You are on page 1of 22

DGMK/ÖGEW-Frühjahrstagung 2013, Fachbereich Aufsuchung und Gewinnung

Celle, 18./19. April 2013

State-of-the-art Review of Liquid Loading in Gas Wells


G. Falcone*, J.R. Barbosa Jr.**
*ITE, TU Clausthal,
**Universidade Federal de Santa Catarina, Department of Mechanical Engineering, Brasilien

Abstract
Gas wells suffering from liquid loading are incapable of removing the liquid associated with
produced gas from the wellbore. This phenomenon is initiated when the upward gas velocity
in the well falls below a critical value at which point the liquid that was initially flowing
upwards, begins to fall back. This liquid accumulates downhole, where it increases the
hydrostatic back-pressure on the reservoir, destabilises the multiphase flow in the well
(following flow regime changes), decreases production rate and, in severe cases, kills the
well. The typical liquid loading sequence begins with a gas flow rate that is high enough to
transport all liquids to surface and there is no liquid fall-back in the well. However, as the gas
velocity slows or the liquid content in the well rises, there is insufficient energy in the well to
carry all liquids to surface and some begins to flow backwards. As the hydrostatic head
downhole increases, the liquid column that has accumulated in the well can re-enter the
near-wellbore region of the reservoir. This results in the well becoming “unloaded” so that it
can flow once more, with the gas carrying all liquids to surface. However, the reinjection of
liquids into the reservoir may cause formation damage, which will impair the well productivity.
This cycle continues, providing the typical intermittent response of liquid-loaded gas wells,
until the reservoir potential starts to fall or the liquid yield rises.
Diagnosing liquid loading is often difficult as the affected well(s) may continue production
without any substantial performance impairment for a long period of time. Typical symptoms
of liquid loading include sharp drops in the cumulative production decline curve, the onset of
liquid slugs in the surface facilities, abrupt changes in the flowing pressure gradient, low
temperature spikes at the wellhead and declining water production or condensate-gas-ratio.
Many remedial lifting options have been developed for use in the field; some unloading
solutions (e.g. velocity strings) rely on the existing natural energy of the system, while others
(e.g. downhole pumps) provide extra energy to bring the water to surface, so reducing the
liquid loading problem. As each of the remedial options have their own technical
characteristics, their applicability varies depending on the characteristics and the status of
the well.
Although a number of established techniques are used to alleviate the effects of liquid
loading, the industry still lacks reliable predictive models to help select the best remedial
option for a particular loading occurrence.
In this paper, an up-to-date critical review of current methods to forecast the onset of liquid
loading and model the subsequent wellbore performance is presented. The review also
includes recent attempts to understand the dynamic interactions between reservoir and
wellbore during liquid loading.

Keywords
Liquid loading; multiphase flow; transient flow; near-wellbore region.

DGMK-Tagungsbericht 2013-1, ISBN 978-3-941721-31-9


129
DGMK/ÖGEW-Frühjahrstagung 2013, Fachbereich Aufsuchung und Gewinnung, Celle

Introduction
Liquid loading is initiated when the upward gas velocity in the well falls below a critical value,
at which point the liquid that was initially flowing upwards begins to fall back. This liquid
accumulates downhole, where it increases the hydrostatic back-pressure on the reservoir,
destabilises the multiphase flow in the well (following flow regime changes), decreases
production rate and, in severe cases, kills the well. The typical liquid loading sequence,
illustrated in Fig. 1, begins with a gas flow rate that is high enough to transport all liquids to
surface and there is no liquid fall-back in the well (1). However, as the gas velocity slows or
the liquid content in the well rises, there is insufficient energy in the well to carry all liquids to
surface and some begins to flow backwards (2). As the hydrostatic head downhole
increases, the liquid column that has accumulated in the well can re-enter the near-wellbore
region of the reservoir (4). This results in the well becoming “unloaded” so that it can flow
once more, with the gas carrying all liquids to surface (5). This cycle continues, providing the
typical intermittent response of liquid-loaded gas wells, until the reservoir potential starts to
fall or the liquid yield rises.

Fig. 1 – Schematic description of liquid loading mechanisms (Veeken et al., 2003)

The presence of the liquid phase in the well is due to:


 Condensation of water vapour along the well.
 Drop-out of condensate in the near-wellbore region of the reservoir (with condensate
saturation above the critical value) or in the wellbore, as pressure and temperature
decrease from bottomhole to surface.
 Ingress of formation water due to declining reservoir pressure.

Liquid loading can occur in low energy reservoirs and tight gas formations as well as in
prolific reservoirs with a high gas-liquid-ratio. Diagnosing liquid loading in the field is often
difficult because the affected well(s) may continue production without any substantial
performance impairment for a long period of time. Typical symptoms of liquid loading include
sharp drops in the cumulative production decline curve, the onset of liquid slugs at the well
surface, abrupt changes in the flowing pressure gradient, lower temperature at the wellhead
and declining water production or condensate-gas-ratio.

Many remedial lifting options have been developed for use in the field; some unloading
solutions (e.g. velocity strings) rely on the existing natural energy of the system, while others
(e.g. downhole pumps) provide extra energy to bring the water to surface, so reducing the
liquid loading problem. As each of the remedial options have their own technical
characteristics, their applicability varies depending on the characteristics and the status of
the well. Lea et al. (2003) presented an extensive technical review of well-established
techniques to alleviate the effects of liquid loading. According to Lea, “the method that is

130 DGMK-Tagungsbericht 2013-1


DGMK/ÖGEW-Frühjahrstagung 2013, Fachbereich Aufsuchung und Gewinnung, Celle

most economic for the longest period of operation is the optimum method. The criteria for
selecting the optimum method are: methods in similar fields that are used successfully,
vendor equipment availability, reliability of equipment, manpower required to operate the
equipment, and lifting capacity.” Park et al. (2009) developed a decision matrix to help
screen the possible remedial options available to the operator. The matrix aims to provide a
critical evaluation of solutions to liquid loading in gas wells vis-à-vis the existing technical and
economic constraints, and also serves as a quick screening tool for the selection of
production optimisation strategies.

Although a number of established techniques are used to alleviate the effects of liquid
loading the oil and gas industry still lacks reliable predictive models to help select the best
remedial option for a particular loading occurrence.

In what follows, a critical review of current methods to forecast the onset of liquid loading and
model the subsequent wellbore performance is presented, in line with the work by Solomon
et al. (2008) and Zhang et al. (2010). The review also includes more recent attempts to
understand the dynamic interactions between reservoir and wellbore during liquid loading.

Mechanisms of flow reversal and its prediction


Understanding the multiphase flows associated with liquid loading continues to elude the oil
and gas industry. Although major efforts have been made to predict the flowing conditions at
which the well remains out of the liquid loading region using the so-called “Turner’s criteria”,
these do not capture the dynamics of the loading sequence. Turner’s criteria are used by
operators to design a production system in such a way that it will flow at gas rates capable of
lifting all liquids out of the well, but they cannot be used to understand how serious the
loading occurrence is or how quickly it will impair production.

In order to model multiphase flows associated with liquid loading in gas wells, it is necessary
to identify the phases that co-exist in the well and characterise the main features of this
phenomenon from a flow mechanics point of view. Regarding the co-existing phases, liquid
loading in gas wells is usually taken as a two-phase (gas-liquid) or a multi-phase (gas and
multiple liquids) flow problem. Within the liquid phases, it is possible to differentiate between
free formation water, condensed vapour and gas condensates, as mentioned earlier.
Depending on the constituents of the liquid phase(s), its physical properties and interactions
with the gas phase will be different.

From a flow mechanics point of view, the characterisation of liquid loading requires an
understanding of flow patterns and flow patterns transitions. The shape and behaviour of the
interfaces between phases in a multiphase mixture dictate what is referred to as “flow
regime” or “flow pattern”. There are competing forces or mechanisms occurring within the
multiphase fluid at the same time. The balance between them determines the flow pattern.

For gas-liquid flows, the factors governing the interfacial distribution are complex. They
include surface tension, wetting, dispersion, coalescence, body forces and heat flux effects.
Nevertheless, it has been possible to classify the type of interfacial distribution in certain
broad categories (flow regimes), although the detailed nature of the flow will still depend on
the relative significance of the influencing factors. Thus, though the classification of flow
regimes is a very useful starting point, it does not in itself allow a complete specification of
the system. It should also be stressed that the relative importance of the influencing factors
changes gradually with phase flow rates, and that the transition from one regime to another
is not usually sharply defined. It is for this reason that the delineation of flow regimes is often
somewhat subjective.

