Professional Documents
Culture Documents
FLOWLINES
by
Luis E. Zerpa
c Copyright by Luis E. Zerpa, 2013
in partial fulfillment of the requirements for the degree of Doctor of Philosophy (Petroleum Engi-
neering).
Golden, Colorado
Date
Signed:
Luis E. Zerpa
Signed:
Dr. Hossein Kazemi
Thesis Advisor
Signed:
Dr. Carolyn A. Koh
Thesis Advisor
Golden, Colorado
Date
Signed:
Dr. William Fleckenstein
Professor and Head
Department of Petroleum Engineering
ii
ABSTRACT
The oil and gas industry is facing very challenging production issues with offshore explorations
in deeper and colder waters. Longer subsea tiebacks will be required to transport hydrocarbon fluids
from the wellhead to production and processing platforms. The formation of solid deposits, such
as gas hydrates, waxes, asphaltenes and scale, may plug the flowlines, preventing production and
generating a safety hazard. The flow assurance of the produced hydrocarbon stream is a technical
discipline that focuses on the design of facilities and procedures for the uninterrupted transport of
reservoir fluids from the reservoir to the point of sale. The rapid formation of gas hydrates, which is
promoted by typical high pressure/low temperature operation conditions in deep subsea facilities,
is considered one of the most challenging flow assurance problems. A transient gas hydrate model,
that predicts when and where hydrate plugs will form in flowlines, will have significant utility for the
flow assurance engineers in the oil and gas industry. The Colorado School of Mines Hydrate Kinetics
model (CSMHyK) was specially designed to predict hydrate formation in oil-dominated systems.
The objective of this research work is to develop a comprehensive hydrate model, extending and
improving the CSMHyK model for the prediction of hydrate formation and transportability in oil,
water and gas-dominated systems. The mechanisms of hydrate formation and transportability in
pipelines is studied through the analysis of experimental data obtained at the Center for Hydrate
Research laboratory of the Colorado School of Mines and two large scale flow loops (ExxonMobil
and Tulsa University flow loops). A set of conceptual pictures is developed to explain the physical
phenomena of gas hydrate formation in flowlines. The mathematical models developed in this work
represent a significant advancement for the prediction of hydrate plugging risk in the pipelines of
oil and gas transport facilities, and can be used as a tool to design flow assurance strategies. These
models improve our capability to predict hydrate formation, by considering dynamic aggregation
phenomena in oil-dominated systems, flow regime transition in high water cut systems, and hydrate
iii
TABLE OF CONTENTS
ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii
NOMENCLATURE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xviii
ACKNOWLEDGMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxvi
DEDICATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxviii
CHAPTER 1 INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.4 Final system design and operation procedures for flow assurance . . . . . . . 14
2.1 Surfactants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
iv
2.1.1 Effects of surfactants in gas hydrate formation . . . . . . . . . . . . . . . . . 31
2.2 Emulsions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.4.2 Extension of a simple hydrodynamic slug flow model for transient hydrate
kinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
3.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
v
4.1 ExxonMobil flow loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
4.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
5.2.2 Implementation of the model into a transient multiphase flow simulator . . . 106
5.3.2 Implementation of the model into a transient multiphase flow simulator . . . 116
vi
6.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
vii
LIST OF FIGURES
Figure 1.1 Wax deposit on a pipe wall narrowing conduit diameter and modifying wall
roughness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Figure 1.3 Common gas hydrate structures; adapted from Sloan (2011, p. 3). . . . . . . . . 8
Figure 1.5 Pressure versus temperature diagram showing the WAT curve (green dashed
curve), upper and lower asphaltene precipitation loci (red dash-dot curves)
and hydrate equilibrium curve (blue continuous curve) from the
characterization of reservoir fluids; adapted from Ayers (2003). . . . . . . . . . 10
Figure 1.6 Pressure versus temperature diagram showing a typical subsea flowline
operation conditions curve, the WAT curve, the lower asphaltene precipitation
locus and hydrate equilibrium curves; adapted from Ayers (2003). . . . . . . . . 11
Figure 1.7 Pressure versus temperature diagram showing a typical subsea flowline
operation conditions curve and hydrate equilibrium curves at different
methanol concentrations; adapted from Sloan and Koh (2008c, p. 646). Points
on the flowline operation conditions curve indicate distance from wellhead in
kilometers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Figure 1.8 Conceptual model for hydrate formation in oil-dominated multiphase flow
systems consisting of gas, oil, and water; adapted from Turner (2005), with
input from J. Abrahamson (U. Canterbury, Christchurch, NZ). . . . . . . . . . 15
Figure 1.9 Bishnoi’s laboratory data used to regress the intrinsic kinetic rate constants . . 17
Figure 1.10 Universal curve of turbulent flow sub-ranges in terms of Weber and Reynolds
dimensionless numbers that affect mean droplet size in an emulsion, and
experimental data obtained in an autoclave mixing cell and a large scale flow
loop, using different crude oils . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Figure 1.11 Illustration of a water droplet with a surrounding hydrate shell showing the
boundary layer around the spherical particle. . . . . . . . . . . . . . . . . . . . 20
viii
Figure 1.13 Illustration of a hydrate particle with a surrounding water film showing the
boundary layer around the spherical particle. . . . . . . . . . . . . . . . . . . . 23
Figure 2.1 Conceptual model for hydrate formation in a multiphase flow systems
consisting of water, oil, and gas; adapted from Turner (2005). . . . . . . . . . . 27
Figure 2.3 Adsorption of a surfactant molecule at: (a) gas/liquid interface, (b)
liquid/liquid interface, (c) liquid/nonpolar-solid interface, (d)
liquid/polar-solid interface, and (e) liquid/polar-solid interface with increased
cationic surfactant concentration. . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Figure 2.7 Balance of interfacial tensions among the three phases at contact point,
defining contact angle of the liquid with the solid surface (left: solid is oil-wet;
right: solid is water-wet). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Figure 2.8 Summary of cohesion forces between cyclopentane hydrates particles in the
presence of Caratinga crude oil, measured by Dieker et al. (2009). . . . . . . . . 35
Figure 2.10 Schematic representation of phase behavior described by Winsor for different
values of the ratio of net interaction energy of the surfactant with the oil
phase to the net interaction energy of the surfactant to the aqueous phase . . . 38
Figure 2.13 PVM images of 68 vol% of water dispersed in crude oil: a) initial W/O
emulsion, b) hydrate formation around water droplets, c) large hydrate
agglomerates during hydrate dissociation, c) w/O/W emulsion after hydrate
dissociation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
Figure 3.1 Flow regimes for gas/liquid flow in a horizontal pipe p. 167]Brennen:2005. . . . 52
ix
Figure 3.2 Flow regimes for gas/liquid flow in a vertical pipe p. 170]Brennen:2005. . . . . 52
Figure 3.3 Flow regimes for slurry (solid particles of sand dispersed in water) flow in a
horizontal pipe p. 169]Brennen:2005. . . . . . . . . . . . . . . . . . . . . . . . . 53
Figure 3.4 Illustration of basic mass fields in a transient multiphase flow model, including
gas, oil and water as continuous phases, and dispersed phases within
continuous phases (i.e., oil and water droplets dispersed in continuous gas,
water droplets and gas bubbles dispersed in continuous oil, and oil droplets
and gas bubbles dispersed in continuous water). . . . . . . . . . . . . . . . . . . 61
Figure 3.5 Pipe segment divided into equally spaced sections, showing node numbering
convention. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Figure 3.6 Liquid holdup distribution along the pipeline length at four different times:
(a) t = 0.04 s, (b) t = 100 s, (c) t = 500 s, (d) t = 700 s. . . . . . . . . . . . . . 66
Figure 3.7 Conceptual picture used to define the phases and multiphase flow concepts. . . 67
Figure 3.8 Cross section of pipe with stratified smooth flow regime. . . . . . . . . . . . . . 70
Figure 3.9 Hydrate equilibrium curve for pure methane (V) and liquid water (L)
calculated with CSMGem. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
Figure 3.10 Distribution of water (continuous blue line) and hydrate (segmented green
line) holdup along the pipeline length at four different times, (a) 40 s, (b)
3230 s, (c) 4000 s, (d) 6000 s, showing the transition from stratified to slug
flow upon hydrate formation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
Figure 3.11 Comparison of predicted results and experimental data of fluid and ambient
temperature as function of time. . . . . . . . . . . . . . . . . . . . . . . . . . . 75
Figure 5.1 Conceptual model for hydrate formation in oil-dominated multiphase flow
systems consisting of gas, oil, and water; adapted from Turner (2005), with
input from J. Abrahamson (U. Canterbury, Christchurch, NZ). . . . . . . . . . 85
Figure 5.2 Illustration of the capillary liquid bridge between two hydrate particles . . . . . 87
x
Figure 5.4 Illustration of mass fields considered in the oil-dominated hydrate formation
model coupled with the OLGA
R
transient multiphase flow simulator,
including gas and oil as continuous phases, water and hydrate is assumed to
be dispersed in the oil phase, and oil and water may be dispersed as droplets
in the continuous gas phase. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
Figure 5.5 Plot of hydrate volume fraction as function of time, comparing experimental
data and oil-dominated model prediction for an ExxonMobil flow loop
constant pressure experiment with Conroe oil, 70% liquid loading, 37% water
cut, and 2.3 m/s fluid mixture velocity (Test # XoM-2004-2). . . . . . . . . . . 92
Figure 5.6 Plot of cumulative gas injection into the loop as function of time, comparing
experimental data and oil-dominated model prediction for a University of
Tulsa flow loop constant pressure experiment with Caratinga oil, 50% liquid
loading, 25% water cut, and 1.2 m/s fluid mixture velocity (Test # TU-2004-3). 92
Figure 5.7 Plot of flow loop pressure as function of time, comparing experimental data
and oil-dominated model prediction for an University of Tulsa flow loop
constant volume experiment with Troika oil, 50% liquid loading, 25% water
cut, and 1.2 m/s fluid mixture velocity (Test # TU-2004-4). . . . . . . . . . . . 93
Figure 5.8 Plot of fluid and ambient temperature as function of time, comparing
experimental data and oil-dominated model prediction for an ExxonMobil
flow loop experiment with Conroe oil, 50% liquid loading, 15% water cut, and
1 m/s fluid mixture velocity (Test # XoM-2011-9). . . . . . . . . . . . . . . . . 94
Figure 5.9 Plot of phases volume fraction as function of time, for an ExxonMobil flow
loop experiment with Conroe oil, 50% liquid loading, 15% water cut, and 1
m/s fluid mixture velocity. The estimated experimental hydrate volume
fraction is also shown for comparison with the model prediction (Test #
XoM-2011-9). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
Figure 5.10 Plot of fluid and ambient temperature as function of time, for an ExxonMobil
flow loop experiment with Conroe oil, 75% liquid loading, 15% water cut, and
1 m/s fluid mixture velocity (Test # XoM-2011-14). . . . . . . . . . . . . . . . 96
Figure 5.11 Plot of phases volume fraction as function of time, for an ExxonMobil flow
loop experiment with Conroe oil, 75% liquid loading, 15% water cut, and 1
m/s fluid mixture velocity. The estimated experimental hydrate volume
fraction is also shown for comparison with the model prediction (solid lines)
(Test # XoM-2011-14). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
Figure 5.12 Plot of hydrate volume fraction as function of time, comparing experimental
data with predictions using the oil-dominated hydrate formation model for an
ExxonMobil flow loop experiment with King Ranch condensate, 50% liquid
loading, 25% water cut, and 1 m/s fluid mixture velocity (Test #
XoM-2011-25). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
xi
Figure 5.13 Plot of hydrate volume fraction as function of time, comparing experimental
data and oil-dominated model prediction for an ExxonMobil flow loop
experiment with Conroe oil, 50% liquid loading, 75% water cut, and 1 m/s
fluid mixture velocity (Test # XoM-2011-23). . . . . . . . . . . . . . . . . . . . 98
Figure 5.15 Schematic illustration of gas-liquid flow regimes in horizontal pipe; adapted
from Brennen (2005). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
Figure 5.16 Schematic illustration of slurry flow (solids dispersion in liquid phase) regimes;
adapted from Brennen (2005) and Kitanovski and Poredos (2002). . . . . . . . 100
Figure 5.17 Typical behavior of measured pressure drop observed after hydrate formation
onset in high water cut flow loop experiments; adapted from Joshi et al.
(2011a). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
Figure 5.20 Cross section of pipe with stratified smooth flow regime. . . . . . . . . . . . . . 103
Figure 5.21 Illustration of gas bubbles entrainment in the liquid slug, due to energy
dissipation in the front of the liquid slug; adapted from Andreussi and
Bendiksen (1989). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
Figure 5.23 Plot of fluid and ambient temperature as function of time, comparing
experimental data with predictions using the model for hydrate formation in
water-dominated systems for an ExxonMobil flow loop experiment with 50%
liquid loading, 100% water cut, and 1 m/s fluid mixture velocity (Test #
XoM-2011-1). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
xii
Figure 5.24 Plot of gas injected into flow loop as function of time, comparing experimental
data with predictions using the model for hydrate formation in
water-dominated systems for an ExxonMobil flow loop experiment with 50%
liquid loading, 100% water cut, and 1 m/s fluid mixture velocity (Test #
XoM-2011-1). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
Figure 5.25 Plot of hydrate volume fraction as function of time, comparing experimental
data with predictions using the model for hydrate formation in
water-dominated systems for an ExxonMobil flow loop experiment with 50%
liquid loading, 100% water cut, and 1 m/s fluid mixture velocity (Test #
XoM-2011-1), also showing the calculated Φtransition . . . . . . . . . . . . . . . . 110
Figure 5.26 Plot of fluid and ambient temperature as function of time, comparing
experimental data with predictions using the model for hydrate formation in
water-dominated systems for an ExxonMobil flow loop experiment with 50%
liquid loading, 100% water cut, and 2.5 m/s fluid mixture velocity (Test #
XoM-2010-2). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
Figure 5.27 Plot of gas injected into flow loop as function of time, comparing experimental
data with predictions using the model for hydrate formation in
water-dominated systems for an ExxonMobil flow loop experiment with 50%
liquid loading, 100% water cut, and 2.5 m/s fluid mixture velocity (Test #
XoM-2010-2). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
Figure 5.28 Plot of hydrate volume fraction as function of time, comparing experimental
data with predictions using the model for hydrate formation in
water-dominated systems for an ExxonMobil flow loop experiment with 50%
liquid loading, 100% water cut, and 2.5 m/s fluid mixture velocity (Test #
XoM-2010-2), also showing the calculated Φtransition . . . . . . . . . . . . . . . . 111
Figure 5.29 Comparison of experimental data with predictions using the model for
hydrate formation in water-dominated systems for Test XoM-2011-21 with
Conroe oil, 50% liquid loading, 90% water cut, and 1.75 m/s fluid mixture
velocity: (a) ambient and fluid temperature as function of time, (b) hydrate
volume fraction as function of time, including the calculated Φtransition . . . . . 112
Figure 5.30 Comparison of experimental data with predictions using the model for
hydrate formation in water-dominated systems for Test XoM-2011-24 with
Conroe oil, 50% liquid loading, 75% water cut, and 2.5 m/s fluid mixture
velocity: (a) ambient and fluid temperature as function of time, (b) hydrate
volume fraction as function of time, including the calculated Φtransition . . . . . 113
Figure 5.31 Conceptual picture for hydrate formation in gas-dominated systems consisting
of gas, some condensate and water; adapted from Lingelem et al. (1994). . . . . 115
xiii
Figure 5.32 Illustration of the mass fields considered in the gas-dominated hydrate
formation model coupled with the OLGA
R
transient multiphase flow
simulator, including a gas phase saturated with water and hydrate film
deposited on the pipe wall. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
Figure 6.1 Geometry of well, flowline, and riser, based on typical geometries from the
Caratinga field located at the Campos Basin in Brazil. . . . . . . . . . . . . . . 120
Figure 6.2 Diagram of pressure versus temperature showing flowline operation conditions
during steady-state flow and hydrate equilibrium curves at different ethanol
concentrations. Points on flowline operation conditions curve indicate distance
from wellhead. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
Figure 6.3 Temperature distribution along the pipeline length, showing system
temperature and hydrate equilibrium temperature corresponding to the
steady-state flow predicted by the oil-dominated gas hydrate formation model
for Case 1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
Figure 6.4 Fluid distribution in terms of volume fraction along the pipeline length,
corresponding to the steady-state flow predicted by the oil-dominated gas
hydrate formation model for Case 1. . . . . . . . . . . . . . . . . . . . . . . . . 124
Figure 6.5 Hydrate slurry relative viscosity along the pipeline length, corresponding to
the steady-state flow predicted by the oil-dominated gas hydrate formation
model for Case 1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
Figure 6.6 Temperature distribution along the pipeline length, showing system
temperature and hydrate equilibrium temperature corresponding to the
steady-state flow predicted by the oil-dominated gas hydrate formation model
for Case 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Figure 6.7 Fluid distribution in terms of volume fraction along the pipeline length,
corresponding to the steady-state flow predicted by the oil-dominated gas
hydrate formation model for Case 2. . . . . . . . . . . . . . . . . . . . . . . . . 126
Figure 6.8 Hydrate slurry relative viscosity along the pipeline length, corresponding to
the steady-state flow predicted by the oil-dominated gas hydrate formation
model for Case 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
Figure 6.9 Performance measures (∆Pf lowline , Φhyd , and µr ) used to establish the
hydrate plugging risk as a function of ethanol concentration, predicted by the
oil-dominated gas hydrate formation model for Case 2. . . . . . . . . . . . . . . 127
Figure 6.10 Temperature distribution along the pipeline length, showing system
temperature and hydrate equilibrium temperature corresponding to the
steady-state flow predicted by the oil-dominated gas hydrate formation model
for Case 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
xiv
Figure 6.11 Fluid distribution in terms of volume fraction along the pipeline length,
corresponding to the steady-state flow predicted by the oil-dominated gas
hydrate formation model for Case 3. . . . . . . . . . . . . . . . . . . . . . . . . 128
Figure 6.12 Hydrate slurry relative viscosity along the pipeline length, corresponding to
the steady-state flow predicted by the oil-dominated gas hydrate formation
model for Case 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
Figure 6.13 Performance measures (∆Pf lowline , Φhyd , and µr ) used to establish the
hydrate plugging risk as a function of ethanol concentration, predicted by the
oil-dominated gas hydrate formation model for Case 3. . . . . . . . . . . . . . . 130
Figure 6.14 Temperature distribution along the pipeline length, showing system
temperature and hydrate equilibrium temperature, after 50 hours of shut-in,
predicted by the oil-dominated gas hydrate formation model for Case 2
without ethanol. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
Figure 6.15 Fluids distribution in terms of volume fraction along the pipeline length, after
50 hours of shut-in, predicted by the oil-dominated gas hydrate formation
model for Case 2 without ethanol. . . . . . . . . . . . . . . . . . . . . . . . . . . 131
Figure 6.16 Valve opening schedule used during the restart of production. . . . . . . . . . . 132
Figure 6.17 Evolution of the production restart in terms of pressure drop along the
pipeline (left plot) and hydrate volume fraction (right plot) at a point in the
middle of the pipeline with respect to time, predicted by the oil-dominated
hydrate formation model for Case 2 with 20 wt% ethanol injected before the
shut-in and during the restart. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
Figure 6.18 Performance measures — max {∆Pf lowline (t)}; max {Φhyd (t)}; and
max {µr (t)} — used to evaluate the hydrate plugging risk as a function of
ethanol concentration in transient restart operations, predicted by the
oil-dominated hydrate formation model for Case 2. . . . . . . . . . . . . . . . . 134
Figure 6.19 Diagram of pressure versus temperature showing flowline operation conditions
during steady-state flow, and hydrate equilibrium curve, predicted by the
water-dominated hydrate formation model for the uninsulated case. . . . . . . . 135
Figure 6.20 Temperature distribution along the pipeline length, showing system
temperature and hydrate equilibrium temperature corresponding to the
steady-state flow, predicted by the water-dominated hydrate formation model
for the uninsulated case. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
Figure 6.21 Fluid distribution in terms of volume fraction along the pipeline length,
corresponding to the steady-state flow, predicted by the water-dominated
hydrate formation model for the uninsulated case. . . . . . . . . . . . . . . . . . 136
xv
Figure 6.22 Hydrate volume fraction along the pipeline length, showing the calculated
Φtransition as plugging criterion, corresponding to the steady-state flow
predicted by the water-dominated hydrate formation model for the
uninsulated case. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
Figure 6.23 Temperature distribution along the pipeline length, showing system
temperature and hydrate equilibrium temperature corresponding to the
steady-state flow, predicted by the water-dominated hydrate formation model
for the thermal insulated case. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
Figure 6.24 Fluid distribution in terms of volume fraction along the pipeline length,
corresponding to the steady-state flow, predicted by the water-dominated
hydrate formation model for the thermal insulated case. . . . . . . . . . . . . . 138
Figure 6.25 Hydrate volume fraction along the pipeline length, showing the calculated
Φtransition as plugging criterion, corresponding to the steady-state flow
predicted by the water-dominated hydrate formation model for the thermal
insulated case. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
Figure 6.26 Plot of gas hydrate deposit thickness along the pipeline length at 30, 60 and
100 days, showing an increase in deposit thickness with time, predicted by the
hydrate formation model for gas-dominated systems. . . . . . . . . . . . . . . . 140
Figure 6.27 Plot of system pressure along pipeline length, showing an increase in inlet
pressure with time required to maintain a constant mass flow rate at the inlet
of the pipeline, predicted by the hydrate formation model for gas-dominated
systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
Figure 6.28 Plot of hydrate deposit thickness along the pipeline length at 100 days, for
different levels of water saturation of the gas phase; 150, 200 and 300 ppm of
water in the gas phase, predicted by the hydrate formation model for
gas-dominated systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
xvi
LIST OF TABLES
Table 3.3 Input parameters for three-phase (gas/liquid water/hydrate) hydrodynamic slug
flow model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
Table 4.1 Experiments performed in the ExxonMobil flow loop in Fall 2010. . . . . . . . . . 79
Table 4.2 Experiments performed in the ExxonMobil flow loop in Fall 2011. . . . . . . . . . 80
Table 4.3 Additional experimental data from the ExxonMobil flow loop . . . . . . . . . . . . 80
Table 4.4 Available experimental data from the University of Tulsa hydrate flow loop . . . . 83
Table 5.1 Experiments from the ExxonMobil flow loop selected for predictions using the
water-dominated hydrate formation model. . . . . . . . . . . . . . . . . . . . . . . 108
Table 5.2 Summary of prediction error of hydrate volume fraction at the hydrate
transition concentration (Φtransition ), for the ten high water cut cases studied. . . 113
xvii
NOMENCLATURE
ACO . . . . . . . . . . . . . . . . . . . net interaction energy of the surfactant with the oil phase
ACW . . . . . . . . . . . . . . . . net interaction energy of the surfactant with the aqueous phase
Ceq . . . . . . . . hydrate guest concentration in water phase in the presence of hydrate (kg/m3 )
xviii
ek . . . . . . . . . . . . . . . . . . . . . . . . . . internal energy per unit mass of phase k (J/kg)
f . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . fractal dimension
h . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . elevation (m)
Hk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . holdup of phase k
xix
LL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . liquid loading (vol.%)
p . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . pressure (Pa)
R . . . . . . . . . Winsor ratio of net interaction energies of surfactant to aqueous and oil phases
Re . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Reynolds number
xx
t . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . time (s)
T . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . temperature (K)
v . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . velocity (m/s)
V . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . volume (m3 )
W e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Weber number
X . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Lockhart-Martinelli parameter
Greek Letters
xxi
β . . . . . . . . . . . . . . mixture Joule-Thomson coefficient multiplied by mixture Cp (m3 /kg)
µr . . . . . . . relative viscosity of the hydrate slurry to the viscosity of the continuous oil phase
ρ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . density (kg/m3 )
xxii
υk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . specific volume of phase k (m3 /kg)
ψg . . . . . . . . . . . . . . . . . . . mass transfer rate between gas and liquid phases (kg/m3 /s)
xxiii
LIST OF ABBREVIATIONS
AA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Anti-Agglomerant
xxiv
THI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thermodynamic Hydrate Inhibitor
xxv
ACKNOWLEDGMENTS
I would like to express my sincere appreciation to Dr. Dendy Sloan, who persuaded me to
work on my PhD degree at the Colorado School of Mines, and for introducing me to the world of
gas hydrates and flow assurance. I would like to thank Dr. Sloan for providing support from the
beginning, and for his excellent advice and guidance during my research work at Mines.
I would like to thank Dr. Carolyn Koh and Dr. Amadeu Sum for all their support, advice and
guidance during my research work at the Center for Hydrate Research, and for their comments,
I would like to acknowledge Dr. Hossein Kazemi for giving me the opportunity to work with
him, and for all his support, advice and guidance during the writing of this dissertation.
I would like to thank my thesis committee members, Dr. Mike Batzle, Dr. Yu-Shu Wu and
Dr. Xiaolong Yin, for their time spent helping me shape my research project and reading this
dissertation.
I acknowledge the financial support from DeepStar and the CSM Hydrate Consortium, which
has been sponsored by BP, Champion Technologies, Chevron, ConocoPhillips, Eni, ExxonMo-
bil, Halliburton, MultiChem, Nalco, Petrobras, Schlumberger, Shell, SPT Group, Statoil, and
Total. Special thanks for the following individuals, who provided direct input to this research
project: Jeff Creek, Douglas Estanga, and Siva Subramanian from Chevron; Tom Danielson from
ConocoPhillips; Larry Talley, Mike Eaton, Doug Turner, and Jason Lachance from ExxonMobil;
Alexander de Freitas formerly from Petrobras; and Inge Wold from SPT Group.
I would like to acknowledge ExxonMobil (Larry Talley, Mike Eaton, Doug Turner, and Jason
Lachance) and University of Tulsa (Mike Volk and Emmanual DelleCase) for access to flow loops,
I would like to acknowledge the Viola Vestal Coulter Foundation for granted me a fellowship
in the first year of my career at Mines, and the Roberto Rocca Education Program for granted me
the Roberto Rocca Fellowship during the last two years of my PhD at Mines.
Very special thanks goes to the present and past graduate and undergraduate students that form
part of the Hydrate Busters, for all the discussions in and out of the hydrate office and specially for
xxvi
your friendship. I would like to particularly thank the following Busters for their direct input to
my work at the Center for Hydrate Research: Zach Aman, John Boxall, Piyush Chaudhari, Simon
Davies, Giovanny Grasso, Eric Grzelak, Sanjeev Joshi, Patrick Lafond, Ishan Rao, Kyle Springer,
I would like to thank my family and friends, who have given me unconditional support all these
years. To my parents, for teaching me the value of perseverance, discipline and honesty, for being
my sources of inspiration and motivation, and for all the support and love they have given me;
Finally, I would like to thank Patricia Cuba, my wife, for being with me through this journey,
for sharing a common goal of seeking new knowledge, for cheering me during tough moments, for
xxvii
This work is dedicated to the loving memory of Solange Martinez, mi Abuela.
xxviii
CHAPTER 1
INTRODUCTION
The offshore oil and gas industry has experienced a continuous growth since its inception in the
1940’s (Sandrea and Sandrea, 2007). Currently, offshore production represents 30% of global oil
and gas production (Serbutoviez, 2012). With the implementation of new exploration and drilling
technologies, it is expected for the offshore industry to continue its growth towards deep (500 - 1500
m) and ultra-deep (> 1500 m) waters. From market predictions, the offshore industry is expected
to invest $210 billion for new developments in five years (2011 - 2015), a 60% increase compared
to the previous period (2006 - 2010), with pipelines and flow systems representing the 38% of this
The offshore industry will face more challenging scenarios with production from deeper and
colder waters. Subsea facilities will require longer subsea tiebacks in satellite fields to transport
hydrocarbon fluids from the wellhead to existing production and processing platforms, and may
require the transportation of processed gas and condensate streams to export facilities through
subsea pipelines. The formation of solid deposits, such as gas hydrates, waxes, asphaltenes and
mineral scale, may plug the flowlines, preventing production and generating a safety hazard. The
flow assurance of the produced hydrocarbon stream has become a technical discipline that focuses
on the design of safe and secure operation techniques for the uninterrupted transport of reservoir
fluids from the reservoir to the point of sale (Kaczmarski and Lorimer, 2001). The rapid formation of
gas hydrates, which is promoted by typical high pressure and low temperature operation conditions
in deep subsea facilities, is considered one of the most challenging flow assurance problems (Sloan,
2005).
It had been recognized that predicting hydrate formation conditions using thermodynamic cal-
culations (with excellent accuracy) is not sufficient to estimate hydrate plugging risk (Sloan, 2005).
Instead, this can be used to design hydrate avoidance methods (e.g., chemical injection, thermal
insulation, active heating, and pressure reduction), keeping the systems out of the hydrate sta-
bility region (Creek et al., 2011). The offshore oil and gas industry has progressed toward using
longer tiebacks to connect subsea wells with platforms (Ronalds, 2005), rendering hydrate avoid-
1
ance methods economically unfeasible. The alternative would be to consider hydrate management
approaches, where hydrates are allowed to form, but the plugging risk is low (Creek et al., 2011;
Sloan, 2005). Models for hydrate formation kinetics coupled with transportability models may be
helpful during the design and assessment of hydrate management approaches, providing estimates
of the amount of hydrates that could form and its transportability in specific scenarios, to estimate
A gas hydrate model that estimates hydrate plugging risk in pipelines is of significant interest
in design and optimization of flowlines and operation procedures. Such a model should account
for transient mechanisms present in different systems of oil/gas production facilities such as: oil-
dominated systems (oil as hydrate carrier phase), water-dominated systems (water as hydrate
carrier phase), and gas-dominated systems (hydrate film growth on pipe walls). By predicting gas
hydrates formation and transportability, the comprehensive model can be applied in the design and
An intense research effort at the Center for Hydrate Research of Colorado School of Mines
(CSM) led to the development of the CSM Hydrate Kinetics model (CSMHyK), a gas hydrate
model specially designed for oil-dominated systems, which has been incorporated as a plug-in
module in a transient multiphase flow simulator (Bendiksen et al., 1991; Boxall et al., 2009; Turner
et al., 2005). The objective of this thesis is to extend and significantly advance the gas hydrate
model (CSMHyK) towards a comprehensive model that includes transient mechanisms present in
oil, water and gas-dominated systems, with the goal of accounting for the principal scenarios of
This introductory chapter presents a general overview of flow assurance focusing on solid de-
posits (waxes, asphaltenes, mineral scale and gas hydrates), a brief description of the current version
of CSMHyK for oil-dominated systems, the thesis objectives and scope, and structure of the thesis.
The flow assurance of produced fluids from offshore wells has become a technical discipline,
critical for the economic success of offshore production projects (Kaczmarski and Lorimer, 2001).
Flow assurance involves the design of facilities and operation procedures for the uninterrupted
2
transport of reservoir fluids from the reservoir to the point of sale, over the life of a project.
The flow assurance contributions to subsea facilities design involve specification of pipeline size,
production facility materials, and pressure boosting requirements (Bai and Bai, 2005). The pipe di-
ameter should guarantee an optimum hydraulic performance of the system, in terms of the pressure
drop versus production flow rate relationship, over a range of operating conditions (i.e., minimum
and maximum allowable flow rates). The selection of materials should consider compatibility with
the fluids to be transported, corrosion issues and thermal insulation requirements. Pressure boost-
ing equipment may be required if the wellhead pressure is not enough to transport fluids up the
The flow assurance design process also involves fluid properties characterization and thermal-
hydraulic calculations, for the development of operation strategies that prevent problems such as
solid blockages, severe slugging, sand erosion and corrosion. Four issues have been identified in the
1. Fluid characterization.
The first stage consists of the characterization of hydrocarbon fluids and produced water, to
obtain the fluid properties required for the thermo-hydraulic calculations and to assess potential
problems with solid deposits. For hydrocarbon fluids the characterization involves pressure-volume-
temperature (PVT) and compositional analysis. The PVT analysis provides properties such as
density, viscosity, oil and gas compressibility, saturation pressure, and gas solubility in oil. The
compositional analyses are typically performed through components with at least C36+ carbon
number, and should include density and molecular weight of the petroleum fractions from C7 and
3
The characterization of hydrocarbon fluids also includes special measurements for the flow
assurance design, to assess potential problems with deposition of waxes or asphaltenes. Wax is
the name given to heavy components of crude oil that consist of linear n-paraffins and branched
iso-paraffins with a carbon number from C20 to C70 (Sarica and Panacharoensawad, 2012). To
determine the amount of wax in a crude oil, a compositional analysis is required that detects
heavier components in the oil, typically using high temperature gas chromatography (HTGC), which
determines paraffin composition as high as C100 (Alboudwarej et al., 2006). A key parameter used
to determine problems of wax deposition is the crude cloud point or wax appearance temperature
(WAT), defined as the temperature below which wax crystals start to form. Another wax parameter
of interest for flow assurance is the pour point, which is related to the oil phase rheology. The pour
point is the temperature at which the oil phase solidifies because of formation of a gel network of
precipitated wax crystals; this could prevent restart of fluid flow during transient shut-down/restart
operations.
The deposition of wax on the pipe walls narrows the flow conduit diameter and changes the wall
roughness increasing pressure drop and reducing production flow rate, and can eventually plug the
pipeline (Figure 1.1). Wax deposition is a slow mechanism mainly dominated by molecular diffusion
of heavy paraffins (Guo et al., 2005). The molecular diffusion process is driven by a concentration
gradient, which is a function of the radial temperature gradient inside the flowing pipeline. The
solubility of heavy paraffins is greater at the higher temperature at the center of the pipe, while
the solubility is lower at the low temperature at the pipe wall. If the temperature at the wall is
lower than the WAT point at a given pressure, then wax will precipitate and deposit on the wall.
Figure 1.1: Wax deposit on a pipe wall narrowing conduit diameter and modifying wall roughness
(Mokhatab and Towler, 2009).
4
Essentially asphaltenes are large polar molecules primarily formed by carbon and hydrogen,
with one to three heteroatoms (nitrogen, oxygen or sulfur) per molecule (McCain, 1990). The het-
eroatoms can be part of aromatic rings clusters or can be in the links between the rings (Figure 1.2).
The polar groups of an asphaltene molecule provides a driving force for interfacial adsoption, due
to the ionization of the polar group at the oil-water interface. For example, the carboxylic group
(COOH), shown to the left of the asphaltene molecule in Figure 1.2, can dissociate in water result-
ing in a carboxylate anion (COO- ), which has a high affinity for the water, and a positively charge
Asphaltenes are petroleum fractions that solubilize in aromatic solvents (e.g., toluene and ben-
zene), but are insoluble in light aliphatic solvents (e.g., pentane and heptane) (Mitchell and Speight,
1973). Asphaltenes are insoluble in the bulk oil phase, but are stabilized as dispersed colloids by
the adsorption of resins. Resins are formed by two groups, a polar group that contains heteroatoms
(e.g., oxygen, sulfur and nitrogen) and a nonpolar paraffinic group. The polar group of the resins is
adsorbed to the surface of asphaltene micelles by hydrogen bonding between the heteroatoms and
dipole-dipole interactions generated from the polarities of the resin and asphaltene. The nonpolar
paraffinic group of the resins acts a transition to the bulk oil phase.
