Escolar Documentos
Profissional Documentos
Cultura Documentos
by
Adam L. Ballard
A thesis submitted to the Faculty and Board of Trustees of the Colorado School of
Mines in partial fulfillment of the requirements for the degree of Doctor of Philosophy
(Chemical and Petroleum-Refining Engineering).
Golden, Colorado
Date_________________
Signed:______________________________
Adam L. Ballard
Approved:______________________________
Dr. E. Dendy Sloan, Jr.
Thesis Advisor
Golden, Colorado
Date_________________
______________________________
Dr. James F. Ely
Professor and Head
Chemical and Petroleum-Refining
Engineering Department
iii
ABSTRACT
The van der Waals and Platteeuw hydrate equation of state, coupled with the classical
thermodynamic equation for hydrates, has been used in the prediction of hydrate
formation for over thirty years. The standard approach in using these models to predict
hydrate equilibrium does not explicitly include them in the flash calculation. Several
limitations of this method have been removed in a new derivation of the model.
In this work, a direct derivation of the standard empty hydrate lattice fugacity has been
given. This allows for description of the hydrate phase itself. The ideal solid solution
assumption is removed by defining a specific volume of the standard hydrate lattice. The
activity of water in the hydrate is a function of the energy difference between the real and
standard lattice, via an activity coefficient. This approach, which allows for distortion of
the hydrate from its standard state, gives a more accurate composition of the hydrate and
significantly improves hydrate formation predictions at high pressure. As a direct result
of accounting for a changing hydrate volume, the cage radii are functions of the hydrate
volume. Direct incorporation of spectroscopic data is crucial for parameter optimization
in the model.
We have included the hydrate phase in a Gibbs energy minimization, multi-phase flash
routine. A total of ten phases (3 fluid, 3 hydrate, and 4 pure solid) are accounted for in
the flash routine. This allows for hydrate phase properties to be calculated at any
temperature and pressure (not just at the formation boundary) and with any other
coexisting phases.
Development of the hydrate prediction program, CSMGem, was also part of this work.
The program was written in Visual Fortran and linked with Visual Basic for ease of use.
A comparison of CSMGem and four commercially available hydrate prediction programs
was made with over 1650 hydrate formation data points and several other types of
hydrate data. CSMGem compares favorably with the other programs. An analysis of all
hydrates of methane, ethane, and propane was performed at seafloor temperatures (277.6
K). Interesting phenomena occur (i.e. structural transitions and pseudo-retrograde
hydrates). The idea of using hydrates to separate close-boiling compounds is also
proposed.
TABLE OF CONTENTS
ABSTRACT
.................................................................................................................v
.......................................................................................................... xxvii
vii
viii
ix
xi
................................................................................................... Enclosed
xii
LIST OF FIGURES
Figure 1.1
Figure 1.2
Figure 1.3
Figure 1.4
Figure 1.5
Figure 1.6
Figure 1.7
Figure 1.8
Figure 2.1
Figure 2.2
Figure 2.3
Figure 2.4
Figure 4.1
Figure 4.2
Figure 4.3
Figure 4.13
Figure 5.1
Figure 5.2
Figure 4.4
Figure 4.5
Figure 4.6
Figure 4.7
Figure 4.8
Figure 4.9
Figure 4.10
Figure 4.11
Figure 4.12
xiii
Figure 5.3
Figure 5.4
Figure 5.5
Figure 5.6
Figure 5.7
Figure 5.8
Figure 5 9
Figure 6.1
Figure 6.2
Figure 6.3
Figure 6.4
Figure 6.5
Figure 6.6
Figure 6.7
Figure 6.8
Figure 6.9
Figure 6.10
Figure 6.11
Figure 6.12
Figure 6.13
Figure 6.14
Figure 6.15
Figure 7.1
Figure 7.2
Figure 7.3
Figure 7.4
Figure 7.5
Figure 7.6
Figure 7.7
Figure 7.8
Figure 7.9
Figure 7.10
Figure 7.11
Figure 7.12
xiv
Figure 7.13
Figure 7.14
Figure 7.15
Figure 7.16
Figure 7.17
Figure 7.18
Figure 7.19
Figure 7.20
Figure 7.21
Figure 7.22
Figure 7.23
Figure 7.24
Figure 7.25
Figure 7.29
Figure 7.30
Figure 7.31
Figure 7.32
Figure 7.33
Figure 7.34
Figure 7.35
Figure 7.36
Figure 7.37
Figure 7.38
Figure 7.39
Figure 7.40
Figure 7.41
Figure 7.42
Figure 7.43
Figure 7.44
Figure 7.45
Figure 8.1
Figure 8.2
Figure 8.3
Figure 7.26
Figure 7.27
Figure 7.28
xv
Figure 8.4
Figure 8.5
Figure 8.6
Figure 8.7
Figure 8.8
Figure 8.9
Figure 8.10
Figure 8.11
Figure 8.12
Figure 8.13
Figure 8.14
Figure 8.15
Figure 8.16
Figure 8.17
Figure 8.18
Figure 8.19
Figure 8.20
Figure 8.21
Figure 8.22
Figure 9.1
Figure 9.2
Figure 9.3
Figure 9.4
Figure 9.5
Figure 9.6
Figure 9.7
Figure 9.8
Figure 9.9
Figure 9.10
Figure 9.11
Figure 9.12
Figure 9.13
Figure 9.14
Figure 9.15
Figure 9.16
Figure 9.17
xvi
Figure 9.18
Figure 9.19
Figure 9.20
Figure 9.21
Figure 9.22
Figure 9.23
Figure 9.24
Figure 9.25
Figure 9.26
Figure 9.27
Figure 10.1
Figure 10.2
Figure 10.3
Figure 10.11
Figure 10.12
Figure 10.13
Figure 10.14
Figure A.1
Figure 10.4
Figure 10.5
Figure 10.6
Figure 10.7
Figure 10.8
Figure 10.9
Figure 10.10
xvii
LIST OF TABLES
Table 3.1
Table 4.1
Types of oxygen atoms at the periphery of both sI and sII hydrate cages
and distance of each to the center of the cages...............................................69
Listing of the major differences between the current and proposed
hydrate fugacity models .................................................................................74
Table 4.2
Table 5.1
Table 5.2
Table 5.3
Table 6.1
Table 6.2
Table 6.3
Table 6.4
Table 6.5
Table 6.6
Table 7.1
Table 7.4
Table 8.1
Table 8.2
Table 8.3
Table 8.4
Table 8.5
Table 8.6
Table 8.7
Table 8.8
Table 8.9
Table 8.10
Table 7.2
Table 7.3
xix
Table 8.11 Constants for Equation 8.23 for ideal gas heat capacity parameters............195
Table 8.12 Constants for Equation 8.23 for formation properties in aqueous phase .....199
Table 8.13 Constants for Equation 8.23 for Born constant ............................................199
Table 8.14 Constants for Equation 8.23 for partial molar volume and heat capacity
parameters ....................................................................................................200
Table 9.1
Table A.1
Table A.2
Table A.3
Table A.4
Table A.5
Table A.6
Table A.7
Table A.8
Table A.9
Table A.10
Table D.1
Table D.2
Table D.3
Table D.4
Table D.5
Table D.6
Table D.7
Table D.8
Table D.9
Table D.10
Table D.11
Table D.12
Table D.13
Table D.14
xx
Table D.15
Table D.16
Table D.17
Table D.18
Table D.19
Table D.20
Table D.21
Table D.22
Table D.23
Table D.24
Table D.25
Table D.26
Table D.27
Table D.28
Table D.29
Table D.30
Table D.31
Table D.32
Table D.33
Table D.34
Table D.35
Table D.36
Table D.37
Table D.38
Table D.39
Table D.40
Table D.41
Table D.42
Table D.43
Table D.44
Table D.45
Table D.46
Table D.47
Table D.48
Table D.49
Table D.50
Table D.51
Table D.52
Table D.53
Table D.54
Table D.55
xxi
Table D.56
Table D.57
Table D.58
Table D.59
Table D.60
Table D.61
Table D.62
Table D.63
Table D.64
Table D.65
Table D.66
Table D.67
Table D.68
Table D.69
Table D.70
Table D.71
Table D.72
Table D.73
Table E.1
Table E.2
Table E.3
Table E.4
Table E.5
Table E.6
Table E.7
Table E.8
Table E.9
Table E.10
Table E.11
Table E.12
Table E.13
Table E.14
Table E.15
Table E.16
Table E.17
Table E.18
Table E.19
Table E.20
xxii
ACKNOWLEDGEMENTS
The completion of this research has involved an enormous amount of help from
many people. I owe particular thanks to:
Dr. E.D. Sloan, Jr. for his wisdom, guidance, enthusiasm, and never-ending
support during many trying times in this research. He has been a mentor to me
not only in the realm of chemical engineering, but also in many other aspects of
life. Thank you for allowing me the freedom to explore hydrate thermodynamics.
the members of my committee: Dr. J.F. Ely, Dr. D.W.M. Marr, Dr. J. Wang, and
Dr. D.M. Wood for their patience, advice, and helpful critique of this work.
the late Dr. S. Selim for showing me the beauty of mathematics and chemical
engineering through his teachings. His absence is felt every day.
Dr. W.R. Parrish for taking the time to discuss hydrate modeling with me. The
many fruitful ideas he brought to my attention were important in this work.
Dr. P.K. Notz for taking the time to help me in understanding the modeling of
heavy oils.
Dr. M.D. Jager for taking the time to review my ideas on many aspects of this
work. Without his generous critiques, this work would not be possible.
Zhongxin Huo, Ramesh Kini, and Dr. S. Subramanian for providing the hydrate
phase measurements needed in the development of the hydrate model.
Phani Bollavaram, Douwe Bruinsma, Erik Freer, Keith Hester, and Douglas
Turner for their friendship and patience.
my Uncle Pete for making sure that I did not spend too much time with research;
there is a life beyond chemical engineering.
xxiii
my wife, Stephanie, for her love. The patience she exhibited during my many
long nights in the office is greatly appreciated.
I would also like to thank all the faculty and students in the Chemical Engineering
department at the Colorado School of Mines for creating a pleasant working atmosphere.
xxiv
xxv
PREFACE
Depending on
the background of the reader, he may choose to skip to more relevant material. A brief
description of each stage will be given.
Theory
This section discusses the development of the Gibbs energy minimization
prediction method. The equations needed for multi-phase thermodynamic equilibrium
and the approach in solving them are given in Chapter 2. Chapter 3 discusses the
fugacity models used in this work. Note that the current hydrate fugacity model is given
in this chapter. Chapter 4 proposes a new derivation of the hydrate fugacity model based
on spectroscopic measurements of the hydrate phase.
A discussion of the fugacity parameter optimization procedure is also given in this
section. Chapter 5 gives the optimization results for all non-hydrate phases as well as the
classical approach to optimizing hydrate parameters. Chapter 6 gives a new approach to
optimizing the hydrate parameters using hydrate phase properties such as composition,
cage occupancy ratios, structural transitions, and volume. The setup for a Gauss-Newton
optimization procedure is developed so that parameters can be regressed simultaneously.
An analysis of the optimized parameters is also given.
This section also considers predictions of hydrate data in efforts to validate the
new hydrate model. All available experimental data on hydrates were predicted using the
xxvii
model in this work (CSMGem) and four commercially available hydrate prediction
programs: CSMHYD, DBRHydrate, Multiflash, and PVTsim.
results. Comparisons are made between the program predictions to see the limitations of
each.
Computer Implementation
The development of a Windows based program, to be used by the funding
companies, was another aspect of this work. Chapter 8 discusses the development of the
CSMGem program and details involved.
xxviii
CHAPTER 1 - INTRODUCTION
1.1
Clathrate Hydrates
Clathrate hydrates are non-stoichiometric crystalline compounds that consist of a
hydrogen bonded network of water molecules and enclathrated molecules. Davy (1811)
first observed clathrate hydrates in the chlorine+water system. It wasnt until 1934,
however, that clathrate hydrates were extensively studied. Hammerschmidt (1934) found
that natural gas transport lines could be blocked by the formation of clathrate hydrates.
This raised a lot of attention in the oil and gas industry, prompting more research to be
performed on clathrate hydrates of natural gas. With the majority of research being done
on clathrate hydrates of natural gas, clathrate hydrates are typically referred to as natural
gas hydrates.
Natural gas hydrates are formed when natural gas is brought into contact with
water, generally at low temperatures and high pressures. The guest molecules most
common in natural gas systems are hydrocarbons ranging from methane to i-pentane.
These gases make up greater than 98 mole percent of a typical natural gas in United
States pipelines. Therefore, the majority of the experimental work performed in the last
70 years has been for hydrates of hydrocarbons ranging from methane to i-pentane.
There are three basic hydrate structures known to form from natural gases:
structure I (sI), structure II (sII), and structure H (sH). The type of hydrate that forms
depends on the size of the gas molecules included in the hydrate. In general, small
molecules such as methane or ethane form sI hydrates as single guests, larger molecules
such as propane and i-butane form sII hydrates, and even larger molecules such as ipentane form sH hydrates in the presence of a small help molecule such as methane.
When sI, sII, and sH hydrate formers are in a mixture, it is not easy to generalize which
hydrate structure will be present. However, the type of hydrate that forms will depend on
the composition, temperature, and pressure of the system.
The basic "building block" of each of these structures is the pentagonal
dodecahedron, which is a 12-sided pentagonal faced polyhedral (512). There are twenty
water molecules in this cage with the oxygen atoms at each vertex and the hydrogen
atoms either chemically or hydrogen bonded between each oxygen atom. The bonds
between the hydrogen and oxygen molecules essentially hold the cage together, and the
guest molecules serve to keep it from collapsing. Depending on what gases are present
and ultimately which hydrate structure is formed; these basic cages stack to form more
complex cages. For sI hydrates they form tetradecahedron cages that have 12 pentagonal
and 2 hexagonal faces (51262). For sII hydrates, hexadecahedron cages are formed; 12
pentagonal and 4 hexagonal faces (51264).
which are, using the previous nomenclature for a cage, 435663 and 51268. Figure 1.1
provides a visual description of each hydrate cage.
Structure I
512 62
2
16
512
46 H2O
136 H 2O
512 64
Structure II
1
34 H2O
4 35 66 3
512 68
Structure H
For example, one unit cell of sI hydrate contains 2 512 cages and 6 51262 cages. Notice
that the relative amount of small (512) cages varies within each hydrate structure. For
instance, sI hydrates are comprised of one-quarter 512 cages, sII hydrates are comprised
of two-thirds, and sH hydrates are comprised of one-half. Differences between the
hydrate structures such as this play a major role in stability considerations.
1.2
hydrates in permafrost and deep ocean sediments has sparked a push to look at hydrates
in natural environments (natural hydrates). This section discusses a few of the current
areas of research in hydrates.
1.2.1 Hydrates in Flow Assurance
Since Hammerschmidt found that natural gas transport lines could be blocked by
the formation of hydrates, nearly all research was focused on natural gas hydrates.
Typical thermodynamic experiments consisted of determining the hydrate formation
temperatures and pressures of given gas mixtures in contact with water. Figure 1.3 is a
simple cartoon of the problem: preventing hydrate blockages.
reservoir
gas processing
hydrate blockage
for a pipeline gas, overlaid with the temperature and pressure profile of a typical offshore
pipeline.
Aq-H-V
Pressure
20
15
10
5 miles
pipeline profile
25
Hydrates Stable
30
45
Aq-V
Hydrates Not Stable
Temperature
Figure 1.4 Plot of offshore pipeline profile and hydrate formation curve
The shaded region represents temperature and pressures in which hydrates are
thermodynamically stable. The thick line is the temperature and pressure profile of the
pipeline, with markers representing the number of miles from the reservoir. At around 5
miles from the reservoir, the pipeline is in the hydrate formation conditions, and
continues to be in this region until around 30 miles. Note that the increase in the
temperature of the pipeline from 23 to 30 miles is due to shallow, warm water.
It is difficult to alter the temperatures and pressures in the pipeline. Therefore,
chemicals are typically injected into the pipeline to inhibit hydrate formation. One
additive typically used is an alcohol (i.e. methanol), which alters the thermodynamic
conditions of hydrate formation (requiring higher pressures to form hydrates). However,
in some cases, so much alcohol is needed that it is economically infeasible to use this
approach.
Another additive commonly used to prevent hydrate blockages in offshore
pipelines is low dosage kinetic inhibitors (KIs). These chemicals are designed such that
they prevent hydrates from growing once nucleation has started. This is much different
than thermodynamic inhibition (i.e. alcohol) in that the formation conditions of the
hydrates are not altered. It has been hypothesized that the KIs prevent hydrate crystals
from growing by adsorption with a part of themselves (the pendant group) in the 51264
cages of the sII hydrate and preventing further growth (Lederhos 1996).
1.2.2 Hydrates as an Energy Resource
Recently, hydrates of predominantly methane have been found in sediments
below the seafloor and in permafrost regions (Haq 1999, Max 2000) as shown in the
cartoon of Figure 1.5.
condensed methane
in hydrates
sediments
hydrates in sediments
sediments
Figure 1.5 Cartoon of hydrates in deep ocean sediments and permafrost regions
The density (amount per volume) of methane is much greater in hydrates than in the
vapor phase. Because of this, the amount of methane in hydrates at the sea floor and
permafrost regions is estimated to be orders of magnitude more than of methane in all
known reservoirs in the world (Collett 2000). It has been proposed that this gas be used
as an energy resource (Yamazaki 1997). Japan, a country with little indigenous energy
sources, has setup a national project on this subject and is currently investigating natural
hydrate energy potential (Denney 2002).
1.2.3
to sub-sea installments. If a pipeline is lying over a bed of hydrates, the heat from the hot
pipeline will transfer to the natural hydrates in sediments. If the temperature of the
hydrates is greater than the formation temperature at the hydrostatic pressure, the
hydrates will dissociate; possibly causing slumping to occur (Dillon et al. 2001, Hovland
and Gudmestad 2001).
pipeline
sediments
hydrates
sediments
Figure 1.6 Cartoon of slumping caused by natural gas pipeline
If this occurs, the pipeline will not be able to support its own weight and break, causing
oil and gas to leak into the ocean (as well as the methane gas trapped in the hydrate).
1.3
composition of the hydrate is not constant. The filling of the cavities in the hydrate
depends on the composition of the gas mixture as well as the temperature and pressure of
the system. This phenomenon poses quite a challenge in the modeling of natural gas
hydrates.
A first approximation for modeling hydrate phase equilibria was developed by
Katz (1945). In this approach, the hydrate formation temperatures and pressures for
several gases were correlated to the specific gravity of the gas mixture. Since it was not
yet determined that there are several types of hydrate structures when this method was
developed, it is independent of hydrate structure. This method can only determine the
hydrate formation temperatures and pressures and not the hydrate composition.
Therefore, another method was developed to account for hydrate composition.
A second and more detailed approximation in modeling hydrate phase equilibria
was conceived by Wilcox et al. (1941). They found that, since hydrates act as a solid
solution, distribution coefficients (K-values) could be used to represent the water-free
mole fraction ratio of hydrate guests in the vapor and hydrate phases. With these Kvalues (water-free vapor composition over the water-free hydrate composition), which
were only a function of temperature and pressure, the formation of hydrates could be
determined by the following mole balance equation:
C
1
i =1
yi
=0
Ki
1.1
where yi is the water-free mole fraction of guest i in the vapor and Ki is the water-free Kvalues determined by Wilcox et al. The temperature and pressure at which Equation 1.1
is true is the formation point of hydrates. This method was also developed independent
of hydrate structure. However, the correlations represent natural gas hydrate formation
points fairly well and can give an approximation of hydrate composition.
In the early 1950s, von Stackelberg and co-workers determined, via X-ray
diffraction, that there were two types of hydrates: sI and sII. With knowledge of the
crystal structure of hydrates, a statistical thermodynamic model of hydrate phase
equilibria was conceived by van der Waals and Platteeuw (1959). In their work, an
expression for the chemical potential of water in any hydrate structure was developed
using an approach analogous to Langmuir adsorption.
The van der Waals and Platteeuw model was difficult to use compared to the
earlier two methods. While the gas gravity and K-value methods could be performed
with a hand calculation, the method of van der Waals and Platteeuw could not. This
posed problems for the applicability of the van der Waals and Platteeuw model in
industrial settings. With the advent of computers, however, this method became the
method of choice.
In 1964, Saito et al. (1964) developed expressions for calculating hydrate phase
equilibria using the chemical potential of water in hydrates developed by van der Waals
and Platteeuw. Their approach allowed the hydrate formation pressures and temperatures
to be determined by equating the chemical potential of water in the hydrate with that in
the aqueous or ice phase. This method was not widely used until Parrish and Prausnitz
(1972) developed an iterative scheme, using the equations developed by van der Waals
and Platteeuw and Saito et al., for use on a computer. The Parrish and Prausnitz scheme
does not explicitly incorporate the hydrate phase. In their method, the thermodynamic
equilibrium of the fluid phases is determined and then compared to the hydrate phases.
The temperature or pressure is then found such that the chemical potential of water is
equal in all phases (including the hydrate phase).
This work was landmark in that it allowed, for the first time, the application of
statistical thermodynamics in an industrial setting. The van der Waals and Platteeuw
model coupled with the Parrish and Prausnitz method to solve for hydrate formation have
remained relatively unchanged in the last 30 years.
1.4
formation pressures and temperature of hydrates has its purpose. The main use of it is to
10
prevent hydrate formation in pipelines. For example, if the temperature and pressure
profiles are known for a given pipeline, the regions of possible hydrate formation can be
determined using the hydrate model (i.e. Figure 1.4). With that determined, chemicals
can be injected into the pipelines to prevent hydrate formation. For this reason, there has
been no need in developing another method for hydrate phase equilibria. However,
current hydrate related problems dictate that another approach to hydrate modeling is
needed.
1.4.1 Use of Hydrate Kinetic Inhibitors
Experiments at the Colorado School of Mines have shown that KIs do not work
as effectively on sI hydrates. This is most likely because the 51262 cage of sI is not large
enough to incorporate the head group of the KIs. This has never been a problem in the
pipeline scenario because sII hydrates are typically formed. However, it has recently
been found, via models and experiments, that some natural gases form sI hydrates at high
pressures. Models and experiments also show that, although sII hydrates form at the
incipient conditions, sI hydrates may form at pressures higher than the formation
pressure, coexisting with sII hydrates. In this case, the KIs may inhibit the sII hydrates
from forming, while allowing sI hydrates to grow.
When using KIs for hydrate inhibition in a pipeline, it could be necessary to
determine if sI hydrates form at conditions above formation conditions. The method of
Parrish and Prausnitz is not suitable for this type of calculation.
1.4.2
sediment. Along a vertical plane of ocean floor, acoustic imaging is used to determine
the depths at which these hydrates are present. A typical hydrate accumulation, shown as
a two-way travel time versus distance plot, is shown in Figure 1.7.
11
that the bottom simulating reflector (BSR), the horizontal thick line, represents the
bottom of a hydrate accumulation.
12
1.5
and pressure, the hydrate phase must be explicitly part of a flash algorithm. The Parrish
and Prausnitz method does not allow this. Therefore, another method is needed which
essentially treats the hydrate phase as just another phase in a flash calculation.
Two tools every engineer needs are mass and energy balances. Therefore the
ideal predictive method would be able give the engineer quantities for which the balances
can be calculated. Quantities that the engineer needs, at any temperature and pressure,
are:
stable phases
phase amounts
composition of each phase
An approach for phase equilibria calculations that can determine each of the above
quantities is needed. In this work, we use a Gibbs energy based flash calculation to
calculate each of these quantities at any temperature and pressure.
Multi-phase flashes are commonly performed for vapor-liquid-liquid (V-L-L)
equilibria and have been around for many years (Baker et al. 1982). These types of
flashes are based on the criterion of minimizing the Gibbs energy of a closed system. A
multi-phase flash calculation is just an extension of the common two-phase flash, if care
is taken. In this sense, it should be a simple task to add the hydrate phase into a flash
calculation. Gupta (1990) was the first to accomplish this feat. In his breakthrough
13
work, Gupta developed the equations and solution procedure necessary to calculate
hydrate phase equilibria at conditions other than hydrate formation.
However, the
1.6
its development in 1959. However, current high pressure hydrate formation data suggest
that the model is not valid at high pressures, which was one of the main driving forces for
this work. We have modified the model by getting rid of two major assumptions used in
its development. We introduce a volume-based hydrate activity coefficient that accounts
for non-ideal effects of the hydrate guests. Predictions with the new model at high
pressure are significantly better.
14
15
XRD
NMR
Raman
pure phase
fugacity
models
hydrocarbon
fugacity
model
hydrate phase
measurements
aqueous
fugacity
model
new hydrate
fugacity
model
Gibbs energy
minimization
next generation
hydrate program
(CSMGem)
Windows
packaging
Figure 1.8 Schematic of the development procedure for the CSMGem program
Each box and circle represents a different aspect of the development while the lines
attaching them represent the process of incorporating them to each other. The dashed
lines and boxes represent work performed by other researchers in this laboratory while
the solid lines and boxes represent this work. The diagram is read in a left-to-right, topto-bottom procedure.
The proposed modifications to the hydrate fugacity model stem from direct
measurements of the hydrate phase using x-ray diffraction (XRD), Raman spectroscopy,
and nuclear magnetic resonance (NMR) spectroscopy. The proposed hydrate fugacity
model incorporates a substantial amount of hydrate phase measurement data via
optimization of the parameters. The hydrate, aqueous, hydrocarbon, and pure phase
fugacity models are all linked using a Gibbs energy minimization (GEM) flash algorithm.
The GEM flash algorithm is coupled with an attractive Windows interface, using Visual
Basic, to make the next generation hydrate program (CSMGem).
16
1.7
are separated into two categories: 1. analysis of hydrates phase equilibria predictions and
2. hydrate modeling. The publications resulting from the analysis of hydrates phase
equilibria predictions are:
1. Ballard, A.L., and Sloan, E.D., Jr. (2000) Optimizing thermodynamic parameters
to match methane and ethane structural transition in natural gas hydrate equilibria.
Annals of the New York Academy of Sciences, 912, 702-712,
2. Subramanian, S., Ballard, A.L., Kini, R., Dec, S.F., and Sloan, E.D., Jr. (2000)
Structural transitions in methane+ethane gas hydrates - Part I: Upper transition
point and applications. Chemical Engineering Science, 55, 5763-5771,
3. Ballard, A.L., and Sloan, E.D., Jr. (2000) Structural transitions in methane+ethane
gas hydrates - Part II: Modeling beyond incipient conditions. Chemical
Engineering Science, 55, 5773-5782,
4. Subramanian, S., Ballard, A.L., Kini, R.A., Dec, S.F., and Sloan, E.D., Jr. (2000)
The phase changes in CH4+C2H6 hydrates, and their impact on oil and gas
production. Energy and Environment: Technological Challenges for the Future,
Edited by Y.H. Mori and K. Ohnishi, Springer, 195-203,
17
5. Ballard, A.L., Jager, M.D., Nasrifar, Kh., Mooijer-van den Heuvel, M.M., Peters,
C.J., and Sloan, E.D., Jr. (2001) Pseudo-retrograde hydrate phenomena at low
pressures. Fluid Phase Equilibria, 185, 77-87,
6. Kini, R., Huo, Z., Ballard, A.L., Jager, M.D., Bollavaram, P., Dec, S.F., and
Sloan, E.D., Jr. (2001) Importance of hydrate phase measurements in flow
assurance and energy storage. 80th Annual GPA Conference, San Antonio, Texas,
March 12-14,
7. Ballard, A.L., and Sloan, E.D., Jr. (2001) Hydrate phase diagrams for methane+
ethane+propane mixtures. Chemical Engineering Science, 56, 6883-6895, and
8. Ballard, A.L., and Sloan, E.D., Jr. (2002) Hydrate separation processes for closeboiling compounds. To be presented at 4th International Conference on Gas
Hydrates, Yokohama, Japan, May 19-23, 2002.
The publications resulting from hydrate modeling are:
1. Pratt, R.M., Ballard, A.L., and Sloan, E.D., Jr. (2001) Beware of singularities
when calculating clathrate hydrate cell potentials. AIChE Journal, 47(8), 18971898,
2. Ballard, A.L., and Sloan, E.D., Jr. (2001) The next generation of hydrate
prediction: I. Hydrate standard states and incorporation of spectroscopy. Fluid
Phase Equilibria, in press,
3. Ballard, A.L., and Sloan, E.D., Jr. (2002) The next generation of hydrate
prediction: III. Gibbs energy minimization formalism. To be extracted from
Chapters 3 of this thesis
4. Ballard, A.L., and Sloan, E.D., Jr. (2002) The next generation of hydrate
prediction: IV. A comparison of available hydrate prediction programs. To be
extracted from Chapter 7 of this thesis.
5. Ballard, A.L., and Sloan, E.D., Jr. (2002) The next generation of hydrate
prediction: An overview. To be presented at 4th International Conference on Gas
Hydrates, Yokohama, Japan, May 19-23, 2002, and
18
This thesis is essentially a combination of each of the above publications. Since this is a
thesis, however, more detail is given. The CD-ROM accompanying this thesis contains
.pdf files of each publication, which can be read using Adobe Acrobat Reader.
19
However,
previous methods to calculate the hydrate phase equilibria can only be used at the
formation/dissociation conditions.
particular hydrate, properties at any pressure and temperature is desired. This chapter
discusses the approach taken to perform these types of calculations using the Gibbs
energy of the system as a criterion.
2.1
Introduction
To calculate thermodynamic equilibrium for a closed system, three fundamental
20
2.2
Implementing Conditions 1 to 3
Implementation of conditions 1 to 3 is done following a similar procedure as that
of a two-phase system (Rachford and Rice 1952). For a system with C components and
possible phases, the following mass balance must be satisfied for each component
k =1
x = zi ,
k ik
i = 1,,C
2.1
where k is the normalized molar amount of phase k (i.e. phase fraction) and xik is the
mole fraction of component i in phase k. We can define a reference phase, r, with the
requirement that it must be present at equilibrium. In this case, Equation 2.1 can be
written as
r xir + k xik = zi ,
i = 1,,C.
2.2
k =1
k r
We can group the mole fraction of component i in the reference phase on the left hand
side of Equation 2.2 to obtain
+ xik x = z
k
ir
i
r
xir
k =1
k r
i = 1,,C.
2.3
21
r = 1 k
2.4a
k =1
k r
and
C
x
i =1
ik
=1
k = 1,,.
2.4b
Equation 2.3, with constraints 2.4a and 2.4b, must be satisfied for a closed system in
order for the mass balance to be conserved.
Imposing constraints of equality of temperature, pressure, and fugacity in all
phases on the system, the following must be true for phases that are present at
equilibrium
f ir xirir P
=
=1
fik xik ik P
i = 1,,C
k = 1,,
2.5
where fik and ik are the fugacity and fugacity coefficient of component i in phase k,
respectively. Equation 2.5 can be equivalently rewritten as
K ik =
ir xik
=
ik xir
i = 1,,C
k = 1,,
2.6
where we have defined Kik as the ratio of fugacity coefficients of component i between
phase k and the reference phase, r. Note that for phases not present at thermodynamic
equilibrium, Equation 2.6 will not be satisfied.
It is common practice to replace the mole fraction ratio in Equation 2.3 with Kik.
However, since Equation 2.6 is only valid for phases that are present at equilibrium, this
can only be done if all of the phases k = 1,, are present. This is the approach
commonly taken for two-phase (vapor and liquid hydrocarbon) systems and can also be
used for multi-phase systems (Rachford and Rice 1952, Michelsen 1982a, 1982b). Note
that if the phases at equilibrium are known a priori, this approach is valid. However, our
problem dictates that the phases are not known a priori so this approach cannot be used
22
in our case. Therefore, a relation similar to that of Equation 2.6 must be derived such that
it is valid for all phases (present or not).
2.3
similar to Equation 2.6 such that it is valid for all phases (present or not). As a starting
point, we look at Equation 2.6
K ik =
ir xik
=
ik xir
i = 1,,C
k = 1,,.
2.6
We would like an equation that reduces to Equation 2.6 for phases that are present but not
for phases that are not present. The ratio
fir
fik
= 1 if phase k present
i = 1,,C
k = 1,,
2.7
will help to obtain our goal. First recall that the reference phase is always present.
Therefore, for any other phase that is present, the fugacity of each component in that
phase must be equal to that in the reference phase. Conversely, for any phase that is not
present, the fugacity of each component in that phase is larger than that in the reference
phase.
Multiplying the mole fraction ratio in Equation 2.6 by the fugacity ratio in
Equation 2.7, we get
K ik =
ir xik f ir xik
f
=
=
exp ln ik
ik xir f ik xir
f ir
i = 1,,C
k = 1,,.
2.8
Gupta (1990) showed that the natural log of the quotient of fugacities in Equation 2.8 is
equal for all components in a given phase k, and refers to this value as the stability of
phase k. Therefore, we define the variable, k, as the stability variable and rewrite
Equation 2.8 in terms of the mole fraction ratio as
23
xik
= K ik e k
xir
i = 1,,C
k = 1,,.
2.9
Equation 2.9 is valid for all phases regardless of that phases presence in the system.
Gupta (1990) showed that defining the mole fraction ratio in this manner is equivalent to
minimizing the Gibbs energy of the system conditional to
Sk =
k k
=0
k +k
k = 1,,.
2.10
That is, if k > 0 then phase k is present and k = 0 and likewise, if k = 0 then phase k is
not present and k 0. Note that Equation 2.10 is always satisfied for the reference phase
by definition of k.
Replacing the mole fraction ratio in Equation 2.3 with Equation 2.9, we obtain
+ K ek
k ik
r
k =1
k r
x = z
i
ir
i = 1,,C
2.11
which is valid for all considered phases. Imposing constraint 2.4a on Equation 2.11, the
following expression for the composition of component i in the reference phase must be
true:
xir =
zi
1 + k K ik ek 1
k =1
k r
i = 1,,C.
2.12
Using Equation 2.9 and recognizing that Kir = 1 and r = 0, Equation 2.12 can be
rewritten for component i in any phase k as
xik =
zi K ik e k
1 + j K ij e 1
j =1
jr
i = 1,,C
k = 1,,.
2.13
Summing Equation 2.13 over all components, and imposing constraint 2.4b for all phases
but the reference phase, we get the objective function, Ek, for each phase
24
zi K ik e k
Ek =
i =1
1 + j K ij e 1
j =1
j r
1 = 0
k = 1,, (k r).
2.14
Substituting constraint 2.4b for the reference phase into Equation 2.14, we get
zi K ik e k
Ek =
i =1
1 + j K ij e 1
j =1
j r
xir = 0
k = 1,,
2.15
=0
k = 1,,
2.16
i =1
Ek =
i =1
zi K ik ek 1
1 + j K ij e 1
j =1
j r
2.4
xik
k
i = 1,,C k = 1,,
k = 1,, ( k r )
k = 1,, ( k r )
{ C unknowns}
{ 1 unknowns} .
{ 1 unknowns}
The number of unknowns to solve for using the above approach is certainly of the same
order of magnitude and so does not require much more computation.
In fact, the
difference in the number of unknowns is only -1. To reiterate, the advantage of not
25
having to know a priori the phases present at thermodynamic equilibrium far outweighs
the fact that this procedure requires a little more computational effort.
The solution procedure is split into two parts: 1. minimizing Gibbs energy by
updating phase amounts and stability variables at a given set of K-values and 2. updating
K-values at a given set of phase amounts and stability variables. This approach is typical
in phase equilibria problems such as this.
2.4.1 Minimizing Gibbs Energy at a Given Set of K-values
The Newton procedure is used to minimize the Gibbs energy of the system at a
given set of K-values (via Equations 2.10 and 2.16). This approach is convergent as long
as the fugacity coefficients are not strong functions of composition. This is certainly the
case for the fluid phases such as vapor and liquid hydrocarbon but is not necessarily true
for the aqueous and hydrate phases. With proper care, which will be discussed later, this
approach is convergent for all systems.
To use the Newton procedure, the derivatives of Equations 2.10 and 2.16 are
needed with respect to each unknown variable. The derivatives of Equation 2.10 are
apparent and therefore only the derivatives of Equation 2.16 with respect to each
unknown variable follow:
)(
j
C z K e k 1 K e 1
i
ik
ij
Ek
=
2
j
i =1
1 + K e m 1
m
im
m =1
m
r
k = 1, , ( k r )
j = 1, ,
( j r)
2.17
and
C
zi K ik ek 1 j K ij e
Ek
=
j
i =1
1 + K em 1
m
im
m =1
mr
k = 1, , ( k r )
2
j = 1, ,
( j r).
2.18
26
It is often desired to solve for the temperature or pressure in which a given phase forms.
For this type of calculation, the following derivative of Equation 2.16 with respect to
temperature (T) or pressure (P) is given:
K ij j
k
K
e
e
1
ik
j
y
j =1
C
Ek
K
j r
= xir ik e m
y
y
i =1
1 + K e j 1
j
ij
j =1
j
r
k = 1,, (k r) 2.19
where y represents T or P. Note that the derivative of the K-value with respect to T or P
is needed. It is assumed, for this derivative, that the K-values are constant in composition
and change only with temperature and pressure. Due to the complexity of the equations
of state used, a numerical centered derivative is used in this work. That is,
K ik K ik ( y + ) K ik ( y )
2
y
i = 1,,C
k = 1,, (k r)
2.20
where = 1E-4 and = 1E-7P are used for the temperature and pressure derivatives,
respectively.
The Newton method gives a correction to the phase amounts and stability
variables based on the gradient of the Gibbs energy at the given set of K-values. Due to
the highly non-ideal behavior of the hydrate phases, all corrections are scaled such that
no phase amount or stability variable is changed more than 25% of its original value.
That is, the entire correction vector is scaled so as to keep the proper direction of
convergence.
2.4.2 Updating K-values and Composition at a Given Gibbs Energy
As discussed above, the Gibbs energy is minimized at a given set of K-values and
then the K-values and composition are updated. Equation 2.13 is used to update the
composition of each phase. With the new composition, the fugacity coefficients are
27
calculated to get the new set of K-values. As is the case for minimizing the Gibbs
energy, this approach is convergent as long as the fugacity coefficients are not strong
functions of composition.
However, the compositions of all non-hydrate phases are determined via Equation 2.13.
All composition corrections of a given phase are scaled such that no composition in that
phase is changed by more than 50% of its original value.
2.4.3 General Algorithm
The solution procedures discussed in Sections 2.4.1 and 2.4.2 are implemented
into the algorithm shown in Figure 2.1.
update k and k
input feed
conditions
estimate
xik, Kik
k, k
Ek = 0 ?
no
yes
xik = 0 ?
no
yes
stop
Figure 2.1 Algorithm to solve for thermodynamic equilibrium
28
As can be seen in Figure 2.1, thermodynamic equilibrium is achieved when the Gibbs
energy is at a minimum (Ek = 0) and the difference in the updated mole fractions and
previous mole fractions (xik) for each phase is zero. On a computer, it is difficult for a
variable to become exactly zero. Therefore, in this work, convergence is met when Ek
and xik are less than 1E-6. Appendix B gives a more detailed discussion of the Newton
scheme for several types of calculations. One of the crucial steps in obtaining a solution
is creating a good initial estimate for the unknown variables. The next section will
discuss initialization of the algorithm.
2.5
(Michelsen 1982a). However, these methods are almost solely for fluid phase systems.
They also require thermodynamic calculations of the fugacity to be performed for each
component in its pure state. These methods are not viable for use with the 7 solid phases
included in this work. Therefore, we use a different approach for determining an initial
estimate of the solution.
The key to obtaining a general routine for estimation of the solution is in
developing composition independent K-values, sometimes referred to as ideal K-values.
With a set of ideal K-values, an initial estimate of all unknown variables can be obtained
by using the algorithm shown in Figure 2.2.
29
input feed
conditions
k = 1/
k = 0
update k and k
Ek = 0 ?
no
yes
stop
Figure 2.2 Algorithm to obtain initial estimates of unknown variables using ideal
K-values
The same equations used to update the phase amounts and stability variables in the
algorithm shown in Figure 2.1 are used in this algorithm. This algorithm is started
assuming all phases are present with an equal amount of each, and therefore the stability
variables are all zero. Convergence to the initial estimate of all unknowns is fairly rapid
since the entire process utilizes the Newton method.
Several authors have determined ideal K-values for component distribution
between V-Lhc (DePriester 1953, Hadden and Grayson 1961) and V-Hydrate (Wilcox et
al. 1941) phases. However, in this work, we need to have a composition independent set
of K-values to determine the component distribution between all possible phases: vapor,
liquid hydrocarbon, aqueous, sI hydrate, sII hydrate, sH hydrate, ice, solid NaCl, solid
KCl, and solid CaCl2. In this work, two forms of the ideal K-values have been derived
and are found in Appendix A. Depending on the type of calculation to perform, one of
the two sets is used to determine an initial estimate for the algorithm in Figure 2.1.
30
2.6
expanding a gas or liquid through a valve or turboexpander. This type of calculation can
be performed using the above algorithm with a slight modification. What follows is a
brief discussion of each type of calculation and the modifications required to perform
them.
2.6.1 Valve Expansions
Expansion through a valve is often referred to as a Joule-Thomson expansion. It
is often the case that a gas is expanded so rapidly through a valve that the change in
enthalpy of the gas is negligible and assumed to be zero (h = 0). The inlet pressure (P1)
and temperature (T1) as well as the outlet pressure (P2) are usually known and the
problem lies in calculating the outlet temperature (T2). An illustration of this process can
be seen in Figure 2.3, where T2 is the temperature to be calculated.
feed gas
P1 T1
P2 T2
h1 = k hk
2.21
k =1
31
Eh = k hk h1 = 0
2.22
k =1
and inserting it into the solution matrix. Note the derivatives needed to perform this
calculation are as follows.
Eh
h
= k k ,
T k =1 T
2.23
Eh
= h j , and
j
2.24
Eh
= 0.
j
2.25
A valve expansion can be performed with any number of phases in the calculation as long
as the enthalpy of each phase can be determined. In this work, we use a centered
derivative to calculate the derivative with respect to temperature in Equation 2.23.
2.6.2
Turboexpansions
The procedure for an expansion through a turboexpander is similar to that of a
valve. In this work, it is assumed that the turboexpander is ideal. In the case of an ideal
expander, the change in entropy of the expanded gas is negligible and assumed to be
zero. As with the valve expansion, the inlet pressure (P1) and temperature (T1) as well as
the outlet pressure (P2) are usually known and the problem lies in calculating the outlet
temperature (T2). An illustration of this process can be seen in Figure 2.4 where T2 is the
temperature to be calculated.
32
W
feed gas
P1 T1
P2 T2
= 100%
33
Fugacity Model
Shock and Helgeson equation with Bromley
activity model
SRK equation of state
solid freeze-out model
solid freeze-out model
modified van der Waals and Platteeuw model
Each fugacity model was chosen based on its ability to describe typical natural gas
pipeline systems, which are of most importance to industry.
Fugacity models typically come in two forms: explicit and implicit. The explicit
fugacity model gives a direct expression for the fugacity of a component while the
implicit fugacity model gives an expression for the chemical potential of a component,
which must then be manipulated to obtain the fugacity.
The derivation of fugacity from chemical potential is simple and can be found in
any introductory thermodynamic text. It revolves around finding a suitable reference on
which to base the chemical potential. The fugacity model for each phase must have the
same reference. Since explicit fugacity models almost always use the ideal gas state of
pure component i at 1 bar and the system temperature as the reference, this same
34
reference is used for implicit fugacity models. The fugacity of any component in the
desired phase, k, can be expressed as:
g io
fik = fio exp ik
RT
3.1
where the subscript o refers to the property of component i in the ideal gas state. fio is the
ideal gas fugacity of component i at 1 bar is simply equal to 1 bar. The Gibbs energy of
component i in the ideal gas state, gio, is calculated using ideal gas properties using the
following classical thermodynamic equation
g i0 o T hio
gio
=
dT
RT RT0 T0 RT 2
3.2
where gi0o is the molar Gibbs energy of formation at T0 and P0 and hi0 is the enthalpy at
P0. The enthalpy of ideal gases is of the form
T
3.3
T0
where hi0o is the molar enthalpy of formation at T0 and P0 and cPi is the heat capacity,
which is of the form
cPi = a0 + a1T + a2T 2 + a3T 3 .
