Você está na página 1de 15

Investigation of the effect of gas entrainment on slug

characteristics in actual fields using the slug capturing


technique
R I Issa, J Castagna
Multiphase Simulation Ltd, UK

ABSTRACT

The slug capturing technique, which is incorporated in the industrial PROMPT code is
used to explore the influence of entrainment of gas bubbles at slug fronts, on slug
characteristics such as slug length and frequency. The computations are compared with
available field data. It is found that the amount of gas entrained within the slug can
substantially affect slug lengths, increasing them (almost doubling) in some cases, while
in another it can decrease them. The paper presents the basic methodology, the models
governing the motion of the gas entrained within slugs and the results of comparison with
the field data.

NOMENCLATURE

A pipe cross-sectional area (m2)  angle of inclination of pipe to the


D pipe diameter (m) horizontal (o)
Dh hydraulic diameter (m)  rate of dissipation of turbulence
d bubble/droplet diameter (m) energy (m2s-3)
F frictional force per unit volume  kinematic viscosity (m2s-1)
(Nm-3)  density (kgm-3)
f friction factor (-)  surface tension (N/m)
g gravitational acceleration (ms-2)  shear stress (Nm-2)
h height of liquid layer (m)
entrainment mass transfer rate Subscripts
(kgm-3s-1) b entrained gas bubble
phase change mass transfer rate e entrained quantity
from liquid to gas (kgm-3s-1) g gas phase (continuous)
p pressure (Nm-2) i interface between continuous phases
Re Reynolds number (-) l liquid phase
U superficial velocity (ms-1) lf liquid film ahead of slug
u velocity (ms-1) ls liquid in slug
ut slug translational velocity (ms-1) o oil phase
t time (s) m mixture phase (oil+water +entrained
x distance along the flow (m) gas)
 Phase fraction (-) w water phase

© BHR Group 2018 Multiphase 11 229


1. INTRODUCTION

Intermittent flow, especially the slug flow regime, occurs very frequently in the transport
of hydrocarbon fluids in pipelines in the oil and gas industry. The slug flow regime is
usually undesirable since the intermittency of slugs causes severe adverse conditions.
Firstly, the flow rates of the gas and oil arriving at the receiving equipment (separators
and slug-catchers) can fluctuate widely thereby undermining the functioning of the
equipment (e.g. separator flooding). To cater for this situation, processing facilities are
usually designed with generous “safety margins” at the cost of capital, weight and size.
Secondly, the flow intermittency results in highly unsteady loading on the piping system
and processing equipment, which can result in serious failure due to metal fatigue. It is
therefore important to be able to predict the onset of slug flow; it would be even more
beneficial if slug characteristics, such as slug length and frequency could be calculated
reliably as well.

Slug flow can be generated in pipelines at all inclinations, ranging from purely horizontal
to fully vertical. There are several mechanisms that lead to this: (i) liquid accumulation
due to instantaneous imbalance between pressure and gravitational forces caused by
undulations in nearly horizontal pipes, (ii) natural growth of hydrodynamic instabilities
in horizontal and slightly inclined pipes and (iii) liquid fall back in vertical flow
accumulating as slugs. In the first, slugs may form at pipe dips due to the retardation and
subsequent accumulation of liquid in the dips leading to the filling up of the cross-section
with liquid. An extreme example of this terrain induced slug flow is called ‘‘severe
slugging’’ and occurs when a slightly inclined pipeline meets a vertical riser (e.g.
Schmidt et. al. [1]). In the case of hydrodynamic instabilities, small random perturbations
of short wavelengths arising naturally may grow into larger and longer waves on the
surface of the liquid. The mechanism behind this growth is attributed to the classical
Kelvin–Helmholtz instability (Lin and Hanratty [2]). Such waves may become stable and
move as roll-waves or they may grow by picking up liquid flowing ahead of them
thereby forming slugs.