The stability of asphaltenes changes with pressure, temperature and composition. Generally,
asphaltene precipitation occurs near the oil bubble point because of the vaporization of light hydro-
carbons; i.e., a change in composition of the liquid hydrocarbon phase (Hammami and Ratulowski,
5
2007). Near the oil bubble point, the different compressibility between lighter and heavier compo-
nents translates into an increase in volume fraction of lighter components, affecting the equilibrium
that exists in the nonpolar portion of the crude oil. This cause desorption of resins from the sur-
face of asphaltene micelles, which reduces the stability of the colloidal dispersion, promoting the
The amount of asphaltenes in an oil sample can be measured with a SARA analysis, which
determines the amount of saturates, aromatics, resins and asphaltenes by a combination of solvent
induced precipitation and column chromatography. Asphaltene stability in dead oils can be mea-
sured using titration tests, where an aliphatic solvent is added and a light transmittance probe
is used to detect the asphaltene flocculation onset, which is marked by a maximum in the light
transmittance signal (Guo et al., 2005). A simple method to determine the asphaltene precipita-
tion tendency of a crude oil is the de Boer plot (de Boer et al., 1995), which identifies three regions
(severe, slight and no problem) based on the difference of reservoir pressure and bubble point pres-
sure as function of oil density at in-situ conditions; however predictions from this method could
be overly pessimistic (Wang et al., 2006). Hammami and Ratulowski (2007) presented a review of
special methods to determine the upper and lower asphaltene precipitation loci in terms of pressure
The precipitation of asphaltenes can cause plugging of wells and pipelines by deposition on
tubular walls (similar to wax deposition showed in Figure 1.1). Additionally, asphaltenes can cause
oil/water separation problems by emulsion stability (Sjöblom et al., 2003) and fouling of surface
The characterization of produced water allows assessment of potential problems with inorganic
precipitates and provides information to estimate the hydrate equilibrium curve. Inorganic pre-
cipitates or scales are mineral solids that precipitate from the water phase, caused due to changes
in pressure and temperature. The most common scales encountered in oil and gas pipelines are
calcium carbonate (calcite, CaCO3 ), barium sulfate (BaSO4 ), strontium sulfate (SrSO4 ), and cal-
cium sulfate (anhydrate, CaSO4 ) (Guo et al., 2005). Carbonate scales form from formation water
with reduced pressure, increased temperature and increased pH. Sulfate scales form with mixing
of different waters (e.g., formation water mixed with sea water). Problems with scale deposition
increase with increasing water cut. Scale deposits can also plug a flowline by reducing the conduit
6
diameter.
Gas hydrates are crystalline compounds formed by cages of hydrogen-bonded water molecules
(host) stabilized by encapsulation of small gas molecules (guest), like methane, ethane, propane
or carbon dioxide (Sloan and Koh, 2008a). The cages of hydrogen-bonded water molecules are
typically formed by pentagonal and hexagonal faces, and sometimes by square faces depending on
the crystal structure, as shown in Figure 1.3 for the most common gas hydrate structures. The
gas hydrate structure I (sI) unit cell consists of two small cages, each formed by twelve pentagonal
faces (512 ) and six large cages, each formed by twelve pentagonal and two hexagonal faces (512 62 ).
The gas hydrate structure II (sII) unit cell consists of sixteen small 512 cages and eight large cages,
each formed by twelve pentagonal and four hexagonal faces (512 64 ). The gas hydrate structure H
(sH) unit cell consists of three small 512 cages, two small cages, each formed by three square, six
pentagonal and three hexagonal faces (43 56 63 ) and one large cage formed by twelve pentagonal and
eight hexagonal faces (512 68 ). The stability of a specific hydrate structure depends on the fit of the
guest molecule inside the host cage. Small hydrocarbon molecules, such as methane and ethane
can stabilize gas hydrate sI. However, sII is the typical gas hydrate structure encountered in oil and
gas pipelines because of the presence of larger hydrocarbon molecules, like propane and iso-butane
that fit in the large cage of sII, but do not fit in the large cage of sI (Sloan and Koh, 2008a).
The pressure versus temperature diagram, shown in Figure 1.4, illustrates a typical phase
water (Sloan and Koh, 2008b). In this diagram the symbols I, LW , H, V, and LHC represent ice,
liquid water, hydrate, vapor, and liquid hydrocarbon, respectively. The quadruple points represent
the coexistence of four phases, Q1 for I–LW –H–V, and Q2 for LW –H–V–LHC . The continuous blue
line formed by line LW –H–LHC at conditions above Q2, line LW –H–V between Q1 and Q2, and line
I–H–V at conditions below Q1, represents the hydrate equilibrium boundary. To the right of this
curve the hydrate phase is not stable (i.e., no hydrates are possible), and at lower temperatures
and higher pressures to the left side of the hydrate equilibrium curve the hydrate phase is stable
(i.e., hydrates can form). The red segmented line in the upper portion of the diagram, represents
the vapor-liquid hydrocarbon equilibrium curve with liquid water to the right of Q2 and hydrate
to the left of Q2. Below the V–LHC equilibrium curve the hydrocarbon will be in the vapor state,
and above the V–LHC equilibrium curve the hydrocarbon will be in the liquid state. The dash-dot
7
Figure 1.3: Common gas hydrate structures; adapted from Sloan (2011, p. 3).
green line is the I–LW equilibrium curve with a vapor phase below Q1 and a hydrate phase above
Q1. To the left of the I–LW curve, water will be in the solid state (ice or hydrate), and to the right
of the I–LW curve, water will be in liquid state or in the hydrate phase. The complexity of the
phase behavior diagram changes with the composition of the system; more details about hydrate
phase diagrams for water-hydrocarbon systems can be found in Sloan and Koh (2008b).
The gas hydrate phase envelope, defined by equilibrium pressure and temperature conditions,
can be determined by thermodynamic flash calculations based on the van der Waals and Plat-
teeuw (1959) model with excellent engineering accuracy (Ballard and Sloan, 2004b). Several com-
PVTSim
R
from Calsep) implement a modified version of this model for the prediction of gas hy-
drate properties (Sloan and Koh, 2008a). Another program developed at the Colorado School of
Mines, based on the Gibbs energy minimization method (CSMGem), can also be used to estimate
the gas hydrate equilibrium conditions (Ballard and Sloan, 2004a). The composition of hydrocar-
bon fluids and produced water, obtained during the fluids characterization, are used as input to
8
HC
L −H−L
Pressure (log scale)
− L HC
w
Lw − V
H−V−L
HC
Q2
V
I−L −H
−
H
−
w
w
L
−V Q1
I−H
I − Lw − V
Temperature
Figure 1.4: Pressure versus temperature diagram for a hydrocarbon–water system; adapted from
Sloan and Koh (2008a, p. 198).
Figure 1.5 shows a pressure versus temperature diagram that includes the WAT curve, the upper
and lower asphaltene precipitation loci and the hydrate equilibrium curve from the characterization
of a Gulf of Mexico reservoir fluid. This kind of diagram allows one to identify conditions corre-
sponding to potential formation of solid deposits. For the wax and hydrate curves, the right side
of the curves represents the regions with no precipitation/formation of solids, and to the left of the
curves are the regions of wax crystals precipitation and hydrate formation, respectively. In the case
of asphaltenes, there are two precipitation curves, an upper and a lower precipitation curve. The
asphaltene precipitation region is found between the upper and lower precipitation curves, while
there are two asphaltene stable regions, one above the upper precipitation curve and other below
The second stage of the flow assurance design process consists of steady-state thermal-hydraulic
9
70
Wax No Wax
60 Asphaltene
stable region
Pressure (MPa)
50 Asphaltene
precipitation region
40
Asphaltene
30 stable region
20 Hydrates
10 No
Hydrates
0
260 280 300 320 340 360
Temperature (K)
Figure 1.5: Pressure versus temperature diagram showing the WAT curve (green dashed curve),
upper and lower asphaltene precipitation loci (red dash-dot curves) and hydrate equilibrium curve
(blue continuous curve) from the characterization of reservoir fluids; adapted from Ayers (2003).
pipeline diameter, considering the geometry of the flowline from the wellhead to the platform
in terms of horizontal distance and elevation. The pressure drop versus production flow rate
relationship is studied through a parametric study modifying initial and boundary conditions. The
pipe diameter selected should guarantee proper hydraulic performance of the system over a range
Potential problems with solid deposits can be determined by comparing the flowline pressure
and temperature conditions obtained from the steady-state thermal-hydraulic calculations with
the conditions of solid precipitation/formation obtained during the fluid characterization studies.
Wherever the flowline operating conditions curve lies to the left of the WAT, asphaltene precip-
itation or hydrate equilibrium curves, the system is considered to potentially have solid deposits
problems. Figure 1.6 shows a pressure versus temperature diagram where a typical subsea flowline
operation conditions curve is compared to the WAT curve, upper and lower asphaltene precipita-
tion loci and hydrate equilibrium curve. The reservoir fluids flowing out of the wellhead through
the horizontal portion of the flowline, start transferring heat to the cold environment (typical sea
floor temperature is 277 K) through the pipe wall, represented by a considerable decrease in tem-
perature and a slight decrease in pressure in the flowline. There is a major pressure decrease when
10
the fluids flow up the riser with an increase in temperature. A Joule-Thomson cooling effect at a
valve near the platform is responsible for an additional decrease in temperature. In this example
the flowline operation conditions intercept the wax precipitation boundary and then the hydrate
equilibrium curve during fluid flow in the horizontal portion of the pipeline, indicating that this
system can have a potential problem with wax and hydrate solid deposits. The operation condi-
tions of the pipeline are below the lower asphaltene precipitation locus (i.e., inside the asphaltene
stable region), indicating that there is no risk of asphaltene precipitation in this flowline. Note that
if the operation pressure is high enough to shift the flowline conditions to inside the asphaltene
this stage, it is possible to study different flow assurance strategies, such as thermal insulation of
the pipeline to prevent the system from entering in the solid precipitation/formation regions.
60
Wax No Wax
50 Asphaltene
Pressure (MPa)
precipitation region
40
Asphaltene
stable region
30
No
Hydrates Hydrates
20
Wellhead
Flowline operation
10 conditions
Platform
0
260 280 300 320 340 360
Temperature (K)
Figure 1.6: Pressure versus temperature diagram showing a typical subsea flowline operation
conditions curve, the WAT curve, the lower asphaltene precipitation locus and hydrate equilibrium
curves; adapted from Ayers (2003).
A flow assurance strategy typically used to avoid hydrate formation is the injection of ther-
modynamic hydrate inhibitors (THI), such as methanol (MeOH) or monoethylene glycol (MEG).
11
Thermodynamic hydrate inhibitors compete for water molecules through hydrogen-bonding, pre-
venting the formation of hydrate cages, thus inhibiting the hydrate formation. This results in the
hydrate equilibrium curve shifting to the left by decreasing the hydrate equilibrium temperature
and increasing the hydrate equilibrium pressure. Figure 1.7 shows a pressure versus temperature
diagram where a typical subsea flowline operation conditions curve is compared to hydrate equi-
librium curves at different methanol concentrations. A significant portion of the flowline is inside
the hydrate forming region when no methanol is injected. A methanol concentration greater than
15 11 km
24 16
32
40
10 Wellhead
48
5 56 Hydrate
Platform 64 72 free
region
80
0
270 280 290 300
Temperature (K)
Figure 1.7: Pressure versus temperature diagram showing a typical subsea flowline operation
conditions curve and hydrate equilibrium curves at different methanol concentrations; adapted
from Sloan and Koh (2008c, p. 646). Points on the flowline operation conditions curve indicate
distance from wellhead in kilometers.
The gas and oil industry is currently in a transition from using hydrate avoidance methods (such
as injection of thermodynamic hydrate inhibitors) to risk management methods (Sloan, 2005). The
use of inhibitors that act at the interfaces of hydrate particles, such as kinetic hydrate inhibitors
(KHI), have been proven to reduce the required concentrations of chemical additives in a more
12
economical and efficient flow assurance strategy. KHIs are low molecular weight polymers that
(PVP). The purpose of KHIs is to delay the hydrate nucleation and crystal growth for a period of
time by adsorbing at the surface of the hydrate crystal (Sloan and Koh, 2008c).
The third stage of the flow assurance design process consists of transient-flow thermal-hydraulic
calculations using transient multiphase flow simulators. This stage involves the study of transient
operations such as planned or unplanned shut-down of production and subsequent restart. As-
suming that the system is originally outside of the hydrate formation region, during a production
shut-down the fluids temperature will decrease, reaching equilibrium with the cold external ambient
temperature, entering into the hydrate formation region. The formation of gas hydrates during a
production shut-down represents a major concern because of its rapid formation at high subcool-
ing (i.e., difference between hydrate equilibrium temperature and fluids temperature). Generally,
wax deposition does not represent a problem during production shut-down because the molecular
diffusion that dominates the deposition mechanism is a slow process requiring a great volume of
estimating the cool-down time, which is the time required for the system to reach the hydrate
formation conditions from the steady-state flow conditions after a production shut-down. The
cool-down time should be long enough to accommodate a no-touch time and an operation time.
During the no-touch time, no means of hydrate mitigation are implemented and the operator should
try to determine the problem that caused the shut-down. The operation time is the time required
to perform a hydrate mitigation strategy during the shut-down, such as depressurization or dead
oil displacement. If the operator is able to find a solution before the no-touch time, then the system
can be restarted without the implementation of any hydrate mitigation strategy. Generally, the
During planned production shut-down operations longer that the no-touch time, it is possible
the other hand, during unplanned production shut-down operations the system will not count with
13
hydrate inhibition, so the process must include other hydrate remediation strategies if the no-
touch time is surpassed. Typically, after the no-touch time hydrate inhibitors are injected in the
system to prevent hydrate formation in the wellbore, wellhead, jumpers and manifolds, followed
blow-down of the pipeline from the top sides, lowering the pressure of the system outside of the
The transient thermal-hydraulic calculations also serve to estimate the pipeline liquids inventory
after a production shut-down, which consist of estimating the amount and distribution of gas, oil
and water inside the pipeline (Hagesaether et al., 2007). The amount of hydrate inhibitor required
can be determined from the amount of water in the system. Also, the transient thermal-hydraulic
calculations can be used to study potential problems with severe slugging. Liquids can accumulate
in local low spots in the pipeline, which can result in a liquid surge coming to the platform during
restart, or generate severe slugging at low restart flow rates. If the upstream pressure is low and
the flow rate is reduced, the gas and liquid velocities might not be enough to continuously carry the
fluids up the riser, resulting in intermittent fluid flow to the top sides (i.e., processing equipment).
Liquid accumulates in the base of the riser until pressure behind the liquid is high enough to push
the slug up the riser resulting in a huge amount of liquids flowing into the separator.
1.1.4 Final system design and operation procedures for flow assurance
The fourth and final stage of the flow assurance design process consists of finalizing the system
design and establishing the operating philosophy for flow assurance issues, by integrating the results
of the previous three stages. The final system design will depend strongly on the operation strategies
adopted to avoid problems with solid blockages. Pigging (passing a cylinder down the line to clean
it) is an alternative to remove solid deposits of wax, asphaltene or mineral scale, by mechanically
removing the deposit from the wall. If pigging is selected as an operation strategy, the system should
include round-trip pigging capabilities in the facilities design. Another alternative to manage solid
deposits is chemical inhibitor injection, which requires the inclusion of chemical inhibitor delivery
The flow assurance design process of an offshore field faces an additional challenge; to balance
capital cost (CAPEX) and operational cost (OPEX) minimizing the overall economic expenses of
14
a project (Guo et al., 2005). The selection of materials (thermal insulation, high grade materials)
and subsea equipment along with the design of solid deposits mitigation systems are related to
CAPEX. Operation procedures like chemical inhibitor injection, pigging, flow monitoring are related
to OPEX.
An intense research effort at the Center for Hydrate Research of the Colorado School of Mines
led to the development of the Colorado School of Mines Hydrate Kinetics model (CSMHyK), a
gas hydrate model specially designed for oil-dominated systems. CSMHyK was developed for oil-
dominated systems based on the conceptual model presented in Figure 1.8, which represents an
approximation to the mechanism of hydrate plug formation and is divided into four main stages:
1. Water entrainment: The water droplets are dispersed in the continuous oil phase by flow
2. Hydrate growth: A hydrate shell forms at the interface between the water droplets and the
3. Agglomeration: The hydrate-encrusted water droplets can agglomerate, increasing into larger
4. Plugging: The agglomeration of hydrate particles leads to an increase in the slurry viscosity,
Figure 1.8: Conceptual model for hydrate formation in oil-dominated multiphase flow systems
consisting of gas, oil, and water; adapted from Turner (2005), with input from J. Abrahamson (U.
Canterbury, Christchurch, NZ).
15
CSMHyK has been incorporated as a plug-in module in a transient multiphase flow simulator
(Bendiksen et al., 1991; Turner et al., 2005). The transient multiphase flow model considers four
phases: gas, oil, water, and hydrate. The governing equations for one-dimensional multiphase flow
in a pipeline consist of the continuity equation, the momentum conservation equation and the energy
equation. These governing equations are constrained by geometrical data; initial and boundary
conditions; fluid properties; and inter-field mass, momentum and heat transfer. The model solves
the system of equations to obtain pressure, phase velocities, phase masses, and temperature. A
detailed description of the governing equations of the transient multiphase flow model is presented
in Section 3.3.3 of Chapter 3, which presents a review of multiphase flow concepts and modeling
approaches.
CSMHyK is used to calculate additional contributions to the conservation equations. The hy-
drate kinetic model (presented in Section 1.2.1) calculates the hydrate formation rate, which is
used to calculate mass transfer terms between the gas, water and hydrate phases. The hydrate
equilibrium curve is used as a thermodynamic constraint to determine the conditions when the hy-
drate phase start to form. A hydrate particle agglomeration model is used to calculate the effective
viscosity of the hydrate carrier phase, which affects the momentum conservation equations. The
modified viscosity will induce modifications in the friction factors. The hydrate heat of formation
acts as an energy source inside the pipeline, affecting the energy equation and the resulting system
temperature.
The current version of the CSMHyK model comprises three sub-models: the kinetics model,
The kinetics model assumes that all the water is dispersed into the continuous oil phase as
water droplets of fixed mean diameter. The surface area between water and hydrocarbon phases is
calculated using the Hinze (1955) correlation. Hydrate nucleation is assumed to occur immediately
after a specified subcooling (∆Tsub = Thyd eq − Tsystem ), and the hydrate growth is calculated using
the following intrinsic kinetics equation of first order with an adjustable rate constant (Boxall,
16
2009),
dmgas k2
= −uk1 exp As ∆Tsub (1.1)
dt Tsystem
The intrinsic kinetics equation gives the rate of gas consumption during hydrate formation as a
function of the intrinsic kinetic rate constants (k1 and k2 ), the surface area (As ) between water and
hydrocarbon phases, and the subcooling as the thermal driving force. Vysniauskas and Bishnoi
(1983) and Englezos et al. (1987) measured the gas consumption rate during the formation of
mixed methane and ethane hydrate at different temperatures and subcooling, in the absence of
either mass or heat transfer resistances. Their experimental data, shown in Figure 1.9, were used
to regressed the intrinsic kinetic rate constants, k1 = 7.3548 × 1017 and k2 = −13600 K. These
constants correspond to the kinetic reaction of hydrate structure I, which represents a limitation
of the model since hydrate structure II is the expected structure in oil and gas flowlines. In order
to match flow loop data, a scaling factor u = 1/500 was introduced, indicating the need to include
mass and heat transfer resistances to the model (Boxall et al., 2009). The need for this large
correction factor (1/500) indicates that our initial model was a fit to the data, rather than a true
Figure 1.9: Bishnoi’s laboratory data used to regress the intrinsic kinetic rate constants (Englezos
et al., 1987; Vysniauskas and Bishnoi, 1983).
17
Once the hydrate particle has formed, the size of the hydrate particle agglomerates is calculated
using a model presented by Camargo and Palermo (2002), based on a steady-state balance between
the inter-particle cohesion force and the fluid flow shear forces. Assuming a constant cohesion force
(Fa ), the following non-linear equation is solved for the hydrate agglomerate diameter (dA ),
" #2
dA 3−f
Φ
4−f Fa 1 −
dA Φmax dP
− " 3−f # = 0 (1.2)
dP dA
d2P µ0 γ̇ 1 − Φ
dP
where dA is the hydrate agglomerate diameter; dP is the hydrate particle diameter; Φ is the
hydrate particle volume fraction; Φmax is the maximum volume fraction to which particles can
pack (assumed to be equal to 4/7, the packing concentration of randomly packed spheres of same
diameter); f is the fractal dimension that accounts for the porosity of the aggregates, assumed to
be equal to 2.5 (a value inside the reported range of fractal dimensions of aggregates under shear
conditions, Hoekstra et al., 1992); Fa is the inter-particle cohesion force, assumed to be equal to 50
mN/m (a value that corresponds to capillary attraction between hydrate particles by a liquid bridge,
obtained from theoretical calculations by Camargo and Palermo, 2002); µ0 is the oil viscosity; and
Due to the fractal structure of the aggregates, the effective volume fraction of the agglomerated
is calculated using a modification of the Mills (1985) equation that considers the effective volume
1 − Φef f
µr = (1.4)
Φef f 2
1 − Φmax
Camargo and Palermo (2002) presented the derivation and description of this model for the
hydrate slurry relative viscosity and compared the predictive performance of the model against
experimental data, showing good agreement between calculation and experimental data.
18
A hydrate plug is identified by a large viscosity increase, which causes an unacceptably large
The subsequent transport model, similarly, assumes that all the water is dispersed by the flow
shear forces into the continuous oil phase as water droplets. The transport model includes a revised
correlation to calculate the water mean droplet size (Boxall, 2009) as a function of the viscosity of
the oil phase and water-oil interfacial tension, which is an improvement from the Hinze correlation.
Boxall (2009) developed a universal curve in terms of two dimensionless numbers, the Weber
number defined as W e = ρo v 2 d/σ, and the Reynolds number defined as Re = ρo vd/µo , that
identified two separate regions in turbulent flow that affect mean droplet size in an emulsion. The
first region is the inertial sub-range, where the droplet size is determined by a balance between
the interfacial stress that holds the drop together, and the turbulent inertial stress that attempts
to break-up the droplet. In this region the droplet sizes are independent of viscosity. The second
region is the viscous sub-range, where the droplet size is determined by a balance between the
interfacial stress and the sub-eddy viscous stress. Figure 1.10 shows this universal curve, which
uses experimental data from an autoclave mixing cell with a Particle Video Microscope probe and
a large scale flow loop, involving different crude oils. The transition from the inertial to the viscous
sub-ranges occurs when W e > 0.0674Re5/4 . A mean droplet size correlation was developed for each
region using the measured mean droplet diameter. For the inertial subrange the mean droplet size
correlation is,
and the droplet size correlation for the viscous sub-range is,
It is important to note that the use of the Weber number is a combination of shear (Re) and
surface properties (σ); this combination is our first quantitative recognition that flow properties
and chemistry affect water droplet size, as opposed to more usual flow properties alone (e.g., Re).
shell at the water droplets interface. For this model, it is assumed that further hydrate growth
19
Figure 1.10: Universal curve of turbulent flow sub-ranges in terms of Weber and Reynolds dimen-
sionless numbers that affect mean droplet size in an emulsion, and experimental data obtained in
an autoclave mixing cell and a large scale flow loop, using different crude oils (Boxall, 2009).
occurs on the interior of the shell (i.e., a shrinking core model), and that hydrate growth is limited
by slowest mass or heat transfer through the boundary layer surrounding the hydrate particle or
Figure 1.11: Illustration of a water droplet with a surrounding hydrate shell showing the boundary
layer around the spherical particle.
The hydrate growth rate from external mass and heat transfer through the boundary layer (i.e.,
and
dmgas hAs ∆Tsub Xgas hyd M Wgas
= (1.8)
dt ext,heat ∆hhyd M Whyd
20
where kmass is the mass transfer coefficient calculated using correlations for the inertial and viscous
sub-ranges (Davies, 2009), As is the surface area of the hydrate particle, Cbulk is the hydrate guest
concentration in the bulk phase, Ceq is the hydrate guest concentration in the water phase in the
presence of hydrate, h is the convective heat transfer coefficient calculated using correlations for the
inertial and viscous sub-ranges (Davies, 2009), Xgas hyd is the guest concentration in the hydrate
phase, M Wgas is the molecular weight of the gas, M Whyd is the molecular weight of the hydrate
The hydrate growth rate from internal mass transfer across the hydrate shell is based on the
diffusion of gas molecules through the hydrate shell, and is calculated using the following equation:
dmgas Dgas As
= (Cbulk − Ceq ) (1.9)
dt int,mass δ rrPw
Here Dgas is the diffusivity of gas molecules through the hydrate shell, δ is the shell thickness,
The hydrate growth rate from internal heat transfer across the hydrate shell is calculated using
Here kcomp is the thermal conductivity of the hydrate shell with porosity ε, assuming that the
The agglomeration of hydrate particles is calculated in the same form as in the kinetics model,
but considering the inter-particle cohesion force as a function of temperature using a correlation
dP
Fa = [0.0017 (7.7 − ∆Tsub ) + 0.0007] (1.12)
2
This correlation captures the increase in the value of inter-particle cohesion forces with increas-
ing temperature, which is an improvement to the previous assumption of a fixed value for the
In a similar way to the kinetics model, a hydrate plug is identified by a large viscosity increase,
21
1.2.3 Cold flow model
The transport model has been adapted to simulate the BP-Statoil-SINTEF stabilized cold flow
concept (Gudmundsson, 2002; Wolden et al., 2005), in which dry hydrate flowable slurries are
generated. In this stabilized flow concept, warm production fluids are contacted with a stream of
dry hydrate particles in a mixing zone. The water from the production fluids coats the dry hydrate
particles and rapidly converts to dry hydrate in the reaction zone. Part of the dry hydrate slurry
can then be transported to the downstream processing facilities and the rest of the slurry is pumped
Figure 1.12: Conceptual picture of the BP-Statoil-SINTEF stabilized flow concept (redrawn from
Davies, 2009).
The cold flow model was specially designed to simulate the hydrate formation in the reaction
zone, where the fundamental difference to the transport model is that dry hydrate particles are
pre-formed, and then coated with a water film. For this model it is assumed that further hydrate
growth occurs on the interior of the water film with a growing hydrate core, and that hydrate growth
is limited by slowest mass or heat transfer through the boundary layer surrounding the hydrate
particle or through the water film, as illustrated in Figure 1.13. A similar calculation procedure to
the one implemented in the transport model is used in the cold flow model to estimate the hydrate
growth rate.
The lower mass transfer resistance across the water film is responsible for rapid conversion of
water into hydrate. This minimizes particles agglomeration, resulting in a transportable hydrate
slurry.
22
Figure 1.13: Illustration of a hydrate particle with a surrounding water film showing the boundary
layer around the spherical particle.
In the cold flow model the number of particles is conserved throughout the simulation. The
particle size is determined from the particle concentration and the volume fraction of water and
hydrate, by dividing the hydrate and water volumes between the particles.
The previous version of CSMHyK was developed for oil-dominated systems, where the oil is
the continuous phase. However, the increasing water cut in the production stream during the life
of a reservoir will eventually lead to a water-dominated system, where water is the hydrate carrier
phase. Gas-dominated systems are also important scenarios for hydrate formation that includes
deposition of hydrates on the pipe wall, sloughing and jamming to form a plug.
The objective of this thesis was to develop a comprehensive model to predict gas hydrate
formation and transportability in oil and gas flowlines. The model includes transient mechanisms
present in oil, water, and gas-dominated systems, to account for the principal scenarios of hydrate
formation found in oil and gas subsea flowlines. As a result, the CSMHyK model was upgraded.
scenarios of oil/gas production facilities, from observations and data of flow loop experiments.
• Built mathematical models, based on experimental observations and data, of the mechanisms
• Included the effect of naturally inhibited oils and the injection of anti-agglomerant surfactants.
23
• Presented a workflow for hydrate flow assurance in the design of subsea transport flowlines
A review of surface chemistry concepts is presented in Chapter 2, with the principal objective to
identify interfacial phenomena and surface chemistry interactions involved in gas hydrate formation
and agglomeration in oil and gas pipelines. There are five types of interfaces where gas hydrates may
form and aggregate: gas/liquid, liquid/liquid, gas/solid, liquid/solid, and solid/solid; where gas is
hydrocarbon gas, liquid is either oil, water or condensate, and solid is either gas hydrate or pipe
wall surface. A review of fundamental interfacial concepts can help create a better understanding
of phenomena at these interfaces. Two areas of surface chemistry have been selected to illustrate
the concepts and mechanisms associated with these systems: surfactants and emulsions. Examples
from the literature pertaining to gas hydrates are presented for each system.
The fluid flow in a pipeline can be classified in single-phase, multiphase and slurry flow (which
includes solid particles and dispersions), by considering the number of phases present. A review
of fundamental concepts of fluid flow in pipes is presented in Chapter 3, along with a review of
multiphase flow modeling approaches. The effect of gas hydrate formation on multiphase flow is
The mechanisms of hydrate formation and transportability in pipelines was studied through
the analysis of experimental data obtained at the Center for Hydrate Research laboratory of the
Colorado School of Mines and two large scale flow loops (i.e., Tulsa University flow loop and
ExxonMobil flow loop). Chapter 4 presents a description of the flow loop facilities available during
this study, and the experimental procedure followed in a flow loop experiment. Chapter 5 presents a
set of conceptual pictures developed to explain the physical phenomena of gas hydrate formation in
flowlines. Different hydrate formation mechanisms are identified for different systems encountered
in oil/gas flowlines. For oil-dominated systems, dispersed water droplets are converted into hydrate
particles, and the agglomeration of these hydrate particles increase the slurry viscosity leading to
that pressure drop in the pipe depends on a flow regime transition involving multiple phases and
solid dispersion in the liquid phase. For gas-dominated systems, hydrates deposit and grow on pipe
24
walls, narrowing the flow diameter and increasing surface roughness.
A description of the mathematical models for hydrate formation in oil, water and gas-dominated
systems is presented in Chapter 5, including the models’ contributions to the transient conservation
equations. This work presents improvements to the hydrate aggregation module used for oil-
dominated systems, based on experimental data, which account for temperature, particle-particle
contact time, excess water, and the presence of surface active compounds. The model for water-
dominated systems consists of a mass transfer-based hydrate growth model and hydrate plugging
criterion, based on fluid velocity and liquid holdup. The model for gas-dominated systems consists
of a combined heat and mass transfer model for hydrate film growth on pipe walls. Examples of
applications of these models applied to a typical subsea well/flowline/riser geometry are presented
in Chapter 6.
Chapter 7 summarizes the main conclusions of this thesis, and Chapter 8 presents suggestions
25
CHAPTER 2
FORMATION
Gas hydrates are crystalline compounds formed by hydrogen-bonded water molecules in a lat-
tice structure that is stabilized by encapsulating a small guest molecule, like methane or ethane
(Sloan and Koh, 2008a). Hydrates form in the presence of appropriate quantities of gas and wa-
ter molecules, typically at high pressures and low temperatures. The hydrocarbon-water interface
represents an ideal location for the formation of gas hydrates in production systems.
Figure 2.1 illustrates a proposed model for hydrate formation in a multiphase flow system
containing water, oil, and gas, where hydrates form at the interface of water droplets entrained in
the oil phase, and/or from gas bubbles entrained in the water phase (Turner, 2005). Frequently
as little as 4 vol% of the water is formed into hydrate, when a plug occurs (Austvik, 1992). In
the oil phase, these hydrate-encrusted water droplets can agglomerate into larger hydrate masses,
leading to an increase in the slurry viscosity, which can eventually form a plug. The surface of water
droplets in water-in-oil (W/O) emulsions is a critical location for the formation and agglomeration
of hydrates. Another aspect that calls the attention of industry and academia are the so-called
naturally inhibited oils, which are characterized by having natural surfactants that seem to prevent
the agglomeration of hydrate particles and allow the transport of hydrate slurries without the
This chapter presents a review of surface chemistry concepts, with the principal objective to
identify interfacial phenomena and surface chemistry interactions involved in gas hydrate formation
and agglomeration in oil and gas pipelines. There are five types of interfaces where gas hydrates may
form and aggregate: gas/liquid, liquid/liquid, gas/solid, liquid/solid, and solid/solid; where gas is
hydrocarbon gas, liquid is either oil, water or condensate, and solid is either gas hydrate or pipe
wall surface. A review of fundamental interfacial concepts can help create a better understanding of
the phenomena at these interfaces. Two areas of surface chemistry have been selected to illustrate
the concepts and mechanisms associated with these systems: surfactants and emulsions. Examples
from the literature pertaining to gas hydrates are presented for each system.
26
Figure 2.1: Conceptual model for hydrate formation in a multiphase flow systems consisting of
water, oil, and gas; adapted from Turner (2005).
2.1 Surfactants
The presence of natural surfactants (acids/salts, alcohols, ethers, esters, and other organic com-
pounds) in oil systems and synthetic surfactants (corrosion inhibitors, hydrate anti-agglomerants,
hydrate kinetic inhibitors, demulsifiers, antifoaming agents, etc.) introduced into the well produc-
tion stream motivated the consideration of surfactants in this review. Surfactants, are amphiphilic
molecules which exhibit a double affinity for polar and non-polar substances. In its simplest con-
figuration, a surfactant molecule has two types of functional groups, a hydrophilic or polar group
chain with at least 12 carbon atoms; one example of a common surfactant is sodium dodecyl sulfate
Figure 2.2: Schematic of an anionic surfactant molecule, identifying the negatively charged hy-
drophilic group and the lipophilic group.
Surfactants can be classified, based on the dissociation of its hydrophilic group in water, in four
general types: anionic, cationic, nonionic and amphoteric (or zwitterionic). Anionic surfactants
have a hydrophilic group negatively charge that is associated to an inorganic metal (a cation, typi-
27
cally sodium). In water, the anionic surfactant molecule is divided in two free ions, the counter-ion
with positive charge and the anionic surfactant molecule with negative charge. Cationic surfac-
tants have a hydrophilic group positively charge associated to an inorganic anion (a counter-ion
with negative charge). The cationic surfactant is ionized in water by the separation of the cationic
surfactant monomer with positive charge and the counter-ion with negative charge. Cationic surfac-
tants are highly adsorbed in negative surfaces. Non-ionic surfactants have no counter-ion, resulting
in no electrical charges. Amphoteric surfactants exhibit two oppositely charged groups in their
hydrophilic part.
The balance of effects produced by the type, size and strength of the hydrophilic and lipophilic
groups will determine the properties of the surfactant molecule. The surfactant molecule will tend
ionized and the lipophilic hydrocarbon chain is relatively short (e.g., <12 carbon atoms in length).
On the other hand, a long hydrocarbon chain (e.g., >16 carbon atoms) will make the surfactant
molecule more oil soluble. The surfactant chemical structure and other parameters depending on
the different fluids, temperature and pressure can alter the affinity of surfactants with the water
Surfactant molecules present two fundamental properties: interfacial adsorption and self asso-
ciation, which are the essential processes leading to surfactants forming structures and enhancing
solubilization. The adsorption of surfactant molecules to interfaces is driven by its double affinity
to polar and non-polar substances, where a minimum in free energy of the system is attained.
The surfactant can diffuse from a bulk phase to an interface, such as gas/liquid (Figure 2.3(a)), liq-
uid/liquid (Figure 2.3(b)) and liquid/solid (Figure 2.3(c), d and e) interfaces, decreasing the surface
or interfacial tension, modifying the contact angle between the phases, modifying the wettability
of solid surfaces, and changing surface charge and surface viscosity (Shah, 1977). The surfactant
adsorption onto solid surfaces depends on the original wettability of the solid. The intrinsic solid
wettability depends on the nature of the solid and liquid substances, particularly their polarity.
