SPE 102583
Optimized Tubing String Design Modeling for Improved Recovery
B.D. Poe Jr., SPE, Schlumberger
Copyright 2006, Society of Petroleum Engineers
This paper was prepared for presentation at the 2006 SPE Annual Technical Conference and
Exhibition held in San Antonio, Texas, U.S.A., 24–27 September 2006.
This paper was selected for presentation by an SPE Program Committee following review of
information contained in an abstract submitted by the author(s). Contents of the paper, as
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
for commercial purposes without the written consent of the Society of Petroleum Engineers is
prohibited. Permission to reproduce in print is restricted to an abstract of not more than
300 words; illustrations may not be copied. The abstract must contain conspicuous
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.
Abstract
This paper presents the results of an investigation of the
design of production tubing string setting depths in gas wells to
optimize gas recovery in wells that produce free liquids in
conjunction with the gas. Particularly important in this work
has been the evaluation of the conditions for which the well
outflow velocity is less than that which would be required to
continuously transport and unload liquids from the well. Sub-
critical velocities are often encountered in low productivity gas
wells that produce liquids, whether the wellbore liquids are
produced directly from the formation and/or condensed from
the gas in the wellbore. The produced liquids considered in the
analysis can be water and/or liquid hydrocarbons.
This paper presents an optimization technique for
determining the most efficient production tubing string setting
depth design that will keep the wellbore continuously unloaded
of produced liquids, yet result in the maximum gas recovery
possible under those conditions. The optimization can be
performed for a single completion operation during the
productive life of the well, or for a series of periodic well
interventions during the productive life of the well for which the
tubing string setting depths are selected.
The wellbore liquid-loading mechanism in a sub-critical
velocity production system has been considered in the analysis
using a mechanistic wellbore outflow model. A commingled
reservoir inflow performance model was utilized in the analysis
to evaluate the fluid inflow of multilayer gas reservoirs that may
produce liquids in conjunction with the gas.
Introduction
A production tubing string setting depth optimization
analysis has been developed to evaluate the optimum tubing
string setting depth design to use for a multilayer commingled
gas reservoir that produces formation or wellbore condensed
liquids. The multiphase flow and pressure traverse models that
are commonly utilized in production systems analysis models
generally assume that steady flow conditions exist in the
wellbore. This in turn generally requires that a sufficient system
outflow rate or fluid velocity exists to continuously transport
any wellbore liquids present in the system to the surface. Gas
flow velocities that are insufficient to continuously transport the
wellbore liquids to the surface are referred to in this work as
sub-critical velocities.
In this research effort, as well as in the discussion presented
in this paper concerning the optimization model, it has generally
been assumed that the wellbore and the flow path under
consideration relatively vertical. The methodology discussed in
this paper, as well as the production optimization model in
general, is not necessarily restricted to use for vertical wells
only. In spite of this fact, it has been noted that laboratory
experiments have clearly demonstrated that only slight changes
in the inclination of the wellbore can have a dramatic effect on
the types of flow and flow regimes exhibited in a multiphase
flow system.
In situations where the outflow velocities are sub-critical the
wellbore liquids are not continuously and uniformly transported
up the wellbore to the surface. Turner et al1 identified that there
are two transport mechanisms that must be considered in
evaluating the transport capability of a system for moving
liquids upward in the well. These are the criteria for moving
the liquid film along the wall of the conduit upward, as well as
the criteria for suspending and transporting entrained liquid
droplets in the flow stream upward. It has been observed by
Turner et al1 and Oudeman9 that the flow stream velocity
required to continuously move the liquid film along the conduit
wall is consistently greater than that required to suspend and lift
the entrained liquid droplets upward in the flow stream. As
noted by Turner et al1, as well as from field observations of
wellbore liquid unloading conditions2-9 that the criteria for the
translation of entrained liquid droplets in the flow stream
generally provides a better indication of the ability of a well to
continuously transport wellbore liquids to the surface.
In cases where the flow stream velocity is not sufficient to
continuously transport the well liquids upward in the wellbore, a
wellbore liquid loading condition begins. In low productivity
gas wells, this condition commonly occurs. In such cases, some
of the wellbore liquid falls back downward in the wellbore. It is
conceived that generally this process occurs in such a way that
2 SPE 102583
some of the liquid in the flow stream coalesces along the
conduit wall and moves downward as a liquid film. If left
unchanged for a period of time, a well that produces at sub-
critical velocities may load up and cease flowing under its own
