UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Effect of minerals on coke formation Sanaie, Nooshafarin 1998

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1999-0087.pdf [ 12.76MB ]
Metadata
JSON: 831-1.0059002.json
JSON-LD: 831-1.0059002-ld.json
RDF/XML (Pretty): 831-1.0059002-rdf.xml
RDF/JSON: 831-1.0059002-rdf.json
Turtle: 831-1.0059002-turtle.txt
N-Triples: 831-1.0059002-rdf-ntriples.txt
Original Record: 831-1.0059002-source.json
Full Text
831-1.0059002-fulltext.txt
Citation
831-1.0059002.ris

Full Text

EFFECT OF MINERALS ON COKE FORMATION by Nooshafarin Sanaie B.A.Sc. Sharif University of Technology, 1995 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF APPLIED SCIENCE in THE F A C U L T Y OF G R A D U A T E STUDIES DEPARTMENT OF C H E M I C A L AND BIO-RESOURCE ENGINEERING We accept this thesis as conforming to the_required standard T H E UNIVERSITY OF BRITISH C O L U M B I A December, 1998 © Nooshafarin Sanaie, In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of QjhemJ CeJL CoV\A Bio- RszSOvrc^ ng /neen'nj The University of British Columbia Vancouver, Canada Date Deo, £ 3 / iqqy DE-6 (2/88) Abstract Petroleum coke is defined as the toluene-insoluble carbonaceous solid which can form in a variety of petroleum processing units, particularly due to heating of residual oils with a high asphaltene content. Coke formation is the objective in some high temperature (500°C) processes such as delayed coking or fluid coking, but is undesirable in refineries or heavy oil upgrading processes, where it gives rise to operating problems. It has been postulated that fine solids in the bitumen could have an effect on the formation of coke precursors which occurs during heating at temperatures above 350 °C. Most of the previous studies on coke formation concentrated mainly on the reaction mechanisms and the development of the coke phase, rather than on the coke yield. There have been very few investigations concerned with the possible effect of either naturally-occurring fines in bitumen or other additives on the coke yield. The objective of this study was to determine the formation of toluene insoluble material from Cold Lake bitumen at temperatures of about 300-400 °C. In particular, the effect of the presence of solids on the formation of toluene insolubles was investigated. The coking reactions were performed at atmospheric pressure in two different batch reactors: a non-stirred tube reactor which initially contained about 1.5 g of bitumen sample, and a stirred autoclave reactor with bitumen samples of 50 or 100 g. Both reactors were purged with 200 ml/min (NPT) of nitrogen throughout the experiments and, therefore, the volatiles which were formed during the bitumen cracking reaction were continuously swept from the remaining fluid. i i In the first stage of experiments, using the non-stirred tube reactor, the toluene insoluble yield and weight loss of bitumen were measured as a function of reaction time at three different temperatures (380, 390 and 400 °C) in the absence of added solids. Volatile material was released over the whole duration of the reaction. An induction time for toluene insolubles formation was observed at the three temperatures; it became shorter as the reaction temperature increased. Similar blank experiments without solids present were performed on the 50 and 100 g samples in the autoclave reactor, but at the single temperature of 390 °C. In the second stage of the experimental program, solids, including molybdenum sulphide, silica, Alberta clay, kaolinite and native clays, were added at 2 wt% to the bitumen samples prior to the coking experiments. A l l of the solids were characterized by scanning electron microscopy (SEM). Molybdenum sulphide, which has shown catalytic activity in cracking reactions, and was reported to produce a lower coke yield in some previous studies, did not have a statistically significant effect on the coke yield or weight loss, in either the tube reactor or the autoclave reactor experiments in this study. Two different batches of silica, having mean sizes of 3.0 and 6.5 p.m were used as inert particles in the coking experiments. Only the 6.5 p_m silica particles had a significant effect, reducing the coke yield from 5.8% for the blank sample to 5.1% in the non-stirred tube reactor after 4.5 h coking at 390 °C and a nitrogen purge rate of 200 ml/min. Neither kaolinite nor southern Alberta clay had any effect on either the coke yield or the weight loss when added to the bitumen at 2.0 wt%. However, native clays, which were originally separated from Athabasca bitumen, did have an influence on the coking reaction. It was found that, when this mixture was coked in the autoclave for 4.5 hour at i i i 390 °C and 200 ml/min nitrogen purge, the coke yield decreased to 3% from the 4% obtained for the blank sample under identical conditions. No effect on volatiles formation was observed. The effect of addition of native solids was also investigated at different concentrations, and it was observed that the more of these solids added to the bitumen, the greater is the reduction in the coke yield. Some possible explanations are given for the role of the native clay in the coke formation reaction. Also the rate of the nitrogen purge rate, whose primary purpose was to remove volatiles from the reactor, had a significant impact on the coke yield both in the presence and absence of added solids. The greater the purge rate, the greater is the coke yield. The results obtained for the toluene insoluble yield and weight loss in the two types of reactors were compared. The results of the tube reactor experiments in the absence of added solids were interpreted using the reaction model of Wiehe (1993, 1997). iv Table of contents Abstract » Table of contents v List of Tables ix List of Figures xiii ACKNOWLEDGMENTS xvi Chapter 1 Introduction 1.1 Background 1 1.2 Coking and Minerals in Coke Formation 1 Chapter 2 Literature review 2.1 Introduction 4 2.1.1 Oil sands of Alberta 4 2.1.2 Extraction of Bitumen from Oil-Sands 5 2.1.3 Fine Solids in Bitumen and their Interaction with Bitumen During Extraction 9 2.2 Heavy Oil Upgrading Processes 11 2.2.1 Coking 12 2.2.2 Hydrocracking 13 2.2.3 Undesired Coking 13 2.3 Mechanism of Coke Formation 15 2.4 Kinetics of Coke Formation 19 2.5 Effect of Fine Minerals on Coke Formation 24 2.5.1 The Role of Fine Solids in Coking , 24 2.5.2 Formation of Coke-in-Oil Emulsions at Reactor Conditions 28 2.6 Objectives 33 V Chapter 3 Experimental Apparatus and Procedures 3.1 Materials 35 3.1.1. Cold-Lake Bitumen 35 3.1.2 Bitumen Characterization by SEM and EDX 36 3.2 Additives 36 3.2.1 Molybdenum Sulfide (MoS2) 36 3.2.2 SiUca Sand (Si02) 42 3.2.3 Kaolin 43 3.2.4 Southern Alberta Clay 45 3.2.5 Native Clays 47 3.2.6.Preparing the Mixture of Additives and Bitumen 50 a) Tube Reactor 50 b) Batch Stirred Reactor 50 3.3 Experimental Apparatus 51 3.3.1 Tube Reactor 51 a) Fluidized Bed Temperature Bath and Pre-Heater 52 b) Tube Reactor, Glass Liner and N 2 Gas Purging 53 c) Experimental Procedures 53 3.3.2 Batch Stirred Reactor 56 a) Experimental Procedure 60 3.3.3 Pressure Filtration 61 3.3.4 Pressure Cylinder 63 3.3.5 Filter Holder 64 a) Loading 65 b) Disassembling and Cleaning the Unit 65 Chapter 4 Results and Discussions 4.1 Tube Reactor 6 7 vi I 4.1.1 Coking in the Absence of Added Solids 67 4.1.2 Effect of Added MoS 2 Particles 72 4.1.3Effect of Added Silica Particles 76 4.2 Autoclave Reactor 80 4.2.1100 Gram Autoclave Samples 81 a) Effect of Coking Time (no solids added) 81 b) Effect of Different Additives 84 4.2.2 50 Gram Autoclave Samples 87 a) Effect of Coking Time 87 b) Effect of MoS 2 on Coking 89 c) Effect of Silica on Coking 90 4.3.Effect of Kaolin and Southern Alberta Clay on Coking 93 4.4 Effect of Native Clays 94 4.5 Effect of Concentration of Netive Clay Additives 96 4.6 Effect of the Purge Rate on Coking 99 4.7 Effect of Purge Rate on Coking in the Presence of Native Solids 101 4.8 Effect of Agitation on Coking 103 4.9 A Phase-Separation Kinetic Model for Coke Formation 104 i 4.9.1 Parallel Reaction Kinetic Model 104 4.9.2 Series Reactions Kinetic Model 106 4.9.3 Finding Parameters of the Wiehe Series Reaction Model for the Tube Reactor Data 107 4.9.4 Summary 115 Chapter 5 Conclusions and Recommendations 5.1 Conclusions 116 5.1.1 Tube Reactor (no solids added) 116 5.1.2 Tube Reactor(solids added) 117 5.1.3 Autoclave Reactor 118 5.2 Recommendations 121 vii Nomenclature 122 References 125 Appendices APPENDIX A Summary of Blank Experiments in the Tube Reactor 131 APPENDIX B Summary of Solids Added Experiments in the Tube Reactor 135 APPENDIX C Summary of Blank and Solids Added Experiments in 100 g Autoclave Reactor 153 APPENDIX D Summary of Blank, MoS 2 and Si0 2 Added Experiments in 50 g Autoclave Reactor 160 APPENDIX E Summary of Statistical Analysis for 50 g Blank and Solids Added Samples in the Autoclave Reactor 164 APPENDIX F Effect of Changing the Native Solids Concentration 170 APPENDIX G Effect of Nitrogen Purge Rate 174 APPENDIX H Derivation of TI and V as Functions of Time from Wiehe's Series Kinetic Model 179 viii L i s t of T a b l e s Table 2.1 Effect of severity of thermal processing 12 Table 2.2 Nomenclature used to describe optical texture in cokes 17 Table 2.3 Mean yields of oil and gases from replicate coking experiments at 430 °C. (Data iromTanabe and Gray, 1996) 27 Table2.4 Coke formed from thermal reactions of gas oil (at 375 °C for 1 h). (Data from Maslyiah et al, 1997.) 30 Table 3.1 Typical physical and chemical properties of Cold Lake bitumen. (Data provided by Esso Resources Canada Ltd.) 35 Table 3.2 Physical characteristics of molybdenum sulfide 39 Table 3.3 Size distribution of silica sand samples 42 Table 3.4 Physical and chemical properties of silica sand samples 42 Table 3.5 Typical chemical analysis of silica sand samples 43 Table 3.6 Whole rock analysis by plasma spectroscopy (ICP) 45 Table 4.1 Batch coking tests with no solid added 69 (a) 380 °C (b) 390 °C (c) 400 °C Table 4.2 Analysis for molybdenum sulfide in the filtered solids from tube reactor experiments at 390 °C where 2% by weight M0S2 was added initially to the sample 75 Table 4.3 Toluene insolubles and weight loss for different times. Coking Coking of 100 sample at 390 °C with a 200 ml/min nitrogen 82 Table 4.4 Toluene insolubles and weight loss results for 100 g bitumen samples with additives. Reaction at 390 °C and 200 ml/min nitrogen purge 85 Table 4.5a Toluene insolubles and weight loss for different times. Coking Coking of 50 g samples at 390 °C and 200 ml/min nitrogen purge 88 ix J Table 4.5b Toluene insolubles and weight loss for different times. Coking Coking of 50 g samples at 390 °C and 200ml/min nitrogen purge 89 Table 4.5c Toluene insolubles and weight loss for different times. Coking Coking of 50 g sample at 390 °C and 200 ml/min nitrogen purge 90 Table 4.6 Comparison between filtration system for different reactors and solid addeitives. Experiments at 390°C and 200 ml/min 92 Table 4.7 Toluene insolubles and weight loss for clay added samples. Experiments at 390 °C and 200 ml/min N 2 purge for 4.5 h coking 93 Table 4.8 Toluene insolubles and weight loss for native clay added samples. Experiments at 390 °C and 200 ml/min N 2 purge for 4.5 h coking 94 Table 4.9 Toluene insolubles and weight loss results for different native clay concentrations. Experiments at 390 °C and 200 ml/min N 2 purge for 4.5 h coking 97 Table 4.10 Weight loss and toluene insolubles for different purge rates. Experiments at 390 °C and for 4.5 h 99 Table 4.11 Six parameters of Wiehe's (1997) series reaction model 107 Table 4.12 Parameters of the Wiehe series reaction model calculated for Cold Lake bitumen using the tube reactor results (Table 4.1) 110 Table 4.13 Parameters of the Wiehe series reaction model used for predicting all of the Cold Lake bitumen data 110 Table 4.14 Experimental and predicted values for % toluene insolubles and %weight loss at 380 °C after the induction time I l l Table 4.15 Experimental and predicted values for % toluene insolubles and %weight loss at 390 °C after the induction time I l l Table 4.16 Experimental and predicted values for % toluene insolubles and %weight loss at 400 °C after the induction time 112 Table Al Toluene insolubles and weight loss results in the tube reactor. Experiments at 380 °C and 200 ml/min nitrogen purge 131 Table A2 Toluene insolubles and weight loss results in the tube reactor. Experiments at 390 °C and 200 ml/min nitrogen purge 132 Table A3 Toluene insolubles and weight loss results in the tube reactor. Experiments at 400 °C and 200 ml/min nitrogen purge 133 Table Bl.l Tube reactor experiments with M0S2 added. Experiments at 390 °C and 200 ml/min nitrogen purge 141 Table B1.2 Tube reactor experiments with M0S2 added. Experiments at 390 °C and 200 ml/min nitrogen purge 143 Table B1.3 Tube reactor experiments with MoS2 added. Experiments at 390 °C and 200 ml/min nitrogen purge 145 Table B2 Silica with the average size of 3 u,m added. Experiment at 390 °C and 200 ml/min nitrogen purge 147 Table B3 Silica with the average size of 6.5 \un added. Experiment at 390 °C and 200 ml/min nitrogen purge 149 Table B4 Toluene insolubles results corrected based on the M0S2 assay 151 Table CI Toluene insolubles and weight loss for 100 g blank sample in autoclave reactor 155 Table C2 Toluene insolubles and weight loss for 100 g sample with M0S2 added in the autoclave reactor 156 Table C3 Toluene insolubles and weight loss for 100 g sample with Si02 added in the autoclave reactor 157 Table C4 Heat-Up data of the autoclave reactor 158 Table C5 Cool down data of the autoclave reactor 159 Table Dl.a Toluene insolubles and weight loss for the blank 50 g. sample in the autoclave reactor 160 Table Dl b Toluene insolubles and weight loss for the blank 50 g. sample in the autoclave reactor 161 xi Table D2 Toluene insolubles and weight loss for 50 g. sample with MoS 2 added in the autoclave reactor 162 Table D3 Toluene insolubles and weight loss for 50 g. sample with Si0 2 added in the autoclave reactor 163 Table E l Coking experiments for 50 g blank sample in the autoclave reactor. Blank sample average values, standard deviations , and 95 % confidence intervals 164 Table E2 Coking experiments on 50 g bitumen samples with 2 wt%Alberta clay added 165 Table E3 Coking experiments on 50 g bitumen samples with 2 wt % Kaolinite added . 166 Table E4 Toluene insolubles and weight loss of the 2 wt% Native solid added experiments 168 Table F l Toluene insolubles and weight loss for 1 wt% Native solid added experiments 170 Table F2 Toluene insolubles and weight loss for 4 wt%Native solid added experiments 172 Table G l Toluene insolubles and weight loss results at 100 ml/min nitrogen purge 174 Table G2 Toluene insolubles and weight loss results at 400 ml/min nitrogen purge 175 Table G3 Toluene insolubles and weight loss results at 100 ml/min Nitrogen purge with 2wt% Native solid added 176 Table G4 Toluene insolubles and weight loss results at 400 ml/min Nitrogen purgewith 2 wt% native solid added 177 Table H. l Rate constants of Cold-Lake residuum for different temperatures 187 xii L i s t o f F i g u r e s Figure 2.1 Schematic diagram of Esso's Cold Lake bitumen extraction process 7 Figure 2.2 Yields of coke from fraction of Arabian vacuum bottoms at 400 °C 22 Figure 2.3 Yields of solubility fractions from pyrolysis of Cold lake residue at 400 °C.(Data from Wiehe, 1993.) 23 Figure 2.4 Coke yields from Athabasca residue at 430 °C. (Data from Tanabe and Gray, 1996.) 26 Figure 2.5 Asphaltene yields from Athabasca residue at 430 °C. (Data from Tanabe and Gray, 1996.) 26 Figure 2.6 Scanning electron microscope pictures of the filtered liquid. Product on the polycarbonate filter washed with xylene then with isopropanol (Pictures from Masliyah, 1997) 31 Figure 3.1 Scanning electron microscope photograph of toluene insoluble fraction of feed 37 Figure 3.2 Energy dispersive x-ray spectrum of toluene insoluble fraction 38 Figure 3.3a Scanning electron micrograph of molybdenum sulfide particles 40 Figure 3.3b Scanning electron micrograph of molybdenum sulfide particles 41 Figure 3.4 Scanning electron microscope image of kaolin particles 44 Figure 3.5 Scaning electron microscope image of southern Alberta clay particles 46 Figure 3.6a Scanning electron microscope image of native clay 48 Figure 3.6b Scanning electron microscope image of native clay 49 Figure 3.7 Tube Reactor 51 Figure 3.8 Filtration system for the tube reactor experiments 55 Figure 3.9 Schematic diagram of batch stirred reactor 56 Figure 3.10a View of the stirrer, thermowell and cooling coil of the autoclave reactor 58 Figure 3.10b View of the ceramic band type heater, two sizes of reactor, and flange which threads reactor to the safety head body 58 Figure 3.10c View of the safety head body and Magne-Drive high speed rotary agitator 59 Figure 3.11 Pressure filtration system 62 Figure 3.12 Pressure cylinder. 63 Figure 3.13 Stainless steel filter holder 64 Figure 4.1 Toluene insoluble vs. time results in the absence of added solids 68 Figure 4.2 % Weight loss vs. time for tube reactor in the absence of added solids...70 Figure 4.3 Toluene insoluble vs. weight loss for all tube reactor results in the absence of added solids 72 Figure 4.4 Toluene insolubles vs. time (solid line corresponds with the blank data at identical conditions, i.e., Fig. 4.1). Experiments at 390 °C and 200 ml/min nitrogen purge 74 Figure 4.5 Weight loss vs. time (solid line corresponds with the blank data at identical conditions, i.e., Fig. 4.2). Experiments at 390 °C and 200 ml/min nitrogen purge 76 Figure 4.6 Toluene insolubles vs. time (solid line corresponds with the blank data at identical conditions, i.e., Fig. 4.1). Experiments at 390 °C and 200 ml/min nitrogen purge 77 Figure 4.7 Weight loss vs. time (solid line corresponds with the blank data at identical conditions, i.e., Fig. 4.2). Experiments at 390 °C and 200 ml/min nitrogen purge 79 Figure 4.8 Toluene insolubles vs. weight loss (solid line is a curve fit of the blank experiments in Fig. 4.3).Points stand for different additives 79 Figure 4.9 Warm up and cool down paths for autoclave reactor containing 100 g feed sample 82 Figure 4.10 Toluene insolubles and weight loss vs. reaction time, experiments at 390 °C and 200 ml/min nitrogen purge. Comparison between tube reactor and 100 g autoclave reactor 83 Figure 4.11 Toluene insolubles vs. time with different additives in 100 g autoclave samples experiments at 390 °C and 200 ml/min nitrogen purge 85 X IV Figure 4.12 Comparison of the Toluene insolubles vs. weight loss for 1.5 g tube and 100 g autoclave. Experiments at 390 °C and 200 ml/min nitrogen purge 86 Figure 4.13 Toluene insolubles vs. time with different additives in 50 g sample Experiments at 390°C and 200 ml/min nitrogen purge 71 Figure 4.14 Toluene insolubles for blank and five different solids added samples. Experiments at 390 °C and 4.5 h with 200 ml/min nitrogen purge 96 Figure 4.15 Toluene insolubles vs. different concentrations of native solids. Experiments at 390 °C and 200 ml/min nitrogen purge for 4.5 h 98 Figure 4.16a Toluene insolubles vs. purge rate. Experiments at 390 °C and 4.5 h coking in 50 g autoclave reactor no solids added 100 Figure 4.16b Weight loss vs. purge rate. Experiments at 390 °C and 4.5 h coking in 50 g autoclave reactor no solids added 100 Figure 4.17 Toluene insolubles vs. purge rate in the presence of Native solids. Experiments at 390 °C and 4.5hours 102 Figure 4.18 Weight loss vs. purge rate in the presence of native solids. Experiments at 390 °C and 4.5 h 103 Figure 4.19 Separation scheme for reaction products used by Wiehe (1993) 105 Figure 4.20 Experimental and predicted % weight loss results for Cold Lake bitumen. Experiments at 200 ml/min nitrogen purge 112 Figure 4.21 Experimental and predicted values of % TI of Cold Lake bitumen. Experiments at 200 ml/min nitrogen purge... 113 Figure 4.22 % TI vs. % weight loss using experimental and predicted values. Experiments at 200 ml/min nitrogen purge 114 X V ACKNOWLEDGMENTS I would like to thank my supervisors, Dr. Paul Watkinson, Dr. Bruce Bowen and Dr. Kevin Smith for their support, valuable advise, and suggestions throughout my research. I would also like to thank people in the workshop and stores, specially Mr. Peter Roberts, for their assistance. Special thanks must go to Mary Mager of the Department of Metal and Materials Engineering, University of British Columbia, for helping me with the S E M and E D X measurements. The financial support of Syncrude Canada Ltd. and Imperial Oil Resources Ltd. are also gratefully acknowledged. I would also like to thank the fellow members of my research group for assisting me with my experimental work, and for creating a pleasant work environment. In closing, I would like to dedicate this thesis to my parents, for all the loving support, advice and encouragement which they have always given me over the years. xvi Chapter 1 Introduction 1. Introduction 1.1 Background As the world's supply of light, sweet crude oil is depleted, the stock of heavy oils and bitumens becomes more and more important as a component in supplying the demand for fuels and petrochemical feeds. Similarly, the need to process the residue fractions of petroleum has also increased in importance. Residue, or residuum, is operationally defined as the fraction of petroleum, heavy oil, or bitumen that does not distill under vacuum. Residues have atmospheric equivalent boiling points over 524 °C, and in refineries would be produced as the bottom product from a vacuum distillation column. Heavy oil and bitumens have residue contents of 40% or more, compared to about 11% in conventional crude oils, and hence require more processing in order to find a market. Interest in non-conventional oil sources such as tar sands and oil shales has grown steadily. These deposits, in which hydrocarbons are dispersed among solids, represent an enormous energy reserve. For example, the Athabasca tar sands deposit is estimated to contain 600 billion barrels (100 km3) of oil, of which from 5 to 10% is recoverable by present surface mining technology. After extracting the bitumen with different techniques such as hot water extraction (Chapter 2), and removal of residual water and solids, the separated bitumen is upgraded by a coking process and subsequent catalytic hydro-treating process to yield a high quality synthetic crude oil. 1.2 Coking and Minerals in Coke Formation Petroleum coke is usually defined as a toluene-insoluble carbonaceous solid, which can 1 Chapter 1 Introduction be formed in a variety of petroleum processing units, particularly due to the heating of residual oils. When petroleum fractions are heated to temperatures over 410 °C, then thermal or free-radical reactions begin to cause the mixture to crack at significant rates. Coking processes have the virtue of eliminating the residue fraction of the feed, at the cost of forming a solid carbonaceous product. The yield of coke in a given coking process tends to be proportional to the carbon residue content of the feed. The formation of large quantities of coke is a severe drawback unless the coke can be put to use. Calcined petroleum coke can be used for making anodes for aluminum manufacture, and a variety of carbon or graphite products such as brushes for electrical equipment. These applications however, require a coke which is low in mineral matter and sulfur. Coke formation may be the process objective, as in delayed coking or fluid coking, or it may be undesirable as in the fouling in furnace tubes, heat exchangers and other process vessels. Coke frequently deposits on the inner surfaces of the process equipment and causes fouling (Garette-Price etal, 1985) particularly in the processing of heavy oils and bitumen. The coke product concentrates some of the components in the feed, such as nitrogen, vanadium and mineral solids, which indicates that those components may have significant effect on the coke formation reaction (Gray, 1994). When metals such as molybdenum are added to heavy oils as hydro-processing catalysts, the resulting metal sulfides are found to be concentrated in the coke fraction (Peureux et al, 1995.). Coking involves the formation of a new phase either via liquid phase separation or by flocculation of asphaltene micelles. In either case the developing coke materials may interact with mineral solids (which exist originally or are added to the coking phase) via 2 Chapter 1 Introduction mechanisms such as nucleation of the coke phase or flocculation with coke solids. The presence of fine solids in an oil stream can provide a large surface area dispersed in proximity to the separating coke phase. Several studies have indicated the interactions between the coke and the solid particle surfaces may be important (Bradford et al, 1971). They observed that the addition of fine solids interfered with the coalescence of the mesophase1 (intermediate phase) coke, which indicated that the ability of coke materials to wet surfaces can lead to the development of coke materials with different properties. This project was undertaken in response to a concern by industry that the clay solids present in bitumen or other particles such as catalyst fines could promote undesired coke formation in the operating plant, leading to enhanced deposition problems. 1 Definition in chapter 2. 3 Chapter 2-Literature Review 2. Literature Review 2.1 Introduction A significant portion of the world's fossil fuel exists as hydrocarbons in the form of heavy oils. UNITAR, the United Nations Institute for Training and Research, formulated definitions for heavy crude oil and tar sands in 1981 (Sanford, 1991). They classified petroleum according to its viscosity and density. Heavy crude oil, where heavy refers to high density or low API gravity, has a viscosity of 100 to 10,000 mPas at the original reservoir temperature, or an API gravity of 20 to 10 at 15.6 °C. Any petroleum, or petroleum-like liquid or semi-solid with a gas-free viscosity greater than 10,000 mPas and API gravity less than 10 is considered to be tar sand oil or bitumen. As North American oil reserves have declined and the price of conventional oil has risen, interest in non-conventional oil sources such as oil sands and oil shale has grown steadily. Oil sands are composed of bitumen, water, quartz sand and clays. The quartz sand makes up the bulk of the mixture with either the oil or the water forming the continuous phase, depending on the grade (Smith and Scharmm., 1984). 2.2 Oil Sands of Alberta Oil sand deposits, in which hydrocarbons are dispersed among solids, represent an enormous energy reserve. Sand beds underlie 48,000 km 2 of the lands in northern Alberta in the four locations of Athabasca, Cold Lake, Wabasca, and Peace River. The total reserves of the 11 3 Alberta deposits are estimated to contain close to 2.15 x 10 m of bitumen (Takamura, 1980). The proven bitumen present in Alberta's deposits is of the order of 900 billion barrels. 