View metadata, citation and similar papers at core.ac.

uk brought to you by CORE

provided by Texas A&M University

PREDICTION OF GAS-HYDRATE FORMATION CONDITIONS IN

PRODUCTION AND SURFACE FACILITIES

A Thesis

by

SHARAREH AMERIPOUR

Submitted to the Office of Graduate Studies of

Texas A&M University
in partial fulfillment of the requirements for the degree of

MASTER OF SCIENCE

August 2005

Major Subject: Petroleum Engineering

PREDICTION OF GAS-HYDRATE FORMATION CONDITIONS IN

PRODUCTION AND SURFACE FACILITIES

A Thesis

by

SHARAREH AMERIPOUR

Submitted to the Office of Graduate Studies of

Texas A&M University
in partial fulfillment of the requirements for the degree of

MASTER OF SCIENCE

Approved by:

Chair of Committee, Maria A. Barrufet

Committee Members, W. John Lee
Mahmood Amani
Malcolm Andrews
Head of Department, Stephen A. Holditch

August 2005

Major Subject: Petroleum Engineering

 iii

ABSTRACT

Prediction of Gas-Hydrate Formation Conditions in

Production and Surface Facilities. (August 2005)
Sharareh Ameripour,
B.S., Amirkabir University of Technology, Tehran, Iran
Chair of Advisory Committee: Dr. Maria. A. Barrufet

Gas hydrates are a well-known problem in the oil and gas industry and cost millions of
dollars in production and transmission pipelines. To prevent this problem, it is important
to predict the temperature and pressure under which gas hydrates will form. Of the
thermodynamic models in the literature, only a couple can predict the hydrate-formation
temperature or pressure for complex systems including inhibitors.

I developed two simple correlations for calculating the hydrate-formation pressure or

temperature for single components or gas mixtures. These correlations are based on over
1,100 published data points of gas-hydrate formation temperatures and pressures with and
without inhibitors. The data include samples ranging from pure-hydrate formers such as
methane, ethane, propane, carbon dioxide and hydrogen sulfide to binary, ternary, and
natural gas mixtures. I used the Statistical Analysis Software (SAS) to find the best
correlations among variables such as specific gravity and pseudoreduced pressure and
temperature of gas mixtures, vapor pressure and liquid viscosity of water, and
concentrations of electrolytes and thermodynamic inhibitors.

These correlations are applicable to temperatures up to 90ºF and pressures up to 12,000

psi. I tested the capability of the correlations for aqueous solutions containing electrolytes
such as sodium, potassium, and calcium chlorides less than 20 wt% and inhibitors such as
methanol less than 20 wt%, ethylene glycol, triethylene glycol, and glycerol less than 40
wt%. The results show an average absolute percentage deviation of 15.93 in pressure and
an average absolute temperature difference of 2.97ºF.
 iv

Portability and simplicity are other advantages of these correlations since they are
applicable even with a simple calculator. The results are in excellent agreement with the
experimental data in most cases and even better than the results from commercial
simulators in some cases. These correlations provide guidelines to help users forecast
gas-hydrate forming conditions for most systems of hydrate formers with and without
inhibitors and to design remediation schemes such as:
• Increasing the operating temperature by insulating the pipelines or applying heat.
• Decreasing the operating pressure when possible.
• Adding a required amount of appropriate inhibitor to reduce the hydrate-
formation temperature and/or increase the hydrate-formation pressure.
 v

DEDICATION

To those whom I think of every moment of my life:

My parents, Hassan and Sareh

My brother and sisters, Shahram, Shoaleh, and Shohreh
My nieces and nephews, Ghazaleh, Raniya, Ala, and Ragheb
My brothers-in-law and sister-in-law, Malek, Emad, and Shayesteh
and
My beloved husband, Hassan

For their love, prayers, and encouragement

 vi

ACKNOWLEDGMENTS

I would like to express my appreciation to Dr. Maria Barrufet, the chair of my advisory
committee, for her great guidance and valuable advice and assistance in my research.

I would like to thank Dr. Stephen Holditch, the department head, for his interest in this
research and his support.

I appreciate Dr. John Lee, Dr. Mahmood Amani, and Dr. Malcolm Andrews for serving
in my advisory committee and for their helpful comments and suggestions on the
manuscript of my thesis.

My special thanks go to my dear parents and brother for their support, sacrifice, and
encouragement.
 vii

TABLE OF CONTENTS
Page

ABSTRACT....................................................................................................................... iii

DEDICATION.....................................................................................................................v

ACKNOWLEDGMENTS ................................................................................................. vi

LIST OF TABLES............................................................................................................. ix

LIST OF FIGURES ............................................................................................................ x

CHAPTER

I INTRODUCTION .......................................................................................1

II BACKGROUND .........................................................................................5

2.1 Gas Hydrate Formation..........................................................................5

2.2 Hydrate Structures .................................................................................5
2.3 Hydrate Phase Equilibrium ....................................................................8
2.4 Gas Hydrates and Problems in the Oil and Gas Industry.....................10
2.5 Ways to Prevent Hydrates Formation ..................................................11
2.6 Experimental Work..............................................................................12
2.7 Correlation Methods ............................................................................17
2.7.1 The K-Value Method ..................................................................18
2.7.2 The Gas Gravity Method ............................................................19
2.7.3 Kobayashi et al. Method .............................................................20
2.7.4 Hammerschmidt Method ............................................................21
2.8 Thermodynamic Methods ....................................................................22
2.9 Equations of State (EOS) .....................................................................27

III METHODOLOGY ....................................................................................30

3.1 Data Collection ....................................................................................30

3.2 Data Observations................................................................................32
3.3 Comments on Data...............................................................................35
3.4 Regression Variables ...........................................................................35
3.4.1 Pseudoreduced Temperature and Pressure..................................35
3.4.2 Gas Specific Gravity ...................................................................38
3.4.3 Water Vapor Pressure .................................................................38
3.4.4 Liquid Water Viscosity ...............................................................39
3.5 Hydrate-Formation Pressure Correlation.............................................39
3.6 Hydrate-Formation Temperature Correlation ......................................40
 viii

CHAPTER Page

IV RESULTS AND DISCUSSION ................................................................42

4.1 Predicted Results versus Experimental................................................42

4.2 Comparison of Predicted Results with a Common Correlation...........48
4.3 Comparison of Predicted Results with Calculated from PVTsim .......48
4.4 Sensitivity Analysis .............................................................................50

V CONCLUSIONS........................................................................................55

5.1 Conclusions from Observations...........................................................56

5.2 Conclusions from Developing the Improved Correlations ..................56

NOMENCLATURE ..........................................................................................................58

REFERENCES ..................................................................................................................60

APPENDIX A EXPERIMENTAL DATA ....................................................................65

APPENDIX B HYDRATE-FORMATION PRESSURE CALCULATION..................66

APPENDIX C HYDRATE-FORMATION TEMPERATURE CALCULATION ........67

VITA ..................................................................................................................................68
 ix

LIST OF TABLES

Page

Table 2.1 Components may enter cavities of hydrates SI and SII ...............................7

Table 2.2 Components may enter cavities of hydrate SH............................................7

Table 2.3 Coefficients for calculating the hydrate-formation temperature

from equation 2.3 .......................................................................................21

Table 2.4 Constants used for evaluating equation 2.7 ...............................................24

Table 2.5 The A and B parameters for calculating the Langmuir

constants (SI & SII) ...................................................................................25

Table 2.6 The A and B parameters for calculating the Langmuir

constants (SH)............................................................................................26

Table 3.1 Range of different independent variables for 1,104 data points ................32

Table 3.2 Effects of gas compositions on hydrate-formation pressure in

systems without inhibitors .........................................................................33

Table 3.3 Values of constants α and β for calculating J and K .................................37

Table 3.4 Range of data for developing the mixing rules..........................................37

Table 3.5 Values of constants for hydrate-formation p and T correlations................41

 x

LIST OF FIGURES

Page

Fig. 2.1 Cavities for hydrates of SI, SII, and SH.......................................................6

Fig. 2.2 Phase diagram for natural gas hydrocarbons which form hydrates .............9

Fig. 2.3 Formation of gas hydrate plugs a subsea hydrocarbon pipeline ................11

Fig. 2.4 Experimental hydrate equilibrium conditions for the ternary

mixture .......................................................................................................14

Fig. 2.5 Experimental hydrate equilibrium conditions for the natural gas
mixture .......................................................................................................15

Fig. 2.6 Experimental hydrate equilibrium conditions for pure carbon

dioxide in presence of pure water, 10.04 wt% EG, and 10 wt%
methanol.....................................................................................................16

Fig. 2.7 Experimental hydrate equilibrium conditions for a carbon

dioxide-rich gas mixture in presence of pure water, 10.04 wt%
EG, and 10 wt% NaCl................................................................................16

Fig. 2.8 Initial hydrate-formation estimation for natural gases based on

gas gravity..................................................................................................20

Fig. 3.1 Hydrate-formation pressure for binaries of CH4 with iC4 and nC4 ............34

Fig. 4.1 Comparison of experimental and calculated values of

hydrate-formation pressure ........................................................................43

Fig. 4.2 Comparison of experimental and calculated values of

hydrate-formation temperature ..................................................................43

Fig. 4.3 Comparison of experimental and calculated results of hydrate-

formation-pressure from p-correlation for pure methane ..........................44

Fig. 4.4 Comparison of experimental and calculated results of hydrate-

formation pressure from p-correlation for a natural gas with
low concentration of propane and nitrogen ...............................................45
 xi

Page

Fig. 4.5 Comparison of experimental and calculated results of hydrate-

formation pressure from p-correlation for a natural gas with
high concentration of propane and nitrogen ..............................................45

Fig. 4.6 Comparison of experimental and calculated results of hydrate-

formation temperature from T-correlation for pure methane.....................46

Fig. 4.7 Comparison of experimental and calculated results of hydrate-

formation temperature from T-correlation for a natural gas with
low concentration of propane and nitrogen ...............................................47

Fig. 4.8 Comparison of experimental and calculated results of hydrate-

formation temperature from T-correlation for a natural gas with
high concentration of propane and nitrogen ..............................................47

Fig. 4.9 Actual differences between predicted and experimental

temperatures for T-correlation and Kobayashi et al. correlation ...............48

Fig. 4.10 Comparison of the calculated hydrate-formation pressure from

PVTsim and p-correlation..........................................................................49

Fig. 4.11 Comparison of the calculated hydrate-formation temperature from

PVTsim and T-correlation..........................................................................50

Fig. 4.12 Calculated results from PVTsim before and after adjusting the
value of A for component C1 in a large cavity of Structure II..................53

Fig. 4.13 Calculated results from PVTsim before and after adjusting the
value of A for component C2 in a large cavity of Structure I ...................53
 1

CHAPTER I

INTRODUCTION

Gas hydrates are ice-like crystalline structures with gas components such as methane and
carbon dioxide as guest molecules entrapped into cavities formed by water molecules.
Whenever a system of natural gas and water exists at specific conditions, especially at
high pressure and low temperature, we expect the formation of hydrates. In the oil and
gas industry, gas hydrates are a serious problem in production and gas-transmission
pipelines because they plug pipelines and process equipment. By applying heat,
insulating the pipelines, and using chemical additives as inhibitors, we can keep the
operating conditions out of the hydrate-formation region.

The most common inhibitors are thermodynamic inhibitors such as methanol and glycols;
however, produced water that contains electrolytes also has inhibiting effects. To
remediate problems caused by hydrates, it is important to calculate the gas-hydrate
formation temperature and pressure accurately; this is more complex when the system
includes alcohols and/or electrolytes.

Hammerschmidt1 first found that the formation of clathrate hydrates could block natural
gas-transport pipelines. Since then, the oil and gas industry has been more willing to
investigate the problem. My work focuses on gas-hydrate formation in three-phase
equilibrium (liquid water, hydrocarbon gas, and solid hydrate) with the objectives of
developing a correlation to predict the gas-hydrate formation at a given temperature, a
correlation to predict the gas-hydrate formation temperature when pressure is available,
and guidelines to calibrate a thermodynamic model by analyzing sensitivity to selective
parameters such as temperature- dependent adsorption constant.