potential due to an increase in the liquid accumulation in the
wellbore. Sutton et al5 and Dousi et al11 have also demonstrated
that gas wells can continue to produce for a significant period of
time at sub-critical velocities. The wells do not necessarily
immediately cease producing once the Turner et al1 criteria for
entrained liquid droplet removal has been reached.
The components of the tubing design optimization analysis
developed in this study are detailed in the following sections of
this paper. These include the specification of a criteria for
establishing the critical gas velocity in the well for continuous
liquid unloading, a sub-critical gas velocity analysis when that
criteria is not met, a discussion of annular flow equivalent pipe
diameter models that are available for use in tubular flow
pressure traverse models since annular flow is commonly
employed in low productivity gas wells that are subject to liquid
loading conditions, and the optimization techniques used in this
analysis to determine the optimum tubing setting depth that
results in the maximum cumulative gas production while
maintaining continuous unloading of the wellbore liquids.
Example results obtained with the critical gas velocity model
employed in this study for the continuous unloading of wellbore
liquids are presented in this paper. A field example is also
presented of a well that has exhibited liquid loading problems.
The application of the optimization analysis described in this
paper results in estimates of the optimum intervention time
when the tubing string should be run in the well, the optimum
tubing setting depth in the wellbore, and the corresponding
ultimate cumulative gas production that would be achieved with
the optimum tubing string design are also summarized for this
field example.
Critical Gas Velocity for Continuous Liquid Unloading
In cases where the gas outflow velocity is less than the
critical velocity (sub-critical velocity), the corresponding
liquid fallback in the wellbore and the evaluation of its effect
on the liquid holdup measurements obtained must be
considered. A commonly used model for evaluating the
minimum gas outflow velocity that is required to continuously
remove liquids from the wellbore is one proposed by Turner et
al1. In the Turner et al1 approach, a balance between the
upward turbulent force and the downward gravitational force
on a liquid droplet in a gas flow stream is resolved in
conjunction with a criteria for the maximum liquid droplet size
to obtain a minimum gas outflow velocity to continuously
transport the liquid droplets to the surface, based on the slip
velocity. This expression is based on a critical Weber number
of 30 and is given mathematically in Eq. 1. The Weber
number is the ratio of the velocity force to the surface tension
force and is defined mathematically (in consistent units) with
Eq. 2.
1.2978 ⎡⎣σ ( ρl − ρ g ) ⎤⎦
0.25
vg =
Cd
0.25
ρ g 0.5
(1)
ρ vg 2 d p
NWe =
g cσ (2)
Turner et al1 suggested that a typical value of the drag
coefficient (Cd) appearing in Eq.1 should be approximately
equal to 0.44 and that the resulting minimum gas velocity used
in the analysis should be adjusted upward by approximately
20% to account not only for the slip velocity, but also for the
velocity to transport the liquid droplets. This is reasoned from
the fact that the minimum gas velocity required for
transporting the liquid droplets upward is the sum of the
minimum gas velocity required for suspending the liquid
droplets (i.e., slip velocity) in the flow stream and the
transport velocity of the droplets. The transport velocity may
be estimated from the liquid production rate, geometry of the
conduit, and the liquid holdup.
The minimum gas velocity required to continuously
transport the liquid droplets to the surface is observed to be a
function of the surface tension and the liquid and gas densities.
Even higher gas velocities than predicted by Eq. 1 are required
to transport the liquid as a film coating the tubing wall to the
surface, therefore at low gas flow rates the liquid production at
the surface has been concluded9 to be primarily controlled by
the liquid droplet transport mechanism described by Eq. 1.
Numerous investigators2-11 have utilized the Turner et al1
basic model in various forms for the evaluation of liquid
loading issues, and some have proposed modifications to the
Turner et al1 model given in Eq. 1. Coleman and co-
workers2-4 have reported that the 20% upward adjustment
proposed by Turner et al1 to account for the transport velocity
component is unnecessary. However, other studies by Guo et
al7 and Sutton et al5 indicate that the 20 % upward adjustment
in the slip velocity relationship to estimate the minimum gas
velocity required for continuous liquid unloading is valid.
Nosseir et al6 have provided a plausible explanation as to why
the conflicting results concerning the use of the Turner et al1
model and the need for an upward adjustment may have been
observed, as well as providing an extension of the Turner et
al1 model for minimum gas velocity for liquid unloading that
includes specific models for use in the transition
(1<NRE<1000) and highly turbulent (1000<NRE<200,000) flow
regimes.
As noted by Nosseir et al6, the original Turner et al1 model
was developed using data with Reynolds numbers mostly in
excess of 200,000, for which the corresponding drag
coefficient (Cd) of a particle with a unit sphericity is actually
about 0.2 (see Fig. 1). Nosseir et al6 also found that most of
the examples reported by Coleman and co-workers2-4 had
