UA

UEIRA

ir
CRANFIELD UNIVERSITY

F
PAULO CESAR RIBEIRO LIMA

Modelling of Transient Gas-liquid Flow and Pigging in Pipes

1,9949

SCHOOL OF MECHANICAL ENGINEERING

PhD THESIS
 CRANFIELD UNIVERSITY

SCHOOL OF MECHANICAL ENGINEERING

PhD THESIS

Academic Year 1998-1999

PCR Lima

MODELLING OF TRANSIENT GAS-LIQUID FLOW

AND PIGGING IN PIPES

Supervisor:Dr H Yeung

August 1999

TMs thesis is submitted m partial submission

for the degreeof Doctor of Philosophy
 - A-

ABSTRACT

More and more transient gas-liquid operations in pipes are being successfully applied in the oil
and gas industry. Pigging in two-phase pipelines, to remove liquid accumulation or for cleaning
purposes, is an important transient operation. Another important operation is the injection of (-)-as
to transport the accumulated liquid in the pipeline to process facilities. Analysis of such transient
two-phase flow in a pipeline is necessary not only for designing the liquid and (Yashandling
facilities, but also for safe operating procedure. In pipeline-fiser system such operations cause
even more severe changesin flow conditions.

A two-fluid model hasbeen developedto determinethe transientbehaviour of fluids during these

operations. The derived for
one-dimensionalset of equations each flow pattern describe the flow
of fluids in all regions. Semi-implicit finite difference to
schemeswere used solve the initial and
boundary value problem for each phase of the process- gas/pig injection, gas shut-in, slug
production and gas flow out of the system.

An extensive experimental program has been carried out to acquire two-phase transient flow and
pigging data on a 67 m long, 0.0525 m diameter, 9.9 m high A
pipeline-riser system. computer
based data acquisition system has been utilised to obtain rapidly changing and detailed information
of the flow behaviour during the transient tests. The model results compare well with the
experimental data for characteristics such as inlet pressure, hold-up and pig velocity.

III
 ACKNOWLEDGENIENTS

I would like to express my thanks to my friends and colleagues who have been directly or
indirectly connected with the work in this thesis. First of all my support panel, Dr. Hoi Yeung for
his invaluable help in the technical aspect of the work, Dr. Will Rawls, Tom Mathew and specially
to Jonathan Montgomery for his great contribution during the experimental stage of this work.
II
I would also like to thank to the Design Office of Cranfield University who designed and
manufactured the pig launcher and receiver and to the CALTEC team, specially Wai Lam Loll for
our chats about the experimental facilities.

The author is very grateful to PETROBRAS for providing him with the opportunity to attend
Cranfield University. PETROBRAS human resource development programme can surely be
considered one of the finest among Petroleum Companies around the world.

This dissertation is dedicated to the my wife, Luiza, whose love, understanding and constant
encouragementmade this work possible, to my parents and to Frederico and Henrique, my sons.

iv
 NOMENCLATURE

Variables Description

A pipe cross sectional area

AF area of flow

CO distribution coefficient

C shear or ffiction coefficient

D diameter

E,F velocity coefficients

G body force

f Fanning ffiction factor

9 by
acceleration caused gravity

h liquid height in stratified flow

k shear force multiplying factor

ni mass

P pressure

P perimeter

time

velocity

position from pipeline inlet

a volume fraction, hold-up

.6 absolute pipe roughness

N,
a liquid film thickness in annular flow

A increment

pipe inclination angle

P viscosity

P density

T
shearstress

Subscripts Description

b bubble

d drift

f film

9 gas

I gas-liquid interface

h hydraulic

one-dimensional cell index

liquid

m mixture

s slug

w pipe wall

Superscripts Description

n time index

0 value in the basic step of SETS method

vi
 CONTENTS Pac-ye

Abstract 111

Acknowledgements V

Nomenclature
Contents vii

CHAPTER I Introduction
Introduction I
1.1 Conventional Pigging I
1.2 Pig Lift and Gas Pump
1.3 Objective of this Work 6

CHAPTER 2 Review of Relevant Literature

Introduction 7
2.1 Flow Regimes and Two-phase Flow Models 7
2.2 Transient Gas-liquid Flow 9
2.3 Pigging 12
2.4 Intermittent Gas Lift
2.5 Summary

CHAPTER 3 Transient Model

Introduefion 14
3.1 Proposed Transient Model 15
3.2 Model Solution 20

CHAPTER 4 Experimental Facilities

Intwduction 30
4.1 Air and Water System 30
4.2 Test Section 32
4.3 Pig Specification 32

CHAPTER 5 Pigging Model

Introduction
5.1 Pig By-pass
5.2 Pigging Nlodel with By-pass 41

vii
CHAPTER 6 Results and Discussions
Introduction 4.)
6.1 Conventional Pigging 4 3)
6.2 Gas Pump without Pig 54
6.3 Gas Pump with Pig 59
6.4 Transient Runs 68

CHAPTER 7 Conclusions and Recommendations

Introduction 76
7.1 Conclusions 76
7.2 Recommendations 78

REFERENCES 79

APPENDICES

A Experimental Data of the Conventional Pigging Runs

B Experimental Data of the Gas Pump without Pig Runs

c Experimental Data of the Gas Pump with Pig Runs

D Experimental Data of the Transient Runs

E Model Prediction of the Conventional Pigging Runs

F Model Prediction of the Gas Pump without Pig Runs

G Model Prediction of the Gas Pump with Pig Runs

H Model Prediction of the Transient Runs

I Steady-State Slug Flow Model

viii
 I

CHAPTER 1. INTRODUCTION

The use of long gas-liquid pipelines is becoming an attractive alternative to conventional

monophase systems, Utilisation of two-phase pipeline-riser systems will continue to grow since
the vast majority of recent discoveries have been located offshore and because of the grovAng
number of horizontal wells. Conventional gas-liquid pipeline-riser pigging, gas pump and pig lift
are severe casesof transient operations.

1.1 Conventional Pigging

A pig is a tool used for various maintenancetasks in pipelines. It is defined as 'a projectile, forced
through the inside of a pipe using either hydraulic or pneumatic pressure, while maintaining a
positive sea] with the pipe wall'. Maintenance tasks include functions such as cleaning the line,
removal of liquid hold-up or condensate from gas pipelines and inspection. In gas-liquid flow the
pig acts as a moving boundary, scooping the liquid ahead of it into an expanding liquid slug
region, and leaving behind a region with almost no liquid as shown in Fig. 1.1, causing a severe
case of transient flow. Knowledge of the dynamic behaviour of the flow is very important to
properly design and operate gas-liquid pipelines and their upstream and downstream facilities. A
frequently pigged wet-gas pipeline is capable of transporting up to 70% more gas as compared to
operation without pigging, according to McDonald and Baker (1964). Gas-liquid flowlines
connecting subsea wellheads or manifolds to surface facilities in deep water are being regularly
pigged to avoid accumulation of organic deposits causing great operational and safety problems
due to the severity of the phenomena according to Lima and Neto (1995) and Lima and Yeung
(1998).

Standard design procedure for gas-liquid pipeline-riser systems subject to pigging still rely on
steady-state empirical correlations and mechanistic models, or sometimes, on simplified quasi-
steady-state pigging models. The use of steady-state empirical correlations and mechanistic
models for predicting pressure, flow rates and liquid hold-up in the pipeline can result in oversized
or undersized facilities.

Upstream Region pig Downstream Region

Figure 1.1 - Conventional gas-liquid pipeline pigging

 I

1.2 Pig Lift and Gas Pump

Petroleumis found below the ground filling porous medium regions called reservoirs.One of the
great challengeof the oil industry is to transport the oil from the reservoir to surface.After an oil
well has been completed the oil flow to the surfacewill only occur if the reservoir pressureis
sufficient to overcomethe back pressureexertedby the headof the production piping system.In
caseswhere the reservoir pressureis not high enoughit then becomesnecessaryto utilise some
in
artificial method order to lift the oil to the surface. Some of these methods are mechanical
pumping systems.A commonfeature sharedby all thesesystemsis that they require the supply of
some kind of energy to drive the pumping equipmentby some physical means.This may be an
electrical cable to feed the motor of a submersedcentrifugalpump, or a string of mechanicalrods
for driving a rod pump or a progressivecavity pump. All of these systemsshare the common
featurethat they consistof a great numberof componentswhich are prone to failure.

A widely employed method is pneumatic pumping, known by 'gas lift' (Brown 1980), which
-
consists basically of injecting gas into the annular space that exists between the production strings
and the well casing as shown in Fig. 1.2. Gas is injected into the production string by means of
special valves with the purpose of gasifying the oil. This gasification reduces the fluid density and
facilitates the flow off to the surface. There are two 'gas lift' systems namely 'continuous gas lift'
and 'intermittent gas lift'. As the very name suggests,gas is injected continually into the annular
it
spaceuntil reachesa valve at the bottom of the well which allows the gas to be injected into the
interior of the production string. An efficient continuous gas lift requires a flow rate of gas 5 times
greater than the flow rate of liquid. In the 'intermittent gas lift', contrary to the previous one, the
well is allowed to produce for some time without injecting gas. Gas is then injected into the
annular space at quite high pressures. Special valves installed in the 'gas lift mandrel' allow the
gas to be injected into the production stnng thus causing the effect of pumping the oil to the
surface.

In spite of some advantagesof the conventional gas lift methods, it has some inconveniences, such
as,, for instance, the high the back pressure exerted by the gas and oil accumulated in vertical or
inclined pipes, which is a limiting factor in oil production. Lima (1996) proposes two methods of
intermittent gas lift systems that eliminate this inconvenience, in which accumulated oil does not
cause back pressure to the production zone and a pig can be utilised to increase the efficiency of
the pneumatic pumping system. The first method, named pig lift, is shown in Fig. 1.3. Two
production strings stretch from the well head to one joint piece which is responsible for their
interconnection downhole. The connection tool is run together with the longest string, which also
carries a standing valve in its lower end. As a consequence of the utilisation of two production
strings, several surface components are utilised pair wise. Certain components that appear in the
portions interconnected with one of the tubing strings have equivalent ones in the portion
interconnected with the other tubing string, with both components performing one single function.
 3

GAS + OIL

GAS VALVES ou
-
MASTER
VALVE

GAS 01

PRODUCTION
STRING

CASING

GAS LIFT VALVE

PACKER

STANDING VALVE PRODUCTION

ZONE

Figure 1.2 - Gas Lift

The pig lift operation begins with the opening of the pig valve of the launching device, allowing
the introduction of at least one pig. Production valves are open while gas valves are closed. This
procedure assures the accumulation of oil in both production strings. Afler a certain period of
time, the pig is launched. For this, either one of the production valves is closed and either one of
the gas valves is opened; they must be the ones installed in the same portion where the pig was
introduced before.

Inasmuch as the pressure of the supply gas is higher than the pressure of the existing production
strings, the pig is pushed by the gas. The pig descends through the whole production string until
it reaches the special coupling; it then enters through the other production string and rises,
causing the displacement to the surface of the liquid accumulated in both production strings. The
upward movement of the pig is naturally interrupted right after its passage through first
derivation, becausethe pressuresupstream and downstream of the pig become equal; the pig then
stops between derivations. A pressure sensor installed in the production line detects the moment
at which oil starts flowng out of the well. At this moment, this sensor commands the closure of
the gas feeding valve. After certain adjusted time, that assuresthe pig arrival, the production valve
that was closed is opened, thus enabling the beginning of a new filling cycle of the two production
strings. After the amount of oil has reached an adequate accumulated value, the pig is launched
again, now in the opposite direction and the same operations has to be done in the opposite side.
 -I

LfNE PRESSURE
SENSOR PIG VALVE
TO
SEPARATOR

GAS VALVES

DERIVATIONS B

MASTER
VALVES

PIG

CASING

PRODUCTION
ZONE

STANDING VALVE

Figure 1.3 - Pig lift system in horizontal well

In the second method,, Lima (1996) proposes a process named gas pump,
in which oil is just
accumulated in the horizontal region of a subsea production flowline. The accumulated oil is
transported to platform by injecting gas at the inlet of the flowline, with or without pig. At this
point the oil coming from the wellhead is deviated to a parallel line or to a subseaseparator. So
the wellhead pressure is kept low all the time, which means high oil production. Fig. 1.4 shows
the subseagas pump system.

COMPRESSOR
\
SEPARATOR

Figure 1.4 - Gas pump system

In both methods after certain time of oil accumulation in the horizontal pipe, gas is injected in
order to transport the oil to the surface. A pig can be utillsed in order to increase the efficiency of
the oil transportation. A is
complex transient gas-liquid modelling required to simulate these two
methods and also for conventional gas-liquid pipeline pigging so common in the oil industry,
 6

1.3 Objective of this Work

There are no previous works on pipeline-riser multiphase pigging and gas pump. The evaluation
of existing tools is difficult because there are no experimental data sets made avallable to tile
public and due to the proprietary nature of these tools. The necessity of a theoretical and
experimental study on gas-liquid pipeline-nser transient flow and piggingyis obvious.

It is the aim of this project to develop a new two-phase pigging model and a lle-transient gas-
C
liquid flow model in pipeline-riser systems and horizontal oil wells. Special attention shall be paid
to flow pattern characterisation in order to avoid the discontinuities generated by the different set
A
of equations. computer programme shall be developed.

Once the computer programme is created, it is verv important to compare the obtained results
with experimental data in order to validate the new models.

Therefore the main targets within this work are:

Complete literature survey and review of the transient gas-fiquid flow and two-phase pI(Y()I in
pipes.
Creation of a new model to smoothly deal with the different gas-liquid flow patterns.
Creation of a pigging model simulating gas and liquid by-passes.
Collection of experimental data on transient gas-liquid flow in a pipeline-riser system with and
without pig.
Comparison of the results of the computer simulator with experimental data.
 7

CHAPTER 2. REVIEW OF RELEVANT LITERATURE

2.1 Flow Regimes and Two-phase Flow Models

During concurrent gas-liquid flow in pipes, a variety of flow patterns can exist. Each pattern
results from the particular manner by which the liquid and gas distribute in the pipe. Authors differ
in
somewhat the name they assign to each of the flow patterns. However, the differences are small
and most agree on four flow regimes- stratified, slug, annular and dispersed bubble as shown in
Fig. 2.1.
" Stratified- in this case the liquid flows at the bottom of the pipe with gas at the top.
" Slug- liquid by
slugs separated gas pockets move violently downstream.
" Annular. the gas flows in the centre of the pipe while liquid flows as an annular.
" Dispersed bubble: the gas is dispersed in the form of small bubbles within a continuous liquid
phase.

1.0

LIA
Dispersed Bubble 1 1

'"
" ""
". ". ""-

Slug

Stratified

Dispersed Slug Annular

Annular
Bubble

Horizontal Flow Vertical flow

Figure 2.1 Flow patterns in gas-liquid flow in pipes

-
 8

Theoretical developments for transient two-phase gas-liquid flow in pipes can be classified into
three categories. no-slip flow or homogeneous models, slip mixture flow or drift-flux models, and
separated flow or two-fluid models.

Homogeneous mixture models are too simplified and in general do not performs well in
comparison with experimental data.

The drift flux models are sometimes called diffusion models. The basic concept of this formulation
is to consider the mixture as a whole, rather than two phasesseparately. Tss hi iii obta ned by us ng
., from liquid linear
a mixture momentum equation that results the combination of the gas and
momentum equations. The mixture momentum equation does not contain the interfacial transfer
terms, because they cancelled out in the summation process. Some additional manipulations to
convert phase velocities into mixture and drift velocities are also done to express the mixture
velocity, the pressure, and the liquid hold-up as dependent variables. This formulation is simpler
than two-fluid models.

Two-fluid formulations are very complex. The equations describing the conservation of mass and
linear momentum equations for each phase are obtained by averaging the respective local
instantaneous partial differential equations over the phase sub-volume in a fixed control volume.
Several closure relationships are needed. These includes relationships for the shear stress at the
pipe wall and at the interface, the mass transfer rate between the phases, which usually depends
on the pressure and temperature. One of the problems that arises when using the two-fluid
formulation is to properly account for the momentum and mass transfer phenomena taking place
at the interface, mainly for flow patterns with a complex interfacial surface. The two-fluid
formulation can be successfully developed, however the computer codes are relatively large and
complex.

While the homogeneous models have been shown to be always well-posed as an initial-value
the drift flux and two-fluid models have been shown to sometimes result in ill-posed
problem,
initial-value problems and convergence is not attainable, as described in item 2.2.

The most relevant works on transient gas-liquid flow in pipes, pigging dynamics of two-phase
intermittent gas lift are reviewed in this section. Special attention is given to those
pipelines and
related to hydrocarbon transportation and production.
 9

2.2 Transient Gas-liquid Flow

A broad description of gas-liquid flow categories was given by Scoggins (1977), who was one of
the first investigators on this subject in the oil industry. Scoggins presented a relatively
comprehensive literature survey, after which he decided to use a drift flux formulation in his
model for horizontal two-phase transient flow. Homogeneous nuixture models were discarded as
being too simplified and for not performing well in comparison with experimental data. Txvo-fluld
models were discarded for the ill-posedness consideration of the equation sets (Lyczkowski et al.
1975), and for the lack of practical and reliable means to account for the flow regime dependent
-
interfacial friction and transient flow forces.

Scoggins determined the slippage between liquid and gas phasesthrough commonly accepted
steady-stateempirical liquid hold-up correlations.The conceptsof slip velocity and slip ratio, both
used in earlier works, were not considered. The Eaton (1967) and Dukler et a]. (1964)
correlations were used as closure relationships. Fluid physical properties and mass transfer
betweenphaseswere calculatedby the black-oil model approach.

The polynomial characteristic equation of the Scoggins model yielded all real-valued roots for a
broad range of operating conditions common to gas-oil two-phase flow pipeline operations. This
non-linear equation was solved using a finite difference method, and an implicit sequential
solution algorithm, which was based upon a Newton-Raphson iterative procedure. Finally, a
comparison between measured transient flow data and the prediction of the model tended to
validate the proposed formulation.

Taitel et al. (1978) developed a theory to predict flow pattern transition under transient conditions
two-fluid flow model equations. Comparison with experimental data was also presented. It
using
found that transient conditions, flow pattern transitions can take place at flow rates
was under
different from those occurring under steady-state conditions. It was also concluded
substantially
that certain unexpected 'spurious' flow with slugging would temporarily occur when the gas and
liquid flow rates were suddenly increased after the establishment of a steady-state flow. This
though the iMtial the final steady-state flow patterns were stratified for both
occurred even and
Taitel et al. also showed that if the flow rates were increased gradually, slugging
conditions.
would not have been observed.
 10

A theoretical and experimental work on two-phase transient flow in pipes was carried out by
Dutta-Roy (1982). He compared the formulations used by Scoggins (1977) and Taitel et al.
(1978), and concluded that the Scoggins formulation did not include all the interfacial terms in the
mixture momentum equation. The two-fluid model formulation used by Taitel el. al. (1978) was
coded and compared with the experimental results. Transients were created by increasing the floN
rates after steady-state condition was reached. The comparison showed that the Taitel et al.
transient flow pattern prediction method did not accurately predict the time period for the slug
formation,, but gave the same trends as the experimental data. The Scoggins transient model and
the two-fluid model for stratified flow were compared with the field data of Cunliffe (1978). The
results of the comparison show that the Scoggins formulation performed better than the two-fluid
model.

The use of a two-fluid model with the inclusion of the pressure differential term was attempted by
Sharma (1983). The inclusion of such a small scale flow property has been shown by several
researchers to improve the stability of the equation set (Banerjee and Chan - 1981, Roy and Ho -
1980). The analysis of the characteristic polynomial equation for the two-fluid equation set
showed that all characteristics were real in the range of parameters investigated, indicating that
the inclusion of the phase pressure difference indeed yielded a hyperbolic and well-posed set of
equations. The numerical results were consistent With other stratified transient flow formulations,
but the predictions were poorer than the ones obtained from the Scoggins drift flux model. No
significant difference in results between unequal and equal-phase pressure formulations was
found, although the last is known to yield an ill-posed set of equations.

Sharma (1985) proposed a transient slug flow model based on the coupling of an unequal phase
velocity and unequal phase pressure two-fluid model with the hydrodynamic slug flow model
developed by Dukler and Hubbard (1975). His method included averaging techniques for the slug
flow parameters in order to allow the use of the separatedflow model. The proposed formulation,
however, was not evaluated against any experimental transient slug flow data.

