bitumen fractionation and multicomponent characterization
TRANSCRIPT
University of Calgary
PRISM: University of Calgary's Digital Repository
Graduate Studies The Vault: Electronic Theses and Dissertations
2018-01-11
Bitumen Fractionation and Multicomponent
Characterization of Solvent-Bitumen Systems
Azinfar, Bahareh
Azinfar, B. (2018). Bitumen Fractionation and Multicomponent Characterization of
Solvent-Bitumen Systems (Unpublished doctoral thesis). University of Calgary, Calgary, AB.
http://hdl.handle.net/1880/106327
doctoral thesis
University of Calgary graduate students retain copyright ownership and moral rights for their
thesis. You may use this material in any way that is permitted by the Copyright Act or through
licensing that has been assigned to the document. For uses that are not allowable under
copyright legislation or licensing, you are required to seek permission.
Downloaded from PRISM: https://prism.ucalgary.ca
UNIVERSITY OF CALGARY
Bitumen Fractionation and Multicomponent Characterization of Solvent-Bitumen Systems
by
Bahareh Azinfar
A THESIS
SUBMITTED TO THE FACULTY OF GRADUATE STUDIES
IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
GRADUATE PROGRAM IN CHEMICAL AND PETROLEUM ENGINEERING
CALGARY, ALBERTA
JANUARY, 2018
© Bahareh Azinfar 2018
ii
Abstract
The phase behaviour and thermo-physical properties of solvent-bitumen systems are critical to
design, optimize, and develop the solvent-based and solvent-aided recovery processes. This study
presents experimental and modeling results on phase behaviour of light hydrocarbon solvents
(methane, ethane, propane, and butane) and Athabasca bitumen fractions. In this work, a modified
vacuum distillation system is developed to fractionate bitumen into four cuts. Boiling point
distribution, density, viscosity, and molecular weight of each obtained cut are measured. Then,
solvent solubility in each distillable bitumen fractions, density, and viscosity of liquid phase are
measured using an in-house developed PVT apparatus at temperature and pressure ranges up to
190 oC and 6 MPa, respectively.
A generalized solubility model is proposed in which the measured data of solvent solubility in
bitumen cuts are used to develop the generalized binary interaction parameters between
hydrocarbon solvents and each bitumen components. In this model, bitumen components are
defined using the simulated distillation results and the binary interaction parameter correlation
found for each solvent are used to calculate the solvent solubility in bitumen sample.
Implementation of this model does not require measurements of solvent solubility in bitumen
sample for tuning. The only input to this model is the boiling point distribution of the bitumen
sample, which is obtained by simulated distillation test. Following the characterization of bitumen,
solvent solubility in bitumen can be calculated using the generalized correlation of binary
interaction parameter proposed in this study. The proposed generalized model showed promising
results in prediction of the solubility of light hydrocarbon solvents in bitumen without need for
extensive and costly solubility measurements.
Another focus of this study is development of a method for characterization of heavy and extra
heavy oils. The gel permeation chromatography (GPC) is combined with the simulated distillation
(ASTM D7169) results and used to provide the molecular weight and boiling point distributions
of very heavy and complex mixtures such as asphaltene and vacuum residue. The proposed model
is properly validated and used to characterize the heaviest bitumen cut obtained by vacuum
distillation.
iii
Acknowledgements
I would like to take advantages of this opportunity to express my appreciation to all the awesome
people who have provided enormous support and encouragement throughout this research effort.
I would like to express my deepest and sincere gratitude to my supervisors, Drs. Jalal Abedi and
Hassan Hassanzadeh for all the continuous support, patience, guidance, and inspiring instructions
in the development and completion of this study. I was really lucky to work under their
supervisions and in their equipped laboratories.
My sincere thanks also go to my supervisory committee members, Drs. Gordon Moore and
Sudarshan Mehta, for their support during my research. I also thank Drs. Robert Martinuzzi and
Hadi Nasrabadi to accept to be in my examination committee.
I also want to extend my gratitude to Dr. Mohsen Zirrahi as a big contribution to this work and
part of my awesome team. I would like to thank all my friends and colleagues at University of
Calgary and SHARP research group, especially Ali Haddadnia.
I wish to express my appreciation for the financial support of all member companies of the NSERC
IRC in Solvent Enhanced Recovery Processes: BP Canada Energy Group ULC, Brion Energy,
Cenovus, ConocoPhillips Canada, Devon Canada Co, Foundation CMG, Husky Energy, Imperial
Oil Limited, Japan Canada Oil Sands Limited, Nexen Energy ULC, Natural Sciences and
Engineering Research Council of Canada (NSERC), N-Solv, Statoil Canada Ltd., Suncor Energy
and Total E&P Canada. The support of the Department of Chemical and Petroleum Engineering
and the Schulich School of Engineering at the University of Calgary is also acknowledged.
A special gratitude and love goes to my parents for showing faith in me and giving me the liberty
to choose what I desired. Saying thank you wouldn’t be enough to express my deepest gratitude to
them. I would never forget the support from my brothers, Babak, Mazdak, and Siamak. They
encouraged me from the first stage of life and expressed confidence in my abilities. I am the
luckiest to have such a lovely and caring family.
Last but never least, I owe thanks to a very special person, my best friend, love of my life, and my
husband, Farzad, for his unbelievable support and understanding during this long path. Thank you
so much for always being there for me.
iv
To My Lovely Husband, Farzad
v
Table of Contents
Abstract ............................................................................................................................... ii Acknowledgements ............................................................................................................ iii
Table of Contents .................................................................................................................v List of Tables ................................................................................................................... viii List of Figures .................................................................................................................... xi List of Symbols, Abbreviations and Nomenclature ....................................................... xviii
CHAPTER ONE: OVERVIEW ...........................................................................................1
1.1 Motivations, Objectives, Steps, and Organization .....................................................1 1.2 Dissertation Outline ...................................................................................................2
CHAPTER TWO: PHASE BEHAVIOUR OF BUTANE-BITUMEN FRACTIONS ........5
2.1 Preface .......................................................................................................................5 2.2 Abstract ......................................................................................................................5 2.3 Introduction ................................................................................................................6
2.4 Materials and Experimental Methods ........................................................................8 2.4.1 Materials .....................................................................................................8
2.4.2 Fractionation Apparatus ..............................................................................8 2.4.3 Phase Behaviour Apparatus ......................................................................11
2.5 Results and Discussion ............................................................................................11
2.5.1 Bitumen Fractionation ..............................................................................11 2.5.2 Experimental Phase Behaviour Data ........................................................12
2.5.3 Modeling Results ......................................................................................16 2.6 Summary and Conclusion ........................................................................................30
Appendix 2.A: Equations Used to Develop the Solubility Model. ................................31 Appendix 2.B: Simulated Distillation Results of Athabasca Bitumen Cut and Whole
Bitumens. ...............................................................................................................34 2.7 References ................................................................................................................39
CHAPTER THREE: PHASE BEHAVIOUR OF PROPANE-BITUMEN FRACTIONS 42
3.1 Preface .....................................................................................................................42 3.2 Abstract ....................................................................................................................42 3.3 Introduction ..............................................................................................................43 3.4 Experimental Section ...............................................................................................47
3.4.1 Materials ...................................................................................................47 3.4.2 Bitumen fractionation ...............................................................................47 3.4.3 PVT tests ...................................................................................................51
3.5 Results and Discussion ............................................................................................52 3.5.1 Experimental Phase Behaviour Data ........................................................52 3.5.2 Solubility Model Description ....................................................................55
3.6 Summary and Conclusion ........................................................................................61
Appendix 3.A: Density of Propane/Bitumen Cuts.........................................................62 Appendix 3.B: Viscosity of Propane/Bitumen Cuts ......................................................65 3.7 References ................................................................................................................68
vi
CHAPTER FOUR: PHASE BEHAVIOUR OF METHANE- AND ETHANE-BITUMEN
FRACTIONS ............................................................................................................72 4.1 Preface .....................................................................................................................72 4.2 Abstract ....................................................................................................................72
4.3 Introduction ..............................................................................................................73 4.3.1 Why is study of solvent-based recovery processes important? ................73 4.3.2 Why is methane and ethane considered in bitumen recovery processes? .74 4.3.3 Why is developing the generalized model to predict k-values of
solvent/bitumen mixture necessary and what is the major contribution of the
present study? ..........................................................................................75 4.4 Experimental Section ...............................................................................................76
4.4.1 Materials ...................................................................................................76 4.4.2 Bitumen Fractionation ..............................................................................76
4.4.3 Phase Behaviour Data Measuring .............................................................77 4.5 Results and Discussion ............................................................................................77
4.5.1 Experimental Results ................................................................................77 4.5.2 Description of k-value Model and Results ...............................................81
4.6 Summary and Conclusion ........................................................................................92 4.7 References ................................................................................................................94
CHAPTER FIVE: EFFECT OF ASPHALTENE ON PHASE BEHAVIOUR AND
THERMO-PHYSICAL PROPERTIES OF SOLVENT-BITUMEN SYSTEMS ....96 5.1 Preface .....................................................................................................................96
5.2 Abstract ....................................................................................................................96 5.3 Introduction ..............................................................................................................97 5.4 Experimental Section ...............................................................................................99
5.4.1 Materials ...................................................................................................99
5.4.2 Deasphalting the Bitumen .........................................................................99 5.4.3 PVT Apparatus .......................................................................................102 5.4.4 Experimental Procedure ..........................................................................103
5.5 Results and Discussion ..........................................................................................104 5.5.1 Solubility of C2H6 and CO2 in Whole and Deasphalted Bitumen ..........106
5.5.2 Effect of C2H6 and CO2 Dissolution on Density and Viscosity of Bitumen and
Deasphalted Bitumen ............................................................................109
5.5.3 Calculation of Asphaltene Density and Viscosity ..................................114 5.6 Summary and Conclusion ......................................................................................117 5.7 References ..............................................................................................................119
CHAPTER SIX: COMBINED GEL PERMEATION CHROMATOGRAPHY AND
SIMULATED DISTILLATION FOR CHARACTERIZATION OF HEAVY CRUDE
OILS AND RESIDUES ..........................................................................................122 6.1 Preface ...................................................................................................................122
6.2 Abstract ..................................................................................................................122 6.3 Introduction ............................................................................................................123 6.4 Methodology ..........................................................................................................126 6.5 Results and Discussion ..........................................................................................136
6.5.1 Validation of the Proposed Model ..........................................................136
vii
6.5.2 Molecular Weight Distributions of Two Bitumen Samples ...................140
6.5.3 Application of the Proposed Model on Athabasca Bitumen Fractions ...143 6.6 Summary and Conclusions ....................................................................................148 6.7 References ..............................................................................................................150
CHAPTER SEVEN: A METHOD FOR CHARACTERIZATION OF BITUMEN .......153 7.1 Preface ...................................................................................................................153 7.2 Abstract ..................................................................................................................153 7.3 Introduction ............................................................................................................154 7.4 Bitumen Characterization Procedure .....................................................................157
7.5 Solubility Modeling ...............................................................................................162 7.6 Results and Discussion ..........................................................................................165 7.7 Summary and Conclusions ....................................................................................174
7.8 References ..............................................................................................................176
CHAPTER EIGHT: CONCLUSIONS AND RECOMMENDATIONS .........................179 8.1 Conclusions ............................................................................................................179
8.1.1 Bitumen Fractionation ............................................................................179 8.1.2 Experimental Phase Behaviour Data of Solvents-Bitumen Fractions ....179
8.1.3 Generalized Solvent Solubility Model ....................................................180 8.1.4 Effect of Asphaltene on Solubility, Density and Viscosity of Solvent-Bitumen
System ...................................................................................................180
8.1.5 Characterization of Heavy and Complex Mixtures Using GPC Coupled to
Simulated Distillation ...........................................................................181
8.1.6 Bitumen Characterization Method Based on Residue Curve Map .........181 8.2 Future Works .........................................................................................................181
APPENDIX A: EXPERIMENTAL APPARATUS DESIGN AND CALIBRATION ...183 A.1 Fractionation Apparatuses ....................................................................................183
A.2 Molecular Weights Measurements .......................................................................187 A.3 Asphaltene Separation ..........................................................................................189 A.4 PVT Apparatus ......................................................................................................190
A.5 Calibration of Densitometer ..................................................................................192 A.6 PVT Apparatus Validation ....................................................................................194
A.6.1. CO2-Toluene System ............................................................................194 A.6.2. Ethane-MacKay River Bitumen System ...............................................194
A.7 References .............................................................................................................196
APPENDIX B: COPYRIGHT PERMISSIONS ..............................................................197
viii
List of Tables
Table 2.1: Athabasca bitumen cuts obtained by modified vacuum distillation system. ............... 11
Table 2.2: Experimental density and viscosity data of Athabasca bitumen cuts. ......................... 13
Table 2.3: Measured solubility, density, and viscosity of butane/Athabasca bitumen cut
systems. ................................................................................................................................. 14
Table 2.4: Characterized Athabasca bitumen cut components and their properties. .................... 17
Table 2.5: Binary interaction parameter coefficients between butane and components of the
three bitumen cuts and AARD between calculated and measured butane solubility in
bitumen cuts. ......................................................................................................................... 19
Table 2.6: Characterized Athabasca and Cold Lake bitumen components and their properties. . 22
Table 2.7: The parameters to implement equation (2.2) and the AARDs between the
measured and calculated density of bitumen cuts. ................................................................ 26
Table 2.8: Required parameters to calculate the effective butane density and AARDs between
calculated and measured density of butane/bitumen cut systems. ........................................ 27
Table 2.9: The fitting parameters to implement equations (2.4 and 2.5) and the AARDs
between measured and calculated viscosity of bitumen cuts. ............................................... 28
Table 2.10: Required parameters to calculate the effective butane viscosity and AARDs
between calculated and measured viscosity of butane/bitumen cut systems. ....................... 29
Table 2.B.1: Simulated distillation results: Boiling point distribution. ........................................ 34
Table 2.B.2: Simulated distillation results: Carbon number distribution. .................................... 36
Table 3.1: Phase behaviour data of propane and Athabasca bitumen cut mixtures. ..................... 53
Table 3.2: The calculated binary interaction parameter coefficients between propane and each
component of three distillable bitumen cuts, and AARD between calculated and
experimental propane solubility in three cuts. ...................................................................... 57
Table 3.A.1: Required parameters for implementation of equation (3.A.1) and the AARDs
between calculated and experimental density of bitumen cuts. ............................................ 62
Table 3.A.2: Required parameters to calculate the effective propane density and AARDs
between calculated and experimental density of propane/bitumen cut systems. .................. 63
Table 3.B.1: Required parameters for implementation of equations (3.B.1 and 3.B.2) and the
AARDs between calculated and experimental viscosity data of bitumen cuts. .................... 65
ix
Table 3.B.2: Required parameters to calculate the effective propane viscosity and AARDs
between calculated and experimental viscosity of propane/bitumen cut systems. ............... 67
Table 4.1: Experimental vapour/liquid equilibrium properties for methane-bitumen cut and
ethane-bitumen cut mixtures. ................................................................................................ 77
Table 4.2: The binary interaction parameter coefficients between solvent (methane and
ethane) and components of bitumen cuts. ............................................................................. 83
Table 4.3: The measured and the predicted k-values of methane- and ethane-bitumen systems
(The measured k-values of Athabasca bitumen mixtures were extracted from (Zirrahi et
al., 2017)) .............................................................................................................................. 88
Table 5.1: Phase behaviour data for C2H6/bitumen, C2H6/deasphalted bitumen, CO2/bitumen,
and CO2/deasphalted bitumen systems. .............................................................................. 105
Table 6.1. The required data to find the calibration correlation obtained by GPC and
simulated distillation. The tests were performed on a bitumen sample from Athabasca
reservoir in Alberta, Canada. .............................................................................................. 133
Table 6.2: The data used in the characterization model for standard oil sample and the
calculated molecular weights along with the ARDs between the calculated molecular
weight using the proposed model and the molecular weights reported in ASTM D7169
test method. ......................................................................................................................... 139
Table 6.3: The required data and the calculated molecular weight distribution of Bitumens A
and B. .................................................................................................................................. 142
Table 6.4: The Athabasca bitumen cuts properties. .................................................................... 144
Table 7.1: Mole fraction and properties of characterized bitumen. ............................................ 166
Table 7.2: Mole fraction and properties of lumped pseudocomponents ..................................... 167
Table 7.3: Parameters for implementation of equation (7.35) to determine the binary
interaction energy parameter of the solvents and pseudocomponents. ............................... 169
Table 7.4: AARDs of the experimental solubility data of light solvents in bitumen and the
results of the proposed model and other predictive models. ............................................... 172
Table A.1: The properties of Athabasca bitumen fractions. ....................................................... 186
Table A.2: Molecular weight of 1-propanol using freezing point depression method (the
molecular weight of 1-propanol is 60.09 g/mol). ............................................................... 188
Table A.3: Molecular weight of (a) tetradecane and (b) hexadecane using freezing point
depression method (the molecular weight of n-tetradecane and hexadecane are 198.39
and 226.44 g/mol, respectively). ......................................................................................... 188
x
Table A.4: Average molecular weights of Athabasca bitumen and its fractions measured
using freezing point depression method. ............................................................................ 189
Table A.5: The obtained coefficients for implementation of equation (A.2). ............................ 193
Table A.6: Phase behaviour data of ethane/MacKay River bitumen measured in this work and
results of (Nourozieh, 2013) at 100 oC. .............................................................................. 195
xi
List of Figures
Figure 2.1: Experimental apparatus implemented for bitumen fractionation. ................................ 9
Figure 2.2: Fractionation scheme: Vacuum batch distillation (at 0.01 MPa) on Athabasca
bitumen. ................................................................................................................................ 10
Figure 2.3: Photograph of four Athabasca bitumen fractions obtained by vacuum distillation. .. 10
Figure 2.4: Simulated distillation results for Athabasca bitumen and its cuts: (a) boiling point
versus %Off; (b) Carbon number distributions. .................................................................... 12
Figure 2.5: Pressure-temperature conditions of measured experimental phase behaviour data
for butane/bitumen cuts. (Solid line shows the vapour pressure of butane). ........................ 14
Figure 2.6: The measured experimental phase behaviour data of butane/Cut 3 system: (a)
butane solubility in Cut 3; (b) density of butane-saturated Cut 3, and (c) viscosity of
butane-saturated Cut 3. ......................................................................................................... 15
Figure 2.7: Phase behaviour data of butane/bitumen cuts at 186 oC; (a) butane solubility; (b)
butane-saturated density, and (c) butane-saturated viscosity. ............................................... 16
Figure 2.8: Calculated (solid lines) and measured (symbols) butane solubility in Athabasca
bitumen cuts. ......................................................................................................................... 20
Figure 2.9: Summary of solubility calculation procedure proposed in this work to calculate
butane solubility in bitumen or heavy oil. ............................................................................ 21
Figure 2.10: Carbon number distribution of Cold Lake bitumen obtained by simulated
distillation (filled symbols) and the extrapolated plot (open symbols) using Pedersen’s
model (Pedersen et al., 1992). ............................................................................................... 21
Figure 2.11: Calculated (solid lines) and measured (symbols) butane solubility in Athabasca
bitumen obtained from (Zirrahi et al., 2017). (AARD, MAD (Maximum Absolute
Deviation), and AAD (Average Absolute Deviation) are 6.7 %, 8.0, and 3.1 mol.%,
respectively.) ......................................................................................................................... 24
Figure 2.12: Calculated (solid lines) butane solubility in Cold Lake bitumen using our
proposed model and measured (symbols) butane solubility in this work. ............................ 25
Figure 2.13: The measured (symbols) and calculated (solid lines) bitumen cuts density. ........... 26
Figure 2.14: Comparison between calculated and measured density of butane/bitumen cut
systems. ................................................................................................................................. 27
Figure 2.15: The measured (symbols) and calculated (solid lines) bitumen cuts viscosity. ......... 28
xii
Figure 2.16: Comparison between calculated and measured viscosity of butane/bitumen cut
systems. ................................................................................................................................. 29
Figure 3.1: Schematic vacuum distillation used for bitumen fractionation in this work: 1,
feeding cell; 2, water tank; 3, Quizix pump; 4, pressure indicator; 5, light fraction
collector; 6, vacuum pump; 7, condenser; 8, heavy fraction collector; 9, heat tape and
insulation; 10, flash cell; 11, oven. ....................................................................................... 48
Figure 3.2: The overall scheme of bitumen fractionation experiments. ....................................... 49
Figure 3.3: Carbon number range of Athabasca bitumen and each cut. ....................................... 50
Figure 3.4: Boiling point versus percent of distilled sample for whole Athabasca bitumen and
each cut. ................................................................................................................................ 50
Figure 3.5: Pressure and temperature conditions of experimental PVT tests for each bitumen
cut with propane. (The line shows the propane vapour pressures.) ...................................... 52
Figure 3.6: Propane solubility in Cut 1, Cut 2, Cut 3, and whole bitumen at 150 oC.
(Experimental solubility data for propane/bitumen system was obtained from (Zirrahi et
al., 2017)) .............................................................................................................................. 54
Figure 3.7: Liquid phase (a) density and (b) viscosity for propane/Cut 1, propane/Cut 2,
propane/Cut 3, and propane/bitumen at 150 oC. (* Experimental data for
propane/bitumen system was obtained from (Zirrahi et al., 2017)) ...................................... 55
Figure 3.8: The comparison between calculated (solid lines) and experimental (symbols)
solubility data of propane in each bitumen cut. .................................................................... 57
Figure 3.9: The procedure of bitumen characterization and calculation of propane solubility
in bitumen and heavy oil. ...................................................................................................... 58
Figure 3.10: The calculated (solid line) and experimental (dots) solubility data of
propane/Athabasca bitumen system. (AARD, MAD, and AAD are 4.2 %, 6.7 mol%, and
1.8 mol.%, respectively.) ...................................................................................................... 59
Figure 3.11: Evaluation of proposed solubility model: (a) carbon number distribution of Cold
Lake bitumen obtained by SimDist; (b) measured solubility data of propane/Cold Lake
bitumen in this work (symbols) and our model predictions (solid lines). (AARD, MAD,
and AAD are 6.5 %, 4.7 mol.%, and 3.0 mol.%, respectively.) ........................................... 60
Figure 3.A.1: Comparison between the experimental (symbols) and calculated (solid lines)
density of bitumen cuts. ........................................................................................................ 63
Figure 3.A.2: Comparison between the calculated and the experimental density of
propane/bitumen cut systems. ............................................................................................... 64
xiii
Figure 3.B.1: Comparison between experimental (symbols) and calculated (solid lines)
viscosity of bitumen cuts. ..................................................................................................... 66
Figure 3.B.2: Comparison between calculated and experimental viscosity of propane/bitumen
cut systems. ........................................................................................................................... 67
Figure 4.1: The bitumen fractionation scheme in this work. ........................................................ 76
Figure 4.2: The experimental phase behaviour data of methane/Cut 3 mixture; (a) methane
solubility, (b) methane-saturated density, and (c) methane-saturated viscosity. .................. 79
Figure 4.3: The experimental phase behaviour data of ethane/Cut 3 mixture; (a) ethane
solubility, (b) ethane-saturated density, and (c) ethane-saturated viscosity. ........................ 80
Figure 4.4: Experimental phase behaviour data of methane/bitumen cut systems at 186 oC; (a)
methane solubility, (b) methane-saturated density, and (c) methane-saturated viscosity. .... 80
Figure 4.5: Experimental phase behaviour data of ethane/bitumen cut systems at 50 oC; (a)
ethane solubility, (b) ethane-saturated density, and (c) ethane-saturated viscosity. ............. 81
Figure 4.6: The simulated distillation results of Athabasca bitumen fractions. ............................ 82
Figure 4.7: Comparison of the calculated (a) methane and (b) ethane solubility in Athabasca
bitumen cuts using tuned model with the measured solubility data in this work. ................ 84
Figure 4.8: The distribution of Athabasca bitumen components obtained by simulated
distillation and extrapolation using Pedersen’s model (Pedersen et al., 1992). .................... 85
Figure 4.9: Solubility calculation using the proposed model in this work; (a) methane and (b)
ethane solubility in Athabasca bitumen. (The experimental methane and ethane
solubility data were extracted from Zirrahi et al. (Zirrahi et al., 2017)) ............................... 86
Figure 4.10: The distribution of Cold Lake bitumen components obtained by simulated
distillation and extrapolation using Pedersen’s model (Pedersen et al., 1992). .................... 87
Figure 4.11. Comparison of the calculated and the measured (a) methane and (b) ethane
solubility in Cold Lake bitumen in this work. ...................................................................... 87
Figure 4.12: The k-values of (a) methane and (b) ethane-Athabasca bitumen systems. (The
symbols are the experimental measured k-value data (Zirrahi et al., 2017) and the solid
lines are the calculated k-values using the proposed method in this work.) ......................... 89
Figure 4.13: The k-values of (a) methane and (b) ethane-Cold Lake bitumen systems. (The
symbols are the experimental measured k-value data in this work and the solid lines are
the calculated k-values using the proposed method.) ........................................................... 90
xiv
Figure 4.14: The comparison between the predicted and the experimental k-value of
hydrocarbon solvent-Athabasca bitumen systems. (Data of propane- and butane-bitumen
mixtures were obtained from Chapters 3 and 2, respectively.) ............................................. 91
Figure 4.15: The comparison between the predicted and the experimental k-value of
hydrocarbon solvent-Cold Lake bitumen systems. (Data of propane- and butane-
bitumen mixtures were obtained from Chapters 3 and 2, respectively.)............................... 92
Figure 5.1: Flame ionization detector signal versus retention time for bitumen, heated
bitumen and deasphalted bitumen. ...................................................................................... 101
Figure 5.2: Boiling point versus mass percentage of distilled bitumen (%Off), deasphalted
bitumen, and asphaltene. ..................................................................................................... 102
Figure 5.3: Schematic PVT setup used in this work: 1, equilibrium cell; 2, densitometer; 3,
viscometer; 4, density measuring unit; 5, viscosity measuring unit; 6, sampling cell; 7,
Quizix pump; 8, water tank; 9, transfer cell; 10, pressure transducer; 11, ISCO pump;
12, gas cylinder; 13, vent valve; 14, oven. .......................................................................... 104
Figure 5.4: Solubility of (a) CO2 and (b) C2H6 in bitumen and in deasphalted bitumen at 70,
100, and 130 C. .................................................................................................................. 107
Figure 5.5: Carbon number distribution for bitumen, deasphalted bitumen, and asphaltene. .... 108
Figure 5.6: (a) Density and (b) viscosity of bitumen and deasphalted bitumen versus pressure
at 70, 100, and 130 ºC. ........................................................................................................ 109
Figure 5.7: Density of (a) CO2-saturated and (b) C2H6-saturated bitumen and deasphalted
bitumen as a function of pressure at temperatures of 70, 100, and 130 ºC. ........................ 111
Figure 5.8: Viscosity of (a) CO2-saturated and (b) C2H6-saturated bitumen and deasphalted
bitumen as a function of pressure at different temperatures. .............................................. 112
Figure 5.9: Comparison of the effect of CO2 and C2H6 dissolution on density of (a) bitumen
and (b) deasphalted bitumen at 70 C. ................................................................................ 113
Figure 5.10: Comparison the effect of CO2 and C2H6 dissolution on viscosity of (a) bitumen
and (b) deasphalted bitumen at 70 C. ................................................................................ 114
Figure 5.11: Density variation versus pressure at different temperatures. (Asterisks(*) denote
the measured density data in this work.) ............................................................................. 115
Figure 5.12: Viscosity variation versus pressure at different temperatures. (Asterisks(*)
denote the measured viscosity data in this work.) .............................................................. 117
Figure 6.1: True boiling point extension of Athabasca bitumen using different correlations;
The green and red lines are the results of Twu (Twu, 1984) and Riazi and Al-Sahhaf
xv
(Riazi and Al-Sahhaf, 1996); The blue circles are the experimental results of Athabasca
bitumen. .............................................................................................................................. 124
Figure 6.2: Refractive index parameter versus molecular weight of n-alkylbenzene, n-
alkylcyclopentane, and n-alkanes, calculated by correlations proposed by Riazi and Al-
Sahhaf (Riazi and Al-Sahhaf, 1996). .................................................................................. 128
Figure 6.3: Schematic of molecular weight distribution of bitumen sample. ............................. 129
Figure 6.4: GPC result of bitumen A: RID signal versus retention time. ................................... 130
Figure 6.5: GPC results of the Bitumen A: The area under curve is divided into various
fractions. .............................................................................................................................. 131
Figure 6.6: Comparison between the measured boiling points reported in ASTM D6352 or
D7169 and the calculated boiling points using Riazi and Al-Sahhaf correlation (Riazi
and Al-Sahhaf, 1996). ......................................................................................................... 132
Figure 6.7: The characterization scheme suggested in this study to obtain the whole
molecular weight distribution of heavy and extra heavy oil samples. ................................ 134
Figure 6.8: Molecular weight variation of Bitumen A versus retention time. ............................ 135
Figure 6.9: The required test results to apply the proposed characterization model on standard
sample; (a) GPC, (b) simulated distillation (the red portion of the curve shows the data
used to generate the calibration curve). .............................................................................. 137
Figure 6.10: The GPC and simulated distillation results of standard oil together; the required
data to apply the proposed model. ...................................................................................... 137
Figure 6.11: Calibration curve for standard sample. (The circles are the points involved in
generating the calibration curve and the cross symbols are those are not considered.) ...... 138
Figure 6.12: The proposed model validation; the molecular weights of standard oil sample
obtained by simulated distillation and coupled GPC to simulated distillation. .................. 139
Figure 6.13: Characterization test results of Bitumens A and B; (a) GPC chromatograms and
(b) Simulated distillation. .................................................................................................... 141
Figure 6.14: The GPC and simulated distillation test results of Bitumens A and B used in
characterization method. ..................................................................................................... 141
Figure 6.15: The calibration curves obtained for Bitumens A and B. ........................................ 142
Figure 6.16: The comparison of the predicted molecular weight distribution of Bitumens A
and B using the proposed method and the distribution obtained by simulated distillation. 143
Figure 6.17: The coupled GPC to SimDist results on whole bitumen and its fractions
obtained by vacuum distillation. ......................................................................................... 144
xvi
Figure 6.18: The moleclar weight distributions of whole bitumen and its fractions obtained
by vacuum distilation using; (a) GPC and (b) Simulated distillation tests. ........................ 145
Figure 6.19: The coupled GPC to SimDist results of whole bitumen and its heavy fractions. .. 146
Figure 6.20: The molecular weight distributions of whole bitumen and its heavy fractions;
Obtained by (a) Coupled GPC to SimDist and (b) SimDist. .............................................. 147
Figure 6.21. The accumulated %Off versus molecular weight. The open symbols are the
predicted molecular weights using the proposed method in this work and the filled
symbols represent the results of simulated distillation test. ................................................ 148
Figure 7.1: Flow chart used to characterize the bitumen. ........................................................... 162
Figure 7.2: Comparison of the Yu et al. (Yu et al., 1989) experimental values and the
calculated simulated distillated curve data.......................................................................... 166
Figure 7.3: Comparisons of the results of the proposed model with the experimental CH4
solubility data in bitumen. Experimental data was obtained from Fu et al. (Fu et al.,
1986) and Mehrotra and Svrcek (Mehrotra and Svrcek, 1988b). ....................................... 169
Figure 7.4: Comparisons between the results of the proposed model and experimental C2H6
solubility data in bitumen. Experimental data was obtained from Fu et al. (Fu et al.,
1986) and Mehrotra and Svrcek (Mehrotra and Svrcek, 1988b). ....................................... 170
Figure 7.5: Comparisons of the results of the proposed model and the experimental CO2
solubility data in bitumen. Experimental data was obtained from Yu et al. (Yu et al.,
1989) and Mehrotra and Svrcek (Mehrotra and Svrcek, 1988b). ....................................... 173
Figure 7.6: Comparisons of the results of the proposed model and the experimental N2
solubility data in bitumen. Experimental data was obtained from Mehrotra and Svrcek
(Mehrotra and Svrcek, 1988b). ........................................................................................... 174
Figure A.1: The scheme of bitumen fractionation considered in this work. ............................... 184
Figure A.2: Schematic of vacuum distillation used for bitumen fractionation in this work: 1,
feeding cell; 2, water tank; 3, Quizix pump; 4, pressure indicator; 5, light fraction
collector; 6, vacuum pump; 7, condenser; 8, heavy fraction collector; 9, heat tape and
insulation; 10, flash cell; 11, oven. ..................................................................................... 185
Figure A.3: The fabricated vacuum distillation setup. ................................................................ 186
Figure A.4: The fractions of Athabasca bitumen using three batch distillations. ....................... 186
Figure A.5: Schematic PVT setup used in this work. 1, equilibrium cell; 2, densitometer; 3,
viscometer; 4, density measuring unit; 5, viscosity measuring unit; 6, sampling cell; 7,
Quizix pump; 8, water tank; 9, transfer cell; 10, pressure transducer; 11, ISCO pump;
12, gas cylinder; 13, vent valve; 14, oven. .......................................................................... 191
xvii
Figure A.6: Measured density versus NIST density for Nitrogen and water.............................. 193
Figure A.7: Comparison between experimental data of CO2 solubility in toluene and
literature data....................................................................................................................... 194
xviii
List of Symbols, Abbreviations and Nomenclature
Abbreviations
AAD Average absolute deviation
AARD Average absolute relative deviation
ARD Absolute relative deviation
ASTM American society for testing and materials
CCE Constant composition expansion
Deasphalted Without asphaltene
DRI Differential refractive index
EoS Equation of state
ES-SAGD Expanding solvent-SAGD
FID Flame ionization detector
GC Gas chromatography
GPC Gel permeation chromatography
HPLC High performance liquid chromatography
MAD Maximum absolute deviation
mol. Mole fraction
MW, M Molecular weight
NRTL Non-random two liquids
N-solv Hot solvent injection process
OF Objective function
PC-SAFT Perturbed-chain statistical associating fluid theory
xix
PR Peng Robinson
PVT Pressure volume temperature
RI Refractive index
SAFT Statistical associating fluid theory
SAGD Steam assisted gravity drainage
SAP Saturate, aromatic, polyaromatic
SARA Saturate aromatic resin asphaltene
SD, SimDist Simulated distillation
SG Specific gravity
VAPEX Vapour extraction
VLE Vapour-liquid equilibrium
VPO Vapour pressure osmometry
wt. Weight fraction
Nomenclature
µ Viscosity
a Equation of state energy parameter
A Intermediate parameter based on equation of state parameter a
a, b Constants in calibration correlation
A, B, C, D Binary interaction energy parameter coefficients
a1-a5 Density correlation coefficients
b Equation of state co-volume parameter
B Intermediate parameter based on equation of state parameter b
xx
b1-b3 Viscosity correlation coefficients
B1-B5 Binary interaction parameter coefficients
c1-c6 Effective solvent viscosity coefficients
Cf Parameter in SØreide correlation
fi Fugacity of component i
Gij Characteristic of interaction energy
gij Energy interaction parameter between i and j
H Liquid hold-up
K Equilibrium ratio
kij Binary interaction parameter between i and j
Kw Watson characterization factor
N Number of experimental data point
Off (%) Accumulated weight percent of distilled sample
P Pressure (Absolute)
R Ideal gas constant
T Temperature
t Retention time
Tb Boiling point temperature
Tsat Saturation temperature
v Molar volume
x Liquid mole fraction
y Gas mole fraction
Z Compressibility factor
xxi
ΔHvap Heat of vaporization
ρ Density
Greek letters
α Temperature-dependent equation of state parameter
γ Activity coefficient
ζ Dimensionless time
Φ Fugacity coefficient
ΩA and ΩB EoS parameter coefficients
Subscripts
b Bitumen
Calc. calculated
Exp. Experimental
m, mix Mixture of solvent and bitumen
p Paraffin
s Solvent
r Reduced property
c Critical property
1
Chapter One: Overview
1.1 Motivations, Objectives, Steps, and Organization
Development of a multicomponent characterization of solvent-bitumen systems is restricted by the
lack of the experimental data for each bitumen fraction in the literature. This data is essential
towards developing a generalized thermodynamic model to predict phase behaviour of solvent-
bitumen systems. The solvent solubility in bitumen at different temperatures and pressures are the
basic data required for modeling and simulation of solvent-based recovery processes of bitumen.
Therefore, developing a predictive model capable of calculating the solubility of solvent in
bitumen is necessary. The current available models are tuned using the experimental solvent
solubility in the ranges of temperature and pressure. The tuned model can only calculate the solvent
solubility in the same bitumen that was used in the experiments. This approach implies that for
another bitumen sample, the new experimental data set of solubility data has to be measured and
used in the model tuning. Obtaining this data set for the bitumen mixtures is always time
consuming, energy intensive, and expensive.
This study investigates the vapour-liquid equilibrium of light hydrocarbons (methane, ethane,
propane and butane)-bitumen mixtures. The main goal of this work is to generate the experimental
phase behaviour data of solvent-bitumen fractions and also to develop a generalized model to
predict the solubility of solvents in bitumen. The target model should be capable of calculating the
solvent solubility in a new bitumen sample without requirement of the experimental solubility data
to tune the model. The proposed model can calculate the solubility in bitumen by having a very
simple analytical characterization test results.
The following main steps have been defined to accomplish this work;
1- Fractionation of a bitumen sample into four fractions using a modified vacuum distillation
system and separation of asphaltene from the heaviest fraction (Cut 4) using solvent fractionation.
2
2- Characterization of each bitumen fraction including boiling point distribution, molecular
weights, and physical property measurements.
3- Design, fabrication, calibration, and validation of the PVT apparatus to collect a comprehensive
phase behaviour data set of solvent-each bitumen fraction.
4- Measuring the vapour-liquid equilibrium data of hydrocarbon solvent (methane, ethane,
propane, and butane)-each bitumen fraction at wide ranges of temperature and pressure
(temperatures up to 190oC and pressures up to 6 MPa).
5- Development of a thermodynamic model to predict the solvent solubility in whole bitumen at
wide ranges of temperature and pressure.
6- Characterization of the heaviest component mixtures, i.e. Cut 4 (residue) and asphaltene.
The measured experimental phase behaviour data in this work fills the data gaps of light
hydrocarbon solvent-bitumen fraction systems in literature. Moreover, the results of the developed
generalized model in this study can properly estimate the solvent solubility in bitumen which are
the required data for the modeling and simulation of solvent-bitumen recovery processes.
The focus of this study is on vapour-liquid equilibria region. Although the proposed model in this
study gives acceptable results considering the distillable fractions of bitumen (Cuts 1, 2, and 3),
the residue or non-distillable fraction of bitumen (Cut 4) has to be well characterized for more
complex conditions such as liquid-liquid equilibria or asphaltene precipitation modeling. For this
reason, this work also attempts to develop a simple, fast, and inexpensive characterization method
of very heavy and complex samples such as bitumen or vacuum residue. Since characterization is
the primary step in all modeling and simulations of processes dealing with bitumen, the developed
characterization methods can find applications in defining bitumen components required for
modeling and simulation works of bitumen recovery methods.
1.2 Dissertation Outline
This dissertation is comprised of eight chapters, which six of them cover the core materials with
an overview at the beginning and conclusions and recommendations at the end. The core materials
(Chapters 2 to 7) consist of two articles published in peer-reviewed journals, three articles
3
submitted for publication in peer-reviewed journals, and one article has been accepted to be
presented in the 2018 SPE Canada Heavy Oil Conference.
Chapter 2 is dedicated to the butane-bitumen fraction systems. The bitumen fractionation apparatus
and procedures, experimental phase behaviour data of butane-bitumen fractions, and the modeling
results including butane solubility calculations and density and viscosity correlations are presented
in Chapter 2. This chapter has been submitted for publication in a peer-reviewed journal.
Chapter 3 presents the experimental phase behaviour data of propane-bitumen fractions including
propane solubility, density, and viscosity of bitumen rich phase. The details of developing the
generalized thermodynamic model to calculate the propane solubility in bitumen are also described
in this Chapter. Moreover, the effective density and viscosity approach are used to represent the
experimental density and viscosity data of propane saturated bitumen fractions. This chapter is
also a modified version of the manuscript submitted for publication.
The experimental and modeling studies of methane- and ethane-bitumen fractions are presented in
Chapter 4. A modified version of Chapter 4 will be presented at 2018 SPE Canada Heavy Oil
Conference held in March 2018.
The effect of existence of asphaltene on solubility of hydrocarbon and non-hydrocarbon solvents
in bitumen are studied in Chapter 5 which has been published in the Journal of Chemical &
Engineering Data. In this chapter, the solubility, density, and viscosity of ethane- and CO2-
bitumen systems are compared with the data of ethane- and CO2-deasphalted bitumen system to
compare the difference in presence and absence of asphaltene.
In Chapter 6, an attempt is made to characterize the very heavy and complex hydrocarbon mixtures
such as bitumen and vacuum residue. In this chapter, a new method based on combining the results
of gel permission chromatography (GPC) and simulated distillation tests are proposed to
characterize the complex components. This chapter has been also submitted for publication in a
peer-reviewed journal.
The developed bitumen characterization method based on residue curve map are offered in
Chapter 7 and has been published in Fuel.
4
Finally, Chapter 8 summarizes the results and contributions of this study and makes
recommendations for future studies.
All the experimental apparatuses and procedures implemented in this work are also described in
Appendix A. Copies of copyright permission for the published papers are provided in Appendix
B.
5
Chapter Two: Phase Behaviour of Butane-Bitumen Fractions
2.1 Preface
This chapter has been submitted for publication in peer-reviewed journal entitled “Phase
Behaviour of Butane/Bitumen Fractions: Experimental and Modeling Studies”. This manuscript
was co-authored by A. Haddadnia, M. Zirrahi, H. Hassanzadeh, and J. Abedi.
Since this dissertation has been prepared on paper-based format, unavoidably, there are some
repetitive parts in each chapter, mainly Chapters 2, 3, and 4, such as bitumen fractionation and
solubility model description.
In this chapter, all the measured experimental data of butane and each bitumen fraction followed
by the results of the model are presented.
2.2 Abstract
Utilization of butane as a solvent in solvent-aided recovery of bitumen has shown to be promising.
In these recovery processes, hot butane or butane and steam mixture is injected into subsurface
bitumen reservoirs to dilute the bitumen. Diluted bitumen drains to production well and is
produced. The primary steps towards an optimized and successful solvent-aided bitumen recovery
method are bitumen characterization and phase behaviour study of solvent/bitumen system. In this
chapter, we develop a generalized equation of state (EoS)-based model to predict the butane
solubility in bitumen and heavy oils. Bitumen is fractionated experimentally and the measured data
for each fraction is used to tune the EoS. The vacuum distillation is utilized to fractionate bitumen
to four cuts. Each bitumen cut is then characterized and the phase behaviour data including
solubility, density and viscosity of butane/bitumen cuts are measured at three temperatures of 100,
150, and 186 oC and pressures up to 4 MPa. The proposed generalized thermodynamic model
predicts butane solubility in bitumen using results of simulated distillation (SimDist) test to define
bitumen components. The developed EoS model is evaluated by comparing the calculated
solubility of butane in Athabasca and Cold Lake bitumens with experimental data. The results
6
show that the generalized model is able to predict butane solubility in bitumen without using
experimental solubility data to tune the EoS with acceptable accuracy. The density and viscosity
of original and butane-saturated bitumen cuts are also correlated. The developed model serves as
a substitute for time consuming and expensive solubility measurements. These results find
applications in design and optimization of solvent-aided recovery process of bitumen.
2.3 Introduction
Characterization of petroleum fluids is always required for process design in upstream and
downstream of oil and gas industries. To characterize heavy oil and bitumen, pseudocomponents
are typically defined based on specific gravity, boiling point, and molecular weight distributions.
Distillation, gas or liquid chromatography, and chemical-based assays are some of the popular
methods used for characterization (Diaz et al., 2014; Woods et al., 2008). For each
pseudocomponent, the physical properties are calculated using the available correlations. Mixing
rules are then applied to estimate these properties for the whole bitumen and heavy oil samples.
To recover bitumen and heavy oil using in situ methods, the viscosity should be reduced by either
increasing the temperature or dilution. The first approach is used in thermal recovery methods such
as Steam Assisted Gravity Drainage (SAGD) and the latter one is the dilution mechanism in non-
thermal recovery method such as Vapour Extraction (VAPEX). The thermal recovery methods
suffer from high energy usage, need for extensive water treatment, and the associated
environmental challenges. In addition, non-thermal recovery methods show a low oil production
rate (Das, 1998). The solvent-aided thermal recovery processes have been recently introduced to
overcome the challenges associated with thermal processes (Albahlani and Babadagli, 2008).
Expanding Solvent-SAGD (ES-SAGD) (Nasr et al., 2002) and N-Solv (Nenniger et al., 2013) are
two examples of these methods. In these methods solvent/steam mixture or heated solvent is
injected into bitumen reservoir. In order to design, develop and optimize these recovery methods,
the phase behaviour data of solvent/bitumen systems including solvent solubility, viscosity and
density of diluted bitumen are always required. However, laboratory measurements of these data
are very time consuming and expensive. Therefore, development of phase behaviour models that
can be used to calculate these properties is essential.