This thesis follows the style of SPE Reservoir Evaluation & Engineering.
 2

It is not practical to measure the gas-hydrate formation pressure and temperature for
every particular gas mixture. The main objective of this research is to develop
correlations to predict these conditions with the least average absolute error. To approach
that, I used over 1,100 experimental points2-18 among over 1,400 points gathered from
published literature from 1940 till 2004. The removed points are those that either
presented off-trend hydrate formation curves or those that contained high concentrations
of inhibitors. The data include samples ranging from single-hydrate formers such as
methane, carbon dioxide, ethane, propane, and hydrogen sulfide to binary, ternary, and
natural gases in the presence of pure water, electrolytes and/or alcohols. Using the
Statistical Analysis Software (SAS),19 I applied a regression model to find the best
correlations among the variables, such as specific gravity and pseudoreduced pressure
and temperature of gas mixtures, vapor pressure and liquid viscosity of water, and
concentrations of electrolytes and thermodynamic inhibitors.

I developed two correlations that are able to predict the hydrate formation pressure for a
given temperature or hydrate formation temperature for a given pressure for a single or
multicomponent gas mixture with and without electrolytes and/or thermodynamic
inhibitors. These correlations are applicable to a range of temperatures up to 90ºF and
pressures up to 12,000 psi. The capability of the correlations has been tested for aqueous
solutions containing electrolytes such as sodium, potassium, and calcium chlorides lower
than 20 wt% and inhibitors such as methanol lower than 20 wt%, ethylene glycol (EG),
triethylene glycol (TEG), and glycerol (GL) lower than 40 wt% since the use of higher
amount of these inhibitors is not practical because is very costly. The results show an
average absolute percentage deviation of 15.93 in pressure and an average absolute
temperature difference of 2.97ºF.

To make the correlations easy to use, I programmed them with Visual Basic. By giving
the gas compositions, the inhibitor concentrations, and either temperature or pressure of
the system, a user can calculate the hydrate-formation pressure or temperature as quickly
as clicking a key.
 3

Gas-hydrate plugging is a challenging and costly problem in gas-gathering systems and

transmission pipelines. Several models have been published in the literature, but not all of
them are applicable for a complex system including gas-hydrate formers and mixed
inhibitors. My correlations will provide guidelines to help the user forecast the gas-
hydrate formation pressure or temperature for a pure or mixed gas with and without
inhibitors at a given temperature or pressure. They will also be able to determine the most
appropriate inhibitor for the given system without the need of doing costly and time-
consuming experimental measurements. The advantage of these correlations is that they
will not require sophisticated calculations or a computer; instead, they are applicable
even with a simple calculator. The disadvantage of these correlations is that they may not
be appropriate in some cases with high concentrations of inhibitors.

Chapter II of this thesis gives general information about the phase equilibrium of forming
hydrates and different types of determined hydrate structures, problems that they may
cause in the oil and gas industry, and solutions that may prevent their formation. This
chapter also reviews the literature in terms of experimental works, the available
correlation methods, and finally the basis of calculating the hydrate-formation conditions
from thermodynamic models. Chapter III explains the methodology for developing the
proposed correlations including my observations from the experimental data, the
regression variables that I used in this work, and an introduction to the new correlations
that improved the estimation of hydrate-formation conditions in systems with and without
inhibitors. Chapter IV includes the results of the regression models for both correlations;
it also shows the comparisons of calculated results from this work with the experimental
data, with a commonly used correlation, and with the results predicted by the PVTsim20
simulator for several gas systems. Chapter V contains the conclusions from this work and
from data observations.

There are three appendixes that come separately in Excel files. Appendix A includes the
experimental data gathered and used in this work. Appendix B contains a Visual Basic
program that calculates hydrate-formation pressure at a given temperature and Appendix
 4

C is a Visual Basic program that calculates hydrate-formation temperature at a given

pressure.
 5

CHAPTER II

BACKGROUND

2.1 Gas Hydrate Formation

Gas hydrates are nonstoichiometric compounds formed from mixtures of water and gas
molecules under suitable pressures and temperatures. Gas molecules with adequate size
become guest molecules entrapped in the cavities of cages formed by water molecules
acting as host molecules. Hydrates are also called clathrates, which in Latin means,
“cage.” When a minimum number of cavities are occupied by the gas molecules, the
crystalline structure stabilizes and solid gas hydrates may form at temperatures above the
water freezing point. Most light molecules such as methane, ethane, propane, isobutane,
normal butane, nitrogen, carbon dioxide, and hydrogen sulfide will form hydrates under
specific conditions of pressure and temperature; however, several heavy hydrocarbons
such as benzene, cyclopentane, cyclohexane, methylcyclopentane, methylcyclohexane,
isopentane and 2,3-dimethylbutane have been recently identified as hydrate formers.21

2.2 Hydrate Structures

Von Stackelberg and Muller22 studied the hydrate structure using X-ray diffraction
methods. Their work along with works by Classen23, 24 identified two hydrate structures,
Structure I (SI) and Structure II (SII) that each has two types of cavities. The SI hydrates
consist of 46 water molecules per eight cavities, two small spherical cavities with 12
pentagonal faces (512) and six large oblate cavities with two hexagonal faces and 12
pentagonal faces (51262).25 The SII hydrates consist of 136 water molecules per 32
cavities, 16 small cavities with 12 pentagonal faces (512) and eight large cavities with 12
pentagonal and four hexagonal faces (51264), all in a spherical shape. Fig. 2.1 shows these
cavities.
 6

Fig. 2.1—Cavities for hydrates of SI, SII, and SH.25

From 1959 to 1967, Jeffrey, McMullan, and Mak26-28 studied crystallography on hydrates
SI and SII. A summary of their experience showed that hydrates are “clathrates”. It is
well known that small gas molecules such as CH4, C2H6, and CO2 form hydrate Structure
I, but gas molecules with larger size such as C3H8 and i-C4H10 form hydrate Structure II.
However, some of the small gas molecules like Ar and Kr form both hydrate structures.7

Table 2.1 and Table 2.2 show the molecules that may enter hydrate cavities.
 7

Table 2.1—COMPONENTS MAY ENTER CAVITIES OF HYDRATES SI AND SII20, 29

Component Structure I Structure II
Small Large Small Large
Cavities Cavities Cavities Cavities
C1 + + + +
C2 _ + _ +
C3 _ _ _ +
nC4 _ _ _ +
iC4 _ _ _ +
CO2 + + + +
N2 + + + +
H 2S + + + +
O2 + + + +
Ar + + + +
2,2 Dimethylpropane _ _ _ +
Cyclopropane _ _ _ +
Cyclohexane _ _ _ +
C 6H 6 _ _ _ +

Table 2.2—COMPONENTS MAY ENTER CAVITIES OF HYDRATE SH20, 29

Component Small/Medium Cavities Huge Cavities
C1 + _
N2 + _
iC5 _ +
Neohexane _ +
2,3-Dimethylbutane _ +
2,2,3-Trimethylbutane _ +
3,3-Dimethylpentane _ +
Methylcyclopentane _ +
1,2- Dimethylcyclohexane _ +
Cis-1,2- Dimethylcyclohexane _ +
Ethylcyclopentane _ +
Cyclooctane _ +
 8

Ripmeester et al.30 discovered a third type of hydrate structure (Structure H). The
formation of hydrate SH requires both small and large molecules to be stabilized. The
hydrates with SH contain 34 water molecules per six cavities, three cavities formed by12
pentagonal (512), two cavities formed by three square, six pentagonal, and three
hexagonal faces (435663), and one large cavity formed by 12 pentagonal and eight
hexagonal faces (51268).25

Hydrate formation of type sH requires large gas molecules such as methylcyclopentane,

which are generally found in gas-condensate and oil systems. My work focuses on
Structure I and mostly Structure II, which are basically formed by natural gas. The
structure type of hydrates does not affect the magnitude of flow blockage in wells or
pipelines; however, most of the thermodynamic models consider the effects of the
hydrate structures and the size of their cavities as we will see in Section 3.3. In this work,
since none of the variables represent the hydrate structures in the regression model, the
structure of hydrates has not been directly involved in the development of the new
correlations; however, because components with different sizes form different types of
hydrate structures, we assume that our correlations have accounted for the hydrate
structure in their specific gravity and pseudoreduced temperature and pressure variables.

Tohidi et al. measured the SII equilibrium data for benzene, cyclohexane, cyclopentane,
and neopentane.31, 32 Becke et al.33 measured SH for methane+methylcyclohexane, and
Ostergaard et al.34 for isopentane and 2,2-dimethylpentane in their binaries and ternaries
with methane and/or nitrogen. Mehta and Sloan35 provided an overview of the state-of-the
art on SH hydrates with an emphasis on its implications for the petroleum industry.

2.3 Hydrate Phase Equilibrium

Fig. 2.22,36 shows the hydrate equilibrium curve (I-H-V), (LW-H-V), (LW-H-LHC) for
several components. The letters H, I, V, LW, and LHC represent hydrate, ice, hydrocarbon
vapor, liquid water, and hydrocarbon liquid respectively. The lower quadruple point, Q1
indicates the point at which the four-phase ice, liquid water, hydrocarbon vapor and
 9

hydrate (I-LW-H-V) are in equilibrium. The temperature at this point approximates the ice
point.

10000

Methane

1000
Pressure, psia

Q2
Ethane

LW-H-V

Propane
100

Q1

Q Isobutane

10
20 30 40 50 60 70 80 90

Temperature, ºF

Fig. 2.2—Phase diagram for natural gas hydrocarbons which form hydrates (after
McCain).36
 10

The point Q2 is the upper quadruple point at which the four-phase water liquid,
hydrocarbon liquid, hydrocarbon vapor, and hydrate (LW-LHC-V-H) are in equilibrium.
The pressures and temperatures at the Q1Q2 line represent the conditions that three-phase
liquid water, hydrocarbon vapor and hydrate are in equilibrium. Therefore, at the right
side of this line no hydrates form; however, hydrates begin to form at this line and
become more stable at a higher pressure and/or lower temperature. Since our objective in
this work is to predict the incipient hydrate-formation pressure or temperature, all the
experimental data gathered and used in developing the new correlations are those that
represent the three-phase equilibrium line (LW-H-V).

2.4 Gas Hydrates and Problems in the Oil and Gas Industry
Hammerschmidt1 determined that natural gas hydrates could block the gas transmission
pipelines sometimes at temperature above the water freezing point. This discovery
highlighted the importance of hydrates to the oil and gas industry and was an introduction
to the modern research era.

Gas hydrates are a very costly problem in petroleum exploration and production
operations. Hydrate clathrates can plug gas gathering systems and transmission pipelines
subsea and on the surface. In offshore explorations, the main concern is the multiphase
transfer lines from the wellhead to the production platform where low seabed
temperatures and high operation pressures promote the formation of gas hydrates. Fig.
2.3 shows plugging of a subsea hydrocarbon pipeline because of hydrate formation.
 11

Fig. 2.3—Formation of gas hydrate plugs a subsea hydrocarbon pipeline.25

2.5 Ways to Prevent Hydrate Formation

The following are the thermodynamic ways to prevent the hydrate formation:
1. Reducing the water concentration from the system.
2. Operating at temperatures above the hydrate-formation temperature for a given
pressure by insulating the pipelines or applying heat.
3. Operating at pressures below the hydrate-formation pressure for a fixed
temperature.
4. Adding inhibitors such as salts, methanol, and glycols to inhibit the hydrate-
formation conditions and shift the equilibrium curve to higher pressure and lower
temperature.

Inhibitors are added into processing lines to inhibit the formation of hydrates. There are
two kinds of inhibitors: thermodynamic inhibitors and low-dosage inhibitors.37 The
thermodynamic inhibitors have been used for long time in the industry and act as
 12

antifreeze. The low-dosage inhibitors have recently been developed and their usage
modifies the rheology of the system rather than changing its thermodynamic states. These
inhibitors work at low concentrations, lower than or equal to 1 wt%; therefore, the use of
this technique reduces the environmental concerns and since no regeneration units are
required, it results in reduction of capital cost. The low-dosage inhibitors are divided into
kinetic inhibitors and antiagglomerants. The kinetic inhibitors are commonly water-
soluble polymers delay the nucleation and growth of hydrate crystals, while the anti-
agglomerants are usually surfactants and miscible in both hydrocarbon and water, so they
impede the agglomeration of hydrate crystals for a period of time without interfering with
crystal formation.