DGMK-Tagungsbericht 2013-1 131


DGMK/ÖGEW-Frühjahrstagung 2013, Fachbereich Aufsuchung und Gewinnung, Celle

To characterise liquid loading in gas wells, it is necessary to predict the transition from an
acceptable flow regime (annular flow) to an unacceptable one (churn flow), which triggers
liquid loading in the well. A schematic description of the most likely flow patterns encountered
in a vertical well suffering from liquid loading is presented in Fig. 2. In upward annular flow,
the liquid phase flows as a thin film on the tube walls and the gas phase flows in the core.
The film interface is covered by ripples (small wavelength, small amplitude) and by
disturbance waves (large wavelength, large amplitude). Disturbance waves travel at
velocities greater than that of the liquid film and have heights many times greater than the
average thickness of the film. Disturbance waves are the main source of droplet entrainment,
a process by which small liquid drops are formed from the tips of disturbance waves and are
drawn into the gas flow. After travelling for a while in the gas core, the droplets eventually re-
deposit back on to the liquid film at a distance downstream. The mass exchange process
between the continuous and dispersed liquid streams is commonly referred to as the
entrainment and deposition mechanism (Hewitt and Hall-Taylor, 1970).

Slug Churn Annular

Fig. 2 – Schematic representation of gas-liquid flow patterns in a vertical pipe. Arrows


indicate the direction of the gas and liquid phases. In slug flow, a falling liquid film is
established around a rising Taylor bubble and the motion of the liquid slug is in the upward
direction. In churn flow, the motion of the liquid substrate between waves is in the downward
direction (Watson and Hewitt, 1999).

Churn flow is unarguably the most complex and least understood regime of gas-liquid flow in
vertical and inclined pipes. As seen in the literature, disagreement between authors concerns
not only to issues regarding the physics of the flow or the mechanisms of flow pattern
transition (Nicklin and Davidson, 1962; Mishima and Ishii, 1984; Dukler and Taitel, 1986), but
also questions about the existence of churn flow as a separate regime (Mao and Dukler,
1993; Hewitt and Jayanti, 1993; Jayanti et al., 1993; Barbosa et al., 2001). In churn flow, the
large Taylor bubbles characteristic of slug flow have disappeared, but characteristic reversals
in flow velocity are observed. Essentially, churn flow has similarities to annular flow in that
there is a continuous gas core in the center of the channel and a liquid layer near the
channel wall. However, it is important to make a distinction between the flow patterns since
their characteristics are substantially different. The main feature of churn flow is that the
liquid film undergoes periodic reversals. Waves are formed that transport the liquids
upwards. The origin of the large waves is similar to those which cause the flooding
phenomenon in counter-current gas-liquid flow (Govan et al., 1991). Between the waves,
there is a film which flows downwards and replenishes the next upstream wave from which
liquid is draining. The effect is one of net upwards liquid transport. Another feature of churn
flow is that the breakdown of slug flow is accompanied by a very rapid increase in pressure
gradient due to intensified wave activity and friction. Eventually, after a transition during

132 DGMK-Tagungsbericht 2013-1


DGMK/ÖGEW-Frühjahrstagung 2013, Fachbereich Aufsuchung und Gewinnung, Celle

which the pressure gradient initially decreases and then passes through a minimum value,
the pressure gradient begins to increase again as annular flow is entered (Owen, 1986;
Barbosa, 2001; Barbosa et al., 2002; Belt et al., 2010). A similar behaviour can be attributed
to the fraction of the liquid flow entrained as droplets in the gas core, which decreases to a
minimum value with increasing gas velocity in what is clearly churn flow, prior to increasing
again in annular flow.

Liquid loading in gas wells corresponds to transient flow conditions where intermittent and
counter-current flows exist. Assuming an initial steady-state configuration characterised by
annular flow in the well, the first flow regime transition is triggered by a reduction in gas flow
rate (or by an increase in liquid hold-up). This can be regarded as the onset of liquid loading
and causes the initial flow regime to develop into churn flow, and the churn flow to develop
into slug flow. There is a clear understanding that the annular flow pattern must prevail along
the entire wellbore to guarantee that there is no liquid loading. Otherwise, the falling liquid
accumulates at the bottom of the well, thus increasing the back-pressure on the reservoir
and reducing its inflow.

In the prediction of liquid loading, the most important flow pattern transitions to be accounted
for are those associated with the slug-to-churn and churn-to-annular flow patterns. It must be
understood that the judgment of the observed regime still remains very subjective, even with
the advent of modern instrumentation, which has made it possible to take high-frequency
measurements of key flow parameters (e.g. phase fraction, pressure, temperature, velocity)
for flow-pattern recognition.

There are three main theories that provide explanations for the slug-to-churn transition in
vertical pipes. Dukler and Taitel (1986) treated churn flow as an entrance phenomenon and
established a relationship, based on visual observations of the flow pattern, for the length of
the entrance region required to form stable liquid slugs. Brauner and Barnea (1986)
recognised the collapse of the liquid slug as related to excessive aeration of a slug unit, i.e.,
as the gas velocity increases, the void fraction in the liquid slug reaches a critical value and
this leads to the collapse of the liquid slugs due to bubble coalescence. Chen and Brill (1997)
also associated the transition to churn flow to the collapse of short and highly aerated liquid
slugs that appear as a result of liquid instabilities in the wake of Taylor bubbles. The third
theory (Nicklin and Davidson, 1962; McQuillan and Whalley, 1985; Jayanti and Hewitt, 1992)
attributed the slug-to-churn transition to flooding of the liquid film surrounding the Taylor
bubble due to the similarity of pressure drop and holdup characteristics between the
phenomena of churn flow following flooding and ‘ordinary’ churn flow (Govan et al., 1991)
and also due to the fact that the gas velocity required to initiate flooding in a falling film is
approximately the same as that at the slug-to-churn transition.

Jayanti and Brauner (1994) evaluated the performance of some of the above methods using
the experimental data of Owen (1986) for vertical, air-water, adiabatic two-phase flows in a
20 m long, 31.8 mm ID. They found that the model by Taitel and Dukler (1986) implied that
‘churn’ flow would disappear after lengths between 200 and 360 tube diameters. However,
Owen (1986) observed churn flow at a distance of 625 diameters downstream of the point of
liquid injection. The predictions of the Brauner and Barnea (1986) model were in line with
Owen’s experimental data at high liquid flow rates, but not at low liquid flow rates. Jayanti
and Brauner (1994) attributed this fact to the limitation of the transition model in predicting
gas holdup in short liquid slugs, but did not rule out the idea of excessive slug aeration as a
possible transition mechanism. The model of Jayanti and Hewitt (1992) presented a much
better prediction of the transition. More recent experimental data (Watson and Hewitt, 1999;
Das et al., 1999; Guedes de Carvalho et al., 2000; Guedes de Carvalho, 2006) in circular
tubes and annuli strengthened the evidence that flooding in the Taylor bubble is the main
mechanism behind the slug to churn transition.

DGMK-Tagungsbericht 2013-1 133


DGMK/ÖGEW-Frühjahrstagung 2013, Fachbereich Aufsuchung und Gewinnung, Celle

At the churn-to-annular flow boundary, as the superficial gas velocity increases, the
interfacial shear stress becomes significant and the flow reversals characteristic of churn flow
cease to exist. The liquid film gets thinner, the flow becomes unidirectional and the annular
flow regime is established. Disturbance waves start to appear and the pressure gradient
eventually begins to increase with increasing gas superficial velocity. A number of criteria for
the transition to annular flow may arise from these processes. These are as follows: (i) at the
flow reversal point, i.e., the point where the gas flow rate is sufficient to carry all the liquid
upwards; (ii) at the pressure drop minimum that occurs due to the decreasing trends of
pressure gradient with increasing gas flow rate observed in fully developed churn flow; (iii) at
the condition of zero mean wall shear stress, which is negative in churn flow and positive in
annular flow; and (iv) at the point of continuously positive local wall shear stress, which must
take place when the liquid film is entirely unidirectional.

Of all the above criteria, the flow reversal condition is the most widely used due to its
relatively easy experimental observation (Hewitt et al., 1985; Belt, 2007) and because there
are simple relationships to correlate it. The wall shear stress conditions are difficult to
evaluate as they require detailed measurements close to the wall. The minimum pressure
gradient criterion is somewhat dubious because for some conditions the pressure drop
minimum occurs in fully developed annular flow (Owen, 1986; Zabaras et al., 1986).The
pressure gradient also includes gravitational and inertia components, and may not wholly
reflect the frictional or interfacial behaviour.