Reynolds numbers in the range of 104<NRE<2x105, for which
the drag coefficient is approximately equal to 0.44, which is
why they reported that no upward adjustment in the Turner et
al1 minimum gas velocity model for continuous liquid
unloading was required.
SPE 102583
Therefore, in the development of a computational analysis
for evaluating the slip velocity of liquids in gas flow, as well
as the corresponding minimum gas velocity for continuous
transport of liquids from the wellbore to the surface, the more
general models reported by Allen12 and Hinze13 that were later
utilized by Nosseir et al6 have been employed in a
computational methodology to evaluate the slip velocity of
liquids in gas flow as a function of the Reynolds number in the
evaluation of the minimum gas flow velocity for continuous
liquid unloading, thereby properly accounting for the effects
of the flow regime in the evaluation of the minimum gas
velocity for continuous liquid unloading.
The drag coefficient as a function of a spherical particle
(liquid droplet) in a flow stream can be obtained from the
graphical presentation in Fig. 1. A set of correlations has been
developed in this study to reliably reproduce the relationship
for the drag coefficient (referred in some references as the
friction factor14) as a function of Reynolds number. This set
of correlations is given in Eq. 3. Note that the first expression
in Eq. 3 (for N RE ≤ 0.04 ) can be simply written as the classic
result for laminar flow, Cd = 24 / N RE .
[1.3802112417116 − log N RE ]
Cd = 10 , N RE ≤ 0.04
⎡1.4503131986 − 0.9043197142log N RE
⎢⎣ +0.0594644891(log N RE )2 + 0.0136391233(log N RE )3 ⎥⎦
Cd = 10
Cd = 0.4,
⎡50.018617877 − 44.799442653log N RE ⎤
⎢ +14.701249478(log N RE )2 − 2.1119561632(log N RE )3 ⎥
⎢ ⎥
Cd = 10 ⎣ +0.11234036938(log N RE )
4
⎦
⎡ −3.1908969689 E 7 + 3.7715081936 E 7 log N RE
⎢ −1.8571668496 E 7(log N RE )2 + 4.876747783 E 6(log N RE )3 ⎥
⎢ 4
−7.2023948102 E 5(log N ) + 5.6724016198 E 4(log N
⎢ RE
⎣⎢ −1.8611888434 E 3(log N RE )
6
Cd = 10
⎡ −59.903600523+ 30.176370129log N RE ⎤ tubular where the velocity is lower). At gas velocities below
⎣⎢ −5.1228268992(log N RE ) + 0.28985939161(log N RE ) ⎦⎥
2 3
Cd = 10
Cd = 0.22,
In the laminar flow regime (NRE<1), Stokes law can be
used to reliably estimate the critical terminal velocity, while a
form of the Nosseir et al6 relationship for the critical gas
velocity for continuous liquid unloading may be used in the
transition flow regime (1<NRE<1000). The transition regime
model reported by Nosseir et al6 contains an exponent for the
gas density term in the denominator of 0.426 in the original
article, when in fact the derivation indicates that a slightly
different value of 0.43155 is more correct. The corrected form
of the relationship for the critical gas velocity for continuous
liquid unloading developed from the Allen12 and Hinze13
relationships expressed in conventional units that is applicable
in the transition flow regime is given by Eq. 4.
0.1186 σ 0.3512 ( ρl − ρ g )
0.2143
vg =
μ 0.1339 ρ g 0.43155
3
However, as noted in Ref. 14, the slip velocity (terminal
settling velocity) of a particle (or liquid droplet in this case) in
a Newtonian fluid can be evaluated using the general form of
the critical velocity relationship given in Eq. 1 (which is in
fact derived in Ref. 14 from an energy balance) for turbulent,
transition, and even (low Reynolds number) laminar flow if
the proper drag coefficient (or friction factor) is used for the
appropriate particle Reynolds number. Therefore, in this
study the general form of the terminal velocity relationship
given in Eq. 1 has been used in conjunction with the drag
coefficient relationship of Eq. 3 to estimate the critical gas
velocity for all values of Reynolds numbers. The particle
Reynolds number in the analysis can be evaluated using Eq. 5,
which was derived from the Hinze13 relationship for the
critical droplet size at a Weber number of 30 and the
fundamental definition of Reynolds number.
98.42518 σ
N RE =
μ g vg
(5)
Sub-Critical Gas Velocity Analysis
In cases where the actual gas velocity is less than the
critical (minimum) gas velocity for continuous liquid removal
⎤
from the well, liquid may accumulate in the wellbore. Also,
, 0.04 < N RE < 2, 000 when there is insufficient gas velocity to continuously remove
the liquid from the well, liquid fallback can occur. This
2, 000 ≤ N RE ≤ 6, 000
results in an increase in the liquid holdup over that which
would be exist if the liquids were continuously removed from
, 6, 000 < N RE ≤ 100, 000 the wellbore at gas velocities above the critical value.
⎤
The liquid holdup at each depth in the well for which sub-
5⎥
critical velocity conditions exist includes the effects of both
)
RE ⎥
⎦⎥
the liquid being transported upward in the flow stream, as well
, 100, 000 < N RE ≤ 230, 000 as the liquid falling back (typically near the walls of the
, 230, 000 < N RE < 600, 000 the critical velocity, a static liquid column may develop.
N RE ≥ 600, 000 Hasan et al8 have developed a mechanistic pressure
traverse model for multiphase flow in wells with sub-critical
(3)
velocities that compares favorably over that range with the
correlation developed by Papadimitriou15. In the bubble flow
regime for pipe flow, the bubble rise velocity may be
computed using Eq. 6.
⎡ gσ ( ρl − ρ g ) ⎤
0.25
vb = 1.5 ⎢ ⎥
⎢⎣ ρl 2 ⎥⎦
(6)
The liquid holdup in tubular flow for the bubble flow regime
can therefore be evaluated using Eq. 7.
vsg
Yl = 1 −
1.97 ( vsg + vsl ) + vb
(7)
The corresponding expression for the liquid holdup in the
(4) bubble flow regime for annular flow is given by Eq. 8.
4 SPE 102583