7
The thermodynamic models which have been developed previously to predict the phase behaviour
properties of bitumen and solvent systems are normally tuned using measured experimental data
of solvent/bitumen systems (Kariznovi et al., 2010; Mehrotra and Svrcek, 1988). For example,
solubility of solvent in bitumen at different temperatures and pressures must be measured for each
bitumen sample. The measured solubility or k-value data are then used to tune the EoS model.
Such tuned models are then applied in reservoir or process simulators to calculate solubility of
solvent in bitumen at temperature and pressure conditions where the experimental data are not
available. However, for a new bitumen or heavy oil sample, a solubility data set for new system
has to be measured and used to tune the model, which is the very time and cost intensive.
In this work, our goal is to develop a generalized solubility model which is capable of calculating
the solubility of butane in bitumen and heavy oils by means of a simple and fast characterization
test (e.g. SimDist). This model can be used in vapour-liquid region for mixture of solvent and
bitumen, which is the dominant phase region of interest in solvent-aided thermal recovery
processes of bitumen such as ES-SAGD. To build this model, bitumen has been experimentally
fractionated to four bitumen cuts using a specialized vacuum distillation technique. Each bitumen
fraction was then characterized using simulated distillation test. Solubility of butane in bitumen
fractions was measured in a wide range of temperatures and pressures and used to develop the
generalized model for butane/whole bitumen system. To apply the proposed model to other
bitumen or heavy oil samples, the boiling point or carbon distribution is the only required data
which can be obtained by simulated distillation test. The carbon number distribution is used to
characterize the bitumen and define the components.
Butane has been proposed as a favorable solvent in solvent-aided processes. Butane not only
dissolves in bitumen and heavy oil more than the other lighter hydrocarbons such as methane,
ethane, and propane but also leads to in situ upgrading of bitumen (Das and Butler, 1995). There
are very limited phase behaviour data of the mixture of butane and heavy oil or bitumen systems
in literature (Luo, 2009; Nourozieh et al., 2017; Yazdani and Maini, 2010) and to the best of our
knowledge, there is no data available for butane/bitumen fractions in the literature.
In this work, a complete data set of vapour-liquid equilibrium data of butane/Athabasca bitumen
cuts are presented. Then, the solubility model is built using the experimental solubility data of
8
butane/cuts to calculate the butane solubility in the whole bitumen. This model is then verified on
two different bitumen samples (Athabasca and Cold Lake reservoirs). The experimental data of
butane solubility in bitumen is compared to the results of the proposed model. The density and
viscosity of the original bitumen cuts and saturated cuts with butane are also correlated.
The rest of this chapter is organized as follows: First the bitumen fractionation and PVT
apparatuses are described and the experimental data are presented. Then, solubility, density, and
viscosity models and their results are discussed followed by summary and conclusion.
2.4 Materials and Experimental Methods
2.4.1 Materials
Butane was supplied by Praxair with purities of 0.995. Water- and sand-free Athabasca (MW=569
g/mol) and Cold Lake (MW=546 g/mol) bitumen samples were provided by oil companies in
Alberta, Canada.
2.4.2 Fractionation Apparatus
The experimental apparatus to fractionate the bitumen is shown in Figure 2.1. The main part of
this set-up is a flash vessel placed in an oven. Feed (raw bitumen) is charged into an injection cell
and then injected into pre-heating coil. The bitumen is heated to the desired temperature in the
heating coil. Then, it is flashed into the flash vessel to separate vapour and liquid phases. These
fluids are condensed, cooled and collected. The implemented fractionation apparatus in this work
offers two main advantages: 1) very small retention time (less than 30 minutes) to minimize the
possibility of thermal cracking reactions (Hassanzadeh et al., 2017; 2016); 2) higher separation
efficiency compared with conventional batch distillation apparatus.
9
Figure 2.1: Experimental apparatus implemented for bitumen fractionation.
The maximum temperature for distillation was 350 oC to prevent the possible thermal cracking
(Riazi, 2005). In the first step and under 350 oC, the bitumen was heated and flashed and heavy
and light fractions were separated as liquid and vapour phases. Then, the light mixture was again
fed into the distillation system at 250 oC and divided to two other fractions. In the next step, by
applying vacuum distillation at 195 oC on the light fraction obtained from the second step, two
other light and heavy fractions were collected. As shown in Figure 2.2, the heavy and light
fractions from the last distillation step are called as Cut 2 and Cut 1, respectively. The heavy
fraction obtained from the second and the first distillation called as Cut 3 and Cut 4, respectively.
Figure 2.2 shows the overall scheme of fractionation conducted in this work.
Pre-Heating Coil
Flash Drum
Condenser
Heating Oven
Pump
Injection Cell (Feed)
Cooler
Vapour Phase
Liquid Phase
Light Fraction Collector
Heavy Fraction Collector
10
Figure 2.2: Fractionation scheme: Vacuum batch distillation (at 0.01 MPa) on Athabasca bitumen.
Figure 2.3 shows a photograph of all four Athabasca bitumen cuts obtained by vacuum distillation.
By increasing the distillation temperature from Cut 1 to Cut 4, bitumen fraction becomes darker
and more viscous. The asphaltene was also separated from Cut 4 which was completely solid-like.
Heptane was considered as solvent for asphaltene separation and the asphaltene separation
procedure was described elsewhere (Azinfar et al., 2017; Diaz et al., 2014). The asphaltene content
was 26 wt.% of Cut 4 (13 wt.% based on the whole bitumen).
Figure 2.3: Photograph of four Athabasca bitumen fractions obtained by vacuum distillation.
The molecular weight of the Athabasca bitumen cuts were measured by cryoscopy method using
the freezing point depression technique. The specifications of each obtained bitumen cut are
summarized in Table 2.1.
Bitumen350 oC
Heavy Fraction (Cut 4)
Light Fraction
Heavy Fraction (Cut 3)
Light Fraction
Heavy Fraction (Cut 2)
Light Fraction (Cut 1)
250 oC
195 oC
11
Table 2.1: Athabasca bitumen cuts obtained by modified vacuum distillation system.
Sample Distillation T (oC) Weight percent
(wt.%)
MW
(g/mol)
Cut 1 195> T 19.4 268.8 ±0.9
Cut 2 195< T <250 11.7 365.5 ±1.8
Cut 3 250<T <350 18.4 464.6 ±2.3
Cut 4 T >350 50.5 906.1±11.7*
* The reported value is molecular weight of deasphalted-Cut 4.
2.4.3 Phase Behaviour Apparatus
A PVT apparatus described in our previous study (Azinfar et al., 2017) has been used to acquire
the solvent solubility data in bitumen cuts, density and viscosity of original bitumen cuts and
butane/bitumen cut systems. This experimental apparatus was designed, fabricated, calibrated, and
validated in our laboratory. More details regarding the calibration and validation of this setup are
presented in Appendix A. The PVT apparatus includes an equilibrium cell equipped with the
rocking system, an Anton Paar densitometer, and a viscometer (Viscopro 2000). After cleaning
the entire system, the butane and bitumen in equilibrium cell were mixed at desired temperature
and pressure until no more butane could dissolve in the bitumen. After reaching equilibrium, by
passing the liquid phase through densitometer and viscometer, the density and viscosity were
measured. The solubility of butane in bitumen was also obtained using the measured volume of
evolved gas when the butane-saturated bitumen is flashed at atmospheric conditions using a
gasometer (Chandler Engineering, Model 2331). Details of the experimental setup and procedure
to measure the thermo-physical properties of solvent/bitumen cut systems have been reported in
our previous study (Azinfar et al., 2017) and are presented in Chapter 5.
2.5 Results and Discussion
2.5.1 Bitumen Fractionation
Vacuum distillation was used to fractionate the bitumen in this work. Figure 2.4 shows the
simulated distillation results of Athabasca bitumen and its fractions obtained by vacuum
12
distillation. Panels a and b illustrate the boiling point versus the weight percent of distilled samples
(Athabasca bitumen and its cuts) and the carbon number distribution of them, respectively.
Figure 2.4: Simulated distillation results for Athabasca bitumen and its cuts: (a) boiling point
versus %Off; (b) Carbon number distributions.
As shown in Figure 2.4 (a), Cut 4 and whole bitumen were not completely distillable, however,
the three bitumen cuts were distilled completely at temperatures up to 700 oC. Therefore, all the
components of these three bitumen cuts are known components and will be used to develop the
solubility model. The efficiency of fractionation can be also observed in Figure 2.4(b) by
considering the difference between distributions of each cut.
2.5.2 Experimental Phase Behaviour Data
The density and viscosity of pure Athabasca bitumen cuts, that are used to develop the density and
viscosity correlations of solvent/bitumen cuts, were measured at four temperatures of 50, 100, 150,
186 oC and pressures from 1.1 to 8.1 MPa and summarized in Table 2.2.
Figure 2.5 shows the pressure and temperature conditions at which phase behaviour experiments
were carried out. The vapour liquid equilibrium (VLE) data including butane solubility, butane-
saturated density, and viscosity of Athabasca bitumen cuts were measured at three temperatures of
100, 150, 186 oC. The butane vapour pressure at 100 oC is 1.5 MPa. Therefore, for PVT
%Off
0 20 40 60 80 100
Te
mp
era
ture
(oC
)
100
200
300
400
500
600
700
800
Cut 1 Cut 2 Cut 3 Cut 4 Bitumen
Carbon Number
20 40 60 80 100
wt.
%0
2
4
6
8
10Cut 1Cut 2Cut 3Cut 4 Bitumen
(a) (b)
13
experiments at 100 oC, the highest pressure was 1 MPa to make sure that liquid/liquid equilibrium
was not formed.
Table 2.2: Experimental density and viscosity data of Athabasca bitumen cuts.
Temperature (ºC)
Pressure (MPa)
Cut 1 Cut 2 Cut 3
Density
(kg/m3)
Viscosity
(mPa.s)
Density
(kg/m3)
Viscosity
(mPa.s)
Density
(kg/m3)
Viscosity
(mPa.s) 50 1.1 893.5 6.112 944.7 94.71 966.3 1434
50 2.1 894.0 6.228 945.2 99.23 966.9 1504
50 3.1 894.7 6.314 945.8 102.4 967.5 1565
50 4.1 895.2 6.473 946.3 106.7 968.0 1628
50 5.1 895.8 6.532 946.8 109.8 968.6 1675
50 6.1 896.5 6.594 947.4 113.8 969.1 1730
50 7.1 897.2 6.711 948.0 117.1 969.6 1795
50 8.1 897.7 6.834 948.6 121.0 970.1 1831
100 1.1 866.0 1.918 913.2 9.467 937.6 42.32
100 2.1 866.7 1.942 913.76 9.585 938.3 43.85
100 3.1 867.3 1.958 914.5 9.696 939.0 46.24
100 4.1 868.2 1.974 915.1 9.969 939.5 47.76
100 5.1 868.4 1.986 915.7 10.09 940.2 50.39
100 6.1 869.6 2.006 916.3 10.34 941.0 52.55
100 7.1 870.3 2.015 917.1 10.58 941.7 54.36
100 8.1 871.2 2.033 917.8 10.86 942.0 56.69
150 1.1 828.6 0.915 882.9 2.681 903.8 6.917
150 2.1 829.6 0.920 883.7 2.729 904.7 7.013
150 3.1 830.4 0.925 884.5 2.787 905.5 7.216
150 4.1 831.3 0.930 885.3 2.822 906.3 7.313
150 5.1 832.2 0.940 886.2 2.859 907.0 7.546
150 6.1 833.1 0.950 887.1 2.893 907.8 7.639
150 7.1 834.0 0.960 887.9 2.944 908.4 7.689
150 8.1 834.9 0.967 888.7 2.998 909.3 7.953
186 1.1 801.1 0.594 859.4 1.503 881.0 3.223
186 2.1 802.6 0.602 860.3 1.530 882.0 3.282
186 3.1 803.7 0.608 861.3 1.562 883.0 3.339
186 4.1 804.9 0.611 862.3 1.582 883.9 3.385
186 5.1 805.9 0.617 863.3 1.614 885.0 3.452
186 6.1 807.0 0.623 864.2 1.621 885.7 3.505
186 7.1 808.1 0.628 865.2 1.634 886.6 3.529
186 8.1 809.4 0.633 866.2 1.661 887.3 3.561
14
Figure 2.5: Pressure-temperature conditions of measured experimental phase behaviour data for
butane/bitumen cuts. (Solid line shows the vapour pressure of butane).
All the measured phase behaviour data of butane/Cut 1, butane/Cut 2, and butane/Cut 3 systems
are summarized in Table 2.3.
Table 2.3: Measured solubility, density, and viscosity of butane/Athabasca bitumen cut systems.
Temperature (ºC)
Pressure (MPa)
Density
(kg/m3)
Viscosity
(mPa.s) Solubility
(wt.%)
Butane/Cut 1
100 0.5 806.5 0.82 11.1
100 1 712.1 0.34 30.5
150 1 773.5 0.80 10.2
150 2 695.0 0.28 25.3
150 3 543.0 0.12 51.2
186 1 765.1 0.42 6.5
186 2 722.8 0.31 15.5
186 3 668.9 0.22 25.7
186 4 589.6 0.14 40.0
Butane/Cut 2
100 0.5 865.4 2.5 9.7
100 1 783.9 0.79 24.4
150 1 836.0 1.2 9.2
150 2 770.0 0.57 21.0
150 3 642.9 0.22 43.1
186 1 827.4 0.95 5.3
186 2 791.6 0.63 11.0
186 3 748.1 0.42 17.1
186 4 692.1 0.28 27.6
Butane/Cut 3
100 0.5 888.7 6.7 7.3
Temperature (oC)
0 50 100 150 200
Pre
ssu
re (
MP
a)
0
1
2
3
4
5
15
100 1 822.1 1.8 17.2
150 1 859.5 2.5 7.3
150 2 802.9 1.1 15.7
150 3 710.3 0.47 28.2
186 1 850.9 1.8 4.3
186 2 819.1 1.2 8.9
186 3 781.8 0.78 14.5
186 4 735.7 0.53 20.9
To study the effect of temperature and pressure on phase behaviour data, butane solubility,
saturated density, and viscosity of butane/Cut 3 were plotted versus pressure at different
temperatures in Figure 2.6. As shown in panel a of this figure, by increasing pressure at constant
temperature or by decreasing temperature at constant pressure, solubility of butane in Cut 3
increases. The same behaviour is also observed for butane/Cut 1, butane/Cut 2 (Table 2.3), and
butane/Athabasca bitumen (Zirrahi et al., 2017) systems. By increasing pressure or decreasing
temperature as a result of more butane dissolution, the saturated density and viscosity are reduced
as shown in Figure 2.6(b and c). The decreasing trend of density and viscosity with solubility is
different depending on the temperature. The observed intersection at different temperatures for
saturated-density and -viscosity is attributed to two competing factors. As pressure increases at
constant temperature, the density and viscosity of gas-free bitumen cuts also increase. On the other
hand, at higher pressures, the density and viscosity of butane-saturated bitumen cuts are decreased
as a result of higher dissolved amount of butane in bitumen cuts.
Figure 2.6: The measured experimental phase behaviour data of butane/Cut 3 system: (a) butane
solubility in Cut 3; (b) density of butane-saturated Cut 3, and (c) viscosity of butane-saturated Cut
3.
Pressure (MPa)
0 1 2 3 4
Bu
tan
e s
olu
bil
ity (
wt.
%)
0
5
10
15
20
25
30
100 oC
150 oC
186 oC
Pressure (MPa)
0 1 2 3 4
De
ns
ity (
kg
/m3)
650
700
750
800
850
900
950
100 oC
150 oC
186 oC
Pressure (MPa)
0 1 2 3 4
Vis
co
sit
y (
mP
a.s
)
0.1
1
10
100 oC
150 oC
186 oC
(a) (b) (c)Cut 3 Cut 3 Cut 3
16
To compare the properties of bitumen cuts, solubility, density, and viscosity of butane/bitumen
cuts are plotted at 186 oC in Figure 2.7(a-c). As shown in panel (a), solubility of butane in Cut 1,
which is the lightest bitumen fraction, is higher than the heavier ones (Cuts 2 and 3). Figure 2.7 (b
and c) show the saturated density and viscosity of butane/bitumen cuts at 186 oC. Cut 1 has lowest
viscosity and density compared to heavy fractions (Cuts 2 and 3). The same trends for solubility,
density, and viscosity can be observed at 100 and 150 oC (Table 2.3).
Figure 2.7: Phase behaviour data of butane/bitumen cuts at 186 oC; (a) butane solubility; (b)
butane-saturated density, and (c) butane-saturated viscosity.
2.5.3 Modeling Results
In this section, the details of solubility model is described. The proposed model is then employed
to calculate the butane solubility in two bitumen samples from two different reservoirs (Athabasca
and Cold Lake bitumen samples). The density and viscosity of pure and butane-saturated bitumen
cuts are also correlated and the results are presented.
2.5.3.1 Butane Solubility Prediction
Solvent solubility in bitumen and heavy oil is one of the key parameters for reservoir simulation
and engineering studies of production, upgrading, and refining. Bitumen is a very complex mixture
containing wide ranges of hydrocarbon molecules (C6 to more than C110) and also non-hydrocarbon
components (Riazi, 2005; Subramanian et al., 1996). To build a model for solvent solubility
prediction, bitumen is defined as some pseudocomponents and the molecular weight and physical
properties of each pseudocomponent are estimated using the available correlations (Kariznovi et
Pressure (MPa)
0 1 2 3 4 5
Vis
co
sit
y (
mP
a.s
)
0.1
1
10
Cut 1Cut 2 Cut 3
186 oC
Pressure (MPa)
0 1 2 3 4 5
Den
sit
y (
kg
/m3)
550
600
650
700
750
800
850
900
Cut 1 Cut 2Cut 3
(b)
Pressure (MPa)
0 1 2 3 4 5
Bu
tan
e s
olu
bil
ity (
wt.
%)
0
10
20
30
40
50
Cut 1
Cut 2
Cut 3
186 oC186
oC(a) (c)
17
al., 2010). Then, the defined pseudocomponents are normally lumped to less number of
components and the properties of the lumped components are estimated using the mixing rules
(Whitson and Brule, 2000). After characterization of bitumen, an appropriate EoS is tuned using
the measured experimental solubility data of solvent in bitumen and heavy oil. However, this
common approach is not only prone to error through the lumping process, but also more
importantly it relies strongly on availability of experimental data for the bitumen sample to tune
the equation of state.
In this work, we propose a solubility model in which bitumen is defined as a mixture of normal
alkanes based on simulated distillation results. The simulated distillation results contain boiling
point and carbon number distributions. Carbon distribution is obtained by comparing the
components in the test sample and components in the standard sample, which is the mixture of
known hydrocarbons. In the proposed characterization method, each sample is assumed as a
mixture of n-alkanes. Based on this characterization method, Cuts 1, 2, and 3 are defined as
mixtures of C6 to C47, C6 to C99, and C6 to C100, respectively, considering the simulated distillation
results shown in Figure 2.4 (b). All components of Athabasca bitumen cuts and corresponded
molecular weights and boiling points of each n-alkane are summarized in Table 2.4.
Table 2.4: Characterized Athabasca bitumen cut components and their properties.
Carbon
number
MW
g/mol
Tb * oC
Cut 1
mol.%
Cut 2
mol.%
Cut 3
mol.%
Carbon
number
MW
g/mol
Tb * oC
Cut 1
mol.%
Cut 2
mol.%
Cut 3
mol.%
6 86 69 0.367 0.262 0.292 54 758 592 - 0.005 0.325
7 100 98 0.289 0.188 0.201 55 772 596 - 0.005 0.260
8 114 126 0.231 0.165 0.176 56 786 600 - 0.005 0.237
9 128 151 0.185 0.147 0.157 57 800 604 - 0.005 0.170
10 142 174 0.148 0.106 0.106 58 814 608 - 0.005 0.154
11 156 196 1.651 0.096 0.097 59 828 612 - 0.005 0.170
12 170 216 3.200 0.089 0.089 60 842 615 - 0.004 0.090
13 184 235 4.971 0.061 0.082 61 856 619 - 0.004 0.076
14 198 254 6.000 0.057 0.076 62 870 622 - 0.004 0.058
15 212 271 7.463 0.053 0.071 63 884 625 - 0.004 0.051
16 226 287 7.454 0.050 0.044 64 898 629 - 0.004 0.073
17 240 302 8.717 0.047 0.042 65 912 632 - 0.004 0.055
18 254 316 8.961 0.711 0.040 66 926 635 - 0.004 0.054
19 268 330 8.856 1.474 0.038 67 940 638 - 0.004 0.053
20 282 344 8.435 2.548 0.036 68 954 641 - 0.004 0.053
21 296 356 8.524 4.473 0.034 69 968 644 - 0.004 0.047
18
22 310 369 7.037 6.030 0.373 70 982 647 - 0.004 0.031
23 324 380 5.508 7.337 0.559 71 996 650 - 0.004 0.025
24 338 391 4.028 8.202 1.026 72 1010 653 - 0.004 0.025
25 352 402 2.822 8.794 1.385 73 1024 655 - 0.004 0.020
26 366 412 1.917 9.075 2.088 74 1038 658 - 0.004 0.024
27 380 422 1.307 9.562 3.095 75 1052 661 - 0.004 0.024
28 394 431 0.780 8.888 3.993 76 1066 664 - 0.004 0.024
29 408 440 0.406 7.219 4.522 77 1080 667 - 0.003 0.023
30 422 449 0.280 5.945 5.015 78 1094 670 - 0.003 0.023
31 436 458 0.163 4.874 5.615 79 1108 673 - 0.003 0.014
32 450 466 0.064 3.669 5.809 80 1122 675 - 0.003 0.022
33 464 474 0.023 2.878 6.370 81 1136 678 - 0.003 0.018
34 478 481 0.022 1.802 5.510 82 1150 681 - 0.003 0.017
35 492 489 0.021 1.491 6.161 83 1164 683 - 0.003 0.013
36 506 496 0.021 0.966 5.285 84 1178 686 - 0.003 0.009
37 520 503 0.020 0.760 5.655 85 1192 688 - 0.003 0.013
38 534 509 0.015 0.451 4.612 86 1206 691 - 0.003 0.008
39 548 516 0.019 0.419 4.907 87 1220 693 - 0.003 0.008
40 562 522 0.014 0.194 3.855 88 1234 695 - 0.003 0.008
41 576 528 0.014 0.189 3.334 89 1248 697 - 0.003 0.012
42 590 534 0.013 0.185 2.982 90 1262 700 - 0.003 0.012
43 604 540 0.013 0.100 2.846 91 1276 702 - 0.003 0.008
44 618 545 0.013 0.061 2.066 92 1290 704 - 0.003 0.008
45 632 550 0.012 0.060 1.710 93 1304 706 - 0.003 0.008
46 646 556 0.012 0.070 1.751 94 1318 708 - 0.003 0.008
47 660 561 0.004 0.040 1.295 95 1332 710 - 0.003 0.011
48 674 566 - 0.011 1.126 96 1346 712 - 0.003 0.011
49 688 570 - 0.005 0.767 97 1360 714 - 0.003 0.007
50 702 575 - 0.011 0.838 98 1374 716 - 0.003 0.007
51 716 579 - 0.005 0.541 99 1388 718 - 0.003 0.007
52 730 584 - 0.010 0.551 100 1402 720 - - 0.007
53 744 588 - 0.005 0.399 * Boiling points were obtained from (SimDist Manual, 2012).
To find the binary interaction parameter between butane and each component of bitumen cuts,
various correlations of binary interaction parameter have been tried and the following correlation
showed the best performance:
)(
3/13/1
6/16/1
321tan
54
tan
tan
)(
)(2
1
jc
jebu
jebu
j
TBB
cc
cc
rjebu TT
TT
BTBBk
(2.1)
19
where B1- B5 are the binary interaction parameter coefficients that can be determined using
experimental solubility data of butane in each cut. In the proposed correlation, the binary
interaction parameter is defined as a function of critical temperature of components and the system
temperature.
The solubility of butane in each bitumen cut is then calculated using the two phase flash
calculations on the particular cut. However, the process of minimization of error for estimation of
the binary interaction coefficients was conducted by integrating all bitumen cuts. All the defined
components of bitumen cuts, summarized in Table 2.4, were used to tune the PR-EoS considering
the details mentioned in Appendix 2.A. To achieve a generalized model, butane solubility data in
each bitumen cut at all available temperatures and pressures have been used to tune the model. The
binary interaction parameters between butane and components of bitumen cuts were optimized to
minimize the sum of squares of the differences between the measured and calculated solubility of
butane in three bitumen cuts. Optimization toolbox of MATLAB R2013a was used for
optimization and regression in this work. Table 2.5 shows the obtained binary interaction
parameter coefficients and the AARD (Average Absolute Relative Deviation) between calculated
and measured butane solubilities in three bitumen cuts.
Table 2.5: Binary interaction parameter coefficients between butane and components of the three
bitumen cuts and AARD between calculated and measured butane solubility in bitumen cuts.
B1 B2 B3 B4 B5 AARD* (%) -0.1181 0.3164 -0.2028 282.17 -0.3233 5.4
* AARD (Average Absolute Relative Deviation)
n
i
calc xxxn 1
expexp /)(1
The measured and calculated butane solubility in bitumen cuts are shown in Figure 2.8 at different
temperature and pressure conditions. This figure confirms that the tuned model can reliably
calculate the butane solubility in bitumen cuts. The AARD between measured and predicted butane
solubility data in Cuts 1, 2, and 3 is 5.4 %.
20
Figure 2.8: Calculated (solid lines) and measured (symbols) butane solubility in Athabasca
bitumen cuts.
The generalized binary interaction coefficients (Table 2.5) have been applied in solubility model
to calculate the butane solubility in whole bitumen. In other words, experimental solubility data of
butane in the whole bitumen is not required to find the binary interaction parameters and tune the
EoS. Since the distillable cuts cover a wide range of carbon distribution in bitumen, it was expected
to obtain reasonable results when these coefficients are considered for butane and whole bitumen
system. Figure 2.9 summarizes the overall proposed method to calculate the solubility of butane
in a bitumen/heavy oil sample. The first step is bitumen characterization. The carbon number or
boiling point distribution of bitumen, which is simply obtained by the simulated distillation test,
are used to characterize bitumen. Because 70-80 wt.% of bitumen is usually distilled using
simulated distillation test, the simple exponential distribution approach proposed by Pedersen
(Pedersen et al., 1992) has been used to extend the carbon number distribution. After defining all
components of the bitumen, these components are used in the PR-EoS model (see Appendix 2.A
for more details). The critical properties and acentric factor of each component are estimated using
equations (2.A.1, 2.A.2, 2.A.4, and 2.A.5) and the binary interaction parameters between the
butane and each component is calculated using equation (2.1). Then solubility of butane in bitumen
can be calculated by employing PR-EoS.
Cut 1
Pressure (MPa)
0 1 2 3 4 5
Bu
tan
e s
olu
bil
ity (
mo
l.%
)
0
20
40
60
80
100
Cut 2
Pressure (MPa)
0 1 2 3 4 5B
uta
ne
so
lub
ilit
y (
mo
l.%
)
20
30
40
50
60
70
80
90
100 oC
150 oC
186 oC
Cut 3
Pressure (MPa)
0 1 2 3 4 5
Bu
tan
e s
olu
bil
ity (
mo
l.%
)
20
30
40
50
60
70
80
90
100 oC
150 oC
186 oC
100 oC
150 oC
186 oC
21
Figure 2.9: Summary of solubility calculation procedure proposed in this work to calculate butane
solubility in bitumen or heavy oil.
We evaluated the proposed model by employing the model to predict the butane solubility in two
bitumen samples from different bitumen reservoirs; Athabasca (which has been used for
fractionation) and Cold Lake bitumen. The only required input is the simulated distillation data.
The carbon number distribution obtained from the simulated distillation test for Athabasca bitumen
was given in Figure 2.4. Figure 2.10 shows the simulated distillation and the extrapolated
distribution for Cold Lake bitumen.
Figure 2.10: Carbon number distribution of Cold Lake bitumen obtained by simulated distillation
(filled symbols) and the extrapolated plot (open symbols) using Pedersen’s model (Pedersen et al.,
1992).
Carbon number distribution
of bitumen/heavy oil
sample using SimDist
Defining sample as mixture
of n-alkanes
Estimating critical properties
and acentric factor
Calculating binary interactions
between butane and each
component of sample
Calculating butane
solubility at each T and P
applying PR-EoS
Carbon number
20 40 60 80 100 120 140
%O
ff
0
20
40
60
80
100
Simulated distillationExtrapolated
22
Using Pedersen’s method, components of Athabasca and Cold Lake bitumen based on simulated
distillation data were obtained and are summarized in Table 2.6.
Table 2.6: Characterized Athabasca and Cold Lake bitumen components and their properties.
Carbon
number
MW
g/mol
Tb*
oC
Athabasca
bitumen
mol.%
Cold Lake
bitumen
mol.%
Carbon
number
MW
g/mol
Tb*
oC
Athabasca
bitumen
mol.%
Cold Lake
bitumen
mol.%
7 100 98 0.534 0.309 79 1108 673 0.212 0.191
8 114 126 0.468 0.497 80 1122 675 0.314 0.294
9 128 151 0.375 0.442 81 1136 678 0.291 0.249
10 142 174 0.338 0.326 82 1150 681 0.274 0.242
11 156 196 0.239 0.297 83 1164 683 0.275 0.257
12 170 216 0.910 1.211 84 1178 686 0.204 0.188
13 184 235 2.117 2.826 85 1192 688 0.273 0.242
14 198 254 2.506 3.406 86 1206 691 0.195 0.175
15 212 271 3.071 4.128 87 1220 693 0.206 0.169
16 226 287 3.187 4.146 88 1234 695 0.221 0.184
17 240 302 3.757 4.505 89 1248 697 0.218 0.206
18 254 316 3.907 4.398 90 1262 700 0.211 0.184
19 268 330 3.843 4.053 91 1276 702 0.192 0.145
20 282 344 3.746 3.761 92 1290 704 0.186 0.144
21 296 356 3.966 3.896 93 1304 706 0.196 0.158
22 310 369 3.632 3.421 94 1318 708 0.162 0.129
23 324 380 3.409 2.971 95 1332 710 0.192 0.155
24 338 391 3.252 2.757 96 1346 712 0.178 0.149
25 352 402 2.835 2.501 97 1360 714 0.141 0.125
26 366 412 2.741 2.434 98 1374 716 0.167 0.142
27 380 422 2.766 2.601 99 1388 718 0.150 0.130
28 394 431 2.600 2.470 100 1402 719 0.169 0.144
29 408 440 2.328 2.082 101 1416 720 0.166 0.142
30 422 449 2.099 1.891 102 1430 722 0.163 0.139
31 436 458 1.995 1.783 103 1444 723 0.160 0.136
32 450 466 1.802 1.625 104 1458 725 0.157 0.133
33 464 474 1.748 1.609 105 1472 726 0.154 0.131
34 478 481 1.418 1.303 106 1486 728 0.151 0.128
35 492 489 1.453 1.381 107 1500 729 0.148 0.126
36 506 496 1.244 1.190 108 1514 731 0.146 0.124
37 520 503 1.252 1.218 109 1528 732 0.143 0.121
38 534 509 1.049 1.012 110 1542 733 0.141 0.119
39 548 516 1.100 1.052 111 1556 734 0.138 0.117
40 562 522 0.902 0.870 112 1570 736 0.136 0.115
41 576 528 0.889 0.867 113 1584 737 0.134 0.113
42 590 534 0.877 0.829 114 1598 738 0.132 0.111
43 604 540 0.848 0.818 115 1612 739 0.129 0.109
23
44 618 545 0.673 0.650 116 1626 741 0.127 0.107
45 632 550 0.642 0.611 117 1640 742 0.125 0.105
46 646 556 0.735 0.701 118 1654 743 0.123 0.103
47 660 561 0.623 0.593 119 1668 744 0.121 0.102
48 674 566 0.641 0.611 120 1682 745 0.119 0.100
49 688 570 0.489 0.464 121 1696 746 0.117 0.098
50 702 575 0.631 0.587 122 1710 747 0.116 0.097
51 716 579 0.477 0.446 123 1724 748 0.114 0.095
52 730 584 0.570 0.536 124 1738 749 0.112 0.093
53 744 588 0.473 0.443 125 1752 750 0.110 0.092
54 758 592 0.493 0.448 126 1766 751 0.109 0.090
55 772 596 0.442 0.413 127 1780 752 0.107 0.089
56 786 600 0.455 0.426 128 1794 753 0.105 0.088
57 800 604 0.454 0.412 129 1808 754 0.104 0.086
58 814 608 0.433 0.392 130 1822 755 0.102 0.085
59 828 612 0.470 0.435 131 1836 756 0.101 0.084
60 842 615 0.342 0.306 132 1850 757 0.099 0.082
61 856 619 0.436 0.409 133 1864 758 0.098 0.081
62 870 622 0.356 0.320 134 1878 759 0.097 0.080
63 884 625 0.320 0.285 135 1892 759 0.095 0.079
64 898 629 0.458 0.407 136 1906 760 0.094 0.077
65 912 632 0.369 0.344 137 1920 761 0.093 0.076
66 926 635 0.369 0.334 138 1934 762 0.091 0.075
67 940 638 0.363 0.323 139 1948 763 - 0.074
68 954 641 0.375 0.335 140 1962 763 - 0.073
69 968 644 0.342 0.308 141 1976 764 - 0.072
70 982 647 0.386 0.351 142 1990 765 - 0.071
71 996 650 0.348 0.315 143 2004 766 - 0.070
72 1010 653 0.370 0.316 144 2018 766 - 0.069
73 1024 655 0.297 0.282 145 2032 767 - 0.068
74 1038 658 0.375 0.332 146 2046 768 - 0.067
75 1052 661 0.350 0.347 147 2060 769 - 0.066
76 1066 664 0.305 0.290 148 2074 769 - 0.065
77 1080 667 0.316 0.300 149 2088 770 - 0.064
78 1094 670 0.317 0.287
* Boiling points were calculated using the equation by Riazi and Al-Sahhaf for n>C100
(Riazi and Al-Sahhaf, 1996).
The calculated solubility in Athabasca bitumen using the proposed model were compared with the
measured experimental data in Figure 2.11. The experimental butane solubility data in Athabasca
bitumen was extracted from (Zirrahi et al., 2017). Figure 2.11 shows that the predicted solubility
data using our model are in good agreement with measured data. The average deviation between
24
the calculated and the measured solubility data is 3.1 mol.%, which is acceptable deviation in case
of butane with high solubility values. At temperature of 180 oC and pressure of 1.78 MPa, the
model overpredicted butane solubility. The same observation for the same bitumen was previously
reported (Zirrahi et al., 2017).
Figure 2.11: Calculated (solid lines) and measured (symbols) butane solubility in Athabasca
bitumen obtained from (Zirrahi et al., 2017). (AARD, MAD (Maximum Absolute Deviation), and
AAD (Average Absolute Deviation) are 6.7 %, 8.0, and 3.1 mol.%, respectively.)
To further examine the validity of the proposed model, butane solubilities have been measured at
three different temperatures for butane/Cold Lake bitumen system. The calculated and measured
solubilities of butane in Cold Lake bitumen were shown in Figure 2.12. This figure shows that the
proposed solubility model can calculate the solubilities of butane in Cold Lake bitumen with
acceptable accuracy while no experimental solubility data was used to tune the model. The average
and the maximum deviations between measured and calculated solubilities are 2.4 and 6.9 mol.%.
Athabasca bitumen
Pressure (MPa)
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5
Bu
tan
e s
olu
bil
ity (
mo
l.%
)
20
40
60
80
100
100 oC
150 oC
180 oC
AARD=6.7 %MAD=8.0 mol%AAD=3.1 mol%
25
Figure 2.12: Calculated (solid lines) butane solubility in Cold Lake bitumen using our proposed
model and measured (symbols) butane solubility in this work.
While measuring the solubility data in laboratory is very expensive and time consuming, the
proposed solubility model only requires the boiling point or carbon number distribution of any
unknown bitumen sample which can be obtained easily by simulated distillation test. The
generalized model combined with the simulated distillation data of a bitumen sample allows
prediction of butane solubility without using experimental data and can be used as a substitute for
time consuming and expensive solubility measurements. The calculated solubility data can be
directly used to generate the fluid models for reservoir simulation when no experimental data is
available.
2.5.3.2 Density Correlations
Prediction of bitumen density in the presence of solvent is always required for design and
optimization of solvent-aided bitumen recovery processes. Correlating the solvent-saturated
density of bitumen assists estimation of density data required for simulation and modeling works.
In this part, the density of the butane-saturated fractions has been correlated. First, the density of
bitumen cuts has been correlated using (Zirrahi et al., 2017),
Pressure (MPa)
0 1 2 3 4 5
Bu
tan
e s
olu
bilit
y (
mo
l.%
)
10
20
30
40
50
60
70
80
90
100 oC
150 oC
186 oC
AARD=8.3 %MAD=6.9 mol%AAD=2.4 mol%
Cold Lake bitumen
26
))exp(exp()( 54
2
321 TaPaTaTaa (2.2)
where, T and P are temperature in K and pressure in MPa, respectively. The coefficients of a1 to
a5 were calculated using the density of bitumen cuts (Table 2.2) and summarized in Table 2.7.
Table 2.7: The parameters to implement equation (2.2) and the AARDs between the measured
and calculated density of bitumen cuts.
Sample a1 a2 a3 a4 a5 AARD (%) Cut 1 902.67 0.42746 -1.42×10-3 1.11×10-4 5.59×10-3 0.095
Cut 2 1146.18 -0.62321 -7.70×10-6 1.06×10-4 5.16×10-3 0.021
Cut 3 1165.54 -0.61155 -2.04×10-5 1.43×10-4 4.32×10-3 0.077
The measured and predicted density of bitumen cuts were compared in Figure 2.13 at two high
and low pressures of 1.1 and 6.1 MPa. This figure and the AARDs in Table 2.7 that are less than
0.1% confirm the reliability of equation (2.2) to calculate the bitumen cuts density.
Figure 2.13: The measured (symbols) and calculated (solid lines) bitumen cuts density.
The typical method to correlate the density of solvent/bitumen mixture using the mixing rules is
considering an effective density for solvent. Equation (2.2) has been also used to estimate the
effective density of butane. The effective density of butane can be calculated using the measured
butane-saturated density of bitumen cuts summarized in Table 2.3 and the following non-linear
mole fraction based mixing rule.
b
s
s
s
mix
xx
)1(1 (2.3)
Cut 1
Temperature (oC)
40 60 80 100 120 140 160 180 200
Den
sit
y (
kg
/m3)
780
800
820
840
860
880
900
920
P= 1.1 MPa
P= 6.1 MPa
Cut 2
Temperature (oC)
40 60 80 100 120 140 160 180 200
Den
sit
y (
kg
/m3)
840
860
880
900
920
940
960
P= 1.1 MPa
P= 6.1 MPa
Cut 3
Temperature (oC)
40 60 80 100 120 140 160 180 200
Den
sit
y (
kg
/m3)
860
880
900
920
940
960
980
P= 1.1 MPa
P= 6.1 MPa
27
where, ρmix, ρs, ρb are the density of mixture, solvent, and bitumen cut, respectively. xs is the solvent
mole fraction.
The tuned parameters of equation (2.2) to calculate the effective density of butane and the AARDs
between measured and calculated butane-saturated density of each cut are summarized in Table
2.8.
Table 2.8: Required parameters to calculate the effective butane density and AARDs between
calculated and measured density of butane/bitumen cut systems.
Sample a1 a2 a3 a4 a5 AARD (%) Cut 1 594.16 1.583 -2.79×10-3 -12.422 -1.11×10-2 1.4
Cut 2 391.11 3.065 -4.70×10-3 -25.804 -1.32×10-2 0.71
Cut 3 391.14 3.058 -4.60×10-3 -23.094 -1.34×10-2 0.44
Using the tuned equations for bitumen cuts and effective density of butane, the density of
butane/bitumen cut has been calculated and compared with the measured density of
butane/bitumen cut in Figure 2.14 at temperatures of 100, 150, and 186 oC. This figure shows that
the aforementioned method based on considering the effective density for butane is the reliable
method to calculate the density of butane/bitumen cut systems.
Figure 2.14: Comparison between calculated and measured density of butane/bitumen cut
systems.
2.5.3.4 Viscosity Correlations
Viscosity of solvent/bitumen system is another important data for simulation and modeling works
of solvent-aided bitumen recovery processes. To correlate the viscosity of butane/bitumen cuts,
Butane/Cut 1
Measured density (kg/m3)
500 600 700 800 900
Calc
ula
ted
den
sit
y (
kg
/m3)
500
600
700
800
900
100 oC150 oC186 oC
Butane/Cut 2
Measured density (kg/m3)
600 650 700 750 800 850 900
Calc
ula
ted
den
sit
y (
kg
/m3)
600
650
700
750
800
850
900
100 oC
150 oC
186 oC
Butane/Cut 3
Measured density (kg/m3)
700 750 800 850 900
Calc
ula
ted
den
sit
y (
kg
/m3)
700
750
800
850
900
100 oC
150 oC
186 oC
28
the effective viscosity method has been also used. The viscosity of Cuts 2 and 3 were correlated
using the equation proposed by Mehrotra and Svrcek (Mehrotra and Svrcek, 1986),
PbTbbb 321
ln)ln(ln (2.4)
where, µb is viscosity of bitumen cut in mPa.s at temperature T (K) and pressure P (MPa).
For Cut 1 (the light fraction of bitumen), the above equation has been modified as single-log type,
PbTbbb 321 lnln (2.5)
The calculated fitting parameters of b1, b2, and b3 were summarized in Table 2.9 along with the
AARDs between the calculated and measured viscosity of bitumen cuts.
Table 2.9: The fitting parameters to implement equations (2.4 and 2.5) and the AARDs between
measured and calculated viscosity of bitumen cuts.
Sample b1 b2 b3 AARD (%) Cut 1 42.717 -7.081 8.5×10-3 8.22
Cut 2 33.761 -5.582 7.3×10-3 10.90
Cut 3 30.174 -4.879 4.9×10-3 8.23
The measured and calculated viscosity of bitumen cuts are compared in Figure 2.15, which
confirms the reliability of the aforementioned equations to correlate the viscosity of bitumen cuts.
Figure 2.15: The measured (symbols) and calculated (solid lines) bitumen cuts viscosity.
To calculate the viscosity of butane/bitumen cut systems, the log-type mixing rule based on mole
fraction was used as given by;
Cut 1
Temperature (oC)
40 60 80 100 120 140 160 180 200
Vis
co
sit
y (
mP
a.s
)
1
10
P= 1.1 MPa
P= 6.1 MPa
Cut 2
Temperature (oC)
40 60 80 100 120 140 160 180 200
Vis
co
sit
y (
mP
a.s
)
1
10
100
1000
P= 1.1 MPa
P= 6.1 MPa
Cut 3
Temperature (oC)
40 60 80 100 120 140 160 180 200
Vis
co
sit
y (
mP
a.s
)
1
10
100
1000
10000
P= 1.1 MPa
P= 6.1 MPa
29
bbssmixxx lnlnln (2.6)
where, µmix, µs, µb are viscosity of mixture, solvent, and bitumen cut, respectively; xs is the solvent
mole fraction. To calculate the viscosity of solvent-saturated bitumen, the effective viscosity of
dissolved butane has to be correlated. The viscosity of dissolved butane was correlated as function
of temperature and pressure as given by,
2
65
2
4321TcPTcPcTcPcc
s (2.7)
where, µs is the viscosity of solvent at pressure P (MPa) and temperature T (K). The fitting
parameters (c1 to c6) have been tuned using the experimental viscosity data of butane/bitumen cuts
(given in Table 2.3) and summarized in Table 2.10. The proposed method calculates the viscosity
of butane/bitumen cut systems with the AARDs less than 6.08 %. The measured and calculated
viscosity of butane/bitumen cut systems were compared in Figure 2.16. A good agreement between
the calculated viscosity and the experimental viscosity has been observed in this Figure.
Table 2.10: Required parameters to calculate the effective butane viscosity and AARDs between
calculated and measured viscosity of butane/bitumen cut systems.
Sample c1 c2 c3 c4 c5 c6 AARD (%) Cut 1 -0.0085 -0.0495 -1.17×10-4 9.03×10-3 -9.67×10-5 1.84×10-6 4.81
Cut 2 0.7504 -0.9399 3.50×10-3 -9.43×10-3 2.14×10-3 -1.07×10-5 6.08
Cut 3 -0.3179 -0.7765 6.59×10-3 -1.17×10-2 1.77×10-3 -1.17×10-5 4.55
Figure 2.16: Comparison between calculated and measured viscosity of butane/bitumen cut
systems.