The orientation of surfactant molecules depend on the system, in particular with respect to the
electric charges on a solid surface (Durbut, 1999). If the solid is oil-wet, the surfactant adsorbs
from the lipophilic tail, regardless of the surfactant type, as shown in Figure 2.3(c). If the solid
is water-wet (i.e., a polar surface), the surfactant molecules can interact with the polar surface
28
through their hydrophilic part, as shown in Figure 2.3(d). In Figure 2.3(d), a negatively charged
hydrophilic solid is shown with cationic surfactants adsorbed to the surface from the hydrophilic
head, attracted by electrostatic forces. This is a typical phenomenon associated with cationic sur-
factant such as, fatty amine salts and quaternary ammonium salts, used in applications of surface
hydrophobation, corrosion inhibition and solid particles dispersant. An increase in the cationic
surfactant concentration promotes the formation of surface bilayers or admicelles (Figure 2.3(e)),
through interactions between the lipophilic part of the surfactant molecules (Esumi, 2006).
Figure 2.3: Adsorption of a surfactant molecule at: (a) gas/liquid interface, (b) liquid/liquid
interface, (c) liquid/nonpolar-solid interface, (d) liquid/polar-solid interface, and (e) liquid/polar-
solid interface with increased cationic surfactant concentration.
Initially, when water soluble surfactants are added to an aqueous solution phase, surfactant
molecules adsorb to any available interface until the interface is saturated. With increasing sur-
factant concentration, surfactant molecules start to associate, forming spherical aggregates called
micelles. The surfactant association is driven by the hydrophobicity of its lipophilic group which
avoids mixing with water. The surfactant concentration at which the first micelle is formed is known
as the critical micelle concentration (CMC), and may be detected from a discontinuity in the change
of several variables such as surface tension, osmotic pressure, viscosity, electrical conductivity and
density (Preston, 1948). Figure 2.4 shows a typical plot of surface tension and electrical conduc-
the CMC.
The CMC depends upon the surfactant molecular structure and physicochemical conditions (i.e.,
temperature, pH, ion concentration). Above the CMC, additional surfactant molecules incorporate
into micelles, while the concentration of un-associated single surfactant molecules (monomers)
29
Figure 2.4: Variation of surface tension and electrical conductivity as a function of surfactant
concentration, indicating the discontinuity at the CMC.
remains nearly constant in a dynamic equilibrium between monomers and micelles, with constant
swapping of molecules between micelles and the bulk phase. A further increase in surfactant
concentration can promote the formation of more complex aggregates in the form of cylindrical,
hexagonal-packed, lamellar structures, worm like micelles, vesicles, liposomes, and other structures
(Shah, 1977). Inverse micelles are formed in oil-soluble surfactants when the hydrophilic part
aggregates to avoid mixing with the oil phase. In the case of inverse micelles of ionic surfactants, a
minimum number of hydrated water molecules is needed that reduces the repulsive forces between
the head groups (Lee, 2008). Figure 2.5 shows a schematic of normal and inverse micelles.
(oil) inside normal micelles in the aqueous phase or water inside inverse micelles in the oleic phase.
Above the CMC, the solubilization increases by incorporation of a non-soluble phase inside the
micelles. When the formulation of the solution is close to the hydrophilic-lipophilic balance of the
surfactant with both oil and water phases, a microemulsion phase can be formed. Bicontinuous
microemulsions are typically defined as containing normal and inverse swollen micelles (filled with
oil and water), which are combined to result in a single phase structure, as shown schematically in
Figure 2.6, where a close to zero curvature of a surfactant layer is limiting the oil and water zones
30
Figure 2.5: Association of anionic surfactant molecules forming normal micelles in an aqueous
phase and inverse micelles in an oleic (oil) phase.
Figure 2.6: Bicontinuous structure of microemulsions formed by associating both types of micelles
and resulting in a zero curvature surfactant film that separates oil and water zones (Salager et al.,
2005).
The association of surfactant molecules in the form of micelles at concentrations above the CMC
increases the gas hydrate formation rate and reduces the induction time in quiescent systems, which
is of special interest for the application of gas hydrates in the storage and transportation of natural
gas (Kalogerakis et al., 1993). Zhong and Rogers (2000) studied the effect of SDS on the hydrate
formation mechanism using ethane and natural gas as guest molecules. They suggested that micelles
act as a nucleation point by increasing the solubility of hydrocarbon gas in the aqueous phase and
by inducing the formation of hydrate crystals around the micelle in the bulk water phase below
31
the gas/water interface. Consequently, the hydrate formation rate was observed to increase by
more than 700 times and the induction time for nucleation decreased compared to systems without
surfactant. Other studies had obtained similar effects with different surfactants, but invoked a
different mechanistic explanation, questioning the CMC formation requirement (Lee et al., 2010;
The morphology of the hydrate film that forms at hydrocarbon/water interfaces is affected
by the adsorption of surfactant molecules at the interface. Luo et al. (2007) studied gas hydrate
formation in a methane bubble column without surfactants, observing the formation of a hydrate
shell around gas bubbles, which hindered further formation of gas hydrates. The effect of SDS,
below the CMC, on hydrate formation on gas bubbles using a static mixer was studied by Tajima
et al. (2009), who reported an increase in hydrate formation rate and a change in the morphology
of the hydrate film with the addition of SDS. The adsorption of surfactant at the bubble interface
promoted the formation of a rougher hydrate film with a weaker structure that easily collapsed;
the increase in hydrate formation rate was attributed to higher mass transfer through the hydrate
In another study, Kalogerakis et al. (1993) compared the effect of anionic and nonionic sur-
factants in a stirred cell, reporting a greater increase in methane hydrate formation rate with an
anionic surfactant (SDS). From the experiments with anionic surfactant, aggregates of hydrate par-
ticles suspended in the liquid phase were observed, increasing the slurry viscosity. Hydrate growth
on the reactor wall was attributed to a more water-wet wall due to the anionic surfactant solution.
The effect of anionic surfactants seems to be detrimental for the transportability of hydrate slurries
and should be avoided in pipelines, while nonionic surfactants seem to prevent the agglomeration
of hydrate particles.
These studies exemplify the changes in the interfacial properties between the gas and water
phases caused by the presence of surfactants. Surfactants tend to promote the formation of hydrates
at the interface by lowering the interfacial tension at the water/gas interface, thus facilitating the
32
2.1.2 Effect of surfactants on contact angle, wettability, and cohesion/adhesion forces
of hydrate particles
The wettability of a solid surface by a liquid is determined by the contact angle, θ, which a
liquid drop makes with the surface at the three-phase contact point (see Figure 2.7). Young’s
equation (eqn. 2.1) represents the relationship of the balance of interfacial tensions between the
solid, i.e., interfacial tension of the liquid phases, σwo , and the solid free energies with the liquids,
σsw and σso . A solid is water-wet if the water drop contact angle is lower than 90o , and it is oil-wet
if the contact angle is greater than 120o . For intermediate values of the contact angle, there is no
Figure 2.7: Balance of interfacial tensions among the three phases at contact point, defining contact
angle of the liquid with the solid surface (left: solid is oil-wet; right: solid is water-wet).
Young’s equation is useful to determine the wetting equilibrium of a solid. However, the wetta-
bility can be determined thermodynamically using the spreading coefficient defined in terms of the
Spontaneous spreading of the water drop occurs when Sw/s > 0, i.e., decreasing the interfacial
tension between water-oil and solid-water. The addition of a surfactant that decreases the interfacial
tension between water-oil and solid-water can make the solid more water-wet. However, the actual
33
It is often difficult to determine the wettability of hydrate surfaces, as a flat and smooth surface
is required for measurements of the contact angle. Høiland et al. (2005) qualitatively studied the
wettability of freon hydrates in water/oil emulsions by comparing the emulsion inversion point
of different crude oils with and without hydrate particles, demonstrating that the wettability of
hydrate particles depends on the oil composition. Oil-wet hydrate particles are attributed to the
adsorption of natural surface active components of crude oils on the hydrate surface, which are
correlated to have a low hydrate plugging risk in pipelines. Intermediate and water-wet hydrate
particles are believed to contain less or none of these oil-wetting components, and are correlated
to have a high hydrate plugging risk. Natural oil components that interact with hydrate surfaces
are characterized to have polar functional groups and to be surface active, such as petroleum acids
Aspenes et al. (2010b) studied the influence of oil composition on the wettability of pipeline
surfaces, suggesting that the same natural components of oil that interact with hydrate surfaces
also influence the pipeline surfaces. It was suggested that oils with high acids content compared to
bases are more likely to form oil-wetting surfaces, resulting in lower hydrate plugging tendency.
The non-plugging hydrate tendency of systems with oil-wet hydrate particles and pipeline sur-
faces is supported by the work of Dieker et al. (2009) where micromechanical force measurements
were used to quantify the cohesion forces between cyclopentane hydrate particles. An increase
in the cohesion forces between hydrate particles was measured when the surface active compo-
nents were removed from the crude oil. Figure 2.8 is a summary of the cohesion force measured
by Dieker et al. (2009) for the Caratinga crude oil, which is known to be a naturally-inhibited,
hydrate-transportable oil. From a SARA analysis the Caratinga oil has an asphaltene content of
6.2%, saturates 39.8%, aromatics 39.8% and resins 14.3% (Sjöblom et al., 2010). Note the increase
in cohesion force when the acidic components where extracted from the Caratinga oil (through a
“pH-14 extraction” process), and a further increase in cohesion forces when the asphaltenes were
extracted. The observed increase in cohesion forces correlates well with an increase in interfacial
tension from its original value of 23 mN/m to 31 mN/m after the removal of acids, which implies
that the absence of surface active compounds on the hydrate surface contribute to the increment
34
Figure 2.8: Summary of cohesion forces between cyclopentane hydrates particles in the presence
of Caratinga crude oil, measured by Dieker et al. (2009).
Additional evidence that associates oil-wet surfaces with low hydrate plugging tendency is
provided by the work of Aspenes et al. (2010a) that studied adhesion forces between cyclopentane
hydrate particles and different solid surface materials. It was concluded that adhesion forces are
higher for water-wet pipeline surfaces, which may be part of the mechanism for hydrate plug
formation, and that addition of petroleum acids to the oil phase drastically reduce adhesion forces.
At this point the importance of the role of surfactants can be seen, with the surfactant funda-
interfaces and defining the wettability of hydrate particle and solid surfaces.
2.2 Emulsions
In this section, a brief discussion of fundamental concepts and mechanisms of emulsions behavior
will set the framework to understand the effect of emulsions on hydrate formation/dissociation and
In the petroleum industry, two types of emulsions are frequently encountered: water-in-oil
(W/O) and oil-in-water (O/W) emulsions. Emulsions are defined as thermodynamically unstable
mixtures of two immiscible liquids (one continuous and one dispersed), which means that they will
separate over time. The separation of the immiscible phases will depend on the rate of coalescence
of the dispersed phase. The stability of emulsions can be extended by kinetics means, making the
separation time longer through complex interfacial mechanisms that delay the coalescence of the
35
dispersed phase.
The formation of an emulsion requires at least three components: two immiscible phases (water
and oil) and an emulsifying agent, which may be a natural component of one phase or an emulsifying
agent may be added to one of the phases. The emulsifying agent usually is a surfactant that adsorbs
at the interface between the immiscible phases; some natural surface-active components in crude oils
include asphaltenes, resins with organic acids and bases, naphthenic acids, and carboxylic acids,
among many others (McLean and Kilpatrick, 1997; Sjöblom et al., 2003, 2007). Solid particles
can also act as stabilizing agents, as in examples of sand, clays, mineral incrustations, corrosion
products, paraffin wax, and precipitated asphaltenes (Sjöblom et al., 2003, 2007; Sullivan and
Kilpatrick, 2002).
The rupture mechanism of an emulsion involves three basic steps: (i) long distance approach of
dispersed drops by sedimentation and creaming, (ii) drainage of the thin film between drops, and
(iii) coalescence of drops to become a single larger drop. The kinetic stability of an emulsion is
attained by mechanisms that influence the thin film drainage. Four mechanisms have been listed as
the principal mechanisms acting to stabilize emulsions: (1) electrostatic repulsion between charged
interfaces, (2) steric repulsion, (3) Marangoni-Gibbs effect, and (4) formation of a rigid cross-linked
network adsorbed at the droplet interface (McLean and Spiecker, 1998; Sullivan and Kilpatrick,
2002). A schematic representation of the emulsion stability mechanisms is presented in Figure 2.9.
When two droplets come into contact, the charged surfaces repel each other preventing ag-
glomeration and coalescence (Becher, 2001; Schramm, 2005); this mechanism plays a significant
role in stabilizing O/W emulsions (McLean and Spiecker, 1998; Schramm, 2005). Steric repulsion
prevents approaching droplets from coalescing due to physical barriers of adsorbed material such as
polymer or surfactant chains (Schramm, 2005). The Marangoni-Gibbs effect describes the process
36
whereby an expanding surface shows a subsequent decrease in surfactant concentration and increase
in interfacial tension; the interfacial tension gradient generated provides the driving force for the
flow of surfactant to areas of high tension, opposing film thinning (McLean and Spiecker, 1998;
Schramm, 2005). The adsorption characteristics of the surfactant has a direct effect on the kinetics
phenomena (Salager, 2000a). The formation of rigid cross-linked networks at droplet interfaces is
associated with asphaltenes (McLean and Spiecker, 1998). Even low amount of resins in the oil
phase can improve the stability of emulsions by breaking up large asphaltene aggregates, which
are then more mobile and hence able to adhere at water-oil interfaces forming a strong interfacial
Emulsion properties, such as emulsion type, droplet size distribution, stability and rheology
depend mostly on the following formulation variables: nature of the oil (equivalent alkane carbon
number, EACN), salinity of the aqueous phase (type and concentration of electrolyte), nature and
the generalized formulation and the trend variation, i.e. the increase or decrease of the different
emulsion properties, whose intensity is influenced by the previous mechanisms (Salager, 2000b).
Other variables which influence emulsion properties are pressure, order of mixing, and energy
input.
With the studies of surfactant-water-oil systems, different formulation approaches have been
developed that can be used to predict or estimate the emulsions type and properties (Solans et al.,
1996). Developed from empirical observations, the Bancroft (1912) rule states that the continuous
phase will be the one in which the surfactant is most soluble. Another empirical method is the
Hydrophilic-Lipophilic Balance (HLB), which assigns a number to a surfactant based on its solu-
bility in water (Griffin, 1949). Oil-soluble (lipophilic) surfactants have an HLB number lower than
9, while water-soluble (hydrophilic) surfactants have an HLB number greater than 11. A general
rule-of-thumb is that a surfactant with an HLB ranging 4-6 will tend to form a water-in-oil (W/O)
emulsion and an HLB ranging 8-18 will tend to form an oil-in-water (O/W) emulsion; this method
correlates well with the Bancroft rule. These methods focus only on the nature of the surfactant
and do not consider the effects of other formulation variables (e.g., temperature, salinity of water,
37
A theoretical concept was introduced by Winsor (1954) that relates the surfactant with its
physico-chemical environment at the interface between water and oil phases, describing three ele-
mentary types of phase behavior. The Winsor ratio (R = ACO /ACW ) is defined as the ratio of
the net interaction energy of the surfactant with the oil phase (ACO ) to the net interaction energy
of the surfactant to the aqueous phase (ACW ). When the interaction of the surfactant with the
aqueous phase is greater than the interactions with the oil phase (R < 1), a type I phase behavior
results, consisting of a pure oil phase and an oil-in-water microemulsion phase with normal micelles
of surfactant solubilizing oil; this type of system will tend to form an O/W emulsion. If the inter-
action of the surfactant with the oil phase is greater than that of water (R > 1), a type II phase
behavior is exhibited, consisting of a pure aqueous phase and a water-in-oil microemulsion phase
with inverse micelles solubilizing water; this type of system will tend to form a W/O emulsion. The
type III phase behavior corresponds to equal interactions of the surfactant with oil and aqueous
phases (R = 1), consisting of three phases in equilibrium: a microemulsion containing the surfac-
tant which solubilizes large amounts of oil and water, and a pure oil and a pure water phases. A
schematic representation of the three Winsor phase behavior types is presented in Figure 2.10.
Figure 2.10: Schematic representation of phase behavior described by Winsor for different values
of the ratio of net interaction energy of the surfactant with the oil phase to the net interaction
energy of the surfactant to the aqueous phase (Salager, 2000a).
A change in the formulation variables can promote a change in the interaction energies of
the surfactant with the oil and aqueous phases, and therefore a change in the phase behavior. For
example, an increase in the salinity concentration of the water phase would decrease the interaction
energy of an anionic surfactant with the aqueous phase, thus promoting a change from type I to
type III, and a further increase in the salinity will eventually lead to a type II phase behavior
where all the surfactant molecules are in the oil phase. The Winsor R ratio can be used to make
38
qualitatively estimations of the changes in phase behavior, and ultimately in the emulsion type,
The surfactant affinity difference (SAD) can be considered as a generalized formulation parame-
ter that provides a quantitative approach of the relationship between the phase behavior described
by Winsor and the formulation variables (Salager, 1999). The concept of the effect of different
formulation variables in a single expression was presented by Salager et al. (1979) and Bourrel
et al. (1980) which was then called surfactant affinity balance (Salager et al., 1982). Using the
definition of surfactant affinity deviation or difference as the negative of the chemical potential,
SAD = µ◦w − µ◦o , where µ◦w and µ◦o are the standard chemical potentials of the surfactant in the
water and oil phases, respectively. When the surfactant affinity for the water phase equals its affin-
ity for the oil phase, SAD = 0, which is equivalent to a Winsor type III phase behavior (R = 1).
When SAD < 0, the affinity of the surfactant towards water dominates, forming O/W emulsions
(Winsor type I). For SAD > 0, the affinity of the surfactant is stronger for the oil phase, forming
W/O emulsions (Winsor type II). For ionic and nonionic (polyethoxylated) surfactants, the SAD
where S is the salinity, K is a constant, ACN the alkane carbon number or its equivalent (EACN )
when the oil phase is not characterized by a pure hydrocarbon chain, f (A) is the contribution of the
type and concentration of co-surfactant like alcohol, σs and αs are the characteristic parameters of
the surfactants, EON is the number of ethylene oxide groups in nonionic surfactants, and (T −Tref )
is the temperature difference relative to some reference temperature (Salager, 1999). Each variable
results in a change of at least one of the interaction energies of the surfactant with either aqueous
or oil phase. For instance, an increase in salinity (S) or surfactant lipophilic group length (and thus
σs ) produces an increase in SAD, while an increase in the ACN of the oil phase or in temperature
produces an opposite effect decreasing SAD in eqn. 2.3 for ionic surfactants.
The effect of the formulation variables on the emulsion properties has been studied using the
SAD, considering similar amounts of water and oil and low quantities of surfactant (Salager, 1996).
39
Figure 2.11 summarizes the effect of the formulation variables on the type of emulsion, stability,
interfacial tension, viscosity, electrolytic conductivity and drop size, indicating the three phase
region (Winsor type III phase behavior) with a shaded band in the middle of each diagram. On
the abscissa of each portion Figure 2.11 is the formulation scan shown in the upper left subplot of
the figure. When the formulation variable is increased from SAD < 0, a transition in the emulsion
type from O/W to W/O and in the emulsion properties occur at SAD = 0, because the affinity of
the surfactant switches from hydrophilic to lipophilic. The emulsion stability and viscosity undergo
a minimum at SAD = 0, which is related to a low interfacial tension. The electrical conductivity
decreases drastically inside the three phase region, indicating the phase inversion.
Figure 2.11: Effect of formulation variables on type and properties of emulsions (Salager, 1999).
The previous representation of the effect of a single generalized formulation variable on the
emulsion properties was extended by Salager et al. (1983) to include the effect of the water-oil
ratio at constant surfactant concentration in the form of a bi-dimensional map. Figure 2.12 shows
a diagram of the bidimensional map, where the ordinate represents the generalized formulation
variable SAD and the abscissa represents the water-oil ratio (WOR). From the diagram, it is
possible to predict the emulsion type and the emulsion inversion based on the formulation variable
and the WOR. The formulation divides the map in two halves, with the upper half corresponding
40
to SAD > 0 suggesting a W/O emulsion, and a lower half corresponding to SAD < 0 suggesting an
O/W emulsion. The WOR divides the map in three vertical sections: the left section represents low
WOR (oil continuous emulsion); the right section represents high water content (water continuous
emulsion), and middle section will be controlled by the formulation. There are two sections where
extreme WOR value causes formation of an emulsion type different than the one suggested by
the formulation; these sections are associated with high instability and varying degrees of multiple
emulsions. For the lower left zone (B-) with oil in water in oil (o/W/O) emulsions, the formulation
suggests an O/W emulsion, but the excess oil becomes the continuous phase, hence resulting in a
multiple emulsion that satisfies both trends. Similarly the upper right zone (C+) with water in
oil in water (w/O/W) emulsions is where excess water is the continuous phase surrounding W/O
emulsion droplets.
In oil dominated flowline systems, water-in-oil emulsions play a critical role in the formation
and agglomeration of hydrate particles in pipelines. Previous studies have shown the formation of
hydrates at the interface of water droplets, forming hydrate shells around the water droplet. In
the study of gas hydrate formation with low water cuts (< 35vol%) in W/O emulsions using a
laboratory-scale high-pressure autoclave cell and an industrial-scale flow loop, Turner (2005) found
41
that dispersed water droplets convert directly to hydrate particles (formed from the gas dissolved
in the oil phase), maintaining the droplet size distribution during hydrate formation, and thus
suggesting the formation of a hydrate shell at the water droplet interface; a particle size analyzer,
based on the Focus Beam Reflectance Method (FBRM, Ruf et al., 2000) was used to measure the
water droplet and hydrate particle size distribution. To explain how the hydrates form in the water
interface, Taylor et al. (2007b) studied hydrate film growth at a flat hydrocarbon/water interface
using video microscopy combined with gas consumption measurements, using cyclopentane and
methane as hydrate formers. From the hydrate film growth experiments, they proposed a hydrate
film formation mechanism which consisted of three stages: (i) initial formation of a thin porous
hydrate film propagating across the interface, (ii) film thickening over time growing into the water
phase (shrinking core) and filling of the pores, and (iii) bulk conversion of hydrate film with filling
of the remaining pores. It was also proposed that the hydrate film formation mechanism could
be extrapolated to a water droplet interface following the same three stages and ending with full
conversion of water (inside the shell) into hydrate over a long period of time.
Taylor et al. (2007b) also developed a model to calculate hydrate growth in W/O emulsions
that considered the hydrate shell formation around the water droplets followed by the conversion
of the remaining water core to hydrate, in a similar mechanism to that described above. This
model considered the mass transfer limitations of methane across the boundary layer at the gas-oil
interface, through the oil phase and across the hydrate shell, to determine the hydrate growth rate
in the interior of the water droplet, with the assumption of constant droplet size and ignoring heat
transfer from the hydrate particles to the surroundings due to the exothermic heat of crystallization.
The methane diffusivity through the hydrate shell was obtained from experimental measurements.
The model gave a reasonably good fit to the experimental data, even though the initial hydrate
growth rate was under-estimated and the long term hydrate growth rate was over-estimated, these
The formation and dissociation of hydrates can affect the stability of emulsions, leading to the
generation of a free water phase from the agglomeration and coalescence of water dissociated from
hydrate particles. The presence of a free water phase increases the risk of forming a hydrate plug
42
in pipelines. Lachance et al. (2008) studied the effect of hydrate formation and dissociation in the
stability of W/O emulsions with different crude oils using differential scanning calorimetry (DSC).
From the comparison of results obtained with four crude oils, they found that hydrate formation
and subsequent dissociation destabilized W/O emulsions in oils with a low fraction of asphaltenes
or when a high concentration of resins prevented the deposition of asphaltenes onto the water
droplet interface, while oils with a higher fraction of asphaltenes resisted destabilization. They
hydrate encrusted water droplets agglomerate, and a dissociation step, where the release of gas
molecules causes the interface to break resulting in coalescence of adjacent droplets. The resistance
to destabilization by oil with higher fractions of asphaltenes can be explained by the formation of
In high water content W/O emulsions (> 35vol%), the formation and dissociation of hydrates
also affect the emulsion stability, leading to an inversion of the emulsion from W/O to O/W or
the generation of multiple emulsions. Greaves et al. (2008) studied emulsions with water content
greater than 60 vol% in a high-pressure autoclave cell equipped with two particle size analyzer
probes (Focus Beam Reflectance Method, FBRM, and Particle Video Microscope, PVM). With
reference to Figure 2.12, these emulsions can be located in the upper half of the bi-dimensional
map (SAD > 0) and near the vertical inversion line. In a similar mechanism to that proposed
by Lachance et al. (2008), hydrate encrusted water droplets first agglomerated and then coalesced
during dissociation, causing the appearance of an aqueous continuous phase due to the large amount
of water in the system, destabilizing the W/O emulsion to form a w/O/W emulsion. Greaves et al.
(2008) also showed using in situ high pressure particle imaging analysis (from the PVM probe)
that hydrate dissociation can result in extensive hydrate agglomeration (see Figure 2.13 of PVM
images of hydrate agglomeration during hydrate dissociation), thereby confirming the calorimetric
results by Lachance et al. (2008) which indicated that hydrate dissociation can lead to emulsion
Different mechanisms have been suggested for the agglomeration of hydrate particles in oil dom-
inated systems. From the expected hydrophilic character of the hydrate surface, it was suggested
43
Figure 2.13: PVM images of 68 vol% of water dispersed in crude oil: a) initial W/O emulsion, b)
hydrate formation around water droplets, c) large hydrate agglomerates during hydrate dissociation,
c) w/O/W emulsion after hydrate dissociation (Greaves et al., 2008).
that capillary forces, formed by a liquid bridge between the particles, hold the particles together
(Austvik et al., 2000). The capillary bridge theory (Israelachvili, 1991) was used as an explanation
two hydrate particles. The inter-particle cohesion force increased when the temperature was raised
toward the melting temperature, and a quasi-liquid water layer formed at the particle surfaces
(Taylor et al., 2007a; Yang et al., 2004). The capillary bridge was also observed in the study of
micromechanical adhesion forces between hydrate particles and pipeline solid surfaces with a water
drop deposited on the solid surface (Aspenes et al., 2010a), increasing the adhesion force to more
than 10 times of that between hydrate particles, indicating possible deposition of hydrates on pipe
According to the capillary bridge theory, the capillary force between two hydrate particles is
proportional to the interfacial tension between the bridging liquid water and the continuous oil
phase. A decrease in the interfacial tension between water and oil decreases the capillary forces
and prevent the agglomeration of hydrate particles that may be separated by shear forces during
flow; this is one of the principles of hydrate anti-agglomeration agents (or anti-agglomerants, AA).
44
From studies of hydrate slurry viscosity (Fidel-Dufour et al., 2006), a hydrate agglomeration
mechanism was proposed which includes crystallization of a water droplet by the contact with a
hydrate particle, and growth of the resulting agglomerate with the subsequent contact of water
droplets. This mechanism was used for the development of a hydrate slurry rheology model by
Camargo and Palermo (2002), in which the viscosity of the hydrate slurry increases with agglom-
erate size. The adsorption of a surfactant on the hydrate particle surface, can prevent the contact
with water droplets and other hydrate particles by changing the wettability of the hydrate particle
surface to oil-wet and by creating steric repulsions; this is another principle of anti-agglomerants.
liquid bridging, has been the principle of new technologies referred to as the stabilized cold flow
concept, in which flowable slurries of dry hydrate particles are generated (Gudmundsson, 2002;
Lund and Larsen, 2000; Talley et al., 2007; Wolden et al., 2005). The conversion of excess free
water into small and dry hydrate particles is the key to eliminate all chances of agglomeration in
Anti-agglomerants are surfactants which have the purpose to disperse hydrate particles in the
oil phase and keep them apart, preventing agglomeration. Some mechanisms that contribute to
the anti-agglomeration of solid hydrate particles are common to stabilization of emulsions, like
steric repulsions. Other mechanisms involved in the anti-agglomeration of hydrates consider the
nature of the agglomeration phenomenon, which can be different from the emulsion rupture mech-
anism; studies on anti-agglomerants indicate that emulsification is not required for an effective
anti-agglomerant (York and Firoozabadi, 2008, 2009). From a theoretical analysis of the forces in
agglomeration and stabilization of hydrate particles (i.e., capillary and dispersion forces), Anklam
et al. (2008) concluded that an effective anti-agglomerant is one that: (a) reduces the size of hy-
drate particles through the reduction of water droplet size in the W/O emulsion, (b) decreases the
interfacial tension between water and oil phases to reduce capillary bridge forces, and (c) modifies
the wettability of the hydrate surface to oil-wet by increasing contact angle through the water
phase. These features for an effective anti-agglomerant agent can be attained considering the effect
45
system, using as a reference a diagram like the one presented in Figure 2.11, where the formulation
conditions that provide minimum droplet size, low interfacial tension, and low emulsion viscosity
can be identified.
number of papers. York and Firoozabadi (2009) studied the effect of adding salt (NaCl and MgCl2 )
model hydrate former). They observed that the addition of salt (NaCl or MgCl2 ) eventually results
in agglomeration of hydrates, with bivalent salts (MgCl2 ) being more detrimental, and non-ionic
biosurfactants being more effective at lower salt concentrations than the anionic surfactant. The
addition of both salts tested increased the SAD, making the surfactant more lipophilic with less
adsorption at the interface. With the anionic surfactant, agglomeration was observed at low salt
(NaCl and MgCl2 ) concentrations and high surfactant concentrations, restoring anti-agglomeration
behavior with the addition of salt until a point where agglomeration was observed again, indicating a
range of concentrations with good anti-agglomeration behavior. Others also studied the effect of salt
the increased concentration of salt in the unconverted water phase (York and Firoozabadi, 2008).
The addition of low concentrations of an alcohol, which acts as a co-surfactant, can change
the formulation improving the anti-agglomeration property of some surfactants. The addition of
methanol (MeOH, 0.5 – 5.0 wt.%) as a co-surfactant has been proven, at laboratory scale, to
decrease the required concentration of anti-agglomerant surfactant, maintaining the anti agglom-
eration properties of the system (Gao, 2009; York and Firoozabadi, 2008). In one study (York
and Firoozabadi, 2008), the improved anti-agglomeration characteristic was explained by a syn-
ergistic contribution of MeOH that increased the surfactant density at the interface, reducing
further the interfacial tension, and thereby decreasing water droplet size conducing to small hy-
drate particles. This change in the formulation modified the emulsion properties to coincide with
a minimum in droplet size as shown in Figure 2.11. In another study (Gao, 2009), it was shown
that anti-agglomeration was improved by adding 15 wt.% of MeOH and 0.5 wt.% of (un-identified)
anti-agglomerant to a system with 80% water cut, explaining the anti-agglomeration improvement
46
by the increase of MeOH concentration in the free water phase that thermodynamically inhibits
2.3 Summary
Typical hydrate prevention methods, based on shifting the hydrate equilibrium phase boundary
by adding thermodynamic inhibitors, focus on the treatment of the aqueous bulk phase requiring
large concentrations of methanol or monoethylene glycol. The gas and oil industry is currently
in a transition from hydrate avoidance to risk management with the use of inhibitors that act at
the interfaces of hydrate particles (kinetic hydrate inhibitors and anti-agglomerants), which has
been proven to reduce the required concentrations of chemical additives, i.e., a more economical
approach. This fact highlights the importance of considering surfaces and interfaces in the study
of gas hydrates.
Revisiting the conceptual model for hydrate formation in a multiphase flow system consisting
of water, oil, and gas presented in Figure 2.1, it is found that interfaces between water and hy-
drocarbon phases are ideal locations for gas hydrate formation, and that interaction mechanisms
between hydrate particles and solid surfaces may or may not result in the blockage of pipelines,
depending on several factors, such as oil composition (presence of natural surfactants), wettability
of solid surfaces, formulation variables and their effect on anti-agglomerants performance, presence
of free water, and particle size among others. Understanding the fundamental concepts of surface
chemistry, in particular surfactants (molecules with preference to interfaces) and emulsions (mix-
tures of two immiscible fluids with abundant surface area), is of great value for the research, design,
and development of novel techniques for the management of gas hydrates in oil and gas pipelines.
In this review the relationship between surface chemistry and gas hydrates was presented using
a number of studies available in the technical literature to provide a clear connection of the current
47
CHAPTER 3
Observations in flow loop experiments of gas hydrate formation in high water cut systems showed
that pressure drop in pipelines depends on a complex flow regime involving multiple phases and solid
dispersion in liquid (Joshi et al., 2011a). These observations require merging two known phenomena
of multiphase flow: (i) multiphase (gas-liquid) flow regimes, which can be useful in estimating
pressure drop in the pipeline and surface area between water and hydrocarbon phases; and (ii)
slurry flow consisting of solid particles dispersed in the liquid phase during hydrate formation.
The purpose of this chapter is to present a review of fundamental concepts and models of steady-
state and transient multiphase flow in pipes, and identify state-of-the-art methods that could be
extended for the prediction of gas hydrate formation and transportability in flowlines.
This chapter starts with a review of single phase flow concepts, presented in Section 3.1. The
second section (Section 3.2) presents an extension of single phase flow concepts to multiple phases,
introducing the fundamental concepts used to describe multiphase flow in pipes. Section 3.3 presents
a discussion of multiphase flow modeling approaches. Finally, Section 3.4 presents the derivation of
a simple hydrodynamic slug flow model and its extension to predict hydrate formation, to illustrate
This section presents the derivation of the steady-state pressure-gradient equation for single-
∂ρ ∂ρv
=− (3.1)
∂t ∂x
For steady-state flow, the transient term disappears and the conservation of mass is given by,
∂ρv
=0 (3.2)
∂x
48
The conservation of momentum is,
∂ρv ∂ρv 2 ∂p πd
=− − − τw − ρg sin θ (3.3)
∂t ∂x ∂x Ap
The pressure-gradient equation is obtained by combining the mass and momentum balance
∂p πd ∂v
= −τw − ρg sin θ − ρv (3.4)
∂x Ap ∂x
where p is pressure, x is the axial coordinate along the length of the pipe, τw is the wall shear
stress, d is the pipe internal diameter, Ap is the pipe cross sectional area, ρ is the fluid density, θ is
the pipe angle of inclination with respect to the horizontal, v is the fluid linear velocity. The first
term on the right side of equation 3.4 represents the friction or shear stress at the pipe wall (5 to
20% of total pressure drop), the second term is the pressure drop due to elevation change (80 to
95% of total pressure drop in vertical flow), the third term represents the change in velocity (i.e.,
The wall shear stress, τw , can be evaluated through the use of a dimensionless friction factor.
In this case, the Moody friction factor is used, defined as the ratio of wall shear to kinetic energy
pressure-gradient equation,
f ρv 2 πd f ρv 2
∂p πd
= τw = = (3.6)
∂x f Ap 8 πd2 /4 2d
Next, the friction factor needs to be determined. For laminar flow, the friction factor is a
istics (i.e., relative roughness, ε/d). For smooth pipe (ε/d → 0) the friction factor can be evaluated
f = 0.0056 + 0.5Re−0.32 , for 3000 < Re < 3 × 106 (Drew and McAdams, 1932) (3.8)
49
or,
For complete turbulence and fully rough wall pipe (i.e., region in Moody diagram where f is a
function of ε/d only), the friction factor can be evaluated using Nikuradse’s (1933) correlation,
1 2ε
√ = 1.74 − 2 log (3.10)
f d
In the transition region between laminar and fully turbulent flow, the friction factor is a function
of both the Reynolds number and relative roughness, and can be evaluated using the Colebrook
direct substitution method is used, where the calculated value is used as the next assumed value.