2.6 Experimental Work

Ng and Robinson38 obtained experimental data on initial hydrate formation conditions for
the nitrogen-propane-water system in the LW-H-V, LW-LHC-H, and LW-LHC-H-V regions,
where LW is the water-rich liquid phase, LHC is the hydrocarbon rich liquid phase, H is the
hydrate and V is the vapor phase. The measurements covered a range of temperatures
from about 275 to 293ºK and pressures from about 0.3 to 17 MPa with the concentrations
of propane from 0.94 to 75 mol% in the gas phase for the LW-H-V region, and from 83.1
to 99 mole percent in the condensed liquid phase for the LW-LHC-H region. Ng and
Robinson used these experimental data to fit a binary interaction parameter to predict
hydrate formation in systems containing nitrogen and propane. Based on their proposed
method, Ng and Robinson39 found the best value of the interaction parameter for
nitrogen-propane mixtures to be 1.03, which is much higher than usual values (-0.5, 0.5).
They reported that using this parameter will reduce the absolute average error from 15.3
to 5.7% for predicting the hydrate-formation pressures at a given temperature in the LW-
H-V region. The importance of this parameter in this system becomes more significant as
the concentration of propane in the gas phase becomes higher.

Most of the experimental studies on gas hydrates have investigated systems in the
presence of pure water but have lacked information on hydrate inhibition. Ng and
Robinson11 studied the hydrate-forming conditions for pure gases, including methane,
 13

ethane, propane, carbon dioxide, and hydrogen sulfide, and for selected binary mixtures
in the presence of solutions up to 20 wt% methanol. This study was carried out in both
the LW-H-V and the LW-H-LHC regions for all the mentioned hydrate formers, but for
methane only in the LW-H-V region. The experimental measurements covered a range of
pressures from about 0.8 to 20 MPa, temperature from -10 to 17ºC, and concentration of
methanol from 5 to 20 wt%. Ng and Robinson11 used the results of these measurements to
compare with the calculated values from the Hammerschmidt equation29 as we will see in
Section 2.7.4. This equation calculates the difference between the temperature of a
system in the presence of water and the temperature of system in an inhibitor solution.
The difference between experimental and calculated hydrate-temperature depression from
their experiment was less than 1ºC for CH4, C2H6, and C3H8 in the gaseous region and
more than 1ºC in the region of liquid. This difference was more than 1ºC for CO2 in
gaseous and liquid regions. The results show that the Hammerschmidt equation over-
estimates the hydrate-temperature depression for H2S in the gaseous region but provides
estimates for this component than the other components in the liquid region.

Inhibitors such as ethylene glycol, methanol, and electrolytes inhibit hydrate formation. It
is important to determine the inhibition effects of these additives to avoid hydrate
formation and select the best inhibitor for a given system and operating conditions.
Bishnoi and Dholabhai40 obtained experimental hydrate equilibrium conditions for
propane hydrate with single and mixed electrolytes. Their work included electrolytes
such as NaCl, KCl, and CaCl2 at pressure and temperature ranges of 133 to 500 KPa and
263 to 276ºK. The results of this work show that for the same concentrations of
electrolytes (5 and 10 wt% in this case), sodium chloride has a greater inhibition effect
than potassium and calcium chlorides.

Bishnoi and Dholabhai5 obtained the hydrate-equilibrium conditions for a ternary mixture
of methane (78 mol%), propane (2 mol%) and carbon dioxide (20 mol%) and a natural
gas mixture in pure water and solutions containing methanol and electrolytes for a
temperature range of 274 to 291ºK and a pressure range of 1.5 to 10.1 MPa. They
observed systems that contain the same total wt% of the inhibitor, for example systems
 14

with 10 wt% of either methanol or sodium chloride and 20 wt% of either methanol or
sodium chloride, 15 wt% of methanol + 5 wt% of sodium chloride, and 5 wt% of
methanol + 15 wt% of sodium chloride. For a given pressure, they reported that the
incipient hydrate-equilibrium conditions for such systems are close to each other, within
3 to 5ºC (Figs. 2.4 and 2.5); one can also conclude from these two figures that sodium
chloride has higher inhibition potential than methanol with the same wt%, a result is
more pronounced at higher pressures. Even in the presence of mixed inhibitors, the
inhibitor with a higher wt% of sodium chloride is more effective than the one with higher
wt% of methanol.

C1=78, C3=2, CO2=20 (mol%)

1400

1200
Pressure, psi

1000

800

600

400

200
30 35 40 45 50 55 60

Temperature, ºF

10w t% CH3OH 10w t% NaCl 20w t% CH3OH 20w t% NaCl

Fig. 2.4—Experimental hydrate equilibrium conditions for the ternary mixture.5

 15

C1=82, CO2=0.5, C2=11.3, C3=4.2, iC4=0.9, nC4=0.6, iC5=0.1,

nC5=0.2, nC6=0.2 (mol%)
1600

1400
Pressure, psi

1200

1000

800

600

400
35 40 45 50 55

Temperature, ºF
15w t% NaCl+5w t% CH3OH 15w t% CH3OH+5w t% NaCl 20w t% NaCl 20w t% CH3OH

Fig. 2.5—Experimental hydrate equilibrium conditions for the natural gas mixture.5

Fan et al.18 measured the hydrate-formation pressures of pure carbon dioxide in water, 10
wt% methanol, and 10 wt% ethylene glycol (EG) solutions and concluded that the
inhibition effect of EG is inferior to that of methanol, as indicated in Fig. 2.6. To
compare the inhibition effects, they also determined the hydrate formation data for a
carbon dioxide-rich quaternary gas mixture containing 88.53 mol% CO2, 6.83 mol%
CH4, 4.26 mol% N2, and 0.38 mol% C2H6 in presence of 10 wt% EG and 10 wt% NaCl.
The results show that the inhibition of EG is less effective (Fig. 2.7).
 16

500

400
Pressure, psi

300

200

100
25 30 35 40 45

Temperature, ºF

Without Inhibitor 10.04w t% EG 10w t% CH3OH

Fig. 2.6—Experimental hydrate equilibrium conditions for pure carbon dioxide in

presence of pure water, 10.04 wt% EG, and 10 wt% methanol.18

C1=6.83, C2=0.38, CO2=88.53, N2=4.26 (mol%)

700

600
Pressure, psi

500

400

300

200

100
20 25 30 35 40 45

Temperature, ºF

Without Inhibitor 10w t% EG 10w t% NaCl

Fig. 2.7—Experimental hydrate equilibrium conditions for a carbon dioxide-rich

gas mixture in presence of pure water, 10.04 wt% EG, and 10 wt% NaCl.18
 17

In another work, Ma et al.16 determined the initial hydrate formation data of pure
ethylene, five binary gas mixtures of methane/ethylene, and four binary gas mixtures of
methane/propylene for temperatures of 273.7 to 287.2ºK and pressures of 0.53 to 6.6
MPa. The ethylene and propylene contents in the mixtures range from 7.13 to 100 mol%
and 0.66 to 71.96 mol%. Their work showed that the model developed by Chen and Guo
represented the measured data; however, my conclusion from the experimental results is
that hydrates could form at higher pressure as the concentration of methane increased in
the mixture. This means that at the same temperature, the lighter the gas specific gravity
of the hydrate former, the higher the pressure at which hydrates form, as indicated in the
literature.36

Sometimes the processed water in pipelines contains electrolytes which also act as
inhibitors. To establish the effect of mixtures of inhibitors on the locus of the three-phase
equilibrium curve, Jager, Peters, and Sloan12 measured data on eight different mixtures of
the quaternary system of methane/water/methanol/sodium chloride. They reported 16
data points at a relative concentration of 2 and 4 mol% sodium chloride combined with
10, 20, 30 and 40 wt% of methanol. Using two different experimental methods, they
measured the incipient hydrate values for pressures from 2 to 14 MPa in a Cailletet
apparatus and from 2 to 70 MPa in a Raman spectroscopy, which had not been used
before to measure hydrate data in complex systems. The results from the two apparatus at
10 MPa are consistent within 0.3 to 1ºK. They compared the data collected in their work
with literature data for the ternary systems of methane/water/sodium chloride and
methane/water/methanol and concluded that the mixtures of inhibitors (sodium chloride +
methanol in this case) have a larger inhibition effect than the sum of the inhibition effect
by each inhibitor; therefore, thermodynamic models must consider the interaction
between electrolytes and methanol to predict hydrate inhibition correctly.

2.7 Correlation Methods

It is well known that Davy discovered hydrates in 1810 and his discovery was confirmed
by Faraday in 1823;2 however, hydrates became a subject of study in the oil and gas
industry after Hammerschmidt1 found that hydrates could plug natural gas pipelines and
 18

process equipment. After Hammerschmidt’s discovery of hydrate blocking in 1934, Katz

et al.41 started an experimental study on hydrates. Because it was impractical to measure
the hydrate formation pressure and temperature for every gas compositions they
estimated the hydrate formation conditions for natural gases using two approaches. Since
these models were developed before discovery of the structure H hydrate, they are only
able to predict the hydrate formation conditions for light hydrate formers that form
Structure I and Structure II hydrates but not Structure H.

2.7.1 The K-Value Method

In the first approach, Wilcox et al.42, 43
initiated the K − Value method based on
distribution coefficients ( K i values) for components on a water-free basis. In the
finalized method, they determined that hydrates were a solid solution that might be
treated similarly to an ideal liquid solution and defined the value as the vapor/solid
equilibrium ratio of a component in LW-H-V equilibrium by the following equation:

K ivs = yi / si ,………………………………………………………………………...... (2.1)

where yi = mol fraction of component i in the vapor phase and s i = mol fraction of

component i in the solid phase. Therefore, similar to the dewpoint calculation in

vapor/liquid equilibria, the K − Value charts are used to calculate the hydrate formation
temperature or pressure of three-phase (LW-H-V) solution in a manner that satisfies the
following equation:

n
yi
= 1 …………………………………………………………………………… (2.2)
i =1 K ivs

The K-Value method was generated before determination of the hydrate-crystal structure
and was improved by Katz and co-workers. The K-Value charts were generated for
methane, ethane, propane, butane, carbon dioxide, hydrogen sulfide, and nitrogen.
Having the K-Value of every component in the mixture at three-phase (LW-H-V)
 19

equilibrium, users could determine the hydrate-formation pressure at a given temperature

or vice versa. This method is limited to the hydrate-formation pressures up to 4,000 psia
for methane, ethane, and propane; up to 2,000 psia for isobutane and hydrogen sulfide;
and up to 1,000 psia for carbon dioxide.

2.7.2 The Gas Gravity Method

In a different approach, Katz and his students2, 43 generated the gas-gravity plot (Fig. 2.8)
that relates the hydrate pressure and temperature with the specific gravity (gas molecular
weight divided by that of air) of natural gases, not including non-hydrocarbons. This plot
was generated from limited experimental data from Deaton and Frost, Wilcox et al.,
Kobayashi and Katz and significant calculations based on the K-Value method.2

This method is simple and may be used for an initial estimation of hydrate formation
conditions. Elgibaly and Elkamel44 have mentioned in their paper that Sloan showed in a
statistical analysis report that this method is not accurate and that the calculated pressure
for the same gas gravity with different mixtures may result in 50% error. Since method
considers only the gas gravity of the components, if two components have equal
molecular weights such as butane and isobutene, the method may estimate the same
hydrate-formation temperature or pressure, although they should be different in reality. I
have shown on page 34 the experimental data for two binary gases with the same gas
specific gravity in the same range of temperature that form hydrates at a very different
range of pressure.