According to Hewitt and Wallis (1963) and Wallis (1969), the flow reversal transition is
relatively insensitive to the liquid superficial velocity, and can occur at an approximately
constant gas superficial velocity. For tubes smaller than 5 cm in diameter, the flow reversal
transition is correlated by the gas phase Froude number, FrG, as follows:
0.5
 U G2  G 
FrG    1
 gD L   G   (1)
For larger diameter tubes, droplet entrainment from the wave tips may occur before the
waves themselves are transported (Hewitt, 2010). In this case, an alternative correlation for
flooding velocity that does not take into account the tube diameter has been suggested. This
has been attributed to Pushkina and Sorokin (1969) and is written in terms of the Kutateladze
number, KuG, as follows:
0.25
 U G4  G2 
KuG     3.2
 g  L   G   (2)
Equation (2) is similar to the well-known Turner et al. (1969) criterion for calculating the
critical velocity for liquid loading in vertical and near vertical wells; the corresponding value of
KuG in Turner et al.’s model is 3.67. Models based on this approach have been tested
against field data, with varying degrees of success reported by each author. In this model,
the minimum velocity required to keep liquid droplets in suspension is the result of a force
balance between the drag exerted by the surrounding gas (see Fig. 3). Thus, for a given set
of wellbore conditions, when the drag force is greater than the buoyancy force, the gas
velocity is higher than the critical velocity necessary to lift the liquid droplets and the flow in
the well is unidirectional upwards. On the contrary, if the drag force is smaller, then the flow
develops into churn or slug flow, leading to liquid falling to the bottom of the well – the onset
of liquid loading.

134 DGMK-Tagungsbericht 2013-1


DGMK/ÖGEW-Frühjahrstagung 2013, Fachbereich Aufsuchung und Gewinnung, Celle

Fig. 3 – Liquid droplet transported in a vertical gas stream (Lea et al., 2003).

In order to achieve a critical value of 3.67 for the Kutateladze number, one has to follow the
Turner et al. (1969) assumption that, for typical oil field conditions, the droplet Reynolds
number ranges from 104 to 2 x105 and the spherical droplet drag coefficient of spherical
droplets in this range corresponds to approximately 0.44. Turner et al. estimated the droplet
diameter via the critical value of 30 for the Weber number given by (Hinze, 1955),

U G2  G d max
Wecrit 
 (3)
Constant values of interfacial tension obtained from standard databases were suggested.
Turner et al. (1969) stated that a correction of approximately 20% to their model was needed
to fit field data (Sutton et al., 2010). This correction corresponded to increasing the critical
Weber number from 30 to 60. Without the correction factor, the Kutateladze number
corresponds to 3.09 (Van’t Westende, 2008).

Coleman et al. (1991) used the approach by Turner et al. (1969) to determine the critical
velocity for gas wells with wellhead pressures of 500 psia (~34 bara) and below. They found
that the 20% adjustment was not required in their case. They also noted that gas gravity,
interfacial tension and temperature have little impact on the accuracy of the critical gas
velocity calculations, with wellbore diameter and wellhead pressure playing a more
significant role. It was observed that wells with slug flow behaviour did not obey the entrained
droplet model.

Nosseir et al. (1997) examined the assumption of turbulent flow made by Turner et al. (1969)
by calculating the Reynolds numbers corresponding to the field data gathered by Turner et
al. (1969) and Coleman et al. (1991). They discovered that almost all the field data from the
first reference work exceeded the assumed range (104 to 2 x105) for Reynolds numbers used
to derive Eq. (2). In particular, it was found that the range was 2 x105 to 106, corresponding
to a drag coefficient of 0.2 and therefore requiring the 20% adjustment to match field data.
The calculated Reynolds numbers from the field data provided by Coleman et al. (1991) was
in the originally assumed range (104 to 2 x105) and therefore requiring little or no adjustment
to match field data. Nosseir et al. (1997) then developed two entrained droplet models: one
for low flow rate systems, and one for high flow rate systems.

The original Turner et al. (1969) equation for critical superficial gas velocity can be written as,

k 0.25  L   G 0.25
UG 
G 0.5 (4)
where k is equal to 1.8949 in field units (with σ in dynes/cm, UG in ft/s and ρ in lbm/ft3)
(Sutton et al., 2010) and equal to 6.494 in SI units (with σ in N/m, UG in m/s and ρ in kg/m3).
Coleman et al. (2001) recommended an updated k value of 1.5934 and 5.464 (in field and SI
units, respectively).

The Turner et al. (1969) criterion was questioned by Guo et al. (2006), who claimed that,
even with 20% adjustments, it still underestimated the minimum gas velocity for liquid

DGMK-Tagungsbericht 2013-1 135


DGMK/ÖGEW-Frühjahrstagung 2013, Fachbereich Aufsuchung und Gewinnung, Celle

carryover using Turner et al.’s original database. A kinetic energy-based model that was
claimed to be more accurate than Turner et al.’s for real field data was proposed. The model
was based on evidence that conditions at the bottomhole are the controlling ones for the
transport of liquid droplets up the well, as opposed to the wellhead conditions adopted in
Turner et al. (1969) criterion.

Belfroid et al. (2008) advanced a criterion for predicting the onset of liquid loading in deviated
wells. In inclined tubes, the liquid film is no longer symmetric, which results in different
characteristics for the waves present in the liquid film and in local rates of droplet
entrainment and deposition (Badie et al., 2000). These mechanisms exert some influence on
the critical velocity that reaches a maximum at intermediate inclinations (30o-50o) from the
horizontal. Belfroid et al. (2008) combined the Turner et al. (1969) criterion with a shape
function of the type suggested by Fiedler (2004) to obtain the following relationship for angles
between 10 o and 90o:

U G  3.1
g  L   G 0.25 sin1.7 0.38
 G 0.5 0.74
(5)
A recent adaptation of the Turner model was proposed by Wang et al. (2010) who
considered the effects of gas compressibility, shape of the droplets and friction factor in the
correlation.

Sutton et al. (2010) revisited the analysis of Turner et al. (1969) and confirmed that a number
of situations exist where the use of the wellhead as the evaluation point can lead to
erroneous conclusions, and provided recommendations as to when the use of downhole or
wellhead conditions is more appropriate in determining the critical gas velocity for a well. In
particular, they evaluated the influence of parameters such as fluid density, surface tension,
evaluation location (downhole or wellhead), temperature gradient, tubing diameter and well
depth. They concluded that, for wellhead pressures greater than 1000 psia (~68 bar), the
wellhead condition should be used to evaluate the critical velocity. For low-pressure cases
(less than 100 psia), the critical velocity should be evaluated using the downhole conditons.
For wells producing free water, the recommendation is that the downhole conditions should
prevail in the calculation of the critical velocity.

Recently, there has been experimental evidence that the point of flow reversal, which is
characteristic of liquid loading, is associated with the oscillatory motion of the liquid film
rather than the droplet flow reversal (Van’t Westende, 2008; Veeken and Belfroid, 2010).
Considering that, in some occurrences, the observation of film flow reversal as the
dominating mechanism of liquid loading does not directly impact on the conventional tools
that are currently employed to predict the phenomenon (Veeken and Belfroid, 2010), it is
expected that the Turner method will continue to serve engineers in the field, perhaps with
gradual modifications and adaptations to incorporate additional physical mechanisms to
make it more widely applicable. On the other hand, one should not forget that the accuracy of
critical gas velocity predictions depends on the accuracy of the predicted flowing gradients.
Thus, if the selected gradient correlation predicted conditions indicative of a given flow
regime, but the calculated critical velocity suggested a different flow regime, the
interpretation of the specific liquid loading occurrence would be erroneous and/or misleading.
Also, the effect of pressure values greater than those for which the entrained droplet model
was originally developed is yet to be assessed against experimental data. Finally, it must be
stressed that the methods to estimate the critical gas velocity and the onset of liquid loading
do not attempt to model the flow in the wellbore during liquid loading. They simply provide
criteria to assess for what value of gas velocity liquid fall-back may begin to occur in the well.
In what follows, a review of methods available to model two-phase flows typical of liquid
loading is presented.

136 DGMK-Tagungsbericht 2013-1


DGMK/ÖGEW-Frühjahrstagung 2013, Fachbereich Aufsuchung und Gewinnung, Celle

Models to characterise the transient multiphase flows in wells typical of liquid


loading
The modelling approaches for the characterisation of liquid loading in gas wells can be
classified as: empirical models, phenomenological models, the two-fluid model and the drift-
flux model. Special attention is paid to ability of these methods to capture the transient nature
of the loading process.

With empirical models, data for frictional pressure gradient and void fraction are related to
system variables through empirical equations. For the development of a reliable empirical
model, a large number of experiments are needed to reproduce a specific problem, but this
may be expensive and, unless a dimensional analysis is carried out, the resulting model will
only apply to a limited set of conditions. In general, empirical models lack the inclusion of
fundamental physical mechanisms, but their main advantage is that they are relatively simple
and fast to run.

An empirical model used for gas-liquid flows at high gas fraction is that of Gray (1974). It is a
vertical flow correlation for gas wells, accounting for entrained fluids, temperature gradient,
fluid acceleration, and non-hydrocarbon gas components to calculate the two-phase
pressure drop in the well. Kabir and Hasan (2006) pointed out that Gray’s correlation is a
pseudo-homogeneous flow approach for computing pressure losses due to friction, gravity
and acceleration. Gray’s model is not fully homogeneous as it includes some liquid hold-up
by considering that liquid droplets may not travel as fast as the gas phase. No flow pattern
differentiation is made with Gray’s model, which is also only applicable to steady-state or
pseudo-steady state flows. A model of this type is therefore not suitable for characterising the
transient phenomena that are typical of liquid loading in gas wells.

