': and,ua = viscos-
ity of air, cp.
Values of~. Values of the coefficient of heat transfer caused by For the surface temperatures at re of about 200°F, McAdams69
deposits of scale and dirt, hJ, seldom are known with any degree of gave the approximate film coefficient as
confidence. Often values are specific to the project and seldom
available unless measurements are made. McAdams 69 used a typi- (B-121)
cal value of 48,000 Btu/ft2-D-°F. This value is recommended in the
absence of better information. In injection and production lines.
where low heat losses are desired, a significant scale deposit be-
L tween pipe and insulation is an asset, because it results in a relatively
high resistance to heat flow. A scale deposit on the inside wall of the
pipe also would be an asset if there were no danger of the deposit
Ul
~
§ 10
'iii
c:
Ql
u, E
o
is
6 6>
~ :.:::
iii ...
i:'
hf, BtuJft2-D-oF
..c:: 1.0
~ 1()2 hf, Btulft-D-OF Ag, Btu/ft-D·oF
..c::
Ij, ft Ij,ft
4v,sc, B/D wg,lbmlD
350°F < T < 600°F Jlg'CP
wg>PgT;
Fig. B-6O--Correlatlon to estimate film coefficient of heat trans- Fig. B-61-Correlatlon to estimate film coefficient of heat trans-
fer for water.s fer for gas flow at high velocities. s
L I. Matthews, C.S. and Russell. D.G.: Pressure Buildup and Flow Tests in
Wells, Monograph Series. SPE. Richardson, Texas (1967) 1.
2. Craig, FF Jr.: The Reservoir Engineering Aspects of Waterflooding,
Inti. (November 1980) 125.
33. Nurnbere, D., Brigham, W, and Standing, M.B.: Correlations for Physi-
cal Properties of Petroleum Reservoir Brines, Petroleum Research Inst.,
Monograph Series, SPE, Richardson, Texas (1971) 3. Stanford U., Palo Alto, California (November 1977) 17.
3. Earlougher, R.C. Jr.: Advances in Well Test Analysis, Monograph Se-
34. van Wingen, N.: Recovery of Oil in the United States, API, New York
ries, SPE, Richardson, Texas (1977) 5.
City (1950) 127.
4. Farouq Ali, S.M.: Oil Recovery by Steam Injection, Producers Publish-
ing Co. Inc., Bradford, Pennsylvania (1970). 35. Frick, Te.: Petroleum Production Handbook, SPE, Richardson, Texas
5. Prats, M.: Thermal Recovery, Monograph Series, SPE, Richardson, (1962) II.
Texas (1982) 7. 36. Hocott, C.R.: "Interfacial Tension Between Water and Oil Under Reser-
6. Engineering Data Book, ninth edition, Gas Processors Suppliers Assn.• voir Conditions," Trans .• AIME (1939) 132, 184.
Tulsa, Oklahoma (1972) Sec. 16. 37. Hough. E.W, Rzasa, MJ., and Wood, B.B.: "Interfacial Tensions at
L 7. Edminster, W.C.: Applied Hydrocarbon Thermodynamics. Gulf Pub- Reservoir Pressures and Temperatures; Apparatus and the Water-Meth-
lishing Co., Houston (1961). ane System." Trans., AIME (1951) 192,57.
8. Trube, A.S.: "Compressibility of Undersaturated Hydrocarbon Reser- 38. Stewart, WF, Burkhard, S.F, and Voo, D.: "Prediction of Pseudo-Criti-
voir Fluids," Trans.. AIME (1957) 210, 341. cal Parameters for Mixtures," paper presented at the 1959 AIChE Meet-
9. Standard 2500, Measuring, Sampling, and Testing Crude Oil, API, Dal- ing, Kansas City, Missouri.
las; reproduced in Petroleum Production Handbook. McGraw-Hili 39. Dranchuk, P.M. and Abu-Kassem, J.H.: "Calculation of Z-Factors for
Book Co. Inc., New York City (1962) 1, Chap. 16. Natural Gases Using Equations-of-State," 1. Cdn. Pet. Tech. (july-Sep-
10. Brown. G. et al.: "Natural Gasoline and the Volatile Hydrocarbons," tember 1975) 14,34.
L 22. Beggs, H.D. and Robinson, J.R.: "Estimating the Viscosity of Crude Oil
Systems," lPT(September 1975) 1140.
23. Baker, O. and Swerdloff, W.: "Finding Surface Tension of Hydrocarbon
5 I. Touloukian, YS. et at.: "Specific Heat-Nonmetallic Liquids and
Gases," Thermophysical Properties of Matter, IFUPlenum, New York
City (1970) Chap. 6.
52. Chemical Engineers' Handbook, R.H. Perry and C.H. Chilton (eds.),
Liquids," Oil & Gas 1. (2 January 1956) 125. fifth edition, McGraw-Hill Book Co. Inc., New York City (1973).
24. Reid. R.C., Prausnitz, J.N .• and Poling, B.E.: The Properties of Gases 53. Steam Tables. Properties of Saturated and Superheated Steam, third
and Liquids. fourth edition, McGraw-Hili Book Co. Inc., New York
edition. Combustion Engineering Inc.. Windsor, Connecticut (1940).
City (1987).
54. Somerton, W.H.: "Some Thermal Characteristics of Porous Rocks."
C.1 Introduction phase, Yi, to the mole fraction of the same component in the liquid
A complex mixture of hydrocarbon compounds flows in producing phase, Xi. Expressed mathematically,
oil and gas wells. This mixture continuously changes its phase com- v.
position during the production cycle, as it flows to the surface and K, = Xi' (C-l)
enters the first stage of separation. A phase is defined as that part of
a mixture that is uniform in physical and chemical properties, homo- For ideal solutions at low pressures (typically below 75 psia), an
geneous in composition, and separated from other coexisting phases ideal equilibrium constant can be derived by combining Raoult's
by well-defined boundary surfaces. 1.2 Depending on the composi- law and Dalton's law. This can be expressed mathematically for a
tion, pressure, and temperature, the hydrocarbon system may exist fixed temperature as
as a single-phase gas, a single-phase liquid, or a mixture of gas and
liquid. In petroleum production fluids, water often is present as an
additional liquid phase.