4 Chapter 2-Literature Review Analysis shows the oil sands contain an average of 12% bitumen and a maximum of 18%. Approximately 275 billion barrels of the bitumen is estimated to be recoverable, which is equivalent to about 80 % of the proven reserves of the Middle East, the largest known reservoir of conventional petroleum. Only about one-tenth of the bitumen is economically recoverable by surface mining techniques. The remainder is too deeply situated (>750 m) for economic retrieval in this way (Speight, 1991). Underground mining for recovery of intermediate depth material and various techniques of in-situ extraction of the bitumen from deposits under more than 100 m of overburden are being vigorously investigated, but none permits as complete a recovery of bitumen as the 90% range which is possible by open pit mining in combination with hot-water extraction. 2.1.2 Extraction of Bitumen from Oil Sands: The most critical factor in the recovery of bitumen from the Alberta oil sands is the extremely high viscosity of bitumen in the reservoir. Two different methods are used to exploit this bitumen: a) In-Situ Extraction: In order to recover the bitumen by an in-situ method, it is necessary to reduce its viscosity or otherwise mobilize it so that it will flow to the producing well bore. The major in-situ test research work was begun in the 1950s by Shell Canada in the Athabasca deposits. They attempted to recover the bitumen by heating in-situ, generally with steam, and pumping the mobilized bitumen to the surface (Warne, 1984). In the 1970s and 1980s, pilot programs of various sizes were carried out to investigate in-situ recovery methods in most of the oil sand 5 Chapter 2-Literature Review deposits. The injection of steam continues to be the most popular method and the most preferable method in currently active projects. In-situ recovery using steam to heat the bitumen is now being used commercially by Esso Resources Canada in its Cold Lake deposit, a 8 x 10s km 2 site located 225 km northeast of Edmonton. Cold Lake bitumen extraction can be divided into four stages: • In-situ bitumen production by cyclic steam injection. • Surface bitumen cleaning and separation. • Surface bitumen upgrading. • Water treatment and disposal. Bitumen is recovered by an in-situ steam displacement process ("steam stimulation"). This involves the injection of a steam-water (80:20) mixture into the reservoir at 300 to 315 °C for weeks to months. The steam-water mixture serves as the drive-fluid which mobilizes the heavy oil, that can then be recovered at the surface. The latent heat of steam mobilizes the highly viscous bitumen, creating a bitumen-water emulsion which can be produced by conventional means. Production of bitumen is achieved through blowdown of the reservoir pressure, then by pumping the well (Fig. 2.1). The cycle is terminated when either production rates or temperatures become too low, or the water/bitumen ratio becomes too high. The entire cycle can be repeated several times. After the bitumen is pumped to the central plant, water and gas are removed in separators, and the residual water is removed by heat and electrostatic dehydration. Raw bitumen from these cleaning and separation steps is first desalted and then distilled in atmospheric or vacuum fractionating units. Then different fractions will be sent out to the refineries off site. 6 Chapter 2-LUerature Review Light ends Bitumen, water and gas Injection and production wells Bitumen separation j 5 t e a m _ ^ Gas Steam generation Fractionation Bitumen Make-up fuel Water Manville sand Water Water from Cold Lake M Water treatment k Water disposal to wells and surface Figure 2.1. Schematic diagram of Esso's Cold Lake bitumen extraction process 7 Chapter 2-LUerature Review b) Mining and Hot Water Extraction: The hot water extraction process was pioneered and developed in the 1940s by Clark (1950). This process is currently used by Syncrude Canada Ltd. to extract bitumen from Athabasca oil sands. There are two objectives in the hot water process: • First it is desired to separate the bitumen from the coarse and fine (clay) solids, which are predominantly water-wet. • Second it is desired to aerate the bitumen globules but not the solids, so that the bitumen will rise quickly in a flotation vessel to form a froth layer. The hot water process consists of three major steps: 4 In the first step, oil sand is agitated in hot water (at about 80 °C) with a small amount of sodium hydroxide, to maintain the pH in the range of 8.0 to 8.5. The role of sodium hydroxide in the process is not to adjust the slurry pH, but mainly to produce carboxylate surfactants from precursors occurring in the bitumen (Smith and Scharmm, 1984). The low iso-electric-point (Le.p.) value of the bitumen implies that a significant amount of acid-type surfactants is contained in the bitumen. The ionization of these surfactants at slightly alkaline pH gives the bitumen a highly negatively-charged and hydrated surface, which is beneficial to bitumen liberation (Chung and Dai, 1995). • In the second step, the sand grains that have settled to the bottom of the settling tank are removed and the oil froth that floats to the top is recovered by slamming the surface. Fine particulate matter, dominated by clay minerals remains in what is called the middling stream. 4 This stream is subjected to a third processing step, which provides incremental recovery of Chapter 2-Literature Review suspended bitumen, and is accomplished by conventional froth flotation (Takamura, 1982). The bitumen recovered from these two steps is treated with naphtha diluent, and passed through settlers or a centrifuge before going to the upgrading plant. There the bitumen is further upgraded by a coking process and subsequent catalytic hydrotreating to yield a high quality synthetic crude oil (Goldman et al, 1980). 2.1.3 Fine Solids in Bitumen and their Interaction with Bitumen During Extraction The solid particles found in the oil sand are predominantly quartz in the 44-500 urn size range. However, appreciable quantities of finely divided clays and small amounts of heavy minerals also occur (Bowman, 1967). The clay minerals are predominantly kaolinite, illite, and a small amount of montmorillonite. These only appear in the fine fraction, which is normally defined as particle diameter smaller than 44 um (Takamura, 1982). The sand grains in the oil sand matrix are surrounded by a film of water, which is, in turn, surrounded by bitumen. The latter is generally the continuous phase. Takumura (1982) concluded that the bitumen and solids are separated for the most part by a water film of about 10 nm in thickness. A portion of the clay and fine silt is associated with the bitumen at the bitumen-water interface. This mineral matter is largely hydrophobic as a result of the adsorption of polar organic molecules onto the mineral surface. When the oil sand matrix is mixed with hot water (during the hot water process), the highly viscous bitumen becomes mobile, and the oil-sand structure breaks down to give water-wet sand grains, which are essentially free of bitumen, and bitumen droplets which are suspended in the aqueous phase, and have associated with them, at the bitumen-water interface, clay and other fine solids. 9 Chapter 2-Literature Review The mechanism by which clay binds to bitumen may be through ionic bonding between clay minerals and the organic acids and natural polar organics, which would naturally migrate to the surface of the bitumen droplets (Levine and Sanford, 1980). In order for bitumen to float and form a froth, it must be freed of most of the minerals and must adsorb enough gas to make the particles less dense than water. It is necessary to apply mechanical energy (shear forces) in the presence of anionic surfactants in order to displace the clay particles from the surface of the bitumen and make the clay hydrophilic. The anionic surfactants result from the interaction of sodium hydroxide with organic acid groups which extend into the water phase. Displacement of clay from the bitumen droplets is the difficult and "processibility determining" step in the hot water extraction process, and is somewhat analogous to removing a hydrophobic coating from a hydrophobic material. Once the bitumen droplets are at least partly free of the clay membrane at the bitumen-water interface (Gowers, 1968), they can then attach to air, water vapor or other gas bubbles, and can also coalesce to form larger bitumen drops. The floated slurry is introduced into the separation vessel. The separation vessel is simply a settling vessel and the fate of the bitumen droplets and inorganic materials in this vessel is largely a function of the rise velocity of individual species and the fluid dynamics of the vesseL Under ideal circumstances, all of the sand would sink, all of the aerated bitumen would float and all of the non-aerated bitumen (associated with clays), water-wet clay and fines with a near zero rise velocity would be removed in the middling stream. In actual fact, some of the coarse sand and some aerated bitumen is removed with the sand tailings. To summarize, current theory considers the transformation of hydrophobic clay on the bitumen 10 Chapter 2-Literature Review surface to hydrophilic clay to be the important processibility step. The separation of bitumen from sand, the coalescence of bitumen droplets and the bitumen-air attachment step are all considered to be fast relative to this processibility determining step (Sanford, 1983). 2.2 Heavy Oil Upgrading Processes It is well known that heavy feeds may contain components, such as sulfur, nitrogen, metals and also considerable amounts of cokable materials (referred to as Conradson Carbon Residue, CCR), which make processing difficult. Upgrading processes convert heavy feeds into lower boiling products and, at the same time, remove metals, heteroatoms and cokable materials. One of the operational problems of upgrading heavy feeds by thermal means is the formation of undesirably large amounts of coke1, which represents a loss in hydrocarbonaceous material, and also may need a costly separation step. Thus distillate production from heavy stocks requires upgrading which improves properties, especially the H/C ratio. This is equivalent to adding hydrogen or rejecting carbon. The processing techniques currently in use for upgrading heavy oil used can be categorized as follows: • Thermal processes (visbreaking, coking). • Catalytic hydroprocesses (hydrocracking). • Non-catalytic hydroprocesses (hydrovisbreaking, donor-solvent processes). • Catalytic cracking. 1 Definition on page 13. 11 Chapter 2-Literature Review Thermal processes suffer from several disadvantages including low distillate conversion with poor quality, low CCR reduction, poor heteroatom removal and production of large volumes of high sulfur and high metal coke (Del Bianco, 1993). For this reason, considerable attention has been paid to hydrogen addition technologies (e.g., hydrocracking) as opposed to carbon rejection technologies (e.g., thermal processes). 2.2.1 Coking Coking processes have the virtue of eliminating the residue fraction of the feed at the cost of forming solid carbonaceous products. The severity of thermal processing, which is determined by the combination of reaction time and temperature, controls the conversion and the product characteristics. Table 2.1 summarizes the severity and conversion characteristics of thermal processes. Table 2.1 Effect of severity of thermal processing Level of severity Process Time (s) Temperature (°C) Conversion Mild Visbreaking 90 425-500 Low Delayed coking (1) 435-480 High High Hydrocracking 3600 420-440 Med-High Fluid coking 25 510-540 High Extreme Ultrapyrolysis 0.5 >540 High (1) Semi -batch process. 12 Chapter 2-LUerature Review Mild and high severity processes are frequently used for the processing of residue fractions, while conditions similar to "Ultrapyrolysis" (high temperature and very short residence time) are only used commercially for cracking ethane, propane, butane, and light distillate fuels to produce ethylene and higher olefins (Gray, 1994). 2.2.2 Hydrocracking Due to favorable liquid yields from hydroconversion processes, these hydrogen addition methods are very popular for primary upgrading. Low hydrogen cost, and a lack of markets for coke products produced by thermal processes, are two factors that favor the hydrogen addition route. The presence of high pressure hydrogen (>5-7 MPa) during conversion of residua has a number of beneficial effects. Hydrogen tends to suppress free-radical addition and hydrogenation reactions. In the presence of hydrogenation catalysts, hydrogen converts aromatic and hetroaromatic species, and further converts heteroatoms to hydrogen sulfide, water, and ammonia. Al l hydroconversion processes involve a complex suite of series and parallel reactions: cracking, hydrogenation, sulfur removal metal removal, etc. Almost all hydroconversion processes use a catalyst or additive to control the formation of coke, and to serve as a surface for deposition of metals, to enhance hydrogenation reactions. 2.2.3 Undesired Coking Coking of residua from petroleum and bitumen is of interest in a number of process technologies, from delayed coking to hydrocracking. Coke is defined operationally in petroleum refining as a hydrocarbon material which is insoluble in an aromatic solvent such as 13 Chapter 2-Literature Review benzene or toluene. Coking takes place by either a semi-batch process (delayed coking) or continuous processes (fluid coking or coking). In coking processes, the objective is to maximize the yield of cracked product while also producing a coke material of desired quality. Coke formation should be confined to the coking reactor, and it is to be avoided in the feed system, in the liquid product lines, and in downstream equipment. In hydrocracking processes, on the other hand, the objective is to avoid the formation of coke deposits throughout (Tanabe and Gray, 1997). In these processes, deposition of coke on the hydrotreating catalyst reduces the catalyst activity. This is likely due to coke blocking the edges of the metal crystallites. At high coke levels, pore blockage would also reduce catalyst activity (Gray et al, 1994). Also the deposition of coke on the inner surfaces of the process equipment in the upgrading plant is costly and undesirable. It leads to increased downtime during scheduled shutdowns, may decrease the feed throughput during the run, causes upsets in the unit operation, and may even force unscheduled shutdowns. In addition, accumulation of unwanted deposits on the surface of heat exchangers represents a resistance to the transfer of heat and therefore reduces the efficiency of the heat exchangers (Bott, 1995) Undesired formation can also block separators and other vessels. In summary, the formation and deposition of coke is known to be a potential operating problem in heavy oil and bitumen treating processes. Therefore rninimizing undesired coke formation is of prime importance for industries producing fuels from residues or heavy materials (Srinivasan andMcKnight, 1994). 14 Chapter 2-LUerature Review 2.3 Mechanism of Coke Formation The formation of coke is usually attributed to the condensation and polymerization of aromatic components, eventually giving a carbon-rich material The chemical species in the coke product will depend on the composition of the starting feed and the process history of the coke. These factors are well understood for mesophase coke, that is for the formation of a structured anisotropic phase within the coke material Mesophase means "intermediate phase", and is a phase which is composed of lamellar molecules. The structure of these molecules is based on a hexagonal network of carbon atoms in a graphite lattice (Marsh, 1973). As early as 1944 people looked for significant changes in the structure between non-graphitizing and graphitizing carbon during the carbonization process, using x-ray diffraction techniques (Blayden et al, 1944). Diffraction patterns were interpreted in terms of the concept that lamellar constituent molecules within the carbonizing system formed stacked units, called crystallites. The carbonization systems leading to graphitizable carbon all pass through a fluid phase and all produce the liquid crystal/mesophase material. The graphitizing carbons are described as soft carbons, having anisotropic properties and possessing low surface areas and little porosity. A powerful analytical tool used in studies of coke structure is the polarized light optical microscope, which either operates with cross-polarized light (Marsh and Smith, 1978) or makes use of reflection interference colors, and definitively categorizes carbon according to its optical texture (Forrest and Marsh, 1977). Optical texture is a measure of the size and coalescence behavior of the mesophase developing from the isotropic parent petroleum and it enables structure and orientation of lamellae within 15 Chapter 2-Literature Review the mesophase to be established. The microscopic appearance of optical texture can be quantified in an arbitrary way by making use of the optical texture index (OTI). The OTI is calculated using the formula (Oyua et al, 1983): O T I = X f i . O T I i (2-1) where f i is the fraction of component of optical texture T in the overall polished surface of the coke, and OTIi is an arbitrary factor for each recognizable component of optical texture related to the relative sizes of the component (Table 2.2). For example, for an isotropic carbon OTI = 0, and for an anisotropic carbon which is composed entirely of domains, OTI = 30. All other anisotropic carbons have OTI values in the range of 0 - 30. The growth unit of mesophase in pitch was detected to be less than 0.1 pjn in diameter using transmission electron microscopy (TEM) (Jhnatowicz et al, 1966). The carbonization of heavy petroleum products shows the initial stacking of molecules which must represent the beginning of mesophase. It is the subsequent growth and coalescence of the planar aromatic structures which dictate coke properties. If polymerization of the feed constituent molecules occurs to form non-lamellar molecules, then random relative orientation will develop, and the resultant coke will become truly non-graphitizable (isotropic). The resultant optical texture of the coke is directly dependent upon the viscosity of the preceding mesophase. Previous studies show that for the systems like pitch, which is rich in oxygen and sulfur and possesses relatively high viscosity, the resultant coke has the smallest size of optical texture. For another pitch system which has lower viscosity, the resultant coke possesses a larger size of optical texture (Yokono and Marsh, 1983). There are many factors which could 16 Chapter 2-LUerature Review influence the size of optical texture in coke: Table 2.2 Nomenclature used to describe optical texture in cokes Component of optical texture seen in microscopy Abbreviation Size Optical textui index (OTI) Mosaics Isotropic I No optical activity 0 Fine-grained M f < 1.5 prn 1 Medium-grained Mm 1.5-5.0 urn 3 Coarse-grained Mc 5.0 - 10.0 j i m 7 Supra Ms Aligned mosaics 10 Small domains SD 10.0 - 60.0 u.m 20 Domains D > 60 pm 30 Anisotropv Medium-flow M F A < 30 u.m length 7 < 5 pm width Coarse-flow C F A 30 - 60 u.m length 20 5-10 u.m width Flow domain FD > 60 \im length" 30 > 10 pjn width 1. Rate of carbonization: It is reported that a decreased rate of carbonization enhances the size of the resultant optical textures. This effect is concerned with the thermolysis of parent coal or pitch and rates of evolution of volatile matter, some of which, if retained, promote the fluidity (decreased viscosity) of the system and hence promote the size of optical texture (Marsh et al., 1980). 2. Heat treatment temperature and soak time: The study of Mochida and Marsh (1979) 17 Chapter 2-LUerature Review indicates, unlike in chemical kinetics, time and temperature for mesophase formation are not interdependent. The reason for this is the controlling influence of viscosity. Maximum size of optical texture and coalescence result if the mesophase is formed under conditions which provide a minimum viscosity as quickly as possible. Probably the rate controlling processes for mesophase growth are not the dehydrogenated polymerization reactions. Therefore, the rapid attainment of temperature (~ 400 °C) provides the necessary molecular size and consequently the resultant mesophase shows minimum viscosity because it is at a high temperature. Mesophase formed at relatively lower temperatures will have a higher viscosity and coalescence behavior can be restricted. 3. Pressure of carbonization: The effect of a pressurized carbonization is to create a closed system preventing the loss of volatile materials. Hence the carbon yields increase. Furthermore the material normally lost as volatiles in the open system is now retained and the effect of this, by reducing turbulence and bubble formation, is to enhance the size of the resultant optical textures. If a high pressure of about 300 MPa (generated hydraulically) is applied to the system, the effect is to enhance the viscosity of the system and thereby prevent coalescence of the mesophase (Marsh and Latham, 1986). There have been many different mechanisms proposed for coke formation. At the temperatures typical of coking and residua hydrocracking processes (in excess of 420 °C), coke materials can exhibit fluid properties such as coalescence and deformation by shear (Mochida et al, 1988). The viscosity of the coke material at reactor conditions will depend on the extent of cross-linking and devolatilization reactions. Depending on the composition of the liquid phase at reactor conditions, separation of the dense aromatic liquid phase may precede actual coke 18 Chapter 2-Literature Review formation (Mochida et al, 1988; Dudekhin-Lalla et al, 1989; Wiehe, 1993). With time, mesophase may form spheres which can coalesce depending on the hydrodynamics of the isotropic liquid and the mesophase itself (Mochida etal, 1988). An alternative mechanistic view of coke formation has been suggested by Storm et al (1995). They suggested that coke formation is a liquid-solid transition, analogous to flocculation of a dispersed polymer in a liquid phase. Based on the size distribution of the dispersed asphaltenes and on rheological measurements, they proposed that the dispersed asphaltene components flocculate as temperature increases over 150 °C. These floes would in turn give coke formation at reactor temperatures. 2.4 Kinetics of Coke Formation Thermal cracking processes are commonly used today to convert petroleum residua into distillable products. Example processes include visbreaking, delayed coking, fluid-coking and flexicoking. In all of these processes the simultaneous formation of coke limits the conversion of residua to distillable liquid products. Although many aspects of the kinetics of residuum thermolysis and the mechanism of coke formation have been published, there has not been much effort made to pull this information together in a single comprehensive model. The available data suggest that coke formation is a complex process involving both chemical reactions and thermodynamic behavior. Like asphaltene, coke should be viewed as a solubility fraction. Its physical state at room temperature is solid, and it is insoluble in toluene and other standard solvents. Many previous investigators follow the lead of Levinter et al (1967) and postulate that coke is produced as a 19 Chapter 2-LUerature Review direct byproduct of residuum thermolysis by a sequence of condensation polymerization steps from the lightest to the heaviest fractions, e.g.: oils • resins • asphaltenes • carbenes • coke However, kinetic models that contains such sequences of direct chemical reaction to coke fail to predict the induction period which has been observed experimentally by many investigators. Magaril et al (1971) were the first to postulate that coke formation is triggered by the phase separation of asphaltenes. Unfortunately their scattered kinetic data led them to use a zeroth-order kinetics rather than the first-order kinetics expected for thermolysis. Yan (1987) described coke formation in visbreaking as resulting from a phase-separation step, but did not include this step in his kinematic model of coke formation. The following two-step mechanism has emerged from studies of whole oils and solubility fractions (Savage and Klein, 1988; Wiehe, 1993): 1. Thermal reactions result in the formation of high molecular weight, aromatic components in solution in the liquid phase. Reactions that contribute to this process are the cracking of side chains from aromatic groups, dehydrogenation of naphthenes to form aromatics, condensation of aliphatic structures to form aromatics, and dimerization/oligomerization reactions. Loss of side chains always accompanies thermal cracking, while dehydrogenation and condensation reactions are favored by hydrogen deficient conditions. The condensation and oligomerization reactions are also enhanced by the presence of Lewis acids, for example AlCb (Mochida et al, 1977). 2. Once the concentration of this material reaches a critical value, phase separation occurs giving a denser aromatic liquid phase. Microscopic examination of coke particles often 20 Chapter 2-Literature Review shows evidence of mesophase; spherical domains that exhibit the anisotropic optical characteristics of a liquid crystal. This phenomenon is consistent with the formation of a second liquid phase; the mesophase liquid is denser than the rest of the hydrocarbon, and has a higher surface tension. From these mechanisms, it has been concluded that coke may not form immediately if the solubility limit is not exceeded, so that an induction time is observed. Also higher molecular weight fractions should give more coke, and a higher asphaltene content in feed will also, in general correlate with higher coke yield (Fig. 2.2) (Banerjee et al, 1986). The phase separation may be very sensitive to surface chemistry, hydrodynamics, and surface to volume ratio, similar to other processes that require nucleation. The data in Figs. (2.2) and (2.3) provide examples of these effects. Banerjee et al (1986) used quartz boats with 200 mg samples, pyrolyzed in flowing inert gas at 400 °C, while Wiehe (1993) used an equivalent method at 400 °C but with 2-3 g samples. In the former case, coking was completed in 10-15 min, while in the latter case, an induction period of up to 45 min was observed before any coke was formed (Fig. 2.3). Wiehe has proposed a simple kinetic model to account for the data of Fig. (2-3) using the following two step approach (see also Chapter 4, Section 4.9.1): (2-2) (2-3) (2-4) (2-5) FT K h ^ a A * + (l-a)V A + K a • bA* + c H -(l-b-c)V A *max = S L ( H + H*) 21 Chapter 2-Literature Review At long reaction times: A * e x • (l-y)TI + yH* (2-6) where H + and A + are maltenes and asphaltenes in the feed, H* and A* are the corresponding products, V is the volatiles produced, and TI is coke. The parameters a, b, c, and y are stoichiometric coefficients. The solubility limit, SL, is given as a fraction of total maltenes, and this limit determines the maximum cracked asphaltene concentration in solution, A max. The excess asphaltene, A* e x , separates as a phase and gives coke. 40 Time (min) Figure.2.2 Yields of coke from fraction of Arabian vacuum bottoms at 400 °C Soft resin was separated from the maltenes by retention on Attapulgus clay and eluted by methyl ethyl ketone. Hard resin was eluted from the clay by tetrahydrofuran. (Data from Banerjee et al., 1986.) 