Butane/Cut 1
Measured viscosity (mPa.s)
0.1 1
Calc
ula
ted
vis
co
sit
y (
mP
a.s
)
0.1
1
100 oC
150 oC
186 oC
Butane/Cut 2
Measured viscosity (mPa.s)
0.1 1 10
Calc
ula
ted
vis
co
sit
y (
mP
a.s
)
0.1
1
10
100 oC
150 oC
186 oC
Butane/Cut 3
Measured viscosity (mPa.s)
0.1 1 10
Calc
ula
ted
vis
co
sit
y (
mP
a.s
)
0.1
1
10
100 oC
150 oC
186 oC
30
2.6 Summary and Conclusion
The Athabasca bitumen was vacuum distilled at three stages into four bitumen fractions. The three
first cuts were completely distillable and the last one was partially distillable and solid like at room
temperature. Butane solubility in bitumen, density, and viscosity of liquid phase for
butane/bitumen cuts were measured at temperatures of 100, 150, and 186 oC and pressures up to 4
MPa. The higher butane solubility in Cut 1, which is the lightest bitumen fraction, was observed
compared to Cuts 2 and 3. Moreover, butane-saturated density and viscosity of each bitumen cut
were measured and compared. Using the measured butane solubilities in bitumen cuts at various
temperature and pressure conditions, the generalized thermodynamic model was proposed to
calculate the butane solubility in whole bitumen. The only required data in this model is carbon
number or boiling point distribution of bitumen or heavy oils using simulated distillation test,
which is much less expensive and fast as compared with the experimental solubility measurements.
The proposed model in this work was examined by calculating the butane solubility in two
different bitumen samples. This model calculated the butane solubility in Athabasca and Cold Lake
bitumens with the average absolute deviations of 3.1 and 2.4 mol.%, respectively. The results
confirm that this model can acceptably calculate the butane solubility in bitumen even when no
experimental solubility data is available to tune the EoS. The proposed model serves a substitute
for time consuming and expensive solubility measurements and will find applications in design
and optimization of solvent-aided recovery process of bitumen. The density and viscosity of
butane-saturated bitumen cuts were correlated using the effective density and viscosity of butane.
This model can correlate the density and viscosity of butane/bitumen cuts with AARDs of less
than 1.4 and 6.8 %, respectively.
31
Appendix 2.A: Equations Used to Develop the Solubility Model.
Perturbation Expansion Correlations to find critical temperature and pressure and specific gravity
(Danesh, 1998; Twu, 1984) are summarized here. The properties of normal paraffins are correlated
using the normal boiling point,
113310
273
])100//(0460774.0)10(65848.1
)10(526167.2)10(343831.0533272.0[
bb
bbbcp
TT
TTTT
(2.A.1)
2422
1
)65163.800546.389698.2099334.0318317.0( cp
P (2.A.2)
1235.1374936159.3128624.0843593.0
pSG (2.A.3)
where, subscript p refers to the properties of normal paraffins, Tc, Pc, and SG are critical
temperature and pressure, and specific gravity, respectively, and cpb TT /1 .
Lee-Kesler correlations was applied to find acentric factor (Danesh, 1998; Lee and Kesler, 1975)
as given:
For 8.0brT ,
)43577.0ln4721.13/6875.152518.15(
/)16934.0ln28862.1/09648.692714.5(ln
6
6
brbrbr
brbrbrbr
TTT
TTTP
(2.A.4)
For 8.0brT ,
brwbrww TKTKK /)01063.0408.1(359.8007465.01352.0904.7 2 (2.A.5)
where, ω is acentric factor, cbpr PPP / , cbpr TTT / , bP is the pressure at which bT (normal
boiling point) is measured, and Kw is Watson characterization factor and calculated as follows,
32
SGTK bw /)8.1( 3
1
(2.A.6)
Peng-Robinson Equation of State (Danesh, 1998; Peng and Robinson, 1972),
)]()(/[)/( bvbbvvabvRTP (2.A.7)
20.5
riici
2
ci
2
i ))T(1m)(1/PT0.457235(Ra (2.A.8)
ciciiPRTb /077796.0 (2.A.9)
For 49.0 ,
226992.05422.137464.0 iiim (2.A.10)
For 49.0 ,
3201667.01644.0485.13796.0 iiiim (2.A.11)
The mixture parameters, a and b, are calculated using mixing rules,
i j
ijjiji kaaxxa )1()( 5.0 (2.A.12)
i
iibxb (2.A.13)
The following equation was solved to find the compressibility factor,
0)()32()1( 32223 BBABZBBAZBZ (2.A.14)
where, RT
PvZ ,
2)(RT
aPA ,
RT
bPB .
After finding the compressibility factor, the following equation was used to calculate the fugacity
coefficient (Danesh, 1998),
33
))21(
)21(ln(]))1()(
2[(
)22(
)ln()1(ln
1
5.0
BZ
BZ
b
bkaax
aB
A
BZZb
b
iN
j
ijjij
i
i
(2.A.15)
where, i is the fugacity coefficient of component i, ia and ib are parameters of PR-EoS and
defined in equations 2.A.8 and 2.A.9, a and b (PR-EoS parameters for mixtures) are also defined
in equations 2.A.12 and 2.A.13, and A and B are defined in equation 2.A.14.
34
Appendix 2.B: Simulated Distillation Results of Athabasca Bitumen Cut and Whole Bitumens.
Table 2.B.1: Simulated distillation results: Boiling point distribution.
%Off Cut 1
Tb(oC)
Cut 2
Tb(oC)
Cut 3
Tb(oC)
Cut 4
Tb(oC)
Athabasca
bitumen
Tb(oC)
Cold Lake
bitumen
Tb(oC)
0 175.4 303.2 360.6 381.2 208.6 205.7
1 188.7 316.8 378.7 432.4 224.8 220.3
2 203.8 329.9 394.9 478.5 247.3 237.7
3 212.6 337.7 404.2 498.7 262.6 251.6
4 219.3 343.5 410.7 511.4 275.4 262.2
5 225.2 348.1 415.7 522 287.1 271.7
6 230.5 351.9 419.8 530.4 296.4 280.9
7 235.3 355.2 423.3 537.7 304.8 289.3
8 239.9 358 426.4 544.2 312.5 296.6
9 244.2 360.8 429.3 550.4 319.8 303.2
10 248.2 363.3 432.0 556.2 327.0 310.0
11 251.8 365.6 434.8 561.5 333.8 316.5
12 254.9 367.7 437.2 566.4 340.5 323.1
13 258.0 369.7 439.5 570.8 346.8 329.6
14 261.0 371.7 441.8 575.1 352.8 336.1
15 263.8 373.6 444.0 579.5 358.7 342.4
16 266.6 375.4 446.0 583.8 364.5 348.4
17 269.2 377.2 448.0 587.9 370.1 354.3
18 271.7 378.9 450.0 591.8 375.8 360.2
19 274.4 380.5 451.8 595.7 381.1 366.0
20 277.1 382.1 453.6 599.4 386.3 372.0
21 279.7 383.7 455.2 603.2 391.9 378.2
22 282.1 385.3 456.8 606.9 397.7 384.4
23 284.5 386.8 458.4 610.5 403.2 390.4
24 286.8 388.2 460.1 614.0 408.7 396.7
25 289.0 389.7 461.6 617.5 414.0 402.8
26 291.0 391.0 463.2 621.1 419.0 408.9
27 293.0 392.4 464.7 624.5 423.9 414.5
28 294.9 393.7 466.2 627.7 428.7 419.5
29 296.7 395 467.6 630.8 433.8 424.4
30 298.5 396.3 469.1 633.8 438.8 429.3
31 300.3 397.6 470.5 636.7 444.2 434.7
32 302 398.9 471.9 639.7 449.5 440.3
33 303.8 400.1 473.2 642.6 454.6 445.9
34 305.6 401.4 474.6 645.5 459.8 451.7
35 307.2 402.6 476 648.3 465.2 457.2
36 308.8 403.8 477.3 650.9 470.6 462.8
37 310.5 405 478.6 653.3 476.1 468.6
35
38 312.2 406.2 479.9 655.7 481.6 474.3
39 313.8 407.3 481.3 658.0 487.6 480.1
40 315.3 408.5 482.6 660.9 493.6 486.1
41 316.9 409.6 484 663.7 499.4 492.2
42 318.5 410.7 485.3 666.7 505.0 498.1
43 320.0 411.8 486.6 669.6 510.9 503.7
44 321.6 412.9 487.9 672.1 517.2 509.4
45 323.1 413.9 489.2 674.6 523.5 515.7
46 324.6 415.0 490.5 677.6 529.7 522.0
47 326.1 416.0 491.9 680.7 536.0 528.2
48 327.7 417.0 493.2 683.5 542.2 534.5
49 329.2 418.0 494.5 686.2 548.7 540.8
50 330.6 419.0 495.8 689.0 555.5 547.2
51 332.1 420.1 497.0 691.7 562.0 554.1
52 333.6 421.1 498.2 693.9 568.3 560.7
53 335.0 422.1 499.4 696.5 574.2 567.1
54 336.5 423.1 500.6 699.4 580.5 573.3
55 337.9 424.1 501.8 701.9 586.8 579.7
56 339.4 425.0 503 704.3 592.8 586.2
57 340.8 426.0 504.2 706.7 598.7 592.5
58 342.2 427.0 505.4 709.4 604.6 598.7
59 343.7 428.0 506.6 712.3 610.4 604.9
60 345.0 429.1 507.9 714.9 616.0 611.0
61 346.4 430.1 509.1 717.7 621.7 616.9
62 347.8 431.1 510.4 - 626.9 622.8
63 349.2 432.3 511.7 - 631.9 628.4
64 350.6 433.4 513 - 636.6 633.5
65 351.9 434.6 514.3 - 641.2 638.5
66 353.3 435.7 515.7 - 645.9 643.5
67 354.7 436.9 517 - 650.2 648.3
68 356.1 438.1 518.4 - 654.1 652.8
69 357.5 439.3 519.8 - 657.9 656.7
70 358.9 440.4 521.2 - 662.6 661.3
71 360.3 441.7 522.6 - 667.5 666.2
72 361.8 443 524.2 - 672.1 671.1
73 363.2 444.3 525.7 - 676.6 675.8
74 364.7 445.6 527.4 - 681.8 681.6
75 366.2 446.9 529 - 686.5 686.5
76 367.7 448.3 530.7 - 691.3 691.6
77 369.2 449.7 532.4 - 695.2 696.2
78 370.8 451.1 534.2 - 700.3 701.7
79 372.4 452.6 535.9 - 704.5 707.0
80 374.1 454.1 537.6 - 708.9 712.6
81 375.8 455.6 539.4 - 713.9 718.0
82 377.6 457.1 541.3 - 718.7 -
36
83 379.4 458.8 543.2 - - -
84 381.3 460.5 545.3 - - -
85 383.3 462.4 547.6 - - -
86 385.3 464.3 550 - - -
87 387.6 466.3 552.6 - - -
88 389.9 468.5 555.3 - - -
89 392.3 470.8 558.1 - - -
90 394.8 473.3 561.2 - - -
91 397.6 476.1 564.5 - - -
92 400.7 479.1 568 - - -
93 404 482.6 571.9 - - -
94 407.8 486.7 576.5 - - -
95 412 491.5 582.1 - - -
96 416.7 497.2 589.2 - - -
97 422.5 504.2 598.5 - - -
98 430.3 514.5 613.6 - - -
99 444.8 535.2 644.6 - - -
100 460.1 559.2 675.4 - - -
Table 2.B.2: Simulated distillation results: Carbon number distribution.
Carbon
range
Cut 1
wt.%
Cut 2
wt.%
Cut 3
wt.%
Cut 4
wt.%
Athabasca
bitumen
wt.%
Cold Lake
bitumen
wt.%
( C5, C6 ) 0.12 0.06 0.05 0 0 0
( C5, C7 ) 0.23 0.11 0.09 0.05 0.10 0.06
( C5, C8 ) 0.33 0.16 0.13 0.09 0.20 0.17
( C5, C9 ) 0.42 0.21 0.17 0.13 0.29 0.28
( C5, C10 ) 0.50 0.25 0.2 0.17 0.38 0.37
( C5, C11 ) 1.48 0.29 0.23 0.20 0.45 0.46
( C5, C12 ) 3.55 0.33 0.26 0.24 0.74 0.86
( C5, C13 ) 7.03 0.36 0.29 0.27 1.47 1.87
( C5, C14 ) 11.55 0.39 0.32 0.30 2.40 3.18
( C5, C15 ) 17.57 0.42 0.35 0.32 3.62 4.88
( C5, C16 ) 23.98 0.45 0.37 0.35 4.97 6.70
( C5, C17 ) 31.94 0.48 0.39 0.37 6.66 8.80
( C5, C18 ) 40.60 0.96 0.41 0.40 8.52 10.97
( C5, C19 ) 49.63 2.01 0.43 0.42 10.45 13.08
( C5, C20 ) 58.68 3.92 0.45 0.44 12.43 15.14
( C5, C21 ) 68.28 7.44 0.47 0.46 14.63 17.38
( C5, C22 ) 76.58 12.41 0.7 0.48 16.74 19.44
( C5, C23 ) 83.37 18.73 1.06 0.50 18.81 21.31
( C5, C24 ) 88.55 26.1 1.75 0.60 20.87 23.12
( C5, C25 ) 92.33 34.33 2.72 0.70 22.74 24.83
37
( C5, C26 ) 95.00 43.16 4.24 0.80 24.62 26.56
( C5, C27 ) 96.89 52.82 6.58 0.90 26.59 28.48
( C5, C28 ) 98.06 62.13 9.71 0.99 28.51 30.37
( C5, C29 ) 98.69 69.96 13.38 1.17 30.29 32.02
( C5, C30 ) 99.14 76.63 17.59 1.36 31.95 33.57
( C5, C31 ) 99.41 82.28 22.46 1.55 33.58 35.08
( C5, C32 ) 99.52 86.67 27.66 1.72 35.10 36.50
( C5, C33 ) 99.56 90.22 33.54 1.90 36.62 37.95
( C5, C34 ) 99.60 92.51 38.78 2.13 37.89 39.16
( C5, C35 ) 99.64 94.46 44.81 2.52 39.23 40.48
( C5, C36 ) 99.68 95.76 50.13 2.87 40.41 41.65
( C5, C37 ) 99.72 96.81 55.98 3.34 41.63 42.88
( C5, C38 ) 99.75 97.45 60.88 3.81 42.68 43.93
( C5, C39 ) 99.79 98.06 66.23 4.43 43.81 45.05
( C5, C40 ) 99.82 98.35 70.54 5.00 44.76 46.00
( C5, C41 ) 99.85 98.64 74.36 5.71 45.72 46.97
( C5, C42 ) 99.88 98.93 77.86 6.49 46.69 47.92
( C5, C43 ) 99.91 99.09 81.28 7.36 47.65 48.88
( C5, C44 ) 99.94 99.19 83.82 8.13 48.43 49.66
( C5, C45 ) 99.97 99.29 85.97 8.94 49.19 50.41
( C5, C46 ) 100 99.41 88.22 9.97 50.08 51.29
( C5, C47 ) - 99.48 89.92 10.9 50.85 52.05
( C5, C48 ) - 99.5 91.43 11.9 51.66 52.85
( C5, C49 ) - 99.51 92.48 12.8 52.29 53.47
( C5, C50 ) - 99.53 93.65 14.00 53.12 54.27
( C5, C51 ) - 99.54 94.42 14.90 53.76 54.89
( C5, C52 ) - 99.56 95.22 16.00 54.54 55.65
( C5, C53 ) - 99.57 95.81 17.00 55.20 56.29
( C5, C54 ) - 99.58 96.3 18.10 55.90 56.95
( C5, C55 ) - 99.59 96.7 19.10 56.54 57.57
( C5, C56 ) - 99.6 97.07 20.20 57.21 58.22
( C5, C57 ) - 99.61 97.34 21.20 57.89 58.86
( C5, C58 ) - 99.62 97.59 22.30 58.55 59.48
( C5, C59 ) - 99.63 97.87 23.40 59.28 60.18
( C5, C60 ) - 99.64 98.02 24.30 59.82 60.68
( C5, C61 ) - 99.65 98.15 25.40 60.52 61.36
( C5, C62 ) - 99.66 98.25 26.30 61.10 61.90
( C5, C63 ) - 99.67 98.34 27.20 61.63 62.39
( C5, C64 ) - 99.68 98.47 28.40 62.40 63.10
( C5, C65 ) - 99.69 98.57 29.40 63.03 63.71
( C5, C66 ) - 99.7 98.67 30.40 63.67 64.31
( C5, C67 ) - 99.71 98.77 31.40 64.31 64.90
( C5, C68 ) - 99.72 98.87 32.50 64.98 65.52
( C5, C69 ) - 99.73 98.96 33.50 65.60 66.10
( C5, C70 ) - 99.74 99.02 34.60 66.31 66.77
38
( C5, C71 ) - 99.75 99.07 35.70 66.96 67.38
( C5, C72 ) - 99.76 99.12 36.80 67.66 68.00
( C5, C73 ) - 99.77 99.16 37.70 68.23 68.56
( C5, C74 ) - 99.78 99.21 38.90 68.96 69.23
( C5, C75 ) - 99.79 99.26 40.10 69.65 69.94
( C5, C76 ) - 99.8 99.31 41.10 70.26 70.54
( C5, C77 ) - 99.81 99.36 42.10 70.90 71.17
( C5, C78 ) - 99.82 99.41 43.20 71.55 71.78
( C5, C79 ) - 99.83 99.44 43.90 71.99 72.19
( C5, C80 ) - 99.84 99.49 45.10 72.65 72.83
( C5, C81 ) - 99.85 99.53 46.10 73.27 73.38
( C5, C82 ) - 99.86 99.57 47.10 73.86 73.92
( C5, C83 ) - 99.87 99.6 48.20 74.46 74.50
( C5, C84 ) - 99.88 99.62 49.00 74.91 74.93
( C5, C85 ) - 99.89 99.65 50.00 75.52 75.49
( C5, C86 ) - 99.9 99.67 50.80 75.96 75.90
( C5, C87 ) - 99.91 99.69 51.60 76.43 76.30
( C5, C88 ) - 99.92 99.71 52.40 76.94 76.74
( C5, C89 ) - 99.93 99.74 53.40 77.45 77.24
( C5, C90 ) - 99.94 99.77 54.20 77.95 77.69
( C5, C91 ) - 99.95 99.79 55.00 78.41 78.05
( C5, C92 ) - 99.96 99.81 55.80 78.86 78.41
( C5, C93 ) - 99.97 99.83 56.70 79.34 78.81
( C5, C94 ) - 99.98 99.85 57.40 79.74 79.14
( C5, C95 ) - 99.98 99.88 58.20 80.22 79.54
( C5, C96 ) - 99.99 99.91 59.00 80.67 79.93
( C5, C97 ) - 99.99 99.93 59.60 81.03 80.26
( C5, C98 ) - 99.99 99.95 60.40 81.46 80.64
( C5, C99 ) - 100 99.97 61.10 81.85 80.99
( C5, C100) - - 99.99 61.10 81.85 80.99
39
2.7 References
Albahlani, A.M., Babadagli, T., 2008. A Critical Review of the Status of SAGD: Where Are We
and What is Next?. SPE Western Regional and Pacific section, Bakersfield, California, 31
March–2 April.
Azinfar, B., Haddadnia, A., Zirrahi, M., Hassanzadeh, H., Abedi, J., 2017. Effect of Asphaltene
on Phase Behavior and Thermophysical Properties of Solvent/Bitumen Systems. J. Chem.
Eng. Data 62, 547–557.
Danesh, A., 1998. PVT and Phase Behaviour of Petroleum Reservoir Fluids. First ed., Elsevier
Science, Amesterdam, The Netherlands.
Das, S.K., 1998. Vapex: An Efficient Process for the Recovery of Heavy Oil and Bitumen. SPE J.
1998; 232–7.
Das, S.K., Butler, R.M., 1995. Extraction of Heavy Oil and Bitumen Using Solvents at Reservior
Pressure. Sixth Petroleum Conference of the South Saskatchewan Section, Regina,
Saskatchewan, 16–18 October.
Diaz, O.C., Sánchez-Lemus, M.C., Schoeggl, F., Satyro, M.A., Taylor, S.D., Yarranton, H.W.,
2014. Deep-Vacuum Fractionation of Heavy Oil and Bitumen, part I: Apparatus and
Standardized Procedure. Energy Fuels 28, 2857–2865.
Hassanzadeh, H., Harding, T.G., Moore, R.G., Mehta, S.A., Ursenbach, M.G., 2016. Gas
Generation during Electrical Heating of Oil Sands. Energy Fuels 30, 7001–7013.
Hassanzadeh, H., Rabiei Faradonbeh, M., Harding, T., 2017. Numerical Simulation of Solvent and
Water Assisted Electrical Heating of Oil Sands Including Aquathermolysis and Thermal
Cracking Reactions. AIChE J. 63, 4243–4258.
Kariznovi, M., Nourozieh, H., Abedi, J., 2010. Bitumen Characterization and Pseudocomponents
Determination for Equation of State Modeling. Energy Fuels 24, 624–633.
Lee, B.I., Kesler, M.G., 1975. A Generalized Thermodynamic Correlation Based on Three-
Parameter Corresponding States. AIChE J. 21, 510–527.
40
Luo, P., 2009. Asphaltene Precipitation and its Effect on a Solvent-Based Heavy Oil Recovery
Process. PhD Thesis, University of Regina, Regina, Saskatchewan.
Mehrotra, A.K., Svrcek, W.Y., 1988. Characterization Of Athabasca Bitumen For Gas Solubility
Calculations. J. Can. Pet. Technol. 27, 107–110.
Mehrotra, A.K., Svrcek, W.Y., 1986. Viscosity of Compressed Athabasca Bitumen. Can. J. Chem.
Eng. 64, 844–847.
Nasr, T.N., Beaulieu, G., Golbeck, H., Heck, G., 2002. Novel Expanding Solvent-SAGD Process
"ES-SAGD". Canadian International Petroleum Conference, Calgary, Alberta, 11–13 June.
Nenniger, J., Holcek, R., Dillon, J., Wolff, V., 2013. Solvent Injection Plant for Enhanced Oil
Recovery and Method of Operating Same. Canadian patent, CA 2777966.
Nourozieh, H., Kariznovi, M., Abedi, J., 2017. Solubility of n-Butane in Athabasca Bitumen and
Saturated Densities and Viscosities at Temperatures Up to 200°C. SPE J. 94–102.
Pedersen, K.S., Blilie, A.L., Meisingset, K.K., 1992. PVT Calculations on Petroleum Reservoir
Fluids Using Measured and Estimated Compositional Data for the Plus Fraction. Ind. Eng.
Chem. Res. 31, 1378–1384.
Peng, D.-Y., Robinson, D.B., 1976. A New Two-Constant Equation of State. J. Ind. Eng. Chem.
Fundam. 15, 59–64.
Riazi, M.R., 2005. Characterization and Properties of Petroleum Fractions. First ed., ASTM
International, U.S.A.
Riazi, M.R., Al-Sahhaf, T.A., 1996. Physical Properties of Heavy Petroleum Fractions and Crude
Oils. Fluid Phase Equilib. 117, 217–224.
SimDist Reporter User Manual, 2012. Bruker, The Netherlands.
Subramanian, M., Hanson, F., 1996. Compositional Analysis of Bitumen and Bitumen-Derived
Products. J. Chromatogr. Sci. 34, 20–6.
Twu, C.H., 1984. An Internally Consistent Correlation for Predicting the Critical Properties and
Molecular Weights of Petroleum and Coal-tar Liquids. Fluid Phase Equilib. 16, 137–150.
41
Whitson, C.H., Brule, M.R., 2000. Phase Behavior. Monograph, vol. 20. Richardson, Texas: SPE,
Henry L. Doherty series.
Woods, J., Kung, J., Kingston, D., Kotlyar, L., Sparks, B., Mccracken, T., 2008. Canadian Crudes:
A Comparative Study of SARA Fractions from a Modified HPLC Separation Technique. Oil
Gas Sci. Technol. 63, 151–163.
Yazdani, A., Maini, B., 2010. Measurements and Modelling of Phase Behaviour and Viscosity of
a Heavy Oil/Butane System. J. Can. Pet. Technol. 49, 9–14.
Zirrahi, M., Hassanzadeh, H., Abedi, J., 2017. Experimental and Modeling Studies of MacKay
River Bitumen and Light n-Alkane Binaries. Can. J. Chem. Eng. 95, 1417–1427.
42
Chapter Three: Phase Behaviour of Propane-Bitumen Fractions
3.1 Preface
This chapter has been submitted for publication in peer-reviewed journal entitled “A Generalized
Thermodynamic Model to Predict Propane Solubility in Bitumen and Heavy Oil Based on
Experimental Fractionation and Characterization”. This manuscript was co-authored by A.
Haddadnia, M. Zirrahi, H. Hassanzadeh, and J. Abedi.
Since this dissertation has been prepared on paper-based format, unavoidably, there are some
repetitive parts in each chapter, mainly Chapters 2, 3, and 4, such as bitumen fractionation and
solubility model description.
In this chapter, all the measured experimental data of propane and each bitumen fraction followed
by the results of the model are presented. The density and viscosity of the pure bitumen fractions
were presented in Chapter 2.
3.2 Abstract
Propane has been suggested as suitable solvent for solvent-aided bitumen recovery methods such
as Vapour Extraction (VAPEX), N-Solv (hot solvent injection), and Expanding Solvent-Steam
Assisted Gravity Drainage (ES-SAGD). Characterization of bitumen and phase behaviour study
of solvent/bitumen systems are the initial steps towards an optimized and successful solvent-aided
bitumen recovery process. In this study, Athabasca bitumen is experimentally fractionated to four
cuts using modified vacuum distillation and then each cut is characterized. The phase behaviour
data including solubility, density, and viscosity of propane/bitumen cuts are measured in wide
ranges of temperature and pressure. Using the measured solubility data of propane/bitumen cuts,
the PR-EoS is tuned and a binary interaction parameter correlation between propane and bitumen
components is developed. The bitumen is then characterized using the boiling point or carbon
number distribution obtained by simulated distillation test (ASTM D7169). The generalized model
is then implemented to calculate the propane solubility in two bitumen samples from different
43
reservoirs (Athabasca and Cold Lake). Our proposed model in this paper, predicted the propane
solubility in Athabasca bitumen sample with an average deviation of 1.8 mol.% without using any
experimental data of propane/bitumen system or tuning parameter. The proposed model was also
evaluated by predicting the propane solubility in Cold Lake bitumen. The propane solubility in
Cold Lake bitumen was calculated with an average deviation of 3.0 mol.%, which shows the
generality of the proposed model.
3.3 Introduction
Although thermal recovery methods are currently the most practical methods for heavy oil and
bitumen recovery, they are energy intensive processes and thus prone to excessive greenhouse gas
(GHG) emission. The high energy intensity and low oil price turn these recovery processes
uneconomical in many reservoirs particularly those with thin pay zone (<10 m), low porosity, high
water saturation, low rock thermal conductivity, and those with an under laying aquifer
(Hassanzadeh and Harding, 2016; Jiang, 1997). Over the last two decades, solvent-based
techniques have been widely introduced as an alternative method promise eliminating the use of
water and reducing the GHG level and energy intensity. To optimize the solvent composition and
operating conditions the phase behaviour data and models of solvent, bitumen and their mixtures
are required. Solvents dilute the oleic phase and reduce viscosity of bitumen, which result in
improved bitumen recovery.
Bitumen is a highly complex mixture and it is practically impossible to define each component of
bitumen individually. Therefore, bitumen needs to be characterized and well-defined. Typically,
bitumen is divided to some pseudocomponents based on distillation assay. Then, the molecular
weight and physical properties of each pseudocomponent are estimated in such a way that all the
pseudocomponents approximate the properties of the whole bitumen. Because there is no complete
set of data available for bitumen fractions, the phase behaviour calculation is then validated using
the measured experimental data of the whole bitumen (Kariznovi et al., 2010; Mehrotra and
Svrcek, 1988; Nourozieh et al., 2015). Pseudocomponents definition and lumping based on the
described approach, introduce errors in modeling studies and reliability of the predictive models
depends strongly on definition method, number of pseudocomponents, and lumping schemes. The
multicomponent characterization of bitumen is currently limited to the aforementioned method.
44
In this work, bitumen is experimentally divided into several light, medium and heavy fractions.
Each fraction covers a range of boiling point distribution and can be well-characterized using
physical property measurement, simulated distillation (SimDist) and molecular weight
measurement. Then, the experimental phase behaviour data of mixtures of solvent and each
individual bitumen cut are used to tune a generalized thermodynamic model. Since the
implemented cuts include a wide range of boiling point distribution, it is expected that tuned model
is general and will have capability to predict the solvent solubility in any type of bitumen without
further tuning. This approach offers a bitumen characterization method which only requires
simulated distillation data of an unknown bitumen and avoids expensive experimental data of
solvent/bitumen system.
Diaz et al. (Diaz et al., 2014) studied deep-vacuum fractionation of heavy oils and bitumen. First,
they separated asphaltene from the bitumen using n-pentane. Then, the maltene was fractioned into
five and eight fractions and the residue using the deep vacuum apparatus. The physical properties
of the fractions were measured. They reported that about 55-56 wt.% of bitumen was distilled
using their apparatus.
The phase behaviour of CO2/Cold Lake bitumen cuts was studied by Eastick et al. (Eastick et al.,
1992), and Huang and Radosz (Huang and Radosz, 1990a). They studied CO2/bitumen cut systems
in which five bitumen fractions were provided by Esso Resources Canada, Ltd.
In 1992, Eastick et al. (Eastick et al., 1992) measured vapour-liquid compositions at temperatures
from 25 to 150C and pressures up to 10 MPa. The equilibrium data was correlated using Peng
Robinson Equation of State (PR-EoS) (Peng and Robinson, 1976). In their work, the use of binary
interaction parameter between two cuts appears to be inevitable. They improved the solubility
predictions by estimating the critical pressure, in terms of molar mass and critical temperature, by
correlating the CO2-saturated bitumen fractions density with Patel-Teja EoS (Patel and Teja,
1982).
Huang and Radosz in 1990 (Huang and Radosz, 1990a) studied the mutual phase equilibrium
solubilities for CO2/bitumen fractions at temperature and pressure ranges up to 250C and 16 MPa.
They measured the solubility of bitumen and the three distillable cuts. Cuts 4 and 5 were only
45
characterized. The solubilities were correlated using the Soave (Soave, 1972) and Perturbed-Hard-
Chain ( PHC) models (Cotterman et al., 1986). They also developed generalized binary parameters
as functions of temperature and molecular weight using the PHC model, which are consistent for
bitumen fractions and the total bitumen. Using their model, equilibrium data for the CO2/bitumen
system can be predicted by having the average molecular weight and aromaticity. However, they
concluded that the Soave binary parameters are dependent on bitumen type and should be fitted to
experimental data for each unknown bitumen (Huang and Radosz, 1990a). In a follow-up study
(Huang and Radosz, 1991), they applied the Statistical Associating Fluid Theory (SAFT) EoS
(Huang and Radosz, 1990b) to predict CO2/bitumen solubilities. They found that the solubility
predictions using SAFT, which considers molecular association, were more accurate than other
EoSs (PHC and Soave) especially at higher molecular weights (Huang and Radosz, 1991).
In another work, Sayegh et al. (Sayegh et al., 1990), measured the physical properties and phase
behaviour of CO2/Lindberg heavy oil mixtures. They measured the phase behaviour data for heavy
oil and its fractions saturated with CO2 at 21 and 140C and pressures up to 15 MPa. First, they
separated the asphaltene and resin from the heavy oil using n-heptane as the solvent. Then, using
vacuum distillation under 1-2 mm Hg, they fractionated the de-asphalted oil into three fractions.
By comparing solubility data for heavy oil, de-asphalted oil, and fraction 2 (intermediate
distillation cut), they observed similar CO2 solubility and concluded that the heavy ends of the
crude oil have similar impacts on the solubility of CO2 (Sayegh et al., 1990).
Kokal et al. (Kokal and Sayegh, 1993), studied the phase behaviour of CO2/Lone Rock heavy oil
and its three fractions. They separated the asphaltene using pentane as a solvent and then used
vacuum distillation under pressure of 1-2 mm Hg to divide the de-asphalted oil into two other
fractions. The phase behaviour data (viscosity, density, and solubility) were measured at 21C and
140C and pressures up to 12.41 MPa. They used the PR-EoS to correlate the experimental phase
behaviour data (Kokal and Sayegh, 1993).
Marufuzzaman and Henni (Marufuzzaman and Henni, 2015) compared the solubility of ethane
and carbon dioxide in Cactus Lake oil fractions including saturate, aromatic, resin, and maltene.
Their results showed that the solubility in saturate and maltene fractions are the highest and the
lowest ones, respectively.
46
In this work, the phase behaviour of propane/bitumen fractions is studied. To the best of our
knowledge, there is no available work on propane and bitumen fraction mixtures in literature.
Propane is one of the favorable solvent in solvent-based processes such as VAPEX and N-Solv
processes. Propane has high solubility in bitumen compared to methane, ethane, carbon dioxide
and nitrogen, which leads to higher viscosity reduction. Moreover, it contributes to in-situ
upgrading, which results in higher oil quality (Nenniger et al., 2013). Propane can be also added
to steam-based processes such as SAGD as an additive to improve the process performance (Nasr
and Ayodele, 2006). The main gains of adding small amount of solvent in steam are; steam usage
is reduced, because solvent is replaced with the fraction of steam, and the produced bitumen offsets
the solvent required to be added for pipeline transportation.
To evaluate the feasibility of a specific recovery method or to find the optimum composition of
injected solvent in the field, the solubility of solvent is always required to calculate the k-values.
However, measuring this data in laboratory is expensive and time consuming. Therefore,
developing a generalized model which is not dependent to this scarce experimental data for
solubility prediction is always the main concern.
In this study, Athabasca bitumen is fractionated to four cuts using a newly designed vacuum
distillation systems. The phase behaviour data including, solubility, density, and viscosity of
propane/bitumen cuts are measured at different temperature and pressure ranges. Using the VL
equilibrium solubility data of the three distillable fractions of bitumen, a generalized binary
interaction (kij) of propane/bitumen components for PR-EoS is developed. The proposed model is
then used to calculate the propane solubility in whole bitumen without using any experimental
data. The propane-saturated density and viscosity of bitumen cuts are also correlated and the results
are presented in Appendices 3.A and 3.B, respectively.
The rest of this chapter is organized as follows: first, the experimental apparatus and procedure for
bitumen fractionation along with simulated distillation results on the obtained bitumen fractions
are described. Then, the PVT apparatus and procedure are presented. In results and discussion
section, the experimental results are presented followed by description of the solubility model and
obtained results. The final part is the summary and conclusion.
47
3.4 Experimental Section
3.4.1 Materials
Propane was supplied by Praxair with purities of 0.995. Water- and sand-free Athabasca and Cold
Lake bitumen samples were provided by oil companies in Alberta, Canada. The molar mass of 569
and 546 g.mol-1 were measured by cryoscopy method (More details regarding MW measurements
are described in Appendix A). All the materials were used in experiments without further
purifications.
3.4.2 Bitumen fractionation
The conventional vacuum distillation has been modified to reach the maximum separation
efficiency in each bitumen cut. Having higher separation efficiency results in less overlap of
carbon distribution of the obtained cuts.
The asphaltene can be first separated from bitumen using solvent and the de-asphalted oil can be
fractionated using vacuum distillation (Diaz et al., 2014; Kokal and Sayegh, 1993, 1989; Sayegh
et al., 1990). Although adding solvent to separate the asphaltene reduces the viscosity of the de-
asphalted oil and facilitates the distillation process, it can introduce impurity to the system. After
separation of asphaltene from bitumen, the solvent should be removed. It is not easily possible to
remove the solvent completely. Therefore, the trace of the added solvent contaminates the
remained oil. In this work, first the whole bitumen is fractionated using vacuum distillation in
which no additive is added to the bitumen. In the next stage, the solvent (such as heptane) was
used to separate asphaltene from the heaviest fraction (in this work, called as Cut 4) out from the
vacuum distillation. Considering the heavy components in this cut, it could be easily recognized if
solvent still exists in the fraction.
After running trial vacuum distillations and analyzing the results, the newly designed vacuum
distillation apparatus for bitumen fractionation has been fabricated. A schematic of the vacuum
distillation set-up used for bitumen fractionation in this work is shown in Figure 3.1. The
fractionation procedure using the modified vacuum distillation is described in the following.
Bitumen is fed to the feeding cell and pumped to the oven. Bitumen is warmed up while passing
through the pre-heating lines in oven until it reaches to the flash cell. In the flash cell, the heated
48
bitumen is flashed and light components (vapour phase) are separated from heavy ones (liquid
phase). Light components (distillate) were collected after passing through the condenser. The
heavy components were also collected from the bottom of the cell in the other collector and called
Cut 4. As it is shown in Figure 3.1, using the liquid level difference pointed at the bottom of the
flash cell, the residence time for liquid phase was set to 20-30 minutes (set by injection rate) to
minimize thermal cracking.
Figure 3.1: Schematic vacuum distillation used for bitumen fractionation in this work: 1, feeding
cell; 2, water tank; 3, Quizix pump; 4, pressure indicator; 5, light fraction collector; 6, vacuum
pump; 7, condenser; 8, heavy fraction collector; 9, heat tape and insulation; 10, flash cell; 11, oven.
The maximum temperature for distillation was 350 oC to minimize the possible thermal cracking
(Riazi, 2005). In the next step after through washing and cleaning of the system, the distillate
collected from the first vacuum distillation is fed to the feeding cell and fractionated to two other
light and heavy fractions using the aforementioned procedure at 250 oC. The heavy fraction in
second vacuum distillation is called Cut 3 and the light fraction is again fed to the system for the
last batch distillation. Applying the third vacuum distillation on distillate out from the second
2
3
5
4
8
11
6
7
9
1
10
49
vacuum distillation at 195 oC results in two more light and heavy fractions called Cut 1 and Cut 2,
respectively.
The overall scheme of fractionation and temperature and pressure at which the distillations were
conducted is shown in Figure 3.2.
Figure 3.2: The overall scheme of bitumen fractionation experiments.
To show the efficiency of the separation, the carbon number distribution obtained from simulated
distillation tests was plotted versus the distilled percent for whole bitumen, and each cut in Figure
3.3. As an example, most of the components in Cut 1 were eluted from the simulated distillation
column at around C20 at which the distillation of Cut 2 is started. It can be concluded that the
distillation was successful and achieved a high fractionation efficiency.
Water Tank
Pump
Light Fraction
Pressure Indicator
Heavy Fraction
Oven
Vacuum Pump
Condensor
Insulation and Heat tape
Bitumen
Bitumen350 oC, 3 inHg
Heavy Fraction (Cut 4)
Light Fraction
Water Tank
Pump
Light Fraction
Pressure Indicator
Heavy Fraction
Oven
Vacuum Pump
Condensor
Insulation and Heat tape
Bitumen
250 oC, 3 inHg
Heavy Fraction (Cut 3)
Light Fraction
Water Tank
Pump
Light Fraction
Pressure Indicator
Heavy Fraction
Oven
Vacuum Pump
Condensor
Insulation and Heat tape
Bitumen
195 oC, 3 inHg
Heavy Fraction (Cut 2)
Light Fraction (Cut 1)
50
Figure 3.3: Carbon number range of Athabasca bitumen and each cut.
The boiling point versus percent of distilled bitumen, and bitumen cuts are shown in Figure 3.4.
This figure illustrates that the Cut 1, Cut 2, and Cut 3 were completely distilled while the whole
bitumen and the Cut 4 were distilled 82 and 61 %, respectively, at temperatures up to 718˚C.
Therefore, all the components of three distillable cuts are known and have been used to build the
generalized model, which is explained in the solubility modeling part.
Figure 3.4: Boiling point versus percent of distilled sample for whole Athabasca bitumen and each
cut.
Carbon Number
20 40 60 80 100
%O
ff
0
20
40
60
80
100
Cut 1Cut 2Cut 3 Cut 4Bitumen
%Off
0 20 40 60 80 100
Tem
pera
ture
(oC
)
100
200
300
400
500
600
700
800
Cut 1 Cut 2 Cut 3
BitumenCut 4
51
3.4.3 PVT tests
The schematic and details of experimental PVT apparatus are presented in Chapter 5 (Azinfar et
al., 2017). The PVT set-up includes an equilibrium cell equipped with rocking system, an Anton
Paar densitometer (model DMA HPM), a Viscopro 2000 viscometer, an ISCO pump, a Quizix
pump, a pressure indicator, one transfer cell, and a sampling cell. The calibrated densitometer,
factory-calibrated viscometer, and the equilibrium cell were placed in an oven. The transfer and
sampling cells are located outside the oven. The ISCO and Quizix pumps are used for solvent
injection to the system and transfer the liquid phase from the equilibrium cell to the transfer and
sampling cells, respectively. Before each measurement, the entire system was washed with toluene
and acetone and then dried and vacuumed. After the cleaning procedure, the sample was charged
into the equilibrium cell inside the oven. The oven temperature had been set to the desired set
point. The gaseous solvent was then injected using an ISCO pump at the desired pressure. The
feed and solvent inside the equilibrium cell were mixed using the rocking system until equilibrium
was reached. The ISCO pump was kept running to inject the makeup solvent and maintain the
pressure constant during the dissolution of solvent in the liquid phase. After reaching equilibrium,
i.e. the solvent could no longer dissolve in the sample, the equilibrium cell was maintained in the
vertical direction for half an hour. Then, the liquid phase was discharged from the bottom of the
equilibrium cell. The liquid phase was passed through the inline density and viscosity measuring
devices and the density and viscosity were recorded. The liquid sample was taken using the
sampling cell and flashed at atmospheric pressure to measure the solubility of gas in the liquid
phase. The evolved gas was measured using the Chandler Engineering Gasometer (Model 2331)
with an accuracy of 0.2% over the range of the readings. The solvent solubility is calculated using
the volume of the evolved gas measured by the gasometer and density of gas at atmospheric
pressure.
The phase behaviour experimental data of propane/bitumen cut systems were measured at four
temperatures of 50, 100, 150, and 186 oC and pressures up to 6 MPa in such a way that VL
equilibrium condition exists in the system. Figure 3.5 illustrates the PT conditions at which the
phase behaviour data were collected.
52
Figure 3.5: Pressure and temperature conditions of experimental PVT tests for each bitumen cut
with propane. (The line shows the propane vapour pressures.)
3.5 Results and Discussion
In this part, first, the experimental phase behaviour data including solvent solubility, density, and
viscosity of liquid phase for propane/Athabasca bitumen cut system are presented. Then, solubility
model is described and the results of solubility prediction employing the proposed model are
presented.
3.5.1 Experimental Phase Behaviour Data
The solubility data of solvent/bitumen systems are the common input in reservoir simulators to
build thermodynamic and phase behaviour models. Solubility of propane in the first three cuts (1,
2, 3) at 50, 100, 150, and 186 oC and pressures up to 6 MPa were measured. The experimental
phase equilibrium data for propane/Athabasca bitumen cut systems are summarized in Table 3.1.
The data of propane solubility in bitumen cuts (measured in this work) and in the whole bitumen
(data from (Zirrahi et al., 2017)) at 150 oC and different pressures are shown in Figure 3.6. The
increase in pressure or decrease in temperature results in higher propane solubility. Moreover,
propane can be dissolved more in lighter bitumen fraction (Cut 1) compared to the heavier fractions
or the whole bitumen.
Temperature (oC)
0 50 100 150 200
Pre
ssu
re (
MP
a)
0
1
2
3
4
5
6
7
L-L
V-L
53
Table 3.1: Phase behaviour data of propane and Athabasca bitumen cut mixtures.
Temperature ºC
Pressure MPa
Density
kg/m3
Viscosity
mPa.s Solubility
wt. %
Propane/Cut 1
50 0.5 853.1 2.25 6.7
50 1.0 790.2 1.14 17.7
100 1.5 804.6 0.82 8.6
100 3.0 707.9 0.34 23.4
150 1.5 796.0 0.61 5.2
150 3 756.2 0.43 11.1
150 4.5 707.2 0.30 16.9
150 6.0 645.4 0.21 25.3
186 1.5 777.3 0.50 3.1
186 3.0 748.4 0.40 7.2
186 4.5 715.5 0.32 11.6
186 6.0 677.8 0.25 17.2
Propane/Cut 2
50 0.5 909.3 15.03 5.2
50 1.0 857.7 3.75 11.9
100 1.5 863.2 2.62 7.0
100 3.0 791.6 0.95 18.0
150 1.5 851.7 1.67 3.4
150 3 822.8 1.04 8.2
150 4.5 787.4 0.72 12.5
150 6.0 744.1 0.46 17.2
186 1.5 837.1 1.10 1.4
186 3.0 813.4 0.84 4.6
186 4.5 788.5 0.63 8.6
186 6.0 761.1 0.48 11.9
Propane/Cut 3
50 0.5 931.9 96.63 4.4
50 1.0 885.6 15.72 10.4
100 1.5 888.5 7.49 5.4
100 3.0 827.0 2.15 13.8
150 1.5 877.1 3.76 3.0
150 3 848.8 2.41 6.4
150 4.5 817.3 2.03 11.6
150 6.0 784.6 1.73 15.6
186 1.5 860.2 2.19 2.6
186 3.0 839.7 1.63 4.8
186 4.5 818.4 1.24 7.6
186 6.0 795.4 0.96 10.9
54
Figure 3.6: Propane solubility in Cut 1, Cut 2, Cut 3, and whole bitumen at 150 oC. (Experimental
solubility data for propane/bitumen system was obtained from (Zirrahi et al., 2017))
Density and viscosity are the important inputs required for simulation studies of bitumen recovery
methods. Figure 3.7 (a and b) illustrate the density and viscosity of propane-saturated bitumen cuts
and whole bitumen at 150 oC and different pressures. At the constant temperature, the density and
viscosity reduce as a result of dissolving more propane in bitumen at higher pressure.