vsg Various annular flow equivalent pipe diameter

Yl = 1 − relationships that have often been used may be found in Ref.
⎛ dt ⎞
⎜1.97 + 0.371 ⎟ ( vsg + vsl ) + vb
14. There is a wide range of opinions in the industry
dc ⎠ concerning the computation of the equivalent tubular diameter
⎝ (8) (de) for annular flow when applied in tubular flow pressure
traverse computations.14,16,and 17 One of the simplest, though
In the slug flow regime, which has been reported by Hasan not necessarily the most accurate annular flow equivalent pipe
et al8 to occur at liquid holdup values of less than 0.75, the flow diameter commonly used in multiphase flow pressure
Taylor bubble rise velocity can be evaluated using Eq. 9 for traverse computational analyses is one that can be derived
tubular flow. from the hydraulic radius concept. In an annulus of two
⎡ gdt ( ρl − ρ g ) ⎤
0.5 concentric cylinders, the hydraulic radius is equal to the cross-
vTb = 0.3 ⎢ sectional area to flow divided by the wetted perimeter of the
⎥
⎣⎢
ρl ⎦⎥
conduit. In this case, the equivalent tubular flow diameter of
(9) an annulus created by an outer tubular (casing) with an inside
diameter of dc, and an inner cylindrical boundary (tubing) with
The liquid holdup in tubular flow in the slug flow regime is an outside diameter of dt, is given by 4 times the hydraulic
therefore evaluated using Eq. 10. radius, or simply as the difference between the diameters of
the outer and inner cylindrical diameters defining the flow
vsg conduit dimensions.
Yl = 1 −
1.182 ( vsg + vsl ) + vTb de = 4rh = d c − dt (13)
(10)

In annular flow, the Taylor bubble rise velocity is Other formulations of the annular flow equivalent pipe
evaluated using Eq. 11 and the corresponding liquid holdup is diameter for use in tubular flow production systems analysis
obtained with Eq. 12. pressure traverse computations may be found in Ref. 14. One
formulation of an equivalent pipe diameter that can be derived
using the concept of an equivalent radius for pipe flow is one
⎞ ⎡ g ( d c − d t ) ( ρl − ρ g ) ⎤
0.5
⎛ d that would result in the same pressure loss as the actual
vTb = ⎜ 0.3 + 0.22 t ⎟⎢ ⎥ annular flow configuration under laminar flow conditions. In
⎝ dc ⎠ ⎢⎣ ρl ⎥⎦
(11) terms of the concentric cylinders that delimit an annulus, the
equivalent pipe diameter is given by Eq. 14.
vsg
Yl = 1 − 2
d c − dt
2
⎛ dt ⎞ 2
de = d c + dt −
2