The well known and one of the few commercially available two-phase transient computational
codes, OLGA, resulted from a joint research program conducted by the Institute for Energy
Technology (IFE) and SINTEF in Norway (Bendiksen et al. - 1986,1991). This code has been
updated since 1983 and is now comprised of tens of thousands of code lines. It is
continuously
based on an 'extended two-fluid model', which assumesthe existence of three separate phases,
namely, gas, liquid film, and liquid droplets. Separate continuity equations are applied to each of
these phases, and two momentum equations are used- a combined equation for the gas and the
liquid droplets, and a separate equation for the liquid film. A mixture energy conservation is also
The flow patterns are grouped into two major categories, separated (stratified and
used. possible
distributed (dispersed bubble and slug). Equations for the interfacial terms and
annular) and
between the phase were given for each of these two categories. Switching between the
slippage
done the minimum slip concept, that is, the roots yielding the
two sets of equations is using
liquid hold-up were picked as the correct ones. Although transition criteria fo r
minimum
determining the flow pattern at a specific location and time were presented, this information was
and was not utilised in the transient calculations.
used as an indication only
 11

Other commercially availabletwo-phase transient codes are PLAC and TACITE. PLAC (AEA
Technology- 1996) was developedfrom the nuclearreactor code TRAC. The PLAC code solves
mass,momentum and energy equationsfor each phaseusing a one-dimensionalfinite difference
scheme. The SETS (Stability-Enhancing Two-Step) method, used in PLAC, is a semi-implicit
method which treats the convective terms implicitly. SETS is a two step method, consistingof a
basic step and a stabillsing step. The basic step is a semi-implicit equation set which provides
information about pressurewave propagation.The secondstep is thus addedas a stabilisiny"step.
it
and provides information about the propagationof density, energy and momentum. PLAC has
flow regime maps for vertical and horizontal pipes. The flow regime boundariesin vertical flow
are mainly based on void fraction: bubbly, plug, chum and annular. In horizontal flow the
transition from stratified flow to other flow regimesis determinedusing the method devisedby
Taitel & Dukler (1976), basedon gas velocity and the transition between slug flow and annular
flow are based simply on void fraction. Studies by Mahaffy (1982) showed that in some
circumstancesnumerical instabilitiescan arise and so the is
method stability enhancingrather than
totally stable.

TACITE (Pauchon et al. - 1993) has been developed under a joint researchprogram between IFP,
TOTAL and ELF AQUITANE. The TACITE code is based on numerical resolution of a drift flux
model. The time is
advancing scheme explicit.

Due to the proprietary nature of OLGA, PLAC and TACITE it is difficult to know the details of
these codes.

Taitel et al. (1989) presented a new simplified approach for modelling two-phase transient flow in
pipes. This model assumesthat the gas phase can be considered in quasi-steady condition. Thus,
the time dependent term in the gas continuity equation can be neglected. Local momentum
equilibrium between the gas and liquid phases is also assumed. In order to compensate for some
inaccuracies incurred in the simplification process, Minami (1991) used mechanistic models for
predicting flow pattern, the slippage between phase and the pressure drop. Minami performed an
program showing this simplified approach is physically sound for some
extensive experimental
flow conditions. However the quasi-steady state gas flow is
assumption considered a serious
in there is a considerable gas accumulation as proposed in this work.
restriction situations where

Vigneron et. al. (1995) carried out an experimental programme to acquire multiphase transient
data. Compansons were presented between the data and predictions with TUFFP simplified
(Minami 1991), PLAC and OLGA. The results show that further work should be done in
model -
have better Even in a simple 420 m horizontal loop, the models predictions
order to a prediction.
were not so good.
 12

2.3 Pigging

McDonald and Baker (1964) were probably the first investigators to present a study on piguing
They assumed a successive steady-state approach to model the phenomena. The pipeline under
the pigging operations was divided into four flowing zones. The front of each zone Xas moved at
every time-step based on a volumetric material balance. Using steady-state correlations a-\-era(-),
e
pressure drop and average liquid hold-up were calculated. The pig velocivy was determined
through a gas volumetric balance, assuming no gas leakage through the p1g. A pressure drop
correlation through the pig was also provided. The successivesteady-stateassumption is the main
weakness of McDonald and Baker pigging model. It fails to predict the hydrodynamic flox'N
behaviour after the delivery of the liquid slug and the pig into the downstream liquid handling
facility.

Barua (1982) pursued an attempt to improve the McDonald and Baker pigging model. He
proposed a procedure to model the liquid slug acceleration during its delivery into the separator.
He considered that the pig was moving at the gas phase velocity immediately behind it, and used
his own empirical correlation for predicting pressure drop across the pig. However, Barua did not
remove the main weakness of successivesteady-stateconditions.

Kohda et. al. (1988a and 1988b) proposed the first pigging model based on full two-phase
transient flow formulation. Their model includes both the Kohda et al. (1987) drift flux transient
code, which is based on the Scoggin's study, and a pigging model. The pigging model composed
of a correlation for the pressure drop across the pig, a correlation for liquid hold-up in the slug
zone, a correlation for the pigging efficiency as a function of the pig to pipe diameter ratio, a pig
velocity model, and a gas and liquid mass flow boundary condition applied to the slug front. The
resulting set of equations was solved numerically by finite difference method, using tv'o co-
ordinate systems, one fixed and the other adaptive. No detail was given on how the difference
equations were coupled and solved simultaneously. In the experimental part of the study, two
pigging test results were reported that were obtained from a 1436.5 m Iong, 105.3 mm diameter,
low pressure horizontal pipeline, using compressed air and water as the two-phase flow mixture.
The experimental data compared relatively well with the predicted values from the numerical
Other than the fact that the Kohda et al. pigging model is still based on a drift flux
simulator.
that it uses flow pattern independent steady-state liquid hold-up and pressure drop
model, and
correlations, no other deficiencies are apparent.

Minami (1991) developed a pigging model and coupled it with the Taitel simplified transient
An Eulerean-Lagrangean approach using a fixed and mo%l co-ordinate sN'stem is used.
model.
He used mechanistic models for predicting flow pattern, the slippage betkeen phases and the
pressure drop. Minaml performed an extensive experimental program showing, this simplified
However the quasi-steady state approach is not sultable for
approach is physically sound.
pigging due to the high accumulation of gas upstream the pig
pipeline-riser svstem gas pump and
 13

2.4 Intermittent Gas Lift

Although intermittent gas lift has been around for almost seventy years, it was not until the work
of Brown and Jessen (1962), that the basis for modem design was established. Increasingly
sophisticated modelling efforts have been realised in the last years (Schimidt et al - 1984,
Zimmerman -1980, and Liao -1991). The work of Liao deserves special attention, since it
represents a complete modelling effort of the whole intermittent cycle. The results of this model
were compared with the experimental data of Brown and Jessen (1962), Brill et al (1967) and
Neely et al (1974), and good matches were obtained. The results of White (1963) regarding the
gas bubble penetration of the liquid slug as being a natural constant of lift system under
consideration are highly encouraging.

Results imply that sophisticated modelling can successfully predict complicated flow behaviour,
provided that the appropriate closure relationships are used.

However all of these studies consider only vertical flow. So there is a need to investigate a similar
in
phenomena a pipeline-riser system or horizontal well.

2.5 Summary

The following conclusions can be made based on the Literature Review-

" The complexity of the gas-liquid transient flow has raised difficulties in the development of
easy-to-use and proven codes for the oil industry to design and operate pipelines under
transient conditions.
The drift-flux model requires the use of empirical correlations to account for the slippage
"
between the phases limiting the degree of confidence of this formulation. It also suffers from
ill-posedness problems.
The state gas flow assumed in the Taitel et al. simplified model is a serious
" quasi-steady
restriction for gas pump and pipeline-riser pigging.
The existing pigging models are not suitable for pipeline-riser systems.
"

The above considerations show that further studies should be done for predicting gas-liquid
transient flow and pigging in pipeline-riser systems. The objectives of the present work areto
collect experimental data on gas-liquid pigging, gas pump with and without pig, transient gas-
liquid flow in pipeline-flow and to develop to
a model predict these complex phenomena. The
be transient two-fluid model avoiding the complexity of highly non-linear
model should a
OLGA or PLAC and the simplicity of Taitel et al., drift flux or
momentum equations of
homogenousflow models.
 14

CHAPTER 3. TRANSIENTMODEL

The two-fluid formulation was assumed to be the more suitable Txo-fluid

in this work.
formulation equations taken from Sharma (1985) are shown below.

The area averaged mass conservation equations, neglecting the mass fluxes across the gas-liquid
interface, are given by:

* Gas continuity equation

(a 0(agpgvg)
9p9 +
64

* Liquid continuity equation

I(alp,
t;. ) 0(alplv,
+
C?
z

where a, and a, are the volumetric fractions, ygand ijare velocities,p,,,.andp, are densitiesand I
and.z.are time and space,respectively.

The area averaged momentum equations neglecting the momentum transfer across the gas-liquid
interface are given by-

Gas momentum equation

d(a,, 2)
(" pgi,
+ +a o G, +r+r ig
=-a 999 ll.g (3)
C cz

* Liquid momentum equation

(al pf "I) (a' P1

+ a, al pi C, +r+ Z'
(4)

G- body force in -- direction and 'rig Til are shear forces per unit volume
where is - -r, g
rather than shear stresses.
 15

3.1 Proposed Transient Model

Taitel et al. (1989) suggested the use of standard flow pattern dependent steady state model,
neglecting the two terms on the left hand 'I
side of momentum equations and -4. However, for
rapid transient events, the inertia terms could not be ignored. As the liquid density is order of
magnitude higher than gas, it is proposed that only the inertia of liquid is included in the model,
Rather than solving the complete equations, the effect of inertia of liquid is taken into account as
a pseudo shear force.

The equation (4) can be written as:

2)
(Miv o(m, v1

0
1) --A, c'p - A, pIg sl'iiO,(' + ri Pi (I- - P,,,,(1- (5)
Cz

where Ti and 'r ,,I are shear stresses, Pi' and P, are the penmeters.
I

The equation (5) above can be rewritten as*

A, t3p+ A, pIg, 5lnO(! --ri Pic?.-+ (I+ (6)

where
/ (M
(m
c3 v v
K+ (7)
cz

For gas, K is taken as zero.

The horizontal (- io )flow pattern is determi by Beggs and Bn *11(1973) criteria. If

c-
-) ,Io0
flow tried. If liquid hold-up is smaller than 0- 1 annular flow is assumed and for
1oo slug is
liquid hold-up greater than 0.9 bubble flow is assumed. Otherwise slug flow is assumed. The slug
flow is treated as a combination of other flow pattems as described in 3.1 4. This new slug floN\
.
approach assuresa smooth two-phase flow structure transition.

3.1.1 Shear Coefricients

The shear coefficient is given by-

0.5 (8)

f is Fanning f 6ction factor and P is the perimeter. The shear stressesare given bv 7-..
where i tile ,, //
O'--VO/ "g-l',, 2.
v/),/) 2 andT-j -f
 I
()

The wall shear coefficients for the gas and liquid phase are calculated from ffiction factors gi\ en
by the Swarnee et al. (1976) explicit equation for turbulent flow (Reynolds number greater than
4000) as follows-

1.325
fw 2
5.74 (9)
In .6
0.9
3.7 Dh Re

where e is the absolute pipe roughness, Dh = 4AFIP and Re = PI'Dh /'U. P and AF are the
perimeter and area of flow, respectively. For annular flow the hydraulic diameter Dh is the liquid
film thickness (5.

If Reynolds number is less than 2000, the friction factor is 64/Re. Between 2000 and 4000 the
friction factor is determined by linear interpolation.

The gas-liquid interface friction factor is calculated by the Andritsos and Hanratty (1987)
correlation for stratified flow given by

fi f., (for v :!- v0)

= S9 S9
and
h VS9
f, (fo rv>v
= 1+15 -I
g
D0 S9 S9
S9

v0+5,
Fjatnr
where, sg in the Sl unit system (m/s).
ppp ,,

by
The correlation recommended Wallis (1969) is used for the calculation of the interface ffiction
factor for annular flow as follows-

fi 0.0 05[ 1+300 (12)

D

where ( is liquid film thickness.

 I-

3.1.2 Annular Flow

* Gas momentum equation

I dp ci I
4- (v -vi v9 -vi +gSillo
dz 9
pg a9

* Liquid momentumequation

I dp C" vl 109 I+
+(I + K) cl. (11
9V9-v, g sinO =0
dz (I -a ag )PI
,o, gI

e Gas continuity equation

(oD, ) A) ]
19 (9[(agog vg

* Liquid continuity equation

]
('[(I - ag)p,
A-

is
where p the pressure, c is the shear coefficient, g is the acceleration caused by gravity, 0 is the
pipe inclination angle and A is the pipe cross sectional area.

3.1.3 Stratified Flow

In the case of stratified flow the term vgvgshould be added to equation 3).
ag

3.1.4 Bubble or Mist Flow

For bubble or mist flow it is assumedthat there is no slippage between gas and liquid phases. O
the homogeneous model equations can be used.
 is

9 Mixture momentum equation

I dp
+ (I + K)cv. IvI + gsinE)
dz
PM

where p,, is the mixture density, c, is the mixture shear coefficient and i,, the mixture velocity.
M

9 Gas continuity equation

0(a, p, ) c3[(agpA) v
A 11
a 02
Liquid continuity equation

]
ag)p, A] v,
a a92

3.1.5 Slug Flow

The slug flow is divided into two regions. Dispersed bubble and stratified flow for horizontal pipe
flow. Otherwise a dispersed bubble and annular combination is assumed.Fig. 3.1 shows the slug
flow structure. This approach smoothly deals with flow pattern transitions avoiding the common
solution difficulties of the previous works on transient flow and pigging.
_gas-liquid

If
: 44 01.4 0:

annular if
bubble: stratified:

bubble

Horizontal Flow

Vertical flow

FiLyure3.1 - Sluty flow structure

 19

9 Gas momentum equation in the film zone

dp wgf Vgf if If
+C v +c (v
- Vjf vgf + [g sit?E)] (20)
gf 9f f =0
Pg dz f agf agf
II

9 Liquid momentumequationin the film zone

I dp (I + k)c-,,, P9
V#,.Vif - Cif - (VO -vu-)v v - +[gs-inE)l -/=0 (2 1)
&f+- 0-
PI (I -allr) (I - a,,, )p,
LL -J L -i
* Momentum equationin the bubblezone

I dp (I + K)cws Ivs I+g Is

+ vs sinE) =0 (22)
ps dz als
s

where v, = qrv,vgf + (I %f) vif and c,,gf is zero for annular flow.
-

The variables a,,, vj, v,,,.,1, and If are determined based on the steady-state slug model based on
Taitel et. al. (1989) as shown in Appendix 1.

The continuity equations are the same as those for bubble, annular or stratified flow.
 1()

3.2 Model Solution

The simplified continuity and momentum equations are solved by a two step sermi-Implicitmethod
(Mahaffy - 1982) using the finite difference equations given below. The basic step is a simply
semi-implicit equation set. A stabilising step is added to provide information about density and
momentum being transported across cell boundaries. The spatial mesh used is staggered, with
thermodynamic properties evaluated at the cell centres and the velocity evaluated at the cell edges
as shown in Fig. 3.2.

Lzi-I 404
LZJ LZI +

V9 V9

J- I
VI Vi 0

paap

J- J+

Figure 3.2 - Spatial mesh

The finite difference divergence operator is

((XV)j+112 )/A.
VAXV) = -(XV)j-, 12 -j (23)

To improve stability, the flux terms at cell edgesuse donor cell averagesof the form
(XV)j+112 'f
= Xj Vj+112 Vj+1/2O

=X j+I vj+1/2 'f vl'+] 2<

03- (24)

where x is any group of state variables.

The finite difference equations for momentum and continuity equations are given below.

3.2.1 Annular Flow

Basic Equations

* Gas momentum equation

n
cl [ n+l n+l
(pOn+ll + vn -vn 1 2(v (Vn _ ,n+g sinO =0
_ pOn+l 9191-v (25)
j+l i. (a n9

9).
j+ /+,
22
 21

* Liquid momentum equation

On+11 On+l (I +K) cvn 1[2

(pj pi wi n n+l n]
(p nI Az +I - vi -v I
i ag)"
j+
22
2

n
P9
n n nl[2 n+l n+l) n+ n)
-C, V9 _ V, V9 V, _( V9 V, gsjnE) =0 (2())
ag)p,
2

9 Gas continuity equation

On+l
(a pg)
A
g - 9p9 On+I n+l 0
+VIJA pna
999
v (27)

9 Liquid continuity equation

]On+l
ag)p, - [(1 -
ag)pi]
[Ao"(I On+l n+l 0
Vj. 191 -a)
v
(28)
3

where the 0
superscript is the value in the basic step of SETS is
method, n the time index and j is
the cell index.

Stabilising Equations

* Gas continuity equation

)n
(a g pg
Vj. [(agpg A) n+l n+l ]=0
A + v9 (29)
A/

o Liquid continuity equation

a, )p, n-I

+ vj. 0 (30)
At
3.2.2 Stratified Flow

n
cug
In the case of stratified flow the term a,,)j+ -ill"1(21,17+1
99g -1,17) be
should added to equation (25).

3.2.3 Bubble or Mist Flow

For bubble flow it is assumed that there is no slippage between gas and liquid phases. So the
homogeneous model equations can be used.

Basic Equations

Mixture momentum equation

017+11 On+l [2v"+'

(Pj+l )+ (I + K)c" In v" +gs'nE)
I =0
-p
_j in in in - in] (-)1)
+2 j+
2

* Gas continuity equation

On+l n
(agpg (agpg
- na On+l n+l
A +V Ap v 0 (2)
199F
At

9 Liquid continuity equation

On+l ]n
ag ), 0,
g)pll jII [Ao n On+l n+l
-- V i19 (1-a ) v (3)3)
"1
At
.

Stabilising Equations
Gas continuity equation

)n
PIA-(a [(a, n+
04)
A 9p9 +V p, A) 1nt
Ai
 * Liquid continuity equation

n+l
ag)p,
1[(I-ag)pA] n-I
n- 10
+V
JI ,, (3)5)

3.2.4 Slug Flow

Momentum equations
I. (Pjil
",,n-i 11 ln+l

Az
-pj )
'o g ,
j+-

cnn Ivgn 1[2 Cif

Vgf
n+l n
Vgf
nn
Vgf
1[2( n4l n4l) (,,, n n ]
VY- Vgf V
n gf if _ 1, +[gsine]
(ae) (a )n e r
J+- J+-

LL- 22

n
cn
lp, ws l[2 n+I
vn VS
(als )n1s
_vn+
s11 gsinG
pg
j+,
(36)
2

On+l On+l
(pj+l pi
(pl)j+,
Az
_
2 1+

+n n
cif
[2 n+l j[20"'- n+l n
v". _n(,
g)nJ+
I-
(P, )n )n -V" y
v
g(
i-at) 2
1(1_a ,,,

L22
-J

(I +, k)Cn
',
21,. " 1-v n+l +gsine (37)
(aL, ) nss

By means of the above momentum equations and steady-state slug flow model of Taltel et al.
(1989) described in Appendix I it is possible to determine the phase velocities as a function of
pressure. The continuity equations of basic and stabillsing steps are treated in the same vvm-as for
annular or stratified flow. The term c,,,, is zero for annular flow.
gf
 24

3.2.5 Linearization of the Finite Difference Equations

The set of finite difference equations is non-linear and hence to solve them at each time step it is
to
necessary use an iterative method, based on Newton's Method. Starting with some estimated
value for the independent variables at the new time step, the derivatives of the equations -Ith
respect to those variables are used to give the next best estimates-basedon linear extrapolation
from the last value - continuing until the latest estimates are equal (within prescribed tolerance) to
the previous ones.

At a time step n+l, given an initial guess of the independent variables (in this casep and a), the
values on the next iteration (variables without primes) are assumed to be related to those at the
last (with primes) by the relations:

nA +OP0-- (3 8)
p

n+l
a,99 +(5a (39)

Since the finite difference equations are functions of p and ag, a Taylor expansion about the last
iteration's value, retaining only the terms linear in 6p and 6o gives:

(p, a, ) f (p + t5p,ag + 5a, )

0,6a, (40)
t3 6p + higher
f (p', a') + + order terms
9 C)9 da
P--
P, 9 g=ag

is
In practice, an expansion performed only on the gas and liquid continuity equations. In the basic
Vn+1 are calculated fTom momentum equation as a function of the dfferent, al
equations v"" 91 and
the last iteration. In the stabilising step the v"' and v"" are kept constant and final
pressure of 9
value of p and ag for the new time step are obtained.