22 Chapter 2-LUerature Review The cracking of the maltenes and asphaltenes (Eqs. 2-2 and 2-3 respectively) was assumed to be first order, following previous work (e.g. Banerjee et al, 1986). The value of the parameter K H was found to be 0 .013 min"1 and K A was 0.026 min 1 from experiments on the respective isolated fractions. The stoichiometric coefficients and solubility limit, SL, were determined from data for pyrolysis of maltenes and the whole residuum The best fit values of the parameters were a = 0.221, b = 0.825, c = 0.02, y = 0.3 and SL = 0.49 for the whole residua and SL = 0.61 when the maltenes were pyrolyzed separately. The curves shown in Fig. 2.3 were caculated from the kinetics implied by Eqs. (2-2) through (2-6) using these stoichiometric parameters and SL = 0.49. The model was consistent with the observed induction behavior, and the maximum in asphaltene concentration. The solubility limit, SL, depended on the initial composition of the reacting mixture as expected. Time (min) Figure 2.3 Yields of solubility fractions from pyrolysis of Cold Lake residue at 400 °C. (Data from Wiehe, 1993.) 23 Chapter 2-LUerature Review 2.5 Effect of Fine Minerals on Coke Formation Coke formation involves the formation of a new phase, either via liquid phase separation (Mochida et al, 1988; Wiehe, 1993) or by flocculation of solids (Storm et al, 1995). In either case we would expect the developing coke material to interact with the mineral solids, via mechanisms such as nucleation of the coke phase or flocculation with the coke solids. Mineral solids tend to be concentrated in the coke products, which indicates significant interaction (Gray, 1994). Fine clays can stabilize oil-water emulsions (Yan and Masliyah, 1995); therefore, by the same analogy, Tanabe and Gray (1996) expected that solids with suitable surface properties should stabilize an emulsion of coke precursors in oil. 2.5.1 The Role of Fine Solids in Coking Tanabe and Gray (1996) investigated the role of mineral solids in the kinetics of coke formation. They were interested in how the concentration of mineral solids might affect the coke yield. Two different feeds were used in their coking experiments, Athabasca bitumen vacuum residua (524 °C *) and residua from extraction of Athabasca vacuum residue with supercritical pentane (Chung and Dai, 1996). Their experimental procedures consisted of the followings steps: • Removal of natural fine solids from the feed. In order to remove the fines, the feed was digested in 40 parts of toluene for 12 h and the final solution was filtered through a 0.25 p_m filter. The recovered solid was dried at 90 °C for 2 h and the solvent was removed from the oil by rotary evaporation, followed by heating on a hot plate for 12 h at 120 °C and then vacuum drying for 12 h at 120 °C. • Coking of the solid-free feed. Three grams of feed were placed in a 15 ml batch micro-24 Chapter 2-Literature Review reactor, which was pressurized to 40 MPa with nitrogen after venting. Then the reactor was heated to 430 °C by the aid of a fluidized sand bath and maintained at that temperature for a specified period of time (ranging from 10 to 90 min) before being cooled down in water. The coking experiment was also repeated on the original feed following the same procedures as outlined above. • Addition of the removed fine solids to the solid-free feed and repeating the same coking experiments. Two methods of sonication and heating plus agitation were used to disperse 1.8 wt% of the dried filtered solids into the feed. Coke, asphaltene, and gas yields were measured after each experiment. Coke was defined as the toluene-insoluble fraction corrected for the solids in the original feed. These solids were predominantly clay minerals. A comparison of the coke yield results for the solid-free feed and original feed showed that removal of solids eliminated the induction time, so that the solid-free vacuum residue gave more coke at every time interval (Fig. 2.4). The trend for the asphaltene was reversed in that the solid-free vacuum residue gave lower asphaltene yields between 20 and 60 min (Fig. 2.5). Tanabe and Gray explained that these differences in coking behavior could have been caused by either of two factors: chemical alteration of the residue due to dilution with toluene, filtration, and subsequent drying, or interaction of the solids with the developing coke phase. The former case was investigated by preparing a sample which was treated with solvent (following the same procedure as in the solid-free feed), but was not filtered, so that the sample contained the original solids. Coking of this sample, resulted the same coke yield as in the original vacuum residue. 25 Chapter 2-LUerature Review 25 20 + 15 1 2 .2 £ 10 o O 5 0 & • - Vacuum residua m - Solid free A Solid addition 10 20 30 40 Time (min) 50 60 Figure 2.4 Coke yields from Athabasca residue at 430 °C. (Data from Tanabe and Gray, 1996.) Figure 2.5 Asphaltene yields from Athabasca residue at 430 °C. (Data from Tanabe and Gray, 1996.) 26 Chapter 2-Literature Review Therefore, they concluded that the effect was likely because of interaction between the fine suspended solids and the developing coke phase. Thus they expected that the addition of solids to the solid-free material should result in behavior similar to that of the original vacuum residue. But further investigation showed that the addition of solids gave similar yields of asphaltene and coke as were obtained for the solid-free vacuum residue (Fig.2.4). The latter result was explained by the poor re-dispersion of the fine solids in the vacuum residue, after they were collected as a dried cake from filtration. They suggested that the inhibiting effect of solids on coke formation could be explained in terms of two possible mechanisms: a) The fine solids could inhibit free radical chain reactions by providing a surface for termination of radicals. If that was the case, the presence of solids would also suppress the formation of gases. However Table 2.3 shows that their presence had no effect on the formation of gas; therefore, fine solids did not suppress the free radical reactions. Table 2.3 Mean yields of oil and gas from replicate coking experiments at 430 °C. (Data from Tanabe and Gray, 1996.) Time (min) Vacuum residue Solids-free vacuum residue Oil gas oil gas 0 70.1 0 66.3 0 10 72.2 1.8 68.4 1.7 20 63.7 3.8 63.9 3.0 30 66.7 5.3 61.8 4.9 40 60.1 6.4 56.5 8.8 60 55.1 9.4 53.4 10.4 90 45.3 13.8 45.3 13.0 27 Chapter 2-LUerature Review b) The fine solids could stabilize highly dispersed coke precursors. If the solids accumulate at the interface between the oil and the coke, then the coalescence of the coke into larger droplets would be inhibited. This mechanism would also account for the effective collection of the mineral solids in the coke products (Peureux et al, 1995). Also, the role of the solids in the oil/coke mixture is somewhat akin to the findings of Beuther et aL (1980), who observed that fine solids inhibited the coalescence of the mesophase in a mixture of isotropic pitch mesophase.-2.5.2 Formation of Coke-in-Oil Emulsions at Reactor Conditions Previous studies suggest that the fluid phase properties of coke at reactor conditions, such as wettability and surface tension, might be important in the development of the coke phase. In an investigation by Gray and Masliyah (1997), the formation of coke and its interactions with solid surfaces of different wettabilities at reactor conditions was studied. The experiments were performed by heating coker gas oil from Athabasca bitumen in a stirred batch reactor at 300-420 °C and initial hydrogen pressure of 7 MPa for 1 hour. Then, after the solids were filtered from the product oil, they were analyzed by scanning electron microscopy (SEM), energy dispersive x-ray spectroscopy (EDX), and elemental analysis. The wettability of the surface of the additives was qualitatively defined as being either hydrophobic or hydrophilic. Kaolin particles (the most abundant clay mineral in the Athabasca oil sands) having two sizes, 5 urn and 0.2 pm, were selected as the hydrophilic material. Carbon black with a mean particle size of 0.2 p:m was selected as the hydrophobic material. Also a portion of the 5 [im kaolin particles were made hydrophobic by coating the particles with asphaltenes, following the 28 Chapter 2-LUerature Review method of Yan and Masliyah (1996). The solids were dispersed in a sample of gas-oil with the aid of a blender to make a 0.39 wt% solid mixture in the gas-oil. The coking reaction was carried out on 240 g of feed at 400 °C for 1 hour in a 500 ml stirred autoclave reactor. The reactor was purged with nitrogen and pressurized with hydrogen to 7 MPa before the reaction started. After the reaction, about 20 ml of the well-mixed liquid product were filtered through a 0.22 prn filter and washed with 100 ml of methylene chloride. The filter cake was weighed and then sent out for analysis. The coke yield was calculated as follows: mB.w. „ . W ( _ -mf .ws f .vv,- f %TI = ' 'P P f (2-7) mf.wsf.wif where mP and ny are the masses of the product and feed, ws„P and vv,,/ are the mass fractions of solids in the product and feed oils, while wip and vv, r are the mass fractions of the tie element in the solids. Carbon was used as a tie component in determining the coke yield in the kaolin case and sulfur was used to determine the coke yield in experiments with carbon black. Carbon and sulfur analyses of the filtered solids were used for this purpose (i.e., via Eq. 2-7), because direct gravimetric determinations were unreliable. Table 2.4 shows that the coke yield on carbon basis, after the addition of either 5 p:m or 0.2 p:m kaolin, is almost the same as in the solid-free case. The yield of coke on a sulfur basis was less reliable due to the low concentration of sulfur in the filtered solid. However the 0.2 p;m carbon black data showed no significant effect of added solids on the overall yield of coke. The coke yields from experiments with asphaltene-coated kaolin could not be determined because the asphaltene interfered with both the carbon and sulfur analyses. 29 Chapter 2-Literature Review Table 2.4 Coke formed from thermal reactions of gas-oil (reaction at 375 ° C for 1 h). (Data from Masliyah et al., 1997.) Solid added Yield of carbon in Yield of sulfur in Mean diameter of coke coke (g/kg feed) coke (g/kg feed) spheres (p.m) No solids 0.181 0.031 2.5 + 0.2 added 5 iim kaolin 0.170 0.063 1.9 + 0.2 0.2 pjm 0.167 0.047 not observed kaolin 0.2 pjn not applicable 0.028 not observed carbon black * Error bounds give the 95% confidence interval based on a count of at least 100 particles Specimens for SEM and EDX analyses were prepared by filtering five or six drops of the well-mixed liquid products through 0.2 pjn polycarbonate filter, and washing with xylene and then isopropanoL The SEM results revealed the presence of spherical coke particles, when the gas-oil sample was reacted at temperatures of 375 °C or higher (Fig. 2.6.a). Also the mean sizes of the coke spheres were measured using SEM. 30 Chapter 2-LUerature Review Figure 2.6 Scanning electron microscope pictures of the filtered liquid. Product on the polycarbonate filter washed with xylene and then with isopropanol. (Pictures from Mashyah, 1997.) a) Gas-oil reacted at 375 °C, b) 0.2 um kaolin added to gas-oil and reacted at 375 °C, c) 0.2 um carbon black particles added to gas-oil and reacted at 375 °C, d) 5 um kaolin particles coated with asphaltene added to gas-oil, reacted at 375 °C 31 Chapter 2-Literature Review The addition of kaolin resulted in a small reduction in the mean size of the coke spheres (Table 2.4), but the coke did not appear to wet the clay surface. Following the same procedure with the addition of 0.2 njn kaolin caused the formation of spheres in the product to disappear (Fig. 2.6.b). Gray and Masliyah concluded that providing a sufficient solids surface area in the samples can suppress the formation of spheres in the product. The addition of hydrophobic particles of 0.2 pm carbon black gave agglomerated solid particles in the product, with adhered spheres of coke. Separate coke spheres were not observed (Fig. 2.6.c). Carbon black particles could change coke formation either by surface chemistry or by providing more surface area than kaolin clay. In order to distinguish between the two factors, the gas-oil was mixed with asphaltene coated 5 pjn kaolin. Following the same procedure, the filtered solid contained no spheres (Fig. 2.6.d). Their conclusion was that the addition of fine solids affected the distributions and the morphology of coke, but not its total yield. Although both the 0.2 pjn kaolin and the 0.2 |jm carbon black prevented the formation of coke spheres, the kaolin samples did not exhibit agglomeration. Similarly, the hydrophobic 5 pjn kaolin (coated with asphaltene) did not exhibit agglomeration, although it also suppressed the formation of coke spheres. Therefore the observed behavior (Wang et al, 1997) of the coke spheres was consistent with the formation of a liquid rnicroemulsion at reactor conditions, in agreement with proposals that liquid-liquid phase separation may be an important step in coke formation (Shaw et al, 1988). There are two points which could be criticized in the work of Gray and Masliyah: First they didn't use direct gravimetric analysis to measure the coke yield because they believed it was 32 Chapter 2-Literature Review unreliable. However, the use of carbon and sulfur analyses to determine coke yield prevented them from measuring the coke yield in the case of the asphaltene-coated kaolin. In this case, a direct gravimetric analysis could have been a useful method to measure the coke yield. Second, they claimed that the presence of asphaltene-coated kaolin prevented the formation of coke spheres. However, for longer reaction times (more than the 1 hour which was the case for this study), it is possible that asphaltene acts as a coke precursor. 2.6 Objectives So far there have been very few studies on the effect of naturally-occurring or added solids on coking of heavy oils. In the few studies done to date, the native fines were first separated by treating the oil samples with standard solvents, a procedure that has the potential to alter characteristics of both the sample and the recovered solids. These studies were primarily concerned with developing and growing the coke phase, rather than with the impact of additives and coking conditions on the coke yield. In order to better understand the role played by these fine solids in the coking process, a less artificial study should be carried out on the feed without changing its characteristics by separating the fines. Also, reactor conditions such as closed versus open vessel top, stirred versus non-stirred vessel and purge rate of the open reactor, etc., should be considered in parallel with the studies of the solids effect. Hence the objectives of the present investigation are: 1. To investigate the effect of time and temperature on the coking of a feed oil in a batch non-stirred reactor. Chapter 2-Literature Review 2. To carry out non-mixed batch coking tests with various solid additives and to compare the results in each case with those obtained using the unaltered feed (without additives). 3. To study the effects of time, temperature, purge rate and solid addition using a stirred reactor. 4. To compare the results obtained using a batch non-stirred reactor with those from a stirred reactor. 34 Chapter 3- Experimental Apparatus and Procedures 3. Experimental Apparatus and Procedures 3.1 Materials 3.1.1 Cold Lake Bitumen Cold Lake de-watered bitumen provided by Esso Resources Canada Ltd., was used as the feedstock in all experiments Some physical and chemical properties of this bitumen are listed in Table 3.1. Table 3.1 Typical physical and chemical properties of Cold Lake bitumen. (Data provided by Esso Resources Canada Ltd.) Physical state Liquid Specific gravity 0.991-1.006 Boiling point range (°C) 168-543 Freezing point (°C) 7-12 API gravity 9.1 Viscosity (cp) at: 15° C 96800 25° C 30200 60° C 1050 Ash (wt%) 0.08 Carbon (wt%) 82.57 Hydrogen (wt%) 10.73 Nitrogen (wt%) 0.54 Sulfur (wt%) 4.41 Oxygen (wt%) 2.42 Vanadium (ppm) 185 Nickel (ppm) 64 Asphaltenes (wt%) 15.4 35 Chapter 3- Experimental Apparatus and Procedures 3.1.2 Bitumen Characterization by SEM and EDX About 1.5 grams of sample were digested in 100 ml of toluene at 25 °C overnight. The suspension was then filtered through a 0.2 pjn Gelman Science TF-200 (Teflon) filter at room temperature with vacuum provided by a water aspirator. The residue on the filters was then dried for 16 hours at 90 °C and at atmospheric pressure. The solids left on the filter divided by the initial sample weight is termed the toluene insoluble fraction (TI), which accounted for about 0.4 wt% of the initial feed. Toluene insolubles measured by this method could include both organic material (coke), and inorganic impurities. Small segments of the dried filter cake were prepared for E D X (energy dispersive x-ray) and S E M (scanning electron microscopy) analysis by coating them with sputtered carbon. Scanning electron micrographs of the sample showed the presence of cubic crystals (Fig. 3.1). These were determined to be sodium chloride crystals by E D X analysis (Fig 3.2). Sodium chloride was present due to the entrapment of this salt from the upstream extraction of bitumen from the oil sands. 3.2 Additives In order to investigate the effect of particles in the coking experiments, six different additives were used. The additives included clays, silica and molybdenum sulfide. 3.2.1 Molybdenum Sulfide (MoS2) Previous studies on the hydrocracking of Athabasca bitumen with dispersed molybdenum based catalysts showed reduced coke formation at certain concentrations of molybdenum 36 Chapter 3- Experimental Apparatus and Procedures Chapter 3- Experimental Apparatus and Procedures Figure 3.2 Energy dispersive x-ray spectrum of the toluene insoluble fraction. 38 Chapter 3- Experimental Apparatus and Procedures (Sanford and Kennepohl, 1996). For this reason the possible effect of molybdenite (M0S2) on the coke yield in the absence of hydrogen was investigated. Molybdenite is composed of molybdenum ions sandwiched between layers of sulfur ions. The sulfur layers are strongly bonded to the molybdenum, but are not strongly bonded to other sulfur layers; hence the mineral is soft and exhibits perfect cleavage. Some properties of molybdenite are listed in Table 3.2. Table 3.2 Physical characteristics of molybdenum sulfide Color Silver metallic Transparency Crystals are opaque Crystal system Hexagonal Fracture Flaky Molecular wt. 160.08 g/mol Specific gravity 4.7 - 4.8 Melting point 2375 °C Sublimation point 450 °C The molybdenite powder provided by the Aldrich Co. (Milwaukee, WI, USA), had a broad size distribution. Particle diameters ranged from 0.2 - 30 pim according to S E M micrographs (Figs. 3.3.a and 3.3.b). The crystals were present as flakes with irregular edges. 39 Chapter 3- Experimental Apparatus and Procedures Chapter 3- Experimental Apparatus and Procedures Figure 3.3b Scanning electron micrograph of molybdenum sulfide particles Chapter 3- Experimental Apparatus and Procedures 3.2.2 Silica Sand (Si02) Silica was chosen as a solid which would be chemically inert in the coking process. However, it might interfere with the coking phase by acting as a nucleation site for the coke formation or providing more surface area for coking. Silica was provided by AGSCO Corp. (Wheeling, IL, USA). Two different size distributions were used, in order to investigate the possible effect of size in coking. The size distributions, physical properties and chemical analyses of the two sand samples are given in Tables 3.3, 3.4 and 3.5 respectively. Table 3.3 Size distribution of silica sand samples. Typical wet sieve analysis based on the percentage finer than the sieve opening. Size (um) Wt% of size less than Typel Type 2 53 100 100 44 99.99 100 37 99.92 100 30 99.10 100 20 92.00 100 15 86.30 100 10 65.40 100 5 19.84 79.00 3 11.64 21.00 1 3.45 13 Ave. Size 6.5 um 3.0 um Table 3.4 Physical and chemical properties of silica sand samples Specific gravity 2.65 pH in distilled water 6.0 - 6.3 Melting point 1671°C to 1871°C Grain shape Well rounded crystalline disc Bulk density 1281-1361 kg/m 3 42 Chapter 3- Experimental Apparatus and Procedures Table 3.5 Typical chemical analysis of silica sand samples Wt% Si0 2 99.49 Fe 2 0 3 0.039 A1 20 3 0.102 TiO z 0.015 CaO 0.014 MgO 0.021 Loss on 0.190 Ignition 3.2.3 Kaolin Kaolin, which is the most abundant clay mineral in the Athabasca oil sands, was provided by Plainsman Clay, Medicine Hat, Alberta. In previous bitumen coking studies, kaolin was selected because of its hydrophilic character (Gray, 1996). Prior to addition of these particles to the bitumen, they were screened on a 325 mesh sieve, to obtain particles of less than 45 pm in diameter. Scanning electron microscopy (Fig 3.4) shows the kaolin particles to have a random shape with a wide size distribution (between 0.5 and 45 p.m). An elemental analysis of the kaolin sample by A C M E Laboratory, Vancouver, B.C. (Table 3.6) shows that this clay contains about 45 % silica and 37 % alumina. 43 Chapter 3- Experimental Apparatus and Procedures Chapter 3- Experimental Apparatus and Procedures Table 3.6 Whole rock analysis by plasma spectroscopy (ICP). Kaolin Alberta clays Native clays Si02(wt%) 45.10 45.47 54.06 Al203(wt%) 37.12 28.40 20.91 Fe203(wt%) 0.67 7.74 3.57 MgO (wt%) 0.12 1.18 0.79 CaO (wt%) 0.49 0.65 0.31 Na 20 (wt%) 0.01 0.27 0.48 K 2 0 (wt%) 0.22 1.02 2.30 Ti02(wt%) 0.32 1.11 1.05 P2Os(wt%) 0.55 0.07 0.05 MnO (wt%) <0.01 0.04 0.07 Cr203(wt%) 0.17 0.032 0.027 Ba (ppm) 463 8134 368 Ni (ppm) 24 60 33 Sr (ppm) 258 164 122 Zr (ppm) 84 85 352 Y(ppm) 25 22 22 Nb (ppm) <10 <10 <10 Sc (ppm) 16 64 <10 Loss on Ignition % 15.2 13.0 16.2 C/TOT % 0.44 0.32 6.45 S/TOT % 0.12 0.06 0.50 Sum % 99.92 99.94 99.93 3.2.4 Southern Alberta Clays A Southern Alberta clay sample in the form of a paste was provided by Plainsman Clay, Medicine Hat, Alberta. In order to get rid of the coarse particles, about 20 grams of the paste were suspended in 500 ml of distilled water and left to settle for half an hour. Then the upper portion of the suspension which contained the fine particles was recovered and dried in air at 100 °C for 24 hours. The dried residue was crushed in a mortar and pestle and then screened through a 45 |im sieve. Chemical analysis of this clay (Table 3.6) shows that it is significantly different from kaolin in terms of its Fe 2 0 3 content, but it has Chapter 3- Experimental Apparatus and Procedures similar contents of the other elements. Its SEM micrograph (Fig. 3.5) showed irregular sharp edged plates, with a wide size distribution, ranging from 0.5 to 45 [im. Figure 3.5 Scanning electron microscope image of Southern Alberta clay particles. 46 Chapter 3- Experimental Apparatus and Procedures 3.2.5 Native Clays A sample of the middling stream obtained during the extraction of bitumen from Athabasca oil sand was provided by Syncrude Canada Ltd., Edmonton, Alberta1. This sample contains most of the fine solids which originally exist with the bitumen in the Athabasca reservoir, water, and traces of bitumen. In order to separate the fine solids from this suspension, it was first dried at 90°C for 24 hours, and then crushed and screened with a 45 p:m sieve. The chemical analysis of the fines fraction of this native clay is listed in Table 3.6. One significant difference between the native clay and the kaolin and other Southern Alberta clay is the carbon content, which is about 6.5% in the native clay compared to less than 0.5% in the latter clays. The S E M micrograph of this clay (Fig 3.6.a and 3.6.b) shows small plate-like crystals, which is more similar to the Southern Alberta clay (Fig 3.5) rather than the kaolin (Fig 3.4). 1 Definition of middling stream on page 10 47 Chapter 3- Experimental Apparatus and Procedures Figure 3.6a Scanning electron microscope image of native clay. 48 Chapter 3- Experimental Apparatus and Procedures Chapter 3- Experimental Apparatus and Procedures 3.2.6 Preparing the Mixture of Additives and Bitumen a) Tube Reactor A sample of 2 wt% of additive in bitumen was prepared prior to each experiment. Two 50-ml beakers were chosen, and about 5 grams of bitumen were placed in one beaker as the blank case, and 10 grams of it in another beaker for mixing with the additive. Then 0.2 gram of solids was added to the latter beaker. Both beakers were placed in a water bath at 80 °C and were stirred until their temperatures reached 60 °C, which took about 10 minutes. Then in order to make sure that good dispersion of solids was attained, the beakers were left in an ultrasonic bath for 5 minutes at 60 °C. For each experiment the coking result of the solid-added case was compared to that of the blank case which was prepared in parallel to the sample with the added solids. In order to check that the prepared suspension was homogeneous, two samples were taken from different parts of the beaker, and soaked in 100 ml of toluene for 16 hours. The suspensions were then filtered through 0.2 pirn TF-200 Gelman Sciences filters and dried. A comparison of the weight of residue on the two filters, which consisted of the toluene insolubles fraction of the bitumen as well as the added solids, indicated the uniformity of dispersion. These samples are listed as samples 4 and 5 in each run (Appendix B). b) Batch Stirred Reactor In this case, because the sample was stirred during the reaction, there was no pre-mixing done prior to the experiment. The required amount of solid to make 2 wt% mixture of solid in bitumen was placed in the reactor after half of the bitumen was added, and the solid was dispersed by stirring of reactor contents. 