⎜1.182 + 0.9 ⎟ ( vsg + vsl ) + vTb

dc ⎠
⎝ (12)
Equivalent Pipe Diameter Models for Annular Flow
Various models for the annular flow equivalent pipe
diameter for use in pressure traverse computations have been
reported in the literature. While the bubble rise velocity and
liquid holdup relationships of the Hasan et al8 mechanistic
model were developed from laboratory measurements
specifically for annular flow conditions and do not require the
evaluation of an equivalent pipe diameter number for annular
flow, other commonly used pressure traverse relationships16,17
often do require an equivalent pipe diameter value for annular
flow computations.
There is not a consensus amongst industry practioners as to
the most appropriate equivalent pipe diameter relationship to
use for annular flow computations to achieve the best
accuracy. It has been found in this investigation that an
equivalent pipe diameter for annular flow using the slot flow
approximation or the hydraulic radius models presented in the
following discussion generally provide reasonable and reliable
results. The annular flow equivalent pipe diameter model
used can be quite significant, since it is utilized in the
traditional multiphase pressure traverse models reported in the
literature.16,17
(14)
⎛d ⎞
ln ⎜ c ⎟
⎝ dt ⎠
An expression for the equivalent annular flow pipe
diameter for use in tubular flow pressure traverse
computations can be derived from the slot flow
approximation. The flow in the annulus is approximated as
flow between two infinite parallel plates. This expression is
given in Eq. 15 and has been found to produce essentially the
same result as the expression given in Eq. 14 for most
practical cases encountered in oilfield completions, with
typical casing and tubing diameters.
2
de = ( d c − dt ) (15)
3
In each of the expressions for the equivalent pipe diameter
for annular flow (Eqs.13-15), the area that is used to compute
the average velocity in the pressure traverse computations is
the actual cross-sectional area of the flow conduit14. The
average velocity of fluid flow in the annulus between the
casing and tubing is given by Eq.16, in which the average
velocity is given in units of ft/s, the superficial in situ flow rate
SPE 102583
is in units of ft3/s and the casing and tubing diameters are in
units of inches.
183.35q
v= 2 2
d c − dt
An alternate annular flow equivalent pipe diameter
relationship reported in the literature that is available for
computing the production systems analysis pressure traverse,
has been derived empirically, yet it is more complex than the
annular flow equivalent pipe diameter expressions given
previously in Eqs.13-15. The equivalent pipe diameter
relationship developed by Crittendon17 was obtained using the
production performance data of hydraulically fractured wells.
This expression is given in Eq. 17.
4
de = d c − d t
4
−
(d c
2
− dt )
2 2
2
+ d c − dt
2
⎛d ⎞
ln ⎜ c ⎟
⎝ dt ⎠
Crittendon’s17 equivalent pipe diameter relationship for
annular flow computations with tubular flow relationships
utilizes the equivalent diameter (de) to compute a fictitious
velocity that is used in the production systems analysis
pressure traverse computations14. The fictitious velocity is
defined as in Eq. 18, in which the equivalent pipe diameter is
given in units of inches, the superficial in situ fluid flow rate is
given in units of ft3/s, and the resulting fictitious velocity has
the units of ft/s.
183.35q
v* =
de
2 generally be well correlated in terms of these parameters.
Critical Velocity Correlations
The simultaneous solution of Eqs. 1, 3, and 5 presented
earlier in this paper provide a means of correlating the critical
gas velocity required to continuously transport entrained liquid
droplets upward in a flow stream. Since the drag coefficient
relationship given in Eq. 3 has been fitted with a piecewise
continuous function that is not varying monotonically with
increasing Reynolds number, a root-solving technique such as
bisection applied in a piecewise manner has been found to be
appropriate.
The fluid properties required for the solution of the critical
gas velocity to upwardly transport entrained liquids are the gas
viscosity and density and the liquid density and surface
tension. Relationships for the viscosity and density of oil, gas,
and water are generally readily available. However,
relationships for the surface tension of water and hydrocarbon
liquids are much more limited in the literature. A correlation
for the surface tension of water18 is presented in Fig. 2. The
surface tension for water is shown to not vary too significantly
over the practical range of interest (from about 75 dynes/cm at
40 oF to about 60 dynes/cm at 200 oF) and a somewhat average
5
value of this property of about 70 dynes/cm is often used in
engineering applications for pressure traverse modeling16.
The surface tension of oil at atmospheric pressure for a
range of oil densities16 is presented in Fig. 3. The ratio of the
(16) surface tension of oil at elevated pressure relative to
atmospheric pressure conditions is given in Fig. 4. The
surface tension for oil at elevated pressure is therefore
determined as a product of the results obtained from Figs. 3
and 4. Note that the oil surface tension is more dependent on
the environmental conditions of pressure and temperature than
water, and a single average value for a wide range of oil
gravity, and pressure and temperature conditions is not
appropriate.
Using the drag coefficient relationship in Fig. 1 and the
fluid surface tension correlations presented in Figs. 2 through
4, correlations for the critical gas velocity required for the
continuous upward transport of entrained liquid droplets as a
function of the gas density and gas viscosity can be generated
for water or oil. In systems that produce only hydrocarbon