3.4
The ideal gas Gibbs energy of formation, enthalpy of formation, and terms in the heat
capacity of each component used in this work were taken from several thermodynamic
texts and converted to the form shown in Equation 3.4. Appendix E lists all constants for
the ideal gas properties.
It is important to note that choosing a given set of fugacity models is arbitrary.
That is, the fugacity models should be chosen based on individual need. For example, if
one desires to only have methane and water in the system, a fairly simple cubic equation
of state and aqueous phase model can be used. A more detailed description of each
fugacity model used in this work follows.
35
3.1
Aqueous Phase
It is common to have a salt-water aqueous phase in a sub-sea natural gas pipeline.
Therefore, the aqueous phase fugacity model must be able to treat the interaction between
salts and water correctly. It is also common procedure to inject an alcohol into the
pipeline to prevent hydrate formation. Therefore, the aqueous phase fugacity model must
also be able to treat the interactions between methanol and water and methanol and salts
correctly.
The Shock and Helgeson Equation (1988) with a coupled Bromley activity model
(1973) was chosen to describe the fugacity of components in the aqueous phase because
of its treatment of salts and alcohols. This model is an implicit fugacity model in that it
gives the chemical potential of each species in the aqueous phase. What follows is a brief
outline of the model. A more detailed description can be found in the thesis of Jager
(2001, pp. 51-63). The aqueous phase model accounts for two types of species in the
aqueous phase: solutes and water. The equations for each are different and need to be
addressed separately.
3.1.1
phase is given by
P
iAq gi0 * T hi*
vi*
dT
dP + ln aiAq
=
+
2
RT RT0 T0 RT
RT
P0
i water
3.5
where the subscripts * and 0 refer to component i in a hypothetical 1 molal solution and
the standard temperature and pressure, respectively.
molar Gibbs energy of formation, enthalpy, and volume of component i in the 1 molal
solution. The partial molar enthalpy is represented in a similar form as Equation 3.3.
Shock and Helgeson developed expressions for the partial molar heat capacity and
volume of solutes in the aqueous phase as
36
cPi* = c1 +
c2
(T )
+ P
c3 ( P P0 ) + c4 ln
+ TX
(T )
+ P0
2T
3.6
and
vi* = v1 +
v2
v 1
+ v3 + 4
2
+P
+ P T P T
3.7
where = 228 K, = 2600 bar, is the dielectric constant of water, is the Born
coefficient, aiAq is the activity of component i in the aqueous phase, and X in Equation 3.6
is described as
2
1 2 ln ln
X =
.
T 2 P T P
3.8
It should be noted that, in this work, the pressure dependent terms in the partial molar
heat capacity are not needed since it is evaluated at P0 (i.e. c3 and c4 are not needed). The
expression for the dielectric constant of water is given by Jager (2001) as
2
3.9
i=0
With this simple expression for the dielectric constant of water, X in Equation 3.8 can
easily be determined.
3.1.2
deviations of the chemical potential are derived from the pure water state.
The
dT + pure dP + ln awAq
2
RT
RT0
RT
RT
T0
P0
3.10
where the subscript Lpure refers to the pure liquid water phase. Since the standard state
for water is pure liquid water, the aqueous phase fugacity model is limited to a water-rich
aqueous phase (xwAq > 0.65). The expression for the enthalpy and heat capacity of pure
37
water is follows the same framework as Equations 3.3 and 3.4. Jager regressed the
following equation for the volume of water as a function of temperature and pressure:
3
3.11
i =0
With the chemical potential defined for the solutes and water, as a function of
temperature and pressure, the activities are then determined to give the compositional
dependence.
3.1.3
Activity of Solutes
The Bromley activity model (1973) is used to represent the activity of all species
in the aqueous phase. It should be noted that both Equation 3.5 and the Bromley activity
model treat salts as separate ionic species. That is, NaCl is treated as two separate
components in the aqueous phase: Na+ and Cl-. Therefore, the fugacity of each ionic
species in the aqueous phase is calculated via Equation 3.5. However, the fugacity of the
salt in the aqueous phase is determined via
f salt = ( f anion )
( f cation )
3.12
where a is the number of anions in the salt and b is the number of cations in the salt.
In this section we discriminate between ions (i), molecular species (m), and water
(w). What follows are the equations needed to determine the activity of all species in the
aqueous phase (except water).
mj
m j*
jAq
j = i and m
3.13
where mj is the molality of species j (moles of species i per kg water), mj* is the molality
at the standard state (1 molal), and jAq is the activity coefficient of species j in the
aqueous phase. Note that this expression is used for all species in the aqueous phase
except water.
38
ln jAq
z j + zk
+
= z 2j LR + mk SR jk
2
k =i
2
z
2 mk P1 jk j 2 mk mn P 2kn
4 I k =i , m n =i , m
k =m
j=i
3.14
where zj is the charge of ion j and I is the ionic strength of the solution, expressed as
I=
1
zk2 mk .
2 k =i
3.15
The long-range interactions (first term on right-hand side of Equation 3.14) are accounted
for using a form of the Debye-Hckel equation:
LR =
ADH I
.
1+ I
3.16
e2
=
0 RT
N A2
2 s
8
3.17
where
e
0
NA
s
=
=
=
=
=
Short-range interactions between ions (second term on RHS of Equation 3.14) are
expressed by
39
SR jk =
(0.13816 + 0.6 B ) z z
jk
3I
1 +
2 z j zk
j k
+ B jk + C jk I + D jk I 2
j,k = i
3.18
where Bij, Cij, and Dij are interactions parameters representing short-range interactions
between ion pairs. Note that Bij, Cij, and Dij are quadratic in temperature and that they are
zero for all ion pairs with the same charge (i.e. zizj > 0).
Interactions between ions and molecular species (third and fourth terms on RHS
of Equation 3.14) are expressed using Pitzer-type relations as:
P1 jk = 0 jk +
1 jk
1 1 + 2 I e2 I
2I
j=m
3.19
j = m.
3.20
and
P 2 jk = 1 jk 1 1 + 2 I + 2 I e2 I
The ion-molecular species interaction parameters, 0jk and 1jk, are linear functions of
temperature and pressure.
Activity coefficient of molecular species in aqueous phase
The activity coefficient for molecular species is expressed using the Pitzer-type
relation given by Equation 3.18:
ln jAq = 2 mk P1 jk
j = m.
3.21
k =i , m
Equations 3.14 through 3.21 represent the activity coefficients of all species in the water
phase, which can be converted to activity using Equation 3.13.
3.1.4 Activity of Water
The activity of water is expressed by explicitly accounting for contributions from
ionic species and molecular species:
40
m m z z
=
m z m z
c
ln awAq
2 2
a c a Ica
2
c c
2
a a
P 2 jk
M w m j mk P1 jk
2I
j =i ,m k =i ,m
+ m j 3.22
j = m
where Mw is the molecular weight of water. The summations in the first term of Equation
3.22 are for cations (c) and anions (a). The total ionic contribution to the water activity,
Ica, is expressed as
2I
Ica = M w
+ DH + Bca
zc za
3.23
DH
1 1+ I
= 2 ADH
1+ I
+ 2 ln 1 + I
3.24
Bca
3I
1 + 3I
ln 1 +
2
z
z
zc z a
0.13816 + 0.6 Bca ) I zc za
(
c a
=
+
2
3
3I
3I
.
2
2 zc za
1 + 2 z z
c a
2 1
2
3
2
3
4
Bca I + Cca I + Dca I
zc za 2
3
4
3.25
Note that all long-range contributions, expressed by Debye-Hckel terms, lump all
contributions into one using the ionic strength. Therefore, all Debye-Hckel terms are
for the ionic solution and not for individual ionic interactions.
3.2
K while the pressure can range from 1 bar to 1000 bar in deep ocean pipelines. Due to
the need for accurate descriptions of hydrocarbon phases in distillation processes,
41
hydrocarbon fugacity model parameters are typically regressed for large temperature
ranges near ambient pressure. For this reason, a cubic equation of state that could that
model well at high pressure was needed. The Soave modification (1972) of the RedlichKwong (1949) cubic equation of state (SRK EoS) was chosen to represent the all
hydrocarbon phase properties in this work.
The SRK EoS is expressed explicitly in pressure as
P=
RT
a
v b v (v + b )
v
T
P
R
a
b
=
=
=
=
=
=
3.26
where,
molar volume,
absolute temperature,
absolute pressure,
universal gas constant,
energy constant, and
volume constant.
The energy constant, representing all attractive forces within the fluid, is expressed as
C
a = xi x j aij
3.27
i =1 j =1
where xi is the mole fraction of component i in the vapor or liquid hydrocarbon. aij,
which represents attractive forces between components i and j, is expressed as
aij = (1 kij ) i ai j a j
3.28
where kij is the interaction parameter between components i and j and ai, representing the
attractive forces between two molecules of component i, is expressed as
ai = 0.42747
R 2Tc2i
Pci
3.29
The i term in Equation 3.28 is a corrective term for the vapor pressure of pure
component i, and is expressed as (Thorwart and Daubert 1993)
42
1 Tri
i = 1 + S1i 1 Tri + S2i
Tri
3.30
where S1i is expressed in terms of the acentric factor as (Graboski and Daubert 1978)
S1i = 0.48508 + 1.55171 i 0.15613 i2
3.31
and S2i is a constant. It should be noted that, for highly polar molecules, S1i is a
regressed constant instead of correlated using Equation 3.31. The volume constant, b, in
Equation 3.26 represents the average hard core volume of the mixture and is expressed
using standard linear mixing rules as
C
b = xibi
3.32
i =1
RTci
Pci
3.33
All critical properties, acentric factors, molecular weights, and interaction parameters for
hydrocarbons are listed in the API Data Book (1997). A thorough check was performed
to determine the accuracy of the SRK EoS for pure, binary, and multi-component
hydrocarbon systems to ensure its applicability in this work. It should be noted that the
SRK EoS was developed for distillation processes in which water is not present.
Therefore, the water-hydrocarbon interaction parameters were still needed. In this work,
we have regressed all water-hydrocarbon interaction parameters to water content data.
The results of the check and regressions are given in Chapter 5.
Development of the fugacity of any component in the vapor and liquid
hydrocarbon phases using a cubic equation of state is explicit in fugacity, which can be
found via the following thermodynamic relationship
P v
1
f i = xi P exp i dP
0 RT P
3.34
43
where the partial molar volume of component i is found from the model. Determining
the partial molar volume and integrating from 0 to P in Equation 3.34, the expression for
fugacity is
2 x j aij
bi
A j =1
bi B
f i = xi P exp
ln 1 +
( z 1) ln ( z B )
b
B
a
b
z
3.35
where,
A=
aP
,
R 2T 2
B=
bP
, and
RT
z=
Pv
.
RT
Note that the SRK EoS is implicitly referenced to the ideal gas state.
3.3
3.36
where the subscript o refers to the property of component i in the ideal gas state. Note
that, since there is only one component in the pure solid phase, all other components are
assumed to have an infinite fugacity in the pure phase. The Gibbs energy of component i
in a solid phase, giS, is given by
T
P
giS gi0 S
h
v
=
iS 2 dT + iS dP
RT RT0 T0 RT
RT
P0
3.37
where gi0 S is the molar Gibbs energy of formation at T0 and P0. The enthalpy of all pure
phases is of the form
T
3.38
44
where hi0 S is the molar enthalpy of formation at T0 and P0 and cPi is the pure solid heat
capacity, which is of the same form as Equation 3.4. The volume of all pure phases is of
the form
2
3
v (T , P ) = v0 exp 1 (T T0 ) + 2 (T T0 ) + 3 (T T0 ) ( P P0 )
3.39
where v0 is the volume at T0 and P0, i are thermal expansion parameters, and is the
compressibility parameter. The Gibbs energy of formation, enthalpy of formation, heat
capacity, and volume parameters for each pure phase are given in Appendix E.
3.4
Hydrate Phases
In 1959, van der Waals and Platteeuw (1959) used statistical thermodynamics to
wH g w
=
+ m ln 1 im
RT
RT
m
i
3.40
where, gw is the Gibbs energy of water in the standard (empty) hydrate lattice at a given
volume and the system temperature and pressure, m is the number of cages of type m
divided by the number of water molecules in the hydrate lattice, and im is the fractional
occupancy of guest i in hydrate cage m. Notice that the fractional occupancy of a given
hydrate cage will always sum to a value less than one. Therefore, the natural log of the
term in parenthisis will always be a number less than zero.
45
summation term in Equation 3.40 lowers the total energy of the hydrate as a result of the
insertion of the guest molecules into the empty hydrate lattice. The fractional occupancy
is described by a Langmuir type expression as
im =
Cim f iH
1 + C jm f jH
3.41
where fiH is the fugacity of component i in the hydrate. Note that, at equilibrium, the
fugacity of component i is equal in all phases, allowing substitution of fiH with the
fugacity of i in any other equilibrium phase. Cim is the Langmuir constant describing the
potential interaction between encaged guest molecule i and the water molecules in
hydrate cage m surrounding it. It is evaluated by assuming a spherically symmetrical
cage and can be described by a spherically symmetrical potential as
4
Cim =
kT
Rm ai
im ( r )
2
kT
r dr
3.42
where Rm is the radius of cage m and im(r) is the potential function. The Kihara
spherical core potential is typically used to calculate the Langmuir constants. McKoy
and Sinanolu (1963) summed all the guest-water binary interactions to yield an overall
cell potential im(r) given by
12 10 ai 11 i6 4 ai 5
+
im (r ) = 2 i zm 11i im
im 5 im +
im
R
r
R
R
r
R
m
m
m
m
3.43a
where,
N
im
1
=
N
N
N
r
ai
r
ai
1
1 +
,
Rm Rm
Rm Rm
3.43b
zm is the coordination number (number of water molecules comprising cavity m), and ai,
i, and i are the hard-core radius, potential well depth, and soft-core radius of guest i in
hydrate cage m, respectively. The potential well depth and soft-core radius are the
unknown parameters that were regressed to data. Note that care should be taken in the
46
computer implementation of Equations 3.42 and 3.43 due to singularities at the limits
(Pratt et al. 2001).
As described in the previous chapter, the fugacity of each component in each
phase is needed to perform thermodynamic equilibrium calculations. Therefore, we still
need to determine the fugacity of each component in the hydrate. The standard approach
to this was to relate the chemical potential of water in the hydrate to that in the aqueous
or ice phase (Parrish and Prausnitz 1972). This was done by considering the energy
change associated with forming a hydrate and an aqueous or ice phase, both from the
empty hydrate lattice. That is,
wH wH g w
=
= m ln 1 im
RT
RT
m
i
3.44
and
wAq / Ice
RT
P
wAq / Ice g w g w0 T hw
vw
dT
+
dP + ln awAq / Ice
2
RT
RT
RT0 T0 RT
RT
P0
3.45
where gw, hw, and vw are the molar Gibbs energy, enthalpy, and volume of pure liquid
water or ice, respectively, and the term refers to the difference of that property between
the pure liquid water or ice and empty hydrate phases.
developed by Saito et al. (1964), and simplified by Holder et al. (1980) and Menten et al.
(1981), utilizes the classical expression for the chemical potential of water in the aqueous
or ice phase and empty hydrate. That is,
P v
wAq / Ice g w0 Lpure / Ice T hwLpure / Ice
wL pure / Ice
=
dT
+
dP + ln awAq / Ice
2
RT
RT0
RT
RT
T0
P0
3.46
and
g w
RT
g w0
RT0
T0
hw
RT
dT +
P0
vw
RT
dP
3.47
are used to describe the energy change in Equation 3.45. This approach takes advantage
of the fact that, at equilibrium, the chemical potential of water is equal in all phases.
47
Using this as a basis, it can easily be shown that Equation 3.44 is equal to Equation 3.45
at equilibrium and that the fugacity of water in the hydrate is
wH wAq / Ice
f wH = f wAq / Ice exp
.
RT
3.48
Since the fugacity of water in the aqueous or ice phase must be known in Equation 3.48,
an aqueous or ice phase must be present to correctly calculate the fugacity of water in the
hydrate. This is a major drawback of this approach in that an aqueous or ice phase is not
present in all hydrate scenarios. Chapter 4 will discuss a new derivation of the fugacity
of water in the hydrate based on Equation 3.1 that eliminates this constraint.
The fugacity of hydrate guests is assumed to be equal to that in the reference
phase during the Gibbs energy minimization procedure, which will be true at equilibrium.
This approach causes the stability of hydrates to be solely a function of the energy or
fugacity of water in the hydrate. For components that are not present in the hydrate, the
fugacity is assumed to be infinite.
As discussed in the previous chapter, Equation 2.13 is not used to update the
composition of hydrates in the Gibbs energy minimization procedure. Depending on the
type of hydrate that forms (sI, sII, sH), the composition of water is constrained due to the
finite number of cages. The lower limit on the composition of water in the hydrates,
assuming that all cages are filled, is 0.852, 0.850, and 0.850 for sI, sII, and sH hydrates,
respectively. The upper limit is not as defined, as it depends on the temperature and
pressure of the system as well as the hydrate guests. However, the composition of water
in the hydrate must be less than 1.0 because, if it were equal to 1.0, the phase would not
be a hydrate, rather some new type of ice phase.
The expression for the composition of species in the hydrate follows from a
simple mass balance is
xiH =
N
+ N
m im
Nw
3.49
jm
48
where the numerator sums up the total amount of species i in each hydrate cage, Nm, and
the denominator is the total number of species in each hydrate cage plus Nw, the number
of water molecules in the hydrate (i.e. everything in the hydrate). Dividing Equation 3.49
by Nw results in the following equation used to determine the successively substituted
composition of the hydrate:
=
1 +
m im
xiH
3.50
jm
where m is the number of hydrate cages of type m per water molecule in the hydrate.
This equation is used in place of Equation 2.13 while determining the thermodynamic
solution.
49
improvements to the model for application at more extreme temperature and pressure
conditions.
50
4.1
model of hydrates. In fact, the model for hydrates is the general form of Langmuir
adsorption with multiple adsorption sites and multiple types of adsorbed particles, as
shown in Figure 4.1.
51
3. the partition function for the motion of a guest in its cage is independent of the
number and types of guests present, and
4. classical statistics are valid.
These assumptions are analogous to those for Langmuir adsorption.
However, for
hydrates, we can have multiple guests and multiple adsorption sites (cages). For hydrates
the canonical partition function, Q, for Nim, the number of guests of type i in cages of type
m, is
Q ( N im ) = e
F
kT
W ( N ) q
im
Nim
im
4.1
where F is the free energy of the empty hydrate lattice, qim is the individual molecular
partition function for a guest of type i in a cavity of type m, and k is the Boltzmann
constant. Note that Equation 4.1 accounts for multiple cages or sites (i.e. product over m)
and multiple types of guests (i.e. product over i).
The degeneracy, W(Nim), is the combinatorial formula for the number of ways to
distribute Nim particles into Nm total, distinguishable cavities of type m,
Nm
" Nim . To
simplify the notation and put Nm in terms of the total water molecules in the hydrate
lattice, Nw, van der Waals and Platteeuw introduce the variable, m, which is the number
of hydrate cavities of type m over the total number of water molecules in the lattice. That
is,
m =
Nm
Nw
Nm = m N w .
or
(m N w )!
N im ! m N w N km !
k
4.2
52
The summation in the denominator accounts for only the available cages. That is, it
accounts for other particles to be in the cages as well. Substitution of Equation 4.2 into
4.1 gives
m N w !
Nim
kT
Q ( N im ) = e
qim
m
i
N ! N N km !
im m w
m N w !
Nim
kT
Q ( N im ) = e
qim .
i
m
N im ! m N w N km !
i
F
4.3
i = e kT
4.4
4.5
m N w !
N im
= e kT
q
( im i )
i
Nim m
N im ! m N w N km !
k
i
F
4.6
F
kT
m N w
1 + qim i
i
m
4.7
Equation 4.7, the expression for the grand canonical partition function, can be used to
obtain the thermodynamic properties of the hydrate. We must first note that, since we
only took into account the guests in the hydrate lattice, Equation 4.7 is the grand
53
canonical partition function for the guests in the hydrate lattice but is the canonical
partition function for water in the hydrate lattice. That is,
Qwater ( N w ) = e
F
kT
m N w
1 + qim i
i
m
= guests
4.8
and
water = Qwater ( N w ) wN w .
4.9
Nw
However, we are only concerned with the grand canonical partition function for the
guests (Equation 4.7). Let's now derive thermodynamic properties from the partition
function.
Since Equation 4.7 is the canonical partition function for water in the hydrate, we
know that the expectation value (from the canonical partition function) for the chemical
potential of water in the hydrate, w, is simply
ln Qwater
ln guests
wH = kT
= kT
.
N w T ,V ,iw
N w T ,V ,iw
4.10
wH = g w kT m ln 1 + qim i .
m
i
4.11
Equation 4.11 is the widely known expression for the chemical potential of water in the
hydrate lattice given by van der Waals and Platteeuw (1959).
Let us now derive the fractional occupancy of the guests in a given hydrate cage.
Note that since we consider the guests, we need expectation values for the grand
canonical partition function (Equation 4.7). The expectation value for Nj (j = 1,,i), the
number of j particles in the hydrate, is
ln
ln
N j = kT
= j
.
j
j
T ,V , i j
T ,V ,i j
Taking this derivative, we get
4.12
54
N j = m N w
m
q jm j
4.13
1 + qim i
i
which is the number of particles of type j in the hydrate. Since this number is just the
sum of particles of type j in each type of hydrate cage, m, the following must be true:
N j = N jm .
4.14
It follows from Equations 4.13 and 4.14 that the number of particles of type j in hydrate
cage m is
N jm = m N w
q jm j
1 + qim i
4.15
jm =
N jm
m N w
q jm j
4.16
1 + qim i
i
wH = g w + kT m ln 1 im .
m
i
4.17
classical thermodynamics
e
i io
kT
f
= i
f io
or
i
kT
=e
io
kT
fi
f io
4.18
where fi is the fugacity of component i and the subscript o indicates a standard state. In
this work, the standard state is the ideal gas state. Then
55
io
kT
pi
kT i (T )
4.19
where, i(T) is the molecular partition function of ideal gas i. Substitution of Equation
4.19 into Equation 4.18, and recalling that the fugacity of an ideal gas is just its partial
pressure, pi, results in
e
i
kT
fi
kT i (T )
4.20
i =
fi
.
kT i (T )
4.21
This is the expression for the absolute activity of guest i in the hydrate. Substituting
Equation 4.21 into the expression for fractional occupancy of guest i in cavity m
(Equation 4.16), we get
qim
fi
kT i (T )
im =
,
q jm
1+
fj
j kT j (T )
4.22
or
im =
Cim fi
1 + C jm f j
4.23
qim (T , V )
.
kT i (T )
4.24
where,
Cim =
Note that Equation 4.23 looks quite similar to a Langmuir expression for adsorption (as it
should). Because of this similarity, we call the variable Cim the Langmuir constant,
which is a function of temperature and volume. Note that the Langmuir constants still
need to be evaluated.
56
qim = i (T )
0 0
im (r )
2
kT
r sin drd d
4.25
where i(T) is the same as that defined in Equation 4.19, im(r) is a function describing
the potential field within the cavity, and Rm is the radius of cavity m. To simplify
Equation 4.25, we make another assumption that
6. the potential energy of a solute molecule when at a distance r from the center
of its cavity is given by the spherically symmetrical potential im(r) proposed
by Lennard-Jones and Devonshire (1937).
Assumption 6 allows the molecular partition function to be expressed in terms of radius
of the hydrate cage only as
qim = 4i (T )
Rm ai
im ( r )
2
kT
r dr .
4.26
Substitution of Equation 4.26 into Equation 4.24 results in the current expression for the
Langmuir constant,
Cim =
4
kT
Rm ai
im ( r )
2
kT
r dr
3.42
57
which was given in Chapter 3. Discussion of the potential model used is also given in
Chapter 3.
4.2
Improvements to the of van der Waals and Platteeuw Model: Hydrates as the
Non-Ideal Solid Solution
As the demand for oil and natural gas becomes greater, there is a need to go to
deeper depths of water to obtain them. Therefore, the pressure and temperature ranges of
thermodynamic models that are the basis of flow assurance strategy need to be extended.
The current hydrate fugacity model described in Chapter 3 applies statistical
thermodynamics and classical thermodynamics.
58
vH (T,P,x)
v (T,P)
.
.
.. ... .
... .... .
Figure 4.2 Current model not allowing for distortion of hydrate due to guests
(v = vH)
Traditionally, the chemical potential of the standard hydrate is assumed to be at a
given volume, independent of the hydrate guests. However, X-ray diffraction (XRD)
measurements of several different hydrates show that the volume of the equilibrium
hydrate is dependent on the hydrate guest(s) (von Stackelberg 1949, 1954, von
Stackelberg and Mller 1954, von Stackelberg and Meinhold 1954, von Stackelberg and
Frhbuss 1954, von Stackelberg and Jahns 1954, Huo 2002). That is, different size and
composition of hydrate guest(s) correspond to different volumes of the hydrate. Figure
4.3 is a plot of sI cubic lattice parameter versus temperature for several sI hydrates.
59
12.02
11.97
11.92
11.87
P = 1 bar
11.82
100
150
200
Temperature (K)
Figure 4.3 sI cubic lattice parameter versus temperature showing different sizes
(volumes) of sI hydrates
If the standard hydrate volume is not the volume of the equilibrium hydrate, there should
be an energy change proportional to the difference in volume (vH = vH-v ). Note that,
in the development of Equation 4.17, vH is assumed to be equal to zero (i.e. all hydrates
of a given structure are at the same volume). However, we can account for the distortion
via an activity coefficient.
Equation 4.17 is considered to be an ideal solid solution model. That is, upon
rearrangement, it can be expressed as
m
wH g w
= RT ln 1 im
i
m
4.27
i gi** = RT ln ai
4.28
60
where gi** is the Gibbs energy of species i in its pure state. With Equations 4.27 and 4.28
in mind, the activity of water in the hydrate can be expressed as
m
awH
= 1 im
i
m
4.29
which is simply a function of the composition of water in the hydrate. Expanding the
product term in Equation 4.29 leads to
S
awH
= 1 iS 1 iL = awS awL
i
i
4.30
where we can see that the activity of water in the hydrate can be expressed in terms of its
activity in each of the cavities. Upon further inspection of Equation 4.30, we can see that
the activity of water in a given hydrate cavity is an effective composition of water in that
cavity. That is, Equation 4.30 gives the decrease in energy of water in the hydrate due to
the inclusion of the guest(s) in the hydrate cavities. Note that, in the derivation of
Equation 4.17, it was assumed that there were no energy changes due to other effects (i.e.
the water and guest(s) in the hydrate act as an ideal solution).
In traditional thermodynamics, the activity of a component in a phase is defined
as the ratio of the fugacity of that component in the phase to the fugacity of that
component in its pure state. Hence, the idea of an empty hydrate lattice. Note, however,
that the Gibbs energy of the component in its pure state is usually known quite well. That
is, the activity is simply a perturbation in the fugacity of the component from its pure
state. For fluid phases, the following expression for the activity of a component is
familiar
aif = xif if
4.31
where xif is the composition and if is the activity coefficient of component i in the fluid
phase. Note that the activity coefficient accounts for non-idealities between component i
and other components in the fluid. We can extend this principle to water in the hydrate
phase.
61
awm
= 1 im wm .
4.32
Substituting Equation 4.32 into Equation 4.29, the activity of water in the hydrate
becomes
m
awH = 1 im wm
i
4.33
awH
= wH 1 im
i
m
4.34
Substitution of Equation 4.34 into
wH = g w + RT m ln 1 im + RT ln wH .
m
i
4.35
Note that Equation 4.35 can be visualized in a similar manner as in Figure 4.2. Figure
4.4 illustrates the processes needed to determine the chemical potential of water in the
hydrate (as given by Equation 4.35).
62
vH (T,P,x)
v (T,P)
v (T,P)
.. ... .
.... ...
. .
.. ..
.. .... .
. .. .
Figure 4.4 Corrected model allowing for distortion of hydrate due to guests
(v vH)
Process (1) in Figure 4.4 is done at constant volume and therefore, the van der Waals and
Platteeuw statistical model can be used. Process (2) in Figure 4.4, the volume change of
the hydrate from its standard state volume, is done at constant composition and is
described by the activity coefficient in Equation 4.35.
An expression must be developed which describes the differences between the
hydrate and the standard state. XRD data show that the volume of the hydrate is a strong
function of the guest(s) present in the hydrate. Therefore, we suggest that the activity
coefficient be a function of the difference in volume between the hydrate and the standard
hydrate, vH. We note, however, that the activity coefficient must have the following
property
wH 1
as
vH 0 .
Implementing the above constraint reduces Equation 4.35 to Equation 4.17 when vH = 0
(i.e. wH = 1).
In the strictest sense, the activity coefficient accounts for the energy change
involved in taking the volume of the standard lattice to the volume of the real hydrate.
That is, it will perturb the Gibbs energy of the standard lattice such as
63
wH = g w + g w + RT m ln 1 im
m
i
4.36
g w0
RT0
T0
hw
RT
vH
dP .
RT
P0
dT +
4.37
In a manner similar to that of Figures 4.2 and 4.4, Equation 4.36 can be seen visually
(Figure 4.5).
vH (T,P)
v (T,P)
vH (T,P,x)
.. ..
.. .... .
. .. .
Figure 4.5 Alternative expression for corrected model allowing for distortion of
hydrate due to guests (v vH)
Process (1) is given by Equation 4.37 and process (2) by the van der Waals and Platteeuw
statistical model, since it is done at constant volume. Since the chemical potential is a
state function, Figures 4.4 and 4.5 are equivalent processes. Inspection of Equations
4.35, 4.36, and 4.37 and assuming that the heat capacity of the hydrate is not affected in
the process, we propose that the activity coefficient of water in the hydrate be expressed
as
ln wH
g w0
hw0 1 1 P vH
=
+
dP
+
RT0
R T T0 P0 RT
4.38
64
In this work, we
arbitrarily define the perturbed Gibbs energy and enthalpy of formation be linear in vH,
g w0 = avH 0
and
hw0 = bvH 0 ,
4.39
satisfying the above constraint on the activity coefficient. Note that the subscript 0 refers
to the volume difference at T0 and P0, the temperature and pressure at which g w0 and
hw0 are evaluated.
It is important to note that the molar volume difference between the standard
hydrate and the real hydrate, vH, is a strong function of composition of the real hydrate
and small function of pressure (as will be shown later). Therefore, the activity coefficient
will be a strong function of composition and a small function of pressure, as is the normal
case for liquids. There will also be a temperature dependence in the activity coefficient
due to the temperature dependent term introduced by the integration of the enthalpy
(Equation 4.38).
4.2.2 Defining a Standard State Hydrate
The best choice for the standard hydrate would be one that is well characterized
and not too different from the real state of the system. If the standard state is well
defined, small perturbations from this standard state can be accounted for correctly. With
this in mind, we turn to the three most common hydrates of sI, sII, and sH: methane,
propane, and methane+neohexane hydrates, respectively. Note that the standard states
for sI, sII, and sH are the empty hydrate lattices and not the actual hydrates. Therefore,
the activity coefficients for sI methane hydrates, sII propane hydrates, and sH
methane+neohexane hydrates will be close to unity.
The Gibbs energy of water in the standard state can be expressed using classical
thermodynamics and is given in Equation 3.47.
g w
RT
g w0
RT0
T0
hw
RT 2
dT +
P0
vw
RT
dP .
3.47
65
Note that g w0 is the molar Gibbs energy of formation of the standard hydrate at the
reference conditions (T0 and P0), hw is the molar enthalpy, and vw is the molar volume.
These three properties are the unknowns that need to be determined in order to specify
the standard state hydrate.
The molar enthalpy of water in the standard hydrate can be expressed as
T
hw = hw0 + cPw dT
4.40
T0
where hw0 is the molar enthalpy of formation of the standard hydrate at the reference
conditions (T0 and P0) and cPw is the heat capacity at P0. The heat capacity of the
standard hydrates for both sI and sII is well approximated by that of ice (Avlonitis
1994a),VV as is that for sH (Mehta 1996). However, the molar enthalpies of formation
are not known and should be regressed to experimental data.
4.2.3 Molar Volume of Hydrates
We define the molar volume of the standard hydrates of sI, sII, and sH as the
molar volumes of methane, propane, and methane+neohexane hydrates, respectively.
The molar volume of these hydrates, and therefore of the standard states, is well
characterized via XRD data (Tse 1990, Huo 2002). The expression for the molar volume
of hydrates was developed in this work, and is given as
#
2
3
vH (T , P, x ) = v0 exp 1 (T T0 ) + 2 (T T0 ) + 3 (T T0 ) ( P P0 )
4.41
where 1, 2, and 3 are functions of the hydrate structure only, is a function of the
hydrate guest and structure, and v0 is the volume of the standard hydrate at T0 and P0.
The compositional dependence of the volume of hydrates is solely in the v0 term. The
compositional dependence was assumed to be a Langmuir type expression that accounts
for a guest molecules repulsive nature with each hydrate cage. The general form of the
equation for v0 is
66
v0 ( x ) = a0* + N m f ( im ) rim
m
i
4.42
where Nm is the number cages of type m in the hydrate, rim is the repulsive constant for
guest molecule i in hydrate cage m, and a0* is denoted as the standard lattice parameter at
T0, P0, and some x0. The function, f(im), is defined to be
f ( im ) =
(1 + m ) im exp D
1 + m im
4.43
(1 + m ) im
1 + m im
4.44
for all other hydrate cages (51262 and 51264). m is the coordination number of cage m
(zm) per water molecule in the hydrate (Nm), im is the fractional occupancy of component
i in cage m, Di is the molecular diameter of component i, and D is the fractional
occupancy average molecular diameter of the guest molecules in the hydrate. Note that
v0 is assumed to be constant for sH hydrates due to lack of experimental data.
Equation 4.41 is of the same form as Equation 3.39 for the volume of pure solids.
The difference, however, is that the compressibility coefficient, , and reference volume,
v0, are solely dependent on hydrate structure and composition of the guest(s) in the
hydrate while the thermal expansion coefficients, 1, 2, and 3 are solely dependent on
the hydrate structure.
Note that Equation 4.41 is used to determine the volume of the real hydrate,
which is dependent on the temperature, pressure, and composition. It is also used to
determine the volume of the standard hydrates; however, in this case, it is only a function
of temperature and pressure. The compositional dependence is removed for the standard
empty hydrates by making v0 a constant. Since the standard hydrates for sI, sII, and sH
are methane, propane, and methane+neohexane hydrates, respectively, v0 is assumed to
be that of methane, propane, and methane+neohexane hydrates.
67
5.5
2
R = 0.8044
5
4.5
3.5
3
0
0.2
0.4
0.6
0.8
1.2
1.4
1.6
68
1000
800
R = 0.9786
600
400
200
0
0
0.2
0.4
0.6
0.8
1.2
1.4
1.6
69
guest
spherical cage
Figure 4.8 Illustration of guest-cage interaction
This approach assumes that all water molecules comprising a hydrate cage are the same
distance from the center. Data by McMullan and Jeffrey (1965) and Mak and McMullan
(1965), shown in Table 4.1, show that the distances of the water molecules from the
center of the cage are not all the same.
Table 4.1 Types of oxygen atoms at the periphery of both sI and sII hydrate cages
and distance of each to the center of the cages
(a) sI ethylene oxide hydrate (a = 12.03 ) (McMullan and Jeffrey 1965)
# waters in cage
average radius
layer type
# waters in layer
radius ()
20
3.908
24
4.326
( k2 )
( k1 )
8
4
4.06
4.645
(i)
8
3.83
(k)
12
3.96
(i)
8
4.47
(c)
4
4.25
(b) sII tetrahydrofuran and hydrogen sulfide hydrate (a = 17.1 ) (Mak and
McMullan 1965)
# waters in cage
average radius
layer type
# waters in layer
radius ()
70
This approach may be ineffective for larger hydrate guests as the guest may fit better in
certain orientations within the cage. That is, a larger guest would more likely be oriented
such that its axis is situated along the largest radius of the cage. Therefore, we propose
another approach that may indirectly account for orientation.
interactions between the guest and water molecules can be better approximated using a
multi-layered cage (Figure 4.9).
multi-layered cage
guest
R1 ai
in ( r )
r 2 dr
exp n
kT
4.45
where the summation is over all layers (n) in cage m. Note that the upper limit of the
integral is evaluated at R1, which is the smallest layer in cage m. Equation 3.43 can still
be used to evaluate the potential for a given layer. In the development of Equation 4.45,
it is assumed that binary interactions between the guest and the water molecules (layers)
are of most importance and that the potential energy is additive between layers.
71
Another crucial change introduced in this work is to make the radii of each layer
functions of temperature, pressure, and composition. Since experimental data show that
the lattice expands and compresses with temperature, pressure and composition, the cages
must also expand or compress. The radii of the layers are assumed to be a linear function
of the cubic hydrate lattice parameter. That is, the radius of a given layer is expressed as
rn = rn0
a
a0
4.46
where a is the cubic lattice parameter at T, P, and x and rn0 and a0 are the layer radius and
cubic lattice parameter given in Table 4.1, respectively. Huo (2001) performed Cerius2
simulations for sI hydrate to ensure that this assumption is reasonable (Figures 4.10 and
4.11).
4
3.95
3.9
3.85
3.8
3.75
layer k
3.7
3.65
layer i
3.6
3.55
3.5
11.1
11.2
11.3
11.4
11.5
11.6
11.7
11.8
11.9
12
Figure 4.10 Predictions of small cage of sI hydrate cage size versus lattice
parameter
12.1
72
4.7
4.6
4.5
4.4
4.3
4.2
layer i
4.1
4
layer c
3.9
3.8
3.7
11.1
layer k1
11.2
11.3
11.4
11.5
11.6
11.7
11.8
11.9
12
12.1
Figure 4.11 Predictions of large cage of sI hydrate cage size versus lattice
parameter
Note that the predictions from Equation 4.46 fall right on the simulation points from
Cerius2. This is most likely because the simulation program makes the same assumption
for the layer size. The only way to definitively confirm this is to perform single crystal
XRD or neutron diffraction experiments for hydrates of different sizes.
4.2.5 New Development of Fugacity
In order to solve for thermodynamic equilibrium, the fugacity of water in the
hydrate must be known. Recall that the current approach to this is to relate the chemical
potential of water in the hydrate to that in the aqueous phase. This approach can be
burdensome; if an aqueous phase is not present, comparing chemical potentials will not
be consistent with thermodynamic equilibrium. Therefore, we suggest the following
approach.
73
RT
4.47
where fwo is 1 bar, gwo is the Gibbs energy of pure water in the ideal gas state at 1 bar, and
wH is given by Equation 4.35. Note that the fugacity of water in the hydrate, as
determined by Equation 4.47, does not require that an aqueous phase be present. We
believe that this approach is much simpler to use and understand since the properties of
the empty hydrate lattice are no longer referenced to pure liquid water. This approach
allows the hydrate phase to be calculated in standard thermodynamic sense. That is,
calculation of the hydrate phase fugacity is no different than any other solid phase, with
the exception of the activity, due to the fact that hydrates are not pure.
4.3
1. accounting for the hydrate volume, 2. indirectly accounting for non-sphericity in the
hydrate cages, and 3. expressing the fugacity of water in the hydrate explicitly via
classical thermodynamics. Table 4.2 lists the major differences between the models.
74
Table 4.2 Listing of the major differences between the current and proposed
hydrate fugacity models
Hydrate Property
volume
formation properties (T0,P0)
cage size
fugacity expression
Current Model
constant
constant
constant
Single shell
difference properties
explicit in aqueous
phase
Proposed Model
f(T,P,x)
f(x)
f(v)
multi-layered shell
actual properties
not explicit in aqueous
phase
75
not widely accepted in the hydrate community due to its empirical nature. We believe
that Holder and co-workers had the correct concept because the Gibbs energy of the
empty hydrate should be a function of the hydrate composition. However, the hydrate
volume at T0 and P0, which we have derived independently of hydrate formation data,
gives that function of composition.
The previous hydrate model has certainly been satisfactory for the last 30 years
for low pressure applications. Otherwise, it would have been scrapped and another
would have been developed. The fact that the same model has been used for the last 30
years is evidence of its applicability. However, the applicability of the model at high
pressure (> 200 bar) has been criticized. Due to the deep-water oil and gas exploration
and production, there is a need for a hydrate model applicable at pressures up to 1000 bar.
Figure 4.12 is a plot of the predicted hydrate formation conditions for methane hydrate.
15%
1000
Standard Volume
Pressure (bar)
5%
100
-5%
-10%
10
274
279
284
289
294
299
304
Temperature (K)
10%
76
The middle solid line is the hydrate prediction for a given, constant hydrate volume while
the two solid lines on either side are the predictions with a 1.5 percent perturbation of
the hydrate volume. The dashed lines show the percent change in pressure for the
perturbed predictions. As can be seen, at low pressure, the pressure prediction is only
changed by approximately 5 percent. However, at higher pressures (> 200 bar) the
percent change in the pressure reaches as high as 15 percent.
To put things in
perspective, data by Huo (2002) show that the volume can change by as much as 3
percent due only to hydrate composition. We believe that, while the change in the
hydrate volume at low pressures may be insignificant, it becomes significant at higher
pressures.
4.3.2 Hydrate Cage Description
A thorough review of the literature on hydrate modeling showed that the radii of
the hydrate cages are assumed to be constant. This goes hand in hand with the fact that
the hydrate volume has always been assumed to be constant. That is, if the volume does
not change, then the cages that make up that volume should not change either. With a
model for the hydrate volume as a function of T, P, and x, it would be inconsistent
thermodynamics to keep the radii of the cages constant. Therefore, this change to the
model is a direct consequence of accounting for the changing hydrate volume.
We also propose changing the description of the hydrate cages to indirectly
account for non-sphericity. It is clear from the data in Table 4.1 that the hydrate cages of
sI and sII are not spherical. However, due to the complexity of the molecular partition
function for a non-spherical cage, we chose to describe this non-sphericity via a
weighted, spherical potential. This approach should allow for preferred orientation of the
guests in the hydrate cages (Figure 4.13).
77
multi-layered cage
large guest
layer 1
layer 2
layer 3
78
therefore altering the parameters that describe its potential (i.e. Kihara parameters).
Evidence of this is the fact that Raman (Subramanian 2000) and NMR (Kini 2002)
spectroscopy can discriminate between given molecules in the vapor and hydrate phases.
Since Raman and NMR spectroscopy give measures of the vibration and rotation of the
molecule, respectively, an unaltered rotational and vibrational property of the molecule
render these tools useless. However, it has been shown that these tools are quite effective
in discriminating between the hydrate and other phases.
parameters determined via second virial coefficients and viscosity data are not relevant
for an encaged molecule.
Second, the addition of the potential energy from the two shells serves only to
scale the Kihara parameters. The shells are far from the center of the hydrate cage and
therefore the energy contribution is a small function of the position of the hydrate guest
within the cage.
potential energy by a constant value for a given hydrate guest, which serves only to adjust
the Kihara parameters, essentially leaving the total energy unchanged. For the above two
reasons, we do not use the concentric shell approach proposed by John and Holder and
Sparks et al.
4.3.3 Expression of Hydrate Fugacity
We believe that the expression for the fugacity of water in the hydrate given in
Equation 4.47 is much simpler to use and understand. The benefits are three-fold:
The aqueous phase does not need to be present to be able to perform hydrate
calculations.
The properties of the empty hydrate are directly used in the model as opposed to
the difference properties suggested by Saito et al.
79
The expression for the fugacity of water in the hydrate follows the same
framework as that in all other phases (i.e. aqueous phase and pure solid phases).
81
Note that the parameters for the Shock and Helgeson equation, Bromley activity model,
and solid salts phases have been regressed and reported by Jager (2001). This chapter
describes the regression procedure for all other parameters in Table 5.1 as well as the
82
results. Note that Section 5.4 discusses the common approach to optimizing hydrate
parameters. Chapter 6 discusses the approach taken in this work.
5.1
predicted phase behavior with experimental data for all hydrocarbon systems available in
the GPA Thermodynamic Database (1996). What follows are a few of the 45 vaporliquid hydrocarbon equilibria (VLE) diagrams showing the validity of the SRK EoS for
the vapor and liquid hydrocarbon phases.
100
Bubble Point
Critical Point
Dew Point
SRK EoS
80
Pressure (bar)
40
20
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
83
31
Bubble Point
Dew Point
SRK EoS
26
Pressure (bar)
16
11
6
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
91
81
Pressure (bar)
71
61
51
Dew Points
Critical Point
Bubble Points
41
SRK EoS
31
21
11
1
150
170
190
210
230
Temperature (K)
250
270
84
For all systems, the predicted K-values were within 20% of the experimental values, with
the most error coming from systems involving hydrogen sulfide and carbon dioxide.