Figure 1: Photograph and schematic of entrained gas bubbles in slug

An important phenomenon associated with slug flow in pipes of all inclinations is the
presence of dispersed gas in the form of small bubbles within the body of the liquid slug
(as depicted in figure 1). The amount of entrained gas varies according to the flow
conditions and can be absent altogether in some circumstances. The presence of
dispersed gas in the slug is known to be due to a process of entrainment at the slug head.
Due to the liquid motion, the entrained gas moves relative to the slug in the opposite

230 © BHR Group 2018 Multiphase 11


direction, exiting at the slug tail. An equilibrium state is reached when the rate of
entrainment at the slug head is equal to that of gas release at the tail; the volume fraction
of the entrained gas then becomes constant (though not necessarily uniform). It is thought
that the presence of this entrained gas would have an influence on the slug
characteristics, and this is the subject of investigation in this paper.

Dynamic multiphase flow computations in industrial pipelines are normally based on the
two-fluid model, which is incorporated in well-established commercial codes like
OLGA. Most of those simulations use coarse grids (pipe segments) of several tens or
hundreds of metres in length; such calculations do not resolve fast transients nor are able
to capture hydrodynamic instabilities automatically. Instead, slugs are usually initiated
artificially using empirical or semi-empirical correlations; those artificially generated
slugs are then tracked within the framework of the two-fluid model. The ability of such
approaches to capture the characteristics of slugs reliably is therefore questionable.

The slug-capturing technique introduced by Issa and Woodburn [3], and Issa and Kempf
[4] was shown to be able to automatically predict the initiation of slugs in horizontal and
slightly inclined pipes. The technique was also shown to be successful in predicting
vertical slug flow by Issa and Galleni [5]. The principle behind this technique is
straightforward in that the transient, one-dimensional, two-fluid model equations must be
solved numerically using very fine meshes (smaller than the pipe diameter) to enable the
capturing of the growth of disturbances leading to the generation and subsequent
development of slugs in an automatic manner. It has also been demonstrated that the
resulting predictions for the main characteristics of slug flow compare astonishingly well
with experimental data including the ability to generate slugs of randomly variable
lengths at different instants in time. Such prediction indeed corresponds to what happens
in reality. Bonizzi & Issa [6] extended the methodology successfully from two to three
phases. The slug capturing technique has been implemented in the industrial PROMPT
code (used in the present study) and an implementation in the commercial LedaFlow
code is also claimed [7].

Accounting for the gas entrained into slugs is made in standard modelling methods by
assuming a fixed gas fraction within slugs, the value of which is obtained from empirical
correlations. However, it is not obvious how this entrained gas can be utilised, if at all, in
accounting for its influence on slug characteristics. In contrast, the slug capturing method
by its nature is able to automatically simulate the presence of this entrained gas
dynamically by solving an additional transport equation for it (Issa et. al. [8]).

The capabilities of the slug capturing technique incorporated in the PROMPT code have
been explored by industry (see Lockett et.al. [9]). The code is used in the present work to
predict two and three-phase flows in field applications to investigate the influence of
entrainment of gas bubbles on slug characteristic such as slug length and frequency. The
computations are compared with published BP field data. The paper presents the basic
methodology, the model used for accounting for the motion of the gas entrained within
slugs and the results of comparison with field data.

2. THE MATHEMATICAL MODEL

2.1 The conservation equations


The basis of the multi-fluid model is the formulation of sets of conservation equations for
the balance of mass, momentum and energy for each of the phases. The one-dimensional
form of the model is obtained by integrating (area averaging) the flow properties over the

© BHR Group 2018 Multiphase 11 231


cross-sectional area of the flow. Transfer of momentum and energy between the walls
and the fluids and between the phases themselves across the interfaces is accounted for
via source terms in the equations; they are formulated using empirical correlations (e.g.
Ishii and Mishima, [10]) for friction factors and heat transfer coefficients. In the case of
gas entrainment, the rate of ingress of gas into the slug is computed from an appropriate
closure relation.