Phenomenological models are flow-pattern based models which rely on the identification of a
specific flow regime and the application of ad hoc pressure drop calculations. Closure
relationships are required (e.g. to define the void fraction), together with the definition of the
criteria for the transitions between flow regimes. Flow pattern classifications are originally
based on visual observations of two-phase flow. Experimental observations can be mapped
on two-dimensional plots (called “flow pattern maps”) and the boundaries between regimes
determined. In the past, different investigators used different coordinates for the maps (e.g.
mass flow rates, momentum fluxes or superficial velocities), in search for parameters
independent of the given experimental set-up. However, the judgment of the observed
regime was, and still remains, very subjective, even with the advent of modern
instrumentation, which has made it possible to take high-frequency measurements of key
flow parameters (e.g. phase fraction, pressure, temperature, pressure drop) to develop flow-
pattern recognition techniques. Examples of phenomenological models are those based on
the unit cell concept for slug flow in vertical pipes, which consider the actual flow
configuration (long Taylor bubbles separated by aerated liquid slugs) in the formulation of ad
hoc models for calculating the void fraction, wall shear stress (Fernandes et al., 1983; Orell
and Rembrand, 1986; Sylvester, 1987; De Cachard and Delhaye, 1996) and heat transfer
coefficient (Barnea and Yacoub, 1983; Barbosa and Hewitt, 2005) in each particular region of
the unit cell.

Criteria for flow regime transitions can be incorporated into computer codes for wellbore
simulation in order to determine the positions at which a certain set of closure relationships
should be used in conjunction with conservation equations that are integrated numerically for
predicting the local phase holdup, the phase temperatures, compositions and the local
pressure. To date, two approaches, the two-fluid model and drift-flux model, have been
widely used in the development of the fundamental equations for steady-state and also
transient flow calculations typical of liquid loading.

DGMK-Tagungsbericht 2013-1 137


DGMK/ÖGEW-Frühjahrstagung 2013, Fachbereich Aufsuchung und Gewinnung, Celle

The two-fluid model treats the fluids and/or phases (gas and liquid) separately as each flow
on its own within the wellbore. Separate continuity, momentum and energy equations are
developed for the multiple fields (e.g. gas, liquid film and liquid droplets). Appropriate closure
laws (usually based on experimental data) are required, which may be subject to the type of
averaging procedure utilized to derive the conservation equations (Drew and Passman,
1999). Transport equations for turbulence quantities may also need to be solved, depending
on the level of refinement of the model (Kataoka and Serizawa, 1989). Comprehensive
discussions and derivations of conservation equations for use in the two-fluid modelling
approach in the context of two-phase flows occurring in the oil industry are given by
Bendiksen et al. (1991) and Shoham (2006).

The drift-flux modelling approach was principally developed by Zuber and Findlay (1965),
Wallis (1969) and Ishii (1977). It treats the two phases as a mixture, though slippage
between gas and liquid is accounted for. It is generally accepted that the drift flux model is
appropriate for the mixture where the dynamics of two components are closely coupled (Ishii
and Hibiki, 2006). One advantage of drift-flux models in the solution of wellbore flows is that
they are generally differentiable and free of discontinuities. Shi et al. (2003, 2004) developed
a drift-flux model that can be used to simulate three-phase flows in wellbores at steady-state.
Closure relationships for the drift velocity have been proposed based on experimental data
obtained for low-pressure three-phase flows in a 15-cm diameter tube at various inclinations.

In the context of their application in the simulation of liquid loading, a critical review of the
two-fluid and the drift-flux approaches reveals the following limitations:

 Flow patterns such as slug flow, which are intermittent in nature, do not fit
satisfactorily into the two-fluid model. In vertical upward slug flow, liquid flows
upwards in the slugs, but downwards in the Taylor bubble region, which makes it
difficult to average the liquid properties.
 Churn flow is also not fully captured by the two-fluid approach. As seen above, this
flow pattern can be regarded as a special type of annular flow, in which the liquid film
undergoes periodic reversals. Waves are formed that transport the liquids upwards.
Between the waves, there is a film which flows downwards and replenishes the next
upstream wave from which liquid is draining. The effect is one of net upwards liquid
transport. This flow intermittency does not fit well into the two-fluid model.
 In most separated flows, the relative motion between phases is directly related to the
pressure and velocity gradients in the two phases. This cannot be captured by a drift
flux model.
 Both the two-fluid and the drift-flux models require closure relationships to
characterise the gas-liquid interface and the wall friction.

The number of studies dealing specifically with modelling of churn flow is rather limited. So
far, no direct account has been taken of the time-dependent nature of churn flow and its
influence on the behaviour of system parameters. The models of Jayanti and Brauner (1994)
and Sawai et al. (2004) have been proposed for calculating the pressure drop and the gas
holdup. Jayanti and Brauner’s model is based on force balances on the gas core and on a
control volume encompassing the entire tube cross section. The average wall shear stress
was computed assuming that only the liquid comes in contact with the wall and that its time-
dependent character can be disregarded. The friction factor associated with the film flow was
estimated from relationships for single phase flows. An interpolation of the correlations of
Bharathan and Wallis (1983) for churn flow and Wallis (1969) for annular flow was
recommended for the interfacial friction factor, based on comparisons with the experimental
results of Govan et al. (1991). Alternative relationships for the interfacial friction factor in
churn flow were proposed by Abe et al. (1991) and Kelly and Freitas (1993). More recently,

138 DGMK-Tagungsbericht 2013-1


DGMK/ÖGEW-Frühjahrstagung 2013, Fachbereich Aufsuchung und Gewinnung, Celle

Sawai et al. (2004) conducted a study on gas-liquid churn flow in order to characterise the
interfacial structure of the liquid film (wave amplitude and speed) and evaluate quantitatively
the behaviour of the pressure gradient as a function of the superficial velocity of the gas.
Correlations were proposed for the pressure gradient and void fraction using a Martinelli two-
phase multiplier and a drift-flux relationship, respectively.

In annular flow, it is common to write mass and momentum balances in an elementary


control volume for the three fields, i.e. liquid film, liquid droplets and gas core (Azzopardi,
2006). Solving transient annular flow equations is not a simple task, which becomes even
more difficult when heat and mass transfer processes come into play. In addition, the
knowledge of boundary conditions for the liquid phase mass conservation (i.e., the amount of
liquid entrained as droplets at the onset of annular flow) is not always available. As
demonstrated by Barbosa and Hewitt (2001a, 2001b), this is a parameter of crucial
importance in some cases, such as in the flow of multicomponent mixtures and in heat
transfer applications (critical heat flux and post-dryout heat transfer).

The annular flow governing equations (or simplified versions of them, for example, for
steady-state) can be seen as the starting point for many annular flow models proposed over
the last 30 or 40 years. Several attempts to simplify these equations by introducing new
hypotheses and closure relationships were proposed by many researchers. Amongst many
of these models, one can cite those by Moeck and Stachiewicz (1972), Dobran (1983),
Schadel et al. (1990), Govan (1990), Nigmatulin et al. (1996), Fu and Klausner (1997) and
Alipchenkov et al. (2004).
A common hypothesis in annular flow modelling is that the droplets and the gas core behave
as a homogeneous mixture (i.e., no slip between the phases). With the homogeneous core
density hypothesis, the interfacial shear stress can be formulated based on an analogy with
single phase flow in a rough pipe.

In search of an expression for the interfacial friction factor that is capable of giving accurate
predictions in annular flow, many authors have proposed empirical correlations between the
interfacial roughness and liquid film thickness (Gill et al., 1963). Examples of such
correlations are those by Wallis (1969) and Hewitt and Whalley (1978). A more refined
relationship was subsequently presented by Owen (1986), who found out that the turbulent
mixing-length in the gas core is reduced in the presence of droplets. More recently, a
correlation for the interfacial friction factor in annular flow have been proposed by Belt et al
(2009) and a model for calculating the average liquid film thickness and the pressure
gradient in annular flow was advanced by Cioncolini et al. (2009).

Another simplifying hypothesis is related to the velocity distribution in the liquid film. Hewitt
and Hall-Taylor (1970) showed that theories for single-phase turbulent flow provide
satisfactory estimates of parameters such as the liquid film flow rate and the film thickness.

Kishore and Jayanti (2004) applied the triangular relationship between the film thickness, the
film flow rate and the pressure gradient (Hewitt and Hall-Taylor, 1970) in a method for
computing the fluid flow in the gas core using a CFD approach that assumes a rough gas-
liquid interface and a velocity distribution in the liquid film given by the universal log-law. The
k-ε turbulence model was used to describe turbulent flow effects in the gas phase. They
included the rates of droplet entrainment and deposition as a source term and as a boundary
condition in a transport equation for the entrained fraction, which was resolved through a
differential model.