However, for systems involving only hydrocarbons, the predicted K-values are within
10% of the experimental values. As seen in Figures 5.1 and 5.3, predictions tend to
deviate from the experimental data near the critical point and retrograde regions. This is
typical of all cubic equations of state.
The molar volumes of several hydrate formers were also predicted to ensure that
the SRK EoS could model them well. Figures 5.4 and 5.5 are plots of the molar volume
of methane and carbon dioxide versus temperature and pressure.
10000
1000
10 bar
100 bar
1000 bar
SRK EoS
100
10
250
260
270
280
290
300
310
320
Temperature (K)
85
10000
1000
SRK EoS
T = 280 K
vapor P = 42 bar
T = 300 K
vapor P = 67.4 bar
100
10
10
100
1000
Pressure (bar)
5.2
Regression of SRK
As described in Chapter 3, the SRK cubic equation of state is explicit in pressure.
The key regression parameter is kij, which is the symmetric interaction parameter between
components i and j. The API Data Book (1999) provided all parameters except for the
interaction parameters for water and hydrocarbons. In this work, we used all parameters
86
given by the API Data Book and we regressed the water-hydrocarbon interaction
parameters to experimental data.
The key type of data used in the regression was the water content (amount) in the
vapor and liquid hydrocarbon phases. Regressing to this type of data ensures that the
saturation point of water is modeled correctly. Note that, because the amount of water in
the hydrocarbon phases is very small (typically < 0.01 mole fraction), the fugacity of
water in all other phases is not changed significantly.
A least squares method was employed for the regression in which the predictions
were compared to data for several values of kij. All parameters were regressed using
binary experimental water content data when available. That is, the water-methane
interaction parameter was determined using water content data in methane gas. When
data was available with two hydrocarbon phases, the water content in the vapor phase
data was used. The best fit interaction parameters were found to 4 significant digits via
linear searches and are given in Table 5.2.
87
kij
0.4965
0.3217
0.5975
0.3815
0.5612
0.5569
0.5382
0.5260
0.5084
0.3238
0.5238
0.5257
0.5135
0.4969
0.5272
0.4816
0.3024
Component
Cycloheptane
Ethylcyclopentane
Methylcyclohexane
n-Heptane
2,2,3-Trimethylbutane
2,2-Dimethylpentane
3,3-Dimethylpentane
cis-1,2-Dimethylcyclohexane
1,1-Dimethylcyclohexane
Ethylcyclohexane
n-Octane
n-Nonane
n-Decane
Nitrogen
Hydrogen Sulfide
Carbon Dioxide
kij
0.4779
0.4823
0.4822
0.4880
0.4886
0.4903
0.4881
0.5108
0.5120
0.5078
0.4871
0.4832
0.4820
0.5063
0.1450
-0.0700
An interaction parameter, kij, equal to zero means that molecule i interacts with molecule
j no differently than it would interact with itself (and vice-versa). Therefore, we expect a
hydrocarbon-hydrocarbon interaction to be relatively small since hydrocarbons do not
have much polarity. Note that the interactions of the molecules with water, shown in
Table 5.2, are fairly large. This is due to the highly polar water molecule. Because the
amount of water content data is great, Figures 5.6 to 5.8 are only a few of the 29 plots
showing the prediction results.
88
0.1
ywater
0.01
0.001
0.0001
10
100
1000
Pressure (bar)
0.01
T = 344.26 K
ywater
0.001
T = 310.93 K
0.0001
3-
region
0.00001
T = 303.15 K
298.15 K
293.15 K
283.15 K
273.15 K
263.15 K
253.15 K
243.15 K
0.000001
10
100
1000
Pressure (bar)
89
0.1
ywater
T = 366.48 K
T = 322.04 K
Methane : 0.712
Propane : 0.037
CO2 : 0.188
H2S : 0.063
0.01
Ng et al., 2001
0.001
10
100
1000
Pressure (bar)
E=
(y
yw prediction )
w experimental
i =1
5.1
was less than 1E-6 with the most error coming from systems involving hydrogen sulfide
and carbon dioxide.
5.3
90
compressibility of ice to the two triple points of Ice I (Ice I-Lw-V and Ice I-Ice III-Lw)
from Bridgman (1912).
Since the model fit the lower phase boundary well, the
compressibility of ice was regressed to fit the upper phase boundary (i.e. 251.17 K and
2099 bar). Figure 5.9 shows the results of the regression.
10000
Ice III
Bridgman, 1912
Douslin and Osborn, 1965
Jancso et al., 1970
Henderson and Speedy, 1987
Chou, 2000
1000
Pressure (bar)
100
10
Lw
Ice I
1
0.1
0.01
V
0.001
250
255
260
265
270
275
280
285
290
Temperature (K)
91
5.4
parameters. Chapter 6 discusses the approach taken in this work. The standard approach
to regressing hydrate model parameters is to set the difference properties between pure
water or ice and the empty hydrate (Equation 3.45) and regress the Kihara potential
parameters independently. Low pressure hydrate formation data were the only data used
in the regression. What follows is a discussion of this approach.
5.4.1 Setting the Difference Properties
Many researchers have reported values for the difference properties in Equation
3.45. The difference in Gibbs energies and enthalpies of formation at T0 and P0
are typically taken to be constants depending only on hydrate structure. Table 5.3
lists the difference properties values (between ice and empty hydrate) reported in
literature.
Table 5.3 Difference properties between ice and empty hydrate at 273.15 K and 1
bar
sI Hydrate (J/mol)
g w0
hw0
699
0
1255
1264
1150
1155
381
1276/1294 941/2213
1297
1389
1299
1861
1120
931
1287
931
1297
1395
1263
1389
1236
1703
0
795
808
0
1025
1400
764
785
1025
Researcher
van der Waals and Platteeuw (1959)
Barrer and Ruzicka (1962)
Child (1964)
Sortland and Robinson (1964)
Parrish and Prausnitz (1972)
Holder (1976)
Holder and Corbin (1979)
Dharmawardhana et al. (1980)
Holder et al. (1980)
John et al. (1985)
Handa and Tse (1986)
Barkan and Shinen (1993)
Sloan (1998)
Cao et al. (2001)
92
As can be seen in the table, there is a lot of scatter between the reported values. The
stability of hydrates is very sensitive to the choice of values for the difference properties.
For this reason, we chose to optimize the formation properties of the standard empty
hydrate and the Kihara potential parameters simultaneously.
The difference in volume between ice and each empty hydrate of sI, sII, and sH is
assumed to be constant values of 3.00, 3.40, and 3.85 cm3/mol, respectively. When the
difference is needed for pure liquid water, a constant value of 1.598 cm3/mol is added to
that of the difference for ice. This approach has two problems. First, it assumes that the
volumes of ice and pure liquid water have the same temperature and pressure
dependence.
dependence of each hydrate structure is the same as that of ice and pure liquid water.
Both of these assumptions are certainly true to a first approximation. However, there is
enough experimental data on the volumes of ice, pure liquid water, and hydrates so that
more accurate expressions can be developed for each phase.
5.4.2
above 200 bar, the model tends to underpredict hydrate formation pressures. Per the
discussion above, it is not a mystery why this is so. For temperatures near 273.15 K and
pressures up to 200 bar, the volumes of ice, pure liquid water, and hydrates are fairly
constant.
expansivity and compressibility of the three phases affects the predictions. This is why
the standard approach is to regress to hydrate formation data at low pressure; because the
assumptions stated above do not hold at high pressure.
In this work, we have developed models for the volumes of each phase that are
valid at pressures beyond 1000 bar. With these models, it is possible to regress the
hydrate parameters to experimental data at more extreme conditions.
93
stringent criterion than pressure, but it is also analytical, permitting a more efficient
optimization approach.
95
This chapter gives a detailed outline of the steps involved in regressing the
hydrate parameters to experimental data. In particular, we will discuss the Gauss-Newton
method, the type of data used in the regression, the objective functions, the parameters to
be optimized, and the derivatives of the objective function with respect to each
parameter.
6.1
1 n
#
#
( p ) = ( F j ( p ))
2 j =1
6.1
#
The vector, p , represents all model parameters that need to be
#
optimized. The objective function, F R n , will be defined in Section 6.3. The
#
equations, Fj ( p ) , are nonlinear and, therefore this problem is commonly called the
is at a minimum.
nonlinear least squares problem. A full Newton method, at a given set of parameters,
involving the step
#
2c sc = c
6.2
#
where sc is the correction vector, involves considerable derivative evaluation since the
gradient of is given by
#
c = J cT Fc
and the second gradient is given by,
6.3
96
c = J J c + Fjc 2 Fjc .
2
T
c
6.4
j =1
#
# #
In this context, Jc is the Jacobian matrix of the function Fc evaluated at p = pc :
Fic
p j
( J c )ij =
1 i n, 1 j m
6.5
or,
F1c
p
1
F2c
J c = p1
$
Fnc
p1
F1c
p2
F1c
pm
F2 c
F2 c
...
p2
pm .
6.6
$
%
$
Fnc
Fnc
...
p2
pm
#
Note that the second gradient of Fc could be considered the Jacobian, or gradient, of this
...
Jacobian matrix. It would require much more evaluation of the derivatives of the
#
function Fc and therefore, we will find a way around having to determine an expression
for it.
If the model we are trying to fit the parameters to fits the data well, then near the
solution we will have Fjc 0 , j = 1,,n, and the summation term in Equation 6.4 will go
away, reducing it to
2c = J cT J c .
6.7
Substitution of Equations 6.3 and 6.7 into 6.2 gives the following expression used to
obtain a solution to the minimization problem stated in Equation 6.1:
#
#
J cT J c sc = J cT Fc .
6.8
We now proceed to find a solution to Equation 6.8 using the Gauss-Newton iterative
method:
97
#
# #
#
1. Evaluate Fc = F ( pc ) and the Jacobian J c = J ( pc ) .
#
#
2. Solve J cT J c sc = J cT Fc .
# # #
3. Update p = pc + sc .
4. Goto Step 1
#
Note that sc is the solution to the normal equations for the least squares problem
# #
min
J
s
.
c + Fc
# m
s R
6.9
Thus, in a reliable implementation of the Gauss-Newton step we can use a least squares
method to compute the step direction.
6.2
formation pressure and temperature points (i.e. the formation pressures and temperatures
of hydrates for a given gas mixture in contact with water). Since the main purpose of a
hydrate prediction program is to predict this type of data, it was used in the optimization.
In all optimization efforts reported in literature, this was the only type of experimental
data used to regress the hydrate parameters. In this work, however, we choose to use
other types of experimental data available.
Note that hydrate formation data is analogous to dew and bubble point data for
vapor-liquid hydrocarbon equilibria. When checking the validity of a cubic equation of
state, however, the predictions of dew and bubble points is checked as well as other type
of experimental vapor and liquid hydrocarbon data such as compositions, azeotropes, and
volumes. We certainly would not be satisfied with a cubic equation of state if it could
predict the dew and bubble points but not other types of data.
With this in mind, in this work, we were not satisfied with the hydrate equation of
state if it could only predict the formation temperatures and pressures. That is, we used
98
6.3
Objective Function
In order to optimize model parameters for the hydrate prediction program, we
must first set up the objective function. As discussed in the previous section, the majority
of the data that we had was formation pressure and temperature points. We know that, at
the hydrate formation point, the fugacity of water in the hydrate is equal to that in the
reference phase (i.e. any other phase present). Since this is a stringent condition on the
stability of the hydrate, we chose the objective function to be of the form
Fj = f wH j f wrj
j = 1,.,n
6.10
where n is number of data points and f wH j and f wrj are the fugacities of water in the
hydrate phase and reference phase, respectively, at the temperature and pressure of data
point j. Due to the expression of fugacity of water in the hydrate phase (Equation 4.47),
and due to the monotonic nature of a function and its natural log, it was more convenient
to express the objective function as
Fj = ln f wH j ln f wrj = 0
j = 1,.,n.
6.11
The choice to make the objective function of this is apparent when the fugacity of water
in the hydrate (Equation 4.47) is substituted into Equation 6.11.
99
Fj =
wH wo
ln f wr = 0
RT
j = 1,.,n
6.12
Recall the complete expression for the chemical potential of water in the hydrate, as
described in Chapter 4:
T
cPw dT
P
v
wH g w0 hw0 1 1 T T0
=
+
+
dT w dP +
2
RT
RT0
R T T0 T0 RT
RT
.
P0
P
v H
g w0 hw0 1 1
m m ln 1 i im + RT + R T T + RT dP
0
0
P0
6.13
Substitution of Equation 6.13 into the expression for the objective function results in
T
cP dT
P
hw0 1 1 T T0 w
v
+
Fj =
dT + w dP + m ln 1 im +
2
RT0
R T T0 T0 RT
RT
m
i
P0
g w0
av H bv H
+
RT0
R
1 1 P v H
dP wo ln f wr = 0
+
RT
T T0 P0 RT
j = 1,.,n.
6.14
Equation 6.14 is the final form of the objective function used in this work. The next
section will discuss the parameters of the hydrate model that were optimized.
6.4
100
g w0 k
hw0 k
ak
bk
ik
rim
i/k
The details of the regression procedure for each set of parameters are discussed in later
sections.
6.5
were used in the Gauss-Newton optimization procedure. In order to use the other types
of hydrate data listed above, another approach was necessary.
101
6.5.1
hydration number (i.e. number of water molecules in the hydrate per guest molecule) and
2. fractional occupancy ratio data (i.e. distribution coefficients for a guest between the
small and large cage of the hydrate). The advantage of having this type of data is that
prediction of them is independent of the formation properties of the standard hydrate
leaving the only dependence on the Kihara potential parameters.
The most abundant of this type of data is for methane hydrate and therefore, the
methane Kihara potential parameters were constrained to fit it. Several experimental
values for the hydration number and fractional occupancy ratio of methane hydrate at
273.15 K have been reported in the literature. We constrained the methane Kihara
potential parameters, i and i/k, to fit both the hydration number and fractional
occupancy data within experimental accuracy as shown in Figure 6.1.
167
162
/k (K)
Constrained Area
157
3.15
3.16
3.17
(A)
3.18
3.19
3.2
102
Figure 6.1 shows the two regions for the potential parameters needed in order to
reproduce the hydration number (light shade) and fractional occupancy ratios (medium
shade) within the experimental error. For example, the hydration number could be
predicted within experimental error for all values of shown with /k is greater than
~155 K, leaving many possible combinations of and /k with which to use. The union
of the two areas is the region in Figure 6.1 labelled Constrained Area (dark shade).
The significance of the constrained area is that the methane Kihara potential parameters
can be constrained to this region during the Gauss-Newton optimization. Note that, if the
data had no experimental error, the constrained regions would be lines intersecting at a
point in the diagram, and determination of the Kihara potential parameters for methane
would be known prior to the Gauss-Newton optimization.
6.5.2
Several
researchers have reported such data. Structural transition in cyclopropane hydrates have
been reported by Hafemann and Miller (1969). They showed that the quadruple points,
Ice-sI-sII-V and Aq-sI-sII-V, of the cyclopropane+water system were at 257.14 K and
274.6 K, respectively.
ethane+propane+water system.
Holder were qualitative in that they deduced the transition points from the slopes of the
hydrate formation lines. Subramanian et al. (2000a, 2000b) reported structural transition
data on the methane+ethane+water system for two compositions, temperatures, and
pressures. In this work, Raman and NMR spectroscopy were used to measure the hydrate
phase directly. Complimentary to the work of Subramanian et al., Jager (2001) reported
the first structural transition data for a processed natural gas using Raman spectroscopy.
The knowledge of hydrate structural transition points can be quite helpful in that
they can be used to fix the parameters, much as one would fix a cubic equation of state to
103
an azeotrope. Another benefit of hydrate structural transition data is that they give an
indication of the stability of the two hydrate structures relative to each other.
In this work, we used only the data by Subramanian et al. (2000a) for the lower
structural transition in the methane+ethane+water system. After all parameters were
optimized, the data reported by other researchers was predicted to ensure the validity of
the model. The reported composition, temperature, and pressure were 0.736 0.014
methane in the vapor phase, 274.15 K, and ~10 bar. This data point was added to the
Gauss-Newton optimization for both sI and sII hydrates. Each point was weighted to
ensure the structural transition was modeled correctly.
6.5.3 Regression to Hydrate Volume
The modifications made to the hydrate fugacity model are based solely on the
knowledge of the hydrate volume.
accounted for the temperature, pressure, and compositional dependencies. The data used
to obtain a model for the volume of hydrates was thermal expansion, compressibility, and
several measurements of hydrate volume with respect to guest composition.
The thermal expansivity of hydrates (i.e. how the volume changes with
temperature) was the first data used to develop the volume model. It has been proposed
by Tse et al. (1987) that the guest molecule plays a subtle role in governing the thermal
expansion of sI hydrates (i.e. the thermal expansion of sI hydrates is independent of
guest).
6.15
where 1, 2, and 3 are functions of the hydrate structure only. Integrating Equation
6.15 from T0 to T and cubing it, we get
2
3
v (T ) = v0 exp 1 (T T0 ) + 2 (T T0 ) + 3 (T T0 )
6.16
104
where v0 is the volume of the hydrate at T0. Huo (2002) has recently reported the lattice
parameters (cube root of volume) for many hydrates of sI and sII at 1 bar. Plotting the
change in lattice parameter versus temperature for both the temperature dependent model
given in Equation 6.16 and experimental data confirms the assumption that all sI hydrates
(Figure 6.2), and sII hydrates as well (Figure 6.3), have the same thermal expansivity.
0.005
[a(T) - a(T0)]/a(T0)
-0.005
-0.01
-0.015
-0.02
0
50
100
150
Temperature (K)
200
250
300
Figure 6.2 Plot of change in lattice parameter versus temperature for sI hydrates
105
0.002
[a(T) - a(T0)]/a(T0)
-0.002
-0.004
-0.006
-0.008
-0.01
-0.012
-0.014
50
100
150
200
Temperature (K)
250
300
Figure 6.3 Plot of change in lattice parameter versus temperature for sII hydrates
Thermal expansion data for sH hydrates of methane+neohexane have been measured by
Tse (1990). Although sH hydrates are hexagonal in structure, the temperature dependent
volume model given by Equation 6.16 is still valid. Due to lack of data, the thermal
expansion of sH hydrates was also assumed to be independent of composition. The
thermal expansion parameters for sH hydrates were regressed using the data by Tse. All
regressed values for the thermal expansivity parameters are given in Table 6.2.
Table 6.2 Regressed volumetric thermal expansion parameters for hydrate volume
(divide by 3 to get linear thermal expansion parameters)
Hydrate
sI
sII
sH
1 (K-1)
3.384960E-4
2.029776E-4
3.575490E-4
2 (K-2)
5.400990E-7
1.851168E-7
6.294390E-7
3 (K-3)
-4.769460E-11
-1.879455E-10
0
106
Note that, since all hydrates of a given structure have the same thermal expansion, the
values in Table 6.2 are also the thermal expansion parameters for the standard empty
hydrates.
Huo (2002) has recently reported XRD data supporting Avlonitis (1994b)
proposal that the volume of the hydrate should depend on the composition. We assume
that the composition dependence is independent of the temperature effects and put it into
the v0 term of Equation 6.16. The model for the compositional dependence was assumed
to be a Langmuir type expression that accounts for a guest molecules repulsive nature
with each hydrate cage. The general form of the model is
#
v0 ( x ) = a0* + N m f ( im ) rim
m
i
6.17
where Nm is the number cages of type m in the hydrate, rim is the repulsive constant for
guest molecule i in hydrate cage m, and a0* is denoted as the standard lattice parameter at
T0, P0, and some x0. The standard lattice parameters for sI, sII, and sH hydrates were
arbitrarily chosen to be 11.99245, 17.10000, and 11.09826 cm3/mol, respectively. The
function, f(im), is defined to be
f ( im ) =
(1 + m ) im exp D
1 + m im
6.18
(1 + m ) im
1 + m im
6.19
for all other hydrate cages. m is the coordination number of cage m (zm) per water
molecule in the hydrate (Nm), im is the fractional occupancy of component i in cage m, Di
is the molecular diameter of component i, and D is the fractional occupancy average
molecular diameter of the guest molecules in the hydrate.
exponential term in Equation 6.18 was suggested by Huo (2001) to account for the
relative effect of each guest in the small cages of the hydrate with respect to its diameter.
107
The regression of the repulsive constants was done all at the same temperature.
That is, all data on the volume of hydrates reported by Huo was taken to T0 using
Equation 6.16 and the repulsive constants were optimized using Newtons method. The
model fits the pseudo-experimental data for hydrate volume well, as shown in Figure 6.4.
23.4
23.3
23.2
GC Gas
Methane + Ethane
Ethane + Propane
Methane + Propane
i-Butane
Propane
Methane + i-Butane
23.1
23
22.9
22.9
23
23.1
23.2
23.3
23.4
Figure 6.4 Predicted hydrate v0 versus experimental hydrate v0 for sII hydrate
Note that the points should lie directly on the one-to-one line in Figure 6.4 if the model
predicted the data with no error. To give an idea of how much error is in the predictions
from Equation 6.17, the dotted lines above and below the one-to-one line are error
bounds of 0.15 percent from the experimental value. To put things in perspective, the
data by Huo suggest that the hydrate volume can change by as much as 3 percent due
only to composition. Table 6.3 lists the repulsive constants for each hydrate guest.
108
Table 6.3 Regressed repulsive constants and guest diameters for hydrate volume
Component
Methane
Ethylene
Ethane
Propylene
Propane
n-Butane
i-Butane
i-Pentane
Benzene
Nitrogen
H2S
CO2
Xenon
Diameter ()
4.247
4.816
5.076
5.522
5.745
6.336
6.306
6.777
6.272
4.177
4.308
4.603
4.404
sI small
1.7668E-2
0
0
0
0
0
0
0
0
1.7377E-2
1.7921E-2
0
1.8321E-2
sI large
0
1.0316E-2
1.5773E-2
2.5154E-2
2.9839E-2
0
0
0
0
0
0
5.8282E-3
0
sII small
2.0998E-3
2.3814E-3
2.5097E-3
0
0
0
0
0
0
2.0652E-3
2.1299E-3
2.2758E-3
2.1774E-3
sII large
1.1383E-2
1.3528E-2
1.4973E-2
2.1346E-2
2.5576E-2
3.6593E-2
3.6000E-2
4.7632E-2
3.5229E-2
1.1295E-2
1.1350E-2
1.2242E-2
1.1524E-2
Guests larger than ethane cannot fit into the small cage of sII and therefore the repulsive
constants are zero. Due to lack of sI compositional data, the repulsive constant for only
one of the hydrate cages could be regressed. Due to lack of sH compositional data, the
volume sH hydrates was assumed to be independent of composition.
With the temperature and compositional dependence of the hydrate modeled well,
we made another assumption that the compressibility is a function of hydrate guest, i, and
hydrate structure, H, and can be expressed as a constant:
1 a
iH = .
a P T , x
6.20
6.21
where v0 is the volume of the hydrate at P0. We can see that the exponential terms in
both Equations 6.16 and 6.21 are just corrections to the hydrate volume at T0 and P0
(Equation 6.17). Assuming that these corrections are independent of each other, the
volume can be expressed as
109
#
2
3
vH (T , P, x ) = v0 exp 1 (T T0 ) + 2 (T T0 ) + 3 (T T0 ) H ( P P0 ) .4.56
Note that the lattice parameter of cubic hydrates is just the cube root of Equation 4.56,
#
2
3
aH (T , P, x ) = a0 exp 1 (T T0 ) + 2 (T T0 ) + 3 (T T0 ) H ( P P0 ) ,
3
3
3
3
Compressibility data is
reported for sI hydrates of ethylene oxide and sII hydrates of krypton, however this data
is not applicable to this work, since we do not account for ethylene oxide or krypton.
We were able to fit the X-ray diffraction compressibility data for methane hydrate
using the hydrate volume model given by Equation 6.21 with a value of 1.0E-5 for the
linear compressibility (or 3.0E-5 for volumetric compressibility). Figure 6.5 shows the
results of the regression for methane hydrate.
110
12
T = 298.15 K
11.9
11.8
= 1.0E-5
11.7
11.6
11.5
11.4
11.3
4000
6000
8000
Pressure (bar)
Figure 6.5 Regressed value of linear compressibility for sI methane hydrate using
data by Hirai et al. (2000)
The neutron diffraction compressibility data for sI and sII hydrates of nitrogen was used
to obtain linear compressibility parameters for nitrogen in sI and sII hydrates, as shown in
Figure 6.6.
111
sII Hydrate
17.2
= 2.9E-6
16.2
15.2
T = 273.15 K
data by Kuhs et al., 1996
14.2
13.2
= 4.4E-6
12.2
11.2
400
900
1400
sI Hydrate
1900
2400
Pressure (bar)
Figure 6.6 Regressed values of linear compressibility for sI and sII nitrogen
hydrates using data by Kuhs et al. (1996)
In their work, Kuhs et al. determined that nitrogen+water forms both sI and sII hydrates
depending on the system pressure. Another, more astonishing, result of their work was
that they found that nitrogen doubly occupies the large cages of sII hydrates at high
pressures. In other work, Kuhs (2001) found that the large cages of sI hydrate are also
doubly occupied at high pressure. This data conflicts with the first assumption made by
van der Waals and Platteeuw in their development of the thermodynamic model for
hydrates. It is possible to account for more than one guest molecule in a hydrate cage;
however, in this work we do not. For this reason, it would be inconsistent to use the data
given in Figure 6.6 for the determination of the compressibility of the nitrogen hydrates.
Therefore, another approach was necessary.
112
Several hydrates have been studied at high pressure using Raman spectroscopy:
carbon dioxide (Nakano et al., 1998), methane (Nakano et al., 1999), ethane (Morita et
al., 2000), and ethylene (Sugahara et al., 2000). Although Raman spectroscopy cannot
determine the hydrate volume, it was used to determine the pressure dependence of the
oxygen-oxygen (O-O) vibration energy for each type of hydrate at pressures up to 5000
bar. The results of their work are shown in Figure 6.7.
230
-1
225
220
215
210
205
200
0
1000
2000
3000
4000
5000
Pressure (bar)
Figure 6.7 Pressure dependence of the O-O vibration energy in four different
hydrates measured via Raman spectroscopy
The actual values of the Raman shifts in Figure 6.7 are not as important as the change in
them with respect to pressure. The slopes corresponding to each hydrate may give some
indication of the compressibility of the hydrates. That is, higher O-O vibration energies
correspond to tighter hydrogen bonded cages (Sugahara et al., 2000). The slopes for each
113
of the hydrates in Figure 6.7 are plotted versus the guest diameter, which was determined
using Spartan, as shown in Figure 6.8.
0.005
methane
0.004
carbon dioxide
0.003
0.002
ethylene
0.001
ethane
0
4.2
4.4
4.6
4.8
Figure 6.8 Correlation of slope of O-O vibration energy with respect to pressure
versus hydrate guest diameter
As seen in Figure 6.8, a linear correlation exists for methane, ethylene, and ethane
hydrates. However, the carbon dioxide slope does not fit well with a linear trend. This
may be because carbon dioxide is a quadrupolar molecule whereas the other three are not.
The general trend to take from Figure 6.8 is that larger hydrate guests have smaller
changes in the O-O vibration energy with respect to pressure, and therefore smaller
compressibility. This is most likely due to the free volume in the hydrate.
Because the optimization of the potential parameters depends on the volume of
the hydrate (and therefore compressibility), an analysis was done to determine the
114
stability of the hydrate as a function of the compressibility. Figure 6.9 is a plot of the
formation pressures and temperatures for sI methane hydrate at three different
compressibilities.
1000
= 1.0E-8
= 1.0E-6
Pressure (bar)
= 1.0E-5
100
285
287
289
291
293
295
297
299
301
303
305
Temperature (K)
formation pressures and temperatures for ethane and ethylene hydrates should not be
affected by the actual value of the compressibility, assuming that the compressibility of
each hydrate is less than 1E-6.
115
With this in mind, the next step taken was to optimize the carbon dioxide,
ethylene, and ethane potential parameters to low pressure hydrate formation data. The
only data used in these regressions was along the Aq-sI-V equilibrium line, which
extends to pressures no higher than 45 bar. For this optimization, the compressibilities
were assumed to be 1E-6. When the predictions of formation pressures and temperatures
at low pressure were satisfactory, the compressibilities were regressed using the higher
pressure hydrate formation data. Although the fluid fugacity models were not regressed
to such high pressures, this was the only approach to take in order to obtain the
compressibilities.
Figure 6.10 shows the result of the regression for the linear
methane
-1
1E-5
8E-6
6E-6
4E-6
2E-6
carbon dioxide
ethylene
ethane
1E-8
4.1
4.2
4.3
4.4
4.5
4.6
4.7
4.8
4.9
5.1
116
Note that, for the non-polar molecules, there seems to be a linear trend of compressibility
with the guest diameter. This was used as a basis for determining the compressibilities of
all other guest molecules in sI hydrates.
The compressibilities for other hydrate guests of sI were regressed in a similar
fashion, however, with the constraint that they fit the linear trend in Figure 6.10. We
assumed that the slope of the linear line in Figure 6.10 for the non-polar molecules could
also be applied to polar molecules. Therefore, the compressibility of hydrogen sulfide,
being a polar molecule, was regressed such that it fit a linear trend with carbon dioxide
(having the same slope as that for non-polar molecules).
nitrogen
methane
xenon
-1
1E-5
8E-6
6E-6
hydrogen sulfide
4E-6
2E-6
ethylene
carbon dioxide
ethane
1E-8
4.1
4.2
4.3
4.4
4.5
4.6
4.7
4.8
4.9
5.1
117
We can see that all of the non-polar molecules fit the linear trend well. Note that the
regressed compressibility of sI nitrogen hydrate is greater than the value determined from
the Kuhs et al. data. This is expected since the current hydrate model does not account
for double occupancy of nitrogen in the large cage of sI hydrate. The effective guest
diameter of a doubly occupied hydrate cage would be larger than than that of a singly
occupied hydrate, and therefore having a smaller compressibility.
The compressibilities of sII hydrate formers could not be determined via
regression to hydrate formation pressures and temperatures due to lack of experimental
data at high pressures. Therefore, all values were arbitrarily chosen between 1E-6 and
1E-8, based on guest size. The compressibility of a mixed hydrate (i.e. more than one
guest molecule) is assumed to be of the form
C
H = iH iL
6.22
i =1
where iL is the fractional occupancy of hydrate guest i in the large cage. Note that for
sH hydrates, no experimental compressibility data exists. Therefore, the compressibility
of the sH hydrate, sH, is assumed to be 1E-7.
parameters used in the hydrate volume model.
118
sI
1.0E-05
2.2E-06
1.0E-08
1.0E-07
1.0E-07
NA
NA
NA
NA
1.1E-05
5.0E-06
1.0E-06
9.0E-06
sII
5.0E-05
2.2E-05
1.0E-07
1.0E-06
1.0E-06
1.0E-08
1.0E-08
1.0E-08
1.0E-08
1.1E-05
1.0E-05
1.0E-05
1.0E-05
Note that the compressibility parameters were not known for sI hydrate formers in the sII
hydrate. Since the large cage of sII hydrate is larger than that of the sI hydrate, and
therefore a larger free volume, it was assumed that the compressibility parameters were
approximately an order of magnitude greater (and vice versa for sII formers in sI
hydrate).
The standard hydrates of sI, sII, and sH were assumed to be the empty hydrates of
methane, propane, and methane+neohexane, respectively. While the thermal expansion
parameters are the same for the real and standard hydrates, the compressibility parameter
and standard volume are not. The volumetric compressibilty of the standard hydrates of
sI, sII, and sH are 3E-5, 3E-6, and 3E-7, respectively. The standard volumes for each are
22.7712, 22.9456, and 24.2126 cm3/mol, respectively.
119
6.6
hydrate parameters, the first derivatives of the objective function, Equation 6.14, with
respect to each parameter is needed. What follows is the derived expressions for the
these derivatives.
6.6.1 Derivatives with Respect to Formation Properties
The derivatives of the objective function with respect to the formation properties
are fairly simple to evaluate.
Fj
g w0
Fj
hw0
Fj
a
Fj
b
6.6.2
1
RT0
j = 1,.,n.
6.23
11 1
R T T0
j = 1,.,n.
6.24
v H
RT0
j = 1,.,n.
6.25
v H 1 1
R T T0
j = 1,.,n.
6.26
parameters are more complex than those for the formation properties. The derivative
with respect to the soft core radius of the hydrate guest is
Fj
k
=
m
1 im
km
k
k = 1,.,C
j = 1,.,n
6.27
k = 1,.,C
j = 1,.,n
6.28
120
= m
m
km Ckm
Ckm k
k = 1,.,C
j = 1,.,n
6.29
d
du
dv
f ( x, y ) dy = f ( x, u ) + f ( x, v ) + f ( x, y ) dy
dx u
dx
dx u x
6.30
for evaluating the derivative of the Langmuir constant expression, Equation 4.66, we get
4
Ckm
=
kT
k
R1 ak
kn ( r )
kn ( r ) 2
r dr
exp n
n k kT
kT
k = 1,.,C
6.31
kn + kn 5 k + kn
k kT
kT Rn11r
Rn
Rn
Rn r
k = 1,.,C
6.32
Since the implicit function knN is not a function of the soft core radius of the hydrate
guest, the evaluation of it is the same (Equation 3.43b).
The derivative of the objective function with respect to the potential well-depth of
the hydrate guest is
Fj
m
km
=
m 1 im k
k
k
k
i
k = 1,.,C
j = 1,.,n
6.33
k = 1,.,C
j = 1,.,n
6.34
k = 1,.,C
j = 1,.,n
6.35
Ckm k
k
k
k
Substitution of Equation 6.34 into 6.33 gives
Fj
Ckm
= m km
Ckm k
m
k
k
k
121
We again use Leibnitz rule to evaluate the derivative of the Langmuir constant
expression to get
Ckm
4
=
kT
j
k
R1 ak
kn ( r )
kn ( r ) 2
exp n
r dr k = 1,.,C 6.36
n k kT
kT
kn (r ) kn (r )
=
kT kT k
k
k
k = 1,.,C
6.37
kT k
k
k
R1 ak
kn ( r )
kn (r ) r 2 dr
exp n
kT
kT
n
k = 1,.,C
6.38
Since the implicit function knN is not a function of the potential well depth of the hydrate
guest, the evaluation of it is the same.
6.7
pressures and temperatures were used. Note, however, that only Aq-H-V data was used
in the optimization (as opposed to Ice-H-V and Aq-H-Lhc data). The reason for this is
that the kinetics of hydrate dissociation are dampened when an ice or liquid hydrocarbon
phase is present, possibly resulting in data of poor quality.
It should be noted that, when optimizing to hydrate formation data, the hydrate
structure needs to be specified. This is not much of a problem for single hydrates such as
methane, ethane, and propane. However, spectroscopic measurements of binary hydrates
show that systems such as methane+ethane show could form either sI or sII. Therefore it
122
was necessary for us to perform an analysis on all binary hydrates to make sure that the
correct structure was assumed.
Although the Gauss-Newton method is a robust method, we found that if too
many parameters were regressed simultaneously, a satisfactory optimization could not be
performed. However, since methane, ethane, and propane are the most abundant gases in
a natural gas pipeline, the formation properties of sI and sII hydrates as well as the
potential parameters for these hydrocarbons were regressed simultaneously. All low
pressure data for methane, ethane, propane, methane+ethane, methane+propane, and
ethane+propane hydrates was used in the regression. A complete listing of the data used
is given in Appendix C. Note that only binary hydrate data was used so that true
predictions could be made for multi-component hydrates.
The hydrate structure of the methane+ethane were assumed based on
spectroscopic measurements and the structure of the methane+propane hydrates were
assumed to be sII for compositions of methane less than 0.99 mole fraction. For the
ethane+propane hydrates, we created pseudo-binary P-x plots to determine the hydrate
structure. Figure 6.12 gives an example for a temperature of 277.6 K.
123
13
12
11
sI Hydrates
Pressure (bar)
10
9
8
7
sII Hydrates
6
5
4
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
124
1000
Pressure (bar)
10
273
275
277
279
281
283
285
287
Temperature (K)
125
sI
-235537.85
sII
-235627.53
sH
-235491.02
hw0 (J/mol)
-291758.77
-292044.10
-291979.26
a (J/cm3)
25.74
260.00
-481.32
-68.64
b (J/cm )
a ()
0.3834
0.4700
0.5651
0.6500
0.6502
0.9379
0.8706
0.9868
1.2000
1.0175
0.7773
1.0054
1.0481
1.0790
1.0576
1.1401
1.0693
1.1288
1.2134
1.2219
1.1494
1.1440
1.1606
0.3526
0.3600
0.6805
0.2357
()
3.14393
3.24461
3.24693
3.33039
3.41670
3.51726
3.41691
3.54550
3.25176
3.55376
3.56184
3.56878
3.54932
3.57910
3.59028
3.60425
3.58776
3.59955
3.59989
3.59117
3.60555
3.60212
3.60932
3.13512
3.10000
2.97638
3.32968
/k (K)
155.593
180.664
188.181
186.082
192.855
197.254
198.333
199.560
223.802
211.924
253.681
229.928
229.832
210.664
250.187
219.083
237.989
232.444
224.609
204.968
233.510
246.996
220.527
127.426
212.047
175.405
193.708
126
Figures 6.14 and 6.15 show the relation between the soft core radius, i, and the
well potential depth, i/k, versus the hard core radius, ai, for the Kihara parameters
determined in this work and those of three other researchers. As can be seen, the set of
Kihara parameters determined in this work are more linear than the other sets, as
suggested by the work of Tee et al. (1966). It should be noted that the set by Erickson
and Avlonitis are for sI and sII hydrate formers, the set by Mehta and Sloan are for sH
hydrates formers, and the set for this work are for sI, sII, and sH hydrate formers.
4.8
Avlonitis, 1994
4.4
Erickson, 1980
4.6
4.2
This Work
R = 0.0557
2
R = 0.0407
2
R = 0.1186
2
R = 0.6133
3.8
3.6
3.4
3.2
3
2.8
0.2
0.4
0.6
0.8
1.2
127
550
Erickson, 1980
R = 0.3834
2
R = 0.1477
Avlonitis, 1994
2
Mehta and Sloan, 1996 R = 0.0015
2
R = 0.4783
This Work
500
450
400
350
300
250
200
150
100
0.2
0.4
0.6
0.8
1.2
128
This difference in
hydrogen position relative to the oxygen will result in different properties of the water
molecule, and therefore the potential parameters. For this reason, each water molecule in
a hydrate cage should have different potential parameters. Note that this will be even
truer for water molecules in other cages. Since the interactions between the encaged
molecules and the water molecules should be different depending on the water molecule,
the Kihara potential parameters should also be different. That is, a set of mixture Kihara
parameters should be regressed for each water molecule in the hydrate lattice, or at least
for each type of hydrate cage. Note that the current approach is to lump all of this into
one set of Kihara potential parameters. However, due to the lack of data and the fact that
the current approach predicts the experimental data within reason, the current approach is
used in this work. That is, each guest-water pair has a single set of Kihara potential
parameters.
129
This chapter gives a quantitative discussion for the results of the hydrate
parameter optimization.
developed in this work, and four other commercial hydrate prediction programs is given
for all hydrate data reported in literature. The four commercial programs compared in
this work are:
CSMHYD
DBRHydrate
Multiflash
PVTsim
The comparisons are broken into two categories: hydrate phase properties and hydrate
formation temperatures and pressures.
7.1
properties with predictions from several hydrate programs. Note that the results of the
volume regression are given in Chapter 6, and therefore are not discussed here.
7.1.1 Fractional Occupancy Ratios
Table 7.1 shows the cage occupancy ratios (fraction of the large cages occupied
by the guest divided by the fraction of the small cages occupied) for methane, hydrogen
sulfide, xenon, and carbon dioxide hydrates at 273.15 K.
130
L/S (Data)
1.156 (e)
1.053 (c)
1.092 (f)
1.000 (d)
8.333 (a)
1.429 (c)
1.493 (d)
1.299 (b)
CSMGem
CSMHYD
1.146
1.118
1.116
1.013
1.908
1.064
1.071
1.347
1.414
1.233
Davidson 1984
Ripmeester and Davidson 1981
Tse and Davidson 1982
Cady 1983
Subramanian, 2000
Ripmeester and Ratcliffe 1988 (T = 233.15 K)
Note that CSMGem and CSMHYD are the only hydrate programs that report the cage
occupancies and are therefore the only program predictions to compare with data. Both
programs give similar results for methane and hydrogen sulfide hydrates. However, for
xenon hydrates, CSMHYD under predicts the data. For the carbon dioxide hydrate data,
both programs under predict the reported experimental value.
The fractional occupancy ratio of methane hydrate has been extensively studied at
temperatures other than 273.15 K. Sum (1997), Subramanian (2000), Jager (2001), and
Kini (2002) have reported the fractional occupancy at several temperatures and pressures
along the three-phase equilibrium line. Figure 7.1 shows the predictions of CSMGem
with their data.
131
Ice-sI-V
1.28
Aq-sI-V
1.23
T = Teq
1.18
1.13
1.08
1.03
0.98
10
100
1000
Pressure (bar)
132
pressures in which the hydrate forms below the ice point. The data of Kini confirm this
prediction.
7.1.2 Hydration Number
The hydration number is the number of water molecules in the hydrate per guest
molecule. Table 7.2 is a comparison of the experimental values for the hydration number
for sI and sII hydrates at 273.15 K. All four hydrate programs predict the hydration
number for methane, ethane, propane, and i-butane hydrates within reason. However,
CSMGem, Multiflash, and CSMHYD all predict the hydration number for hydrogen
sulfide within reason but PVTsim predicts a much larger value. This is because PVTsim
predicts hydrogen sulfide to form sII hydrates (which has less large cages to
accommodate the hydrogen sulfide molecule) whereas the other hydrate programs predict
hydrogen sulfide to form sI hydrates. The experimental data for the hydration number
suggest that hydrogen sulfide forms sI hydrates. CSMGem predicts the xenon hydration
number well, but CSMHYD under predicts the experimental value. Note that Multiflash
and PVTsim do not contain xenon as a component so predictions could not be made.
Also note that DBRHydrate does not report the overall composition of the hydrate so that
comparisons could not be made.
133
Methane
Ethane
Propane
i-Butane
H2S
Xenon
a
b
c
d
e
f
Data
5.77 (d)
6.00 (c,h)
6.30 (e)
7.4 (e)
7.00 (f)
7.18 (g)
7.67 (c)
7.00 (f)
8.25 (g)
8.24 (h)
17.0 (a,c)
5.7 (i)
17.95 (g)
18.0 (j)
17.0 (c)
17.1 (b)
17.5 (k)
6.12 (a)
6.48 (a)
6.29 (b)
CSMGem
CSMHYD
Multiflash
PVTsim
6.300
6.073
6.100
6.046
7.939
7.775
7.767
7.774
17.02
17.02
17.01
17.02
17.01
17.02
17.01
17.02
6.241 (sI)
6.027 (sI)
6.106 (sI)
6.632 (sII)
6.529
6.181
NA
NA
Cady 1983
Uchida and Hayano 1964
Handa 1986
Glew 1962
de Roo et al. 1983
Roberts et al. 1941
g
h
i
j
k
In order to get hydration number, hydrates must be formed and then dissociated.
Due to the highly metastable nature of forming hydrates (i.e. disordered system !
ordered system), typically the pressure is increased above the equilibrium pressure (Peq)
until hydrates form. At that point, the hydrates are dissociated at T and Peq. Figure 7.2
gives an illustrative description of this.
134
1000
Aq-H
Aq-H-V
formation conditions
Texp, Pexp
Pressure
Teq, Pexp
dissociation
conditions
Aq-V
Texp, Peq
100
Temperature
135
6.35
6.3
Hydration Number
6.25
6.2
T = Teq
6.15
6.1
T = 273.15 K
6.05
6
5.95
5.9
26
36
46
56
66
76
86
96
Pressure (bar)
136
7.1.3
Structural Transitions
Another crucial piece of hydrate data is the structural transition composition.
Hydrate structural transition data gives the composition, temperature, and pressure at
which a two hydrates are present at the formation conditions. Table 7.3 is a comparison
of the experimental values for the structural transition compositions of the methane+
ethane+water system at 274.15 K.