For three-phase (LW-H-V) conditions, Kobayashi et al.43 developed an empirical equation

that predicts the hydrate temperatures at given pressures for systems including only
hydrocarbons in limited range of temperatures, pressures, and gas specific gravities. I
have compared the calculated results from Kobayashi et al. with the calculated results
from our new correlation in Chapter IV.
 20

10000

Methane
1000
Pressure, psia

0.7
0.6
0.8
0.9
1.0
100

10
30 40 50 60 70 80

Temperature, ºF

Fig. 2.8—Initial hydrate-formation estimation for natural gases based on gas gravity
(after McCain).36

2.7.3 Kobayashi et al. Method

Kobayashi et al.43 developed Eq. 2.3 on the basis of the gas-gravity plot to estimate the
hydrate-formation temperature at a given pressure. The reference did not give the units
for temperature and pressure; therefore, I had to try different combinations of units for
temperature and pressure to find the units that best predicted the temperature of the
experimental data. I found that by having pressure in bar and temperature in ºC, the
Kobayashi et al. correlation would have the best results with experimental data.
 21

T = 1 /[c1 + c 2 (ln p ) + c3 (ln γ ) + c 4 (ln p ) 2 + c5 (ln p )(ln γ ) + c 6 (ln γ ) 2 + c 7 (ln p ) 3

+ c8 (ln γ )(ln p ) 2 + c9 (ln γ ) 2 (ln p ) + c10 (ln γ ) 3 + c11 (ln p) 4 + c12 (ln γ )(ln p ) 3
+ c13 (ln γ ) 2 (ln p ) 2 + c14 (ln γ ) 3 (ln p ) + c15 (ln γ ) 4 ] ............................................... (2.3)

This method is a regression method that correlates temperature (T ), pressure ( p ), and

gas specific gravity ( γ ) at three-phase equilibrium. The equation is applicable in the
temperature range of 34 to 60ºF, the pressure range of 65 to 1,500 psia, and gas gravity
range from 0.552 to 0.9. The three-phase condition was fit only for hydrocarbons and not
gases containing CO2 and H2S. Table 2.3 shows the coefficients for this correlation.

Table 2.3—COEFFICIENTS FOR CALCULATING THE HYDRATE-FORMATION

TEMPERATURE FROM EQUATION 2.343
c1 = 2.7707715 × 10 −3 c 2 = − 2.782238 × 10 −3 c3 = −5.649288 × 10 −4
c 4 − 1.298593 × 10 −3 c5 = 1.407119 × 10 −3 c6 = 1.785744 × 10 −4
c7 = 1.130284 × 10 −3 c8 = 5.9728235 × 10 −4 c9 = −2.3279181 × 10 −4
c10 = −2.6840758 × 10 −5 c11 = 4.6610555 × 10 −3 c12 = 5.5542412 × 10 −4
c13 = −1.4727765 × 10 −5 c14 = 1.3938082 × 10 −5 c15 = 1.4885010 × 10 −6

2.7.4 Hammerschmidt Method

Hammerschmidt2, 29 proposed an empirical equation to calculate the effect of alcohols on
hydrate formation; however, his work includes no experimental data on the effect of
inhibitors added to the water and it cannot be used unless the hydrate-formation
conditions in the presence of pure water have been determined.

kjxj
∆T = ......................................................................................... (2.4)
(100 × M j − x j × M j )

The following are the values of constant k for different inhibitors29:

 22

k = 2,335 for methanol.

k = 2,700 for ethylene glycol.
k = 5,400 for triethylene glycol.

∆T is the difference in ºC between the hydrate-formation temperatures in the presence of

pure water and in a methanol solution, M j is the molecular weight of the inhibitor j ,

and x j is the concentration of inhibitor j in weight percent.

2.8 Thermodynamic Methods

Parrish and Prausnitz45 developed the first thermodynamic model for calculating hydrate-
formation conditions based on a statistical method by van der Waals and Platteeuw. Du
and Guo10 developed a model to predict the hydrate-formation conditions for systems
including alcohol solutions. The model by Javanmardi and Moshfeghian4 can predict the
hydrate-formation conditions for systems including electrolyte solutions. If the system
includes electrolytes and alcohol, the model by Nasrifar et al.8 and the model by Nasrifar
and Moshfeghian3 can be used to predict the hydrate formation conditions.

The transformation from a pure-water state to a hydrate state can be considered in two
steps:

1) pure water (aq ) → empty hydrate lattice ( β ), and

2) empty hydrate lattice ( β ) → filled hydrate lattice ( H ),

where aq indicates the state of pure water, H the filled hydrate lattice, and β indicates
the empty hydrate lattice, which is hypothetical but used to facilitate the hydrate
calculations.

In a system at three-phase equilibrium of vapor/hydrate/aqueous, the chemical potential

of water in hydrate and aqueous phases is equal and can be expressed as:
 23

µ aq = µ H ……………………………………………………………………………. (2.5)
If µ β is the indication of the hypothetical empty-hydrate phase, then Eq. 2.5 can be
written as:

∆µ aq = ∆µ H ,……………………………………………………………………….... (2.6)

where ∆µ aq = µ β − µ aq and ∆µ H = µ β − µ H .

The term of ∆µ aq at a given temperature and pressure has been defined by Holder et al.46
as:

T
∆µ aq / RT = ∆µ o (To , p = 0 atm) / RTo − (∆H W / RT 2 )dT + (∆VW / RT ) p − ln aW .... (2.7)
To

T and p are hydrate-formation temperature and pressure, To indicates the reference

temperature, 273.15º K , R is the universal gas constant, and aW is the water activity in

the aqueous phase. The term of ∆VW is molar volume associated with transition and

∆H W (molar enthalpy difference) is independent of pressure and is defined by:

T
∆H W = ∆H o + ∆C p dT …………………………………………………………..... (2.8)
To

The term ∆C p is a function of temperature and is given by:

∆C p = a + b(T − To ) .................................................................................................. (2.9)

The values of the constants needed for calculation of ∆µ aq are given in Table 2.4.
 24

TABLE 2.4—CONSTANTS USED FOR EVALUATING EQUATION 2.720, 29

Property Unit Structure I Structure II

∆µ o (liq ) J/mol 1264 883

∆H o (liq ) J/mol -4858 -5201

∆H o (ice) J/mol 1151 808

∆Vo (liq ) Cm3/mol 4.6 5.0

∆Vo (ice) Cm3/mol 3.0 3.4

∆C p (liq ) J/mol/K 39.16 39.16

The chemical potential difference of water in the empty and the filled hydrate lattice was
derived by van der Waals and Platteeuw47 as follows:

∆µ H = RT ni ln(1 − f j C ji ) ,............................................................................... (2.10)

i j

where ni is the number of cavities of type i per water molecules, and f j is fugacity of

the component j in the gas phase and is determined by an equation of state, EOS. The

parameter C ji is the Langmuir adsorption constant, a function of temperature and specific

for the cavity of type i and for component j .

C ji = A / T exp( B / T ) ................................................................................................. (2.11)

Constants A and B are unique for each component j that is capable of entering into a
cavity of type i and must be determined from experimental data. These parameters are
specific for the selected EOS and according to PVTsim,20 for Structures I and II are
mostly calculated by Munck et al.,20 Rasmussen and Pederson,20 and for Structure H by
Madsen et al.20
 25

Table 2.5 and Table 2.6 give the values of the A and B parameters used in PVTsim.20

TABLE 2.5—THE A AND B PARAMETERS FOR CALCULATING THE LANGMUIR

CONSTANTS (SI & SII)20
Gas Structure Small Cavity Large Cavity

A × 10 3 B A × 10 3 B
(K/atm) (K) (K/atm) (K)
I 0.7228 3187 23.35 2653
C1
II 0.2207 3453 100 1916
I 0 0 3.039 3861
C2
II 0 0 240 2967
C3 II 0 0 5.455 4638
iC4 II 0 0 189.3 3800
nC4 II 0 0 30.51 3699
I 1.671 2905 6.078 2431
N2
II 0.1742 3082 18 1728
I 0.00588 5410 3.36 3202
CO2
II 0.0846 3602 846 2030
I 10.06 2999 16.34 3737
H 2S
II 0.065 4613 252.3 2920
I 17.4 2289 57.7 1935
O2
II 14.4 2383 154 1519
I 25.8 2227 75.4 1918
Ar
II 21.9 2315 1866 1539
 26

TABLE 2.6—THE A AND B PARAMETERS FOR CALCULATING THE LANGMUIR

CONSTANTS (SH)20

Compound Small Cavity Large Cavity

A × 10 3 B A × 10 3 B
(K/atm) (K) (K/atm) (K)
C1 2.800 × 10 −4 3390
N2 1.336 × 10 −5 3795
iC5 1.661 × 10 4 1699

Neohexane 1.627 × 10 3 3175

2,3-Dimethylbutane 1.747 × 10 2 3608

2,2,3-Trimethylbutane 8.066 × 10 8 − 39

3,3-Dimethylpentane 2.826 × 10 3 3183

Methylcyclopentane 6.420 × 101 4024

1,2- Dimethylcyclohexane 3.912 × 101 5050

Cis-1,2- 1.826 × 10 3 3604

Dimethylcyclohexane
Ethylcyclopentane 1.332 × 10 2 4207

Cyclooctane 1.647 × 10 3 4135

Replacing Eqs. 2.7 and 2.10 in Eq. 2.6 results in the following equation:
 27

T
∆µ o (To , p = 0 atm) / RTo − (∆H W / RT 2 )dT + (∆VW / RT ) p − ni ln(1 + C ji f j )
To
i j

− ln aW = 0 .................................................................................................................. (2.12)

After calculating the water activity aW from one of the equations derived by Javanmardi
and Moshfeghian4 or Nasrifar and Moshfeghian,3, 8 Eq. 2.12 can be used to calculate the
gas hydrate-formation temperature for a given pressure or gas hydrate pressure at a given
temperature for a system containing aqueous electrolytes only or in the presence of both
electrolytes and alcohol.

2.9 Equations of State (EOS)

An equation of state (EOS) relates the pressure (p), temperature (T), and volume (V) of a
given system mathematically and is a tool that can predict the phase behavior and the
volumetric properties of fluids. In hydrate prediction, an EOS can be used to determine
the fugacity of each component in the gas phase (Eq. 2.10). Section 4.4 will discuss
fugacity and its calculation using an EOS in more detail. Several EOS are available in the
literature and each is useful for different applications. Cubic EOSes, the most commonly
used in petroleum engineering, are cubic polynomials in volume. They are explicit in
pressure and can be written as a sum of repulsion and attraction terms.

p = p repulsion + p attraction ................................................................................................ (2.13)

The following equation has been defined by Soave-Redlich-Kwong (SRK)48:

RT a (T )
p= − ,............................................................................................... (2.14)
V − b V (V + b)

where T , p, V , and R are the temperature, pressure, molar volume, and universal gas
constant. The EOS parameters, a and b , have different values depending on the EOS; for
a pure component, they are evaluated at the critical temperature using the following two
equations:
 28

∂p
( ) T = 0 ................................................................................................................. (2.15)
∂V c
∂2 p
( ) T = 0 ............................................................................................................... (2.16)
∂V 2 c

For component i , the parameter a at the critical point and parameter b are defined by:

R 2Tci2
a ci = Ω a ,......................................................................................................... (2.17)
Pci
and
RTci
bi = Ω b ,............................................................................................................. (2.18)
Pci

with Ω a = 0.42748 and Ω b = 0.08664 .

The parameter a is a function of temperature and can be defined as:

a i (T ) = a ciα i (T ) ,....................................................................................................... (2.19)

where α i (T ) = [1 + m(1 − Tr )]2 ................................................................................ (2.20)

The term Tr is the reduced temperature (temperature divided by critical temperature) and
m is given by:

mi = 0.480 + 1.574ω i − 0.176ω i2 ................................................................................ (2.21)

At the critical temperature, the right-hand side of the Eq. 2.20 is equal to 1 and
consequently, in Eq. 2.19, a i (T ) will be equal to a ci . In Eq. 2.21 ω i is the acentric factor
and defined by Pitzer49:
 29

ω i = − log p riv (at Tri = 0.7) − 1 ,................................................................................... (2.22)

p riv is the reduced vapor pressure of component i ( p / p ci ).

The parameters a and b for mixtures can be determined from the following mixing rules:

aM = z i z j (ai a j ) 0.5 (1 − K ij ) ,.............................................................................. (2.23)

i j

and

bM = z i bi ,.............................................................................................................. (2.24)
i

where z i and z j represent the mole fractions of components i and j in the mixture and

K ij is the binary interaction coefficient between those components.

 30

CHAPTER III
METHODOLOGY

3.1 Data Collection

In this research, I gathered over 1,400 data point at the three-phase equilibrium of
different gas systems and used over 1,100 of them. The points that have not been
considered in this development are mostly the points with high concentrations of
inhibitors. For example, I removed those samples with electrolyte and methanol
concentrations equal to or higher than 20 wt%, because adding inhibitors with higher
concentrations is neither practical nor economic. I collected the data from Sloan2 and
literature published from 19402-18 to 2004. My collection included data from pure
components such as methane, carbon dioxide, ethane, propane, and hydrogen sulfide to
natural gas systems in the presence of pure water, electrolytes and/or alcohols. A total of
12 hydrocarbons, three nonhydrocarbons, three electrolytes, and four thermodynamic
inhibitors were involved in this development.