A one-dimensional transient annular flow model has been developed recently with the
specific purpose of simulating annular and sub-annular flows in wellbores (Gessner and
Barbosa, 2010; Waltrich et al., 2011). The Hyperbolic Transient Annular Flow (HyTAF) model

DGMK-Tagungsbericht 2013-1 139


DGMK/ÖGEW-Frühjahrstagung 2013, Fachbereich Aufsuchung und Gewinnung, Celle

consists of mass and energy balance equations for the three fields (gas, liquid film and
droplets) and momentum balance equations for the liquid film and for the homogeneous gas
(entrained droplets plus gas) core. As the model is intended also for simulation of fast
transients, a finite difference based solution algorithm derived from the Split Coefficient
Matrix Method (Chakravarthy et al., 1980) was developed in order to deal with discontinuities
in the solution domain, such as shocks and void waves. Waltrich et al. (2011) generated
steady-state and transient two-phase flow experimental data using the TowerLAB facility at
Texas A&M University, which has a transparent vertical test section of 42 m in length, and
0.048 m ID. Air and water were used as working fluids. Comparisons between a benchmark
code for multiphase flow in pipes, experimental data and HyTAF were conducted,
highlighting the need for further investigations in this area. Alves et al. (2012) presented a
comparison between the experimental pressures gradients of Waltrich et al. (2011) and
those predicted by HyTAF for an induced transient case, from steady-state annular flow to
steady-state churn flow, with satisfactory agreement (Fig. 4).

Fig. 4 – Prediction of local pressures during a flow-rate and pressure induced transient from
steady-state annular flow to steady-state churn flow (Alves et al., 2012),

Models to characterise the dynamic interactions between reservoir and


wellbore during liquid loading.
At present, the models used to predict and diagnose liquid loading problems tend to be
based on steady-state analysis, and so cannot handle the transient phenomena typical of
liquid loading, such as the gradual buildup of static fluid column in the well and the
associated increase in back-pressure on the formation. Efforts are being made across the
industry and within academia to link the observed well dynamics with the intermittent
response of a reservoir that typifies liquid loading in gas wells. Even when transient
multiphase wellbore models are used, the problem is still incorrectly defined as a steady-
state Inflow Performance Relationship (IPR) is used to characterise the reservoir, which
involves the wrong boundary conditions between the well and the reservoir.
A more reliable approach would be to use a transient multiphase flow wellbore model that
includes the transient features of the near-wellbore region of the reservoir as boundary
conditions. This dynamic, integrated modelling approach is essential to investigate the
loading-unloading sequence described in Fig. 1, including the possible re-injection of the
heavier phase into the reservoir, as shown in Fig. 5.

140 DGMK-Tagungsbericht 2013-1


DGMK/ÖGEW-Frühjahrstagung 2013, Fachbereich Aufsuchung und Gewinnung, Celle

Wellbore

Near-wellbore
region

Reservoir

Fig. 5 – An integrated system approach to investigate liquid loading (Costantini, 2005).

The conventional approach to characterising the dynamic interaction between reservoir and
wellbore is to combine a steady-state or pseudo steady-state reservoir performance model
with a steady-state or pseudo steady-state wellbore performance model and to determine the
point of stable operating conditions for the integrated system by solving the models together
(e.g. by iteration, until both models converge to the same flowing bottomhole conditions).
Fig. 6 shows a typical IPR, which characterises the inflow from the reservoir, together with a
typical Tubing Performance Relationship (TPR), characterising the outflow from the well.
Flowing bottomhole pressure

Inflow

Outflow

Operating rate

Gas rate

Fig. 6 – IPR-TPR intersection to obtain the operating flow rate under stable flowing
conditions. Solomon et al. (2008).

If the operating gas rate determined from the approach, shown in Fig. 6, corresponds to the
critical gas velocity for that specific system, liquid loading may begin to occur.
This diagnostic approach ignores both the transient flows in the well and those in the near-
wellbore region of the reservoir, which are typical of liquid loading in gas wells. Realistic
transient boundary conditions at the interface between reservoir and wellbore must be
defined in order to couple the two sub-systems while capturing the intermittent features of
liquid loading.
The conventional pressure profile in the near-wellbore region of a flowing reservoir (Fig. 7a)
is not suitable to characterise the transient phenomena that take place during liquid loading.
Due to the wellbore phase redistribution that occurs during liquid loading, the bottomhole
pressure changes with time. The frequency and amplitude of these changes vary with the
magnitude of the liquid loading occurrence. If the reservoir were capable of providing an
instantaneous response to the bottomhole pressure fluctuations, the pressure profile in the
near wellbore would quickly re-adjust to the new wellbore conditions (Fig. 7b). However, due
to a combination of inertia and compressibility effects, the reservoir response is not
instantaneous and can be particularly slow for tight formations. A sequence of conventional

DGMK-Tagungsbericht 2013-1 141


DGMK/ÖGEW-Frühjahrstagung 2013, Fachbereich Aufsuchung und Gewinnung, Celle

pressure profiles (from a to b in Fig. 7) could be assumed, but this would imply a temporary
discontinuity of the pressure function at the wellbore, which is unphysical. Thus, Zhang et al.
(2010) proposed a U-shaped pressure profile (Fig. 7c), which could also explain the
possibility of re-injection of the heavier phase into the reservoir.

Fig. 7 – As the well starts to load up with liquid, backpressure is imposed on the formation.
The fat arrow points to the minimum pressure existing in the near-wellbore region,
where the pressure profile is a U-shaped curve. Zhang et al. (2010).

Transient fluctuations of flowing bottomhole pressure can be caused by dominating wellbore


effects, such as flow reversal, operational adjustments to wellhead conditions (including shut-
ins and start-ups), and wellbore phase redistribution (including gas rise/expansion).
Ali et al. (2005) experimentally investigated the effects of wellbore phase redistribution and
phase re-injection on downhole pressure build-up data.

Al-Darmaki et al. (2008) performed further experimental investigations along the same
direction, coupled with data analyses using drift-flux modelling techniques, and verified the
occurrence of phase redistribution in the wellbore under transient flowing conditions, which
leads to downhole pressure fluctuations that could trigger transient flow in the near-wellbore.
Attempts have recently been made to characterise the dynamic interaction between reservoir
and wellbore under transient flow conditions.

Sturm et al. (2004) presented an investigation of unstable production from a vertical, gas-
lifted well tapping into segregated black oil and gas layers. For their study, they described the
two-phase flow in the tubing by means of a drift-flux model, and modeled single-phase gas
flow in the annulus. They characterised the gas and liquid reservoir inflow by applying the
radial Darcy’s equation to the liquid flow, and the radial Forcheimer equation to the gas flow,
respectively. The authors also investigated the reservoir inflow response to sinusoidal
variations of the drawdown and obtained an ellipsis on a drawdown vs. liquid production plot.

Dousi et al. (2005) defined metastable gas flow in a liquid loaded well as the flow that occurs
when a dynamic equilibrium is attained between liquid produced out of the well and liquid
falling downward and being re-injected into the reservoir. This results in a stable liquid
column at the bottom of the well. To mimic the gas metastable flow, they assumed a
reservoir made of two layers, with the top layer producing gas and the bottom one taking in
liquid from the wellbore. They used steady-state relationships to characterise both the
reservoir and the wellbore. Fig. 8 depicts this modelling approach to liquid loading in gas
wells.

142 DGMK-Tagungsbericht 2013-1


DGMK/ÖGEW-Frühjahrstagung 2013, Fachbereich Aufsuchung und Gewinnung, Celle

Fig. 8 – Liquid loading modelling with re-injection of heavier phase (Dousi et al., 2005).

Initially, (a) the well is producing at a constant rate sufficient to produce all the fluid streams
to the surface. As soon as the tubing head pressure increases or the reservoir pressure
decreases, (b) the rate drops below the critical Turner value, which triggers the reversal flow
of the produced fluids towards the bottom of the well. As the denser fluid builds up in the
wellbore, a higher hydrostatic pressure acts upon the formation, thus reducing the pressure
drawdown in the reservoir, which subsequently reduces its gas and liquid production.
According to Dousi et al. (2005), a stabilized condition, (c) is reached before the demise of
the production well, (d) when the liquid production rate becomes equal to the liquid injection
rate. The rates of both the produced gas and the re-injected liquid are assumed to take place
under steady-state flow conditions. In particular, a single-phase (gas) inflow performance
relationship is used to characterise the reservoir flow, and the co-produced water is assumed
to be a fixed percentage of the gas flow. The Cullender-Smith well flow model is used as the
tubing performance relationship. The Turner’s criterion for estimating the critical gas velocity
required to lift all fluids to surface is used to calculate the required gas flowrate to achieve
liquid unloading. By adjusting the water-gas-ratio, the flowing tubing pressure and the water
re-injection factor (a variable introduced by the authors), and re-arranging the reservoir inflow
equation to model liquid re-injection into the reservoir during meta-stable gas flowrates,
Dousi et al. were able to match the flow rates below the Turner criteria observed from field
data. The authors concluded that understanding the interaction between the reservoir and
the well is of crucial importance in the modelling liquid loading in gas wells.
Although this analytical modelling approach is yet to be proven by fluid mechanics, it
represents one of the very few attempts to model and quantify the re-injection of the heavier
phase in the reservoir, which is thought to occur during liquid loading.