Table 7.3 Comparison of experimental and predicted structural transition
composition at 274.15 K for the methane+ethane+water system
Data
0.736 0.014 (a)
0.993 0.006 (b)
a
b
CSMGem
0.722
0.994
Multiflash
0.600
0.994
PVTsim
0.668
0.992
Note that Multiflash and PVTsim both under predict the lower composition whereas
CSMGem predicts the value within experimental error.
predicted well by all three programs. This is important to note since it could have
applications for natural hydrates in the permafrost and deep-sea sediments. It also should
be noted that both CSMHYD and DBRHydrate predict that the methane+ethane+water
system will not form sII hydrates along the entire composition extremes.
7.2
pressures. This type of data is most important to natural gas pipeline situations. We have
compiled a list of all experimental data of this type and compared it with predictions from
CSMGem, CSMHYD, DBRHydrate, Multiflash, and PVTsim. It should be noted that
DBRHydrate has a prediction limit to pressures below 680 bar and so comparisons could
not be made for all data points.
137
What follows is a series of tables giving the average errors in temperature for
several types of hydrates based on the following categories:
7.2.1
7.2.2
7.2.3
7.2.4
7.2.5
7.2.6
7.2.7
single hydrates,
binary hydrates,
ternary hydrates,
multi-component hydrates (natural gas hydrates),
hydrates in black oils and gas condensates,
sH hydrates, and
groupings of all hydrates.
The results are given as the average error either in temperature or pressure. The average
error in temperature is calculated as
(T
predicted
TError =
# data points
-Texperimental )
# data points
and is given as a number (in Kelvin). The temperature and pressure ranges for all results
are reported in K and bar, respectively. At the end of each sub-section, an analysis is
given, discussing major differences in the models. A more complete listing of the results
can be found in Appendix D.
7.2.1 Hydrate Formation Conditions for Single Hydrates
The most common hydrate studied is the single hydrate (i.e. methane hydrate).
This is because hydrate model parameters are typically regressed to simple hydrates
(single and binary) and then extrapolated to more complicated hydrates (multicomponent). Figures 7.4 through 7.13 show the average temperature error, broken into
several different ranges of pressure along the three-phase equilibrium line, for each single
hydrate included in this work. The ranges of the temperatures and pressures are given at
the bottom of the figures. Also reported at the top of the figures is the three-phase
equilibrium line in which the prediction errors are given (i.e. Ice-H-V, Aq-H-V, and AqH-Lhc).
138
Ice-H-V
Aq-H-V
Aq-H-V
Aq-H-V
CSMGem
CSMHYD
DBRHydrate
Multiflash
PVTsim
4
3
NA
NA
0
-1
-2
T < 273
P < 26
Ice-H-V
Aq-H-V
Aq-H-V
Aq-H-V
CSMGem
CSMHYD
DBRHydrate
5
4
Multiflash
PVTsim
NA
NA
NA
NA
NA
NA
0
-1
-2
T < 273
P < 5.3
139
2
1
0
Aq-H-V
Aq-H-Lhc
Aq-H-Lhc
Aq-H-Lhc
CSMGem
CSMHYD
DBRHydrate
Multiflash
PVTsim
NA
Ice-H-V
NA
-1
-2
-3
-4
T < 273
P<5
288 < T < 288.5 288.5 < T < 299 299 < T < 304
33 < P < 100
100 < P < 1000 1000 < P < 5000
1.0
Aq-H-V
Aq-H-Lhc
Multiflash
PVTsim
0.0
-0.5
-1.0
-1.5
NA
NA
NA
0.5
NA
1.5
CSMGem
CSMHYD
DBRHydrate
140
1.0
Aq-H-V
Aq-H-Lhc
0.5
0.0
-0.5
-1.0
CSMGem
CSMHYD
DBRHydrate
Multiflash
PVTsim
273 < T < 278
1.5 < P < 5.4
-1.5
Ice-H-V
Aq-H-V
Aq-H-Lhc
NA
NA
-1
CSMGem
CSMHYD
DBRHydrate
-2
-3
Multiflash
PVTsim
-4
-5
-6
-7
T < 273
P<1
141
Ice-H-V
Aq-H-V
Aq-H-V
NA
0.5
NA
1.0
0.0
CSMGem
CSMHYD
DBRHydrate
-0.5
-1.0
Multiflash
PVTsim
-1.5
T < 272
P < 141
-2.0
Ice-H-V
Aq-H-Lhc
Aq-H-Lhc
CSMGem
CSMHYD
DBRHydrate
Multiflash
PVTsim
1
0
-1
NA
Aq-H-V
T < 273
P<1
-2
142
4
2
0
Aq-H-V
Aq-H-Lhc
Aq-H-Lhc
Aq-H-Lhc
CSMGem
CSMHYD
DBRHydrate
Multiflash
PVTsim
NA
Ice-H-V
NA
-2
-4
-6
T < 272
P < 10.5
-8
4
3
Aq-H-V
Aq-H-V
Aq-H-V
Aq-H-V
CSMGem
CSMHYD
DBRHydrate
Multiflash
PVTsim
-1
T < 268
P<1
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
Ice-H-V
-2
143
In this study, it was found that the most reliable type of experimental data is the data
along the Aq-H-V equilibrium line. Data along the Ice-H-V and Aq-H-Lhc equilibrium
lines tended to be much more scattered and not consistent among different experimenters.
This may be due to the hydrate not being in equilibrium and could account for much of
the large errors seen in the above figures. Figures 7.14 and 7.15 show the average
temperature error along the Aq-H-V (P < 1000 bar) line for hydrates of sI and sII,
respectively.
1.0
CSMGem
CSMHYD
DBRHydrate
Multiflash
PVTsim
0.0
NA
NA
0.5
NA
1.5
-0.5
Methane
-1.0
Ethylene
Ethane
Xenon
sI Hydrate Guest
Figure 7.14 Hydrate formation T error along the Aq-H-V line for sI hydrates (P <
1000 bar)
144
CSMGem
CSMHYD
DBRHydrate
Multiflash
PVTsim
NA
0.00
NA
0.25
NA
0.50
-0.25
Propylene
Propane
i-Butane
-0.50
Figure 7.15 Hydrate formation T error along the Aq-H-V line for sII hydrates
CSMGem is quite comparable to the other hydrate programs and considerably better for
most systems at pressure greater than 1000 bar.
The most notable difference between CSMGem the other hydrate programs is in
the methane and carbon dioxide single hydrates. As shown in Figures 7.16 and 7.17, the
methane+water and carbon dioxide+water pressure versus temperature phase diagrams,
the experimental data suggest a maximum temperature at which hydrates form. This
temperature is approximately 321 K and 294 K for methane hydrates and carbon dioxide
hydrates, respectively. This phenomenon is called retrograde behavior, in which hydrates
will dissociate upon pressurization at a given temperature.
145
10000
Aq-sI
1000
Pressure (bar)
100
Aq-V
10
273
283
293
303
313
323
Temperature (K)
Pressure (bar)
1000
100
H-Lhc
Aq-Lhc
H-V
Aq-V
10
271
276
281
286
Temperature (K)
291
146
To keep the figures easy to read, predictions from DBRHydrate, PVTsim, and CSMHYD
are not given, however predictions from these programs are similar to those of
Multiflash. Note that the CSMGem predictions give the same trend as the data whereas
the Multiflash predictions do not.
between CSMGem and the other programs is due to the treatment of the hydrate volume
as a function of temperature and pressure. As the pressure is increased, the hydrate
volume compresses, therefore reducing the size of the hydrate cages.
At some
temperature and pressure (as determined by the data), the small cages become so small
that the guest molecule can no longer fit into them to stabilize the hydrate.
Another interesting difference between the models is in the nitrogen+water
system.
approximately 280.2 K and 334.8 bar, as shown in Figure 7.18. CSMHYD predicts sII
hydrate for the entire three-phase equilibrium line for nitrogen hydrates, whereas
Multiflash and PVTsim predict sI hydrates. The structure of nitrogen hydrates has been a
question that has never been completely answered. Kuhs et al. (1996) have done several
neutron diffraction measurement on nitrogen hydrates at 273.15 K and many pressures up
to 2500 bar. However, they found both sI and sII hydrates of nitrogen present at these
conditions.
147
1000
Ice-sII
Pressure (bar)
10000
100
Aq-sII
Aq-V
Ice-V
269
274
279
284
289
294
299
304
Temperature (K)
beyond 3800 bar in pressure. However, as shown in Figure 7.19, CSMGem predicts that
xenon hydrates do have a maximum temperature of formation at 356 K (8578 bar).
148
10000
Pressure (bar)
1000
100
10
sI-Lhc
de Forcrand, 1925
Braun, 1938
van der Waals and Platteeuw, 1959
Barrer and Edge, 1967
Aaldijk, 1971
Ohgaki et al., 2000
CSMGem
sI-V
Aq-V
1
273
283
293
303
313
323
333
343
353
Temperature (K)
149
model in order to extrapolate predictions to more complicated hydrates (multicomponent). Figures 7.20 through 7.23 show the average temperature error for several
binary hydrates. The x-axis is the primary hydrate former and each category is the
second hydrate component. It should be noted that only experimental hydrate data along
the Aq-H-V hydrate formation line is used in the comparison.
CSMGem
CSMHYD
DBRHydrate
Multiflash
PVTsim
0.5
NA
1.0
0.0
-0.5
-4.4
-6.3
-1.0
Ethane
Propane
i-Butane n-Butane
-1.5
Figure 7.20 Hydrate formation T error for binary hydrates with methane
150
1.0
0.5
CSMGem
CSMHYD
DBRHydrate
Multiflash
PVTsim
0.0
-0.5
-1.0
Propane
Carbon Dioxide
-1.5
Figure 7.21 Hydrate formation T error for binary hydrates with ethane
2.5
1.5
1.0
NA
0.0
NA
0.5
NA
2.0
CSMGem
CSMHYD
DBRHydrate
Multiflash
PVTsim
-0.5
-1.0
Propylene
i-Butane
n-Butane
Nitrogen
-1.5
Carbon
Dioxide
Figure 7.22 Hydrate formation T error for binary hydrates with propane
151
CSMGem
CSMHYD
DBRHydrate
Multiflash
PVTsim
NA
NA
-4 -9
-1
-2
i-Butane + n-Butane
i-Butane + Carbon
Dioxide
n-Butane + Carbon
Dioxide
-3
Figure 7.23 Hydrate formation T error for binary hydrates with butanes
Figures 7.20 to 7.23 give only the average temperature error for all reported data of these
binary hydrates. It should be noted that the error analysis given in the figures can be
misleading for some systems. For instance, in the propane+n-butane binary hydrate,
shown in Figure 7.22, CSMHYD could only predict 8 of 54 data points. For these 8 data
points, the error is small. However, most of the error in this system comes from other
data points. A similar example can be made for the ethane+propane system shown in
Figure 7.21. Therefore, a complete listing of the number of data points and the number
of predicted points for each program is given in Appendix D as well.
There have been little experimental data for binary systems at pressures greater
than 100 bar (less than 1 percent) and no data have been taken at pressures greater than
1000 bar. Therefore, a determination of whether the CSMGem volume model correctly
accounts for high pressure in binary systems is not done. However, as can be seen from
152
Figures 7.20 to 7.23, CSMGem compares favorably with the other hydrate programs for
most systems.
The most notable difference between CSMGem and all other hydrate programs is
found in the methane+benzene hydrate. Benzene is a very large molecule and therefore
cannot form sII hydrates without the presence of a help molecule such as methane.
Figure 7.24 shows the pressure versus temperature phase diagram for this system. Note
that the overall composition of the system is not needed since the four-phase equilibrium
line (Aq-sII-V-Lhc) is set by Gibbs phase rule.
100
CSMGem
DBRHydrate
Multiflash
PVTsim
Pressure (bar)
sII-V-Lhc
Aq-V-Lhc
277
279
281
283
Temperature (K)
285
287
153
therefore, only predictions from these two programs are given in the figure. DBRHydrate
and PVTsim predict that sI hydrates are the most stable, and therefore, these programs
have large errors in the predicted formation temperature.
An interesting prediction to be noted is that of the methane+n-butane hydrates.
Predictions show that there are several structural transitions: sII hydrates at low pressure
and sI hydrates at higher pressure. Figure 7.25 is a phase equilibria plot for this system.
1000
Pressure (bar)
Aq-sI-V
Aq-sI-sII
100
Aq-sII-V
Aq-V
10
273
278
283
288
293
298
Temperature (K)
154
predictions such as this one that show the benefits of using the Gibbs energy
minimization technique to predict at pressures above the hydrate formation conditions.
7.2.3 Hydrate Formation Conditions for Ternary Hydrates
Hydrate formation data for ternary hydrates is the first type of data in which a
check on the extrapolation from the regression of pure and binary hydrates can be made.
However, very few ternary hydrate data have been taken. What follows is a comparison
of the formation temperatures and pressures for ternary hydrates of methane+ethane+
propane.
1.0
0.5
CSMGem
CSMHYD
DBRHydrate
Multiflash
PVTsim
3.5
3.6
3.1
3.5
3.3
0.0
-0.5
-1.0
155
methane, all programs have large error. The data at low concentrations of methane were
taken by Deaton and Frost (1946) and Hammerschmidt (1934) while the data at higher
concentrations were taken by Holder (1976). It is unlikely that all programs describe the
methane+ethane+propane+water system incorrectly.
between the predictions and the data may be due to experimental procedure of Holder.
One other set of ternary hydrate data has been reported. Figure 7.27 shows the
error analysis for ternary hydrates of methane+carbon dioxide+hydrogen sulfide, broken
into three different categories of methane composition.
0.5
0.0
-0.5
CSMGem
CSMHYD
DBRHydrate
Multiflash
PVTsim
-1.0
-1.5
-2.0
156
7.2.4
data from an industrial perspective. It is also important for the purpose of validating the
hydrate model. However, very few data have been taken for hydrates of natural gases.
Therefore, it is difficult to determine the accuracy of hydrate programs for real
systems.
Deaton and Frost (1946) have determined hydrate formation points for several
natural gas hydrates, however, they did not report the butane analysis for each gas
mixture (i.e. separate compositions of i-butane and n-butane).
Because of this, we
performed a sensitivity analysis to determine the affect of the composition of i- and nbutane on hydrate stability. Figure 7.28 is a pressure versus temperature phase diagram
for one of their typical natural gas hydrates.
100
Pressure (bar)
10
Gas Composition
methane
0.906
ethane
0.038
propane
0.015
butanes
0.020
nitrogen
0.012
CO2
0.009
1
273
275
277
279
281
283
285
287
289
291
Temperature (K)
Figure 7.28 Pressure versus temperature phase diagram for a natural gas hydrate
showing the sensitivity of the predictions to butane content
157
Note that, in the report of the gas composition, the individual butane amounts were not
specified. Two cases were studied: 1. assuming the entire butane amount was i-butane
(dashed line) and 2. assuming the entire butane amount was n-butane (solid line). As
seen in Figure 7.28, the hydrate formation temperatures and pressures are very sensitive
to the relative amount of butanes in the natural gas mixture. In fact, the difference in
predicted hydrate formation pressure for both cases ranged from 15-50% (1.0-4.0 K in
predicted temperature) depending on the overall gas composition and relative amount of
butanes. Therefore, the Deaton and Frost data could not be used to compare the different
hydrate program predictions.
A complete listing of all researchers reporting hydrate formation points for natural
gases is given in Table 7.4.
Table 7.4 Complete listing of natural gas hydrate formation data
Researcher
Wilcox et al.
Deaton and Frost
Kobayashi et al.
McLeod and Campbell
Lapin and Cinnamon
Adisasmito and Sloan
Bishnoi and Dholabhai
Jager
Number of Gases
3
11
2
1
3
5
1
1
27
Data Points
36
90
10
7
3
20
3
17
186
Year
1941
1946
1951
1961
1969
1992
1999
2001
Unfortunately, the number of data points measured by Deaton and Frost comprise almost
50 percent of the total amount of data for natural gas hydrates in the past 60 years.
Figures 7.29 through 7.33 give the errors for multi-component hydrate formation
temperatures for the natural gas hydrate data reported in Table 7.4, with the exception of
the data by Deaton and Frost.
158
1.0
0.5
0.0
CSMGem
CSMHYD
DBRHydrate
Multiflash
PVTsim
-0.5
-1.0
-1.5
-2.0
Natural Gas B
(9 data points)
Natural Gas C
(15 data points)
Natural Gas D
(12 data points)
Figure 7.29 Hydrate formation T error for Wilcox et al. (1941) data
0.5
0.0
-0.5
-1.0
-1.5
-2.0
-2.5
Hugoton Gas
(5 data points)
CSMGem
CSMHYD
DBRHydrate
Multiflash
PVTsim
Michigan Gas
(5 data points)
Figure 7.30 Hydrate formation T error for Kobayashi et al. (1951) data
159
2.0
1.5
CSMGem
CSMHYD
DBRHydrate
Multiflash
PVTsim
1.0
0.5
0.0
-0.5
Natural Gas 1
(1 data point)
Natural Gas 2
(1 data point)
Natural Gas 3
(1 data point)
Figure 7.31 Hydrate formation T error for Lapin and Cinnamon (1969) data
CSMGem
CSMHYD
DBRHydrate
Multiflash
PVTsim
-1
-2
Natural Gas A
(4 data points)
Natural Gas B
(4 data points)
Natural Gas C
(4 data points)
Natural Gas D
(4 data points)
Natural Gas E
(4 data points)
-3
Figure 7.32 Hydrate formation T error for Adisasmito and Sloan (1992) data
160
1.0
CSMGem
CSMHYD
DBRHydrate
Multiflash
PVTsim
0.5
0.0
-0.5
McLeod and
Campbell (1961)
(7 data points)
Bishnoi and
Dholabhai (1999)
(3 data points)
Jager (2001)
(17 data points)
-1.0
Natural Gases
By
considering the errors associated with each hydrate program, it is a possible conclusion
that the data by Kobayashi et al. (1951) is in error.
It is interesting to note the large amounts of error between the predictions and data
for the majority of the systems, especially when compared with pure and binary hydrates.
Work by Bollavaram (2002) suggests that the kinetics of hydrate dissociation are slower
for sII hydrates than for sI. Other work at the Colorado School of Mines (Sloan, 2002)
shows that the isobaric heating rate at which the hydrate is dissociated may have a
significant effect on the experimental hydrate formation conditions. If care was not taken
in the experimental procedures, the data for natural gas hydrate formation may be in
error. This may account for some of the large errors seen in Figures 7.29 to 7.33.
It should be noted that the data by McLeod and Campbell (1961) and Jager (2001)
is at fairly high pressure. This type of data is helpful in that the trend of the data at high
161
pressure can be compared to the prediction programs. Figure 7.34 is a pressure versus
temperature plot of the data by McLeod and Campbell.
1000
Pressure (bar)
Aq-sII-V
Aq-V
Gas Composition
methane
0.906
ethane
0.066
propane
0.018
i-butane
0.005
n-butane
0.005
100
293
295
297
299
Temperature (K)
301
303
Figure 7.34 P versus T phase diagram for natural gas of McLeod and Campbell
(1961)
The high pressure trend of the predictions by DBRHydrate is incorrect relative to the
experimental data while that of CSMGem is consistent with the data.
This is an
indication of the correct high pressure treatment of the hydrate phase via the volume
model. It should be noted that, while predictions from the other programs are not shown
in Figure 7.34, CSMHYD gives a similar trend to that of DBRHydrate while the PVTsim
and Multiflash predictions give a similar trend to that of CSMGem.
An interesting anomaly associated with the processed natural gas hydrate data by
Jager (2001) is that, at high pressure, the equilibrium hydrate changes from sII to sI. This
phenomenon has been observed in many systems such as methane+ethane+water and
162
methane+i-butane+water. However, it has not been observed for a natural gas. Figure
7.35 is the pressure versus temperature phase diagram for the gas.
1000
Pressure (bar)
sII
sI/sII Jager, 2001
sI
CSMGem
CSMHYD
DBRHydrate
sI-V
sI-sII-V
100
Gas Composition
methane
0.9753
ethane
0.0088
propane
0.0014
i-butane
0.0001
n-butane
0.0002
i-pentane
0.0002
n-pentane
0.0002
hexane
0.0002
heptane
0.0001
nitrogen
0.0093
CO2
0.0041
sII-V
Aq-V
10
278
283
288
Temperature (K)
293
298
Figure 7.35 P versus T phase diagram for natural gas of Jager (2001)
While all programs predict the structural transition to occur, the high pressure behavior of
CSMHYD and DBRHydrate, shown in Figure 7.35 as the long- and short-dashed lines,
respectively, are inconsistent with the experimental data. Again, this is indicative of the
correct high pressure treatment of the hydrate phase by CSMGem.
Predictions by
163
because of the high carbon dioxide content, and the Jager data cannot be used because it
is for a processed natural gas. This leaves only 59 (out of 189) data points for 10 (out of
27) natural gases to make the comparisons. Because of this, more data for relevant
natural gases is needed.
7.2.5 Hydrate Formation Conditions for Hydrates in Black Oils and Gas Condensates
The most stringent test for a hydrate program is to compare predictions with
experimental hydrate data in black oils and gas condensates. These types of predictions
are more of a test on the hydrocarbon equation of state than the hydrate model because of
the complex hydrocarbon mixtures involved. Figure 7.36 gives comparisons of the
hydrate programs with experimental data for black oil and gas condensate systems (Notz,
2001).
1.5
0.0
-0.5
-1.0
-1.5
NA
0.5
NA
1.0
CSMGem
CSMHYD
DBRHydrate
Multiflash
PVTsim
-2.0
-2.5
-3.0
System
Figure 7.36 Hydrate formation T error for two heavy hydrocarbon systems (Notz,
2001)
164
The models perform sufficiently for the black oil but there is substantial error in the gas
condensate system.
transition from sII to sI for the gas condensate hydrates. Figure 7.37 is the pressure
versus temperature phase diagram for the gas condensate system.
1000
Pressure (bar)
Aq-H-Lhc
Aq-sI-V-Lhc
Aq-Lhc
Aq-H-V-Lhc
100
Aq-sII-V-Lhc
Data
Aq-V-Lhc
CSMGem
DBRHydrate
PVTsim
10
283
288
293
298
Temperature (K)
Figure 7.37 P versus T phase diagram for gas condensate #1a-a (Notz, 2001)
The structural transition pressures are predicted to occur at 340 bar (CSMGem), 120 bar
(DBRHydrate), 290 bar (PVTsim), and 260 bar (Multiflash) for each prediction program.
Note that predictions from DBRHydrate and CSMGem are essentially identical in the sII
region (low P) while predictions from PVTsim and CSMGem are essentially identical in
the sI region (high P). Predictions from Multiflash are not shown but are similar to those
of CSMGem. Note that DBRHydrate straddles the data, which leads to a small overall
error, but incorrectly models the trend. This is another indication of how the bar charts
presenting the predicted temperature error can be misleading for some systems.
165
7.2.6
when a large molecule such as i-pentane is in the presence of a help molecule such as
methane. To be complete, we give a comparison of the hydrate programs for all sH
hydrate data, with methane as a help molecule, in the literature. Figure 7.38 shows the
results. All data is for methane as the help molecule while the large molecules are listed
in each column.
Cyclopentane
2,3-Dimethylbutane
Neohexane
Methylcyclopentane
NA
3,3-Dimethylbutene
NA
-0.25
2,3-Dimethylbutene
0.00
-0.50
0.7
CSMGem
CSMHYD
Multiflash
PVTsim
i-Pentane
0.25
166
0.3
NA
CSMGem
CSMHYD
Multiflash
PVTsim
-1.0
Ethylcyclohexane
1,1-Dimethylcyclohexane
3,3-Dimethylpentane
cis-1,2-Dimethylcyclohexane
-0.50
2,2,3-Trimethylbutane
-0.25
Methylcyclohexane
0.00
Ethylcyclopentane
0.25
167
Pressure (bar)
100
Aq-sH-V-Lhc
10
274
278
280
282
284
286
288
290
Temperature (K)
hydrates of Sections 7.2.1 to 7.2.6 as well as the errors for all hydrates. Figures 7.40 and
7.41 are the average temperature and pressure errors for the different groupings.
168
0.25
-0.25
-0.5
CSMGem
CSMHYD
DBRHydrate
Multiflash
PVTsim
Single
(632 pts)
NA
0.79
NA
0.59
-1.14
Binary
(747 pts)
BO &
Ternary Natural Gas GC
(89 pts)
(72 pts) (10 pts)
sH
(135 pts)
Types of Hydrates
21%
0%
CSMGem
CSMHYD
DBRHydrate
Multiflash
PVTsim
-5%
-10%
NA
5%
NA
10%
Single
(632 pts)
Binary
(747 pts)
Ternary
(89 pts)
Types of Hydrates
sH
(135 pts)
169
PError =
(P
predicted
-Pexperimental )
Pexperimental
# data points
# data points
100%
where the units of pressure are arbitrary. The average error in temperature and pressure
for all hydrates is given in Figure 7.42.
0.24
0.1
4%
6%
CSMGem
CSMHYD
DBRHydrate
Multiflash
PVTsim
2%
0%
-0.1
-2%
T Error
-0.2
0.2
P Error
-4%
Figure 7.42 Hydrate formation T and P errors for all hydrates (1685 pts)
CSMGem compares quite well with the other programs. However, to be complete, we
list the average absolute errors in temperature and pressure. In this case, the absolute
value of the error is taken for each data point, and an average is taken. This ensures that
the small errors for CSMGem, shown in Figures 7.40 through 7.42, are not due to
cancellation of errors. Figures 7.43 through 7.45 show the average absolute errors for the
same systems as those in Figures 7.40 through 7.42.
170
0.75
CSMGem
CSMHYD
DBRHydrate
Multiflash
PVTsim
All Program
Errors > 1.04
All Program
1.36 Errors > 1.40
0.5
-0.25
Single
(632 pts)
Binary
(747 pts)
Ternary
(89 pts)
NA
0.25
NA
sH
(135 pts)
Types of Hydrates
Figure 7.43 Hydrate formation T error (absolute) for all types of hydrates
15%
27%
CSMGem
CSMHYD
DBRHydrate
Multiflash
PVTsim
29%
27%
10%
0%
-5%
Single
(632 pts)
Binary
(747 pts)
Ternary
(89 pts)
NA
5%
NA
20%
sH
(135 pts)
Types of Hydrates
Figure 7.44 Hydrate formation P error (absolute) for all types of hydrates
171
T Error
0.6
CSMGem
CSMHYD
DBRHydrate
Multiflash
PVTsim
P Error
12%
10%
0.5
8%
0.4
6%
0.3
4%
0.2
2%
0.1
0.7
0%
Figure 7.45 Hydrate formation T and P errors (absolute) for all hydrates (1685 pts)
All programs give large error for the ternary, black oil (BO), and gas condensate (GC)
hydrates. This is most likely due to error in the data for the ternary hydrates. However,
for the BO and GC hydrates, the error is most likely in the characterization of the heavy
hydrocarbon components.
There is certainly cancellation of errors for all programs given the larger absolute
errors in Figures 7.43 to 7.45. However, CSMGem gives the smallest prediction errors
for nearly every type of hydrate (both relative and absolute).
7.3
programs available.
172
173
The objective of this work was to create a user-friendly program that could easily
be used by a person with little knowledge of gas hydrates. The program needed to have
the appearance of a simple program but have the capability to perform many different
types of calculations. We did not want to bog down the user with all kinds of buttons
and options. Therefore, we sat down with the funding companies on several occasions to
ensure that the program included only the necessary options.
The Gibbs energy minimization program was written in Visual Fortran code for
optimal speed of the calculations. However, it was necessary to package the Gibbs
energy minimization program with a Windows interface so that the companies funding
this work could easily use it. This chapter describes the integration of the Fortran code
into a Visual Basic interface. The final program was named CSMGem.
8.1
Layout of CSMGem
The layout of CSMGem was a minor but very important step. We had available
174
8.2
minimization program. Later in the work, however, it was desired to create a graphical
175
user interface (GUI) for ease of use. Since the GUI capabilities of Visual Fortran are
quite limited, Visual Basic was used to develop the GUI. The Visual Fortran program
was then converted to a dynamic link library (.dll) so that the GUI could access it. The
parameters needed for the .dll file to perform a calculation are:
calculation,
T and/or P,
feed components,
feed composition,
feed component SRK parameters (Tc, Pc, Vc, , MW, and kij),
petroleum fraction properties (Tb, SG, MW), and
phases to account for (specified phase and fraction if needed).
Therefore, these parameters are determined in the GUI prior to calling the .dll file. Note
that default SRK parameters are used in the program (given in Appendix E). However,
the user has the option to change them if desired. The first 6 sets of parameters are
known from the users input. However, the last set, the phases to account for in the
calculation, is more detailed to determine and will be discussed later in this appendix.
8.3
Calculations to Perform
The main objective was to create a program that could calculate hydrate phase
equilibria. These types of calculations included: the hydrate formation T and P of the
most stable hydrate and phases present in equilibrium with hydrates at a given T and P
above the hydrate formation conditions. However, we decided to make CSMGem a more
versatile phase equilibria program and go beyond only the hydrate calculations.
The Gibbs energy minimization method allows for calculations of the formation
conditions for any phase (including the hydrate). It also allows for the calculation of
phases present at any T and P (whether hydrates are present or not). Therefore, we
included the option to perform all thermodynamic calculations with every phase and not
just the hydrate. The types of calculations that we included in CSMGem are:
176
1.
2.
3.
4.
5.
Each of these calculations is treated differently within the Visual Basic and Fortran code.
The rest of this chapter will discuss the steps taken to perform each calculation.
8.4
Name
V
Lhc
Aq
I
sI
sII
sH
NaCl
KCl
CaCl2
Key Components
all components but NaCl, KCl, and CaCl2
hydrocarbons and alcohols
water
water
water and hydrate former
water and hydrate former
water and sH hydrate former
NaCl
KCl
CaCl2
177
For example, for an ethane+water mixture, the following phases would be included in the
calculation: V, Lhc, Aq, I, sI, and sII. Before the Visual Basic front-end sends the feed
information to the Visual Fortran library, the initial phase selection is made to ensure that
all phases included can possibly form. Note that the user also has the option to turn off
any phase. For example, if the user does not want to include hydrates in a flash, she may
deselect the hydrate phases in another menu.
8.5
calculations for industrial use. When planning flow assurance strategies for a natural gas
pipeline, the injection rate of hydrate inhibitor is based on aqueous concentrations to
maintain the system outside the hydrate formation region. In performing this calculation,
the temperature or pressure is specified and the pressure or temperature at which a
hydrate phase first forms (i.e. H = 0) is calculated. It should be noted that this type of
calculation is iterative in that there is no thermodynamic expression to explicitly
determine the most stable hydrate structure.
The most common way to solve this problem is to use the method developed by
Parrish and Prausnitz (1972), in which the hydrate formation temperature or pressure is
determined for each hydrate.
formation P is the stable hydrate. An example of this can be seen in Figure 8.2.
178
100
Metastable sH
Metastable sII
sI
Pressure (bar)
P = 54 bar
methane
ethane
i-pentane
water
P = 17 bar
P = 13 bar
10
274
276
278
280
282
284
0.33
0.33
0.33
0.01
286
Temperature (K)
179
100
V-Lhc-sII
Pressure (bar)
V-Lhc-sI-sII
P = 44 bar
methane
ethane
i-pentane
water
0.33
0.33
0.33
0.01
V-Lhc-sI
Aq-V-Lhc
P = 13 bar
10
274
276
278
280
282
284
286
Temperature (K)
180
in this work, we found a way around this problem by only considering one hydrate phase
in the calculations at a time.
The approach we took is to assume that only one hydrate phase is present in the
system (whether metastable or not). This is done by turning off all other hydrate
phases. With this approach, the GEM essentially finds the formation T or P in a similar
manner as the Parrish and Prausnitz approach. Also, with only one hydrate phase to
account for, convergence is fairly rapid. We do decrease the solution time for the most
stable hydrate ordering the most likely hydrates that could form with a given feed gas
mixture. After the first hydrate formation T or P is determined, an internal check on the
fugacity of the other hydrates is made to determine if they are more or less stable than the
given hydrate. If they are less stable, then the hydrate formation T or P is not determined,
and vice versa.
8.6
Expansion Calculations
When a gas is expanded through a valve or turboexpander, the inlet T and P and
the outlet P is usually known. Therefore, the problem lies in calculating the outlet T.
This type of calculation can be performed for any number of phases in the system as long
as the enthalpy or entropy of the phase can be determined. The most common reason for
performing an expansion calculation is to determine whether condensation of the liquid
hydrocarbon phase will occur during the expansion. Therefore, in this work, expansion
calculations are performed only for the water-free feed. That is, the feed without aqueous
phase components (i.e. water, alcohols, salts). This simplifies the calculation procedure
dramatically and allows for swift convergence to a solution.
181
8.7
Plotting
Another option available in CSMGem is to create plots of a given calculation. In
this work, 3 types of plots are available: two-phase VLE envelope, specified phase
boundary, expansion. The specified phase boundary and expansion plots are calculated
by stepping off in pressure for a specified number of intervals. That is, the phase
boundary is calculated at equal pressure intervals from the lower specified pressure to the
upper specified pressure. The solutions at the previous two pressures are used as an
initial guess for the next pressure using a logarithmic extrapolation of the temperature
(i.e. assuming the phase boundary is exponential in pressure).
The two-phase VLE envelope is calculated using the water-free feed. The dew
points (i.e. Lhc = 0) are calculated at given pressures, with an initial pressure of 1 bar.
The pressure intervals are calculated using the previous solutions based on the curvature
of the envelope. The dew points are calculated at pressures ranging from 1 bar to the
critical pressure of the mixture.
calculated at given temperatures stepping off from the dew T at 1 bar to the critical T.
The method of Heidemann and Khalil (1980) is used to determine the critical point of the
mixture.
8.8
essential for some pipelines. Therefore, the characterization of a given crude oil based on
boiling temperature, specific gravity, and molecular weight was implemented into
CSMGem. In order to include undefined heavy components (petroleum fractions) in a
flash calculation, correlations were needed to describe these components in each phase.
Typical oils contain greater than 100 components, and so it is often necessary to lump
certain groups of these into petroleum fractions or pseudo-components. The properties of
182
the petroleum fractions are determined one of two ways: 1) from a distillation curve and
2) from gas chromatographic results that simulate the distillation curve.
8.8.1
determined as a function of the volume percent of the stock tank oil distilled. Figure 8.4
is a typical curve.
Temperature
Specific Gravity
FBP
IBP
200
0
10
20
30
40
50
60
70
80
90
0.6
100
% Distilled
183
Petroleum Fraction #1
Petroleum Fraction #2
SG1
SG2
NBP2
NBP1
amount
amount
200
0.6
0
% Distilled
amounts of the different molecular species in the oil. In addition, the specific gravity and
molecular weight of the stock tank oil are determined by simple laboratory measurements
(Notz 2002). The oil is usually split into ranges of boiling components. For example, all
components that boil off around that of n-heptane (carbon number equals 7) are lumped
into a C7 fraction. This is typically continued for carbon numbers up to C29, at which
point, the remainder of the oil is lumped into a C30+ fraction. The average specific
gravity, molecular weight, and relative amount are determined for each fraction by
correlation.
184
8.1
Equation 8.1 reproduces experimental values for molecular weight to within an average
error of 3.4 percent when MW < 300 and 4.7 percent for MW > 300 when tested against
635 data points (API, 1987a). Equation 8.1 is safe to use for 305 K < Tb < 1089 K, 70 <
MW < 700, and 0.63 < SG < 0.97.
If specific gravity and molecular weight are known Equation 8.1 can be
rearranged such that a successive substitution iterative method can be applied to solve for
the normal boiling point in K (Equation 8.2).
Tb ( K ) =
ln
MW
a3 SG
a1Tba5 SG a6
a2 + a4 SG
8.2
It was found that an initial guess for the normal boiling point of 500 K gives reasonable
convergence to the solution. The limitations of Equation 8.2 are the same as that for
Equation 8.1. However, the reliability of Equation 8.2 is not as good as Equation 8.1
(Spencer 2000). The constants for Equations 8.1 and 8.2 are given in the Table 8.2.
185
42.965354
2.097E-4
-7.78712
2.08476E-3
1.26007
4.98308
With the properties of the petroleum fractions, we then determine the needed parameters
for phase equilibria calculations.
8.9
aqueous phases, equation of state specific properties need to be determined for them in
each phase. In this work, correlations have been developed or taken from literature
sources to obtain these properties. All correlations are based on the normal boiling point,
specific gravity, and molecular weight of the petroleum fractions.
8.9.1 Vapor and Liquid Hydrocarbon Phases
To determine the fugacity of the petroleum fraction in the vapor and liquid
hydrocarbon phases, the critical temperature, critical pressure, and acentric factor are
needed. Equations 8.3 and 8.4 are API modifications (1987b, 1987c) of the correlations
developed by Riazi (1979) for the pseudo-critical temperature and pressure of petroleum
fractions, respectively.
Tc ( K ) = a1 exp [a2Tb + a3 SG + a4Tb SG ]Tba5 SG a6
8.3
8.4
186
Equations 8.3 and 8.4 reproduce pure hydrocarbon experimental values for critical
temperature and pressure to within average errors of 0.8 and 2.6 percent, respectively.
Note that the correlations were not evaluated using petroleum fraction data as
pseudocritical temperature and pressure are defined rather than measured. Equations 8.3
and 8.4 are safe to use for 300 K < Tb < 616 K, 70 < MW < 295, and 0.625 < SG < 1.025.
The constants for Equations 8.3 and 8.4 are given in Table 8.3.
Table 8.3 Constants for Equation 8.3 and 8.4
a1
a2
a3
a4
a5
a6
Equation 8.3
9.523276
-9.31446E-4
-0.54444
6.4791E-4
0.81067
0.53691
Equation 8.4
319584
-8.505E-3
-4.8014
5.74902E-3
-0.4844
4.0846
8.5
where, P* is the vapor pressure of the petroleum fraction at T = 0.7Tc. The vapor
pressure of a petroleum fraction can be determined via an iterative scheme developed by
API (1978a). The governing equations are
a X + a2
P* (bar ) = exp 1
,
a3 X + a4
Tb'
+ a1Tb'
T
X= b
, and
a2 + a3Tb'
Tb' = Tb a1 f ( KW a2 ) ln
8.6
8.7
P*
a3
8.8
187
where Tb is a corrected normal boiling point for the petroleum fraction, KW is the Watson
characterization factor defined as
1
KW
(1.8Tb ) 3 ,
=
8.9
SG
T a
f = b 1
a2
1
Tb < 366.5 K
366.5 K Tb 477.6 K .
8.10
Tb > 477.6 K
Equation 8.6 was obtained from Exxon Research and Engineering Company (1977) and
Equations 8.7 and 8.8 were adapted from Maxwell and Bonnell (1955). Equations 8.6
through 8.10 are applicable for narrow-boiling fractions, i.e., those having less than 10 K
difference in a true-boiling point distillation. The method is most reliable for vapor
pressures near atmospheric pressure; however, it has not been tested with petroleum
fraction data (only pure hydrocarbons). The constants for Equations 8.6 through 8.10 are
given in Table 8.4.
Table 8.4 Constants for Equation 8.6 through 8.10
X < 0.0013
a1
a2
a3
a4
6140.031
-8.213800
36.00
-0.989679
Eq. 8.6
0.0013 X
0.0022
5498.135
-7.363971
95.76
-0.972546
X > 0.0022
6624.327
-9.030525
43.00
-0.987672
Eq. 8.7
Eq. 8.8
Eq. 8.10
-5.1606E-4
748.1
-0.3861
--
1.0857362
12.0
1.01325
--
366.5
200.0
---
Figure 8.6 is the iterative routine suggested for the above procedure.
188
read Tb and SG
calculate KW and f
Tb = Tb
calculate X
calculate P*
calculate Tb
done
yes
T b = 0 ?
no
8.11
xa = a7 + a8 Ri + a9VG
where, Ri is the refractivity of the petroleum fraction and VG is a viscosity correlation.
The refractivity is given by
Ri = n
d
,
2
8.12
189
where, n is the refractive index and d is the liquid density (g/cc), both at 20 oC and 1
atmosphere. The viscosity correlation is given by
a1 + a2 SG + a3 ln v100,210
VG =
a1SG + a2 + a3 ln (V100,210 a4 )
a5 + a6 ln (V100,210 a4 )
MW 200
,
8.13
MW > 200
where, v100,210 is the kinematic viscosity in centistokes (cS) and V100,210 is the Saybolt
universal viscosity in Saybolt universal seconds, at either 100 oF (37.8 oC) or 210 oF (98.9
o
C), respectively. Equations 8.11 through 8.13 are safe to use for: Light Fractions - 78 <
MW < 214, 1.04 < Ri < 1.08, 0.57 < VG < 1.52, 0.02 < xp < 0.93, 0.02 < xn < 0.46, and
0.01 < xa < 0.93; Heavy Fractions - 233 < MW < 571, 1.04 < Ri < 1.06, 0.79 < VG < 0.98,
0.10 < xp < 0.81, 0.13 < xn < 0.64, and 0.00 < xa < 0.31. Equations 8.11 through 8.13
reproduce experimental values for to within average deviations for light fractions of 0.04
and 0.06 mole fraction and for heavy fractions of 0.02 and 0.04 mole fraction for xp and
xn, respectively. The constants for Equations 8.11 and 8.13 are given in Table 8.5.
Table 8.5 Constants for Equation 8.11 and 8.13
a1
a2
a3
a4
a5
a6
a7
a8
a9
Eq. 8.11
MW < 200 MW > 200
-13.359
2.5737
14.4591
1.0133
-1.41344
-3.573
23.9825
2.464
-23.333
-3.6701
0.81517
1.96312
-9.6235
-4.0377
8.8739
2.6568
0.59827
1.60988
The following correlations give the refractive index, n, and the liquid density, d,
190
n=
1 + 2I
,
1 I
8.14
d = a1MW a2 I a3 ,
8.15
8.16
where
Equations 8.14 through 8.16 are safe to use for 309 K < Tb < 811 K, 70 < MW < 600, and
0.63 < SG < 1.1.
Equation 8.14 to within an average error of 0.6% (MW<300) and 0.2% (MW>300). The
constants for Equations 8.15 and 8.16 are given in Table 8.6.
Table 8.6 Constants for Equation 8.15 and 8.16
Eq. 8.16
MW < 300 MW > 300
2.34348E-2 1.8429E-2
7.029E-4
1.16352E-3
2.468
5.144
-1.02672E-3 -5.9202E-4
5.72E-2
-0.407
-0.72
-3.333
Eq. 8.15
a1
a2
a3
a4
a5
a6
2.83085
0.03975
1.13543
----
The following correlation has been developed by Abbott et al. (1971) for the kinematic
viscosity, v, at 100 and/or 210 oF,
ln 100,210 = a1 + a2 KW + a3 KW2 + a4 ( o API ) + a5 ( o API ) + a6 KW ( o API ) +
2
2
a7 KW + a8 KW2 + a9 ( o API ) + a10 ( o API ) + a11 KW ( o API ) , 8.17
where the Watson characterization parameter, KW, is defined by Equation 8.9, and the
API gravity, oAPI, is defined as
o
API =
141.5
131.5 .
SG
8.18
191
Equation 8.17 reproduces experimental data for petroleum fractions with an average error
of 16.7% and 15.5% for v100 and v210, respectively. Care should be taken when KW 10
and oAPI 0.0.
1 + a4T
VT = (1 + a1 (T a2 )) a3T +
2
3
a5 + a6T + a7T + a8T
8.19
In the range of 294 K to 422 K, the average deviation between calculated and
experimental Saybolt universal viscosities is 0.17% with a maximum deviation of 0.5%.
The constants for Equations 8.17 and 8.19 are given in Table 8.7.
Table 8.7 Constants for Equation 8.17 and 8.19
Eq. 8.17
T = 100 F
T = 210 oF
10.11689
-1.067557
-4.483893
0
0.2940171
0
0
-0.3834541
7.5131E-4
1.1822554E-3
-2.722715E-2 -1.954883E-2
0
0.18495515
0.395163
0
25.315311
2.8759058
0.218898
0.455175
-1.980725
0
50.3642
26.786
-4.78231
-2.6296
o
a1
a2
a3
a4
a5
a6
a7
a8
a9
a10
a11
a12
a13
Eq. 8.19
1.098E-4
310.928
4.6324
3.264E-2
3.9302E-2
2.627E-3
2.397E-4
1.646E-5
------
With the molecular type of the petroleum fraction defined, the next step was to determine
the interaction of the petroleum fraction with other components in the feed.