50 Chapter 3- Experimental Apparatus and Procedures 3.3 Experimental Apparatus 3.3.1 Tube Reactor Figure 3.7 shows a schematic diagram of the batch reactor system, which basically consists of a fluidized bed heating system, and reaction tubes containing glass liners. This system was designed and used by Dr. C. Yue in another research project. Figure 3.7 Tube reactor 51 Chapter 3- Experimental Apparatus and Procedures a) Fluidized Bed Temperature Bath and Pre-Heater The temperature bath, which provides rapid heating and good temperature control consists of two electrically heated sections: an air pre-heater and a fluidized bed as shown in Figure 3.7. The pre-heater is constructed of 96.5 cm long, 76 mm schedule 40, 316 stainless steel pipe having an inside diameter of 7.8 cm and a 0.3 cm wall thickness. Semi-cylindrical heaters (4.8 kW, 2-phase, 240 volts) and a 2.54 cm thick layer of ceramic fiber insulation are mounted on the pipe. The pre-heater is packed with 1.27 cm ceramic Raschig rings to retain heat and to increase the rate of heat transfer to the gas. Air passes through both the pre-heater and a loop of 0.99-cm ID stainless steel tubing to the conical section of the fluidized bed. The temperature near the exit of the pre-heater is measured by a 4.8 mm OD type K thermocouple. The 32 cm long fluidized-bed temperature bath is constructed of 316 stainless steel having an inside diameter of 14.2 cm and a 0.5 cm wall thickness. It is heated by 16 cm long semi-cylindrical electrical heaters (3.0 kW, 2-phase, 240 volts) and is insulated by a 2.54 cm thick ceramic fiber blanket. Silica sand particles having a 430 pm mean diameter are fluidized to achieve a uniform temperature distribution. The distributor of the fluidized bed is made of a perforated stainless steel plate with a voidage of 1.08 % and a perforation diameter of 1 mm. The preheated air is used as the fluidizing gas. The temperature of the fluidized bed is measured by two thermocouples, one placed 1 cm from the bed wall and 5 cm above the distributor, and another positioned at the center of the fluidized bed. The fixed bed depth was 22 cm. The thermocouples placed close to the fluidized bed wall and at the pre-heater exit are wired to two PID controllers, which independently monitor and control the temperatures. 52 Chapter 3- Experimental Apparatus and Procedures The power supplied to the heaters is dependent upon the difference between the set point temperatures and the actual temperatures read by the thermocouples, b) Tube Reactor, Glass Liner and N 2 Gas Purging The tubular reactor, shown in Figure 3.7, is constructed of two concentric cylindrical stainless steel tubes of OD 12.7 mm, and OD 6.4 mm, both 30.5 cm long connected by 2 three-way fittings. The outer tube is closed at the bottom. A Type K thermocouple of OD 1.6 mm is positioned at the centerline of the tubes and 2.5 cm from the bottom of the outer tube to monitor the reactant temperature. Nitrogen gas introduced at NPT from the upper union, passes through the center tube and then through the annular space between the two tubes before exiting through the lower union. A glass liner is used to contain the sample in order to prevent any catalytic effect of the stainless steel tube wall, and to permit recovery of the reaction residues. The glass tube has an OD of 10 mm and a wall thickness of 0.8 mm, and is 7.6 cm long. The nitrogen entrance tube was 3 cm from the top of the glass liner. The bottom of the tube reactor is equipped with a cap to prevent water leaking into the reactor during quenching which would otherwise cause the glass liner to break and contaminate the samples. The tube reactor was designed to be of sufficient length that sand is prevented from entering into the glass liner. 3.3.2 Experimental Procedures A clean glass liner was weighed and then about 1.5 g of the sample (bitumen or mixture of bitumen and added solids) was placed in the glass liner. The glass liner with its bitumen sample was then weighed again. The glass liner was purged with nitrogen at 200 53 Chapter 3- Experimental Apparatus and Procedures ml/min for 20 minutes before starting, and throughout each run. After the initial purge period was over, the tube reactor was plunged into the fluidized bed, which had already been preheated to the required temperature. When the reaction was carried out for the desired time period, the tube reactor was removed from the fluidized bed and quenched by plunging it in cold water (25 °C). Then after cooling down the tube reactor, the glass liner was removed and weighed accurately. The weight of the glass liner (at the beginning of the experiment) was subtracted from the weight of the glass liner with the bitumen sample in it, before and after the reaction, in order to obtain the initial and final weights of the bitumen sample. These weights were used in the calculation of the volatile weight loss (Chapter 4). Then the glass liner was placed in a sealed flask with 100 ml toluene without agitation for 16 hours. The flask was placed in a water bath to keep its contents at 25 °C overnight. The resulting suspension was filtered through a 0.2 pm Gelman Science TF-200 (Teflon) filter at room temperature using vacuum provided by a water aspirator, as shown in Figure 3.8. After filtering all the suspension in the flask, the glass liner and flask container were flushed with 50 ml toluene and put in the ultrasonic bath for 2 minutes, in order to remove any possible adhering insolubles from the wall of the glass liner and flask. This suspension was also filtered through the same filter. The wall of the glass funnel (Fig. 3.8) was also flushed with toluene and the residue on the filter was washed with toluene until the filtrate became clear. Then the filter and its residue were removed from the filtration system using a forceps into a watch glass, which were then dried for 16 hours at 90 °C in air at atmospheric pressure before weighing. Also after removing the filter from the filtration system, any possible residue stuck to the glass funnel was scraped 54 Chapter 3- Experimental Apparatus and Procedures and collected with the residue on the filter in the watch glass prior to drying. By using these precautions, all the solids precipitated by the toluene were collected on the filter. The solids left on the filter are termed the toluene insolubles (TI), or coke. Figure 3.8 Filtration system for the tube reactor experiments. 1) Funnel, 300 ml, ground glass seal 2) Glass base and tubulated cap 3) Anodized aluminum spring clamp 4) Ground joint flask, 1 liter 55 Chapter 3- Experimental Apparatus and Procedures 3.3.2 Batch Stirred Reactor Coking tests were also performed in 300 and 100 ml batch reactors (Autoclave Engineers, Erie, PA, USA). The reactor consisted of a 300 or 100 ml stainless steel vessel, a removable furnace, magnetic stirrer, cooling coil, safety valve, pressure gauge, speed controller, and temperature controller as shown in Fig 3.9. Gas outlet Reactor effluent outlet I Speed contro l ler Water outlet Pressure gauge •+ K> Water inlet Safety valve Thermocouple Temperature control ler Heating element Figure 3.9 Schematic diagram of batch stirred reactor 56 Chapter 3- Experimental Apparatus and Procedures The reactor has a safety head assembly which consists of the following: 1) The safety head body, which threads into the vessel cover. 2) The body gasket, which provides a seal at the cover connection. 3) The rupture disc. 4) The hold down ring, which conforms to the shape of the rupture disc and presses it against its seat. 5) The hold down gland, which exerts the force to seat the rupture disc. Figure 3.10 shows the internal components of the autoclave reator. The vessel's body can be removed from the flange for charging and cleaning. The internal volume of the reactor is interchangeable within a specific pressure range and for a specific vessel I.D. When a vessel size change is desired, the vessel body, internals and heater must be replaced. The heater assembly is a clamp-on, ceramic band-type heater with a quick release clamp and a thermocouple female adapter. The thermocouple assembly and electrical leads are six feet long and will connect directly to the modular temperature/speed controller. A single element type ' K ' thermocouple functions as a heater control thermocouple. A thermowell (0.3175 cm O.D. x 0.1576 cm I.D) provides a location for the internal process temperature measurement. A pencil type thermocouple sheath 0.1016 cm in diameter and 30.48 cm long was immersed in the thermowell. The Magne-Drive high-speed rotary agitation is effected by the rotation of external magnets, which actuate the internal magnets, fastened to the shaft. The external drive magnet assembly consists of an outer aluminum holder containing the stator magnet and bearings. This outer holder is placed over a pressure-sealed housing containing the encapsulated rotator magnets, which are mounted on a center rod. A strong magnetic field 57 Chapter 3- Experimental Apparatus and Procedures makes the inner center rod rotate at the same speed as the outer holder. Figure 3.10a View of the stirrer, thermowell and cooling coil of the autoclave reactor Coil Figure 3.10b View of the ceramic band-type heater, two sizes of reactor and flange which threads reactor to the safety head body. 58 Chapter 3- Experimental Apparatus and Procedures Figure 3.10 c View of the safety head body and Magne-Drive high-speed rotary agitator. The 'sample' and 'vent' valves stand for the gas inlet and outlet to the reactor, respectively. 59 Chapter 3- Experimental Apparatus and Procedures a) Experimental Procedure A sample (50 or 100 g) of the feed (bitumen or mixture of bitumen and added solids) was placed in the reactor vessel, which was secured to the body of the reactor via 6 hexagonal screws, and wrapped with a removable electric furnace. First the reactor was pressurized to 700 psig (47.6 x 10 2 kPa) hydrogen for a pressure test and then purged with UHP (ultra-high purity) nitrogen at the desired purge rate for 20 minutes before the reaction. Also the reactor was purged with nitrogen at the same rate during the reaction. The gas outlet valve was left open; therefore gas produced during the reaction mixed with the nitrogen purge stream and left the reactor to exit to the fume hood. Prior to heating the reactant mixture, it was important to cool the bearing of the magnetic stirrer using tap water to prevent hydrocarbon deposition between the reactor stirrer shaft and the bearing. Since solid hydrocarbon deposition was not completely prevented, the stirring assembly had to be removed and cleaned occasionally. The stirrer speed was set at 350 rpm and the reactor vessel was heated to the reaction temperature (390 °C) at a ramp rate of 5 °C/min. The heat-up time was approximately 1 h and 15 minutes, during which the reactor temperature was recorded. Then the final reaction temperature was maintained for different reaction times. After the reaction period, the heating element was removed and the reactor vessel was first air cooled to 160 °C over half an hour, and then cooled to room temperature by passing cold water through the cooling coil in the reactor vessel. The temperature was also recorded during the cooling time period. The reactor was then disassembled. The weights of the reactant and products were recorded. Solid carbonaceous deposits that adhered to the reactor wall and the impeller were removed by scraping followed by washing with toluene. The amounts of coke and 60 Chapter 3- Experimental Apparatus and Procedures liquid product recovered were weighed, so that an overall mass balance calculation could be completed. Products from the coking experiment consisted of solid coke, liquid residue, and gas. The liquid product was mixed with 500 ml of toluene and then placed in a sealed flask without agitation for 16 hours. Flasks were placed in a water bath to keep them at 25 °C overnight. In the experiments using 100 grams of sample, the liquid product was filtered with the aid of a pressure filtration system (Section 3.5) and a series of filters down to 0.2 \im pore-size. But in the case of the 50 gram sample reactions, filtration of the liquid product was done by vacuum from a water aspirator using filters of 3, 0.45, and 0.2 p:m pore diameters. The filtrations were carried out in sequence starting with the 3 \im filter, then passing the filtrate through the 0.45 ujn filter and so on. Washings from the impeller and vessel product were filtered separately. Coke yield, expressed as a weight percent of the feed bitumen, was based on the solid coke recovered from both the reactor and the reactor washings. 3.3.3 Pressure Filtration In order to improve the rate of vacuum filtration relative to that attainable using the water aspirator, a new system was designed. For the toluene-treated liquid product from the 100 g autoclave reactor experiments, pressure filtration with a larger surface area filter (90 mm diameter compared to 47 mm for the water aspirator system) was employed to decrease the filtration time. The pressure filtration system includes a cylindrical sample container (pressure cylinder), a pressure gage, the 90 mm diameter stainless steel filter holder, and a nitrogen gas source to pressurize the cylinder (Fig 3.11). 61 Chapter 3- Experimental Apparatus and Procedures Figure 3.11 Pressure filtration system 62 Chapter 3- Experimental Apparatus and Procedures 3.3.4 Pressure Cylinder A stainless steel cylinder (Fig. 3.12) with an internal volume of 1 liter, approximate weight of 14 lb. (6.342 kg) and wall thickness of 0.206 inch (5.232 mm) was obtained from the Swaglok Co. (Vancouver, B.C.). The cylinder (model SS-3A 1800) is capable of withstanding pressures up to 120 bars (1800 psig). The upper part of the cylinder was connected to a pressure gauge (range 0-300 psi) by means of a male tee connector with a reducing bushing of ( lA NPT to Vz NPT). The third outlet of the tee connector was connected to a nitrogen cylinder, through which the gas flow could be controlled by a needle valve (Swagelok, SS- 3VS4). I Figure 3.12 Pressure cylinder A = 3.5 inch (8.89 cm) O.D. B = 10.88 inch (27.64 cm) T (minimum wall thickness) = 0.18 inches (4.572 mm) P (female pipe) = Vi inch (1.27 cm) 63 Chapter 3- Experimental Apparatus and Procedures 3.3.5 Filter Holder The stainless steel 90 mm diameter filter holder (Fig. 3.13) is connected to the bottom of the pressure cylinder by a male connector. The components of the filter holder are listed in Fig. 3.13. Figure 3.13 Stainless steel filter holder 1. Connector to 3/8 I D hose 2. Vent valve 3. Allen-head cap screws 4. Inlet plate 5. O-ring 6. 90 mm diameter filter 7. Filter support screen 8. Underdrain support 9. Outlet plate 10. Allen set screws 11. Legs with caps 64 Chapter 3- Experimental apparatus and procedures a) Loading For filtering the suspension of the liquid product in toluene, M F filters (Millipore membrane filter, made from mixed cellulose esters) with various pore sizes (8, 5, 3, 1.2, 0.45, and 0.2 pjn) were used. Filtration started with the largest pore size filter and then filtrate from each step was collected in a flask and then passed through the next smaller filter down to 0.2 p:m. The desired filter was centered on the support screen. Then the three screw holes in the inlet plate were aligned with the corresponding threaded holes in the outlet plate and the inlet plate was lowered to the outlet, so that the sealing O-ring rests on the filter. The cap screws were inserted through clearance holes and then hand-tightened with the Allen wrench supplied. Then the filter was attached to the cylinder, which contains the unfiltered fluid. A l l the connections were checked for leaks and after that the suspension was poured into the vessel. A flask was placed at the outlet of the filter holder to collect the liquid. The top part of the cylinder was then attached to the pressure gauge, which was connected through a needle valve with suitable tubing to the nitrogen cylinder regulator. The needle valve was closed, and so was the nitrogen cylinder regulator valve. Again all the connections were examined for leaks. First the regulator valve was opened, then the needle valve opened slowly to the nitrogen cylinder regulator, which was adjusted to the pressure of 50 psi (3.45 bar) b) Disassembling and Cleaning the Unit After making sure that all the liquid passed through the filter, first the nitrogen cylinder valve was closed. Then the needle valve was closed and then the tubing between the N2 source and the needle valve was disconnected. The needle valve was then slowly opened 65 Chapter 3- Experimental apparatus and procedures to the atmosphere to release any pressure in the vessel. Then the pressure gage and vessel were disconnected from the filter holder and, after opening the filter holder, the filter and the residue were carefully collected for weighing. 66 Chapter 4 -Results and Discussion 4. Results and Discussion 4.1 Tube Reactor 4.1.1 Coking in the Absence of Added Solids The objective of these sets of experiments was to establish a relationship between the two dependent variables, toluene insolubles (coke) yield and weight loss, and the two independent variables, temperature and time, in the absence of added solids. The reaction time was increased from 0.5 to 5.5 hours, by half hour intervals. Also, at each time, the coking reaction was investigated at three different temperatures, 380 °C, 390 °C, and 400 °C with a 200 ml/min nitrogen purge. Each run was repeated three times using three reactors, each containing a glass tube, which held the heavy oil sample. The three reactors were inserted into the fluidized bed at intervals, in order to allow sufficient time for each reactor to reach the bed temperature before the next one was introduced. The primary dependent variables of this study, i.e., weight loss and toluene insolubles, were defined as follows. Weight loss, which is a measure of volatiles that leave the reactor during the reaction, was calculated as: The toluene insolubles is the weight of residue collected on the filter after dissolution of the reacted sample in excess toluene for 17 hours. The percentage toluene insolubles is given by: % Wtloss = weight of the sample before reaction - weight of the sample after reaction x 100 (4-1) weight of the sample before reaction % TI = Weight of toluene insolubles x l 0 0 (4.2) Weight of the initial sample 67 Chapter 4 -Results and Discussion Some experiments at different reaction times were repeated to investigate the reproducibility of the results for the three different reaction temperatures (Appendix A). The results for all experiments are listed in Table 4.1. The % toluene insolubles and % weight loss are plotted as functions of coking time, in Figures 4.1 and 4.2, respectively. The error bars shown on these figures are the standard deviations of the three measurements made for each reaction condition. Tables 4.1(a) and Figure 4.1 indicate that the initial feed contains about 0.4% TI, a normal occurring component of heavy oil likely consisting of inorganic and carbonaceous solids. 14 , Time (h) Figure 4.1 Toluene insoluble vs. time in the absence of added solids. 68 Chapter 4 -Results and Discussion Table 4.1 Batch coking tests with no solids added (a) 380 °C Time (h) %Wt loss Standard Deviation %TI Standard Deviation 1 31.0 0.28 0.42 0.035 1.5 36.6 1.21 0.42 0.053 2 46.9 1.92 0.44 0.065 2.5 47.4 0 0.45 0 3 49.7 1.85 1.16 0.370 3.5 50.8 0.99 1.70 0.276 4 54.6 0.88 2.07 0.907 4.5 55.8 0.36 1.85 0.212 5 57.7 0.78 2.70 0 (b) 390 °C Time (h) %Wt loss Standard Deviation % TI Standard Deviation 0.5 27.1 0 0.40 0 1 34.6 1.13 0.48 0.042 1.5 41.7 1.21 0.47 0.050 2 46.3 0.57 0.61 0.072 2.5 53.7 2.13 2.54 0.166 3 58.2 1.59 3.43 0.295 3.5 59.5 0.71 4.47 0.240 4 61.8 0.61 5.60 0.169 4.5 63.1 0.88 6.38 0.213 5 65.2 0.66 7.12 0.277 (c) 400 °C Time (h) %Wt loss Standard Deviation %TI Standard Deviation 0.5 37.0 0 0.44 0 1 46.6 0 0.52 0 1.5 52.4 3.26 2.03 1.039 2 58.6 3.96 4.35 0.544 2.5 59.9 0.33 5.18 0.338 3 65.2 0.54 7.53 0.167 3.5 67.6 0.81 8.57 0.185 4 69.9 0.29 9.96 0.552 4.5 72.0 0.60 11.19 0.186 5 74.2 3.05 11.06 0.494 Chapter 4 -Results and Discussion At 380 °C the toluene insolubles fraction remains constant until about 2.5 hours of reaction time and then increases, approximately linearly with time, over the next 3.5 hours. Similar induction times for toluene insolubles production have been seen in other coking experiments with Cold Lake bitumen (Wiehe, 1993). As the coking temperature is increased to 390 °C and 400 °C, the induction times decreased to about 2 hours and 1 hour respectively. Also, at each time interval, the toluene insolubles increased as the temperature increased, reaching a maximum of about 11% for the highest temperature and longest reaction time. Figure 4.2 shows that the % weight loss continues to increase with reaction time, apparently approaching an asymptotic value after about 5 hours. At any given time, the % weight loss increases as the reactor temperature is raised. w (0 o D) £ 90 80 70 60 50 40 30 20 10 0 m i i . • * B • • * • * 4 Temperature 380 ° C ^Temperature 390 ° C « Temperature 400 0 C I I Time (h) Figure 4.2 % Weight loss vs. time for tube reactor tests in the absence of added solids. 70 Chapter 4 -Results and Discussion Based on the work of Srinivasan and McKnight (1993), who investigated a similar reaction system, the above experimental results can be explained as follows. As the feed is heated, the volatiles in the feed can be expected to be released. When bitumen is cracked, light species are formed, and some of the organic compounds present in the feed or produced due to cracking may agglomerate or chemically react to form heavier species. In an open system where lighter products are purged, the reaction matrix becomes more viscous and concentrated in the heavier and more aromatic compounds, which can form TI. As the heavier species grow in size, a second liquid or solid phase may separate from the reaction matrix. The slow step in the process could be the physical mechanism of phase separation. The growth of the heavier species is also promoted by longer reaction times resulting in higher yield. The toluene insolubles, which approximates the coke yield, is plotted in Fig. 4.3 as a function of weight loss for each time and temperature. As can be seen from the figure, when the data are plotted in this way, regardless of time and temperature, all the points lie more or less on the same curve. This somewhat surprising behavior can be explained as follows. At low value of weight loss, no coking takes place, and %TI remains at its background level of about 0.4%. The threshold of coking is reached at a certain weight loss (about 47%), after which the coke yield increases linearly as a function of weight loss. This result presumably occurs because the concentration of heavy aromatics must reach a certain level by the loss of volatiles, in order to form a second phase. After reaching this level, the coke phase separates from the rest of the reaction matrix and continues to increase by further weight loss and growth of the heavier species. 71 Chapter 4 -Results and Discussion 40 % W t Loss Figure 4.3 %Toluene insoluble vs. %weight loss for all tube reactor results in the absence of added solids 4.1.2 Effect of Added MoS2 Particles Finely dispersed metal sulfides, formed from a variety of precursors, are commonly used to suppress coke formation during the hydrocracking of petroleum residues (Sanford and Kennepohl, 1996). The pioneering work of Bearden and Aldridge (1981) showed that coke formation during the cracking of Cold Lake crude decreased as the molybdenum concentration was increased up to approximately 500-800 ppm in the feed, under a relatively high reactor hydrogen pressure of 17.3 MPa. Although the experimental conditions are very different in this work, an objective of this study was to investigate the possible effect of fine MoS 2 particles on coking at atmospheric pressure in a nitrogen purge, rather than in a hydrogen environment. 72 Chapter 4 -Results and Discussion The experiments were carried out using three reactors concurrently. Two of them contained the same mixtures of bitumen and 2 wt% MoS 2 , while the third contained the blank feed which underwent the same thermal treatment as the solids added cases during the preparation procedure (see Section 3.2.6). The coke yields of the solids-added samples in each set of experiments were compared to that of the blank case in that set, in the event that there was any possible alteration of samples during preparation. The conditions selected for these experiments were a temperature of 390 °C, a nitrogen purge rate of 200 ml/min and three coking times of 2.5, 3.5 and 4.5 hours. Also two additional samples were collected during the preparation procedure in order to investigate the dispersion of the solids (see Section 3.2.6). The experiments were repeated 3-4 times at each reaction time, to check the reproducibility of the results. In the solids-added cases, the % toluene insolubles was corrected for the initial solids added to the sample, assuming that they were completely removed by filter, i.e., <y JI - Weight of toluene insolubles - Weight of added solids x JQQ ^ Weight of the initial sample - Weight of added solids Appendix B tabulates all of the results obtained for the three different coking times and also includes sample calculations showing how the %TI and %weight loss values as well as their 95% confidence intervals were determined. Figures 4.4 and 4.5 show, respectively, the average coke yield and weight loss at each reaction time. 73 Chapter 4 -Results and Discussion 0 1 2 3 4 5 6 Time( h) Figure 4.4 Toluene insolubles vs. time (solid line corresponds with the blank data at identical conditions, i.e., Fig. 4.1). Experiments at 390 °C and 200 ml/min nitrogen purge. *Error bars are the 95 % confidence intervals of the average value based on the number of experiments. A statistical comparison, using the 't' test (Mendenhall, and Sincich, 1992), of the average value %TI from the two samples with MoS 2 added with the single sample without MoS 2 is carried out in Appendix B . This comparison confirmed that, at the 95% confidence level, there was no difference between two cases. Based on the previous studies by Sanford and Kennepohl (1996), and many others, it was expected that MoS 2 would reduce the coke yield. Therefore it was necessary to investigate thoroughly the reliability of the results. Since MoS 2 adheres to the most of the glassware during the filtration operation, 100 % recovery of MoS 2 on the filter was the first concern. Less than complete recovery would introduce an error in the numerator of 74 Chapter 4 -Results and Discussion Eq. 4.3. In order to check that the assumption of complete recovery was reasonable, five different filtered solids and filtrate samples were assayed for MoS 2 , two samples from the 2.5 hour coking experiments, two from the 4.5 hour coking runs and one from a mixture of bitumen and MoS 2 which was not coked. The results of the assays, carried out by the CANTEST Laboratory (Vancouver, B.C.), using plasma spectroscopy (ICP), confirmed that less than 0.2 ppm MoS 2 exists in the filtrate, which was negligible. The results for the filtered solids, in terms of the % recovery of MoS 2 in the filter coke, are listed in Table 4.2. Table.4.2 Analysis for molybdenum sulfide in the filtered solids from the tube reactor experiments at 390 °C where 2% by weight M0S2 was added initially to the sample. Sample wt. (g) Coking time (h) % Recovery % T I based on %100 Recovery % T I based on actual recovery* 2.1387 4.5 91 5.636 5.819 1.618 2.5 62.4 1.887 2.648 1.9811 4.5 76.4 6.562 7.046 1.1861 2.5 92 2.503 2.668 1.653 0 83 0.615 0.967 * Detailed calculation and results of the MoS 2 assay are included in Appendix B in Table B4. Poor recovery of MoS 2 in some cases could be due to its sticking and eventual loss to the filtration glass funnel. Even though the percent recovery was low in some cases, it did not affect the results significantly. This is due to small amount of MoS 2 , which was initially added to the feed samples (about 0.2 g). 