The flows considered in this work are both two and three-phase. The equations governing
the flow are for mass and momentum for the gas and a liquid mixture phase comprising
of water, oil and entrained gas bubbles. Within the mixture phase, mass conservation
equations are solved for water and entrained gas contents. Slip between oil and water
within the mixture is accounted for by a drift-flux model; another drift flux model is used
to calculate the velocity of the gas bubbles within the slug. The cases considered are
isothermal, hence the energy equation is not solved. The governing transport equations
are:

Gas Continuity:

( )
+ = − (1)

Mixture Continuity:

( ) ( )
+ =− + (2)

Gas Momentum:

( )
+ =− − + − − (3)

Liquid Mixture Momentum:

( )
+ =− − + − +
+Ψ (4)

In these equations, α represents the volume fraction with the condition that: αm+αg=1.
The quantities and stand for the rate of mass transfer from the liquid hydrocarbon
to the gas due to phase transfer (obtained from PVT tables) and the rate of entrainment of
gas into the slug respectively. The Fm and Fg quantities denote respectively the frictional
forces at the wall acting on the liquid mixture phase and gas phase; Fi stands for the
inter-phase force between the continuous gas and mixture phases. The term  in equation
(4) represents the hydrostatic pressure difference in the oil and water when the liquid
phases flow in a stratified state, as well as an additional term due to the difference in
velocities of the two liquids and gas bubbles in the mixture; the expressions for these
terms are given by Barbeau [11].

The mixture phase is composed of oil, water and dispersed gas bubbles, the fraction of
each is given by o, w and b such that: o+w+b=m. To compute each of those
quantities, two additional mass conservation equations need to be solved:

232 © BHR Group 2018 Multiphase 11


Water:

( ) ( )
+ =0 (5)

Entrained gas:

( ) ( )
+ = (6)

The densities of the oil, water and gas are obtained from PVT tables, and those determine
the mixture density, which is given by:

= + + (7)

Furthermore the mixture velocity is determined from the individual velocities of oil,
water and entrained bubbles as:

= + + (8)

The respective velocities of the liquids and the entrained bubbles are related to the
mixture velocity by slip relations (drift flux model) to be described in the following
section.

2.2 Closure relations


Closure models are needed to determine:
 The frictional forces in the momentum equations
 Whether the liquids are in stratified or dispersed state
 Which liquid is in a continuous state and the point of phase inversion
 The sizes of dispersed droplets in the mixture (if dispersed)
 The slip velocity between the two liquid phases in the mixture
 The size of entrained bubbles in the slug and their velocity
 The entrainment rate of gas into the slug body

All of these closures are invoked dynamically at every point along the pipeline.

The frictional forces per unit volume, Fg, Fm and Fi between each phase and the wall and
between the phases at the interface respectively are prescribed by the following closure
relations:

= /
= | | / (9)
= − ( − ) /

where Sg and Sl are the perimeters of the contact surfaces between the wall and gas and
between the wall and liquid (denoted by l) which can either be oil (o) or water (w)
depending on which phase is in contact with the wall; Si is the interfacial contact length
between the continuous phases, which also depends on which liquid is in contact with the
continuous gas phase (hence the subscript l).

The friction factors, f, in equations (9) are to be determined from correlations that depend
on pipe inclination. In PROMPT, several friction factor correlations (such as those of

© BHR Group 2018 Multiphase 11 233


Taitel & Dukler [12] and Spedding & Hand [13]) are available; typically they take the
form:

= . (10)

where the Reynolds number is based on the hydraulic diameter Dh defined as:

= (11)

The coefficients C and n take different values depending on the model used.