The
192
type = a1 + a2 ln 1 3
MW
8.20
Equation 8.20 was developed for pure paraffins, naphthenes, and aromatics.
The
paraffins
8.2551
-0.7992
-180.1
naphthenes
8.502
-0.81122
-100
aromatics
8.4294
0.1175
-136.7
Figure 8.7 is a plot of the solubility parameter versus molecular weight with the
predictions given by Equation 8.20. As can be seen, the correlation does best at high
molecular weights, which is the case for petroleum fractions. The solubility parameters
for the various components in Figure 8.7 are from the API Data Book (1978b).
Paraffins
Naphthenes
Aromatics
1/2
8.8
8.6
8.4
8.2
8
7.8
7.6
7.4
7.2
7
100
1000
Molecular Weight
193
The solubility parameter of the petroleum fractions will be calculated using a mole
fraction average of Equation 8.20 for each molecular type. That is,
PF = x p p + xn n + xa a
8.21
The interaction parameter, kij, for the petroleum fraction with other components in the
feed can then be determined using API correlations (1999) for CH4, CO2, N2, and H2S.
For the interaction between the petroleum fraction and water, the following correlation
has been developed in this work,
kij = a1 PF w exp a2 + a3 PF w
8.22
8.31514E-2
-5.40178
0.29890
The constants in Equation 8.22 were regressed using the water-hydrocarbon interaction
parameters from this work. Figure 8.8 gives the predictions of Equation 8.22 versus pure
paraffins, naphthenes, and aromatics.
194
0.6
kij (water-component)
0.55
Paraffins
Naphthenes
Aromatics
0.5
0.45
0.4
0.35
0.3
13.5
14
14.5
15
15.5
16
|PF-w|
8.23
195
With a given property defined for each molecular type of compound, a fraction average
of the property will be determined using the molecular type analysis described
previously. Therefore, a given property will be determined via the following equation.
Property = x p Property p + xn Property n + xa Property a .
8.24
Table 8.10 lists the constants for each formation property and type of compound.
Table 8.10 Constants for Equation 8.23 for ideal gas formation properties
gi0o (J/mol)
paraffin
-50903.6
589.509
a1
a2
naphthene
-16754.0
580.183
hi0o (J/mol)
aromatic
66070.2
590.298
paraffin
-40595.2
-1468.13
naphthene
12000.0
-1507.21
aromatic
185180
-1477.15
= c p1 + c p 2T + c p3T 2 + c p 4T 3
8.25
The constants in Equation 8.25 are determined for the petroleum fraction with an
equation similar to that of Equation 8.23. The constants are given in Table 8.11.
Table 8.11 Constants for Equation 8.23 for ideal gas heat capacity parameters
cp1
naphthene
-5.460852
-0.017476
a1
a2
paraffin
1.35677
-0.017476
aromatic
-2.44248
-0.017476
a1
a2
Figures 8.9 through 8.14 are the predictions of Equation 8.23 for each property described
above. The data points in the figures are for pure hydrocarbons of each molecular type.
196
300000
250000
Aromatics
Naphthenes
Paraffins
gi0o (J/mol)
200000
150000
100000
50000
0
-50000
-100000
10
100
1000
Molecular Weight
Figure 8.9 Gibbs energy of formation for ideal gas at 298.15 K and 1 bar
200000
Aromatics
Naphthenes
Paraffins
100000
hi0o (J/mol)
0
-100000
-200000
-300000
-400000
-500000
10
100
Molecular Weight
Figure 8.10 Enthalpy of formation for ideal gas at 298.15 K and 1 bar
1000
197
Paraffins
Aromatics
Naphthenes
cp1
-5
-10
-15
-20
-25
10
100
1000
Molecular Weight
40
Naphthenes
Aromatics
Paraffins
35
30
cp2
25
20
15
10
5
0
10
100
Molecular Weight
1000
198
10
0
-10
-20
cp3
Paraffins
Aromatics
Naphthenes
-30
-40
-50
-60
10
100
1000
Molecular Weight
160
Naphthenes
Aromatics
Paraffins
140
120
100
cp4
80
60
40
20
0
-20
10
100
Molecular Weight
1000
199
With the ideal gas properties of the petroleum fraction defined, expansion calculations
can be performed as well as fugacity calculations in the aqueous phase.
8.9.3 Aqueous Phase
In order to determine the fugacity of a petroleum fraction in the water phase, the
Gibbs energy and enthalpy of solution in the standard state at reference conditions are
required. As with the ideal gas properties, the water phase property of the petroleum
fraction will be determined via a fraction average using the molecular type and Equation
8.24. Correlations with the general form of Equation 8.23 describe these properties as
functions of molecular weight for each molecular type. Table 8.12 gives the constants for
the formation properties in the standard state.
Table 8.12 Constants for Equation 8.23 for formation properties in aqueous
phase
gi0 * (J/mol)
a1
a2
paraffin
-39605.3
669.68
naphthene
-15259.3
669.68
hi0 * (J/mol)
aromatic
60651.2
669.68
paraffin
-48084.1
-1754.78
naphthene
8500.09
-1754.78
aromatic
177519
-1754.78
The Born constant, , is also a required parameter to determine for the petroleum
fraction. In the same form as Equation 8.23, the constants for the Born constant are given
in Table 8.13.
Table 8.13 Constants for Equation 8.23 for Born constant
a1
a2
paraffin
-84820.0
-2875.8
naphthene
-92970.0
-2875.8
aromatic
143269
-2981.7
200
The partial molar volume of and heat capacity are also required in the standard state.
Equations 8.26 and 8.27 are the forms of the partial molar volume and heat capacity,
respectively,
vi* = v1 +
cPi* = c1 +
v2
v 1
+ v3 + 4
2
and
+P
+ P T P T
c2
(T )
3.7
+ P
c3 ( P P0 ) + c4 ln
+ TX ,
(T )
+ P0
2T
3.6
where, = 228 K, = 2600 bar, and is the dielectric constant of pure water. The
constants for each equation, vi and ci, for a petroleum fraction can be found via Equation
8.23 and the following constants given in Table 8.14. They were each regressed to the
parameters for known components in this work.
Table 8.14 Constants for Equation 8.23 for partial molar volume and heat
capacity parameters
a1
a2
paraffin
1.70404
6.34695E-2
a1
a2
paraffin
20.4615
-0.608039
a1
a2
Paraffin
121.810
3.50906
v1 (J/mol-bar)
naphthene
aromatic
1.86816
0.457662
6.34695E-2 6.34695E-2
v3 (J-K/mol-bar)
naphthene
aromatic
19.1737
41.0918
-0.608039
0.105082
c1 (J/mol-K)
naphthene
128.228
3.50906
aromatic
-84.6320
5.17567
paraffin
905.770
154.941
v2 (J/mol)
naphthene
1213.13
154.941
paraffin
-120032
-640.126
v4 (J-K/mol)
naphthene aromatic
-121147
-106863
-640.126
-518.628
paraffin
254210
12733.4
c2 (J-K/mol)
naphthene aromatic
301113
-273809
12733.4
4274.36
aromatic
-3559.96
133.513
201
Figures 8.15 through 8.22 are the predictions of Equation 8.23 for each property
described above. The data points in the figures are for pure hydrocarbons of each
molecular type.
800000
700000
Aromatics
Naphthenes
Paraffins
600000
gi0* (J/mol)
500000
400000
300000
200000
100000
0
-100000
10
100
1000
Molecular Weight
202
500000
hi0* (J/mol)
-500000
Aromatics
Naphthenes
Paraffins
-1000000
-1500000
-2000000
10
100
1000
Molecular Weight
Naphthenes
Paraffins
Aromatics
25
v1 (J/mol-bar)
20
15
10
0
10
100
Molecular Weight
1000
203
60000
Naphthenes
Paraffins
Aromatics
50000
v2 (J/mol)
40000
30000
20000
10000
0
10
100
1000
Molecular Weight
v3 (J-K/mol-bar)
50
0
-50
Aromatics
Naphthenes
Paraffins
-100
-150
-200
-250
-300
10
100
Molecular Weight
1000
204
-100000
-150000
Aromatics
Naphthenes
Paraffins
v4 (J-K/mol)
-200000
-250000
-300000
-350000
-400000
10
100
1000
Molecular Weight
Aromatics
Naphthenes
Paraffins
c1 (J/mol-K)
1400
1200
1000
800
600
400
200
0
10
100
Molecular Weight
1000
205
5000000
4000000
Aromatics
Naphthenes
Paraffins
c2 (J-K/mol)
3000000
2000000
1000000
-1000000
10
100
1000
Molecular Weight
207
In the beginning stages of this work, an analysis of the methane, ethane, and
propane systems was performed. Since these three components make up nearly 97 mole
percent of a typical natural gas mixture, the hydrate phase behavior of a natural gas
mixture in contact with water will likely be approximated by that of a simple mixture of
methane, ethane, and propane in contact with water. For this reason, we generated phase
diagrams with all possible combinations of methane, ethane, and propane at one
isotherm, 277.6 K.
This chapter will relate the phase equilibria of pure and binary gas mixtures of
methane, ethane, and propane in contact with water to the phase equilibria of the ternary
gas mixture in contact with water at 277.6 K, the most common temperature a natural gas
will encounter in a pipeline on the ocean floor at water depths beyond 600 meters.
Discussion will be given for some of the interesting systems. It should be noted that this
work was performed prior to the development of the new hydrate model. However, the
predictions presented in this chapter are approximately the same as predictions with the
new model.
9.1
208
natural gas is mainly comprised of these components, the phase equilibria of each of the
first three normal paraffins with water will be quite different from that of a natural gas
with water. Therefore, the evaluation of phase equilibria of a pure gas hydrate is for
fundamental understanding.
The hydrate phase equilibria of pure gases with water form the basis for
understanding the phase equilibria of water with binary and ternary mixtures of gases.
Figure 9.1 is the pressure versus temperature phase diagram for the methane+water
system. Note that excess water is present so that, as hydrates form, all gas is incorporated
into the hydrate phase. The phase equilibria of methane hydrates is well predicted as can
be seen by a comparison of the prediction and data in Figure 9.1. It is important to note
for later discussions that the predicted hydrate formation pressure for methane hydrates at
277.6 K is 40.6 bar.
Pressure (bar)
1000
100
Aq-sI
Aq-V
10
273
275
277
279
281
283
285
287
289
Temperature (K)
291
209
Figure 9.2 is the pressure versus temperature phase diagram for the ethane+water
system. Again, the predictions represent the data well, especially along the Aq-sI-V
three-phase equilibrium line. Note that the Aq-sI-V line intersects the Aq-V-Lhc line at
287.8 K and 35 bar. Due to differences in the volume and enthalpy of the vapor and
liquid hydrocarbon, the three-phase hydrate formation line changes from Aq-sI-V to AqsI-Lhc. Mathematically, this change in the three-phase hydrate formation line can be
reconsidered by applying the Clapeyron equation to the phase change (Jager and Sloan,
2000).
Note that the hydrate formation pressure for ethane hydrates at 277.6 K is
Pressure (bar)
1000
Aq-sI
Aq-Lhc
100
10
Aq-V
1
273
283
293
303
313
323
Temperature (K)
210
equilibrium line. However, note that the data are quite scattered along the Aq-sII-Lhc line
due to difficulties in measuring hydrate equilibria with three relatively incompressible
phases. As with the ethane+water system in Figure 9.2, the slope of the three-phase
hydrate formation line changes drastically when the Aq-sII-V line intersects the Aq-V-Lhc
line. In fact, the Aq-sII-Lhc line is nearly vertical but comes back to lower temperature at
high pressure. The predictions suggest "pseudo-retrograde" phenomena for the propane
hydrates in which the sII hydrate is predicted to dissociate by pressurization at a constant
temperature.
1000
Pressure (bar)
100
10
Aq-Lhc
Aq-sII
Aq-V
1
273
274
275
276
277
Temperature (K)
278
279
211
phenomena can be found in Section 9.2.4. Note that the hydrate formation pressure of
propane hydrates along the Aq-sII-V line at 277.6 K is predicted to be 4.3 bar.
9.2
212
40.65
41
36
Aq-sI-V
40.61
40.59
31
Pressure (bar)
Aq-sI-sII
Aq-sI-sII
40.63
Aq-sII-V
Aq-V
40.57
40.55
0.9992
26
21
0.9994
0.9996
0.9998
16
11
Aq-sII-V
Aq-sII
Aq-V
1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
213
(Figure 9.5). Since both pure components form sI hydrates, the hydrate structure is fixed
at the composition extremes.
214
incorrectly measured gas composition by Deaton and Frost. However, Hendriks et al.
(1996) reported predictions of sII forming in the methane+ethane+water system and this
was later experimentally confirmed by Subramanian et al. (2000a, 2000b) via Raman and
NMR spectroscopy. The experimentally determined composition in which the incipient
hydrate structure transitions from sI to sII is between 0.736 0.014 mole percent
methane in the vapor at a temperature of 274.2 K. Ballard and Sloan (2000) indicated
that, by fitting model parameters to incipient hydrate equilibrium data as well as this
structural transition point, the predictions fit the methane+ethane+water data much better
than if only sI was assumed as the incipient hydrate structure over the entire composition
range.
What follows is a qualitative discussion of the effects of temperature, pressure,
and composition on the methane+ethane+water system in detail, and will help to better
understand the abbreviated discussion of the hydrogen sulfide+difluoroethane+water
system by van der Waals and Platteeuw. The phase diagrams generated for this system
contain several unexpected predictions for such a simple gas mixture.
Figure 9.6 is the pseudo-binary pressure versus water-free composition diagram
for the methane+ethane+water system at a temperature of 277.6 K. In the diagram, pure
ethane and pure methane both form sI hydrates in the presence of water at pressures of
8.2 bar and 40.6 bar, respectively. However, it is interesting to note that between the
compositions of 0.74 and 0.994 mole fraction methane, sII hydrates form at the incipient
formation pressure.
spectroscopy (Subramanian et al. 2000a, 2000b) and XRD (Huo, 2002). Similar to the
methane+propane+water system, only a small amount of ethane added to pure methane
will form sII hydrates.
215
41
36
Aq-sII
Aq-sI-sII
31
Pressure (bar)
Aq-sI
Aq-sI-sII
Aq-sI-V
26
Aq-sII-V
Aq-sI
21
16
Aq-sI-V
11
Aq-V
6
1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
composition of 50 mole percent methane and 50 mole percent ethane is fed to a cell at
277.6 K, sI hydrates will initially form at approximately 11 bar. Ethane, being the larger
guest, will preferentially stabilize the large cages in the sI hydrate lattice so that the
water-free composition of the hydrate will contain about 24 mole percent methane and 76
mole percent ethane (Figure 9.6).
216
temperature dependence as well. Table 9.1 shows the predicted effect of temperature on
incipient hydrate structure for a water-free gas mixture of 73 mole percent methane and
27 mole percent ethane.
Table 9.1 Effect of temperature on hydrate structure in the methane(0.73)+
ethane(0.27)+water(excess) system
Temperature (K)
273 - 275
275-292
292-301
> 301
As temperature increases, the incipient hydrate structure changes from sII to sI to sII and
back to sI. This change in structure can also be seen in the pressure versus temperature
diagram in Figure 9.7.
217
100
Pressure (bar)
Aq-sII
Aq-sII-V
Aq-sI-V
Aq-V
10
273
275
277
279
281
283
285
287
289
291
293
Temperature (K)
In many practical
situations such as flow assurance in natural gas pipelines and hydrates in oceanic and
permafrost regions it is essential to know what phases are present. Subramanian et al.
(2000a, 2000b) discuss the practical applications of these predictions.
218
determined for the remaining phases in the diagram just as they would for a traditional
binary P-x diagram.
Figure 9.8 is the pressure versus water-free composition phase diagram at 288.05
K (which is very similar to Figure 9.6). The lower and upper structural transition points
are at 76 and 98.2 mole percent methane, respectively. However, a difference in this
diagram is that there is an aqueous-sI hydrate-vapor (Aq-sI-V) negative homogeneous
azeotrope at 14 mole percent methane. At the azeotrope, the water-free composition of
the hydrate and the vapor are the same (not of the aqueous phase, however). This
phenomenon has been discussed by Holder and Grigoriou (1980) who found that the
dissociation pressures of a 10 mole percent methane and 90 mole percent ethane mixture
fell below the dissociation pressure of pure ethane at temperatures greater than 281 K.
Azeotropes in hydrate systems are not uncommon and can be attributed to increased
stability of the hydrate phase as, in this case, methane is added to pure ethane.
219
120
Aq-sII
Pressure (bar)
100
Aq-sI-sII
Aq-sI-V
Aq-sI
Aq-sI-sII
Aq-sII-V
Aq-sI
80
60
Negative homogeneous
azeotrope
40
Aq-sI-V
Aq-sI-V
20
Aq-V
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
9.1
This means that on a water-free basis, ethane vapor with water will condense before it
forms sI hydrates.
220
At 288.5 K, pure ethane is predicted to flash before hydrates are formed as seen in
Figure 9.9. On the whole, Figure 9.9 is quite similar to Figure 9.8, but if the pressure and
composition axes are focused near pure ethane it is quite different (inset diagram of
Figure 9.9). The intersection of the Aq-V-Lhc, Aq-sI-V, and Aq-sI-Lhc envelopes results
in a four-phase quadruple line pressure at 36.1 atmospheres.
Aq-sI-sII
Aq-sI-V
Aq-sI
120
Aq-sII
Aq-sI
100
Aq-sI-Lhc
Aq-sI-sII
Pressure (bar)
Aq-Lhc
Aq-sII-V
Aq-sI-V
80
Quadruple Line
Aq-V-Lhc
60
0.01
0.02
0.03
Aq-V
0.04
0.05
Aq-sI
0.06
Aq-sI-V
40
Negative homogeneous
azeotrope
20
Aq-V
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
221
Because, on a water-free basis, pure methane is above its critical temperature, the
Aq-V-Lhc envelope seen in Figure 9.9 will never extend to the methane axis. This can be
seen visually in the pressure versus composition diagram at 292.9 K (Figure 9.10). The
hydrate phase envelopes have "broken away" from the Aq-V-Lhc envelope at this
temperature leaving an unusual phase diagram. The Aq-V-Lhc envelope extends into the
binary range to about 20 mole percent methane.
250
Pressure (bar)
Aq-sII
Aq-sI-sII
Aq-sI
200
Aq-sI-Lhc
150
Aq-sI-sII
300
Aq-sII-V
Negative heterogeneous
azeotrope
Aq-sI
Aq-sI-V
Aq-sI-V
100
Aq-Lhc
50
Aq-V-Lhc
Aq-V
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
222
structural transition points at this temperature are at 72.3 and 96.3 mole percent methane,
respectively.
To this point, the more interesting phenomena have occurred near the pure ethane
axis (i.e. azeotropes and affect of vapor pressure). In Figure 9.11, a pressure versus
composition diagram at 296.65 K, the more interesting phenomena occur near the pure
methane axis. The lower and upper structural transition points at this temperature are at
68.5 and 94 mole percent methane, respectively. Figure 9.11 is essentially the same as
Figure 9.10; however, the inset diagram in Figure 9.11 focuses on the phase space near
the lower structural transition point. The formation of a negative homogeneous azeotrope
of Aq-sII-V occurs at 68.86 mole percent methane and 205.7 bar. Note the highly
expanded composition scale in the inset. Because of the small composition range over
which the azeotrope occurs, experimental confirmation is unlikely.
Aq-sII
600
Aq-sI-sII
Aq-sII-V
500
Aq-sII-V
Aq-sI-V
Aq-V
0.683
0.685
0.687
0.689
0.691
Aq-sI-Lhc
300
Aq-sI
Aq-sII
Aq-sI-sII
0.681
Aq-sI-sII
400
Pressure (bar)
Aq-sI
Negative homogeneous
azeotrope
Aq-sI-V
Aq-sII-V
200
Negative heterogeneous
azeotrope
Aq-Lhc
100
Aq-sI-V
Aq-V
Aq-V-Lhc
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
223
1200
Homogeneous
peritectic
1000
Aq-sI-sII
Aq-sI-sII
Aq-sI
Aq-sI-V
Aq-sII
Aq-sII-V
Pressure (bar)
Aq-sI
800
Aq-sI-V
Aq-sI-Lhc
0.7
600
0.72
0.74
Aq-V
0.76
0.78
0.8
0.82
0.84
Aq-sI-V
400
Aq-sI-V
Aq-Lhc
Negative heterogeneous
azeotrope
200
Aq-V
Aq-V-Lhc
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
224
transition point, the mole fraction of methane in sI hydrate is more than that in sII
hydrate. The significance of the peritectic is that this is the composition and pressure at
which the mole fraction of methane is the same in both sI and sII hydrates.
homogeneous peritectic also occurs in Figure 9.6 and Figures 9.8 through 9.11. The
peritectic occurs at a higher pressure as the temperature is lowered and therefore, by
including them in these figures, the pressure scale would be so inflated that the lower
portion of the hydrate phase diagrams would be indistinguishable.
The predicted maximum temperature at which sII hydrates can occur in the
methane+ethane+water system is predicted to be 303.1 K.
1200
1000
Pressure (bar)
Aq-sI
800
Aq-sI-Lhc
Negative heterogeneous
azeotrope
600
Aq-sI-V
400
Aq-Lhc
200
Aq-V-Lhc
Aq-V
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
225
The negative
226
12
10
Aq-sI-sII
Pressure (bar)
Aq-sII
Aq-sI-V
Aq-sII-V
4
Aq-V
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
227
Pressure (bar)
Aq-sII-Lhc
Aq-Lhc
Aq-sII
15
Aq-sI-Lhc
Aq-V-Lhc
10
Aq-sI-V
Aq-sII-V
Aq-V
Holder and Hand, 1982
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
228
phase rule there is one degree of freedom (3 components, 4 phases), namely temperature,
which is set at 277.6 K. This point of intersection creates a four-phase line, Aq-sII-V-Lhc,
in the pseudo-P-x diagram. Therefore, the pressure at which the quadruple line occurs in
Figure 9.15 is unique. That is, if pressure is increased, one of the phases must disappear.
In this case, the sII phase dissociates and an Aq-V-Lhc region remains.
The validity of the predictions in Figure 9.15 can be shown with a comparison of
the data taken by Holder and Hand (1982) for this system. The sI and sII hydrate
formation data points all compare quite well with the predictions with the exception of
the point at 0.66 mole fraction ethane. Holder and Hand state that the data point at 0.66
mole fraction is sII but note that it could be at Aq-sII-V-Lhc conditions. The predictions
in Figure 9.15 support their observation of possible 4-phase conditions but suggest that
the data point may be at metastable Aq-sI-V-Lhc conditions.
To test the predictions, experiments were carried out at the Technical University
of Delft (TUD) (Ballard et al., 2001). In this work, the pressure versus temperature phase
diagram was generated using the model and then confirmed by experimental data. Figure
9.16 is the pressure versus temperature diagram for a 30/70 mixture of ethane and
propane in contact with excess water.
229
21
17
Pressure (bar)
15
Aq-sI-sII
13
Aq-sII-Lhc
19
Aq-Lhc
11
Aq-V-Lhc
Aq-sII-V
7
5
Aq-V
3
1
276
276.5
277
277.5
278
278.5
279
279.5
280
Temperature (K)
However, note that the Aq-sII-Lhc data deviate 0.2 K from the
predictions.
We have made predictions for this system using the new hydrate model and
parameters. Figure 9.17 is the phase diagram for the same system as in Figure 9.16 with
the exception of the model used to make the predictions.
230
21
Aq-sII-Lhc
19
17
Pressure (bar)
15
Aq-sI-sII
Aq-Lhc
13
11
9
Aq-V-Lhc
Aq-sII-V
7
5
Aq-V
3
1
276
276.5
277
277.5
278
278.5
279
279.5
280
Temperature (K)
difficulty modeling the pseudo-retrograde region, it could be concluded that the data is in
error along the Aq-sII-V-Lhc line.
Mooijer (2002) has recently reported data for the system ethane(0.5)+propane
(0.5)+water(excess).
comparing the data.
231
100
Aq-sI-Lhc
Pressure (bar)
Aq-sI-sII
Aq-Lhc
Aq-V-Lhc
10
Aq-sII-V
sI Hydrate Dissociation
sII Hydrate Dissociation
Dew and Bubble Points
CSMGem
Aq-V
Data by Mooijer, 2002
1
276
277
278
279
280
281
Temperature (K)
282
283
284
232
9.3
consider phase equilibria of the ternary gas mixture with water. Pseudo-ternary phase
diagrams were created at a temperature of 277.6 K and pressures ranging from 10 to 45
bar. The pseudo-ternary phase diagrams are similar to true ternary phase diagrams except
that water is in excess and therefore all compositions are given on a water-free basis. The
pseudo-ternary phase diagrams are composites of the phase diagrams discussed earlier: PT diagrams for the pure hydrates and pseudo-P-x diagrams for the binary hydrates. That
is, the corners represent the intersection of an isotherm and isobar in the pure hydrate P-T
diagrams while the edges represent an isobar in the pseudo-P-x phase diagrams at 277.6
K.
Figure 9.19 is a pseudo-ternary phase diagram for the methane+ethane+propane+
water system at a temperature and pressure of 277.6 K and 10.13 bar, respectively. An
example of how this diagram is related to the previous diagrams is given. The ethanepropane edge of the phase diagram in Figure 9.19 can be directly compared to the
pseudo-P-x phase diagram for the ethane+propane+water system in Figure 9.15 at a
pressure of 10.13 bar. At 10.13 bar in Figure 9.15, the composition range for the Aq-sII
phase region is between 0 and 0.16 mole fraction ethane. This is the same composition
range for the Aq-sII phase region on the ethane-propane edge of Figure 9.19. Similar
comparisons can be made with each edge of the pseudo-ternary phase diagram and the
corresponding pseudo-P-x phase diagram.
233
propane
0
1
0.1
0.9
0.2
0.8
0.7
0.3
0.4
0.6
Aq-sII
0.5
0.5
0.6
0.4
0.7
0.3
Aq-sII-V
0.8
0.2
0.9
0.1
Aq-V
Aq-sI-V
1
ethane
0
0.1
0.2
0
methane
0.3
0.4
0.5
0.6
0.7
0.8
0.9
234
pseudo-ternary phase diagram with excess water throughout, as is the case of the other
diagrams, the composition of water in any of the phases cannot be determined.
The predicted phase diagram in Figure 9.19 indicates that sII hydrate is the
predominate hydrate that forms.
composition range. In Figure 9.19 four major phase regions appear from the top to the
bottom of the diagram: Aq-sII, Aq-sII-V, Aq-V, and Aq-sI-V. Three of the four phase
regions contain hydrates and encompass approximately eighty-percent of the overall
phase diagram. In other words at a temperature and pressure of 277.6 K and 10.13 bar,
respectively, the likelihood of hydrate formation is large given all possible mixture
compositions.
As pressure is increased, the likelihood of hydrate formation will be greater.
Figure 9.20 is the pseudo-ternary phase diagram for the methane+ethane+propane+water
system at 277.6 K and 11.04 bar. There are still the same four phase regions present as in
Figure 9.19. However, the Aq-sII-V and Aq-sI-V phase regions have expanded in size
and are nearly intersecting.
235
propane
0
1
0.1
0.9
0.2
0.8
0.7
0.3
0.4
0.6
Aq-sII
0.5
0.5
0.6
0.4
0.7
0.3
Aq-V
Aq-sII-V
0.8
0.9
0.2
0.1
Aq-sI-V
Aq-V
1
ethane
0
0.1
0.2
0.3
0.4
0.5
0
methane
0.6
0.7
0.8
0.9
236
propane
0
1
0.1
0.9
0.2
0.8
0.7
0.3
0.4
0.6
Aq-sII
0.5
0.5
Aq-sII-V
0.6
0.4
Aq-sII-V
0.7
0.3
Aq-V
0.8
0.2
Aq-sI-V
Aq-sI-V
1
ethane
0
0.1
0.2
Aq-sII-V
Aq-sI-sII-V
0.9
0.3
0.1
Aq-V
0.4
0.5
0.6
0
methane
0.7
0.8
0.9
composition puts the system in a four-phase region, the composition of the four phases is
the same anywhere within that region. For example, if the overall water-free composition
(CH4, C2H6, C3H8) of the feed is (0.25, 0.65, 0.10), the system is in the lower four-phase
region of Figure 9.21. The water-free composition of the vapor phase is (0.30, 0.59,
0.11), the sI phase is (0.17, 0.83, 0.00), and the sII phase is (0.42, 0.24, 0.34). If the
overall water-free composition puts the system anywhere within this four-phase region,
237
the composition of each of the phases will be the same, however the amounts of each
phase will change accordingly.
The pseudo-retrograde hydrate phenomenon observed in the ethane+propane+
water system in Figure 9.15 occurs on the pseudo-ternary phase diagram at a pressure of
11.45 bar in Figure 9.22.
propane
0
1
0.1
0.9
Aq-sII-L hc 0.2
Aq-L hc
Aq-sII-V-L hc
Aq-V-L hc
0.8
0.7
0.3
Aq-sII
0.4
0.6
Aq-sII-V
0.5
0.5
0.6
0.4
0.7
0.3
Aq-sI-sII
Aq-V
0.8
-V
0.1
Aq-sI-V
1
ethane
0
Aq-sII-V
Aq
- sI
0.9
0.2
Aq-sI-sII-V
0.1
0.2
0.3
0.4
Aq-V
0.5
0
methane
0.6
0.7
0.8
0.9
238
ethane-propane edge.
239
propane
0
1
0.1
0.9
0.2
0.8
Aq-sII-L hc
Aq-L hc
0.7
0.3
0.4
0.6
Aq-sII
0.5
0.5
Aq-sI-sII-L hc
0.6
0.4
Aq-sI-L hc 0.7
0.3
Aq-sI-sII
Aq-sI-V
0.8
0.9
0.2
Aq-V
Aq-sII-V
Aq-sI-sII-V
0
methane
1
ethane
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
240
sI-V phase region extends into the pseudo-ternary phase diagram from the pseudo-P-x
methane-ethane edge. At 15.2 bar, the pseudo-P-x in Figure 9.6 clearly shows an Aq-sIV region. However, as pressure is increased to 16.72 bar, the Aq-sI-V phase region is
replaced with Aq-sI-sII and Aq-sII-V phase regions. Figure 9.24 shows the impact of the
pseudo-P-x phase diagram on the pseudo-ternary diagram. Near the methane-ethane
edge, the four-phase region, Aq-sI-sII-V, is no longer present.
propane
0
1
0.1
0.2
Aq-sII-L hc
Aq-L hc
Aq-sI-sII-L hc
0.9
0.8
0.7
0.3
0.4
0.6
Aq-sII
0.5
0.5
0.6
Aq-sI-L hc
0.4
0.7
0.3
Aq-sI-sII
0.8
0.2
Aq-sII-V
0.9
Aq-V
0
methane
1
ethane
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
241
essentially incompressible and therefore changes very little with pressure. The vertical
lines in the pseudo-binary phase diagrams are representative of this. The only phase
regions in Figure 9.24 that have compressible phases (i.e. vapor) are near the pure
methane corner. Therefore, as pressure is increased, the only significant change in the
pseudo-ternary phase diagram will be near that corner. Figure 9.25 is the pseudo-ternary
phase diagram at 39.01 bar.
propane
0
1
0.1
Aq-sII-L hc
Aq-L hc
0.9
0.2
0.8
0.7
0.3
Aq-sI-sII-L hc 0.4
0.6
0.5
0.5
0.6
0.4
Aq-sII
0.7
0.3
Aq-sI-sII
0.8
Aq-sI-L hc
0.2
Aq-V
0.9
Aq-sII-V
0
methane
1
ethane
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
242
As the pressure is increased beyond 39.11 bar, which is the pressure at which an
Aq-sI-V phase region appears in the pseudo-P-x phase diagram for the methane+
ethane+water system, another four-phase region, Aq-sI-sII-V, appears in Figure 9.26 (P =
39.52 bar). This four-phase region appears as the result of the Aq-sI-V and Aq-sII-V
phase regions intersecting. Since only a small fraction of the phase diagram is different
from Figure 9.25, only the phase regions near high methane concentrations are shown in
Figure 9.26.
propane
0.08
Aq-sII
0.05
Aq-sII-V
Aq-sI-V
Aq-V
Aq-sI-sII
Aq-sI-sII
ethane
0.9
0.95
-V
methane
1
243
contact with water is contaminated with any mixture of ethane and propane by as little as
0.005 mole fraction in the vapor, hydrates will form.
The highest pressure in which an Aq-sI-V phase region is present is 40.62 bar and
it last occurs in the methane+propane+water pseudo-binary phase diagram in Figure 9.4.
Therefore at pressures higher than 40.62 bar, there will no longer be a Aq-sI-sII-V fourphase region. Figure 9.27 is the pseudo-ternary phase diagram at 277.6 K and 45.6 bar.
propane
0
1
0.1
Aq-sII-L hc
0.9
0.2
0.8
0.7
Aq-L hc 0.3
Aq-sI-sII-L hc
0.4
0.6
0.5
0.5
0.6
0.4
Aq-sII
0.7
0.3
Aq-sI-sII
0.8
Aq-sI-L hc
0.2
0.9
Aq-sI-sII
0.1
0
methane
1
ethane
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
percent of the diagram and sI hydrates in over sixty percent of the diagram. Because all
244
phase regions present in this figure are incompressible, the phase diagrams at higher
pressures do not change significantly. Therefore, we only give predictions at pressures
up to 45.6 bar.
9.4
Industrial Implications
Determining the stable hydrate structure at a given temperature, pressure, and
composition is not a simple task, even for such a simple systems as the ones discussed
here. The fact that such basic mixtures of methane, ethane, propane, and water exhibit
such complex phase behavior leads us to believe that industrial mixtures of ternary and
multi-component gases with water will exhibit even more complex behavior.
Spectroscopic methods are candidates to observe such complex systems because, as
discussed earlier, pressure and temperature measurements of the incipient hydrate
structure are not enough.
Experimental work is required to confirm predictions for the majority of these
systems at temperatures and pressures above the incipient conditions, and techniques
such as XRD, Raman, and NMR are well suited to do this. Spectroscopic measurements
will allow hydrate model parameters to be fit to hydrate composition and structural data.
Corrected model predictions can then guide experiments (Subramanian et al., 2000b).
The methane+ethane+propane+water system is the simplest approximation of a
natural gas mixture. As shown in Figures 9.19 through 9.27, the phase equilibria of such
a simple mixture is quite complicated at pressures above incipient hydrate formation
conditions. The two most interesting phenomena found are the coexistence of sI and sII
hydrates and the lack of the pseudo-retrograde phenomenon in the interior of the pseudoternary phase diagrams. These two phenomena could have industrial implications as
indicated in the below paragraphs.
Hydrates in the permafrost and beneath the sea floor are formed from nearly pure
methane and are therefore assumed to be sI hydrates. Calculations of the amount of
245
methane gas trapped in these in situ hydrates are based on this assumption. At low
pressures, as little as a 0.005 mole fraction contamination of ethane and/or propane could
force the stable hydrate to be sII. At higher pressures, any contamination of ethane
and/or propane (< 0.005) will force the sII hydrates to be in equilibrium with sI hydrates.
The amount of methane trapped in sI hydrates versus sII hydrates is different
(Subramanian et al., 2000) and the incorrect assumed structure could significantly affect
the calculations for the amount of methane trapped in natural hydrates.
Chemicals such as kinetic inhibitors or anti-agglomerates are added to natural gas
pipelines to prevent hydrate plugs.
formation kinetics while anti-agglomerates are designed to prevent hydrate particles from
agglomerating. Typical natural gas hydrates are assumed to be sII and therefore these
chemicals are designed to prevent sII hydrates from plugging a pipeline. Figure 9.27
suggests that if a natural gas mixture is rich in methane or fairly rich in ethane, sI and sII
hydrates will coexist. Therefore, a kinetic inhibitor or anti-agglomerate may prevent the
sII hydrates from plugging the pipeline but not the sI hydrates.
Since most natural gases are rich in methane, it is not likely that the pseudoretrograde hydrate phenomena will occur in a natural gas pipeline. However, in a gas
mixture with very little methane, such as the bottoms product of a demethanizer, the
pseudo-retrograde hydrate phenomenon could be possible.
demethanizer bottoms product pipeline could be operated at pressures above the incipient
hydrate formation pressure without hydrate formation.
247
CHAPTER 10 -
The Gibbs energy minimization method for predicting hydrate phase equilibria is
a powerful tool. However, the power cannot be realized until predictions are made and
complete hydrate phase diagrams are generated. The reason that the potential for the
method of Gupta (1990) was not realized is that no phase diagrams were made. The
previous chapter, analyzing methane, ethane, and propane hydrates, is the first of its kind
and is solely due to the Gibbs energy minimization procedure.
The approach to hydrate modeling taken in this work is another powerful tool that
cannot be realized until predictions of hydrate phase equilibria are made. In this work,
several measurements of the hydrate phase were incorporated into the optimization
procedure. Extrapolations of the current hydrate database were made using the modified
hydrate model, resulting in several interesting predictions. It was these predictions that
lead to further experiments.
In this chapter, we present several examples of how predictions from the hydrate
model guided experimental efforts. We also present interesting predictions that have not
yet been experimentally determined.
248
parameters, predicted the hydrate formation temperatures with greater accuracy than any
other current hydrate model. However, of more importance, it predicted the structural
transition as well.
To test the quality of the predictions, the upper structural transition point was
modeled at 274.15 K and predicted to occur at 99.4 mole percent methane.
This
prediction was used to guide work in our laboratory to measure the upper transition point
using Raman spectroscopy. The experimental result was found to be between 99.3 0.06
mole percent methane. Figure 10.1 shows pseudo-P-x diagram at 274.15 K with the two
Raman spectroscopy data points at the upper and lower structural transition points.
Experimental Structural
Transition Points at
~73 and ~99.3 mole%
30
Pressure (bar)
25
Aq-sII
Aq-sI-sII
20
Aq-sI-sII
Aq-sI
Aq-sI-V
Aq-sII-V
15
Aq-sI
10
Aq-sI-V
Aq-V
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
249
transition composition, the model predictions were used as an initial guess for it. This
saved substantial time and effort on the part of Subramanian (2000).
The structural transition found by Jager (2001) in a natural gas system was also
predicted prior to the experimental confirmation (Figure 7.35). Just as the predictions led
to experiments in methane+ethane hydrates, they did so with the lean natural gas. Jager
used the predictions as a starting point for his experiments, saving time and effort.
250
10
9
Aq-Lhc
Aq-sII-Lhc
Aq-sI-V
Pressure (bar)
Aq-sI-Lhc
Aq-sII
Aq-V-Lhc
6
5
4
Aq-sII-V
Aq-V
2
1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
251
between 1 and 6.5 bar, aqueous, vapor, and liquid hydrocarbon phases are present. At 6.5
bar sII hydrates form. As pressure is increased, the amount of sII increases. As this is
occurring, the vapor phase is being depleted of light components (ethane and propane)
while the liquid hydrocarbon phase is increasing due to the increase in pressure
(condensation of vapor).
Aq-V-Lhc
Aq-sII-V-Lhc
Aq-Lhc
Aq-V-Lhc
sII Hydrate
Vapor
Liquid Hydrocarbon
10
13
Pressure (bar)
Figure 10.3 Water-free phase amount versus pressure diagram for ethane+
propane+n-decane+water at 277.5 K
The amount of sII in the system is predicted to reach a maximum amount at ~8 bar. The
sII hydrates then start to dissociate as pressure is increased, completely dissociating at
~9.6 bar. After sII dissociates, there is a small pressure window in which Aq-V-Lhc are
present (9.6-10 bar). The only phases predicted to be present at pressures greater than 10
bar are the aqueous and liquid hydrocarbon phases. Note that the pseudo-retrograde
252
separation processes are used for such compounds. However, close-boiling binaries
require so many equilibrium stages that cost dictates investigation of other methods.
The following close-boiling binary systems were evaluated: 1. propane+
propylene, 2. ethane+ethylene, 3. n-butane+i-butane, and 4. n-pentane+i-pentane. Here
we show the advantage of idealized, equilibrium, separation processes for the above four
binaries using aqueous-hydrate-vapor (Aq-H-V) and aqueous-hydrate-liquid hydrocarbon
separations (Aq-H-Lhc). For each binary pair, pseudo-T-x and pseudo-x-y diagrams have
been generated at optimal pressures for standard McCabe-Thiele analysis.
10.3.1 Propane+Propylene System
The optimal pressure suggested by Glew (1966) was 6.2 bar. This pressure was
chosen such that, at the hydrate formation temperatures for all mixtures of propane and
propylene, no vapor phase exists. We performed an analysis at this pressure as well.
Figure 10.4 is a pseudo-T-x diagram at 6.2 bar.
253
282
281
Aq-V
(V)
Temperature (K)
280
279
Aq-Lhc
(Lhc)
278
277
Aq-V-Lhc
(V-Lhc)
Aq-sII-Lhc
276
275
Aq-sII
274
0
0.25
0.5
0.75
formation, all mixtures of propane and propylene have condensed to a liquid hydrocarbon
phase. The hydrate envelope (Aq-sII-Lhc) is much larger than the hydrocarbon envelope
(V-Lhc). This is the main indicator that hydrates could potentially separate the two
hydrocarbons more efficiently than the standard approach.
An alternative to the pseudo-T-x diagram shown in Figure 10.4 is a pseudo-x-y
diagram at the same pressure. Pseudo-x-y diagrams are typically used to determine the
ideal, equilibrium number of stages needed in a distillation column to separate two
compounds. Figure 10.5 is the analogous pseudo-x-y diagram for the propane-propylene
pair at 6.2 bar.
254
V-Lhc
sII-Lhc
yC3= (V/sII)
0.75
0.5
0.25
0
0
0.25
0.5
xC3= (Lhc)
0.75
255
xT > 0.99
xB < 0.01
yC3= (V/sII)
0.75
V-Lhc
sII-Lhc
V-Lhc = 54 stages
sII-Lhc = 9 stages
0.5
Feed Line
0.25
Lhc
30 mol% C3H8
70 mol% C3H6
0
0
0.25
0.5
xC3= (Lhc)
0.75
256
V-Lhc
V-sI
yC2= (V)
0.75
0.5
0.25
0
0
0.25
0.5
0.75
xC2= (Lhc/sI)
preferentially distribute much within the hydrate. Therefore, the standard hydrocarbon
distillation should be used to perform the separation of ethane and ethylene.
257
290
Aq-V-Lhc
(V-Lhc)
Temperature (K)
285
280
Aq-Lhc
(Lhc)
Aq-sII-Lhc
Aq-sII
275
270
Ice-Lhc
Ice-sII-Lhc
Ice-sII
265
260
0
0.25
0.5
0.75
258
fraction i-butane, the hydrate envelope is much larger than the hydrocarbon envelope.
Therefore, we consider the pseudo-x-y diagram at this pressure (Figure 10.9).
V-Lhc
sII-Lhc
Ice Point
yi-C4 (V/sII)
0.75
0.5
0.25
0
0
0.25
0.5
0.75
xi-C4 (Lhc)
259
V-Lhc
sII-Lhc
Ice Point
yi-C4 (V/sII)
0.9
xT > 0.99
0.8
V-Lhc = 11 stages
sII-Lhc = 2 stages
0.7
Feed Line
Lhc
30 mol% n-C4H10
70 mol% i-C4H10
0.6
0.6
0.7
0.8
0.9
xi-C4 (Lhc)
260
V-Lhc
V-sH
Ice Point
sI Hydrate Point
yi-C5 (V)
0.75
0.5
0.25
0
0
0.25
0.5
0.75
xi-C5 (Lhc/sH)
261
The compositions shown in Figure 10.11 are on a water-free, methane-free basis. That is,
we only look at how the pentanes distribute between phases. We have chosen to add 9
moles of methane to every 1 mole of the pentane binary pair. The amount of methane
added to the pentane binary system affects the distribution very little. Note that for
water-, methane-free compositions below 0.33 i-pentane, sI hydrates are predicted to
form. sH hydrates form with ice for compositions below 0.38 i-pentane. If sI hydrate
forms, i-pentane will not be included in the hydrate and we do not desire to perform the
separation with ice present. Therefore, at this pressure we must keep the feed greater
than 0.38 i-pentane. It should be noted that the standard hydrocarbon envelope is not
shown above. The V-Lhc envelope shown represents the liquid hydrocarbon composition
when hydrates are present. However, the standard V-Lhc envelope is almost identical to
the V-Lhc envelope with hydrates present.