The density and viscosity of propane/bitumen cuts have been correlated and the results are
presented in Appendices 3.A and 3.B, respectively.
150 oC
Pressure (MPa)
1 2 3 4 5 6 7
Pro
pan
e s
olu
bil
ity (
wt.
%)
0
5
10
15
20
25
30
Cut 1
Cut 2
Cut 3
Bitumen*
55
Figure 3.7: Liquid phase (a) density and (b) viscosity for propane/Cut 1, propane/Cut 2,
propane/Cut 3, and propane/bitumen at 150 oC. (* Experimental data for propane/bitumen system
was obtained from (Zirrahi et al., 2017))
3.5.2 Solubility Model Description
The objective is to develop a generalized model for prediction of propane solubility in the whole
bitumen by tuning the PR-EoS (Peng and Robinson, 1972) using the measured solubility data of
propane in the three distillable cuts.
An important factor in characterizing the ill-defined mixtures such as bitumen is definition of their
heavy end. The simulated distillation and GC test results on bitumen and heavy oil are presented
up to about 700 oC and usually about 70-80 wt.% of bitumen can be eluted from the GC column.
For this reason, characterization methods have been proposed in the literature to describe the
undistillable fraction of the mixture (20-30 wt.%) (Pedersen et al., 1992; Riazi, 1989; Whitson,
1983). After characterization and definition of the pseudocomponents, the EoS should be then
tuned to find the binary interaction parameters. The experimental solubility data of
solvent/bitumen system should be used to tune the model (Kariznovi et al., 2010). This approach
not only needs very expensive and time consuming experimental PVT data to predict the solvent
Pressure (MPa)
1 2 3 4 5 6 7
Den
sit
y (
kg
/m3)
600
650
700
750
800
850
900
950
Propane/Cut 1
Propane/Cut 2
Propane/Cut 3
Propane/Bitumen*
Pressure (MPa)
1 2 3 4 5 6 7
Vis
co
sit
y (
mP
a.s
)
0.1
1
10
100
Propane/Cut 1Propane/Cut 2Propane/Cut 3Propane/Bitumen*
150 oC150
oC
(a) (b)
56
solubility, but also the tuned model is not generalized and can only be used for the type of bitumen
used in that experiments. In other words, the results are not applicable to other bitumen samples.
This study concerns development of a generalized thermodynamic model to predict solvent
solubility in bitumen without the need for experimental solubility data. To achieve this goal, first,
each bitumen cut is characterized. Cut 1, Cut 2, and Cut 3 were assumed as the mixtures of normal
alkanes based on simulated distillation results presented in Figure 3.3. For example according to
simulated distillation results of Cut 1 presented in Figure 3.3, this cut was considered as the mixture
of C6 to C47. The boiling point and molecular weight of each normal alkane have been used in the
calculations. All the considered components of Cuts 1, 2, and 3 were shown in Table 2.4. After
defining the components in each bitumen cut, the solubility model is developed as described in the
following.
Perturbation Expansion Correlations were employed to calculate the critical properties of each
component (Twu, 1984). Acentric factor was calculated by Lee-Kesler correlation (Lee and Kesler,
1975). Equations used for developing the solubility model were presented in Appendix 2.A. To
find compressibility factor (equation 2.A.14), the root which results in lower Gibbs free energy
was selected. The criterion to select among all the roots was developed by Michelsen (Michelsen,
1982a, 1982b). Then, the selected root was used to calculate fugacity coefficient (equation 2.A.15).
Optimization toolbox of MATLAB R2013a was used for optimization and regression in this work.
Various types of binary interaction coefficients can be found in different forms and as functions
of different properties (Chueh and Prausnitz, 1967; Gao et al., 1992; Varotsis et al., 1986). After
screening several types, we propose the binary interaction coefficient between propane and
hydrocarbon molecules as functions of component critical temperature and temperature (equation
(3.1)).
)(
3/13/1
6/16/1
321
54
)(
)(2
1
jc
jpropane
jpropane
j
TBB
cc
cc
rjpropane TT
TT
BTBBk
(3.1)
57
where Tc is the critical temperature. B1, B2, B3, B4, and B5 are the binary interaction parameter
coefficients. The binary interaction coefficients of propane/bitumen cut components have been
tuned to match the experimental solubility data of propane in bitumen cuts.
The binary interaction parameters between propane and hydrocarbon components present in
bitumen cuts were adjusted to minimize the sum of squares of the differences between the
experimental and calculated solubility of propane in all three bitumen cuts. The calculated binary
interaction parameters and AARD are listed in Table 3.2.
Table 3.2: The calculated binary interaction parameter coefficients between propane and each
component of three distillable bitumen cuts, and AARD between calculated and experimental
propane solubility in three cuts.
B1 B2 B3 B4 B5 AARD* (%) -0.1837 0.3529 0.1799 -0.2543 0.0714 5.4
* AARD (Average Absolute Relative Deviation)
n
i
calc xxxn 1
expexp /)(1
The calculated and experimental solubility data of propane in bitumen cuts are shown in Figure
3.8. This figure and the value of AARD in Table 3.2 show that the tuned model can acceptably
calculate the solubility of propane in all bitumen cuts. To achieve more accurate and generalized
model, all the solubility data at all available temperatures and pressures for three bitumen cuts
were used to tune the model.
Figure 3.8: The comparison between calculated (solid lines) and experimental (symbols)
solubility data of propane in each bitumen cut.
Cut 1
Pressure (MPa)
0 1 2 3 4 5 6 7
Pro
pa
ne
so
lub
ilit
y (
mo
l.%
)
10
20
30
40
50
60
70
80
Cut 2
Pressure (MPa)
0 1 2 3 4 5 6 7
Pro
pa
ne
so
lub
ilit
y (
mo
l.%
)
0
10
20
30
40
50
60
70
Cut 3
Pressure (MPa)
0 1 2 3 4 5 6 7
Pro
pa
ne
so
lub
ilit
y (
mo
l.%
)
10
20
30
40
50
60
70
50 oC
100 oC
150 oC
186 oC
50 oC
100 oC
150 oC
186 oC
50 oC
100 oC
150 oC
186 oC
58
In the next step in order to implement the tuned model for propane/whole bitumen sample, the
bitumen sample has to be characterized. The simple exponential distribution for carbon number
proposed by Pedersen et al. (Pedersen et al., 1992) was used to characterize the bitumen. In this
work, we just assumed the bitumen components as n-alkanes based on simulated distillation result
which does not mean that bitumen is only comprised of n-alkanes.
The procedure of bitumen characterization and calculation of propane solubility in bitumen and
heavy oil using our proposed model is summarized in Figure 3.9.
Figure 3.9: The procedure of bitumen characterization and calculation of propane solubility in
bitumen and heavy oil.
Here, we evaluate the proposed model to calculate propane solubility in two bitumen samples.
These two sample include Athabasca bitumen, which was used in the fractionation process and
Cold Lake bitumen, which is used to test the validity of the developed characterization procedure
and solubility model.
All Athabasca bitumen components (C7 to C138) obtained by Pedersen characterization method
(Pedersen et al., 1992) and their properties were summarized in Table 2.6. The characterized
Athabasca bitumen is then inputted to the model in which the generalized binary interaction
parameters summarized in Table 3.2 were used. In the other words, the tuned model on bitumen
cuts has been employed to calculate the solubility of propane in the whole bitumen sample without
any tuning parameters.
The calculated propane solubility in Athabasca bitumen were compared with experimental
solubility data (Zirrahi et al., 2017) in Figure 3.10.
Carbon distribution of sample
(bitumen/heavy oil) is
obtained from SimDist
Sample is completely
distillable
Sample is not
completely distillable
The exponential distribution of
carbon number is used for
characterization of sample
All components of sample
(assumed as n-alkanes) are
inputted into our proposed model
Critical properties and acentric
factor of each component are
obtained by equations (2.A.1-2.A.5)
Binary interaction coefficients between propane
and each component of sample are calculated
using equation (3.1) and Table 3.2
Applying PR-EoS, propane solubility
in each temperature and pressure is
calculated
59
Figure 3.10: The calculated (solid line) and experimental (dots) solubility data of
propane/Athabasca bitumen system. (AARD, MAD, and AAD are 4.2 %, 6.7 mol%, and 1.8
mol.%, respectively.)
Figure 3.10 confirms that proposed model can predict the propane solubility in bitumen without
requiring any tuning parameters. Although no experimental data of propane/ bitumen was used in
this model, the solubilities have been calculated with AARD of 4.2 %. The average deviation
between our model predictions and experimental data is only 1.8 mol% even when no experimental
solubility data in bitumen has been used for modeling. In the calculations, the carbon number or
boiling point distribution, which is the outcome of simulated distillation, has only been used.
Therefore, having the carbon number or boiling point distribution for a bitumen sample, the
propane solubility can be predicted using the proposed model in this work.
Pressure (MPa)
1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Pro
pa
ne
so
lub
ilit
y (
mo
l.%
)
0
20
40
60
80
100
100 oC
150 oC
190 oC
AARD=4.2 %MAD= 6.7 mol.%AAD=1.8 mol.%
60
Figure 3.11: Evaluation of proposed solubility model: (a) carbon number distribution of Cold
Lake bitumen obtained by SimDist; (b) measured solubility data of propane/Cold Lake bitumen in
this work (symbols) and our model predictions (solid lines). (AARD, MAD, and AAD are 6.5 %,
4.7 mol.%, and 3.0 mol.%, respectively.)
To evaluate the capability of our proposed model to be applied on another bitumen type, a bitumen
sample from Cold Lake reservoir has been considered. The experimental solubility data of propane
in Cold Lake bitumen were measured at three temperatures of 100, 150, and 186 oC and pressures
of 2, 4, and 6 MPa as shown in Figure 3.11(b). The carbon number distribution for Cold Lake
bitumen obtained by simulated distillation was shown in Figure 3.11(a). Applying our
characterization model, Cold Lake bitumen was characterized as a mixture of C7 to C149 (Table
2.6). Then, the proposed solubility model was employed to calculate propane solubility in Cold
Lake bitumen. The measured and predicted solubilities are shown in Figure 3.11(b). The results
shown in Figure 3.11(b) confirmed that the proposed solubility model in this work can adequately
predict the propane solubility in bitumen without using any experimental solubility data of
propane/Cold Lake bitumen system. The maximum and average deviations from experimental
solubility are 4.7 and 3.0 mol.%, respectively, which are acceptable deviations in case of propane
with high solubility in bitumen. The proposed model finds applications in generation of fluid
models for reservoir simulation where no experimental data is available. Simply, by having carbon
number distribution of any bitumen type obtained from simulated distillation, our developed model
Pressure (MPa)
1 2 3 4 5 6 7
Pro
pa
ne
so
lub
ilit
y (
mo
l.%
)
10
20
30
40
50
60
70
80
90
100 oC
150 oC
186 oC
Carbon number
20 40 60 80 100
%O
ff
0
20
40
60
80
100AARD=6.5 %MAD=4.7 mol.%AAD=3.0 mol.%
(a) (b)
61
can calculate the propane solubility in bitumen without requiring experimental data of
propane/bitumen system.
3.6 Summary and Conclusion
In this work, Athabasca bitumen was fractionated to four cuts using a modified vacuum distillation
method. Each obtained bitumen fraction was individually characterized. The three first cuts (called
as Cuts 1, 2, and 3) were completely distillable and the fourth cut was solid like at room
temperature.
In the wide ranges of temperature and pressure (including four temperatures of 50, 100, 150, 186
oC, and pressures up to 6 MPa), solubility, density, and viscosity of propane/bitumen cuts were
measured. The solubility of propane in each bitumen cut increased with increasing pressure and as
a result of propane dissolution, density and viscosity decreased. The propane solubility in the light
fraction (Cut 1) was found more than the heavy ones (Cuts 2 and 3).
The PR-EoS was tuned considering the measured propane solubility in the three distillable cuts at
all available temperatures and pressures. A binary interaction parameter correlation of propane and
bitumen components was proposed as a function of temperature and critical temperatures of
bitumen components. The tuned model represented the propane solubilities in all bitumen cuts
with AARD of 5.4 %. The propane solubility in Athabasca bitumen sample was then calculated
using the tuned model with AARD of 4.2 % without using experimental data of propane/bitumen
system. Therefore, the proposed model in this study, can acceptably predict the propane solubility
in bitumen without requiring experimental data. The only data set used to calculate the propane
solubility was carbon number or boiling point distribution, which is obtained by simulated
distillation test. The capability of proposed model was also confirmed by employing this model to
predict propane solubility in Cold Lake bitumen sample. Our developed model in this work
predicted the propane solubility in Cold Lake bitumen by average deviation of 3 mol.% without
using experimental data of this bitumen.
62
Appendix 3.A: Density of Propane/Bitumen Cuts
Using the density of each component at different temperature and pressure and after calculating
the phase compositions, simulators calculate the density of mixture employing mixing rules. For
propane/bitumen system, density of bitumen can be experimentally measured and input to the
model. However, for propane, effective liquid density is used such that the use of mixing rule
results in the mixture density.
To correlate the density of bitumen cuts, the following correlation is used (Zirrahi et al., 2017);
))exp(exp()(54
2
321TaPaTaTaa (3.A.1)
where, T and P are temperature and pressure in K and MPa, respectively. The fitting parameters
(a1 to a5) are tuned using experimental density data of bitumen cuts (given in Table 2.2) at all
available temperatures and pressures and summarized in Table 3.A.1.
Table 3.A.1: Required parameters for implementation of equation (3.A.1) and the AARDs
between calculated and experimental density of bitumen cuts.
Sample a1 a2 a3 a4 a5 AARD (%) Cut 1 902.67 0.42746 -1.42×10-3 1.11×10-4 5.59×10-3 0.095
Cut 2 1146.18 -0.62321 -7.70×10-6 1.06×10-4 5.16×10-3 0.021
Cut 3 1165.54 -0.61155 -2.04×10-5 1.43×10-4 4.32×10-3 0.077
The comparison between predicted and measured density data of bitumen cuts at different
temperatures and two pressures are shown in Figure 3.A.1.
The densities of bitumen cuts were represented using equation (3.A.1) with AARDs less than 0.095
%. This shows that the developed correlation can accurately calculate the densities of whole
bitumen (Zirrahi et al., 2017) and bitumen cuts (in this work) by capturing the effect of temperature
and pressure on density.
63
Figure 3.A.1: Comparison between the experimental (symbols) and calculated (solid lines) density
of bitumen cuts.
To estimate the effective density of propane, equation (3.A.1) has been considered. The non-linear
mixing rule base on mole fraction which is common in reservoir simulators has been used to
calculate the density of propane/bitumen cuts as follows;
b
s
s
s xx
)1(1 (3.A.2)
where, ρ, ρs, ρb, and xs are mixture density, solvent density, density of bitumen cut, and solvent
mole fraction, respectively. For bitumen cuts, equation (3.A.1) was used to calculate the density.
The effective density of propane was also estimated using equation (3.A.1). The measured density
of propane/bitumen cuts are given in Table 3.1. The tuned parameters to calculate the propane
effective density are summarized in Table 3.A.2.
Table 3.A.2: Required parameters to calculate the effective propane density and AARDs between
calculated and experimental density of propane/bitumen cut systems.
Sample a1 a2 a3 a4 a5 AARD (%) Cut 1 397.12 2.862 -4.74×10-3 -5.332 -1.14×10-2 0.22
Cut 2 565.20 2.487 -4.57×10-3 -22.088 -1.62×10-2 0.35
Cut 3 481.57 2.710 -4.36×10-3 -5.083 -1.26×10-2 0.17
The comparison between calculated and measured density of propane/bitumen cuts were shown in
Figure 3.A.2. The experimental data has been represented by AARDs less than 0.35 %, which
Cut 1
Temperature (oC)
40 60 80 100 120 140 160 180 200
De
ns
ity (
kg
/m3)
780
800
820
840
860
880
900
920
P= 1.1 MPaP= 6.1 MPa
Cut 2
Temperature (oC)
40 60 80 100 120 140 160 180 200D
en
sit
y (
kg
/m3)
840
860
880
900
920
940
960
P= 1.1 MPaP= 6.1 MPa
Cut 3
Temperature (oC)
40 60 80 100 120 140 160 180 200
De
ns
ity (
kg
/m3)
860
880
900
920
940
960
980
P= 1.1 MPaP= 6.1 MPa
64
shows that the above mentioned method based on assuming the effective density of propane is the
reliable method to correlate the density of propane/bitumen cut systems.
Figure 3.A.2: Comparison between the calculated and the experimental density of
propane/bitumen cut systems.
Propane/Cut 1
Experimental density (kg/m3)
600 650 700 750 800 850 900
Pre
dic
ted
den
sit
y (
kg
/m3)
600
650
700
750
800
850
90050 oC100 oC150 oC186 oC
Propane/Cut 2
Experimental density (kg/m3)
750 800 850 900
Pre
dic
ted
den
sit
y (
kg
/m3)
750
800
850
90050
oC
100 oC
150 oC
186 oC
Propane/Cut 3
Experimental density (kg/m3)
750 800 850 900 950
Pre
dic
ted
den
sit
y (
kg
/m3)
750
800
850
900
95050
oC
100 oC
150 oC
186 oC
65
Appendix 3.B: Viscosity of Propane/Bitumen Cuts
To correlate the viscosity of pure samples, the correlation proposed by Mehrotra and Svrcek
(Mehrotra and Svrcek, 1986) was used for Cuts 2 and 3.
PbTbbb 321 ln)ln(ln (3.B.1)
For the light fraction (Cut 1) viscosity calculation, this correlation has been used as one-log type,
PbTbbb 321 lnln (3.B.2)
In the above equations, µb is viscosity of bitumen cuts in mPa.s at absolute temperature T (K) and
pressure P (MPa). These correlations were tuned using experimental viscosity data of bitumen cuts
given in Table 2.2. The values of parameters b1, b2, and b3 were summarized for three bitumen
cuts in Table 3.B.1.
Table 3.B.1: Required parameters for implementation of equations (3.B.1 and 3.B.2) and the
AARDs between calculated and experimental viscosity data of bitumen cuts.
Sample b1 b2 b3 AARD (%) Cut 1 42.717 -7.081 8.5×10-3 8.22
Cut 2 33.761 -5.582 7.3×10-3 10.90
Cut 3 30.174 -4.879 4.9×10-3 8.23
The experimental and correlated viscosity data for Cut 1, Cut 2, and Cut 3 has been compared in
Figure 3.B.1. This figure shows that the tuned model can correlate the viscosity data of bitumen
cuts.
66
Figure 3.B.1: Comparison between experimental (symbols) and calculated (solid lines) viscosity
of bitumen cuts.
Several mixing rules were proposed in literature to calculate the viscosity of solvent/bitumen
system (Glandt and Chapman, 1995). In this work, the log mixing rule based on mole fraction
which is the popular mixing rule for simulators was used as follows:
bbssm xx lnlnln (3.B.3)
where, x is mole fraction and ‘m’, ‘s’, and ‘b’ refer to mixture, solvent, and bitumen cut,
respectively.
To calculate the viscosity of saturated oil using mixing rules, the dissolved solvent viscosity in oil
must be correlated. The effective viscosity for solvent in solvent/oil system was correlated to be
functions of temperature and pressure as follows:
2
65
2
4321 . TcTPcPcTcPccs (3.B.4)
where, µs is viscosity of solvent at pressure P (MPa) and temperature T (K). The fitting parameters
(c1 to c6) have been tuned using the experimental viscosity data of propane/bitumen cuts (given in
Table 3.1) and summarized in Table 3.B.2.
Cut 1
Temperature (oC)
40 60 80 100 120 140 160 180 200
Vis
co
sit
y (
mP
a.s
)
1
10
P= 1.1 MPa
P= 6.1 MPa
Cut 2
Temperature (oC)
40 60 80 100 120 140 160 180 200V
isc
os
ity (
mP
a.s
)
1
10
100
1000
P= 1.1 MPa
P= 6.1 MPa
Cut 3
Temperature (oC)
40 60 80 100 120 140 160 180 200
Vis
co
sit
y (
mP
a.s
)
1
10
100
1000
10000
P= 1.1 MPa
P= 6.1 MPa
67
Table 3.B.2: Required parameters to calculate the effective propane viscosity and AARDs
between calculated and experimental viscosity of propane/bitumen cut systems.
Sample c1 c2 c3 c4 c5 c6 AARD (%) Cut 1 6.583 0.3959 -3.51×10-2 9.96×10-3 -1.16×10-3 4.797×10-5 5.82
Cut 2 -9.690 -0.6919 5.55×10-2 -1.01×10-2 1.73×10-3 -7.548×10-5 3.45
Cut 3 -7.113 0.2063 3.85×10-2 2.14×10-2 -7.5×10-4 -4.756×10-5 7.47
The comparison between calculated and experimental viscosity of propane/bitumen cut systems is
shown in Figure 3.B.2. This figure and the values of AARD in Table 3.B.2 confirm that the
effective viscosity model can properly predict the viscosity of propane/bitumen cut systems.
Figure 3.B.2: Comparison between calculated and experimental viscosity of propane/bitumen cut
systems.
Propane/Cut 1
Experimental viscosity (mPa.s)
0.1 1 10
Pre
dic
ted
vis
co
sit
y (
mP
a.s
)
0.1
1
10
Propane/Cut 2
Experimental viscosity (mPa.s)
0.1 1 10
Pre
dic
ted
vis
co
sit
y (
mP
a.s
)
0.1
1
10
Propane/Cut 3
Experimental viscosity (mPa.s)
0.1 1 10 100
Pre
dic
ted
vis
co
sit
y (
mP
a.s
)
0.1
1
10
100
68
3.7 References
Azinfar, B., Haddadnia, A., Zirrahi, M., Hassanzadeh, H., Abedi, J., 2017. Effect of Asphaltene
on Phase Behavior and Thermophysical Properties of Solvent/Bitumen Systems. J. Chem.
Eng. Data 62, 547–557.
Chueh, P.L., Prausnitz, J.M., 1967. Vapor-liquid equilibria at high pressures: Calculation of Partial
Molar Volumes in Nonpolar Liquid Mixtures. AIChE J. 13, 1099–1107.
Cotterman, R.L., Schwarz, B.J., Prausnitz, J.M., 1986. Molecular Thermodynamics for Fluids at
Low and High Densities. Part I: Pure Fluids Containing Small or Large Molecules. AIChE J.
32, 1787–1798.
Danesh, A., 1998. PVT and Phase Behaviour of Petroleum Reservoir Fluids, First ed., Elsevier
Science, Amesterdam, The Netherlands.
Diaz, O.C., Sánchez-Lemus, M.C., Schoeggl, F.F., Satyro, M.A., Taylor, S.D., Yarranton, H.W.,
2014. Deep-vacuum Fractionation of Heavy Oil and Bitumen, part I: Apparatus and
Standardized Procedure. Energy and Fuels 28, 2857–2865.
Eastick, R.R., Svrcek, W.Y., Mehrotra, A.K., 1992. Phase Behaviour of CO2 -Bitumen Fractions.
Can. J. Chem. Eng. 70, 159–164.
Gao, G., Daridon, J.L., Saint-Guirons, H., Xans, P., Montel, F., 1992. A Simple Correlation to
Evaluate Binary Interaction Parameters of the Peng-Robinson Equation of State: Binary Light
Hydrocarbon Systems. Fluid Phase Equilib. 74, 85–93.
Glandt, C.A., Chapman, W.G., 1995. Effect of Water Dissolution on Oil Viscosity. SPE Reserv.
Eng. 10, 59–64.
Hassanzadeh, H., Harding, T., 2016. Analysis of Conductive Heat Transfer During in-situ
Electrical Heating of Oil Sands. Fuel 178, 290–299.
Huang, S.H., Radosz, M., 1991. Phase Behavior of Reservoir Fluids V: SAFT Model of CO2 and
Bitumen Systems. Fluid Phase Equilib. 70, 33–54.
Huang, S.H., Radosz, M., 1990a. Phase Behavior of Reservoir Fluids II: Supercritical Carbon
69
Dioxide and Bitumen Fractions. Fluid Phase Equilib. 60, 81–98.
Huang, S.H., Radosz, M., 1990b. Equation of State for Small, Large, Polydisperse, and
Associating Molecules. Ind. Eng. Chem. Res. 29, 2284–2294.
Jiang, Q., 1997. Recovery of Heavy Oil and Bitumen Using Vapex Process in Homogeneous and
Heterogenous Reservoirs. PhD Thesis, University of Calgary, Calgary, Alberta, Canada.
Kariznovi, M., Nourozieh, H., Abedi, J., 2010. Bitumen Characterization and Pseudocomponents
Determination for Equation of State Modeling. Energy Fuels 24, 624–633.
Kokal, S.L., Sayegh, S.G., 1993. Phase Behavior and Physical Properties of CO2-saturated Heavy
Oil and its Constitutive Fractions: Experimental Data and Correlations. J. Pet. Sci. Eng. 9,
289–302.
Kokal, S.L., Sayegh, S.G., 1989. Phase Behavior Correlation of CO2/Heavy Oil Mixtures for
Enhanced Oil Recovery. Fluid Phase Equilib. 52, 283–290.
Lee, B.I., Kesler, M.G., 1975. A Generalized Thermodynamic Correlation Based on Three-
Parameter Corresponding States. AIChE J. 21, 510–527.
Marufuzzaman, M., Henni, A., 2015. Solubility of CO2 and C2 H6 in Heavy Oil and its SARA
Fractions. Can. J. Chem. Eng. 93, 553–564.
Mehrotra, A.K., Svrcek, W.Y., 1988. Characterization of Athabasca Bitumen for Gas Solubility
Calculations. J. Can. Pet. Technol. 27, 107–110.
Mehrotra, A.K., Svrcek, W.Y., 1986. Viscosity of Compressed Athabasca Bitumen. Can. J. Chem.
Eng. 64, 844–847.
Michelsen, M.L., 1982a. The Isothermal Flash Problem. Part I. Stability. Fluid Phase Equilib. 9,
1–19.
Michelsen, M.L., 1982b. The Isothermal Flash Problem. Part II. Phase-split Calculation. Fluid
Phase Equilib. 9, 21–40.
Nasr, T.N., Ayodele, O.R., 2006. New Hybrid Steam-Solvent Processes for the Recovery of Heavy
Oil and Bitumen, Paper SPE 101717 Presented at International Petroleum Exhibition and
70
Conference, Abu Dhabi, U.A.E., 5–8 November.
Nenniger J., Holcek R., Dillon J., Wolff V., 2013. Solvent Injection Plant for Enhanced Oil
Recovery and Method of Operating same, Canadian Patent CA2777966 A1.
Nourozieh, H., Kariznovi, M., Abedi, J., 2015. Experimental and Modeling Studies of Phase
Behavior for Propane/Athabasca Bitumen Mixtures. Fluid Phase Equilib. 397, 37–43.
Patel, N.C., Teja, A.S., 1982. A New Cubic Equation of State for Fluids and Fluid Mixtures. Chem.
Eng. Sci. 37, 463–473.
Pedersen, K.S., Blilie, A.L., Meisingset, K.K., 1992. PVT Calculations on Petroleum Reservoir
Fluids Using Measured and Estimated Compositional Data for the Plus Fraction. Ind. Eng.
Chem. Res. 31, 1378–1384.
Peng, D.Y., Robinson, D.B., 1976. A New Two-Constant Equation of State. Ind. Eng. Chem.
Fundam. 15, 59–64.
Riazi, M.R., 1989. Distribution Model for Properties of Hydrocarbon-Plus Fractions. Ind. Eng.
Chem. Res. 28, 1731–1735.
Riazi, M.R., 2005. Characterization and Properties of Petroleum Fractions, First ed., ASTM
International, U.S.A.
Riazi, M.R., Al-Sahhaf, T.A., 1996. Physical Properties of Heavy Petroleum Fractions and Crude
Oils. Fluid Phase Equilib. 117, 217–224.
Sayegh, S.G., Rao, D.N., Kokal, S., Najman, J., 1990. Phase Behaviour and Physical Properties
Of Lindbergh Heavy Oil/CO2 Mixtures. J. Can. Pet. Technol. 29, 31–39.
Soave, G., 1972. Equilibrium Constants from a Modified Redlich-Kwong Equation of State.
Chem. Eng. Sci. 27, 1197–1203.
Twu, C.H., 1984. An Internally Consistent Correlation for Predicting The Critical Properties and
Molecular Weights of Petroleum and Coal-tar Liquids. Fluid Phase Equilib. 16, 137–150.
Varotsis, N., Stewart, G., Todd, A.C., Clancy, M., 1986. Phase Behavior of Systems Comprising
North Sea Reservoir Fluids and Injection Gases. J. Pet. Technol. 38, 1221–1233.
71
Whitson, C.H., 1983. Characterizing Hydrocarbon Plus Fractions. Soc. Pet. Eng. J. 23, 683–694.
Zirrahi, M., Hassanzadeh, H., Abedi, J., 2017. Experimental and Modeling Studies of MacKay
River Bitumen and Light n-Alkane Binaries. Can. J. Chem. Eng. 95, 1417–1427.
72
Chapter Four: Phase Behaviour of Methane- and Ethane-Bitumen Fractions
4.1 Preface
This chapter will be presented at the 2018 SPE Canada Heavy Oil Conference entitled
“Generalized Approach to Predict k-values of Hydrocarbon Solvent/Bitumen Mixtures”. This
manuscript was co-authored by M. Zirrahi, H. Hassanzadeh, and J. Abedi.
Since this dissertation has been prepared on paper-based format, unavoidably, there are some
repetitive parts in each chapter, mainly Chapters 2, 3, and 4, such as bitumen fractionation and
solubility model description.
In this chapter, all the measured experimental data of methane-each bitumen fraction and ethane-
each bitumen fraction followed by the results of the model are presented. The density and viscosity
of the pure bitumen fractions and more details regarding fractionation procedure were presented
in Chapter 2.
4.2 Abstract
A generalized model is presented to calculate the k-values of methane/bitumen and ethane/bitumen
systems. These data are required for phase behaviour modeling and simulation of solvent-aided
bitumen recovery processes. The proposed model is evaluated by comparing the calculated results
with the measured experimental k-values.
The proposed model provides generalized binary interaction parameters between hydrocarbon
solvents (methane and ethane) and the defined components in bitumen and calculates the k-values
of solvent/bitumen systems. Unlike the existing common approaches, experimental solubility data
are not required to tune the model. The boiling point or carbon number distribution of bitumen or
heavy oil obtained by simulated distillation (SimDist) test is the only required data to characterize
and define the components of heavy oil or bitumen. The SimDist test is a very fast test and much
less expensive than the common solubility measurements.
73
This model has been developed based on the experimental fractionation of bitumen. The Athabasca
bitumen was experimentally fractioned to four bitumen cuts applying vacuum distillation method
and the solubility of solvent in each bitumen cut were measured at wide ranges of temperature and
pressure. The measured solubility data of methane and ethane in each bitumen cut have been used
to tune the PR-EoS and the generalized binary interaction parameter coefficients for each solvent
and bitumen components have been found. To calculate the k-values of solvent/bitumen mixtures,
the bitumen is defined as a mixture of n-alkanes based on simulated distillation results. The
properties of n-alkanes have been assumed for each component. Employing the obtained binary
interaction parameters in PR-EoS using experimental data of solvent/bitumen cut systems and
considering the defined bitumen components as input to the proposed model, the k-values of
solvent and any bitumen or heavy oil mixtures are calculated. The validity of the proposed model
has been confirmed by calculating the k-values of methane and ethane with two different bitumen
samples with an average deviation of less than 3.0 %. The outputs of this model can be directly
used as k-values in simulation of solvent-aided thermal recovery processes.
4.3 Introduction
4.3.1 Why is study of solvent-based recovery processes important?
Although steam-based recovery methods have been successfully used in the Canadian Oil Sands
for the past decades, they require very high energy to operate, which makes them uneconomical
especially with the current low oil price. In addition, these processes result in high level of
greenhouse gas emission. Recent environmental restriction, such as Carbon Tax regulations,
requires the Oil Sands companies develop new technology to lower the fuel consumptions and
increase the fuel efficiency. Therefore, high level of greenhouse gas emission, substantial water
requirement to generate steam, and cost of water treatment make the steam injection technology
environmentally and economically not efficient.
The concerns regarding the thermal processes directed more attention to the solvent-based
processes. Solvent reduces the bitumen viscosity while using less energy and producing less
greenhouse gas. In addition, the injected solvent can be recovered from the produced bitumen and
reinjected to the reservoir, which makes this process economically attractive. However, low
74
production rate is always the challenge in solvent injection processes. To mitigate this concern,
the hybrid processes are suggested in which solvent and steam are co-injected to take the
advantages of both steam and solvent injections. The design and development of economical
solvent injection processes require phase behaviour study of the solvent/bitumen system.
Moreover, the phase behaviour studies are vital for pipeline transportation, surface upgrading and
refinery processes. The amount of dissolved solvent in the oil phase at particular temperature and
pressure, the thermos-physical properties of oil such as density and viscosity in the presence of
solvent are the required data to develop the fluid flow model and simulation of the solvent-based
recovery processes.
4.3.2 Why is methane and ethane considered in bitumen recovery processes?
Non-condensable gases can be co-injected with steam. Addition of non-condensable gases to steam
may cause insulation effect at the top of the steam chamber, which reduces the heat loss to the
overburden (Canbolat et al., 2004). The non-condensable gas are mainly methane and carbon
dioxide. Methane was also used to improve the SAGD efficiency. Butler (Butler, 1999) and Jiang
et al. (Jiang et al., 1998) suggested that addition of a small amount of methane improves the SAGD
performance by providing high production and recoveries at lower steam-oil-ratio (SOR).
However, the idea of addition of non-condensable gases to SAGD is still controversial with regards
to the behaviour of these gases in the steam chamber and their impact on ultimate bitumen recovery
(Heron, 2008).
Ethane is also a potential candidate for additive of steam-based recovery processes. Ethane with
higher solubility than methane can diffuse and dissolve in bitumen resulting in lower viscosity and
higher oil production and recovery rates (Kariznovi et al., 2017). In addition to improved oil
recovery, ethane can be also involved in oil upgrading and supercritical extraction processes (Rose
et al., 2001).
To study the co-injection process of methane and ethane to steam-based recovery processes, their
effects on bitumen properties, and role of ethane in supercritical extraction processes for surface
upgrading, the phase behaviour and thermodynamic properties of methane- and ethane-bitumen
mixtures have to be determined.
75
4.3.3 Why is developing the generalized model to predict k-values of solvent/bitumen mixture
necessary and what is the major contribution of the present study?
One of the necessary data used to design the solvent-based recovery methods is the k-value of
components in the mixture. The typical approach for prediction of k-values in solvent/bitumen
system is tuning an equation of state using the available solubility data of solvent in bitumen
(Kariznovi et al., 2010; Zirrahi et al., 2017) and finding the binary interaction parameters (BIP)
between solvent and bitumen. This way, a model is tuned using the measured experimental
solubility data in the range of temperature and pressure and can be then employed to find the
solubility data at other pressure and temperature conditions where the required experimental data
are not available. This typical approach is hampered by limited, time consuming, and expensive
experimental solubility data of solvent/bitumen system.
The aim of this study is to generate the phase behaviour data of solvent (methane and
ethane)/bitumen fractions at wide ranges of temperature and pressure and more importantly to
develop a generalized model capable of predicting the k-values of solvent/bitumen system. The
idea is to fractionate a bitumen sample into several cuts, find carbon number and boiling point
distributions of each individual cut using simulated distillation (SimDist) and then study phase
behaviour of each solvent/bitumen cut separately. Once, all components are defined for all cuts,
using experimental solubility data of solvent and each bitumen cut, the generalized binary
interaction parameters between the solvent (methane and ethane) and the bitumen components are
established. The solubility of solvent in bitumen and k-values are then calculated using the
generalized binary interaction parameters. In this work, we show that the binary interaction
parameters obtained using this approach can be generalized and applicable to other solvent
/bitumen systems. In other words, once the generalized binary interaction parameters are
established the time consuming and costly experimental measurements of solubility data can be
avoided.
In this chapter, the experimental phase behaviour data of methane- and ethane-bitumen fractions
at temperatures up to 186 oC and pressures up to 6 MPa are measured. The measured solubility
data of methane and ethane in bitumen fractions are used to develop the generalized model to
predict the k-values of methane-bitumen and ethane-bitumen mixtures. Moreover, the calculated
76
k-values of propane-bitumen and butane-bitumen systems (data from Chapters 2 and 3) are
compared with the experimental values.
4.4 Experimental Section
4.4.1 Materials
Methane and ethane were supplied by Praxair with purities of 99.999 and 99.95 mol.%,
respectively. Athabasca (MW=569 g/mol) and Cold Lake (MW=546 g/mol) bitumens were
provided by oil production companies in Alberta, Canada.
4.4.2 Bitumen Fractionation
The Athabasca bitumen was fractionated to four cuts using the modified vacuum distillation
system. Figure 4.1 summarizes the fractionation scheme considered in this work.
Figure 4.1: The bitumen fractionation scheme in this work.
Applying three batch vacuum distillations at 350, 250, and 195 oC, bitumen was fractionated to
four cuts. More details regarding the distillation apparatus, and each fraction’s properties can be
found in Chapter 2.
Bitumen
Light fraction Heavy fraction (Cut 4)
Light fraction Heavy fraction (Cut 3)
Light fraction (Cut 1) Heavy fraction (Cut 2)
350 oC
250 oC
195 oC
77
4.4.3 Phase Behaviour Data Measuring
The phase behaviour data of solvent/bitumen fractions including solvent solubility, density, and
viscosity of liquid phase were measured using our PVT apparatus described in Chapter 5 (Azinfar
et al., 2017). The solvent and bitumen fraction were mixed in equilibrium cell using a rocking
system. The equilibrium properties were measured once the mixture of gaseous solvent and liquid
(bitumen or its cuts) reached an equilibrium. The density and viscosity of liquid phase were
measured using an inline densitometer (Anton Paar) and viscometer (Viscopro 2000). The k-value
was calculated using the measured solvent solubility in bitumen fractions. The solvent solubility
in liquid phase was found by measuring the volume of evolved gas when the saturated liquid is
flashed at atmospheric conditions using a gasometer (Chandler Engineering, Model 2331).
4.5 Results and Discussion
4.5.1 Experimental Results
The phase behaviour data measured in this work for methane- and ethane-Athabasca bitumen cuts
are summarized in Table 4.1. The measured density and viscosity data of pure bitumen fractions
at temperatures up to 186 oC and pressures up to 8 MPa can be found in Chapter 2.
Table 4.1: Experimental vapour/liquid equilibrium properties for methane-bitumen cut and
ethane-bitumen cut mixtures.
Temperature
(ºC)
Pressure
(MPa)
Density
(kg/m3)
Viscosity
(mPa.s)
Solubility
(wt.%)
Density
(kg/m3)
Viscosity
(mPa.s)
Solubility
(wt.%)
Methane/Cut 1 Ethane/Cut 1
50 1.5 893.8 5.25 0.35 856.6 2.63 4.60
50 3.0 889.0 4.67 0.70 816.3 1.33 9.81
50 4.5 884.5 4.21 1.01 765.4 0.79 14.90
50 6.0 880.0 3.80 1.39 691.6 0.42 23.91
100 1.5 862.0 1.65 0.32 837.2 1.37 2.55
100 3.0 857.7 1.55 0.65 815.8 1.19 4.11
100 4.5 853.7 1.51 0.90 794.4 0.77 6.92
100 6.0 849.8 1.43 1.22 774.2 0.59 10.08
150 1.5 827.3 0.88 0.27 815.5 1.00 1.36
150 3.0 823.2 0.82 0.56 799.4 0.62 3.17
150 4.5 818.9 0.77 0.83 784.6 0.55 4.81
150 6.0 815.0 0.74 1.10 765.75 0.47 6.45
186 1.5 799.0 0.58 0.28 788.8 0.53 1.32
78
186 3.0 795.0 0.54 0.56 775.7 0.46 2.47
186 4.5 791.1 0.53 0.85 762.8 0.414 3.72
186 6.0 787.3 0.51 1.11 747.4 0.37 5.44
Methane/Cut 2 Ethane/Cut 2
50 1.5 944.4 73.19 0.21 918.5 19.48 3.19
50 3.0 940.5 59.50 0.43 887.62 8.86 5.94
50 4.5 937.0 49.74 0.77 853.5 3.94 9.73
50 6.0 933.3 42.00 1.03 814.8 1.74 13.57
100 1.5 913.6 8.18 0.13 895.0 5.49 1.58
100 3.0 910.3 7.88 0.36 878.5 3.99 3.11
100 4.5 906.9 6.88 0.67 861.1 2.68 4.62
100 6.0 903.6 6.05 0.93 843.7 2.12 6.53
150 1.5 880.0 2.60 0.19 870.2 2.15 0.79
150 3.0 876.8 2.43 0.39 858.1 1.82 2.01
150 4.5 873.3 2.29 0.59 845.4 1.45 3.57
150 6.0 870.2 2.18 0.81 830.1 0.81 4.54
186 1.5 855.1 1.43 0.21 848.9 1.35 0.72
186 3.0 852.2 1.35 0.44 838.6 1.11 1.75
186 4.5 849.1 1.27 0.66 827.1 0.95 2.95
186 6.0 846.4 1.25 0.84 816.2 0.82 3.80
Methane/Cut 3 Ethane/Cut 3
50 1.5 961.9 966.8 0.24 941.0 147.4 2.38
50 3.0 958.1 902.7 0.47 912.0 37.65 5.46
50 4.5 955.0 640.2 0.69 881.5 12.92 8.89
50 6.0 951.8 551.9 0.90 850.3 6.72 12.52
100 1.5 932.4 31.53 0.21 919.5 20.15 1.62
100 3.0 929.2 27.99 0.44 904.0 12.06 3.34
100 4.5 926.4 23.61 0.62 888.4 7.19 4.92
100 6.0 923.7 21.57 0.81 872.7 5.07 5.96
150 1.5 900.5 6.30 0.21 892.3 5.14 0.90
150 3.0 897.6 5.85 0.41 881.1 4.05 2.22
150 4.5 894.9 5.65 0.64 870.1 3.60 3.35
150 6.0 892.4 5.59 0.75 858.9 2.63 4.16
186 1.5 877.3 3.00 0.19 870.4 2.63 0.82
186 3.0 874.3 2.80 0.39 861.0 2.26 1.82
186 4.5 871.7 2.51 0.57 851.5 2.21 2.76
186 6.0 868.9 2.45 0.73 842.4 1.68 3.43
To show the trends of the measured data, the phase behaviour data of methane- and ethane-Cut 3
systems including solvent solubility, the density, and viscosity of liquid phase are plotted in
Figures 4.2 and 4.3 panels (a to c), respectively. By increasing the pressure, more solvent is
dissolved in bitumen cuts and the solubility increases as depicted in Figures 4.2(a) and 4.3(a).
79
However, compare to the heavier solvents, the solubility of methane is lower, specifically at higher
temperature. To describe the density and viscosity variations of solvent-saturated bitumen cuts,
two competitive factors should be considered including; temperature and solvent dissolution.
Lower solvent solubility at higher temperatures results in higher density and viscosity of the liquid
phases. However, the effect of temperature in lowering density and viscosity is dominant.
Therefore, the overall effect of solubility and temperature results in lower density and viscosity of
bitumen cuts at higher temperatures as shown in Figures 4.2(b) and 4.2(c). Since solubility of
methane in Cut 3 is very low, the dissolution effect appears to be insignificant and the effect of
temperature on viscosity and density is dominant. Thus, it is expected to observe lower density
and viscosity at higher temperatures. However, when ethane is considered as a solvent (Figure 4.3)
and as pressure increases the liquid phase density at 50 oC is lower than 100 oC (Figure 4.3(b))
indicating the importance of ethane dissolution effect in lowering density of the liquid phase.
Figure 4.2: The experimental phase behaviour data of methane/Cut 3 mixture; (a) methane
solubility, (b) methane-saturated density, and (c) methane-saturated viscosity.
Pressure (MPa)
1 2 3 4 5 6 7
Meth
an
e s
olu
bil
ity (
wt.
%)
0.0
0.2
0.4
0.6
0.8
1.0
50 oC100 oC150 oC186 oC
Pressure (MPa)
1 2 3 4 5 6 7
Den
sit
y (
kg
/m3
)
860
880
900
920
940
960
980
100050 oC100 oC150 oC186 oC
Pressure (MPa)
1 2 3 4 5 6 7
Vis
co
sit
y (
mP
a.s
)
0.1
1
10
100
1000
100 oC
150 oC
186 oC
50 oC
80
Figure 4.3: The experimental phase behaviour data of ethane/Cut 3 mixture; (a) ethane solubility,
(b) ethane-saturated density, and (c) ethane-saturated viscosity.