The initial guess value can be obtained from one of the explicit smooth-pipe equations or from the
following explicit approximation to the Colebrook equation (Zigrang and Sylvester, 1985),
1 2ε/d 5.02 2 ε/d 13
√ = −2 log − log + (3.12)
f 3.7 Re 3.7 Re
Initial values for roughness, ε, are needed for design calculations, some suggested values found
• For tubing exposed to an environment that causes significant changes in roughness, very dirty
The term multiphase flow refers to any fluid flow consisting of more than one phase. One of the
simpler cases of multiphase flow, which have been the subject of numerous studies, consists of two-
phase gas/liquid flow (Brill and Mukherjee, 1999). This case is used to introduce the fundamental
multiphase flow concepts and the different approaches used to model these systems. The presence
50
of dispersed phases within continuous phases presents a subsequent level of complexity. Three
phases exist during the transport of hydrocarbons in pipelines; gas, oil and water. Depending on
the amount and flow rates of each phase, different configurations can be observed; these will be
The holdup for each phase (gas or liquid) is defined as the fraction of pipe cross sectional area
Ak
Hk = (3.13)
Ap
where the subscript k denotes the phase, either gas (g) or liquid (L). In general, empirical correla-
tions for multiphase flow are based on the prediction of liquid holdup (HL ).
HL + Hg = 1 (3.14)
An important correlating parameter is the superficial velocity, which assumes that a given phase
occupies the entire pipe area, and it is defined as the volumetric flow rate of the phase (Qk ) divided
Qk
vSk = (3.15)
Ap
The mixture velocity is defined as the sum of the superficial velocities.
QL + Qg
vm = = vSL + vSg (3.16)
Ap
The phase linear velocities are related to the superficial velocities and the liquid hold up by,
vSk
vk = (3.17)
Hk
The gas-liquid slip velocity is defined as the difference between the linear velocities,
vSg vSL
vs = vg − vL = − (3.18)
1 − HL HL
A condition of no-slip between the phases indicates that both the gas and liquid would flow at
Flow pattern or flow regime is a classification of the different physical distributions of the
phases in the flow conduit. Figure 3.1 shows different flow regimes identified for gas/liquid flow in
horizontal pipes, the flow regimes for vertical pipes are shown in Figure 3.2.
51
Figure 3.1: Flow regimes for gas/liquid flow in a horizontal pipe (Brennen, 2005, p. 167).
Figure 3.2: Flow regimes for gas/liquid flow in a vertical pipe (Brennen, 2005, p. 170).
52
Another type of two-phase flow is solid/liquid mixtures or slurry flow, which consists of a
dispersion of solid particles in a continuous liquid phase. With increasing solid particle volume
fraction, different slurry flow regimes can occur, as illustrated in Figure 3.3 for sand and water.
From a homogeneous distribution of well-mixed solid particles in the liquid phase at low particle
ating a vertical gradient in the particle concentration, to a moving bed of particles, and finally to
a stationary bed where the packing of particles prevents its movement in the pipe.
Figure 3.3: Flow regimes for slurry (solid particles of sand dispersed in water) flow in a horizontal
pipe (Brennen, 2005, p. 169).
The methods used to model multiphase flow vary widely according to the assumptions made
in their development. The homogeneous fluid model approach assumes that the two-phase flow
can be treated as single phase flow by using average properties for the mixture. In the separated-
flow approach, a steady-state momentum balance equation is developed for gas and liquid phases,
assuming that the two phases flow through separate conduits, and either equation can be used to
calculate pressure drop. The phenomenological or mechanistic approach used to develop steady-
state models represents a compromise between empirical correlations and conservation laws, where
basic physical laws are used to model flow phenomena, such as flow pattern prediction. The
transient two-phase flow modeling approach involves the simultaneous solution of conservation
equations of mass, momentum and energy for each phase (gas and liquid), and uses empirical
correlations and simplified constitutive relationships for flow parameters, such as interfacial friction,
fraction of liquid entrained in the free gas phase, and liquid holdup in the slug body.
53
3.3.1 Homogeneous fluid model approach
The single-phase pressure gradient equation can be modified for multiphase flow by considering
dp f ρf vf2 dvf
= + ρf g sin θ + ρf vf (3.19)
dx 2d dx
where the subscript f denotes the homogeneous fluid. The homogeneous fluid properties are cal-
culated using different mixing rules depending on the author of the method (Brill and Mukherjee,
1999). The pressure-drop component caused by friction losses requires evaluation of a two phase
friction factor. The pressure-drop caused by elevation change is dependent on the two phase mixture
The assumption of homogeneous fluid ignore the fact that gas generally flows faster than liquid
(i.e., no-slip conditions), which results in under-prediction of the pressure drop because the liquid
holdup predicted in the pipe is smaller than the actual liquid holdup (Brill and Mukherjee, 1999).
Some methods based on empirical correlations were designed to account for the slip between the
gas and liquid phases, by calculating the liquid holdup as function of the multiphase flow regime.
Brill and Mukherjee (1999) presented a classification of different empirical correlations based on
the homogeneous fluid model approach, to predict pressure drop of multiphase flow in pipes. The
first class was called Category “a”, the correlations in this class assume the no-slip condition (gas
and liquid have equal velocity) and does not consider flow patterns. The properties of the mixture
The second class was called Category “b”, the methods in this class consider slip between the
gas and liquid phases, but do not consider flow patterns. The methods in Category “b” offer
The third class was called Category “c”, the methods in this class consider gas-liquid slip and
flow patterns. A procedure to determine the flow pattern is required before the correlations for
liquid holdup and two-phase friction factor can be used. In these methods the acceleration pressure-
drop term is also dependent of flow pattern. Table 3.1 presents a list of the empirical correlations
54
Table 3.1: Classification of empirical correlations for pressure drop prediction in multiphase flow
based on the homogeneous fluid model (Brill and Mukherjee, 1999).
There is another approach for the homogeneous fluid model that, instead of using a friction factor
correlation uses a two-phase correction factor multiplier. This approach consists of estimating the
single-phase frictional pressure drop for the liquid phase only, and multiplying this by a two-phase
correction factor,
dp dp
= φ2LO (3.20)
dx f dx f,LO
where φ2LO is the two-phase correction factor, and the subscript LO denotes liquid only. The
superficial velocity of the liquid phase is used in the calculation of the single-phase frictional pressure
µL − µg −1/4
ρL ρL − ρg
φ2LO = 1 + Hg 1 + Hg (3.21)
ρf ρg µg
The separated-flow model approach is specific for stratified flow. In the separated-flow model
approach the gas and liquid phases are assumed to flow in separate conduits with cross sectional
area equal to the area occupied by the phases. Steady-state momentum balance equations are
dp (τw,k Pk + τi Pi )
=− − ρk g sin θ (3.22)
dx Ak
55
where τw,k is the wall shear stress of phase k, evaluated assuming single-phase flow in a conduit
with cross sectional area equal to the area occupied by the phase (Ak ), Pk is the wetted perimeter
of phase k, τi is the interfacial shear stress, Pi is the width of the flat interface between the phases.
An empirical correlation based on the separate flow model approach is the Lockhart and Mar-
tinelli (1949) correlation. The Lockhart-Martinelli correlation was developed from horizontal flow
of air-liquid mixtures at near-atmospheric pressure. It assumes the acceleration term of the pres-
sure gradient equation (eqn. 3.19) negligible, the pressure drop component by elevation changes
vanishes, leaving only the friction component of the pressure drop equation. From the calculation
of “only gas” and “only liquid” reference flows they derived equations for the frictional pressure
rameter (X),
The Lockhart-Martinelli parameter is defined as the ratio of the single-phase frictional pressure
(dp/dx)f,LO
X= (3.25)
(dp/dx)f,GO
The previous two-phase flow modeling approaches used separate empirical correlations for flow
regime, liquid holdup and friction factor, but it is important to note that these parameters are
physically inter-related. During the development of transient two-phase flow models an attempt
was made to unify the treatment of liquid holdup, pressure drop and flow regimes, as an integral
The transient two-phase modeling approach involves the simultaneous solution of conservation
equations of mass, momentum and energy for each phase (gas and liquid), and uses empirical
correlations and simplified constitutive relationships for flow parameters, such as interfacial friction,
fraction of liquid entrained in the free gas phase, and liquid holdup in the slug body. Bendiksen
56
et al. (1991) presented a transient two-phase (gas and liquid) flow model that considered three
separate mass conservation equations for the gas, liquid bulk and liquid droplets dispersed in the
gas phase, which are coupled by interfacial mass transfer; a combined momentum equation for the
gas plus liquid droplets and a separate momentum equation for the liquid film; and a simplified
mixture energy equation considering all the phases as a homogeneous fluid. The mass conservation
∂ 1 ∂
(Hg ρg ) = − (Ap Hg ρg vg ) + ψg + Sg (3.26)
∂t Ap ∂ x
For the liquid film the mass conservation equation is,
∂ 1 ∂ HL
(HL ρL ) = − (Ap HL ρL vL ) + ψg − ψe + ψd + SL (3.27)
∂t Ap ∂x HL + HD
The mass conservation equation for liquid droplets is,
∂ 1 ∂ HD
(HD ρL ) = − (Ap HD ρL vD ) + ψg + ψe − ψd + SD (3.28)
∂t Ap ∂ x HL + HD
where the phases are denoted by subscripts g, L and D for gas, liquid film and liquid droplets,
respectively; Sk is the external source/sink term of phase k; ψg is the mass transfer rate between
gas and liquid phases; ψe is the liquid droplet entrainment rate into the gas phase; and ψd is the
The linear momentum conservation equation for the gas phase is,
∂ ∂p 1 ∂ fg ρg vg2 Pg
(Hg ρg vg ) = −Hg − Ap Hg ρg vg2 −
∂t ∂x Ap ∂x 8 Ap
2
fi ρg vs Pi
− − Hg ρg g sin θ + ψg va − FD (3.29)
8 Ap
∂ ∂p 1 ∂ 2
(HD ρL vD ) = −HD − Ap HD ρL vD − HD ρL g sin θ
∂t ∂x Ap ∂x
HD
−ψg va + ψe vi − ψd vD + FD (3.30)
HL + HD
The combined momentum equation for the gas plus liquid droplets result from adding equations
57
∂ ∂p 1 ∂
Ap Hg ρg vg2 + Ap HD ρL vD
2
(Hg ρg vg + HD ρL vD ) = − (Hg + HD ) −
∂t ∂x Ap ∂x
2
fg ρg vg Pg 2
fi ρg vs Pi
− − − (Hg ρg + HD ρL ) g sin θ
8 Ap 8 Ap
HL
+ψg va + ψe vi − ψd vD (3.31)
HL + HD
Note that the gas/droplet drag terms (FD ) cancel in the combined gas/liquid droplet momentum
equation.
The linear momentum conservation equation for the liquid film is,
∂ ∂p 1 ∂ fL ρL vL2 PL fi ρg vs2 Pi
(HL ρL vL ) = −HL − Ap HL ρL vL2 − +
∂t ∂x Ap ∂x 8 Ap 8 Ap
HL
−HL ρL g sin θ − ψg va − ψe vi + ψd vD
HL + HD
∂HL
+HL d (ρL − ρg ) g cos θ (3.32)
∂x
Here vs is the slip velocity between the gas and liquid film; Pg , PL , and Pi are the wetted
perimeter of the gas, liquid and interface, respectively; fg , fL and fi are the friction factors of gas,
liquid and interface, respectively. The velocity, va , associated with the mass transfer rate term, ψg ,
where v0D is the fall velocity of droplets. The interface velocity, vi , is approximated by the liquid
velocity, vL .
A pressure equation is obtained by expanding the mass conservation equations with respect to
pressure, temperature and composition, dividing the expansions by each phase density and adding
the equations.
58
Hg ∂ρg 1 − Hg ∂ρL ∂p 1 ∂
+ =− (Ap Hg ρg vg )
ρg ∂p T ρL ∂p T ∂t Ap ρg ∂x
1 ∂ 1 ∂ 1 1
− (Ap HL ρL vL ) − (Ap HD ρL vD ) + ψg −
Ap ρL ∂x Ap ρL ∂x ρg ρL
Sg SL SD
+ + + (3.34)
ρg ρL ρL
This pressure equation along with the momentum conservation equations can be solved simul-
∂ 1 2 1 2
Hg ρg eg + vg + gh + HL ρL eL + vL + gh
∂t 2 2
1 2 ∂ 1 2
+HD ρL eD + vD + gh =− Hg ρg vg hg + vg + gh
2 ∂x 2
1 2 1 2
+HL ρL vL hL + vL + gh + HD ρL vD hD + vD + gh + hS + q (3.35)
2 2
where ek is internal energy per unit mass of phase k, hk is enthalpy of phase k, hS is enthalpy from
mass sources, q is heat transfer through the pipe wall, and h is elevation. The energy conservation
equation is used to determine the average fluid temperature along the length of a pipeline.
The mass transfer rate between gas and liquid phases is a function of the gas mass fraction
(Rs ),
∂Rs ∂p ∂Rs ∂p ∂x
ψg = +
∂p T ∂t ∂p T ∂x ∂t
#
∂Rs ∂T ∂Rs ∂T ∂x
+ + (Hg ρg + HL ρL + HD ρL ) (3.36)
∂T p ∂t ∂T p ∂x ∂t
Hg ρg
Rs = (3.37)
(Hg ρg + HL ρL + HD ρL )
In the mass transfer rate equation (3.36), the first term represents the mass transfer caused
by a pressure change in the same section, the second term represents the mass transfer caused by
mass flowing out of the section driven by a pressure gradient, the third term represents the mass
transfer caused by a temperature change in the same section, and the fourth term represents the
59
mass transfer caused by mass flowing out of the section driven by a temperature gradient.
For this model, the fluid properties (density, compressibility, viscosity, surface tension, enthalpy,
heat capacity, and thermal conductivity) are generated a priori using a thermodynamic fluid prop-
temperatures. For a given pressure and temperature condition the properties are interpolated from
the tables.
Bendiksen et al. (1991) considered four multiphase flow regimes during the development of this
model, which are stratified, annular, bubble and slug. The transition between the flow regimes
is determined by assuming a continuous average gas holdup and selecting the flow regime with
the minimum gas velocity (i.e., a minimum slip constraint). Since friction factors and wetted
perimeters are dependent on flow regime, they implemented a series of proprietary semi-empirical
relations to calculate flow parameters such as: average wave height in stratified wavy flow, wall and
interfacial friction factors, droplet deposition and entrainment rates, slip velocity between gas and
liquid phases, slug bubble velocity, and gas fraction in liquid slugs.
The transient two-phase model presented by Bendiksen et al. (1991) have evolved into a com-
ers more than two phases. This model accounts for three pre-defined phases, gas, oil and water,
which are distributed into different mass fields. Figure 3.4 presents a schematic representation of
the mass fields considered, which includes gas, oil and water as continuous phases, and dispersed
phases within continuous phases (i.e., oil and water droplets dispersed in continuous gas, water
droplets and gas bubbles dispersed in continuous oil, and oil droplets and gas bubbles dispersed in
continuous water). This allows for a closer representation of the production stream of hydrocarbon
pipelines.
Similarly to the two-phase model presented by Bendiksen et al. (1991), the OLGA
R
model
considers separate mass conservation equations for each mass field, a combined momentum con-
servation equation for each flowing layer (i.e., each continuous phase with its dispersed phases),
one pressure equation, and one mixture energy conservation equation. The model uses a sequential
splitting scheme to solve the conservation equations in subsequent steps for each time step, starting
with the simultaneous solution of the momentum and pressure equations to obtain pressure and
velocities, then solving the set of mass conservation equations to get phase masses, and finally the
60
Figure 3.4: Illustration of basic mass fields in a transient multiphase flow model, including gas, oil
and water as continuous phases, and dispersed phases within continuous phases (i.e., oil and water
droplets dispersed in continuous gas, water droplets and gas bubbles dispersed in continuous oil,
and oil droplets and gas bubbles dispersed in continuous water).
A more flexible flow model framework was developed in the HORIZON JIP program for OLGA
R
that allows one to include the effects of a user-defined dispersed phase on the fluid flow behavior.
The new flow model framework adds two additional layers to the model, in the form of a bedding
layer and a deposited wall layer. A user-defined plug-in program can be coupled with OLGA
R
to define a new phase and calculate additional contributions to the conservation equations. The
• The PVT module, used to calculate properties (such as density, viscosity, enthalpy, heat
capacity, and thermal conductivity) for the additional user-defined phases, and can modify
• The rheology module, used to calculate the apparent viscosity of a layer as a function of the
• The flash module, used to calculate the mass transfer rate between the phases and mass fields.
• The entrainment/deposition module, used to calculate the mass transfer rate of the same
phase between different mass fields, for example, mass transfer from a continuous phase to a
61
• The heat transfer module, used to calculate the heat transfer coefficient of a deposited wall
layer, and the corrected inner heat transfer coefficient; this only applies when the user-defined
phase can deposit on the pipe wall and the flash module accounts for this phenomena.
The user must set PVT data for user-defined phases, and can modify the properties of the pre-
defined phases (gas, oil and water) using the PVT module. It is possible to have balance equations
for statistical moments for each mass field, which are used for tracking particle size distribution.
The main purpose of the plug-in modules is to calculate contributions to the conservation
equations. A plug-in can contribute to the mass conservation equations through the inter-field mass
transfer terms, either by flashing (change of phase and mass-field) or entrainment/deposition (same
done through the calculation of modified viscosities for the flow layers, accounting for concentration
of particles. The modified viscosity will induce modifications in the friction factors. The particle
slip velocity with respect to the continuous phase may be calculated in the plug-in, which will
modify the resulting friction factor as well. The only contribution of a plug-in to the pressure
and also calculates the temperature of the pipe wall, assuming radial heat conduction through
concentric wall layers. An internal heat transfer coefficient is calculated for the heat transfer from
the fluid to the inner pipe wall. The presence of an additional wall layer due to deposition of the
user-defined phase could add resistance to the overall heat transfer, and this is the contribution of
This section illustrates how multiphase flow characteristics for simple systems (gas and water)
may change when a gas hydrate solid phase is introduced. It has been observed experimentally
that, for a system already near the stratified/slug transition point, the formation of a small quantity
of hydrate may immediately lead to slug flow onset (Joshi, 2012). The ability to provide a first-
principles estimation of flow regime transitions and slugging dynamics represents a critical step
A simple multiphase flow model, developed by Danielson (2011), is used as a starting point,
which is capable of predicting stable hydrodynamic slug flow based on fundamental multiphase flow
62
concepts. The hydrodynamic slug flow model is extended by including an additional mass balance
equation for the gas hydrate phase, allowing the introduction of the effects of gas hydrates as a
third phase. A transient hydrate kinetics model is used to calculate the gas hydrate growth rate,
and forms the basis for the mass transfer rate between the phases. The resulting model represents
an improvement on the commonly used steady-state assumption for both pressure drop and liquid
holdup in slugging systems, using instead the transient pressure drop and liquid holdup. Using fluid
properties and flow parameters, a mixture energy conservation equation is introduced to update the
temperature profile of the pipeline, accounting for the exothermic nature of the hydrate formation
reaction. The performance of the proposed model is compared against experimental data obtained
A new approach was presented by Danielson (2011) to model two-phase flow, which is based on
the multiphase flow concepts presented in Section 3.2. The definition of slip velocity, vs = vg − vL ,
The mass conservation equation for the liquid phase can be written as,
0
vSL = vSL + vg HL (3.42)
63
Assuming constant slug propagation is equal to the gas bubble velocity, vg , the volume conser-
convention shown in Figure 3.5, where node i is the solution node, the volume conservation equation
Figure 3.5: Pipe segment divided into equally spaced sections, showing node numbering convention.
Note that n indicates the time step level of the temporal discretization and i indicates the node
0 ,
index for the spatial discretization. It is assumed that the superficial liquid velocity term, vSL
moves from right to left in the moving reference frame, and the vg HL terms move from left to right.
Using a first order upwind scheme to approximate the derivative of the superficial liquid velocity
terms in the moving reference frame, and a third order upwind scheme to approximate the derivate
(n) (n−1)
HL,i − HL,i (n−1)
0 0
∆x = vSL,i − vSL,i+1
∆t
(−HL,i−2 + 6HL,i−1 − 3HL,i − 2HL,i+1 )(n−1)
+vg (3.45)
6
Equation 3.45 can be solve explicitly for the liquid holdup in the new time step n,
(n) ∆t n 0
(n−1) 0
(n−1)
HL,i = HL,i + vSL,i − vSL,i+1
∆x )
(−HL,i−2 + 6HL,i−1 − 3HL,i − 2HL,i+1 )(n−1)
+vg (3.46)
6
64
The superficial liquid velocity at the moving reference frame could be expressed as a function
0 0
= vs HL2 + vm
vSL − vs HL − vg HL (3.47)
0
vm = vm − vg (3.48)
From a drift-flux formulation, the mixture velocity in the moving reference frame can be ex-
0
vm = −vo (3.49)
Substituting,
0
vSL = vs HL2 − (vo + vs ) HL (3.50)
Note that the slip velocity is also a function of liquid holdup, and the following relationship
could be used,
where a, b and c are coefficients that can be determined from experimental data.
This hydrodynamic slug model was coded in Matlab, to reproduce the results reported by
Danielson (2011). Table 3.2 presents the input parameters used in the example. In this case,
a periodic boundary condition is used, where the fluids that exit at the right boundary cell is
reintroduced at the left boundary cell. The case is initialized with a liquid holdup of 0.4 and a
small perturbation (HL = 0.41) is introduced in the 20th node. Figure 3.6 shows the distribution of
liquid holdup along the pipeline length at four different times obtained using the hydrodynamic slug
model. The small perturbation in the interface between gas and fluid grows into a hydrodynamic
3.4.2 Extension of a simple hydrodynamic slug flow model for transient hydrate ki-
netics
slug flow model. This new model is an extension of the simple two-phase flow model presented in
the previous section (Section 3.4.1), to study the effect of hydrate formation on the flow regime in
65
Table 3.2: Input parameters for hydrodynamic slug flow model.
1 1
0.8 0.8
Liquid Holdup
Liquid Holdup
0.4 0.4
0.2 0.2
0 0
0 100 200 300 400 500 0 100 200 300 400 500
Pipeline length (m) Pipeline length (m)
(a) Time = 0 s. (b) Time = 100 s.
1 1
0.8 0.8
Liquid Holdup
Liquid Holdup
0.6 0.6
0.4 0.4
0.2 0.2
0 0
0 100 200 300 400 500 0 100 200 300 400 500
Pipeline length (m) Pipeline length (m)
(c) Time = 500 s. (d) Time = 700 s.
Figure 3.6: Liquid holdup distribution along the pipeline length at four different times: (a) t = 0.04
s, (b) t = 100 s, (c) t = 500 s, (d) t = 700 s.
66
flowlines. The extended model includes a transient hydrate kinetics model (Turner et al., 2005),
allowing the prediction of hydrate formation, and is the base for mass transport between the phases.
For hydrate formation it is necessary to estimate the pressure and temperature conditions
along the flowline length. In this work, a simple pressure drop model based on the Lockhart and
Martinelli’s (1949) correlation is used to predict the pressure distribution along the pipeline and
In order to introduce gas hydrate as a third phase in the flow model, the multiphase flow
fundamental concepts presented in Section 3.2 are modified based on the conceptual picture shown
in Figure 3.7.
Figure 3.7: Conceptual picture used to define the phases and multiphase flow concepts.
First, the holdup for the gas hydrate phase (Hhyd ) is defined as the fraction of pipe cross
sectional area (Ap ) or volume increment that is occupied by the gas hydrate phase (Ahyd ),
Ahyd
Hhyd = (3.52)
Ap
The sum of the phases holdup equals one, Hg + Hw + Hhyd = 1. This phase summation allows
specifying the gas holdup in terms of the other two phases (Hg = 1 − Hw − Hhyd ).
The mixture velocity (vm ) is defined as the sum of the superficial velocities,
qg + qw + qhyd
vm = = vSg + vSw + vShyd (3.53)
Ap
The slip velocity is defined as the difference between the phase velocities, and we can define a
and a slip velocity between liquid water and hydrate (Us,(w−hyd) ) as,
67
vSw vShyd
vs,(w−hyd) = vw − vhyd = − (3.55)
Hw Hhyd
Equations 3.54 and 3.55 for the gas-water and water-hydrate slip velocities respectively, can be
The mass conservation equation for the water phase can be written as,
nH dmgas
ψw = (3.59)
∆x dt
where nH is the hydration number (kg of water/kg of gas), dmgas /dt is the gas mass consumption
Assuming the water density, pipe cross sectional area, and section length are each separate
constants, the mass conservation equation simplifies to the volume conservation equation,
∂Hw ∂vSw 0
= + ψw (3.60)
∂t ∂x
where,
0 nH dmgas
ψw = (3.61)
ρw Ap ∆x dt
The water volume conservation equation (3.60) can be expanded using finite differences following
a similar procedure presented in the previous section (Section 3.4.1), yielding an explicit expression
(n) (n−1) ∆t n 0 0
(n−1)
Hw,i = Hw,i + vSw,i − vSw,i+1
∆x )
(−Hw,i−2 + 6Hw,i−1 − 3Hw,i − 2Hw,i+1 )(n−1) 0
+vg + ∆tψw (3.62)
6
68
Following a procedure similar to the one used for the water phase, the mass conservation
equation for the gas hydrate phase can be expanded using finite differences resulting in the following
expression,
(n) (n−1) ∆t n 0 0
(n−1)
Hhyd,i = Hhyd,i + vShyd,i − vShyd,i+1
∆x )
(−Hhyd,i−2 + 6Hhyd,i−1 − 3Hhyd,i − 2Hhyd,i+1 )(n−1) 0
+vg + ∆tψhyd (3.63)
6
0
where ψhyd is the gas hydrate source term, which is related to the gas hydrate formation rate and
is defined as,
0 (nH + 1) dmgas
ψhyd = (3.64)
ρhyd Ap ∆x dt
The superficial velocities in equations 3.62 and 3.63 are a function of the phases holdup and
can be obtained from the simultaneous solution of equations 3.56 and 3.57.
Note that the gas-water slip velocity is considered to be function of liquid holdup, and the
where a, b, c are coefficients that can be determined from experimental data. No slip between
The hydrate formation rate is calculated using the following intrinsic kinetics equation (Turner
et al., 2005),
dmgas k2
= −k1 exp As ∆Tsub (3.66)
dt Tsys
The intrinsic kinetics equation gives the rate of gas consumption during hydrate formation
(dmgas /dt) as a function of the intrinsic rate constants (k1 and k2 ) regressed from Vysniauskas
and Bishnoi (1983) data; the surface area (As ) between water and gas phases; and the subcooling
(∆Tsub ) as the thermal driving force. The subcooling is defined as the difference of the hydrate equi-
librium temperature and system temperature. Hydrate nucleation is assumed to occur immediately
69
The surface area between water and gas is calculated assuming stratified smooth flow in each
pipe section (Figure 3.8). The cross sectional area of the water layer (As ) as a function of pipe
r2
As = (2α − sin 2α) (3.67)
2
Figure 3.8: Cross section of pipe with stratified smooth flow regime.
The relationship between water holdup (Hw ) and wetted angle in stratified flow is obtained by
2α − sin 2α
Hw = (3.68)
2π
This is a non-linear function of the wetted angle. Given the liquid hold up, this equation can be
solved for the wetted angle using a numerical method. Alternatively, an explicit approximation for
the wetted angle as function of liquid hold up can be used to avoid iterative calculations (Biberg,
1999),
π
α≈ 1 − (1 − Hw )1/3 + Hw1/3 (3.69)
2
Then, the surface area between water and gas in a pipe section is,
As = 2r sin(α)∆x (3.70)
The mixture energy conservation equation used in this work is based on a homogeneous fluid
model, where the gas, liquid and hydrate phases are treated as a pseudo-phase, and the properties
are the volume average of the independent phases (Cazarez-Candia and Vasquez-Cruz, 2005).
70
The energy balance equation includes terms for the transient behavior of the system enthalpy,
the convection of energy, the transient change in system pressure, fluid viscous dissipation, change
in potential energy, pipe wall heat transfer to the environment and a source term for the exothermic
nature of the hydrate formation reaction. The energy balance equation in terms of the pseudo-phase
enthalpy is,
∂h ∂h 1 ∂p ∂v ∂v q̇πd
+v = −v − vv − vg sin θ − + SE,hyd (3.71)
∂t ∂x ρ ∂t ∂t ∂x ρAp
where h is mixture enthalpy per unit mass, v is volume average velocity, ρ is volume average density,
q̇ is the wall heat flux, d is pipe diameter, and SE,hyd is the energy source term and is related with
nH + 1 dmgas
SE,hyd = ∆hhyd (3.72)
ρhyd Ap ∆x dt
where, ∆hhyd is the enthalpy change of hydrate formation (for methane hydrate, 4175.4×103 J/kg).
When hydrates are forming, the heat of formation (∆hhyd ) acts as an energy source inside a pipe
section.
Following the derivation presented in Appendix A, the energy balance equation (eqn. 3.71) can
∂T ∂T ∂p ∂p 1 ∂p ∂v ∂v q̇πd
Cp +v −β +v − +v +v = −vg sin θ − + SE,hyd (3.73)
∂t ∂x ∂t ∂x ρ ∂t ∂t ∂x ρAp
Equation 3.73 can also be discretized using finite differences obtaining an explicit solution for
the temperature in the solution node i at the new time step n (see Appendix A).
The performance of the three-phase (gas/liquid water/hydrate) hydrodynamic slug flow model
presented in the previous section (Section 3.4.2) is tested against experimental data obtained from
an industrial scale flow loop (Joshi et al., 2011a). In this case, a periodic boundary condition is
used, where the fluids exiting the right boundary cell are reintroduced at the left boundary cell,
with the objective of modeling the flow behavior of the loop. The virtual connection between the
71
first and last cell represents, in terms of pressure, the positive displacement pump used in the flow
loop.
The flow loop is located inside a temperature controlled room, where the temperature can be
varied between 264 and 323 K. The average pressure in the flow loop is maintained constant at a
desired set point by injecting gas from a gas-filled piston accumulator connected to the flow loop.
Table 3.3 presents the input parameters used in the case study. The case is initialized with a water
holdup of 0.5. The temperature of the room housing the flow loop measured during the experiment
is an input to the model, and it is used to calculate the heat transfer between the fluids and the
Table 3.3: Input parameters for three-phase (gas/liquid water/hydrate) hydrodynamic slug flow
model.
We assume that there is no slip between water and hydrate, i.e., the slip velocity is equal to zero.
The hydrate equilibrium conditions (pressure and temperature), assuming pure methane as the gas
phase (Figure 3.9), were calculated using CSMGem (Ballard and Sloan, 2004a). The nucleation
of hydrates was observed at a subcooling of 1.1 K in the flow loop experiment. In the simulation,
Figure 3.10 shows the distribution of water and hydrate holdup along the flowline length at four
different times obtained using the three-phase hydrodynamic slug model. At the beginning of the
simulation, the system was forced to remain in a stratified flow regime and no hydrates (Hhyd = 0)
are present in the system, as shown in Figure 3.10(a). Upon hydrate nucleation, the solid hydrate
72
12
10
Pressure (MPa)
L−H
6 L−H−V
4 L−V
2
270 275 280 285 290
Temperature (K)
Figure 3.9: Hydrate equilibrium curve for pure methane (V) and liquid water (L) calculated with
CSMGem.
particles act as a perturbation that promotes a flow regime transition from stratified to slug flow;
the beginning of this transition is shown Figure 3.10(b). The amount of hydrates continues to
The temperature predicted by the model is compared against measured data from the flow
loop experiment (Figure 3.11). Note that the input to the model was an approximation to the
ambient temperature measured during the experiment (curve labeled as “Ambient”), and the fluid
temperature was matched by adjusting the overall heat transfer coefficient. With the three-phase
hydrodynamic slug flow model, it was possible to match the cooling rate of fluids inside the loop
and to capture the exothermic nature of the hydrate formation reaction shown as an increase in
3.5 Summary
This chapter presented a review of the fundamental concepts of steady-state and transient
multiphase flow modeling. In order to understand the origin of multiphase flow modeling techniques
it is necessary to review the single-phase flow pressure-gradient equation, which resulted from the
analytical solution of the mass and momentum conservation equations. There is one empirical
component in the pressure gradient equation for single-phase flow, which is the dimensionless
73
1 1
0.8 0.8
Hold−up
Hold−up
0.6 0.6
0.4 0.4
0.2 0.2
0 0
0 20 40 60 80 0 20 40 60 80
Pipeline length (m) Pipeline length (m)
(a) Time = 40 s. (b) Time = 3230 s.
1 1
0.8 0.8
Hold−up
Hold−up
0.6 0.6
0.4 0.4
0.2 0.2
0 0
0 20 40 60 80 0 20 40 60 80
Pipeline length (m) Pipeline length (m)
(c) Time = 4000 s. (d) Time = 6000 s.
Figure 3.10: Distribution of water (continuous blue line) and hydrate (segmented green line) holdup
along the pipeline length at four different times, (a) 40 s, (b) 3230 s, (c) 4000 s, (d) 6000 s, showing
the transition from stratified to slug flow upon hydrate formation.
74
Fluid
300 Ambient
Experiment fluid
Temperature (K)
Experiment ambient
290
280 Hydrate
formation onset
270
0 0.5 1 1.5
Time (hr)
Figure 3.11: Comparison of predicted results and experimental data of fluid and ambient temper-
ature as function of time.
The ultimate objective in flow modeling is to estimate the pressure drop in the pipeline, an
important element for the design of pipeline systems to guarantee the effective transport of fluids
In general, the multiphase flow modeling methods are extensions to the well known single-phase
flow models. The complexity of the multiphase flow requires empirical correlations to estimate the
effect of unknown mechanisms affecting the pressure gradient. The homogeneous fluid modeling
method considers the multiphase mixture as a homogeneous phase and uses modeling approaches
similar to that of single phase flow, but this method typically underestimate the pressure drop.
Other methods consider separate flow and try to estimate the flow regimes based on predefined
flow regime maps. Most of the flow regime maps were developed for air/water flow at near atmo-
spheric pressure conditions, which could not be representative of gas/oil systems encountered in
offshore production. The transient multiphase modeling method involves the simultaneous solution
of conservation equations of mass, momentum and energy for each phase, and uses empirical cor-
relations and simplified constitutive relationships for flow parameters, such as interfacial friction,
entrainment of dispersed phases in continuous phases, and liquid holdup in the slug body. The
transient multiphase flow modeling method form the basis of the commercially available OLGA
R
transient multiphase flow simulator, which is considered the standard for multiphase flow model-
75
ing in the oil and gas industry. It was shown how the flexible flow model framework of OLGA
R
allows to define new phases and introduce additional contributions to the conservation equations,
A new simple modeling approach based on fundamental concepts of multiphase flow was re-
viewed in the final section of this chapter, and is known as the hydrodynamic slug flow model. This
hydrodynamic slug flow model predicts the formation of liquid slugs. The hydrodynamic slug flow
model was extended including an additional mass balanced equation for the gas hydrate phase,
and an energy conservation equation. The hydrate formation rate was calculated using a transient
hydrate kinetics model. The mass transfer between the phases (gas, water and hydrate) is a func-
tion of the hydrate formation rate, and was included as a source/sink term in the mass balance
equations.