Similar to the n-butane-i-butane system, a separation for the n-pentane-i-pentane
system will only serve to purify i-pentane. However, in this system, since n-pentane does
not enter sH hydrates, we have pure i-pentane at every stage of the column (on a pentane
basis). The i-pentane will contain methane, which can easily be separated. Therefore, we
need not perform an equilibrium stage calculation.
10.3.5 Feasibility of a Hydrate Separation Technique
As can be seen from the above figures, by adding water to a hydrocarbon mixture
and allowing hydrates to form, the number of stages required to separate the
hydrocarbons is greatly reduced. However, this is from a theoretical point of view. The
next question to ask is whether it is practicle to use hydrates as a separation tool. In
answering this question, several concerns must be addressed.
By adding water to the hydrocarbon mixture, three potential problems appear:
1. water will distribute between phases,
2. the hydrocarbons will distribute to the aqueous phase, and
262
Lhc (x1)
To Formation
Chamber k-1
P = 6.2 bar
T = Tk
Lhc
hydrate
hydrates
To Dissociation
Chamber
water
Recycled water
from Dissociation
Chamber
Figure 10.12 Possible schematic for hydrate formation chamber
263
The liquid hydrocarbon is fed to the formation chamber, which is at 6.2 bar and T1,
containing water. Since nucleation of hydrates may be a problem, a highly agitated
system or the addition of a seed hydrate crystal would be recommended. When a given
amount of hydrate is formed, they can then be extracted, with any excess water being
recycled back to the formation chamber.
P = 1 bar
hydrates
From Formation
Chamber k
Lhc to Formation
Chamber k+1
HC
Fluid
water
Q
Recycle water
to Formation
Chamber k
discussed above dissociate at temperatures near the ice point, ambient temperature should
be sufficient and will reduce the cost of heating.
The water recovered from the dissociated hydrates can be recycled back to the
formation chamber at T1. The Lhc recovered, which will have a higher concentration of
the most volatile component, can then be sent to another formation chamber at another
temperature, T2. This process will be continued until the desired separation is achieved.
264
The number of formation chambers needed will be the same as the number of equilibrium
stages while only one dissociation chamber is needed.
10.3.7 Question of Hydrate Equilibrium
Another question to ask is, How accurate it is to assume that each stage of the
hydrate separation process is an equilibrium stage?
measurements by Kini (2002) show that the formation temperature dictates the
composition of binary hydrates. For example, in the propane-propylene system shown in
Figure 10.14, a 30/70 mixture forms hydrates at 276 K and 6.2 bar. If the hydrates were
formed at temperatures below 276 K (i.e. 274 K), the hydrate composition will be the
same as that at 276 K.
279
278
Temperature (K)
Aq-Lhc
277
Feed
Aq-sII-Lhc
Hydrate
Formation T
276
Hydrate
Composition
275
Aq-sII
274
0
0.25
0.5
0.75
265
267
11.1 Conclusions
This work consisted of several separate steps of research: 1. development of the
hydrate model, 2. implementation into a multi-phase equilibria program, and 3. analysis
of the hydrate model. Conclusions will be stated from each step of the research:
1. Non-Ideal Solid Solution Hydrate Model: In this work, we have developed the first
expression for the hydrate volume as a function of temperature, pressure, and
composition.
thermodynamic model for hydrates, consequently allowing for two of the basic
assumptions by van der Waals and Platteeuw to be removed:
a.
b.
Assumption (1) was removed by accounting for the energy change, via an activity
coefficient, associated with the change in hydrate volume from the standard state to the
real state. Assumption (2) was removed by making the cage radii (equivalently Langmuir
constants) functions of the hydrate volume. The removal of these assumptions was a
direct result of obtaining an expression for the hydrate volume and lead to the non-ideal
solid solution hydrate model. The proposed description of the hydrate cages (i.e. multilayered) served to indirectly remove the assumption of spherical cages and provide more
realistic Kihara potential parameters. The new hydrate model compared favorably to 4
commercial hydrate prediction programs for all hydrate literature data.
268
269
11.2 Recommendations
The following studies are recommendations, based on conclusions made in the
previous chapters, for extension of this work:
1. Investigation of Langmuir constants for non-spherical guests: The model proposed in
this work does very satisfactory predicting hydrate formation temperatures and pressures
and hydrate phase properties. However, it is clear that the error in predictions is greater
with the addition of larger guests in the sII hydrate. We believe that this work has
accounted for most of the previous inadequacies in the hydrate model (i.e. high pressure,
non-spherical cages, and allowing for distortion of lattice and cages). The predictions for
xenon hydrate in Figure 7.19 are evidence of the first of these two. However, one area
which still needs to be addressed is the assumption of spherical hydrate guests. When
McKoy and Sinanolu extended the Kihara potential model for use in the hydrate model,
they derived an alternative expression for rod-like molecules (1963, Eq. 17). Their
expression was not used in this work, or any other to date. It may be worthwhile to try
this expression for non-spherical hydrate guests.
2. Pseudo-retrograde hydrate phenomenon: The applications of the pseudo-retrograde
hydrate behavior described in Chapters 8 and 9 are still scarce.
However, this is
270
Although the
fractional occupancy does not tell us how much of the guest in the hydrate, it does tell us
about how the composition of that guest changes with each experiment.
271
5. More hydrate phase measurements: The development of the hydrate model in this
work was made possible by several measurements of the hydrate phase via X-ray
diffraction and Raman and NMR spectroscopy. Without the data obtained from these
experiments, we would not have had the theoretical validation for improving the model.
We believe that this is the type of data that will enable future improvements of the model.
6. More natural gas hydrate data: More measurements of natural gas hydrates are
needed. Natural gas hydrate data are the most relevant for flow assurance applications;
however, it is interesting that these types of data are scarce (59 points for 10 natural
gases). For modeling purposes, natural gas hydrate data at low and high pressure are
needed. Hydrate phase measurements of these hydrates using XRD and Raman and
NMR spectroscopy are also recommended.
The future of hydrate modeling will be very dependent on hydrate phase
measurements. Likewise, we believe that the future of hydrate phase measurements will
be dependent on hydrate modeling. That is, as the model gets better, more complicated
systems and extreme conditions will need to be experimentally investigated to test its
limits.
273
LIST OF SYMBOLS
Latin Letters
a
c
C
D
E
f
F
g
h
J
k
K
P
r
R
s
S
T
v
x
z
Greek Letters
274
Subscripts
Aq
c
f
H
i
J
k
Lw
m
o
r
S
w
0
aqueous phase
critical property
arbitrary fluid phase
hydrate phase
arbitrary component
hydrate guest component
arbitrary phase
pure liquid water
hydrate cavity
ideal gas at 1 bar
reference phase
solid salt phase
water (as a component)
property at reference conditions (298.15 K, 1 bar)
standard empty hydrate lattice
hypothetical 1 molal solution
Superscript
__
partial property
275
REFERENCES CITED
Abbott, M.M, Kaufmann, T.G., and Domash, L. (1971) A correlation for predicting
liquid viscosities of petroleum fractions. Can. J. Chem. Eng., 49, 379
Adisasmito, S., and Sloan, E.D., Jr. (1992) Hydrates of hydrocarbon gases containing
carbon dioxide. J. Chem. Eng. Data, 37(3), 343-348
API Data Book (1978a) Vapor pressures of pure hydrocarbons and narrow-boiling
petroleum fractions. Proc. 5A1.13, (5)19
(1978b) Solubility parameters and effective molar volumes for
hydrocarbons and nonhydrocarbons. Table 8B1.7, (8)47
(1987a) Molecular weight of petroleum fractions. Proc. 2B2.1, (2)15
(1987b) Method for the pseudocritical temperature of petroleum
fractions. Proc. 4D3.1, (4)57
(1987c) Method for the pseudocritical pressure of petroleum fractions.
Proc. 4D4.1, (4)65
(1994) Alternate (computer) method for hydrocarbon-hydrocarbon and
hydrocarbon-nonhydrocarbon vapor-liquid equilibrium K-values.
Proc. 8D1.1, (8)53-62
(1997) Primary properties. Tables 1C1.1-1C1.8, (1)54-127
(1999) Recommended interaction coefficients for the Soave procedure.
Table 8D1.4, 8-63
Avlonitis, D.A. (1994a) A scheme for reducing experimental heat capacity data of gas
hydrates. Ind. Eng. Chem. Res., 33(12), 3247-3255
Avlonitis, D.A. (1994b) The determination of Kihara potential parameters from gas
hydrate data. Chem. Eng. Sci., 49(8), 1161-1173
276
Baker, L.E., Pierce, A.C., and Luks, K.D. (1982) Gibbs energy analysis of phase
equilibria. Soc. Petr. Eng. J., October, 731-742
Ballard, A.L., and Sloan, E.D., Jr. (2000) Optimizing thermodynamic parameters to
match methane and ethane structural transition in natural gas hydrate equilibria. Ann.
NY Acad. of Sci., 912, 702-712
Ballard, A.L., Jager, M.D., Nasrifar, Kh., Mooijer-van den Heuvel, M.M., Peters, C.J.,
and Sloan, E.D., Jr. (2001) Pseudo-retrograde hydrate phenomena at low pressures.
Fluid Phase Equil., 185, 77-87
Barkan, E.S., and Sheinin, D.A. (1993) A general technique for the calculation of
formation conditions of natural gas hydrates. Fluid Phase Equil., 86, 111-136
Barrer, R.M., and Ruzicka, D.J. (1962) Non-stoichiometric clathrate compounds of water
Part 2 Formation and properties of double hydrates. Trans. Faraday Soc., 58, 22392252
Bertie, J.E., Bates, F.E., and Hendricksen, D.K. (1975) The far infrared spectra and x-ray
powder diffraction patterns of the structure I hydrates of cyclopropane and ethylene
oxide at 100 oK. Can. J. Chem., 53, 71-75
Bishnoi, P.R., and Dholabhai, P.D. (1999) Equilibrium conditions for hydrate formation
for a ternary mixture of methane, propane and carbon dioxide, and a natural gas
mixture in the presence of electrolytes and methanol. Fluid Phase Equil., in press
Bollavaram, P., Ph.D. Thesis (2002) Colorado School of Mines, Golden, CO, USA
Bridgman, P.W. (1912) Water, in the liquid and five solid forms, under pressure. Proc.
Am. Acad. Arts Sci., 47(13), 530-533
Bromley, L.A. (1973) Thermodynamic properties of strong electrolytes in aqueous
solutions. AIChE J., 19(2), 313
Cady, G.H. (1983) Composition of clathrate gas hydrates of H2S, Xe, SO2, Cl2, CH3Cl,
CH3Br, CHClF2, CCl2F2, and C3H8. J. Phys. Chem., 87, 4437-4441
Cao, Z., Tester, J.W., Sloan, E.D., Jr., and Trout, B.L. (2001) Sensitivity analysis of
hydrate thermodynamic reference properties using experimental data and ab initio
methods. J. Phys. Chem. B, Submitted August, 2001
277
Carson, D.B., and Katz, D.L. (1942) Natural gas hydrates. Trans. AIME, 146, 150-158
Child, W.C. (1964) Thermodynamic functions for metastable ice structures I and II. J.
Phys. Chem., 68(7), 1834-1838
Chou, I.-M., Sharma, A., Burruss, R.C., Shu, J., Mao, H.-K., Hemley, R.J., Goncharov,
A.F., and Stern, L.A. (2000) Transformations in methane hydrates. Proc. Natl. Acad.
Sci., 97(25), 13484-13487
Collett, T.S. (2000) Natural gas hydrate as a potential energy resource. Natural Gas
Hydrate in Oceanic and Permafrost Environments, Edited by M.D. Max, Kluwer
Academic, 123-136
Danesh, A., Tohidi, B., Burgass, R.W., and Todd, A.C. (1993) Benzene can form gas
hydrates. Trans. IChemE, 71, A, 457-459
Davidson, D.W. (1984) Unpublished results
Davidson, D.W., Garg, S.K., Gough, S.R., Handa, Y.P., Ratcliffe, C.I., Tse, J.S., and
Ripmeester, J.A. (1984) Some structural and thermodynamic studies of clathrate
hydrates. J. Incl. Phen., 2, 231-238
Davy, H. (1811) The Bakerian lecture. On some of the combinations of oxymuriatic gas
and oxygen, and on the chemical relations of these principles to inflammable bodies.
Phil. Trans. Roy. Soc. (London), 101, 1-35
Deaton, W.M., and Frost, E.M. (1946) Gas hydrates and their relation to the operation of
natural-gas pipelines. U.S. Bureau of Mines Monograph 8, 101
de Loos, Th.W. (1994) Applied Sciences, 273, 89-113.
Denney, D. (2002) Exploration for natural hydrate offshore Japan. J. Petr. Tech., 54(1),
41-42
DePriester, C.L. (1953) Light-hydrocarbon vapor-liquid distribution coefficients:
Pressure-temperature-composition charts and pressure-temperature nomographs.
Chem. Eng. Prog. Symp. Ser., 49(7), 1-43
de Roo, J.L., Peters, C.J., Lichtenthaler, R.N., and Diepen, G.A.M. (1983) Occurrence of
methane hydrate in saturated and unsaturated solutions of sodium chloride and water
in dependence of temperature and pressure. AIChE J., 29(4), 651-657
278
Dharmawardhana, P.B., Parrish, W.R., and Sloan, E.D., Jr. (1980) Experimental
thermodynamic parameters for the prediction of natural gas hydrate dissociation
conditions. Ind. Eng. Chem. Fund., 19(4), 410-414
Dillon, W.P., Nealon, J.W., and Taylor, M.H. (2001) Seafloor collapse and methane
venting associated with gas hydrate on the Blake Ridge Causes and implications to
seafloor stability and methane release. Natural Gas Hydrates: Occurrence,
Distribution, and Detection, Edited by C.K. Paull and W.P. Dillon, American
Geophysical Union, 211-233
Douslin, D.R., and Osborn, A. (1965) Pressure measurements in the 0.01-30 torr range
with an inclined-piston gauge. J. Sci. Instrum., 42, 369-373
Dyadin, U.A., Larionov, E.G., Mirinskij, D.S., Mikina, T.V., Aladko, E.Y., and
Starostina, L.I. (1997) Phase diagram of the Xe-H2O system up to 15 kbar. J. Incl.
Phen., 28(4), 271-285
Erickson, D.D., M.Sc. Thesis (1980) Development of a natural gas hydrate prediction
computer program. Colorado School of Mines, Golden, CO, USA
Esso Research and Engineering Company (1977) Personal communication with API
Freer, E.M., M.Sc. Thesis (2000) Methane hydrate growth kinetics. Colorado School of
Mines, Golden, CO, USA
Friend, D.G., Ely, J.F., and Ingham, H. (1989) Thermophysical properties of methane. J.
Phys. Chem. Ref. Data, 18(2), 586-630
Frost, E.M., Jr., and Deaton, W.M. (1946) Gas hydrate composition and equilibrium data.
Oil and Gas J., 45(12), 170-178
Galloway, T.J., Ruska, W., Chappelear, P.S., and Kobayashi, R. (1970) Experimental
measurements of hydrate numbers for methane and ethane and comparison with
theoretical values. Ind. Eng. Chem. Fund., 9, 237-243
Gas Processors Association, Thermodynamic Database Version 2.0 (1996) Oklahoma
State University
Gibbs, J.W. (1875) On the equilibrium of heterogeneous substances. Reprinted in The
Scientific Papers of J. Willard Gibbs, Vol. I, Dover Publications, Inc., N.Y., 1961
279
Glew, D.N. (1962) Aqueous solubility and the gas hydrates The methane-water system.
J. Phys. Chem., 66, 605
Glew, D.N. (1966) Liquid fractionation process using gas hydrates. U.S. Patent
#3,231,630
Graboski, M.S., and Daubert, T.E. (1978) A modified soave equation of state for phase
equilibrium calculations 1. Hydrocarbon systems. Ind. Eng. Chem. Proc. Des. Dev.,
17, 443-448
Gupta, A.K., Ph.D. Thesis (1990) Steady state simulation of chemical processes.
University of Calgary, Calgary, Canada
Gutt, C., Asmussen, B., Press, W., Johnson, M.R., Handa, Y.P., and Tse, J.S. (2000) The
structure of deuterated methane hydrate. J. Chem. Phys., 113, 4713
Hadden, S.T., and Grayson, H.G. (1961) New charts for hydrocarbon vapor-liquid
equilibria. Hydrocarbon Proc. and Petrol. Refiner, 40(9), 207-218
Haffemann, D.R., and Miller, S.L. (1969) The clathrate hydrates of cyclopropane. J.
Phys. Chem., 73(5), 1392-1397
Hammerschmidt, E.G. (1934) Formation of gas hydrates in natural gas transmission lines.
Ind. Eng. Chem., 26, 851-855
Handa, Y.P. (1986) Calorimetric studies of laboratory synthesized and naturally
occurring gas hydrates. AIChE Meeting, November 2-7, Miami Beach, Florida
Handa, Y.P., and Tse, J.S. (1986) Thermodynamic properties of empty lattices of
structure I and structure II clathrate hydrates. J. Phys. Chem., 90(22), 5917-5921
Haq, B.U. (1999) Methane in the deep blue sea. Science, 285, 543-544
Heidemann, R.A., and Khalil, A.M. (1980) The calculation of critical points. AIChE J.,
26(5), 769-779
Henderson, S.J., and Speedy, R.J. (1987) Melting temperature of ice at positive and
negative pressures. J. Phys. Chem., 91(11), 3069-3072
Hendriks, E.M., Edmonds, B., Moorwood, R.A.S., and Szczepanski, R. (1996) Hydrate
structure stability in simple and mixed hydrates. Fluid Phase Equil., 117, 193-200
280
Hirai, H., Kondo, T., Hasegawa, M., Yagi, T., Yamamoto, Y., Komai, T., Nagashima, K.,
Sakashita, M., Fujihisa, H., and Aoki, K. (2000) Methane hydrate behavior under
high pressure. J. Phys. Chem. B, 104(7), 1429-1433
Holder, G.D., Ph.D. Thesis (1976) Multi-phase equilibria in methane-ethane-propanewater hydrate forming systems. University of Michigan, Ann Arbor, Michigan, USA
Holder, G.D., and Corbin, G. (1979) The prediction of phase equilibria and
thermodynamic properties of multi-component gas-water hydrate forming systems.
86th National AIChE Meeting, April 1-6, Houston, Texas
Holder G.D., and Grigoriou, G.C. (1980) Hydrate dissociation pressures of methane +
ethane + water. Existence of a locus of minimum pressures. J. Chem. Thermo., 10931104
Holder, G.D., Corbin, G., and Papadopoulos, K.D. (1980) Thermodynamic and molecular
properties of gas hydrates from mixtures containing methane, argon and krypton.
Ind. Eng. Chem. Fund., 19(3), 282-286
Holder, G.D., and Hand, J.H. (1982) Multiple-phase equilibria in hydrates from methane,
ethane, propane and water mixtures. AIChE J., 28(3), 440-447
Hollander, F., and Jeffrey, G.A. (1977) Neutron diffraction study of the crystal structure
of ethylene oxide deuterohydrate at 80 oK. J. Chem. Phys., 66(10), 4699-4705
Hovland, M., and Gudmestad, O.T. (2001) Potential influence of gas hydrates on seabed
installations. Natural Gas Hydrates: Occurrence, Distribution, and Detection, Edited
by C.K. Paull and W.P. Dillon, American Geophysical Union, 307-315
Huo, Z. (2001) Personal communication
Huo, Z., Ph.D. Thesis (2002) Colorado School of Mines, Golden, CO, USA
Hutz, U., and Englezos, P. (1996) Measurement of structure H hydrate phase equilibrium
and effect of electrolytes. Fluid Phase Equil., 117, 178-185
Hyndman, R.D., Spence, G.D., Chapman, R., Riedel, M., and Edwards, R.N. (2001)
Geophysical studies of marine gas hydrate in northern Cascadia. Natural Gas
Hydrates: Occurrence, Distribution, and Detection, Edited by C.K. Paull and W.P.
Dillon, American Geophysical Union, 273-295
281
Ikeda, T., Yamamuro, O., Matsuo, T., Mori, K., Torii, S., Kamiyama, T., Izumi, F.,
Ikeda, S., and Mae, S. (1999) Neutron diffraction study of carbon dioxide clathrate
hydrate. J. Phys. Chem. Solids, 60, 1527-1529
Jager, M.D., Ph.D. Thesis (2001) High pressure studies of hydrate phase inhibition using
Raman spectroscopy. Colorado School of Mines, Golden, CO, USA
Jager, M.D., and Sloan, E.D., Jr. (2001) The effect of pressure on methane hydration in
pure water and sodium chloride solutions. Fluid Phase Equil., 185, 88-99
Jancso, G., Pupezin, J., and Van Hook, W.A. (1970) The vapor pressure of ice between
+10-2 and 10+2o. J. Phys. Chem., 74(15), 2984-2989
Jhaveri, J., and Robinson, D.B. (1965) Hydrates in the methane-nitrogen system. Can. J.
Chem. Eng., 43, 75-78
John, V.T., and Holder, G.D. (1981) Choice of cell size in the cell theory of hydrate
phase gas-water interactions. J. Phys. Chem., 85(13), 1811-1814
John, V.T., and Holder, G.D. (1982) Contribution of second and subsequent water shells
to the potential energy of guest-host interactions in clathrate hydrates. J. Phys. Chem.,
86(4), 455-459
John, V.T., Papadopoulos, K.D., and Holder, G.D. (1985) A generalized model for
predicting equilibrium conditions for gas hydrates. AIChE J., 31(2), 252-259
Katz, D.L. (1945) Prediction of conditions for hydrate formation in natural gases. Trans.
AIME, 160, 140-149
Kini, R., Ph.D. Thesis (2002) Colorado School of Mines, Golden, CO, USA
Knox, W.G., Hess, M., Jones, G.E., and Smith, H.B. (1961) The hydrate process. Chem.
Eng. Prog., 57(2), 66
Kobayashi, R., Withrow, H.J., Williams, G.B., and Katz, D.L. (1951) Gas hydrate
formation with brine and ethanol solutions. Proc. Conf. NGAA, 30, 27-31
Kremer, H. Ph.D. Thesis (1982) Experimentelle untersuchung und berechnung von
hochdruck-flussigkeits-dampf-und-flussigkeits-flussigkeits-dampf-glwichgewichten
fur tiefsiedende gemishche. University of Berlin, Berlin, Germany
282
Kuhs, W.F., Chazallion, B., Radaelli, P., Pauer, F., and Kipfstuhl, J. (1996) Raman
spectroscopic and neutron diffraction studies on natural and synthetic clathrates of air
and nitrogen. Proc. Second Int. Conf. Natural Gas Hydrates, June 2-6, Toulouse,
France, 9-16
Kuhs, W.F. (2001) Personal communication with E. D. Sloan, Jr.
Lapin, A., and Cinnamon, S.J. (1969) Problem of hydrate formation in natural gas
systems. Proc. Liquid Natural Gas: its Production, Handling, and Use, London, 198
Leadbetter, A.J. (1965) The thermodynamic and vibrational properties of H2O ice and
D2O ice. Proc. R. Soc. London, A, 287, 403-425
Lederhos, J.P., Mehta, A.P., Nyberg, G.B., Warn, K.J., and Sloan, E.D., Jr. (1992)
Structure H clathrate hydrate equilibria of methane and adamantane. AIChE J., 38(7),
1045-1048
Lederhos, J.P., Ph.D. Thesis (1996) The transferability of hydrate kinetic inhibition
results between bench scale apparatuses and a pilot scale flow loop. Colorado School
of Mines, Golden, CO, USA
Lee, S.-Y., and Holder, G.D. (2002) Model for gas hydrate equilibria using a variable
reference chemical potential: Part 1. AIChE J., 48(1), 161-167
Lee, B.I., and Kesler, M.G. (1975) A generalized thermodynamic correlation based on
three-parameter corresponding states. AIChE J., 21, 510-527
Lennard-Jones, J.E., and Devonshire, A.F. (1937) Critical phenomena in gases I. Proc.
Roy. Soc. (London), 163, 53-70
Mak, T.C. W., and McMullan, R.K. (1965) Polyhedral clathrate hydrates. X. Structure of
the double hydrate of tetrahydrofuran and hydrogen sulfide. J. Chem. Phys., 42(8),
2732-2737
Mann, S.L., McClure, L.M., Poettmann, F.H., & Sloan, E.D., Jr. (1989) Vapor-solid
equilibrium ratios for structure I and II natural gas hydrates. Proceedings of the SixtyEighth GPA Annual convention, 60-74
Marshall, D.R., Saito, S., & Kobayashi, R. (1964) Hydrates at high pressure: Part I
methane-water, argon-water, and nitrogen-water systems. AIChE J., 10(2), 202-205
283
Matschke, D.E., and Thodos, G. (1962) Vapor-liquid equilibria for the ethane-propane
system. J. Chem. Eng. Data, 7(2), 232-234
Max, M.D. (2000) Preface. Natural Gas Hydrate in Oceanic and Permafrost
Environments, Edited by M.D. Max, Kluwer Academic, xi-xiv
Maxwell and Bonnell (1955) Vapor pressure charts for petroleum engineers. Esso
Research and Engineering Company, Linden, New Jersey
McKoy, V., and Sinanolu, O. (1963) Theory of dissociation pressures of some gas
hydrates. J. Chem. Phys., 38(12), 2946-2956
McLeod, H.O., and Campbell, J.M. (1961) Natural gas hydrates at pressures to 10,000
psia. J. Petrol. Tech., 222, 590-597
McMullan, R.K., and Jeffrey, G.A. (1965) Polyhedral clathrate hydrates. IX. Structure of
ethylene oxide hydrate. J. Chem. Phys., 42(8), 2725-2732
Mehta, A.P., Ph.D. Thesis (1996) A thermodynamic investigation of structure H clathrate
hydrates. Colorado School of Mines, Golden, CO, USA
Mehta, A.P., and Sloan, E.D., Jr. (1993) Structure H hydrate phase equilibria of methane
+ liquid hydrocarbon mixtures. J. Chem. Eng. Data, 38(4), 580-582
Mehta, A.P., and Sloan, E.D., Jr. (1994) Structure H hydrate phase equilibria of paraffins,
naphthenes, and olefins with methane. J. Chem. Eng. Data, 39(4), 887-890
Mehta, A.P., and Sloan, E.D., Jr. (1996) Improved thermodynamic parameters for
prediction of structure H hydrate equilibria. AIChE J., 42(7), 2036-2046
Menten, P.D., Parrish, W.R., and Sloan, E.D., Jr. (1981) Effect of inhibitors on hydrate
formation. Ind. Eng. Chem. Proc. Des. Dev., 20(2), 399-401
Michelsen, M.L. (1982a) The isothermal flash problem. Part I. Stability. Fluid Phase
Equil., 9, 1-19
Michelsen, M.L. (1982b) The isothermal flash problem. Part II. Phase-split calculation.
Fluid Phase Equil., 9, 21-40
Miller, B., and Strong, E.R., Jr. (1946) Hydrate storage of natural gas. Am. Gas Assoc.
Monthly, 28(2), 63
284
285
Pratt, R.M., Ballard, A.L., and Sloan, E.D., Jr. (2001) Beware of singularities when
calculating clathrate hydrate cell potentials. AIChE J., 47 (8), 1897-1898
Rachford, H.H., and Rice, J.D. (1952) Procedure for use of electronic digital computers
in calculating flash vaporization hydrocarbon equilibrium. J. Petrol. Tech., 4(10), 1019
Reamer, H.H., Olds, R.H., Sage, B.H., and Lacey, W.N. (1943) Phase equilibria in
hydrocarbon systems: Composition of dew-point gas in ethane-water system. Ind.
Eng. Chem., 35(7), 790-793
Reamer, H.H., Sage, B.H., and Lacey, W.N. (1950) Phase equilibria in hydrocarbon
systems: Volumetric and phase behavior of the methane-propane system. Ind. Eng.
Chem., 42(3), 534
Redlich, O., and Kwong, J.N.S. (1949) On the thermodynamics of solutions V: An
equation of state. Fugacities of gaseous solutions. Chem. Rev., 44, 233-244
Riazi, M.R., Ph.D. Thesis (1979) Prediction of thermophysical properties of petroleum
fractions. The Pennsylvania State University, University Park, PA, USA
Richards, T.W., and Speyers, C.L. (1914) The compressibility of ice. J. Am. Chem. Soc.,
36(3), 491-494
Ripmeester, J.A., and Davidson, D.W. (1981) 129Xe nuclear magnetic resonance in the
clathrate hydrate of xenon. J. Mol. Struct., 75, 67-72
Ripmeester, J.A., Tse, J.A., Ratcliffe, C.I., and Powell, B.M. (1987) A new clathrate
hydrate structure. Nature, 325, 135-136
Ripmeester, J.A., and Ratcliffe, C.I. (1988) Low-temperature cross-polarization/magic
angle spinning 13C NMR of solid methane hydrates: structure, cage occupancy, and
hydration number. J. Phys. Chem., 92, 337-339
Roberts, O.L., Brownscombe, E.R., Howe, L.S., and Ramser, H. (1941) Phase diagrams
of methane and ethane hydrates. Petrol. Eng., 12(6), 56
Robinson, D.B., and Hutton, J.M. (1967) Hydrate formation in systems containing
methane, hydrogen sulfide, and carbon dioxide. J. Can. Petr. Tech., 6(1), 6
286
Rouher, O.S., and Barduhn, A.J. (1969) Hydrates of iso- and n-butane and their mixtures.
Desalination, 6, 57
Saito, S., Marshall, D.R., and Kobayashi, R. (1964) Hydrates at high pressures: Part II.
Application of statistical mechanics to the study of hydrates of methane, argon, and
nitrogen. AIChE J., 10(5), 734-740
Sargent, D.F., and Calvert, L.D. (1966) Crystallographic data for some new type II
clathrate hydrates. J. Phys. Chem., 70(8), 2689-2691
Shock, E.L., and Helgeson, H.C. (1988) Calculation of the thermodynamic and transport
properties of aqueous species at high pressures and temperatures: Correlation
algorithms for ionic species and equation of state predictions to 5 kbar and 1000C.
Geochim. Et Cosmochim. Acta, 52, 2009-2036
Sloan, E.D., Jr., Sparks, K.A., and Johnson, J.J. (1987) Two-phase liquid hydrocarbonhydrate equilibrium for ethane and propane. Ind. Eng. Chem. Res., 26(6), 1173-1179
Sloan, E.D., Jr. (1998) Clathrate Hydrates of Natural Gases, Marcel Dekker, Inc., NY
Sloan, E.D., Jr. (2002) Personal communication
Soave, G. (1972) Equilibrium constants from a modified Redlich-Kwong equation of
state. Chem. Eng. Sci., 27, 1197-1203
Sortland, L.D., and Robinson, D.B. (1964) The hydrates of methane and sulphur
hexafluoride. Can. J. Chem. Eng., 42, 38-42
Span, R., and Wagner, W. (1996) A new equation of state for carbon dioxide covering the
fluid region from triple-point temperature to 1100 K at pressures up to 800 MPa. J.
Phys. Chem. Ref. Data, 25(6), 1509-1596
Sparks, K.A., Tester, J.W., Cao, Z., and Trout, B.L. (1999) Configuration properties of
water clathrates: Monte Carlo and multidimensional integration versus the LennardJones and Devonshire approximation. J. Phys. Chem. B, 103(30), 6300-6308
Spencer, C. (2001) Personal communication
Stackelberg, M. von (1949) Feste Gashydrate. Die Naturwissenschatten, 36, 327-333
287
Stackelberg, M. von, and Mller, H.R. (1954) Fest Gashydrate II: Struktur und
Raumchemie. Zeitschrift fuer Elektrochemie, 58, 25-39
Stackelberg, M. von, and Meinhold, W. (1954) Fest Gashydrate III: Mischhydrate.
Zeitschrift fuer Elektrochemie, 58, 40-45
Stackelberg, M. von, and Frhbuss, H. (1954) Fest Gashydrate IV: Doppelhydrate.
Zeitschrift fuer Elektrochemie, 58, 99-104
Stackelberg, M. von (1954) Fest Gashydrate V: Die Bindungsenergien. Zeitschrift fuer
Elektrochemie, 58, 104-109
Stackelberg, M. von, and Jahns, W. (1954) Feste gashydrate
gitteraufweitungsarbeit. Zeitschrift fuer Elektrochemie, 58, 162-164
VI:
Die
Struik, D.J. (1969) A Source Book of Mathematics 1200-1800, Harvard University Press,
Cambridge, MA
Subramanian, S., Ph.D. Thesis (2000) Measurements of clathrate hydrates containing
methane and ethane using Raman spectroscopy. Colorado School of Mines, Golden,
CO, USA
Subramanian S., Kini, R., Dec, S.F., and Sloan, E.D., Jr. (2000a) Evidence of structure II
hydrate formation from methane + ethane mixtures. Chem. Eng. Sci., 55(11), 19811999
Subramanian, S., Ballard, A.L., Kini, R., Dec, S.F., and Sloan, E.D., Jr. (2000b)
Structural transitions in methane + ethane gas hydrates - Part I: Upper transition point
and applications. Chem. Eng. Sci., 55, 5763-5771
Sugahara, T., Morita, K., and Ohgaki, K. (2000) Stability boundaries and small hydratecage occupancy of ethylene hydrate system. Chem. Eng. Sci., 55, 6015-6020
Sum, A.K., Burruss, R.C., and Sloan, E.D., Jr. (1997) Measurement of clathrate hydrates
via Raman spectroscopy. J. Phys. Chem. B, 101(38), 7371-7377
Takeya, S., Nagaya, H., Matsuyama, T., Hondoh, T., and Lipenkov, V.Y. (2000) Lattice
constants and thermal expansion coefficient of air clathrate hydrate in deep ice cores
from Vostok, Antarctica. J. Phys. Chem. B, 104(4), 668-670
288
Tee, L.S., Gotoh, S., and Stewart, W.E. (1966) Molecular parameters for normal fluids.
Ind. Eng. Chem. Fund., 5(3), 363-366
Thomas, M., and Behar, E. (1994) Structure H hydrate equilibria of methane and
intermediate hydrocarbon molecules. Proc. 73rd GPA Conv., March 7-9, New
Orleans, Louisiana, 100
Thorwart, and Daubert, T.E. (1993) Unpublished results
Tohidi, B., Danesh, A., Tabatabaei, A.R., & Todd, A.C. (1997) Vapor-hydrate
equilibrium ratio charts for heavy hydrocarbon compounds. 1. Structure-II hydrates:
benzene, cyclopentane, cyclohexane, and neopentane. Ind. and Eng. Chem. Res.,
36(7), 2871-2874
Tse, J.S. (1987) Thermal expansion of the clathrate hydrates of ethylene oxide and
tetrahydrofuran. J. de Phys., 48(3), C1, 543-549
Tse, J.S. (1990) Thermal expansion of structure-H clathrate hydrates. J. Incl. Phen. Mol.
Rec. Chem., 8, 25-32
Tse, J.S., and Davidson, D.W. (1982) Intermolecular potentials in gas hydrates. Proc. 4th
Can. Permafrost Conf., March 2-6, 1981, Calgary, Alberta, Canada, 329-334
Tsonopoulos, C. (2001) Thermodynamic analysis of the mutual solubilities of
hydrocarbons and water. Fluid Phase Equil., 186, 185-206
Uchida, T., and Hayano, I. (1964) Studies on saline water conversion by freezing method
(III) Some observations on the formation of isobutene hydrate and its thermodynamic
properties. Reports Govt. Chem. Ind. Res. Inst., Tokyo, 59, 382-387
Unruh, C.H., and Katz, D.L. (1949) Gas hydrates of carbon dioxide-methane mixtures.
Trans. AIME, 186, 83
Waals, J.H. van der, and Platteeuw, J.C. (1959) Clathrate solutions. Adv. Phys. Chem., 2,
1-57
Wagner, W., Saul, A., and Pruss, A. (1994) International equations for the pressure along
the melting and along the sublimation curve of ordinary water substance. J. Phys.
Chem. Ref. Data, 23(3), 515-527
289
Wilcox, W.I., Carson, D.B., and Katz, D.L. (1941) Natural gas hydrates. Ind. Eng.
Chem., 33, 662
Wilson, G.M., Cunningham, J.R., Lofgren, B.S., and Messick, V.E. (1977) Measurement
of ethane and propane recovery and total fraction condensed for simulated natural gas
mixtures. GPA Research Report, RR-23, May 1977
Wu, B.-J., Robinson, D.B., and Ng, H.-J. (1976) Three and four-phase hydrate forming
conditions in the methane-isobutane-water system. J. Chem. Therm., 8, 461-469
Yamazaki, A. (1997) MITIs plan of R&D for technology of methane hydrate
development as domestic gas resources. Production, treatment, and underground
storage of natural gas. International Gas Union, 20th World Gas Conf., 355-370
Zele, S.R., Lee, S.-Y., and Holder, G.D. (1999) A theory of lattice distortion in gas
hydrates. J. Phys. Chem. B, 103(46), 10250-10257
291
The initialization of the Gibbs energy minimization algorithm requires a set of Kvalues that are composition independent. That is, we are seeking expressions for the
distribution of components between each phase in the calculation that is only dependent
on the component properties.
always effective. Therefore, two sets of ideal K-values are generated in the CSMGem
program, depending on the type of calculation to perform. The first set is generated using
composition independent correlations while the second is generated based on incipient
solid phases. Each set will be discussed in this appendix.
Ki =
Pc
Tc
xiV
= i exp 5.373 (1 + i ) 1 i
xiLhc
P
T
i salt, water.
A.1
292
between the vapor and liquid hydrocarbon phases. For water, we have regressed the ideal
K-values to the following equation
Ki =
xiV
a +a T
= 1 2 + a3 P
xiLhc
P
i = water.
A.2
The values for ai have been regressed to predictions of the actual K-value for many
water-hydrocarbon systems. Table A.1 lists the regressed values of ai (i=1,2,3).
Table A.1 Parameters for Equation A.2
a1
-133.67
a2
0.63288
a3
3.19211E-3
Since we do not know how salts distribute between the vapor and liquid hydrocarbon
phases, we make the arbitrary selection of
Ki =
xiV
=1
xiLhc
i = salts.
A.3
Note that an accurate expression for the distribution of the salts between the hydrocarbon
phases is not necessary since salts are infinitesimally present in either of them.
A.1.2 Vapor and Aqueous Phases
The approach taken to describe the distribution of components between the vapor
and aqueous phases follows. We can take the standard Raoults Law
P v
A.4
293
xiV Pi sat
iAq .
=
xiAq
P
A.5
Note that the saturation pressure, Pisat, and the infinite dilution activity coefficient, iAq
,
can be expressed with well-known correlations. Lee and Kesler (1975) used a twoparameter corresponding states equation to express the saturation pressure as
Pi sat = Pci exp [a1 + i a2 ]
A.6
where
a1 = a11 + a12
a2 = a21 + a22
Tci
T
Tci
T
+ a13 ln
T
T6
+ a14 6 ,
Tci
Tci
+ a23 ln
T
T6
+ a24 6 ,
Tci
Tci
T is in Kelvin, and Pisat is in bar. The parameters for Equation A.6 are given in Table
A.2.
Table A.2 Parameters for Equation A.6
a11
a12
a13
a14
5.927140
-6.096480
-1.288620
0.169347
a21
a22
a23
a24
15.25180
-15.68750
-13.47210
0.43577
Note that Equation A.6 is good for all hydrocarbons, predicting experimental data for
saturation pressure within 1 to 2 percent. For alcohols and water, however, the standard
Antoine equation is used for the saturation pressure. It can be expressed as
a2i
Pi sat = exp a1i
T + a3i
A.7
294
where T is in K and Pisat is in bar. The constants can be found in most thermodynamics
book. The set used in this work is given in Table A.3.
Table A.3 Parameters for Equation A.7
Water
Methanol
Ethanol
Ethylene Glycol
a1
12.048399
11.521061
11.904003
11.904003
a2
4030.18245
3391.96113
3578.90801
3578.90801
a3
-38.15
-43.15
-50.50
-50.50
Note that the parameters for ethylene glycol are assumed to be the same as those of
ethanol because we could not find a source for them. For salts, the saturation pressure is
assumed to be zero.
The correlation for the infinite dilution activity coefficient is given by Pierotti et
al. (1959) as
a
i = exp a1 + a2 N1 + 3
N1
A.8
where N1 is the total number of carbon atoms in molecule i (the solute in water). The
parameters for Equation A.8 are given in Table A.4.
Table A.4 Parameters for Equation A.8
n-paraffins
n-primary alcohols
a1
0.688
-0.995
a2
0.642
0.622
a3
0
0.558
The infinite dilution activity coefficient for water is assumed to be equal to one. Note
that the infinite dilution activity coefficient is not needed for salts since the saturation
pressure of salts is equal to zero in Equation A.5.
295
if i water
if i = water
A.9
We have used freezing point depression data to regress a pressure and temperature
dependent equation for the composition of water, which is
xwAq = 1 + a1 (T Tice ) + a2 (T Tice )
A.10
Tice = T0 + b1 ( P P0 ) + b2 ( P P0 ) .
A.11
where
2
In the above equation, T0 = 273.1576 K and P0 = 6.11657E-3 bar. The constants are
given below.
Table A.5 Parameters for Equations A.10 and A.11
a1
8.33076E-3
a2
3.91416E-5
b1
-7.404E-3
b2
-1.461E-6
Note that Equation A.11 represents the pressure dependence of the pure water ice point
while Equation A.10 represents the depression of the ice point due to composition of the
aqueous phase.
A.1.4 Salts and Aqueous Phases
Since solid salts are pure, we can define the following for the ideal K-value for
components in the solid salt phases
0
xiSalt
Ki =
= 1
xiAq
xiAq
if i salt
if i = salt
A.12
296
Note that the composition of the salt in water in Equation A.12 needs to be the saturation
composition of the salt at the given T and P. In this work, we have used salting out data
to regress a pressure and temperature dependent equation for saturation composition of
several salts, which is
xiAq = a1 + a2T + a3T 2 + a4T 3
A.13
ai = bi + 1 exp ci ( P 1)
i = 1,,4
A.14
The constants, bi and ci, are listed for each salt in Table A.6.
Table A.6 Parameters for Equation A.14
NaCl
KCl
CaCl2
b1
7.1811E-4
3.9347E-1
10.9
b2
1.1215E-3
-5.3151E-3
-9.6298E-2
b3
-4.3613E-6
2.2103E-5
2.9116E-4
b4
5.7463E-9
2.6414E-8
-2.9591E-7
NaCl
KCl
CaCl2
c1
-1.70E-5
-2.10E-5
-2.53E-5
c2
1.00E-9
1.00E-9
1.00E-9
c3
1.00E-10
1.00E-10
1.00E-10
c4
1.00E-14
1.00E-14
1.00E-14
Note that the bi parameters account for the temperature dependence while the ci
parameters account for the pressure dependence.
A.1.5 Vapor and Hydrate Phases
Several researchers have developed expressions for the distribution of guest
molecules between the vapor and hydrate phases. In 1942, Carson and Katz (1942)
produced K-value charts for several hydrate formers based solely on experimental data.
It should be noted that their charts were developed without the knowledge of different
hydrate structures. That is, their K-values were structure independent. Following this
same approach, Unruh and Katz (1949), Noaker and Katz (1954), Jhaveri and Robinson
297
(1965), Wu et al. (1976), and Poettmann (1984) produced K-value charts for other
hydrate formers to compliment Carson and Katz original work.
summarized this work via an equation. In 1989, Mann et al. (1989) developed structure
dependent K-values for sI and sII hydrates with Tohidi et al. (1997) complimenting this
work for some obscure sII hydrate formers.
The K-values given in each of these works is in water-free form. That is, the
composition of the vapor and the hydrate are on a water-free basis.
K iwf =
xiVwf
xiHwf
A.15
For this work, however, we need a K-value that is not water-free. What follows is a brief
outline of the method we used to convert these K-values from a water-free basis.