Basic Step
From the momentum equation phasevelocities, as shown is Appendix J, be
can written as-

vnf
OnA
11)
'In
+F (41)
g( Pil 19 -
9
(POn 0?
nil
E"I
, 41 11)+
(42)
11
1= i+l A
 -)i

9 Gas Phase
(agpg )j+1/2 (agpgA
On+l A vg i
(Aa g )n
0 pg) _ (Aagpg g)j_ll_
fg' (p, ag )=-
- (431)
At AZ
-.Jj

[pg, On+l (pgAvg)on+l '01'1+1

0 4]
3fg -(pgAvg
j+1/2 j-i/2
+
(44)
Oag A/ Az

On+l
Apg
a9 A (agpgA 0?
1+1 (agpgA 011+1
E(pj+j pj ))j Eg (p,
fg, Ap - - 1-1))j_ll
+1 /2
-j .
(45)
A A

9 Liquid Phase
0"+l
)j-H/2 o"+1
)j-i/2)
On+l n (1-ag), o,Av, -((1-ag)pAv,
(A(I - ag)pl) ag)pl)
0(p,
ag) (46)
A/ Az

On+l
On+l 011+1
0 [-piAl
Of 1-i- j+1/2 j-i/2
(47)
Oag At Az

-On+l
Ap,
ag) o"t"I
A El (pjl - pj) -a g)pl AF (pj - pj
Ap -a g)pl

A Az
rp
 26

Stabiliser Step
9 Gas Phase
((agpgA n+l
)j+112 (avg)J ii+1
(Aagpg) n+l vg -

-(Aa gp
At (49)
Az

n+l (pgA n+l n+l

Apgjj vg A
- pg vg)
4r% j+112 j-1/2
(50)
0 ag At Az

( n+l
)j+1/2 n+l
)j-1,2,1
Apg Ap An
ag A gAv KAv
a a7
Ap 9 Ap 9 9 Ap 9
Of
,
(51)
Op A/ Az

o Liquid Phase
n+l
)j+1/2 n+l
)j-]/, )
ag),iA v, A
ag),c v,
(A(I (Al
- ag),q)t'+1 - - ag),q)" )
(p, ag)
(52)
At Az

n+l
(-pl ) n+l
-Ap,
A
-pl v, - A v,
j+1/2 j-1/2
(53)
t-ag A/ Az

- FIA )j-1
Ap1 Ap,
(]-a) a )API v, ag) A
Ap AP
Ap j+l 2 2
1-
(54)
At
 Performing the differentiation indicated in equation (40) in the continuity equation leads to the
following linear system of equations-

CIT9
1 4,91 cygi Of9i Ig, (p, a,

Op 2 ia2
Cf,I 6a, (p,
-a,
Arl 0'2 a2
071 I
07-X CTg Cyg 2 Org (P-2.
2 2 2 -fg2 a-'
(51),
_Cfg2 _Cfg2
tP3 'r
C
2 t9a2 3

(TI 2 Cy 12 oyl 2 Cyl 2 of, 2 (P2. a-,

_OY-12
t5a2 -f12
41
C71 C-4
CP 2 C 3 da3
(Tg 3 oyg Cyg Cfg (Pl', )
3 3 3 45P3 -fg3 'Cf-i
_Ofg3 _CTg3
0'P2 0't2 0'P3 olC3 C' Ca4
4

Cyl 3 O)rl 3 Cyl 3 oyl 3 Cyl 3 ct3)

(5a3 -fl3(p3,
_C-Y-13
t'P 2 CJ93 C't3 cp 04
-a2 4

ffgj-i C-Tgj- cjgj-, 6T9 CT9 )I

-fg, l. aj
-(fgj-,
j-
45pj -I(p,
Arj-1 terj c
631j-2 '51aj-2 (jqj-1 j
cr
(Tli-l CY, (3 ga -flj- I (pj aj )I
li-i J-1 -.
(Pi-I e-P c9a
I L-----i
Tgj &P (pj
oyg j CY9 -f
gj aj )I
j ,
-,
C)?i-l c6ti da
I
cp j
of, i oy-I CY, CY, 6a (p,. a_, ) I
j -fl,

cpj cia
or
DU y R
LDU y R
LDU R

LDU
LDU R
LD R

where,

LI D and U are block sub-matrices on the diagonals of the Jacobian trIdiagonal matrix. Y is the
solution vector containing the increment values (6p and 6odfor each cell and R is the vector
containing the independent terms.

3.2.6 Inertial Force Adjustment

The liquid momentum equations contain the inertial force adjustment term, K, which is lumped
together with the shear forces. The value of K for the present time step is estimated from the
previous time step by the following equation-

n
(M pi- I'll
n-
ttln ,n
(cls. 0.
K= / v, 11
lAp, 1 (55)
Af Az

2
 29

3.2.7 Two-phase Transient Model Flow Chart

The solution strategy used in the code is shown in the Fig. 3.3.

Initial Conditions
4,
vg and v, from momentum equations as function of Ap

41
Taylor expansion of basic continuity equ
I
Solution of the block trigiagonal syst
I

P: --P+(49
No
INO-
< paTdt5ag:! -to ler
;:::7 er
ag
-a g
I Yes

Final value of vg and v, and preliminary value of pando,

I
Taylor expansion of stabilising continuity equations

41
Solution of the block trigiagonal system
4
New time step
No
toler NoIT
+ -;: j-
4-
ag ::::ag
I Yes

Final value of p and a time ! final iinie

Yes

Figure 3.3 - Code flow chart

 30

CHAPTER 4. EXPERIMENTAL FACILITIES

An extensiveexperimentalprogram hasbeen carried out to acquiretwo-phasetransientflow, gas

pump and pigging data on a 67 m long, 0.0525 m diameter,9.9 m high pipeline-risersystem.Fig-
4.1 shows a schematic diagram of the experimental facility. Description of the various
componentsof the system follows.

4.1 Air and Water System

Air was supplied by two compressors CI and C2. Compressor CI is a screw compressor rated at
0.0717 M3/s at 8.5 Bara. Compressor C2 is a reciprocating compressor rated at 0.1887 at 19
Bara. The compressors CI and C2 were connected to the buffer vessel V2. Upstream from the
mixing section the compressed air can pass through two turbine meters. FG I and/or FG2. Beside
these flowmeters a thermocouple was wired into the data acquisition system to provide the
temperature signal.

Water was pumped from a storage tank by pump P 1. Pump PI is a progressive cavity pump with
0.00972 M3/Scapacity and its maximum discharge pressure is 7 Bara. The water phase can be
measured by two magnetic flowmeters. FLI and/or FL2. The electrical signal from the turbine
meter was sent to the analogue-to-digital (A/D) converter of the data acquisition system.

Before the test section air and water are mixed through a mixing section.

After the test section, the mixture flows into a separator SI at the top of the riser. At the upper
part of the top separator there is a gas line and at the lower part there is a liquid line. These two
lines merge into a two-phase line that connects the separator SI to the big horizontal separator S2
on the floor. The differential pressure signal from the upper to the lower part of the separator SI
was sent to the level control valve LCVI installed in the liquid line. A turbine flowmeter FG3, a
pressure transducer and a thermocouple was installed in the gas line. At the top of the floor
separator S2 there was a line to pernut venting of air to the atmosphere through a back pressure
valve and there was also a liquid level control valve LCV2. The separated liquid was sent from S2
to water coalescer vessel WC1 and the to its storage tank WTI. The main specifications of the
equipmentand vessels are shown in Table 4. land 4.2.
 31

IN

a., 99
1":

1-4
i- ,
V-4 W

4.1 Schematic diagram of the experimental facility

Figure -
 32

4.2 Test Section

A schematicdiagram of the flow loop with all dimensionsand the locations of the instrumentsis
presentedin Fig. 4.2 and Fig 4.3. The pipeline-riserline of the test loop is a 0.0525 m diameter,
69 m long steelpipe.

The bulk of the instruments for the experimental study were located in the test section. The test
section instrumentation consists of 5 pressure transducers, I thermocouple at the inlet and I
gamma densitometer at the bottom of the riser. Pressure at the gas flowmeter section and at the
separator were measured by transuducers. The flowrates from the 2 turbine meters, the air
flowing temperature, and the 2 magnetic flowmeters complete the set of signals to be logged.

All the analogue signals from the transducers were converted to 0-5 volt outputs, which were
then wired to the data acquisition system.

A total of 21 analogue signals were captured by the data acquisition system, comprising-
-9 pressure transducer
turbine flowmeters (air)
-3
flowmeters (water)
-2 magnetic
-4 thermocouples
gamma densitometer
-I
-I differential pressure transducer

The software was written using LabVIIEW 4.0 and was run on a P166 PC computer. The
analogue signals were sampled using a signal conditioning extensions for instrumentation system
(SCXI) and passedto the parallel port of the microcomputer.

The data was sampled at 10 samples/sec for each channel. Data was collected to the hard disk
whilst viewing the signals on screen. The main elementsof the system are shown in the Fig. 4.4.

4.3 Pig Specification

A 60 mm diameter ball pig, made of 240 kg/m' density polyurethane foam was used in the
experimentalprogramme.
 33

Cl,
'

Figure 4.2 - Test section dimensions

 34

71

(A
cz

u m u
Ell
1- 73

L.T.

--1

1--11 110
"-

.1- CD

,' CL.

I-
k,r)

00 IND
C)

N
N

00

Test of the instruments

Figure 4.3 - section position
 35

P166
SCM

parallel port cable 000000

A/D card 000000
(ATMI016E-10) 000000
000

21

from sensors

Figure 4.4 - Data Acquisition System

 36

Table 4.1 - Equipment

COMPRESSORS

ITEM TYPE FLOW RATE DISCHARGE MAXIMUM

CAPACITY PRESSURE PRESSURE

Cl screw 0.0717 M3/S 8.5 Bara 8.5 Bara

C2 reciprocating 0.1887 M3/S 19 Bara 19 Bara

PUMP

ITEM TYPE FLOW RATE DISCHARGE MAXIMUM

CAPACITY PRESSURE PRESSURE

Pi moyneau (PD) 0.00972 ni'/s 4 Bara 7 Bara

FLOWMETERS

ITEM TYPE FLOW RATE MAXIMUM

CAPACITY PRESSURE

FGI turbine 0.00022 - 0.0022 M3/S 20 Bara

3/S
FG2 turbine 0.0025 - 0.025 111 25 Bara

FG3 turbine 0.00139 - 0.0167 M3/S 25 Bara

FLI magnetic 0.00011 - 0.00167 n, 3/S 12 Bara

FL2 magnetic 0.0008-3 - 0.0 125 M3/S 12 Ba

 Table 4.2 - Vessels

TWO-PHASE SEPARATORS

ITEM POSITION DIAMETER VOLUME MAXIMUM PRESSURE

(m) (m) (Bara)

si UPSTAIRS/ 0.5 0.33

VERTICAL

S2 FLOOR/ 1.5 12
HORIZONTAL

BUFFER VESSELS

ITEM POSITION DIAMETER VOLUME MAXIMUM PRESSURE

(m) (M)
(Bara)

V2 FLOOR/ 1.7 2.76 26.

HORIZONTAL

WATER SYSTEM

ITEM TYPE MAXIMUM DIMENSIONS

PRESSURE
(Bar)
DIAMETER HEIGHT
wcI COELESCER 25 0.915
-3.277
VESSEL
HEIGHT WIDTH LENGTH
WTI STORAGE TANK atni 1.68 2.2 2.75
(10 M)
i
 38

CHAPTER 5. PIGGING MODEL

To simulate the pig motion in the pipe, the pipeline is divided into two sections as shown is Fig.
5.1. The first section is ftom the inlet to the pig, and the second ftom the pig to the outlet of the
pipeline. The velocity of the pig is given by the velocity of the mixture pushingthe pig in the
previous time step. The pig is assumedto run one cell in each pig time step. As the mass of the
cell in front of the pig is known and also the time to run it, we know the mass flow rate entering
the section in front of the pig. This mass flow rate will be the boundary condition for the transient
calculation from the pig to the outlet. This transient calculation gives the pressure in the pig which
is the boundary condition for the transient calculation ftom the inlet to the pig. Thus the transient
calculation throughout the pipeline in this pig time step is finished. This procedure is repeated
until the pig reaches the end of the pipeline. After that the normal transient calculation can be
done. The pigging model described above makes it unnecessaryto track the position of the slug
front. The pressure drop across the pig is neglected.

UPSTREAM PIG DOKWSTREAM

PIPELINE CELL PIPELINE

------------------------------------------------------- -----------------------------------------------------

Figure 5.1 - Pigging model

 39

5.1 Pig By-pass

The pig by-pass introduced in the pigging model comes from the runs of gas pump with pip. In
these runs the test section is filled with water and followed by pig and gas injection In the Inlet.
Experimental results, shown in Appendix C, indicate that there is gas by-pass in the same
direction of pig motion and that liquid is left behind the pig. The gas by-pass was evaluated ftom
the gamma densitometer signals shown in Fig. 5.2.

Figure 5.2 - Density at the bottom of the riser

 40

The gas by-pass is evaluated by comparing the gas mass in front of the pig to the total mass of gas
injected up to the time when the pig reach the gamma densitometer.

The liquid by-pass (liquid left behind the pig) is evaluated by draining the system after the plo
arrival at the receiver and comparing the measuredliquid to the total volume of the test section.

Table 5.1 presents the gas by-pass and the liquid in the test section for the gas pump runs with pig
as a function of the average pig velocity from the inlet to the densitometer.

Table 5.1 - Pig By-pass

RUN AVERAGE GAS BY-PASS LIQUID LEFT IN

CODE PIG VELOCITY M THE TEST SECTION
(m/s) M

LPPIA 1.52 18.3 2.46

LGPP2 1.85 11.1 237

LGPP3 2.00 8.2 1.1

37

The relationship between liquid left behind and average pig velocity was not clear as the gas by-
pass. A liquid by-pass of 2.5% was adopted in the model to take into account for leakage through
the pig. This value is based on the highest value (2.46%) of liquid left in the test section as shown
in table 5.1. Fig. 5.3 show how gas by-pass is affected by the average pig velocity and the
correlation used in the pigging model.

2.4

Gas by-pass (%) = 50 - 21),

2 p
.2
I lp("I/S) >0
0< Gas by-pass (%) < 50
2.0 +

0 +
1
.8

CM 1-6
CL
+

1A

1 20 25
.2 0 5 10 15

By-pass (%)

Figure 5.3 - Pig By-pass

 41

5.2 Pigging Model with By-pass

The velocity of the pig is 91 iven by the corrected velocity of the mixture pushing the pilij. This
velocity is calculated considering the flow rate reduction due to the gas by-pass (Fig. 5.31)and the
reduction of the area (2.5%) due to liquid left behind the pig as a film. The p1g), is assumedto run
one cell in each pig time step. The mass of gas in the cell in front of pig should be added hy the
by-pass mass of gas. The mass of liquid should be reduced by the mass of liquid left behind the
pig. As the mass of the cell in front of the pig is known and also the time to run it, % e knox the
mass flow rate entering the section in front of the pig. This mass flow rate will be the boundar).
condition for the transient calculation from the pig to the outlet. This transient calculation gives
the pressure at the outlet of the pig cell. The initial condition of the pig cell is assumed to be
annular steady flow due to the film left behind the pig. This steady state calculation gyi'v-es the
pressure at the inlet of the pig cell which will be the boundary condition for the transient
calculation from the inlet of the pipeline to the inlet of the pig cell. The transient model calculation
in the pig cell gives a new inlet pressure which will be the new boundary condition for the
transient calculation from the inlet of the pipeline to the inlet of the pig cell. This process is
repeated until pressure convergence is obtained. Thus the transient calculation throughout the
entire pipeline in this pig time step is finished. This is
procedure repeated until the pig reachesthe
end of the pipeline. Fig. 5.4 shows the pigging model flow chart.
 42

itial Conditions

vp = v. at the inlet of the cell

I
time to run the cell (pig time
A
liquid and gas mass flow rate at the outlet of the cell
I
transient model from the outlet of the cell to the end
of the pipeline by the pig time step

I pressure at the outlet of the cell after the pig time step I

assume steady state annular flow in the cell

pressure at the inlet of the cell

transient model from the inlet of the pipeline

to the inlet of the cell hy the pig time step
; T-
liquid and gas mass flow rate at the inlet of the cell

new pressure from the transient model in the pig cell

I dif = new inlet pressure previous inlet pressure I

-

VP for Ihe next pig time step time

Yes
;
---7<
dif < toler
final tirne, -'
v. at the outlet of the -,
cell (vp = v. ) LsI vc
Endi_____)

Figure 5.4 - Pigging

modeinowchart
 43

CHAPTER 6. RESULTS AND

-DISCUSSIONS
In this chapterthe transient data collected in the experimentalprogrammeare comparedwith the
model predictions. Severaltypes of transientflow behaviourare observedfor the different runs,
I. ConventionalPigging- a pig is launchedfrom the inlet of the test sectionafter the establishment
of a steady-stategas-liquid flow.
2. Gas Pump without pig- the test section is filled with water and followed by injection of gas
from the inlet.
3. Gas Pump with pig: the test section is filled with water and followed by pig and gas injection
ftom the inlet.
4. TransientFlow: changeof boundaryconditionssuchas inlet gasflow rate, inlet liquid flow rate
and/or outlet pressure.

A total of 32 experimental runs were carried out summary of the expenmental.results of each
-A
run can be found in Appendixes A, B, C and D. The Appendixes E, F, G and H show the model
results of these runs.

The transient flow parameters of interest to the pipeline and oil industry, such as pressure, liquid
hold-up, and flow rates are analysed and discussed based on the experimental data and the
prediction of the model. In some tests the gas flow rate at the outlet of the is
separator greater
than the maximum value measuredby the flowmeter (0.0 167 m'/s). A change of Im in the water
level of the separator corresponds to 5 mV.

The use of a fast computer data acquisition system produced experimental transient data sets of
unique characteristics. Very detailed information of the flow structure can be determined from
these data sets. This includes, for instance, flow pattern, slug frequency, slug hold-up, film height
and pig velocity. The sampling rate of the data sets are 10 samplesper second per each channel.

6.1 Conventional Pigging

Table 6.1 presents the experimental conditions and comparison between the experimental data and
the model prediction.

A total of 13 experimental conventional pigging runs are carried out. A summary of the
experimental results of each run can be found in Appendix A. The Appendix E shows the model
results of theses runs. Once the pig is launched the separator pressure tends to decrease because
the pig starts to block to
the gas and generate the liquid slug in its front. The inlet pressure does
not change too much while the liquid slug front is in the near horizontal pipe. As soon the liquid
slug front the riser the inlet pressure starts to increase due to the head exerted by the
reaches
liquid in the riser. This period of gas accumulation means low pig velocity. When the back of the
liquid slug reaches the riser, the head reduces dramatically due to gas phase in the riser. At this
high flow rate of liquid is observed in the separator followed by high
point the pig accelerates and
flow rate of gas, causing a sharp increase and decreaseof the separator pressure. .-\fter some time
the system reaches a stable condition again.
 44

Table 6.1 - Conventional pigging experimental data and model results

INITIAL PIG TRAVEL

RUN SUP. GAS SUP. LIQ. INLET PRESSURE TIME
CODE VELOCITY VELOCITY Exp. Model Exp. Model
(m/s) (m/s) (Bara) (Bara) (s) (s)

LPG3WI 2.6 0.16 2.3 2.3 48 40.6

LPG4WI 6.2 0.16 2.4 2.6 25 26.6

LPG3W2 2.2 0.48 2.7 2.5 42 34.0

LPG4W2 5.9 0.45 3.2 2.9 18 17.0

LPG3W3 1.7 0.81 2.8 2.8 39 33.0

LPG4W3 4.0 0.77 3.1 3.2 23 21.0

HPG2W] 0.8 0.27 4.7 4.6 79 78.0

HGP3W1 2.8 0.16 5.6 5.7 34 32.6

HPG4WI 5.9 0.24 5.8 6.0 16 18.0

HPG4W2 5.4 0.34 6.2 6.2 18 20.4

HPG2W3 0.5 0.99 6.2 6.1 40 44.0

HPG3W3 2.5 0.94 6.3 6.3 27 25.3

HPG4W3 6.4 0.67 6.9 6.8 14 13.6

Fig. 6.1 and Fgi. 6.2 show the predicted pig travel time and the initial inlet pressure respectively.
In these diagrams no special trends of the error can be detected. Good agreement is observed.
 45

50.0

40.5

cu
L-
F-
31.0
.T
CL

0
2 21.5

12.0 "-
12.0 21.5 31.0 40.5 50.0
Exp. Pig Travel Time (s)

Figure 6.1 - Measured and predicted time between pig launching and receiving

w
U) 6
U)
w
0
-5
C

C5

G)
"0
0

2
2 45
Exp. Initial Inlet Pressure (Bara)

Figure 6.2 - Measured and predicted initial inlet pressure

 46

Experimental runs LPG4WI, LPG3W-'), RPG4WI, and RPG3W3 are wsed to analvse
conventional pigging case.

Figures 6.3,6.5,6.7 and 6.9 show the results of the model simulation of bottom nser densitv
along with the experimental measurements. All these figures show a high hold-up for a certain
time. This time is proportional to the slug length in front of the pig. After the pIcy.a very low
liquid hold-up is observed. This means that the pig is mainly pushed by gas. After certain time,
liquid start flowing again at the bottom of the riser.

Figures 6.4,6.6,6.8 and 6.10 show the results of the model simulation of inlet pressure along
with the experimental measurements. All these figures show an increase of the inlet pressure
followed by a sharp decrease.After some time the pressure stabIlises.

6.1.1 Run LPG4WI

The initial and final flow conditions are in stratified flow pattern. The initial inlet gas and liquid
superficial velocities are 6.2 m/s and 0.16 m/s, respectively. The inlet pressure is 2.4 Bara. The pig
is launched at 458 s. Fig. 6.3 and Fig. 6.4 show the transient behaviour of the bottom riser density
and the inlet pressure during and after the pigging run.