In three-phase flow, the two liquids may flow either in segregated state or dispersed into
each other in a mixture. In the latter case, there are two different possibilities: the oil may
be dispersed into a continuous water phase, or water dispersed in the oil phase. This can
change dynamically within the pipeline as phase inversion may occur. The mixture in the
presence of gas entrainment is further complicated by the presence of bubbles. The
schematic in Figure 2 illustrates the various possible flow configurations in stratified
flow. To cater for all of these possibilities, closure models are used to determine if the
liquids are dispersed into each other or flow in a segregated state, and if dispersed which
of the liquids is continuous and which is dispersed. Such models had been introduced and
validated in [11], and are implemented in the PROMPT code; because of their
complexity, the details are omitted here for brevity.

Figure 2: Possible flow patterns in stratified flow

The slip velocity between the liquids (oil and water) plays an important role in
determining the water-cut at every point in the pipe. The value of that velocity depends
on whether the liquids flow in a segregated or dispersed state; in both cases its value can
be found from a balance of forces, these being the local pressure gradient (computed as
an outcome of the solution procedure), wall shear (if in contact) and inter-phase force,
assuming steady state (see [11]). Similarly, the bubble velocity is computed from a
balance between the local pressure gradient and the drag force acting on the bubble. An
example of the applicable equation for a bubble is as follows:

= ( − ) (12)

from which the velocity ub can be calculated. In arriving at equation (12) it is assumed
that the bubble is spherical of diameter d. The quantity Cd is a drag coefficient whose
value can be determined from any one of standard correlations existing in the literature

234 © BHR Group 2018 Multiphase 11


such as those due to Haberman and Morton or Peebles and Garber (see Wallis [14]
relating it to the bubble/droplet Reynolds number. The nominal bubble size (or droplet
when the two liquids flow in a dispersed state) is estimated from relationships such as
that given by Wallis [14]:

= 0.725 (13)

where  is the surface tension, and  is the rate of dissipation of turbulence energy that
may be estimated from the wall friction force acting on the liquid phase.

The entrainment rate at the head of the slug is an essential parameter for the simulation of
the entrainment process and it needs specification. There are several correlations that
estimate this value, almost all of which relate the entrainment to the difference between
the velocities of the slug and of the liquid in the film ahead of it. Some sophisticated
models such as that due to Brauner & Ullman [15] exist and that model is incorporated in
PROMPT. However, here the simpler closure model, also incorporated in PROMPT,
which assumes the phenomenon of entrainment to be analogous to a hydraulic-jump is
used. The entrainment rate expression due to Chanson [16] gives the value as:

= − ( − 1) / (14)

In this equation the translational velocity of the slug ut, the liquid film velocity ulf and the
liquid film fraction lf can all be obtained dynamically from the actual computations
themselves, as they are an outcome of the solution process. The quantity V is the volume
of the computational cell surrounding the slug front and A is the area of the pipe. The
Froude number Fr is given by:

= (15)
/

The coefficient  and the index m in equation (14) take difference values depending on
the Froude number (see [8]). Also, entrainment is supposed to occur only when Fr>1,
otherwise it is zero.

2.3 Numerical solution technique


The system of equations (1-6) is solved numerically. The equations are discretised on a
staggered grid using a conservative finite volume method following the description by
Issa & Kempf [4.] The upwind difference scheme is invoked to represent the spatial
derivatives while the Euler implicit scheme is used for temporal integration. The
resulting set of simultaneous non-linear algebraic equations is solved in PROMPT
utilising a novel coupled iterative procedure to handle the non-linearity of the equations.

The boundary conditions used in all the calculations are as follows. At the pipe inlet, the
liquid and gas flow rates and volume fractions were specified and were assumed to
remain constant at those values. The pressure at the pipe outlet was fixed at the given
value.

As was mentioned earlier, in order to capture the fine details of the flow leading to the
automatic generation of intermittent/slug flow, small mesh sizes (grid intervals of the
order D/3) and short time steps of (of the order of milliseconds) have to be used. This
obviously demands very high computing resources to perform calculations that
commence from arbitrarily assumed flow conditions, then proceed to capture the

© BHR Group 2018 Multiphase 11 235


initiation and subsequent development of waves/slugs automatically, and finally reach a
statistically steady state. Typical simulations cover several hours of actual flow time.