75 Chapter 4 -Results and Discussion Although Figure 4.5 suggests an effect of MoS 2 addition on weight loss, a statistical analysis of the data (Appendix B) determined that there was also no significant difference in the results obtained for the added solid and blank cases at any of the three reaction times investigated (Appendix B). in (A o 70 60 50 J. 40 J. 30 J. 20 » No Solid added • MoS2added I + + 3 4 Time (h) Figure 4.5 Weight loss vs. time (solid line corresponds with the blank data at identical conditions, i.e., Fig. 4.2). Experiments at 390 °C and 200 ml/min nitrogen purge. *Error bars are the 95 % confidence intervals of the average value based on the number of experiments. 4.1.3 Effect of Added Silica Particles The objective of these experiments was to study the effect of chemically inert silica sand particles on the coking yield and to investigate i f the size of the particles has any impact on coking. The experimental conditions selected were a temperature of 390 °C and a reaction time of 4.5 hours. Two batches of silica sand, having average sizes of 6.5 and 76 Chapter 4 -Results and Discussion 3.0 um, were used in the coking experiments (see Section 3.2.2). The preparation procedure for dispersing 2 wt% of solids in the bitumen samples was the same as with MoS 2 (Section 3.2.6). Again a 'blank' sample (without S i 0 2 particles) was processed along with a pair of solids-added samples. The weight losses and toluene insolubles were calculated, respectively, from Eqs. 4.1 and 4.2, as in the case of the experiments with MoS 2 particles. 8 0 1 2 3 4 5 6 Time( h) Figure 4.6 Toluene insolubles vs. time (solid line corresponds with the blank data at identical conditions, i.e., Fig. 4.1). Experiments at 390 °C and 200 ml/min nitrogen purge. *Error bars are the 95 % confidence intervals based on the number of experiments. 77 Chapter 4 -Results and Discussion Each experiment was repeated 4-5 times to check the reproducibility of the results. The average %TI and % weight loss values are shown on Figures 4.6 and 4.7, respectively. A l l of the raw data from these experiments are tabulated in Appendix B. A statistical analysis (t test) of the results shows that, within a 95% confidence interval, adding 3 um silica particles to the heavy oil samples has no significant effect on the yield of toluene insolubles (Appendix B). But for the 6.5 u,m silica sand experiments, a statistical analysis shows that the addition of solids significantly reduces the coke yield. This effect is likely a physical one (because silica particles are chemically inert), and it may be due to the interference of these solids with the developing viscous phase during coke formation. Thus, the solids may collect at the interface between the two liquid phases thereby preventing the agglomeration and growth of the heavy nucleus aromatics which otherwise form coke, as has been suggested by Gray (1997). For the same wt% suspensions, the 3.0 u,m silica particles actually offer more than twice the surface area as the 6.5 um particles. Also, the 3.0 pm particles yield a lower average %TI than the 6.5 um particles. The effect of the former, however, was not as statistically significant as that of the latter, simply because the standard deviation of the former was so much larger. The average weight loss results measured for these cases are plotted in Fig. 4.7. A similar statistical analysis showed that the addition of silica particles of either size has no significant effect on the weight loss under the conditions of these experiments (Appendix B). Figure 4.8 provides a summary comparison of the coke yield and weight loss results obtained in the presence and absence of added solids. The solid line represents the results shown in Fig. 4.3, where the experiments involved no addition of solids. Note that the 78 Figure 4.7 Weight loss vs. time (solid line corresponds with the blank data at identical conditions, i.e., Fig. 4.2). Experiments at 390 °C and 200 ml/min nitrogen purge. *Error bars are the 95 % confidence intervals based on the number of experiments. 8 0 I i i i i i I 0 10 20 30 40 50 60 70 % W t Loss Figure 4.8 Toluene insolubles vs. weight loss (solid line is a curve fit of the blank experiments in Fig. 4.3). Points stand for different additives. 79 Chapter 4 -Results and Discussion results obtained by adding silica particles to the heavy oil lie virtually on the same curve as the non-addition experiments, but those obtained in the presence of MoS 2 particles show significant scatter but are largely grouped to the left of the line for no solids addition. 4.2 Autoclave Reactor In the second stage of this study, coking experiments were carried out in the stirred autoclave reactor. Mixing the sample during the reaction has the advantages of keeping the additives well dispersed in the bitumen as well as generally maintaining homogeneous conditions. However, the reactor environment is very different from that of the tube reactor. In the autoclave reactor, for example, it was found that the coke yield and weight loss results were influenced by the amount of the sample, the N 2 purge rate and the presence or absence of agitation. Thus a secondary objective of this part of the work was to investigate how each of these factors affected the results, in the absence of any solid additives. The autoclave experiments were carried out using two different feed sample sizes, namely 50 and 100 grams. For the 100 g samples, blank, MoS 2 , and silica sand cases were studied at 390 °C. For the 50 g samples, not only were the effects of MoS 2 and silica additives studied at this temperature, but also the influences of adding southern Alberta clays, kaolin and native clays. In addition, the effects of additive concentration and N 2 purge rate were investigated in a few cases. The experimental procedures followed in this work were similar to those used by Strausz, et al., (1988). Coke yield and weight loss were calculated slightly differently than for the 80 Chapter 4 -Results and Discussion tube reactor experiments. Liquid products and washings from the impeller were digested in toluene and then filtered separately. For cleaning and digesting, 500 ml of toluene for the liquid product and 100 ml for the impeller were used. Therefore, the toluene insolubles in each portion was recovered and weighed separately, and the total toluene insolubles calculated as: o/ 0 TI = TI in vessel+ TI in washings x m ( 4 > 4 ) Weight of the feed After the reaction, it was generally found that a small portion of the product adhered to the impeller. Therefore the weight loss could not be calculated directly from the product drained from the reactor. If it is assumed that only toluene insoluble material adheres to the impeller, then the weight loss can be determined as follows: % Weight loss = 100 - % Recovery (4.5) where: „, _ Weight of the product + Toluene insolubles in washings . . . , . % Recovery = — x 1 0 0 (4.6) Weight of the feed 4.2.1100 Gram Autoclave Samples a) Effect of Coking Time (no solids added) In this set of experiments, the objective was to determine the behaviors of the coke yield and weight loss as functions of time. The temperature and nitrogen purge rate were chosen to be 390 °C and 200 ml/min, respectively, for all of the experiments. Three different coking times, namely 2.5, 3.5, and 4.5 hours, were used. Figure 4.9 shows the 81 Chapter 4 -Results and Discussion warming up and cooling down paths for the autoclave reactor prior to and after reaction. Because the amount of the produced coke is fairly large compared to that obtained in the tube reactor, pressure filtration was used to recover the toluene insolubles fraction. W a r m u p a n d c o o l d o w n c u r v e s 450 -. 0 20 40 60 80 T i m e (min) Figure 4.9 Warm up and cool down paths for autoclave reactor containing 100 g feed sample. The coke yield and weight loss data obtained in these experiments are summarized in Table 4.3. Sample calculations as well as a listing of the raw data are given in Appendix C. Table 4.3 Toluene insolubles and weight loss for different times. Coking of 100 g sample at 390 °C with a 200 ml/min nitrogen purge. No sol id added Time (h) %TI % Wt. loss 2.5 0.35 30.15 3.5 0.36 33.62 4.5 5.23 39.98 4.5 5.1 38.81 82 Chapter 4 -Results and Discussion As is obvious from the table, the induction time prior to significant coke yield seems to be between 3.5 and 4.5 hours. For the same conditions, the induction time was only 2.0 hours in the tube reactor. I I 1 1 1 1 r 0 1 2 3 4 5 6 Time (h) (a) 7 0 6 0 . 5 0 - Tube reactor • (0 o 4 0 -% 3 0 - tf' • * s~ 2 0 1 0 Autoclave 100 g 0 I l I I I r 0 1 2 3 4 5 Time (h) (b) Figure 4.10 Toluene insoluble and weight loss vs. reaction time, for Experiments at 390 °C and 200 ml/min nitrogen purge. Comparison between tube reactor and 100 g autoclave reactor. 83 Chapter 4 -Results and Discussion The weight loss range is also higher for the tube reactor (53-63 %) compared to 30-39 % for the 100 g autoclave. Figure 4.10(a) and 4.10(b) show, respectively, the coke yields and weight losses as functions of time for both the tube and autoclave reactors. These differences in the weight loss, induction times and coke yield between tube and autoclave reactors could be because of the different geometry, initial sample weight and degree of homogeneity of the two reactors, but are undoubtedly at least partially due to relatively long warm up time (>1 h) of the autoclave reactor. The purge gas velocity rate over the liquid surface is also markedly different, b) Effect of Different Additives In order to investigate the possible effect of solid additives in the stirred reactor, experiments were undertaken with 2 wt% of MoS 2 , 3 um S i 0 2 or 6.5 um S i 0 2 added to the 100 g of bitumen feed. The solids were not pre-dispersed in these experiments, but simply added to the autoclave reactor with the bitumen sample. The reactor was stirred at 350 rpm. Each experiment was compared with the blank case (no solid added) at the same reaction times (2.5 and 3.5 h), temperature (390 °C) and N 2 purge rate (200 ml/min). The toluene insolubles and weight loss results are shown in Table 4.4. As was the case for the tube reactor, the %TI was corrected for the addition of solids, i.e., „, ^  TI m vessel + TI in washing - Weight of the solids added . _ n %TI = - — x 100 (4.7) Weight of the feed - Weight of the solids added 84 Chapter 4 -Results and Discussion This calculation assumes the 100% recovery of solids either in the liquid product or in the washing solvent. A sample calculation and the detailed results are provided in Appendix c Table 4.4 Toluene insoluble and weight loss results for 100 gram bitumen samples with additives. Reaction at 390 °C and 200 ml/min nitrogen purge. Time (h) % Solid added %TI % Wt. loss 2.5 M0S2,1.84 0.38 28.69 2.5 MoS2,1.85 0.38 28.72 2.5 3 um Si02,1.81 0.41 28.30 2.5 6.5 iim Si02, 2.01 0.34 29.29 3.5 M0S2,1.84 0.61 33.37 3.5 3 Lim Si02, 2.00 0.47 33.08 j 1 _ > - N o sol id a d d e d • M o S 2 4 S i l i ca (3 m ic rome te r ) % S i l i ca (6.5 m i c r o m e t e r ) / / / / / / / / / / / / / / / / / / / / / / / rr It / / I f Time (h) Figure 4.11 Toluene insolubles vs. time with different additives in 100 g autoclave samples. Experiments at 390 °C and 200 ml/min nitrogen purge. 85 Chapter 4 -Results and Discussion For this set of experiments, which are still in the induction period, there was no discernable differences in either the coke yield or the weight loss between the solids-added and blank cases (compare Tables 4.3 and 4.4). Figure 4.11 shows the coke yield vs. time results obtained using different additives. It is obvious from the figure that, during the induction period, i.e., at 3.5 h or less, there were no significant differences between the solids-added and blank cases. Comparison of the tube reactor and 100 g autoclave results as a graph of % toluene insolubles vs. weight loss is shown in Figure 4.12. (A _ si _ o (A _ C o 3 10 20 30 40 50 60 70 80 % Wt Los  Figure 4.12 Comparison of the toluene insolubles vs. weight loss for 1.5 g tube and 100 g autoclave experiments at 390 °C and 200 ml/min nitrogen purge. For each reactor size, the presence of silica and MoS 2 has no effect on TI vs. Wt loss curves. 8 6 Chapter 4 -Results and Discussion Filtration of the suspensions produced for all of these runs was carried out with the aid of the pressure filtration system. The filter holder was rated for up to 85 psi differential pressure between the inlet and outlet. In a couple of the experiments, while investigating the effect of added solids for 4.5 hours coking at 390 °C, the pressure difference in the filter holder exceeded 200 psi for the 3 um pore size membrane. Therefore, it was not possible to carry out a filtration for the 0.2 um pore size filter. Even for the 3 urn filter, it proved to be impossible to finish the filtration so that the results for the solids-added and the blank cases could not be compared at 4.5 hours of coking (which, according to Table 4.3, is beyond the induction time). Thus, to make it feasible to obtain coking results past the induction time (> 3.5 h), the autoclave sample size was reduced to 50 g in order to lower the amount of solids requiring filtration. 4.2.2 50 Gram Autoclave Samples a) Effect of Coking Time Coking experiments were carried out on 50 g samples at 390 °C and 200 ml/min nitrogen purge for different periods of time (2.5, 3.5 and 4.5 hours). The objective of this set of experiments again was to develop relationships for the coke yield and weight loss as functions of time, which could then be compared to the results of the tube and 100 gram autoclave reactor experiments. Following the same procedures as were used for 100 g autoclave reactor tests, it was found that pressure filtration was successful for the products from the 2.5 and 4.5 h tests, but was unsuccessful for the 3.5 h coking reaction because of excessive pressure buildup. As an alternative, for the 3.5 h coking test, 87 Chapter 4 -Results and Discussion filtration was attempted with a standard 47 mm filter connected to a water aspirator using a series of filters (3, 0.45 and 0.2 pm pore sizes); but this was not successful. It proved impossible to complete the filtration with the 0.2 um pore size membrane. Therefore, only experiments for 2.5 and 4.5 hours of coking were finished and further studied with the addition of solids. Table 4.5a summarizes the coke yield and weight loss results obtained in the absence of added solids. Appendix D provides detailed results including the weights of toluene insolubles collected on the different filter sizes for each experiment. Table 4.5a Toluene insolubles and weight loss for different times. Coking of 50 g samples at 390 °C and 200 ml/min nitrogen purge. No solid added Coking time (h) % Toluene insolubles % Wt. loss 2.5 0.43 40.2 2.5 0.34 46.11 4.5 3.83 47.6 4.5 4.22 45.04 4.5 4.21 53.29 4.5 3.96 46.02 Based on the qualitative results from the 3.5 h experiment and the quantitative results shown in Table 4.5a, the induction time seems to be between 2.5 and 3.5 hours. By comparison, Table 4.3 and Figure 4.11 show that, for the 100 gram sample, the induction time lies between 3.5 and 4.5 hour. Comparison of the results of all three reactors (1.5 g tube reactor, 50 g autoclave and 100 g autoclave) suggests that, at the same purge rate and 88 Chapter 4 -Results and Discussion temperature, even though there are variations in reactor geometry, the induction time increases with the sample size, b) Effect of MoS2 on Coking This set of experiments was carried out in order to investigate the possible effect of adding MoS 2 particles on the coking reaction. The autoclave conditions were set at 390 °C and 200 ml/min N 2 purge rate, and the reaction times were 2.5 and 4.5 h after the operating temperature had been reached. It was only at these two times that blank (i.e., no solids added) results were available for comparison. Mixtures of 2 wt% MoS 2 in bitumen were prepared for all of these coking experiments. The results are summarized in Table 4.5b. Table 4.5b Toluene insolubles and weight loss for different times. Coking of 50 g samples at 390 °C and 200 ml/min nitrogen purge. 2% MoS 2 added Coking time (h) %Toluene Insolubles %Wt. Loss 2.5 0.413 54.2 2.5 0.4 46.29 4.5 4.56 48.79 4.5 4.08 45.08 Appendix D gives the detailed results including the weight of residue on each filter (different pore sizes were used). Comparison between Tables 4.5b and 4.5a shows that, for the 2.5 h coking experiments, there was no discernable difference in the coke yield results, but, for the 4.5 h runs, the average coke yield values appeared to be different. However a't ' test analysis reported in Appendix E, shows that, within a 95% confidence 89 Chapter 4 -Results and Discussion interval, the 4.5 h blank and MoS2-added %TI and %wt loss results are not statistically different. c) Effect of Silica on Coking The addition of chemically inert silica particles, with an average size of 3 um, was studied in the 50 g autoclave reactor under the standard operating conditions and for 2.5 and 4.5 h of coking. The weight loss and coke yield results are summarized in Table 4.5c. The same conclusion as in the MoS 2 case could be drawn here. For the 2.5 h experiments, there was no significant difference between the solids-added and blank results (Tables 4.5a and 4.5c), and, for the 4.5 h runs, a ' t ' test analysis, presented in appendix E, shows that they are not statistically different within a 95% confidence interval. Appendix E also includes other details about the results. Table 4.5c Toluene insolubles and weight loss for different times. Coking of a 50 g sample at 390 °C and 200 ml/min nitrogen purge. 3 L i m silica added Coking time (h) % Toluene insolubles % Weight loss 2.5 0.32 47.53 2.5 0.34 47.07 4.5 4.93 53.24 4.5 4.62 51.59 Figure 4.13 illustrates the coke yield results for the blank and solids-added experiments in the 50 g autoclave reactor. There appears to be a higher %TI for the silica at 4.5 h than for the blank. Table 4.6 summarizes the relative ease of filtering the products obtained from all the 100 g and 50 g autoclave experiments. As is obvious from this table, in some 90 Chapter 4 -Results and Discussion cases filtration was very difficult and, in fact impossible to finish, whereas in other cases it was fairly easy and successful. Therefore, based on its relative ease of filtering and the fact that it produces a significant amount of coke (approximately 4%), a 4.5 hour coking time was chosen for all subsequent studies. 6 5 4 1 0 • Silic • MoJ :a 3 microt 52 added neter . . • - - No Solid added 4 II WW Time (h) Figure 4.13 Toluene insolubles vs. time with different additives in 50 g sample. Experiments at 390°C and 200 ml/min nitrogen purge. 91 Chapter 4 -Results and Discussion Table 4.6 Comparison between filtration systems for different reactors and solid additives. Experiments at 390 °C and 200 ml/min purge. Time (h) Sample(g) Filtration Pore size used (|im) Plugged filter Filtration 2.5 100, Blank Aspirator 0.45, 0.2 - Easy 2.5 100,2% MoS2 Aspirator 0.45,0.2 - Easy 2.5 100,2% Si0 2,3 um Aspirator 0.45, 0.2 - Easy 2.5 100,2% Si0 2, 6.5 um Aspirator 0.45,0.2 - Easy 3.5 100, Blank Pressurized 0.22 - Easy 3.5 100,2% MoS2 Pressurized 8,1.2,0.22 - Easy 3.5 100,2% Si0 2,3 iim Pressurized 8,1.2, 0.22 - Easy 4.5 100, Blank Pressurized 8,1.2, 0.22 - Hard 4.5 100,2% MoS2 Pressurized 8,1.2 1.2 Hard 4.5 100,2% MoS2 Pressurized 8,5,3,1.2, 0.65, 0.22 0.22 Hard 2.5 50, Blank Aspirator 0.45, 0.2 - Easy 2.5 50, 2% MoS2 Aspirator 0.45, 0.2 Easy 2.5 50,2% Si0 2,3 um Aspirator 0.45,0.2 - Easy 3.5 50, Blank Pressurized 3,1.2, 0.65,0.22 0.22 Hard 4.5 50, Blank Pressurized 8, 5,3,1.2, 0.65, 0.22 Easy 4.5 50,2% MoS2 Pressurized 8,5,3,1.2,0.65, 0.22 Easy 4.5 50,2% Si0 2,3 um Pressurized 8, 5,3,1.2, 0.65, 0.22 - Easy 4.5 50, Blank Aspirator 0.45, 0.22 Easy 4.5 50, 2% MoS2 Aspirator 0.45,0.22 Easy 4.5 50,2% Si0 2,3 um Aspirator 0.45, 0.22 Easy 92 Chapter 4 -Results and Discussion 4.3 Effect of Kaolin and Southern Alberta Clay on Coking Studies were carried out using the 50 g autoclave operating for 4.5 hours at 390 °C and a 200 ml/min N 2 purge rate. Sample mixtures containing either 2 wt% southern Alberta clay or 2 wt% kaolin (<0.45 um) in bitumen were used. After most of the coking experiments, the toluene insolubles were measured by vacuum filtration using first a 0.45 urn then a 0.2 urn pore size filter. In a few of these experiments, a 3 um pore size filter was also used prior to the 0.45 um in order to reduce the total filtration time. The experiments were repeated three times for each type of clay and the results were compared with those obtained for the blank case at the same conditions. Table 4.7 summarizes the results of these six experiments. Statistical analyses in Appendix E show that there is no significant difference in the coke yield or weight loss obtained in the presence of either type of solid compared to the results obtained (Table 4.5a) when no solid was added. Table 4.7 Toluene insolubles and weight loss for clay added samples. Experiments at 390 °C and 200 ml/min N 2 purge for 4.5 h coking. Solid added % Toluene Insolubles % Weight Loss 2% Alberta clay 4.74 51.35 2% Alberta clay 4.24 48.76 2% Alberta clay 4.19 50.93 Mean value ± Std. dev 4.39 ± 0.35 50.35 ± 1.39 2% Kaolin 4.23 49.80 2% Kaolin 3.93 48.80 2% Kaolin 4.25 54.48 Mean value ± Std. dev 4.14 + 0.18 51.02 ± 3.03 No solid added 3.83 47.6 No solid added 4.22 45.04 No solid added 4.21 53.29 No solid added 3.96 46.02 Mean value ± Std. dev 4.05 ± 0.19 47.98 ± 3.69 93 Chapter 4 -Results and Discussion 4.4 Effect of Native Clays Native clays provided by Syncrude Canada Ltd. were prepared as discussed in Section 3.2.5 and added to bitumen to make a 2 wt% mixture. The coking experiments, carried out on the mixture for 4.5 hours at 390°C and 200 ml/min nitrogen purge, were repeated three times. The toluene insoluble percentage and the weight loss results are tabulated in Table 4.8. Table 4.8 Toluene insolubles and weight loss for native clay added samples. Experiments at 390 °C and 200 ml/min N 2 purge for 4.5 h coking. Solid added % Toluene Insolubles % Weight Loss 2% Native clay 3.02 47.84 2% Native clay 2.95 46.81 2% Native clay 3.22 48.84 Mean value ± Std. dev 3.06 ± 0.14 47.83 ± 1.01 Statistical analysis in appendix E confirms that the coke yield obtained in the presence of the native solids is significantly different from that of the blank experiments (Table 4.5a) within a 95% confidence interval. The toluene insoluble yield was calculated according to Eq. 4.7 which assumes the 100% recovery of the added solids during the filtration of the liquid product and washing liquid. Silica is the major component of these clays, comprising more than 50%. Thus, in order to determine i f this assumption was correct and that there was no significant loss of clays during filtration, a sample of the filtrate was sent out for silicon analysis. 94 Chapter 4 -Results and Discussion The analysis, performed using plasma spectroscopy (ICP) by CANTEST Laboratory (Vancouver, B.C.), showed that there was less than 8 ppm silicon in the filtrate, which represented a negligible amount compared to the mass of added clay retained by the filter. A statistical analysis of the weight loss measurements (Appendix E) confirms that, contrary to the coke yield, there was no significant difference, within 95% confidence, between the blank and solids-added experiments. Characterization of the native clays (Chapter 3) shows that its morphology is similar to that of the other clays which didn't have any effect on coking (i.e., kaolin, Southern Alberta clay). The major discernible difference between this clay and the others is that it has a much higher carbon content, (6.45 wt% C compared to 0.44% and 0.32% for kaolin and Southern Alberta clay, respectively - see Table 3.6). Test showed the native solids did not easily disperse in water. Figure 4.14 compares the coke yield results for the six different solid additives at 390 °C and 4.5 hour coking with a 200 ml/min nitrogen purge. Among these additives, only the Athabasca native clays decreased the coke yield, while none of the rest had any statistically significant effect. This behavior is likely because of interaction of these solids with the developing coke phase. For the hydrophilic kaolin particles with the same morphology, there was no observed effect on coke yield. Therefore the higher carbon content of the native clays, which likely makes them hydrophobic, is suggested to be the reason for their different behavior from other solids. Native clays with a higher carbon content and a hydrophobic character could be better dispersed than hydrophilic clays in the bitumen, and therefore interfere with the developing coke phase. This implies that the surface character of the added solids plays the major role in their interaction with the coke phase. 95 Chapter 4 -Results and Discussion Blank Kaolin Southern Silica sand Native M o s 2 Alberta clay 3 ^ m s o l i d s Figure 4.14 Toluene insolubles for blank and five different solids added samples. Experiments at 390 °C and 4.5 h with 200 ml/min nitrogen purge. 