Kj = ~ = p
i/' ... 0 •••••• 0 0 • 0 0 • 0 0 0 0 •••••••••••• (C-2)
Unlike a single-component fluid, a multicomponent mixture ex- where Pvi = vapor pressure of Component i, psia, and P = total sys-
hibits a phase envelope rather than a distinct monotonic phase tran- tem pressure, psia.
sition between gas and liquid phases defined uniquely by pressure The vapor pressure of any pure hydrocarbon is a unique function
and temperature. Within the two-phase region bounded by the phase of temperature. Eq. C-2 assumes the liquid phase behaves as an ideal
envelope for multicomponent systems, both liquid and gas phases solution following Raoult's law and the gas phase behaves as an
exist in equilibrium at constant pressures and temperatures. The ideal solution following Dalton's law.
shape of the phase envelope and the location of the critical point are For a real solution, the equilibrium constants are not only func-
unique properties of the fluid composition. tions of pressure and temperature but also of the composition of the
Fig. C-I permits a qualitative classification of the types of reser- hydrocarbon phases. In compositional modeling, the engineering
voirs that are encountered in oil and gas production systems. Typical objective is to determine the physical properties of the individual
oil reservoirs have temperatures below the critical temperature of gas and liquid phases. Consequently, the equilibrium constants (in-
the hydrocarbon mixture. Volatile oil and condensate reservoirs nor- dicating partitioning of each component between the liquid and gas
mally have temperatures between the critical temperature and the phases) must be known. Sec. C.7 provides a detailed discussion on
cricondentherm for the hydrocarbon mixture. Dry gas reservoirs determination of equilibrium constants by use of equations of state
have temperatures above the cricondentherm. (EOS 's). An alternative method to determine approximate equilibri-
Many condensate fluids exhibit retrograde condensation, a phe- um constants uses charts for pure components, such as those given
nomenon in which condensation occurs during pressure reduction. in Fig. C-2. However, it is appropriate at this time to introduce flash
In Fig. C-I, this abnormal (or retrograde) behavior occurs within the calculation techniques to determine the mole fractions of each com-
two-phase envelope between the critical point and the criconden- ponent in each equilibrium phase.
therm or cricondenbar.
When volatile oils and condensate fluids flow through pipes, their
phase behavior is best predicted with a variable composition model. C.3 Flash Calculations
The procedure requires a series of vapor/liquid equilibrium or Flash calculations are based on simple overall and component molal
"flash" calculations. These calculations are based on the concept of balances. It is important to define the variables required in the mate-
an equilibrium constant, otherwise known as the distribution coeffi- rial balance:
cient or K value. F = number of moles of feed,
L = number of moles of liquid,
Co2 Equilibrium Constant V = number of moles of vapor,
For a multicomponent system, such as petroleum fluids, the com- Zj = mole fraction of Component i in feed,
position, pressure, and temperature uniquely define the system Xi = mole fraction of Component i in liquid phase,
phase behavior. The equilibrium constant, Kj. of a Component i is Yi = mole fraction of Component i in vapor phase, and
defined as the ratio of the mole fraction of the component in the gas n = total number of components in the feed.
L
U".r..t.uraUci Ot1 1...rYo1r or
n
L(yj - Xj) = o.
i= I
," , .: , , 1'1
,,
, , ',I
" I ,I
,
1001
Uqui
,, , " )' .' I '" " = f(*) = o. .. (C-IO)
," ,
I Cu-Cap
le_nai,'
,
I
'"
I
i
'"-
I
,',' I., This equation is often called the Rachford and Rice equation. It
I I '"
, 150% / " is implicit in VIF. To solve it, K values must be known. Note that K
I I I values are themselves phase-composition-dependent. Because
, / I I
Two-Phase I phase compositions are unknown in flash calculations, so are the de-
ba i a n /
I pendent variables for which a solution is needed. This imposes a se-
cond level of implicitness in Eq. C-IO. Therefore, solution of Eq.
C-IO is computation intensi ve.
Temperature This section presents a solution scheme for the Rachford and
Rice equation, assuming K values can be calculated at each itera-
Fig. C-l-Typical phase diagram. 2 tion step for VIF.
A second-order Newton-Raphson convergence scheme'[ can be
The overall material balance is used to sol ve the Rachford and Rice equation for VIF with a reason-
able (10- 6) convergence tolerance.? The procedure requires that
F = L + V. . (C-3)
these variables be known:
Indi vidual component balances are • Number of components, n.
• Mole fraction of each component in the feed mixture, «.
zjF = xjL + yjV. (C-4) • Equilibrium constant for each component, Ki.
• A first guess for the mole ratio, (VIF)j, where the subscriptj re-
Eq. C-3 may be rewritten as fers to the iteration step count.
Given a first guess, for instance 0.5 for (VIF)j, an improved value
L V
F - F' (C-S) of (VlF) can be estimated from
C3.1 Liquid Phase. Solving Eq. C-4 for Xi, and replacing Yi and UF
...................... (C-II)
with Eqs. Col and C-S, respectively, results in
V V
Zj - YjF Zj - xjK jF where the derivative, !,(VIF), is obtained by differentiating Eq .
........ (C-6)
L I - Y C-IO with respect to VIF.