I used Marisoft Digitizer50 software to translate data presented in graphic form only to
tabulated data. To do this, I provided a JPEG file of those data reported in graphs, then by
opening the file in Marisoft Digitizer environment and selecting the ranges for both X
and Y axes, pointed on each experimental data point and transferred the digitized points
to an Excel file. To have the temperature and pressure for all data in the same units, I
converted all the different units of temperatures to ºF and all the different units of
pressures to psi, because these units are commonly used in the industry.

In this work, I used the Statistical Analysis Software (SAS)19 to find the best correlations
among the variables, such as gas specific gravity and pseudoreduced pressure and
temperature of gas mixtures, vapor pressure and liquid viscosity of water, and
concentrations of electrolytes and thermodynamic inhibitors. Because of large number of
independent variables, particularly hydrocarbons and nonhydrocarbons (15 components),
I reduced those to only three variables, pseudoreduced temperature and pressure and gas
 31

specific gravity, to represent all the gas components. The observations from the data
gathered in this work show that in systems without inhibitors the gases with lower
specific gravity form hydrates at higher pressure or lower temperature; these observations
also show that the inclusion of some components such as propane, isobutene, and
nitrogen causes different behavior for these systems. Taking into account all of these
observations, I considered the use of specific gravity and pseudoreduced temperature and
pressure as regression variables for developing these correlations.

The other regression variables, vapor pressure and liquid viscosity of water (especially
water vapor pressure), were helpful in modeling the p-correlation, because the plot of
vapor pressure of water versus temperature has the same shape as that for hydrate
pressure versus temperature. Since the hydrate-formation process is considered to be a
physical rather than a chemical process36 (the guest molecules can rotate within the void
spaces and no strong chemical bonds are formed between the hydrocarbon and water
molecules), and because water is the most important element in this process, the physical
properties of water such as liquid water viscosity at equilibrium could contribute to
allowing the gas molecule to enter the void space as a guest.

The data include about 250 samples from pure components to binary, ternary and
mixtures of gases in the presence of pure water, single and mixed inhibitors. The
concentration ranges of each gas component and inhibitor along with the ranges of
specific gravity, pressure, and temperature are summarized in Table 3.1.
 32

TABLE 3.1—RANGE OF DIFFERENT INDEPENDENT VARIABLES FOR 1,104

DATA POINTS
Variable Mean Minimum Maximum
C1 66.59 0 100
C2 4.98 0 100
C3 3.33 0 100
i-C4 0.50 0 63.60
nC4 0.38 0 5.82
i-C5 0.004 0 0.20
nC5 0.045 0 1.01
nC6 0.005 0 0.25
nC7 0.001 0 0.10
nC8 0.0006 0 0.05
CO2 18.66 0 100
N2 3.45 0 89.20
H 2S 1.07 0 100
C 2H 4 0.92 0 94.40
C 3H 6 0.04 0 7.60
NaCl 1.98 0 17.17
KCl 0.53 0 15
CaCl2 0.53 0 15.03
CH3OH 1.81 0 19.99
Ethylene Glycol 0.69 0 30
Triethylene Glycol 0.16 0 20.20
Glycerol 0.43 0 30
Temperature, ºF 46.33 10.29 89.33
Pressure, psi 1448.42 9.86 11240.42
Gas Specific
0.829 0.5531 1.52
Gravity (γ air = 1)

3.2 Data Observations

By screening the experimental data, I made some interesting observations. For example,
for systems without inhibitors and at the same temperature, a system with lighter gas
specific gravity usually forms hydrates at a higher pressure; this is consistent with the
 33

results in the literature. However, that is not always the case and some of the data do not
follow this pattern. By screening the experimental data and inspecting the compositions, I
learned that some components have significant effects on the hydrate-formation pressure
or temperature. For example, propane and isobutane will decrease the hydrate-formation
pressure dramatically; however, nitrogen will increase it and act like an inhibitor.

Table 3.2 shows how the binaries of methane with propane, isobutane, or nitrogen
decrease or increase the hydrate formation pressures at the same temperature for systems
without inhibitors.

TABLE 3.2—EFFECTS OF GAS COMPOSITIONS ON HYDRATE-FORMATION PRESSURE IN

SYSTEMS WITHOUT INHIBITORS
Gas Compounds Composition Hydrate Hydrate Gas Specific
(mol%) Formation Formation Gravity
Temperature (ºF) Pressure (psi)

C1 100 39.11 552.59 0.5531

C1 37.10
39.11 60.77 1.16172
C3 62.90

C1 27.20
39.11 1473.58 0.85359
N2 72.80

C1 100 39.29 565.65 0.5531

C1 71.4
39.29 51.63 0.9681
iC4 28.6

C1 50.25
39.29 889.08 0.75845
N2 49.75

C1 10.80
39.29 2300.30 0.92128
N2 89.20
 34

This table shows that at 39.11ºF, pure methane with specific gravity of 0.5531 forms
hydrate at a pressure of 552.59 psi. At the same temperature for a binary of methane and
nitrogen with higher specific gravity (0.85359), we expect hydrates to form at lower
pressure, but the pressure is acually1473.58 psi, which is higher than for the case of pure
methane. This indicates that the hydrate-formation pressure does not always correlate
negatively with specific gravity, and the presence of some components such as nitrogen
in a mixture increases the hydrate-formation pressure.

Another example shows two binary systems of 97.50 mol% C1 + 2.5 mol% i-C4 and 97.50
mol% C1 + 2.5 mol% nC4, which both have the same gas specific gravity of 0.589, but the
first binary will form hydrates at much lower pressures in a temperature range of 37.85 to
55.85ºF. Fig. 3.1 compares the hydrate-formation pressures for these systems. Although
these two binary systems have the same molecular weight, they behave differently
because the presence of some components such as isobutane in a mixture decreases the
hydrate-formation pressure. Therefore, two systems with equal gas specific gravities do
not necessarily form hydrates at equal pressures, but the presence of some components in
a mixture has a very significant effect on determining the hydrate-formation pressure or
temperature.

Experimental Hydrate Formation Data

Comparison Between Binary of CH4 with iC4 and nC4

1600

1200
Pressure, psi

800

400

0
30 35 40 45 50 55 60 65 70

Temperature, ºF

97.50 C1+2.5 iC4 (mol%) 97.52 C1+2.48 nC4 (mol%)

Fig. 3.1—Hydrate-formation pressure for binaries of CH4 with iC4 and nC4.2
 35

3.3 Comments on Data

1. Since Hammerschmidt discovered in 1934 that hydrates plug pipelines, a number
of experiments have been done by different researchers; and less data have been
measured in recent years than before. In addition, there are more data in the
literature for the systems in the presence of pure water than with inhibitors.
2. I observed compositions in some data sets that did not total 100 mol%. People
have used these data over and over to develop their predictive models or have
reported them in their books and papers without noting that. For these types of
data sets, I preferred to make up the deficiency of the compositions by adding to
the methane mol% to normalize the compositions. I did that because in a natural
gas methane usually has the highest fraction and adding one or two more mol% to
methane does not affect the predicted results for hydrate-formation conditions.
3. Substantial existing data are the hydrate-formation conditions for gas systems that
never or rarely exist in reality, such as pure hydrogen sulfide, pure propane, or
pure ethane.
4. The experimental data reported in the literature, either graphically or digitized,
have different units for temperature and pressure; to use these data I used digitizer
software to translate those reported in graphs to tables and then to have the same
units for all data I converted the different units to field units, ºF for temperature
and psi for pressure.

3.4 Regression Variables

The following equations show the calculation of the regression variables for developing
the proposed correlations.

3.4.1 Pseudoreduced Temperature and Pressure

As we learned from the observation of experimental data, the gas compositions play an
important role in determination of hydrate-formation pressure or temperature. By
calculating the pseudoreduced temperature and pressure, we can take into account the
effect of each component in the mixture. The pseudoreduced temperature and pressure
 36

are defined as temperature or pressure of a system divided by pseudocritical temperature

or pressure of the mixture of gas:

T pr = T / T pc ,................................................................................................................. (3.1)

and
p pr = p / p pc ,............................................................................................................... (3.2)

where T pr and p pr are the pseudoreduced temperature and pressure, and T pc and p pc are

the pseudocritical temperature and pressure of gas mixtures.

To calculate T pc and p pc , I used the mixing rules of Piper et al.51 as follows:

K2
T pc = ,..................................................................................................................... (3.3)
J
and
T pc
p pc = ,.................................................................................................................... (3.4)
J

where J and K are defined by:

3
Tc Tc
J = α0 + α i yi ( ) +α4 yj( ) j ,....................................................................... (3.5)
i =1 pc j pc
and
3
Tc Tc
K = β0 + β i yi ( ) + β4 yj( ) j ,................................................................ (3.6)
i =1 pc j pc

where yi is the mole fraction of nonhydrocarbon, yj is the mole fraction of hydrocarbon

components, α 0 − α 4 and β 0 − β 4 are constants as given in Table 3.3.
 37

TABLE 3.3—VALUES OF CONSTANTS α AND β FOR

CALCULATING J AND K51
i αi βi
0 5.2073 × 10 −2 − 3.9741 × 10 −1
1 1.0160 × 10 0 1.0503 × 10 0
2 8.6961 × 10 −1 9.6592 × 10 −1
3 7.2646 × 10 −1 7.8569 × 10 −1
4 8.5101 × 10 −1 9.8211 × 10 −1

The range of data for the mixing rules is summarized in Table 3.4.

TABLE 3.4—RANGE OF DATA FOR DEVELOPING THE MIXING RULES51

Variable Mean Minimum Maximum
H 2S 2.45 0.00 51.37
CO2 3.38 0.00 67.16
N2 1.87 0.00 15.68
C1 71.15 19.37 94.73
C2 8.21 2.30 18.40
C3 4.04 0.06 12.74
iC4 0.90 0.00 2.60
C4 1.55 0.00 6.04
iC5 0.64 0.00 2.24
C5 0.88 0.00 3.92
C6 0.65 0.00 4.78
Temperature, ºF 243.8 78 326
Pressure, psia 3758.6 514 12814
Gas Specific Gravity
0.972 0.613 1.821
(γ air = 1)
 38

3.4.2 Gas Specific Gravity

The specific gravity of the gas mixture is expressed as the molecular weight of the gas
mixture divided by the molecular weight of air and is given by:

y i MWi
γ = i
,............................................................................................................ (3.7)
MWair

where MWi and y i are the molecular weight and mole fraction of component i in the

mixture, and γ is the specific gravity of the mixture of gas.

3.4.3 Water Vapor Pressure

The vapor pressure measures the ability of molecules to escape from the surface of a
solid or liquid. A common equation to estimate the vapor pressure of a component (in this
case, water) is given by52:

log10 (p vw ) = a wv + bwv / T + c wv log10 T + d wv T + e wv T 2 ,...................................................... (3.8)

where T is the temperature of the system in ºK, p wv is the vapor pressure of water in

mmHg (must be converted to psi when used in Eq. 3.10) and a wv , bwv , c wv , d wv , and e wv are
constants which for water at these units specified as follows:

a wv = 29.8605

bwv = −3.1522 × 10 3

c wv = −7.3037

d wv = 2.4247 × 10 −9

ewv = 1.8090 × 10 −6

This equation is valid for the range of temperature from 273.16 to 647.13ºK.
 39

3.4.4 Liquid Water Viscosity

The viscosity measures the resistance of a substance to flow. Viscosity is affected by both
temperature and pressure; for example, it decreases as temperature increases or pressure
decreases. The following equation calculates the viscosity of liquid (in this case, water)52:

log10 ( µ wl ) = a wl + bwl / T + c wl T + d wl T 2 ,........................................................................ (3.9)

where T is the temperature of the system in Kelvin, ºK, µ wl is the viscosity of liquid

water in centipoise, cp and a wl , bwl , c wl , and d wl are constants which for water at these
units are as follows:
a wl = −10.2158

bwl = 1.7925 × 10 3

c wl = 1.7730 × 10 −2

d wl = −1.2631 × 10 −5
This equation is valid for the range of temperature from 273 to 643ºK.