4.5 Effect of the Concentration of Native Clay Additives Native clays were shown to yield a significant reduction in the coke yield, when they were added to the bitumen at a concentration of 2 wt%. The objective of this set of experiments was to investigate how the coke yield varied with the concentration of native solids in the bitumen. For this reason, additional coking experiments were carried out with native solids added to the bitumen at two other concentrations, namely 1 and 4 wt%. The standard operating conditions of 200 ml/min nitrogen purge, 390 °C and 4.5 hours Chapter 4 -Results and Discussion were employed. Table 4.9 summarizes the coke yield and weight loss results obtained at different concentrations of the native clay solids. Table 4.9 Toluene insolubles and weight loss results for different native clay concentrations. Experiments at 390 °C and 200 ml/min N 2 purge for 4.5 hour coking. % Added solid % Toluene insolubles % Weight loss 0 3.82 47.6 0 4.22 45.04 0 4.21 53.29 0 3.96 46.02 Mean value + Std. dev 4.05 ± 0.19 47.98 + 3.69 1 3.1 47.35 1 3.25 45.38 1 3.11 47.52 Mean value ± Std. dev 3.15 ± 0.08 46.75 ± 1.19 2 3.02 47.84 2 2.95 46.81 2 3.22 48.84 Mean value + Std. dev 3.06 + 0.14 47.83 ± 1.01 4 2.76 46.11 4 2.93 49.6 4 2.73 46.34 Mean value ± Std. dev 2.81 ± 0.11 47.35± 1.95 The coke yield results are also plotted in Figure 4.15 as a function of the concentration of the native solids. It is obvious from the figure that the more native solids added the greater the reduction in the coke yield. Chapter 4 -Results and Discussion 4.5 3.5 2.5 i 2 3 % Native sol id Figure 4.15 Toluene insolubles vs. different concentrations of native solids. Experiments at 390 °C and 200 ml/min nitrogen purge for 4.5 h. A statistical analysis presented in appendix F shows that the coke yield results at any concentration of the native clays, are significantly different, within a 95 % confidence interval, from the blank experiments (0 % native solids). But, by similar analysis, it was determined that the addition of these solids had no significant impact on the weight loss. 98 Chapter 4 -Results and Discussion 4.6 Effect of the Purge Rate on Coking A l l of the autoclave experiments described thus far were carried out at a constant nitrogen purge rate of 200 ml/min. In order to investigate the possible effect of purge rate, two additional purge rates of 100 and 400 ml/min were chosen for study. For these experiments, no solids were added and a coking time of 4.5 hours and a temperature of 390 °C were employed. The experiments at each purge rate were repeated three times and the results are summarized in Table 4.10. The detailed experimental results, including average values and standard deviations, are provided in Appendix G. Table 4.10 Weight loss and toluene insolubles for different purge rates. Experiments at 390 °C and for 4.5 h. Purge rate (ml/min) % TI % Weight loss 100 3.45 42.6 100 3.63 46.2 100 3.72 44.9 Mean value ± Std. dev. 3.59 ± 0.14 44.5 ± 1.82 200 3.83 47.6 200 4.22 45.04 200 4.21 53.29 200 3.96 46.02 Mean value ± Std. dev. 4.05 ± 0.19 47.98 ±3 .69 400 4.93 58.81 400 4.78 58.68 400 4.71 60.79 Mean value ± Std. dev. 4.81+0.11 59.43 ± 1.18 The above table shows that, under otherwise identical conditions, increasing the purge rate increases both the coke yield (Fig. 4.16a) and the weight loss (Fig. 4.16b). Therefore, at a constant temperature and coking time, the weight loss depends on the Chapter 4 -Results and Discussion purge rate. Furthermore, since there is an apparent relationship between the coke yield and the weight loss (see Fig. 4.3), the weight loss also decreases with the purge rate. 6 -5 -4 -1 0 0 100 200 300 400 500 Purge rate (ml/min) Figure 4.16a Toluene insolubles vs. purge rate. Experiments at 390 °C and 4.5 h coking in 50 g autoclave.reactor with no solids added. The weight loss trend can be explained as follows. As the purge rate increases, more volatiles will be swept from the reactor, thereby, yielding an increased weight loss (Fig. 4.16b). Figure 4.16b Weight loss vs. purge rate. Experiments at 390 °C and 4.5 h coking in 50 g autoclave reactor with no solids added. 100 Chapter 4 -Results and Discussion The coking reaction involves devolatilization. Increasing the purge rate results in the stripping of more volatiles from the reaction matrix and hence should enhance coke formation. As well, from a chemical equilibrium point of view, more volatiles should form to compensate the increased loss from higher purge rates. This also would result in more weight loss at higher purge rates. 4.7 Effect of Purge Rate on Coking in the Presence of Native Solids In this set of experiments the objective was to study the effect of the purge rate in parallel with the effect of the native solids. Hence, three sets of experiments were carried out at different purge rates (100, 200 and 400 ml/min) in the presence of 2 wt% native clays. As demonstrated in Fig. 4.17, the trend of increasing coke yield with increasing purge rate is almost the same in both the presence and absence of native solids. However, the coke yield is significantly lower when the solids are present. For each purge rate, the results of solid added and blank experiments were statistically compared (Appendix G). This comparison shows that, within 95% confidence, the addition of solids, independent of the purge rate, decreases the coke yield. 101 Chapter 4 -Results and Discussion Figure 4.17 Toluene insolubles vs. purge rate in the presence of native solids. Experiments at 390 °C and 4.5 h. As shown in Fig. 4.18, the weight loss trend was almost unaffected by the presence of the native solids. This is confirmed by the statistical analysis presented in Appendix G. Therefore, the parallel lines in Figure 4.17 suggest that the presence of native solids only reduces the coke yield without effecting the weight loss (Fig. 4.18). 102 Chapter 4 -Results and Discussion 500 P u r g e Ra te (ml/min) Figure 4.18 Weight loss vs. purge rate in the presence of native solids. Experiments at 390 °C and 4.5 h. 4.8 Effect of Agitation on Coking A few of the autoclave experiments were carried out in the absence of stirring. The objective of these experiments was to provide further information about differences in behaviour noted for the non-stirred tube reactor and the stirred autoclave reactor. Three different coking times (2.5, 3.5 and 4.5 hours) at 390 °C and 200 ml/min nitrogen purge were chosen for these experiments. In contrast to all of the other experiments, the coke materials precipitated out of solution as a black brittle solid, which was very difficult to remove from the reactor. It can be concluded that stirring not only keeps the bitumen sample homogenized, but also prevents any temperature stress during the reaction. It 103 Chapter 4 -Results and Discussion seems that, in all of these experiments, the extremely high temperature existing near the walls carbonized portions of the sample. 4.9 A Phase-Separation Kinetic Model for Coke Formation 4.9.1 Parallel Reactions Kinetic Model A phase separation kinetic model for coke formation was studied by Wiehe in 1993 (Chapter 2). In this model it is postulated that the coke formation mechanism involves the liquid-liquid phase separation of reacted asphaltenes to form a phase that is lean in abstractable hydrogen. Cold Lake vacuum residue, containing 25 wt% asphaltene (heptane insoluble) and 75 wt% maltenes (heptane soluble), with a normal boiling point above 566 °C, was used as the reactant. Either 3 grams of the heptane soluble fraction or 3 grams of the full residue (containing both the heptane solubles and asphaltene) were reacted at various times at 400 °C in both open and closed reactors and the products were separated according to the scheme shown in Fig. 4.19. The variation with time of the concentration of each of the solvent fractions obtained, for example, by reacting the full residue in an open rector, was given in Fig. 2.3. This plot shows four common features of residuum thermal conversion kinetics, which led Wiehe to formulate the parallel reactions model. These features are listed as follows: 1. Coke induction time: In the initial stage of the reaction, the coke concentration is constant and the same as the initial toluene insoluble content of the feed. 104 Chapter 4 -Results and Discussion Volatiles Residuum • Insolubles (Coke) • Step 1 Thermal reaction Step 2 Toluene filtration Insolubles (Asphaltenes) Step 3 Heptane filtration Heptane solubles (Maltenes) Figure 4.19 Separation scheme for reaction products used by Wiehe (1993). 2. Asphaltene maximum: The concentration of asphaltene increased from its initial value to a maximum and then decreased. The maximum occurred at the same reaction time as the end of coke induction period. It is a result of heptane solubles reacting to form asphaltenes, which in turn react to form coke. This maximum in asphaltene concentration has been observed by other investigators (Levinter et al, 1966; Magaril and Aksenova, 1968; Valyvinetal, 1980; Takatsuka et al, 1989). 3. Decrease of asphaltene parallels decrease of heptane solubles: In Figure 2.3, during the period when coke is formed, the ratio of the asphaltene concentration to the concentration of heptane solubles approaches a constant. Wiehe suggested that this ratio is the solubility limit of converted asphaltenes in the heptane solubles. 4. High reactivity of converted asphaltenes: A first order reaction rate constant of 0.013 min"1 was obtained by fitting the decrease in heptane solubles with reaction time. When the reactant is the asphaltene fraction, the disappearance of the asphaltene fraction can be described using a first order kinetic model with a reaction rate constant of 0.026 min"1. Therefore, the unconverted asphaltenes are actually the most thermally reactive fraction of the residua. Chapter 4 -Results and Discussion Wiehe's parallel reaction model, based on these four common features, was described in Section 2.4. According to Eqs. 2.2 - 2.6, seven parameters, namely K H , K A , SL, a, b, c, and y, are required to describe the temporal variation of the four fractions. Wiehe obtained these parameters by curve fitting the Cold Lake residua conversion data, for both the open and closed reactors, to these equations. These parameters are also given in Section 2.4. This parallel reaction model with its fitted parameters described the Cold Lake residuum thermolysis behavior very well. However, further studies by Wiehe on other feed stocks, such as Arab heavy oil and Hondo oil, showed that the parallel model does only a poor job of describing their thermolysis data. 4.9.2 Series Reactions Kinetic Model Further studies by Wiehe (1997) on the molecular weight distribution of the products showed that asphaltenes from the thermolysis of heptane solubles have properties between those of A + and A*. Also, heptane solubles do not suddenly increase aromaticity when coke formation is initiated. Therefore he improved his parallel reaction model to a series reaction model. He also concluded that H* is not a co-product of coke formation. Thus the series reactions model consists of the following sequence of reactions: (4.8) (4.9) (4.10) (4.11) 106 H + K " ^ bA + + n-bW A + K a • c A * + d H ' - ( l - c - d ) V A * m a x = S L ° ( H + H') A * = A * - A * •"-ex ^ max Chapter 4 -Results and Discussion At long reaction times: A * e x ^ T I (4.12) Wiehe also used curve fitting to estimate the six parameters of the series reaction model for the three different feed stocks, in open and closed reactors, and for different temperatures. These parameters are listed in Table 4.11. Table 4.11 Six parameters of Wiehe's (1997) series reaction model. Residuum Reactor T(°C) K^mhV 1) K^mm 1 ) s°L b C d Cold Lake Open 370 0.00754 0.0286 0.349 0.302 0.602 0.108 Open 400 0.0260 0.0134 0.036 0.302 0.602 0.108 Closed 400 0.0260 0.0134 0.137 0.302 0.602 0.108 Open 420 0.0553 0.0350 0.284 0.302 0.602 0.108 Arab Heavy Open 400 0.0240 0.00773 0.291 0.383 0.611 0.019 Closed 400 0.0240 0.00773 0.126 0.383 0.611 0.019 Hondo Closed 400 0.315 0.0353 0.144 0.263 0.640 0.315 The three stoichiometric coefficients, b, c, and d, were found to be independent of temperature. The rate constants, K H and K A , were described by Arrhenius relationships, in which the activation energies were E H = 185,400 J/mol (44.3 kcal/g-mol) and E A = 147,700 J/mol (35.3 kcal/g-mol) (Wiehe, 1997). 4.9.3 Finding Parameters of the Wiehe Series Reaction Model for the Tube Reactor Data In the tube reactor experiments there were two factors measured, %toluene insolubles and %weight loss, over different periods of time. These variables correspond to the volatiles 107 Chapter 4 -Results and Discussion and coke fractions in the Wiehe model. Here there is an attempt made to use Wiehe's series reactions model and determine the parameters of the equations which dominate the volatiles and coke yield changes as functions of time. The derivation, based on the series model, of the differential equation which describes the toluene insolubles fraction as a function of time is given in Appendix H. The equation obtained is as follows: dTI dt (c - SL°d)KA b \ + SL°H0(l + db) - bcH0 V K A — Air + KA ~ KH J x KH exp(-KHt) xKAexp(-KAt) (4.13) This equation assumes that the reactions have passed the induction time and the asphaltene concentration has reached its solubility limit. Therefore, the equation is only valid for the period of reaction that follows the induction time. Integrating the above equation yields: TI (SL°d-c) K, SL°H0(l + db) + bcHQ x exp(-KHt) V \ KA — KH J x e x p ( - A V ) + A'I ( 4 - 1 4 > Also, the following equations for volatiles were obtained (Appendix H): dV_ dt (l-2b + bc + bd) + (\-c-d) bK. KA ~ KH + (l-c-d)KA A _ bKHHQ kHHQexp(-KHt) expf-AV) (4.15) KA ~KH 108 Chapter 4 -Results and Discussion V= (2b-bc-bd-l) + (c + d-l) bKA H0 exp(-KHt) KA~KH\ + (c + d-l)\ A{ o H exp(-KAt) + K2 (4.16) There are six unknowns in the above equations: the three stoichiometric coefficients, b, c, and d, which are independent of temperature, the solubility limit, S°L, which is temperature dependent and the two integration constants, K[ and K 2 , also assumed to be independent of temperature. It is assumed that the reaction rate constants, K H , and K A , at 4 0 0 °C and the activation energies given for the Cold Lake residuum in Wiehe's work are also valid for the Cold Lake bitumen used in this study. By knowing the activation energies, rate constants were calculated for 3 8 0 °C and 3 9 0 °C using the Arrhenius equation (Appendix H). AQ and H 0 are the initial asphaltene and heptane soluble contents (wt%) of the feed. Knowing that the initial feed contains 0.4 wt% toluene insolubles (TI0) and 15.4 wt% asphaltene (AQ) (refer to the separation schema, Fig 4 .19) , results in H 0 = 84.2 wt%, since Using the above values, the % toluene insolubles and the % weight loss data in the tube reactor for three different experimental temperatures (380, 3 9 0 and 4 0 0 °C) were curve fitted to Eqs. 4 .14 and 4.16, respectively, using Mathcad subroutine Linfit. The detailed calculations are presented in Appendix H . Table 4 . 1 2 summarizes the results obtained for the stoichiometric coefficients, b, c, and d, and solubility limit, S°L, along with the rate constants. It should be noted that these parameters were obtained from the experimental A^+Ho+TI^lOO (4.17) 109 Chapter 4 -Results and Discussion data after the induction time and at a 200 ml/min nitrogen purge rate. Table 4.12 also gives the correlation factors between the predicted and experimental V and TI values. These values may be different for the other experimental conditions and feeds. Table 4.12 Parameters of the Wiehe series reaction model calculated for Cold Lake bitumen using the tube reactor resultsfTable 4.1)*. Temperature (°C) b c d sL° K H (min l ) R\ R 2 T , 380 0.762 -5.8 E-3 0.774 0.050 0.00486 0.00486 0.991 0.959 390 0.640 0.284 0.844 0.075 0.00813 0.00813 0.994 0.996 400 0.743 0.282 0.318 0.088 0.01340 0.01340 0.992 0.993 * Detailed calculations in Appendix H . According to the Wiehe series model, the solubility limit is a function of temperature, but the stiochiometric coefficients should be independent of temperature. In order to use an identical set of stoichiometric coefficients for all the experimental temperatures, the arithmetic averages of b, c, and d were used for predicting the results. Also, to improve the curve fitting, slightly different solubility limits from the ones in Table 4.13 were used along with the set of average parameters (Table 4.13). Table 4.13 Parameters of the Wiehe series reaction model used for predicting all of the Cold Lake bitumen data. Temperature (°C) b c d s\ KH(min') K^min 1 ) 380 0.716 0.1867 0.6453 0.065 0.004858 0.01158 390 0.716 0.1867 0.6453 0.080 0.008131 0.01746 400 0.716 0.1867 0.6453 0.099 0.013400 0.02600 110 Chapter 4 -Results and Discussion Then Equations 4.14 and 4.16 with the average parameters (Table 4.14) were used to predict the % weight loss and % toluene insolubles for the three different temperatures and times which exceeded the induction times. The predicted and experimental results, are compared in Tables 4.14-4.16 and in Figures 4.20 and 4.21. Table 4.14 Experimental and predicted values for % toluene insolubles and % weight loss at 380 °C after the induction time. Time after Induction (min) %TI predicted % T I experimental %Wt. loss predicted %WL loss experimental 0 0.40 0.45 47.40 47.4 30 0.80 1.16 50.25 49.7 60 1.34 1.70 52.93 50.8 90 1.93 2.07 55.40 54.6 120 2.54 1.85 57.64 55.8 150 3.14 2.70 59.66 57.7 Table 4.15 Experimental and predicted values for % toluene insolubles and % weight loss at 390 °C after the induction time. Time after Induction (min) %TI predicted % T I experimental %Wt. loss predicted % Wt. loss experimental 0 0.40 0.40 41.70 41.7 30 1.35 0.61 46.24 46.3 60 2.48 2.54 50.29 53.7 90 3.60 3.43 53.75 58.2 120 4.62 4.47 56.63 59.5 150 5.50 5.60 58.99 61.8 180 6.23 6.38 60.89 63.1 210 6.84 7.12 62.42 65.2 111 Chapter 4 -Results and Discussion Table 4.16 Experimental and predicted values for % toluene insolubles and % weight loss at 400 °C after the induction time. Time after Induction (min) %TI predicted % T I experimental % Wt. loss predicted %Wt. loss experimental 0 0.40 0.40 46.60 46.6 30 2.53 2.03 53.65 52.4 60 4.63 4.35 59.30 58.6 90 6.35 5.18 63.50 59.9 120 7.64 7.53 66.51 65.2 150 8.58 8.57 68.62 67.7 180 9.23 9.96 70.07 69.9 210 9.68 11.19 71.05 72.0 240 9.99 11.06 71.72 74.2 Figure 4.20 Experimental and predicted % weight loss results for Cold Lake bitumen. Experiments at 200 ml/min nitrogen purge. 112 Chapter 4 -Results and Discussion At 400 °C and 380 °C, the measured weight losses are lower than the predicted values over most reaction times. At 390 °C, the experimental results are slightly higher than those predicted. 12 10 380 °C model —a— 380 °C exp. 390 °C model •—A— 390 °C exp. 400 °C model — • — 4 0 0 °C exp. 50 100 250 300 350 150 200 Time (min) Figure 4.21 Experimental and predicted values of % TI for the Cold Lake bitumen Experiments at 200 ml/min nitrogen purge. The %TI fit is quite good for all three temperatures tested. In contrast to the weight loss data, the TI data appears to be randomly scattered about the predicted values. The largest deviation appears at the most severe condition of high temperature and long reaction time. 113 Chapter 4 -Results and Discussion Also, plotting both the experimental and calculated %TI vs. % Wt. loss results in Fig. 4.22, shows that the series model approximates the temperature-independent linear behavior observed for the experimental results. In contrast to the single curve suggested by the data, the model predictions are a family of curves which depend on temperature. The curve for 400 °C lies between those for 380 °C and 390 °C. 12 10 380 °C model • 380 °C exp. 390 °C model A 390 °C exp. 400 °C model • 400 °C exp. 380 °C 30 40 i 50 60 70 80 %Wt. loss Figure 4.22 % T I vs. % weight loss using experimental and predicted values. Experiments at 200 ml/min nitrogen purge. Wiehe (1997) reported behavior for the Cold Lake vacuum residuum roughly similar to that in Figure 4.22. However, the range of the %weight loss and % toluene insoluble data 114 Chapter 4 -Results and Discussion was slightly different for his Cold Lake vacuum residuum experiments (with H 0 = 75% = 25% and Tl 0 = 0). Wiehe used temperatures of 370, 400 and 420 °C in his experiments and found that the % weight loss was ranged from 30 to 65%, while the %TI was 0 to 27%. 4.9.4 Summary The Wiehe series kinetic model with the new calculated set of parameters was able to predict the toluene insoluble yield and weight loss for Cold Lake bitumen after the induction time at different temperatures. The correlation coefficients between the experimental and predicted %TI and %Weight loss values ranged from 0.93 to 0.99, which indicates a good correspondence. However, it should be kept in mind that the kinetic parameters might be different for other experimental conditions and for other types of feed. Also, the Wiehe model might not be reliable in the presence of the additives such as native solids. Further work would be necessary to determine how the presence of native solids would change the model parameters (Solubility limit, the rate constants, etc.). Coking in the presence of native solids reduced the coke yield without any impact on the weight loss. However, according to the Wiehe model, weight loss and coke yield are related and any change in one of them should affect the other. 115 Chapter 5 - Conclusions and Recommendations 5. Conclusions and Recommendations 5.1 Conclusions Thermal treatments of Cold Lake bitumen was carried out at temperatures from 380-400 °C in different batch reactors in the presence or absence of different solids in order to study the effects of added solids. Toluene insolubles, an indicator of coke yield, and percent weight loss were measured in all the experiments. 5.1.1 Tube Reactor (no solids added) a) The toluene insolubles yield and weight loss of the Cold Lake bitumen were determined as functions of time and temperature in the absence of added solids in the non-stirred tube reactor. After a certain induction period, the toluene insolubles content increased from 0.4% (initial feed content) to a maximum of about 11% for the highest temperature and longest reaction time. The induction time, during which the rate of formation of TI is zero, decreased from 2.5 h to 2 h and 1.5 h, as the reaction temperature was increased from 380 °C to 390 °C and 400 °C, respectively. b) After the induction period, the yield of toluene insolubles increased approximately linearly with time. The greater the reaction temperature, the greater was the rate of coke formation for any time interval beyond the induction period. c) The percentage of weight loss increased continuously with reaction time, exhibiting no induction period, and approaching an apparent asymptotic value after about 5 hours. For a given reaction time, weight loss increased with temperature. d) Regardless of time and temperature, all the data of TI yield vs. weight loss lie essentially on the same curve. The threshold of TI formation was reached at a 116 Chapter 5 - Conclusions and Recommendations specific weight loss of about 47%, after which the coke yield increased linearly as a function of weight loss. This phenomenon is understandable, given that the devolatilization is mechanistically linked to the coking reaction. Previous results by others have suggested that dealkylation is the predominant reaction at low volatile yields, resulting in volatiles and condensed aromatic cores. At higher volatile yields, aromatization becomes predominant, e) A set of stoichiometric parameters as well as a solubility limit were calculated for the Wiehe series kinetic model, which was able to predict reasonably well the weight loss and toluene insoluble yield, after the induction time at a 200 ml/min nitrogen purge rate. 5.1.2.Tube Reactor (solids added) a) Addition of 2 wt % MoS 2 (with a size distribution 0.2 p;m to 20 p.m) did not result in a statistically significant difference for either the coke yield or weight loss compared to the blank sample (no solids added). b) Silica particles (i.e., chemically inert particles), with an average size of 3.0 |im at the level of 2 wt% in the bitumen, did not have a measurable effect on either the coke yield or weight loss at 390 °C and 4.5 h. Addition of 6.5 pm silica sand at the same concentration significantly reduced the coke yield from that found with the blank. The average coke yield with the 3.0 [im solids was lower than the yield with the 6.5 p.m particles, but it was not statistically different from the blank results, simply because the standard deviation of the former measurements was much larger. 117 Chapter 5 - Conclusions and Recommendations 5.1.3 Autoclave Reactor a) Coking tests on blank samples, showed that under identical conditions of 390 °C and 200 ml/min nitrogen purge, the induction time in a 100 g stirred autoclave reactor was between 3.5 and 4.5 hours, in comparison to 2 hours in the non-stirred 1.5 g tube reactor. The weight loss range obtained in the autoclave reactor (30-39%) was also lower than that in the tube reactor (53-63%). These differences in results were attributed to the effects of different geometry and different stirring conditions of the two reactors. b) Experiments for 2.5 h and 3.5 h at 390 °C with addition of 2 wt% of silica or MoS 2 particles to a 100 g bitumen sample, showed no significant difference from blank runs with no solid added, for either coke yield or weight loss. For 4.5 h of reaction, which exceeds the induction time, a significant amount (about 5%) of coke formed. However, for 4.5 h of coking of a 100 g sample, a quantitative study of the effect of solids was not possible due to filtration problems in the presence of the solids. c) Comparison of experiments with a 50 g and 100 g samples, under otherwise identical conditions of 390 °C and 200 ml/min of nitrogen purge showed that the induction time increased from 2.5-3.5 h to 3.5-4.5 h. This was also ascribed to the nature of the chemical reactions and volatilization involved. In order to reach the coke formation threshold, the sample must first lose a certain amount of volatiles and pass through the reaction phase where dealkylation is predominant. After this point is reached, aromatization becomes predominant and coke starts to form. Therefore, the larger the sample size, the greater the mass of volatiles which must leave the reactor before 118 Chapter 5 - Conclusions and Recommendations coke can form. This means a longer reaction time is required to reach this level or, in other words, the induction time will increase with the sample size. d) The addition of 2 wt% of Southern Alberta clay, kaolinite (with a size distribution of < 0.45 p:m), MoS 2 (with a size distribution of 0.2-20 (im) or silica (with an average size of 3 Jim), had no measurable impact on either the coke yield or the weight loss for 4.5 h of coking at 390 °C and 200 ml/min nitrogen purge. It was concluded that there was no significant chemical or physical interaction between the bitumen and these solids during the reaction. e) Athabasca native clays, which were separated from a Athabasca bitumen during the extraction process and were recovered from middling stream of Syncrude Canada Ltd.'s operations, were also applied as an additive in several coking experiments. Characterization of these clays showed that their morphology is similar to that of the other clays tested (i.e., kaolinite and Southern Alberta clay). The only significant difference among them is the higher carbon content of the native clays (6.45 wt% compared to 0.44 and 0.32 wt% for kaolin and Southern Alberta clays, respectively). Coking of a 50 g bitumen sample, at 390 °C for 4.5 h with 2 wt% of native clays, gave a significant decrease from 4 wt% to 3 wt% in the toluene insolubles content compared to what was found in the absence of these clays. This behavior is probably caused by the interaction of these solids with the developing coke phase. For the hydrophilic kaolin particles with the same morphology there was no observed effect on the coke yield. Therefore, the higher carbon content of the native clays, which makes them hydrophobic, is suggested to be the reason for their different behavior compared to the other solids. Native clays with a high carbon content and a 119 Chapter 5 - Conclusions and Recommendations hydrophobic character could be better dispersed in the bitumen than hydrophilic clays and therefore might interfere more with the developing coke phase. This implies that the surface character of added solids may play a major role in their interaction with the coke phase. There was no effect of native clays on the weight loss. f) The effect of concentration of native solids (added from 1 to 4 wt%) on the coke yield was studied at 390 °C and a reaction time of 4.5 h. The more native solids added to the mixture, the greater was the reduction in coke yield obtained. Hence removal of native solids is expected to enhance the yield of toluene insolubles. Increasing the concentration of the native solids to 4 wt% had no impact on the weight loss. g) The effect of the nitrogen purge rate was studied in the absence of native solids. It was concluded that increasing the purge rate increased both the weight loss and the coke yield. The coke yield increased linearly with the purge rate. The devolatilization reaction is an important first step in the small coking process. Increasing the purge rate results in the stripping of more volatiles from the headspace of the reactor, reducing the concentration of these species on the gas side of the vapor-liquid interface. Since this concentration is presumed to be in equilibrium with the liquid-side interfacial concentration, the devolatilization rate will be increased. Thus for a given reaction time, a greater weight loss will occur at a higher purge rate. h) The effect of the purge rate was also studied in the presence of 2 wt% native solids. The trend of the toluene insolubles vs. purge rate is the same in the presence of native solids as in their absence. It was concluded that regardless of the purge rate, the addition of the native solids at 2 wt% reduces the toluene insolubles. 