F F
On further simplification, this equation to determine the mole frac- .......... (C-12)
tion of the ith component in the liquid phase can be written as
1)(¥) +
J
x, = V Zj .••.•...••.••.•...•.•••.•. (C-7)
I+F(K j I) Convergence is achieved when
C.3.2 Vapor Phase. Similarly, solving Eq. C-4 for Yi and using Eqs.
I(¥)j+l - (¥)J < 1.0 x 10- 6
• • (C-13)
C-I and C-S gives
Once convergence is obtained for VlF, the composition of each
phase can be determined from Eq. C-7 and either Eq. C-I or C-9.
z.F - Yj
zjF - xjL I Kj The previous procedure requires values for Ki at the pressure,
Yj = V .................... (C-8) temperature, and composition of each phase. As mentioned earlier,
V
several methods exist to determine K values, including use of charts,
Eq. C-8 can be simplified as such as those given in Fig. C-2. The recommended way to determine
K values, however, is from EOS's. Sec. C.9 gives this procedure.
Yj =
s,«,
V . . (C-9)
I+F(Kj-l) C.4 Bubblepoint and Dewpoint Curves
For temperatures less than the critical temperature, the bubblepoint
curve separates a single-phase liquid from the two-phase region.
C.3.3 Solution Procedure. It is important to note that the solutions
Thus, a bubblepoint pressure is the minimum pressure at which a
for Xi andYi, based on Eqs. C-7 and C-9, respectively, require the de- liquid will not vaporize. At the bubblepoint pressure, there is no va-
termination of the VlF value. To solve for VIF, Eqs. C-7 and C-9 are por and all feed is in the saturated-liquid phase. Thus, V = 0 and
combined by use of this definition of mole fractions. Zi =Xi· Therefore, Eq. C-9 simplifies to Yi = «K, and
n
L i.«, = 1. (C-14)
i=l
Ethane
Conversion Pressure, 2,000 psia
Fig. C-2-Example K-value chart. (After Katz et al. 3 Reproduced with permission ofthe McGraw-
Hill Cos.)
The dewpoint curve in Fig. C-I separates the single-phase gas and dewpoint pressure, the hydrocarbon mixture exists in a single phase.
two-phase regions. At a given temperature, two dewpoint pressures The known feed composition is also the phase composition. The
can exist. At a dewpoint pressure, L = 0, V = F, and z, = Yi : problem is to determine iteratively the pressure and the K values of
Therefore, Eq. C-7 simplifies to each component in the feed at that calculated pressure to satisfy Eq.
C-14 or C-15, depending on whether a bubblepoint or a dewpoint
Xi
z,
= K ••••••••••••••••••••••••••••••••••• (C-15a) pressure is to be calculated. The Wilson 5 correlation can be used to
i
estimate K values initially. Wilson proposed this simplified thermo-
I
~KZi. = I.
L (C-15b) K, = pp'ex p[ 5.37(1 + W,)( I - ~i) ], (C-16)
I ;=1 i-I I
Bubblepoint and dewpoint pressures also are calculated with it- where Pci = critical pressure of Component i, psia: p = system pres-
erative methods. This procedure is iterative in pressure at any tem- sure, psia: J;'i = critical temperature of Component i, OR; T= system
perature below the critical point. Note that, at the bubblepoint or temperature, OR; and co, = acentric factor of Component i.
I
I..-
tially zero. For methane it is very small. For higher molecular-
weight hydrocarbons, w increases. Acentric factors for pure
'" z, I,
LK> ................................ (C-23a)
i= I I
where a = -In Pc - 5.97214 + 6.09648 0- 1 + 1.28862 In 0 then the assumed dewpoint pressure is lower than the actual dew-
-0.169374 06 and /3=15.2518-15.6875 0-1-13.4721 In
point pressure.
(J + 0.43577 06 , where 0 = Tb/Tc = reduced boiling point; Pc = criti-
cal pressure, in atmospheres; Tc = critical temperature, "R; and
C.7 Equilibrium Constants From EOS's
Tb = normal boiling point at I atm, OR.
For hypothetical or pseudocomponents, such as those used to For most applications in the multiphase flow of hydrocarbon mix-
characterize the heavy ends (C7+) in a multicomponent hydrocar- tures in pipes. the ideal assumptions used in deriving Eq. C-2 are un-
bon mixture, the acentric factors also can be estimated from the Ed- realistic. A hydrocarbonlliquid mixture at any point in the pipe is in
misterf correlation as equilibrium with a vapor mixture at the local pressure and tempera-
ture. The important parameters controlling the phase behavior are
W
= 2T
i:»
[IOgpe
T-T
- 1.167] _ 10
" ............... (C-19) pressure, temperature, and the composition of both phases. An im-
c b portant objective in flash calculations is to determine the individual
phase compositions, given the pressure and temperature at the com-
putation node and the overall mixture composition. The equilibrium
C.S Bubblepoint Pressure
constants, or K values, thus depend on all these parameters; i.e.,
Bubblepoint-pressure calculations can be initialized with the Wil-
son correlation (Eq, C-16) for Ki. At the bubblepoint, the system K = !Cp, T,zj)' (C-24)
pressure is denoted by Pb. A good starting value for Pb can be ob- These equilibrium constants can be calculated based on the ther-
tained from modynamics of vapor/liquid equilibria. Such equilibrium depends
on the fugacity of each component in each of the phases. Physically,
equal fugacities of a component in each phase implies zero net mass
transfer of that component between the phases, resulting in thermo-
dynamic equilibrium of phases. Thus, the criteria for thermodynam-
Starting with a first guess of bubblepoint pressure from Eq. C-20,
ic vapor/liquid equilibrium is that the fugacities of every component
K values are determined by use of EOS'so These K values are substi-
in each phase must be equal, or
tuted into Eq. C-14. IfEq. C-14 is satisfied within a reasonable toler-
ance, then the iteration is stopped and the last assumed bubblepoint f;V =f;L, i = I, ,n, (C-25)
pressure is accepted as the solution. Otherwise, a new bubblepoint
pressure is assumed based on these criteria: where!= fugacity, V = vapor phase, L = liquid phase, and n = num-
During the iterative process, if ber of components in the system.