The predicted steady state inlet pressure is higher than observed. The measured density at the
riser bottom is higher than predicted. The time and the maximum value of inlet pressure are very
well predicted. The model predicts a faster recovery of pressure, but the time interval of the low
pressureafter the is
arrival of pig well predicted. The time interval of the liquid front to reach the
gamma densitometer after the passageof the pig is also well predicted. The measured slug length
is smaller than predicted, however the total volume of liquid in the slug shows good agreement.
The predictions of the model can be considered good.
 4-

pig launching time 458s I

LPG4NN'l
000

0
4.0
0 300
m

moo

450 470 550 570 590

im e (s)

Figure 6.3 - Measured values and model prediction for the riser bottom density with time

ILPG4 -WV7
l

r
ft

a- 2.4
4.1
0)
2.2

2.0 Experimental

450 470 490 510 530 550 570 590

Time (s)

Figure 6.4 - Measured values and model prediction for the inlet pressure with time
 48

6.1.2 Run LPG3W3

Slug flow is observed for run LPG3W3. The initial gas and liquid superficial velocities are 1.7 n-Ys
and 0.81 m/s, respectively. The inlet pressure is 2.8 Bara. The pig is launched at 193s. Fig. 6.5
and Fig. 6.6 show the transient behaviour of the bottom riser density and inlet pressure during and
after the pigging run.

The pigging model performed well. The predicted steady state inlet pressure shows good
agreement. The measured average liquid hold-up at the riser bottom is well predicted. The model
predicts a total hold-up of the pipe cell, not the slug frequency. The model slug frequency is given
by the number of pipe cells, which is an input data. The model predicts a sharper increase and
decreaseof the inlet pressure due to the pigging operation. The time interval for the liquid front to
reach the gamma densitometer after the passageof the pig is well predicted. The measured slug
length and the total volume of liquid in the slug shows good agreement. The model predicts an
inlet pressure reduction after the pig arrival that surprisingly was not observed.
 49

900

800
700

60o
0

40G

300

20

100 jLPG3W M
odel
170 180 190 200 210 22 240 250 260 270

Time (s)

Figure 6.5 - Measured values and model prediction for the riser bottom density with time

3 mocial
3
m
3,

2.9

CL

". 3 V'V 0

200 2lu 220 230 240 J zuu u

Time (s)

Figure 6.6 - Measured values and model prediction for the inlet pressure with time
 -50
6.1.3 Run "PG4Wl

Stratified flow is observed for run HPG4WI. The initial gas and liquid superficial velocities are
5.9 m/s and 0.24 m/s, respectively. The inlet pressure is 5.8 Bara. The pig is launched at 118 s.
Fig. 6.7 and Fig. 6.7 show the transient behaviour of the bottom riser density and inlet pressure
during and after the pigging operation.

The pigging model shows good predictions. The predicted steady state inlet pressure is higher
than observed. The measured and predicted liquid hold-up show good agreement. The tinle and
the maximum value of inlet pressure is well predicted. The model predicts a sharper recovery of
pressure, but the time interval of the low pressure after the arrival of pig shows good agreement.
The model predicted a longer time interval for the liquid front to reach the a densitometer
gamma
after the passage of the pig. The measured slug length is smaller than observed, however the total
volume of liquid in the slug shows good agreement. The liquid hold-up of the slug in front ot'the
pig Is smaller than predicted.
 51

1000

E 9,90 L JHPG4
Experimen
800

>, 700

606,

0 500 Model
E
0 400

0 3 Fin
cr)

Lr-
0
100 110 120 1"
Time (s)

Figure 6.7 - Measured values and model prediction for the riser bottom density with time

(D
CL 5.7

5 5
.

pig launching time 118s

I Ob iI C) 120 1JU i4o

Time (s)

Figure 6.8 - Measured values and model prediction for the inlet pressure with time
6.1.4 Run LPG3W3

The initial and final flow conditions of run HPG.')W') are in the slug flow pattern. The initial gas
and liquid superficial velocities are 2.5 m/s and 0.94 rn/s, respectively. The inlet pressure is 6.9
Bara. The pig is launched at 408 s. Fig. 6.9 and Fig. 6.10 show the transient behaviour of the
bottom riser density and inlet pressure during and after the pigging test.

The predicted and measured steady state inlet pressure shows good agreement. The measured
average liquid hold-up at the riser bottom is well predicted. The model predicts a sharper increase
and a higher maximum value of the inlet pressure due to the pigging operation, The time interval
for the liquid front to reach the gamma densitometer after the passageof the pig is NNiell predicted.
The measured slug length is shorter than predicted, however the total volume of liquid in the SILI(_Y
shows good agreement. The inlet pressure oscillations is well predicted. The pressure oscillation
equates to the slug frequency.
 53

1000

cn
E 90()

800

>% 70G,

Nil
300

200

10C JHPG3W-
'Model
464, j 0440 480 490 500

Time (s)

Figure 6.9 - Measured values and model prediction for the riser bottom density with time

6.9
ni rf iin r-h in ry t- i Tnc. =AnR
6.8
6.7
6.6
6.
.5

6
6
CL
6.1
6.0

4UO 410 2Uu 4140 4 Aj 4b j

-4

Time (s)

Figure 6.10 - Measured values and model prediction for the inlet pressure with time
 54

6.2 Gas Pump without Pig

This type of transient is created by filling the system with water and injecting gas from the inlet of
the test section. After some time the gas valve is closed, the test section is blocked and the
remaining liquid is drained from the bottom of the riser and measured. The ratio between the
remaining volume of liquid to the initial volume of liquid is defined as efficiency of the gas pump.
The initial volume of liquid in the test section is 0.14 m'. The ratio between the standard volume
of injected gas and the transported liquid is called gas-liquid ratio.

A total of 6 experimental gas pump runs without pig are carried out. A summary of the
experimental results of each run can be found in Appendix B. The Appendix F shows the model
results of these runs. Once the gas is injected the inlet pressure tends to increase because the
liquid start flowing to the separator. As soon the gas front reachesthe riser the inlet pressure start
decreasingdue to the low density of the gas. At this point the liquid acceleratesand high flow rate
of liquid is detected in the separator. The separator pressure tends to increase due to the high
volumetric flow rate. However the inlet pressure decreasesrapidly and gets near to the separator
pressure.

Table 6.2 presents data to allow comparison between the experimental data and the model
prediction.

Table 6.2 - Gas pump without pig experimental and model results

RUN AVERAGE SUP. VOLUME OF EFFICIENCY

CODE GAS VELOCITY GAS
AT INLET INJECTED Exp. Model
(M/S) (SM3)

LGPI 1.8 0.233 90.66 87.7

LGP2 5.2 0.521 91.65 91.9

LGP3 7.0 0.721 92.88 92.7

HGPI problems in gas measurement 93.59 -

I
HGP2 problems in gas measurement 90.10 -
1
HGP3 3.7 0.627 90.07 87.3

In Fig. 6.11 the efficiency of the gas pump Without pig is reported. Good agreement is observed.
However this diagram shows that the predictions are lower than measured values. This indicates
that more liquid is transported than predicted, possibly liquid droplets not accounted for by the
model.
 100

98

96

94

92

90

,z88
86

84

82

80
1

Average sup. gas velocity (*)

Figure 6.11 - Measured values and model predictions for the efficiency

6.2.1 Rijn LGPI

At 162sgas is injected at an average flow rate of 0.0080") Sm/s. At 191s the gas valve is closed.
At 180s the hold-up at the bottom of the riser decreasessharply as show in Fig. 6.12. The model
prediction is good, however a slower decrease was predicted. The final hold-up is well predicted
as shown in Fig. 6.12. Fig. 6.13 shows the inlet pressure increase due to the liquid flow. When the
gas reaches the riser the pressure start dropping. The model prediction of the inlet pressure is
good, although the peak pressure was higher than predicted. At 210s the test section is blocked
and the remaining liquid is drained and measured to evaluate the efficiency. The predicted
87.7% and the measured efficiency is 90.66%. The volume of injected gas is 0.233 m"
efficiency is
which gives a oras-liquid ratio of 1.84 Sm/m. This value is very small compared to the
lift system used in the oil industry. A gas-liquid ratio of 9.2 Sm/m' is required
conventional gas
for an efficient continuous gas lift. Even though the bottom riser pressure will be 0.3 Bar greater
than the separator pressure according to Haggedorn-Brown correlation ( 1980).
 56

400

300

200

100

0L
150 160 170 180 190 200
Time (s)

Figure 6.12 - Measured values and model prediction for the riser bottom density with time

Experimental
3.2

150 160 170 180 190 200

Time (s)

Figure 6.13 - Measured values and model prediction for the inlet pressure with time
 3

6.2.2 Run HGP3

At 131s gas is injected at an average flow rate of 0.02239 Sm/s. At 159s the gas valve is closed.
At 140s the hold-up at the bottom of the riser decreases sharply as shown in 6.12. The uYas
reaches the bottom of the riser earlier than predicted. The final bottom riser density is xell
predicted as show in Fig. 6.14. Fig. 6.15 shows the inlet pressure increase due to the liquid flow.
When the gas reaches the riser the pressure start dropping. The model prediction of the inlet
is
pressure satisfactory, although the peak pressure is smaller than predicted. The test section is
closed at 165s and the remaining liquid is drained and measured to evaluate the efficienc'. The
volume of injected gas is 0.627 m' which gives a gas-liquid ratio of 4,97 Sm/m.

6.2.3 Summary

The reduction rate of the riser bottom density is sharper than predicted. The model averaging
approach is the main reason for the smoother reduction of the liquid hold-up. However the
general agreement is good. The measured efficiency is higher than predicted, but the difference is
small.

The inlet pressure prediction is good. In these diagrams no special error trends can be detected.

In run LGPI is injected J*ust 0.233 Sm; of gas and the efficiency is 90.66%. Increasing the oas
volume to 0.721 SM3 according to LGP3 there is a small increase of the efficiency to 92.88%.
This meansthat levels of efficiency of around 90% can be achieved by small volume of gas, there
is no need to inject high volumes of gas. The increased efficiency to 93% requires a three fold
increase in gas which is not econon-k from the production operation point of view. The 1.84
Sm.,/M-' gas-liquid ratio of run LGP I is 5 times smaller when compared with the conventional gas
lift system.
 58

E
77;
Model

60U

50o

0 400
w
-4.0
o
300
I
co
200

100 I

0
120 130 140 150 160
Time (s)

Figure 6.14 - Measured values and model prediction for the riser bottom density with time

'-6

'.
"

'I)
U)
a)
'-6

.4-I
C)

120 130 140 150 160

Time (s)

Figure 6.15 - Measured values and model prediction for the inlet pressure with time
 59

6.3 Gas Pump with Pig

This type of transient is created by filling the system with water and injecting orasand pig from the
inlet of the test section. After some time the gas valve is closed, the test secti'on Iis blocked and the
remaining liquid is drained from the bottom of the riser and measured. The ratio between the
remaining volume of liquid to initial volume of liquid is called efficiency. The initial volume of
3
liquid in the test section is 0.14 m. The ratio between the standard volume of irliected gas and the
transported volume of liquid is called gas-liquid ratio.

A total of 3 experimental gas pump runs with pig were carried out. A summary of the
experimental results of each run can be found in Appendix C. The Appendix G shows the model
results of theses runs. Once the gas is injected the inlet pressure tends to increase because the
liquid start flowing to the separator. As soon the gas front and the pig reachesthe riser the inlet
pressure start to decrease due to the small density of the gas. At this point the pig acceleratesand
high flow rate of liquid is observed in the separator followed by high flow rate of gas. The
separator pressure tends to increase due to the high volumetric flow rate. However the inlet
pressure decreases rapidly and gets near to the separator pressure because a small volume of
liquid remains inside the test section.

Table 6.3 presents data to allow comparison between the experimental data and the model
prediction.

Table 6.3 - Gas pump with pig experimental and model results

RUN AVERAGE SUP. VOLUME PIG TRAVEL EFFICIENCY

CODE GAS VE LOC I TY OF GAS TIME(s)
AT INLET INJECTED Exp. Model Exp. Model
(m/s) (SM3)

LPPIA 1.97 0.466 40 38.6 97.54 99.18

LGPP2 3.45 0.641 32 28.2 97.63 98.62

LGPP) 3.95 0.824 25 25 98.63 99.7

In Fig. 6.16 is reported the measured and predicted efficiency of the gas pump with pig. Good
agreementis observed. However this diagram shows that the predicted efficiencies are higher than
the measuredvalues. All measured efficiencies are above 97.5%. This shows how efficient is foam
pig to transport liquid. Based on this value it is assumed in the model calculation that the liquid
hold-up behind the pio- is 0.025 (2.5%). The measured efficiencies are greater than 97.5%
becausesome liquid left behind the pig is transported by the gas phase. The model predicts that a
higher than observed volume of liquid is transported by the gas. The reason for this is the model
assumption of mist flow in case of extremely low liquid hold-up.
 60

In Fig. 6.17 is reported the measured and predicted pig travel time. Good agreement is observed.
However this diagram shows that the predicted times are smaller when compared with the
measuredvalues.

100

99

98

97

96

95

94

93
Experim enta I
Model
92

91

on
234
Average sup. gas velocity (ms)

Figure 6.16 - Measured values and model predictions for the efficiency

42

40
I,
38
(1)
36
E
34

32

30

28

26
CL
24

22

20
400 500 600 700 800 900
Volume of gas injected (S/m3)

Figure 6.17 - Measured values and model predictions for pig travel time with gas injected
 61

Figures 6.19,6.22 and 6.25 show the results of the model simulation of the bottom riser density
along with the experimental measurements.The measured liquid hold-up shows a non predicted
behaviour as shown in Fig. 6.18. The experimental results show a low density region (2) between
two high densities regions (1) and (3). It is postulated that the low density region (2) is due to
high leakage through the pig during and after the launching when the pig velocity is low. The
leakage rate reduces as the pig speedsup generating a liquid slug in front of the pig, region (1).

(2) (3)

Flow "-X
X1

r
N

Upstream Region pig Downstream Region

Figure 6.18 - Slug pattern in front of the pig

6.3.1 Run LPPIA

The results of run LPPIA are shown in Figures 6.19,6.20 and 6.21. At 307s gas and pig are
inJ.ected at the inlet of the test section. A volume of 0.466 SM3of gas is injected. At 339s the gas
valve is closed. At 345s the pig reaches the receiver. At 327s the hold-up at the bottom of the
riser decreases sharply as shown in Fig. 6.19. The evaluated gas by-pass is 18.3%. Fig. 6.20
shows the inlet pressure increase due to the liquid flow. At 33 Is the pressure start dropping due
to a high volume of gas inside the riser, The test section is blocked at 360s and the remaining
liquid is drained and measured to evaluate the efficiency. The measured efficiency is higher than
observed. The gas volume of 0.466 m' gives a gas-liquid ratio of 3.40 Sm'/m'. The predicted inlet
pressure matches the measured data very well. The model predicts a maximum pig velocity of
18.2 m/s as shown in Fig. 6.2 1.
 ($2

10 11;
I
(0)

60C

500

400

300

200

100
L-
0
300 310

Time (s)

Figure 6.19 - Measured values and model prediction for the riser bottom density with time

2.9

2.7

Jut)
Time (s)

Figure 6.20 - Measured values and model prediction for the inlet pressure with time
 63

20

18

16

14

0 12

10

CD 8

()M
305.0 314.4 323.8 333.2 342.6

Time (s)

Figure 6.21 - Model prediction for pig velocity with time

6.3.2 Run LGPP2

The results of run LGPP2 are shown in Figures 6.22,6.23 and 6.24. At 242s gas and pig are
injected at the inlet of the test A
section. volume of 0.641 Sm' of gas is injected. At 271 s the gas
valve is closed. At 274s the pig reaches the receiver. At 262s the bottom riser density decreases
sharply as shown in Fig. 6.22. The evaluated gas by-pass is 11.1%. Fig. 6.23 shows an increase
the inlet due to the liquid flow. At 265s the pressure started dropping due to high
of pressure
inside the riser. The test section is blocked at 285s and the remaining liquid is
volume of gas
drained and measured to evaluate the efficiency. The measured efficiency is smaller than
The transient behaviour of the inlet pressure is well predicted by the model. The model
predicted.
predicts a maximum pig velocity of 27 m/s as shown in Fig. 6.24,
 64

;t
6

E
400
0
-W I
0 300
cc

40

Time (s)

Figure 6.22 - Measured values and model prediction for the riser bottom density with time

Model
3 F

-
a)'

230 240 250 260 270 280 290

Time (s)

Figure 6.23 - Measured values and model prediction for the inlet pressure with time
 6.5

30

25

20

15

10

0 4mm
243.00 250.05 257.10 264.15 271.20

Time (s)

Figure 6.24 - Model prediction for pig velocity with time

6.3.3 Run LGPP3

The results of run LGPP3 are shown in Figures 6.25,6.26 and 6.27. At 195s gas and pig are
injected at the inlet of the test section. A volume of 0.824 SM3 of gas is injected. At 223s the gas
valve is closed. At 220s, before closing the gas valve, the pig has already reached the receiver. At
21 Is the hold-up at the bottom of the riser decreased sharply as shown in Fig. 6.25. The
evaluated gas by-pass is 8.2%. Fig. 6.26 shows the increase of the inlet pressure due to the liquid
flow. At 214 s the pressure started dropping due to high volume of gas inside the riser. The test
section is closed at 240s and the remaining hquid is drained and measured to evaluate the
efficiency. The measured efficiency is smaller than predicted. The model prediction of the inlet
pressure is very good. The model predicts a maximum pig velocity of 35 m/s as shown in Fig.
6.27.
 66

101,

_x

0)
5

300

2'-

IL

180 190 200 210

Time (s)

Figure 6.25 - Measured values and model prediction for the riser bottom density with time

F xperimenta I

180 190 2uo 10

Time (s)

Figure 6.26 - Measured values and model prediction for the inlet pressure with time
 40

LGPP3

20

CL
10

0 40
195 200 205 210 215 220

Time (s)

Figure 6.27 - Model prediction for pig velocity with time

 61

6.4 Transient runs

A total of 10 experimental transient runs without pig were carried out. Transients caused bN.
changes in liquid flow rate, gas flow rate and separator pressure have been considered. A
summary of the experimental results of each run can be found in Appendix D. Appendix H shows
the model results of theses runs.

Table 6.4 presents the initial and final inlet pressure to allow comparison between the
1
experimental data and the model prediction.

Table 6.4 - Transient runs without pig

RUN TYPE OF INITIAL INLET PRESSURE FINAL INLET PRESSURE

CODE TRJALNSIENT Exp. Model EXP. Model
(Bara) (Bara) (Bara) (Bara)

HTG I 2W3 Gas flow rate 5.75 5.78 5.75 5.80

increase

HTG34W I Gas flow rate 4.76 4.70 5.93 6.30

increase

HTG34W'? Gas flow rate 6.55 7.00 5.60 5.60

decrease

HTG34W-33 Liquid flow 6.13 6.20 6.85 7.30

rate decrease

LTG12WI Gas flow rate 2.60 2.80 2.60 2.48

increase

LTG23W2 Gas flow rate 2.80 2.63 2.80 2.75

increase

LTG-')4W2 Gas flow rate 2.70 2.70 3.20 3.20

increase

LTG34W-') Liquid flow 2.85 2.75 2.90 3.00

rate decrease

LTG4-')WI Gas flow rate 2.35 2.60 2.30 2.40

decrease

LTG-')2WI Gas flow rate 2.35 2.40 2.25 2.27

decrease
 69

In Figures 6.28 and 6.29 are reported the prediction of the initial and final inlet pressure. Good
agreementis observed. There is no special error trend for these runs.

cu

0
1..

U)

CL

.m

456

Exp. Initial Inlet Pressure (Bara)

Figure 6.28 - Measured and predicted initial inlet pressure

LL

3
10

2 8
2 456
Exp. Final Inlet Pressure (Bara)

Figure 6.29 - Measured and predicted final inlet pressure

Transient runs HTG12W3, LTG12WI, LTG34W-,' and HTG-)34W2 are used to analvse the
transient tests.

Figures 6.30,6.32,6.34 and 6.36 show the results of the model simulation of the bottom nser
density along with the experimental measurements.

Figures 6.31,6.33,6.35 and 6.37 show the results of the model simulation of the inlet pressure
along with the experimental measurements.

6.4.1 Run HTG12W3

This transient run is created by increasing the gas flow rate at the inlet. The initial inlet gas flow
rate is 0.00034 m'/s and the liquid flow rate is 0.00235 m-')/s.The separator pressure is 4.4 Bara.
From 68s to 95s the gas flow rate is increased to 0.0012 m-/s. The density at the bottom of the
riser decreases as shown in Fig. 6.30. The model prediction is good. The initial liquid hold-up is
smaller than predicted. The inlet pressure prediction is also good, however a slightly higher
temporary increase of the inlet pressure is observed. Fig 6.331 shows the that initial and final inlet
pressureare well predicted.