120 Chapter 5 - Conclusions and Recommendations 5.2 Recommendations In order to further investigate the effect of the native clays, which reduced the coke yield under different conditions, additional studies are suggested. 1. Since this effect appears to be due to the surface characteristics of the native clays (which may be hydrophobic because of their high carbon content), then it might diminish if the carbon is first removed from the solids. This could be carried out by treating the native solids with a strong solvent such as tetrahydrofuran which should elute all the resins from the native clays (e.g.,following the ASTM-D2007-80 method). 2. Also, other particles such as kaolin, which had no effect on the coke yield, could be coated by a surfactant to investigate whether the coke yield is reduced in a similar way as with the native solids. For example, polyoxyethylene-4-lauryl ether (PE4LE, Ci2H25(OC 2H4)40H2, could be used as a hydrophobic surfactant which would adsorb to the surface of the particles. Also, Yan and Masliyah (1996) give a recipe for making hydrophilic kaolin particles hydrophobic by coating them with asphaltene. A similar procedure could be performed with kaolin or Southern Alberta clays in order to change their surface characteristics before adding those coated particles to the bitumen samples and then determining their effects of those in the coking experiment. 3. Particle size effect of all solids should be investigated. 4. A model, which describes the effect of the purge rate should be developed, to provide quantitative predictions for scale up. 121 Nomenclature a Stoichiometric coefficient A Heptane insoluble asphaltenes A* Concentration of asphaltene core fraction (wt%) A*e* Excess asphaltenes cores beyond what can be held in solution (wt%) A*maX Maximum concentration of asphaltene cores that can be held in solution (wt%) A + Concentration of reactant asphaltene fraction (wt%) Ao Initial asphaltene concentration (wt%) b Stoichiometric coefficient c Stoichiometric coefficient d Stoichiometric coefficient E a Activation energy (J/mol) E A Activation energy of the first order reaction for the thermolysis of asphaltenes (J/mol) E H Activation energy of the first order reaction for the thermolysis of heptane solubles (J/mol) H Concentration of nonvolatile heptane solubles (wt%) H* Concentration of nonvolatile heptane solubles product (wt%) H + Concentration of nonvolatile heptane solubles reactantt (wt%) H 0 Initial concentration of non volatile heptane solubles, wt% K i Integration constant 122 K 2 Integration constant K A First order reaction rate constant for the thermolysis of reactant asphaltene, min"1 K H First order reaction rate constant for the thermolysis of reactant nonvolatile heptane solubles, min"1 K T i Rate constant at temperature T i (min1) K-T2 Rate constant at temperature T 2 (min 1) m Stoichiometric coefficient n Stoichiometric coefficient ni Number of data n 2 Number of data R 2 V Correlation factor between experimental and predicted volatiles yield R 2 n Correlation factor between experimental and predicted Toluene insolubles yield 51 Standard deviation 5 2 Standard deviation S L° Solubility limit S L Solubility limit t Calculated't' value for the V test statistics t Time (min) tan. 't' value obtained from the student t table for the't' test statistics Ti Temperature (K) 123 T 2 TI V y Greek letters v Abbreviations E D X ICP NPT SEM TI V Temperature (K) Concentration of toluene insoluble coke (wt%) Concentration of volatiles (wt%) Stoichiometric coefficient Population mean Degree of freedom Energy Dispersive X-Ray Inductively Coupled Argon Plasma Spectroscopy Normal Pressure and Temperature Scanning Electron Microscope Toluene Insolubles Volatiles References: 1. Banerjee, D.K.; Laidler, K.J.; Nandi B.N. ; Patmore, D.J., 'Kinetic studies of coke formation in hydrocarbon fractions of heavy crude', Fuel, 1986, 65, 480-484. 2. Beaerden, R.; Aldridge, C.L., US patent 4,134,825,1979. 3. Beuther, H. ; Larson, O.A.; Perotta, A.J., The mechanism of coke formation on the catalysts ', In Catalyst Deactivation, Delmon, B.; Forment, G. F., Eds., Elsevier, Amsterdam, 1980, 271-282. 4. Blayden, H.E.; Gibson, J.; Reily, H.L., 'Proc. Conf. Ultrafine Structure of Coals and Cokes', B U C U R A (London) 1944, 176-182. 5. Bott, T.R., 'Fouling of the Heat Exchangers' Elsevier., 1995, 5-10. 6. Bowman, C.W., 'Molecular and interfacial properties of Athabasca tar sands', Proc. 7th World Petrol. Congr., 1967, 3, 583-603. 7. Bradford, D.J.; Greenhalgh, E.; Kindshott, R.; Senior, A. ; Baily, P.A. 'Proc. 3 r d Conf. Ind. Carbons and Graphite', London, 1971, 520-571. 8. Chung, K . H . ; Dai, Q., 'Bitumen - sand interaction in oil sand processing', Fuel, 1995, 74, 1858-1864. 9. Clark, K .A. , 'The hot water washing method for recovery of oil from the Alberta tar sands', Can. Oil and Gas Industries, 1950, 3(6), 46-49. 10. Del Bianco, A. ; Nicoletta, P., 'Thermo-catalytic hydroconversion of heavy petroleum cuts with dispersed catalyst', Applied catalyst, A. General, Elsevier Sciences, 1993, 94, 1-16. 11. Dukhedin-Lalla, L . ; Yushun, S.; Shaw, J.M., 'Phase splitting of complex hydrocarbon mixtures', Fluid Phase Equilibria, 1989, 53, 415-422. 125 12. Forrest, R.A.; Marsh, H. , 'Determination of the penetration rate of impregnating pitches', Carbon, 1977, 15, 348-349. 13. Garett-Price, B.A. ; Smith, S.A.; Watts, R.L.; Knudsen, J.G.; Marner, W.J.; Suitor, J.W., 'Fouling of Heat Exchangers', Noyes Publications, Park Ridge, New Jersy, USA, 1985. 14. Goldman, J .M.; Flynn, P.C.; Goldman, J.M., 'Direct coking of Athabasca tar sands, Can. J. ofChem. Eng., 1980, 58, 77-83. 15. Gowers, C.W.W., 'Colloid and surface chemical problems in non-conventional Heavy Oil Recovery', Can. Petrol. Techn., 1968, 1(2), 85-90. 16. Gray, M.R.; Ghorpadkar, S.; Vincente, A. , 'Characterization of coke on spent hydro-processing catalyst by optical microscopy', Energy and Fuels, 1994, 8, 426-434. 17. Gray, R .M. , 'Upgrading petroleum residues and heavy oils', Marcel Dekker Inc. New York, 1994, 237-262. 18. Ihnatowicz, M . ; Chiche, P.; Dediut, J.; Pregermain, S.; Tournant, R., Carbon, 1966, 4, 41-50. 19. Levine, S.; Sanford, E., 'Clay-bitumen interactions and their Relevance to tar sand Processing', Proc. 30 t h Can. Chem. Eng. Conf., 1980, 6, 1112-1124. 20. Levinter, M , E.; Meduedeva, M . I.; Panchenkov, G. M . ; Agapov, G. I; Galiakbarov, M.F.; Galikeev, R.K., 'The mutual effect of group components during coking', Khim., Technl. Topi. Masel, 1967, No 4, 20-22. 21. Levinter, m.E.; Medvedeva,M.L; Panchenkov, Gm. ; Aseev, Y . G . ; Nedoshivin, Y . N . ; Finkelshtein, g.B.; Galiakbarov, M.F. 'Mechanism of coke formation in the cracking of 126 component groups of petroleum during coking' Khim. Tekhol. Topi. Masel, 1966, No 9,31-35. 22. Magaril, R.Z.; Ramazeava, L.F.; Asenora, E.I., 'Kinetics of the formation of coke in thermal processing of crude oil', Int. Chem. Eng., 1971, 11, 250-151. 23. Magaril, R.Z.; Ramazeava, L.F.; Asenora, E.I., 'Study of the mechanism of coke formation in the cracking of petroleum resins' Int. Chem. Eng., 1968, 8, 727-729. 24. Marsh H.; Gerus-Piasecka, I.; Grint, A., 'Carbonization and liquid crystal (mesophase) development', Fuel, 1980, 59, 343-348. 25. Marsh, H., 'Carbonization and liquid crystal (mesophase) development: part 1. The significance of the mesophase during carbonization of coals', Fuel, 1973, 52, 205-301. 26. Marsh, H.; Latham, C.S., 'Chemistry of mesophase formation' Petroleum-Derived Carbons, Bacha, J.D; Newman, J.W.; White, J.L. Eds. American Chemical Society, Washington, DC, 1986, 1-28. 27. Marsh, H.; Smith, J., 'Analytical methods for coal and coal products', Karr, C , Ed., Academic Press. New York, 1978, p.371. 28. Mendenhall, W.; Sincich, T., 'Statistics for Engineering and the Sciences', Third Edition, Dellen Publishing Company, 1992,351-356, 282-284. 29. Mochida, I., Oyama, T.; Korai, Y.; Fei, Y.Q., 'Study of carbonization using a tube bomb:Evaluation of lump needle coke, carbonization mechanism and optimization', Fuel, 1988,67, 1171-1181. 30. Mochida, I.; Marsh, H., 'Carbonization and Liquid Crystal Development. 13. Kinetic considerations of the co-carbonization of coals with organic additives', Fuel, 1979, 58, 809-814. 127 31. Mochida, I.; Ihaue, S.I.; Maeda, K.; Takeshita, K., ' carbonization of aromatic hydrocarbon', Carbon, 1977, 15, 9-16. 32. Oyua, A.; Qian, Z.; Marsh, H., 'Structural study of cokes using optical microscopy and x-ray diffraction.', Fuel, 1983, 62, 274-278. 33. Peureux, S.; Bonnamy, S.; Fixari, B.; Lambert, F.; Le perchec; P.; Pepin-Donat; B.; Vrinat, M. , 'Deep hydroconversion of heavy oil residues with dispersed catalyst: analysis of the transformation', Bui. Soc. Chim. Belg., 1995, 104, 359-366. 34. Sanford, E.; Kennepohl, D., ' Conversion of Athabasca bitumen with dispersed and supported Mo-based catalysts as a function of dispersed catalysts concentration.', Energy and Fuels, 1996, 10, 229-234. 35. Sanford, E.C., ' Processibility of Athabasca Oil Sand', 'Can. J. Chem. Eng., 1983, 6, 554-567. 36. Sanford, E.C., 'The nomenclature and jargon of residuum', AOSTRA J. Res., 1991, 7, 201-204. 37. Savage, P.E.; Klein, M.T., 'Asphaltene reaction pathways, Three effects of reaction Enviroment', Energy and Fuels, 1988, 2, 619-628. 38. Shaw, J.M.; Gaikwad, R.P.; Douglas, A.S., 'Phase splitting of pyrene-tetralin mixtures under coal liquefaction conditions', Fuel, 1988, 67, 1554-1559. 39. Smith, R.G.; Scharmm, L.L., 'The influence of natural surfactant on interfacial charges in hot water process for recovering bitumen from Athabasca oil sands', Colloids and surfaces, 1985, 14, 67-85. 40. Speight, J.G., 'The Chemistry and Technology Of Petroleum', Marcel Dekker, Inc., 1991, 176-185. 128 41. Srinivasan, N.S.; McKnight, C.A., ' Mechanism of coke formation from hydrocracked Athabasca residuum', Fuel, 1994, 73, 1511-1517. 42. Storm, D.A.; Barresi, R.J.; Shaw, E.Y. , 'Rheological study of ratawi vacuum residue in the 298 -673 0 K ' , Energy and Fuels, 1995, 9, 168-176. 43. Strausz, O.P.; Montogomery, D.S.; Chen, H.H., 'Hydrocracking of Athabasca bitumen using oil-soluble organometallic catalysts', AOSTRA J. of Res., 1988, 4, 45-57. 44. Takamura, K., 'Microscopic structure of Athabasca oil sands', Can. J. Chem. Eng., 1982, 60, 538-545. 45. Takatsuka, T.; Kajiyama, R.; Hashimoto, H. ; Mastsuo, I.;Miwa, S., ' A practical model of thermal cracking of residual oil. ' J. Chem. Eng. Jpn. 1989a, 22, 304-310. 46. Tanabe, K.; Gray, M.R., 'The role of fine solids in the coking of vacuum residues', Energy and Fuels, 1997, 11, 1040-1043. 47. Valyavin, G.G.; Fryazinov, V . V . ; Gimaev, R.H.; Syunyaev, Z.I.; Vyatkin, Y . L . ; Mulyukov, S.F., 'Kinetics and mechanism of the macromolecular part of crude oil. ' , Khim. Tekhol. Topi. Masel.., 1979, 8-11. 48. Wang, S.; Chung, K.; Masliyah, J.H. and Gray, R., 'Formation of a coke-in-oil Emulsion at reactor conditions: Interactions with dispersed solids', 1997, 1-20, Preprints for submission to Fuel. 49. Warne, G.A, 'Administration of bituminous oil sands of Alberta', J. of Can. Petroleum Tech., Jan- Feb 1984, 45-49. 50. Wiehe, I.A., ' A phase separation kinetic model for coke formation', Ind. Eng. Chem. Res., 1993, 32, No. 11, 2447-2454. 129 51. Wiehe, I.A., 'Thermal Reactivity of Heavy Oils' Presented at the Exxon Research and Engineering Company meeting, Division of Petroleum Chemistry, May, 1997, 149-177. 52. Yan, N . ; Masliyah, J., 'Characterization and demulsification of sohds-stabilized oil-water emulsions. Part 1. Partitioning of clay particles and preparation of emulsions, Colloids Surfaces, A . Physicochem. Engin. Aspects, 1995, 96, 229-242. 53. Yan, T .Y, 'Coke Formation in Visbreaking Process', Prepr. Pap. Am. Chem. Soc. Div. Pet. Chem., 1987, 32, 490-495. 54. Yokono, T.; Marsh, H. , 'Coal Liquefaction Products', Scultz, H.D. Ed., John Wiley & Sons, New York, 1983, 1, p. 125. 130 Appendix A • APPENDIX A - Summary of Blank Coking Experiments in the Tube Reactor This appendix includes all of the results obtained by coking blank samples in the tube reactor at different times and temperatures. An example of the calculation of % weight loss and % toluene insolubles (coke yield) is given at the end. Table A l Toluene insolubles and weight loss results in the tube reactor. Experiments at 380 °C and 200 ml/min nitrogen purge Run- Time Sample Weight % Weight TI (g) %TI Sample 00 wt.(g) after coking (g) loss (TI(g)/feed(g)) 9-1 1 1.3735 0.9511 30.8 0.0061 0.44 9-2 1 1.6835 1.1580 31.2 0.0065 0.39 1-1 2 0.3748 0.2055 45.2 0.0018 0.48 1-2 2 0.3686 0.2039 44.7 0.0018 0.49 1-3 2 0.2424 0.1219 49.7 0.0008 0.33 2-1 2 0.2777 0.1472 47.0 0.0012 0.43 2-2 2 0.2953 0.1575 46.7 0.0015 0.51 2-3 2 0.2490 0.1281 48.6 0.0011 0.44 3-3 2.5 0.7750 0.4078 47.4 0.0026 0.45 4-3 2:40 0.8662 0.4624 46.6 0.0045 0.52 3-2 3 0.4616 0.2348 50.9 0.0026 0.56 4-2 3 1.0685 0.5593 47.6 0.0006 0.56 7-3 3 0.7985 0.3936 50.7 0.0098 1.20 4-1 3:20 0.6290 0.3069 46.6 0.0051 0.80 7-2 3.5 1.1906 0.5943 50.1 0.0180 1.51 3-1 3.5 0.8348 0.4044 51.5 0.0164 1.90 5-1 3:40 1.0001 0.4817 51.8 0.0142 1.40 5-2 3:50 0.7648 0.3469 54.6 0.0203 2.70 5-3 4 0.7020 0.3121 55.6 0.0088 1.20 6-3 4 1.0321 0.4802 53.5 0.0143 1.40 7-1 4 0.9064 0.4140 54.3 0.0270 3.00 9-3 4 1.4908 0.6731 54.8 0.0324 2.70 6-2 4.5 1.0271 0.4575 55.5 0.0182 1.70 8-2 4.5 0.9965 0.4386 56.0 0.0199 2.00 6-1 5 0.8774 0.3670 57.1 0.0240 2.70 8-1 5 1.0790 0.4623 58.2 0.0292 2.70 10-1 6 1.1664 0.4743 59.3 0.0398 3.41 10-2 6 1.4648 0.6049 58.7 0.0450 3.07 10-3 6 1.5578 0.6382 59.0 0.0558 3.58 11-1 6 1.2073 0.4953 58.9 0.0496 4.10 131 Appendix A Table A2 Toluene insolubles and weight loss results in the tube reactor. Experiments at 390 °C and 200 ml/min nitrogen purge Run- Time Sample Weight after % Weight TI(g) %TI Sample (h) wt.(g) coking (g) loss TI(g)/feed(g)) 9-3 0.5 1.1911 0.8689 27.1 0.0047 0.40 9-2 1 1.7982 1.1710 34.9 0.0091 0.51 16-3 1 1.6826 1.1226 33.3 0.0076 0.45 9 -1 1.5 1.1493 0.6725 41.5 0.0054 0.47 11-1 1.5 1.0764 0.6094 43.4 0.0058 0.54 12-3 1.5 1.3097 0.7727 41.0 0.0055 0.42 16-1 1.5 1.6259 0.9642 40.7 0.0074 0.46 12-2 2 0.8153 0.4334 46.9 0.0055 0.69 15-2 2 0.9731 0.5242 46.1 0.0053 0.55 15-3 2 1.3409 0.7271 45.8 0.0078 0.59 2-3 2.5 1.3034 0.5824 53.2 0.0331 2.53 7-2 2.5 1.0835 0.4928 55.3 0.0291 2.69 11-2 2.5 0.7958 0.3631 54.4 0.0184 2.31 12-1 2.5 1.1147 0.5525 50.4 0.0291 2.62 2-1 3 1.0363 0.4225 59.2 0.0384 3.71 5-1 3 1.2295 0.5034 59.1 0.0423 3.44 6-3 3 1.5822 0.6891 56.4 0.0493 3.12 6-2 3.5 1.4339 0.5740 60.0 0.0617 4.30 4-2 3.5 1.2539 0.5141 59.0 0.0709 4.64 3-3 4 1.5800 0.6125 61.2 0.0091 5.76 6-1 4 1.5921 0.6108 61.6 0.0919 5.80 7-1 4 1.3489 0.5070 62.4 0.0741 5.49 3 -2 4.5 1.2405 0.4471 63.9 0.0823 6.63 8 -2 4.5 1.2914 0.4904 62.0 0.0785 6.07 10-1 4.5 1.0764 0.4000 63.7 0.0692 6.28 10-2 4.5 1.0224 0.5202 62.2 0.0893 6.49 10-3 4.5 0.7958 0.6039 62.5 0.1032 6.41. 13-2 5 1.1512 0.3994 65.3 0.0845 7.34 13-3 5 0.9179 0.3210 66.1 0.0708 7.47 14-1 5 1.5920 0.5632 64.6 0.1092 6.86 14-2 5 1.0514 0.3629 65.5 0.0745 7.09 14-3 5 1.3540 0.4813 64.5 0.0929 6.86 132 Appendix A Table A3 Toluene insolubles and weight loss results in the tube reactor. Experiments at 400 °C and 200 ml/min nitrogen purge Run Time Sample Weight after % Weight TI (g) %TI -Sample 00 weight(g) coking (g) loss (TI(g)/feed(g)) 1-1 0.5 0.8000 0.5038 37.0 0.0035 0.44 1-2 1 1.7564 0.9385 46.6 0.0091 0.52 1-3 1.5 0.8612 0.3899 54.7 0.0238 2.76 11-3 1.5 1.5904 0.7937 50.1 0.0206 1.29 3.-3 2 0.5327 0.2057 61.4 0.0252 4.73 11-1 2 1.5990 0.7070 55.8 0.0634 3.96 4 -2 2.5 1.7156 0.6840 60.1 0.0291 5.57 11-2 2.5 1.4326 0.5792 59.6 0.0685 4.78 2.-.1 3 0.7826 0.2676 65.8 0.0589 7.53 3 -1 3 1.3886 0.4819 65.3 0.1066 7.68 5 -3 3 1.4414 0.5051 65.0 0.1099 7.62 6 -3 3 1.5070 0.5347 64.5 0.1100 7.30 5-2 3.5 1.2701 0.4101 67.7 0.1107 8.71 6-2 3.5 1.6137 0.5366 66.7 0.0846 8.36 7-3 3.5 1.4796 0.4683 68.3 0.2247 8.64 6-1 4 0.9233 0.2806 69.6 0.0846 9.16 7-2 4 1.6572 0.4962 70.1 0.1641 9.90 8-3 4 1.3880 0.4206 69.7 0.1352 9.75 9-3 4 1.2212 0.3617 70.3 0.1281 10.50 10-3* 4 1.3497 0.4054 69.9 0.1410 10.45 8-2 4.5 1.4220 0.4081 71.3 0.1557 10.94 7-1 4.5 1.4450 0.3994 72.4 0.1641 11.36 9-2 4.5 0.7751 0.2127 72.6 0.0875 11.29 10-2 4.5 0.2127 0.3309 71.8 0.1309 11.14 8-1 5 1.2372 0.3348 75.8 0.1425 11.51 9-1 5 1.1108 0.3365 69.7 0.1170 10.53 10-1 5 1.7096 0.4659 72.6 0.1892 11.13 * Used for sample calculation. 133 Appendix A Sample calculation (No Solid added): Example calculations for Run 10 -Sample 3 Weight of the glass liner = 4.1058 g Weight of the glass liner and bitumen before coking = 5.4555 g Weight of the sample = Weight of the glass liner and bitumen before coking - Weight of the glass liner => Weight of the sample before coking = 5.4555-4.1058 =1.3497 g Weight of the glass liner and sample after coking = 4.5112 g Weight of the sample after coking = 4.5112 - 4.1058 = 0.4054 g Weight of toluene insoluble on the filter = 0.1410 a) % Weight loss: a n r t t weight of the sample before coking - weight of the sample after coking % Wt loss = weight of the sample before coking Therefore: % Wt loss = 1.3497-0.4054 0.4054 x 100 =69.9% b) % Toluene insoluble: % TI = Weight of toluene insoluble Weight of the initial sample X 1 0 0 Therefore: %TI = 0.1410 1.3497 x 100=10.45 % 134 Appendix B APPENDIX B - Summary of Solids Added Experiments in the Tube Reactor This appendix gives sample calculations for and summarizes the coke yield and weight loss results obtained when various solids were added to the bitumen samples in the tube reactor. Sample calculations (solids added): Weight loss was calculated in a similar manner to the blank experiments (Appendix A). For the solids added experiments (sample 1 and 2 in each run), % TI was calculated as follow: Weight of toluene insolubles - Weight of added solid x J Q Q Weight of initial sample - Weight of added solid Run 14-SampIe 2 in Table B1.3: Weight of the sample = 1.5051 g Weight of the M0S2 added = Weight of the sample x percentage of the solid added Weight of the MoS 2 added = 2.015/100 x 1.5051= 0.0303 g Weight of the toluene insoluble = 0.1192 g 0.1199 - 0.0303 t n n r n - n „ % T l = 15051-0.0303 *">0 = 6.073% Average value and Standard Deviation For each set of experiments, average value of the % TI and % weight loss for the blank and solids added experiments were calculated separately. For example, in Table B l . l there are 4 blank experiments (sample 3 in each run) and 8 M0S2 added experiments (samples 1&2 in each run). Thus, for this table, the average and standard deviations were calculated as follows: 135 Appendix B %TI average (Blank) = y = - 2.94 + 2.56 + 2.44 + 2.01 = 2.485% The standard deviation of a sample of n measurements is given by: l a - y ) 2 i (2.94-2.485)2 +(2.56-2.485)2 +(2.44 -2.485)2 +(2.01 -2.485)2 3 '=1 = 0.38 n-1 The same calculations were carried out for the %TI in the solids added cases and the weight losses in all cases. Small-Sample (l-ce)xl00% confidence interval for population mean,|i Assumption: The population from which the sample is selected has an approximate normal distribution. A confidence interval for p. (population interval) is constructed based on the t distribution, Le., if y is the sample mean, then the interval which encloses \i (the population mean) by (l-a)100% confidence is: where the distribution of t is based on (n-1) degrees of freedom and n is the number of observations in the sample. For example, in Table B1.1 and %TI (blank): n = 4, y = 2.485 and s = 0.38. For a = 0.05 and 3 degrees of freedom, from the t test table (Mendenhill and Sincich, 1992, page 903): t<x/2 = to.025= 3.182 Thus, the 95 % confidence interval which encloses the mean value is (Bl.l) 136 Appendix B 1= [2.485 - 3.182X 0.38/A/4 , 2.485 + 3.182x 0.38/^4 ] = [2.485 - 0.6045,2.485+0.6045] s )is In most of the tables in this appendix, only the value of the second part (i.e., t^x^-— recorded. For example in Table B 1.1, only 0.61 (=0.6045) is recorded. Testing the Difference Between two population means: independent samples. Small-sample test of hypothesis about (}Li-\Li) for independent samples: Consider independent random samples from two populations with means Hi and pi2. When the samples sizes are small (ie., ni,n 2 < 30), a test of hypotheses for the difference between the population means (Hi-p:2) is based on the t test. Assumptions: 1. The populations from which the samples are selected both have approximately normal relative frequency distributions. 2. The random samples are selected in an independent manner from the two populations. In order to determine if the means of the two samples are different or not within (1-a) 100 % confidence interval (or at the a level of significance), start with the null hypothesis and assume that they are not different. If Si and S 2 are the standard deviations and yi and y 2 are the means of populations 1 and 2, respectively, then the t test is as follows: a) If n i = n 2 = n: Degrees of freedom: v - n t+ n2-2 = 2.(n-l) , = <>OlL (B1.2) 137 Appendix B From the t table, t an is obtained, and from Eq. B1.2, t is calculated. Then the Null hypothesis is rejected if 11 | > t an • b) If ni * n 2 Degrees of freedom: t = yry2 V = ni 112 2 2 ( 2 > L ( 2 \ \ ) —+• n, - 1 n 2 - l (B1.3) (B1.4) Note: the value of v will generally not be an integer. Therefore.v should be rounded to the nearest integer in order to use the t table. From the t table, tan. is obtained and from Eq. B 1.3, t is calculated: Then the Null hypothesis is rejected if 1 1 | > t an • For example, Table B1.2 includes 5 runs, which contain 5 blank samples (sample 3 in each run) and 10 MoS2-added samples (samples 1 and 2 in each run). Therefore, n< = 5 ,n 2 = 10 and the average values and standard deviations were calculated as shown in the following table: Average %TI %Wt Loss Std. Dev.%TI Std. Dev. %Wt. Loss Blank 2.485 54.14 0.76 2.03 MoS 2 added 3.125 50.56 1.13 1.42 138 Appendix B In order to find out if there is any difference between the blank sample and MoS2-added samples for either the coke yield or weight loss results, t test statistics were carried out as follows: 1) t test for %TI According to Eq. B1.4 ,0J6 2 1.13\ 2 ( + ) 5 10 f0.762_Y f U 3 _ 2 V 10 = 11.5262 Rounded down to the nearest integer: v=l Looking up in the t table statistics for a=0.05 level or (t ^ = t 0 . 0 2 5 ) and 11 degrees of freedom: to/2 = 2.201 Also, calculating t from Eq. B1.3 yields: = 2.0288 t< tan., therefore, is not at the rejection limit of the null hypothesis, and hence mean values of the two groups ( %TI for blank and added solids) are not statistically different. 139 Appendix B 2) t test for %wt loss: According to Eq. B1.4: 2.262 ( 2.35* (2.26"] 2 f 2.352 "I 5 J 4- . { 10 J — 4 9 Rounded down to the nearest integer: v=8 Looking up in the t table statistics for cc=0.05 level or (t «/2 = t 0.025) and 8 degrees of freedom: t ^ = 2.306 Also, calculating t from Eq. B1.3 gives: = 0.7979 t< t«/2, therefore, is not at the rejection limit. The null hypothesis is true, and mean values of the two groups (% Wt loss for blank and added solids) are not statistically different. 140 Appendix B Effect of M0S2 at different times and temperatures. Table B l . l Tube reactor experiments with M0S2 added. Experiments at 390 °C and 200 ml/min nitrogen purge. Run -Sample %MoS2 Sample wt.(g) Coking Timefh) TI(g) % TI % Wt Loss 6-1 2 1.2929 2.5 0.058 2.54 52.84 6-2 2 1.587 2.5 0.1054 4.74 51.94 6-3 0 1.432 2.5 0.0421 2.94 57.14 6-4 2 1.7469 0 0.044 0.53 0 6-5 2 0.9311 0 0.0061 0.66 0 7-1 1.96 13806 2.5 0.0796 3.91 48.76 7-2 1.96 1.6715 2.5 0.0648 1.96 48.69 7-3 0 1.452 2.5 0.0372 2.56 53.27 7-4 1.96 2.0149 0 0.042 0.13 0 7-5 1.96 13128 0 0.0346 0.70 0 9-1 1.92 1.9261 2.5 0.0962 5.09 50.73 9-2 1.92 1.9904 2.5 0.0384 1.97 50.92 9-3 0 1.2559 2.5 0.0306 2.44 53.52 9-4 1.92 1.55492 0 0.0112 0.74 0 9-5 1.92 1.5338 0 0.0093 0.62 0 10-1 1.95 1.6046 2.5 0.0401 2.55 50.18 10-2 1.95 1.4749 2.5 0.032 2.20 5037 10-3 0 1.164 2.5 0.0234 2.01 52.63 10-4 1.95 2.1188 0 0.0137 0.66 0 10-5 1.95 0.9399 0 0.0062 0.67 0 Average values and standard deviations: % TT % Wt Loss Std.Dev. % WtLoss Std.Dev. % T I Blank 2.485 54.14 2.03 0.38 MoS 2 added 3.125 50.56 1.42 1.27 Statistical analysis for 95% confidence interval Small Sample 95% Confidence Interval Blank MoS 2 added % TI 0.61 1.06 % Wt Loss 3.24 1.19 141 Appendix B Comparison of the results with the four blank experiments. t test statistic with 95% confidence interval Toluene Insolubles u (degrees of freedom ) = 9.0743 I Rounded down to the nearest integer = 9 10.025 - 2.262 t = -1.3096 Therefore, t is not in the rejection region and the average % TI value is not statistically different from the blank case at a 0.05 level of significance. WeiehtLoss v (degrees of freedom) = 1 4.5309 | Rounded down to the nearest integer = 4 t 0.025 — 2.776 t= -0.5639 Therefore, t is not in the rejection region and the average weight loss value is not statistically different from the blank case at a 0.05 level of significance. 142 Appendix B Table B1.2 Tube reactor experiments with MoS 2 added. Experiments at 390 °C and 200 ml/min nitrogen purge. Run -Sample % MoS2 Sample wt.