n
Mathematically, the fugacity of a pure component is defined as
L(zjKJ < I, (C-2Ia)
I i=1
L then the assumed bubblepoint pressure is high and the next guess
must be a lower pressure. If
n where!= fugacity, psia: P = system pressure, psia; and Z = com-
L(zjK > I, j) (C-2Ib) pressibility factor.
j=l It is evident from this definition that fugacity has units of pres-
then the assumed bubblepoint pressure is low and the next guess sure. Fugacity also can be considered as vapor pressure modified
must be a higher pressure. to represent the escaping tendency of molecules from one phase
into the other. The ratio of the fugacity to the system pressure is
C.6 Dewpoint Pressure called the fugacity coefficient, <1>, and is calculated from Eq. C-26
for pure components.
The first guess of dewpoint pressure can be obtained by combining
i
where <I> = fugacity coefficient of Component i in the vapor phase (:~)TC = o (C-3Ib)
and <1>; = fugacity coefficient of Component i in the liquid phase.
It is clear from the definition of mixture fugacity coefficient that,
Soave.'! and Peng and Robinsonl- used these criteriato deter-
because at thermodynamic vapor/liquid equilibrium IV = I L , the
mine Parameter a and then modified the resulting expression empir-
equilibrium constant may be found with "
ically by use of the acentric factor to reproduce pure hydrocarbon
v, <1>; vapor pressures. In Table C-I, 'Fe and Pc are the critical temperature
Kj = x: =I
mV'
'l',
•••••••••••••••••••••••••••••• (C-28) and critical pressure of the pure component, respectively, and T, is
I
the reduced temperature. w is the acentric factor of the pure compo-
The fugacity coefficients for each component in each phase are de- nent. Alternative forms of these EOS's, in terms of compressibility
fined by use of EOS's. factor, are used often to calculate phase compressibility factors and
fugacity coefficients.
C.B Cubic EOS's Sees. C.8.1 and C.8.2 present the SRK and PR EOS's, the two
An EOS is an analytical expression relating the pressure to the tem- most widely used in terms of the phase compressibility factors. Both
perature and molar volume of fluids. The ideal and real gas laws pre- have been modified with suggested mixing rules recommended by
sented in Appendix B are two EOS's. Ideal gas laws assume that the Soave II and Peng and Robinson. 12 These modifications allow the
volume of molecules is insignificant compared to the volume of the use of SRK and PR EOS's for mixtures of hydrocarbons, such as
container, and that there are no attractive or repulsive forces be- volatile oils and gas condensates.
tween the molecules.
Cubic EOS 's have been developed that do not use these restrictive C.S.I SRK EOS. Eqs. C-32a and C-32b give the alternatives for liq-
assumptions. Thus, it is possible to apply these equations to both va- uid and vapor phases, respectively, from the SRK EOS.
por and liquid phases. The term cubic EOS implies that the volume
terms in it can be cubic. A generalized form for a cubic EOS is zi - zi + ZL(AL - B L - BD - ALBL = 0 ....... (C-32a)
RT a and
P = V - b - VZ + ubV + wb 2 ' ••••••••••••••• (C-29)
n ..................... (C-34)
, CI./pct.5
"L xp.0.5T 0
;=1
•
I U Cl
;=1
L ......................... (C-36)
Xi
Calculate the following
with Eq. C-?
YI with Eq. C-9
At.vw~h Eqs. C-38 and C-39
" Bt. v with Eqs, C-40 and C·41
IX;TcJpci
i= I
Determine q and Zv from
appropriate roots of EOS
......................... (C-37)
L "
IYiTcJPci
;=1
Calculate the following
1'?
r with Eq. C·?
¢ With Eq. C-9
K , With Eq. C-38
................ (C-38)
L BL = c, ~ I
;=1
xjTjPCj , (C-40)
Fig. C-3-Algorithm to perform vaporlliquld equilibrium cal-
culations.
in the single-phase region and three real roots in the two-phase region.
s; = Cb~ IY;Tc/Pcj, (C-4I)
The largest root for the vapor cubic equations corresponds to the com-
pressibility factor of the vapor phase, Zv, and the smallest root for the
i=l
liquid cubic equations corresponds to that of the liquid phase, 4..
and It is important to note that the compressibility factors used in the
fugacity coefficient for each phase are phase-composition-depen-
a, = [I + fWj(1 - ~/lt, (C-42) dent. Thus, the equilibrium constants also are composition-depen-
dent. Consequently, the solution procedures for mixture equilibrium
where Ca = 0.42748 and Cb = 0.08664. The functionjte, can be cal-
L culated on the basis of the equation for Soave presented in Table
C-l. The equilibrium constants, K;, can be calculated from Eq. C-28.
constants are highly iterative. The following is a step-by-step algo-
rithm 2 to calculate the equilibrium constants.
I. The input data required for this calculation are the system pres-
sure, p, temperature, T, and the overall system composition, Z<, for
C.8.2 PR EOS. Eqs. C-43a and C-43b give cubic equations for the each component.
liquid and vapor phases, respectively, based on the PR EOS's.12-14
2. On the basis ofEq. C-16, estimate or assume Kt values for each
ZI - (l - BL)ZI + (AL - 3BI - 2BdZL component.
3. On the basis of assumed Kt values from Eq. C-16 and the
- (ALB L - BI - Bi) = 0 (C-43a) known Z< values, use Eq. Cvl O to perform flash calculations.