(17)
liquids and no water, oil is the entrained liquid phase that is
used in the analysis for the critical gas velocity. In systems
that produce water, whether or not liquid hydrocarbons are
present, water is the entrained liquid phase that is used in the
analysis since the gas velocity required to suspend and lift the
more dense liquid (water) upward is greater than that required
to transport an entrained hydrocarbon liquid phase upward in
the system. This criterion determines which liquid phase
transport analysis defines the critical gas velocity of the
system.
Figure 5 is an illustration depicting the critical gas velocity
required to suspend and transport upward entrained water in
the gas flow stream as a function of the gas viscosity and gas
density. This correlation could also have been developed for
the critical gas velocity as a function of the gas specific
gravity, and pressure and temperature conditions since the gas
pseudocritical and physical properties (ρg and μg) can
(18)
Figure 6 is presents the results of the critical gas velocity
computations for a 40 oAPI oil at a temperature of 100 oF and
pressure of 100 psia. Note that there does not appear to be a
significant effect on the critical gas velocity due to the effects
of gas viscosity as there was in the entrained water example
presented in Fig. 5.
Note also that this behavior is also generally apparent in
Figs. 7 and 8 for pressures of 500 and 1,000 psia, respectively.
Therefore, a more appropriate characterization of the critical
gas velocity for an entrained 40o API oil analysis is presented
in Fig. 9 in which the critical gas velocity is correlated as a
function of pressure and gas density. The oil density in this
example was chosen to be somewhat representative of a
volatile oil or condensate fluid. Similar graphical correlations
can be readily developed for other hydrocarbon liquid
gravities.
These examples have been presented to demonstrate the
use of the critical gas velocity criteria in the optimization
analysis. Examples of the mechanistic model liquid holdup
relationships for the bubble and slug flow regimes could also
be developed to demonstrate the effect of that component of
the sub-critical velocity.
6 SPE 102583
Optimization Technique
The optimization techniques used in this work and the
corresponding optimization constraints employed were
selected to determine the maximum gas cumulative production
that could be obtained while maintaining the continuous
transport of liquids from the wellbore. A comprehensive
multilayer reservoir inflow performance model capable of
generating the commingled system response of a gas reservoir
that may completed with a combination of unfractured,
vertically fractured, or horizontal drainhole completions was
employed in this study.
The reservoir inflow performance model was coupled to a
wellbore outflow performance model that contained both a
fairly comprehensive set of steady state multiphase pressure
traverse correlations and a mechanistic model for sub-critical
velocity analyses. The coupled system response that was
maximized in this analysis was the cumulative gas production.
The constraints applied in the analysis were (a) continuous
liquid unloading was required, defined by the critical gas
velocity criteria, (b) the range of the permissible tubing setting
depths was between a depth halfway between the surface and
the top of the shallowest completed interval in the well and the
depth of the wellbore, and (c) either a single completion
tubing design for the entire well life or a finite number of
periodic interventions over the well’s productive life at which
the tubing string design optimization could be performed.
The mathematical technique employed in this study for
determining the optimum tubing string setting depth that
corresponds to the maximum gas cumulative recovery was the
Golden Section. A more efficient maximization procedure
(Brent’s method) was later added to the analysis model as an
option to further improve the computational efficiency of the
optimization analysis.
Example Application
A field example was selected to illustrate the application of
the optimization analysis described in this paper. The
production flow rate history of the example gas well is
presented in Fig. 10, along with the production history match
obtained using a production systems analysis model generally
described in this paper. The corresponding cumulative
production and model match results are given in Fig. 11.
The reservoir in this example is relatively thick (150 ft)
and consists of a number of thin low-permeability sand layers,
but which generally tend to behave in this case as a single
productive sand body. A single stage hydraulic fracture
treatment was performed on this reservoir and the production
matches given in Figs. 10 and 11 correspond to that of an
equivalent single layer reservoir model. The well was initially
completed without a production tubing string, with production
initially via the casing.
The optimization analysis of this well performance
indicated that the optimum intervention time at which the
tubing string should have been run was approximately equal to
256 days. The optimum tubing setting depth obtained for a
single intervention production history of 10 years was
determined to be equal to 6822 ft, which resulted in an
optimized cumulative gas recovery of 0.877 bcf.
Summary
The tubing string setting depth optimization analysis
presented in this paper requires the use of a significant number
of components to properly accomplish the requirements
needed. These are a critical velocity criteria for establishing
the minimum gas flow rate for continuous liquid unloading
from the wellbore, a coupled comprehensive production
systems analysis model of the reservoir inflow and wellbore