3.5 Hydrate-Formation Pressure Correlation

I applied a regression model in SAS19 software to find the best relationships among the
above mentioned regression variables. Eq. 3.10 is the result of this regression, the p-
correlation, which predicts the hydrate-formation pressure at a given temperature:

xi xi
ln p pr = a 0 + a1 ln T pr + a 2 [( )T pr ] / γ 2 + a 3 ( )T pr
i Mi i Mi
xj 1 2 xj 1
+ a 4 [( )( xCO 2 + x H 2 S + x N 2 ]( ) + a5 ( )( ) + a 6 [(µ wl ) 4 (T pr ) 2 ( p wv )]
j Mj T pr j Mj T pr
1
+ a 7 [( p wv )( )] + a8 (T pr ) 2 + a 9 [(100 − xi )(T pr ) 2 ] + a10 [(ln γ )(T pr )]
T pr i

+ a11 [(ln γ )(ln T pr )( p )] + a12 [(ln γ ) (T pr ) 2 ] + a13 [(ln γ )(T pr )( p wv ) 2 ]

v
w
2

1
+ a14 [( xC 3 + xiC 4 )(T pr ) 6 ] + a15 [( xC 3 + xiC 4 )( p wv )(T pr ) 6 ] + a16 [( xCO 2 + x H 2 S )(γ )( )]
T pr
1 2
+ a17 [( x N 2 )(γ )( ) ] ,...................................................................................... (3.10)
T pr
 40

where Tpr and ppr are the pseudoreduced temperature and pressure, γ is the specific

gravity of the gas. The variable xi indicates concentration of electrolytes such as sodium
chloride (NaCl), potassium chloride (KCl), or calcium chloride (CaCl2), and the
variable x j indicates concentration of thermodynamic inhibitors such as methanol,

ethylene glycol, triethylene glycol, or glycerol, and both are expressed in weight percent.
The variables p wv and µ wl are the water vapor pressure and liquid water viscosity, and a0
to a17 are the coefficients of this correlation.

3.6 Hydrate-Formation Temperature Correlation

The following equation is the result of regression, the T-correlation, which predicts the
hydrate-formation temperature when a pressure is given:

xi xj
ln T pr = b0 + b1 (ln p ) 2 + b2 [( ) / γ 2 ] + b3 [( ) / γ 2 ] + b4 (γ 2 ) + b5 [(100 − xi )(γ 3 )]
i Mi j Mj i

xj
+ b6 ( xCO 2 + x H 2 S + x N 2 ) + b7 [( )( xCO 2 + x H 2 S + x N 2 )] + b8 [( xC 3 + xiC 4 ) / γ 6 ]
j Mj
+ b9 [(ln γ )(ln p )] + b10 [(ln γ )(ln p ) 4 ] + b11 [(ln p ) / γ ] + b12 [(ln p ) 2 / γ ] + b13 [(ln p ) 3 / γ ]

+ b14 [(ln p ) 4 / γ ] + b15 (ln p pr ) + b16 (ln p pr ) 2 ,........................................................ (3.11)

where constants b0 to b16 are the coefficients of this correlation. The values of coefficients
a 0 to a17 and b0 to b16 are given in Table 3.5.
 41

TABLE 3.5—VALUES OF CONSTANTS FOR HYDRATE-FORMATION p AND T

CORRELATIONS
i ai Standard Error bi Standard Error
0 − 2.924729 × 10 0 8.031 × 10 −2 3.1113797464 × 10 0 2.319 × 10 −2

1 7.069408 × 10 0 3.424 × 10 −1 − 6.121811 × 10 −2 5.4841 × 10 −4

2 − 6.71674 × 10 −1 8.15 × 10 −2 − 3.4581592 × 10 −2 1.3 × 10 −3

3 2.158912 × 10 0 3.0819 × 10 −1 − 2.2257841 × 10 −2 1.06 × 10 −3

4 − 1.4446 × 10 −2 1.6 × 10 −3 − 1.61387206 × 10 −1 9.5 × 10 −3

5 3.367516 × 10 0 9.032 × 10 −2 4.644864 × 10 −4 3.249 × 10 −5

6 − 1.68816 × 10 −1 3.947 × 10 −2 6.0870675 × 10 −3 2.101 × 10 −5

7 1.3213962 × 101 3.0521 × 10 −1 − 4.9726 × 10 −4 4.62 × 10 −5

8 2.365031 × 10 0 3.4994 × 10 −1 1.682281 × 10 −4 1.282 × 10 −5

9 − 2.5796 × 10 −2 3.41 × 10 −3 − 1.93610096 × 10 −1 5.68 × 10 −3

10 2.461102 × 10 0 2.3531 × 10 −1 1.963793 × 10 −4 8.61 × 10 −6

11 − 7.100059 × 10 0 1.50553 × 10 0 1.324677497 × 10 −1 1.1 × 10 −2

12 1.820312 × 10 0 1.6222 × 10 −1 − 7.8512137 × 10 −2 4.03 × 10 −3

13 7.517561 × 10 0 6.8072 × 10 −1 9.232805 × 10 −3 4.9397 × 10 −4

14 − 1.8793 × 10 −2 9.1908 × 10 −4 − 2.32276 × 10 −4 2.148 × 10 −5

15 1.9029 × 10 −2 2.78 × 10 −3 8.054836679 × 10 −1 3.98 × 10 −3

16 − 5.307 × 10 −3 8.8911 × 10 −4 6.3403148 × 10 −3 1.04 × 10 −3

17 − 3.2564 × 10 −2 5.44 × 10 −3
 42

CHAPTER IV
RESULTS AND DISCUSSION

Using the statistical analysis software (SAS),19 I applied a regression model for 1,104
experimental data points to find the best correlations among the variables. The data points
include different samples from pure hydrocarbon and nonhydrocarbon components to gas
mixtures. As can be seen from Table 3.1, besides temperature and pressure, 15 gas
components, 3 electrolytes, and 4 thermodynamic inhibitors have entered the regression
as independent variables. Section 3.3 included the calculations of the variables included
in this regression. To check the accuracy of the correlations and compare the predicted
results with the experimental data, I applied a statistical error analysis for both
correlations.

4.1 Predicted Results Versus Experimental

Figs. 4.1 and 4.2 show the calculated results versus experimental for 1,104 data points.
Using the following equation, the average absolute percentage error on pressure (paae)
measures the statistical error for the p-correlation of 15.93 with the R2 equal of 0.968:

n
100
p aae = p (cal.)i − p (exp .)i / p (exp .) ,.............................................................. (4.1)
n i

The average of absolute temperature difference (Taad) measures the statistical error from
the following equation for the T-correlation of 2.97ºF with the R2 equal of 0.999:

n
Taad = T (cal.)i − T (exp .)i / n ,................................................................................... (4.2)
i

where n is the total number of the data points that have been used in developing the two
correlations.
 43

Predicted vs. Experimental

100000

10000
Pred., Psi

1000

100

10
10 100 1000 10000 100000

Pexp. , psi

Fig. 4.1—Comparison of experimental and calculated values of hydrate-formation

pressure (number of data points: 1,104, R2= 0.968, paae = 15.93).

Experimental vs. Predicted

100
Tpred. , Psi

10
10 100

Texp., ºF

Fig. 4.2—Comparison of experimental and calculated values of hydrate-formation

temperature (number of data points: 1,104, R2= 0.999, Taad = 2.97).
 44

Figs. 4.3 through 4.5 show the calculated hydrate-formation pressure from pressure
correlation versus the experimental. Methane has been chosen for this comparison since it
is a key component of any natural gas mixtures, and hydrates of methane are the most
commonly found hydrates. Fig. 4.3 shows an excellent agreement between the calculated
hydrate-formation pressure and experimental data except a little deviation for the system
of gas and electrolytes, which occurs only at high pressures. Figs. 4.4 and 4.5 compare
the calculated and experimental hydrate-formation pressure for two natural gas mixtures
with the same components. In Fig. 4.4 where the gas mixture contains small
concentrations of propane and nitrogen, the predicted results represent the experimental
perfectly; however, Fig 4.5 shows slight deviation for predicted results from experimental
data when the mixture contains higher fractions of these two components.

12000

10000
Pressure, psi

8000

6000

4000

2000

0
30 40 50 60 70 80 90

Temperature, ºF

Without Inhibitor 6.21w t% NaCl 6.21w t% NaCl+10w t% CH3OH

Fig. 4.3—Comparison of experimental and calculated results of hydrate-formation

pressure from p-correlation for pure methane.
 45

C1=87.9, CO2=0.2, N2=1.1, C2=4.4, C3=4.9, nC4=1.5 (mol%)

(Without Inhibitor)
500

Pressure, psi 400

300

200

100

0
30 35 40 45 50 55 60

Temperature, ºF

Fig. 4.4—Comparison of experimental and calculated results of hydrate-formation

pressure from p-correlation for a natural gas with low concentration of propane and
nitrogen.

C1=65.4, CO2=0.2, N2=7.7, C2=12.7, C3=10.3, nC4=3.7 (mol%)

(Without Inhibitor)
1400

1200

1000
Pressure, psi

800

600

400

200

0
30 40 50 60 70 80

Temperature, ºF

Fig. 4.5—Comparison of experimental and calculated results of hydrate-formation

pressure from p-correlation for a natural gas with high concentration of propane
and nitrogen.
 46

Figs. 4.6 through 4.8 show the calculated hydrate-formation temperature from
temperature correlation versus the experimental. Fig. 4.6 shows the calculated results are
in excellent agreement with experimental data except slight deviation at low temperatures
for the complex system of gas and mixed inhibitors.

Figs. 4.7 and 4.8 compare the calculated and experimental hydrate-formation temperature
for the same natural gas systems. In Fig. 4.7 where the gas mixture contains small
concentrations of propane and nitrogen, the predicted results represent the experimental
perfectly; however, Fig 4.8 shows deviation for predicted results from experimental when
the mixture contains higher fractions of these two components. Despite of the deviation
for this case, the calculated values still follow the same trend as the experimental.

90

80
Temperature, ºF

70

60

50

40

30
0 2000 4000 6000 8000 10000 12000

Pressure, psi

Without Inhibitor 6.21w t% NaCl 6.21w t% NaCl+10w t% CH3OH)

Fig. 4.6—Comparison of experimental and calculated results of hydrate-formation

temperature from T-correlation for pure methane.
 47

C1=87.9, CO2=0.2, N2=1.1, C2=4.4, C3=4.9, nC4=1.5 (mol%)

(Without Inhibitor)
60

Temperature, ºF 55

50

45

40

35

30
0 100 200 300 400 500

Pressure, psi

Fig. 4.7—Comparison of experimental and calculated results of hydrate-formation

temperature from T-correlation for a natural gas with low concentration of propane
and nitrogen.

C1=65.4, CO2=0.2, N2=7.7, C2=12.7, C3=10.3, nC4=3.7 (mol%)

(Without Inhibitor)

90

80
Temperature, ºF

70

60

50

40

30
0 200 400 600 800 1000 1200 1400

Pressure, psi

Fig. 4.8—Comparison of experimental and calculated results of hydrate-formation

temperature from T-correlation for a natural gas with high concentration of
propane and nitrogen.
 48

4.2 Comparison of Predicted Results with a Common Correlation

As I mentioned earlier in Section 2.7.3, Kobayashi et al.43 developed Eq. 2.3 from the
gas-gravity plot from Katz. To compare the accuracy of the T-correlation with that of
Kobayashi et al. equation, considering all the limitations for this equation (data included
only hydrocarbons; data without inhibitors; data at pressures lower than 1,500 psia,
specific gravities lower than 0.9, and temperatures between 0.34 and 62ºF), I calculated
the hydrate-formation temperature from both methods. For 173 data points from my
collection, the results showed that my equation is superior to the Kobayashi et al.
equation with an average absolute temperature difference of 2.87ºF versus 13.02ºF. Fig.
4.9 shows the actual difference between predicted and experimental temperatures for my
T-correlation and the Kobayashi et al. correlation.

Actual Temperature Difference from Experimental

40
30
20
Texp. -Tpred., ºF

10
0
-10 32 42 52 62

-20
-30
-40
-50

Texp., ºF

Predicted Kobayashi et al.