M(
900
800

700

600

500
FI Ex p chrim enta I
400

300

200
IHTG12W3]

ui4U
50 60 70 80 90 10 u001 1-)

Time (s)

Figure 6.30 - Measured values and model prediction for the riser bottom density with time
 -1

6.00 - Experimental

5,9 5
cu
m
5.90
A
:35.85
0
V)
IILI

I
HTG12W31

5U 60 70 -4 U

Time (s)

Figure 6.31 - Measured values and model prediction for the inlet pressure with time

6.4.2 Run HTG34W2

This transient run is created by reducing the inlet gas flow rate and reducing the separator
pressure. The initial inlet gas flow rate is O.ol45 m-/s and separator pressure is 5.8 Bara. From
130sto 170s the gas flow rate is reduced to 0.012 m,.'/s. The final separator pressure is 4.9 Bara.
The initial and final density at the bottom of the nser is well predicted, however the model
predicted some oscillation around 170s that was not observed as shown in Fig. 6.32. The inlet
pressure prediction is higher than predicted, however the final inlet pressure is well predicted as
shown in Fig. 6.33.
 -

1000

80 IHTG34W2]
7

(n 600
r r57

500
E40, i
0
bulh il
o3C,

M0deI
0L
50 70 90 110 130 150 170
Time (s)

Figure 6.32 - Measured values and model prediction for the riser bottom density with time

-
-r; IA-. Ixk7lv-l
7

6.4

6.2

6.0

5.2

50 70 90 110 130 ISO 170 190 210 230 250

Time (s)

Figure 6.33 - Measured values and model prediction for the inlet pressure with time
6.4.3 Run LTG12WI

The initial condition of this run was severe slugging as shown in Figures 6.34 and 6.35. The
average gas flow rate at the inlet is 0.00025 m /s and the average liquid flow rate is 0.000-333)rn/s.
The separator pressure is 1.9 Bara. The model prediction of the severe slugging conditions such
as frequency and maximum pressure is good, however the minimum pressure is smaller than
predicted. From 430s to 490s the gas flow rate is increased to 0.00125 m`/s, Even thou(-)rhthe
system continued operating under severe lugging condition. The model failed to predict this final
A
condition. stable inlet pressure and density at the bottom of the riser is predicted,

10 C,
Cl)
E- 9

CD
8

>- 700

600

500

400

o 300 4 w-
co
:m odel
4)
200

100
Expe rim on t,,,.
a
0
100
200 300 700 800 900 1000

Time (s)

Figure 6.34 - Measured values and model prediction for the riser bottom density with time
 74

3,2
3.
m 3,:
I-
CU 2
co
2 ps
2
2.6
2.5
0-
2.4
2.3
2.2

Figure 6.35 - Measured values and model prediction for the inlet pressure with time

6.4.4 Run LTG34W3

This transient run is created by increasing the gas flow rate and reducing the liquid flow rate at the
inlet. The initial inlet gas flow rate Is 0.0037 '
m/s and the liquid flow rate is 0.0017 m`/s. The
separator pressure is 1.9 Bara. From 80s to 120s the gas flow rate is increased to 0.006 m/s and
the liquid flow rate reduced to 0.0015 m/s. The density at the bottom of the nser is reduced as
shown in Fig. 6.36. The model well predicts the average liquid hold-up for the slug unit. The inlet
pressureprediction is also (,Yood, however a slightly higher inlet pressure is observed as shown in
Fig, 6.37.
 -i

900 k

800

>% 700 I
cn
c 600

I
300 I

4f
ITT

200
W
,
n 100
n Model
xperm... e
0-20 40 60 --------
Time (S)

Figure 6.36 - Measured values and model prediction for the riser bottom density with time

3.4

3.3 Experimental
3.2
Ah
A.
3.1

3.0
U)
2.9
CL

ILTG34W73
------- -- -- -- ---- --1--- -- ---i
60 81 140 160 180 200
20 40

Time (s)

Figure 6.37 - Measured values and model prediction for the inlet pressure with time
 76

7. CONCLUSIONS AND RECOMMENDATIONS

An experimental programme was conducted and a new model to simulate pigging and gas
pump
in pipeline-riser system was developed. Transients data were collected and computer simulations
were made for conventional pigging, gas pump with and without pig, and also for transients
created by gas flow rate, liquid flow rate and separator pressure changes. The gas accumulation
term is important in most of the test runs. So the assumption of quasi-steady state for the gas
phase proposed by Taitel et al. (1989) and used by Minami (1991) is not suitable to model the
pigging dynamics and transient flow tests presented in this study. Simple transient gas-liquid flow
tests in a 420m, 77.9mm diameter, can not be considered well predicted by TUFFP simplified
model (Minami - 1991), PLAC and OLGA according to experimental data collected by Vigneron
et al. (1995). The following conclusions and recommendations are in 7.1 and 7.2.
_given

7.1 Conclusions

The following conclusions regarding the model and the experimental work are made-

Transient gas-liquid flow and pigging data in a 0.0525 m diameter, 67 m long, 9.9 m high
pipeline-riser system has been collected using a high speed computer based data acquisition
system. Unprecedented detailed information on pipeline-fiser transient and pigging
dynamics was obtained.

(2) A new transient two-fluid model based on flow pattern dependent set of equations was
successfully developed in this work. The semi-implicit numerical solution coupled to a new
slug flow is
approach very stable The simplified gas and liquid momentum equations were
found to be justified. The treatment of the inertial terms as pseudo shear force was found
sound. The Andritsos and Hanratty (1987) and Wallis (1969) gas-liquid interface ffiction
factor correlations yielded good results. The Gregory et al.(1978) correlation yielded low
slug liquid hold-up.

(3) A new flow pattern transition based on slug flow as a combination of annular, stratified and
bubble flow was developed. This approach was very successful from the physical and
numerical point of view, generating smooth transitions between flow patterns avoiding
discontinuities in the pressure and hold-up calculations.

(4) A new pigging model was developed. The pipeline is divided into two sectionsupstream
and downstream the pig. Conventional transient calculation with proper boundary
is
conditions made in each section. Gas and liquid by-passes are considered. Neglecting the
pressure drop across the pig was found not to cause any deficiency.
(5) For conventional pigging the model predictions of the pig travel time and the inlet pressure
are good. A smaller hold-up in front of the pig is predicted. The use of Gregory et al.
correlation for the calculation of the liquid hold-up in the slug region is the main reason for
this. It should be expected a high increase of the inlet pressure and high liquid and gas now
rates at the separator. Operational be
problems can caused.

In case of gas pump without pig the simulated results matches the experimental data vell-
well. A slightly greater than predicted efficiency was observed. This means that the gas
phase transports more liquid than expected for liquid hold-up near 0.10. The reason for this
is the fact that liquid droplets in the gas phase is neglected by the model. The gas-liquid
ratio are smaller when compared with the conventional gas lift system.

(7) For gas pump with pig runs the model predictions of inlet pressure are very good. However
the model fails to predict a low density region between two high densities regions in front of
the pig. However the model predicts well the average density and length of the gas-liquid
mixture region in front of the pig. The predicted inlet pressure values matched the measured
data very well.

(8) Gas pump can be a efficient process of liquid transportation. However the observed and
theoretical flow behaviours indicate that high liquid and gas flow rates at the separator
should be expected, specially when pig is used. Efficiencies of 90% was obtained without
pig, even for low gas-liquid ratio.
7.2 Recommendations

The following recommendations regarding the model and the experimental work are made

The pigging model was found to be inaccurate in the prediction of the m-o regions of
different liquid hold-up in front of the pig. The development of these regions can be better
investigated by using gamma-densitometer in other points of the pipeline section.

(2) Pig by-pass should be further investigated specially during the pig launching. A transparent
pig launcher can provide a better understanding of the phenomena.

(3) The model should be improved by using a more accurate pig by-pass model. A correlation
based on a wider range of experimental data can be developed. There is a need to develop a
liquid by-pass correlation based on pig velocity, fluid properties and pipe wall conditions.

(4) In this work it was used just one type of pig. In the oil industry a huge variety of pl(ys is
Z'
used. The evaluation of the effect induced in the gas-liquid flow by different types of pig Is
essential. The gas and liquid by-passes models for different pigs are important issues for
pipeline-riser systems. High gas by-pass through the pig can reduce the severity of tile
problems detected in this work, allowing a safer pigging operation.

(5) Different riser shapes, other inclination angles of the pipeline and different fluids should be
tested. Due to the high level of instability of pipeline-riser systems,the flow regimes could
be tremendously affected by different combinations of geometrical configurations and fluid
properties.

(6) The transient gas-liquid flow and pigging models developed in this work should be further
tested against experimental data collected from larger and higher pressure pipeline-nser
systems. The increase of pressure tends to eliminate the pressure and hold-up fluctuation.
So it is important to perform a detailed investigation on the pressure effects.

(7) Additional experimental verification of the proposed models is needed for transient flow
in
conditions not covered this work such as severe slugging. The simultaneous combination
be
of severe slugging and pigging should also investigated.
 -9

REFERENCES

AEA Technology 1996. PLAC - An advanced computer program for the analysis of transient
multiphase hydrocarbon flows. Technical Manual AEAT 1996.
-0096.,

Andritsos, N., Hanratty, T. J., 1987. Influence of interfacial waves in stratified gas-liquid flows.
AlChE J., pp. 444-454.

Banerjee, S., Chan, A. M. C., 1981. Refilling and rewetting of a hot horizontal tube 11:
-
Structure of a two-fluid model. J. Heat Transfer Vol. 103, pp. 287-292.

Barua, S., 1982. An experimental verification and modification of the McDonald-Baker pigging
model for horizontal flow. MS Thesis U. of Tulsa.

Barnea, D., Brauner, N., 1985. Hold-up of the liquid slug in two phase intermittent flow. Int. J.
Multiphase Flow 11, pp. 43-49.

Barnea, D., Taitel, Y., 1986. Flow pattern transition in two-phase gas-liquid flows. Encyclopedia
of Fluid Mechanics, Vol. 3, Gas Liquid Flows. Gulf Publishing, Houston.

Beggs, H. D., Brill, I P., 1973. A study of a two-phase flow in inclined pipes. I Pet.
Technology, pp. 607-617.

Bendiksen, K. H., Brandt, I., Fuchs, P., Linga, H., MaInes, D., Moe, PL, 1986. Two-phase
flow research at SINTEF and IFE- some experimental results and a demonstration of the dynamic
two-phase flow simulator OLGA. Offshore Northern SeasConference.

Bendiksen, K. H., MaInes, D., Moe, &, Nuland, S., 1991. The dynamic two-fluid model
OLGA: theory and applications. SPE Production Engng. J., pp, 171-180.

Brill, I P., Doerr, T. C., Brown, K. E., 1967. An analytical description of liquid slug flow is
small diameter vertical conduits. I Pet. Technology, pp. 419-432.

Brown, K. E, Jessen, F. W., 1962. Evaluation of valve port size, surface chokes and fluid fall-
back in intermittent gas-lift installations. I Pet. Technology, pp. 315-322.

Brown, K. E., 1980. The Technology of Artificial Lift. Penwell Publishing Co..

Cunfiffe, & S., 1978. Prediction of condensate flow rate in large diameter high pressure wet-gas
pipelines. APEA J. Vol. 18, pp. 171-177.

Dukler, A. E., Wicks, M., Cleveland, P, G., 1964. Frictional pressure drop in two-phase flow- a
comparison of existing correlations for pressure loss and holdup, AlChE I Vol, 10, pp. 3 8-43.
 80

Dukler, A. E., Hubbard, M. G., 1975. A model for_gas-liquid slug flow in horizontal and near
horizontal tubes. Ind. Eng. Chem. Fund. Vol. 14, pp. 337-347.

Dutta-Roy, K., 1982. An investigation of transient Phenomena in two-phase flow. MS Thesis U.

of Tulsa.

Eaton, B. A., 1967. The prediction of flow patterns, liquid holdup, and pressure losses occurTing
during continuous two-phase flow in horizontal pipelines. I Pet. Technology Vol. 19, pp. 815-
828.

Gregory, G. A., Nicholson, M., Aziz, K., 1978. Correlation of the liquid volume fraction in the
slug for horizontal gas-liquid slug flow. Int. I Multiphase Flow 4, pp. 33-39.

"ewitt, G. F., 1,982.Handbook of multiphase systems.Hemisphere Publishing Corp.

Kohda, K., Suzukawa, Y., Furukawa, H., 1988. New method for analysin!
g transient flow after
pigging scores well. Oil and Gas J., Vol. 86, pp.40-43,46-47.

Kohda, K., Suzukawa, Y., Furukawa, H., 1988. Pigging analysisfor gas-liquid two-phase flow
in pipelines. II th Annual Energy-resources Technology Conference & Exhibition.

Lima, P. C. R., Neto, S. I A., 1995. Foam Pigs Solve Pipe Cleaning Problems Offshore Brazil.
Oil and Gas J., Vol. 93, pp. 64-67.

Lima, P. C. R., 1996. Pig Lift. I Pet. Technology, 930-93 1.

_pp.
Lima, P. C. R., 1996. Method and equipment for offshore oil production by intermittent gas
injection. Brazilian Patent Application PI 9602747-9.

Lima, P. C. R., Yeung, H., 1998. Modelling of Transient Two-Phase Flow Operations and
Offshore Pigging. 1998 SPE Annual Technical Conference and Exhibition, New Orleans.

Liao, T., 1991. Mechanistic modeling of internuttent gas lift. PhD Thesis U. of Tulsa.

Lyczkowski, R. W., Gidaspow, D., Solbrig, C. W., Hughes, E. D., 1975. Characteristics and
stability analysis of transient one-dimensional two-phase flow equations and their finite difference
equations. Paper 75-WA/HT-23ASME Winter Annual Meeting.

Mahaffyq J. H., 1982. A stability-enhancing two-step method for fluid flow calculations. I
Comput. Physics Vol. 46, pp. 329-341.

McDonald, A. E., Baker, 0., 1964. Multiphase flow in pipe lines. Oil and Gas J. pp. 68-71 (June
15), pp. 171-175 (June 22), pp, 118-119 (July 6).
 81

Minami, K., 1991. Transient flow and pigging dynamics in two-phase pipelines. PhD Thesis U.
of Tulsa.

Neely, A. B., Montgomery, I W., Vogel, J. W., 1974. A field test and analytical
study of
intermittent gas lift. Soc. Pet. Eng. I pp. 502-512,1974

Nicholson, M. K., Aziz, K., Gregory, G. A., 1978. Intermittent two phase flow in horizontal
pipes: predictive models. Can. J. Chem. Engng 56, pp. 653-663.

Nicklin, A J., Wilkes, J. 0. Davidson, J. F. 1962. Two-phase flow in vertical tubes. Trans.
, ,
Instn. Chem. Engrs 40, pp. 61-68.

Pauchon, C., Mulesia, H., Lopez, D., Fabre, J., 1993. A comprehensivemechanistic model for
two-phase flow. 6h International Conference on Multiphase Production, Cannes, pp. 29-59,

Roy, R. P. Ho, S., 1980. Influence of transverse intraphase velocity profiles and phase fraction
distributions on the character of two-phase flow equations. Int. J. Heat and Mass Transfer Vol.
23, pp. 1162-1167.

Schmidt, Z. et al., 1984. Hydrodynamic model for intermittent gas lifting of viscous oil, J. Pet.
Technology, pp. 475-483.

Scoggins Jr., M. W., 1977. A Numerical Simulation Model for Transient Two-Phase Flow in a
Pipeline. PhD Dissertation University of Tulsa.

Sharma, Y., 1983. Modelling transient two-phase flow in stratified flow pattern. MS Thesis U.
of Tulsa.

Sharma, Y., 1985. Modelling two-phase slug flow. PhD Thesis U. of Tulsa.

Swamee, P. K., Jain, A. K., 1976. Explicit equation for pipe-flow problems. Hydr. Div., Proc.
ASCE, pp. 657-664.

Taitel, Y., Dukler, A. E., 1976. A model for predicting flow regime transitions in horizontal and
near-horizontal gas-liquid flow. AlChE J. Vol. 44, pp. 920-935.

Taitel, Y., Lee, N., Dukler, A. E., 1978. Transient gas-liquid flow in horizontal pipes - modelling
flow pattern transitions. AlChE I Vol. 22, pp. 47-55.

Taitel, Y., Shoham, 0., Brill, I P., 1989. Simplified transient solution and simulation of two-
phase flow in pipelines. Chem. Eng. Sci. Vol. 44, No. 6, 1353-1359.
_pp.
 81

Vigneron, F., Sarica, C., Brill, I P., 1995. Experimental analysis of imposed two-phase flow
in
transients horizontal pipelines. 7h International Conference, Multiphase 95, Cannes, pp. 199-
217.

Wallis, G. B., 1969. One-dimensional two-phase flow. McGraw-fEll.

White, G. W. et al., 1963. An analytical concept of the static and dynamic parameters of
intermittent gas lift. J. Pet. Technology, pp. 301-308.
 41
.

APPENDIX A

EXPERIMENTAL DATA OF THE CONVENTIONAL PIGGINCIF Rt-Vs

 A2

A. EXPERIMENTAL DATA OF THE CONVENTIONAL PIGGING RUNS

A total of 13 conventional pigging runs are surnmarised in this appendix. Table A. I presents the
initial conditions of the steady state gas-liquid flow, pig launching schedules and other transient
characteristics related to the runs. Figs A. I to A. 13 show the transient flow behaviour by
caused a
conventional pigging operation.

Table A. 1 - Summary of the conventional pigging

RUN IN SITU IN SITU PRESSURE PIG SCHEDULE

CODE GAS FLOW LIQUID Inlet Separator Launching Arrival
RATE FLOW RATE (Bara) (Bara) (s) (s)
(M3/S) (M3/S)

LPG3W] 0.0053 0.00033 2.3 1.9 510 558

LPG4W] 0.0125 0.00032 2.4 2.0 458 483

L,PG3W2 0.0045 0.00098 2.7 2.0 288 330

LPG4W2 0.0120 0.00091 3.2 2.2 520 538

LPG3W3 0.0035 0.00170 2.8 1.9 193 232

LPG4W3 0.0080 0.00160 3.1 2.0 265 288

HPG2W] 0.0016 0.00054 4.7 4.2 271 350

[IGP3WI 0.0057 0.00033 5.6 5.6 322 356

HPG4W1 0.0120 0.00049 5.8 5.8 118 134

HPG4W2 0.0110 0.00069 6.2 6.1 123 141

HPG2W3 0.0011 0.00200 6.2 6.2 440 480

HPG3W3 0.0050 0.00190 6.3 6.3 408 435

0.0130 0.00135 6.9 6.7 94 108

HPG4W3
, 1
 A3

0.9
E
X 0.8
8 0.7
6 0.6
0

OA

LL 0.3

0.2
-di 2
0.1

0 0.0
500 520 540 560 580 600 620 640 500 520 540 560 580 600 620 640
Time (s) Time (s)

1000 3.2

i; - 900 10
E
6,800
Z8
700
(A 2.6
c 600

500 2.4
E
0 4A
r. 400 (A
4) 2.2
0
300 IL

ZO
LA 200

Ji 1.8
100- Separator
0 t6
500 520 540 560 580 600 620 640 500 520 540 560 580 600 520 640

Time (s) Time (s)

3.5
Cl)
E
5 3.4
x
16 E
T 3.3
14
m 3.2
12 L .C
u
3.1
lo
0
LL 3.0

2.9 L
CL
4 w
U) 2.8

2
2.7
0
500 520 540 560 580 600 620 640 500 520 540 560 580 Boo 620 ,4 0
Time (s) Time (s)

Figure A-1 - Summary of run LPG3W I

 A4

18-
mE 14
1.6 -
121-
8 1.4 -

iv 1,2 -
1.0
09
0
iL 06
A 0.4
m 4
02
.