It is clear that in an industrial application where pipelines are of the order of tens, or even
hundreds, of kilometres in length, meshes of the order of tens of thousand nodes together
with millions of time steps may be necessary to perform the kind of slug-capturing
calculations shown here. In order for the computations to be completed within feasible
times (and to be of use to industry), PROMPT utilises the power of parallel computing to
accelerate the computations. This is achieved by utilising either multi-core desktop
computers or machines that are equipped with a GPU (Graphics Processing Unit) both of
which are supported by the code. The computations presented herein were performed
using both.

3. RESULTS

3.1 Cases considered


Three BP field cases have been studied for which some measurements are published by
Lockett et. al. [9] and Fan et. al. [17]. The cases are denoted by the letters A, F and V.
Different data are provided for each case, mainly liquid density histories measured with
gamma densitometers. Only in case A, a histogram of slug lengths is provided. PROMPT
is run for all those cases with and without gas entrainment activated and the results are
compared against available data. All simulations were made using a mesh size of D/3.

Case A
This case involves three-phase flow in a pipeline that is about 11km long the geometry of
which is shown in Figure 3 (taken from ref. [17]). The inner diameter is 0.743 m (30
inches). The pipeline is connected to a 24-in header, and then split into two headers. Each
bank is connected to a separator. The separators have the same operating pressure of 11.2
barg. In the present simulation an equivalent pipe is used to model the two banks as was
3
done in the work presented in [17]. The gas flow rate is 66.859 sm /s, which includes
3
both the lift gas and produced gas. The oil flow rate is 0.078 sm /s, and the water flow
3 3 3
rate is 0.151sm /s. The GOR is calculated to be 856 sm /sm , and the water cut is 65.8%.

Figure 3: Pipeline profile in case A

236 © BHR Group 2018 Multiphase 11


Figure 4: Computed mixture hold-up distribution; case A

Figure 4 shows a snapshot of the predicted mixture hold-up distribution along the
pipeline, from which it is clear that fairly long slugs are being captured. These slugs
change in length as they proceed along the pipeline and encounter terrain undulations. A
typical predicted pressure-time history at a point 7 km downstream of inlet is compared
with the measurements of ref. [17] in figure 5. It can be observed that pressure
fluctuations of more than 2 bar have been measured and these are due to the passage of
slugs; this is also predicted by the simulation although the average pressure is some 7%
smaller than that measured. It is interesting to note that earlier predictions presented in
ref. [17] (denoted by “Anonymous code” in figure 5) using standard modelling technique
(i.e. non slug capturing) fail to predict these very large pressure fluctuations.

Figure 5: Pressure history at a point 4 km from inlet; case A

Of interest in this study is to assess the influence of gas entrainment on slug


characteristics and this is illustrated in figures 6 and 7 where the slug length distributions
are plotted respectively against field data. The measured slug lengths are determined

© BHR Group 2018 Multiphase 11 237


from measurements by two densitometers, located 7 km downstream of inlet, that detect
the apparent density of the flowing fluids. Such measurements cannot distinguish clearly
between waves and slugs containing entrained gas. Hence, the slug histogram obtained
has some uncertainties, which need to be borne in mind.

It is clear from figures 6 and 7 that at least for this case, gas entrainment has a major
effect on the predicted slug lengths, almost doubling the size of the longest slug.
Inclusion of gas entrainment brings the computations much closer to the measured
spectrum of slugs. This is thought to be a direct consequence of the swelling of the slug
body caused by the presence of the entrained gas. Such pronounced effect is likely to be
due to the high entrained-gas fraction (around 30%). It may therefore be concluded that it
is important to account for the entrainment phenomenon if correct slug length predictions
are to be obtained.