Fig. 4.9—Actual differences between predicted and experimental temperatures for

T-correlation and Kobayashi et al. correlation.

4.3 Comparison of Predicted Results with Calculated from PVTsim

The results of the two improved correlations make them competitive even with the
commercial software. To prove this claim, I have compared the results of the improved
 49

correlations with the results from the available commercial software, PVTsim20. For most
sets of data, the results from the new correlations are as good as the results from PVTsim;
however, for some of the data sets, these correlations predict the hydrate-formation
conditions even better than PVTsim. Figs. 4.10 and 4.11 compare the predicted hydrate-
formation pressure and temperature from the correlations and PVTsim for two different
natural gas systems.

C1=78.4, CO2=0.2, N2=9.4, C2=6, C3=3.6, nC4=2.4 (mol%)

(Without Inhibitor)
350

300
Pressure, psi

250

200

150

100
30 40 50 60

Temperature, ºF
Experimental Predicted PVTsim

Fig. 4.10—Comparison of the calculated hydrate-formation pressure from PVTsim

and p-correlation.
 50

C1=88.5, N2=3.4, C2=4.3, C3=2, nC4=1.7 (mol%)

(Without Inhibitor)

320

270
Pressure, psi

220

170

120

70

20
30 32 34 36 38 40 42 44 46

Temperature, ºF
Experimental Predicted PVTsim

Fig. 4.11—Comparison of the calculated hydrate-formation temperature from

PVTsim and T-correlation.

4.4 Sensitivity Analysis

The best way to determine the hydrate formation conditions is to measure the pressure
and temperature experimentally. Since this is expensive and time consuming, the
available correlations or thermodynamics methods can predict these conditions. When
using a thermodynamics model, if the predicted results have significant deviations from
the experimental data, it is possible to improve the model by adjusting the parameters
entering into the model. Recall that one of the objectives of this work is to provide
guidelines to calibrate a thermodynamics model by applying a sensitivity analysis to
selective parameters entering into the model. To conduct the sensitivity analysis, we
determined two possible parameters, binary interaction coefficients ( K ij ) and the

Langmuir adsorption constant ( C ji ) in a thermodynamics model so that adjusting those

parameters will cause the model to behave differently.

 51

To predict the hydrate-formation conditions from a thermodynamics model, the chemical

potential difference of water in the empty and in the filled hydrate lattice ( ∆µ H ) requires
calculation of fugacity for each component in the gas phase (Eq. 2.10). The fugacity of
component i ( f i ) is defined by:

f i = φ i y i p ,.................................................................................................................. (4.3)

where φ i = fugacity coefficient, yi = mole fraction of component i in the gas phase, and

p = pressure of the system. The fugacity coefficients φ i can be determined from an

EOS. For a given component the difference of its fugacity in one phase with respect to
another phase is a value that measures the transfer potential of that component between
the phases. Therefore, at equilibrium conditions when the fugacity of a component in the
two or more phases is equal, there is no mass transformation across the phases; pressure,
temperature, and the compositions within the phases remain constant.

The following equation is a thermodynamic relationship that determines the fugacity

coefficient of component i in the mixture:

V ∂p RT
ln φ i = −1 / RT (( ) T ,V ,n j − )dV − ln Z ,............................................................. (4.4)
∞ ∂ni V

where ni is the number of moles of type i .

In case of the SRK-equation29:

b V
ln φi = ( i )( Z − 1) − ln Z + ln
b (V − b )
Z j (1 − K ij )(a i a j ) (V + b )
0.5
a bi
+ −2 ln ............................................. (4.5)
bRT b j a V
 52

PVTsim20 has different options to calculate the binary interaction coefficient K ij ;

however, this software supports the following temperature-dependent binary

interaction K ij :

K ij = K ij A + cn j [ K ij B + K ij C (T − T0 )] ,..................................................................... (4.6)

where K ij A , K ij B , and K ij C are user input; T0 is a reference temperature of 288.15ºK;

and cn j is the carbon number of component j . The attempt to adjust the variables K ij A ,

K ij B , and K ij C failed because the software does not accept the changes.

PVTsim considers three types of hydrate lattices, SI, SII and SH. Hydrates with structures
I and II consist of two different sizes of cavities, small and large. Structure H consists of
three different sizes of cavities which in PVTsim are modeled as two cavity sizes,
small/medium and huge. As we saw in Section 2.8, the Langmuir adsorption constant C ji

is a temperature-dependent parameter and the values of A and B (Eq. 2.11) are unique
for each component j that is capable of entering into a cavity of type i . For example, the

value of A for component C1 that enters a large cavity of SII is equal to 1.335 × 10 −1
ºF/psia, and with this value we saw the prediction from PVTsim in Fig. 4.10 that deviated
from experimental. Multiplying the value of A by 10, PVTsim shows better results, as
can be seen in Fig. 4.12. Similarly, the value of A for component C2 that enters a large
cavity of SI is equal to 3.722 × 10 −4 ºF/psia, and by making this value 10 times larger,
PVTsim shows better results, as we see in Fig. 4.13.
 53

C1=88.5, N2=3.4, C2=4.3, C3=2, nC4=1.7 (mol%)

(Without Inhibitor)

320
Pressure, psi 270

220

170

120

70

20
30 35 40 45 50

Temperature, ºF
Experimental Predicted PVTsim pvtsim(AIIL,C1*10)

Fig. 4.12—Calculated results from PVTsim before and after adjusting the value of
A for component C1 in a large cavity of Structure II.

C1=88.5, N2=3.4, C2=4.3, C3=2, nC4=1.7 (mol%)

(Without Inhibitor)

320

270
Pressure, psi

220

170

120

70

20
30 35 40 45 50
Temperature, ºF

Experimental Predicted PVTsim pvtsim(AIL,C2*10)

Fig. 4.13—Calculated results from PVTsim before and after adjusting the value of
A for component C2 in a large cavity of Structure I.
 54

Therefore, the predicted results from PVTsim are sensitive to the A parameter for some
components in large cavities of SI and SII. Although the hydrate types and the sizes of
cavities do not affect our correlation, from the sensitivity analysis for PVTsim, we can
reach the conclusion that our correlations could be sensitive to some or all of the
variables involved in their development. This means that we can probably improve the
new correlations by adjusting the current variables, by entering new variables such as
density of water, or even by using improved mixing rules with a wider range of
hydrocarbon, temperatures, and pressures. The mixing rule that I used to calculate the
pseudoreduced temperature and pressure is limited to some ranges of composition,
temperature, and pressure which do not cover all the ranges for the experimental data;
that could affect the accuracy of the new correlations to some extent. By having more
experimental data and considering other regression variables, it is possible to improve
these correlations more and more.
 55

CHAPTER V
CONCLUSIONS

Gas hydrates are a costly problem when they plug oil and gas pipelines. The best way to
determine the hydrate-formation temperature and pressure is to measure these conditions
experimentally for every gas system. Since this is not practical in terms of time and
money, correlations are the other alternative tool. Only a couple of the thermodynamics
methods in the literature are applicable for systems including inhibitors. In this work, we
introduced two improved correlations that calculate the hydrate-formation pressure or
temperature for single gases or mixtures of gases with or without inhibitors. These
correlations are based on over 1,100 published data points of gas-hydrate formation
temperatures and pressures with and without inhibitors. The data include samples ranging
from pure-hydrate formers such as methane, ethane, propane, carbon dioxide, and
hydrogen sulfide to binary, ternary, and natural gases. Using the Statistical Analysis
Software (SAS),19 we found the best correlations among the variables including gas
specific gravity, pseudoreduced pressure and temperature of gas mixtures, vapor pressure
and liquid viscosity of water, and concentrations of electrolytes and thermodynamic
inhibitors.

These correlations are applicable to temperatures up to 90ºF and pressures up to 12,000

psi and they are capable of handling aqueous solutions containing electrolytes such as
sodium, potassium, and calcium chlorides lower than 20 wt% and inhibitors such as
methanol lower than 20 wt%, ethylene glycol, triethylene glycol, and glycerol lower than
40 wt%. The results show an average absolute percentage deviation of 15.93 in pressure
and an average absolute temperature difference of 2.97ºF.

The improved correlations are simple and portable since they are applicable even with a
simple calculator. The results are in excellent agreement with the experimental data in
most cases and even better than the results from commercial simulators in some cases.
These correlations provide guidelines to help the users to forecast gas-hydrate forming
conditions for most systems of hydrate formers with and without inhibitors.
 56

My conclusions of this research come in two parts, conclusions from my observations of

experimental data which are consistent with what I read in the literature and conclusions
from my work in developing the improved correlations.

5.1 Conclusions from Observations

• In absence of inhibitors and at the same temperature, a gas with lighter specific
gravity forms hydrates at higher pressure, but we should consider the presence of
components such as propane, isobutane, and nitrogen. The presence of propane
and isobutane in a gas mixture decreases the hydrate-formation pressure and
increases the hydrate-formation temperature, while the presence of nitrogen in a
gas mixture increases the hydrate-formation pressure and decreases the hydrate-
formation temperature.
• Compositions of a gas system play a very important role in determining the
hydrate-formation temperature or pressure. That means two gas systems with
equal specific gravity may form hydrates at very different conditions. For
instance, a binary mixture of methane and isobutane forms hydrates at lower
pressure and higher temperature than a mixture of methane and butane with the
same composition and specific gravity.
• Sodium chloride has a higher inhibition effect than methanol at the same
concentration; this is very obvious at higher pressures. In the presence of mixed
inhibitors, the inhibitor with higher concentration of sodium chloride is more
effective than the one with higher concentration of methanol.
• The inhibition effect of ethylene glycol is inferior to that of methanol at the same
concentration.

5.2 Conclusions from Developing the Improved Correlations

• The improved correlations estimate the hydrate-formation temperature or pressure
for a variety of gas-hydrate formers in the presence or absence of inhibitors.
• These correlations are easy to use and they are applicable even with a simple
calculator.
 57

• The predicted results are in a good agreement with the experimental data in most
cases.
• These correlations are very accurate and in some cases they can predict the
hydrate-formation conditions even better than the commercial software, PVTsim.
• The improved correlations are unique since none of the available correlations in
the literature can predict the hydrate-formation conditions for complex systems
including inhibitors; in addition, the new correlations proved to be much more
accurate than the common correlations.
• The results from 1,104 data points show an average absolute percentage deviation
of 15.93 in pressure for the p-correlation and an average absolute temperature
difference of 2.97ºF for the T-correlation.
• The correlations are useful for a wide range of temperature (to 90ºF) and pressure
(to 12,000 psi).
• A sensitivity analysis on parameter A of the Langmuir adsorption constant
showed that the value of this parameter for hydrocarbons entering in large cavities
of Structures I and II has significant effect on the calculated hydrate-formation
temperature and pressure by PVTsim simulator.
 58

NOMENCLATURE

aw = water activity

a.a.T .d . = average of absolute temperature difference

a.a. p.e. = average of absolute percentage error on pressure
C ji = Langmuir adsorption constant

Cp = molar heat capacity

fi = fugacity of gas component in a gas mixture

H= molar enthalpy
k ij = binary interaction coefficient

K vs = vapor/solid equilibrium ratio

M air = molecular weight of air

Mi = molecular weight of electrolyte i

Mj= molecular weight of thermodynamic inhibitor j

MWi = molecular weight of component i in a gas mixture

n= number of data points

ni = number of cavities of type i per water molecules

p= pressure of the system

pc = critical pressure

p pc = pseudocritical pressure

p pr = pseudoreduced pressure

p rv = reduced vapor pressure

p wv = vapor pressure of water

R= universal gas constant
si = mol fraction of component i in solid phase
T= temperature of the system
Tc = critical temperature
 59

T pc = pseudocritical temperature

T pr = pseudoreduced temperature

Tr = reduced temperature

T0 = reference temperature

V= molar volume
yi = mol fraction of component i in vapor phase

xi = concentration of electrolyte i

xj = concentration of thermodynamic inhibitor j

∆= difference in properties
φ= fugacity coefficient
γ= gas specific gravity
µ= chemical potential