L 4. With appropriate EOS 's, the compositions of the liquid and gas
and phases obtained from flash calculations can be used to determine the
fugacity coefficients of each component in each phase, cI>f and cI> i-
z~ - (1 - Bv)Z~ + (Av - 3B~ - 2B vlZv 5. Using fugacity coefficient ratios from Eq. C-28, calculate the
equilibrium constants or Kf
values for each component.
- (AvB v - B~ - Btl = 0, (C-43b) 6. Compare the equilibrium constants in Step 2 with the calcu-
lated values in Step 5 using the following convergence criteria:
where 4. and Zv = liquid-phase and gas-phase compressibility factors,
i(~~ - 1)2
respectively. The fugacity coefficients are calculated with Eqs, C-33
through C-42, as previously discussed for the SRK EOS. Only the Ca , :5 E, .......................... (C-44)
[ Cb ,andfivi values are different For the PR EOS, these are Ca = 0.45724, 1=1 I
Cb = 0.07780, and fivi = 0.37464 + 1.54226wi - 0.26992 W f. The equi-
L librium constants, K;, are calculated with Eq. C-28.
where E = preassigned convergence tolerance ( :5 10 - 4) and
n = number of components in the system.
7. If the convergence criterion in Step 6 is satisfied for all compo-
e.9 Solution Algorithm nents. the values of the equilibrium constants ace used to calculate
To determine the compressibility factor, Z, in the liquid or gas phase, phase compositions required in determining phase physical proper-
the appropriate EOS can be solved either by direct or iterative meth- ties. Otherwise, these calculated values are used as the new guesses,
ods. These equations are cubic equations that yield a single real root and Steps 3 through 6 are repeated until convergence is achieved.
Fig. C-3 is a computer flow chart to perform a flash calculation with where V· is the characteristic molal volume of the liquid and V~o)
L
the algorithm. and V~cl) are two correlating parameters given by
Soave II suggested some modifications to calculations of the A and
B parameters. These parameters consider molecular interactions. The VIol = I - 1.52816(1 - T,)1/3 + 1.43907(1 - Ti /3
R
modifications were based on an empirically determined correction
factor called the binary interaction coefficient.P Soave II and Gra- - 0.81446(1 - Tr) + 0.190454(1 - Tl/3, ... (C-50a)
boski and Daubert'" suggested that these corrections are not neces- where 0.25 < T, < 0.95, and
sary for pure hydrocarbon systems. In the presence of associated non-
hydrocarbons, such as HzS and COz, this correction can greatly im- V~cl) = ( - 0.296123 + 0.386914T, - 0.0427258T~
prove the volumetric- and phase-behavior predictions.
- 0.0480645T~ )/(T, - 1.00001), (C-50b)
C.10 Densities
where 0.25 < T, < 1.0, and T, = TITc ·
Once the compressibility factors of each phase are determined, the Hankinson and Thomsonlf provided the values of V· and WSRK
vapor and liquid densities can be determined from for a list of organic and inorganic pure components relevant to
hydrocarbon fluids. WSRK is the acentric factor derived from the
SRK EOS.ll It can be calculated, independent of any EOS, by meth-
i=1 ods proposed by Lee and Kesler? and presented in Sec. CA. Thus,
WSRK can be replaced in Eq. C-49 by W from a more general source.
The relevant independent variables in Eq. C-49 are calculated with
and the following mixing rules.
n
PLxjM j
;=1
L" L" XjXjVZTcjj
j= 1 j= 1
(C-46) ....................... (C-5t)
ZLRT ' V*
where, Mj = molecular weight of the ith component.
Liquid densities calculated from Eq. C-46 are usually not suffi- W = L" XjWj, •••••••••••••••••••••••••••••••• (C-52)
ciently accurate. Alani and Kennedy'? also developed an equation i=1
similar to the van der Waals equation to determine the molal liquid and
volume. A better approach is to use the correlation proposed by
(~x,v;,n) ]
and their mixtures. 18 The method can be applied to a variety of liq-
uids with sufficient accuracy and reliability. From Eq. C-46, the liq- x (C-53)
uid density can be defined as
L
i=1
L where a, b, and c are determined from Table C-2. The last resort in
determining the characteristic molal volume, V', is to equate it to
i-C 4
n-C4
i-C s
181.5
189.9
225.0
the critical volume, Vc ; i.e., n-C s 231.5
V' = Vc ' ••••.•••••••••••••••••••••••••••••• (C-58) n-C6 271.0
n-C7 312.5
T2
0
1/4
= ~](PCh);<Axj - BY;>], (C-70) where A = Helmholtz energy at any specified state, (Jzmol):
i= I AO= Helmholtz energy at any reference state or an ideal gas state;
p=system pressure; T=system temperature; V=system volume;
with
and yO = reference volume.
The entropy departure function, 5, can be calculated as
A = 62 ~~) (C-7Ia)
. aL v
and S- SO = - d~(A - A"IV = J[(~i.), - ~]dV + RI';;.
B = 62.~~aS (C-7Ib)
(C-75)
where P a = density of the oil phase.Ibm/ft-: MaL = apparent molecu- where SO = entropy at reference state, (l/mol - K).
lar weight of the oil phase; P g = density of the gas phase, Ibmlft3; The departure function for enthalpy, H, is presented as
Mag = apparent molecular weight of the gas phase; Xi = mole fraction H - W = (A - A 0) + 1(5 - SO) + RT(Z - 1).
of Component i in the oil phase; Yi = mole fraction of Component i
in the gas phase; and n = total number of components in the system. (C-76)
Note that the SRK EOSIO can be presented as
C.13 Thermodynamic Properties 1
In this section, we present a method based on EOS's to estimate the p = /! b - V(/+ b)' (C-77)
enthalpy and entropy departure from ideality. These variables are
very useful thermodynamic properties] and often can be related to where the values of Parameters a and b are presented in Table C-l.