In this work, we have developed an expression to convert the water-free K-values
to actual K-values using simple mole fraction balances.
approximation, the content of water in the vapor phase is negligible and therefore, we
need not convert the vapor composition. However, the water content of the hydrate is
significant. The water-free composition of the hydrate is given as
xiHwf =
xiH
x
j =1
j w
jH
xiH
.
1 xwH
A.16
A.17
xiV
x
x wf
K iwf
= wf iV
wf iV
=
.
xiH xiH (1 xwH ) xiH (1 xwH ) (1 xwH )
A.18
Equation A.18 gives the actual distribution of components between the vapor and hydrate
phases. For components that are not present in the hydrate phase, the K-value is assumed
to be infinite (1E100). Note that Kiwf is the water-free K-value and xwH is the composition
298
of water in the hydrate. In this work, it is sufficient to assume that the composition of
water in the hydrate phase is constant: 0.88, 0.90, and 0.94 for sI, sII, and sH hydrates,
respectively. The K-value for water must be treated in a different manner, since we only
have the composition of it in the hydrate phase. In this case, we use the K-values for the
distribution of water between the vapor and aqueous phases and ice and aqueous phases.
That is,
Kw =
K wV Aq
xwV
=
xwH xwH K wIce Aq
A.19
where KwV-Aq and KwIce-Aq are found from Equations A.5 and A.9, respectively. What
follows are the expressions for the water-free K-values used in this work.
Vapor and sI Hydrate Phases
The expression for sI hydrate K-values reported by Mann et al. is used in this
work. Note that, in their original work, Mann et al. gave separate expressions for each sI
hydrate former. Each expression given in log2 form and was in units of Fahrenheit and
psia for the temperature and pressure, respectively.
expressions and developed our own expression for the water-free K-values in term of the
natural log and in units of Kelvin and bar. Equation A.20 is the expression we use in this
work.
K iwf =
xiVwf
wf
xisI
a1 + a2 ln P + a3 (ln P )2
2
3
a4 + a5 ln P + a6 ( ln P ) + a7 ( ln P )
= exp
+ .
a8 a9
ln P a
P P
T
T
A.20
299
CH4
C2H6
CO2
H2S
N2 (w/out H2S)
N2 (w/ H2S)
27.474169
-0.8587468
0
6604.6088
50.8806
1.57577
-1.4011858
0
0
0
0
0
0
14.81962
6.813994
0
3463.9937
2215.3
0
0
0
0
0
0
0
0
15.8336435
3.119
0
3760.6324
1090.27777
0
0
0
0
0
0
0
0
31.209396
-4.20751374
0.761087
8340.62535
-751.895
182.905
0
0
0
0
0
0
0
173.2164
-0.5996
0
24751.6667
0
0
0
1.441
-37.0696
-0.287334
-2.07405E-5
0
0
71.67484
-1.75377
-0.32788
25180.56
0
0
0
56.219655
-140.5394
0
8.0641E-4
366006.5
978852
Note that two sets of parameters are given for nitrogen: one set if hydrogen sulfide is
present and one set if hydrogen sulfide is not present. Values for xenon were not
reported and were therefore taken to be those of methane in this work.
Vapor and sII Hydrate Phases
The expression for sII hydrate K-values reported by Mann et al. were not used in
this work due to the fact that they gave unrealistic predictions for propane and i-butane
hydrates. Instead, the expression developed by Sloan was used. Although the expression
developed by Sloan is structure independent, the predictions for sII hydrates are sufficient
for the initialization procedure. We also use the expression given by Tohidi et al. for
benzene. The expression given by Sloan is given in units of Fahrenheit and psia for the
temperature and pressure, respectively, while that by Tohidi et al. is given in Kelvin and
MPa. We have combined both expressions into one, which is in units of Fahrenheit and
bar. Equation A.21 is the expression we use in this work.
300
K iwf =
xiVwf
wf
xisII
a4 a5
2
2
a1 + a2T + a3 P + T + P + a6TP + a7T + a8 P +
P
P a11
T
T2
P
= exp
a9 + a10 ln + 2 + a12 + a13
+ a14 2 + .
T
T P
P
P
T
3
T
P
3
4
P
T
A.21
CH4
C2H6
C3H8
n-C4H10
i-C4H10
N2
CO2
H 2S
Benzene
-0.45872
0
0
31.6621
-3.4028
-7.702E-5
0
0
1.8641
-0.78338
0
0
0
-77.6955
0
-2.3E-7
-6.102E-5
0
0
3.21799
0
0
-290.283
181.2694
0
0
-1.893E-5
1.882
-1.19703
-402.166
-4.897688
0.0411205
-68.8018
25.6306
0
0
0
0
-7.51966
0
0
47.056
0
-1.697E-5
7.145E-4
0
0
0.12348
79.34
0
0.0160778
0
-14.684
5.50E-6
0
0
0
-37.211
0.86564
0
732.2
0
0
0
1.9711E-3
-15.6144
0
0
-4.56576
0
0
300.5535
0
0.0151942
-1.26E-6
0
-9.55128
0
0
0
0
0
0.001251
2.1036E-6
2.40904
-2.75945
0
0
0
0
-0.28974
0
-1.6476E-3
-1.0E-8
0
1.78857
0
-0.019667
-6.187
0
0
0
5.259E-5
0
0
0
0
0
0
192.39
0
3.051E-5
1.1E-7
0
9.0242
0
0
-207.033
0
6.7588E-4
-6.992E-3
-6.0794E-4
-9.026E-2
0
0
0
0.0186833
0
0
8.82E-5
7.78015E-3
0
0
-6.42956
0.06192
0
82.627
0
-1.0718E-4
0
0
3.493522
-0.64405
0
0
0
-184.257
0
-1.30E-6
0
0
0
-5.9757
-1.02387E-3
0
0
12.96295
0
7.40216E-4
0
0
0
0
0.0282
0
0
0
1.11529E-6
0
0
-0.12898
301
the small guest (sI former), it does not give that of the large sH hydrate former.
However, this approach gives a reasonable estimate.
A.1.6 Derivatives of Ideal K-Values for Each Phase
For flashes in which we are seeking the temperature or pressure at which a given
phase forms, we need expressions for the derivatives of each K-value with respect to the
temperature or pressure. What follows are each of these derivatives of each of the Ideal
K-values given previously.
Vapor and Liquid Hydrocarbon Phases
Derivatives of Equation A.1
Tc
K i
= 5.373K i (1 + i ) 2i
T
T
A.22a
K i
1
= Ki
P
P
A.22b
K i a2
=
T
P
for i = water
A.23a
K i
a +a T
= a3 1 2 2
P
P
for i = water
A.23b
i = hydrocarbons
A.24a
Tc
( a12 + a22 i ) 2i +
T
K i
1
= K i ( a13 + a23 i ) +
T
T
T5
6 ( a14 + a24 i ) 6
Tci
302
a2i
K i
= Ki
T
T + a3i
K i
1
= Ki
P
P
A.24b
A.24c
if i water
0
K i
= 1 xwAq
P 2
xwAq P
if i water
if i = water
if i = water
A.25a
A.25b
= a1 + 2a2 (T Tice )
= a1 + 2a2 (T Tice )
A.25c
Tice
P
A.25d
where
Tice
= b1 + 2b2 ( P 1) .
P
A.25e
if i salt
if i = salt
A.26a
if i salt
if i = salt
A.26b
303
= a2 + 2a3T + 3a4T 2
=
A.26c
a1 a2
a
a
+
T + 3 T2 + 4 T3
P P
P
P
A.26d
where
ai
= ci exp ci ( P 1)
P
i = 1,,4.
A.26e
K i
1
K iwf
=
T (1 xwsI ) T
i water
A.27a
K i
K iwf
1
=
P (1 xwsI ) P
i water
A.27b
2
T
xwH K wIce Aq T
xwH K wIce
T
Aq
A. 27c
K wV Aq
K wV Aq K wIce Aq
K w
1
=
2
P
xwH K wIce Aq P
xwH K wIce
P
Aq
A. 27d
a + a ln P + a ( ln P )2 + a ( ln P )3
6
7
4 5
T2
ln P
a13
a10 2a12 3 2 3
T
T
A. 27e
304
a5 + 2a6 ln P + 3a7 ( ln P )
a
ln
P
2 + 2a
3
P
P
PT
a8 2a9
a12
+
a
+
11
P 2 P3
PT 2
A. 27f
K
= K iwf
T
a4
a9 P
1 a12
a2 T 2 + a6 P + 2a7T T 2 a10 T + P +
3
2a13T 2a14 P + a15 + 3a T 2 2a17 P + 4a T 3
16
18
P
T3
P3
T3
A. 27g
K
= K iwf
T
a5
a9 a10 2a11
a3 P 2 + a6T + 2a8 P + T + P P 3
2 + 2 4 +
+
P2
P
T
P
T2
P
A. 27h
K
= K iwf
P
wf
i
wf
i
xiV
,
xiLhc
K i ,2 =
xiV
,
xiAq
K i ,6 =
xiIce
,
xiAq
K i ,7 =
xiNaCl
x
, K i ,8 = iKCl ,
xiAq
xiAq
K i ,3 =
xiV
,
xisI
K i ,4 =
K i ,9 =
xiV
,
xisII
K i ,5 =
xiV
,
xisH
xiCaCl2
xiAq
The form of the K-values needed for the initialization of the Gibbs energy minimization
is
K im =
xim
.
xir
A.28
305
Therefore, we seek expressions of the form in Equation A.28 with respect to each phase.
During the iterative process, several reference phases may be tested. Therefore, it is
necessary to determine the reference phase specific form of the K-values for each
possible reference phase. Note that all phases may be a reference phase except for the
hydrate phases. Table A.9 gives the forms of the K-values for each possible reference
phase.
Table A.9 K-values for each phase with respect to each possible reference phase
V
Lhc
Aq
KiV
K i ,1
K i ,2
KiLhc
1
K i ,1
KiAq
1
K i ,2
K i ,2
K i ,6
K i ,1 K i ,6
K i ,2
K i ,2
KisI
1
K i ,3
KisII
1
K i ,4
KisH
1
K i ,5
K i ,7
K i ,1 K i ,7
K i ,2
K i ,2
K i ,8
K i ,1 K i ,8
K i ,2
K i ,2
K i ,9
K i ,1 K i ,9
K i ,2
K i ,2
KiIce
KiNaCl
KiKCl
KiCaCl2
Reference Phase
Ice
NaCl
K i ,2
K i ,2
K i ,7
K i ,6
KCl
K i ,2
CaCl2
K i ,2
K i ,8
K i ,9
K i ,2
K i ,2
K i ,2
K i ,2
K i ,2
K i ,1
K i ,1 K i ,6
K i ,1 K i ,7
K i ,1 K i ,8
K i ,1 K i ,9
1
K i ,6
1
K i ,7
1
K i ,8
1
K i ,9
K i ,6
K i ,6
K i ,6
K i ,6
K i ,7
K i ,8
K i ,9
K i ,1
K i ,2
K i ,2
K i ,2
K i ,2
K i ,2
K i ,3
K i ,3
K i ,3 K i ,6
K i ,3 K i ,7
K i ,3 K i ,8
K i ,3 K i ,9
K i ,1
K i ,2
K i ,2
K i ,2
K i ,2
K i ,2
K i ,4
K i ,4
K i ,4 K i ,6
K i ,4 K i ,7
K i ,4 K i ,8
K i ,4 K i ,9
K i ,1
K i ,2
K i ,2
K i ,2
K i ,2
K i ,2
K i ,5
K i ,5
K i ,5 K i ,6
K i ,5 K i ,7
K i ,5 K i ,8
K i ,5 K i ,9
K i ,7
K i ,7
K i ,8
K i ,9
K i ,1
K i ,7
K i ,8
K i ,9
K i ,7
K i ,6
K i ,8
K i ,8
K i ,6
K i ,7
K i ,9
K i ,9
K i ,9
K i ,6
K i ,7
K i ,8
K i ,8
K i ,9
1
306
V
K iV
y
KiLhc
y
K iAq
y
K iIce
y
Aq
K i ,2
y
1 K i ,2 K i ,2 K i ,1
2
K i ,1 y
K i ,1 y
1 K i ,1
K i2,1 y
1 K i ,2
K i2,2 y
1 K i ,1 K i ,1 K i ,2
2
K i ,2 y
K i ,2 y
1 K i ,6 K i ,6 K i ,2
2
K i ,2 y
K i ,2 y
K i ,6 K i ,1
K i ,2 y
K i ,1 K i ,6
K i ,2 y
K i ,1K i ,6 K i ,2
K
2
i ,2
K i ,6
y
K isI
y
1 K i ,3
K i2,3 y
1 K i ,1 K i ,1 K i ,3
2
K i ,3 y
K i ,3 y
1 K i ,2 K i ,2 K i ,3
2
K i ,3 y
K i ,3 y
K isII
y
1 K i ,4
K i2,4 y
1 K i ,1 K i ,1 K i ,4
2
K i ,4 y
K i ,4 y
1 K i ,2 K i ,2 K i ,4
2
K i ,4 y
K i ,4 y
KisH
y
1 K i ,5
K i2,5 y
1 K i ,1 K i ,1 K i ,5
2
K i ,5 y
K i ,5 y
1 K i ,2 K i ,2 K i ,5
2
K i ,5 y
K i ,5 y
K iNaCl
y
1 K i ,7 K i ,7 K i ,2
2
K i ,2 y
K i ,2 y
K i ,7 K i ,1
K iKCl
y
1 K i ,8 K i ,8 K i ,2
2
K i ,2 y
K i ,2 y
K i ,8 K i ,1
K iCaCl2
1 K i ,9 K i ,9 K i ,2
2
K i ,2 y
K i ,2 y
K i ,2 y
K i ,2 y
K i ,9 K i ,1
K i ,2 y
+
+
+
K i ,1 K i ,7
K i ,2 y
K i ,1 K i ,8
K i ,2 y
K i ,1 K i ,9
K i ,2 y
K i ,1 K i ,7 K i ,2
K
2
i ,2
K i ,1 K i ,8 K i ,2
K
2
i ,2
K i ,1 K i ,9 K i ,2
K
2
i ,2
K i ,7
y
K i ,8
y
K i ,9
y
307
Table A.10 Derivatives of K-values for each phase and each possible reference
phase with respect to y (temperature or pressure) (continued)
K iV
y
KiLhc
y
K iAq
y
K i ,1 K i ,r y
K i ,1 K i ,r y
K i ,1 K i2,r y
1 K i ,r
K i2,r y
K iIce
y
1 K i ,6 K i ,6 K i ,r
2
(= 0 if r = 6)
K i , r y
K i , r y
K isI
y
K
K i ,3
K i ,2 K i ,r
1 K i ,2
2 i ,2
K i ,3 K i ,r y
K i ,3 K i ,r y
K i ,3 K i2,r y
K isII
y
K
K i ,4
K i ,2 K i ,r
1 K i ,2
2 i ,2
K i ,4 K i ,r y
K i ,4 K i , r y
K i ,4 K i2,r y
KisH
y
K
K i ,5
K i ,2 K i ,r
1 K i ,2
2 i ,2
K i ,5 K i ,r y
K i ,5 K i ,r y
K i ,5 K i2,r y
K iNaCl
y
1 K i ,7 K i ,7 K i ,r
2
(= 0 if r = 7)
K i , r y
K i ,r y
K iKCl
y
1 K i ,8 K i ,8 K i , r
2
(= 0 if r = 8)
K i ,r y
K i , r y
K iCaCl2
1 K i ,9 K i ,9 K i ,r
2
(= 0 if r = 9)
K i , r y
K i ,r y
308
strict of an assumption. However, it may not work well if solid phases are present in
large amounts.
In obtaining these ideal K-values, the fugacity of components in each phase must
be calculated. As stated in Chapter 2 (Section 2.4), this approach can be time-intensive
and therefore is not used unless necessary. What follows is the approach in obtaining the
K-values.
A.2.1 Vapor and Liquid Hydrocarbon Water-Free Compositions
The feed is first converted to a water-free feed via the following equation:
zi =
zi
i, j water, salts.
z
j =1
A.28
To initialize the composition and phase fractions of the vapor and liquid hydrocarbon
phases Equation A.1 is used to give the distribution of the components. The following
procedure is used:
1. find Pdew using Rachford-Rice equations
z
2. if P < Pdew then xiV = zi , xiL = i , and V = 1 (stop)
Ki
bub
3. find P using Rachford-Rice equations
4. if P > Pbub then xiL = zi , xiV = zi Ki , and V = 0 (stop)
5. find V using Rachford-Rice equations
zi
6. xiL =
and xiV = xiL K i (stop)
V ( K i 1) + 1
This procedure gives the distribution of all components but water and salts in the
hydrocarbon phases. It also gives the relative amounts of V (V) and Lhc (1-V). After
this has been done, the feed is converted back to the normal feed.
309
xiV
1 zwater
i water
A.29
i water
A.30
Lhc-Aq Flash
zi =
xiLhc
1 zwater
where zwater is the overall composition of water in the feed. Note that alcohols will
preferentially go into the aqueous phase at most conditions. However, if salts are present
in the feed, the following adjustments are made to the aqueous phase composition:
xiAq = xiAq 1 z j
j = salts
i salts
A.31
xiAq = zi
i = salts.
A.32
and
310
xik
.
xir
A.33
The choice of the reference phase is arbitrary. However, an aqueous, vapor, or liquid
hydrocarbon phase is suggested.
A.2.6 A Note on the Use of the Incipient Solid Phase Based Ideal K-Values
Since the procedure used in obtaining this set of ideal K-values is an iterative one,
derivatives cannot be taken. This is not a problem for a T-P flash. However, it does pose
problems in T- and P- flashes, in which the T or P is iteratively calculated. If this set
of ideal K-values is needed for this type of flash calculation, the T or P is calculated
independent of the ideal K-values. What follows is a description of the procedure used to
determine temperature or pressure at which a phase first forms.
Saturation Point of Aqueous Phase (Aq = 0)
The saturation T/P at a given P/T is calculated via rearrangement of Equation A.5.
The composition of water in the vapor phase is assumed to be the salt-free composition of
the feed. The composition of water in the aqueous phase is assumed to the hydrocarbonfree composition of the feed. The saturation T is of the form:
311
T sat =
a2
a3
xwV
a1 ln
P
xwL
w
A.34
xwAq
xwV
a
exp a1 2
T + a3
A.35
The
A.36
where ai (i=1,2) are given in Table A.5. To determine the ice point P at T, calculate Tice
from
Tice = T +
A.37
using the assumed xwAq. Plug this into Equation A.11 for Tice and solve for P.
P = 1
A.38
312
a x
a3 2 a2
T + T + 1 SAq = 0
a4
a4
a4
A.39
The method of Cardan (Struik, 1969) is used to solve for T. To determine the salt point P
at T, solve for P using a combined form of Equations A.13 and A.14:
P = 1+
A.40
where bi and ci (i=1,2,3,4) are given in Table A.6 and the expressions for ai (i=2,3,4) are
given by Equation A.13. Note that the ai terms are pressure dependent, making the
solution to Equation A.40 an iterative approach.
Hydrate Point (H = 0)
The initial guess for the hydrate formation T/P uses the generalization of gas
gravity suggested by Katz (1945). Looking at most of the hydrate formation data in the
literature as a function of specific gravity, we have developed an equation that gives a
decent initial guess for the T or P based solely on the gravity of the feed mixture. The
equation, explicit in P or T, is of the form
P = exp [a + bT ]
A.41
ln P a
b
A.42
T=
where,
a ( SG ) = a1 + a2 SG + a3 SG 2 + a4 SG 3 ,
A.43
313
b ( SG ) = b1 + b2 SG + b3 SG 2 + b4 SG 3 ,
SG =
zi MWi
MWair
i = guest
j = guest
zj
A.44
A.45
and MWair = 28.95 g/mol. Katz had originally used the gas gravity, in which zi in
Equation A.45 was the composition of the vapor phase. However, in this work, we do
not know the composition of the vapor phase, and therefore use the normalized feed of
the hydrate guests. The constants used in Equations A.43 and A.44 are given in Table
A.11.
Table A.11 Parameters for Equations A.43 and A.44
a
b
1
-0.2068
74.473
2
1.0960
-347.130
3
-1.1790
367.240
4
0.4236
-129.750
Figure A.1 shows the predictions from Equation A.41 versus known experimental data.
The hydrate formation pressure is plotted versus the gravity (from Equation A.45) of the
hydrate formers in the feed at two temperatures.
314
100
Methane Hydrates
Natural Gas
Hydrates
Ethane Hydrates
10
T = 277.6 K
T = 280.4 K
Equation A.41
Propane Hydrates
0.5
0.7
0.9
1.1
1.3
1.5
Gravity
Figure A.1 Plot of hydrate formation pressure versus gravity of hydrate formers
As can be seen, Equation A.41 gives a decent approximation of the hydrate formation
temperatures and pressures.
It should be noted that all equations used in this section for the initial formation
temperature or pressure of a phase are used as a starting point for the initialization
procedure described in Figure 2.2.
315
In minimizing the Gibbs energy at a given set of K-values via Newtons method,
the following equation needs to be solved
# #
J x = f
B.1
which is derived from the least squares problem. The matrix J is the Jacobian matrix, x
is the correction vector, and f is the objective vector. Note that we are solving for x
such that f = 0. The Jacobian matrix is just the derivative of the objective functions with
respect to each element in x. For a flash at a given temperature and pressure, Equation
B.1 is of the form
E1
1
1
S1
&
E1
E1
1
&
&
E
1
&
&
S1
S1
1
&
&
S
1
&
E1
E
1 1
E $ $
=
S1 1 S1
$ $
S
$
B.2
where the objective functions Ek and Sk are defined by Equations 2.16 and 2.10,
respectively. The terms preceded by are corrections to the actual terms. For example,
316
j(n+1) = j(n)-j, where the superscripts refer to the iteration number. The derivatives
shown in Equation B.2 are given in Chapter 2 of this thesis. The Jacobian matrix size is
2 x 2. In this work, Kramers rule is used to solve for the correction vector.
For a temperature or pressure and specified phase fraction, j, flash, the objective
functions to solve for are E1E and S1Sj-1, Sj+1S. The variables to solve for are
1 j-1, T or P, j+1 , and 1 j-1, j+1 . Note that j is equal to the specified
phase fraction and j is equal to zero, and therefore do not need to be solved for.
However, the objective function Ej is still needed as a constraint whereas Sj is not. The
Jacobian matrix size is 2-1 x 2-1.
For an expansion calculation described in Chapter 2, the objective functions to
solve for are E1E, S1S, and Eh
or s.
or s
is defined by
Equation 2.22. The variables to solve for are 1 , 1 , and T. The Jacobian
matrix size is 2+1 x 2+1.
317
This appendix lists each data set used in the optimization of parameters in the
fugacity models. The listing is here, as opposed to the body of the thesis, for ease of
reading. Two types of data are listed here: water content data and hydrate formation data.
All other types of data used in the optimization are listed in the body of the thesis as it
was used (i.e. triple points of ice and hydrate phase properties).
Researcher
Olds et al. (1942), Kremer (1982)
Anthony and McKetta (1968)
Reamer et al. (1943), Parrish et al. (1982)
Parrish et al. (1982)
Parrish et al. (1982), Sloan et al. (1987)
Black et al. (1948), Brooks et al. (1951)
Black et al. (1948), Parrish et al. (1982)
Black et al. (1948), Polak and Lu (1973)
Black et al. (1948)
318
Table C.1 Water content data used for optimization of interaction parameters
(continued)
Gas
Benzene
n-Hexane
Neohexane
2,3-Dimethylbutane
Toluene
n-Heptane
Ethylcyclohexane
n-Octane
n-Nonane
n-Decane
Nitrogen
Hydrogen Sulfide
Carbon Dioxide
Researcher
Staveley et al. (1943), Black et al. (1948), Englin et al. (1965),
Polak and Lu (1973), Tsonopoulos and Wilson (1983)
Black et al. (1948), Englin et al. (1965), Polak and Lu (1973),
Tsonopoulos and Wilson (1983)
Polak and Lu (1973)
Englin et al. (1965), Polak and Lu (1973)
Englin et al. (1965), Polak and Lu (1973)
Black et al. (1948), Englin et al. (1965), Schatzberg (1963),
Polak and Lu (1973)
Brady et al. (1982), Heidman et al. (1985)
Black et al. (1948), Englin et al. (1965), Polak and Lu (1973),
Brady et al. (1982), Heidman et al. (1985)
Schatzberg (1963)
Schatzberg (1963)
Gillespie and Wilson (1980)
Gillespie and Wilson (1980), Kremer (1982)
Kremer (1982)
It should be noted that the data by Anthony and McKetta (1968) was for ethylene+ethane
gas mixtures. However, the interaction parameter for ethane-water was determined first
using data for ethane+water mixtures, and then the ethylene-water interaction parameter
was regressed to the data by Anthony and McKetta. For components that are not listed in
Table C.1, the interaction parameters were determined via molecular weight and carbon
number correlations (Tsonopoulos, 2001).
319
Hydrate
Methane
sI
Ethylene
sI
Ethane
sI
Propylene
sII
Propane
sII
i-Butane
sII
Nitrogen
sI and sII
Hydrogen
Sulfide
sI
T Range (K)
273.2-286.7
273.7-285.9
295.7-302.0
285.7-301.6
290.2-301.6
273.2-294.3
283.2-288.7
275.2-291.2
273.3-286.0
275.4-281.2
273.4-286.4
273.2-290.8
279.5-303.8
273.4-287.0
273.7-286.5
279.9-287.4
277.6-282.5
277.5-286.5
278.8-288.2
277.8-287.2
273.0-274.1
273.2-278.0
273.7-277.1
274.3-277.2
274.3-278.9
274.2-278.4
273.9-278.4
274.2-278.2
273.6-278.0
273.2-275.1
273.2-275.0
274.4-274.6
272.0-291.0
277.6-286.7
273.2-281.1
283.2-302.7
272.8-302.7
P Range (bar)
26.4-108.0
27.7-97.8
339.9-775.0
96.2-680.9
159.0-654.0
26.5-285.7
71.0-131.1
30.2-185.5
26.9-100.4
28.7-61.0
26.8-105.7
5.6-49.1
11.7-920.0
5.5-30.5
5.1-27.3
9.7-33.0
8.1-15.5
7.8-26.2
9.5-33.6
8.5-30.8
4.7-6.0
1.7-4.7
1.8-3.9
2.4-4.1
2.1-5.5
2.1-5.4
1.9-5.6
2.2-5.1
2.1-5.1
1.1-1.7
1.2-1.7
1.3-1.6
144.8-958.6
249.3-637.1
162.7-351.6
3.1-22.4
0.9-22.4
Researcher
Roberts et al., 1940
Deaton and Frost, 1946
Kobayashi and Katz, 1949
McLeod and Campbell, 1961
Marshall et al., 1964
Jhaveri and Robinson, 1965
Galloway et al., 1970
Verma, 1974
de Roo et al., 1983
Thakore and Holder, 1987
Adisasmito et al., 1991
Diepen and Scheffer, 1950
Sugahara et al., 2000
Roberts et al., 1940
Deaton and Frost, 1946
Reamer et al., 1952
Galloway et al., 1970
Holder and Grigoriou, 1980
Holder and Hand, 1982
Avlonitis, 1988
Clark et al., 1964
Miller and Strong, 1946
Deaton and Frost, 1946
Reamer et al., 1952
Robinson and Mehta, 1971
Kubota et al., 1972
Verma, 1974
Thakore and Holder, 1987
Patil, 1987
Schneider and Farrar, 1968
Rouher and Barduhn, 1969
Thakore and Holder, 1987
Van Cleeff and Diepen, 1960
Marshall et al., 1964
Jhaveri and Robinson, 1965
Bond and Russell, 1949
Sellek et al., 1952
320
Hydrate
Carbon Dioxide
sI
Xenon
sI
T Range (K)
273.7-282.9
277.2-283.1
271.8-283.2
273.9-283.3
271.6-283.2
279.6-282.8
274.3-282.9
273.2-323.4
290.5-320.6
P Range (bar)
13.2-43.2
20.4-45.0
10.5-45.0
13.8-44.7
10.4-45.1
27.4-43.6
14.2-43.7
1.5-942.5
9.3-690.2
Researcher
Deaton and Frost, 1946
Unruh and Katz, 1949
Larson, 1955
Robinson and Mehta, 1971
Vlahakis et al., 1972
Ng and Robinson, 1985
Adisasmito et al., 1991
Aaldijk, 1971
Ohgaki et al., 2000
Table C.3 Data used for binary guest hydrates of sI and sII
Hydrate Formers
Hydrate
Methane, Ethane
sI and sII
Methane, Propane
sII
Methane, i-Butane
sII
Methane,
n-Butane
Methane, Nitrogen
Methane, CO2
Methane, H2S
Methane, Benzene
Ethane, Propane
Ethane, CO2
Propane, i-Butane
Propane, Nitrogen
Propane, CO2
sII
sI
sI
sI
sII
sI and sII
sI
sII
sII
sII
T Range (K)
274.8-283.2
279.4-287.8
274.8-283.2
274.4-282.3
274.8-277.6
273.8-290.9
274.8-280.4
276.0-287.6
273.2-295.2
275.5-285.7
273.7-287.6
276.5-295.4
275.5-287-5
273.1-283.3
273.5-287.8
275.3-277.9
275.1-289.2
P Range (bar)
9.5-56.7
9.9-30.8
2.7-30.1
2.6-9.5
13.2-18.4
1.6-100.7
20.5-40.8
20.5-110.5
39.0-343.3
19.9-70.0
14.5-109.5
10.3-67.9
15.1-85.7
4.4-20.3
5.7-40.8
21.3-49.0
2.6-180.9
Researcher
Deaton and Frost, 1946
Holder and Grigoriou, 1980
Deaton and Frost, 1946
Verma et al., 1974
Deaton and Frost, 1946
Wu et al., 1976
Deaton and Frost, 1946
Ng and Robinson, 1976
Jhaveri and Robinson, 1965
Unruh and Katz, 1949
Adisasmito et al., 1991
Noaker and Katz, 1954
Danesh et al., 1993
Holder and Hand, 1982
Adisasmito and Sloan, 1992
Paranjpe et al., 1987
Ng et al., 1977, 1978
273.8-286.2
273.7-282.0
3.2-42.7
2.2-38.2
321
T Range (K)
275.2-279.0
274.0-277.4
275.7-280.8
276.2-281.4
276.5-280.8
279.2-287.8
278.2-287.0
276.0-282.2
285.4-288.2
275.0-282.8
275.9-280.8
282.6-286.4
281.4-290.4
280.2-287.4
280.2-290.4
275.6-281.2
282.6-290.3
277.1-287.1
273.6-290.6
275.6-280.9
286.6-290.0
274.8-281.3
280.6-286.4
275.8-281.0
282.0-290.0
280.2-293.2
283.6-286.0
P Range (bar)
26.5-41.5
22.4-35.0
25.3-48.1
20.2-38.7
22.0-38.1
32.2-100.1
26.4-86.3
16.0-33.4
52.2-75.1
14.2-37.5
20.8-38.0
49.5-81.9
33.9-109.3
35.9-91.3
30.0-112.0
16.0-33.6
39.9-105.0
20.4-73.9
13.5-119.3
14.8-27.0
37.9-71.5
17.3-39.3
36.2-72.8
18.7-34.3
40.0-113.2
20.0-115.3
63.0-89.0
Researcher
b
g
c
c
c
d
e
b
d
g
c
d
f
d
a
c
d
h
i
c
d
c
d
c
d
f
f
323
The results are given as the average error either in temperature or pressure. The
average error in temperature and pressure are calculated as
(T
predicted
TError =
# data points
-Texperimental )
PError =
# data points
(P
predicted
# data points
-Pexperimental )
Pexperimental
100%
# data points
and are given in Kelvin and bar, respectively. The temperature and pressure ranges for
all results are reported in K and bar, respectively. The following abbreviations are used
for each hydrate program: CG (CSMGem), CH (CSMHYD), DB (DBRHydrate), MF
(Multiflash), and PV (PVTsim). The number of data points used in the comparisons is
given in parenthesis.
CG
1.89
-0.04
0.02
1.95
0.67
1.65
-9%
0%
0%
-17%
0%
-6%
5%
-2%
-12%
-40%
-6%
-16%
NA
0.25
1.15
4.73
-0.44
-0.25
1.14
-0.18
-0.82
NA
0.08
2.09
-0.01
-0.08
0.30
6.48
-0.30
0.83
0.09
-0.33
-0.26
3.13
0.67
1.65
-7%
2%
12%
NA
6%
4%
-1%
1%
-3%
-28%
-1%
-9%
PV
-16%
4%
3%
NA
3%
1%
324
Table D.2 Ethylene hydrate formation T/P at experimental P/T (78 pts)
T and P Range
T
P
269-273
0-5.3
273-292
5.3-100
292-304
100-1000
304-328
1000-5000
All Aq-H-V Data<1000
All Data
NA
0.08
0.69
7.71
0.27
2.38
0.57
-0.11
-0.01
NA
-0.09
-0.01
0.61
-0.43
-0.31
5.45
-0.40
1.25
NA
NA
NA
NA
NA
NA
CG
PV
-2%
6%
-4%
-1%
3%
1%
-6%
-2%
-9%
-27%
-4%
-10%
-3%
9%
7%
-17%
9%
1%
NA
NA
NA
NA
NA
NA
PV
3%
5%
61%
7%
NA
5%
14%
-3%
2%
0%
NA
1%
1%
Table D.3 Ethane hydrate formation T/P at experimental P/T (112 pts)
T and P Range
T
P
200-273
0-5
273-288
5-33
288-289
33-100
289-299
100-1000
299-324
1000-5000
All Aq-H-V Data
All Data
CG
0.63
0.08
-0.09
-0.08
-1.24
0.08
-0.12
-3%
-1%
11%
0%
8%
-1%
2%
2%
-9%
-32%
-38%
-17%
-9%
-13%
NA
0.69
0.95
1.81
3.23
0.69
1.20
-0.02
0.05
-0.77
-2.28
NA
0.05
-0.17
0.71
0.43
0.32
0.62
-0.78
0.43
0.25
-0.24
-0.34
-0.59
-0.06
-0.99
-0.34
-0.47
1%
1%
27%
15%
NA
1%
6%
-2%
-6%
-21%
-17%
3%
-6%
-6%
Table D.4 Propylene hydrate formation T/P at experimental P/T (19 pts)
T and P Range
T
P
273-274.3
4.6-6.1
274.5-275
6.1-11
All Aq-H-V Data
All Data
CG
-0.04
-0.57
-0.04
-0.12
1%
NA
1%
1%
NA
NA
NA
NA
-0.08
1.62
-0.08
0.18
0.27
-0.30
0.27
0.18
NA
NA
NA
NA
1%
NA
1%
1%
PV
-6%
NA
-6%
-6%
NA
NA
NA
NA
PV
0%
15%
0%
2%
Table D.5 Propane hydrate formation T/P at experimental P/T (92 pts)
T and P Range
T
P
273-278
1.5-5.4
278-278
5.4-170
All Aq-H-V Data
All Data
CG
0.17
0.04
0.17
0.12
-4%
-10%
-4%
-6%
-1%
650%
-1%
221%
0.09
0.19
0.09
0.10
-0.02
-0.30
-0.02
-0.13
-0.25
-0.52
-0.25
-0.36
0.01
-0.13
0.01
-0.04
0%
-36%
0%
-8%
6%
NA
6%
6%
325
Table D.6 i-Butane hydrate formation T/P at experimental P/T (52 pts)
T and P Range
T
P
240-273
0-1
273-275
1-1.7
275-276
2-150
All Aq-H-V Data
All Data
NA
NA
0.02
NA
0.02
-4.63
-0.11
-0.34
-0.11
-1.00
-0.10
0.20
-0.12
0.20
0.10
-5.97
0.45
0.20
0.45
-0.82
CG
PV
-10%
-3%
-73%
-3%
-12%
14%
59%
1%
-6%
55%
-4%
205% 1049% 511%
-6%
55%
-4%
23% 170% 56%
30%
-9%
-65%
-9%
-8%
Table D.7 Nitrogen hydrate formation T/P at experimental P/T (79 pts)
T and P Range
T
P
268-272
129-141
272-291
158-1000
291-307
1000-3500
All Aq-H-V Data<1000
All Data
NA
0.57
0.35
0.57
0.50
0.30
0.05
NA
0.05
0.11
-0.01
0.27
0.36
0.27
0.24
-0.38
-0.08
-1.21
-0.08
-0.36
CG
PV
-1%
0%
6%
0%
1%
-10%
-5%
-3%
-5%
-6%
1%
1%
NA
1%
1%
NA
0%
NA
0%
0%
0%
-3%
-3%
-3%
-2%
Table D.8 H2S hydrate formation T/P at experimental P/T (54 pts)
T and P Range
T
P
CG
250-273
0-1
283-301
3-21
301-303
21-100
303-306
100-1000
All Aq-H-V Data
All Data
1.36
0.73
-0.11
0.16
0.73
0.64
-0.23
0.47
-0.32
-0.74
0.47
0.01
-0.12
0.85
0.27
0.17
0.85
0.45
4.79
0.70
-0.20
-0.11
0.70
1.27
CG
-7%
-8%
41%
-7%
-8%
0%
13%
-4%
0%
-10%
-4%
-1%
1%
-5%
38%
60%
-5%
13%
1%
-9%
-20%
-13%
-9%
-9%
PV
-24%
-8%
74%
7%
-8%
4%
Table D.9 CO2 hydrate formation T/P at experimental P/T (195 pts)
T and P Range
T
P
150-272
0-10.5
272-283
10.5-45.1
283-283
45.1-100
283-289
100-1000
289-292
1000-8000
All Aq-H-V Data
All Data
CG
0.81
-0.16
0.11
-0.02
-4.54
-0.16
-0.66
-7%
4%
-11%
0%
12%
4%
2%
13%
-1%
-35%
-41%
-58%
-1%
-11%
NA
0.11
0.79
1.49
7.96
0.11
1.60
0.72
0.01
0.16
-0.14
NA
0.01
0.11
-0.62
-0.65
-0.45
-0.24
5.76
-0.65
0.30
-0.23
0.04
0.23
0.43
6.48
0.04
1.04
-5%
1%
-14%
5%
2%
1%
0%
5%
19%
52%
16%
-40%
19%
10%
PV
-15%
1%
-19%
-13%
-25%
1%
-4%
326
Table D.10 Xenon hydrate formation T/P at experimental P/T (139 pts)
T and P Range
T
P
211-268
0-1
273-291
1-10
292-311
10-100
311-323
100-1000
323-344
1000-3800
All Data < 1000
All Data
NA
0.40
1.17
2.48
5.35
1.24
1.97
0.18
0.72
1.23
-0.13
NA
0.78
0.75
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
CG
PV
-6%
1%
0%
-1%
1%
0%
0%
9%
-2%
-13%
-31%
-26%
-13%
-14%
NA
NA
NA
NA
NA
NA
NA
3%
-6%
-13%
5%
NA
-7%
-6%
NA
NA
NA
NA
NA
NA
NA
CG
0.02-0.60
0.80-0.95
0.95-0.99
All Data
0.11
0.29
0.11
0.20
-0.01
-0.23
-0.12
-0.13
0.24
0.89
0.22
0.53
-0.37
-0.03
-0.29
-0.19
CG
PV
-1%
-4%
-1%
-3%
-8%
-2%
6%
-3%
6%
0%
4%
3%
1%
4%
1%
2%
-3%
-11%
-2%
-7%
Table D.12 Methane+propane hydrate formation T/P at experimental P/T (54 pts)
Methane
x Range
0.20-0.40
0.70-0.95
0.95-0.99
All Data
-0.14
0.09
0.63
0.26
0.39
0.46
0.80
0.59
0.11
-0.72
-0.56
-0.42
-0.04
-0.35
-0.05
-0.13
0.44
-0.35
-0.08
-0.02
CG
PV
2%
-1%
-7%
-3%
-6%
-5%
-9%
-7%
-7%
8%
3%
1%
-2%
13%
9%
7%
1%
5%
1%
2%
Table D.13 Methane+i-butane hydrate formation T/P at experimental P/T (87 pts)
Methane
x Range
CG
0.04-0.30
0.30-0.80
0.80-1.00
All Data
-0.54
-0.39
-0.52
-0.51
-0.69
0.05
-0.44
-0.40
-0.46
-0.64
-1.28
-1.07
-0.20
0.06
-0.84
-0.62
CG
PV
13%
6%
10%
10%
8%
6%
10%
9%
4%
-1%
16%
12%
16%
-1%
8%
8%
10%
10%
21%
18%
327
CG
0.94-0.97
0.97-1.00
All Data
0.02
0.08
0.05
-0.44
-0.20
-0.30
0.23
0.09
0.15
-0.29
-0.26
-0.27
CG
PV
1%
0%
0%
-5%
-4%
-4%
5%
4%
4%
7%
3%
4%
-4%
-1%
-2%
Table D.15 Methane+benzene hydrate formation T/P at experimental P/T (15 pts)
All Data
CG
-0.02
0%
NA
-4.39
-0.31
-6.28
69%
4%
PV
70%
CG
0.06-0.30
0.30-0.60
0.65-0.87
All Data
-0.36
-0.82
-0.38
-0.45
5%
11%
6%
6%
-0.42
-1.23
-0.33
-0.54
-0.24
-0.33
0.63
0.01
-0.22
-0.82
-0.25
-0.35
0.39
-0.35
-0.30
0.04
4%
5%
-6%
1%
4%
10%
4%
5%
PV
-3%
5%
5%
1%
Table D.17 Methane+H2S hydrate formation T/P at experimental P/T (20 pts)
Methane
x Range
CG
0.70-0.90
0.90-0.95
0.95-0.99
All Data
1.30
0.78
-0.16
0.58
1.37
0.85
-0.06
0.66
1.18
0.58
-0.41
0.38
1.47
0.91
-0.21
0.66
CG
-13%
-9%
2%
-6%
-9%
-5%
4%
-3%
-14%
-9%
1%
-7%
-12%
-6%
5%
-4%
PV
-16%
-10%
3%
-7%
Table D.18 Methane+CO2 hydrate formation T/P at experimental P/T (89 pts)
Methane
x Range
CG
0.15-0.30
0.30-0.60
0.60-1.00
All Data
0.17
0.12
0.05
0.08
0.05
0.04
0.22
0.14
-0.53
-0.51
-0.21
-0.35
0.07
-0.10
-0.22
-0.15
CG
PV
-2%
-1%
0%
-1%
0%
0%
-2%
-1%
-1%
1%
2%
2%
-1%
0%
-2%
-1%
9%
9%
4%
6%
328
Table D.19 Ethane+propane hydrate formation T/P at experimental P/T (96 pts)
Ethane
x Range
CG
0.00-0.40
0.40-0.60
0.60-0.80
0.80-0.90
All Data
-0.15
-0.11
-0.19
-0.26
-0.19
1%
1%
2%
3%
2%
7%
17%
11%
-3%
8%
-0.12
-0.44
-0.13
0.35
-0.05
-0.12
-2.26
-1.24
-0.32
-0.93
-0.64
-0.47
-0.24
0.08
-0.28
-0.48
-0.49
-0.72
-0.75
-0.65
3%
31%
16%
3%
12%
10%
12%
4%
-1%
5%
PV
204%
11%
11%
8%
48%
Table D.20 Ethane+CO2 hydrate formation T/P at experimental P/T (40 pts)
Ethane
x Range
0.01-0.30
0.30-0.60
0.60-0.81
All Data
-0.16
-0.34
-0.23
-0.23
0.55
0.77
0.71
0.64
0.26
0.12
0.03
0.17
0.27
0.30
0.32
0.29
0.27
0.00
-0.20
0.08
CG
PV
2%
4%
3%
3%
-6%
-8%
-9%
-7%
-3%
0%
3%
-1%
-3%
-1%
0%
-2%
-2%
-5%
-5%
-4%
CG
0.31
0.52
All Data
0.44
0.73
0.64
NA
NA
NA
0.42
0.56
0.52
NA
NA
NA
CG
PV
-10%
-16%
-14%
-8%
-13%
-12%
NA
NA
NA
NA
NA
NA
-9%
-12%
-11%
Table D.22 Propane+i-butane hydrate formation T/P at experimental P/T (17 pts)
T Range
CG
< 273.15
> 273.15
All Data
-0.41
0.07
-0.24
2%
-7%
-1%
1%
-22%
-7%
NA
NA
NA
NA
1.10
1.10
-0.43
0.28
-0.18
0.20
0.66
0.36
NA
-26%
-26%
2%
-17%
-4%
PV
-1%
-24%
-9%
CG
0.60-0.70
0.70-0.90
0.90-1.00
All Data
0.41
2.48
2.58
2.09
0%
-15%
-16%
-12%
NA
0.30
0.11
0.25
1.95
2.29
2.77
2.36
0.14
1.83
1.73
1.46
0.19
1.49
1.55
1.24
-5%
-16%
-18%
-15%
-1%
-10%
-9%
-8%
PV
0%
-10%
-9%
-8%
329
Table D.24 Propane+nitrogen hydrate formation T/P at experimental P/T (39 pts)
Propane
x Range
0.01-0.30
0.45-0.75
All Data
-0.13
-0.37
-0.19
1.74
0.46
1.45
-0.10
0.07
-0.06
0.06
-0.26
-0.01
-0.07
-0.03
-0.06
CG
PV
2%
9%
4%
-21%
-5%
-18%
2%
1%
1%
2%
-2%
1%
0%
6%
1%
Table D.25 Propane+CO2 hydrate formation T/P at experimental P/T (92 pts)
Propane
x Range
0.01-0.30
0.30-0.70
0.70-0.90
All Data
1.20
0.57
0.04
0.97
-0.85
-0.15
-0.01
-0.60
1.10
0.56
0.09
0.87
-0.14
0.26
0.02
-0.05
0.53
0.55
0.29
0.51
CG
PV
-14%
-8%
0%
-11%
15%
4%
-1%
12%
1%
-9%
-5%
-2%
-10%
-10%
-1%
-9%
3%
-5%
0%
1%
-2.96
NA
-4.20
-8.76
CG
PV
20%
41%
40%
38%
25%
Table D.27 i-Butane+CO2 hydrate formation T/P at experimental P/T (53 pts)
i-Butane
x Range
CG
0.00-0.10
0.10-0.30
0.30-0.79
All Data
1.65
0.78
0.43
1.25
0.78
-0.51
0.08
0.37
-0.96
-0.91
-0.31
-0.84
0.05
-0.35
0.17
-0.02
CG
PV
-21%
-10%
-7%
-16%
25%
36%
19%
26%
-1%
6%
-3%
0%
-4%
9%
-2%
0%
5%
11%
4%
6%
Table D.28 n-Butane+CO2 hydrate formation T/P at experimental P/T (21 pts)
n-Butane
x Range
0.92-0.95
0.95-1.00
All Data
-0.91
-0.06
-0.26
1.75
0.37
0.67
1.84
3.24
2.91
-1.85
-0.56
-0.87
-1.16
0.15
-0.16
CG
PV
13%
1%
4%
-4%
-2%
-3%
14%
-2%
2%
-18%
-31%
-28%
24%
8%
12%
330
CG
0.00-0.01
0.17-0.50
All Data
-0.12
3.47
1.11
2%
-35%
-11%
-6%
-34%
-15%
0.46
3.56
1.60
-0.19
3.05
0.93
0.19
3.54
1.34
-0.56
3.30
0.77
2%
-30%
-9%
-3%
-36%
-14%
PV
8%
-35%
-7%
0.68-0.70
0.70-0.80
0.80-0.84
All Data
-0.28
-0.24
-1.38
-0.75
-0.88
-0.69
-1.62
-1.10
-0.41
-0.35
-1.48
-0.85
-0.76
-0.66
-1.80
-1.17
-0.62
-0.46
-1.54
-0.95
CG
PV
3%
3%
21%
11%
10%
8%
24%
15%
8%
7%
24%
14%
7%
5%
22%
13%
9%
9%
27%
17%
CG
B
C
D
-0.07
0.25
0.78
-0.16
-1.60
0.47
-0.21
-0.02
0.68
-0.50
-0.46
0.28
CG
PV
1%
-4%
-12%
4%
4%
-6%
11%
9%
-4%
3%
30%
-7%
3%
0%
-11%
Table D.32 Kobayashi et al. (1951) hydrate formation T/P at experimental P/T
(10 pts)
NG
Hugoton
Michigan
-1.56
-1.96
-1.01
-1.55
-1.11
-1.35
-0.95
-1.32
CG
PV
17%
21%
23%
30%
15%
21%
15%
24%
16%
19%
331
Table D.33 McLeod and Campbell (1961) hydrate formation T/P at experimental
P/T (7 pts)
CG
All Data
0.87
0.50
0.87
-0.40
CG
PV
-12%
-8%
6%
-6%
-12%
Table D.34 Lapin and Cinnamon (1969) hydrate formation T/P at experimental
P/T (3 pts)
NG
1
2
3
CG
0.35
0.58
1.25
-5%
-8%
-17%
-2%
-9%
-19%
0.14
0.65
1.43
0.32
0.81
1.54
0.38
0.47
1.08
0.28
0.41
1.04
-5%
-11%
-21%
-5%
-7%
-15%
PV
-4%
-6%
-15%
Table D.35 Adisasmito and Sloan (1992) hydrate formation T/P at experimental
P/T (20 pts)
NG
A
B
C
D
E
-0.80
-1.54
-2.72
-0.20
-0.25
-0.09
-0.92
-1.52
2.92
2.99
-0.56
-0.84
-1.35
-0.26
-0.84
0.00
-0.25
-0.60
0.74
-0.22
CG
PV
9%
11%
7%
-15%
4%
10%
22%
44%
2%
3%
0%
3%
9%
-9%
3%
1%
13%
22%
-28%
-29%
7%
12%
21%
4%
12%
Table D.36 Bishnoi and Dholabhai (1999) hydrate formation T/P at experimental
P/T (3 pts)
CG
All Data
-0.29
-0.25
-0.28
-0.10
CG
4%
4%
4%
PV
2%
Table D.37 Jager (2001) hydrate formation T/P at experimental P/T (17 pts)
All Data
0.49
2.66
0.40
-0.10
CG
PV
-4%
-11%
1%
-27%
-4%
332
0.8641
0.0647
0.0357
0.0099
0.0114
0.0078
0
0.0064
0
0.9320
0.0425
0.0161
0
0
0
0
0.0043
0.0051
Kobayashi et al.