µ wl = viscosity of liquid water

ω= acentric factor
 60

REFERENCES

1. Hammerschmidt, E.G.: “Formation of Gas Hydrates in Natural Gas Transmission

Lines,” Industrial & Engineering Chemistry (1934) 26, 851.
2. Sloan, E.D. Jr.: Clathrate Hydrates of Natural Gases, Marcel Dekker, New York
(1990).
3. Nasrifar, K. and Moshfeghian, M.: “A Model for Prediction of Gas Hydrate
Formation Conditions in Aqueous Solutions Containing Electrolytes and/or
Alcohol,” J. Chem. Thermodynamics (2001) 33, 999.
4. Javanmardi, J. and Moshfeghian, M.: “A New Approach for Prediction of Gas
Hydrate Formation Conditions in Aqueous Electrolyte Solutions,” Fluid Phase
Equilibria (2000) 168, 135.
5. Bishnoi, P.R. and Dholabhai, P.D.: “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
Equilibria (1999) 158-160, 821.
6. Dholabhai, P.D., Parent, J.S., and Bishnoi, P.R.: “Equilibrium Conditions for
Hydrate Formation from Binary Mixture of Methane and Carbon Dioxide in the
Presence of Electrolytes, Methanol and Ethylene Glycol,” Fluid Phase Equilibria
(1997) 141, 235.
7. Nasrifar, K. and Moshfeghian, M.: “Computation of Equilibrium Hydrate
Formation Temperature for CO2 and Hydrocarbon Gases Containing CO2 in the
Presence of an Alcohol, Electrolytes and Their Mixtures,” J. of Petroleum Science
and Engineering (2000) 26, 143.
8. Nasrifar, K., Moshfeghian, M., and Maddox, R.N.: “Prediction of Equilibrium
Conditions for Gas Hydrate Formation in the Mixtures of Both Electrolytes and
Alcohol,” Fluid Phase Equilibria (1998) 146, 1.
9. Dimitrios, A. and Varotsis, N.: “Modeling Gas Hydrate Thermodynamic
Behavior: Theoretical Basis and Computational Methods,” Fluid Phase Equilibria
(1996) 123, 107.
 61

10. Du, Y. and Guo, T.: “Prediction of Hydrate Formation for Systems Containing
Methanol,” Chemical Engineering Science (1990) 45, 893.
11. Ng, H.J. and Robinson, D.B.: “Hydrate Formation in Systems Containing
Methane, Ethane, Propane, Carbon Dioxide or Hydrogen Sulfide in the Presence
of Methanol,” Fluid Phase Equilibria (1985) 21, 145.
12. Jager, M.D., Peters, C.J., and Sloan, E.D.: “Experimental Determination of
Methane Hydrate Stability in Methanol and Electrolyte Solutions,” Fluid Phase
Equilibria (2002) 193, 17.
13. Jager, M.D. and Sloan, E.D.: “The Effect of Pressure on Methane Hydration in
Pure Water and Sodium Chloride Solutions,” Fluid Phase Equilibria (2001) 185,
89.
14. Jossang, A. and Stange, E.: “A New Predictive Activity Model for Aqueous Salt
Solutions,” Fluid Phase Equilibria (2001) 181, 33.
15. Ma, Q.L., Chen, G.J., and Guo, T.M.: “Modeling the Gas Hydrate Formation of
Inhibitor Containing Systems,” Fluid Phase Equilibria (2003) 205, 291.
16. Ma, C.F., Chen, G.J., Wang, F., Sun, C.Y., and Guo, T.M.: “Hydrate Formation
of (CH4+C2H4) and (CH4+C3H6) Gas Mixtures,” Fluid Phase Equilibria (2001)
191, 41.
17. Chen, G.J. and Guo, T.M.: “Thermodynamic Modeling of Hydrate Formation
Based on New Concepts,” Fluid Phase Equilibria (1996) 122, 43.
18. Fan, S.S., Chen, G.J., Ma, Q.L., and Guo, T.M.: “Experimental and Modeling
Studies on the Hydrate Formation of CO2 and CO2-Rich Gas Mixtures,” Chemical
Engineering J. (2000) 78, 173.
19. SAS software, Version 8.02, SAS Institute Inc., Cary, North Carolina, (2001).
20. PVTsim software, Version 14, Calsep, Houston, Texas, (2004).
21. Tohidi, B., Danesh, A., Burgass, R.W., and Todd, A.C.: “Effect of Heavy Hydrate
Formation on the Hydrate Free Zone of Real Reservoir Fluids,” paper SPE 35568
presented at the 1996 European Production Operations Conference and
Exhibition, Stavanger, Norway, 16-17 April.
22. von Stackelberg, M. and Muller, H.R.: “On the Structure of Gas Hydrates,” J. of
Chemical Physics (1951) 19, 1319.
 62

23. Claussen, W.F.: “Suggested Structures of Water in Inert Gas Hydrates,” J. of

Chemical Physics (1951) 19, 259.
24. Claussen, W.F.: “A Second Water Structure for Inert Gas Hydrates,” J. of
Chemical Physics (1951) 19, 1425.
25. Heriot Watt University, Institute of Petroleum Engineering, “What Are Gas
Hydrates?” www.pet.hw.ac.uk/hydrate/hydrates_what_html, (2005).
26. McMullan, R.K. and Jeffrey, G.A.: “Hydrates of the Tetra n-Butyl and Tetra i-
Amyl Quaternary Ammonium Salts,” J. of Chemical Physics (1959) 31, 1231.
27. McMullan, R.K. and Jeffrey, G.A.: “Polyhedral Clathrate Hydrates. IX. Structure
of Ethylene Oxide Hydrate,” J. of Chemical Physics (1965) 42, 2725.
28. Mak, T.C.W. and McMullan, R.K.: “Polyhedral Clathrate Hydrates. X. Structure
of the Double Hydrate of Tetrahydrofuran and Hydrogen Sulfide,” J. of Chemical
Physics (1965) 42, 2732.
29. Pederson, K.S., Fredenslund, A.A, and Thomassen, P.: Properties of Oils and
Natural Gases, Gulf Publishing Co., Houston, Texas, (1989).
30. Ripmeester, J.A., Tse, J.S., Ratcliffe, C.I., and Powell, B.M.: “A New Clathrate
Hydrate Structure,” Nature (1987) 325, 135.
31. Tohidi, B., Danesh, A., Burgass, R.W., and Todd, A.C.: “Equilibrium Data and
Thermodynamic Modeling of Cyclohexane Gas Hydrates,” Chemical Engineering
Science (1996) 51, 159.
32. Tohidi, B., Danesh, A., Todd, A.C., Burgass, R.W., and Ostergaard, K.K.:
“Equilibrium Data and Thermodynamic Modeling of Cyclopentane and
Neopentane Hydrates,” Fluid Phase Equilibria (1997) 138, 241.
33. Becke, P., Kessel, D., and Rahimian, I.: “Influence of Liquid Hydrocarbons on
Gas Hydrate Equilibrium,” paper SPE 25032 presented at the 1992 European
Petroleum Conference, Cannes, France, 16-18 November.
34. Ostergaard, K.K., Tohidi, B., Danesh, A., Burgass, R.W., and Todd, A.C.:
“Equilibrium Data and Thermodynamic Modeling of Isopentane and 2, 2-
Dimethylpentane Hydrates,” Fluid Phase Equilibria (2000) 169, 101.
 63

35. Mehta, A.P. and Sloan, E.D.: “Structure H Hydrates: Implications for the
Petroleum Industry,” paper SPE 36742 presented at the 1996 Annual Technical
Conference and Exhibition Denver, Colorado, 6-9 October.
36. McCain, W.D. Jr.: The Properties of Petroleum Fluids, PennWell Publishing Co.,
Tulsa, Oklahoma (1990).
37. Paez, J.E., Blok, R, Vaziri, H., and Islam, M.R.: “Problems in Gas Hydrates:
Practical Guidelines for Field Remediation,” paper SPE 69424 presented at the
2001 Latin American and Caribbean Petroleum Engineering Conference, Buenos
Aires, Argentina, 25-28 March.
38. Ng, H.J., Petrunia, J.P., and Robinson, D.B.: “Experimental Measurement and
Prediction of Hydrate Forming Conditions in Nitrogen-Propane-Water,” Fluid
Phase Equilibria (1977/1978) 1, 283.
39. Ng, H.J and Robinson, D.B.: “The Measurement and Prediction of Hydrate
Formation in Liquid Hydrocarbon-Water System,” Industrial & Engineering
Chemistry Fundamentals (1976) 15, 293.
40. Bishnoi, P.R. and Dholabhai, P.D.: “Experimental Study on Propane Hydrate
Equilibrium Conditions in Aqueous Electrolyte Solutions,” Fluid Phase
Equilibria (1993) 83, 455.
41. Katz, D.L., Cornell, D., Kobayashi, R., Poettmann, F.H., Vary, J.A. et al.:
“Water/Hydrocarbon Systems,” Handbook of Natural Gas Engineering, McGraw-
Hill, New York (1959).
42. Wilcox, W.I., Carson, D.B., and Katz, D.L.: “Natural Gas Hydrates,” Industrial &
Engineering Chemistry (1941) 33, 662.
43. Kobayashi, R., Song, K.Y., and Sloan, E.D.: “Phase Behavior of
Water/Hydrocarbon Systems,” Petroleum Engineering Handbook, H.B. Bradley
(ed.) SPE, Richardson, Dallas, Texas (1987).
44. Elgibaly, A.A. and Elkamel, A.M.: “A New Correlation for Predicting Hydrate
Formation Conditions for Various Gas Mixtures and Inhibitors,” Fluid Phase
Equilibria (1998) 152, 23.
 64

45. Parrish, W.R. and Prausnitz, J.M.: “Dissociation Pressures of Gas Hydrates
Formed by Gas Mixtures,” Industrial & Engineering Chemistry Process Design
and Development (1972) 11, 26.
46. Holder, G.D., Zetts, S.P., and Pradhan, N.: “Phase Behavior in Systems
Containing Clathrate Hydrates,” Review in Chemical Engineering (1988) 5, 1.
47. van der Waals, J.H. and Platteeuw, J.C.: “Clathrate Solutions,” Adv. Chem. Phys.
(1952) 2, 1.
48. Soave, G.: “Equilibrium Constants from a Modified Redlich-Kwong Equation of
State,” Chemical Engineering Science (1972) 27, 1197.
49. Pitzer, K.S.: “The Volumetric and Thermodynamic Properties of Fluids. I.
Theoretical Basis and Virial Coefficients,” J. of the American Chemical Society
(1955) 77, 3427.
50. Marisoft Digitizer, Version 3.3, Mark Mitchell, (1997). [Available at
http://home.earthlink.net/~mmc1919].
51. Piper, L.D., McCain, W.D., and Corredor, J.H.: “Compressibility Factors for
Naturally Occurring Petroleum Gases,” paper SPE 69424 presented at the 1993
Annual Technical Conference and Exhibition, Houston, 3-6 October.
52. Yaws, C.L.: Chemical Properties Handbook, McGraw-Hill, New York (1999).
 65

APPENDIX A

EXPERIMENTAL DATA

The entire experimental data gathered and used in this work is in a separate Excel file.
 66

APPENDIX B

HYDRATE-FORMATION PRESSURE CALCULATION

This Excel file includes a Visual Basic program that calculates the hydrate-formation
pressure at a given temperature. By giving the gas composition and the inhibitor
concentration as input data to this program, a user can calculate the hydrate-formation
pressure at a given temperature for that system. This calculation can be done very quick
and only by clicking on the Hydrate Pressure button in the Data sheet.
 67

APPENDIX C

HYDRATE-FORMATION TEMPERATURE CALCULATION

This Excel file includes a Visual Basic program that calculates the hydrate-formation
temperature at a given pressure. By giving the gas composition and the inhibitor
concentration as input data to this program, a user can calculate the hydrate-formation
temperature at a given pressure for that system. This calculation can be done very quick
and only by clicking on the Hydrate Temperature button in the Data sheet.
 68

VITA

Name: Sharareh Ameripour

Permanent Address: 22015 Velicata

Woodland Hills, CA 91364

Education: M.S., Petroleum Engineering, 2005

Texas A&M University
College Station, Texas

B.S., Chemical Engineering, 1987

Amirkabir University of Technology
Tehran, Iran

Work Experience: Senior Gas Process Engineer, 1988-1998

National Iranian Oil Company
Ahwaz, Iran