The measured phase behaviour data of methane- and ethane-each bitumen cut at constant
temperature of 186 and 50 oC are compared in Figures 4.4 and 4.5, respectively. As shown in
Figures 4.4(a) and 4.5(a), more solvent (methane and ethane) is dissolved in lighter bitumen
fraction. Moreover, the density and viscosity of methane- and ethane-saturated Cut 1 are lower
than Cuts 2 and 3 as shown in Figures 4.4(b,c) and 4.5(b,c).
Figure 4.4: Experimental phase behaviour data of methane/bitumen cut systems at 186 oC; (a)
methane solubility, (b) methane-saturated density, and (c) methane-saturated viscosity.
Pressure (MPa)
1 2 3 4 5 6 7
Eth
an
e s
olu
bil
ity (
wt.
%)
0
2
4
6
8
10
12
14
T=50 oCT=100 oCT=150 oCT=186 oC
Pressure (MPa)
1 2 3 4 5 6 7
De
ns
ity (
kg
/m3)
820
840
860
880
900
920
940
960
T=50 oCT=100 oCT=150 oCT=186 oC
Pressure (MPa)
1 2 3 4 5 6 7
Vis
co
sit
y (
mP
a.s
)
1
10
100
1000
T=50 oCT=100 oCT=150 oCT=186 oC
(a) (b) (c)
Pressure (MPa)
1 2 3 4 5 6 7
Vis
co
sit
y (
mP
a.s
)
0.1
1
10
Pressure (MPa)
1 2 3 4 5 6 7
De
ns
ity (
kg
/m3)
760
780
800
820
840
860
880
900
920
940
Cut 1
Cut 2
Cut 3
Pressure (MPa)
1 2 3 4 5 6 7
Me
tha
ne
so
lub
ilit
y (
wt.
%)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
(a) (b) (c)
Cut 1
Cut 2
Cut 3
Cut 1
Cut 2
Cut 3
81
Figure 4.5: Experimental phase behaviour data of ethane/bitumen cut systems at 50 oC; (a) ethane
solubility, (b) ethane-saturated density, and (c) ethane-saturated viscosity.
4.5.2 Description of k-value Model and Results
Normally, k-values are measured experimentally for a range of temperature and pressure. Then,
using these measured data, a thermodynamic model is tuned and used to calculate the k-values
where the experimental data are not available. However, obtaining the experimental k-value of
components for the systems containing bitumen is very expensive and time consuming. Therefore,
developing a model that can calculate these values without requiring extensive PVT data is
essential.
Our aim in this work is developing a model that can predict the k-values of methane- and ethane-
bitumen systems using the simple and fast characterization tests. The experimental data and
modeling results of propane-bitumen and butane-bitumen system were presented in Chapters 3
and 2, respectively. The generalized binary interaction parameter coefficients between solvent
(methane or ethane) and each component of bitumen are developed in this work. Boiling
point/carbon number distribution of bitumen, which is simply obtained by simulated distillation
test, is the only required data to calculate the k-values using the proposed model. After defining
the bitumen components using simulated distillation results and establishing the binary interaction
parameter coefficients based on experimental solubility data of solvent in each individual cut, the
k-values are calculated. Since the three distillable bitumen fractions (Cuts 1, 2, and 3) cover wide
ranges of components in bitumen, it is expected to obtain a generalized model that can be applied
Pressure (MPa)
1 2 3 4 5 6 7
Vis
co
sit
y (
mP
a.s
)
0.1
1
10
100
1000
Cut 1
Cut 2
Cut 3
Pressure (MPa)
1 2 3 4 5 6 7D
en
sit
y (
kg
/m3)
650
700
750
800
850
900
950
1000
Cut 1
Cut 2
Cut 3
Pressure (MPa)
1 2 3 4 5 6 7
Eth
an
e s
olu
bil
ity (
wt.
%)
0
5
10
15
20
25
Cut 1
Cut 2
Cut 3
(a) (b) (c)
82
to the vapour-liquid equilibrium calculations of any solvent/bitumen systems. Later, we validate
this hypothesis by applying the developed model to other bitumens.
In the following, first, the procedure to develop the binary interaction coefficients between solvent
and bitumen components are described. Then, the developed coefficients and the simulated
distillation results of two bitumen types are used to calculate the solubility and k-values of
solvent/bitumen system.
The simulated distillation results of distillable bitumen fractions are shown in Figure 4.6. These
fractions are completely distillable and can be defined as mixtures of n-alkanes because the
simulated distillation test has been developed and calibrated based on the mixture of n-alkanes
known as test standard. That way, the properties of n-alkanes such as boiling point and molecular
weight are assumed for each component in the mixture. The details of each component and its
properties are presented in Chapter 2. After defining the components, the critical properties and
acentric factor should be calculated for each component. Perturbation Expansion Correlations were
used to calculate the critical properties (Twu, 1984). The acentric factor was estimated using Lee-
Kesler correlation (Lee and Kesler, 1975). The details of the equation used in this model was also
presented in Chapter 2. Optimization toolbox of MATLAB R2013a was used for optimization in
this work.
Figure 4.6: The simulated distillation results of Athabasca bitumen fractions.
Carbon Number
20 40 60 80 100
wt.
%
0
2
4
6
8
10
Cut 1Cut 2Cut 3
83
The binary interaction parameter between solvent (methane/ethane) and each component of
bitumen cuts is calculated using the following equation;
)(
3/13/1
6/16/1
321
54
624
624
624 )(
)(2
1
jc
jHorCCH
jHorCCH
j
TBB
cc
cc
rjHorCCH TT
TT
BTBBk
(4.1)
where Tc (K) is the critical temperature. B1 to B5 are the coefficients. The coefficients in equation
(4.1) were adjusted to minimize the sum of square of the differences between the measured and
the calculated solubility data of solvent in three bitumen cuts. The obtained parameters for methane
and ethane using the solubility data (summarized in Table 4.1) are given in Table 4.2 for methane
and ethane as solvent.
Table 4.2: The binary interaction parameter coefficients between solvent (methane and ethane)
and components of bitumen cuts.
Solvent B1 B2 B3 B4 B5 AARD*
(%)
Methane -2.1487 0.2071 -2.6402 -46.6555 0.0314 5.97
Ethane -0.6581 0.2506 -4.1999 -28.0113 0.0215 5.57
* AARD (Average Absolute Relative Deviation)
n
i
calc xxxn 1
expexp /)(1
Figures 4.7(a) and 4.7(b) compare the experimental and calculated methane and ethane solubilities
in Athabasca bitumen fractions, respectively. These figures show the good agreement between the
calculated and the measured solvent solubility in the bitumen fractions.
84
Figure 4.7: Comparison of the calculated (a) methane and (b) ethane solubility in Athabasca
bitumen cuts using tuned model with the measured solubility data in this work.
After finding the binary interaction parameters, the solubility of solvent (methane/ethane) in
bitumen sample can be calculated. The procedure of calculating the solvent solubility in any
bitumen sample is summarized in the following steps;
Step 1. Provide the extrapolated simulated distillation results of bitumen sample to cover 100%
Off.
Step 2. Define the bitumen components as mixture of n-alkanes based on carbon number
distribution obtained by simulated distillation.
Step 3. Input all the defined components with estimation of their critical properties and the acentric
factor into EoS model.
Cut 1
Pressure (MPa)
1 2 3 4 5 6 7
Meth
an
e s
olu
bil
ity (
mo
l.%
)
0
5
10
15
20
25
T=50 oC
T=100 oC
T=186 oC
T=150 oC
Cut 2
Pressure (MPa)
1 2 3 4 5 6 7M
eth
an
e s
olu
bil
ity (
mo
l.%
)
0
5
10
15
20
25
T=50 oC
T=100 oC
T=186 oC
T=150 oC
Cut 3
Pressure (MPa)
1 2 3 4 5 6 7
Meth
an
e s
olu
bil
ity (
mo
l.%
)
0
5
10
15
20
25
T=50 oC
T=100 oC
T=186 oC
T=150 oC
(a) Methane solubility
Cut 1
Pressure (MPa)
1 2 3 4 5 6 7
Eth
an
e s
olu
bil
ity (
mo
l.%
)
0
10
20
30
40
50
60
70
80
T=50 oC
T=100 oC
T=150 oC
T=186 oC
Cut 2
Pressure (MPa)
1 2 3 4 5 6 7
Eth
an
e s
olu
bil
ity (
mo
l.%
)
0
10
20
30
40
50
60
70
80
T=50 oC
T=100 oC
T=150 oC
T=186 oC
Cut 3
Pressure (MPa)
1 2 3 4 5 6 7E
tha
ne
so
lub
ilit
y (
mo
l.%
)0
10
20
30
40
50
60
70
80
T=50 oC
T=100 oC
T=150 oC
T=186 oC
(b) Ethane solubility
85
Step 4. Calculate the binary interaction parameter between solvent and each component of sample
using the equation (4.1) and Table 4.2.
Step 5. Carry out the flash calculation and calculate the mole fraction of solvent in liquid phase.
The simulated distillation results of Athabasca bitumen and the extrapolated distribution using the
Pedersen’s model (Pedersen et al., 1992) are shown in Figure 4.8. The obtained distribution shown
in this figure and the binary interaction parameters in Table 4.2 are used to calculate the solvent
solubility in bitumen.
Figure 4.8: The distribution of Athabasca bitumen components obtained by simulated distillation
and extrapolation using Pedersen’s model (Pedersen et al., 1992).
The calculated methane and ethane solubility in Athabasca bitumen and the measured ones are
compared in Figures 4.9(a) and 4.9(b), respectively. The proposed model predicts the methane and
ethane solubilities in Athabasca bitumen with the average absolute deviation of 1.0 and 1.7 mol.%,
respectively, even when no experimental solubility data were used to tune the model. The results
show that the proposed methodology provides a reliable tool to calculate the solubility of methane
and ethane in bitumen. It is worth noting that previously we have shown that our model is able to
predict the solubility of propane and butane in bitumen (Chapters 3 and 2).
Carbon number
0 20 40 60 80 100 120 140
Mo
le f
rac
tio
n
0.00
0.01
0.02
0.03
0.04
0.05
Simulated distillation resultExtrapolated
86
Figure 4.9: Solubility calculation using the proposed model in this work; (a) methane and (b)
ethane solubility in Athabasca bitumen. (The experimental methane and ethane solubility data
were extracted from Zirrahi et al. (Zirrahi et al., 2017))
To evaluate the proposed model, we test it using different bitumen type and predict the methane
and ethane solubility in Cold Lake bitumen sample. Solubilities of methane and ethane in Cold
Lake bitumen have been measured using our PVT apparatus described in Chapter 5 (Azinfar et al.,
2017). Figure 4.10 shows the carbon number distribution of Cold Lake bitumen obtained from
simulated distillation test and the calculated one using Pedersen’s model (Pedersen et al., 1992).
Pressure (MPa)
1 2 3 4 5
Me
tha
ne
so
lub
ilit
y (
mo
l.%
)
0
5
10
15
20
25
30
T=100 oC
T=150 oC
T=190 oC
AAD=1.0 mol.%MAD=2.4 mol.%AARD=7.8 %
Pressure (MPa)
1 2 3 4
Eth
an
e s
olu
bilit
y (
mo
l.%
)
0
10
20
30
40
50
60
T=100 oC
T=150 oC
T=190 oC
AAD=1.7 mol.%MAD=3.6 mol.%AARD=9.4 %
(a) (b)
87
Figure 4.10: The distribution of Cold Lake bitumen components obtained by simulated distillation
and extrapolation using Pedersen’s model (Pedersen et al., 1992).
The experimental data along with the results of the proposed model are shown in Figure 4.11. The
calculated and experimental solubilities of methane and ethane in bitumen show 0.89 and 2.0
mol.% deviation (AAD), respectively. These levels of deviations confirm the ability of our model
to predict the methane and ethane solubility in bitumen.
Figure 4.11. Comparison of the calculated and the measured (a) methane and (b) ethane solubility
in Cold Lake bitumen in this work.
Carbon number
0 20 40 60 80 100 120 140 160
Mo
le f
racti
on
0.00
0.01
0.02
0.03
0.04
0.05
Simulated distillationExtrapolated
Pressure (MPa)
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5
Me
tha
ne
so
lub
ilit
y (
mo
l.%
)
0
5
10
15
20
25
30
T=100 oC
T=150 oC
T=186 oC
Pressure (MPa)
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5
Eth
an
e s
olu
bilit
y (
mo
l.%
)
0
10
20
30
40
50
60
T=100 oC
T=150 oC
T=186 oC
(a) (b)AAD=0.89 mol.%MAD=1.8 mol.%AARD=9.6 %
AAD=2.0 mol.%MAD=4.9 mol.%AARD=10.9 %
88
The k-values of solvent/bitumen systems are used for thermodynamic modeling and equilibrium
calculations. Therefore, the k-values of components are measured based on PVT experiments and
applied into reservoir simulation software. The k-values are calculated by:
624
624
HorCCH
HorCCH
x
yk
(4.2)
where y and x are the mole fractions of solvent in equilibrium vapour and liquid phases,
respectively. For solvent/bitumen systems, the k-values of bitumen components were considered
zero because bitumen is commonly assumed non-volatile (Kariznovi et al., 2017; Nourozieh et al.,
2016). The vapour phases were analyzed by Varian GC 3900 gas chromatography system and the
results showed that the vapour phases were virtually pure solvent (methane or ethane), which
means ysolvent =1 (Kariznovi et al., 2017; Nourozieh et al., 2016). The measured solvent solubility
(Table 4.1) along with the measured molecular weight of each sample are used to calculate xsolvent.
The measured and the calculated k-values for the methane- and ethane-Athabasca bitumen are
summarized in Table 4.3 and plotted as a function of pressure in Figure 4.12.
Table 4.3: The measured and the predicted k-values of methane- and ethane-bitumen systems (The
measured k-values of Athabasca bitumen mixtures were extracted from (Zirrahi et al., 2017))
T
(ºC)
P
(MPa)
Measured
k-value
Predicted
k-value
Absolute
deviation
T
(ºC)
P
(MPa)
Measured
k-value
Predicted
k-value
Absolute
deviation
Athabasca Bitumen-Methane Athabasca Bitumen-Ethane
100 1.69 11.38 12.53 1.15 100 1.14 7.43 6.34 1.09
100 2.48 8.28 8.75 0.47 100 2.03 3.86 3.77 0.09
100 3.17 6.15 6.99 0.84 100 2.86 2.96 2.83 0.13
100 4.41 4.61 5.21 0.6 100 4.17 2.10 2.11 0.01
150 1.17 15.83 19.12 3.29 150 1.28 8.94 7.57 1.37
150 1.93 10.60 11.85 1.25 150 2.17 5.02 4.68 0.34
150 2.66 7.88 8.78 0.9 150 3.17 3.61 3.38 0.23
150 3.93 5.67 6.15 0.48 150 4.24 2.92 2.67 0.25
190 1.86 12.12 12.45 0.33 190 1.79 8.40 6.43 1.97
190 2.69 8.70 8.81 0.11 190 2.69 4.77 4.46 0.31
190 3.72 6.39 6.56 0.17 190 3.52 3.73 3.54 0.19
190 4.76 5.47 5.27 0.2 190 4.28 3.11 3.01 0.1
Cold Lake Bitumen-Methane Cold Lake Bitumen-Ethane
100 1.0 21.92 20.62 1.3 100 1.0 7.59 7.10 0.49
100 2.0 10.92 10.63 0.29 100 2.0 4.20 3.80 0.4
89
100 4.0 6.28 5.64 0.64 100 4.0 2.42 2.16 0.26
150 1.0 25.11 22.17 2.94 150 1.0 10.48 9.47 1.01
150 2.0 12.63 11.39 1.24 150 2.0 5.08 4.99 0.09
150 4.0 6.46 6.02 0.44 150 4.0 2.83 2.77 0.06
186 1.0 26.67 22.57 4.1 186 1.0 13.84 10.92 2.92
186 2.0 13.69 11.57 2.12 186 2.0 6.57 5.71 0.86
186 4.0 6.84 6.11 0.73 186 4.0 3.54 3.13 0.41
Figure 4.12: The k-values of (a) methane and (b) ethane-Athabasca bitumen systems. (The
symbols are the experimental measured k-value data (Zirrahi et al., 2017) and the solid lines are
the calculated k-values using the proposed method in this work.)
As shown in Figure 4.12, a higher solubility (as a result of higher pressure) results in lower k-
value. The measured and the predicted k-values of methane- and ethane-Cold Lake bitumen
systems are also compared in Figure 4.13. The generated k-values are needed for the basic
assessment of any solvent-based recovery processes. However, the emphasis of the present work
is on generating these values for solvent/bitumen systems where no experimental data is available.
The only required data used in this model is carbon number or boiling point distribution and can
be obtained easily by simulated distillation test.
a) Methane-Athabasca Bitumen
Pressure (MPa)
1 2 3 4 5
k-v
alu
e
5
10
15
20
25
T=100 oC
T=150 oC
T=190 oC
b) Ethane-Athabasca Bitumen
Pressure (MPa)
1 2 3 4 5
k-v
alu
e
2
4
6
8
10
12
14 T=100 oC
T=150 oC
T=190 oC
90
Figure 4.13: The k-values of (a) methane and (b) ethane-Cold Lake bitumen systems. (The
symbols are the experimental measured k-value data in this work and the solid lines are the
calculated k-values using the proposed method.)
Figures 4.14 and 4.15 compare the calculated and the measured k-values of the hydrocarbon
solvents-Athabasca and -Cold Lake bitumen systems, respectively. These figures show the
reliability of the proposed model to predict the k-values of light hydrocarbon solvents-bitumen
systems, even when no experimental solubility data of solvent in bitumen are available.
a) Methane-Cold Lake Bitumen
Pressure (MPa)
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5
k-v
alu
e
0
10
20
30
40
T=100 oC
T=150 oC
T=186 oC
b) Ethane-Cold Lake Bitumen
Pressure (MPa)
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5
k-v
alu
e
0
2
4
6
8
10
12
14
16
18
20
T=100 oC
T=150 oC
T=186 oC
91
Figure 4.14: The comparison between the predicted and the experimental k-value of hydrocarbon
solvent-Athabasca bitumen systems. (Data of propane- and butane-bitumen mixtures were
obtained from Chapters 3 and 2, respectively.)
Methane-Athabasca bitumen
Experimental k-value
1 10 100
Pre
dic
ted
k-v
alu
e
1
10
100
Ethane-Athabasca bitumen
Experimental k-value
1 10
Pre
dic
ted
k-v
alu
e
1
10
Propane-Athabasca bitumen
Experimental k-value
1 10
Pre
dic
ted
k-v
alu
e
1
10
Butane-Athabasca bitumen
Experimental k-value
1 10
Pre
dic
ted
k-v
alu
e
1
10
92
Figure 4.15: The comparison between the predicted and the experimental k-value of hydrocarbon
solvent-Cold Lake bitumen systems. (Data of propane- and butane- bitumen mixtures were
obtained from Chapters 3 and 2, respectively.)
4.6 Summary and Conclusion
A thermodynamic model was developed to calculate the k-values of methane-bitumen and ethane-
bitumen systems at wide ranges of temperature and pressure. To develop this model, the bitumen
was vacuum distilled to four fractions. The simulated distillation results showed that three of cuts
Methane-Cold Lake bitumen
Experimental k-value
1 10 100
Pre
dic
ted
k-v
alu
e
1
10
100
Ethane-Cold Lake bitumen
Experimental k-value
1 10 100
Pre
dic
ted
k-v
alu
e
1
10
100
Propane-Cold Lake bitumen
Experimental k-value
1 10
Pre
dic
ted
k-v
alu
e
1
10
Butane-Cold Lake bitumen
Experimental k-value
1 10
Pre
dic
ted
k-v
alu
e
1
10
93
were completely distillable. Phase behaviour data including solubility, density, and viscosity of
methane- and ethane-bitumen fraction mixtures were measured up to 186 oC and 6 MPa. The
generalized binary interaction parameters were found using the measured solubility of methane
and ethane in distillable bitumen fractions. Applying the developed binary interaction parameters
in this work and having the simulated distillation results of any bitumen sample, the k-values of
solvent/bitumen system can be calculated. The proposed model calculated the methane solubility
in Athabasca and Cold Lake bitumen samples with average absolute deviation of 1.0 and 0.89
mol.%, respectively, when no experimental solubility data has been applied to tune the model.
Moreover, the acceptable deviations of 1.7 and 2.0 mol.% for prediction of ethane solubility in
Athabasca and Cold Lake bitumen samples confirm the ability of this model to calculate the
solubility and k-values of solvent/bitumen mixtures required for design and simulation of solvent
injection recovery processes. The proposed methodology in this work provides a useful tool to find
k-values of hydrocarbon solvents (methane, ethane, propane, and butane) for applications in
solvent-aided thermal bitumen recovery processes.
94
4.7 References
Azinfar, B., Haddadnia, A., Zirrahi, M., Hassanzadeh, H., Abedi, J., 2017. Effect of Asphaltene
on Phase Behavior and Thermophysical Properties of Solvent/Bitumen Systems. J. Chem. Eng.
Data 62, 547–557.
Butler, R., 1999. The Steam and Gas Push (SAGP). J. Can. Pet. Technol. 54–61.
Canbolat, S., Akin, S., Polikar, M., 2004. Evaluation of SAGD Performance in the Presence of
Non-Condensable Gases. 5th Canadian International Petroleum Conference, Calgary, Alberta,
Canada, 8–10 June.
Heron, C., Thimm, H., Sullivan, L., Atkinson, I., 2008. NCG Behavior in SAGD -A Numerical
Simulation Analysis. SPE International Thermal Operations and Heavy Oil Symposium,
Calgary, Alberta, Canda, 20–23 October.
Jiang, Q., Butler, R., Yee, C.T., 1998. The Steam and Gas Push (SAGP)-2: Mechanism Analysis
and Physical Model Testing. 49 th Annual Technical Meeting of The Petroleum Society,
Calgary, Alberta, Canada, 8–10 June.
Kariznovi, M., Nourozieh, H., Abedi, J., 2017. Vapor−Liquid Equilibrium of Bitumen−Ethane
Mixtures for Three Athabasca Bitumen Samples. J. Chem. Eng. Data. 62, 2198–2207.
Kariznovi, M., Nourozieh, H., Abedi, J., 2010. Bitumen Characterization and Pseudocomponents
Determination for Equation of State Modeling. Energy Fuels 24, 624–633.
Lee, B.I., Kesler, M.G., 1975. A Generalized Thermodynamic Correlation Based on Three-
parameter Corresponding States. AIChE J. 21, 510–527.
Nourozieh, H., Kariznovi, M., Abedi, J., 2016. Measurement and Modeling of Solubility and
Saturated-Liquid Density and Viscosity for Methane/Athabasca-Bitumen Mixtures. SPE J.,
180–189.
Pedersen, K.S., Blilie, A.L., Meisingset, K.K., 1992. PVT Calculations on Petroleum Reservoir
Fluids Using Measured and Estimated Compositional Data for the Plus Fraction. Ind. Eng.
Chem. Res. 31, 1378–1384.
95
Rose, J.L., Monnery, W.D., Chong, K., Svrcek, W.Y., 2001. Experimental Data For The Extraction
of Peace River Bitumen Using Supercritical Ethane. Fuel 80, 1101–1110.
Twu, C.H., 1984. An Internally Consistent Correlation for Predicting The Critical Properties and
Molecular Weights of Petroleum and Coal-tar Liquids. Fluid Phase Equilib. 16, 137–150.
Zirrahi, M., Hassanzadeh, H., Abedi, J., 2017. Experimental and Modeling Studies of MacKay
River Bitumen and Light n-Alkane Binaries. Can. J. Chem. Eng. 95, 1417–1427.
96
Chapter Five: Effect of Asphaltene on Phase Behaviour and Thermo-Physical
Properties of Solvent-Bitumen Systems
5.1 Preface
This chapter was published in the Journal of Chemical & Engineering Data, 2017, volume 65,
547-557. This manuscript was co-authored by A. Haddadnia, M. Zirrahi, H. Hassanzadeh, and J.
Abedi. A copy of the copyright permission from the publisher to reproduce this manuscript in the
present thesis is provided in Appendix B.
The PVT apparatus described in this chapter has been also used to collect the phase behaviour data
of solvent-bitumen fractions presented in Chapters 2, 3, and 4.
In this chapter, the measured experimental data of CO2- and ethane-bitumen and CO2- and ethane-
deasphalted bitumen are compared.
5.2 Abstract
Solvent-aided bitumen production from oil sands has shown promise as an alternative to thermal
recovery methods. Phase behaviour studies of solvent/bitumen mixtures are necessary for reservoir
simulation of recovery methods, process design and operation of surface facilities, and
transportation. Bitumen and heavy crudes comprise a different weight fraction of asphaltene. In
this study, the effect of asphaltene on phase behaviour, viscosity, and density of solvent/bitumen
systems is studied. Ethane (C2H6) and carbon dioxide (CO2) are considered as solvents. Phase
behaviour studies and property measurements are conducted on solvent/bitumen and
solvent/deasphalted bitumen systems. Solubility of C2H6 and CO2 in the original and deasphalted
bitumen are measured. The viscosity and density of the liquid phase are also measured by inline
viscometer and densitometer at temperature and pressure ranges of 70−130 C and 2−8 MPa,
respectively. The measured data showed that the asphaltene has no significant effect on C2H6
solubility in bitumen. However, the solubility of CO2 in the original bitumen differs from that of
the deasphalted bitumen. The significant effect of asphaltene on density and viscosity of bitumen
97
is also quantified. Mixing rules are also employed to estimate the density and viscosity of
asphaltene using the density and viscosity of bitumen and deasphalted bitumen.
5.3 Introduction
Thermal recovery methods are the most practical methods for heavy oil and bitumen recovery in
Alberta, Canada. High energy, water consumption, water pollution, greenhouse gas emission, and
the need to further dilute the produced fluid are the associated concerns, which necessitate
development of alternative methods. Solvent-aided processes are potential alternatives to the
thermal recovery methods (Nasr and Ayodele, 2006). Co-injecting solvent into a bitumen reservoir
further reduces the bitumen viscosity and improves the recovery process. Further development and
optimization of these processes require advanced understanding of the phase behaviour of
solvent/oil systems (Kokal and Sayegh, 1993). The composition, density, and viscosity of each
phase in a solvent/bitumen system are the basic data required for the engineering studies and
simulation of solvent-aided thermal recovery processes.
Bitumen contains usually more than 10 wt.% asphaltene, which can significantly affect the
physicochemical properties of heavy oil and bitumen (Letcher, 2013). The role of asphaltene in
the phase behaviour of solvent/heavy oil or bitumen is a challenging topic in this context. A clear
understanding of the effect of asphaltene on phase behaviour of a solvent/bitumen system is
lacking in the literature. While the importance of asphaltene on phase behaviour has been
emphasized in a number of studies (Shirani et al., 2012; Vargas et al., 2009), its effect on the
solubility of CO2 has often been ignored (Foroughi et al., 2011; Kokal and Sayegh, 1993;
Marufuzzaman and Henni, 2015).
Asphaltene is generally defined as the nonvolatile and polar fraction of crudes, which is insoluble
in n-alkanes (i.e., pentane or heptane) (Alian et al., 2011). Because the definition of asphaltene is
based on insolubility rather than chemical structure, a wide distribution of molecular structure is
expected to be covered (Luo et al., 2010; Yarranton et al., 2013). Moreover, different solvents and
procedures used to separate the asphaltene from bitumen can lead to different asphaltene contents,
properties, and structures (Luo et al., 2010). For example, the asphaltene separated by heptane has
H/C ratios lower than those obtained by pentane (Kokal and Sayegh, 1995). Therefore, the
98
asphaltene precipitated by heptane shows a higher degree of aromaticity as compared to that
extracted by pentane. The asphaltene yield also decreases by increasing the carbon number of
alkane precipitant (Luo et al., 2010). Moreover, the degree of washing can affect the yield of
asphaltene. Alboudwarej et al. (Alboudwarej et al., 2002) compared the yield of asphaltene in four
different washing levels; unwashed, filter-washed, sonicator-washed, and Soxhlet washed. They
reported a decrease of asphaltene yield from unwashed to Soxhlet-washed systems. They noted
that removing the trapped maltene and high molar mass resins due to increased washing may be
attributed to the decreasing of yield (Alboudwarej et al., 2002).
In some of the available phase behaviour studies on CO2/bitumen systems, the effect of asphaltene
on the solubility of CO2 in bitumen has been ignored. Kokal and Sayegh (Kokal and Sayegh, 1993)
studied phase behaviour of CO2 and a Canadian heavy oil mixture. In their work, asphaltene and
resin were separated from heavy oil. Then, the phase behaviour data of maltene or deasphalted oil
with CO2 including solubilities, swelling factors, densities, gas/oil ratios and viscosities were
measured at 21 and 140 ºC and pressures up to 12.41 MPa. The listed phase behaviour data were
also measured for fractions of heavy oil obtained by batch distillation. They concluded that the
removal of asphaltene and solid from the heavy oil has very little effect on the solubility of CO2.
Foroughi et al. (Foroughi et al., 2011) also compared solubility of CO2 in maltene and bitumen at
22 and 35 ºC and a pressure range of 15 MPa and noted that asphaltene has a negligible effect on
the solubility of CO2 in bitumen. Marufuzzaman and Henni (Marufuzzaman and Henni, 2015)
measured the solubilities of CO2 and C2H6 in heavy oil and its SARA (Saturate, Aromatic, Resin,
and Asphaltene) fractions. They observed that asphaltene content affects C2H6 solubility in heavy
oil significantly as compared to CO2 at the same equilibrium condition. The reported solubilities
of C2H6 and CO2 were higher in maltene than heavy oil (Marufuzzaman and Henni, 2015). The
temperature and pressure ranges of the reported experimental data were 15−30 ºC and 0.2−2 MPa,
respectively. Pentane (Kokal and Sayegh, 1993; Marufuzzaman and Henni, 2015) and heptane
(Foroughi et al., 2011) were used as solvent for separation of asphaltene in the mentioned works.
Asphaltene consists of aromatic rings, O, S, and N elements that can show associating behaviour
with CO2. Therefore, considering asphaltene as an inert component in the study of CO2 solubility
in heavy oil or bitumen may not be a valid assumption.
99
In this work, we investigate the effect of asphaltene on phase behaviour and thermophysical
properties of solvent/bitumen systems. The solubility of C2H6 (as hydrocarbon solvent) and CO2
(as nonhydrocarbon solvent) in bitumen and deasphalted bitumen as well as viscosity and density
of the liquid phase are measured in wide ranges of temperature and pressure. Simulated distillation
(SimDist) test based on ASTM D7169 is used to analyze the whole and deasphalted bitumen, and
asphaltene.
The density and viscosity data of asphaltene is highly valuable for both simulation and modeling
of many processes in which asphaltene presents as a main constitute. For example, incompatible
blends of crudes might precipitate asphaltene during transportation and refining (Wiehe and
Kennedy, 2000). Furthermore, in N-Solv process (Nenniger et al., 2013) liquid solvent is injected
into a bitumen reservoir that precipitates a portion of asphaltene in the porous media. To model or
simulate such a process and predict the asphaltene precipitation, the ill-defined nature of
asphaltene is the main challenge. In this study, density and viscosity of asphaltene, which are the
important inputs to these models, are also predicted by applying mixing rules.
The rest of this work is organized as follows: first, the experimental apparatus and procedures to
separate the asphaltene and measure the experimental phase behaviour data are described. Then,
the results and discussion are presented followed by the conclusion.
5.4 Experimental Section
5.4.1 Materials
C2H6 and CO2 were supplied by Praxair with purities of 99.9 and 99.5 mol %, respectively.
Heptane was supplied by British Drug Houses (BDH), and the purity of n-heptane was greater than
99.0 mol %. Water- and sand-free Athabasca bitumen sample was provided by an oil company in
Alberta, Canada.
5.4.2 Deasphalting the Bitumen
Heptane was considered as solvent for separation of asphaltene from bitumen. Heptane was
suggested as the logical solvent to obtain asphaltene, because the most insoluble material is
precipitated by heptane and heavier solvents (Kokal and Sayegh, 1995; Mitchell and Speight,
100
1972). The asphaltene separation process is described as follows (Alboudwarej et al., 2002; Diaz
et al., 2014).
First, a specific amount of bitumen with heptane at the ratio of 40 mL of solvent per 1 g of bitumen
was added to a beaker. Then, the solution was sonicated for 50 min and allowed to settle for 24 h.
After 1 day, about 75 vol.% of the solution was decanted into the funnel and filtered using filter
paper with #2 mesh size. Then, fresh solvent at an amount equivalent to about 10 vol.% of the
initial solvent was added to the beaker and sonicated for 45 min. After sonication, solution was
settled for about 15 h. After that, all the settled material inside the beaker was filtered using the
same filter paper. The final step is washing the asphaltene remaining on the filter paper using fresh
n-heptane until the liquid leaving the filter paper was colorless. The weight fraction of asphaltene,
which is the weight of the asphaltene after washing and drying, divided by the weight of the initial
bitumen was about 14 wt.%. The next step is separation of heptane from the filtrate. All of the
filtrate was placed in the rotary evaporator to evaporate the solvent and recover the deasphalted
bitumen. The solvent evaporation was done for 2 days at 150 ºC and under vacuum to ensure that
there was no heptane that remained in the sample.
During evaporation of heptane from deasphalted bitumen some light components may be removed.
In order to relate the difference in properties of solvent/bitumen and solvent/deasphalted bitumen
systems to the asphaltene, light ends of bitumen and deasphalted bitumen should be the same.
Therefore, the whole bitumen was also heated to 150 ºC under vacuum for 2 days.
101
Figure 5.1: Flame ionization detector signal versus retention time for bitumen, heated bitumen
and deasphalted bitumen.
The signals of Flame Ionization detector (FID) for bitumen before and after heating, and
deasphalted bitumen versus retention time is shown in Figure 5.1. The overlap of the heated and
deasphalted bitumen chromatograms at low retention time shows that light components of whole
and deasphalted bitumen are almost the same after heating. It also indicates that thermal cracking
has not occurred during the heating period of bitumen at 150 ºC. The existence of asphaltene in
bitumen is evident at the end of the chromatograms (retention time > 26 min), which indicates
higher FID signal for bitumen as compared to the deasphalted one. Therefore, the observed
differences between thermo-physical properties and solubilities of C2H6 or CO2 in bitumen and
deasphalted bitumen can be attributed to the asphaltene. Moreover, no peaks for heptane have been
observed confirming that heptane has been completely removed from the deasphalted bitumen.
The SimDist results of whole and deasphalted bitumen and asphaltene are shown in Figure 5.2. It
can be seen that 88 and 79% of deasphalted bitumen and bitumen were distilled, respectively,
while only 30% of asphaltene was eluted from the column at temperature up to 710 C.
0
10000
20000
30000
40000
50000
60000
70000
80000
1 6 11 16 21 26 31
Vo
ltag
e(µ
V)
Retention time (min)
Bitumen
Heated bitumen
Deasphalted bitumen
102
Figure 5.2: Boiling point versus mass percentage of distilled bitumen (%Off), deasphalted
bitumen, and asphaltene.
The molecular weights of whole and deasphalted bitumen were measured by freezing point
osmometer (model 5009, Precision Systems Inc., Natick, MA, USA) where benzene was
considered as solvent. The molecular weights of whole and deasphalted bitumen are 541 and 470
g/mol, respectively. For asphaltene, molecular weight was measured using vapor pressure
osmometry (VPO) method and it was obtained as 5220 g/mol.
5.4.3 PVT Apparatus
A PVT apparatus was designed, fabricated, and validated to measure the solubility of hydrocarbon
and nonhydrocarbon gases in whole and deasphalted bitumen as well as the density and viscosity
of liquid phase over wide ranges of temperature and pressure. Density measurement is carried out
using an Anton Paar densitometer (Model DMA HPM) with accuracy of 0.001−0.0001 g/cm3
applicable for the density range of 0-3 g/cm3. The operating temperatures and pressures for the
densitometer are up to 200 ºC and 70 MPa, respectively.
The Viscopro 2000 with a SPL 440 sensor was used to measure viscosity of the liquid phase. The
viscometer can measure viscosities from 0.2 to 20000 mPa.s up to a maximum temperature of
% Mass distilled
0 20 40 60 80 100
Bo
ilin
g p
oin
t (o
C)
100
200
300
400
500
600
700
800
BitumenDeasphalted bitumenAsphaltene
103
190°C and maximum pressure of 13.6 MPa. The accuracy of the measured viscosity is 1.0% of the
full scale of the piston for the piston ranges of 0.2−2, 1−20, and 50−1000 mPa.s, the accuracies
were 0.02, 0.2, and 10 mPa.s, respectively.
Figure 5.3 shows the schematic of the PVT apparatus used in this work. The equilibrium cell
equipped with an automated rocking system, a densitometer, and a viscometer is placed in the
oven. The transfer and sampling cells are located outside the oven. The ISCO and Quizix pumps
are used for solvent injection to the system and transfer the liquid phase from the equilibrium cell
to the transfer and sampling cells, respectively.
Before conducting the experiments, the deasphalted bitumen was filtered with two different mesh
sizes with the smallest size of 0.5 µm in order to remove the smallest possible particles existing in
the sample. Existence of these particles can cause errors in viscosity and density measurements.
5.4.4 Experimental Procedure
The entire system, including cells, densitometer, viscometer, and lines, were washed with toluene
and acetone and then vacuumed. To ensure no contaminants were left, pure solvent was purged
into the system. After cleaning, whole or deasphalted bitumen was charged into the equilibrium
cell using a transfer cell and Quizix pump through the top of the equilibrium cell. The oven
temperature had been set to the desired set point (70, 100, and 130 C). The gaseous solvent was
then injected using an ISCO pump at the desired pressure. The pressure of the system was
measured by a Rosemount 3051 pressure transmitter. The feed and solvent were mixed using the
rocking system at a specific temperature and pressure until an equilibrium was reached. While
solvent was dissolving in the liquid phase, the ISCO pump was injecting the makeup solvent to
keep the pressure constant. The equilibrium was determined whenever the injection was stopped;
i.e., the solvent could no longer dissolve in the sample. After reaching equilibrium, the bitumen-
rich phase was discharged from the bottom of the equilibrium cell by receiving it using the Quizix
pump and the transfer cell. During sampling, the liquid phase was passed through the inline density
and viscosity measuring devices and the physical properties were recorded. During the receiving
process, the system pressure was kept constant using the ISCO pump to prevent the gaseous solvent
evolving from the liquid phase. To measure the solubility of gas in the liquid phase, the liquid
sample was taken using the sampling cell and flashed at atmospheric pressure. The evolved gas
104
was measured using the Chandler Engineering Gasometer (Model 2331) with an accuracy of 0.2%
over the range of the readings.
Figure 5.3: Schematic PVT setup used in this work: 1, equilibrium cell; 2, densitometer; 3,
viscometer; 4, density measuring unit; 5, viscosity measuring unit; 6, sampling cell; 7, Quizix
pump; 8, water tank; 9, transfer cell; 10, pressure transducer; 11, ISCO pump; 12, gas cylinder;
13, vent valve; 14, oven.
5.5 Results and Discussion
The solubilities of C2H6 and CO2 in bitumen and deasphalted bitumen, liquid phase densities, and
viscosities were measured at 70, 100, and 130 ºC. The experimental results are reported in Table
5.1.
57 cp
0.874 gr/cc
98 OC
1
2 3
6
9
7
4
5
11
12
14
10
88 751 PSI
8
13
105
Table 5.1: Phase behaviour data for C2H6/bitumen, C2H6/deasphalted bitumen, CO2/bitumen, and
CO2/deasphalted bitumen systems.
Temperature ºC
Pressure MPa
Density kg/m3
Viscosity mPa.s
Solubility mol.%
C2H6/bitumen
70.2 8.3 868.1 23.1 70.3
70.5 6.0 894.5 43.3 63.0
70.6 4.1 923.0 104.9 49.6
70.1 2.1 951.1 − 32.9
100.2 8.1 879.4 18.9 62.2
100.2 6.1 898.5 29.7 54.1
100.1 3.9 920.1 52.3 42.6
100.1 2.1 939.5 110.4 27.1
130.4 8.0 877.0 12.6 55.4
130.4 5.9 892.5 17.6 47.8
130.4 4.0 908.4 26.3 36.2
130.4 2.1 924.5 42.0 22.0
C2H6/deasphalted bitumen
70.9 8.2 843.1 7.3 69.5
70.9 6.0 870.5 12.1 62.4
70.9 4.0 900.2 27.8 49.5
70.9 2.0 928.9 82.5 31.7
100.2 8.1 855.5 6.4 61.2
100.2 6.1 875.4 9.4 52.1
100.2 3.9 898.1 16.7 41.5
100.2 2.1 917.7 30.2 25.9
130.4 8.0 853.1 4.7 53.6
130.4 6.0 869.2 6.1 45.7
130.4 4.1 885.6 9.0 35.4
130.4 2.1 902.1 12.6 20.5
CO2/bitumen
70.9 8.1 975.8 153.4 55.3
70.9 6.0 976.1 205.2 48.4
70.9 4.1 976.4 377.9 38.4
70.9 2.1 976.8 774.5 22.5
100.2 8.0 957.1 59.2 50.5
100.2 6.0 957.9 85.4 42.0
100.2 4.1 958.3 118.4 32.7
100.2 2.1 958.8 176.1 18.9
130.3 8.0 938.8 27.4 45.1
130.4 6.0 939.2 33.6 38.1
130.4 4.0 939.8 42.8 30.2
130.4 2.0 940.5 55.3 17.8
CO2/deasphalted bitumen
70.8 8.0 955.3 36.6 49.6
70.9 6.0 955.3 55.9 41.8
70.9 4.0 955.6 90.2 32.6
70.9 2.0 955.8 164.6 19.6
100.2 8.0 938.4 19.3 43.3
100.2 6.0 938.6 23.7 36.9
106
100.2 4.0 939.0 34.1 28.0
100.2 2.1 939.3 48.1 15.9
130.3 8.0 917.6 9.1 38.8
130.3 6.0 917.9 11.4 32.7
130.3 4.1 918.5 13.2 24.9
130.3 2.1 919.0 17.1 13.1
To observe the effect of asphaltene on the phase behaviour data of solvent/bitumen systems, the
solubilities of CO2 and C2H6 in bitumen and in deasphalted bitumen, and liquid phase densities
and viscosities are compared in the following.
5.5.1 Solubility of C2H6 and CO2 in Whole and Deasphalted Bitumen
Solubilities of CO2 and C2H6 in bitumen and deasphalted bitumen were measured at temperatures
of 70, 100, and 130 ºC and at four pressure values up to 8.2 MPa. The results are shown in Figures
5.4 (a and b) in which the filled and open symbols represent the bitumen and deasphalted bitumen
data, respectively. Solubilities of CO2 and C2H6 in bitumen and deasphalted bitumen increase with
increasing pressure at constant temperature and decrease with increasing temperature. CO2 shows
lower solubility in bitumen than C2H6 at a given temperature and pressure. C2H6 usually has higher
solubility compared to CO2 in bitumen and heavy oils. These two solvents have close molecular
weight and critical properties. C2H6 is a hydrocarbon molecule which is more similar to
hydrocarbon reservoir fluids. Therefore, C2H6 solubility in bitumen is higher as compared to CO2.
Pressure (MPa)
1 2 3 4 5 6 7 8 9
CO
2 s
olu
bili
ty (
mole
fra
ction)
0.1
0.2
0.3
0.4
0.5
0.6
BitumenDeasphalted bitumen
70 oC 100
oC 130
oC
Pressure (MPa)
1 2 3 4 5 6 7 8 9
CO
2 s
olu
bili
ty (
mole
fra
ction)
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
0.50
BitumenDeasphalted bitumen
Pressure (MPa)
1 2 3 4 5 6 7 8 9
CO
2 s
olu
bili
ty (
mole
fra
ction)
0.1
0.2
0.3
0.4
0.5
0.6
BitumenDeasphalted bitumen
(a) CO2
107
Figure 5.4: Solubility of (a) CO2 and (b) C2H6 in bitumen and in deasphalted bitumen at 70, 100,
and 130 C.
Solubility of CO2 and C2H6 in bitumen was compared to those in deasphalted bitumen to
investigate the effect of the presence of asphaltene on the solubility of solvents. It is observed that
the presence of asphaltene in the bitumen results in higher CO2 dissolution. However, the effect of
asphaltene on C2H6 solubility in bitumen is marginal. This difference in the effect of asphaltene
on solubility can be possibly explained by the following.
CO2 is a linear molecule with a permanent quadrupolar moment and no permanent dipolar moment.
The permanent quadrupole is strong enough to affect the thermodynamic properties of CO2
molecules, which differs significantly from those of other nonpolar molecules such as C2H6
(Lansangan and Smith, 1993). On the other hand, asphaltene definition based on solubility rather
than chemical class makes study of asphaltene more difficult than simple and light components.