-E 0.0
0
400 420 440 460 480 500 520 540 560 580 600 400 420 440 460 480 500 520 540 5K 580 600
Time (s) Time (s)

10DO 4.0

3.8
i; goo
E 3.6
-6800
3.4 G's FL. ! ): pr

700 A
3.2

a 600 S3.0-

500 2.8
E
0 400 2.6
Z
0
co 300
-
2.2
200
2.0
Am
'00 1 .9
Separator
0
400 420 "o 460 480 500 520 540 560 580 600 400 420 440 460 480 500 520 540 566 58c, 600
Time (s) Time (s)

E 18 3.6
x
3.4
8 16
6
14 3.2
w0
M 12 u
w 3.0
3: lo
rL 2.8
8
2.6

4, U) 2.4

2 2.2

o 0, 520 540 560 580 600

480 500 520 540 560 580 600 400 420 440 460 480 500
400 420 "0 460

Time (s) Time (s)

Figure A-2 - Summary of run LPG4W I

 A5

12

10 E
x
8
ci
6 W
ea

01
290 290 300 310 320 330 340 350 360 370 380 280 290 300 310 320 330 350 360 31-0 3 so
Time (s)

1000 3.6

i; goo 3.4

-6800 3.2 - Gas Flowrnet, i

>1 700 1-i 3,0
600 2.8
500 2.6
400 24
3 22
20 20-0,
2,0

100 t8 -
Separator
0280 290 300 310 320 330 340 350 360 370 380
1.6
280 290 300 310 320 330 40 - --
-150 360 370 380
Time (s) Time (s)

34
E
3.3
16

14
3.2
12,
3.1

U-
A 3.0
6
ci. 2.9
4,

2
2.8
0
280 290 300 310 320 330 340 350 360 370 380 280 290 300 310 320 330 340 150 "60 -170 180
Time (s) Time (s)

Figure A-3 - Summary of run LPG3W2

 A6

14 x

12 1.0

10,

6 LL 0,9

0.8
500 510 520 530 540 550 560 570 580 590 600 500 510 520 530 540 550 56 10 570 580 5 90 boc
Time (s) Time (s)

1000 4.0

jT 900 3.8
Gas FlowrneV
E 3.6 -
900
3.4 -
700
3.2
600 S3.0-

500 2.8

76 -
400
0
OD 300 CIL ZA
-1
2.2
200
2.0
100
1.8 Separator
0
500 510 520 530 540 550 560 570 580 590 600 500 510 520 530 0 550 560 5-0 580 590 600
-"
Time (s) Time (s)

18 3.8
-> 3.7
16 E
8 3.6
S14 35
.
34
12! .
3.3 L
lo 3.2
31
3.0
6 2.9
Problems in measurement
2.8
4 U)
2.7 Ir
2L 2.6

590 600 500 510 520 530 540 550 560 570 580 590 600
500 510 520 530 540 550 560 570 580
Time (s)
Time (s)

Figure A-4 - Summary of run LPG4W2

 A7

(. )

2.6

2.4

2.0
9:
1.8 -

1,2

01
170 180 190 200 210 220 230 240 250 260 270 170 180 190 200 210 220 230 240 250 260 270
Time (s) TirnA s)

1000 116

j;,- 900
800 12

>, 700 3.0

:t-_
LA
600 28

500 2.6
E
0 400 2,4
:z
0
E0 300 22
4)
T 200 2.0
it
100 1.8 1w,
Separator
1.6
170 i8o 190 200 210 220 230 240 250 260 270 170 180 190 200 210 220 230 240 250 260,270

Time (s) Time (s)

33
14
3.2
12,L

10 3.1

4)
>
3.0

cr
Z 2.9
CL 4 JW V v IV

2.8
2'

:n
0 o 2.7 '
170 180 190 200 210 220 230 240 250 260 270 170 180 190 200 210 220 220 240 250 260 270

Time (s) Time (s)

Figure A-5 - Summary of run LPG3W3

 A8

1.9
E E
14 x 1.8

8 12
1.7
ci
0 10
M - 111116,111
cr- 1-5
.,:: 8
0 0 1.4
i,
6- LL
. 1.3
4 1.2

2
1.1
0 1.01
250 260 270 280 290 300 310 320 330 340 350 170 180 190 200 210 220 230 240 250 260

Time (s) Time (s)

1000 '1 0
3.8
i; 900
E 3.6 Gas Flowmetef
6)800 3.4 -
All
700
3.2
600 3.0

500 2.8
E 2.6
0 400
r.0
co 300 (L Z4
2,2
V 200
2.0
100 1.8
Separator
0
250 260 270 280 290 300 310 320 330 340 350 250 260 270 280 290 3IJU 310 32u liti ;4L, nj
-
Time (s) Ti nie (s

E 3.7
18
E
16
3.5
S14
3.4
cu12
L) 3.3
cr
.16 10 3.2

3.1
LL
LA
8
cr
3.0
,
6
0.2.9 ,
CL , 0
u) 2.8
2- 2.7

0, 290 300 310 320 "30 340250

250 260 270 280 290 300 310 320 330 340 350 250 260 270 280
Time (s)
Time (s)

Figure A-6 - Summary of run LPG4W3

 A9

r)
I
-2
vi
i
E
x
1.7 &7
8
0.6 .....................

1,5 0.5
0 0 0A
iL
U.
1.3 0.3

1.2, 0.2

=11- 0

'(P 00
200 300 400 500 200
Tirre (s) Time (s

1000- 5,5

i; 900 - Gas Flowmf-el

E
-6800- 5.0
700
LA 600
c
43)
- IM 4.5
a
500
E
0 400 40
r0
300 Separa
200 35

100

3.0 -
200 300 400 500 200 co 400 500

Time (s) Time (s)

W
7 3.3
(0
E
x >
6 3.2
E

(U 3.1

Ir 4 L)

> 3.0
LL 3
VI
Z9
2

0 o 27
200 300 400 500 200 1100 400

Time (s) Time is)

Figure A-7 - Summary of run HPG2WI

 AIO

0,9
E
x 0.8
7 8 0.7
6
6 0.6
5 0.5
4 ;t0.4
.2
3- LL. 0.3

2 0.2

0.1

0.0
300 320 340 360 380 400 420 440 460 480 500 300 320 3AO 360 280 400 420
Time (s) Time ; s)

1000
6.4
j; 900
E
6.2
700
6.0
IA
a 600 Co
4)
a 500
E
0 400

300
5.4
200
5.2
Separator
DO 420 500
5.0
300 320 340 360 380 400 440 460 480 500

0
300 320 340 260 380 400 440 460 480 420

Time (s) Time is

E 3.7
14
E 3.6 r
12 3.5
3.4
W 10
3.3
3.2
0 3.1
-j 3.0
6
r
0 2.9
-J
CL 4 0- 2.8 -
Q)
U) 2.7
2
2.6

0 o
300 360 380 400 420 440 460 480 500 300 320 UO 160 380 400 420 440 460 4480 SOG
'320 3.40

Time (s) Time (s)

Figure A-8 - Summary of run HPG3WI

 All

V) is

E
14
x
12 0.7
6
10
It 0.5

0
0 0.4
U- 6 U.
(A 0.3
4
0.2
2

0 0.0
100 110 120 130 140 150 160 170 180 190 200 100 110 120 1co Ido 150 160 1 SO
Time (s) Time is)

6,6 -
E
goo- 6.4

700 6.2
(A m
600 - S6.0
500 4)
E
0 400 to
V.
0
co 300
5.4
200
5.2
100
separator
0 5.0
IN 110 120 130 140 150 160 170 180 190 200 100 110 120 130 140 150 1P7,0 1-(Il 180 i9o 200

Time (s) Timp f,; )

Il)

18 3.6
x 3.5
3.4
S14
3.3

12 3.2
3.1
lo
3.0
LL
8 2.9

6 2.8
CL t 2.7
2.6
2.5

0 0, 110 120 130 140 150 160 170 190 190 200 100 110 120 130 140 150 160 170 180 190 20G
1LIO
Time (s)
Time (s)

Figure A-9 - Summary of run HPG4WI

 A]2

0.9

14 x 0.8

OJ
12
Ci 6
0.6

0.5

CA
0
U. 6 ILL
& 0.11 -
4
0.2
41
2
0.1
0
50 70 90 110 130 150 170 190 0.050 70 90 110 15
Time (s) Time (si

1000 7.0

j; 900 Gas Flowmeter

E 800
6.5

700

600 6.0
4)
a
500
E
0 400
1-, 5.5
0
300

200 5.0 Separator

100

0 4.5
50 70 90 110 130 150 170 i9o so TO 90 110 130 150 1-0 190

Time (s) Time (s)

E 18

3.7
16
S14
w 3.5:
m 12

- lo 3.3
0
LL

2.7
2

o 2.5 1-
50 70 90 110 130 150 170 190 50 70 90 110 130 150 17C,

Time (s) Time (s)

Figure A-10- Summary of run HPG4W2

 A 13

21

2.4

1.2 2,2

1.0 0 z0
0.8
LL LL
L4 0.6

0.4 1.4
0.2 E t2 -

0.0 1.0 1
400 420 440 460 480 500 520 540 400 420 440 460 480 500 520 540

Time (s) Time (s

1000 7. C

j; 900 Gas F; o\Ntneter

E
-6800
700
600 Im 60 -
0
500
F-
0 400 V> 55
=0
300

200
Separator
100

4.5
420 440 460 480 Soo 520 540 400 420 440 460 480 500 520 540
400
Time (s) Time (s)

Ln 6
5
E 3.5
X 3.4
E
3.3

3.2

L) 3.1
w
3 3.0
3 >
0
2.9

2.8

CL 2.7
U) 2.6

2,5
0
400 420 440 460 480 500 520 540 400 420 440 460 480 500 520 540

Time (s) Time (s)

Figure A-1 I- Summary of run HPG2W3

 A 14

10
2.8

2,6

7 8 14

Z2

0: 2,0

3 LL
1.6

1'0
400 410 420 430 440 450 460 470 480 490 500 400 410 420 430 440 450 460 470 490 500
.180
Time (s) mle (S)

1000 rl 7,0

900 - Gas Flowmetef

E
-6800- 6.5 -

>, 700 A J'o

600 - 6.0 -
4)
a
500
E
0 400 5,5
*:
0 A
300
200 5,0
Separator
100 -
01 4,5
400 410 420 430 440 450 460 470 480 490 500 400 410 420 430 440 450 460 470 480 490 1,00

Tinne (s) Time (s)

12 3,6
E 3,5
x
10 3A

3.3

8 3,2

31
6
0
FL 30

29
0 4
2.8

2 2.7

2.6
0.
400 410 420 430 440 450 460 470 480 490 500 400 410 420 430 440 450 460 470 480 490 Soo

Time (s)
Time (s)

Figure A-12- Summary of run HPG3W3

 A 15

to Is -
E
E
14 1.6
x
6 12
$ 1.4
10,
4) 1.2
10
cr 1.0
3:
0 0 0.8
LL 6
(A
LL
0.6 -
4
0.4
2 0.2 -

0 0.0 1
60 70 80 90 100 110 120 130 140 150 160 so 70 80 90 101,110 120
Time (s) Time (s)

1000 8
j; 900 Gas Flo%vineter
E
800 A
700
(A
600
PD-
0
500
E
0 400
=0
co 300 0

200
5 -%1
100
I Separator
7 so 70 so 90 100 110 130 14 0 150
60 so 90 100 110 120 130 140 150 160 120 16 f)
,o
Tirne (s) Ti rn t? is

1
r)
3.7
x E 16
;; 3.5 7
14
3.4
12 3.3
0 3-2
10 Problems in measureemnt 3.1
3.0
2.9
6 2.8

4 2.7
2.6
2 2.5

150 160 60 70 80 90 100 110 120 IN 140 150 160

60 70 80 90 100 110 120 130 140
Time (s) Time (s)

Figure A-13- Summary of run HPG4W3

 BI

APPENDIX B

EXPERIMENTAL DATA OF THE GAS PUNIP WITHOUT PIG

 B2

B. EXPERIMENTAL DATA OF THE GAS PUMP WITHOUT PIG

A total of 6 gas pump runs without pig are summansed in this appendix. Table B. I presents the
initial conditions of the test section,, the gas injection conditions and the efficiency of liquid
transportation. Figs BA to B. 6 show the transient flow behaviour caused by a gas injection,
without pig, in a pipeline-riser system full of liquid.

Table B. 1 - Summary of the gas pump without pig

INITIAL GAS-
RUN INLET INJECTED GAS AVERAGE EFFICIENCY LIQUID
CODE PRESSURE Volume Time* GAS INJECTION RATIO
(SM3/S)
(Bara) (SM3) (s) (SMI/

LGPI 2.52 0.233 29 0.00803 90.66 1.84

LGP2 2.51 0.521 29 0.01797 91.65 4.06

LGP3 2.60 0.721 28 0.02575 92.88 5.54

HGPI 6.03 55 - 93.59 -

-

HGP2 5.91 44 - 90.10 -

-

HGP3 5.90 0.627 28 0.02239 90.07 4.97

*Duration of gas injection

 B3

3.4
7
3.2
E
3.0
LA

2-8

40 3 2.6
riz
E
3: 2.4
0
.22
LL. 2.2

0 2.0

1.8
0
162 90 169 25 176.50 183,75 191.00 16Z 00 169.25 176.50 183.75 191 001

Time (s) Time fs

1000
TF, 3.3
E
6) 800
goo Bottom
Bot to", f
,, Rise!,
3.1
-,f,
700 2.9

2.7
600
2.5
500
400
2.3 Inlet
2.1 -
300
1.9
200
0 1.7
03 oo
.
15
.
Separator
15000 163 75 177.50 191,25 205.00 150.00 16375 17 50 191 25 205
Time (s) Time (s)

-ir
E 3.2 -
x > 3.0
16
E
Z 2.8
14 r m 2.6
r
tu 2.4 ,
M 121-
2.2 L
lo 2.0
1.8-,
cr 1.6 -
1.4
CL
4) 1.2
(n 1.0
21- 0.8

0 o-
165 175 185 195 205 150.00 163 75 177.50 191 25 2D5001

Time (s) Time (s)

Figure B-1 - Summary of run LGPI

 B4

3.6

3.4

3-2
IA
in
41
3.0
05

4 E 2-6
3.1
0
2.4

2 22
0
2,0

1a
4700 54 25 61.50 68.75 76.00 47 00 54 25 61 51 68 -5 - F ,^I
Tirne (s)
Time s

1000
900 3.7

800 35
3,3
700
31
600 2.9

500 2,7 0
2
.W1 400 2-5

300
2. Inlet
2.1
200 1.9

100 1.7
1.5
0 Separator
30 40 50 60 70 so 90 30 40 50 60 70 8n qn
Tirne (s) Time (s

E 3.3

16
a
Z8
14 h

M 12
ir L) 2.3
: lo
0

a 1.3
4

2
0-
0.2 o
so 60 TO 80 90 30 40 50 60 10 80
Time (s) Time (s)

Figure B-2 - Summary of run LGP2

 B5

lo 4,2
E 4.0
X
3.8
3.6
6
CL 14
3.2
Ir
3.0
4 E
o
3: 2-9
0
jL 2.6
Q) 2
2.4
2.2
0 2.0 I
47 51 55 59 63 67 71 75 47 51 55 59 63 67 71 75
Time (s) Time (s)

1000
M goo 42
E
800
Z, 3.7
7OU
Z, ef, -)1t(DT7i o Riser
600 m 3.2
in
500

400 2.7

300 0.2 2 Inlet

E
0 200
:t::
0 1.7
rn 100
Separator
030 40 so 60 70 80 90 1,2
30 40 50 60 71)
Time (s) Time (, !

4,5
E
x
16 E
8 3.5
14 Cb

12

lo 2.5

a,

6
1.5
4

21-
:3 o
0 01- 0.5 '
50 60 70 so 90 30 40 50 60 70 80

Time (s) Time (s)

Figure B-3 - Summary of run LGP3

 B6

SA
7
E 6.2
6
6.0
5 IA
0
41
IL 5.8
4-
5.6
3: 3
C,
LL 2 5A
2

5.2

0 5.0
476 5 `6 16 I
461 00 474 75 488 50 502 25 111600 456 466 486 -:96
Time (s) 7i rie
js

1000
Bottom of
900 6.3
E
goo 6.1

TOO 5 !3-
Inlet
Soo 5.7
C3
500 1! 5.5
0)
400 5.3
0 300 CL 5.1
E
0 200
49
Separator
0
Co 4.7 -

520 4.5 L
450 460 470 480 490 500 510 450 460 4-0 480 490 500 510 520

Time (s) Time (s)

E
x > 2.6
E
8 2.4
Ci
72
m5 2-0

04
LL 1.6
3 0, 1.4

1.2
cL 2
(D
U) 1.0
0,8

0 450 460 470 480 490 50C 510 520

462 0 476.5 491.0 505.5 520.0

Time (s) Time (s)

Figure B-4 - Summary of run HGPI

 B7

Co 6.6
20
6.4

6.2
15
6.0
S 0.

0
LL 5.4
W
5.2

0 5.0
40 62 73 84 40 51 62 B4 I
Time (s) Time sl

1000 7.0

900 Bottoi-r 3f Piser

Soo
700

4,600
0
im 6.0
500
L)
cr 400 Inlet
0
300
E
0 200 5.0
v
0 Separator
Ca I Oo

040 4.5
51 62 73 84 40 51 62 7a 84

Time (s) Twie (s

IA
i 8
E 2.8,
x E 2.6
8 6L
2.4

cj 2.2
5 10
0

6 4
cu
1.6
3 1.4
1.2
cL 2
(n 1.0
0.8
0 0 84 0 40 51 62
65.0 74.5

"I
A60 555 .
Time (s)
Time (s)

Figure B-5 Summary of run HGP2

-
 B8

8-

77- 68
i
E 00 6.6
4u 6.4

6.2

4 6.0

5.8
3 P
0 3: 5,6
LL 0
02
LL 5.4
LM
ta
0 5.2

0 5.0 -
135 139 143 147 151 155 159 131 136 141 146 151 156
Time (s) Time s)

1000 r IsIf 70

E IT11'f11
900 F
F1
6.5
4,700 goo
600 6.0

500

400 5.5

300 a.
E
0 200
:z 5.0
0 Separator
00 loo

0 4.5 L- ---
11000 123,75 137.50 151,25 i6s. o() 11000 12375 137 50 51 2 165
Time js Time (s)

Ln
co

E 3.0 ,
X :' 2.8
E
8 z 2.6
C5 0 2.4
21
.c
2.0
4
1.8

1.4
2 1.2

0 1.0

127,0 136.5 146.0 155.5 165.0 12175 137.50 151 25 165

Time (s) Time (s)

Figure B-6 - Summary of run HGP3

 Cl

APPENDIX C

EXPERIMENTAL DATA OF THE GAS PUNIP WITH PIG

 2

C. EXPERIMENTAL DATA OF THE GAS PUMP WITH PIG

A total of 3 gas pump runs with pig are summarisedin this appendix. Table C. I presents the
conditions of the test section, the gas and pig injection conditions and the efficiency of liquid
i
transportation, Figs CA to C.3 show the transient flow behaviour caused by a gas and pig
injection in a pipeline-riser system full of liquid.

Table C. 1 - Summary of the gas pump with pig

RUN INITIAL PRESSURE VOLUME OF PIG SCHEDULE EFFICIENCY

CODE Inlet Separator GAS INJECTED Launching Arrival M
(SM3) (s) (s)
(Bara) (Bara)

LPPIA 2.4 1.5 0.466 305 345 97.54

LGPP2 2.4 1.5 0.641 242 274 97.63

LGPP3 2.5 1.6 0.824 195 220 98.63

 C3

-7
32
tu
E
4u 3.0 INS

ci (D 2.8 4.00
40 CL
m
It
2.6
3:
0 E
LL :
2,4
W 0
IL
LA
22

01
307 315 2.0 '
323 331 339 307 1115
Time (s)
Time (s)

1000

900 32-

800 3.0

IUU 2.8

10
600 m 2.6 -
in
500 2.4

400 2.2
Inlet
2.0
300 CL
1.8 -
200
0 1.6
in 100
1.4 Separator
0
300 310 320 330 340 350 360
300 310 320 330 240 1, 50 360
Time (s)
Time (s)

LI)
18
E 12 -
x > 3.0
8 14 2.8
43
cri 2.6
12, c 2.4
M
m
10 (3 2.2
3:
0 IV 2.0
:,,
1.9
IA
m6

1.4
CL
4 1.2

2 11)1.0
:3
0.8 -

312 323 334 345 356 300 310 320 330 340 350 360
Time (s) Time (s)

Figure C-1 - Summary of run LPPIA

 C4

38

to 3.6

3.4
6
ci V, 3.2
5 0- 3.0
4
3: E
0 3
E 3:
0
2.6
W)
rJ 2 LL 2.4
0
2.2

2.0
2,11, 2'1 2 556 261 266 271 246 251 256 1
16 *1 266 ri
Time (s) Time jsj

1000

900
E
800 3.5
Bottom of
700

c 3.0 -
600

500
2,5
400

0 300 CL 2.0
E Inlet
0
V 200
0
Im 100 1 15 It
Separator
23 0 240 250 260 270 280 290 1.0
230 24 01 250 1.
260 -1
270 280 290
Time (s) Time ,s

3.5
16
x >
M
E 3.0

(U 12
2.5
10

2.0

6 cr
1.5 iL

4
LO 1.0
2

00 0.5
250.00 25&75 267.50 276,25 285.00 230 240 250 260 270 280 290

Time (s) Time (s)

Figure C-2 - Summary of run LGPP2

 C5

45

12 ,Z
E m 4.0
10 '00,

3.5
8
(L

W6 3.0
3.
0 E
LL 4 3:
0
IL. 2.5
LA
2 m
0

0 2 0
19 199 202 205 2,08 211 214 217 . 196
220 22 199 202 35 208 21l 214

Time (s) Ti me (s

'9"0"'0'

800 L 4

700

600

500

400 (A
0
91
300 CL

200
r0
CID loo
Separatoi
0
180 190 200 210 220 230 240 iso 190 200 210
Time (s)
Time

LA
() 45
E
16"
x
14
3.5

12

cr
2.5
0
vo 6
CL 4 1.5

U)
2
I
0.5
20000 209,75 217-50 226.25 235.00 180 190 200 210 220 230 240

Time (s) Time (s)

Figure C-3 - Summary of run LGPP3

 I)i

APPENDIX D

EXPERIMENTAL DATA OF THE TRANSIENT RUNS

 1)2

D. EXPERIMENTAL DATA OF TRANSIENT RUNS

A total of 10 transient runs with pig are summarised in this appendix. Table D. I presents the
initial and final conditions of the test section. Figs D. I to D. 10 show the transient flow behaviour
by
caused a change of boundary conditions.

