bitumen fractionation and multicomponent characterization

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Solvent-Bitumen Systems (Unpublished doctoral thesis).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


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.,


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.,


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


Light Fraction


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


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).


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


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


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


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


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.




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


Light Fraction

Water Tank

Pump

Light Fraction

Pressure Indicator

Heavy Fraction

Oven

Vacuum Pump

Condensor


Bitumen

250 oC, 3 inHg


Light Fraction

Water Tank

Pump

Light Fraction

Pressure Indicator

Heavy Fraction

Oven

Vacuum Pump

Condensor


Bitumen

195 oC, 3 inHg


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.)


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


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.


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


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


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


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


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


0.1 1 10

Pre

dic

ted

vis

co

sit

y (

mP

a.s

)

0.1

1

10

Propane/Cut 3


0.1 1 10 100

Pre

dic

ted

vis

co

sit

y (

mP

a.s

)

0.1

1

10

100

68

3.7 References



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.



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.


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.



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




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


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.



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.




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.


4.4.1 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.


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).


4.5.1 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


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


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


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


(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 %


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


1 10

Pre

dic

ted

k-v

alu

e

1

10

Propane-Athabasca bitumen


1 10

Pre

dic

ted

k-v

alu

e

1

10

Butane-Athabasca bitumen


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.)


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


1 10 100

Pre

dic

ted

k-v

alu

e

1

10

100

Ethane-Cold Lake bitumen


1 10 100

Pre

dic

ted

k-v

alu

e

1

10

100

Propane-Cold Lake bitumen


1 10

Pre

dic

ted

k-v

alu

e

1

10

Butane-Cold Lake bitumen


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


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.




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.



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.





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.


5.4.1 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.


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


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


(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


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


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


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


100 oC 130

oC

(a)

Pressure (MPa)

0 2 4 6 8 10

Vis

co

sity (

mP

a.s

)

10

100

1000

10000


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


Pressure (MPa)

0 2 4 6 8 10

Vis

co

sity (

mP

a.s

)

10

100

1000

10000


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


Pressure (MPa)

0 2 4 6 8 10

De

nsity (

kg/m

3)

910

920

930

940

950

960

970

980


Pressure (MPa)

0 2 4 6 8 10

De

nsity (

kg/m

3)

920

930

940

950

960

970

980


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


Pressure (MPa)

1 2 3 4 5 6 7 8 9

De

nsity (

kg/m

3)

820

840

860

880

900

920

940

960


Pressure (MPa)

1 2 3 4 5 6 7 8 9

De

nsity (

kg/m

3)

820

840

860

880

900

920

940

960


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


Pressure (MPa)

1 2 3 4 5 6 7 8 9

Vis

co

sity (

mP

a.s

)

1

10

100

1000


Pressure (MPa)

1 2 3 4 5 6 7 8 9

Vis

co

sity (

mP

a.s

)

1

10

100

1000


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


Pressure (MPa)

1 2 3 4 5 6 7 8 9

Vis

co

sity (

mP

a.s

)

1

10

100

1000


Pressure (MPa)

1 2 3 4 5 6 7 8 9

Vis

co

sity (

mP

a.s

)

1

10

100

1000


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.)


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.


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)


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.


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


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


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


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)


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


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


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


Bitumen A

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 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


20 22 24 26 28 30 32

Lo

g(M

W)

2.2

2.4

2.6

2.8

3.0

3.2


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.



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.


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.



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.



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.


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


(Fu et al., 1986) 4.82

C2H6


1988b)

Proposed model 5.7


(Kariznovi et al., 2010) 7.64

(Fu et al., 1986)

Proposed model 7.2


(Fu et al., 1986) 4.72

CO2


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


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.



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.


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


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



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.


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

bitumen fractionation and multicomponent characterization

Documents