The extended hydrodynamic slug flow model was applied to simulate a flow loop experiment,
where the temperature of the fluids inside the flow loop is decreased until hydrates start to form.
The model captured the flow regime transition, from stratified to slug flow, observed in flow loop
experiments upon hydrate formation. The temperature behavior of the system was also captured
with the extended hydrodynamic slug flow model, matching the cooling rate of fluids inside the
loop and the exothermic nature of the hydrate formation reaction that increases the temperature
76
CHAPTER 4
Two large scale flow loop facilities were available during the development of this study, the
ExxonMobil Upstream Research Company flow loop located at Friendswood, Texas, and the Uni-
versity of Tulsa hydrate flow loop located at Tulsa, Oklahoma (the loop was originally designed
and constructed at the Marathon Oil Technology Center in Littleton, Colorado). This chapter
presents a description of the two large scale flow loops, the experimental procedure and materials
used, and the experimental matrix that represents the available data. These data are critical to
Figure 4.1 shows a schematic of the ExxonMobil flow loop used to study hydrate formation in
multiphase flow systems. The ExxonMobil flow loop has an internal diameter of 3.826 inches (4
in. nominal size) with an axial length of 314 ft (95 m). The fluids are displaced by a multiphase
sliding-vane pump, specially designed to minimize shear on hydrate particles, which can operate
at rotation speeds between 300 to 1500 rpm (equivalent to 0.8 to 3.8 m/s fluid mixture velocities).
The flow loop is located inside a temperature controlled room, which can operate in the range
from 90 to 25 o F (273 to 269 K). The flow loop can be operated at pressures up to 2000 psi (12.41
MPa). A piston-driven gas accumulator with maximum volume of 120 gallons is connected to the
flow loop near the suction side of the pump, and allows gas injection into the loop to maintain
constant pressure across the loop during hydrate formation. The position of the piston inside the
gas accumulator is controlled by a hydraulic power unit (HPU). The flow loop incorporates a port
where a particle size analyzer probe (FBRM or PVM) can be installed for visualization of entrained
gas bubbles, droplets and hydrate particles in the continuous phase. A mass flow and density meter
The parameters that can be measured in this flow loop are average pressure of the loop, pressure
drop across the loop, temperature inside the loop, and gas injection into the loop. The loop
pressure is measured with an accuracy of 0.06 psig using a pressure transducer located on the line
77
Figure 4.1: Schematic of ExxonMobil flow loop.
connecting the gas accumulator to the loop. The pressure drop across the loop is measured using
a differential pressure transducer with an accuracy of 0.12 psi, which is located across the pump.
The temperature is measured using two thermocouples with an accuracy of 0.02 o F, located at the
intake and discharge of the pump toward the center of the pipe. The gas injected into the loop is
indirectly measured by the position of the piston, which can be related to the accumulator volume.
The flow loop incorporates two viewports, each equipped with a digital camera and a light source
on the opposite side for video recording. The viewports are cubic expansion boxes with vertical
The experimental procedure followed for the tests performed at the ExxonMobil flow loop
consisted of two stages. The first stage consists of charging fluids into the loop by adding the
desired amount of oil and water, and pressurizing with methane gas to the desired operation
pressure. The fluids are circulated in the loop using a pump speed of 600 rpm for at least two
hours, to mix and saturate the phases at the ambient temperature. The second stage consists of
setting the experimental conditions, by setting the constant operation pressure in the HPU, the
pump speed, and the cold room temperature. When a test is complete, the temperature is increased
above 70 o F (294 K) and held at this temperature for the next test.
Two sets of experiments were performed in the ExxonMobil flow loop for this study. The first
set of experiments was performed in the Fall of 2010, and consisted of pure water (100% water
cut) and methane gas experiments to study the hydrate formation mechanisms in water-dominated
78
systems. Table 4.1 presents the parameters of the experiments performed in the Fall of 2010. The
liquid loading values used in this set of experiments were 50, 75 and 90 % of the flow loop volume;
the liquid loading is the fraction of flow loop volume occupied by liquid phases. The mixture
velocities used in the experiments were 1.0, 1.75 and 2.0 m/s; the mixture velocity of fluids inside
the flow loop is set by the pump speed. In all the experiments the operation pressure was set
to 1000 psi (6.89 MPa) and maintained constant by injecting gas from the gas accumulator. The
temperature of the flow loop room is decreased from an initial ambient temperature to a set point
of 34 o F (274 K).
Table 4.1: Experiments performed in the ExxonMobil flow loop in Fall 2010.
The second set of experiments was performed in the Fall of 2011, and consisted of a variety
of water- and oil-dominated systems. For the oil-dominated systems Conroe oil or King Ranch
Condensate were used as the oil phase. Table 4.2 presents the parameters of the experiments
performed in the Fall of 2011. The liquid loading values used in this set of experiments were 50,
75, 80, 90 and 95 % of the flow loop volume. The mixture velocities used in the experiments were
1.0, 1.75, 2.0, 2.5 and 3.0 m/s. In all the experiments the operation pressure was set to 1000 psi
(6.89 MPa) and maintained constant by injecting gas from the gas accumulator. The temperature
of the flow loop room is decreased from an initial ambient temperature to a set point of 34 o F (274
K).
Additional data from oil-dominated experiments (Table 4.3) performed in the ExxonMobil flow
79
Table 4.2: Experiments performed in the ExxonMobil flow loop in Fall 2011.
Table 4.3: Additional experimental data from the ExxonMobil flow loop (Turner, 2005).
80
4.2 Tulsa University flow loop
The hydrate flow loop available at the University of Tulsa in Tulsa, Oklahoma, was originally
designed and constructed at the Marathon Oil Technology Center in Littleton, Colorado. Figure 4.2
shows a schematic of this hydrate flow loop used to study hydrate formation in multiphase flow
systems. The University of Tulsa hydrate flow loop has an internal diameter of 2.9 inches (3 in.
nominal size) with an axial length of 164 ft (50 m). The fluids are displaced by a Leistritz twin-
screw multiphase pump, which can operate at rotation speeds up to 2000 rpm (equivalent to 3.7
m/s fluid mixture velocities). The flow loop has a cooling jacket over most of its length. Glycol
is used as the coolant fluid, and the glycol temperature can be controlled between the range from
32 o F to 80 o F (273 to 300 K) by measuring the glycol inlet and outlet temperatures. The flow
loop can be operated at pressures up to 2000 psi (12.41 MPa). A high-pressure piston pump is
connected to the loop at the discharge side of the pump, and allows gas injection into the loop to
maintain constant pressure across the loop during hydrate formation. A gamma densitometer is
The parameters that can be measured in this flow loop are pump suction and discharge pressures,
pressure drop, temperature, fluid density and gas injection into the loop. The pump suction and
discharge pressures are measured with an accuracy of 0.06 psig using pressure transducers. The
pressure drop is measured across each section of the loop and across the pump, using differential
pressure transducers with an accuracy of 0.025 psi. Several thermocouples are installed on the
outside pipe wall and measure the temperature of the pipe surface with an accuracy of 0.01 o F.
The gas injected into the loop is indirectly measured by the volume of oil pumped into the other
side of the piston of the high-pressure piston pump. The mass of gas injected is calculated using
equations of state and the gas composition. The equations of state available to calculate the mass
of gas injected are Peng-Robinson, Redlich-Kwong and Benedict-Webb-Rubin. The flow loop has
four viewports at the end of each section, each equipped with a digital camera for video recording.
Each view port has three sapphire windows at a 120o angle of each other around the pipe.
The experimental procedure used for the tests performed at the University of Tulsa flow loop
is similar to the one used for the ExxonMobil flow loop. In the first stage, fluids are charged into
the flow loop, adding oil, water and pressurizing with gas at ambient temperature. In the second
81
Figure 4.2: Schematic of University of Tulsa hydrate flow loop.
stage the experimental conditions (pressure, pump speed and final temperature) are set.
Table 4.4 presents the parameters of experiments performed in the University of Tulsa flow loop.
The experimental data from these experiments were obtained from previous studies (Boxall, 2009;
Joshi, 2012). The first five experiments consisted in oil-dominated systems using one model oil
(Citgo) and three different crude oils (Buttermilk, Caratinga and Troika). The liquid loading used
in the oil-dominated experiments was 50%. The mixture velocity used was 1.2 m/s. The first three
experiments in Table 4.4 were performed at a constant pressure of 2000 psi (13.79 MPa), while the
experiments using Troika oil were at constant volume with an initial pressure of 1500 psi (10.34
MPa). In the constant volume experiments no gas is injected into the loop, and the system pressure
is allowed to decrease during cooling (due to contraction of gas phase) and hydrate formation (due
to consumption of gas phase). The temperature of the system is decreased to a final temperature
4.3 Summary
This chapter presented a description of the two large scale flow loop facilities used in this
study; the ExxonMobil Upstream Research Company flow loop located at Houston, Texas, and
the University of Tulsa hydrate flow loop located at Tulsa, Oklahoma. The experiments performed
at the flow loop facilities allow one to study the effects of hydrate formation in oil and water-
dominated multiphase flow systems. The multiphase flow observed in the flow loops is a physical
82
Table 4.4: Available experimental data from the University of Tulsa hydrate flow loop (Joshi,
2012).
simulation of the expected multiphase flow in the pipelines of subsea transport facilities. The flow
loops provide a controlled environment, where the pressure, temperature and mixture fluid velocity
can be specified. The system conditions can be placed inside the hydrate formation region by
decreasing the temperature. The amount of hydrates formed can be estimated by measuring the
gas consumption (or gas injected into the flow loop to maintain constant pressure). Measurements
of pressure drop across the flow loop can be used to estimate the transportability of the fluids inside
the pipeline as function of the hydrate volume fraction, until a hydrate plug is formed. Next chapter
(Chapter 5) presents conceptual and mathematical models developed from the data obtained from
the experiments performed in these flow loops. These data is also used for the validation of the
mathematical models.
83
CHAPTER 5
Three different systems of oil and gas production facilities, such as, oil-dominated systems (oil
as hydrate carrier phase), water-dominated systems (water as hydrate carrier phase), and gas-
dominated systems (hydrate film growth on pipe walls), have been considered for the development
of conceptual and mathematical models of gas hydrate formation in pipelines. A set of conceptual
pictures was developed to explain the physical phenomena of gas hydrate formation in these sys-
tems. Different hydrate formation mechanisms were identified for the different systems encountered
in oil/gas flowlines. For oil-dominated systems, dispersed water droplets are converted into hydrate
particles, and the agglomeration of these hydrate particles increase the slurry viscosity leading to
ExxonMobil flow loop, showed that pressure drop in the pipe depends on a flow regime transi-
tion involving multiple phases and solid dispersion in the liquid phase. For gas-dominated systems,
hydrates deposit and grow on pipe walls, narrowing the flow diameter and increasing surface rough-
ness.
This chapter presents a description of the three sub-models that constitute the comprehensive
hydrate formation model, including the mathematical models’ contributions to the multiphase flow
transient conservation equations presented in Section 3.3.3. This work presents improvements to
the hydrate particle aggregation module used for oil-dominated systems, based on experimental
data, which account for temperature, particle-particle contact time, excess water, and the presence
of surface active compounds. The model for water-dominated systems consists of a mass transfer-
based hydrate growth model and hydrate plugging criterion, based on fluid velocity and liquid
holdup. The model for gas-dominated systems consists of a combined heat and mass transfer
A previous version of the model for gas hydrate formation in oil-dominated systems was pre-
sented in Section 1.2 of the introductory chapter. The gas hydrate formation mechanism in oil-
84
dominated systems was described using the conceptual model shown in Figure 5.1 (reproduced
from Figure 1.8), where water is assumed to be entrained in the continuous oil phase as water
droplets, and the surface area between water and hydrocarbon phases is calculated using the Hinze
(1955) correlation. An alternative approach to calculate the surface area of water droplets is based
on an empirical correlation developed by Boxall (2009) for mean droplet size as function of fluid
properties and flow regime. Hydrate nucleation is assumed to occur immediately after a specified
subcooling (∆Tsub = Thyd eq − Tsystem ), and the hydrate growth is calculated using an intrinsic first
order kinetics equation with an adjustable rate constant (Boxall et al., 2009), or using the transport
model developed by Davies (2009) based on mass and heat transfer resistances. Once the hydrate
particle has formed, the particle agglomerate size is calculated using a steady-state balance between
the inter-particle cohesion force and fluid flow shear forces (Camargo and Palermo, 2002), assuming
a constant cohesion force. The viscosity of the slurry is calculated using a modified Mills equation
(Camargo and Palermo, 2002; Mills, 1985), using the effective volume fraction of the agglomerated
particles. A hydrate plug is identified by a large viscosity increase, which causes an unacceptably
Figure 5.1: Conceptual model for hydrate formation in oil-dominated multiphase flow systems
consisting of gas, oil, and water; adapted from Turner (2005), with input from J. Abrahamson (U.
Canterbury, Christchurch, NZ).
Improvements have been made to the agglomeration section of the conceptual model by remov-
ing the assumption of a constant particle cohesion force and introducing a dynamic cohesion force
model. The dynamic cohesion force model is based on particle capillary attraction due to a liquid
85
5.1.1 Description of dynamic particle cohesion force model
showed that a liquid bridge between hydrate particles can be responsible for hydrate particle ag-
glomeration in oil-dominated systems. The liquid bridge forms from a liquid layer present on the
hydrate particle surface when the particle is surrounded by a secondary immiscible phase (i.e.,
the oil phase). Similar liquid layers have been observed on the surface of ice (Bienfait, 1992;
Döppenschmidt and Butt, 2000) and are considered thermodynamically stable, since the presence
Based on the schematic representation of the capillary liquid bridge between two hydrate par-
ticles, shown in Figure 5.2, Aman et al. (2011) derived an equation for the particle cohesion force
where R∗ is the harmonic mean radius of particle radii R1 and R2 , σ is the water-oil interfacial
tension, θp is the wetting angle of water on the particle surface, H is the total height of the liquid
bridge, d is the immersion depth (i.e., how far the bridge extends into the particle), α is the
embracing angle, and fslv is a generic three phase free energy term. The first term represents
the capillary pressure across the liquid bridge, the second term represents the contribution of the
surface tension of the liquid bridge interfacial area, and the third term represents the specific free
energy of the three phase contact line. The last term is typically small, in comparison to the first
For the implementation of the model, it is assumed that hydrate particles are spherical and that
all unconverted water is evenly distributed on the surface of the hydrate particles as a uniform water
hydrate formation models mentioned above), less water is available to participate in the capillary
bridge between hydrate particles, resulting in a decrease of cohesive force. The liquid bridge height
(H) is defined as a thermodynamic stable liquid layer plus the excess water. The thermodynamic
stable liquid layer is assumed to be equal to that measured for ice, approximately 50 nm thick
(Döppenschmidt and Butt, 2000). The wetting angle of water on the hydrate particle surface (θp )
is assumed to be equal to 29o (a value measured by Asserson et al., 2009 for Freon hydrates in
86
Figure 5.2: Illustration of the capillary liquid bridge between two hydrate particles (Aman et al.,
2011).
brine). The embracing angle (α) is assumed to be equal to 0.1o . This value of embracing angle was
calculated by Aman et al. (2011) considering the baseline cohesive force of 4.3 mN/m measured for
cyclopentane hydrate, with a water/cyclopentane interfacial tension of 51 mN/m and the assumed
values for liquid layer thickness and wetting angle of water on hydrate surface. Rabinovich et al.
(2005) presented relations to estimate the volume (V ) and immersion depth (d) of the capillary
r !
H 2V
d= −1 + 1+ (5.3)
2 πRH 2
From cyclopentane hydrate cohesion experiments using a micromechanical force apparatus,
Aman et al. (2011) observed that inter-particle cohesion forces increased with contact time, at-
tributing this behavior to the liquid bridge converting into hydrate, a phenomenon called sintering.
They observed that the cohesion force started to increase when the contact time was greater than
30 seconds. The required force to fracture the solid hydrate bridge is defined by the tensile strength
FS = τt Ai (5.4)
87
where FS is the inter-particle force attributed to sintering, τt is the hydrate tensile strength, and
Ai is the cross sectional area of the solid bridge at the weakest point (the middle).
Due to the increase in measured cohesion force with contact time, Aman et al. (2011) deter-
mined that the hydrate bridge area (Ai ) must be increasing over the course of contact time, while
the hydrate tensile strength is a constant mechanical property. The cyclopentane hydrate tensile
strength (τt ) was estimated to be 0.91 MPa, for the baseline cohesive force of 4.3 mN/m. Using
the estimated hydrate tensile strength, they calculated the hydrate bridge area corresponding to
experimental cohesive force measurements for different contact times, in terms of the radius of con-
tact (χ). Then, the time-dependence of the hydrate bridge area was approximated by the following
χ = 1.2147t0.1249 (5.5)
The dynamic hydrate particle cohesion force model results from combining the capillary cohesion
force equation (5.1) and the sintering cohesion force equation (5.4), considering that the capillary
force dominates below about 30 s, and the sintering force dominates above 30 s,
2πσ cos θ
p
H
+ 2πσ sin (α) sin (θp + α) , tcontact < 30 s
FIP = 1 + 2d (5.6)
2
τt π 1.2147t0.1249 , tcontact ≥ 30 s
where FIP is the inter-particle cohesion force, and tcontact is the contact time between hydrate
particles. Figure 5.3 shows the results of this model compared against experimental data for
cohesion forces as function of contact time of cyclopentane hydrate particles. The model offers a
The dynamic cohesion force model, represented by equation (5.6), is based on five key quanti-
ties: (i) water-oil interfacial tension; (ii) hydrate surface wettability; (iii) hydrate particle radius;
(iv) volume of liquid water available, arising from either a thermodynamic quasi-liquid layer or
unconverted water; and (v) contact time between the hydrate particles. The water-oil interfacial
tension is a characteristic property of the system, and allows one to account for the presence of
surface active compounds (or surfactants). Surfactants that adhere to the liquid bridge interface
may reduce the interfacial tension, hence reducing the cohesion force between hydrate particles
(Aman et al., 2010). Surfactants can also affect the hydrate surface wettability, as was observed
by Aman et al. (2012) during micromechanical force measurements between cyclopentane hydrate
88
Figure 5.3: Comparison of experimental measurements (points) and model prediction (solid line)
of cyclopentane hydrate cohesive forces as function of contact time (Aman et al., 2011).
particles in the presence of different carboxylic acids. However, there are no means available to
predict the effect of surfactants on hydrate surface wettability, so this parameter is assumed to be
constant in the model. The volume of liquid water available to form a liquid bridge has a direct
effect on the magnitude of the cohesion force. As the liquid water is converted into hydrate, the
liquid bridge volume decreases, resulting in a decrease of cohesive force. The sintering of the liquid
bridge starts when the particles have been in contact for a period greater than 30 seconds.
The model for hydrate formation in oil-dominated systems was implemented in this thesis work
in the OLGA
R
transient multiphase flow simulator, through the development of a user-defined
hydrate phase that will be dispersed into the continuous oil phase. Figure 5.4 shows a schematic
representation of the mass fields considered for oil-dominated systems, which includes gas and oil as
continuous phases, water and hydrate as dispersed phases in the oil continuous phase, and oil and
water may be dispersed as droplets in the continuous gas phase. The user-defined plug-in program
calculates additional contributions to the conservation equations using the PVT, flash and rheology
modules.
89
Figure 5.4: Illustration of mass fields considered in the oil-dominated hydrate formation model
coupled with the OLGA
R
transient multiphase flow simulator, including gas and oil as continuous
phases, water and hydrate is assumed to be dispersed in the oil phase, and oil and water may be
dispersed as droplets in the continuous gas phase.
The PVT module calculates the properties for the hydrate phase, making additional contri-
butions to the mass, momentum and energy conservation equations. Typical values of hydrate
structure II (Sloan and Koh, 2008a) are assigned for density (914 kg/m3 ), heat capacity (498.83
J/(kg K)), thermal conductivity (0.7 W/(m K)). The enthalpy of the hydrate phase (hhyd ) is a
function of the enthalpy of the gas (hg ) and water (hw ) phases at the system temperature and
pressure, and the heat of reaction (∆hhyd ), to account for the exothermic nature of the hydrate
formation reaction,
hg + nH hw + ∆hhyd
hhyd = (5.7)
1 + nH
where nH is the hydration number (water to gas ratio in the hydrate phase).
The flash module calculates the gas consumption rate during hydrate formation using the in-
trinsic first order kinetics equation with an adjustable rate constant (Boxall et al., 2009) or using
the transport model developed by Davies (2009) based on mass and heat transfer resistances. The
flash module makes direct contributions to the mass conservation equations by calculating the mass
transfer rates from gas and water phases to the hydrate phase during hydrate formation, and during
hydrate dissociation where gas and water are released from the hydrate phase. The mass transfer
rates are used to update the phase masses of gas, water and hydrate.
90
The rheology module calculates the apparent viscosity of the oil layer as a function of the hydrate
particles agglomerate diameter, making contributions to the momentum conservation equation. The
hydrate particles agglomerate diameter is calculated using a force balance between the hydrate
particles cohesion force and the flow shear forces, based on the Camargo and Palermo (2002)
model. The particles cohesion force is calculated using the dynamic cohesion force model described
in Section 5.6. The flow shear forces are calculated in the OLGA
R
model. The effective viscosity
of the oil phase, considering the hydrate particle dispersion, is calculated using a modified Mills
equation (Camargo and Palermo, 2002; Mills, 1985), based on the effective volume fraction of the
agglomerated particles. The modified viscosity will induce modifications in the friction factors.
This section presents the comparison of prediction results obtained with the model for hydrate
Figure 5.5 to Figure 5.7 show results corresponding to three different experiments, one performed
at the ExxonMobil flow loop and two at the University of Tulsa flow loop, using three different
oils. These three cases show good agreement between model prediction and the experimental data.
The intrinsic kinetics equation was used to estimate the hydrate formation rate, using a correction
factor of 0.002. Figure 5.5 shows the hydrate volume fraction as function of time of an ExxonMobil
flow loop constant pressure experiment using Conroe oil, 70% liquid loading, 37% water cut, and
2.3 m/s fluid mixture velocity. The hydrate formation starts around 0.7 hours with a high hydrate
formation rate, which decreases slightly in a few minutes. The hydrate formation rate remains
constant for a long portion of the experiment, and slows down at the end, when almost all the
Figure 5.6 shows the cumulative gas injection into the loop (a measure of the amount of hy-
drates formed) as function of time of a University of Tulsa flow loop constant pressure experiment,
using Caratinga oil, 50% liquid loading, 25% water cut and 1.2 m/s fluid mixture velocity. The
gas injection in the first five hours of the experiment is required to maintain constant pressure
inside the loop, due to contraction of the gas phase during cooling of the flow loop; there is no
hydrate formation in this period. After five hours in the experiment, the hydrate formation onset
corresponds to a discontinuity in the slope of the cumulative gas injection curve, where the rate of
91
0.1
0.04 Experiment
0.02
0
0 0.5 1 1.5 2
Time (hr)
Figure 5.5: Plot of hydrate volume fraction as function of time, comparing experimental data
and oil-dominated model prediction for an ExxonMobil flow loop constant pressure experiment
with Conroe oil, 70% liquid loading, 37% water cut, and 2.3 m/s fluid mixture velocity (Test #
XoM-2004-2).
gas injection increases due to gas consumption during hydrate formation. The gas injection stops
when all the water is converted into hydrate. The intrinsic kinetics equation was used to estimate
the hydrate formation rate, using a kinetic rate correction factor of 0.002, which also offers a good
prediction of the hydrate formation rate in this University of Tulsa flow loop experiment.
5 CSMHyK
Cumulative gas injection
oil−dom.
4 model
into loop (kg)
3
Experiment
2
0
0 5 10 15
Time (hr)
Figure 5.6: Plot of cumulative gas injection into the loop as function of time, comparing ex-
perimental data and oil-dominated model prediction for a University of Tulsa flow loop constant
pressure experiment with Caratinga oil, 50% liquid loading, 25% water cut, and 1.2 m/s fluid
mixture velocity (Test # TU-2004-3).
Figure 5.7 shows the flow loop pressure as function of time of a University of Tulsa flow loop
constant volume experiment using Troika oil, 50% liquid loading, 25% water cut, and 1.2 m/s
92
fluid mixture velocity. Since this is a constant volume experiment, the pressure decreases due to
contraction of the gas phase during cooling of the flow loop. Upon the hydrate formation onset, a
discontinuity in the rate of change of the pressure is observed, with a greater decrease in the pressure
due to consumption of the gas phase during hydrate formation. Considering the different conditions
of these three experiments, the model predicts with good agreement the behavior observed.
14
Flow Loop Pressure (MPa)
13 CSMHyK
oil−dominated
model
12
Experiment
11
10
0 5 10 15 20
Time (hr)
Figure 5.7: Plot of flow loop pressure as function of time, comparing experimental data and
oil-dominated model prediction for an University of Tulsa flow loop constant volume experiment
with Troika oil, 50% liquid loading, 25% water cut, and 1.2 m/s fluid mixture velocity (Test #
TU-2004-4).
compared against selected experimental data obtained at the ExxonMobil flow loop in the Fall of
2011 (Table 4.2). This set of experiments includes a range of low and high water cuts that can be
used to test the limits of model applicability. Three test cases with low water cut are presented;
Test XoM-2011-9 with Conroe oil, liquid loading of 50 vol.%, and 15% water cut; Test XoM-2011-14
with Conroe oil, liquid loading of 75 vol.%, and 15% water cut; and Test XoM-2011-25, with King
Ranch condensate, 50 % liquid loading, and 25% water cut. One test case with a higher water
cut of 75 vol.% (Test XoM-2011-23) is presented that seems to indicate the transition from oil to
water-dominated systems. The thermal response observed in the flow loop experiments during the
cooling stage was captured by adjusting the overall heat transfer coefficient (an input to the model).
An overall heat transfer coefficient of 80 W/(m2 K) (14 BTU/(ft2 hr K)) was used to match the
cooling stage of the ExxonMobil flow loop experiments. The intrinsic kinetics equation was used
93
Figure 5.8 shows a plot of temperature as function of time of Test XoM-2011-9, with Conroe
oil, 50% liquid loading, 15% water cut, and 1 m/s fluid mixture velocity, where the green curve
is the ambient temperature outside the flow loop, the blue dashed curve is the measured fluid
temperature, and the red curve is the fluid temperature predicted by the oil-dominated hydrate
formation model. The experiment starts with a cooling stage, decreasing the ambient temperature.
There is heat transfer from the fluids inside the pipe to the cooled air outside the loop. In the
model, an overall heat transfer coefficient of 80 W/(m2 K) (14 BTU/(ft2 hr K)) was used to match
the cooling stage of the experiment. Upon the hydrate formation onset, around 2 hours, an increase
in the fluid temperature is observed due to the exothermic nature of the hydrate formation reaction,
310
CSMHyK
oil−dominated
Temperature (K)
300 model
290
Experiment
280 Ambient
temp.
0 1 2 3 4
Time (hr)
Figure 5.8: Plot of fluid and ambient temperature as function of time, comparing experimental
data and oil-dominated model prediction for an ExxonMobil flow loop experiment with Conroe oil,
50% liquid loading, 15% water cut, and 1 m/s fluid mixture velocity (Test # XoM-2011-9).
Figure 5.9 shows the predicted fluid distribution, in terms of volume fraction, inside the flow loop
as function of time for Test XoM-2011-9, the experimental hydrate volume fraction is also shown
(gray dashed curve). The volume fraction of total liquid and gas remains constant throughout
the experiment, while the water volume fraction decreases after hydrates start to form. Water is
almost completely consumed after 4 hours. The model underpredicts the hydrate formation rate
in this low water cut experiment, with a 20% relative error in the amount of hydrates with respect
to the experimental value at the end of the experiment. The intrinsic kinetics equation was used
94
Total liquid
Volume fraction
0.4 Gas
Experiment hyd.
0.2 Water
Hydrate
0
0 1 2 3 4
Time (hr)
Figure 5.9: Plot of phases volume fraction as function of time, for an ExxonMobil flow loop
experiment with Conroe oil, 50% liquid loading, 15% water cut, and 1 m/s fluid mixture velocity.
The estimated experimental hydrate volume fraction is also shown for comparison with the model
prediction (Test # XoM-2011-9).
For Test XoM-2011-14 with a higher liquid loading of 75 vol.% and same low water cut of 15% as
in the previous case, the model also capture the exothermic response at the hydrate formation onset
after the experimental cooling period, as observed in Figure 5.10. The predicted fluid distribution
as function of time is shown in Figure 5.11; the experimental hydrate volume fraction is also
shown (gray dashed curve). The volume fraction of gas remains almost constant throughout the
experiment, with a slight decrease during the hydrate formation. Note that this is a constant
pressure experiment and gas is being injected into the loop to maintain a constant pressure. The
water volume fraction decreases after hydrates start to form. In this experiment, water is not
completely consumed because of the greater amount of liquids. The model overpredicts the hydrate
volume fraction with a relative error of 10% with respect to the experimental value, for this low
water cut experiment. However, the predicted hydrate formation rate seems to be in agreement
with the experiment, since the hydrate volume fraction curves are nearly parallel.
With a different oil phase (King Ranch condensate) used in Test XoM-2011-25, with 50 % liq-
uid loading, 25% water cut, and 1 m/s fluid mixture velocity, the model for hydrate formation in
oil-dominated systems approximates the hydrate formation rate at earlier formation times (Fig-
ure 5.12). For the hydrate volume fraction estimates, there is an increase in the hydrate formation
rate observed in the experiment (at around 2 hours) that is not capture by the model (see Fig-
ure 5.12). However, in general the model does capture the trend of the experimental data and can
95
310
CSMHyK
oil−dominated
Temperature (K)
300 model
290 Experiment
280
Ambient
temp.
0 1 2 3
Time (hr)
Figure 5.10: Plot of fluid and ambient temperature as function of time, for an ExxonMobil flow
loop experiment with Conroe oil, 75% liquid loading, 15% water cut, and 1 m/s fluid mixture
velocity (Test # XoM-2011-14).
0.4
Gas
0.3
Volume fraction
0.2
Water
Hydrate
0.1
Experimental
hyd.
0
0 0.5 1 1.5 2 2.5 3 3.5
Time (hr)
Figure 5.11: Plot of phases volume fraction as function of time, for an ExxonMobil flow loop
experiment with Conroe oil, 75% liquid loading, 15% water cut, and 1 m/s fluid mixture velocity.
The estimated experimental hydrate volume fraction is also shown for comparison with the model
prediction (solid lines) (Test # XoM-2011-14).
96
be used for estimation purposes, though underpredicts hydrate volume fraction at later times.
0.2
0.1
CSMHyK
0.05 oil−dominated
model
0
0 1 2 3 4
Time (hr)
Figure 5.12: Plot of hydrate volume fraction as function of time, comparing experimental data
with predictions using the oil-dominated hydrate formation model for an ExxonMobil flow loop
experiment with King Ranch condensate, 50% liquid loading, 25% water cut, and 1 m/s fluid
mixture velocity (Test # XoM-2011-25).
For Test XoM-2011-23, with Conroe oil, 50% liquid loading, 75% water cut and 1 m/s fluid
mixture velocity, the plot of hydrate volume fraction as function of time (Figure 5.13) shows good
agreement between the predicted and measured hydrate formation rate, up to about 3 hours, where
the model experiences some instabilities in the calculated hydrate volume fraction. The assumption
of all water dispersed as droplets in a continuous oil phase, results in all hydrate particles being
carried by the oil phase. At a certain hydrate concentration the model predicts a huge increase
in the oil effective viscosity, predicting a hydrate plug, which results in the instabilities observed.
This is related with the assumption of maximum packing fraction of spherical particles in the
oil phase, which is reached considering the small amount of oil in this case. The 75% water cut
represents a limiting condition for the oil-dominated model, where the main assumptions made in
the development of the model are no longer valid. This high water cut value could represent the
Water-dominated systems are characterized by a free and continuous water layer. As the oil/gas
fields mature, the produced water content increases, excess water cannot be further dispersed in
the oil, and a free water layer may form. Some oil and gas will also be dispersed in the free water
97
0.5
CSMHyK
oil−dominated
0.3
0.2
Experiment
0.1
0
0 1 2 3 4 5
Time (hr)
Figure 5.13: Plot of hydrate volume fraction as function of time, comparing experimental data
and oil-dominated model prediction for an ExxonMobil flow loop experiment with Conroe oil, 50%
liquid loading, 75% water cut, and 1 m/s fluid mixture velocity (Test # XoM-2011-23).
A simplified scenario that considers a liquid aqueous phase and a hydrocarbon gaseous phase is
used for the development of a conceptual model for hydrate formation in high water cut systems.
For modeling purposes, the problem is divided into three stages (see Figure 5.14). The first stage
consists of estimating surface area between water and gas phases, which can be obtained from
reported multiphase flow correlations that estimate the entrainment of gas bubbles in the continuous
water phase. Second, assuming immediate hydrate nucleation to occur after a specified subcooling,
the hydrate growth rate can be calculated using a mass transfer-based growth model. Third, the
plugging criterion can be estimated as function of hydrate volume fraction, fluid velocity and liquid
holdup.
Figure 5.14: Conceptual picture for hydrate formation in water-dominated systems consisting of
gas and water phases.
98
To complement the stages identified in Figure 5.14 for the mechanism of hydrate formation in
water-dominated systems, it is required to merge two known phenomena of multiphase flow. First,
the determination of multiphase gas–liquid flow regimes, as illustrated in Figure 5.15, can be useful
in estimating pressure drop in the pipeline and surface area between water and gas phases. Second,
solid particles that are dispersed in the liquid phase upon hydrate formation resemble slurry flow.
Figure 5.15: Schematic illustration of gas-liquid flow regimes in horizontal pipe; adapted from
Brennen (2005).
With increasing solid particle volume fraction, four different slurry flow regimes can occur in
water–hydrate systems, as illustrated in Figure 5.16 (Brennen, 2005). The first solid–liquid flow
regime consists of a homogeneous distribution of well-mixed solid particles in the liquid phase
at low particle concentration. The second solid–liquid flow regime consists of a heterogeneous
distribution with faster sedimentation of larger particles generating a vertical gradient in the particle
concentration. The third solid–liquid flow regime consists of a moving bed of particles. The fourth
The hydrate formation mechanism in water-dominated systems differs from the one proposed for
oil-dominated systems. Joshi et al. (2011b) studied hydrate-hydrate cohesion forces in water using a
micromechanical force apparatus, concluding that the magnitude of hydrate cohesion forces in water
are relatively low and cannot explain particle agglomeration as part of the plugging mechanism.
Particle agglomeration is a major part of the hydrate plugging mechanism in oil-dominated systems.