With the SRK EOS, it can be shown that the departure functions for
well-operating variables, such as temperature change in the fluid. It
the Helmholtz free energy, entropy, and enthalpy for a pure material
is important to estimate thermodynamic-property variations as the or for a mixture of constant composition can be written as
other independent variables (pressure, temperature, and other such
parameters) in the system change. A - A 0 = _ RTln V - b _ ~ In V +b - RTln J::.
The variation of thermodynamic properties between two equilib- V b V YO'
rium states is independent of the path chosen from one state to the (C-78a)
other, as long as the pressure and temperature at each of the states
are known. For example, the enthalpy of a fluid system, H, can be rlJ V-b a V+b IV
5 - o = Rln-V- - 2bTln-V- + R nyo,
represented at any pressure, p ; and temperature, T, as H=fl.p,T).
Between two states at (Ph TI) and (PZ, Tz) the enthalpy change (C-78b)
can be represented as
and
L References
I. Reid, R.C., Prausnitz, J.N., and Poling, B.E.: The Properties ofGases
and Liquids, fourth edition, McGraw-Hill Book Co. Inc., New York
ries, American Chemical Soc., Washington, DC (1980) No. 133.
15. Slot-Petersen, C.: "A Systematic and Consistent Approach To Deter-
mine Binary Interaction Coefficients for the Peng-Robinson Equation
City (1987). of State," SPERE (November 1989) 488.
2. Ahmed, TH.: Hydrocarbon Phase Behavior, Gulf Publishing Co., 16. Graboski, M.S. and Daubert. TE.: "A Modified Soave Equation of
Houston (1989). State for Phase Equilibrium Calculations I: Hydrocarbon System," Ind.
3. Katz, D.L. et. al.: Handbook ofNatural Gas Engineering, McGraw-Hill Eng. Chem. hoc. Des. Dev. (1978) 17,443.
Book Co. Inc., New York City (1959). 17. Alani, G.H. and Kennedy, H.T: "Volume of Liquid Hydrocarbons at
t 4. Carnahan, B., Luther, H.A., and Wilkes, J.O.: Applied Numerical Meth- High Temperatures and Pressures," Trans., AIME (1960) 219, 288.
L ods, John Wiley & Sons, New York City (1969).
5. Wilson, G.: "A Modified Redlich-Kwong EOS Application to General
18. Hankinson, R.w. and Thomson, G.H.: "A New Correlation for Saturated
Densities of Liquids and Their Mixtures," AlChE J. (1979) 25, 653.
Physical Data Calculations," paper 15C presented at the 1968 AIChE 19. Carr, N.L., Kobayashi, R., and Burrows, D.B.: "Viscosity of Hydrocar-
Annual Meeting, Cleveland, Ohio, 4-7 May. bon Gases Under Pressure," Trans., AIME (1954) 201,264.
6. Pitzer, K.S.: 'The Volumetric and Thennodynarnic Properties of 20. Lee, A.L., Gonzalez, M.H., and Eakin, B.E.: "The Viscosity of Natural
Fluids," J. Am Chem. Soc. (1955) 77, No. 13,3427. Gases," JPT (August 1966) 997; Trans., AIME, 237,
7. Lee, B.I. and Kesler, M.G.: "A Generalized Thermodynamics Correla- 21. Lohrenz, J., Bray, B.G., and Clark, C.R.: "Calculating Viscosities of
tion Based on Three-Parameter Corresponding States," AIChE J. Reservoir Fluids From Their Compositions." JPT (October 1964)
(1975) 21, 510. 1171; Trans., AIME, 231.
8. Edmister, We.: "Applied Hydrocarbon Thermodynamics, Part 4: 22. Sugden, S.: 'The Variation of Surface Tension, VI. The Variation of
Compressibility Factors and Equations of State," Petroleum Refiner Surface Tension with Temperature and Some Related Functions," J.
(1958) 37, No.4, 173. Chern. Soc. (1924) 125,32.
9. van der Waals, J.D.: "On the Continuity of the Liquid and Gaseous 23. Weinaug, C. and Katz, D.L.: "Surface Tension of Methane-Propane
State," PhD dissertation, Sigthoff, Leiden, The Netherlands (1873). Mixtures," Ind. Eng. Chem. (1943) 25, 35.
\
'-
1
L 144 MULTIPHASE FLOW IN WELLS
t
Appendix D
L Tubing and Casing Properties
Tables D-I and D-2 provide the standard tubing and casing informa- from Production Operations by T.O. Allen and A.P. Roberts, Oil &
tion required for pipe-flow calculations. The information is taken Gas Consultants IntI. Inc., Tulsa, Oklahoma (1978).