Hugoton Michigan
0.8836
0.0682
0.0254
0.0038
0.0089
0.0101
0
0
0
0.7329
0.0670
0.0390
0.0036
0.0055
0.0020
0
0.1500
0
0.7964
0.0938
0.0322
0.0018
0.0058
0.0015
0.0005
0.0680
0
McLeod &
Campbell
0.906
0.066
0.018
0.005
0.005
0
0
0
0
0.7278
0.1450
0.0763
0
0.0265
0.0063
0.0011
0.0170
0
0.6769
0.1350
0.1410
0
0.0245
0.0058
0.0010
0.0158
0
0.6249
0.1247
0.2062
0
0.0226
0.0058
0.0011
0.0147
0
A
0.7662
0.1199
0.0691
0.0182
0.0266
0
0
0
0
Methane
Ethane
Propane
i-Butane
n-Butane
i-Pentane
n-Pentane
n-Hexane
n-Heptane
Nitrogen
CO2
Bishnoi &
Dholabhai
Jager
0.820
0.113
0.042
0.009
0.006
0.001
0.002
0.002
0
0
0.005
0.9753
0.0088
0.0014
0.0001
0.0002
0.0002
0.0002
0.0002
0.0001
0.0093
0.0041
0.2442
0.0399
0.0307
0.0075
0.0092
0
0
0
0.6685
0.1238
0.0196
0.0166
0.0037
0.0048
0
0
0
0.8315
E
0.0786
0.0113
0.0086
0.0020
0.0033
0
0
0
0.8962
333
BO #1a
GC #1a-a
CG
1.09
-1.78
-15%
29%
NA
NA
0.45
0.09
0.89
-1.79
0.13
-2.40
-11%
7%
-11%
30%
PV
-2%
43%
D.6 sH Hydrates
Table D.40 sH hydrate formation T/P at experimental P/T (methane+sH guest)
sH Guest
i-Pentane
2,3-Dimethyl-1-Butene
3,3-Dimethyl-1-Butene
Methylcyclopentane
Neohexane
2,3-Dimethylbutane
Cycloheptane
2,2-Dimethylpentane
Ethylcyclopentane
Methylcyclohexane
2,2,3-Trimethylbutane
3,3-Dimethylpentane
cis-1,2Dimethylcyclohexane
1,1-Dimethylcyclohexane
Ethylcyclohexane
0.02
0.00
0.00
0.00
0.00
0.00
0.00
6.41
0.00
0.00
-0.29
0.00
0.03
-0.04
0.01
-0.05
0.03
0.00
0.00
-1.72
0.00
0.16
0.07
0.04
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
0.09
-0.01
0.01
0.06
-0.42
0.00
0.01
-1.24
-0.01
0.20
0.01
-0.02
-0.04
NA
NA
-0.36
-0.06
-0.03
0.66
NA
-0.04
0.09
0.31
0.00
-1%
0%
0%
0%
0%
0%
0%
-51%
0%
0%
4%
0%
-2%
1%
0%
1%
-1%
0%
0%
33%
0%
-2%
-1%
-1%
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
11
4
4
21
11
6
7
16
6
29
6
7
0.00
0.00
NA
0.05
-0.01
0%
0%
NA
10
0.00
0.00
0.01
-0.96
NA
NA
-0.03
0.06
-0.03
NA
0%
0%
0%
11%
NA
NA
11
2
# pts
Data
Points
190-273
0-24
273-286
26-100
286-306
100-1000
306-321
1000-5000
All Data
14
63
47
57
181
0 (14)
54 (63)
47 (47)
32 (57)
133 (181)
14 (14)
63 (63)
42 (45)
0 (1)
119 (123)
14 (14)
63 (63)
47 (47)
57 (57)
181 (181)
9 (14)
63 (63)
47 (46)
57 (0)
176 (123)
334
Data
Points
269-273
0-5.3
273-292
5.3-100
292-304
100-1000
304-328
1000-5000
All Data
6
38
13
21
78
0 (6)
30 (38)
13 (13)
17 (21)
60 (78)
6 (6)
37 (38)
9 (13)
0 (0)
52 (57)
6 (6)
38 (38)
13 (13)
21 (21)
78 (78)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
Data
Points
200-273
0-5
273-288
5-33
288-289
33-100
289-299
100-1000
299-324
1000-5000
All Data
12
61
15
5
19
112
0 (12)
58 (61)
15 (15)
5 (5)
15 (19)
93 (112)
12 (12)
61 (61)
15 (15)
3 (3)
0 (0)
91 (91)
12 (12)
61 (61)
15 (15)
5 (5)
19 (19)
112 (112)
10 (12)
61 (61)
15 (15)
5 (5)
19 (0)
110 (93)
Data
Points
13
6
19
0 (11)
0 (2)
0 (13)
13 (11)
6 (2)
19 (13)
13 (13)
6 (3)
19 (16)
0 (0)
0 (0)
0 (0)
Data
Points
273-278
1.5-5.4
278-278
5.4-170
All Data
56
36
92
37 (56)
1 (29)
38 (85)
56 (56)
36 (18)
92 (74)
56 (56)
36 (0)
92 (56)
56 (56)
36 (11)
92 (92)
Data
Points
240-273
0-1
273-275
1-1.7
275-276
2-150
All Data
10
36
6
52
0 (7)
0 (36)
4 (6)
4 (49)
10 (10)
36 (36)
6 (6)
52 (52)
10 (10)
36 (36)
6 (6)
52 (52)
10 (10)
36 (36)
6 (6)
52 (52)
335
Data
Points
268-272
129-141
272-291
158-1000
291-307
1000-3500
All Data
12
51
16
79
0 (12)
39 (51)
16 (16)
55 (79)
12 (0)
43 (47)
0 (0)
55 (47)
12 (12)
51 (51)
14 (14)
77 (77)
12 (12)
51 (51)
16 (0)
79 (63)
Data
Points
250-273
0-1
283-301
3-21
301-303
21-100
303-306
100-1000
All Data
11
26
8
9
54
0 (11)
26 (26)
8 (8)
9 (9)
43 (54)
11 (11)
26 (26)
8 (8)
9 (9)
54 (54)
11 (11)
26 (26)
8 (8)
9 (9)
54 (54)
11 (11)
26 (26)
8 (8)
9 (9)
54 (54)
Data
Points
150-272
0-10.5
272-283
10.5-45.1
283-283
45.1-100
283-289
100-1000
289-292
1000-8000
All Data
24
129
5
9
28
195
0 (18)
95 (129)
5 (5)
9 (9)
23 (28)
132 (189)
24 (24)
128 (129)
5 (5)
7 (8)
0 (1)
164 (167)
24 (24)
129 (129)
5 (5)
9 (9)
28 (28)
195 (195)
11 (23)
129 (128)
5 (5)
9 (9)
28 (6)
182 (171)
Data
Points
211-268
0-1
273-291
1-10
292-311
10-100
311-323
100-1000
323-344
1000-3800
All Data
5
42
44
26
22
139
0 (4)
33 (42)
44 (44)
26 (26)
22 (22)
125 (138)
5 (5)
42 (42)
44 (44)
19 (22)
0 (0)
110 (113)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
336
Data
Points
0.02-0.60
0.80-0.95
0.95-0.99
All Data
18
25
11
54
18 (18)
22 (25)
8 (11)
48 (54)
18 (18)
23 (25)
11 (11)
52 (54)
18 (18)
25 (25)
11 (11)
54 (54)
18 (18)
25 (25)
11 (11)
54 (54)
Data
Points
0.20-0.40
0.70-0.95
0.95-0.99
All Data
15
16
23
54
13 (15)
14 (16)
22 (23)
49 (54)
15 (15)
16 (16)
23 (23)
54 (54)
15 (15)
16 (16)
23 (23)
54 (54)
15 (15)
16 (16)
23 (23)
54 (54)
Data
Points
0.04-0.30
0.30-0.80
0.80-1.00
All Data
12
13
62
87
0 (10)
6 (13)
54 (62)
60 (85)
12 (12)
13 (13)
62 (62)
87 (87)
12 (12)
13 (13)
62 (62)
87 (87)
12 (12)
13 (13)
62 (62)
87 (87)
Data
Points
0.94-0.97
0.97-1.00
All Data
18
24
42
4 (18)
19 (24)
23 (42)
18 (18)
24 (24)
42 (42)
18 (18)
24 (24)
42 (42)
18 (18)
24 (24)
42 (42)
15
0 (0)
12 (15)
15 (15)
15 (15)
337
Data
Points
0.06-0.30
0.30-0.60
0.65-0.87
All Data
32
12
19
63
23 (32)
9 (12)
16 (19)
48 (63)
32 (32)
12 (12)
19 (19)
63 (63)
32 (32)
12 (12)
19 (19)
63 (63)
32 (32)
12 (12)
19 (19)
63 (63)
Data
Points
0.70-0.90
0.90-0.95
0.95-0.99
All Data
5
8
7
20
5 (5)
8 (8)
7 (7)
20 (20)
5 (5)
8 (8)
7 (7)
20 (20)
5 (5)
8 (8)
7 (7)
20 (20)
5 (5)
8 (8)
7 (7)
20 (20)
Data
Points
0.15-0.30
0.30-0.60
0.60-1.00
All Data
9
31
49
89
7 (9)
27 (30)
43 (49)
77 (88)
9 (9)
31 (31)
49 (49)
89 (89)
9 (9)
31 (31)
49 (49)
89 (89)
9 (9)
31 (31)
49 (49)
89 (89)
Data
Points
0.00-0.40
0.40-0.60
0.60-0.80
0.80-0.90
All Data
19
14
41
22
96
11 (19)
11 (14)
18 (40)
16 (19)
56 (92)
6 (6)
4 (4)
41 (41)
22 (22)
73 (73)
19 (19)
14 (14)
41 (41)
22 (22)
96 (96)
19 (19)
14 (14)
41 (41)
22 (22)
96 (96)
Data
Points
0.01-0.30
0.30-0.60
0.60-0.81
All Data
19
12
9
40
15 (19)
7 (12)
6 (8)
28 (39)
19 (19)
12 (12)
8 (8)
39 (39)
19 (19)
12 (12)
9 (9)
40 (40)
19 (19)
12 (12)
9 (9)
40 (40)
338
Data
Points
0.31
0.52
All Data
8
21
29
3 (8)
10 (21)
13 (29)
0 (0)
0 (0)
0 (0)
8 (8)
21 (17)
29 (25)
0 (0)
0 (0)
0 (0)
Data
Points
< 273.15
> 273.15
All Data
11
6
17
0 (11)
0 (6)
0 (17)
0 (0)
4 (4)
4 (4)
11 (11)
6 (5)
17 (16)
11 (11)
6 (6)
17 (16)
Data
Points
0.60-0.70
0.70-0.90
0.90-1.00
All Data
11
28
15
54
0 (11)
6 (28)
2 (15)
8 (54)
7 (7)
28 (28)
12 (12)
47 (47)
11 (11)
28 (28)
15 (14)
54 (53)
11 (11)
28 (28)
15 (15)
54 (54)
Data
Points
0.01-0.30
0.45-0.75
All Data
30
9
39
24 (30)
7 (9)
31 (39)
30 (30)
9 (9)
39 (39)
30 (30)
9 (9)
39 (39)
30 (30)
9 (9)
39 (39)
Data
Points
0.01-0.30
0.30-0.70
0.70-0.90
All Data
66
18
8
92
40 (65)
13 (17)
7 (8)
60 (90)
52 (53)
18 (18)
8 (8)
78 (79)
65 (66)
18 (18)
8 (8)
91 (92)
66 (66)
18 (18)
8 (8)
92 (92)
339
0 (3)
0 (5)
5 (5)
5 (5)
Data
Points
0.00-0.10
0.10-0.30
0.30-0.79
All Data
32
12
9
53
15 (31)
6 (11)
5 (8)
26 (50)
32 (32)
12 (12)
9 (9)
53 (53)
32 (32)
12 (12)
9 (9)
53 (53)
32 (32)
12 (12)
9 (9)
53 (53)
Data
Points
0.92-0.95
0.95-1.00
All Data
16
5
21
11 (16)
3 (5)
14 (21)
16 (16)
5 (5)
21 (21)
16 (16)
5 (5)
21 (21)
16 (16)
5 (5)
21 (21)
Data
Points
0.00-0.01
0.17-0.50
All Data
21
11
32
19 (21)
11 (11)
30 (32)
21 (21)
11 (11)
32 (32)
21 (21)
11 (11)
32 (32)
21 (21)
11 (11)
32 (32)
Data
Points
0.68-0.70
0.70-0.80
0.80-0.84
All Data
6
26
25
57
6 (6)
26 (26)
23 (25)
55 (57)
6 (6)
26 (26)
25 (25)
57 (57)
6 (6)
26 (26)
25 (25)
57 (57)
6 (6)
26 (26)
25 (25)
57 (57)
340
Table D.71 Number of data points predicted for natural gas hydrates
Data
Points
Researcher NG Type
Wilcox et
B
al.,
C
1941
D
Kobayashi
Hugoton
et al., 1951
Michigan
McLeod and Campbell,
1961
Lapin and
1
Cinnamon,
2
1969
3
Adisasmito
A
and
B
Sloan,
C
D
1992
E
Bishnoi and Dholabhai,
1999
Jager, 2001
9
15
12
5
5
9 (9)
15 (15)
12 (12)
5 (5)
5 (5)
6 (9)
15 (15)
12 (12)
5 (5)
5 (5)
9 (9)
15 (15)
12 (12)
5 (5)
5 (5)
9 (9)
15 (15)
12 (12)
5 (5)
5 (5)
9 (9)
15 (15)
12 (12)
5 (5)
5 (5)
7 (7)
7 (7)
7 (7)
7 (7)
7 (7)
1
1
1
4
4
4
4
4
1 (1)
1 (1)
1 (1)
4 (4)
4 (4)
4 (4)
4 (4)
4 (4)
1 (1)
1 (1)
1 (1)
3 (4)
2 (4)
2 (4)
3 (4)
3 (4)
1 (1)
1 (1)
1 (1)
4 (4)
4 (4)
4 (4)
4 (4)
4 (4)
1 (1)
1 (1)
1 (1)
4 (4)
4 (4)
4 (4)
4 (4)
4 (4)
1 (1)
1 (1)
1 (1)
4 (4)
4 (4)
4 (4)
4 (4)
4 (4)
3 (3)
2 (3)
3 (3)
3 (3)
3 (3)
17
17 (17)
17 (17)
17 (17)
17 (17)
17 (17)
Table D.72 Number of data points predicted for black oil and gas condensate
hydrates
Data
Points
BO #1a
GC #1a-a
5
5
0 (0)
0 (0)
4 (5)
5 (5)
5 (5)
5 (5)
5 (5)
5 (5)
341
Data
Points
11
4
4
21
11
6
7
16
6
29
6
7
11 (11)
4 (4)
4 (4)
21 (21)
11 (11)
6 (6)
7 (7)
16 (16)
6 (6)
29 (29)
6 (6)
7 (7)
10 (11)
4 (4)
4 (4)
21 (21)
11 (11)
6 (6)
7 (7)
15 (16)
6 (6)
28 (29)
6 (6)
6 (7)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
11 (11)
4 (4)
4 (4)
21 (21)
11 (11)
6 (6)
7 (7)
16 (16)
6 (6)
29 (29)
6 (6)
7 (7)
11 (11)
0 (0)
0 (0)
21 (21)
11 (11)
6 (6)
7 (7)
0 (0)
6 (6)
29 (29)
6 (6)
7 (7)
10
10 (10)
10 (10)
0 (0)
10 (10)
10 (10)
11
2
11 (11)
2 (2)
11 (11)
2 (2)
0 (0)
0 (0)
11 (11)
2 (2)
11 (11)
0 (0)
343
What follows is a complete listing of fugacity model parameters. Note that all
parameters with a subscript, 0, are at reference conditions (T0 = 298.15 K and P0 = 1 bar).
gi0o
(J/mol)
hi0o
(J/mol)
Component
Methane
Ethylene
Ethane
Propylene
Propane
n-Butane
i-Butane
n-Pentane
i-Pentane
Benzene
2,3-Dimethyl-1-Butene
3,3-Dimethyl-1-Butene
Methylcyclopentane
n-Hexane
Neohexane
2,3-Dimethylbutane
Toluene
Cycloheptane
Ethylcyclopentane
-50830
68170
-32900
62760
-23500
-17200
-20900
-8370
-15229
129700
79034
98155
35773
-290
-9623
-4100
122100
63010
44559
-74900
52320
-84720
20400
-103900
-126200
-134600
-146500
-165976
82980
-55730
-43136
-106690
-167300
-185557
-177775
50030
-119326
-127066
Methylcyclohexane
n-Heptane
2,2,3-Trimethylbutane
2,2-Dimethylpentane
3,3-Dimethylpentane
cis-1,2-Dimethylcyclohexane
1,1-Dimethylcyclohexane
Ethylcyclohexane
n-Octane
n-Nonane
n-Decane
Nitrogen
Hydrogen Sulfide
Carbon Dioxide
Xenon
Water
Methanol
Ethanol
Ethylene Glycol
gi0o
hi0o
(J/mol) (J/mol)
27279
8120
4268
84
2636
41212
35229
39245
16500
24811
33220
0
-33100
-394600
0
-228700
-162600
-168400
-304464
-154763
-187900
-204803
-206142
-201540
-172169
-180997
-171750
-208600
-229028
-249655
0
-20200
-393800
0
-242000
-201300
-235000
-389314
344
a0 /R
a1 /R (K-1)
(102)
a2 /R (K-2)
(105)
a3 /R (K-3)
(109)
2.3902
0.4750
0.8293
0.3789
-0.4861
0.4755
-0.9511
0.8142
-1.9942
-4.3528
0.8444
-1.5091
-6.0224
0.8338
-1.9992
-1.7557
-4.1328
-9.1569
-6.6479
-7.4419
-0.6184
-2.7576
-6.0214
-0.8469
-8.2174
-8.6663
-7.6790
-0.7327
0.3779
-0.9511
3.4736
3.5577
2.6751
2.5000
3.8747
2.2896
2.3902
4.2904
0.6039
1.8795
2.0752
2.8638
3.6629
4.4650
4.9999
5.4598
6.6725
5.8261
6.7128
6.5920
7.6689
6.6373
7.5632
7.3921
6.7214
9.4553
9.0276
9.4251
8.1268
9.0376
10.7640
8.2174
10.7840
10.8140
10.6880
9.2691
8.1419
11.5490
-0.0189
0.1574
0.7188
0
0.0231
1.1000
2.5191
2.9845
0.1525
-1.0029
-0.7699
-1.4643
-1.8895
-2.2041
-2.7651
-2.6997
-3.9738
-3.7942
-4.4423
-3.5038
-4.3774
-3.444
-4.1837
-4.0574
-4.1414
-5.0522
-5.2837
-5.334
-4.388
-5.3139
-7.6437
-4.4881
-6.1744
-6.0335
-6.1391
-5.0421
-2.3178
-6.3555
0.0971
0.0686
-0.4208
0
0.1269
-0.1464
-1.2475
-1.7995
-1.3234
2.1235
0.8756
2.9589
3.8143
4.2068
5.9982
5.0824
9.1735
9.3295
12.7765
6.2599
9.6314
6.9342
8.2325
8.1973
9.6616
9.0879
12.0669
11.2568
9.2037
12.0770
20.8630
9.4150
13.2042
12.3840
13.2545
10.6429
-3.5833
13.5917
-0.3453
-0.3959
0.8977
0
-0.4321
-0.9662
2.4104
3.6181
345
Table E.3 Partial molar formation properties and Born constants of solutes
(Equations 3.5, 3.6, and 3.7)
Component
Methane
Ethylene
Ethane
Propylene
Propane
n-Butane
i-Butane
n-Pentane
i-Pentane
Benzene
2,3-Dimethyl-1-Butene
3,3-Dimethyl-1-Butene
Methylcyclopentane
n-Hexane
Neohexane
2,3-Dimethylbutane
Toluene
Cycloheptane
Ethylcyclopentane
Methylcyclohexane
n-Heptane
2,2,3-Trimethylbutane
2,2-Dimethylpentane
3,3-Dimethylpentane
cis-1,2-Dimethylcyclohexane
1,1-Dimethylcyclohexane
Ethylcyclohexane
n-Octane
n-Nonane
n-Decane
Nitrogen
Hydrogen Sulfide
Carbon Dioxide
Xenon
Methanol
Ethanol
Ethylene Glycol
Na+
K+
Ca2+
Cl-
gi0 *
hi0 *
(J/mol) (J/mol)
(J/mol)
-34451
81379
-17000
74935
-7550
-940
-2900
9160
1541
133888
97020
116140
53760
18200
8367
13890
126608
81190
62740
27280
27500
22450
18260
20820
59910
53930
57950
36700
45200
54140
18188
-27920
-385974
13493
-175937
-181293
-325000
-261881
-282462
-552790
-131039
-133009
-167360
-169870
-232547
-211418
-253592
-253592
-300955
-301000
-82676
-335000
-335000
-335000
-335180
-335000
-335000
-135896
-380000
-380000
-380000
-380158
-380000
-380000
-380000
-404000
-404000
-404000
-404258
-453600
-494000
-145101
-41840
-8368
-92634
-61756
-85228
406762
138323
80626
517142
609190
-87906
35857
-103136
-1213
-131000
-152000
-156000
-173887
-193476
51170
-86430
-73840
-137390
-199200
-216257
-208480
13724
-153430
-161170
-154760
-225000
-238900
-240240
-235640
-212570
-221400
-212150
-250500
-273150
-297330
-10439
-37660
-413798
-18870
-246312
-287232
-320000
-240300
-252170
-543083
-167080
346
Table E.4 Partial molar heat capacity (at 1 bar) and volume terms of solutes
(Equations 3.6 and 3.7)
Component
Methane
Ethylene
Ethane
Propylene
Propane
n-Butane
i-Butane
n-Pentane
i-Pentane
Benzene
2,3-Dimethyl-1-Butene
3,3-Dimethyl-1-Butene
Methylcyclopentane
n-Hexane
Neohexane
2,3-Dimethylbutane
Toluene
Cycloheptane
Ethylcyclopentane
Methylcyclohexane
n-Heptane
2,2,3-Trimethylbutane
2,2-Dimethylpentane
3,3-Dimethylpentane
cis-1,2-Dimethylcyclohexane
1,1-Dimethylcyclohexane
Ethylcyclohexane
n-Octane
n-Nonane
n-Decane
Nitrogen
Hydrogen Sulfide
Carbon Dioxide
Xenon
Methanol
Ethanol
Ethylene Glycol
Na+
K+
Ca2+
Cl-
c1
(J/mol-K)
176.12
163.59
226.67
209.62
277.52
330.77
330.77
373.24
373.00
338.33
425.00
425.00
425.00
424.53
425.00
425.00
392.98
472.00
472.00
472.00
472.37
472.00
472.00
472.00
522.00
522.00
522.00
522.13
571.90
621.10
149.75
135.14
167.50
176.16
165.21
251.11
-2.55
76.065
30.962
37.656
-18.410
c2
v1
(J-K/mol) (J/mol-bar)
6310762
2.829
5846257
3.287
9011737
3.612
8447000
4.170
11749531
4.503
14610096
5.500
14610096
5.500
16955051
6.282
16997948
6.300
1072758
5.491
19734906
7.200
19734906
7.200
19734906
7.200
19680558
7.175
19734908
7.200
19734908
7.200
1745012
6.287
22327815
8.100
22327815
8.100
22327815
8.100
22283347
8.064
22327914
8.100
22327914
8.100
22327914
8.100
24920724
9.000
24920724
9.000
24920724
9.000
24886075
8.961
27607264
9.830
30256344
10.710
5046230
2.596
2850801
2.724
5304066
2.614
6126637
3.146
-903211
2.903
90828
3.863
5711758
4.000
-593488
0.7694
-1079442
1.4891
-1520020
-0.0815
-1176604
1.6870
v2
(J/mol)
3651.8
5288.2
5565.2
6925.3
7738.2
11014.4
11014.4
12082.2
12100.0
7579.0
14300.0
14300.0
14300.0
14264.4
14300.0
14300.0
9169.4
16400.0
16400.0
16400.0
16435.2
16400.0
16400.0
16400.0
18600.0
18600.0
18600.0
18624.1
20730.0
22890.0
3083.0
2833.6
3125.9
4426.4
2307.3
4166.5
0
-956.04
-616.30
-3034.24
2008.74
v3
v4
(J-K/mol-bar) (J-K/mol)
9.7119
-131365
-7.8396
-138131
2.1778
-139277
-3.1648
-144900
-6.3316
-148260
-14.9298
-157256
-14.9298
-157256
-23.4091
-166218
-23.4000
-166000
49.4653
-147599
-32.0000
-175000
-32.0000
-175000
-32.0000
-175000
-32.0202
-175238
-32.0000
-175000
-32.0000
-175000
50.7724
-154176
-40.5000
-184000
-40.5000
-184000
-40.5000
-184000
-40.5342
-184213
-40.5000
-184000
-40.5000
-184000
-40.5000
-184000
-49.1000
-193000
-49.1000
-193000
-49.1000
-193000
-49.1356
-193263
-57.4000
-201950
-65.9000
-210850
11.9407
-129018
24.9559
-127989
11.7721
-129198
6.6605
-134570
47.7051
-125809
50.8126
-133495
0
0
13.6231
-114056
22.7400
-113470
22.1610
-103721
23.2756
-119118
347
1 (K-1)
2 (K-2)
243.9576
0.039037
-1.01261E-5
-0.7520846
-2.12309E-4
6.04961E-8
6.60648E-4
3.18021E-7
-9.33341E-11
i=
a1i
a2i (bar-1)
a3i (bar-2)
Table E.6 Formation properties and heat capacity terms of pure water (Equations
3.3, 3.4, and 3.10)
g w0 Lpure (J/mol) hw0 Lpure (J/mol)
-237129
-285830
a0 /R
8.712
a1 /R (K-1)
(102)
0.125
a2 /R (K-2)
(105)
-0.018
a3 /R (K-3)
(109)
0
1 (K-1)
2 (K-2)
3 (K-3)
3.10034E-4
31.1251
-1.14154E-1
-2.48318E-7
a1i
-1
a2i (bar ) -2.46176E-2 2.15663E-4 -6.48160E-7 6.47521E-10
a3i (bar-2) 8.69425E-6 -7.96939E-8 2.45391E-10 -2.51773E-13
a4i (bar-3) -6.03348E-10 5.57791E-12 -1.72577E-14 1.77978E-17
Dij = d1+d2T+d3T2
D1
d2
d3
-1.12161E-3
2.49474E-5
-4.603E-8
7.12650E-5
-6.04659E-7
1.327E-9
1.08383E-4
-1.03524E-6
2.3878E-9
348
Table E.9 Aqueous species interaction parameters (Equations 3.19 and 3.20)
Species Pair
H2S-H2S
CO2-CO2
CH3OH-CH3OH
CH3OH-CH4
CH3OH-C3H8
Na+-CH4
Cl--CH4
Na+-C3H8
Cl--C3H8
Interaction Parameter
0 = a + bT + cP
a
b
c
0.1
-4.7E-4
0
0.107
-4.5E-4
0
-0.02214 7.4E-5
-2.6E-6
0.1
-4.48E-4
0
0.247
-1.0E-3
0
0.025
0
0
0.025
0
0
-0.09809 4.19E-4
-6.2E-6
-0.09809 4.19E-4
-6.2E-6
1
0
0
0
0
0
-5.0E-3
-5.0E-3
0
0
349
Tc (K)
Pc (bar)
Vc
(L/mol)
MW
S2
190.56
282.34
305.32
365.57
369.83
425.12
408.14
469.70
460.43
562.16
500.00
480.00
532.79
507.60
488.78
499.98
591.80
604.30
569.52
572.19
540.20
531.17
520.50
536.40
606.15
591.15
609.15
568.70
594.60
617.70
126.20
373.53
304.21
289.74
647.30
512.58
516.25
645.00
45.991
50.401
48.721
46.650
42.481
37.960
36.481
33.701
33.810
48.981
32.200
32.900
37.851
30.251
30.810
31.270
41.060
38.403
33.981
34.710
27.400
29.536
27.733
29.456
29.385
29.385
30.400
24.900
22.900
21.100
34.001
89.630
73.831
58.405
220.483
80.959
63.835
75.300
0.09865
0.13100
0.14548
0.18836
0.20014
0.25509
0.26271
0.31305
0.30584
0.25894
0.34887
0.33311
0.31892
0.37122
0.35884
0.35777
0.31580
0.35920
0.37514
0.36779
0.42788
0.39786
0.41600
0.41412
0.46026
0.44975
0.43013
0.48635
0.54367
0.59958
0.08919
0.09849
0.09396
0.11803
0.05595
0.11781
0.16681
0.0000
16.043
28.054
30.070
42.081
44.097
58.124
58.124
72.151
72.151
78.114
84.162
84.162
84.162
86.178
86.178
86.178
92.141
98.189
98.189
98.189
100.205
100.205
100.205
100.205
112.216
112.216
112.216
114.232
128.259
142.286
28.013
34.076
44.010
131.300
18.015
32.040
46.070
62.068
0.0115
0.0865
0.0995
0.1398
0.1523
0.2002
0.1808
0.2515
0.2275
0.2100
0.2269
0.2257
0.2302
0.3013
0.2350
0.2461
0.2621
0.2430
0.2716
0.2350
0.3495
0.2504
0.2879
0.2672
0.2324
0.2326
0.2455
0.3996
0.4435
0.4923
0.0377
0.0942
0.2236
0.0000
0.3440
0.5656
0.6371
0.2300
-0.012223
-0.002805
-0.012416
-0.006163
-0.003791
0.003010
0.006209
-0.000636
-0.005345
-0.000318
0.004265
-0.008468
-0.001461
-0.007459
-0.006195
-0.004579
-0.005125
-0.004421
-0.002891
-0.000172
-0.003031
0
-0.003786
-0.004779
0.014432
0.007389
0
-0.000821
0.005435
0.003324
-0.011016
0.010699
-0.004474
0
-0.201789
-0.430885
-0.216396
0
350
The constants, S1 are calculated using Equation 3.31 for all components except for the
following, in which a constant is used:
Water
Methanol
Ethanol
1.2440
1.8283
1.6787
Methane
0.0371
0.0258
0.0612
0.0148
0.0544
0.0417
0.0291
0.0912
0.0936
0.4965
N2
0.0291
0.0441
0.0082
0.0852
0.0862
0.0596
0.0845
0.0917
0.1070
0.1697
0.1552
0.2193
0.1243
0.1475
-0.0462
0.5063
H2S
0.0912
0.0846
0.0874
0.0564
0.0549
0.0655
0.0142
0.0191
0.0543
0.0033
0.1475
0.1093
0.1450
CO2
0.0936
0.0568
0.1320
0.0686
0.1300
0.1336
0.1289
0.1454
0.1371
0.0734
0.1167
0.0935
0.1209
0.1440
-0.0462
0.1093
-0.0700
Water
0.4965
0.3217
0.5975
0.3815
0.5612
0.5569
0.5382
0.5260
0.5084
0.3238
0.5238
0.5257
0.5135
0.4969
0.5272
0.4816
0.3024
0.4779
0.4823
0.4822
0.4880
0.4886
0.4903
0.4881
0.5108
0.5120
0.5078
0.4871
0.4832
0.4820
0.5063
0.1450
-0.0700
-
MeOH
0.2015
0.2411
0.1746
0.1147
0.0980
0.0932
0.0777
0.1330
0.0716
0.0786
-
EtOH
0.0709
0.1108
0.0852
0.1145
0.0774
0.0759
0.0960
0.0863
-
351
For binary pairs in which an interaction parameter is not given, the following correlations
are used (API, 1999):
CH4
kiCH 4 = 0.014 i CH 4
E.1
N2
kiN2 = 0.0403 i N2
E.2
H2S
kiH 2 S = 0.0316 i H 2 S
E.3
CO2
kij = 0.10
E.4
Where i is the solubility parameter in calories per cubic centimeter. Note that Equation
E.1 should not be used for solubility parameter differences greater than 3.5 cal/cm3.
Table E.12 gives the solubility parameters for each component.
Table E.12 Solubility parameters (cal/cm3) for Equations E.1 to E.4
Component
Methane
Ethylene
Ethane
Propylene
Propane
n-Butane
i-Butane
n-Pentane
i-Pentane
Benzene
2,3-Dimethyl-1-Butene
3,3-Dimethyl-1-Butene
Methylcyclopentane
n-Hexane
Neohexane
2,3-Dimethylbutane
Toluene
Cycloheptane
Ethylcyclopentane
i
(cal/cm3)
5.680
6.080
6.050
6.313
6.400
6.768
6.376
7.059
6.750
9.144
7.172
6.664
7.907
7.328
6.730
7.017
8.969
8.416
7.988
Component
Methylcyclohexane
n-Heptane
2,2,3-Trimethylbutane
2,2-Dimethylpentane
3,3-Dimethylpentane
cis-1,2-Dimethylcyclohexane
1,1-Dimethylcyclohexane
Ethylcyclohexane
n-Octane
n-Nonane
n-Decane
Nitrogen
Hydrogen Sulfide
Carbon Dioxide
Xenon
Water
Methanol
Ethanol
Ethylene Glycol
i
(cal/cm3)
7.954
7.433
6.965
6.945
7.105
7.944
7.662
7.998
7.506
7.562
7.556
4.440
8.800
7.120
7.780
22.380
0
0
0
352
gi0o (J/mol)
hi0o (J/mol)
-236539.24
-235537.85
-235627.53
-235491.02
-384138.00
-409140.00
-748100.00
-292714.43
-291758.77
-292044.10
-291979.26
-411153.00
-436747.00
-795800.00
Table E.14 Heat capacity parameters of pure phases at 1 bar (Equation 3.4)
Phase
Ice
sI-
sII-
sH-
NaCl
KCl
CaCl2
a0 /R
0.735409713
0.735409713
0.735409713
0.735409713
5.526
6.17
8.646
a1 /R (K-1)
(102)
1.4180551
1.4180551
1.4180551
1.4180551
0.1963
0
0.153
a2 /R (K-2)
(105)
-1.72746
-1.72746
-1.72746
-1.72746
0
0
0
a3 /R (K-3)
(109)
63.5104
63.5104
63.5104
63.5104
0
0
0
v0 (cm3/mole)
19.7254
22.7712
22.9456
24.2126
26.9880
37.5760
51.5270
1 (K-1)
1.522300E-4
3.384960E-4
2.029776E-4
3.575490E-4
2.000000E-5
2.000000E-5
2.000000E-5
2 (K-2)
1.660000E-8
5.400990E-7
1.851168E-7
6.294390E-7
0
0
0
3 (K-3)
0
-4.769460E-11
-1.879455E-10
0
0
0
0
(bar-1)
1.3357E-5
3.0000E-5
3.0000E-6
3.0000E-7
2.0000E-6
2.0000E-6
2.0000E-6
353
g w0 (J/mol)
hw0 (J/mol)
-235537.85
-235627.53
-235491.02
-291758.77
-292044.10
-291979.26
a (J/cm3)
25.74
260.00
0
b (J/cm3)
-481.32
-68.64
0
1 (K-1)
3.384960E-4
2.029776E-4
3.575490E-4
2 (K-2)
5.400990E-7
1.851168E-7
6.294390E-7
3 (K-3)
-4.769460E-11
-1.879455E-10
0
sI
1.0E-5
2.2E-6
1.0E-8
1.0E-7
1.0E-7
NA
NA
NA
NA
1.1E-5
5.0E-6
1.0E-6
9.0E-6
sII
5.0E-5
2.2E-5
1.0E-7
1.0E-6
1.0E-6
1.0E-8
1.0E-8
1.0E-8
1.0E-8
1.1E-5
1.0E-5
1.0E-5
1.0E-5
354
Table E.19 Repulsive constants and guest diameters for Equation 4.42
Component
Methane
Ethylene
Ethane
Propylene
Propane
n-Butane
i-Butane
i-Pentane
Benzene
Nitrogen
H 2S
CO2
Xenon
Diameter ()
sI small
sI large
sII small
sII large
4.247
4.816
5.076
5.522
5.745
6.336
6.306
6.777
6.272
4.177
4.308
4.603
4.404
1.7668E-2
0
0
0
0
0
0
0
0
1.7377E-2
1.7921E-2
0
1.8321E-2
0
1.0316E-2
1.5773E-2
2.5154E-2
2.9839E-2
0
0
0
0
0
0
5.8282E-3
0
2.0998E-3
2.3814E-3
2.5097E-3
0
0
0
0
0
0
2.0652E-3
2.1299E-3
2.2758E-3
2.1774E-3
1.1383E-2
1.3528E-2
1.4973E-2
2.1346E-2
2.5576E-2
3.6593E-2
3.6000E-2
4.7632E-2
3.5229E-2
1.1295E-2
1.1350E-2
1.2242E-2
1.1524E-2
Table E.20 Kihara potential parameters (Equation 3.43) (a is from Sloan 1998)
Component
Methane
Ethylene
Ethane
Propylene
Propane
n-Butane
i-Butane
i-Pentane
Benzene
2,3-Dimethyl-1-Butene
3,3-Dimethyl-1-Butene
Methylcyclopentane
Neohexane
2,3-Dimethylbutane
Cycloheptane
Ethylcyclopentane
Methylcyclohexane
2,2,3-Trimethylbutane
2,2-Dimethylpentane
3,3-Dimethylpentane
cis-1,2-Dimethylcyclohexane
1,1-Dimethylcyclohexane
Ethylcyclohexane
Nitrogen
Hydrogen Sulfide
Carbon Dioxide
Xenon
a ()
()
/k (K)
0.3834
0.4700
0.5651
0.6500
0.6502
0.9379
0.8706
0.9868
1.2000
1.0175
0.7773
1.0054
1.0481
1.0790
1.0576
1.1401
1.0693
1.1288
1.2134
1.2219
1.1494
1.1440
1.1606
0.3526
0.3600
0.6805
0.2357
3.14393
3.24461
3.24693
3.33039
3.41670
3.51726
3.41691
3.54550
3.25176
3.55376
3.56184
3.56878
3.54932
3.57910
3.59028
3.60425
3.58776
3.59955
3.59989
3.59117
3.60555
3.60212
3.60932
3.13512
3.10000
2.97638
3.32968
155.593
180.664
188.181
186.082
192.855
197.254
198.333
199.560
223.802
211.924
253.681
229.928
229.832
210.664
250.187
219.083
237.989
232.444
224.609
204.968
233.510
246.996
220.527
127.426
212.047
175.405
193.708