Asphaltene consists of polynuclear aromatics, a small amount of heteroatoms (S, N, and O), nickel,
vanadium, and other metal elements (Luo et al., 2010). Existence of components such as
heteroatoms in asphaltene structure can give the molecule polarity (Akbarzadeh et al., 2007; Aslan
and Firoozabadi, 2014).
For the CO2/bitumen system in the presence of asphaltene, CO2 is more soluble in polar molecules
of asphaltene because of dipole/quadrupole interactions. For C2H6/bitumen and C2H6/deasphalted
systems, higher solubility of C2H6 in deasphalted bitumen as compared to bitumen is expected
because of the presence of lighter components in deasphalted bitumen as reported by
Pressure (MPa)
1 2 3 4 5 6 7 8 9
C2H
6 s
olu
bili
ty (
mole
fra
ctio
n)
0.2
0.3
0.4
0.5
0.6
0.7
BitumenDeasphalted bitumen
Pressure (MPa)
1 2 3 4 5 6 7 8 9
C2H
6 s
olu
bili
ty (
mole
fra
ctio
n)
0.1
0.2
0.3
0.4
0.5
0.6
BitumenDeasphalted bitumen
Pressure (MPa)
1 2 3 4 5 6 7 8 9
C2H
6 s
olu
bili
ty (
mole
fra
ctio
n)
0.2
0.3
0.4
0.5
0.6
0.7
0.8
BitumenDeasphalted bitumen
70 oC 100
oC 130
oC
(b) C2H6
108
Marufuzzaman and Henni (Marufuzzaman and Henni, 2015). However, our experimental results
show that the absence of asphaltene does not have a noticeable effect on the solubility of C2H6 in
bitumen (Figure 5.4(b)).
Asphaltene was dissolved in carbon disulfide (CS2) and tested by SimDist to see its carbon
distribution. The weight percent of distilled samples versus carbon number is shown in Figure 5.5.
This figure shows a broad range of carbon number for asphaltene, which means light components
are also present in the asphaltene. C2H6 may be dissolved in the light components of asphaltene,
and this could compensate for the difference between solubility of C2H6 in the absence and
presence of asphaltene. While the effect of asphaltene on solubility of C2H6 in bitumen is not
noticeable, it will be shown later that its effect on phase density and viscosity is evident.
Asphaltene properties can be altered by temperature, contact time with de-asphaltening solvent,
solvent to oil ratio, solvent type, washing procedure, and filter pore size (Luo et al., 2010;
Nikooyeh et al., 2012). Consequently, the effect of asphaltene on solubility can be changed based
on the type of bitumen or heavy oil, amount of asphaltene, asphaltene separation process, solvent
used for asphaltene separation and the type of gas for phase behaviour studies.
Figure 5.5: Carbon number distribution for bitumen, deasphalted bitumen, and asphaltene.
Carbon number
20 40 60 80 100
wt.
%
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5BitumenDeasphalted bitumen Asphaltene
109
5.5.2 Effect of C2H6 and CO2 Dissolution on Density and Viscosity of Bitumen and Deasphalted
Bitumen
The high viscosity of bitumen and heavy oil is mainly due to the presence of asphaltene. In order
to demonstrate the effect of asphaltene on viscosity and density of bitumen, the density and
viscosity of the whole and deasphalted bitumen at 70, 100, and 130 ºC and pressures up to 8 MPa
are shown in Figures 5.6 (a and b). It is clearly seen that after separation of asphaltene, the viscosity
and density of the deasphalted bitumen are significantly lower than those of the whole bitumen.
Although asphaltene content of bitumen is only 14 wt %, it has significantly affected bitumen
density and viscosity.
Figure 5.6: (a) Density and (b) viscosity of bitumen and deasphalted bitumen versus pressure at
70, 100, and 130 ºC.
Pressure (MPa)
0 2 4 6 8 10
De
nsity (
kg/m
3)
920
940
960
980BitumenDeasphalted bitumen
70 oC
Pressure (MPa)
0 2 4 6 8 10
De
nsity (
kg/m
3)
920
940
960
980BitumenDeasphalted bitumen
Pressure (MPa)
0 2 4 6 8 10
De
nsity (
kg/m
3)
920
940
960
980
BitumenDeasphalted bitumen
100 oC 130
oC
(a)
Pressure (MPa)
0 2 4 6 8 10
Vis
co
sity (
mP
a.s
)
10
100
1000
10000
BitumenDeasphalted bitumen
70 oC 100
oC 130
oC
(b)
Pressure (MPa)
0 2 4 6 8 10
Vis
co
sity (
mP
a.s
)
10
100
1000
10000
BitumenDeasphalted bitumen
Pressure (MPa)
0 2 4 6 8 10
Vis
co
sity (
mP
a.s
)
10
100
1000
10000
BitumenDeasphalted bitumen
110
The densities of CO2- and C2H6-saturated whole and the deasphalted bitumen are plotted in Figures
5.7 (a and b). As shown in Figure 5.7(a), the density of deasphalted bitumen saturated with CO2 is
lower than the whole bitumen saturated with CO2. There is a very slight decrease in density of
CO2-saturated bitumen and deasphalted bitumen as pressure increases. This behaviour was
observed at all temperatures. When a gas molecule with low molecular weight is dissolved in liquid
phase with higher molecular weight and there is no attraction between gas and liquid molecules,
the density of the liquid phase decreases because of liquid phase swelling. However, when there
is association between gas and liquid molecules, the liquid density increases or remains constant.
CO2 molecules can associate with aromatic molecules. Increasing density because of association
of molecules was compensated by decreasing density due to dissolving gas in the bitumen.
Therefore, the density variations with pressure are very marginal for CO2/bitumen and
CO2/deasphalted bitumen systems.
The density variation of the C2H6-saturated bitumen and the deasphalted bitumen is shown in
Figure 5.7(b). Density varies linearly with pressure and the slope of density versus pressure
decreases by increasing temperature for both bitumen and deasphalted bitumen as a result of lower
gas dissolution at higher temperature. The density of C2H6-saturated bitumen is significantly
higher than that of deasphalted bitumen.
Pressure (MPa)
0 2 4 6 8 10
De
nsity (
kg/m
3)
920
930
940
950
960
970
980
BitumenDeasphalted bitumen
Pressure (MPa)
0 2 4 6 8 10
De
nsity (
kg/m
3)
910
920
930
940
950
960
970
980
BitumenDeasphalted bitumen
Pressure (MPa)
0 2 4 6 8 10
De
nsity (
kg/m
3)
920
930
940
950
960
970
980
BitumenDeasphalted bitumen
70 oC 100
oC 130
oC
(a) CO2
111
Figure 5.7: Density of (a) CO2-saturated and (b) C2H6-saturated bitumen and deasphalted bitumen
as a function of pressure at temperatures of 70, 100, and 130 ºC.
Panels a and b of Figure 5.8 show the viscosity of CO2- and C2H6-saturated bitumen viscosity
versus pressure. The decreasing trend of viscosity reduces at higher temperatures. C2H6 dissolution
reduces the viscosity of whole and deasphalted bitumen more than CO2. The viscosity of CO2- and
C2H6-saturated bitumen is extremely higher than those of deasphalted bitumen. Understanding the
asphaltene effect on bitumen viscosity value helps in designing effective and economical bitumen
recovery methods. Although there are many empirical correlations and models to predict the crude
oil viscosity, they cannot be easily applied on bitumen because of its complex structure (Luo and
Gu, 2007).
Pressure (MPa)
1 2 3 4 5 6 7 8 9D
en
sity (
kg/m
3)
820
840
860
880
900
920
940
960
BitumenDeasphalted bitumen
Pressure (MPa)
1 2 3 4 5 6 7 8 9
De
nsity (
kg/m
3)
820
840
860
880
900
920
940
960
BitumenDeasphalted bitumen
Pressure (MPa)
1 2 3 4 5 6 7 8 9
De
nsity (
kg/m
3)
820
840
860
880
900
920
940
960
BitumenDeasphalted bitumen
70 oC 100
oC 130
oC
(b) C2H6
Pressure (MPa)
1 2 3 4 5 6 7 8 9
Vis
co
sity (
mP
a.s
)
1
10
100
1000
BitumenDeasphalted bitumen
Pressure (MPa)
1 2 3 4 5 6 7 8 9
Vis
co
sity (
mP
a.s
)
1
10
100
1000
BitumenDeasphalted bitumen
Pressure (MPa)
1 2 3 4 5 6 7 8 9
Vis
co
sity (
mP
a.s
)
1
10
100
1000
BitumenDeasphalted bitumen
70 oC 100
oC 130
oC
(a) CO2
112
Figure 5.8: Viscosity of (a) CO2-saturated and (b) C2H6-saturated bitumen and deasphalted
bitumen as a function of pressure at different temperatures.
Panels a and b of Figure 5.9 compare the effects of CO2 and C2H6 dissolution on the density of
bitumen and deasphalted bitumen at 70 C. C2H6 dissolution decreases density significantly for
both whole and deasphalted bitumen while the change of density for the whole and deasphalted
bitumen saturated with CO2 is marginal. This behaviour is also because of the difference in the
nature of C2H6 and CO2. The dominant factor for the whole or deasphalted C2H6/bitumen systems
is dilution with dissolved C2H6. However, in the case of CO2, there are two competing factors that
tend to offset each other leading to negligible change in the liquid density. These two effects, as
described in Figure 5.7(a), are dilution because of dissolved CO2, which decreases the density, and
association bonding between CO2 and complex hydrocarbon molecules, which increases the
density as shown in Figure 5.7. This behaviour for CO2 was also previously reported elsewhere
(Lansangan and Smith, 1993; Zirrahi et al., 2015).
Pressure (MPa)
1 2 3 4 5 6 7 8 9
Vis
co
sity (
mP
a.s
)1
10
100
1000
BitumenDeasphalted bitumen
Pressure (MPa)
1 2 3 4 5 6 7 8 9
Vis
co
sity (
mP
a.s
)
1
10
100
1000
BitumenDeasphalted bitumen
Pressure (MPa)
1 2 3 4 5 6 7 8 9
Vis
co
sity (
mP
a.s
)
1
10
100
1000
BitumenDeasphalted bitumen
70 oC 100
oC 130
oC
(b) C2H6
113
Figure 5.9: Comparison of the effect of CO2 and C2H6 dissolution on density of (a) bitumen and
(b) deasphalted bitumen at 70 C.
Panels a and b of Figure 5.10 shows the viscosity variation of bitumen and deasphalted bitumen
saturated with CO2 and C2H6 and without solvent. As it was expected, the reduction of viscosity in
the presence of both gases for the bitumen system is more pronounced than for the deasphalted
bitumen system. As mentioned in Figure 5.8, the viscosity reduction in the case of C2H6 is more
evident than CO2, because of higher solubility of C2H6 than CO2 in bitumen and deasphalted
bitumen.
Pressure (MPa)
1 2 3 4 5 6 7 8 9
De
nsity (
kg/m
3)
820
840
860
880
900
920
940
960
980
1000
Bitumen C2H6-saturated bitumen
CO2-saturated bitumen
Pressure (MPa)
1 2 3 4 5 6 7 8 9
De
nsity (
kg/m
3)
820
840
860
880
900
920
940
960
980
1000
Deasphalted bitumenC2H6-saturated deasphalted bitumen
CO2-saturated deasphalted bitumen
(a) (b)
114
Figure 5.10: Comparison the effect of CO2 and C2H6 dissolution on viscosity of (a) bitumen and
(b) deasphalted bitumen at 70 C.
5.5.3 Calculation of Asphaltene Density and Viscosity
Density and viscosity of asphaltene (as pseudo-liquid fraction) are the required data for simulation
and optimization of the processes involving the asphaltene precipitation such as N-solv (Nenniger
et al., 2013) and insitu upgrading (Mokrys and Butler, 1993) processes. In these processes, heavy
solvents are injected into the bitumen reservoir resulting in precipitation of asphaltene. The
produced oil in these processes has better quality and is lighter than the original bitumen in place.
The density of asphaltene cannot be directly determined. There are some methods based on
dissolution of asphaltene in Heptol (heptane and toluene) or toluene (Barrera et al., 2013; Stratiev
et al., 2016; Yarranton and Masliyah, 1996). In these methods, the asphaltene is dissolved in
solvent at different concentrations and the density of the mixture is measured. Then, the density of
mixture is extrapolated to zero concentration of solvent. The density of asphaltene is calculated at
the point of zero solvent concentration. Another method is based on utilizing of mixing rules to
determine the density of solute (asphaltene) by knowing the density of solution and solvent. Most
of the similar works were done at low temperature (20 C) and atmospheric pressure (Barrera et
al., 2013). Furthermore, the main issue associated with this method is re-dissolution of asphaltene
Pressure (MPa)
1 2 3 4 5 6 7 8 9
Vis
cosity (
mP
a.s
)
1
10
100
1000
10000
Bitumen
C2H6-saturated bitumen
CO2-saturated bitumen
Pressure (MPa)
1 2 3 4 5 6 7 8 9
Vis
cosity (
mP
a.s
)
1
10
100
1000
10000
Deasphalted bitumen
C2H6-saturated deasphalted bitumen
CO2-saturated deasphalted bitumen
(a) (b)
115
in solvent that may change the structure and properties of the original asphaltene in bitumen or
heavy oil.
In this study, densities of whole and deasphalted bitumen were measured in wide ranges of
temperature and pressure. Then, density of asphaltene was calculated using the following mixing
rule as given by:(Luo and Gu, 2007)
bitumen dDeasphalte
bitumen dDeasphalte
Bitumen
Asphaltene
Asphaltene1
(5.1)
where ρ and ω are the density and mass fraction, respectively.
Figure 5.11 shows density of asphaltene obtained by the equation (5.1) versus pressure at three
different temperatures. As Figure 5.11 shows, the density of asphaltene varies slightly compared
to whole and deasphalted bitumen at a constant temperature. This figure also reveals the asphaltene
density is much higher than whole and deasphalted bitumen. For example, at pressure of 2.2 MPa
and temperature of 130 C, the densities of whole, deasphalted bitumen, and asphaltene are 946.3,
920.3, and 1144.7 kg/m3, respectively. Although the weight fraction of asphaltene in bitumen is
small (around 14 wt %), asphaltene highly affects the density of bitumen.
Figure 5.11: Density variation versus pressure at different temperatures. (Asterisks(*) denote the
measured density data in this work.)
Viscosity data of asphaltene are essential for quantitative studies of fluid flow in porous media.
The effect of asphaltene on heavy oil and bitumen viscosity were previously studied (Dealy, 1979;
100 o
C
Pressure (MPa)
1 2 3 4 5 6 7 8 9
De
nsity (
kg
/m3)
800
900
1000
1100
1200
Bitumen*
Deasphalted bitumen*Asphaltene (Calculated density)
Pressure (MPa)
1 2 3 4 5 6 7 8 9
De
nsity (
kg
/m)
800
900
1000
1100
1200
Bitumen*Deasphalted bitumen*Asphaltene (Calculated density)
130 o
C70 oC
Pressure (MPa)
1 2 3 4 5 6 7 8 9
De
nsity (
kg
/m3)
800
900
1000
1100
1200
Bitumen*
Deasphalted bitumen*Asphaltene (Calculated density)
116
Luo and Gu, 2007; Mack, 1932). Using the similar mentioned method to calculate the density of
asphaltene, asphaltene viscosity can be determined. This means that asphaltene is dissolved in
proper solvent and the viscosity of solution is recorded. Applying mixing rules which consider
solvent and asphaltene as constituents that form the solution, the viscosity of asphaltene can be
calculated. However, as noted earlier, re-dissolution of asphaltene in solvent after precipitation
alters the structure and properties of asphaltene. Moreover, when the components of solution have
great differences in viscosity, applying the mixing rule to find the viscosity of the mixture may
lead to significant errors (Centeno et al., 2011). The large viscosity contrast between solvent and
asphaltene results in large errors. In the following method, the original mixture of asphaltene and
deasphalted bitumen which constitutes the whole bitumen is considered. The Arrhenius mixing
rule was used to calculate the viscosity of asphaltene as follows (Zirrahi et al., 2012)
Asphaltene
bitumen dDeasphaltebitumen dDeasphalteBitumen
Asphaltene
)ln()ln(exp
(5.2)
µ and ω are the viscosity and mass fraction, respectively. The calculated viscosity of asphaltene
and measured viscosity of whole and deasphalted bitumen in three temperatures are plotted in
Figure 5.12. The remarkable effect of asphaltene on the viscosity of bitumen can be interpreted by
seeing the high viscosity of asphaltene in this figure. At saturation pressure of 2.2 MPa and
temperature of 130 oC, the viscosities of whole, deasphalted bitumen, and asphaltene are 84.1,
26.0, and 1.1×105 mPa.s, respectively. It can be concluded that separation of asphaltene from crude
oil will increase the mobility of crudes. This results in higher oil production from reservoirs and
lower cost of transportation with pipeline.
117
Figure 5.12: Viscosity variation versus pressure at different temperatures. (Asterisks(*) denote
the measured viscosity data in this work.)
5.6 Summary and Conclusion
The effect of asphaltene on phase behaviour and thermophysical properties of C2H6/bitumen and
CO2/bitumen systems was studied. The deasphalted bitumen was obtained by separation of
asphaltene from bitumen using heptane. Solubility of C2H6 and CO2 in bitumen and deasphalted
bitumen as well as viscosity and density of liquid phase were measured at temperatures of 70, 100
and, 130 C and pressures of 2−8 MPa. The major conclusions drawn from this study are
summarized as follows:
1. At the same operating conditions, solubility of C2H6 in bitumen and deasphalted bitumen
is more than CO2. As a result, the density and viscosity reduction using C2H6 is greater that
CO2.
2. The solubility of CO2 in bitumen was significantly more than that of in deasphalted
bitumen because of the molecular interaction of CO2 and asphaltene.
3. In contrast to CO2 where its solubility in bitumen and deasphalted bitumen are rather
different, C2H6 solubility in bitumen was slightly different than its solubility in deasphalted
bitumen. This different solubility behaviour is attributed to the difference in the molecule
structure of CO2 and C2H6.
4. The significant effect of asphaltene on viscosity and density of bitumen in the presence and
absence of C2H6 and CO2 were quantified.
130 oC
Pressure (MPa)
1 2 3 4 5 6 7 8 9
Vis
co
sity (
mP
a.s
)
1e+0
1e+1
1e+2
1e+3
1e+4
1e+5
1e+6
1e+7
1e+8
1e+9
1e+10Bitumen*Deasphalted bitumen*Asphaltene (Calculated viscosity)
100 oC
Pressure (MPa)
1 2 3 4 5 6 7 8 9V
isco
sity (
mP
a.s
)
1e+0
1e+1
1e+2
1e+3
1e+4
1e+5
1e+6
1e+7
1e+8
1e+9
1e+10
Bitumen*Deasphalted bitumen*Asphaltene (Calculated viscosity)
70 oC
Pressure (MPa)
1 2 3 4 5 6 7 8 9
Vis
co
sity (
mP
a.s
)
1e+0
1e+1
1e+2
1e+3
1e+4
1e+5
1e+6
1e+7
1e+8
1e+9
1e+10
Bitumen*Deasphalted bitumen*Asphaltene (Calculated viscosity)
118
5. The density and viscosity of asphaltene in wide ranges of temperature and pressure was
calculated using mixing rules applicable for simulation and engineering studies of
processes in which asphaltene precipitation occurs.
119
5.7 References
Akbarzadeh, K., Hammami, A., Kharrat, A., Zhang, D., Allenson, S., Creek, J., Kabir, S.,
Jamaluddin, A., Marshall, A.G., Rodgers, R.P., Mullins, O.C., Solbakken, T., 2007.
Asphaltenes -Problematic but Rich in Potential. Oilfield Rev. 19, 22–43.
Alboudwarej, H., Beck, J., Svrcek, W.Y., Yarranton, H.W., Akbarzadeh, K., 2002. Sensitivity of
Asphaltene Properties to Separation Techniques. Energy Fuels 16, 462–469.
Alian, S.S., Omar, A.A., Altaee, A.F., Hani, I., 2011. Study of Asphaltene Precipitation Induced
Formation Damage during CO2 Injection for a Malaysian light oil. Int. J. Chem. Mol. Nucl.
Mater. Metall. Eng. 5, 45–49.
Aslan, S., Firoozabadi, A., 2014. Effect of Water on Deposition, Aggregate Size, and Viscosity of
Asphaltenes. Langmuir 30, 3658–3664.
Barrera, D.M., Ortiz, D.P., Yarranton, H.W., 2013. Molecular Weight and Density Distributions
of Asphaltenes from Crude Oils. Energy Fuels 27, 2474–2487.
Centeno, G., Sanchez-Reyna, G., Ancheyta, J., Munoz, J.A.D., Cardona, N., 2011. Testing Various
Mixing Rules for Calculation of Viscosity of Petroleum Blends. Fuel 90, 3561–3570.
Dealy, J.M., 1979. Rheological Properties of Oil sand Bitumens. Can. J. Chem. Eng. 57, 677–683.
Diaz, O.C., Sánchez-Lemus, M.C., Schoeggl, F.F., Taylor, S.D., Yarranton, H.W., 2014. Deep-
vacuum Fractionation of Heavy Oil and Bitumen, part I: Apparatus and Standardized
Procedure. Energy Fuels 28, 2857–2865.
Foroughi, H., Acosta, E.J., Kawaji, M., 2011. A Miniature Cell for Gas Solubility Measurements
in Oils and Bitumen. Rev. Sci. Instrum. 82, 035104.
Kokal, S.L. Sayegh, S.G., 1995. Asphaltenes: The Cholesterol of Petroleum. SPE Middle East Oil
Show, Bahrain, 11-14 March.
Kokal, S.L., Sayegh, S.G., 1993. Phase Behaviour and Physical Properties of CO2-saturated Heavy
Oil and its Constitutive Fractions: Experimental Data and Correlations. J. Pet. Sci. Eng. 9,
289–302.
120
Lansangan, R.M., Smith, J.L., 1993. Viscosity, Density, and Composition Measurements of
CO2/West Texas Oil Systems. SPE Reserv. Eng. 8, 175–182.
Letcher, M. T., 2013. Future Energy: Improved, Sustainable and Clean Options for Our Planet; 2nd
ed., Elsevier.
Luo, P., Gu, Y., 2007. Effects of Asphaltene Content on the Heavy Oil Viscosity at Different
Temperatures. Fuel 86, 1069–1078.
Luo, P., Wang, X., Gu, Y., 2010. Characterization of Asphaltenes Precipitated with Three Light
Alkanes Under Different Experimental Conditions. Fluid Phase Equilib. 291, 103–110.
Mack, C., 1932. Colloid Chemistry of Asphalts. J. Phys. Chem. 36, 2901–2914.
Marufuzzaman, M., Henni, A., 2015. Solubility Of CO2 and C2H6 in heavy Oil and its SARA
Fractions. Can. J. Chem. Eng. 93, 553–564.
Mitchell, D.L., Speight, J.G., 1972. The Solubility of Asphaltenes in Hydrocarbon Solvents.
Energy Fuels 52, 149–152.
Mokrys, I.J., Butler, R.M., 1993. In-Situ Upgrading of Heavy Oils and Bitumen by Propane
Deasphalting : The Vapex Process. The Production Operation Symposium, Oklahoma City,
Oklahoma, USA 21-23 March.
Nasr, T.N., Ayodele, O.R., 2006. New Hybrid Steam-Solvent Processes for the Recovery of Heavy
Oil and Bitumen. International Petroleum Exhibition and Conference, Abu Dhabi, U.A.E, 5–
8 November.
Nenniger, J., Holcek, R., Dillon, J., Wolff, V., 2013. Solvent Injection Plant for Enhanced Oil
Recovery and Method of Operating Same, Canadian Patent, WO2013173907.
Nikooyeh, K., Bagheri, S.R., Shaw, J.M., 2012. Interactions Between Athabasca Pentane
Asphaltenes and n-alkanes at Low Concentrations. Energy Fuels 26, 1756–1766.
Shirani, B., Nikazar, M., Mousavi-Dehghani, S.A., 2012. Prediction of Asphaltene Phase
Behaviour in Live Oil with CPA Equation of State. Fuel 97, 89–96.
Stratiev, D., Shishkova, I., Tsaneva, T., Mitkova, M., Yordanov, D., 2016. Investigation of
121
Relations Between Properties of Vacuum Residual Oils from Different Origin, and of Their
Deasphalted and Asphaltene Fractions. Fuel 170, 115–129.
Vargas, F.M., Gonzalez, D.L., Hirasaki, G.J., Chapman, W.G., 2009. Modeling Asphaltene Phase
Behaviour in Crude Oil Systems Using the Perturbed Chain Form of the Statistical
Associating Fluid Theory (PC-SAFT) Equation of State. Energy Fuels 23, 1140–1146.
Wiehe, I.A., Kennedy, R.J., 2000. The Oil Compatibility Model and Crude Oil Incompatibility.
Energy Fuels 14, 56–59.
Yarranton, H.W., Masliyah, J.H., 1996. Molar Mass Distribution and Solubility Modeling of
Asphaltenes. AIChE J. 42, 3533–3543.
Yarranton, H.W., Ortiz, D.P., Barrera, D.M., Baydak, E.N., Barre, L., Frot, D., Eyssautier, J.,
Zeng, H., Xu, Z., Dechaine, G., Becerra, M., Shaw, J.M., Mckenna, A.M., Mapolelo, M.M.,
Bohne, C., Yang, Z, Oake, J., 2013. On the Size Distribution of Self-Associated Asphaltenes.
Energy Fuels, 27, 5083–5106.
Zirrahi, M., Azinfar, B., Hassanzadeh, H., Abedi, J., 2015. Measuring and Modeling the Solubility
and Density for CO2-Toluene and C2H6-Toluene Systems. J. Chem. Eng. Data 60, 1592–
1599.
Zirrahi, M., Hassanzadeh, H., Abedi, J., 2012. Prediction of Bitumen and Solvent Mixture
Viscosity Using Cubic-Plus-Association Equation of State, SPE Heavy Oil Conference,
Calgary, Alberta, Canada. 12-14 June.
122
Chapter Six: Combined Gel Permeation Chromatography and Simulated
Distillation for Characterization of Heavy Crude Oils and Residues
6.1 Preface
This chapter has been submitted for publication in peer-reviewed journal. This manuscript was co-
authored by M. Zirrahi, H. Hassanzadeh, and J. Abedi.
In this Chapter, new method is introduced to characterize the very and extra heavy oil samples
based on gel permeation chromatography and simulated distillation test results.
6.2 Abstract
Characterization of high molecular weight hydrocarbon mixtures such as heavy oil, bitumen, and
vacuum residue is essential to design and optimization of the recovery, upgrading, and
transportation processes. Characterization provides the information about the boiling point and
molecular weight distributions. To obtain these information, developing the simple, fast, and
consistent characterization method is needed. In this study, for the first time, the GPC and
simulated distillation (SimDist) test results are combined to characterize very heavy samples. In
the proposed characterization method, each sample is used individually as a standard sample for
GPC calibration by properly matching simulated distillation and GPC results. In other words, use
of a standard sample, which is usually polymer compound and not appropriate representative for
oil samples, is avoided. The GPC and simulated distillation results are coupled to generate the
calibration curve that correlates the retention time and the molecular weight. Employing the
obtained calibration curve for each sample and having the GPC results, the whole molecular weight
and boiling point distributions are obtained regardless of the complexity of the extra heavy sample.
The proposed characterization method is validated using a known standard sample (Reference Oil
ASTM D6352). Next, the proposed method is applied to obtain molecular weight distribution of
several bitumen samples and the bitumen fractions. The developed characterization method
provides a tool to find a better understanding of the molecular weight and boiling point
distributions of complex mixtures.
123
6.3 Introduction
Bitumen and heavy oils are very complex mixtures of light, medium, and heavy hydrocarbon
compounds with no specific recognized formula or structure. Therefore, it should be characterized
and defined as a well-defined mixture to be used in modeling and simulation studies of production
and refinery processes. Molecular weight distribution is a very useful tool for characterization of
complex mixtures such as heavy oil and bitumen samples. This information is the important data
for computation of thermodynamic properties and phase equilibria of the hydrocarbon systems.
There are several well-established methods for characterization of conventional crudes such as
ASTM D2887 and D1160, in which the conventional oils can be well characterized. The
recognized method to characterize the heavy crudes is simulated distillation (SimDist) under
standard of ASTM D7196. The boiling point distribution up to a temperature of 720°C can be
obtained using ASTM D7196. However, heavy and extra heavy crudes such as bitumen and
vacuum residue do not elute completely from the gas chromatography (GC) column. For example,
only about 80% of Athabasca bitumen elutes from GC column (Azinfar et al., 2017). The un-
distilled part plays a critical role in study of asphaltene precipitation, coke formation, and thermal
cracking processes.
Crude characterization is also important in thermodynamic and phase behavior calculations. The
oil is usually defined as some pseudocomponents. To estimate the critical properties and acentric
factor of the defined pseudocomponents in heavy oil and bitumen samples, boiling point or
molecular weight distributions are used. The obtained boiling point distribution by simulated
distillation is usually extended using the available correlations. Figure 6.1 compares the boiling
point distribution obtained by simulated distillation and correlations. As shown in this figure,
employing various correlations to extend the un-distilled part of bitumen results in different boiling
point distributions and inconsistent outcomes. Therefore, developing a general method to give a
consistent method and completely characterize the heavy oil, bitumen, and vacuum residue is
needed.
124
Figure 6.1: True boiling point extension of Athabasca bitumen using different correlations; The
green and red lines are the results of Twu (Twu, 1984) and Riazi and Al-Sahhaf (Riazi and Al-
Sahhaf, 1996); The blue circles are the experimental results of Athabasca bitumen.
Although measuring boiling point distribution is preferred to molecular weight distribution, it is
not feasible in case of complex mixtures such as extra heavy oil and bitumen samples and almost
half of the sample remains undetected in the form of residue. Moreover, some of hydrocarbon
molecules crack before boiling at high temperatures. For these components, degradation
temperature is lower than boiling temperature. Thus, they cannot be recognized and characterized
using the methods implementing at high temperature conditions.
Many attempts have been conducted in literature to characterize crude oils such as atmospheric
and vacuum distillation (Diaz et al., 2014), simulated distillation (Azinfar et al., 2015), gel
permeation chromatography (Such et al., 1979; Xu et al., 2014; Zhang et al., 2014), high
performance liquid chromatography (Such et al., 1979), SARA analysis (Woods et al., 2008), and
supercritical fluid extraction (Xu et al., 2014; Yang and Wang, 1999; Zhang et al., 2014).
Distillation is the preferred method in refining industry as it is feasible and economical and also
provides the extensive information about the oil fractions (Diaz et al., 2014). However, laboratory
distillation methods can be carried out at temperatures up to 350 oC to prevent the possible thermal
Tb (
oC)
200 400 600 800
wt.
%
0
20
40
60
80
100
125
cracking and decomposing problems (Riazi, 2005; Yang and Wang, 1999). For conventional oil
samples, this type of distillation can be more practical as the atmospheric and vacuum distillation
methods such as ASTM D86, D1160, and D2892 characterize 80-95 wt.% of the oil (Diaz et al.,
2014). However, for heavy and extra heavy oil, the mentioned distillation methods can provide the
information about a small fraction of the sample. Even the new developed high vacuum distillation
methods cannot distill more than 60 wt.% of heavy oil and natural bitumen samples (Diaz et al.,
2014). Increase of solvation power of a solvent near its critical region was employed to fractionate
heavy oil and vacuum residue samples in supercritical fluid extraction method. However, to
characterize the obtained fractions, various analyses must be performed including elemental
analysis, density and viscosity/temperature relationship, molecular weight by vapour pressure
osmometry (VPO) and GPC, boiling point curve by GC, SARA (saturate aromatic resin
asphaltene) analysis, and proton nuclear magnetic resonance (H NMR) (Xu et al., 2014; Yang and
Wang, 1999; Zhang et al., 2014). In other fractionation methods such as SARA analysis, the
compositional analysis of oil samples also produces inadequate and inconsistent information.
Chemical-based analysis of heavy oil and bitumen is not only inconsistent, but also applying
further characterization tests on obtained fractions is time and cost intensive.
Rodgers et al. (Rodgers et al., 1987) presented a correlation for predicting the molecular weight
distribution of high boiling hydrocarbon mixtures. They used GPC elution volume, hydrogen to
carbon ratio obtained by elemental analysis, and hydrogen distribution (α, β, γ) from NMR
spectroscopy. This method seems expensive and complicated. Because it is necessary to have
sufficient fractions of oil from GPC and obtain hydrogen to carbon ratios and hydrogen
distributions, which requires extensive analytical data including NMR and elemental analysis
measurements. Peramanu et al.(Peramanu et al., 1999) measured the molecular weight distribution
of Athabasca and Cold Lake bitumen samples and their SARA fractions using calibrated GPC with
polystyrene standards. They verified the results using VPO measurements and found the correction
factor for GPC distributions. Huang and Radosz also corrected the obtained molecular weights
from GPC to compensate the impropriety of polystyrene as standard and make the GPC-derived
results consistent with VPO (Huang and Radosz, 1991).
126
Champagne et al.(Champagne et al., 1985) compared the Athabasca bitumen molecular weight
distributions obtained by different solvent types and measurement techniques. They reported that
the GPC method is not suitable for determining the molecular weight distribution of bitumen as a
mixture of many different structures because no single calibration curve could be obtained. GPC
was also employed to characterize crude oils by Oelert et al.(Oelert et al., 1970). GPC together
with other methods such as VPO, high performance liquid chromatography (HPLC) and various
spectroscopy methods has been also used for characterization of asphaltene and oil heavy
component (Dettman et al., 2005; Leontaritis and Mansoori, 1989; Seidl et al., 2004). Moreover,
GPC coupled to aerosol matrix-assisted laser desorption/ionization (MALDI) (Fei and Murray,
1996; Hanton and Liu, 2000; Williams et al., 2003) and Fourier transform mass spectrometry
(FTMS) (Aaserud et al., 1999) were used for characterization and obtaining the molecular weight
distribution of polymers.
In this chapter, our goal is establishing a general method in which the whole molecular weight or
boiling point distribution of complex mixtures such as bitumen and heavy oils can be obtained by
coupling the results of the simple, fast, and low cost analytical tests; simulated distillation and
GPC. Combining the two characterization methods, the molecular weight or boiling point
distribution of bitumen and heavy oil samples and even more heavy compounds like asphaltene
and vacuum residue is achieved. GPC benefits by high reproducibility of runs, short run time which
results in low labor costs, and ability to be applied on wide variety of sample types even thermally
sensitive compounds (Oelert et al., 1970).
In the following parts of this chapter, the methodology of the present characterization method is
described, validated, and employed to characterize two bitumen samples. Then, Athabasca
bitumen and its fractions obtained by vacuum distillation and solvent fractionation methods are
characterized using the mentioned method.
6.4 Methodology
Gel permeation chromatography is a form of liquid chromatography and one of the versatile
analytical techniques for characterization of fluids specifically polymer fluids. Separation
mechanism in this chromatography method relies on the size of molecules in the solution (carrier
phase). A stagnant liquid in the pores of the beads in the chromatography column plays the role as
127
the stationary phase when the flowing liquid passes through the beads as mobile phase. If the
molecules are much larger than the biggest pores, they cannot enter to the pores and so are eluted
from the column by the carrier phase. On the other hand, small molecules can enter into many
pores in the beads and it takes a long time to pass through the column. Therefore, the small
molecules are eluted from the column slowly and the larger molecules, as can access few pores,
are eluted from the column quickly. To carry out the GPC test on oil samples, a calibration curve
has to be first generated. Typically, to determine the molecular weight of polymer samples,
calibration is carried out using standard polymers such as polystyrene and polyethylene with
known molecular weight. The polymers are not an appropriate calibration standard for petroleum
fluids characterization because of the difference between molecular structures of polymer and oil
samples. GPC separates molecules based on molecular size not molecular weight (Ferworn, 1995).
Therefore, molecules of different structure but the same molecular weight may elute at different
times which results in inaccurate molecular weight distribution of the oil samples. To cure the
mentioned problem, correction factors were used in previous studies (Ferworn, 1995; Huang and
Radosz, 1991; Peramanu et al., 1999). In the present work, each individual sample itself is
implemented as the calibration standard. The simulated distillation and GPC results of each sample
have been coupled to relate the molecular weight of components to retention time in order to create
the calibration correlation.
Differential refractive index is one of the common GPC detector, which works by measuring the
difference in refractive index of the mobile phase and the reference phase (pure solvent or carrier
phase). The refractive index parameters for n-alkylbenzene, n-alkylcyclopentane, and n-alkane
hydrocarbons are plotted versus the molecular weight in Figure 6.2 using the proposed correlation
by Riazi and Al-Sahhaf (Riazi and Al-Sahhaf, 1996). As shown in this figure, by increasing the
hydrocarbon molecular weight, the refractive indices approach a constant value. Therefore, the
detector response can be directly proportional to the sample concentration for heavy hydrocarbons
independent from their molecular structure. Therefore, with this explanation and considering the
behaviour shown in Figure 6.2, the light hydrocarbons should be ignored and the medium to heavy
hydrocarbons are suggested to use in generating the calibration curve. It should be noted that
normal alkane properties such as refractive index, molecular weight, and boiling point are used in
this work. Because the simulated distillation test (ASTM D7169) has been developed and
128
calibrated based on properties of normal alkane mixtures. The retention time standard is a mixture
of normal alkane hydrocarbons from C5 to C100. The crude samples contain millions of isomers
which cannot be detected and quantified. The justification behind using the normal alkane
properties for retention time standards is that the ASTM tests such as ASTM D2887 and D7169
are calibrated based on the properties of normal alkanes.
Figure 6.2: Refractive index parameter versus molecular weight of n-alkylbenzene, n-
alkylcyclopentane, and n-alkanes, calculated by correlations proposed by Riazi and Al-Sahhaf
(Riazi and Al-Sahhaf, 1996).
Figure 6.3 shows schematic of molecular weight distribution of a heavy oil sample. This plot has
been divided into three regions; (1) light to medium, (2) medium to heavy, and (3) heavy
hydrocarbon fractions. The first region indicates the light to medium components consisted of the
hydrocarbons with molecular weight less than about 400 g/mol. This section can be characterized
properly using simulated distillation (ASTM D7169) or ASTM D2887. The second region
corresponds the medium to heavy components and can be well-analyzed using the simulated
distillation results (ASTM D7169). The normal alkane hydrocarbons in this region have very close
refractive index (Figure 6.2). A linear relationship can be obtained between the measured refractive
index signal and concentration of sample in the carrier phase. Therefore, this region is used to
MW (g/mol)
10 100 1000
Re
fra
cti
ve
in
de
x p
ara
me
ter
0.20
0.22
0.24
0.26
0.28
0.30
0.32
n-alkanesn-alkylcyclopentanesn-alkylbenzenes
129
develop the GPC calibration curve in this work. The last region, which is shown by region (3), is
the main focus of this study. The simulated distillation results cannot give the insight of this part.
We use the GPC results to define and analyze this region.
Figure 6.3: Schematic of molecular weight distribution of bitumen sample.
As mentioned earlier, region (1) which corresponds to the light to medium hydrocarbon is not
considered for generating calibration correlation. As noted earlier, the proposed method in this
work uses the simulated distillation and GPC results of medium to heavy components (region (2)
in Figure 6.3) to develop the calibration curve. In other words, the calibration correlation is
developed using the data of region (2) where both GPC and simulated distillation tests are reliable.
After obtaining the calibration correlation, the region (3) can be characterized and presented as the
molecular weight of heavy components in sample.
In the proposed method, the output chromatograph of GPC has to be analyzed and converted to
the boiling point or molecular weight distribution using the calibration curve/correlation. Figure
6.4 shows a typical chromatograph obtained from GPC for a crude sample. This figure has been
obtained using a bitumen sample collected from an oil sand reservoir in Alberta, Canada. The
small peak at the end of the chromatograph shows the adsorbed water in tetrahydrofuran (THF)
MW (g/mol)
100 1000 10000
%O
ff
0
20
40
60
80
100
(1)
(2)
(3)
ASTM D7169ASTM D2887
ASTM D7169
GPC
130
during preparation of the samples. The first peak shows the existence of very heavy component
such as asphaltene in sample which will be described by detail in Results and Discussion.
Figure 6.4: GPC result of bitumen A: RID signal versus retention time.
The GPC result has been analyzed in the similar way as simulated distillation result. The area
under curve of each component is proportional to the amount of eluted sample from the GPC
column. Therefore, the total area of sample (area under the curve in Figure 6.4) is first calculated.
Then, the total area has been divided to meet the different weight percent of eluted components as
indicated in Figure 6.5. As mentioned earlier, the heavy components are eluted before the light
ones. Therefore, moving from the right to the left side of the chromatogram, the components
become heavier and the percentage of the eluted samples goes from 1 to 100 wt.%. The eluted
percentage in Figure 6.5 has been calculated using the following equation;
100amchromatogr of area Total
part each of area dAccumulate)Cumulative(% Off (6.1)
Retention time (min)
14 16 18 20 22 24 26 28 30 32 34
RID
sig
nal
(mV
)
0
5
10
15
20
25
131
Figure 6.5: GPC results of the Bitumen A: The area under curve is divided into various fractions.
Therefore, the percentage of the eluted sample (%Off) versus retention time is obtained after
dividing the area under the curve of chromatogram (Figure 6.5). The percentage of the eluted
sample (%Off) and the corresponded boiling point is the typical output of the simulated distillation
test. The boiling point distribution can be easily transformed to the molecular weight distribution
using the equation proposed by Riazi and Al-Sahhaf (Riazi and Al-Sahhaf, 1996) as given by;
])14
2(11193.09955.6exp[1090 3
2
MW
Tb
(6.2)
where Tb is boiling point in K. The experimental boiling points of n-alkanes (from ASTM D6352
or D7169) and the calculated values using correlation proposed by Riazi and Al-Sahhaf (Riazi and
Al-Sahhaf, 1996) are compared in Figure 6.6 (Absolute average relative deviation of 0.58 %). As
this figure shows, the proposed correlation can properly calculate the boiling points of the normal
paraffin.
Retention time (min)
14 16 18 20 22 24 26 28 30 32 34
RID
sig
na
l (m
V)
0
5
10
15
20
25
1%
5%
10%
15%
20%
25%
30%35%40%
45%50%
55%60%
65%70%
75%
80%
85%
90%95%
100%
132
Figure 6.6: Comparison between the measured boiling points reported in ASTM D6352 or D7169
and the calculated boiling points using Riazi and Al-Sahhaf correlation (Riazi and Al-Sahhaf,
1996).
The GPC and simulated distillation results of Bitumen A, used to generate the calibration curve,
are summarized in Table 6.1. The calibration curve is obtained by matching the overlapped %Off
in GPC and the simulated distillation results. At each percentage of eluted sample (%Off), the
corresponded retention time and the molecular weight are obtained from GPC and simulated
distillation results, respectively. Then, the retention time and molecular weight (Table 6.1) are
used to find the calibration correlation. Therefore, combining these two test results, the correlation
between molecular weight and the retention time is obtained.
MW (g/mol)
0 500 1000 1500 2000 2500
Tb
(K
)
200
400
600
800
1000
1200
Calculated boiling points using equation (6.2)Measured boiling points
133
Table 6.1. The required data to find the calibration correlation obtained by GPC and simulated
distillation. The tests were performed on a bitumen sample from Athabasca reservoir in Alberta,
Canada.
Off (%) GPC
SimDist
(ASTM
D7169)
t (min) MW (g/mol)
1 30.9500 186.3
5 29.5167 238.7
10 28.6500 279.0
15 28.0000 315.9
20 27.4833 354.1
25 27.0167 392.8
30 26.5833 435.8
35 26.1333 486.7
40 25.6833 549.3
45 25.2167 627.7
50 24.7500 721.8
55 24.2667 828.1
60 23.7333 934.4
65 23.1833 1039.3
70 22.5667 1187.9
75 21.8833 1407.9
80 21.1000 -
85 20.1667 -
90 19.0667 -
95 17.8167 -
100 15.2000 -
The general form of calibration correlation for GPC is as follow (Hanton and Liu, 2000; Stringano
et al., 2011);
batMW )log( (6.3)
where, MW and t are the molecular weight of each component and the retention time; a and b are
the constants and must be determined using simulated distillation and GPC results of each sample
(such as data presented in Table 6.1 for Bitumen A).
134
This correlation is called as calibration correlation that has to be obtained for each oil sample using
both simulated distillation and GPC tests. Then, the whole molecular weight distribution can be
achieved by applying calibration correlation for the sample.
Figure 6.7: The characterization scheme suggested in this study to obtain the whole molecular
weight distribution of heavy and extra heavy oil samples.
The summarized scheme of characterization proposed in this work was illustrated in Figure 6.7
and summarized in the following steps:
1. Performing GPC test on the crude sample and obtaining the sample chromatogram,
2. Transforming the GPC chromatogram to the percent of eluted sample data versus retention
time by dividing the area under the curve of chromatogram,
2- GPC: 4- SimDist:
7- Analyzed GPC:
1- GPC:
5- Calibration Curve:
batMW )log(
6- Calibration Correlation:
“a” and “b” are constant and have to be
determined for each sample.