I
I TABLE D-1-TUBING MINIMUM PERFORMANCE PROPERTIES
I
l... Threaded and Coupled (T&C) Integral Joint
Coupled Outside
Tubing SIZe Nominal Weight Diameter Joint Yield Strength
1.050 1.14 1.20 H40 0.113 0824 0.730 1.313 1.660 7,680 7,530 6,360 13,30
Y. 1.050 1.14 1.20 J55 0.113 0.824 0.730 1.313 1.660 10,560 10,360 8,740 18,290
1.050 1.14 1.20 C75 0.113 0.824 0.730 1.313 1660 14,410 14,120 11,920 24,940
1.050 1.14 120 N80 0.113 0.824 0730 1.313 1.660 15,370 15,070 12,710 26,610
1315 1.70 1.80 1.72 H40 0.133 1.049 0955 1.660 1.900 0.955 1.550 7,270 7,270 10,960 19,760 15,970
1 1.315 1.70 1.80 1.72 J55 0.133 1.049 0.955 1.660 1.900 0.955 1.550 10,000 10,000 15,060 27,160 21,960
1.315 1.70 180 1.72 C75 0.133 1.049 0955 1660 1.900 0.955 1550 13,640 13,640 20,540 37,040 29,940
1.315 1.70 1.80 1.72 N80 0.133 1.049 0.955 1.660 1.900 0.955 1.550 14,550 14,160 21,910 39,510 31,940
1.660 2.10 H40 0.125 1,410 1.286 1.880 5,570 5,270 22,180
1.660 2.30 2.40 2.33 H40 0.140 1380 1.286 2.054 2.200 1.286 1.880 6,80 5,900 15,530 26,740 22,180
1V4 1.660 2.10 J55 0.125 1.410 1.286 1.880 7,660 7,250 30,500
1.660 2.30 2.40 2.33 J55 0.140 1.380 1.286 2054 2.200 1.286 1.880 8,500 8,120 21,360 36,770 30,500
1.660 2.30 2.40 2.33 C75 0.140 1.380 1.286 2054 2200 1.286 1.880 11,580 11,070 29,120 50,140 41,600
1.660 2,30 2.40 2.33 N80 0.140 1.380 1.286 2054 2.200 1.286 1880 12,360 11,810 31,060 53,480 44,370
1.900 2.40 H40 0.125 1.650 1.516 2.110 4,920 4,610 26,890
! 1.900 2.75 2.90 2.76 H40 0.145 1.610 1.516 2200 2.500 1.516 2.110 5,640 5,340 19,090 31,980 26,890
36,970
1.516 2.110 6,640 6,330
I '--
1% 1.900
1.900 2.75 2.90
2.40
2.76
J55
J55
0.125
0.145
1.650
1610 1.516 2.200 2.500 1.516 2.110 7,750 7,350 26,250 43,970 36,970
50,420
1.900 2.75 2.90 2.76 C75 0.145 1610 1.516 2.200 2.500 1.516 2.110 10,750 10,020 35,800 59,960
1.900 2.75 2.90 2.76 N80 0.145 1.610 1.516 2.200 2.500 1.516 2.110 11,280 10,680 38,180 63,960 53,780
I 2.063 3.25 H40 0.156 1.751 1.657 2.325 5,590 5,290 35,690
2.063 3.25 J55 0.156 1,751 1657 2325 7,690 7,280 49,070
I 2'/'6
2.063 3.25 C75 0.156 1.751 1.657 2.325 10,480 9,920 66,910
L 2.063 3.25 N80 0.156 1.751 1.657 2325 11,180 10,590 71,370
2.375 4.00 H40 0.167 2.041 1.947 2.875 5,230 4,920 30,130
2.375 4.60 4.70 H40 0.190 1.995 1.901 2.875 3.063 2.910 5,890 5,600 35,960 52,170
2.375 4.00 J55 0.167 2.041 1947 2.875 7,190 6,770 41,430
2.375 4.60 4.70 J55 0.190 1.995 1.901 2.875 3.063 2,910 8,100 7,700 49,450 71,730
2.375 4.00 C75 0.167 2.041 1.947 2.875 9,520 9,230 56,500
2.375 4.60 4.70 C75 0.190 1.995 1.901 2.875 3063 2.910 11,040 10,500 67,430 97,820
23/8 2.375 5.80 5.95 C75 0.254 1.867 1.773 2.875 3063 2.910 14,330 14,040 96,560 126,940
2.375 4.00 N80 0.167 2.041 1.947 2.875 9,980 9,840 60,260
l 2.375 4.60 4.70 N80 0.190 1.995 1.901 2.875 3.063 2,910 11,780 11,200 71,930 103,340
il... 2.375
2.375
5.80
4.60
5.95
4.70
N80
Pl05
0.254
0.190
1.867
1.995
1.773
1.901
2875
2.875
3.063
3.063
2.910
2.910
15,280
15,460
14,970
14,700
102,990
94,410
135,400
136,490
2.375 5.80 5.95 Pl05 0.254 1.867 1.773 2.875 3.063 2.910 20,060 19,650 135,180 177,710
2875 6.40 6.50 H40 0.217 2.441 2.347 3.500 3.668 3.460 5,580 5,290 52,780 72,480
2.875 6.40 6.50 J55 0.217 2.441 2.347 3.500 3.668 3.460 7.680 7.260 72,580 99,660
2.675 6.40 6.50 C75 0.217 2.441 2.347 3500 3.668 3460 10,470 9,910 98,970 135,900
2 7/8 2.875 8,60 8.70 C75 0.308 2.259 2.165 3.500 3.668 3.460 14,350 14,060 149,360 186,290
2.875 6.40 650 N80 0.217 2.441 2.347 3.500 3.668 3.460 11,160 10,570 105,570 144,960
2.875 860 8.70 N80 0308 2.259 2.165 3.500 3.668 3.460 15,300 15,000 159,310 198,710
2.875 6.40 6.50 Pl05 0.217 2.441 2.347 3.500 3.668 3.460 14,010 13,870 138,560 190,260
2.875 8.60 8.70 Pl05 0.308 2.259 2.165 3.500 3.668 3.460 20,090 19,960 209,100 260,810
TABLE D-1-TUBING MINIMUM PERFORMANCE PROPERTIES (continued)
Threaded and Coupled (T&C) Integral Joint