Further studies of water-dominated systems by Joshi et al. (2011a) using the ExxonMobil flow loop
99
Figure 5.16: Schematic illustration of slurry flow (solids dispersion in liquid phase) regimes; adapted
from Brennen (2005) and Kitanovski and Poredos (2002).
showed that pressure drop in the pipe depends on a complex flow regime transition, involving
multiple phases and solid dispersion in the liquid phase. Three different regions were identified
based on pressure drop measurements across the flow loop after hydrate nucleation, as shown in
Figure 5.17. Region I consists of constant pressure drop upon hydrate formation, which is associated
with a homogeneous distribution of hydrate particles in the liquid phase. Region II consists of a
sharp increase in pressure drop with increasing hydrate concentration. The hydrate concentration
at which the pressure drop starts increasing is identified as Φtransition , and denotes the transition
to a heterogeneous distribution of hydrate particles in the liquid phase. Region III consists of large
fluctuations in pressure drop, representing a plugged condition in the flow loop. It is hypothesized
that a particle bed and a wall deposit forms in Region III, increasing pressure drop.
results when the multiphase gas–liquid flow regimes are combined with slurry flow regimes, as il-
lustrated in Figure 5.18. The proposed conceptual model for water-dominated systems starts with
a homogeneous distribution of hydrate particles in a stratified wavy gas–liquid flow regime. In-
creasing the hydrate volume fraction, the system evolves to a heterogeneous distribution of hydrate
particles in the liquid phase, due to buoyant forces. An experimentally observed increase in hydrate
volume fraction marks the transition to a more complicated flow regime, which exhibits combined
characteristics of slug flow with a stationary bed at the tail of the slug and a heterogeneous distri-
bution of hydrate particles into the liquid slug; this seems to be responsible for large fluctuations
100
Figure 5.17: Typical behavior of measured pressure drop observed after hydrate formation onset
in high water cut flow loop experiments; adapted from Joshi et al. (2011a).
From high water cut experiments performed in the ExxonMobil flow loop varying the pump
speed, Joshi et al. (2011a) observed that the transition from Region I to Region II is irreversible
with respect to mixture velocity, suggesting that the transition hydrate concentration (Φtransition )
would be an indication of eventual plugging with continued hydrate growth. This transition may
Reynolds number and liquid loading was developed by Grasso (presented in the August 2012 CSM
hydrate consortium meeting) from experimental data obtained at the ExxonMobil and University
of Tulsa flow loops at constant mixture velocities (Figure 5.19). The mixture Reynolds number is
defined as,
ρw vm d
Re = (5.8)
µw
This Φtransition correlation can be applied as a prediction tool to estimate the maximum amount
of hydrates that a system is able to flow without a plugging risk. This correlation is applicable
between two boundaries, the lower boundary is the deposition velocity and the upper boundary is
101
Figure 5.18: Conceptual model for hydrate formation mechanism in water-dominated systems
consisting of an aqueous phase, a gaseous phase and solid hydrate particles, showing transition of
flow regimes with increasing hydrate volume fraction.
102
5.2.1 Description of the model
A mass transfer model proposed by Skovborg and Rasmussen (1994) is used to predict the gas
dmgas
= −kmass As (Csol − Csol−hyd ) (5.9)
dt
The mass transfer model uses as driving force for hydrate formation, the difference between gas
solubility in water (Csol ) and gas concentration in water in equilibrium with hydrates (Csol−hyd ).
The mass transfer model relies on two key quantities: the gas-water surface area (As ) and the mass
The gas-water surface area is determined using empirical correlations that depend on the mul-
tiphase flow regime (stratified flow and slug flow). For stratified flow, the gas-water surface area is
calculated assuming a stratified smooth interface, according to the illustration shown in Figure 5.20.
The cross sectional area of the water layer (Aw ) as a function of pipe radius (r) and wetted angle
r2
Aw = (2α − sin 2α) (5.10)
2
Figure 5.20: Cross section of pipe with stratified smooth flow regime.
Ap = πr2 (5.11)
103
The relationship between liquid holdup (HL ) and wetted angle (α) in stratified flow is obtained
Aw 2α − sin 2α
HL = = (5.12)
Ap 2π
This is a non-linear function of the wetted angle. Given the liquid hold up, this equation can be
solved for the wetted angle using a numerical method. Alternatively, an explicit approximation for
the wetted angle as function of liquid hold up can be used to avoid iterative calculations (Biberg,
1999),
π 1/3
α≈ 1 − (1 − HL )1/3 + HL (5.13)
2
Then, the gas-water surface area per unit volume for stratified flow is,
dissipation of energy at the front of the liquid slug generates surface area by the entrainment of
gas bubbles in the liquid slug, as illustrated in Figure 5.21 (Andreussi and Bendiksen, 1989). The
gas bubble dispersion in the water layer can be calculated using a correlation for liquid holdup in
the slug (HLs ) as a function of the mixture velocity (Gregory et al., 1978),
1
HLs = (5.15)
vm 1.39
1+ 8.66
where vm is the mixture velocity.
Figure 5.21: Illustration of gas bubbles entrainment in the liquid slug, due to energy dissipation
in the front of the liquid slug; adapted from Andreussi and Bendiksen (1989).
There is an alternative correlation to estimate the liquid holdup in the slug by Gomez et al.
(2000), which can be used in horizontal and upward vertical flow and is a function of the slug
104
The liquid slug Reynolds number is defined as,
ρL vm d
ReLs = (5.17)
µL
The gas bubble size in the slug body can be estimated using a correlation for horizontal pipelines
under turbulent liquid flow conditions based on the theory proposed by Levich (1962), as presented
by Hesketh et al. (1987). The maximum stable gas bubble diameter is given by,
σ 0.6
0.5
d
db,max = 1.38 (W ecrit )0.6 (5.18)
ρc0.3 ρd0.2 µc0.1 1.1
vSc
where W ecrit is the critical Weber number, which is equal to 1.0 if the forces tending to break up
and the forces tending to restore the gas bubble are equal (i.e., equilibrium bubble size); σ is the
interfacial tension; ρc is the density of the continuous phase, ρd is the density of disperse phase, µc
is the viscosity of continuous phase; d is the internal pipe diameter; and vSc is superficial velocity
The Sauter mean diameter (db,32 ) is related to the interfacial area of the dispersed phase and
should be used for the purposes of mass and heat transfer calculations,
The number of bubbles per unit volume (Nbubbles ) is obtained from the liquid slug void fraction
(1 − HLs ) divided by the volume of one gas bubble, considering the Sauter mean diameter,
6(1 − HLs )
Nbubbles = (5.20)
πd3b,32
Then the gas-water surface area per unit volume for slug flow is,
The mass transfer coefficient is calculated using a correlation presented by Hanratty (1991),
vL∗
kmass = 0.5 (5.22)
NSC
where vL∗ is the characteristic liquid phase velocity given by,
r
τi
vL∗ = (5.23)
ρL
and NSC is the Schmidt number defined as,
µL
NSC = (5.24)
ρL Dgas−water
105
where Dgas−water is diffusivity coefficient of gas in water. Here the gas is assumed to be mostly
methane and the diffusivity coefficient of methane in water can be calculated using the Wilke-Chang
ρg (vg − vL )2
τi = fi (5.25)
2
The interfacial friction factor (fi ) is assumed constant and equal to 0.022.
The concentrations of methane in water (Csol and Csol−hyd ), used in equation 5.9, are calculated
pressure and temperature values. A look-up table is used to interpolate the methane concentrations
The model for hydrate formation in water-dominated systems was implemented in this thesis
plug-in program, that allows to define a gas hydrate phase in the OLGA
R
’s flow model framework.
Figure 5.22 shows a schematic representation of the mass fields considered for water-dominated
systems, which includes gas and water as continuous phases, and hydrate and gas as dispersed
phases in the water continuous phase. The user-defined plug-in program calculates additional
contributions to the multiphase flow conservation equations using the PVT and flash modules.
Figure 5.22: Illustration of mass fields considered in the water-dominated hydrate formation
model coupled with the OLGA
R
transient multiphase flow simulator, including gas and water as
continuous phases, and hydrate and gas as dispersed phases in the water continuous phase.
106
The PVT module calculates the properties for the hydrate phase, making additional contri-
butions to the mass, momentum and energy conservation equations. Typical values of hydrate
structure I (Sloan and Koh, 2008a) are assigned for density (910 kg/m3 ), heat capacity (2080 J/(kg
K)), thermal conductivity (0.49 W/(m K)). The enthalpy of the hydrate phase (hhyd ) is a function
of the enthalpy of the gas (hg ) and water (hw ) phases at the system temperature and pressure,
and the heat of reaction (∆hhyd ), to account for the exothermic nature of the hydrate formation
reaction,
hg + nH hw + ∆hhyd
hhyd = (5.26)
1 + nH
where nH is the hydration number (water to gas ratio in the hydrate phase).
The flash module calculates the gas consumption rate during hydrate formation using the mass
transfer model (equation 5.9). The flash module makes direct contributions to the mass conservation
equations by calculating the mass transfer rates from gas and water phases to the hydrate phase
during hydrate formation, and the opposite during hydrate dissociation where gas and water are
released from the hydrate phase. The mass transfer rates are used to update the phase masses of
The prediction performance of the CSMHyK water-dominated model was compared against
experimental data obtained using the ExxonMobil flow loop for high water cut systems. Ten high
water cut experiments were selected for predictions using the water-dominated hydrate formation
model (Table 5.1). These experiments cover the range of high water cut values from 100 to 75
vol.%, and different mixture velocities from 0.75 to 2.5 m/s. The cases with 100 vol.% water cut,
only included hydrocarbon gas (methane) and liquid water, no liquid hydrocarbon phase was loaded
The water-dominated hydrate formation model showed excellent agreement with experimental
data for two cases with 50 vol.% liquid loading and 100 vol.% water cut; Test XoM-2011-1 with 1
m/s fluid mixture velocity, and Test XoM-2010-2 with 2.5 m/s fluid mixture velocity. Figure 5.23
to Figure 5.25 show the comparison between the experimental data and simulation results from
Test XoM-2011-1 with 50% liquid loading, 100% water cut, and 1 m/s fluid mixture velocity.
107
Table 5.1: Experiments from the ExxonMobil flow loop selected for predictions using the water-
dominated hydrate formation model.
Figure 5.23 shows the temperature of the system as function of time; in the model, an overall heat
transfer coefficient of 80 W/(m2 K) (14 BTU/(ft2 hr K)) was used to match the cooling stage of
the experiment. This case did not show a noticeable temperature increase at the time of hydrate
nucleation. Figure 5.24 shows the gas injected into the loop, required to maintain a constant
pressure inside the flow loop, as function of time. The initial increase in gas injection corresponds
to the cooling of fluids, with a contraction of the gas phase. There is a change in the slope of
the gas injection curve that indicates the hydrate formation onset. The slope of the gas injection
curve during hydrate formation can be related to the hydrate formation rate. There is a good
match between the experimental data and the simulation results. Figure 5.25 shows the hydrate
volume fraction as function of time with the calculated Φtransition . When the hydrate volume
fraction is greater than Φtransition the system enters in Region II and the pressure drop increases,
eventually leading to a hydrate plug. A deviation of the estimated hydrate volume fraction from
the experimental values above the Φtransition is observed, since the model does not considers the
phenomena affecting the pressure drop, which tend to decrease the hydrate formation rate.
For Test XoM-2010-2, with a higher fluid mixture velocity (2.5 m/s), the experiment exhibited
a slight increase in temperature at the point of hydrate formation onset, which is captured by the
model, as shown in Figure 5.26. Figure 5.27 shows the gas injected into the loop, with an excellent
agreement between the model prediction and the experimental values during the cooling stage of the
108
310
CSMHyK
Temperature (K)
300 water−dominated
model
290
Experiment
280
Ambient (input)
270
0 0.5 1 1.5 2 2.5
Time (hr)
Figure 5.23: Plot of fluid and ambient temperature as function of time, comparing experimental
data with predictions using the model for hydrate formation in water-dominated systems for an
ExxonMobil flow loop experiment with 50% liquid loading, 100% water cut, and 1 m/s fluid mixture
velocity (Test # XoM-2011-1).
10 CSMHyK
Gas injected (kg)
water−dominated
8
model
6
4
Experiment
2
0
0 0.5 1 1.5 2 2.5
Time (hr)
Figure 5.24: Plot of gas injected into flow loop as function of time, comparing experimental
data with predictions using the model for hydrate formation in water-dominated systems for an
ExxonMobil flow loop experiment with 50% liquid loading, 100% water cut, and 1 m/s fluid mixture
velocity (Test # XoM-2011-1).
109
Hydrate volume fraction
0.25
Experiment
0.2
CSMHyK
in liquid
0.15 water−dominated
model
0.1
Φtransition
0.05
0
0 0.5 1 1.5 2 2.5
Time (hr)
Figure 5.25: Plot of hydrate volume fraction as function of time, comparing experimental data with
predictions using the model for hydrate formation in water-dominated systems for an ExxonMobil
flow loop experiment with 50% liquid loading, 100% water cut, and 1 m/s fluid mixture velocity
(Test # XoM-2011-1), also showing the calculated Φtransition .
experiment (i.e., gas injection required to maintain constant pressure due to gas phase contraction
during cooling). Upon hydrate formation the gas injection rate increases, and the model matches
the measured gas injection rate. Figure 5.28 shows the hydrate volume fraction as function of time
with the calculated Φtransition . Similar to the previous case, the model predicts a hydrate formation
rate in agreement with the experimental values up to the transition hydrate concentration, where
predicted and measured values diverge. In this case, the hydrate formation rate increases slightly
above Φtransition .
CSMHyK
Temperature (K)
300
water−dominated
model
290
Experiment
270
0 0.5 1 1.5 2
Time (hr)
Figure 5.26: Plot of fluid and ambient temperature as function of time, comparing experimental
data with predictions using the model for hydrate formation in water-dominated systems for an
ExxonMobil flow loop experiment with 50% liquid loading, 100% water cut, and 2.5 m/s fluid
mixture velocity (Test # XoM-2010-2).
110
20
Experiment
0
0 0.5 1 1.5 2
Time (hr)
Figure 5.27: Plot of gas injected into flow loop as function of time, comparing experimental
data with predictions using the model for hydrate formation in water-dominated systems for an
ExxonMobil flow loop experiment with 50% liquid loading, 100% water cut, and 2.5 m/s fluid
mixture velocity (Test # XoM-2010-2).
Hydrate volume fraction
Experiment
in liquid
0.4
0.2 CSMHyK
water−dom. Φtransition
model
0
0 0.5 1 1.5 2
Time (hr)
Figure 5.28: Plot of hydrate volume fraction as function of time, comparing experimental data with
predictions using the model for hydrate formation in water-dominated systems for an ExxonMobil
flow loop experiment with 50% liquid loading, 100% water cut, and 2.5 m/s fluid mixture velocity
(Test # XoM-2010-2), also showing the calculated Φtransition .
111
For experiments with liquid hydrocarbon (Conroe oil), the model also offers a good approxima-
tion of the experimental data. Results for tests XoM-2011-21 (90 vol.% water cut) and XoM-2011-24
(75 vol.% water cut) are shown in Figure 5.29 and Figure 5.30. In both cases the model predicts
with acceptable agreement the thermal behavior of the system, and underpredicts the amount of
310
Experiment
300 CSMHyK
CSMHyK water−dominated
water−dominated
in liquid
0.4 model
290 model
0.2
280
Φtransition
Ambient Experiment
270 0
0 1 2 3 0 1 2 3
Time (hr) Time (hr)
(a) Temperature (b) Hydrate volume fraction
Figure 5.29: Comparison of experimental data with predictions using the model for hydrate forma-
tion in water-dominated systems for Test XoM-2011-21 with Conroe oil, 50% liquid loading, 90%
water cut, and 1.75 m/s fluid mixture velocity: (a) ambient and fluid temperature as function of
time, (b) hydrate volume fraction as function of time, including the calculated Φtransition .
To summarize the prediction errors of the high water cut cases studied, Table 5.2 presents the
measured and predicted hydrate transition concentrations (Φtransition ), and the relative percentage
error. The hydrate transition concentration is considered a criterion to determine the hydrate plug
formation in the pipeline. The maximum prediction error for the hydrate transition concentration
obtained from the ten cases studied is about 12%, with an average error of 4.5%.
amounts of liquid hydrocarbon (i.e., condensate) and water. Hammerschmidt (1934) was the first to
determine that gas hydrates were plugging natural gas transmission lines well above the ice forma-
tion temperatures. Dorstewitz and Mewes (1994) published the first flow loop study investigating
gas systems. Hydrates were observed to first form on the pipe wall at the water-gas interface. The
112
Hydrate volume fraction
CSMHyK
Temperature (K)
300
water−dominated Experiment
model
in liquid
290 0.4
Experiment
Figure 5.30: Comparison of experimental data with predictions using the model for hydrate forma-
tion in water-dominated systems for Test XoM-2011-24 with Conroe oil, 50% liquid loading, 75%
water cut, and 2.5 m/s fluid mixture velocity: (a) ambient and fluid temperature as function of
time, (b) hydrate volume fraction as function of time, including the calculated Φtransition .
Table 5.2: Summary of prediction error of hydrate volume fraction at the hydrate transition
concentration (Φtransition ), for the ten high water cut cases studied.
113
hydrate layer then grew along the pipe wall, until the entire perimeter of the pipe was covered with
hydrates. Hatton and Kruka (2002) conducted a flow loop study, observing hydrate buildup at the
pipe wall, as well as hydrate formation in the bulk liquid in a horizontal 3-inch pipe. The observed
buildup on the pipe wall suggests a possible plugging mechanism in gas pipelines.
Rao et al. (2011) presented a laboratory scale study of methane hydrate deposition and growth.
Using a high pressure Jerguson visual cell and maintaining a continuous flow of water-saturated
methane gas, they were able to quantify hydrate growth on the outside of a small cold pipe placed
inside the visual cell. From the measurements of thickness and volume of the hydrate deposits as
function of time, Rao et al. (2011) developed a model based on heat and mass balances.
Figure 5.31 shows a proposed conceptual model for gas hydrate formation in gas-dominated
systems, consisting of four main stages. In the first stage, based on the assumption of immediate
hydrate nucleation after a specified subcooling, gas hydrate forms on the pipe wall at the gas-water
interface. Free or condensed water may be present from the saturated gas phase due to a decrease
in temperature. A layer of gas hydrates grow to cover the perimeter of the pipe. The second
stage consists of the inward growth of annular hydrate deposits (Dorstewitz and Mewes, 1994;
Hatton and Kruka, 2002). An increase in shear forces, related to increased flow velocities because
of narrowed conduit diameters, could promote sloughing of hydrate deposits from the pipe wall,
releasing hydrate particles that can accumulate in a jamming process plugging the pipe (Hatton
and Kruka, 2002). However, more fundamental work is required to understand the conditions for
sloughing of hydrates deposits from the pipe wall and subsequent jamming to form a plug. These
last two phenomena are currently under investigation in the Center for Hydrate Research at the
The model for hydrate formation in gas-dominated systems was specially designed for cases
with a water-saturated gas phase flowing through a pipe without free liquid phases. There is no
free water initially in the system, all the water is dissolved in the gas phase. If the pipe surface
temperature falls below the dew point temperature, water condenses on the cold pipe surface.
The amount of liquid water condensing out of the water-saturated gas phase is determined by the
decrease in water concentration in the gas phase, through a phase equilibrium calculation.
114
Figure 5.31: Conceptual picture for hydrate formation in gas-dominated systems consisting of gas,
some condensate and water; adapted from Lingelem et al. (1994).
The condensed liquid water is immediately converted to gas hydrate. The hydrate growth
rate is calculated using an intrinsic kinetic equation, of first order in temperature departure from
equilibrium,
dmgas k2
= −k1 exp As ∆Tsub (5.27)
dt Tsystem
The intrinsic kinetics equation gives the rate of gas consumption during hydrate formation as
a function of the kinetics rate constants (k1 and k2 ), the surface area (As ) between water and
the gas phase, and the subcooling (∆Tsub ) as the thermal driving force. Vysniauskas and Bishnoi
(1983) measured the gas consumption rate during the formation of methane hydrate at different
temperatures and subcooling, in the absence of either mass or heat transfer resistances. Their
experimental data, were used to regressed the intrinsic rate constants, k1 = 7.3548 × 1017 and
k 2 = −13600 K.
The amount of hydrate phase formed is evenly distributed in a deposited layer on the pipe wall.
The hydrate deposit is assumed to have a constant porosity. The porosity of the hydrate deposit
(φhyd deposit ) is used to determine the deposit thickness, through the definition of the hydrate deposit
The presence of a hydrate deposited layer on the pipe wall will add thermal resistance to the
overall heat transfer between the fluid and the inner pipe wall. The heat transfer coefficient of
115
the hydrate deposit (HT Chyd deposit ) is calculated, assuming a steady-state heat balance across
the hydrate wall layer, as the hydrate thermal conductivity (khyd ) divided by the hydrate deposit
khyd
HT Chyd deposit = (5.29)
δhyd deposit
The model for hydrate formation in gas-dominated systems was implemented in this thesis work
in the OLGA
R
transient multiphase flow simulator, through the development of a user-defined plug-
in program, that allows one to define a gas hydrate phase in the OLGA
R
’s flow model framework.
(Figure 5.32) shows a schematic representation of the mass fields considered for gas-dominated
systems, which includes a water-saturated gas phase and a hydrate film deposited on the pipe
wall. The user-defined plug-in program calculates additional contributions to the multiphase flow
conservation equations using the PVT, flash and heat transfer modules.
Figure 5.32: Illustration of the mass fields considered in the gas-dominated hydrate formation
model coupled with the OLGA
R
transient multiphase flow simulator, including a gas phase satu-
rated with water and hydrate film deposited on the pipe wall.
The PVT module calculates the properties for the hydrate deposited layer based on properties
of the hydrate and gas phases. The PVT module makes contributions to the mass, momentum
and energy conservation equations. Typical values of hydrate structure II (Sloan and Koh, 2008a)
are assigned for the hydrate phase properties, such as, density (914 kg/m3 ), heat capacity (498.83
J/(kg K)), thermal conductivity (0.5 W/(m K)). The density of the hydrate deposit is calculated
using equation 5.28. The enthalpy of the hydrate phase (hhyd ) is a function of the enthalpy of the
gas (hg ) and water (hw ) phases at the system temperature and pressure, and the heat of reaction
116
(∆hhyd ), to account for the exothermic nature of the hydrate formation reaction,
hg + nH hw + ∆hhyd
hhyd = (5.30)
1 + nH
where nH is the hydration number (water to gas ratio in the hydrate phase).
The flash module calculates the gas consumption rate during hydrate formation using the in-
trinsic kinetics equation 5.27. The amount of hydrates formed is limited by the available amount
of condensed water. The flash module makes direct contributions to the mass conservation equa-
tions by calculating the mass transfer rates from gas and water phases to the hydrate phase during
hydrate formation, and the opposite during hydrate dissociation where gas and water are released
from the hydrate phase. The mass transfer rates are used to update the phase masses of gas, water
and hydrate.
The heat transfer module calculates the heat transfer coefficient of a deposited hydrate layer
using equation 5.29, and the corrected inner heat transfer coefficient as,
wall, assuming radial heat conduction through concentric wall layers. An internal heat transfer
coefficient (HT Cinner,OLGA ) is calculated for the heat transfer from the fluid to the inner pipe wall.
The presence of a hydrate wall layer adds thermal resistance to the overall heat transfer, and this
This model has not been compared against experimental data, because of the scarcity of ex-
perimental data for gas-dominated systems. This represents the first attempt to implement the
capability of predicting hydrate film growth on pipe wall in a transient multiphase flow simulator,
and should be considered as a first pass model. In the next chapter, an example application of this
model is presented, in the context of water-saturated gas flowing through a horizontal pipeline,
5.4 Summary
This chapter shows how the CSMHyK model have been improved and extended to water and
117
gas hydrate phase on the fluid flow behavior. The plug-in program can be coupled with OLGA
R
,
defining a hydrate phase and calculating additional contributions to the conservation equations.
A PVT module is used to calculate properties for the hydrate phase, a flash module is used to
calculate mass transfer rates for the mass conservation equations, and a heat transfer module is
used to calculate the heat transfer coefficient of a deposited hydrate layer, affecting the energy
equation.
The hydrate formation model for oil-dominated systems includes a new dynamic cohesion force
model. The model predicts the hydrate volume fraction measured in flow loop experiments with
good agreement. The water-dominated model uses a mass transfer-based hydrate growth model
to calculate the amount of hydrate formed in the system. A correlation to estimate the hydrate
transition concentration as function of Reynolds number and liquid loading is used as a plugging
criterion. The model predictions matched experimental data obtained in the ExxonMobil flow
loop. The gas-dominated model was also implemented as a plug-in for OLGA
R
. The model for
gas-dominated systems was specially designed for cases with a water-saturated gas phase flowing
through a pipe without free liquid phases. Water condenses on the cold pipe surface and imme-
diately is converted to gas hydrate. The gas-dominated hydrate formation model has not been
compared against experimental data due to the lack of such data, and should be considered as a
118
CHAPTER 6
Offshore explorations in deeper and colder waters impose more challenging scenarios to the flow
assurance of the produced streams. High pressures and low temperatures of operation of subsea
production facilities with longer tiebacks will promote the formation of natural gas hydrates. Gas
hydrates form in the presence of appropriate quantities of gas and water, and are considered one of
the most challenging problems in deep subsea facilities because of their rapid formation compared
A transient gas hydrate model that predicts when and where gas hydrates form in flowlines will
have significant utility for the flow assurance engineer in the oil and gas industry. By predicting
gas hydrate formation and transportability, the gas hydrate model can be applied to design and
optimize oil and gas transport facilities, focusing on prevention, management, or remediation of
This chapter presents examples of applications of the hydrate formation models developed in
this work and presented in Chapter 5 for oil, water and gas-dominated systems. The hydrate
plugging risk is studied, using the typical geometry and fluid properties of an offshore well from
the Caratinga field located in the Campos basin, Brazil (provided by Petrobras).
This section focuses on the application of the hydrate formation model for oil-dominated sys-
tems, to study the hydrate plugging risk of an offshore well. Three different periods of the well
life are considered: an early stage with low produced gas/oil ratio (GOR) and low water cut, a
middle stage with an increased GOR, and a late stage with higher GOR and higher water cut.
The hydrate plugging risk is estimated from the calculation of three performance measures (pres-
sure drop along the flowline, hydrate volume fraction, and hydrate slurry relative viscosity) in an
attempt to quantify the plugging risk. Once the hydrate plugging risk is estimated for the three
different periods of the well life, the gas hydrate model is used to optimize the concentration of
thermodynamic hydrate inhibitor (THI) to be injected in order to reach a safe operation condition.
119
In this study, ethanol is considered as the THI because it is the typical hydrate inhibitor used in
South America (because of the large industrial production of ethanol in the region).
Additionally, the hydrate model for oil-dominated systems is used to study the effect of hydrates
in transient shut-in/restart operations of the offshore well and the performance of hydrate inhibition
in terms of ethanol concentration. In general, ethanol injection decreased the hydrate-plugging risk
This section presents the approach followed to estimate the hydrate plugging risk in an offshore
well using the gas hydrate prediction model for oil-dominated systems. The simulation case consists
of typical well/flowline/riser geometry and fluid properties from the Caratinga field located at the
Campos Basin in Brazil. Figure 6.1 shows the geometry in terms of horizontal length and depth
from sea level, which consists of a slanted well with a diameter of 5 in., a departure of 2000 m, and
a depth of 1800 m. The well is connected to a straight flowline with a slight uphill slope, a diameter
of 6 in., and a horizontal length of approximately 13 km, followed by the bend into the riser that
leads up to the platform. Table 6.1 presents a summary of the Caratinga crude-oil properties and
0
Platform
500 Wellhead
1000
Depth (m)
Flowline
1500
2000 Well
2500
3000
0 5 10 15
Horizontal distance (km)
Figure 6.1: Geometry of well, flowline, and riser, based on typical geometries from the Caratinga
field located at the Campos Basin in Brazil.
As a first approach to estimate hydrate plugging risk, a pressure versus temperature diagram,
shown in Figure 6.2, was constructed overlaying the flowline operation conditions under steady-state
120
Table 6.1: Properties and composition of Caratinga crude oil (Sjöblom et al., 2010).
Property Value
Density (kg/m3 ) 914.0
Viscosity dead oil (Pa.s) 0.262
Interfacial tension (mN/m) 23
Asphaltene content (%) 6.2
Saturates content (%) 39.8
Aromatic content (%) 39.8
Resin content (%) 14.3
flow with hydrate equilibrium curves at different ethanol concentrations. Whenever the flowline
operating conditions curve lies to the left of the hydrate equilibrium curves is considered a region
for hydrate formation. It can be observed that a portion of the flowline operating conditions curve
is located inside the hydrate formation region, confirming the feasibility of forming hydrates in this
system, and that more than 30 wt% ethanol is required to totally inhibit the system. However,
this diagram does not provide information about the amount of hydrates formed and hydrate
transportability; hence nothing can be said about the actual plugging risk level of this system.
14
Ethanol concentration (wt%)
40 30 20 10 0
12 Wellhead
Pressure (MPa)
630 m 210 m
10 3.2 km
Flowline operation
8 conditions
6 14.1 km
4
2
Topsides
0
260 280 300 320
Temperature (K)
Figure 6.2: Diagram of pressure versus temperature showing flowline operation conditions during
steady-state flow and hydrate equilibrium curves at different ethanol concentrations. Points on
flowline operation conditions curve indicate distance from wellhead.
121
The hydrate formation model for oil-dominated systems coupled with the OLGA
R
transient
multiphase flow simulator can be used to determine the flowline operating conditions and estimate
the amount and transportability of gas hydrates at different periods of the well life. During the
well life, an increase in the GOR and in the water cut is expected, which represents an increase
in the amount of fundamental components for gas hydrate formation. This means that different
periods in the well life are characterized by different magnitudes of GOR and water cut, resulting in
different amounts and transportability of hydrates in the system. On the basis of this information,
hydrate remediation techniques can also differ for different periods in the well life. In this study,
we consider three different scenarios representing periods in the well life: (i) an early stage with
low GOR and low water cut (Case 1), (ii) a middle stage with an increased GOR (Case 2), and
(iii) a late stage with higher GOR and higher water cut (Case 3).
The hydrate plugging risk is determined from the calculation of three performance measures
(∆Pf lowline is the pressure drop along the flowline, Φhyd is the hydrate volume fraction in pipe, and
µr is the hydrate slurry relative viscosity) and is classified in three qualitative levels as follows:
• Low-risk: ∆Pf lowline < 2 MPa (300 psi); Φhyd < 0.10; µr < 10. At this level, formed hydrates
are considered to be easily transported through the pipeline and up to the riser.
• Intermediate-risk: 2 MPa (300 psi) < ∆Pf lowline < 3.45 MPa (500 psi); 0.10 < Φhyd < 0.40;
10 < µr < 100. At this level, hydrates could still flow through the pipeline but represent a
risk where there are restrictions or changes in flow direction, where hydrates can accumulate,
• High-risk: ∆Pf lowline > 3.45 MPa (500 psi); Φhyd > 0.40; µr > 100. At this level, a very
viscous hydrate slurry forms, fluid flow stops, and the pipeline plugs.
Note that the performance measure values used to delimit the three levels of hydrate plugging
risk in this study were determined from the current version of the gas hydrate model and for the
considered well/flowline/riser geometry, and may or may not underestimate the effect of hydrates
on fluid flow in more general scenarios. These values could be adjusted by improving the accuracy
of the model with further knowledge of the mechanisms of hydrate formation and transportability
122
Case 1: Low Production GOR and Low Water Cut
Case 1 represents an early stage in the well life characterized by a liquid loading of 90 vol%, a
water cut of 30%, and a production GOR of 570 scf/STB. The calculated temperature distribution
along the flowline length is presented in Figure 6.3 with the hydrate equilibrium temperature.
The warm fluids flowing from the well cool down at the beginning of the flowline because of heat
transfer to the ocean (ambient temperature of 277 K). Hydrates start to form when the fluids
temperature decreases 3.6 K below the hydrate equilibrium temperature (i.e., a subcooling of 3.6
K). The hydrate heat of formation raises the temperature of the system. Then the heat transfer to
the ocean balance with the heat of formation, making the system temperature equal to the hydrate
equilibrium temperature, in a condition known as heat transfer limited. When all the free gas phase
is converted into hydrates, the hydrate formation reaction stops. The system cools down until the
seafloor temperature is reached, with no further hydrate formation. Then, the system temperature
340
Temperature (K)
320
280
Fluids temperature
260
0 5 10 15
Pipeline Length (km)
Figure 6.3: Temperature distribution along the pipeline length, showing system temperature
and hydrate equilibrium temperature corresponding to the steady-state flow predicted by the oil-
dominated gas hydrate formation model for Case 1.
The plot of the fluids distribution shown in Figure 6.4 supports the explanation previously
presented for the system temperature behavior. It can be observed at the beginning of the flowline
that the liquid volume fraction started at 0.9 (i.e., 90 vol% liquid loading) and increases up to a
100% liquid loading in the flowline, because of free gas being consumed during hydrate formation.
Some hydrates continue to form, due to the pressure drop along the flowline resulting in gas evolving
123
from the oil phase. Gas hydrates are transported and melted up the riser. Figure 6.5 shows the
ratio of hydrate slurry viscosity to the viscosity of the continuous oil phase; a maximum hydrate
slurry viscosity of 3.4 times the oil viscosity can be observed, with a location that coincides with
0.6
0.4
Water
0.2
Gas Hydrate
0
0 5 10 15
Pipeline Length (km)
Figure 6.4: Fluid distribution in terms of volume fraction along the pipeline length, corresponding
to the steady-state flow predicted by the oil-dominated gas hydrate formation model for Case 1.
3.5
Slurry Relative Viscosity
2.5
1.5
1
0 5 10 15
Pipeline Length (km)
Figure 6.5: Hydrate slurry relative viscosity along the pipeline length, corresponding to the steady-
state flow predicted by the oil-dominated gas hydrate formation model for Case 1.
In general, Case 1 is considered to have a low risk of hydrate plugging because the calculated
flowline pressure drop of 1.6 MPa (230 psi) is less than the specified limit of 2 MPa (300 psi),
the maximum hydrate fraction in pipe of 0.09 is less than 0.10, and the maximum hydrate-slurry
124
relative viscosity of 3.4 is less than 10.
Case 2 represents a middle stage in the well life characterized by a liquid loading of 79 vol%,
a water cut of 32%, and a production GOR of 894 scf/STB. Figure 6.6 shows the calculated tem-
perature distribution along the flowline length and the hydrate equilibrium temperature. Similar
to that of Case 1, the warm fluids flowing from the well cool down at the beginning of the flowline
until a subcooling temperature of 3.6 K is reached when hydrates nucleate. The hydrate heat of
formation raises the temperature, and the system becomes heat transfer limited until all the ad-
ditional gas is converted into hydrates, consuming the water phase. When the hydrate formation
reaction stops, the system cools down, reaching the seafloor temperature with no further hydrate
formation. Then, the system temperature increases and hydrates dissociate, flowing up the riser.
340
Temperature (K)
320
280
Fluids temperature
260
0 5 10 15
Pipeline Length (km)
Figure 6.6: Temperature distribution along the pipeline length, showing system temperature
and hydrate equilibrium temperature corresponding to the steady-state flow predicted by the oil-
dominated gas hydrate formation model for Case 2.
Figure 6.7 shows a plot of fluid distribution along the pipeline length. It can be observed at
the beginning of the flowline that the liquid volume fraction started at nearly 0.8 and increases
until the flowline is filled by liquid; simultaneously, the gas volume fraction is reduced because
of a decrease in temperature and gas consumption during hydrate formation. In this case, water
is consumed almost completely during hydrate formation because of the increased GOR from the
well. Hydrates are melted on their way up the riser. But now a greater amount of hydrates, when
compared to Case 1, is present in the flowline, and could represent a higher risk of accumulating,
125
jamming, and plugging during flow through a restriction or change in flow direction. Figure 6.8
shows the ratio of hydrate slurry viscosity to the viscosity of the continuous oil phase along the
flowline length; a maximum hydrate slurry viscosity of 8.9 times the oil viscosity can be observed.