Nennie et al. (2007) simulated the flow from a horizontal well with three inflow sections
located in a thin oil rim, each individually regulated by an inflow control valve (ICV’s). They
used a commercial transient wellbore simulator to mimic the well’s response, and a full-field
numerical reservoir simulator to model the reservoir, including the near wellbore. Nennie et
al. integrated the two simulators by explicit coupling, assuming no capillary effects in the
reservoir. Among their case studies, they included a sinusoidal ICV setting variation and
obtained an ellipsis on a bottomhole pressure vs. mass rate plot.

Sagen et al. (2007) proposed an implicit coupling approach based on a finite difference
model for the near-wellbore region of the reservoir, which simulates fluid transport through
the porous medium as a plug-in into a transient wellbore model (originally developed for
pipeline flow simulations (Bendiksen et al., 1991). Outputs from the reservoir model are
pressures and saturations in time and space, while the inputs are permeability and porosity
for the porous medium and fluid properties of the flowing phases. The wellbore model
provides the pressure boundary for the reservoir model, while the reservoir model provides
the flow into the pipeline together with fluid temperature.

DGMK-Tagungsbericht 2013-1 143


DGMK/ÖGEW-Frühjahrstagung 2013, Fachbereich Aufsuchung und Gewinnung, Celle

Chupin et al. (2007) investigated liquid loading in gas wells by implicit coupling of a
commercial transient wellbore simulator and a near wellbore model based on the mass
conservation equations and Darcy’s law written for the liquid and gas phases. However, they
did not provide details of the model and solution procedure. For the coupling, they suggested
using “sensitivity coefficients” to be determined based on the size of the near wellbore
region, estimated a priori as a user input.

Zhang et al. (2009) identified the U-shaped pressure profile for a three-phase (oil, gas, and
water) synthetic case. This was achieved by imposing a transient oscillating bottomhole
pressure (BHP) as the inner-boundary. A sensitivity analysis was performed to evaluate the
length of the “disturbed” distance from the wellbore under multiphase flow condition. It was
observed that BHP high frequency oscillation, great fluid compressibility, and low reservoir
permeability led to a shorter length of the U-shaped pressure profile. The fluctuating
amplitude showed no significant effect on the penetration distance of the U-shape.

Zhang et al. (2010) further investigated the backflow rates associated with a possible pattern
of BHP oscillations for a low permeability gas reservoir. A synthetic, 1D blackoil gas reservoir
model was built using a commercial software package. After simulating steady-state
production, the bottomhole pressure (BHP) was raised, with the inner boundary condition
being changed from constant production rate to target BHP. With the commercial software
package, independently of the timestep chosen for the simulations, it was not possible to
generate a U-shaped pressure profile because the commercial simulator defines the
perturbed BHP as a target, not as a fixed constraint, and so introduces a gradual pressure
build-up process. Also, when defining a wellbore, the user has to specify a priori if it is going
to be a producer or an injector, and the well will not be allowed to automatically switch mode
unless the user states so in a subsequent production schedule. This showed that backflow
rates cannot be detected with a conventional numerical simulator. Zhang et al. went on to
proposing a novel approach, based on conventional numerical reservoir modelling
techniques, but using them in a way that accommodates transient pressure and rate
boundary conditions at the wellbore and also along the distance from the well. They then
focused on a two-phase (gas and water) scenario, considered to be more representative of
natural gas occurrences. Their results showed a U-shaped pressure distribution in the near-
wellbore region for both gas and water, and backflow rates. On the other hand, the
commercial simulator did not seem to capture the same transient scenarios.

Conclusions
A critical review of liquid loading in gas wells was presented, which highlighted the
differences between the traditional Turner’s criteria to model the onset of this common
phenomenon and the experimental evidence of the transition from annular to churn flow, with
suggests that liquid film flow reversal could be the root cause of liquid loading.
The review included modelling approaches to predict the onset of liquid loading and model
subsequent flows in a production system, including the more recent attempts to model the
dynamic interactions between reservoir and wellbore.
The conventional pressure profile in the near-wellbore region of a flowing reservoir is not
suitable to characterise the transient phenomena that take place during liquid loading. Due to
a combination of inertia and compressibility effects, the reservoir response to wellbore phase
redistribution effects is not instantaneous.
Despite the wide range of two-phase flow modelling techniques available, it still remains
crucial to capture the transition from annular flow into churn flow, from the latter into slug flow
and, ultimately, from slug flow to bubble flow, which may lead to the end of the well’s life.
Coupling the transient reservoir response to the transient wellbore response is a complex
problem, especially considering the intrinsic difficulty in characterising the dynamic boundary
conditions between the two domains and the need for further experimental validation.
The industry needs reliable predictive models that can help select the best remedial option
for a given occurrence of liquid loading and more research is needed in this direction.

144 DGMK-Tagungsbericht 2013-1


DGMK/ÖGEW-Frühjahrstagung 2013, Fachbereich Aufsuchung und Gewinnung, Celle

Nomenclature
Roman

d = Droplet diameter, m (in.)


D = Wellbore diameter, m (in.)
Fr = Froude number
g = Acceleration due to gravity, m/s2 (ft/s2)
k = Constant in Turner’s equation, m0.25/s0.5 (ft0.25/d0.5)
Ku = Kutateladze number
U = Superficial velocity, m/s (ft/s)
We = Weber number

Greek

φ = Inclination angle from horizontal, degrees


ρ = Density, kg/m3 (lbm/ft3)
σ = Interfacial tension, N/m (dyn/cm)

Subscripts

crit = critical
G = Gas
L = Liquid
max = maximum

References
Abe, Y., Akimoto, H., Murao, Y. (1991), Estimation of shear stress in countercurrent
annular flow. Journal of Nuclear Science and Technology, 28(3):208-217.
Alves, M.V.C., Falcone, G., Barbosa, J. R. (2012), Modeling the Transient Behaviour of
Churn-Annular Flow in a Vertical Pipe, 3rd EBECEM (3º Encontro Brasileiro sobre
Ebulição, Condensação e Escoamentos Multifásicos), Curitiba, Brazil, 7 - 8 May 2012.
Al-Darmaki, S., Falcone, G., Hale, C.P., Hewitt, G.F. (2008), Experimental Investigation
and Modelling of the Effects of Rising Gas Bubbles in a Closed Pipe, SPE Journal,
September 2008, 354-365.
Ali, A.M., Falcone, G., Hewitt, G.F., Bozorgzadeh, M., Gringarten, A.C. (2005),
Experimental Investigation of Wellbore Phase Redistribution Effect on Pressure Transient
Data, SPE 96587, SPE Annual Technical Conference and Exhibition, Dallas, Texas, 9-12
Oct 2005.
Alipchenkov, V.M., Nigmatulin, R.I., Soloviev, S.L., Stonik, O.G., Zaichik, L.I., Zeigarnik,
Y.A. (2004), A three-fluid model for two-phase dispersed annular flow, International
Journal of Multiphase Flow, 47: 5323-5338.
Azzopardi, B.J. (2006), Gas-Liquid Flow, Begell House Inc., NY.
Badie, S., Lawrence, C.J., Hewitt, G.F. (2001), Axial viewing studies of horizontal gas-
liquid flows with low liquid loading, International Journal of Multiphase Flow, 2001, Vol:27,
Pages:1259-1269.
Barbosa Jr., J. R. (2001), Phase Change of Single Component Fluids and Mixtures in
Annular Flow, Ph.D. thesis, Imperial College of Science, Technology and Medicine, UK,
2001.