Table D. 1 - Summary of the transient runs

RUN INITIAL PRESSURE FINAL PRESSURE

CODE Inlet Separator Inlet Separator
(Bara) (Bara) (Bara) (Bara

HTG I 2W3 5.75 4.65 5.75 4.75

HTG34W] 4.76 4.31 5.93 5.37

HTG34W2 6.55 5.88 5.60 4.90

HTG34W3 6.13 4.94 6.85 5.95

LTG12WI 2.60 1.80 2.60 2.00

LTG23W2 2.80 2.00 2.80 2.10

LTG34W2 2.70 2.05 3.20 2.20

LTG34W3 2.85 1.90 2.90 2.05

LTG43WI 2.35 2.05 2.30 2.00

LTG32WI 2.35 2.00 2.25 1.95

 D3

1.2
2.9

2.8

io 2.7

C;0.9 16
08 u 2-5 -
ckt
07
3:
0 2.4
0.6 EL
0,5
7- 2.3

*6 0.4 c
. 2.2

2.1 -
0
2.0 '
50 60 70 80 90 100 110 120 130 140 150 50 60 70 80 90 CO 11.3 121J 133
Time (s) Time fs

1000r 6.5

Vuu In let
goo 6.0 -A

700

600

500
co 55

400 50
(L
z
300

200 45 Separator
100

0 4.0 -- I
50 70 90 110 130 150 170 190 50 60 70 80 90 100 110 120 0 150

Time (s) Time (s)

3.14
-
>
E 3.12
E 3.10
14 3.08
8 3.06
12
3.04
Z
lo 3.02
3.00

LL 6 2.98
2,96
2.94
4
2.92
CL
4) 2 190
in y
0' 60 -0 80 90 100 110 120 130 140 150 50 60 70 80 90 100 110 120 130 140 150
so
Time (s) Time (s)

Figure D-1 - Summary of run HTG12W3

 D4

14
t
E E 06
12 x
8 0.5
8
ci 10
0.4
8 m
cr
3: 0.3
0 8 0
LL LL
CA 0.2
4

2 0.1

0 0.0
4 ri 0 500 600 700 400 5co 6&,
Time (s) Time js

900
1 cl
Boo
6.5 -
700 fr

600
w
500 5.5 -
I',

400 CL

300
IV
.=4P8.0-
200 Separator

1rjo

3,5 ---
600 700 400 500 600 TOO
300 400 500 100
Time (s) Time (s)

33
0 is , >
E 3.2
E
3.1
x
14
0
C!
S12 2.9
4) -E 2.8 -
16 10
2.7
8 26
U- 6 2-5
24
4 U)
2,3
2 22

0 400 500 600 700

300 400 500 Soo 700 300

Time (s) Time (s)

Figure D-2 Summary of run HTG34W I

-
 D5

2.0

;, 1.8 -
15

1.4
14

at 13 1.0
3:
0 0'a
1L. 12
IA 0.6
m
0
0.4
11
0.2

1050 0,0
70 90 110 130 150 170 190 210 230 250 50 70 90 110 130 150 170 190 210 230
Time (s) Time (s)

1000
6,8
900
&6
900 L
6.4
700
6.2
600 6.0

500 5.8

400 5.6
CL 5.4 Separator
0 -
300
5,2 -
200
5O -
100 4.8 -
0
50 70 90 110 130 150 170 190 210 230 250 50 70 90 110 130 150 170 190 210 2'0 250

Time (s) Time (s)

L4 18
34
(4 1
E 3.3
x
14 r 12

12 'L t) 3.1

3.0 jj
lo
29

.2
- 28

2.7
0 un
a- 2.6 -
2
2.5
01
50 70 90 110 130 150 170 190 210 230 250 5,0 70 90 110 130 150 170 190 210 230 250

Time (s)
Time (s)

Figure D-3 - Summary of run HTG34W2

 D6

14

13 2.0
E
12 x 1.8
8
6 1.6
io
9 1.4

2 1.2
7 LL
6 1.o
5
0.8

0 20 40 60 so 100 120
0.6 L
140 0 20 40 60 80 100 120 140
Time (s) Time fs',

1000
7.5 Inlet
900

Boo L
7.0
700
6.5
Soo
500
6.0
400 CL

300 5.5

200
5.0
100 Separator
0 4.5
0 20 40 60 80 100 120 140 0 20 40 60 so 100 120 1.40

Time (s) Time (s)

s i
E
x
14 r 3.2

12 L 3.1

di
lo > 3.0
Ir
!; 8 2.9

2-8

0 2.7
13.
2.6

2.5
01
0 20 40 60 80 100 120 140 0 20 40 60 80 100 120 140

Time (s) Time (s)

Figure D-4 - Summary of run HTG34W3

 D7

E 23 0.7
r-
x
2.0
ci
ci
0
m 0.4

0 1.0 0 0.3
U. EL

0.5 0.2

01
0,0
0.0 1 I
100 200 300 400 500 600 700 800 900 1000 0 100 200 300 400 500 600 700 800 900 1000
Time (s) Time (s)

1000
3.2 - In I-#
100
3.0

i 800
iuu
Co
2.8
600 ,30 2.6

500
(A
0 2.4
C 400
a) CL
0 2.2
300
'E
200 2.0

100 -1001,
Separator
0
0 100 200 300 400 500 600 700 800 900 1000 0 100 200 300 400 500 600 700 800 900 1000
Time (s) Time (s)

16 3.4
> 3.3
E'141 E
3.2
x
3.1
12,
3.0
ci
0)10 - 2.9
2.8
2.7
0 2-6
rL 6
2.5
4 2.4

CL 2.3
4) 2 21 -
U)

100 200 300 400 500 600 700 800 900 1000 0 100 200 300 400 500 600 700 800 900 1000
Time (s) Time (s)

Figure D-5 - Summary of run LTG12WI

 D8

E E
X
1.4

g.
0 0,8 -
LL

0.4
Q1
0.2

075 95 115 135 11115 175 195 215

0,0
235 255 275 75 95 115 135 155 175 145 22 2
Time (s) Time- is)

1000

900 3.4 -

goo 3.2 - in oi

700 3.0
IM
600 2*8

500 Z6
Ln 400
CL
300
2.2
200
2.0
100
1.8 - Separator

75 95 115 135 155 175 195 215 235 255 275 75 95 115 135 155 175 195 215 235 255 275
Time (s) Time s)

12

3.3
E 10
3.2

0 3.1
(3) 1

3.01
0
a'. 4
#A 2.9
CL

rL 2.8
W
V)
2.7 '
75 95 115 135 155 175 195 215 235 255 275 75 95 115 135 155 175 195 215 235 255 275
Time (s) Time (s)

Figure D-6 - Summary of run LTG23W2

 D9

20

1.8
E
14 1.6
8 1.4
12
ci
1.2 -
io
1.0 -

0 0.8
LL 06

04

D2
0.0
850 900 950 1000 1050 1100 1150 850 00 95 aOf 5c
Time (s) Time (s

1000
16
900
3.4 A'
800
3.2
700
3.0
600
2.8
500
2.6
400
24
300
zz-
.s
200 2.0

100 Separator
0 1100 1150
850 900 950 1000 1050 1100 1150 850 900 950 1000 1050

Time (s) Time is)

E 3.2
x
14
31
12

lo
10

LL 6
2.9

2 2.8

0 950 900 950 1000 1050 11 OP

850 900 950 1000 1050 1 i0o 1150

Time (s) Time (s)

Figure D-7 - Summary of run LTG34W2

 DIO

E
x
1.7

tu / ftop +ft
m
cc 1.5
.: 1A -
2
LL 1,3

0 ri 20 40 60 80 100 120 140 160 180 200

1.0
0 20 40 60 80 CIO 1120 1.41) l6j 81
Time (s) Tim, ts)

1000r
900 3.4

800 3.2 -
Inlet
M
i 700 3.0 -A

600 2.8 -

500 2.6
(A
40 0 Z4
c.L
300 2.2

200 ZOA

100

0
1.8 Separator
0 20 40 60 80 100 120 140 160 180 200 0 20 40 60 80 100 120 140 16C, 180 200
Time (s) Time (s)

16 33

0 14
s 3.2
E
12 3.1

ci 10 3.0
w
8 2.9

06 2.8
iL
0
m4 2.7
0
CL
0) 2.6 -
Y)
00 20 40 60 80 100 120 140 160 180 200 0 20 40 60 80 100 120 140 160 180 200
Time (s) Time (s)

Figure D-8 - Summary of run LTG34W3

 DII

14

13
E co
04, E 05 -
x 12

8 0.4
64) lu
to m
0.3
Ix

o8 0
LL 0.2 -
7-

0.1
5-

4 0.0
100 200 300 400 500 Soo 700 800 900 100 200 300 400 500 600 700 80C

Time (s) Time (s)

1000
2,6 -
900
2.5 -
800 Inlet
700 2.4 A
co
600 2.3

5oo 2,2
,A
(D
400
0-
300 V"
2.0

200 Separator
100

0
100 200 300 400 500 600 700 800 900 100 200 300 400 500 600 700 800 900
Time (s) Time is)

3.2

5
E 16
X 3.1
14

S12-
3.0
lo

8
0
LL 6 2.9
CL
4
CL
4)
U) le.0 I I

01
100 200 300 400 500 600 700 800 900 100 200 300 400 Soo 600 700 80() 900

Time (s)
Time (s)

Figure D-9 - Summary of run LTG43WI

 DJ2

E E 05 -

8 0.4
ci
(D
m
cc 0.3

0
LL 02

0
0

C
4 oO L 1
95U 970 990 1010 103JD 1050 1070 1090 1110 1130 115C 950 970 990 1010 11030 1050 10-0 10-0 1110 113Z i
Time (s) Time is

1000
2.6
900

800 2.5

700 2.4 A
-"Or -O rN-e-,
Go - .
2.3 lk-; F
600
Soo 21

400 2.1
IL
300 2.0

200
191
Separartor
100
1.8 -

0
950 970 990 1010 1030 1050 1070 1090 1110 1130 1150 950 970 990 1010 IOSO 1050 1070 1090 1110 1130 1150

Time (s) Time s)

is 14 5
E
x 12 L 3.1

o 10 L)
3.0
8

06 sr
LL .-J 2.9
0; 0.
m4
0
CL 2
0) 2.8
U) I

950 970 990 1010 1030 1050 1070 1090 1110 1130 1150 950 970 990 1010 1030 1050 1070 1090 1110 1130 1150
Time (s) Time (s)

Figure D-10 - Summary of run LTG32WI

 EI

APPENDIX E

MODEL PREDICTION OF THE CONVENTIONAL PIGGING RUNS

 2

E. MODEL PREDICTION OF THE CONVENTIONAL PIGGING RUNS

The model prediction of the 13 conventional pigging runs are shown in this appendix. Table E. I
presents the prediction of initial conditions of the steady state gas-liquid flow, superficial
velocities and pig launching schedules of the runs. Figs E. I to E. 13 show the transient flow
prediction due to conventional pigging operations.

Table E. 1 - Conventional pigging results

RUN SUP. GAS SUP. LIQ. PRESSURE PIG SCHEDULE

CODE VELOCITY VELOCITY Inlet Separator Launching Arrival
(m/s) (m'/s) (Bara) (Bara) (s) (s)
Exp. Exp. Exp. Exp.

LPG3WI 2.6 0.16 2.3 1.9 510 550.6

LPG4WI 6.2 0.16 2.6 2.0 458 484.6

LPG3W2 2.2 0.48 2.5 2.0 288 322.0

LPG4W2 5.9 0.45 2.9 2.2 520 537.0

[-PG3W3 1.7 0.81 2.8 1.9 193 226.0

LPG4W3 4.0 0.77 3.2 2.0 265 286.0

HPG2W] 0.8 0.27 4.6 4.8 271 349.0

HGP3W 1 2.8 0.16 5.7 5.6 322 354.6

HPG4W] 5.9 0.24 6.0 5.8 118 136.0

HPG4W2 5.4 0.34 6.2 6.1 123 143.4

HPG2W3 0.5 0.99 6.1 6.2 440 484.0

HPG3W3 2.5 0.94 6.3 6.3 508 533.3

6.4 0.67 6.8 6.7 94 107.6

HPG4W3
 E3

If, --. --
goo -
E 25
Soo x

700
20
goo

a 500 w 15
E
0 400- 0
IL lo
0 300
co
W
2000
5

loo 500 520 540 560 580 600 620 640

0
500 520 540 560 580 Soo 132" b43
Time (s) Ti me (s

2,8 6

2.7 .e
ei
E5
2,6

2.5

2.4
Ln
fcs 3
2.3
CL 0
- 2.2 LL

2.1

2.0

19 0
500 520 540 560 580 600 620 640 500 520 540 560 580 Soo 620 S40

Time (s) Time (,

12

10

17 N

Lc: 4

0
51 000 520 15 530 3JO 540 4t 610

TimL- s)

Figure E-I - Model results of run LPG3WI

 E4

I lk r-------- -- -- -- ---

900 - 25

Soo,
20-
700

Soo

500
E
0 400
z 10.
0 300-

200 5
of
1001

0
450 470 490 510 530 550 570 590 450 470 490 510 5-10 5,0 57C, 590

Time (s) Time (s)

33.2 :-
6-
co
3.1 E
3.0 5-
2.9

2.8 - 4
13)
2.7 -
2.6 3
2.5 0
2.4 2

2,3
lj
2.2 CL 1
4)
2.1 cn

450 470 490 510 530 550 570 590 450 470 490 510 530 550 570 590

Time (s) Time (s)

IA

12

10

do
0
.0)

Pe
z:
2

- 41
-I'-eI 458 X 46465 471.30 47795 48460.

Time (s)

Figure E-2 - Model results of run LPG4WI

 E5

10fk

900
i; E
9 800,
X
M
14
Too
12-
800

500 10-

FI
0 400,
.2a
LL
0 300 06
to
200 4-
CL
4)
2-
100 in

290 0 ri 310 3,20 330 340 350 360 370 380 280 290 300 310 320 330 4" 1-50 35c, : -D --sc,
280 33
Time (s) Time (s)

36
g7
i; 5
3.4
E
3.2 6

3.0

2-8
41-
2.6

2.4 3

2.2 - 2'
2.0
4) 11
1.8 0

1,6 0 280
300 310 320 330 340 350 360 370 380 280 290 300 310 320 330 ', 0 50 360 370
280 290
Time (s)
Time (s)

10
9

04

Zito
AR; 2

"Wo

288 0 296 5 305 0

Time (s)

Figure E-3 Model results of run LPG3W2

-
 E6

11)06

900

Boo-

700

600- 40

50o
E 30-
0 400-
v
0 300 20 -
w -
200
cc
100 -

550 560 570 580 590 600 500 510 520 530 540 550 560 570 580 590 600
5rjo 51ri ') 205" C) 540
Time (s) Time (s)

40

3.8
E
3.6 x
3.4

3.2 41,

3.0 :
03
CL
U.
2.8
2'

2.6
CL

24
0
580 590 600 500 510 520 530 540 550 560 570 580 590 600
500 510 520 530 540 550 560 570
Time (s) Time (s)

20

10

0 I
52000 52425 52850 75- 53, 00

, 1-55, Time (s)

Figure E-4 Model results of run LPG4W2

-
 E7

1000

900 16
E 800-
6) U

700
12
600
10
a 500 it
E : 8
400- ,
0
t:
0 300 -
.2
LL 6

200

DO

0
170 10 190 200 210 220 230 240 250
0
260 270 170 180 190 200 M 220 230 240
Time (s) Time (s)

7
3.2 -
.e
m
3,1 E6
3.0 -
x
5
Co 2.9 -

2.8 -
(A
2.7

IL 2.6 0
LL
E 2.5

2.4

2.3 l. ,
2,2 0
170 180 190 200 210 220 230 240 250 260 270 170 180 190 200 210 220 230 240 250 260 270
Time Time (s)

10

7t
7

ol 6
-11, ,tI
5
>
"j; CD 4
iL
3

0
193.00 201 25 209.50 217 75 22SO01

Time (s)

Figure E-5 - Model results of run LPG3W3

 E8

0 9"

900

800 x
30 -
700

800

500 Ir 20

0 400
.2
0 3DO LL
10
200

loo, W

0 0
250 260 270 280 290 300 310 320 330 340 350 250 260 270 280 290 300 310 '120 330 340 50
21
Time (s) Time (s)

Z8-
36
34 E
X7-
2

3.0

2.8
Lo
2.6

IL 2.4 0
3
-E 2.2
2
2O
cL
1,8 V)

250 260 270 280 290 300 310 320 330 3O 350
ol
250 260 270
.
280
L-J,
290 300 310 3320 330 340 350

Tone (-, ) Time is)

, C)

16

14

12

tM 8

-let,
4

0
2 5 266 271 274 2 77 280 28-1 286

Time (s)

Figure E-6 - Model results of run LPG4W3

 E9

900
E6-
800 x
700 5-

600-
4
500

0 400 3-
.2
LL
300
to
m2
200

II-
250 3,50 400 450 500
Iloo 250 300 350

Time (s) Time (s)

5.1

50 - E25

4.9 -

4.8 2.0!
41 L
4.6 1.5
4.5
0
4.4 1.0
4.3

4.2 0.5

4.1 W

0,01
250 300 350 400 450 500 250 300 350 400 450 500

Time (s) Time (s)

2-

CL

271.0 290.5 310.0 329.5 34-

Time (s)

Figure E-7 - Model results of run HPG2WI

 EIO

900 -
P
800- x 14
700
12 -
,:;
600-
10
500

400
LL
CO 300 - W
m
200
CL
4) 2
cn

'00 320 4 Ol 150 380 400 420 440 320, : 4C 60 280 42"
-100
Time (s) Time ts)

6" 4-
5
6.2 E
6.1
3-
6.0

5.9

5.9 2

IL 5.7 0
LL
5.6

5.5

54

5.3 0300
00 320 340 360 380 400 420 440 320 340 360 380 4)0 310 44,7)

Time (s) Time (s)

12

10
9
ca 8
7
0
7@ 6
>5

CL 4
"1
3 1W I
2
I

322- N 131 15 Z40 IN 1,4945 60

58
Time (s)

Figure E-8 - Model results of run HPG3W I

 Ell

900 30

8oo-
25
700

600 - 20-

a 500 w
15- .
0 400-
V .2
LL
300 - tA 10
200
-------------
CL

fj 0
100 11 ri 12 0 130 1140 150 160 170 180 190 200 100 110 120 130 140 150

Time (s) Time (s)

67

65-

6.3
4

6.1

5.9

5.7 LL

53

51
100 ilo 120 M 140 150 160 170 180 190 200 100 110 120 130 140 150 160 , -LI 1BC, 190 2,io

Time (s) Time (s)

14

12
do

10-
0

CL

118 120 122 124 12F 126 lie 1,12 2-4

T: ri- s)

Figure E-9 Model results of run HPG4W I

-
 E12

900
25
800

700 20-

600
C:
wW 15 -
500

7 400
U- lo
-
'00 LA

200 5
CL
100-

00 W 150
100 110 120 130 140 150 160 170 180 190 200 100 110 120 130

Time (s) T nne is

68
6
6
6,5
M
Co 6.4 2 1,
3
6.3
6.2 CA:
V) 6.1 I
2
CL
0
6.0- lz ,
5.9
58
57
16

200 loo 110 120 130 140 150 " 6c 1_0 i8c 19Q 200
100 110 120 130 140 15() 1rso Wo 180 190

(s) Time (s)

Time

14

12

lo

8
0
>

CL
,j, -,
"4p*- I.k'-
--I?

2-

12' 0 128 1

Time fs)

Figure E-10 Model results of run HPG4W2

-
 E13

1.rf, , - -- ---- - -- I----- ----

7
900 tj Z4
E
2.2
800
2-0 -
700

600
c
w 1.4
500
1.2
F, 400 -
z
1.0
0 0.8
co 300 m
w 200 0 0.6-
ci 0.4
e -
loci U) 0.2
-

400 420 440 460 480 Soo 520 540 400 420 "0 460
Time (s) Ti

70- -----
-1
68

6,6
0
6.4

6.2
4)

6.0

5.8
0
LL
56-

CL
52
0
50
400 420 440 460 480 500 520 540 400 420 440 460 500 520 1-4C)
-180
Time (s) Time s)

1.7

1.5
Of > 1.4
SM
IL 1.3

12

II
j
4Z
440 451 462 473 484

Time (s)

Figure E-11 - Model results of run HPG2W3

 E 14

900 -5
iZ3 10- k
8oo
E
0) x
700 o 8-
R 7
600 -

500 M. J\o

400
0
4
0 300-
M
200
CL 2
w
W 1. -

400 410 420 4330 440 450 460 470 480 490 500 400 '1 C- 12) 44-C .45o 6C
Time (s)
Tirne (s

!-
6.8 co
E
6.7
6.6
6.5
6.4
6.3

L
6,0 C
59
58

'400 410 420 430 440 450 460 470 480 490 500 400 410 420 430 "0 450 4 60 470 -180 49 Cl 500
Time (s) Time (s)

V,

ej
0

\L.;
CL

7" Zi
I
408.000 414.320 420650 4261970 433 W
47
Time (s)

Figure E-12 - Model results of run HPG3W3

 EJ5

goo - so-

8oo-
40 -
700

Z; 800-
30 -
500

C> 400 20
t:
0 300
-

2001 10
100

1Ili 1,ri C, 110 120 130 140 150 90 100 110 120

Time (s)

9
Z
78

76- E
x 7
74 -

7.2 9 6
co
7.0 5
6.8 fw
4
6.6-
CL 2
6.4- .
LL 3
C
6.2 2
r, 0 CL

0 - .
140 150 90 100 lio 120 1110 110 15c,
90 100 110 120 130

Time (s) Time (s)

4 20
4.
Y, j,

10
CL

- le 0
I 940 974 2
4il
Tirno- (s)

Figure E-13 Model results of run HPG4W3

-
 F]

APPENDIX F

NIODEL PREDICTION OF THE GAS PUMP WITHOUT PIG

,
 F2

F. MODEL PREDICTION OF THE GAS PUMP WITHOUT PIG

The model predictions of the six gas pump runs without pig are shown in this appendix. Table F. I
presentsthe initial conditions of the test section, the gas injection conditions and the efficiency of
liquid transportation. Figs F. I to F.6 show the transient flow behaviour predicted by the model
when gas is injected, without pig, in a pipeline-riser system full of liquid.