Figure 6: Measured and Predicted slug length distribution with entrainment;


case A

Figure 7: Measured and Predicted slug length distribution without entrainment;


case A

Case F
This is an oil field in moderate water depth; the pipeline has 10 inch internal diameter
and is 7.5 km long, with about 300m flexible riser; the pipeline profile is shown in figure
8. The flow rates of the only two phases are: 13800 stbd of oil and 6.7 mmscfd of gas.

238 © BHR Group 2018 Multiphase 11


Figure 8: Case F pipeline profile

A snapshot of the liquid hold-up close to the riser is shown in figure 9. The figure
displays a large slug flowing through the flexible riser where it is broken up into shorter
slugs. This case was also simulated with and without entrainment, and the results are
presented in figures 10 and 11 respectively for slugs arriving at the bottom of the riser.

Figure 9: Instantaneous liquid hold-up; case F

Figure 10: Slug length distribution with entrainment; case F

Figure 11: Slug length distribution without entrainment; case F

© BHR Group 2018 Multiphase 11 239


Here again, the inclusion of entrainment is shown to increase the slug lengths
considerably; unfortunately there are no data to compare against. However, ref. [9]
presents temporal measurements of the density as measured by a densitometer. The
results are shown in figure 12, while the corresponding simulation is depicted in figure
13. It appears that slugs of frequency similar to that measured are predicted (judged by
the peaks). However, the predicted densities are higher than those measured perhaps
indicating that there is more entrained gas than predicted. It is worth noting however that
Lockett et.al. [9] point out in their paper (from which the field data are extracted) that
there is uncertainty in the actual field conditions, hence the computations may not reflect
reality exactly.

Figure 12: Measured density variations; case F

Figure 13: Predicted density variations; case F

Case V
This case is of thee phase flow from an oil field in shallow water with top of riser
choking. The pipeline is almost entirely horizontal of length 6.5 km with a vertical riser
of approximately 70 m and a similar length downcomer; the internal diameter is 9.6
inches. The flow rates of oil, water and gas are: 5574 stbd, 1038 stbd and 11.6 mmscfd.

240 © BHR Group 2018 Multiphase 11


Here the simulations showed an opposite effect to the previous two cases, in that the
slugs became shorter when entrainment was accounted for as shown in figures 14 and 15.

Figure 14: Slug length distribution with entrainment; case V

Figure 15: Slug length distribution without entrainment; case V

Figures 16 and 17, which display respectively the time variation of measured and
predicted density with no entrainment indeed show reasonable agreement between data
and computation for slug frequency. However, the lowest predicted densities (pertaining
to regions of gas bubbles) are much lower than what is measured. This may indicate that
the computed liquid content in those regions is under-predicted; however it should be
borne in mind that this only represents an under-prediction of 12% in liquid content in
those regions. Moreover, Lockett et. al. point out that the measurements are obtained
from historical data and there were uncertainties around their accuracy.

Figure 16: Measured density variation; case V

© BHR Group 2018 Multiphase 11 241


Figure 17: Predicted density variation with no entrainment; case V

In this particular case, the computed results have yielded fewer and longer slugs without
entrainment whereas many more but shorter slugs are obtained when entrainment is
included. This suggests that the process of entrainment may be breaking up the larger
slugs into many shorter ones. It is possible that the entrainment rate model presented
earlier in section 2.2 leads to too high an entrainment (30%) as shown in figure 18, where
the phase fraction distribution of the entrained gas in a typical slug is presented. Perhaps
a better entrainment model is required, but this needs further investigation.