outflow, a sub-critical gas velocity analysis that can be
employed for non-steady flow conditions, and the use of a
robust and stable optimization numerical technique for
determining the optimum tubing string design.
Each component of the analysis is necessary. A fully
coupled production systems analysis is required to properly
determine the system productivity. Simply employing a
criterion such as the critical gas velocity alone is not
sufficient, since a well will not necessarily immediately load
up and cease to flow the instant the gas velocity drops below
the critical value. Instead, the well will commonly continue to
produce for a period of time after that condition has been
satisfied, during which time liquid will continue to accumulate
in the wellbore until the well will ultimately cease flowing
continuously under natural means.
As a result of this research effort, it seems reasonably clear
that production tubing string setting depths can normally be
optimized to achieve the maximum cumulative gas recovery
possible for a well, while maintaining the continuous removal
of wellbore liquids. With a complete production systems
analysis approach, the maximization of the gas recovery can
be modeled fairly reliably.
Conclusions
Based upon the results of this research effort the following
pertinent conclusions have been obtained:
1. An optimization analysis for determining a
production tubing string setting depth that achieves
the maximum cumulative recovery of a gas well,
producing liquids and subject to liquid loading
conditions, has been developed.
2. Graphical correlations have been developed for the
critical gas velocity in systems with entrained water
and oil liquid phases.
3. A coupled reservoir inflow and wellbore outflow
production systems analysis is necessary to properly
evaluate the long-term performance of a gas well that
is subject to liquid loading and for which an optimum
tubing string setting depth design is to be derived.
4. The optimum intervention time and ultimate gas
cumulative recovery can be obtained with the
optimization analysis described in this work for gas
wells subject to liquid loading conditions.
Nomenclature
Variable
Cd Drag coefficient (friction factor in some references)
dc Inside diameter of casing, ft
dt Outside diameter of tubing, ft
de Annular flow equivalent pipe diameter for use in
tubular flow correlation determination of multiphase
flow pressure traverse, in
SPE 102583
dp Liquid droplet diameter, ft
g Acceleration due to gravity, ft/s2
gc Gravitational constant, lbm-ft/lbf-s2
NRE Reynolds number
NWE Weber number
q Flow rate, ft3/s
rh Hydraulic radius, ft
v In situ average fluid velocity, ft/s
v* Fictitious fluid velocity in Critendon14 analysis, ft/s
vb Bubble rise velocity, ft/s
vg In situ gas phase velocity, ft/s
vl In situ liquid phase velocity, ft/s
vs Slip velocity, ft/s
vsg Superficial gas velocity, ft/s
vsl Superficial liquid velocity, ft/s
vTb Taylor bubble rise velocity, ft/s
z Depth in the well, ft
Greek
σ Interfacial (surface) tension between liquid and gas,
dynes/cm
μg Gas viscosity, cp
ρg Gas density, lb/ft3
ρl Liquid density, lb/ft3
References
1. Turner, R.G., Hubbard, M.G, and Dukler, A.E.:
“Analysis and Prediction of Minimum Flow Rate for
the Continuous Removal of Liquids from Gas
Wells,” JPT (Nov. 1969) 1475-82; Trans., AIME,
246.
2. Coleman, S.B., Hartley, B.C., McCurdy, D.G, and
Norris, H.L. III: “Understanding Gas-Well Load-Up
Behavior,” JPT (Mar. 1991), 334-338.
3. Coleman, B.G., Hartley, B.C., McCurdy, D.G., and
Norris, H.L. III: “A New Look at Predicting Gas-
Well Load-Up,” JPT (Mar. 1991) 329-333.
4. Coleman, S.B., Hartley, B.C., McCurdy, D.G., and
Norris, H.L. III: “Applying Gas-Well Load-Up
Technology,” JPT (Mar. 1991) 344-349.
5. Sutton, R.P., Cox, S.A., Williams, E.G. Jr., Stolz,
R.P., and Gilbert, J.V.: “Gas Well Performance at
Subcritical Rates,” paper SPE 80887 presented at the
2003 SPE Production and Operations Symposium,
Oklahoma City, OK, March 22-25.
7
6. Nosseir, M.A., Darwich, T.A., Sayyouh, M.H., and
El Sallaly, M.: “A New Approach for Accurate
Prediction of Loading in Gas Wells Under Different
Flowing Conditions,” SPE Prod. & Facilities (Nov.
2000), 241-246.
7. Guo, B, Ghalambor, A., and Xu, C.: “A Systematic
Approach to Predicting Liquid Loading in Gas
Wells,” paper SPE 94081 presented at the 2005 SPE
Production and Operations Symposium, Oklahoma
City, OK, April 17-19.
8. Hasan, A.R., Kabir, C.S., and Rahman, R.:
“Predicting Liquid Gradient in a Pumping-Well
Annulus,” SPEPE (Feb. 1988) 113-20.
9. Oudeman, P.: “Improved Prediction of Wet-Gas-Well
Performance,” SPEPE (Aug. 1990) 212-216.
10. Duggan, J.O.: Estimating Flow Rates Required to
Keep Gas Wells Unloaded,” JPT (Dec. 1961), 1173-
1176.
11. Dousi, N., Veeken, C.A.M., and Currie, P.K.:
“Modelling the Gas Well Liquid Loading Process,”
paper SPE 95282 presented at the 2005 Offshore
Europe, Aberdeen, Scotland, Sept. 6-9.
12. Allen, H.S.: Philos. Mag. (1900) 50, 323.
13. Hinze, J.O.: “Critical Speeds and Sizes of Liquid
Globules,” Appl. Sci. Res., Sect. A, 1, No. 4, 273
(1949).
14. Applied Drilling Engineering,, A.T. Bourgoyne Jr.,
M.E. Chenevert, K.K. Millheim, and F.S. Young Jr.,
eds., SPE Textbook Series, Vol. 2, SPE, Richardson,
TX, 1986.
15. Papadimitriou, D.A.: A Mechanistic Model for
Predicting Annulus Bottomhole Pressures for Zero
Net Liquid Flow in Pumping Wells, MS Thesis, U. of
Tulsa (1990).
16. The Technology of Artificial Lift Methods, K.E.
Brown and H.D. Beggs, eds., Vol. 1, PennWell
Publishing Company, Tulsa Oklahoma, 1977.
17. Crittendon, B.C.: “The Mechanics of Design and
Interpretation of Hydraulic Fracture Treatment,” JPT
(Oct. 1959) 21-29.
18. Handbook of Chemistry and Physics, Weast, W.C.,
ed., CRC Press, 56th Edition, 1975-1976, Cleveland,
Ohio.
8 SPE 102583