Table F. 1 - Model prediction of the gas pump without pig

RUN INITIAL GAS INJECTED EFFICIENCY

CODE INLET Volume Time M
PRESSURE (SM3) (s)

(Bara)

LGPI 2.52 0.233 29 87.7

LGP2 2.51 0.521 29 91.9

LGP3 2.60 0.721 28 92.7

HGPI 6.03 0.777* 55 83.6

HGP2 5.91 0.750* 44 84.1

HGP3 5.90 0.627 28 87.3

* assumed value (problems in gas measurement)
 F3

900 -

800,
8
e
700

600

500 v0

400-
r
0
300 -

LA 200 ,
.
100

C,
150 160 170 180 190 200
Time (s)

1.0
Z9

28 -
CD 2.7 -
2.6
2.5
2.4
IL 2.3
22 go
21
20

150 160 170 180 190 200

Time (s)

- 6
12 E

10

8
m
3
6
0
LL 21-
V, 4
m

2 W

0 160 170 180

0
180 190 200 150
150 160 170
Time (s) I
Time (s)

Figure F-I - Model results of run LGPI

 F4

r. ry, r

900

800

700

600 -

500
E 400-
0
0
300.-
IA 200 " -2
.
w
100 - -"

0
40 50 60 70 so 90
Time (s)

36

3.4

3
-2

3.0

Ln
2.8
1 41

-,4
Ln
2
IL
2.6

2.4
10
AT

2.0

1a
30 40 50 60 70 80 90
Time (s)

40
9
8

7-
30
6-

20
4

10 2

0 0
40 so 60 70 so 90 30 40 50 60 _0 a0 9-

Time (s) Time (s)

Figure F-2 - Model results of run LGP2

 F5

900

soo -

700

600 -

500
I
400 10 r,
loo -

200

100

I 10 40 50 60 70 so 90

Time (s)

40

' 8

3.4
3.2

3.0

2.8
CL
2.6

2A

2.2

2.0

30 40 50 60 70 80 90
Time

14
30 L
E
12
25

101-
cs C5
201-
4)
m a'-

15
0 06
LL.
10-
4

CL 5 CL
2-

06 I
30 40 50 60 70 80 90 30 40 50 60 70 9c
Time (s) Time (s)

Figure F-3 - Model results of run LGP3

 T-6

900
P
-Z 800
8
IUU f

Soo r
a
500
0 400
0
300

CO 200

100-
-1

450 460 470 480 490 500 510 520

Time (s)

61

62

61

O.v

V) 5.9
LA

5,7

56

55 Ae
4T
500 510 520 1-0
450 460 470 480 490

Time (s)

3-
3
OW

21
:
0 0
LL
V)
m

-450 460 470 480 49C 2r

4 460 470 480 490 500 sio 520

Time (s) Time (s)

Figure F-4 - Model results of run HGPI

 F7

900-

wo -
s
IOU e
600

500

400
0
ED 300 -
i-
LA 200
.

40 50 60 70 so
Time (s)

6.2 4
CO 6.1
6.0
5.9
5.8 5 v'
5.7
56

40 so 60 70 80

Time (s)

5
0
9
4

3
02
LL
in
2

70 80 40 50 60
40 50 60

Time (s) Time (s)

Figure F-5 - Model results of run HGP2

 F8

ork
4,
900

goo-
700

goo

500
400,,

Co 300 -

0 200 ,
loo

0
120 130 140 150 160
Timr- (s)

T2

7.0

CD 6.6

6.4

6.2

IL 6.0
Z '\ '' *-

5.8

56

120 1A 140 150 160

Tone (s)

9
E
10
x 9
7, -

6
7

w 6 5
:
5- 4
0
LL 4
0 3
ra 3
(D
2
2

0
120 130 140 150 160 120 130 140
Time (s) Time (s)

Figure F-6 - Model results of run HGP3

 (f I

APPENDIX G

NXIODEL PREDICTION OF THE GAS PUMP NVITH PIG

.
 GI
-

G. MODEL PREDICTION OF THE GAS PUMP WITH PIG

The model prediction of the three gas pump runs with pig is shown in this appendix. Table G. I
presents the initial conditions of the test section, the gas and pig injection schedules and the
predicted efficiency of liquid transportation. Figs G. I to G.3 show the predicted transient flow
behaviour causedby a gas and pig injection in a pipeline-riser system full of liquid.

Table G. 1 - Model results of the gas pump with pig

RUN INITIAL PRESSURE VOLUME OF PIG SCHEDULE EFFICIENCY

CODE Inlet Separator GAS INJECTED Launching Arrival M
(SM3) (s)
(Bara) (Bara) (s)

LPPIA 2.4 1.5 0.466 305 343.6 99.18

LGPP2 2.4 1.5 0.641 242 270.2 98.62

LGPP3 2.5 1.6 0.824 195 217.0 99.7

 G3

UUV
8
700 e

ko
600- r-

500
E
0 400
0
03 300

200
A
00 310 30 330 340 350 360
-1;5q, -ev
Time (s)

3.1

2.9

2.7

2.5

2.3
5 0,
21
I'

1.7

.-

1,20 330 340 350 360

Tione (-,

40,

E6
* 1-
30

41

W 20
3
0
EL

lo

0 0
320 330 340 350 360 300 310 320 330 240
300 310

Time ts T; mp s

Figure G-1 - Model results of run LPPIA

 G4

r, rr, ,- -- :7---I----- ---- - --

900 -
P
Soo

700

Soo-
ol
a
Soo

400
0
Co 300 -

200
.0
100
240 250 260 270 280 290
Timf-, (s)

4-
3.2

3.0 cp

2.8

2.6

2.4

22

2,0 4p

18

230 240 250 260 270 280 290

Time (s)

60 10

E.

7
401
6-

30 5

4
20 3-

2,
0.10

0
280 290 230 240 250 260 270 28C
2 240 250 260 270
19
Time (s)
Time (s)

Figure G-2 - Model results of run LGPP2

 G5

900

Z 800-
a)
700

Boo -

500

400
v
0
(D 300-

tA 200
3
100 -

0
80 190
7-
200 210 220 230 24A
Time (s)

3.3 -

Ln
2.8 -
a.

21

i8o 190 AIC, 210 220 230 240

-%.

Tmic (s)

80
P E
70 1
x X 12
0
6011

50'-
Ix
3: 40
0 06
30 U-
4
0 20-
2

0
230 240 180 190 200 210
iso 190 200 210 220 12e
Time (s)
Time (s)

Figure G-3 - Model results of run LGPP3

 HI

APPENDIX H

MODEL PREDICTION OF THE TRANSIENT WAS

 H2

H. MODEL PREDICTION OF TRANSIENT RUNS

The model prediction results of the ten transient runs is shown in this appendix. Table H. I
presents the predicted results for initial and final conditions of the test section. Figs H. I to H. 10
show the predicted transient flow behaviour causedby a change of boundary conditions.

Table H. 1 - Model results of the transient runs

RUN INITIAL PRESSURE FINAL PRESSURE

CODE Inlet Separator Inlet Separator
(Bara) (Bara) (Bara) (Bara

HTG I 2W3 5.78 4.65 5.80 4.75

HTG34WI 4.70 4.31 6.30 5.37

HTG34W2 7.00 5.88 5.60 4.90

HTG34W3 6.20 4.94 7.30 5.95

LTG12WI 2.80 1.80 2.48 2.00

LTG23W2 2.63 2.00 2.75 2.10

LTG. 334W2 2.70 2.05 3.20 2.20

LTG34W3 2.75 1.90 3.00 2.05

LTG43W] 2.60 2.05 2.40 2.00

LTG32WI 2.40 2.00 2.27 1.95

 H3

900

700
60
6oo - r- ,

500

0 400-
0
CO 300 -

tA 200 ,
w
loll -

70 80 90 100 110 120 130 140 150

Time (s)

6.0 - -- -- -- --

1
5.8 -
0)

5,7

5.6

50 60 70 80 90 100 110 120 130 140 150

Time (s)

j-

E
1.4

1.2

1.0

6 0
iL 0.6
- LL

0 0.4

0.2
(0

10 1
60 90 q() 100 110 120 130 140 150 50 60 10 so 90 100 110 120 1230 140 150
10
Time s Tf-Ip s

Figure H-1 - Model results of run HTG12W3

 H4

700

600 -
cc
0 500

400
V
(a 300 -

tA 200

300 400 500 600 700 41,

Time (s)

6,9
6.7
6.5
6,3
6.1
49
5.9
5.7
ell
5.5
5,3
5.1
4.9
47
4S

400 500 600 700

Time.

E
16 0,9
x
0.8''
14
0.71
12,1
0.6,
W lo

o 8'
0 4'
. .
6
0.3
4 0.2
0. ci
w q
U) 01

0.0
300 400 500 600 700 300 4 00 5130 600 700

Time (s) Tini- (s)

Figure H-2 - Model results of run HTG34WI

 H-5

" Y,

900

800

700

soo
0 500
P
400

to 300 -

200
'b
100
9
C,
10 70 90 110 1110 150 170 190 210 23.0 250

Time (s)

72

7.0 v

6.8

6.6
-
CD
6A

V) 6.2
vi
ID
6.0

d) 5.8

5.6

14

70 90 110 130 150 170 190 210 230 250 %"

.

Time (s)

22 5
E 20 E
x
4
0

14 L 3
121- Ir

0 101-
L-L
8 .22
0 6

0. 4-
v

0
170 190 210 230 250 50 70 90 110 130 150 170 190 210 2,0 25'
50 90 110 150
N
Time (s) Tirn- (s)

Figure H-3 - Model results of run HTG34W2

 H6

0 rf)

900

uuu

700, $1
600-

500
E
0
z 400 15
0
ca
300 -

200

100 -

0
0 20 40 60 80 100 120 140
Time (s)

75

7 ', A-

71

6.7
""/__\
' ;e.
6.5

20 40 b,01 so 100 120 140

Time (s)

i 4
16,,
E
i
14 r x
12
0

lot

6 LL
4

cu

20 40 60 80 100 120 140 40 8 80 1Cl 12r, 140

Time (s) Time (s)

Figure H-4 - Model results of run HTG34W3

 H

I OC
900

800

700

600-
a "v
500: -
0 400-
0 0
cc 300
200

100 - $v -, v

0 100 200 300 400 500 600 700 800 900 1000

Time (s)

3.1

2.9

2.8

2.7 0
V) A

2.6

2.5

23 -C -
10 200 300 500 600 700 Boo 9 l0 00
'100

Time (s)

0005
0.0030

0 0.004 -4 0.0025
CO
E

0.0020
w 0.003

0 0o 0015
LL i
0.002
0.0010
CL
fi) 0001 (n
0.0005

0Oclo
C too 200 300 rk II
400 500 600 700 800 900 1000
0.0000
0 100 200 300 400 500

Trn-- (sl
000 700 goo 900 0

Time js

Figure H-5 Model results of run LTG12WI

-
 HS

1000---
900

700

C
Soo r
-a -' -

0 50o

400

cc 300
w
200
.0
Ir
100-

0L I-
95 . 115 1315 155 175 195 215 235 255 275
75
Time (s)

2,9

28

2.7

2.6
db

25
1114
241- -- -- --
T5 95 i15 1-15 155 175 195 215 235 255 275

Time (, )
-,

Ln
3.0
E
x 2.5
0

2.0

1.5,
0
UL

I
0 V)

UM
255 275 75 95 115 135 155 175
95 115 135 155 175 195 215 235
Tirre (s) Tim- si

Figure H-6 - Model results of run LTG23W2

 H9

900

700
.0.11-17-
Boo rl - " - 't -
01

500

400-

300

700

100

900 950 1000 1050 1100 1150

'rime (s)

3.3
f-rwr\
3.2 - IV

31 -
! Cl
3.0 -- i , .,,

2.9
Ln
2.8 4
2.7
2.6
2,5
24
3
Ak/

0 950 1000 1050 1100 1150

Time (s)

20

18
E1-
16
1.6
14
cj 1.4-
12
1.2 I,
m,
10
1.0
8
-2 0.8
LL
6 0.6

4 0.4-

0.2

0,0
900 950 1000 DO
900 950 1000 1050 1100 1150

Time (s) Time (s)

Figure H-7 Model results of run LTG34W2

-
 HIO

1orio

9 DO -

Uuu -

700
, I ai "-
600 10
7j)
0 500
E
0 400
0
300

IA 200

100
'-ir

0
20 40 60 80 100 120 140 160 SO 200

Time (s)

3.2

11

3.0-

2.9

2.8
CL

2.7
"E

2.6

2.5
20 40 60 80 100 120 140 160 180 200
Time (s-)

E E
10
X
0 91 -0
81-
7 2
6

0 5
iL
4L .2
U.
(9 3

2
ci

I
20 40 60 80 100 120 140 160 180 200 20 40 60 80 100 120 140 160 180 200
Time (s) Time (s)

Figure H-8 - Model results of run LTG34W3

 HII

Boo

700

6oo -
0 5oo

400 - 100'
1100-

4)
200
'(A
ft
100 -

100 200 Mo 400 500 Soo 700 800 900

Time (s)

2.6

fl,
fi
im
Z. D
G)
L4

Ci
1 Ci
74
0.

2.3

22
'100 200 300 loo 500 600 700 800 900

Time (,, )

0,7
E 20
x
-0 18 0,6

16 ;:
0.5
14
Ir 12 0.4

0 10,
03

6 02

4
0.1
V)

0.0
100 200 400 500 600 700 800 900 100 200 300 41JO 500 600 loo 800 900
1100
Time (s) Time (s)

Figure H-9 - Model results of run LTG43WI

 H12

7' 800 -

700 -

Boo
ol
D 50o S'll
E
400 .0 01

Co 300 -

ILA 200

100

0. I. I11
950 970 990 1010 103ri 1050 1070 1090 1110 1,3() l1r, ()
Time (s)

243

2.41

-@'2 39
.
S2.37

2.35
,..
2-33 C, -,
-..
101 ;
CL
2.31 -

227

225

50 970 490 110 030 1050 1070 1090 1110 1130 1150

Tirne (s)

14 0,6

12, E 0.5
x
io
0.4

W 0.3
6
0 0
AL LL. 0.2
0 41 C
m

9 -1C, 990 1010 D

1,07, 1050 1070 1090 1110 1130 li5O

Tirre s) Tn),- (s)

Figure H-10 - Model results of run LTG32WI

 11

APPENDIX I

STEADY-STATE SLUG FLOW MODEt,

 12

1. STEADY-STATE SLUG FLOW MODEL

For slug flow the average liquid flow rate is given by:

Is if
v,al = alsvs + alf vlf (1- 1)
11
The vif in eq. (I- I) can be eliminated using a liquid continuity balance relative to a moving co-
ordinate system that travels with the slug translational velocity v,-

alf (vt + vlf )=a,, (v, )

- v,

The liquid hold-up for a slug unit can be defined as

ls 1f
als + a,,
11

Solving eq. 1-3 for alf and combining the result with eq. 1-1 and 1-2 yields

vlq vlq -%(vt - 10

The translational velocity is correlated in terms of the slug velocity v, as follows-

Vt : -- (o'v + Vd

Substituting v, and vlal = v, - vgag into eq. 1-4 yields-

V9ag - Vd(aL. )
-'61
Vs
-
1
-a, C,, + (C,, - I)als

Eq. 1-6 allows the calculation of the average slug velocity for given liquid hold-up for a slug unit
and gas flow is
rate. a,, correlated in terms of slug velocity by Gregory et al. (1978) as follows:

aL,= 1.38
(1-7)
I+ (vs / 8.66)

for a,, > 0.48; otherwise, a minimum value of 0.48 was used (Bamea and Brauner - 1985).
 Ll

The drift velocity for vertical flow was determined from Nicklin et al. (1962)

vd = 0.35 (1-8)

For inclined flow, the value was multiplied by sin 0.

A value of 1.2 was taken for the constant C,, (Nicholson et al. 1978).

Oncev, is known, v, can readily be obtainedfrom v, = ft, - v9(I -ad] / al.

Applying the momentum balance on the film region yields-

wgr
pg (v [pg
pg -c v2+ -v2+ g sinE)l
agf
9f 9f
LL JL a9f

Cwf Pg
2 (Vgf )2
Vif Cif - -v If +[Plgs. in E)] =0
) (I - agf )-
-agf

Eq. (1-9) is solved by a trial-and-error procedure for the gas volume fraction in the film zone ae.
is
For a given ae, the liquid film velocity v1f calculated using eq. 1-2. The gas velocity in the film
zone, ve, is calculated using the following mass balance:

(I )
v,,a,, - v,, - a. = V, (I-JO)

The shear coefficients used here are also the sameas for stratified or annular flow.

1,,
The ratios 'I and If,/ can be calculated from eq. 1-3.
 11
.

APPENDIX J

VELOCITY COEFFICIENTS
 12
.

J. VELOCITY COEFFICIENTS

The momentum equations in the Basic Step are:

9 Gas momentum equation

I n
On+l Ci
(pOn+ll +( 2(v n+l n+l )-(v n-vn )I+ gsinE)
j+l - pi vn _Vn -v =0
(p n991g
Az a 9)i+-,
g I
22 2

* Liquid momentumequation
n
On+11 On+l (, +K) CW1 IVI 1[2v,
n n+l nI
(pj+l - pi V,
ni n
Az )j+
I -ag
222

n
P9 1[2( n+l
n n_n n+l) n_ n) (J-2)
vg V, vg _ V, vg V, +gsinO=O
-Ci
g)pl
2

Equations(J- I) and (J-2) can be rewritten as:

On+ll On+l (J-3)

I
GI(pj+1 )+G2[2 (Vn+l
_ Vn+l )-G3]+G4=0
-pj gI

On+11 On+l n+l L3) L4 [2(v,n+l n+l

)-L5]+G 0 (J-4)
I, (pi
- pi )+1.2(2 V, _ _ _ v, 4:: 7-
'-
+1

I cnivn-v I'll
G2 G3 (Vn Vn) G4 = gSin(3
where G nI 1. 91 91,_
IL, (P
(p IAz
Az a g)n
i+- 22 j+
j+-22 j+- 2

(I + K) cw", Pg n).
nj, L5 V,
1-2 == vn-v =(n_ vg
vnL
14i( == cn 91
(I - ag)p,
I-a X)n I 2
P-
2
 3

Equations (J-3) and Q-4) can be rewritten as:

G, (p On+11 On+])+2G n+l n+l (J-5)

o j+l _p i2921234 v _Gv _GG +G =0

On+I On+I n+I n+I

o L, (pl+I -pj )-2L4Vg +(L2+L4)Vl -
L2L3 +L4L, +G4 =o (J-6)

or
On-0 On+l
G, (pj+l pj + 2G vn+l - G2 v'+l Q-7)
2g1 -G5 =0

On+11 On+l n+l n+l

9 L, (pj+l pj )-2L4Vg +L6VI L7 :-- 0
- (J-8)

G5 G2G3 G4 L6 = L2 + L4 L7 L2 L3
where = - and = - LJ,
7
-G 4'

Multiplying eq. (J-7) by L4and eq. (J-8) by G2yields-

On+II On+I
9 G, L4 (Pj+I P, )+2L4Cr2vg n+I n+I
(J-9)
-L4G2v, -L4G, =O

On+II On+I n+I n+I

LIG 2 (Pj+I )-2L Gv G2 L
- Pj 42g6217 +L Gv _ (1- 10)

Adding eq. Q-9) to eq. (J- 10) yields:

O. On+l )+(L6G.
(GIL4+LIG, oj+, n+ll )Vn+l (L4G5 + G2 L7)
-pj -L4G. 1-

Eq. (J- 11) can be rewtitten as-

On+ll On+l n+l

L8(pj+l -pj )+Lgvl -L, 0=0 (J- 12)

L4 LA
where L, = G, +LIG, Lq = L6G, -L4G., y LIO = +G2L7.

Multiplying eq. (J-7) by L6yields:

On+11 On+l I "+I I

GL-I )+2L6G2Vn+ - L6G2 VI L 0 (J- 13)
1 6(Pl+ -pj 9A-::::

Adding eq. (J- 13) to eq. (J- 10) Yields-

On+I ort+l ?I'+ I10 (1,665

)+2(L6(; v +G 2
((i, I, 2-L4G2)g -
6+LIG2)(pj',,
 14
,

Eq. (J-14) can be rewritten as:

On+II On+I n+I

G6 (pj+, Pj )+G7Vg 15)
-Gg =O

G6 G7 G8
where =GIL6+LG2. 2(L6G2 ), = LA +G2L7.
= -L4G2

Equations (J- 12) and Q- 15) can be rewritten respectively as-

nf1( Onl+1 "2n+1)

v, =EI pi pl
+F (J-16)

(P, 'I+n
n+l 41 On +I
)+F 17)
=E pj
919

L8 G6 G8
Lio.
where E F
I
E =-- and F, -
Lq Lq 9 G7 G7