DGMK-Tagungsbericht 2013-1 145


DGMK/ÖGEW-Frühjahrstagung 2013, Fachbereich Aufsuchung und Gewinnung, Celle

Barbosa Jr., J. R. Hewitt, G.F. (2001a), Forced convective boiling of binary mixtures in
annular flow. Part I: liquid phase mass transport, International Journal of Heat and Mass
Transfer 44: 1465-1474.
Barbosa Jr. J. R., Hewitt, G.F. (2001b), Forced convective boiling of binary mixtures in
annular flow. Part II: heat and mass transfer, International Journal of Heat and Mass
Transfer 44: 1475-1484,.
Barbosa Jr. J. R., Hewitt, G.F. (2005) A thermodynamic nonequilibrium slug flow model.
Journal of Heat Transfer, 127(3): 323-331.
Barbosa Jr., J. R., Hewitt, G.F., Konig, G., Richardson, S.M. (2002), Liquid entrainment,
droplet concentration and pressure gradient at the onset of annular flow in a vertical pipe,
International Journal of Multiphase Flow, 28(6): 943-961.
Barbosa, J. R. Jr., Hewitt, G.F., Richardson, S.M. (2001), Churn flow: Myth, mystery and
magic. In: 39th European Two-Phase Flow Group Meeting, Aveiro, Portugal.
Barnea, D., Yacoub, N. (1983) Heat transfer in vertical upwards gas-liquid slug flow,
International Journal of Heat and Mass Transfer, 26(9): 1365-1376.
Belfroid, S.P.C. , Schiferli, W. , Alberts, G.J.N. , Veeken, C.A.M. , Biezen, E. (2008),
Prediction onset and dynamic behaviour of liquid loading gas wells, Proceedings - SPE
Annual Technical Conference and Exhibition, 3, pp. 1528-1536.
Belt, R.J. (2007), On the liquid film in inclined annular flow, PhD thesis, Delft University of
Technology.
Belt, R.J., Van’t Westende, J.M.C., Portela, L.M. (2009), Prediction of the interfacial
shear-stress in vertical annular flow, International Journal of Multiphase Flow 35 (2009)
689–697
Bharathan, D., Wallis, G.B. (1983) Air-water countercurrent annular flow. International
Journal of Multiphase Flow, 9(4):349-366.
Brauner, N., Barnea, D. (1986), Slug/churn transition in upward gas liquid flow, Chemical
Engineering Science, 41(1):159-163.
Chakravarthy S.R., Anderson D.A., Salas, M.D. (1980), The split coefficient matrix
method for hyperbolic systems of gasdynamic equations, In: Proc. AIAA 18th Science
Meeting, Paper 80-0268.
Chen, X.T., Brill, J.P. (1997), Slug to churn transition in upward vertical two-phase flow,
Chemical Engineering Science, 52(23):4269-4272.
Chupin, G., Hu, B., Haugset, T. and Claudel, M. (2007), Integrated Wellbore/Reservoir
Model Predicts Flow Transient in Liquid-Loading Gas Wells, paper SPE 110461
presented at the SPE Annual Technical Conference and Exhibition, Anaheim, California,
11-14 November 2007.
Cioncolini, A., Thome, J. R., Lombardi, C. (2009), Algebraic turbulence modelling in
adiabatic gas–liquid annular two-phase flow, International Journal of Multiphase Flow 35,
580-596.
Coleman, S.B., Clay, H.B., McCurdy, D.G., and Lee Norris, H. III (1991), A New Look at
Predicting Gas-Well Load Up”, J. Pet. Tech., March 1991, 329-333.
Costantini, A. (2005), Dynamic interaction between the reservoir and the well during well
testing, Dip.Ing. thesis, University Sapienza of Rome & Imperial College London, October
2005.
Das, G., Das, P.K., Purohit, N.K., Mitra, A.K. (1999). Flow pattern transition during gas
liquid upflow through vertical concentric annuli. 1: Experimental investigations. Journal of
Fluids Engineering, Transactions of the ASME, 121:895-901.

146 DGMK-Tagungsbericht 2013-1


DGMK/ÖGEW-Frühjahrstagung 2013, Fachbereich Aufsuchung und Gewinnung, Celle

De Cachard, F., Delhaye, J.M. (1996) A slug-churn flow model for small-diameter air-lift
pumps, International Journal of Multiphase Flow, 22(4): 627-649.
Dobran, F. (1983), Hydrodynamic and heat transfer analysis of two-phase annular flow
with a new liquid film model of turbulence, International Journal of Heat Mass Transfer,
26(8):1159-1171.
Dousi, N., Veeken, C.A.M., Currie, P.K. (2005), Modelling the Gas Well Liquid Loading
Process, paper SPE 95282, presented at the Offshore Europe Conference 2005 held in
Aberdeen, Scotland, UK, 6-9 September 2005.
Drew, A., Passman, S.L. (1999), Theory of multicomponent fluids, Vol. 135 Applied
Mathematical Sciences. Springer-Verlag, NY.
Dukler, A.E., Taitel, Y. (1986), Flow pattern transition in gas-liquid systems:
measurement and modeling, In G.F. Hewitt, J.M. Delhaye, and N. Zuber, editors,
Multiphase Science and Technology, volume 2, pages 1-94. Hemisphere Publishing Co.,
New York.
Fernandes, R.C., Semiat, R., Dukler, A.E. (1983) Hydrodynamic model for gas-liquid slug
flow in vertical tubes. AIChE Journal, 29(6): 981-989.
Fiedler, S., Auracher, H. (2004), Experimental and theoretical investigation of reflux
condensation in an inclined small diameter tube, International Journal of Heat Mass
Transfer 47, 4031-4043.
Fu, F., Klausner, J.F. (1997), A separated flow model for predicting two-phase pressure
drop and evapourative heat transfer for vertical annular flow, International Journal of Heat
and Fluid Flow, 18(6):541-549.
Gessner, T.R., Barbosa Jr., J.R. (2010), A three-field model for transient annular flow in a
vertical tube, In: 2nd Brazilian Meeting on Boiling, Condensation and Multiphase Flow
(EBECEM 2010), São Carlos, SP, 3-4 May.
Gill, L.E., Hewitt, G.F., Hitchon, J.W., Lacey, P.M.C. (1963), Sampling probe studies of
the gas core in two-phase flow: I, the effect of length on phase velocity distribution,
Technical report, UKAEA Report AERE-R 3954, 1963. (also: Chemical Engineering
Science, 18, pp. 525-535.)
Govan, A.H. (1990), Modelling of vertical annular and dispersed two-phase flows, PhD
thesis, University of London, Imperial College.
Govan, A.H., Hewitt, G.F., Richter, H.J., Scott, A. (1991), Flooding and churn flow in
vertical pipes. International Journal of Multiphase Flow, 17:27-44.
Gray, H. E. (1974), Vertical Flow Correlation in Gas Wells." User manual for API14B,
Subsurface controlled safety valve sizing computer program, App. B, June 1974.
Guedes de Carvalho, J.R.F. (2006), Experimental study of the slug/churn flow transition
in a single Taylor bubble, Chemical Engineering Science, 61:3632-3642.
Guedes de Carvalho, J.R.F., Talaia, M.A.R., Ferreira, M.J.F. (2000), Flooding instability
of high-density gas slugs rising in vertical tubes filled with water, Chemical Engineering
Science 55: 3785–3802.
Guo, B., Ghalambor, A., Xu, C.(2006), A systematic approach to predicting liquid loading
in gas wells, SPE Production and Operations, 21 (1), pp. 81-88.
Hasan, A.R., Kabir, C.S. (2004), Simplified Wellbore Flow Modelling in Gas/Condensate
Systems, paper SPE 89754, presented at the 2004 Asia Pacific Conference on
Integrated Modelling for Asset Management, Kuala Lumpur, Malaysia, 29-30 March
2004.
Hewitt, G.F. (2010), Flooding and flow reversal, In: Thermopedia, G.F. Hewitt and J.R.

DGMK-Tagungsbericht 2013-1 147


DGMK/ÖGEW-Frühjahrstagung 2013, Fachbereich Aufsuchung und Gewinnung, Celle

Barbosa (Eds.), DOI:10.1615/AtoZ.f.flooding_and_flow_reversal


Hewitt, G.F., Hall-Taylor, N.S. (1970), Annular two-phase flow, Pergamon Press, Oxford,
UK.
Hewitt, G.F., Jayanti, S. (1993), To churn or not to churn? International Journal of
Multiphase Flow, 19:527-529.
Hewitt, G.F., Martin, C.J., Wilkes, N.S. (1985), Experimental and modeling studies of
annular flow in the region between flow reversal and pressure drop minimum.
PhysicoChemical Hydrodynamics, 69(1/2): 69-86.
Hewitt G.F., Wallis, G.B. (1963), Flooding and associated phenomena in falling film in a
vertical tube, In: Proc. Multi-Phase Flow Symposium, pages 62-74, Philadelphia, PA.
Hewitt, G.F., Whalley, P.B. (1978), The correlation of entrained fraction and entrainment
rate in annular two-phase flow, Technical report, UKAEFA Report AERE-9187.
Hinze, J.O. (1955), Fundamentals of the Hydrodynamic Mechanism of Splitting in
Dispersion Processes, ALCHE Journal, September 1955, 1, No. 3, 289.
Ishii, M. (1977): One-Dimensional Drift-Flux Model and Constitutive Equations for
Relative Motion between Phases in Various Two-Phase Flow Regimes, Argonne National
Lab Report, ANL 77-47, October 1977.
Ishii, M., Hibiki, T. (2006), Thermo-fluid Dynamics of Two-Phase Flow, Chapter 13, p.
345, Springer.
Jayanti, S., Brauner, N. (1994), Churn flow, In G.F. Hewitt, J.M. Delhaye, N. Zuber, J.H.
Kim, and R.T. Lahey, Jr., editors, Multiphase Science and Technology - Two-phase flow
fundamentals, volume 8, pages 471-521. Begell-House, Inc., New York.
Jayanti, S., Hewitt, G.F. (1992), Prediction of the slug to churn flow transition in vertical
two-phase flow, International Journal of Multiphase Flow 18:847-860.
Jayanti, S., Hewitt, G.F., Low, D.E.F., Hervieu, E. (1993), Observation of flooding in the
Taylor bubble of co-current upwards slug flow, International Journal of Multiphase Flow,
19:531-534.
Kataoka, I., Serizawa, A. (1989), Basic equations of turbulence in gas-liquid two-phase
flow, International Journal of Multiphase Flow 15, 843-855.
Kelly, J.M., Freitas, R.L. (1993), Interfacial fiction in low flow rate vertical annular flow. In
Proceedings of the Sixth International Meeting on Nuclear Reactor Thermal Hydraulics,
volume 1, 154-160.
Kishore, B.N., Jayanti, S. (2004), A multidimensional model for annular gas-liquid flow,
Chemical Engineering Science, 59: 3577-3589.
Lea, J.F., Nickens, H.V., and Wells, M. (2003), Gas Well De-Liquefaction, first edition,
Elsevier Press, Cambridge, MA.
Mao, Z.S., Dukler, A.E. (1993), The myth of churn flow, International Journal of
Multiphase Flow, 19:377-383.
McQuillan, K.W., Whalley, P.B. (1985), Flow patterns in vertical two-phase flow,
International Journal of Multiphase Flow, 11:161-175.
Mishima, K., Ishii, M. (1984), Flow regime transition criteria for two-phase flow in vertical
tubes, International Journal of Heat Mass Transfer, 27:723-734.
Moeck, E.O., Stachiewicz, J.W. (1972), A droplet interchange model for annular-
dispersed, two-phase flow. International Journal of Heat Mass Transfer, 15:637-653.