100000

10000

1000

100
Cd

10

0.1
0.001 0.01 0.1 1 10 100 1000 10000 100000 1000000
N RE

Fig. 1 – Drag coefficient as a function of particle (unit sphericity) Reynolds number.

80

75

70
σw, dynes/cm

65

60

55
40 60 80 100 120 140 160 180 200
T, deg F

Fig. 2 – Surface tension of water.

SPE 102583 9

36

34 68 deg F
100 deg F

32

30
σoa, dynes/cm

28

26

24

22

20
20 25 30 35 40 45
ρ o , deg API

Fig. 3 – Surface tension of oil at atmospheric pressure.

1.2

0.8
σo /σoa

0.6

0.4

0.2

0
0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 2600 2800
P, psia

Fig. 4 – Surface tension of oil at elevated pressure relative to surface tension at atmospheric pressure.
10 SPE 102583

Entrained Liquid: Water (σ w =70 dynes/cm)

10

6
vg , ft/s

5 0.01 cp
0.02 cp
4
0.03 cp
3 0.04 cp

0
0 10 20 30 40 50 60 70
3
ρ g , lbs/ft

Fig. 5 – Critical gas velocity for systems with entrained water as a function of gas density and viscosity.

o
Entrained Liquid: 40 API Oil, P=100 psia

12

10
0.01 cp
0.02 cp
8
0.03 cp
0.04 cp
vg , ft/s

0
0 10 20 30 40 50 60
3
ρ g , lbs/ft

Fig. 6 – Critical gas velocity for entrained 40o API oil as a function of gas viscosity and density, P=100 psia.
SPE 102583 11

o
Entrained Liquid: 40 API Oil, P=500 psia

14

12

0.01 cp
10 0.02 cp
0.03 cp
8 0.04 cp
vg , ft/s

0
0 10 20 30 40 50 60
3
ρ g , lbs/ft

Fig. 7 – Critical gas velocity for entrained 40o API oil as a function of gas viscosity and density, P=500 psia.

o
Entrained Liquid: 40 API Oil, P=1000 psia

12

10
0.01 cp
0.02 cp
8
0.03 cp
0.04 cp
vg , ft/s

0
0 10 20 30 40 50 60
3
ρ g , lbs/ft

Fig. 8 – Critical gas velocity for entrained 40o API oil as a function of gas viscosity and density, P=1000 psia.
12 SPE 102583

o
Entrained Liquid: 40 API Oil, μ g=0.02 cp

6 100 psia
500 psia

5 1000 psia
1500 psia
2000 psia
4
2500 psia
vg , ft/s

0
0 10 20 30 40 50 60
3
ρ g , lbs/ft

Fig. 9 – Critical gas velocity for entrained 40o API oil as a function of gas density and pressure level.

6000 Actual qo (STB/D)

Actual qg (Mscf/D)
Actual qw (STB/D)
Model qo (STB/D)
5000 Model qg (Mscf/D)
Model qw (STB/D)

4000
Rate

3000

2000

1000

0
0 100 200 300 400 500 600
Time (day s)

Fig. 10 - Liquid loading modeling and production rate history match.

SPE 102583 13

10000 Actual Np (STB)

Actual Gp (MMscf)
Actual Wp (STB)
Model Np (STB)
Model Gp (MMscf)
Model Wp (STB)
1000
Cumulativ e Production

100

10

1
0 100 200 300 400 500 600
Time (day s)

Fig. 11 – Cumulative production modeling and history match.