Coupled Outside
Tubing Size Nominal Weight Diameter Joint Yield Strength
3.500 7.70 H40 0.216 3.068 2.943 4.250 4,630 4,320 65,070
3500 9.20 9.30 H40 0.254 2.992 2.867 4.250 4.500 4.180 5,380 5,080 79,540 103.610
3.500 10.20 H40 0,289 2.922 2.797 4.250 6,060 5,780 92,550
3.500 7.70 J55 0.216 3.068 2.943 4.250 5,970 5,940 89,470
3500 9.20 9.30 J55 0.254 2.992 2,867 4.250 4.500 4.180 7,400 6,980 109,370 142.460
3500 10.20 J55 0.289 2.922 2.797 4,250 8,330 7,950 127,250
3.500 7.70 C75 0.216 3.068 2.943 4.250 7,540 8,100 122,010
3500 9.20 9.30 C75 0.254 2.992 2.867 4.250 4.500 4.180 10,040 9,510 149.140 194,260
3 1/ 2 3.500 10.20 C75 0.289 2.922 2.797 4.250 11,360 10,840 173,530
3500 12.70 12.95 C75 0.375 2.750 2.625 4.250 4.500 4.180 14,350 14,060 230,990 276,120
3500 7.70 N80 0.216 3.068 2.943 4.250 7,870 8,640 130,140
3500 9.20 9.30 N80 0.254 2.992 2.867 4.250 4.500 4.180 10,530 10,160 159,090 207,220
3500 10.20 N80 0.289 2.922 2.797 4.250 12,120 11,560 185,100
3500 12,70 12.95 N80 0.375 2.750 2.625 4.250 4.500 4.180 15,310 15,000 246,390 294,530
3500 9.20 9.30 Pl05 0.254 2.992 2.867 4.250 4.500 4.180 13,050 13.330 208,800 271,970
3500 12.70 12.95 Pl05 0.375 2.750 2.625 4.250 4.500 4.180 20,090 19,690 323,390 386,570
4.000 9.50 H40 0.226 3.548 3.423 4.750 4,060 3,960 72,000
4.000 11.00 H40 0.262 3.476 3351 5.000 4,900 4,580 123,070
4000 9.50 J55 0.226 3.548 3.423 4,750 5,110 5,440 99,010
4 4.000 11.00 J55 0.262 3.476 3.351 5.000 6,590 6,300 169,220
4.000 9,50 C75 0,226 3548 3.423 4.750 6,350 7,420 135,010
4.000 11.00 C75 0.262 3.476 3.351 5.000 8,410 8,600 230,750
4000 9.50 N80 0.226 3.548 3.423 4.750 6,590 7,910 144,010
4.000 11.00 N80 0.262 3.376 3.351 5.000 8,800 9,170 246,140
4.500 12.60 12.75 H40 0.271 3.958 3.833 5.200 5.563 4,500 4,220 104,360 144,020
4V2 4.500 12.60 12.75 J55 0.271 3.958 3.833 5.200 5.563 5,720 5,800 143,500 298,030
4.500 12.60 12.75 C75 0.271 3.958 3.833 5.200 5.563 7,200 7,900 195,680 270,040
4500 12.60 12.75 N80 0.271 3.958 3.833 5.200 5.563 7,500 8,430 208,730 288,040
00 Nominal Coupling
Size Weight Wall 10 Drift 00
(in.) (Ibf/ft) (in.) (in.) (in) (in.) H40 J55 N80 PlIO H40 J55 N80 PlIO H40 J55 N80 PlIO J55 N80 PlIO
9.50 0.205 4.090 3965 5.000 2,nO 3,310 3,190 4,380 77 101
4V2 11.60 0.250 4.000 3.875 5000 4,960 6,350 7,560 5,350 7,780 10,690 154 162 223 279
13.50 0290 3920 3.795 5.000 8,540 10,670 9,020 12,410 270 338
15.10 0.337 3.826 3.701 5.000 14,320 14,420 406
1400 0244 5.012 4.887 6.050 2,630 3,120 3,110 130 172
15.50 0.275 4.950 4.825 6.050 4,040 4,270 202 217
5V2 17.00 0.304 4.892 4.767 6.050 4,910 6,280 7,460 4,810 1,740 10,840 229 247 348 445
20.00 0.361 4.778 4.653 6.050 8,830 11,080 5,320 9,190 12,640 428 548
23.00 0,415 4.670 4.545 6.050 11,160 14,520 9,880 13,580 502 643
20.00 0.288 6.049 5.924 7.390 2,520 2,970 3,040 4,180 164 245 266
65fa 24.00 0352 5.921 5.796 7390 4,560 5,760 6,710 5,110 7,440 10,230 314 340 481 641
28.00 0,417 5.791 5.666 7.390 8,170 10,140 8,810 12,120 586 781
32.00 0.475 5675 5.550 7.390 10,320 13,200 10,040 13,800 677 904
l
L
146 MI'lT1Pl-l",<;F FI OW Tl\IWFT T <;
TABLE D-2-CASING MINIMUM PERFORMANCE PRO.,ERTIES (continued)
Joint Yield Strength (1,000 Ibf)
00 Nominal Coupling
Size Weitt Wall 10 Orltt 00
(in.) (Ibftt) (in.) (in.) (in.) (in.) H40 J55 N80 Pll0 H40 J55 N80 Pll0 H40 J55 N80 Pl10 J55 N80 Pll0
L 133/8
48.00 0.330
54.50 0.380
61.00 0.430
68.00 0.480
72.00 0.514
12,715
12.615
12515
12.415
12.347
12.559
12.459
12.359
12259
12.191
14.375
14.375
14.375
14375
14.375
740
1,130
1,540
1,950
2,670
1,730
2,730
3,090
3,450
5,380
322
514
595
675
1,040
185fa 87.50 0.435 17.755 17.567 19.625 650 630 1,630 2,250 559 754
94.00 0.438 19.124 18.936 21000 520 520 1,530 2,110 581 784 907
20 106.50 0500 19.000 18.812 21.000 770 2,410 913 1,057
133.00 0.635 18.730 18.542 21.000 1,500 3,060 1,192 1,380
L
L
I
L TUBING AND CASING PROPERTIES ".,