3- SimDist:
MW (g/mol)
0 200 400 600 800 1000 1200 1400 1600
%O
ff
0
20
40
60
80
100
Tb(K)
500 600 700 800 900 1000
%O
ff
0
20
40
60
80
100
Retention time (min)
14 16 18 20 22 24 26 28 30 32
%O
ff
0
20
40
60
80
100
MW (g/mol)
100 1000 10000
%O
ff
0
20
40
60
80
100
Retention time (min)
14 16 18 20 22 24 26 28 30 32 34
RID
sig
nal (m
V)
0
5
10
15
20
25
t (min)
21 22 23 24 25 26 27 28 29
Lo
g (
MW
)
2.4
2.6
2.8
3.0
3.2
135
3. Performing simulated distillation test on the oil sample and obtaining percent of distilled
sample data versus boiling point,
4. Transforming the boiling point distribution to the molecular weight distribution using
equation (6.2),
5. Generating the calibration curve (MW versus retention time) by matching the %Off-
Retention time and %Off-MW data sets obtained from steps 2 and 4, respectively,
6. Obtaining the calibration correlation; the light components are not considered in calibration
curve and the rest is correlated as a line and the constants (a and b) are obtained,
7. Calculating the molecular weight distribution using the GPC results (step 2) and the
obtained calibration correlation (step 6).
The molecular weight variation of Bitumen A as function of retention time is illustrated in Figure
6.8. It was mentioned earlier that the refractive index parameter of light hydrocarbons is not
constant. Therefore, these points (as a rule of thumb for bitumen MW<400 g/mol) have not been
considered in generating the calibration curve. The points between the two dashed lines in Figure
6.8 show the linear trend and are considered as the points to build the correlation between the
molecular weight and the retention time.
Figure 6.8: Molecular weight variation of Bitumen A versus retention time.
t (min)
20 22 24 26 28 30 32
Lo
g (
MW
)
2.2
2.4
2.6
2.8
3.0
3.2
3.4 log(MW)=-0.1093t+5.5512
R2=0.9975
136
6.5 Results and Discussion
The GPC tests were conducted using a Waters Breeze 2 system liquid chromatograph equipped
with a differential refractometer detector (DRI) model 2414. Three equal size (7.8 mm×300 mm)
styragel columns (HR-1, 2, and 3; 5 μm) and a styragel guard column (20 μm, 4.6 mm×30 mm) in
series were used to obtain the distributions. The sample was diluted in tetrahydrofuran (THF) as
solvent at 10 mg/mL concentration. The operation was under a 0.1 mL/min flow rate at room
temperature.
In the following, first the proposed model is validated using a reference oil sample (Standard
5010). Then, the characterization method is employed to characterize two bitumen samples. In
addition, this method is applied on bitumen fractions obtained by vacuum distillation and solvent
fractionation.
6.5.1 Validation of the Proposed Model
The proposed method is validated using the known standard sample. The molecular weight
distribution of the standard oil sample generated using the proposed model is compared with the
experimental data reported in ASTM test method. We used the standard oil 5010, the Reference
oil for standard test ASTM D6352 and D7169, for the method validation. The GPC and simulated
distillation (ASTM D7169) results of the standard oil sample are shown in Figure 6.9 panels a and
b, respectively. In the validation process, the %Off data up to 80 % were used to generate the
calibration correlation (the red line in Figure 6.9(b)) and the rest of the data from 80 to 100% (the
green line in Figure 6.9(b)) was used to evaluate the proposed model by comparison of the
predicted and experimental data of %Off.
137
Figure 6.9: The required test results to apply the proposed characterization model on standard
sample; (a) GPC, (b) simulated distillation (the red portion of the curve shows the data used to
generate the calibration curve).
The chromatogram obtained from GPC were analyzed using the explained method in the
methodology part and the percentage of eluted sample versus the retention time was plotted as
presented in Figure 6.10 (the green triangles). In this figure, the simulated distillation results (the
pink circles) are also added to illustrate all the required data to apply the proposed model.
Figure 6.10: The GPC and simulated distillation results of standard oil together; the required data
to apply the proposed model.
(b)
Tb (K)
700 750 800 850 900 950
%O
ff
0
20
40
60
80
100(a)
Retention time (min)
20 22 24 26 28 30
RID
sig
nal (m
V)
0
10
20
30
40
50
60
MW (g/mol)
350 400 450 500 550 600 650 700 750
%O
ff
0
20
40
60
80
100
Retention time (min)
21 22 23 24 25 26 27 28
Simulated distillation
GPC
138
The calibration curve for the standard oil obtained using the data from simulated distillation and
GPC (Table 6.2 and Figure 6.10) are shown in Figure 6.11. Although all the simulated distillation
results (1 to 100 wt.%) are given in Table 6.2, the molecular weights corresponded to the 1 to 80
wt.% of the standard sample were only used in generating the calibration curve (the pink circles in
Figure 6.10).
Figure 6.11: Calibration curve for standard sample. (The circles are the points involved in
generating the calibration curve and the cross symbols are those are not considered.)
The four points on the right side in Figure 6.11 (showed by cross sign) which correspond to low
molecular weight hydrocarbons have not been considered and the remained points that fall on the
straight line were used to develop the calibration curve. The calibration curve of the standard oil
is obtained as;
1352.50957.0)log( tMW (6.4)
Using the obtained calibration correlation, %Off-Retention time curve (the green triangles in
Figure 6.10 obtained by GPC) can be converted to the %Off-MW data, which is the desired
molecular weight distribution. This distribution is plotted in Figure 6.12. This figure confirms the
validity of the proposed method. The blue symbols show the calculated molecular weight values
using the present method and the line plot is the molecular weight reported for the standard D6352
test. Considering this figure and the Absolute Relative Deviation (ARD) in Table 6.2, it can be
t (min)
23.5 24.0 24.5 25.0 25.5 26.0 26.5 27.0 27.5
Lo
g (
MW
)
2.55
2.60
2.65
2.70
2.75
2.80
2.85
2.90
log(MW)=-0.0957t+5.1352
R2=0.9981
139
concluded that the calculated molecular weight distribution for each point are well matched with
the reported values in the standard test.
Figure 6.12: The proposed model validation; the molecular weights of standard oil sample
obtained by simulated distillation and coupled GPC to simulated distillation.
Table 6.2: The data used in the characterization model for standard oil sample and the calculated
molecular weights along with the ARDs between the calculated molecular weight using the
proposed model and the molecular weights reported in ASTM D7169 test method.
Off (%)
GPC SimDist
(ASTM D7169)
Calculated using
the proposed
method ARD* (%)
t (min) MW (g/mol) MW
(g/mol)
1 27.2667 394.1 335.6 14.86
5 26.0833 473.3 435.5 7.98
10 25.6333 502.8 480.9 4.35
15 25.3667 520.4 510.1 1.98
20 25.1667 536.5 533.0 0.65
25 25.0000 553.3 553.0 0.06
30 24.8667 566.3 569.5 0.56
35 24.7500 581.9 584.3 0.40
40 24.6333 595.8 599.5 0.63
45 24.5333 610.0 612.9 0.47
50 24.4333 622.2 626.5 0.69
MW (g/mol)
200 400 600 800 1000 1200
%O
ff
0
20
40
60
80
100
Simulated distillation (<80%)Simulated distillation (>80%)Coupled GPC to simulated distillation
140
55 24.3333 637.2 640.5 0.51
60 24.2333 652.8 654.7 0.30
65 24.1333 668.8 669.3 0.08
70 24.0167 685.3 686.8 0.21
75 23.9167 702.4 702.1 0.04
80 23.8000 723.0 720.3 0.37
85 23.6500 747.7 744.5 0.42
90 23.4833 776.8 772.4 0.57
95 23.2333 825.5 816.1 1.13
100 21.4000 988.0 1222.4 23.73
*SimDistCalculatedSimDist
MWMWMWARD /)(100
6.5.2 Molecular Weight Distributions of Two Bitumen Samples
In this part, the proposed model is employed to predict the molecular weight distributions of two
bitumen samples, called as Bitumens A and B. The GPC chromatograms and the simulated
distillation results of two bitumen samples are illustrated in Figure 6.13 (a and b), respectively.
The peak observed at the beginning of the chromatographs shows the existence of very heavy
components such as asphaltene in the bitumen samples. Bitumen B is expected to have more
asphaltene content compared to Bitumen A, as seen in Figure 6.13(a). The asphaltene contents of
the bitumen samples were measured by implementation of ASTM D2007. The results showed the
asphaltene contents of 11.8 and 14.7 wt.% for Bitumens A and B, respectively, which is in
agreement with the results shown in Figure 6.13(a). As shown in Figure 6.13(b), the maximum
percentage of eluted components for Bitumens A and B using test ASTM D7169 are 75 and 73
wt.%, respectively.
141
Figure 6.13: Characterization test results of Bitumens A and B; (a) GPC chromatograms and (b)
Simulated distillation.
Applying the mentioned method in this work, the area under the curves of Figure 6.13(a) was
divided to different percentages. The GPC and simulated distillation data, used in this
characterization method, are shown in Figure 6.14 for Bitumens A and B.
Figure 6.14: The GPC and simulated distillation test results of Bitumens A and B used in
characterization method.
Tb (K)
500 600 700 800 900 1000
%O
ff
0
10
20
30
40
50
60
70
80
Bitumen ABitumen B
Retention time (min)
16 18 20 22 24 26 28 30 32 34
RID
sig
nal (m
V)
0
5
10
15
20
25
Bitumen A
Bitumen B
(a) (b)
MW (g/mol)
0 200 400 600 800 1000 1200 1400
%O
ff
0
20
40
60
80
100
14 16 18 20 22 24 26 28 30 32
Simulated distillation
GPC
Retention time (min)
Bitumen A
MW (g/mol)
0 200 400 600 800 1000 1200 1400
%O
ff
0
20
40
60
80
100
Retention time (min)
14 16 18 20 22 24 26 28 30 32 34
Simulated distillationGPC
Bitumen B
142
The obtained calibration curves using the proposed method are plotted in Figure 6.15. This figure
shows a well R-squared (R2) between the data and regressed calibration curve. The obtained
retention time from GPC test, the corresponding molecular weight using simulated distillation test,
and the calculated molecular weight distribution using the calibration curves are summarized in
Table 6.3.
Figure 6.15: The calibration curves obtained for Bitumens A and B.
Employing the obtained calibration curves for bitumen samples, the molecular weight for each
point is calculated and summarized in Table 6.3. The molecular weight distributions calculated in
this work along with the simulated distillation test are compared in Figure 6.16.
Table 6.3: The required data and the calculated molecular weight distribution of Bitumens A and
B.
Off (%)
Bitumen A Bitumen B
GPC SimDist
Calculated using
the proposed
method ARD
(%)
GPC SimDist
Calculated using
the proposed
method ARD
(%)
t (min) MW
(g/mol) MW (g/mol) t (min)
MW
(g/mol) MW (g/mol)
1 30.9500 186.3 147.4 20.92 31.7000 176.2 122.8 30.32
5 29.5167 238.7 211.4 11.46 30.0667 224.1 184.2 17.79
10 28.6500 279.0 262.9 5.79 29.0833 262.2 235.2 10.31
15 28.0000 315.9 309.6 1.98 28.3667 297.8 281.0 5.65
Retention time (min)
20 22 24 26 28 30 32
Lo
g(M
W)
2.2
2.4
2.6
2.8
3.0
3.2
Retention time (min)
20 22 24 26 28 30 32 34
Lo
g(M
W)
2.2
2.4
2.6
2.8
3.0
3.2
Bitumen A Bitumen B
log(MW)=-0.1093t+5.5512
R2=0.9975
log(MW)=-0.1079t+5.5095
R2=0.9979
143
20 27.4833 354.1 352.6 0.43 27.7833 334.2 324.8 2.79
25 27.0167 392.8 396.5 0.95 27.2833 375.0 367.8 1.91
30 26.5833 435.8 442.2 1.47 26.7833 415.2 416.5 0.31
35 26.1333 486.7 495.3 1.75 26.2833 461.9 471.6 2.09
40 25.6833 549.3 554.6 0.98 25.7833 518.8 533.9 2.92
45 25.2167 627.7 623.8 0.62 25.2833 587.6 604.5 2.88
50 24.7500 721.8 701.5 2.82 24.7667 675.9 687.3 1.70
55 24.2667 828.1 792.2 4.33 24.2333 781.5 784.7 0.41
60 23.7333 934.4 906.0 3.03 23.6667 909.9 903.4 0.72
65 23.1833 1039.3 1040.5 0.12 23.0500 1067.5 1052.9 1.37
70 22.5667 1187.9 1215.2 2.30 22.4000 1255.2 1237.5 1.41
75 21.8833 1407.9 1443.3 2.51 21.6500 - 1491.0 -
80 21.1000 - 1757.8 - 20.8000 - 1841.5 -
85 20.1667 - 2223.2 - 19.8333 - 2341.5 -
90 19.0667 - 2932.3 - 18.7167 - 3090.1 -
95 17.8167 - 4016.4 - 17.7167 - 3961.6 -
100 15.2000 - 7759.6 - 14.9167 - 7943.1 -
Figure 6.16: The comparison of the predicted molecular weight distribution of Bitumens A and B
using the proposed method and the distribution obtained by simulated distillation.
6.5.3 Application of the Proposed Model on Athabasca Bitumen Fractions
The GPC and simulated distillation results on two sets of Athabasca bitumen fractions described
in Chapters 2,3, and 4 are compared in this part. Table 6.4 summarizes bitumen cut specifications.
MW (g/mol)
100 1000 10000
%O
ff
0
20
40
60
80
100
SimDistCoupled GPC to SimDist
Bitumen A
MW (g/mol)
100 1000 10000
%O
ff
0
20
40
60
80
100
SimDistCoupled GPC to SimDist
Bitumen B
144
Table 6.4: The Athabasca bitumen cuts properties.
Sample Distillation T (oC) Weight percent
(wt.%)
MW
(g/mol)a
MW
(g/mol)b Cut 1 195> T 19.4 268.8 276.7
Cut 2 195< T <250 11.7 365.5 379.4
Cut 3 250<T <350 18.4 464.6 509.3
Cut 4-deasphaltedc T >350 37.5 906.1 1169.0
Asphaltene - 13.0 - 2731.5 a The measured molecular weight using cryoscope b The measured weight-average molecular weight using GPC c Cut 4 without asphaltene
The obtained GPC chromatograms of the bitumen sample and its fractions using the proposed
method in this study can be analyzed. The calibration curves of each sample was found and the
analyzed GPC results are shown in Figure 6.17. As expected, moving from left to the right side,
Cuts 1, 2, 3, and 4 were respectively observed. An interesting observation in Figure 6.17 is the
molecular weight distribution of Cut 4 contradicts a common opinion that considers the residues
(Cut 4) composed mostly of very high molecular weight components. The results reveal a broad
molecular weight distribution for the heavy fraction.
Figure 6.17: The coupled GPC to SimDist results on whole bitumen and its fractions obtained by
vacuum distillation.
MW (g/mol)
100 1000
RID
sig
na
l (m
V)
0
10
20
30
40
50
Cut 1
Cut 2
Cut 3
Cut 4
Whole Bitumen
145
To compare the GPC and the simulated results, the %Off versus the molecular weight for each
sample is plotted in Figure 6.18. The simulated distillation results (ASTM D7169) cover the
molecular weight ranges up to 1400 g/mol while GPC goes up to about 5000 g/mol for the
components in Cut 4. The covered ranges by the simulated distillation for all samples in Figure
6.18 are in very good agreement with the one obtained by GPC.
Figure 6.18: The moleclar weight distributions of whole bitumen and its fractions obtained by
vacuum distilation using; (a) GPC and (b) Simulated distillation tests.
The obtained average molecular weights in this work (presented in Table 6.4) and the previous
reported results (Huang and Radosz, 1991; Seidl et al., 2004; Zhang et al., 2014) reveal that the
MW measured by GPC is between the MW measured by cryoscopy (based on freezing point
depression concept) and the VPO methods. However, cryscopy and VPO only give an average
MW, not the molecular weight distribution.
The advantage of GPC compared to simulated distillation is more pronounced when the samples
are getting heavier. To see the difference, the GPC and simulated distillation tests on whole
bitumen, Cut 4, de-asphalted Cut 4, and asphaltene itself are carried out and the GPC results are
shown in Figure 6.19. Comparing the choromatograms of the whole bitumen, Cut 4 and Cut 4-
deasphalted, an interesting point is observed. For Cut 4 and the whole bitumen (red and black,
respectively), a peak signal is observed at the end of the choromatograms while for the Cut 4-
a) Obtained by Coupled GPC to SimDist
MW (g/mol)
100 1000 10000
%O
ff
0
2
4
6
8
10
MW (g/mol)
100 1000
%O
ff
0
2
4
6
8
10
b) Obtained by SimDist
Cut 1
Cut 2
Cut 3
Cut 4
Whole bitumen
Cut 1Cut 2
Cut 3
Cut 4Whole bitumen
146
deasphalted (green) the peak disappears. This signal peak can be attributed to the existence of
very heavy components in whole bitumen and Cut 4. Since the asphaltene is separated in Cut 4-
deasphalted, these heavy components are removed and the peak is vanished.
Figure 6.19: The coupled GPC to SimDist results of whole bitumen and its heavy fractions.
The obtained weight fraction of the recovered samples versus molecular weights of components
using proposed method and simulated distillation test are illustrated in Figure 6.20 (a and b),
respectively. The results reveal that the only very small ranges of components have been covered
in simulated distillation results for asphaltene, Cut 4, and Cut 4-deasphalted. However, GPC was
successfully able to extend the distribution and provide more detailed information about the
molecular weight distribution of asphaltene and the extra heavy cuts.
MW (g/mol)
100 1000 10000
RID
Sig
na
l (m
V)
0
5
10
15
20
25
Whole Bitumen
Cut 4-deasphalted
Cut 4
Asphaltene
147
Figure 6.20: The molecular weight distributions of whole bitumen and its heavy fractions;
Obtained by (a) Coupled GPC to SimDist and (b) SimDist.
The comparison between the predicted molecular weights and the molecular weights obtained by
simulated distillation test are shown in Figure 6.21 for Cut 4, Cut 4-deasphalted, whole bitumen,
and asphaltene. This figure shows the good agreement between the simulated distillation results
and the calculated results by the proposed model. The distillable fractions of bitumen (Cuts 1, 2,
and 3) are well characterized using simulated distillation only and applying GPC test is not
necessary to characterize them. Therefore, the similar graphs as Figure 6.21 for these fractions do
not add more information and are not included here.
a) Obtained by coupled GPC to SimDist
MW (g/mol)
100 1000 10000
%O
ff
0.0
0.5
1.0
1.5
2.0
b) Obtained by SimDist
MW (g/mol)
100 1000
%O
ff
0.0
0.5
1.0
1.5
2.0
Whole bitumen
Cut 4-deasphalted
Cut 4
Asphaltene
Whole bitumen
Cut 4-deasphalted
Cut 4
Asphaltene
148
Figure 6.21. The accumulated %Off versus molecular weight. The open symbols are the predicted
molecular weights using the proposed method in this work and the filled symbols represent the
results of simulated distillation test.
6.6 Summary and Conclusions
A new characterization method of very complex mixtures was proposed by combining gel
permeation chromatography and simulated distillation for characterization of complex mixtures.
The developed characterization method offers a complete analysis of molecular weight distribution
Cut 4
MW (g/mol)
100 1000 10000
%O
ff
0
20
40
60
80
100
Cut 4-deasphalted
MW (g/mol)
100 1000 10000
%O
ff
0
20
40
60
80
100
Asphaltene
MW (g/mol)
1e+2 1e+3 1e+4 1e+5
%O
ff
0
20
40
60
80
100
Whole bitumen
MW (g/mol)
100 1000 10000
%O
ff
0
20
40
60
80
100
149
for heavy oil and extra heavy oil such as bitumen and vacuum residue samples. The result of the
detailed simulated distillation and the gel permeation chromatography analysis were combined to
build the calibration curve. Employing the generated calibration correlation and the GPC results,
the molecular weight distribution can be then obtained. This method is able to provide useful
information on the molecular weight distribution for the very heavy mixtures such as vacuum
residue and asphaltene, not obtainable otherwise using simulated distillation. The model was
properly validated using a standard reference sample and then was employed to calculate the
molecular weight distribution of several bitumen samples and bitumen fractions. The results of the
present method shed light on molecular weight distribution of very complex mixtures such as
asphaltene and vacuum residue and will find applications in many engineering and science field
related to bitumen and heavy oil production, transportation, and refinery. The developed approach
can also be applied to other areas of materials characterization where partial information on
molecular distribution of complex mixtures can be extended by combining gel permeation
chromatography and simulated distillation to obtain more detailed information.
150
6.7 References
Aaserud, D.J., Prokai, L., Simonsick, W.J., 1999. Gel Permeation Chromatography Coupled to
Fourier Transform Mass Spectrometry for Polymer Characterization. Anal. Chem. 71, 4793–
4799.
Azinfar, B., Haddadnia, A., Zirrahi, M., Hassanzadeh, H., Abedi, J., 2017. Effect of Asphaltene
on Phase Behavior and Thermophysical Properties of Solvent/Bitumen Systems. J. Chem.
Eng. Data 62, 547–557.
Azinfar, B., Zirrahi, M., Hassanzadeh, H., Abedi, J., 2015. A Method for Characterization of
Bitumen. Fuel 153, 240–248.
Champagne, P.J., Manolakis, E., Ternan, M., 1985. Molecular Weight Distribution of Athabasca
Bitumen. Fuel 64, 423–425.
Dettman, H., Inman, A., Salmon, S., Scott, K., Fuhr, B., 2005. Chemical Characterization of GPC
Fractions of Athabasca Bitumen Asphaltenes Isolated Before and After Thermal Treatment.
Energy Fuels 19, 1399–1404.
Diaz, C., Sánchez-Lemus, M.C., Schoeggl, F., Satyro, M.A., Taylor, S.D., Yarranton, H.W., 2014.
Deep-vacuum Fractionation of Heavy Oil and Bitumen, part I: Apparatus and Standardized
Procedure. Energy Fuels 28, 2857–2865.
Fei, X., Murray, K.K., 1996. On-Line Coupling of Gel Permeation Chromatography with MALDI
Mass Spectrometry. Anal. Chem. 68, 3555–3560.
Ferworn, K.A., 1995. Thermodynamic and Kinetic Modeling of Asphaltene Precipitation from
Heavy Oils and Bitumen. PhD Thesis, University Of Calgary, Calgary, Alberta.
Hanton, S.D., Liu, X.M., 2000. GPC Separation of Polymer Samples for MALDI Analysis. Anal.
Chem. 72, 4550–4554.
Huang, S.H., Radosz, M., 1991. Phase Behavior of Reservoir Fluids IV: Molecular Weight
Distributions for Thermodynamic Modeling. Fluid Phase Equilib. 66, 23–40.
Leontaritis, K.J., Mansoori, G.A., 1989. Fast Crude-Oil Heavy-Component Characterization Using
151
Combination of ASTM, HPLC, and GPC Methods. J. Pet. Sci. Eng. 2, 1–12.
Oelert, H.H., Latham, D.R., Haines, W.E., 1970. Characterization of Crude Oils by Gel Permeation
Chromatography. Sep. Sci. 5, 657–668.
Peramanu, S., Pruden, B.B., Rahimi, P., 1999. Molecular Weight and Specific Gravity
Distributions for Athabasca and Cold Lake Bitumens and Their Saturate, Aromatic, Resin,
and Asphaltene Fractions. Ind. Eng. Chem. Res. 38, 3121–3130.
Riazi, M.R., 2005. Characterization and Properties of Petroleum Fractions, 1st ed. ASTM
International, USA.
Riazi, M.R., Al-Sahhaf, T.A., 1996. Physical Properties of Heavy Petroleum Fractions and Crude
Oils. Fluid Phase Equilib. 117, 217–224.
Rodgers, P.A., Creagh, A.L., Prange, M.M., Prausnitz, J.M., 1987. Molecular Weight Distribution
for Heavy Fossil Fuels from Gel-Permeation Chromatography and Characterization Data.
Ind. Eng. Chem. Res. 26, 2312–2318.
Seidl, P.R., Chrisman, E.C.A.N., Silva, R.C., de Menezes, S.M.C., Teixeira, M.A.G., 2004.
Critical Variables for the Characterization of Asphaltenes Extracted from Vacuum Residues.
Pet. Sci. Technol. 22, 961–971.
Stringano, E., Gea, A., Salminen, J.-P., Mueller-Harvey, I., 2011. Simple Solution for a Complex
Problem: Proanthocyanidins, Galloyl Glucoses and Ellagitannins Fit on a Single Calibration
Curve in High Performance-Gel Permeation Chromatography. J. Chromatogr. A 1218, 7804–
7812.
Such, C., Brule, B., Baluja-santos, C., 1979. Characterization of a Road Asphalt by
Chromatographic Techniques (GPC and HPLC). J. Liq. Chromatogr. 2, 437–453.
Twu, C.H., 1984. An Internally Consistent Correlation for Predicting The Critical Properties and
Molecular Weights of Petroleum and Coal-tar Liquids. Fluid Phase Equilib. 16, 137–150.
Williams, J.B., Chapman, T.M., Hercules, D.M., 2003. Matrix-Assisted Laser
Desorption/Ionization Mass Spectrometry of Discrete Mass Poly(butylene glutarate)
152
Oligomers. Anal. Chem. 75, 3092–3100.
Woods, J., Kung, J., Kingston, D., Kotlyar, L., Sparks, B., Mccracken, T., 2008. Canadian Crudes:
A Comparative Study of SARA Fractions From a Modified HPLC Separation Technique. Oil
Gas Sci. Technol. 63, 151–163.
Xu, Z., van den Berg, F.G.A., Sun, X., Xu, C., Zhao, S., 2014. Detailed Characterization of Virgin
Heavy Oil Resid and Its Thermally Cracked Resid. Energy Fuels 28, 1664–1673.
Yang, G., Wang, R.A., 1999. The Supercritical Fluid Extractive Fractionation and The
Characterization of Heavy Oils and Petroleum Residua. J. Pet. Sci. Eng. 22, 47–52.
Zhang, Y., Zhang, L., Xu, Z., Zhang, N., Chung, K.H., Zhao, S., Xu, C., Shi, Q., 2014. Molecular
Characterization of Vacuum Resid and Its Fractions by Fourier Transform Ion Cyclotron
Resonance Mass Spectrometry with Various Ionization Techniques. Energy Fuels 28, 7448–
7456.
153
Chapter Seven: A Method for Characterization of Bitumen
7.1 Preface
This chapter was published in Fuel, 2015, volume 153, 240-248. This manuscript was co-authored
by M. Zirrahi, H. Hassanzadeh, and J. Abedi. A copy of the copyright permission from the
publisher to reproduce this manuscript in the present thesis is provided in Appendix B.
In this chapter, residue curve map is used to characterize bitumen and the results of the proposed
model are compared with the experimental data and the results of previous models.
7.2 Abstract
Characterization of bitumen is a necessary step to perform phase equilibrium computations
involved in bitumen production and processing. This study presents a methodology for bitumen
characterization using a residue curve map. A fugacity-activity coefficient approach is applied to
model the thermodynamic equilibrium of the species in the gas and liquid phases. The Peng-
Robinson equation of state (PR-EoS) and NRTL activity model are utilized to calculate the
fugacity and activity coefficients in the gas and liquid phases, respectively. The proposed model
was evaluated using experimental simulated distillation (SD) data as well as experimental
solubility data of light hydrocarbon (CH4, C2H6) and non-hydrocarbon gases (CO2, N2) in bitumen.
The molecular weight, specific gravity, and SD curves were represented using defined
pseudocomponents. The tuned EoS was able to regenerate the solubility data with an acceptable
accuracy (AARD less than 5.6 and 4.5% for hydrocarbon and non-hydrocarbon solvents,
respectively). The proposed method can be applied for bitumen characterization and to predict the
solubility of the light gases in bitumen.
154
7.3 Introduction
Increase of worldwide energy demand besides the decline of conventional oil reserves resulted
more attention toward bitumen and heavy oil recourses. Currently, thermal oil recovery is the
leading processes for bitumen and heavy oil extraction in Alberta. Disadvantages of thermal
recovery methods such as high energy and water consumption, water pollution and greenhouse gas
emission have raised the interests to find alternative methods such as solvent injection.
Characterization of bitumen is the initial step for all thermodynamic and phase behaviour
calculations. Crude oil assay obtained from distillation experiments is the common tool to
characterize hydrocarbon mixtures and petroleum fractions. The underlying idea is that
hydrocarbon molecules, which boil at a range of temperature, are lumped into a pseudocomponent
and assigned an average boiling temperature (Riazi, 2005; Satyro and Yarranton, 2009).
Bitumen and extra heavy oil assays should be obtained under deep vacuum distillation to avoid
possible thermal cracking at temperatures higher than 523 K. Standard tests such as ASTM D5236
test and based on gas chromatography (GC) have been developed to characterize high boiling point
hydrocarbon mixtures. Simulated distillation is fast, repeatable, and inexpensive. Additionally, a
small amount of bitumen/heavy oil sample is required for the analysis (Riazi, 2005).
Saajanlehto and Alopaeus (Saajanlehto and Alopaeus, 2014) and Saajanlehto et al. (Saajanlehto et
al., 2014) have used distillation curve data to develop a characterization method for Perturbed-
Chain Statistical Associating Fluid Theory (PC-SAFT). Their model has been verified to represent
density, saturation pressures for propane-Athabasca bitumen and CO2-propane-Athabasca bitumen
systems and solubility of hydrogen in heavy oil. They proposed a single carbon number type
characterization method for heavy oils and bitumen. In their model, each cut of crude oil was split
into saturate, aromatic and polyaromatic (SAP) fractions based on its boiling point. Therefore,
distribution of SAP in crude oils is assumed universal.
Doherty and Perkins developed the residue curve map to model the simple distillation assay
(Doherty and Perkins, 1979, 1978a, 1978b). This model calculates the loci of saturation
temperatures of the mixture versus the composition of the liquid phase in the flask during the
distillation. To apply this model there is no need to know the exact condition and procedure of the
155
distillation process. Only pressure at which the test has been carried out and the resulting
experimental distillation curve are required. This is the remarkable advantage of the model
developed by Doherty and Perkins.
Residue curve map has been used by Satyro and Yarranton to characterize several oil samples
using distillation curve data (Satyro and Yarranton, 2009). They assumed ideal gas and liquid
phases in their thermodynamic equilibrium calculations. While this assumption is quite well
applicable for the gas phase, it may not be valid for the liquid phase especially for heavy oil and
bitumen. Ortiz et al. used the residue curve map upon well-defined mixtures to tune equation of
state parameters and then calculated the phase behaviour data (Ortiz et al., 2013).
One of the important parameters in heavy oil and bitumen applications is the solubility of light
components in bitumen or heavy oil. Prediction of solubility of gaseous solvents in bitumen is key
for reservoir simulation and engineering studies of bitumen production, upgrading, fractionation
and refining. Once the solubility of solvent in bitumen is defined, density, viscosity and other
physical properties can be calculated using the established methods. In previous solubility models
bitumen has been considered as single pseudocomponent (such as (Zirrahi et al., 2014); (Fu et al.,
1986)) or multicomponents (such as (Kariznovi et al., 2010); (Eastick et al., 1992); (Huang and
Radosz, 1991); (Mehrotra and Svrcek, 1988a)).
After definition of the pseudocomponents, a proper equation of state should be tuned to find binary
interaction parameters (kij) between presented molecules in the system. The tuned equation of state
then can be used to find the thermodynamic and thermophysical properties required for
engineering calculations. Tuning of equation of state for conventional oils is typically performed
using the data obtained from PVT experiments such as constant composition expansion (CCE),
differential liberation and separator tests. For bitumen, aforementioned laboratory tests are not
common. Therefore, there are no independent calculations to find and tune the interaction
parameters between pseudocomponents. Some studies estimated the interaction parameters using
correlations while some ignored these interaction parameters (Huang and Radosz, 1991; Mehrotra
and Svrcek, 1988a).
Kariznovi et al. defined the pseudocomponents and their properties such as molecular weight,
specific gravity to match the average molecular weight and specific gravity of bitumen (Kariznovi
156
et al., 2010). They tuned the normal boiling point, acentric factor, and EoS parameters (ΩA and
ΩB) of pseudocomponents and the binary interaction parameters between solvent-
pseudocomponent (ks-pc) to match the experimental solubility data.
Eastick et al. (Eastick et al., 1992) and Huang and Radosz (Huang and Radosz, 1991) characterized
Cold Lake bitumen using five pseudocomponents to calculate solubility of CO2 in the bitumen and
its pseudocomponents. They applied Peng-Robinson equation of state (PR-EoS) and statistical
associating fluid theory (SAFT), respectively. Mehrotra and Svrcek divided Cold Lake bitumen
into three pseudocomponents, including asphaltene, distillable and undistillable maltene (Mehrotra
and Svrcek, 1988a). They used PR-EoS to model the gas and liquid phases and ignored the
interaction between pseudocomponents.
In this study, we propose an approach using the simulated distillation (SD) data to characterize the
bitumen by applying the residue curve map. We apply the φ-γ (φ is the fugacity coefficient of
component in the gas phase and γ is the activity coefficient of component in the liquid phase)
approach to model the thermodynamic equilibrium of gas and liquid phases. We use the NRTL
activity model to capture the non-ideal behaviour of bitumen molecules. The major difficulty to
apply NRTL model is the introduction of additional binary interaction energy parameters. While
some methods have been presented in literature to find these parameters, their application to
complex mixture such as bitumen has not been confirmed. To overcome this problem, we
determine the interaction energy parameters between the pseudocomponents of bitumen using the
residue curve map method.
After characterization of the bitumen, we follow the φ-γ approach to calculate the solubility of
light hydrocarbon and non-hydrocarbon gases (solvents) in bitumen. Using this approach, the
energy interaction parameters between molecules of the solvent and the pseudocomponents are the
only required parameters to be tuned for solubility calculations.
In summary, we propose a general approach with a solid thermodynamic background to
characterize bitumen. The developed method can be used to calculate the required interaction
parameters and solubility of light solvents in bitumen. Application of this methodology can be an
alternative for expensive laboratory PVT tests. Therefore, it will decrease the costs of bitumen
157
characterization and find many applications in phase behaviour studies related to bitumen and
extra heavy oil production and refining.
The rest of this chapter is organized as follow: first, we present our methodology to characterize
bitumen. Then, solubility model is presented. Next, accuracy of the approach is evaluated using
experimental data and discussed followed by the summary and conclusion.
7.4 Bitumen Characterization Procedure
We use the results obtained from simulated distillation assay to characterize the bitumen. The SD
test results in mass percent of the distilled (%Off) versus the temperature. Residue curve map,
which is the trajectory of bubble point temperature of the mixture from the initial to the end state
of batch distillation, is applied to represent the simulated distillation results. As described by
Doherty and Perkins (Doherty and Perkins, 1979, 1978a, 1978b) liquid composition changes
during the simple batch distillation can be represented by:
xT
d
d,,
Fxyx
x (7.1)
where x = [x1, x2, xs, …., xN] and y = [y1, y2, ys, …., yN] are the liquid and gas mole fraction arrays,
respectively, N is the number of pseudocomponents, ξ is the dimensionless time called warped
time varying from 0 to +∞ where ξ=0 corresponds to the start of the distillation (t=0) and ξ=+∞
corresponds to the time when the distillation flask becomes empty. Liquid hold-up in the flask is
calculated by (Doherty and Perkins, 1978a):
)exp(0 HH (7.2)
where H is the liquid hold-up and subscript 0 denotes the initial hold-up. At each warped time, the
liquid phase is at saturation condition (T=Tsat) and the gas phase composition can be obtained
using the saturation temperature equilibrium calculations. The set of equations 7.1 and 7.2 was
solved using fourth-order Runge-Kutta method with the warped time interval Δξ as:
)22(6
143211 kkkkxx nn (7.3)
where
158
nn xfk ,1
(7.4)
12
2
1,
2
1kxfk nn
(7.5)
23
2
1,
2
1kxfk nn
(7.6)
34 , kxfk nn
(7.7)
),( PxfT n
sat (7.8)
nn
sat
n xxTKy ),( (7.9)
1nn (7.10)
nn HH exp0 (7.11)
100%
0
0
MH
MHMHOff nn (7.12)
where M is the molecular weight, n is the step index and Δξ is the integration step of the warped
time. This parameter (Δξ) was set to 0.005 as the optimum integration step for the sake of accuracy
and low run time for calculations. Tsat and K are the saturation temperature at the distillation
pressure and equilibrium constant at Tsat and P, respectively. To find the equilibrium constant,
ideal gas phase was assumed (=1). Therefore, thermodynamic equilibrium can be written as:
sat
iiii PxPy (7.13)
where Psat and γi are the saturation pressure and the activity coefficient of component i in the liquid
phase, respectively. To find the saturation pressure, Lee-Kesler equation was used as given by
(Lee and Kesler, 1975):
10 ln sat
rP (7.14)
159
60 169347.0ln28862.109648.6
92714.5 rr
r
TTT
(7.15)
61 43577.0ln4721.136875.15
2518.15 rr
r
TTT
(7.16)
where Tr and Pr are the reduced temperature (Tr=T/Tc) and pressure (Pr=P/Pc), respectively, and
ω is the acentric factor. Furthermore, NRTL model was applied to calculate the activity coefficient
of components in the mixture as given by (Wong and Sandler, 1992):
k
kjk
l
ljljl
ij
j
k
kjk
jjj
k
kik
j
jijij
iGx
Gx
Gx
Gx
Gx
Gx
)ln( (7.17)
where Gij is a characteristic of interaction energies between molecules of i and j given by (Wong
and Sandler, 1992):
)exp( ijijijG (7.18)
RT
gg jjij
ij
(7.19)
where g is the energy interaction parameter and α specifies the randomness in the mixture. In this
work α=0.3 was assumed for all pseudocomponents of bitumen (Prausnitz et al., 1986). Energy
interaction parameter between molecules i and j (gij) was calculated assuming the geometric mean
proposed by Vetere (Vetere, 2004) as:
iijjij ggg
(7.20)
Tassios proposed defining the energy parameter of molecule i by the heat of vaporization (ΔHvap)
as follows (Tassios, 1971):
)( RTHUg vap
ij (7.21)
Applying Clausius-Clapeyron equation we reach to:
160
dT
Pd
RT
H satvap ln
(7.22)
We used the central finite difference to find the right hand of equation (7.22) with temperature
difference of 1 K (ΔT=1). Other properties of bitumen and heavy oil that should be considered for
characterization are the average molecular weight (M) and specific gravity (SG). We used the
following mixing rules to find the average molecular weight and specific gravity of bitumen (Riazi,
2005).
aveii MMx (7.23)
ave
ave
i
ii
SG
M
SG
Mx
(7.24)
Correlation of Rao and Bardon was applied to relate the boiling point of each pseudocomponent
to its molecular weight as given by (Rao and Bardon, 1985):
w
biwi
K
TKM
68.131.22ln)071.027.1(ln (7.25)
where Tb is the boiling temperature and Kw is the Watson characterization factor. Then, we used
SØreide correlation (Soreide, 1989) as the relation between molecular weight and specific gravity
of each pseudocomponent. This correlation has been developed using 843 samples obtained from
68 reservoir fluids. This correlation is given by:
13.0)66(2855.0 ifi MCSG (7.26)
where Cf obtained to satisfy the equation (7.26) for representation of experimental specific gravity
of reservoir fluid using its molecular weight. This parameter is typically between 0.27 and 0.31.
To perform characterization, first, 28 pseudocomponents were defined with boiling point range
from 300 to 1110 K with interval of 30 K. The lower bound (300 K) is between boiling point of
C4H10 and C5H12 which are the lightest possible hydrocarbon in bitumen. The lightest hydrocarbon
component in bitumen is usually C7H16 to C10H22. The upper limit that simulated distillation can
capture is C120H242 with boiling point of 1023 K. Therefore, the range of 300 to 1110 K can safely
161
capture the hydrocarbon components presented in bitumen. Since the simulated distillation
apparatus is calibrated with normal paraffin (n-alkanes) components, we used Twu correlations to
calculate the molecular weight, critical temperature and pressure for each pseudocomponent using
its boiling point (Twu, 1984). Also, acentric factor was obtained using Lee-Kesler equation which
has been presented specifically for petroleum fractions (Whitson and Brule, 2000). Next, mole
fraction of each pseudocomponent was optimized by solving equations (7.3-7.12) to find the best
match of the SD assay results.
After finding the mole fraction distribution of the pseudocomponents present in bitumen, the
average Watson characterization factor is obtained to find the average molecular weight of the
bitumen using equations (7.23) and (7.25). Then, SØreide correlation (equation (7.26)) is tuned
with Cf =0.308 to match the experimental specific gravity of bitumen or heavy oil using equations
(7.24) and (7.26). Characterization procedure is described in a diagram shown in Figure 7.1.
162
Figure 7.1: Flow chart used to characterize the bitumen.
7.5 Solubility Modeling
The output of characterization scheme is a multicomponent system, which is not suitable for
computationally demanding thermal reservoir simulators. Reservoir fluids are usually lumped into
manageable number of pseudocomponents to decrease the run time required for compositional and
thermal reservoir simulations (Riazi, 2005; Whitson and Brule, 2000). Molecular weight, Mi,
which separates the groups is obtained as given by (Whitson and Brule, 2000):
Assume the initial mole fraction guesses
Give the pseudo components from Table 7.1
Find Tsat and yi
)exp(0 nn HH
0
0 )(100%
HMW
HMWHMWOFF nn
%OFF < 90Yes
NO
Adjust the initial mole fraction of
liquid phase
Compare the results with
experimental SD data
Non-acceptable deviation
Acceptable deviation
43211 226
1kkkkxx nn
1nn
H0=1
Assume average Watson
characterizationfactor
Calculate M of each pseudocomponent using eq.
(7.25) and then find the average molecular weight
Compare
calculated Mave
with experimental
Mave
Non-acceptable AcceptableAdjust the average Watson characterization
factor
Compare calculated SGave
with experimental SGave
Acceptable
Non-acceptable
Adjust the Cf
Print the mole fraction, M and SG of each
pseudocomponent
Assume the Cf and calculate SGi and then find
SGave using eq. (7.26)
163
HNi
ni MMMM/
11 )/( (7.27)
where NH is the number of groups (pseudocomponents), i=1, …, NH, M1 is the molecular weight
of the lightest component and Mn is the molecular weight of the heaviest component. Components
with molecular weight falling between Mi-1 and Mi are lumped in group i. Next, mixing rules
proposed by Lee and Kesler were applied to find the critical properties and acentric factor of each
group as follows (Lee and Kesler, 1975):
2
33/13/1 )(8
1
i
i
i j
cjciji
cg
x
VVxx
V
(7.28)
2
33/13/15.0 )()(8
1
i
i
i j
cjcicjciji
cg
cg
x
VVTTxxV
T
(7.29)
i
i
i
ii
gx
x
(7.30)
gcgZ 085.02905.0
(7.31)
cg
cgcg
cgV
RTZP
(7.32)
where Tc, Pc, Vc and Zc are critical temperature, pressure, volume and compressibility factor of the
group, respectively. R is the universal gas constant and ω is the acentric factor.
After grouping of the components using the scheme described above, we use the φ-γ approach to
calculate the solubility of gases in bitumen for the sake the consistency with the characterization
164
procedure described earlier. Therefore, the thermodynamic equilibrium of a component in gas and
liquid phases can be written as:
L
iiiii fxPy (7.33)
where φ denotes the fugacity coefficient of specie i in the gas phase. f L is the fugacity in
hypothetical liquid state given by (Prausnitz et al., 1986):
RT
PVPff
L
ici
L
i
)013.1(exp0 (7.34)
and f 0 is the reduced hypothetical liquid fugacity at atmospheric pressure. ViL is the partial molar
volume of component i in the liquid phase.
Prausnitz et al. (Prausnitz et al., 1986) reported f 0 in a graphical format, which was later presented
by Riazi and Vera (Riazi and Vera, 2005) using the correlation given by:
)ln(08.3
19643.8902.7exp0
r
r
i TT
f (7.35)
where Tr is the reduced pressure (Tr=T/Tc). Fugacity coefficient of specie i was calculated using
Peng-Robinson equation of state (PR-EoS) as given by(Peng and Robinson, 1976; Prausnitz et al.,
1986):
)()(
)(
bVbbVV
Ta
bV
RTP
(7.36)
where
C
C
P
TRa
2245724.0
(7.37)
C
C
P
TRb
07780.0
(7.38)
25.0 )1(1)( rTkT (7.39)
165
0.49 0.016671644.0485.13796.0
0.49 26992.054226.137464.0
32
2
k
k
(7.40)
where ω is the acentric factor. Activity coefficient of solvent molecules in liquid phase has been
obtained using NRTL model as presented in equations (7.17-7.19). Interaction energy parameters
of the pseudocomponents of bitumen were calculated using equations (7.20-7.22). The only
parameter that should be tuned is the interaction energy parameter between solvent and
pseudocomponents. Interaction energy parameter between solvent and pseudocomponents
molecules is a function of temperature and changes for different pseudocomponents (Kontogeorgis
and Folas, 2010). Therefore, a generalized equation was assumed to calculate the interaction
energy parameter between solvent molecules and pseudocomponent i as given by:
riisi TDMCTBA (7.41)
Interaction energy parameters between solvent molecule and pseudocomponent i were assumed
symmetric (τis=τsi). To tune the interaction energy parameters, experimental solubility data at
maximum pressure of each isotherm was applied as proposed by Huang and Radosz (Huang and
Radosz, 1991). Objective function was defined and minimized as given by:
N
x
xx
OF
N
i i
i
calc
i
exp
exp
(7.42)
where N is the number of experimental data points. Optimization toolbox of MATLAB R2012a
was used for regression and optimization in this work.