0.4 Hydrate
Gas
0.2
Water
0
0 5 10 15
Pipeline Length (km)
Figure 6.7: Fluid distribution in terms of volume fraction along the pipeline length, corresponding
to the steady-state flow predicted by the oil-dominated gas hydrate formation model for Case 2.
10
Slurry Relative Viscosity
0
0 5 10 15
Pipeline Length (km)
Figure 6.8: Hydrate slurry relative viscosity along the pipeline length, corresponding to the steady-
state flow predicted by the oil-dominated gas hydrate formation model for Case 2.
Case 2 is considered to have an intermediate risk of plugging with hydrates because the calcu-
lated flowline pressure drop of 2.7 MPa (398 psi) is between the specified lower limit of 2 MPa and
the upper limit of 3.45 MPa, and the maximum hydrate fraction in pipe of 0.30 is greater than
0.10.
126
Considering the previous results and the estimated hydrate plugging risk, the gas hydrate model
for oil-dominated systems was used to study the performance of THI injection, with the objective
of determining the optimum ethanol concentration needed to lower the hydrate plugging risk of
Case 2 to a low risk level. Figure 6.9 shows three plots of the performance measures (∆Pf lowline ,
Φhyd , and µr ) used to establish the hydrate plugging risk as function of ethanol concentration. It
can be observed that the injection of ethanol reduces the performance measures up to an ethanol
concentration of 30 wt%. The pressure drop along the flowline has a minimum value at 30 wt%,
while the hydrate volume fraction in pipe and the hydrate slurry relative viscosity become constant,
indicating a complete dissociation of hydrates. The ethanol injection reduces the hydrate plugging
risk to a safer operation point if the injected concentration is greater than or equal to 30 wt% of
ethanol.
2.75 0.4 10
Pressure drop along flowline (MPa)
2.7 8
0.3
2.65 6
0.2
2.6 4
0.1
2.55 2
2.5 0 0
0 10 20 30 40 0 10 20 30 40 0 10 20 30 40
Ethanol concentration (wt%) Ethanol concentration (wt%) Ethanol concentration (wt%)
Figure 6.9: Performance measures (∆Pf lowline , Φhyd , and µr ) used to establish the hydrate plugging
risk as a function of ethanol concentration, predicted by the oil-dominated gas hydrate formation
model for Case 2.
Case 3 represents a late stage in the well life characterized by a liquid loading of 67.4 vol%,
a water cut of 52.5%, and a production GOR of 1,677 scf/STB. Figure 6.10 shows the calculated
temperature distribution along the flowline length and the hydrate equilibrium temperature at the
plugged condition. A great portion of the flowline length is at very low temperatures. The system
temperature increases and follows the hydrate equilibrium temperature in the direction of the riser.
Figure 6.11 shows the plot of fluids distribution along the pipeline length. It can be observed
that the liquid loading remains fairly high and the gas volume fraction decreases slightly, but the
127
310
300
Temperature (K)
Hydrate equilibrium temperature
290
280
260
0 5 10 15
Pipeline Length (km)
Figure 6.10: Temperature distribution along the pipeline length, showing system temperature
and hydrate equilibrium temperature corresponding to the steady-state flow predicted by the oil-
dominated gas hydrate formation model for Case 3.
water volume fraction decreases from the beginning of the flowline until it is completely consumed
in the hydrate formation reaction, resulting in a massive accumulation of hydrates in the middle of
the flowline. Figure 6.12 shows the ratio of hydrate slurry viscosity to the viscosity of the continuous
oil phase (in a logarithmic scale) along the pipeline length; it can be observed that the maximum
hydrate slurry viscosity reaches values greater than 1,000 times the oil viscosity, confirming the
0.8
Volume Fraction
Total liquids
0.6 Hydrate
0.4
Gas
0.2 Water
0
0 5 10 15
Pipeline Length (km)
Figure 6.11: Fluid distribution in terms of volume fraction along the pipeline length, corresponding
to the steady-state flow predicted by the oil-dominated gas hydrate formation model for Case 3.
In Case 3, the viscous hydrate slurry formed prevents the fluid from flowing. This case has
a high risk of plugging with hydrates because the calculated flowline pressure drop of 9 MPa is
128
4
10
2
10
1
10
0
10
0 5 10 15
Pipeline Length (km)
Figure 6.12: Hydrate slurry relative viscosity along the pipeline length, corresponding to the
steady-state flow predicted by the oil-dominated gas hydrate formation model for Case 3.
greater than 3.45 MPa, the maximum hydrate fraction in pipe of 0.55 is greater than 0.40, and the
The oil-dominated hydrate formation model was used to study the performance of THI injection
at the conditions of Case 3, with the objective of determining the optimum ethanol concentration
needed to lower the hydrate plugging risk to a low risk level. Figure 6.13 shows three plots of the
performance measures (∆Pf lowline , Φhyd , and µr ) used to establish the hydrate plugging risk as
a function of ethanol concentration. It can be observed that the injection of ethanol reduces the
performance measures, preventing the formation of a hydrate plug. For 10 wt% of ethanol injected,
the models estimates a minimum value of pressure drop along the flowline for this systems, but
the amount of hydrates (about 40 vol.%) and slurry relative viscosity (> 10) indicates that the
system is at an intermediate hydrate plugging risk. The pressure drop along the flowline increases
with further increase in ethanol injection, which is related to the increase in liquid content and
the released gas from hydrate dissociation that changes the multiphase flow regime. The hydrate
volume fraction and the hydrate slurry relative viscosity decrease with increasing injected ethanol
Ethanol injection prevents formation of a hydrate plug and remediates the hydrates completely if
the injected concentration is greater than or equal to 30 wt% of ethanol, reducing the hydrate
plugging risk.
129
4
0.8
Pressure drop along flowline (MPa)
10 10
2
6 0.4 10
1
4 0.2 10
0
2 0 10
0 10 20 30 0 10 20 30 0 10 20 30
Ethanol concentration (wt%) Ethanol concentration (wt%) Ethanol concentration (wt%)
Figure 6.13: Performance measures (∆Pf lowline , Φhyd , and µr ) used to establish the hydrate
plugging risk as a function of ethanol concentration, predicted by the oil-dominated gas hydrate
formation model for Case 3.
Considering Case 2, which showed an intermediate risk of plugging with hydrates, the gas
hydrate prediction model for oil-dominated systems is used to study the effect of hydrates in
transient shut-in/restart operations and the system performance as a function of THI concentration
in the shut-in/restart procedure. Two valves are included in the model to control the shut-in and
restart operations, one valve located at the wellhead and another located at the platform end.
During the shutdown of production, the valves are closed for 50 hours, allowing the gravitational
segregation of fluids in the system toward a hydrostatic and thermal equilibrium. Figure 6.14
shows the system temperature distribution after 50 hours of shut-in and the corresponding hydrate
is observed over a great portion of the flowline. However, the amount of hydrates formed during
the rearrangement of fluids in the flowline following the shutdown of production is not significant,
as shown by the fluid distribution profile in Figure 6.15, which is similar to the one presented in
Figure 6.7 for the steady-state operation of the well. An accumulation of water toward the low spot
near the wellhead and the migration of the gas toward the topside in the riser can be observed;
some accumulation of unconverted water also takes place at the base of the riser, which is separated
130
30
Temperature (ºC)
20
10
0
Fluids temperature
−10
0 5 10 15
Pipeline Length (km)
Figure 6.14: Temperature distribution along the pipeline length, showing system temperature and
hydrate equilibrium temperature, after 50 hours of shut-in, predicted by the oil-dominated gas
hydrate formation model for Case 2 without ethanol.
1
Total liquids
0.8
Volume Fraction
0.6
0.4 Hydrate
0.2 Gas
Water
0
0 5 10 15
Pipeline Length (km)
Figure 6.15: Fluids distribution in terms of volume fraction along the pipeline length, after 50
hours of shut-in, predicted by the oil-dominated gas hydrate formation model for Case 2 without
ethanol.
131
After 50 hours of shut-in, the production is restarted following the valve opening schedule shown
in Figure 6.16. This valve opening schedule is similar to an optimized valve opening schedule
designed to minimize hydrate formation upon restart by Zerpa et al. (2011) using the same hydrate
prediction model, where production is restarted using a valve opening of 1% for 60 minutes, then
1
Valve opening (fraction)
0.8
0.6
0.4
0.2
0
0 50 100 150
Time (min)
Figure 6.16: Valve opening schedule used during the restart of production.
During the restart of production at different THI concentrations (from 0 to 40 wt% of ethanol),
it was found that an accumulation of hydrates was formed by the initial movement of fluids, and
this accumulation is transported through the flowline by a wave front. Figure 6.17 shows the
evolution of the restart of production in terms of pressure drop along the flowline as a function of
time (left plot) and the hydrate volume fraction at a fixed point in the middle of the flowline as a
function of time (right plot), for Case 2 with 20 wt% ethanol injected before the shut-in and during
the restart. It can be observed that pressure drop increases with time, exhibiting a large peak at
approximately 200 minutes, which coincides with the large peak of hydrate volume fraction. This
peak represents the hydrate accumulation that travels along the flowline transported by the wave
front and is followed by a bank of fluids with no hydrates; then, the system equilibrates, reaching
a stable value. This hydrate accumulation that moves through the flowline upon restart represents
The system performance during the restart procedure is evaluated for different ethanol con-
centrations using, as performance measures, the maximum values reached during the initial stage
of restart of the following variables: pressure drop along the pipeline, hydrate-volume fraction,
132
5 0.25
3 0.15
2 0.1
1 0.05
0 0
0 500 1000 0 500 1000
Time (min) Time (min)
Figure 6.17: Evolution of the production restart in terms of pressure drop along the pipeline (left
plot) and hydrate volume fraction (right plot) at a point in the middle of the pipeline with respect
to time, predicted by the oil-dominated hydrate formation model for Case 2 with 20 wt% ethanol
injected before the shut-in and during the restart.
and hydrate-slurry relative viscosity. Figure 6.18 shows three plots of the performance measures
(max {∆Pf lowline (t)}, max {Φhyd (t)}, and max {µr (t)}) used to establish the hydrate plugging risk
as a function of ethanol concentration. It can be observed that the injection of ethanol reduces the
performance measures. The maximum pressure drop along the pipeline exhibits a minimum value
at 20 wt% ethanol injected, and increases with further increase in ethanol injection, which is related
to the balance between the decrease in hydrate slurry viscosity and the increase in liquid loading by
the inhibitor injection. The maximum hydrate volume fraction in the pipe decreases and remains
nearly constant between 10 and 20 wt% ethanol, then increases, showing a local maximum at 25
wt% ethanol, and decreases with further increase in ethanol concentration. The hydrate slurry
relative viscosity shows a behavior similar to that of the hydrate volume fraction, exhibiting a local
maximum at 25 wt% ethanol. Ethanol injection improves the system performance upon restart,
reducing the hydrate plugging risk. A concentration of 20 wt% ethanol seems to be the optimum
133
Max. Hydrate slurry relative viscosity
3
5.5 0.4 10
0.35
2
5 10
0.3
0.25 1
4.5 10
0.2
0
4 10
0 10 20 30 40 0 10 20 30 40 0 10 20 30 40
Ethanol concentration (wt%) Ethanol concentration (wt%) Ethanol concentration (wt%)
Figure 6.18: Performance measures — max {∆Pf lowline (t)}; max {Φhyd (t)}; and max {µr (t)} —
used to evaluate the hydrate plugging risk as a function of ethanol concentration in transient restart
operations, predicted by the oil-dominated hydrate formation model for Case 2.
The hydrate formation model for water-dominated systems, presented in Section 5.2, was applied
to estimate the hydrate plugging risk in an offshore well. The geometry of the well/flowline/riser
used was presented in Figure 6.1. The fluid properties used are from the Caratinga field (Table 6.1),
with a high produced water cut of 85 vol.%. Two simulation cases are studied, the first case
considers a bare pipe without thermal insulation, and the second case accounts for a thermal
The flowline operation conditions for the uninsulated case are plotted in a pressure versus
temperature diagram, including the hydrate equilibrium curve for this system, to determine if this
subsea flowline enters the hydrate formation region (Figure 6.19). Produced reservoir fluids transfer
heat to the cold environment resulting in the fluids being cooled down, and the system enters in
the hydrate formation region (left of hydrate equilibrium curve), with a significant portion of the
pipeline being within hydrate forming conditions. Then, fluids flow up the riser with a significant
decrease in pressure, until the system is out of the hydrate formation region near the platform.
Figure 6.20 shows the temperature distribution along the length of the pipeline, predicted by
the water-dominated hydrate formation model. The temperature of the produced fluids decreases
below the hydrate equilibrium temperature, and hydrates start to form when the system reaches
134
20
Hydrate
equilibrium Wellhead
Pressure (MPa)
15 curve 590 m from wellhead
Flowline operation
10 conditions
5
Topsides
0
280 300 320
Temperature (K)
Figure 6.19: Diagram of pressure versus temperature showing flowline operation conditions dur-
ing steady-state flow, and hydrate equilibrium curve, predicted by the water-dominated hydrate
formation model for the uninsulated case.
a subcooling of 3.6 K. The temperature remains constant during hydrate formation, and decreases
around 5 km of pipeline length when all free gas is converted into hydrate (see Figure 6.21).
Then, the fluids temperature decreases further and equals the ambient temperature. The hydrate
equilibrium temperature decreases because of the decrease in system pressure in the riser, and the
Figure 6.21 shows the predicted fluid distribution in terms of volume fraction along the pipeline
length. The volume fraction of gas and water decreases while hydrate increases, and all the free
gas is consumed. The maximum amount of hydrates in the pipeline is 35 vol.%, approximately.
The hydrate plugging risk is established by comparing the hydrate transition concentration
(Φtransition , the hydrate volume fraction transitioning to plugging) with the predicted hydrate
volume fraction (see Figure 6.22). For steady-state operations, if the maximum predicted hydrate
volume fraction is lower than Φtransition , the fluids can be produced with the presence of hydrates
in the line without risk of plugging. Figure 6.22 shows the distribution of hydrate volume fraction
along the pipeline with the corresponding Φtransition , calculated using the pipeline conditions. In
this case, the estimated hydrate volume fraction is greater than Φtransition , and the system is
135
340
320
Temperature (K)
Hydrate equilibrium temperature
300
280
240
0 5 10 15
Pipeline Length (km)
Figure 6.20: Temperature distribution along the pipeline length, showing system temperature and
hydrate equilibrium temperature corresponding to the steady-state flow, predicted by the water-
dominated hydrate formation model for the uninsulated case.
0.8
Volume Fraction
0.6
Water
0.4 Hydrate
Gas
0.2 Oil
0
0 5 10 15
Pipeline Length (km)
Figure 6.21: Fluid distribution in terms of volume fraction along the pipeline length, correspond-
ing to the steady-state flow, predicted by the water-dominated hydrate formation model for the
uninsulated case.
136
considered to have a high hydrate plugging risk.
The Φtransition correlation is a function of mixture Reynolds number and liquid loading. Since
the mixture Reynolds number is a function of fluid properties, it is also a function of temperature.
The decrease of Φtransition at the beginning of the flowline is related to the sharp decrease in
temperature observed in Figure 6.20. The continuing decrease in Φtransition is related with the
consumption of the gas phase, which reduces the mixture velocity. Φtransition reaches a constant
value when all the free gas phase is converted into hydrate.
0.8
Volume Fraction
0.6
0.4 Hydrate
0.2 Φtransition
0
0 5 10 15
Pipeline Length (km)
Figure 6.22: Hydrate volume fraction along the pipeline length, showing the calculated Φtransition
as plugging criterion, corresponding to the steady-state flow predicted by the water-dominated
hydrate formation model for the uninsulated case.
In this case, a thermal insulation layer is included on the quasi-horizontal portion of the pipeline,
reducing the overall heat transfer coefficient of the system by 46%. The thermal insulation reduces
the rate of heat transfer from the fluids to the external environment, as observed in Figure 6.23,
where the system temperature reaches the hydrate equilibrium temperature in a longer distance
from the wellhead than in the uninsulated case. Hydrates start to form when the system reaches
a subcooling of 3.6 K, and the temperature remains constant during hydrate formation. Hydrates
start to dissociate in the riser, with the increase in ambient temperature, and the system temper-
Figure 6.24 shows the predicted fluid distribution in terms of volume fraction along the pipeline
length. The volume fraction of gas and water decreases while the hydrate volume fraction increases.
137
340
320
Temperature (K)
Hydrate equilibrium temperature
300
280
Fluids temperature
260
240
0 5 10 15
Pipeline Length (km)
Figure 6.23: Temperature distribution along the pipeline length, showing system temperature and
hydrate equilibrium temperature corresponding to the steady-state flow, predicted by the water-
dominated hydrate formation model for the thermal insulated case.
Not all the free gas is consumed in this case, because of the lower heat transfer rate to the envi-
ronment, reducing the hydrate formation rate. The maximum amount of hydrates in the pipeline
is 24 vol.%, approximately.
0.8
Water
Volume Fraction
0.6
0.4 Gas
Hydrate
0.2
Oil
0
0 5 10 15
Pipeline Length (km)
Figure 6.24: Fluid distribution in terms of volume fraction along the pipeline length, corresponding
to the steady-state flow, predicted by the water-dominated hydrate formation model for the thermal
insulated case.
For the thermal insulated case, the hydrate volume fraction is lower than the calculated hydrate
transition concentration (see Figure 6.25). This case can be considered to have a lower hydrate
138
1
0.8
Volume Fraction
0.6
Φtransition
0.4
0.2
Hydrate
0
0 5 10 15
Pipeline Length (km)
Figure 6.25: Hydrate volume fraction along the pipeline length, showing the calculated Φtransition
as plugging criterion, corresponding to the steady-state flow predicted by the water-dominated
hydrate formation model for the thermal insulated case.
The hydrate formation model for gas-dominated systems was applied to a case with water-
saturated natural gas flowing through a horizontal pipeline of length equal to 5 km and inner
diameter of 5 inches. Figure 6.26 shows the estimated thickness of the hydrate deposit along the
pipeline length for three different times during hydrate formation, with a maximum hydrate deposit
thickness of 12 mm formed over 100 days. The increase in deposit thickness is associated with an
increase in pressure drop across the pipeline, as shown in Figure 6.27. In this case, the simulation
was set with a constant mass flow rate at the inlet and a constant outlet pressure. In order to
maintain the constant mass flow rate, it is required to increase the inlet pressure (see Figure 6.27).
Figure 6.28 shows the deposit thickness along the pipeline length at 100 days for different levels
of water saturation of the gas phase. The hydrate deposit thickness increases with increasing water
6.4 Summary
This chapter presented examples of application of the three versions of CSMHyK developed in
this work for oil, water, and gas-dominated systems. The hydrate formation model for oil-dominated
systems was used to estimate the hydrate plugging risk of an offshore well with geometry and fluid
properties typical of those in the Caratinga field located in the Campos basin, Brazil, allowing
139
14
10
8
60 days
6
4 30 days
0
0 1000 2000 3000 4000 5000
Pipeline length (m)
Figure 6.26: Plot of gas hydrate deposit thickness along the pipeline length at 30, 60 and 100 days,
showing an increase in deposit thickness with time, predicted by the hydrate formation model for
gas-dominated systems.
8.4
100 days
8.2
Pressure (MPa)
60 days
8
30 days
7.8
7.6
7.4
0 1000 2000 3000 4000 5000
Pipeline length (m)
Figure 6.27: Plot of system pressure along pipeline length, showing an increase in inlet pressure
with time required to maintain a constant mass flow rate at the inlet of the pipeline, predicted by
the hydrate formation model for gas-dominated systems.
140
25
15
200 ppm
10
5 150 ppm
0
0 1000 2000 3000 4000 5000
Pipeline length (m)
Figure 6.28: Plot of hydrate deposit thickness along the pipeline length at 100 days, for different
levels of water saturation of the gas phase; 150, 200 and 300 ppm of water in the gas phase, predicted
by the hydrate formation model for gas-dominated systems.
the estimation of the hydrate plugging risk at three different periods of the well life. The hydrate
plugging risk was estimated from the calculation of three performance measures (pressure drop
along the flowline, hydrate volume fraction in pipe, and hydrate slurry relative viscosity) in an
attempt to quantify the plugging risk. The hydrate plugging risk depends on the amount of water
and gas flowing from the well; therefore, the hydrate plugging risk could increase with well life
because different periods in the well life are characterized by different magnitudes of GOR and
water cut.
Additionally, the hydrate formation model for oil-dominated systems was applied to study the
transient shut-in/restart operations. In general, ethanol injection decreased the hydrate plugging
risk in steady-state and transient operations. The application of the oil-dominated hydrate forma-
tion model proved to be useful in determining the optimal ethanol concentration that minimized
The hydrate formation model for water-dominated systems was applied to the same well/flow-
line/riser geometry from the Caratinga field with and without thermal insulation, and with a high
water cut of 85 vol.%. The hydrate plugging risk of the water-dominated cases was estimated using
a correlation to estimate the hydrate transition concentration as function of Reynolds number and
liquid loading. The hydrate transition concentration represents the hydrate concentration at which
141
an irreversible flow pattern transition occurs that eventually leads to the hydrate plug formation.
The hydrate formation model for water-dominated systems allows one to estimate the thermal
insulation needed to reduce the hydrate plugging risk in offshore production facilities.
The model for gas-dominated systems was specially designed for cases with a water-saturated
gas phase flowing through a pipe without free liquid phases. A case with water-saturated natural
gas flowing through a horizontal pipeline was studied. The hydrate deposit thickness increases
with time, which causes an increase in the pressure drop across the pipeline. An increase in water
saturation of the gas phase increases the thickness of the hydrate deposit. These results provide
the basis and motivation for future experimental studies of gas-dominated systems for validation
of this first pass model, which represents the first model to address gas-dominated systems.
142
CHAPTER 7
This dissertation presents a compilation of the latest knowledge in the areas of flow assurance
and gas hydrates. Flow assurance is an applied science that focuses on the design and optimization
of facilities and procedures to maintain non-interrupted flow in transport facilities of the oil and
gas industry, in wells and pipelines to the sales point. Gas hydrates are crystalline compounds (i.e.,
solid phase), that form by combining small gas molecules with liquid water at high pressures and
low temperatures. The rapid formation of gas hydrates is a challenging problem in flow assurance,
particularly in subsea transport facilities of the offshore oil and gas industry, where high pressure
The prediction of hydrate formation conditions using thermodynamic calculations (with ex-
cellent accuracy) is not sufficient to estimate hydrate plugging risk in transport facilities. The
objective of this research was to provide a solution to estimate hydrate plugging risk, by extending
and improving a first-pass hydrate formation model developed at the Center for Hydrate Research
of the Colorado School of Mines, to predict gas hydrate formation, dissociation and transportability
in oil and gas flowlines. The resulting prediction tool may enhance the design process of production
facilities and flow assurance procedures, through the estimation of hydrate plugging risk in specific
scenarios. Through this thesis work, the initial hydrate model has been transformed from a kinetic
model for oil-dominated systems only, to a comprehensive modeling approach which can be applied
The main conclusions that can be drawn from the research conducted for this thesis are:
1. Improved flow assurance predictive models for gas hydrates can be constructed by including
interphase mass transfer, interphase momentum transfer, reaction kinetic heat generation,
and modified phase properties (e.g., effective viscosity, corrected heat transfer coefficients),
2. Fluid-fluid and fluid-solid interfaces and the bulk flow of multiple phases in pipelines are key
143
3. Different hydrate formation mechanisms between oil and water-dominated systems were iden-
tified from experiments in large scale flow loop and other laboratory scale apparatuses (e.g.,
4. The hydrate formation model for oil-dominated systems was improved by the implementation
of a new dynamic cohesion force model. The dynamic cohesion force model is based on particle
capillary attraction due to a liquid bridge between particles and sintering of the liquid bridge
(i.e., liquid bridge converting into hydrate). The model predicts well the hydrate volume
of a mass transfer-based hydrate growth model and hydrate plugging criterion, based on fluid
velocity and liquid holdup. The model predicts the amount of hydrate formed, while the
onset of hydrate plug formation can be predicted using the correlation of hydrate transition
that eventually leads to the hydrate plug formation). The model predictions matched hydrate
formation rates from experimental data from gas-water flow loop experiments, and the model
predicted fairly well the amount of hydrates observed in high water cut (75% and 90% water
cut) experiments with crude oil. The model predicted the experimentally observed hydrate
6. The model for hydrate formation in gas-dominated systems was specially designed for cases
with a water-saturated gas phase flowing through a pipe without free liquid phases. The gas-
dominated model consists of a combined heat and mass transfer model to predict hydrate film
growth on pipe walls. The gas-dominated hydrate formation model has not been compared
against experimental data, and should be considered as a first pass model for hydrate film
7. In oil-dominated systems, the hydrate-plugging risk increases with the amount of water and
gas produced from the well; therefore, the hydrate-plugging risk could increase with well
life, which is characterized by increasing GOR and water cut. Prediction of the amount of
hydrates formed and their transportability through pipelines results in an estimate of hydrate
144
plugging risk.
8. In gas-dominated systems, the hydrate deposit thickness increases with time, which causes
an increase in the pressure drop across the pipeline. An increase in water saturation of the
9. The gas hydrate models developed in this thesis can be used in the design of flow assurance
strategies.
145
CHAPTER 8
In this work, a gas hydrate model originally designed to predict hydrate formation in oil-
dominated systems was improved and extended to water and gas-dominated systems. The models
developed in this work represent a significant advancement for the prediction of hydrate plugging
risk in the pipelines of oil and gas transport facilities, and can be used as a tool to design flow
assurance strategies. However, the accuracy of these models in representing hydrate formation
The hydrate formation model for oil-dominated systems is based on the kinetics for hydrate
structure I, where the hydrate kinetic rate constants were regressed from experimental data of
methane hydrates. The kinetics for structure I may not be applicable in pipeline modeling, where
hydrate structure II is typically expected. Further investigations are needed to determine how the
kinetics for hydrate structure I affect the accuracy of the predictions, and to include the kinetics
for the hydrate structure II formed with heavier components (e.g., propane, iso-butane) which are
framework of OLGA
R
allows defining an additional gas hydrate phase and calculating additional
contributions to the conservation equations. Improved pressure drop predictions could be obtained
for phenomena like bedding or deposition of hydrate particles, which were not considered in this
work, could improve pressure drop predictions. Alternatively, the development of improved hydrate
An attempt to quantify the hydrate plugging risk of a subsea pipeline with an oil-dominated
system was presented in Chapter 6. Three performance measures (pressure drop, hydrate volume
fraction and hydrate slurry relative viscosity) were calculated to determine the hydrate plugging
risk. Depending on the values of these parameters the hydrate plugging risk was classified in low,
146
intermediate, and high plugging risk. The limitation of this approach is that the different risk levels
are specific to a particular system geometry and fluid properties. A more general quantification
parameter for hydrate plugging risk is desirable, which should be independent of the characteristics
of a given system. The quantification parameter could be developed using a dimensionless number
that combines parameters such as pressure drop, hydrate volume fraction, slurry relative viscosity,
and flow velocity, which can be then applied to different scenarios to quantify hydrate plugging
risk.
The gas and oil industry is currently in a transition from using hydrate avoidance methods to risk
management methods. The use of inhibitors that act at the interfaces of hydrate particles, such as
kinetic hydrate inhibitors (KHI) and anti-agglomerant (AA) surfactants, have been proven to reduce
the required concentrations of chemical additives in a more economical and efficient flow assurance
strategy. The effect of AA surfactants may be considered through the dynamic cohesion force
model by a decrease in water-oil interfacial tension. However, this has not been compared against
experimental data. On the other hand, the purpose of KHIs is to delay the hydrate nucleation.
Currently, the stochastic nature of hydrate nucleation, even with the injection of kinetic inhibitors,
has made its implementation into the hydrate formation model difficult. Introducing the capabilities
of modeling the effect of AAs and KHIs on the hydrate formation models could be a subject of
future research.
The current work has put more emphasis on the hydrate formation models for oil, water and
gas-dominated systems, and has overlooked the importance of dissociation models. Hydrate dis-
sociation represents an important phenomenon that is present in every subsea transport facility,
since hydrates will dissociate when flowing in the vertical riser toward the platform. Also, after
the formation of a hydrate plug, the hydrate dissociation mechanism forms an important part of
the remediation strategy. Most of the flow loop experiments should have dissociation data after
the hydrate formation stage. Analysis of the dissociation data could result in the suggestion of
mechanisms that can be translated into mathematical models. The models for hydrate dissociation
can be easily coupled with the hydrate formation models in the same computational program.
The transition from oil to water-dominated systems needs to be studied in more detail. Based
on the flow loop experiments performed in this work, the case with 75% water cut represents the
transition between oil and water-dominated systems for the values of water cut tested. Additional
147
flow loop experiments are needed that include several water cut values between 60 and 90%, which
could help in defining hydrate formation mechanisms during the transition from oil to water-
dominated systems. The transition from oil to water-dominated systems could be correlated with
laboratory scale experiments (e.g., bottle tests, high pressure autoclave tests). The dependence of
Hydrate deposition on the pipe wall was considered in the hydrate formation model for gas-
dominated systems, which considered a gas phase saturated with water flowing in the pipeline (i.e.,
single phase flow). However hydrate deposition on the pipe wall in multiphase flow has not been
considered in this study. In the models developed in this work, it is assumed that hydrate particles
form in the liquid phases and can either agglomerate (oil-dominated systems) or promote a flow
regime transition (water-dominated systems). But deposition of hydrate particles on a pipe wall
surface was not accounted for in the oil and water-dominated models. Hydrate particles deposition
on a pipe wall is possible through adhesion forces between the hydrate particle and pipe surface,
148
REFERENCES CITED
Alboudwarej, H., Huo, Z., Kempton, E.C., 2006. Flow-assurance aspects of subsea systems design
for production of waxy crude oils, in: SPE Annual Technical Conference and Exhibition, Society
of Petroleum Engineers, San Antonio, Texas, USA, 24–27 Sep. Doi: 10.2118/103242-MS.
Aman, Z.M., Brown, E.P., Sloan, E.D., Sum, A.K., Koh, C.A., 2011. Interfacial mechanisms gov-
erning cyclopentane clathrate hydrate adhesion/cohesion. Physical Chemistry Chemical Physics,
13(44):19796–19806. Doi: 10.1039/C1CP21907C.
Aman, Z.M., Dieker, L.E., Aspenes, G., Sum, A.K., Sloan, E.D., Koh, C.A., 2010. Influence of
model oil with surfactants and amphiphilic polymers on cyclopentane hydrate adhesion forces.
Energy & Fuels, 24(10):5441–5445. Doi: 10.1021/ef100762r.
Aman, Z.M., Sloan, E.D., Sum, A.K., Koh, C.A., 2012. Lowering of clathrate hydrate cohesive forces
by surface active carboxylic acids. Energy & Fuels, 26(8):5102–5108. Doi: 10.1021/ef300707u.
http://pubs.acs.org/doi/pdf/10.1021/ef300707u.
Andreussi, P., Bendiksen, K., 1989. An investigation of void fraction in liquid slugs for horizontal
and inclined gas–liquid pipe flow. International Journal of Multiphase Flow, 15(6):937 – 946.
Doi: 10.1016/0301-9322(89)90022-0.
Anklam, M.R., York, J.D., Helmerich, L., Firoozabadi, A., 2008. Effects of antiagglomerants on the
interactions between hydrate particles. AIChE Journal, 54(2):565–574. Doi: 10.1002/aic.11378.
Asheim, H., 1986. Mona, an accurate two-phase well flow model based on phase slippage. SPE
Production Engineering, 1(3):221–230. Doi: 10.2118/12989-PA.
Aspenes, G., Dieker, L.E., Aman, Z.M., Høiland, S., Sum, A.K., Koh, C.A., Sloan, E.D., 2010a.
Adhesion force between cyclopentane hydrates and solid surface materials. Journal of Colloid
and Interface Science, 343(2):529–536. Doi: 10.1016/j.jcis.2009.11.071.
Aspenes, G., Høiland, S., Borgund, A.E., Barth, T., 2010b. Wettability of petroleum pipelines:
Influence of crude oil and pipeline material in relation to hydrate deposition. Energy & Fuels, 24
(1):483–491. Doi: 10.1021/ef900809r.
Asserson, R.B., Hoffmann, A.C., Høiland, S., Asvik, K.M., 2009. Interfacial tension measurement
of freon hydrates by droplet deposition and contact angle measurements. Journal of Petroleum
Science and Engineering, 68(3-4):209–217. Doi: 10.1016/j.petrol.2009.06.012.
Austvik, T., 1992. Hydrate formation and behavior in pipes. Ph.D. thesis. Norwegian University of
Science and Technology. Trondheim, Norway.
149
Austvik, T., Li, X., Gjertsen, L.H., 2000. Hydrate plug properties: Formation and removal of
plugs. Annals of the New York Academy of Sciences, 912(1):294–303. Doi: 10.1111/j.1749-
6632.2000.tb06783.x.
Ayers, R.R., 2003. Thermally managed flowline engineering. Technical Report - DeepStar CTR
5307. DeepStar.
Aziz, K., Govier, G.W., Fogarasi, M., 1972. Pressure drop in wells producing oil and gas. Journal
of Canadian Petroleum Technology, 11(3). Doi: 10.2118/72-03-04.
Bai, Y., Bai, Q., 2005. Subsea systems engineering, in: Subsea Pipelines and Risers - Part III:
Flow Assurance. Elsevier. chapter 17, pp. 263 – 276.
Balasubramaniam, R., Rame, E., Kizito, J., Kassemi, M., 2006. Two Phase Flow Modeling: Sum-
mary of FlowRegimes and Pressure Drop Correlations inReduced and Partial Gravity. Tech-
nical Report NASA/CR2006-214085. NASA - Glenn Research Center. Cleveland, Ohio, USA.
(http://hdl.handle.net/2060/20060008906, accessed: 10/19/2012).
Ballard, A.L., Sloan, E.D., 2004a. The next generation of hydrate prediction: Part III. gibbs energy
minimization formalism. Fluid Phase Equilibria, 218(1):15–31. Doi: 10.1016/j.fluid.2003.08.005.
Ballard, L., Sloan, E.D., 2004b. The next generation of hydrate prediction IV: A compari-
son of available hydrate prediction programs. Fluid Phase Equilibria, 216(2):257–270. Doi:
10.1016/j.fluid.2003.11.004.
Bancroft, W.D., 1912. The theory of emulsification, i. The Journal of Physical Chemistry, 16(3):
177–233. Doi: 10.1021/j150129a001.
Baxendell, P.B., Thomas, R., 1961. The calculation of pressure gradients in high-rate flowing wells.
Journal of Petroleum Technology, 13(10):1023–1028. Doi: 10.2118/2-PA.
Becher, P., 2001. Emulsion stability: simple theories, in: Emulsions: Theory and Practice. Oxford
University Press, New York. chapter 4, Third edition. pp. 119–148.
Beggs, D.H., Brill, J.P., 1973. A study of two-phase flow in inclined pipes. Journal of Petroleum
Technology, 25(5):607–617. Doi: 10.2118/4007-PA.
Bendiksen, K.H., Maines, D., Moe, R., Nuland, S., 1991. The dynamic two-fluid model OLGA:
Theory and application. SPE Production Engineering, 6(2):171–180. Doi: 10.2118/19451-PA.