Figure 18: Phase distribution in vicinity of typical slug; case V

4. CONCLUSIONS

The slug capturing technique incorporated in the PROMPT code has been applied to
actual field pipelines for which data are published. The study into the effect of gas
entrainment into slugs shows that the process has a dramatic effect on slug
characteristics, namely length and possibly frequency. In two cases, entrainment leads to
much longer slugs (thought to be due to slug swelling with entrained gas), and that drives
the simulations much closer to field measurements. In another case entrainment shortens
the slugs by breaking them up but this is thought to be an anomaly since it is at variance
with the measurements. Closer agreement is obtained in that case with no entrainment,
which leads to the conclusion that the entrainment rate closure model used is not perfect;
different closures such as that of ref. [15] should be further investigated.

242 © BHR Group 2018 Multiphase 11


ACKNOWLEDGEMENT

The authors wish to thank Dr. Tim Lockett of BP and his team for providing
clarifications to the data presented in their published work and which were needed in
setting up the cases.

REFERENCES

1. Schmidt, Z., Doty, D.R. & Dutta-Roy, K., “Severe slugging in offshore pipeline
riser-pipe systems”, J. SPE, 27–38, 1985.
2. Lin, P.Y. & Hanratty, T.J., “Prediction of the initiation of slugs with linear stability
theory”, Int. J. Multiphase Flow, 12, 79–98, 1986.
3. Issa, R. & Woodburn, P.J., “Numerical prediction of instabilities and slug
formation in horizontal two-phase flows”, Proc. Third Int. Conf. on Multiphase
Flow, ICMF98, Lyon, France, 1998.
4. Issa, R.I. & Kempf, M.H.W., “Simulation of slug flow in horizontal and nearly
horizontal pipes with the two-fluid model”, Int. J. Multiphase Flow, 29, 60-95,
2003.
5. Issa, R.I. & Galleni, F., “Mechanistic simulation of slug flow in vertical pipes using
the one-dimensional two-fluid model”, Multiphase Science and Technology, 27 (2-
4): 229–245, 2015.
6. Bonizzi, M. & Issa, R.I., “On the simulation of three-phase slug flow in nearly
horizontal pipes using the multi-fluid model”, International Journal of Multiphase
Flow 29, 1719–1747, 2003.
7. www.ledaflow.com
8. Issa, R. I., Bonizzi, M. and Barbeau, S. “Improved closure models for gas
entrainment and interfacial shear for slug flow modeling in horizontal pipes”, Int. J.
Multiphase Flow, 32, 1287–1293, 2006.
9. Lockett, T.J., Fan, Y. & Ajani, A., “Advances in modelling hydrodynamic slug
flow in production systems”, Proc. 18th Conf. on Multiphase Production
Technology, 293-307, Cannes 2017.
10. Ishii, M. & Mishima, K., “Two-fluid model and hydrodynamic constitutive
relations”, Nucl. Eng. Des. 107–126, 1984.
11. Barbeau, S. “Improved models for transient one-dimensional simulations of
multiphase slug flows”, PhD Thesis, Imperial College London, 2008.
12. Taitel, Y. & Dukler, A.E., “A model for predicting flow regime transitions in
horizontal and near horizontal gas–liquid flow”. AIChE J., 22, 47–55, 1976.
13. Spedding, P.L. & Hand, N.P., “Prediction of stratified gas-liquid co-current flow in
horizontal pipelines”, Int. J. Heat Mass Transfer, 40, 1923-1935, 1997.
14. Wallis, G.B., “One-dimensional two-phase flow”, McGraw-Hill Book Co., 1969.
15. Brauner, N. & Ullmann, A., “Modelling of gas entrainment from Taylor bubbles,
Part A: Slug flow”, Int. J. Multiphase Flow, 30, 239–272, 2004.
16. Chanson, H., “Air Bubble Entrainment in Free-surface Turbulent Shear Flow”,
Academic Press, 1996.
17. Fan, Y., Clausen, R., Lockett, T. & Schutte, R. “Measurement and simulation of
slug flow in a large diameter three-phase pipeline”. Proc. 16th Conf. on Multiphase
Production Technology, Cannes, 2013.

© BHR Group 2018 Multiphase 11 243

Você também pode gostar