7.6 Results and Discussion
The characterization procedure described in the previous section (Figure 7.4) was applied to the
Cold Lake bitumen sample. The experimental SD data curve was extracted from the work of Yu
et al. (Yu et al., 1989). The original data was reported in a graphical format and as a result, a
deviation approximately 2 K is observed in the extracted data. The mole fraction, molecular
weight, specific gravity, and boiling point of the pseudocomponents are given in Table 7.1. A
comparison of the experimental SD curve and the results calculated by the presented model is
166
shown in Figure 7.2. A good agreement can be found from this figure and the results confirm that
the presented model can accurately regenerate the experimental SD curve.
Figure 7.2: Comparison of the Yu et al. (Yu et al., 1989) experimental values and the calculated
simulated distillated curve data.
Both the calculated and measured values for the specific gravity and molecular weight of Cold
Lake bitumen are given in Table 7.1. The Watson characterization factor (Kw) was 15.8 and the Cf
of the Søreide equation was 0.308. The very small deviation between the calculated and
experimental values shows that the obtained pseudocomponents can accurately represent the Cold
Lake bitumen.
Table 7.1: Mole fraction and properties of characterized bitumen.
Pseudocomponents Normal boiling
point (K)
Mole
fraction
(%)
Molecular
weight
(g/mol)
Specific gravity
PC1 300 0.24 76.8 0.7051
PC2 330 0.80 96.4 0.7657
PC3 360 1.43 118.8 0.8013
PC4 390 4.83 143.8 0.8280
PC5 420 3.38 171.7 0.8500
% Mass Distilled
0 20 40 60
Tem
pera
ture
(K
)
400
500
600
700
800
900
1000
Experimental DataProposed model
167
PC6 450 2.00 202.5 0.8691
PC7 480 13.19 236.3 0.8861
PC8 510 3.54 273.2 0.9016
PC9 540 6.86 313.2 0.9159
PC10 570 10.65 356.4 0.9293
PC11 600 4.97 403.0 0.9419
PC12 630 8.04 452.8 0.9537
PC13 660 6.09 506.1 0.9650
PC14 690 3.30 562.9 0.9758
PC15 720 8.83 623.2 0.9862
PC16 750 2.66 687.2 0.9962
PC17 780 0.14 754.7 1.0058
PC18 810 1.52 826.0 1.0151
PC19 840 3.13 901.1 1.0240
PC20 870 1.19 980.0 1.0328
PC21 900 2.55 1062.8 1.0412
PC22 930 2.99 1149.5 1.0495
PC23 960 0.25 1240.2 1.0575
PC24 990 1.56 1334.9 1.0653
PC25 1020 2.08 1433.7 1.0730
PC26 1050 2.43 1536.6 1.0804
PC27 1080 1.31 1643.7 1.0877
PC28 1110 0.04 1755.1 1.0949
Property of the Cold Lake bitumen
Exp.(Fu et al., 1986) Calculated AARD (%)
Molecular weight 533 527.23 1.08
Specific gravity 0.986 0.987 0.10
Table 7.2: Mole fraction and properties of lumped pseudocomponents
Pseudocomponents Mole
fraction
Molecular
weight
Critical
Temperature
(K)
Critical
Pressure
(MPa)
Acentric
factor
PC1 0.1268 153.4 570.85 2.517 0.3965
PC2 0.3922 306.8 700.12 1.572 0.6750
PC3 0.2905 551.2 820.60 0.909 1.0563
PC4 0.1163 1005.9 958.19 0.453 1.5650
PC5 0.0742 1485.4 1061.0 0.274 1.9135
After finding the pseudocomponents, grouping calculations performed to lump them into five new
pseudocomponents. The mole fraction, average molecular weight, critical properties and acentric
factor of each group were obtained and are given in Table 7.2.
168
The characterized bitumen was introduced into the presented solubility model to calculate the
solubility of the light hydrocarbon solvents (CH4 and C2H6) and non-hydrocarbons (CO2 and N2).
Equation (7.41) presents the binary interaction energy parameters as a linear function of
temperature, molecular weight, and reduced temperature. The linear function successfully captured
the interaction between the pseudocomponents and the solvent in the mixture. Table 7.3 lists the
parameters required to calculate binary interaction parameters for light hydrocarbon solvents (CH4
and C2H6) and non-hydrocarbons (CO2 and N2).
The experimental data for solvent solubility in bitumen is very scarce in the open literature. Here,
we apply the reference data sets of (Fu et al., 1986; Mehrotra and Svrcek, 1988b) to validate the
developed characterization procedure as well as the solubility model. Figure 7.3 compares the
experimental CH4 solubility data with the results of the proposed model. The experimental data
was recorded at temperatures to 423.15 K and pressures up to 11 MPa. The calculated CH4
solubility results are very close to the experimental data, which was represented by the proposed
model with an Absolute Average Relative Deviation (AARD) of less than 4.7%. The experimental
data reported by Mehrotra and Svrcek (Mehrotra and Svrcek, 1988b) for temperatures up to 376.15
K was represented with an AARD of 3.4%. Data of Fu et al. (Fu et al., 1986) produced a data set
covering temperatures up to 423.15 K. The Fu et al. (Fu et al., 1986) data at 373.15 K, 10 MPa and
423.15 K, 6 MPa show scatter and do not follow the trend found in the other experimental data
sets. Therefore, these data points were not considered in the calculation of the AARD between the
experimental and calculated CH4 solubility data. Lower methane solubility was found in the Fu et
al. (Fu et al., 1986) data at 343.15 K than the Mehrotra and Svrcek (Mehrotra and Svrcek, 1988b)
data at 350 K. Since higher solubility is expected at lower temperatures, it seems plausible to
observe more deviation between the results of our proposed model and the experimental data of
Fu et al. (Fu et al., 1986). We found an AARD of 6.2% between the Fu et al. (Fu et al., 1986) data
and the results of our model.
169
Table 7.3: Parameters for implementation of equation (7.35) to determine the binary interaction
energy parameter of the solvents and pseudocomponents.
Solvent A B C D
CH4 1.593 8.8262×10-4 -2.9303×10-3 1.0354×10-3
C2H6 9.1211×10-2 -8.6345×10-4 -9.9354×10-5 2.0314×10-3
CO2(supercritical) 0.8329 -2.7713×10-3 1.9092×10-6 0.2809
CO2(subcritical) 5.9255 -2.2605×10-2 -4.0392×10-3 10.801
N2 4.7267 1.1687×10-3 -3.6593×10-3 -7.129×10-2
Figure 7.3: Comparisons of the results of the proposed model with the experimental CH4 solubility
data in bitumen. Experimental data was obtained from Fu et al. (Fu et al., 1986) and Mehrotra and
Svrcek (Mehrotra and Svrcek, 1988b).
Pressure (MPa)2 4 6 8 10
Mo
le f
ractio
n o
f C
H4
0.1
0.2
0.3
0.4299.95 K (Mehrotra and Svrcek, 1988b)319.15 K (Mehrotra and Svrcek, 1988b)376.55 K (Mehrotra and Svrcek, 1988b)373.15 K (Fu et al., 1986)Proposed Model
Pressure (MPa)
2 4 6 8 10
Mo
le f
ractio
n o
f C
H4
0.1
0.2
0.3
0.4
343.15 K (Fu et al., 1986)
350.55 K (Mehrotra and Svrcek, 1988b)423.15 K (Fu et al., 1986)Proposed Model
(a)
(b)
170
Figure 7.4 (a and b) compare the C2H6 solubility experimental data and the results of the proposed
model. An AARD equal to 6.3% was found, which indicates that the proposed model is accurate.
At high temperature and pressure conditions, our proposed model slightly underestimates the
experimental C2H6 solubility data. A similar deficiency for methane solubility at 343.15 K was
observed when ethane solubility data from Fu et al. (Fu et al., 1986) at 343.15 K and Mehrotra and
Svrcek (Mehrotra and Svrcek, 1988b) at 350 K were compared. However, compared with other
models available in the literature, our model predicts the experimental data with acceptable
accuracy (see Table 7.4).
Figure 7.4: Comparisons between the results of the proposed model and experimental C2H6
solubility data in bitumen. Experimental data was obtained from Fu et al. (Fu et al., 1986) and
Mehrotra and Svrcek (Mehrotra and Svrcek, 1988b).
Pressure (MPa)
2 4 6 8 10
Mo
le fra
ctio
n o
f C
2H
6
0.2
0.4
0.6
0.8
297.35 K (Mehrotra and Svrcek, 1988b)325.25 K (Mehrotra and Svrcek, 1988b)349.35 K (Mehrotra and Svrcek, 1988b)343.15 K (Fu et al., 1986)Proposed Model
Pressure (MPa)
2 4 6 8 10
Mo
le fra
ctio
n o
f C
2H
6
0.2
0.4
0.6
0.8
373.15 K (Fu et al., 1986)374.25 K (Mehrotra and Svrcek, 1988b) 423.15 K (Fu et al., 1986)Proposed Model
(a)
(b)
171
For non-hydrocarbon solvents, the experimental solubility data for CO2 and N2 was obtained from
the literature (Mehrotra and Svrcek, 1988b; Yu et al., 1989). Figures 7.5 (a and b) illustrate an
evaluation of ability of the proposed model to reproduce the experimental CO2 solubility data for
Cold Lake bitumen. An AARD of less than 2.0% was found when the Mehrotra and Svrcek
(Mehrotra and Svrcek, 1988b) experimental data was regenerated at subcritical and supercritical
conditions. In the subcritical and supercritical regions, the energy interaction parameters were
different. Table 7.3 presents these parameters for CO2 molecules in different regions. The model
is more accurate in the subcritical region where the experimental data for the solubility of CO2 in
Cold Lake bitumen had an AARD less than 1.5%. This change in the accuracy of the model may
be related to the change of the energy interaction parameters between pseudocomponents with sub-
and supercritical CO2 molecules.
Yu et al. reported an extensive data set including temperatures up to 523.1 K and pressures up to
16 MPa (Yu et al., 1989). As mentioned in Table 7.4, the model accurately reproduced this data
set with an AARD of less than 5.8%. The model predictions of the solubility of N2 in bitumen
were compared with the Mehrotra and Svrcek (Mehrotra and Svrcek, 1988b) experimental N2
solubility data, resulting in an AARD of 7.6%. Figure 7.6 shows the comparison between results
of the proposed model and experimental data of N2 solubility in bitumen. This experimental data
does not show consistent trend with temperature while increasing N2 solubility by increasing
pressure can be found. Lack of non-scatter experimental data for N2 solubility in bitumen is
observed in open literature.
We also compared the predictions of the proposed model with those of the models reported in the
literature (Table 7.4). Our proposed model proved to be more accurate in some cases. Table 7.4
highlights the applicability and accuracy of the proposed model for calculating the solubility of
light hydrocarbon and non-hydrocarbon solvents in bitumen over a wide range of temperatures
and pressures. This method can be used for bitumen characterization and solubility calculations
involved in bitumen recovery and processing.
172
Table 7.4: AARDs of the experimental solubility data of light solvents in bitumen and the results
of the proposed model and other predictive models.
Exp. data Predictive models AARD
(%)
Temperature range
(K)
Pressure range
(MPa)
CH4
(Mehrotra and Svrcek,
1988b)
Proposed model 3.4
299.85–376.95 2.57–10.07 (Mehrotra and Svrcek, 1988a) 2.7
(Kariznovi et al., 2010) 3.74
(Fu et al., 1986)
Proposed model 6.2
343.15–423.15 2.17–11.83 (Mehrotra and Svrcek, 1988a) 4.5
(Fu et al., 1986) 4.82
C2H6
(Mehrotra and Svrcek,
1988b)
Proposed model 5.7
296.05–375.85 1.02–10.07 (Mehrotra and Svrcek, 1988a) 5.7
(Kariznovi et al., 2010) 7.64
(Fu et al., 1986)
Proposed model 7.2
343.15–423.15 2.62–10.96 (Mehrotra and Svrcek, 1988a) 6.5
(Fu et al., 1986) 4.72
CO2
(Mehrotra and Svrcek,
1988b)
Proposed model 2.0
287.85–371.35 2.14–10.95 (Kariznovi et al., 2010) 8.89
(Eastick et al., 1992) 6.8
(Mehrotra and Svrcek, 1988a) 1.9
(Yu et al., 1989)
Proposed model 5.8
323.15–523.1 4–16.04 (Huang and Radosz, 1991) 5.67
(Huang and Radosz, 1990) 5.44
N2
(Mehrotra and Svrcek,
1988b)
Proposed model 7.6
303.85–371.35 2.46–10.66 (Kariznovi et al., 2010) 5.62
(Mehrotra and Svrcek, 1988a) 8.1
AARD (Absolute Average Relative Deviation) =
erimental
erimentalcalculated
x
xx
N exp
exp1
173
Figure 7.5: Comparisons of the results of the proposed model and the experimental CO2 solubility
data in bitumen. Experimental data was obtained from Yu et al. (Yu et al., 1989) and Mehrotra and
Svrcek (Mehrotra and Svrcek, 1988b).
Pressure (MPa)
2 4 6 8 10 12
Mole
fra
ctio
n o
f C
O2
0.1
0.2
0.3
0.4
0.5
0.6
0.7
288.15 K (Mehrotra and Svrcek, 1988b)299.15 K (Mehrotra and Svrcek, 1988b)350.05 K (Mehrotra and Svrcek, 1988b)Proposed Model
Pressure (MPa)
2 4 6 8 10 12
Mole
fra
ctio
n o
f C
O2
0.2
0.4
0.6
0.8
325.65 K (Mehrotra and Svrcek, 1988b)323.3 K (Yu et al., 1989)371.05 K (Mehrotra and Svrcek, 1988b)373.3 K (Yu et al., 1989)473.7 K (Yu et al., 1989)523.1 K (Yu et al., 1989)Proposed Model
(a)
(b)
174
Figure 7.6: Comparisons of the results of the proposed model and the experimental N2 solubility
data in bitumen. Experimental data was obtained from Mehrotra and Svrcek (Mehrotra and Svrcek,
1988b).
7.7 Summary and Conclusions
The ability to accurately characterize bitumen and predict the solubility of solvents in bitumen are
essential to engineering calculations for bitumen production, upgrading, fractionation, and
refining. Due to the high molecular weight and boiling point of the hydrocarbon molecules present
in bitumen, conventional characterization methods are not applicable or are not straightforward.
In this work, we proposed a fast, accurate, and inexpensive methodology based on solid theoretical
background to characterize bitumen and tune an equation of state to predict the solubility of light
solvents in bitumen. The residue curve map method was applied to characterize the bitumen using
simulated distillation (SD) data. The Peng-Robinson equation of state (PR-EoS) and the NRTL
activity model were applied to capture the departure from the ideal state in the gas and liquid
phases. The proposed method successfully characterized the bitumen and was applied to estimate
the interaction parameters of the pseudocomponents.
The results of the characterization compared favorably with the experimental molecular weight,
specific gravity, and simulated distillation (SD) curve of the bitumen. We verified the applicability
of the proposed model for calculation of the solubility of light solvents in bitumen by comparison
Pressure (MPa)
2 4 6 8 10
Mole
fra
ction o
f N
2
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
304.15 K (Mehrotra and Svrcek, 1988b)329.95 K (Mehrotra and Svrcek, 1988b)370.95 K (Mehrotra and Svrcek, 1988b)Proposed Model (333.15 K)
175
with the experimental data. The calculated and the experimental solubility data of CH4, C2H6, CO2
and N2 were in good agreement. The accuracy of the proposed model was also compared with that
of some of the predictive models reported in the literature and acceptable accuracy was found.
This model can be applied to characterize bitumen and extra heavy oils. Application of the
proposed model for bitumen characterization eliminates the need for expensive physical
distillation data and will find many applications in bitumen production, upgrading, and refining.
176
7.8 References
Doherty, M.F., Perkins, J.D., 1979. On The Dynamics of Distillation Processes—III: The
Topological Structure of Ternary Residue Curve Maps. Chem. Eng. Sci. 34, 1401–1414.
Doherty, M.F., Perkins, J.D., 1978a. On The Dynamics of Distillation Processes—I: The Simple
Distillation of Multicomponent Non-reacting, Homogeneous Liquid Mixtures. Chem. Eng.
Sci. 33, 281–301.
Doherty, M.F., Perkins, J.D., 1978b. On The Dynamics of Distillation Processes—II: The Simple
Distillation of Model Solutions. Chem. Eng. Sci. 33, 569–578.
Eastick, R.R., Svrcek, W.Y., Mehrotra, A.K., 1992. Phase Behaviour of CO2-Bitumen Fractions.
Can. J. Chem. Eng. 70, 159–164.
Fu, C.T., Puttagunta, R., Vilcsak, G., 1986. Vapour-Liquid Equilibrium Properties for Gas-Cold
Lake Bitumen. The 37th Annual Technical Meeting of the Petroleum Society of CIM,
Calgary, Alberta, Canada, 8–11 June.
Huang, S.H., Radosz, M., 1991. Phase Behaviour of Reservoir Fluids V: SAFT Model of CO2 and
Bitumen Systems. Fluid Phase Equilib. 70, 33–54.
Huang, S.H., Radosz, M., 1990. Phase Behaviour of Reservoir Fluids II: Supercritical Carbon
Dioxide and Bitumen Fractions. Fluid Phase Equilib. 60, 81–98.
Kariznovi, M., Nourozieh, H., Abedi, J., 2010. Bitumen Characterization and Pseudocomponents
Determination for Equation of State Modeling. Energy Fuels 24, 624–633.
Kontogeorgis, G.M., Folas, G.K., 2010. Thermodynamic Models for Industrial Applications: From
Classical and Advanced Mixing Rules to Association Theories. First ed., JohnWiley & Sons,
Ltd., NewYork.
Lee, B.I., Kesler, M.G., 1975. A Generalized Thermodynamic Correlation Based on Three-
parameter Corresponding States. AIChE J. 21, 510–527.
Mehrotra, A.K., Svrcek, W.Y., 1988a. Correlation and Prediction of Gas Solubility in Cold Lake
Bitumen. Can. J. Chem. Eng. 66, 666–670.
177
Mehrotra, A.K., Svrcek, W.Y., 1988b. Properties of Cold Lake Bitumen Saturated with Pure Gases
and Gas Mixtures. Can. J. Chem. Eng. 66, 656–665.
Ortiz, D.P., Satyro, M.A., Yarranton, H.W., 2013. Thermodynamics and Fluid Characterization
Using Trajectory Optimization. Fluid Phase Equilib. 351, 34–42.
Peng, D.Y., Robinson, D.B., 1976. A New Two-Constant Equation of State. Ind. Eng. Chem.
Fundam.15, 59–64.
Prausnitz, J.M., Lichtenthaller, R.N., Azedevo, E.G.D., 1986. Molecular Thermodynamics of
Fluid Phase Equilibria. Prentice Hall, New York.
Rao, V.K., Bardon, M.F., 1985. Estimating The Molecular Weight of Petroleum Fractions. Ind.
Eng. Chem. Process Des. Dev. 24, 498–500.
Riazi, M.R., 2005. Characterization and Properties of Petroleum Fractions. First ed., ASTM
International, U.S.A.
Riazi, M.R., Vera, J.H., 2005. Method to Calculate the Solubilities of Light Gases in Petroleum
and Coal Liquid Fractions on the Basis of Their P/N/A Composition. Ind. Eng. Chem. Res.
44:186–192.
Saajanlehto, M., Alopaeus, V., 2014. Heavy Oil Characterization Method for PC-SAFT. Fuel 133,
216–223.
Saajanlehto, M., Uusi-Kyyny, P., Alopaeus, V., 2014. Hydrogen Solubility in Heavy Oil Systems:
Experiments and Modeling. Fuel 137, 393–404.
Satyro, M.A., Yarranton, H., 2009. Oil Characterization from Simulation of Experimental
Distillation Data. Energy Fuels 23, 3960–3970.
Søreide, I., 1989. Improved Phase Behaviour Predictions of Petroleum Reservoir Fluids from a
Cubic Equation of State. PhD dissertation, Norwegian Inst. of Technology, Trondheim,
Norway.
Tassios, D., 1971. A Single-parameter Equation for Isothermal Vapor-Liquid Equilibrium
Correlations. AIChE J. 17, 1367–1371.
178
Twu, C.H., 1984. An Internally Consistent Correlation for Predicting The Critical Properties and
Molecular Weights of Petroleum and Coal-tar Liquids. Fluid Phase Equilib. 16, 137–150.
Vetere, A., 2004. The NRTL Equation as a Predictive Tool for Vapor–Liquid Equilibria. Fluid
Phase Equilib. 218, 33–39.
Whitson, C.H., Brule, M.R., 2000. Phase Behaviour. Monograph, vol. 20. Richardson, Texas: SPE,
Henry L. Doherty series.
Wong, D.S.H., Sandler, S.I., 1992. A Theoretically Correct Mixing Rule for Cubic Equations of
State. AIChE J. 38, 671–680.
Yu, J.M., Huang, S.H., Radosz, M., 1989. Phase Behaviour of Reservoir Fluids: Supercritical
Carbon Dioxide and Cold Lake Bitumen. Fluid Phase Equilib. 53, 429–438.
Zirrahi, M., Hassanzadeh, H., Abedi, J., Moshfeghian, M., 2014. Prediction of Solubility of CH4 ,
C2H6 , CO2 , N2 and CO in Bitumen. Can. J. Chem. Eng. 92, 563–572.
179
Chapter Eight: Conclusions and Recommendations
8.1 Conclusions
The multicomponent characterization of light hydrocarbon solvents (methane, ethane, propane,
and butane)-Athabasca bitumen fractions have been studied in this thesis. Two experimental
apparatuses were designed and fabricated to fractionate a bitumen and to measure the phase
behaviour data of solvent-bitumen fractions including solvent solubility, density, and viscosity of
liquid phase. Moreover, various bitumen characterization approaches were developed and
evaluated.
8.1.1 Bitumen Fractionation
The bitumen fractionation and characterization was carried out in this work. The vacuum
distillation was utilized to fractionate a bitumen sample to four fractions by applying three batch
distillation at 350, 250, and 195 oC under vacuum pressure of 0.01 MPa. Then, asphaltene was
separated from the heaviest cut (Cut 4) using heptane. The molecular weight of each obtained cut
was measured using cryoscope based on freezing point depression concept. The average molecular
weight of heavy fraction and asphaltene were also calculated using the gel permission
chromatography (GPC). The simulated distillation tests (ASTM D7169) was carried out on
bitumen fraction to obtain the boiling point distribution of each cut. The boiling point distribution
and carbon number distribution of each bitumen fraction showed the good separation efficiency
achieved by the modified vacuum distillation. Three bitumen fractions (Cuts 1, 2, and 3) were
distillable and the heavy fraction (Cut 4) was completely solid like. The density and viscosity of
each cut in the wide ranges of temperature and pressure were also measured in which the
decreasing trends were obvious from heavy to light fractions.
8.1.2 Experimental Phase Behaviour Data of Solvents-Bitumen Fractions
The comprehensive phase behaviour data set of methane, ethane, propane, and butane-each
bitumen fraction systems including solvent solubility, density, and viscosity at temperatures up to
190 oC and pressures up to 6 MPa were generated. The experimental solubility data showed the
180
increasing solvent solubility with increasing pressure and decreasing temperature. Heavier
solvents dissolved more in bitumen fractions. Moreover, the solvent solubility in the light fraction
is more than the heavy one. The viscosity and density of the bitumen rich phase decreased as a
results of solvent dissolution.
8.1.3 Generalized Solvent Solubility Model
The generalized model to calculate solvent solubility in bitumen and heavy oil has been proposed
in this work. Using the measured solvent solubilities in the distillable bitumen fractions, the
generalized binary interaction parameter correlation has been developed. For each solvent, i.e.,
methane, ethane, propane, and butane, the generalized binary interaction parameter coefficients
for solvent/bitumen systems have been established. The only input to this model to predict the
solubility is the boiling point or carbon number distribution, which is simply obtained by simulated
distillation test. Using this distribution, the bitumen components were defined and applying the
obtained binary interaction parameters developed in this study, the solvent solubility in bitumen
can be calculated. The comparison between the model calculations and the measured solubility
data in different bitumen samples confirmed the ability of the proposed model in this work to
estimate the solvent solubility in bitumen required to design, model, and simulate the solvent-
based recovery processes. Therefore, instead of measuring the solvent solubility data in bitumen
samples and using them to tune the EoS, the simple, fast, and inexpensive analytical test, i.e.
simulated distillation, is the only required test needs to be carried out on bitumen samples.
8.1.4 Effect of Asphaltene on Solubility, Density and Viscosity of Solvent-Bitumen System
The effect of existence of asphaltene in bitumen on phase behaviour data of hydrocarbon and non-
hydrocarbon solvents (i.e., ethane and carbon dioxide) has been also investigated in this work. The
solubility, density, and viscosity of solvent-bitumen and -deasphalted bitumen systems were
compared in the temperature and pressure ranges of 70-130 oC and 2-8 MPa, respectively. The
significant effect of asphaltene on the density and viscosity of bitumen was quantified. Moreover,
using the measured data, the pseudo-liquid density and viscosity of asphaltene were also estimated
that are the data required for studying processes involving asphaltene. The solubility measurement
results showed that ethane solubility in the presence and absence of asphaltene is almost the same.
However, solubility of CO2 in the whole bitumen is more than the deasphalted bitumen. These
181
observations contradicted the available results in literature which implied the asphaltene has a
negligible effect on the CO2 solubility in bitumen.
8.1.5 Characterization of Heavy and Complex Mixtures Using GPC Coupled to Simulated
Distillation
A new characterization method to provide the molecular weight and boiling point distributions of
extra heavy hydrocarbon samples such as bitumen and vacuum residue was proposed by
combining the gel permission chromatography (GPC) and simulated distillation results. In this
model, we used the simulated distillation results of each sample to generate the calibration curve
for GPC test. Having the calibration correlation, the whole molecular weight distribution of sample
could be obtained. For the very heavy and complex mixtures, such as asphaltene, most of the
components cannot be characterized by simulated distillation. The GPC, which is the common tool
to characterize the polymers, was employed to find the complete molecular weight distribution
and more detailed information of heavy hydrocarbon components.
8.1.6 Bitumen Characterization Method Based on Residue Curve Map
The new bitumen characterization method using simulated distillation results and based on residue
curve map was also proposed in this work. This proposed model could properly regenerate the
simulated distillation results and estimate the molecular weight and specific gravity of bitumen.
The PR-EoS and the NRTL activity model were used to model the gas and liquid phases. This
proposed model was used to predict the light hydrocarbon (methane and ethane) and non-
hydrocarbon (carbon dioxide and nitrogen) solvent solubilities in bitumen. Comparison between
measured and the calculated solubilities confirmed the applicability of this model.
8.2 Future Works
The solubility model has been developed in this work by considering the vapour-liquid equilibria
of solvent and bitumen fraction mixtures. The results of this study showed that binary interaction
parameters obtained using the solubility data of solvent and distillable bitumen fractions (Cuts 1,
2, and 3) provide reasonable predictions of solvent solubility in whole bitumen. On the other hand,
the un-distillable fraction of bitumen (Cut 4) has also been characterized in this study.
182
Implementing the characterized Cut 4 and asphaltene components in the model is recommended
to understand the effects of these heavy components on predictions of solvent solubility in
bitumen. Moreover, studying the liquid-liquid equilibrium of solvent-bitumen cut systems will add
more insight to the characterization of bitumen/solvent systems. Advanced EoS such as cubic plus
association (CPA) can be also used to model the experimental data and compared the results with
PR-EoS. The mixtures of non-hydrocarbon solvents, such as carbon dioxide and dimethyl ether,
and solvent mixtures such as condensate can be also considered for future studies of solvent-
bitumen fraction systems.
183
Appendix A: Experimental Apparatus Design and Calibration
In this appendix, the experimental apparatuses utilized in this work including the fractionation
apparatuses and the PVT setup are presented. Moreover, calibration of densitometer, molecular
weight measuring apparatus, and validation of PVT setup are described.
A.1 Fractionation Apparatuses
After running trial vacuum distillations and analyzing the results, the modified vacuum distillation
was designed to fractionate bitumen into different cuts with the highest possible separation
efficiency. A schematic of the vacuum distillation scheme is shown in Figure A.1. First, vacuum
distillation (350 C and 0.01 MPa) was used to fractionate whole bitumen into two fractions. The
residue of the first vacuum distillation was called Cut 4 and it was composed of the heaviest
components of the bitumen. The distillate of 350 C and 0.01 MPa vacuum distillation was a
mixture of the lighter components. Once the heavy components were separated from the bitumen,
what remains resembles the properties of conventional oil. In the next step, the distillate from the
first vacuum distillation (350 C and 0.01 MPa) was fractionated using the vacuum distillation at
250 C and 0.01 MPa into two other fractions including Cut 3 and a distillate. Finally, the distillate
obtained at 250 oC was distilled at 195 C and 0.01 MPa into Cut 1 and Cut 2. Three flash
vapourization systems were used to fractionate the bitumen into four distinct cuts. Cut 4 was
further divided into Cut 4-deasphalted and asphaltene for further analysis.
184
Figure A.1: The scheme of bitumen fractionation considered in this work.
Figure A.2 shows the designed and fabricated vacuum distillation setup used for bitumen
fractionation. Bitumen from the feeding cell was pumped and warmed up by passing it through the
lines in the oven set to the desired temperature (as mentioned in Figure A.1). In the flash cell, the
warmed bitumen was flashed and the light components were separated from the heavy ones. Light
components from the top of the flash cell were collected after passing through the condenser.
Heavy components from bottom of the cell were also collected. This fractionation set-up requires
the hydrocarbon fluid to be maintained at high temperatures for only 30 minutes. This reduces the
chance of thermal cracking of bitumen at high temperatures.
185
Figure A.2: Schematic of vacuum distillation used for bitumen fractionation in this work: 1,
feeding cell; 2, water tank; 3, Quizix pump; 4, pressure indicator; 5, light fraction collector; 6,
vacuum pump; 7, condenser; 8, heavy fraction collector; 9, heat tape and insulation; 10, flash cell;
11, oven.
The fabricated vacuum distillation set-up and obtained four Athabasca bitumen fractions are
shown in Figures A.3 and A.4, respectively. As the molecular weight of the fraction increases, it
becomes darker and more viscous.
2
3
5
4
8
11
6
7
9
1
10
186
Figure A.3: The fabricated vacuum distillation setup.
Figure A.4: The fractions of Athabasca bitumen using three batch distillations.
The temperatures at which the distillation was conducted and the weight fractions of each
obtained cut are shown in the Table A.1.
Table A.1: The properties of Athabasca bitumen fractions.
Sample Distillation T (oC) wt.%
Cut 1 195> T 19.4
Cut 2 195< T <250 11.7
Cut 3 250<T <350 18.4
Cut 4 T >350 50.5
Feeding cell
Collector for light fraction
Vacuum pump
High temperature oven
Collector for heavy fraction
Flash cell inside the high
temperature oven
187
A.2 Molecular Weights Measurements
The molecular weight of oil samples is an essential input for characterization and thermodynamic
modeling. In this work, the cryoscopy method using the freezing point depression concept was
utilized to measure the molecular weight of bitumen and its fractions. By dissolving the solute in
solvent, the freezing point is reduced and this depression can be correlated to the concentration
and molecular weight of the solute.
Using this method, the molecular weight is calculated by measuring the freezing point depression
because this depression is linearly related to the solute concentration (Cryscope Instruction
Manual, 2004).
FPW
KWMW
Solvent
fSolute
1000 (A.1)
In this equation, ΔFP is the reading obtained from the instrument, W is the weight, and Kf is the
freezing point depression constant. The Kf values for benzene and water are 5.12 and 1.86 oC/mole,
respectively.
Before starting the molecular weight measurements, the apparatus should be calibrated using a
solution of a known solute and solvent. Generally, two solutions, aqueous and non-aqueous, can
be used for molecular weight measurements depending on the type of the solute. For the aqueous
solution, water is used as the solvent and for the non-aqueous solution, benzene is used. For
aqueous solutions, the standards were provided by factory which are different concentrations of
salt in water. However when benzene is used as the solvent, it is necessary to make standards daily
to calibrate and validate the system because the concentration of the benzene solution may change.
First, the aqueous solution was used to calibrate and validate the apparatus. The factory provided
standards were used to calibrate the system. Two different molal solutions of 1-propanol in
distilled water were made to examine the accuracy of the system and the effect of concentration
on the molecular weight of the samples. Table A.2 summarizes the measured molecular weight of
1-propanol and the error of the measurements. All of the errors are less than 1% and the results are
independent of the concentration of samples.
188
Table A.2: Molecular weight of 1-propanol using freezing point depression method (the molecular
weight of 1-propanol is 60.09 g/mol).
Sample Test No. Cryette Reading (ΔT) MW (g/mol) Error* (%)
1
1 745 59.72 0.63
2 742 59.96 0.23
3 738 60.28 0.31
2
1 1014 59.95 0.25
2 1017 59.77 0.54
3 1015 59.89 0.35
* 100/)((%) actualmeasuredactual
MWMWMWError
To measure the molecular weight of the bitumen samples, benzene (99.5 mol.%) was used as the
solvent. First, n-tetradecane was used to calibrate the system following the calibration procedure
in the Cryette instruction manual. Then, in order to test the accuracy of the measured molecular
weight data by Cryette, another concentration of n-tetradecane in benzene and two different
concentrations of hexadecane in benzene were prepared and their molecular weights were
measured by the calibrated Cryette. Table A.3 presents the measured molecular weights of n-
tetradecane and hexadecane and the deviations from the actual values.
Table A.3: Molecular weight of (a) tetradecane and (b) hexadecane using freezing point
depression method (the molecular weight of n-tetradecane and hexadecane are 198.39 and 226.44
g/mol, respectively).
(a)
Sample Test No. Cryette Reading (ΔT) MW (g/mol) Error (%)
1
1 583 196.89 0.76
2 581 197.56 0.42
3 583 196.89 0.76
(b)
Sample Test No. Cryette Reading (ΔT) MW (g/mol) Error (%)
1
1 527 229.31 1.25
2 530 228.01 0.68
3 530 228.01 0.68
2
1 644 230.54 1.77
2 647 229.47 1.32
3 647 229.47 1.32
189
As shown in Tables A.3 (a and b), the Cryette measures molecular weight within about 1% error.
After calibrating and validating the experimental apparatus, the molecular weight of the oil
samples can be measured. A benzene solution was prepared for each sample (bitumen, Cut 1, Cut
2, Cut 3, and deasphalted-Cut 4) within the concentration range recommended by the factory (0.1
to 0.2 molal). Five runs were conducted for each sample. The final molecular weight is the average
of all of the runs. Table A.4 shows the average molecular weight measurements for whole bitumen
and the four fractions. The increasing trend from Cut 1 to Cut 4 is obvious in the results.
Table A.4: Average molecular weights of Athabasca bitumen and its fractions measured using
freezing point depression method.
Sample MW (g/mol)
Bitumen 568.8±4.7
Cut 1 268.8±0.9
Cut 2 365.5±1.8
Cut 3 464.6±2.3
Deasphalted-Cut 4 906.1±11.7
A.3 Asphaltene Separation
To characterize the heavy fraction of bitumen (Cut 4), asphaltene is removed and studied
separately. For asphaltene separation, n-heptane was used as the paraffinic solvent. The asphaltene
separation process based on ASTM D2007 is described below.
First, a specific amount of Cut 4 was weighed and 40 times the weight as volume of the solvent
(e.g., 40 g of sample requires 1600 mL of solvent) was added to the sample. Then, the solution
was sonicated for 50 minutes and allowed to settle for 24 hours. After one day, about 75 vol.% of
the solution was decanted into the funnel and filtered using filter paper with #2 mesh size. Then,
about 10 vol.% of the fresh solvent was added to the beaker and sonicated for 45 minutes. After
sonication, solution was settled for 5 hours. After that, all of the settled material inside the beaker
was filtered using the same filter paper. The final step is washing the asphaltene remaining on the
filter paper using fresh n-heptane until the filter paper is clean and white. The weight fraction of
asphaltene, which is the weight of the asphaltene divided by the weight of the initial sample (Cut
4) was about 26%. The same procedure was applied to whole bitumen to estimate the asphaltene
190
content of the whole bitumen. The asphaltene weight fraction for bitumen was about 13%. This
percent of asphaltene was expected based on the weight percent of Cut 4 to the whole bitumen.
The asphaltene content of a mixture of the three first cuts was also measured and found to be less
than 1 wt.% of the cuts.
A.4 PVT Apparatus
The composition, density and viscosity of each phase in a solvent/bitumen fraction system
are the basic data required for the phase behaviour study and simulation of solvent-aided thermal
recovery methods. To acquire the experimental data of various solvent/bitumen fraction systems
over a wide range of temperatures and pressures, a new PVT apparatus was designed. Since the
bitumen was divided into different fractions, we have only a small amount of each fraction. This
is a limitation for traditional PVT tests. The newly designed system requires only 100 ml of sample
for 5-6 runs. Moreover, the taken sample for the solubility measurement can be reused in the
system for other runs.
Figure A.5 shows the schematic of the designed PVT apparatus. This PVT apparatus was
fabricated by the machine shop at the University of Calgary. In this setup, the equilibrium cell,
which is rocked to mix the solvent and bitumen, densitometer, and viscometer are placed in the
oven. The transfer and sampling cells are located outside the oven. Two pumps are used for solvent
injection and transfer the liquid phase from the equilibrium cell to the transfer cell.
191
57 cp
0.874 gr/cc
98 OC
1
2 3
6
9
7
4
5
11
12
14
10
88 751 PSI
8
13
Figure A.5: Schematic PVT setup used in this work. 1, equilibrium cell; 2, densitometer; 3,
viscometer; 4, density measuring unit; 5, viscosity measuring unit; 6, sampling cell; 7, Quizix
pump; 8, water tank; 9, transfer cell; 10, pressure transducer; 11, ISCO pump; 12, gas cylinder;
13, vent valve; 14, oven.
A factory calibrated Cambridge viscometer is used to measure the viscosity. This viscometer is
calibrated for temperatures up to 200ºC and pressures up to 10.2 MPa. Evaluation of this
viscometer to measure the viscosity of hydrocarbons reveals a deviation of less than 5%.
The density measurements are conducted using an Anton Paar densitometer with accuracy of
0.001-0.0001 g/cm3 applicable for the density range of 0-3 g/cm3. The working temperatures and
pressures for the densitometer are up to 200ºC and 70 MPa, respectively. The calibration procedure
is presented in the next section.
The experiment procedure is as follows. First, the system is vacuumed to remove all contaminants.
Then, the equilibrium cell is fed a sample of bitumen or any of the bitumen cuts. Next, the solvent
is injected into the equilibrium cell. At the desired temperature, the equilibrium cell begins to rock
until the system reaches equilibrium. The equilibrium condition is achieved when the solvent can
no longer dissolve in the sample. After reaching equilibrium, the bitumen-rich phase is discharged
192
from the bottom of the equilibrium cell and passed through the viscometer and densitometer to
record the physical properties. The discharged phase is then received by the transfer cell to ensure
that there is bitumen rich phase in the lines. Using the sampling cell, the sample is taken from the
system and flashed at atmospheric pressure and the evolved gas is measured using the Chandler
Engineering Gasometer (Model 2331) with an accuracy of 0.2% over the range of the readings.
Two sets of experiments for propane/Cut 3 and methane/Cut 2 at three temperatures of 50, 100,
and 150 oC were repeated to examine the repeatability of the measured experimental data. The
measured data were in good agreement. The deviation of less than 0.2 kg/m3 for saturated density
and less than 3% for solubility of gas in liquid phase had been observed.
A.5 Calibration of Densitometer
The Anton Paar densitometer has been used to measure the density of liquid phase. This
densitometer consists of a U-shaped tube in which the fluid is passed through and the vibration
frequency is measured. Using a polynomial equation, the density is calculated as a function of
temperature, pressure, and frequency of the density measuring cell. The following procedure was
used to calibrate the density measuring cell.
1- Nitrogen and water were selected as the reference gas and liquid fluids, respectively.
2- The density measuring frequencies were recorded over a wide range of temperatures and
pressures. For each data point at the specified temperature and pressure, the corresponding density
was obtained from the National institute of Standards and Technology (NIST) data bank.
3- The NIST density data and the density measuring cell frequencies were used to determine the
coefficients in the following calibration curve (Anton Paar, Instruction manual, L-Dens 313/323).
4222
22
)( PAKPddAJTAIddAHTAGAF
ddAETADddACTABAA
(A.2)
where T is the temperature (oC), P is the pressure (psig), and dd is the density measuring cell
frequency. AA to AK are correlation coefficients that have to be determined using the NIST
density data.
193
The temperature and pressure ranges considered are 20-190 oC and 10-120 bar (1-12 MPa) with
increments of 15 oC and 10 bar (1 MPa). Table A.5 summarizes the obtained coefficients.
Table A.5: The obtained coefficients for implementation of equation (A.2).
Coefficient Value
AA -1.64E+01
AB 1.08E-03
AC 1.02E-04
AD -1.99E-06
AE -1.91E-07
AF 2.45E-06
AG -7.84E-10
AH -1.56E-11
AI 2.00E-13
AJ 2.84E-14
AK 0
Figure A.6 shows the measured and NIST density data for nitrogen and water. All of the data is
fitted on a line (y = x) with an average deviation of 0.19267 kg/m3. It means that there is good
agreement between the measurements and the reference data which states the densitometer was
well calibrated.
Figure A.6: Measured density versus NIST density for Nitrogen and water.
Nitrogen
NIST density (kg/m3)
0 20 40 60 80 100 120
Mea
su
red
den
sit
y (
kg
/m3)
0
20
40
60
80
100
120
Water
NIST density (kg/m3)
880 900 920 940 960 980 1000
Mea
su
red
den
sit
y (
kg
/m3)
880
900
920
940
960
980
1000
194
A.6 PVT Apparatus Validation
The reliability of the designed PVT apparatus in this work has to be confirmed. This was done by
comparing the measured data using our PVT apparatus with those of in literature. Two systems of
CO2-toluene and ethane-MacKay River bitumen were considered.
A.6.1. CO2-Toluene System
To ensure the reliability of the solubility measurements, the solubility of CO2 in toluene was
measured at 35oC. A comparison between the measured solubility data in this work and the data
reported by (Nemati Lay et al., 2006) and (Finkt and Hershey, 1990) were carried out in Figure
A.7, which confirms the accuracy and reliability of the new apparatus.
Figure A.7: Comparison between experimental data of CO2 solubility in toluene and literature
data.
A.6.2. Ethane-MacKay River Bitumen System
Table A.6 summarizes the measured solubilities, densities, and viscosities of ethane-MacKay
River bitumen system in this work and in the study done by (Nourozieh, 2013). The measured data
using our PVT apparatus and the reported data by Nourozieh are in good agreement, which
confirms the reliability of the new PVT apparatus.
Pressure (MPa)
0 1 2 3 4 5 6 7 8
CO
2 m
ole
fra
cti
on
0.0
0.2
0.4
0.6
0.8
1.0
Nemati Lay et al., 2006
Fink and Hershey, 1990
Measured in this work
195
Table A.6: Phase behaviour data of ethane/MacKay River bitumen measured in this work and
results of (Nourozieh, 2013) at 100 oC.
Experimental data Pressure
(MPa) This work Nourozieh (2013)
Solubility (mol.%) 4.0 41.0 41.4
2.1 24.6 26.7
Density (kg/m3) 4.0 912.0 911.2
2.1 933.3 932.8
Viscosity (mPa.s) 4.0 46.0 38.1
2.1 78.3 80.4
196
A.7 References
Cryscope Instruction Manual, 2004. Precision System Inc., MA, USA.
Finkt, S.D., Hershey, H.C., 1990. Modeling the Vapor-Liquid Equilibria of 1,1,1 -Trichloroethane
+ Carbon Dioxide and Toluene + Carbon Dioxide at 308, 323, and 353 K. Ind. Eng. Chem.
Res 29, 295–306.
Nemati Lay, E., Taghikhani, V., Ghotbi, C., 2006. Measurement and Correlation of CO2 Solubility
in the Systems of CO2 + Toluene, CO2 + Benzene, and CO2 + n-Hexane at Near-Critical and
Supercritical Conditions. J. Chem. Eng. Data 51, 2197–2200.
Nourozieh, H., 2013. Phase Partitioning and Thermo-physical Properties of Athabasca Bitumen /
Solvent. PhD Thesis, University of Calgary, Alberta, Canada.
197
Appendix B: Copyright Permissions
198