Biberg, D., 1999. An explicit approximation for the wetted angle in two-phase stratified pipe flow.
Canadian Journal of Chemical Engineering, 77(6):1221–1224. Doi:10.1002/cjce.5450770619.
Bienfait, M., 1992. Roughening and surface melting transitions: consequences on crystal growth.
Surface Science, 272(1-3):1–9. Doi: 10.1016/0039-6028(92)91415-8.
150
Blasius, H., 1908. Grenzschichten in flssigkeiten mit kleiner reibung (boundary layersin fluidswith-
littlefriction). Z. Math. Phys., 56.
Borgund, A.E., Høiland, S., Barth, T., Fotland, P., Askvik, K.M., 2009. Molecular analysis of
petroleum derived compounds that adsorb onto gas hydrate surfaces. Applied Geochemistry, 24
(5):777–786. Doi: 10.1016/j.apgeochem.2009.01.004.
Bourrel, M., Salager, J.L., Schechter, R.S., Wade, W.H., 1980. A correlation for phase behavior of
nonionic surfactants. Journal of Colloid and Interface Science, 75(2):451–461. Doi: 10.1016/0021-
9797(80)90470-1.
Boxall, J., 2009. Hydrate plug formation from less than 50% water content water-in-oil emulsions.
Ph.D. thesis. Colorado School of Mines. Golden, Colorado.
Boxall, J., Davies, S., Koh, C.A., Sloan, E.D., 2009. Predicting when and where hydrate plugs
form in oil-dominated flowlines. SPE Projects, Facilities & Construction, 4(3):80–86. Doi:
10.2118/129538-PA.
Brennen, C.E., 2005. Fundamentals of multiphase flow. Cambridge University Press, New York,
NY.
Brill, J.P., Mukherjee, H., 1999. Multiphase Flow in Wells. SPE Monograph Series, Society of
Petroleum Engineers, Richardson, Texas.
Camargo, R., Palermo, T., 2002. Rheological properties of hydrate suspensions in an asphaltenic
crude oil, in: Mori, Y.H. (Ed.), 4th International Conference on Gas Hydrates, ICGH, Yokohama,
Japan, 19-23 May. pp. 880–885.
Cazarez-Candia, O., Vasquez-Cruz, M.A., 2005. Prediction of pressure, temperature, and velocity
distribution of two-phase flow in oil wells. Journal of Petroleum Science and Engineering, 46(3):
195–208. Doi: 10.1016/j.petrol.2004.11.003.
Chierici, G.L., Ciucci, G.M., Sclocchi, G., 1974. Two-phase vertical flow in oil wells - prediction of
pressure drop. Journal of Petroleum Technology, 26(8):927–938. Doi: 10.2118/4316-PA.
Clausse, M., Peyrelasse, J., Heil, J., Boned, C., Lagourette, B., 1981. Bicontinuous structure zones
in microemulsions. Nature, 293(5834):636–638. Doi: 10.1038/293636a0.
Coelho, R.R., Hovell, I., Rajagopal, K., 2012. Elucidation of the functional sulphur chemical struc-
ture in asphaltenes using first principles and deconvolution of mid-infrared vibrational spectra.
Fuel Processing Technology, 97:85–92. Doi: 10.1016/j.fuproc.2011.12.041.
Colebrook, C.F., 1939. Turbulent flow in pipes, with particular reference to the transition region
between the smooth and rough pipe laws. J. Inst. Civil Engrs, 11(4):133 – 156. Doi: 10.1680/i-
joti.1939.13150.
151
Creek, J.L., Subramanian, S., Estanga, D., 2011. New method for managing hydrates in deepwater
tiebacks, in: 2011 Offshore Technology Conference, Offshore Technology Conference, 2-5 May.
Doi: 10.4043/22017-MS.
Danielson, T.J., 2011. A simple model for hydrodynamic slug flow, in: 2011 Offshore Tech-
nology Conference, Offshore Technology Conference, Houston, Texas, USA, 2-5 May. Doi:
10.4043/21255-MS.
Davies, S., 2009. The role of transport resistances in the formation and remediation of hydrate
plugs. Ph.D. thesis. Colorado School of Mines. Golden, Colorado, USA.
Davis, L., 2011. Offshore tide on the rise, in: E&P Magazine. Hart Energy, Houston, Texas, USA.
(http://www.epmag.com/item/Offshore-Tide-The-Rise 92764, accessed: 09/17/2012).
de Boer, R.B., Leerlooyer, K., Eigner, M.R.P., van Bergen, A.R.D., 1995. Screening of crude oils
for asphalt precipitation: Theory, practice, and the selection of inhibitors. SPE Production &
Operations, 10(1):55–61. Doi: 10.2118/24987-PA.
Dickakian, G., Seay, S., 1988. Asphaltene precipitation primary crude exchanger fouling mechanism.
Oil & Gas Journal, 86(10):47–50.
Dieker, L.E., Aman, Z.M., George, N.C., Sum, A.K., Sloan, E.D., Koh, C.A., 2009. Microme-
chanical adhesion force measurements between hydrate particles in hydrocarbon oils and their
modifications. Energy & Fuels, 23(12):5966–5971. Doi: 10.1021/ef9006615.
Dieker, L.E., Taylor, C.J., Koh, C.A., Sloan, E.D., 2008. Micromechanical adhesion force measure-
ments between cyclopentane hydrate particles, in: 6th International Conference of Gas Hydrates,
ICGH, Vancouver, British Columbia, Canada, 6-10 Jul.
Döppenschmidt, A., Butt, H.J., 2000. Measuring the thickness of the liquid-like layer on ice surfaces
with atomic force microscopy. Langmuir, 16(16):6709–6714. Doi: 10.1021/la990799w.
Dorstewitz, F., Mewes, D., 1994. The influence of heat transfer on the formation of hydrate layers in
pipes. Int. J. Heat and Mass Transfer, 37(14):2131–2137. Doi: 10.1016/0017-9310(94)90314-X.
Drew, T. B., K.E.C., McAdams, W.H., 1932. The friction factor for clean round pipes. Trans.
AIChE, 28(1932):56–72.
Duns, H., Ros, N.C.J., 1963. Vertical flow of gas and liquid mixtures in wells, in: 6th World
Petroleum Congress, World Petroleum Congress, Frankfurt, Germany, 19-26 Jun.
Durbut, P., 1999. Surface activity. Marcel Dekker, New York. chapter 3. Surfactant Science, pp.
47–97. Doi: 10.1201/b10985-4.
Englezos, P., Kalogerakis, N., Dholabhai, P.D., Bishnoi, P.R., 1987. Kinetics of gas hydrate for-
mation from mixtures of methane and ethane. Chemical Engineering Science, 42(11):2659–2666.
Doi: 10.1016/0009-2509(87)87016-1.
152
Esumi, K., 2006. Adsolubilization of organic pollutants, in: Somasundaran, P. (Ed.), Encyclopedia
of Surface and Colloid Science: Absorption - Adsirotuib. Taylor & Francis Group. volume 1, 2nd
edition. pp. 124–127.
Fancher, G.H., Brown, K.E., 1963. Prediction of pressure gradients for multiphase flow in tubing.
SPE Journal, 3(1):59–69. Doi: 10.2118/440-PA.
Fidel-Dufour, A., Gruy, F., Herri, J.M., 2006. Rheology of methane hydrate slurries during their
crystallization in a water in dodecane emulsion under flowing. Chemical Engineering Science, 61
(2):505–515. Doi: 10.1016/j.ces.2005.07.001.
Gao, S., 2009. Hydrate risk management at high watercuts with anti-agglomerant hydrate in-
hibitors. Energy & Fuels, 23(4):2118–2121. Doi: 10.1021/ef8009876.
Gomez, L.E., Shoham, O., Taitel, Y., 2000. Prediction of slug liquid holdup: horizontal to upward
vertical flow. International Journal of Multiphase Flow, 26(3):517–521. Doi: 10.1016/S0301-
9322(99)00025-7.
Greaves, D., Boxall, J., Mulligan, J., Sloan, E.D., Koh, C.A., 2008. Hydrate formation from
high water content-crude oil emulsions. Chemical Engineering Science, 63(18):4570–4579. Doi:
10.1016/j.ces.2008.06.025.
Gregory, G.A., Nicholson, M.K., Aziz, K., 1978. Correlation of the liquid volume fraction in the
slug for horizontal gas-liquid slug flow. International Journal of Multiphase Flow, 4(1):33–39.
Doi: 10.1016/0301-9322(78)90023-X.
Griffin, W.C., 1949. Classification of surface active agents by hlb. J. Soc. Cosmet. Chem., 1(5):
311–326.
Gudmundsson, J.S., 2002. Cold flow hydrate technology, in: Mori, Y.H. (Ed.), 4th International
Conference on Gas Hydrates, ICGH, Yokohama, Japan, 19-23 May. pp. 912–916.
Guo, B., Song, S., Chacko, J., Ghalambor, A., 2005. Chapter 15 - flow assurance, in: Offshore
pipelines. Gulf Professional Publishing, Burlington. chapter 15, pp. 169–214. Doi: 10.1016/B978-
075067847-6/50072-X.
Hagedorn, A.R., Brown, K.E., 1965. Experimental study of pressure gradients occurring during
continuous two-phase flow in small-diameter vertical conduits. Journal of Petroleum Technology,
17(4):475–484. Doi: 10.2118/940-PA.
Hagesaether, L., Lunde, K., Nygaard, F., Eidsmoen, H., 2007. Flow-assurance modeling: Reality
check and aspects of transient operations of gas/condensate pipelines. SPE Projects, Facilities
& Construction, 2(1):1–17. Doi: 10.2118/108834-PA.
Hammami, A., Ratulowski, J., 2007. Precipitation and deposition of asphaltenes in production
systems: A flow assurance overview, in: Mullins, O.C., Sheu, E.Y., Hammami, A., Marshall,
A.G. (Eds.), Asphaltenes, Heavy Oils, and Petroleomics. Springer, New York. chapter 23, pp.
617–660. Doi: 978-0-387-68903-6.
153
Hammerschmidt, E.G., 1934. Formation of gas hydrates in natural gas transmission lines. Industrial
& Engineering Chemistry, 26(8):851–855. Doi: 10.1021/ie50296a010.
Hanratty, T.J., 1991. Separated flow modelling and interfacial transport phenomena. Applied
Scientific Research, 48(3):353–390. Doi: 10.1007/BF02008206.
Hatton, G.J., Kruka, V.R., 2002. Hydrate Blockage Formation - Analysis of Werner Bolley Field
Test Data. Technical Report. DeepStar.
Hesketh, R.P., Fraser Russell, T.W., Etchells, A.W., 1987. Bubble size in horizontal pipelines.
AIChE Journal, 33(4):663–667. Doi: 10.1002/aic.690330414.
Hinze, J.O., 1955. Fundamentals of the hydrodynamic mechanism of splitting in dispersion pro-
cesses. AIChE Journal, 1(3):289–295. Doi: 10.1002/aic.690010303.
Hoekstra, L.L., Vreeker, R., Agterof, W.G.M., 1992. Aggregation of colloidal nickel hydroxycar-
bonate studied by light-scattering. Journal of Colloid and Interface Science, 151(1):17–25. Doi:
10.1016/0021-9797(92)90234-d.
Høiland, S., Askvik, K.M., Fotland, P., Alagic, E., Barth, T., Fadnes, F., 2005. Wettability of freon
hydrates in crude oil/brine emulsions. Journal of Colloid and Interface Science, 287(1):217–225.
Doi: 10.1016/j.jcis.2005.01.080.
Israelachvili, J., 1991. Intermolecular and surface forces. Academic Press Inc., San Diego, CA,
USA. 2nd edition.
Joshi, S., 2012. Experimental investigation and modeling of gas hydrate formation in high water
cut producing oil pipelines. Ph.D. thesis. Colorado School of Mines. Golden, Colorado, USA.
Joshi, S., Rao, I., Zerpa, L.E., Webb, E., Lafond, P., Sloan, E.D., Sum, A.K., Koh, C.A., 2011a.
Understanding hydrate plug formation from high water cut systems with a four inch flowloop,
in: 7th International Conference of Gas Hydrates, ICGH, Edinburgh, Scotland, 17-21 Jula.
Joshi, S., Sloan, E.D., Koh, C.A., Sum, A.K., 2011b. Micromechanical adhesion force measure-
ments between cyclopentane hydrate particles in water, in: 7th International Conference of Gas
Hydrates, ICGH, Edinburgh, Scotland, 17-21 Julb.
Kaczmarski, A.A., Lorimer, S.E., 2001. Emergence of flow assurance as a technical discipline
specific to deepwater: Technical challenges and integration into subsea systems engineering, in:
2001 Offshore Technology Conference, Offshore Technology Conference, Houston, Texas, USA,
30 Apr-3 May. Doi: 10.4043/13123-MS.
Kalogerakis, N., Jamaluddin, A.K.M., Dholabhai, P.D., Bishnoi, P.R., 1993. Effect of surfactants
on hydrate formation kinetics, in: SPE International Symposium on Oilfield Chemistry, Society
of Petroleum Engineers, New Orleans, Louisiana, 2-5 Mar. Doi:10.2118/25188-ms.
154
Kanicky, J.R., Lopez-Montilla, J.C., Pandey, S., Shah, O.D., 2001. Surface chemistry in the
petroleum industry, in: Holmberg, K. (Ed.), Handbook of Applied Surface and Colloid Chem-
istry. John Wiley & Sons. chapter 11, pp. 251–267.
Kitanovski, A., Poredos, A., 2002. Concentration distribution and viscosity of ice-slurry in het-
erogeneous flow. International Journal of Refrigeration, 25(6):827–835. Doi: 10.1016/S0140-
7007(01)00091-3.
Kudsen, J.G., Hottel, H.C., Sarofim, A.F., Wankat, P.C., Knaebel, K.S., 1999. Heat and mass trans-
fer, in: Perry, R.H., Green, D.W., Maloney, J.O. (Eds.), Perry’s Chemical Engineers’ Handbook.
McGraw-Hill, 7th edition. pp. 5–51.
Lachance, J.W., Sloan, E.D., Koh, C.A., 2008. Effect of hydrate formation/dissociation on emulsion
stability using dsc and visual techniques. Chemical Engineering Science, 63(15):3942–3947. Doi:
10.1016/j.ces.2008.04.049.
Lee, J., Shin, C., Lee, Y., 2010. Experimental investigation to improve the storage poten-
tials of gas hydrate under the unstirring condition. Energy & Fuels, 24(2):1129–1134. Doi:
10.1021/ef901020g.
Lee, Y.S., 2008. Reverse micelles, in: Self-assembly and nanotechnology - A force balance approach.
John Wiley & Sons. chapter 4, pp. 95–96.
Leporcher, E.M., Peytavy, J.L., Mollier, Y., Sjöblom, J., Labes-Carrier, C., 1998. Multi-
phase transportation: Hydrate plugging prevention through crude oil natural surfactants, in:
SPE Annual Technical Conference and Exhibition, Society of Petroleum Engineers, 27-30 Sep.
Doi:10.2118/49172-ms.
Levich, V.G., 1962. Physicochemical hydrodynamics. Prentice-Hall, Englewood Cliffs, N.J., USA.
Lingelem, M.N., Majeed, A.I., Stange, E., 1994. Industrial experience in evaluation of hydrate
formation, inhibition, and dissociation in pipeline design and operation. Annals of the New York
Academy of Sciences, 715(1):75–93. Doi: 10.1111/j.1749-6632.1994.tb38825.x.
Lockhart, R.W., Martinelli, R.C., 1949. Proposed correlation of data for isothermal two-phase,
two-component flow in pipes. Chemical Engineering Progress, 45(1):3943.
Lund, A., Larsen, R., 2000. Conversion of water to hydrate particles theory and application, in:
14th Symposium on Thermophysical Properties, Elsevier, Boulder, Colorado, .
Luo, Y.T., Zhu, J.H., Fan, S.S., Chen, G.J., 2007. Study on the kinetics of hydrate formation in a
bubble column. Chemical Engineering Science, 62(4):1000–1009. Doi: 10.1016/j.ces.2006.11.004.
McCain, W.D., 1990. Components of naturally occurring petroleum fluids, in: The properties of
petroleum fluids. PennWell. chapter 1, 2nd edition edition. p. 40.
155
McLean, J.D., Kilpatrick, P.K., 1997. Effects of asphaltene aggregation in model heptane-toluene
mixtures on stability of water-in-oil emulsions. Journal of Colloid and Interface Science, 196(1):
23–34. Doi:10.1006/jcis.1997.5177.
McLean, J.D., Spiecker, P.M., 1998. The role of petroleum asphaltenes in the stabilization of water-
in-oil emulsions, in: Mullins, O.C., Sheu, E.Y. (Eds.), Structure and Dynamics of Asphaltenes.
Plenum, New York. chapter 12, pp. 377–422.
Mills, P., 1985. Non-newtonian behavior of flocculated suspensions. Journal de Physique Lettres,
46(7):301–309. Doi: 10.1051/jphyslet:01985004607030100.
Mitchell, D.L., Speight, J.G., 1973. The solubility of asphaltenes in hydrocarbon solvents. Fuel, 52
(2):149–152. Doi: 10.1016/0016-2361(73)90040-9.
Mokhatab, S., Towler, B., 2009. Wax prevention and remediation in subsea pipelines and flowlines.
World Oil, 230(11).
Mukherjee, H., Brill, J., 1983. Liquid holdup correlations for inclined two-phase flow. Journal of
Petroleum Technology, 35(5):1003–1008. Doi: 10.2118/10923-PA.
Orkiszewski, J., 1967. Predicting two-phase pressure drops in vertical pipe. Journal of Petroleum
Technology, 19(6):829–838. Doi: 10.2118/1546-PA.
Poettman, F.H., Carpenter, P.G., 1952. The multiphase flow of gas, oil, and water through vertical
flow strings with application to the design of gas-lift installations, in: Drilling and Production
Practice, American Petroleum Institute, Wichita, Kansas, USA, Mar.
Preston, W.C., 1948. Some correlating principles of detergent action. Journal of Physical Chemistry,
52(1):84–97. Doi: 10.1021/j150457a010.
Rabinovich, Y.I., Esayanur, M.S., Moudgil, B.M., 2005. Capillary forces between two spheres
with a fixed volume liquid bridge: theory and experiment. Langmuir, 21(24):10992–10997. Doi:
10.1021/la0517639.
Rao, I., Sloan, E.D., Koh, C.A., Sum, A.K., 2011. Laboratory experiments and modeling for hydrate
formation and deposition from water saturated gas systems, in: 7th International Conference of
Gas Hydrates, ICGH, Edinburgh, Scotland, 17-21 Jul.
Ronalds, B.F., 2005. Applicability ranges for offshore oil and gas production facilities. Marine
Structures, 18(3):251–263. Doi: 10.1016/j.marstruc.2005.06.001.
Ruf, A., Worlitschek, J., Mazzotti, M., 2000. Modeling and experimental analysis of PSD mea-
surements through FBRM. Particle & Particle Systems Characterization, 17(4):167–179. Doi:
10.1002/1521-4117(200012)17:4<167::AID-PPSC167>3.0.CO;2-T.
156
Salager, J.L., 1996. Guidelines for the formulation, composition and stirring to attain desiredemul-
sion properties, in: Chattopadhyay, A.K., Mittal, K.L. (Eds.), Surfactants in solution. Marcel
Dekker, New York. Surfactant Science. chapter 16, pp. 261–295.
Salager, J.L., 1999. Microemulsions, in: Broze, G. (Ed.), Handbook of detergents - Part A: Prop-
erties. Marcel Dekker, New York. Surfactant Science. chapter 8, pp. 253–302.
Salager, J.L., 2000a. Emulsion properties and related know-how to attain them, in: Nielloud, F.,
Marti-Mestres, G. (Eds.), Pharmaceutical emulsions and suspensions. Marcel Dekker, New York.
Drugs and Pharmaceutical Science. chapter 3, pp. 73–125.
Salager, J.L., 2000b. Formulation concepts for the emulsion maker, in: Nielloud, F., Marti-Mestres,
G. (Eds.), Pharmaceutical emulsions and suspensions. Marcel Dekker, New York. Drugs and
Pharmaceutical Science. chapter 2, pp. 19–72.
Salager, J.L., Anton, R., Sabatini, D., Harwell, J., Acosta, E., Tolosa, L., 2005. Enhancing sol-
ubilization in microemulsionsstate of the art and current trends. Journal of Surfactants and
Detergents, 8(1):3–21. Doi: 10.1007/s11743-005-0328-4.
Salager, J.L., Loaiza-Maldonado, I., Minana-Perez, M., Silva, F., 1982. Surfactant-oil-water sys-
tems near the affinity inversion part i: Relationship between equilibrium phase behavior and
emulsion type and stability. Journal of Dispersion Science and Technology, 3(3):279 – 292. Doi:
10.1080/01932698208943642.
Salager, J.L., Minana-Perez, M., Perez-Sanchez, M., Ramfrez-Gouveia, M., Rojas, C.I., 1983.
Surfactant-oil-water systems near the affinity inversion part III: The two kinds of emul-
sion inversion. Journal of Dispersion Science and Technology, 4(3):313 – 329. Doi:
10.1080/01932698308943373.
Salager, J.L., Morgan, J.C., Schechter, R.S., Wade, W.H., Vasquez, E., 1979. Optimum formulation
of surfactant/water/oil systems for minimum interfacial tension or phase behavior. Society of
Petroleum Engineers Journal, 19(2):107–115. Doi: 10.2118/7054-PA.
Sandrea, I., Sandrea, R., 2007. Global offshore oil: Exploration trends show continued promise in
world’s offshore basins. Oil & Gas Journal, 105(9).
Sarica, C., Panacharoensawad, E., 2012. Review of paraffin deposition research under multiphase
flow conditions. Energy & Fuels, 26(7):3968–3978. Doi: 10.1021/ef300164q.
Schramm, L.L., 2005. Interfacial energetics, in: Emulsions, Foams, and Suspensions: Fundamentals
and Applications. Wiley-VCH, Germany. chapter 3, pp. 53–97.
157
Serbutoviez, S., 2012. Offshore hydrocarbons, in: Panorama. IFP Energies nouvelles,
Rueil-Malmaison Cedex, France. (http://www.ifpenergiesnouvelles.com/publications/notes-de-
synthese-panorama/panorama-2012, accessed: 09/17/2012).
Shah, D.O., 1977. The world of surface science. Chemical Engineering Education, American Sociey
of Engineering Education.
Sjöblom, J., Aske, N., Auflem, I.H., Brandal, Ø., Havre, T.E., Sæther, Ø., Westvik, A., Johnsen,
E.E., Kallevik, H., 2003. Our current understanding of water-in-crude oil emulsions.: Recent
characterization techniques and high pressure performance. Advances in Colloid and Interface
Science, 100-102:399–473. Doi: 10.1016/S0001-8686(02)00066-0.
Sjöblom, J., Hemmingsen, P.V., Kallevik, H., 2007. The role of asphaltenes in stabilizing water-
in-crude oil emulsions, in: Mullins, O.C., Sheu, E.Y., Hammami, A., Marshall, A.G. (Eds.),
Asphaltenes, Heavy Oils, and Petroleomics. Springer, New York. chapter 21, pp. 549–587. Doi:
10.1007/0-387-68903-6 21.
Sjöblom, J., Øvrevoll, B., Jentoft, G., Lesaint, C., Palermo, T., Sinquin, A., Gateau, P., Barré,
L., Subramanian, S., Boxall, J., Davies, S., Dieker, L., Greaves, D., Lachance, J., Rensing,
P., Miller, K., Sloan, E.D., Koh, C.A., 2010. Investigation of the hydrate plugging and non-
plugging properties of oils. Journal of Dispersion Science and Technology, 31(8):1100 – 1119.
Doi: 10.1080/01932690903224698.
Skovborg, P., Rasmussen, P., 1994. A mass transport limited model for the growth of methane
and ethane gas hydrates. Chemical Engineering Science, 49(8):1131–1143. Doi: 10.1016/0009-
2509(94)85085-2.
Sloan, E.D., 2005. A changing hydrate paradigm - from apprehension to avoidance to risk man-
agement. Fluid Phase Equilibria, 228-229:67–74. Doi: 10.1016/j.fluid.2004.08.009.
Sloan, E.D., 2011. Chapter one - introduction, in: Sloan, E.D., Koh, C.A., Sum, A.K. (Eds.),
Natural Gas Hydrates in Flow Assurance. Gulf Professional Publishing, Boston. chapter 1, pp.
1–11.
Sloan, E.D., Koh, C.A., 2008a. Clathrate hydrates of natural gases. CRC Press, Boca Raton, FL.
3rd edition.
Sloan, E.D., Koh, C.A., 2008b. Estimation techniques of phase equilibria of natural gas hydrates,
in: Clathrate hydrates of natural gases. CRC Press, Boca Raton, FL. chapter 4, 3rd edition. pp.
189–256.
Sloan, E.D., Koh, C.A., 2008c. Hydrates in production, processing, and transportation, in:
Clathrate hydrates of natural gases. CRC Press, Boca Raton, FL. chapter 8, 3rd edition. pp.
643–683.
158
Solans, C., Infante, M., Garcia-Celma, M., Salager, J., 1996. Quantifying the concept of physico-
chemical formulation in surfactant-oil-water systems state of the art, in: Trends in Colloid and
Interface Science X. Springer Berlin / Heidelberg. volume 100 of Progress in Colloid and Polymer
Science, pp. 137–142.
Spiecker, P.M., Gawrys, K.L., Trail, C.B., Kilpatrick, P.K., 2003. Effects of petroleum resins on
asphaltene aggregation and water-in-oil emulsion formation. Colloids and Surfaces A: Physico-
chemical and Engineering Aspects, 220(13):9–27. Doi: 10.1016/S0927-7757(03)00079-7.
Sullivan, A.P., Kilpatrick, P.K., 2002. The effects of inorganic solid particles on water and crude
oil emulsion stability. Industrial & Engineering Chemistry Research, 41(14):3389–3404. Doi:
10.1021/ie010927n.
Tajima, H., Kiyono, F., Yamasaki, A., 2009. Direct observation of the effect of sodium dodecyl
sulfate (SDS) on the gas hydrate formation process in a static mixer. Energy & Fuels, 24(1):
432–438. Doi: 10.1021/ef900863y.
Talley, L.D., Turner, D.J., Priedeman, D.K., 2007. Method of generating a non-plugging hydrate
slurry. Patent WO/2007/095399. ExxonMobil Upstream Research Company. USA.
Taylor, C.J., Dieker, L.E., Miller, K.T., Koh, C.A., Sloan, E.D., 2007a. Micromechanical adhesion
force measurements between tetrahydrofuran hydrate particles. Journal of Colloid and Interface
Science, 306(2):255–261. Doi: 10.1016/j.jcis.2006.10.078.
Taylor, C.J., Miller, K.T., Koh, C.A., Sloan, E.D., 2007b. Macroscopic investigation of hydrate film
growth at the hydrocarbon/water interface. Chemical Engineering Science, 62(23):6524–6533.
Doi: 10.1016/j.ces.2007.07.038.
Turner, D., 2005. Clathrate hydrate formation in water-in-oil dispersions. Ph.D. thesis. Colorado
School of Mines. Golden, Colorado, USA.
Turner, D.J., Boxall, J., Yang, S., Kleehammer, D.M., Koh, C.A., Miller, K.T., Sloan, E.D., 2005.
Development of a hydrate kinetic model and its incorporation into the OLGA2000
R
transient
multiphase flow simulator, in: 5th International Conference on Gas Hydrates, ICGH, Trondheim,
Norway, 13-16 Jun. pp. 1231–1240.
van der Waals, J.H., Platteeuw, J.C., 1959. Clathrate solutions, in: Prigogine, I. (Ed.), Advances
in Chemical Physics I. John Wiley & Sons, Inc., pp. 1–57.
Verruto, V.J., Kilpatrick, P.K., 2008. Water-in-model oil emulsions studied by small-angle neutron
scattering: Interfacial film thickness and composition. Langmuir, 24(22):12807–12822. Doi:
10.1021/la802095m. http://pubs.acs.org/doi/pdf/10.1021/la802095m.
Vysniauskas, A., Bishnoi, P.R., 1983. A kinetic study of methane hydrate formation. Chemical
Engineering Science, 38(7):1061–1072. Doi: 10.1016/0009-2509(83)80027-X.
159
Wang, J., Creek, J.L., Buckley, J.S., 2006. Screening for potential asphaltene problems, in: SPE An-
nual Technical Conference and Exhibition, Society of Petroleum Engineers, San Antonio, Texas,
USA, 24-27 Sep. Doi: 10.2118/103137-MS.
Winsor, P.A., 1954. Solvent properties of amphiphilic compounds. Butterworths Scientific Publica-
tions, London.
Wolden, M., Lund, A., Oza, N., Makogon, T., Argo, C., Larsen, T., 2005. Cold flow black oil
slurry transport of suspended hydrate and wax solids, in: 5th International Conference on Gas
Hydrates, ICGH, Trondheim, Norway, 13-16 Jun. pp. 1101–1106.
Yang, S.O., Kleehammer, D.M., Huo, Z., Sloan, E.D., Miller, K.T., 2004. Temperature dependence
of particle-particle adherence forces in ice and clathrate hydrates. Journal of Colloid and Interface
Science, 277(2):335–341. Doi: 10.1016/j.jcis.2004.04.049.
York, J.D., Firoozabadi, A., 2008. Alcohol cosurfactants in hydrate antiagglomeration. Journal of
Physical Chemistry B, 112(34):10455–10465. Doi: 10.1021/jp8017265.
York, J.D., Firoozabadi, A., 2009. Effect of brine on hydrate antiagglomeration. Energy & Fuels,
23(6):2937–2946. Doi: 10.1021/ef800937p.
Zerpa, L.E., Sloan, E.D., Sum, A., Koh, C., 2011. Generation of best practices in flow assurance
using a transient hydrate kinetics model, in: 2011 Offshore Technology Conference, Offshore
Technology Conference, Houston, TX, USA, 2-5 May. Doi: 10.4043/21644-MS.
Zhang, C.S., Fan, S.S., Liang, D.Q., Guo, K.H., 2004. Effect of additives on formation of natural
gas hydrate. Fuel, 83(16):2115–2121. Doi: 10.1016/j.fuel.2004.06.002.
Zhong, Y., Rogers, R.E., 2000. Surfactant effects on gas hydrate formation. Chemical Engineering
Science, 55(19):4175–4187. Doi: 10.1016/S0009-2509(00)00072-5.
Zigrang, D.J., Sylvester, N.D., 1985. A review of explicit friction factor equations. Journal of
Energy Resources Technology, 107(2):280–283. Doi: 10.1115/1.3231190.
160
APPENDIX - DERIVATION OF THE MIXTURE ENERGY CONSERVATION EQUATION
The mixture energy conservation equation presented in Section 3.4.2.2 is based on a homoge-
neous fluid model, where the gas, liquid and hydrate phases are treated as a pseudo-phase. The
∂h ∂h 1 ∂p ∂v ∂v q̇πd
+v = −v − vv − vg sin θ − + SE,hyd (A.3)
∂t ∂x ρ ∂t ∂t ∂x ρAp
where h is mixture enthalpy per unit mass, v is volume average velocity, ρ is volume average density,
q̇ is the wall heat flux, d is pipe diameter, and SE,hyd is the energy source term and is related with
nH + 1 dmgas
SE,hyd = ∆hhyd (A.4)
ρhyd Ap ∆x dt
where, ∆hhyd is the enthalpy change of hydrate formation (for methane hydrate, 4175.4×103 J/kg).
When hydrates are forming, the heat of formation (∆hhyd ) acts as an energy source inside a pipe
section.
The following is a procedure to express the mixture energy conservation equation (eqn. A.3)
explicitly in terms of temperature. From the thermodynamic theory, entropy (s) is a function of
161
For simple compressible systems of fixed chemical composition subject to an internally reversible
dhk υk dpk
dsk = − (A.7)
Tk Tk
From equations A.5 and A.8, the following equation can be obtained,
∂sk Cp,k
= (A.9)
∂Tk pk Tk
From Maxwell’s relationships,
∂υk ∂sk
=− (A.10)
∂Tk pk ∂pk Tk
162
Combining A.15 with A.3, the energy balance equation in terms of temperature is obtained,
∂T ∂T ∂p ∂p 1 ∂p ∂v ∂v q̇πd
Cp +v −β +v − +v +v = −vg sin θ − + SE,hyd (A.16)
∂t ∂x ∂t ∂x ρ ∂t ∂t ∂x ρAp
where,
ρg Hg ηg Cp,g + ρw Hw ηw Cp,w
β= (A.17)
ρ
ρg Hg Cp,g + ρw Hw Cp,w
Cp = (A.18)
ρ
The Joule-Thomson coefficient for the gas phase can be estimated using the following correlation,
18
0.0048823Tpc 2
−1
Tpr
ηg = (A.19)
ppc Cp,g γg
The Joule-Thomson coefficient for the liquid water phase can be estimated from,
1
ηw = − (A.20)
ρw Cp,w
The energy balance equation in terms of temperature can be re-written as,
∂T ∂T 1 ∂p β ∂p ∂p v ∂v ∂v vg sin θ q̇πd SE,hyd
+v − − +v + +v =− − + (A.21)
∂t ∂x ρCp ∂t Cp ∂t ∂x Cp ∂t ∂x Cp Cp ρAp Cp
The energy balance equation in terms of temperature can be discretized applying a first-order
downstream scheme for spatial derivatives and a first-order upstream scheme for time derivatives,
(n+1) (n)
" (n+1) (n+1)
# " (n+1) (n)
#
Ti − Ti Ti − Ti−1 1 pi − pi
+v −
∆t ∆x ρCp ∆t
(n+1) (n+1) (n+1) (n+1)
" (n+1) (n)
# " (n+1) (n)
#
β pi − pi pi − pi−1 v vi − vi vi − vi−1
− +v + +v
Cp ∆t ∆x Cp ∆t ∆x
vg sin θ q̇πd SE,hyd
=− − + (A.22)
Cp Cp ρAp Cp
Re-ordering,
163
1 v (n+1) 1 β 1 v (n+1) v 1 v (n+1)
+ T − + + pi + + v
∆t ∆x i ρCp ∆t Cp ∆t ∆x Cp ∆t ∆x i
(n)
Ti v (n+1) 1 β (n) βv (n+1) v (n) vv (n+1)
= + Ti−1 − + pi − pi−1 + vi + v
∆t ∆x ρCp ∆t Cp ∆t Cp ∆x Cp ∆t Cp ∆x i−1
vg sin θ q̇πd SE,hyd
− − + (A.23)
Cp Cp ρAp Cp
For this model we assume that pressure and velocity profile are known and the only unknown
(n+1)
is Ti , which is solved explicitly and sequentially from node 1 to n.
−1
(n+1) 1 v 1 β 1 v (n+1) v 1 v (n+1)
Ti = + + + pi − − v
∆t ∆x ρCp ∆t Cp ∆t ∆x Cp ∆t ∆x i
(n)
Ti v (n+1) 1 β (n) βv (n+1)
+ + Ti−1 − + pi − p
∆t ∆x ρCp ∆t Cp ∆t Cp ∆x i−1
v (n) vv (n+1) vg sin θ q̇πd SE,hyd
+ v + v − − + (A.24)
Cp ∆t i Cp ∆x i−1 Cp Cp ρAp Cp
164