148 DGMK-Tagungsbericht 2013-1


DGMK/ÖGEW-Frühjahrstagung 2013, Fachbereich Aufsuchung und Gewinnung, Celle

Nennie, E.D., Alberts, G.J.N., Belfroid, S.P.C., Peters, E., and Joosten, G.J.P. (2007), An
Investigation Into the Need of a Dynamic Coupled Well-Reservoir Simulator, paper SPE
110316 presented at the SPE Annual Conference and Exhibition, Anaheim, California,
11-14 November 2007.
Nicklin, D.J., Davidson, J.F. (1962), The onset of instability in two-phase slug flow, In
proceedings of the Symposium on Two-Phase Flow, UK, Institution of Mechanical
Engineers. 7 February, Paper No. 4.
Nigmatulin, R.I., Nigmatulin, B.I., Khodzaev, Y.A., Kroshilin, V.E. (1996), Entrainment
and deposition rates in a dispersed-film flow, International Journal of Multiphase Flow,
22:19-30.
Nosseir, M.A., Darwich, T.A., Sayyouh, M.H., and El Sallaly, M. (1997), A New Approach
for Accurate Prediction of Loading in Gas Wells Under Different Flowing Conditions,
paper SPE 37408, presented at the SPE Production Operations Symposium, Oklahoma
City, OK, 9-11 March 1997.
Owen, D.G. (1986), An experimental and theoretical analysis of equilibrium annular flow,
PhD thesis, University of Birmingham, UK,.
Orell, A., Rembrand, R. (1986), A model for gas-liquid slug flow in a vertical tube,
Industrial and Engineering Chemistry: Fundamentals, 25: 196-206.
Park, H.Y., Falcone, G., Teodoriu C. (2009), Decision matrix for liquid loading in gas
wells for cost/benefit analyses of lifting options, Journal of Natural Gas Science and
Engineering, Volume 1, Issue 3, September 2009, 72-83.
Pushkina, O.L., Sorokin, Y.L. (1969), Breakdown of liquid film motion in vertical tubes,
Heat Transfer Soviet Research 1, 56-64.
Sagen, J., Sira, T., Ek, A., Selberg, S., Chaib, M., and Eidsmoen, H. (2007), A Coupled
Dynamic Reservoir and Pipeline Model–Development and Initial Experience, 13th
International Multiphase Conference on Multiphase Production Technology, Edinburgh,
UK, 13-15 June 2007.
Sawai, T., Kaji, M., Kasugai, T., Nakashima, H., Mori, T. (2004), Gas–liquid interfacial
structure and pressure drop characteristics of churn flow, Experimental Thermal and
Fluid Science 28:597–606.
Shi, H., Holmes, J.A., Durlofsky, L.J., Aziz, K., Diaz, L.R., Alkaya, B., Oddie, G. (2003),
Drift-flux modelling of multiphase flow in wellbores, SPE 84228, SPE Annual Technical
Conference and Exhibition, Denver, 5-8 Oct.
Shi, H., Holmes, J.A., Diaz, L.R., Durlofsky, L.J., Aziz, K. (2004), Drift-flux parameters for
three-phase steady-state flow in wellbores, SPE 89836, SPE Annual Technical
Conference and Exhibition, Houston, 26-29 September 2004.
Shoham O. (2006), Mechanistic Modelling of Gas/liquid Two Phase Flows in Pipes,
Chapter 4, pp. 146-147, Society of Petroleum Engineers.
Sturm, W.L., Belfroid, S.P.C., van Wolfswinkel, O., Peters, M.C.A.M., and Verhelst, F.
(2004), Dynamic Reservoir Well Interaction, paper SPE 96256, presented at the SPE
Annual Technology Conference and Exhibition held in Houston, USA, 26-29 September
2004.
Sutton, R.P., Cox, S.A., Lea, J.F., Rowlan, O.L. (2010), Guidelines for the Proper
Application of Critical Velocity Calculations, SPE Production & Operations, Vol. 25, No. 2,
pp. 182-194, May 2010.
Sutton, R.P., Cox, S.A., Williams, E.G., Stoltz, R.P., and Gilbert, J.V., (2003), Gas Well
Performance at Subcritical Rates, paper SPE 80887, presented at the SPE Production
and Operations Symposium, held in Oklahoma City, Oklahoma, USA, 22-25 March 2003.

DGMK-Tagungsbericht 2013-1 149


DGMK/ÖGEW-Frühjahrstagung 2013, Fachbereich Aufsuchung und Gewinnung, Celle

Sylvester, N.D. (1987) A mechanistic model for two-phase vertical slug flow in pipes.
Journal of Energy Resources Technology, 109: 206-213.
Toma, P., Vargas, E., and, Kuru, E. (2006), Predicting Slug to Annular Flow Pattern
Transitions for Reducing Risks of Gas Lift Instabilities and Effective Gas/Liquid Transport
from Low Pressure Reservoirs, paper SPE 100615 presented at the Gas Technology
Symposium held in Calgary, Alberta, Canada, 15-17 May 2006.
Turner, R.G., Hubbard, M.G., and Dukler, A.E. (1969), Analysis and Prediction of
Minimum Flow Rate for The Continuous Removal Of Liquids from Gas Wells, J. Pet.
Tech., Nov. 1969, 1475-1482.
Van’t Westende, J.M.C. (2008), Droplets in annular-dispersed gas-liquid pipe flows, PhD
thesis, Delft University of Technology.
Veeken, C.A.M., Belfroid, S.P.C. (2010), New perspective on gas-well liquid loading and
unloading, SPE 134483, Presented at the SPE Annual Technical Conference and
Exhibition, Florence, Italy 19-22 Sept.
Wallis, G.B. (1969), One Dimensional Two-Phase Flow. McGraw-Hill, New York.
Waltrich, P.J., Falcone, G.F., Barbosa Jr., J.R. (2011), Performance of Vertical Transient
Two-Phase Flow Models Applied to Liquid Loading in Gas Wells, paper SPE 147128,
presented at the SPE Annual Technical Conference and Exhibition, Denver, CO, USA, 30
October – 2 November 2011
Wang, Y.W., Zhang, S.C., Yan, J., Chen, W.B. (2010), A New Calculation Method For
Gas-Well Liquid Loading Capacity, Journal of Hydrodynamics, 22(6):823-826
Watson, M.J., Hewitt, G.F. (1999), Pressure effects on the slug to churn transition,
International Journal of Multiphase Flow, 25:1225-1241.
Zabaras, G., Dukler, A. E. and Moalem-Maron, D. (1986), Vertical upward cocurrent gas-
liquid annular flow, AIChE J., 32(5), pp. 829-843.
Zhang, H., Falcone, G., Teodoriu, C. (2010), Modelling Fully-Transient Two-phase Flow
in the Near-Wellbore Region During Liquid Loading in Gas Wells, Journal of Natural Gas
Science & Engineering, Vol. 2, Issues 2-3, July 2010, 122-131.
Zhang, H., Falcone, G., Valko, P., and Teodoriu, C. (2009), Numerical Modelling of Fully-
Transient Flow in the Near-Wellbore Region during Liquid Loading in Gas Wells, paper
SPE 122785 presented at SPE Latin American and Caribbean Petroleum Engineering
Conference held in Cartagena, Colombia, 31 May–3 June.

150 DGMK-Tagungsbericht 2013-1

You might also like