(g) Coking Time (h) TI(g) %TI % Wt Loss 1-1 1.75 1.4782 3.5 0.063 2.54 52.84 1-2 1.75 1.4779 3.5 0.049 4.74 51.94 1-3 0 1.9688 3.5 0.0562 2.94 57.14 2-1 1.84 1.7942 3.5 0.0925 3.91 48.76 2-2 1.84 13096 3.5 0.0704 1.96 48.69 2-3 0 1.7845 3.5 0.0797 2.56 53.27 2-4 1.84 L5205 0 0.0368 0.13 0 2-5 0 1.2759 0 0.005 0.70 0 3-1 2 2.0371 3.5 0.0951 5.09 50.73 3-2 2 2.0092 3.5 0.0798 1.97 50.92 3-3 0 1.6234 3.5 0.0733 2.44 53.52 3-4 2 1.6308 0 0.0446 0.74 0 3-5 0 0.5903 0 0.0016 0.62 0 4-1 2 2.5608 3.5 0.1229 2.55 50.18 4-2 2 1.8609 3.5 0.0987 2.20 5037 4-3 0 1.8449 3.5 0.0683 2.01 52.63 4-4 2 1.4908 0 0.0378 0.66 0 4-5 0 1.2322 0 0.0057 0.67 0 5-1 1.98 2.5608 3.5 0.0977 3.57 55.4 5-2 1.98 1.8609 3.5 0.1079 3.35 55.1 5-3 0 1.8449 3.5 0.0662 4.01 573 5-4 1.98 1.4908 0 0.0423 0.39 0 5-5 0 1.2322 0 0.0028 0.30 0 Average values and standard deviations: % TT % Wt Loss Std.Dev. % Wt Loss Std.Dev. % TI Blank 2.79 55.46 2.26 0.76 MoS 2 added 3.18 54.86 2.35 1.13 Statistical analysis for 95% confidence interval: Small Sample 95% Confidence Interval Blank MoS 2 added %TI 0.94 0.81 % Wt Loss 2.80 1.68 143 Appendix B Comparison of the results with the three blank experiments. t test statistic with 95 % confidence interval Toluene Insolubles u (Degree of freedom) = 11.5262 Rounded down to the nearest intege r= 11 • 0.025 ~ t = 2.201 -0.7912 Therefore t is not in the rejection region and the average % TI value is not statistically different from the blank case at a 0.051evel of significance. Weight Loss D ODegree of freedom) = | 8.3912 Rounded down to the nearest integer = 8 t 0.025 — 2.306 t = -0.3112 Therefore, t is not in the rejection region and the average weight loss value is not statistically different from the blank case at a 0.05 level of significance. 144 Appendix B Table B1.3 Tube reactor experiments with MoS2 added. Experiments at 390 °C and 200 ml/min nitrogen purge. Run -Sample %MoS2 Sample wt(g) Coking Time(h) TI(g) % TI % Wt Loss 13-1 1.96 1.2916 4.5 0.0726 5.73 60.8 13-2 1.96 13555 4.5 0.0805 6.06 60.79 13-3 0 13082 4.5 0.0788 6.02 62.69 13-4 1.96 L5709 0 0.0017 0.11 0 13-5 1.96 1.4619 0 0.0013 0.09 0 14-1 2.02 1.8482 4.5 0.1052 5.81 60.72 14-2 2.02 1.5051 4.5 0.1192 6.07 61.21 14-3 0 13883 4.5 0.0788 5.55 63.22 14-4 2.02 1.4106 0 0.0067 0.49 0 14-5 2.02 1.4094 0 0.007 0.51 0 15-1 2 1.0776 4.5 0.0623 5.91 61.36 15-2 2 13842 4.5 0.0842 6.21 61.1 15-3 0 1.4558 4.5 0.0733 6.20 62.82 15-4 2 13662 0 0.0058 0.44 0 15-5 2 1354 0 0.0055 0.42 0 Average values and standard deviations: % TI % Wt Loss Std.Dev. % Wt. Loss Std.Dev. %TI Blank 5.92 62.84 0.28 0.34 MoS 2 added 5.97 60.99 0.26 0.18 Statistical analysis for 95 % confidence interval Small Sample 95% Confidence Interval Blank MoS 2 added %TI 0.84 0.19 % Wt Loss 0.69 0.28 145 Appendix B Comparison of the results with the three blank experiments. t test statistic with 95 % confidence interval Toluene Insolubles X) (Degree of freedom) = | 2.5699 |Rounded down to the nearest integer = 2 10.025 - 4.303 t = -0.2394 Therefore, t is not in the rejection region and the average % TI value is not statistically different from the blank case at a 0.05 level of significance. Weight Loss u (Degree of freedom) = 1 3.9048 | Rounded down to the nearest intege r = 3 10.025- 3.182 t = -0.2600 Therefore, t is not in the rejection region and the weight loss average value is not statistically different from the blank case at a 0.05 level of significance. 146 Appendix B Table B2 Silica with the average size of 3 um added. Experiment at 390 °C and 200 ml/min nitrogen purge Run -Sample %Si02 Sample wt(g) Coking Time (h) TI(g) % TI % Wt. Loss 1-1 2.03 1.4067 4.5 0.0862 4.18 57.74 1-2 2.03 1.5935 4.5 0.1043 4.61 58.19 1-3 0 1.4484 4£ 0.0657 4.54 59.89 1-4 2.03 1.4624 0 0.0383 0.60 0 1-5 2.03 13946 0 0.0344 0.45 0 2-1 2 13481 4.5 0.0788 4.96 57.68 2-2 2 1.6993 4.5 0.1025 4.54 58.02 2-3 0 13941 4.5 0.0686 4.92 59.5 2-4 2 13275 0 0.0285 0.15 0 2-5 2 1.8023 0 0.0391 0.17 0 3-1 2 2.1036 4.5 0.1386 4.68 58.226 3-2 2 1.7453 4.5 0.1262 534 58.82 3-3 0 1.5754 4.5 0.0886 5.62 61.18 3-4 2 43813 0 0.0327 037 0 3-5 2 1.0173 0 0.0228 0.25 0 4-1 2 13485 4.5 0.099 5.52 59.48 4-2 2 1.8764 4.5 0.1404 5.59 59.76 4-3 0 1.203 4.5 0.0695 5.78 62.43 4-4 2 0.9751 0 0.0336 032 0 4-5 2 1.8481 0 0.0438 038 0 Average values and standard deviations: Average % T I % Wt Loss Std.Dev. % Wt Loss Std.Dev. % TI Blank 5.21 60.75 133 0.59 SiOzadded 4.8 58.49 0.78 1.49 Statistical analysis for 9 5 % confidence Interval: Small Sample 95% Confidence Interval Blank Si0 2 added %TI 0.93 0.43 % Wt Loss 2.12 0.66 147 Appendix B Comparison of the results with the four blank experiments. t test statistic with 95% confidence interval Toluene Insolubles \) (Degree of freedom) = | 5.3804 |Rounded down to the nearest integer = 9 t 0 . 0 2 5 - 2.262 t = 0.8427 Therefore t is not in the rejection region and the average % TI value is not statistically different from the blank case at a 0.05 level of significance. WeieM Loss •ofDegree of freedom) = 1 4.0793 iRounded down to the nearest integer = 4 t 0.025 = 2.776 t = 0.4024 Therefore t is not in the rejection region and the average weight loss value is not statistically significantly different from the blank case at a 0.05 level of significance. 148 Appendix B Table B3 Silica with the average size of 6.5 |im added. Experiment at 390 °C and 200 ml/min nitrogen purge. Run -Sample % Silica Sample wt(g) Coking Time 00 TI(g) % T I % Wt Loss 1-1 1.8802 4.5 0.1304 5.00 58.41 1-2 2.04 0.9466 4.5 0.0742 5.81 61.08 1-3 0 13548 4.5 0.0764 5.64 61.52 1-4 2.04 1.638 0 0.0388 034 0 1-5 2.04 2.007 0 0.0509 0.51 0 2-1 1.98 1.4893 4.5 0.097 4.62 58.51 2-2 1.98 1.435 4.5 0.096 4.60 58.89 2-3 0 0.8971 4.5 0.0473 5.27 62.24 2-4 1.98 1.2618 0 0.0317 0.53 0 2-5 1.98 1.4243 0 0.0346 0.45 0 3-1 2.02 2.1608 4.5 0.0167 5.86 60.61 3-2 2.02 2.1373 4.5 0.0172 6.15 59.94 3-3 0 1.4426 4.5 0.0912 632 62.98 3-4 2.02 1.4052 0 0.033 033 0 3-5 2.02 1.2756 0 0.025 0 4-1 2.03 2.5925 4.5 0.1604 4.22 60.01 4-2 2.03 1.7556 4.5 0.1258 5.22 58.77 4-3 0 1.4197 4.5 0.077 5.42 61.08 4-4 2.03 1.0876 0 0.0269 0.43 0 4-5 2.03 1.087 0 0.027 0.44 0 5-1 2.06 1.5265 4.5 0.1043 4.87 59.98 5-2 2.06 1.5125 4.5 0.1112 5.40 59.7 5-3 0 1.1952 4.5 0.07 5.86 61.86 5-4 2.06 1.6417 0 0.0445 0.66 0 5-5 2.06 1.6413 0 0.0399 038 0 Average values and standard deviations: %TI % Wt Loss Std.Dev. % Wt Loss Std.Dev. % TI Blank 5.70 61.94 0.72 0.41 Si0 2 added 5.08 59.63 0.91 0.63 Statistical analysis for 95% confidence interval Small Sample 95% Confidence Interval Blank Si0 2 added %TI 0.51 0.46 % Wt Loss 0.90 0.65 149 Appendix B Comparison of the results with the five blank experiments. t test statistic wi th 95% confidence interval Toluene Insolubles x> (Degree of freedom) = | 11.7182 |Rounded down to the nearest integer = 1 1 •0.025-t = 2.201 2.2922 Since the calculated t value of t falls in the rejection region, the statistical analysis indicates that the mean response is different from that of the blank case at the 0.05 level of significance. Weight Loss •o (Degree of freedom) | 10.0339 |Rounded down to the nearest integer = 10 t a o „ = 2.228 t = 1.4321 Therefore t is not in the rejection region and the average weight loss value is not statistically different from the blank case at a 0.05 level of significance. 150 Appendix B Table B4 summarizes the experiments, for MoS 2 assays. Also, an example of the calculations is given. Table B4 Toluene insoluble results corrected based on the MoS2 assay. Sample No. WLofthe sample (g) Wtofthe TI te) Coking time(h) % Recovery MoS 2 MoS 2 added(g) 1 2.1387 0.1609 4.5 91 0.0427 2 1.6180 0.0620 2.5 62.4 0.0321 3 1.9811 0.1673 4.5 76.4 0.0399 4 1.1861 0.0503 2.5 92 0.0239 5 1.6533 0.0433 0 83 0.0333 Sample No. MoS 2 recovered on the filter based on analysis % TI based on 100 recovery %TI based on actual recovery Loss 1 0.03886 5.636 5.819 59.03 2 0.02003 1.887 2.648 50.56 3 0.03051 6.562 7.046 61.47 4 0.02199 2.503 2.668 53.83 5 0.02763 0.615 0.967 0 Sample calculation: Sample 4 Based on the definition of the %TI Weight of toluene insolubles - Weight of added solid Weight of initial sample - Weight of added solid 151 Appendix B a) Based on 100% recovery: %77 = 0.053 - 0.02391 1.1861-0.02391 = 2.503% b) Based on actual recovery: According to the M0S2 assay (Table B4) 92% of the initial added M0S2 was recovered on the filter, therefore the real amount of M0S2 which should be subtracted from the toluene insolubles is as follows: Actual amount of M0S2 on the filter (g) = %Recovery x Amount of TT initially added (g) For sample 4 it would be equal to: Actual amount of M o S 2 on the filter (g) = 0.92 x 0.02391= 0.02199 % TI based on the actual recovery = 0.053 - 0.02199 0.1861-0.02391 = 2.668% 152 Appendix C A P P E N D I X C - Summary of Blank and Solids Added Experiments in 100 g Autoclave Reactor This appendix summarizes the coke yield and weight loss results for 100 g blank, MoS 2 , and S i 0 2 samples in the autoclave reactor. A Sample calculations for the blank and a so lid-added (i.e. M0S2 case) is also given. Sample calculations (Run-o-6): a) Blank sample: Weight of the sample before reaction = 100 g. Weight of the sample after reaction = 69.8 g. Toluene insoluble from washing liquid = 0.0541 g. Toluene insoluble from liquid product = 0.297 g. From the definition of %Recovery: % Recovery = Weight of the product + Toluene insoluble in washing Weight of the feed xlOO % Recovery = 69.8 + 0.0541 100 xlOO = 69.85% % Wt. loss = 100-% Recovery = 100-69.85 = 30.15 % % n = TI in vessel + TI in washing Weight of the feed xlOO %TI = 0.297 + 0.0541 100 xlOO = 0.3511% 153 Appendix C b) Sohd added (MoS 2 added, Run-o-7) Weight of the sample before reaction = 100 g. Weight of the sample after reaction = 71.304 g. Toluene insoluble from washing liquid = 0.1041 g. Toluene insoluble from liquid product = 2.1270 g. From the definition of %Recovery: Weight of the product + Toluene insoluble in washing % Recovery = - - x 100 Weight of the feed 71.3041 + 0.1041 1 A A _ 1 Q 1 C 7 % Recovery = x l00 = 71.31% 100 % Wt. loss = 100-71.31= 28.69 % % M o S 2 added = 1.85 % of the initial sample = 1.85 g in the initial 100 g sample. y , _ _ Tl in vessel + TI in washing - Weight of the solid added ^ Weight of the feed - Weight of the solid added % T I m 2.127 + 0 .104 1 -1 .85 x = ( _ 1 100 100 154 Appendix C Table CI Toluene insolubles and weight loss for 100 g blank sample in autoclave reactor 390 °C, Blank coking, purge with 200 ml/min nitrogen, 100 gram sample Time (h) Filters TI (g) % TI % Wt loss liquid product Liquid washing Run-0-6 2.5 0.45p.m 0.2pm 0.2756 0.0214 0.0502 0.0039 0.351 30.15 TI total (g) 0.3511 Filters TI (g) % TI % Wt loss Liquid product Liquid washing Run-0-8 3.5 0.2pm 0.2729 0.0851 0.358 33.62 TI total (g) 0.0358 Filters TI (g) % TI % Wt loss Liquid product Liquid washing Run-0-13 4.5 8 (im 1.2 um 0.22 um 2.9839 1.3536 0.0144 0.7654 0.1127 5.23 39.98 TI total (g.) 5.23 Filters TI (g) % TI % Wt loss Liquid product Liquid washing Run-0-14 4.5 8 urn 1.2 um 0.22 um 2.8821 1.2897 0.0118 0.7285 0.1879 5.1 38.81 TI total (g) 5.10 155 Appendix C Table C2 Toluene insolubles and weight loss for 100 g sample with MoS 2 added in the autoclave reactor % MOS 2 390 °C, MoS 2 added, purge with 200 ml/min nitrogen, 100 gram sample Time (h) Filters l i fe) % Wt loss liquid product Liquid washing Run-0-4 1.84 2.5 0.45um 1.6452 0.0731 038 28.69 0.2um 0.4926 0.0023 TI total (g) 2.1378 Filters TI(g) % TI % Wt loss Liquid product Liquid washing Run-0-7 1.85 2.5 0.45um 1.5661 0.0714 038 28.72 0.2um 0.5609 0.0327 TI total (g) 2.2311 FUters TI(g) % TI % Wt loss Liquid product Liquid washing Run-0-11 1.84 3.5 8 um 1.2 Um 0.22 um 1.6568 0.6436 0.1643 0.1337 0.61 33.37 TI total (g) 2.5984 156 Appendix C Table C3 Toluene insolubles and weight loss for 100 g sample with Si0 2 added in the autoclave reactor %Si02 390 °C, Si0 2 added, purge with 200 ml/min nitrogen, 100 gram sample Time (h) Filters 11(g) % TI % Wt loss % 1.81 liquid product Liquid washing Run-0-4 of IS 3 um 0.45um 1.5574 0.0106 0.41 28.3 silica 0.2um 0.551 0.0917 TI total (g) 2.2051 Filters TI(g) % Wt loss % 2.01 liquid product Liquid washing Run-0-5 of IS 6.5 |im 0 . 4 5 u M n 2.1958 0.108 034 29.29 silica 0.2um 0.0333 0.0081 TI total (g) 2.3452 Filters TI(g) % TI % Wt loss % 2 Liquid product Liquid washing Run-0-10 of 3.5 8 um 1.7851 3 |xm 1.2 um 0.4561 0.47 33.08 silica 0.22 um 0.099 0.1254 TI total (g) 2.4656 157 Table C 4 Heat -Up data of the autoclave reactor Heat-up data Time (min) Temp (°C) Time Temp (°C) 0 25 43 260 12 126 44 265 13 133 45 272 14 137 46 277 15 140 47 282 16 140 48 288 17 140 49 292 18 141 50 297 19 142 51 302 20 144 52 307 21 148 53 312 23 160 54 317 24 167 55 322 25 173 56 327 26 178 57 332 27 182 58 337 28 187 59 342 29 192 60 347 30 197 61 352 31 202 62 357 32 208 63 361 34 215 66 371 36 225 67 373 37 230 70 379 39 240 71 381 40 245 72 385 41 250 73 388 42 255 74 390 Table C5 Cool down data of the autoclave reactor Cooling in air Cooling in water Time (min) Temp(°C) Time(min) Temp(°C) 0 390 30 157 1 375 32 145 2 360 33 140 4 333 34 136 5 321 35 132 6 309 36 129 7 298 37 125 9 279 38 120 10 270 39 114 11 260 40 111 13 242 41 107 15 226 42 104 17 217 43 102 19 206 45 96 20 201 46 93 21 196 47 89 22 191 48 87 23 186 49 86 24 181 50 84 25 176 55 74 26 173 56 72 27 169 57 70 28 165 58 68 29 157 59 67 30 157 60 66 Appendix D APPENDIX D-Summary of Blank, MoS2 and Si0 2 Added Experiments in 50 g Autoclave Reactor This appendix summarizes the results of the blank, MoS 2 and S i 0 2 added experiments for 50 g samples in the autoclave reactor. Table D l a Toluene insolubles and weight loss for the blank 50 g. sample in the autoclave reactor 390 °C, 2.5 hours coking, purge with 200 ml/min nitrogen, 50 g of sample filters TI te) Liquid product Liquid washing % TI %Wt Loss Run-O-29 Blank 0.45um 0.135 0.0535 0.403 40.2 0.2um 0.0105 0.0025 Total TI te) 1.2038 filters TI te) Liquid product Liquid washing % TI % Wt Loss Run-O-36 Blank 0.45|im 0.1151 0.0452 0.34 46.11 0.2um 0.0049 0.0023 Total TI te) 1.197 160 Appendix D Table D l b Toluene insolubles and weight loss for the blank 50 g. sample in the autoclave reactor 390 °C, 4.5 hour coking, purge with 200 ml/min nitrogen, 50 g sample Filters TI(g) % TI % Wt loss liquid product Liquid washing Run-0-23 Blank 0.45um 1.5247 0.2837 3.83 47.6 0.2um 0.1029 0.0035 TI total 1.9148 Filters TI (g) % TI % Wt loss Liquid product Liquid washing 8 Um 0.3016 5 Um 0.0884 Run-0-21 Blank 3 um 0.1197 0.2311 4.22 45.04 1.2 um 0.126 0.65 um 0.263 0.22 um 0.9319 0.0513 Total TI (g) 2.113 Filters TI (g) % TI % Wt loss Liquid product Liquid washing Run-0-43 Blank 0.45um 1.8499 0.2506 4.21 53.29 0.2um 0.0211 0.0043 Total TI (H) 2.1048 Filters TI(g) % TI % Wt loss Liquid product Liquid washing Run-0-46 Blank 0.45um 1.5574 0.1848 3.96 46.02 0.2um 0.2295 0.0087 Total TI (g) 1.9804 161 Appendix D Table D2 Toluene insolubles and weight loss for 50 g sample with MoS 2 added in the autoclave reactor. 390 °C, 2.5 hours coking, purge with 200 ml/min nitrogen, 50 g of sampl e filters TI(g) < Liquid product Liquid washing % TI % Wt Loss MoS 2 Run-O-30 2% added 0.45micron 0.9985 0.1858 0.413 54.2 0.2 micron 0.0087 0.0108 Total TI(g) 1.2038 filters TI(g) Liquid product Liquid washing % TI %Wt Loss MoS 2 Run-O-34 2% added 0.45micron 0.9722 0.0954 0.4 46.29 0.2 micron 0.0719 0.0575 Total TI (g) 1.197 390 °C, 4.5 hours coking, purge with 200 ml/min, 50 g of sample filters TI(g) Liquid product Liquid washing % TI %Wt Loss 8 um 0.7803 0.3827 MoS 2 5 pi m 0.173 0.1192 Run-O-19 2% added 3 urn 0.1949 1.2 urn 0.2025 4.56 48.79 0.65 um 0.5474 0.22 um 0.8383 0 Total TI(g) 3.2383 filters TI(g) Liquid product Liquid washing % TI %Wt Loss MoS 2 Run-O-25 2% added 0.45 |Xm 1.5305 0.3479 4.0829 45.88 0.2 um 0.8081 0.0101 Total TI(g) 2.9966 162 Appendix D Table D3 Toluene insolubles and weight loss for 50 g. sample with Si0 2 added in the autoclave reactor. 390 °C, 2.5 hours coking, purge with 200 ml/min nitrogen, 50 g of sample filters TI(g) Liquid product Liquid washing % TI % Wt Loss Silica Run-O-31 3 um 0.45pm 0.9076 0.1222 0.32 47.53 2% added 0.2pm 0.1161 0.009 Total TKe) 1.1549 filters TI(g) Liquid product Liquid washing % TI %Wt Loss Silica Run-O-35 3 um 0.45pm 0.9225 0.1273 0.34 47.07 2% added 0.2pm 0.0871 0.0352 Total TI (e) 1.1721 390 °C, 4.5 hours coking, purge with 200 ml/min, 50 g of sample filters TI(g) Liquid product Liquid washing % TI %Wt Loss Silica Run-O-26 3 Urn 0.45pm 2.4966 0.3392 4.93 53.24 2% added 0.2pm 0.5422 0.0399 Total TKe) 3.1418 filters TI(g) Liquid product Liquid washing % TI %Wt Loss 8 pm 0.4697 Silica 5 um 0.2462 Run-O-22 3 um 3 um 0.4312 0.3786 2% added 1.2 um 0.1934 4.62 51.59 0.65 um 0.1113 0.22 um 1.4107 0.0229 Total TI(e) 3.6299 163 J Appendix E APPENDIX E-Summary of Statistical Analysis for 50 g Blank and Solids Added Samples in the Autoclave Reactor. This appendix tabulates all the coke yield and weight loss results for 50 g. autoclave samples coked for 4.5 hour experiments at 390 °C, including blank, M0S2, Kaolin, SiC«2, and native solids added experiments. Average values, standard deviations and 95% confidence intervals are also given. In addition, Y test was also performed to determine if there is any difference between the results of the blank experiments and those in which solids were added under otherwise identical conditions. Table E l Coking experiments for 50 g blank sample in the autoclave reactor. Blank sample average values, standard deviations, and 95 % confidence intervals Average Standard Deviations 95% confidence interval % TI 4.05 0.19 0.31 % Wt Loss 47.98 3.69 5.91 164 Appendix E Table E2 Coking experiments on 50 g bitumen samples with 2 wt%Alberta clay added 390 °C, 4.5 hour coking, purge with 200 ml/min nitrogen, SO g sample Filters TI(g) % TI % Wtloss liquid product Liquid washing Run-0-37 %2 clay 3um 2.2356 added < 0.45um 0.7587 0.3196 4.74 51.35 45um 0.2um 0.0111 0.0053 TI total (g) 3.3303 Filters TI(g) % TI % Wt loss Liquid product Liquid washing Run-0-39 %2 clay 3jim 0.8667 added < 0.45nm 1.3678 0.4149 4.24 48.76 45u.m 0.2um 0.4301 0.0042 TI total (g) 3.0837 Filters TI(g) % TI % Wt loss Liquid product Liquid washing Run-0-40 %2 clay added < 0.45|im 2.0357 0.3267 4.19 50.93 45u.m 0.2um 0.6902 0.0058 TI total (g) 3.0584 Average values, standard deviations and 95% confidence intervals: Average Standard deviations 95% Confidence interval % TI 4.39 0.35 0.65 % Wt Loss 50.35 1.39 2.55 Comparing the results with the four blank experiments (See Appendix B for procedures and sample calculations) t test statistic with 95% confidence interval Toluene Insolubles -o(Degree of freedom) 2.8907 | Rounded down to the nearest integer = 2 t 0.025 - 4.303 t = 1.5069 Therefore t is not in the rejection region and the average value is not statistically different from that obtained in the blank case at the 0.05 level of significance 165 Appendix E Weight Loss t> (Degree of freedom) = 4.0269 |Rounded down to the nearest integer = 4 t 0.025 - 2.776 t = 1.1762 Therefore t is not in the rejection region and the average value is not statistically different from that obtained for the blank case at the 0.05 level of significance Table E3 Coking experiments on 50 g bitumen samples with 2 wt % Kaolinite added 390 °C, 4.5 hour coking, purge with 200 ml/min nitrogen, 50 g sam )IeRun-0-37 2% Kaolinite added < 45u.m Filters TI(g) % TI % Wt loss 0.45u.m 0.2um liquid product Liquid washing 4.23 49.8 2.6438 0.0542 0.3113 0.0682 Total TI(g) 3.0775 Run-0-39 2% Kaolinite added < 45(xm Filters TI(g) % TI % Wt loss 0.45u.m 0.2|im Liquid product Liquid washing 3.93 48.8 2.5434 0.0893 0.2818 0.0141 Total TI(g) 2.9286 Run-0-40 2% Kaolinite added < 45(xm Filters TIOO % TI % Wt loss 0.45|im 0.2|xni Liquid .product Liquid washing 4.25 54.48 2.6795 0.6902 0.3388 0.02 Total TI(g) 3.0934 166 Appendix E Average values, standard deviations and 95 % confidence intervals Average Standard deviations 95 % Confidence Interval %TI 4.136 0.18 0.33 % Wt Loss 51.02 3.03 5.57 Comparing the results with the four blank experiments( See Appendix B for procedures and sample calculations) t test statistic with 95% confidence interval Toluene Insolubles \)(Degree of freedom) = 4.6372 Rounded down to the nearest integer = 4 0.025 • 2.776 0.6088 Therefore t is not in the rejection region and the average value is not statistically different from that obtained in the blank case at the 0.05 level of significance. Weight Loss -o (Degree of freedom) = 4.8887 |Rounded down to the nearest integer = 4 t 0.025 - 2.776 t = 1.1954 Therefore t is not in the rejection region and the average value is not statistically different from that obtained in the blank case at the 0.05 level of significance. 167 Appendix E Table E4 Toluene insolubles and weight loss of the 2 wt% Native solid added experiments 390 °C, 4.5 hour coking, mrge with 200 ml/min nitrogen, 50 g sample Run-0-54 2% Native clay added < 45um Filters TT(g) % Wt loss 0.45u.m 0.2um liquid product Liquid washing 3.02 47.84 2.797 0.0362 0.0147 0.2699 0.0037 0.0038 Total TI(g) 2.5070 Run-0-55 2% Native clay added < 45um Filters TI(g) %TI % Wt loss 3p.ni 0.45(Xm 0.2um Liquid product Liquid washing 2.95 46.81 2.1471 0.0173 0.0128 0.2568 0.0318 0.0089 TotalTT (g) 2.4747 Run-0-56 2% Native clay added < 45um Filters TI(g) %TI % Wt loss 3p.m 0.45|im 0.2um Liquid product Liquid washing 3.22 48.84 2.3059 0.0196 0.0089 0.2733 0.0041 0.0023 Total TI (g) 2.6141 Average values, standard deviations and 95% confidence intervals: Average Standard deviations 95% Confidence Interval %TI 3.06 0.14 0.26 % Wt Loss 47.83 1.02 1.86 Comparing the results with the four blank experiments( See Appendix B for procedures and sample calculations) 168 Appendix E t test statistic with 95% confidence interval Toluene Insolubles | 4.9938 JRounded down to the nearest integer=4 D (Degree of freedom) = 0.025 t = 2.776 -7.8794 Since the calculated t value of t falls in the rejection region, the statistical analysis indicates that the mean response is different from that of the blank case at the 0.05 level of significance. Weight Loss xi (Degree of freedom) = | 3.5815 |Rounded down to the nearest integer=3 t 0.025 - 3.182 t = -0.0775 Therefore t is not in the rejection region and the average value is not statistically different from that obtained in the blank case at the 0.05 level of significance. 169 Appendix F APPENDIX F - Effect of Changing the Native Solids Concentration This appendix summarizes the results of the native solid added experiments at different concentrations. Following to the each table there is a 't' test analysis table which includes the comparison between the blank sample (390 °C, 200 ml/min nitrogen purge and 4.5 hours coking, Table E l ) and the results in that table. For example't' test in Table F l is comparison between Table F l and Table E l results. Table F l Toluene insolubles and weight loss for 1 wt% Native solid added experiments. 390 °C, 4.5 hour coking, purge with 200 ml/min nitrogen, 50 g sample Filters TI(g) % TI % Wt loss liquid product Liquid washing Run-0-64 %1 day added < 0.45 p.m 1.8086 0.2175 3.1 47.35 45u.ni 0.2 um 0.0122 0.0051 TI total (g) 2.0434 Filters TI(g) % TI % Wt loss Liquid product Liquid washing Run-0-65 1% clay added < 0.45 \im 1.8837 0.205 3.25 45.38 45u.m 0.2 um 0.0293 0.0058 TI total (g) 2.1238 Filters TI(g) %TI % Wt loss Liquid product Liquid washing Run-0-66 1% clay added < 0.45 \im 1.7971 0.2328 3.11 47.52 45u.ni 0.2 um 0.017 0.0082 TI Total (g) 2.0551 Average values, standard deviations and 95% confidence intervals: Average Standard Deviations 95 % Confidence Interval % TI 3.15 0.08 0.15 % Wt Loss 46.75 1.19 2.19 170 Appendix F Comparing the results with the four blank experiments (see Appendix B for procedures and sample calculation). t test statistic with 95% confidence interval Toluene Insolubles i)(Degree of freedom) = | 4.2976 JRounded down to the nearest integer = 4 t 0.025 ' t = 2.776 -8.3302 Since the calculated t value of t falls in the rejection region, the statistical analysis indicates that the mean response is different from that of the blank case at the 0.05 level of significance. Weight Loss x> (Degree of freedom) = | 3.7804 |Rounded down to the nearest integer = 3 10.025 _ 3.182 t = -0.6249 Therefore t is not in the rejection region and the average value is not statistically different from that obtained in the blank case at the 0.05 level of significance. 171 Appendix F Table F2 Toluene insolubles and weight loss for 4 wt%Native solid added experiments. 390 °C, 4.5 hour coking, purge with 200 ml/min nitrogen, 50 g sample FQters TI(g) %TI % Wtloss 4% Native liquid product Liquid washing Run-0-67 clay added < 0.45 mm 2.9854 03275 2.76 46.11 45um 0.2 mm 0.053 0.0173 TI total (g) 3.3832 Filters TI(g) %TI % Wt loss 4% Native Liquid product Liquid washing Run-0-68 clay added < 0.45 mm 3.10136 03889 2.93 49.6 45um 0.2 mm 0.0518 0.011 TI total (g) 3.4653 Filters TI(g) %TI % Wt loss 4% Native Liquid product Liquid washing Run-0-69 clay added < 0.45 mm 3.0182 03153 2.73 4634 45uni 0.2 mm 0.0153 0.0144 TI total (g) 3.3632 Average values, standard deviations and 95% confidence intervals: Average Standard Deviations 95% Confidence Interval %TI 2.81 0.11 0.20 % Wt Loss 47.35 1.95 3.59 Comparing the results with the four blank experiments (see Appendix B for procedures and sample calculation). t test statistic with 95% confidence interval Toluene Insolubles | 4.7803 |Rounded down to the nearest intege r = 4| u(Degree of freedom) t 0.025 ' 2.776 •10.8258 172 Appendix F Since the calculated t value of t falls in the rejection region, the statistical analysis indicates that the mean response is different from that of the blank case at the 0.05 level of significance. Weight Loss -o (Degree of freedom) = | 4.6797 |Rounded down to the nearest integer = 3 * 0.025 - 3.182 t = -0.2915 Weight Loss x> (Degree of freedom) = | 3.7804 |Rounded down to the nearest integer = 3 10.025 - 3.182 t = -0.6249 Therefore t is not in the rejection region and the average value is not statistically different from that obtained in the blank case at the 0.05 level of significance. 173 Appendix G APPENDIX G- Effect of Nitrogen Purge Rate This appendix summarizes the results of the blank experiments at different purge rate (Table G1-G2). Second part (Table G3-G4) includes the results of the native solid added at the identical conditions (purge rate, etc.) to the first part. A sample of t test is also given. Table G l Toluene insolubles and weight loss results at 100 ml/min nitrogen purge 390 °C, 4.5 hour coking, purge with 100 ml/min nitrogen, 50 g sample Filters TI(g) | % TI % Wt loss liquid product Liquid washing j Run-0-47 Blank 0.45jMni 1.4113 0.1585 3.45 42.6 0.2(im 0.1485 0.006 Total TI(g) 1.7243 Filters TI(g) % TI % Wt loss Liquid product Liquid washing Run-0-48 Blank 0.45um 1.6314 0.1487 3.63 46.2 0.2um 0.0332 0.0059 TI Total (g) 1.8192 Filters Total TI (g) % TI % Wt loss Liquid product Liquid washing Run-0-49 Blank 0.45u,m 1.3006 0.1482 3.72 44.9 0.2u.m 0.406 0.0025 Total TI(g) 1.8573 Average values, standard deviations and 95% confidence intervals: Average Standard Deviations 95% Confidence Interval %TI 3.59 0.14 0.25 % Wt Loss 44.5 1.82 3.35 174 Appendix G Table G2 Toluene insolubles and weight loss results at 400 ml/min nitrogen purge 390 °C, 4.5 hour coking, purge with 400 ml/min nitrogen, 50 g sample Filters TI(g) %TI % Wt loss Uquid product Liquid washing Run-0-51 Blank 2.2319 0.45um 0.0261 0.1875 4.93 58.81 0.2um 0.0136 0.0023 Total TT(g) 2.4655 Filters TI (g) %TI % Wt loss Liquid product Liquid product Liquid washing Run-0-52 Blank 3u,m 1.9999 0.2504 0.45um 0.1233 0.0064 4.78 58.68 0.2um 0.0074 0.0014 Total TT(g) 23888 Filters TI(g) %TI % Wt loss Liquid product Liquid washing Run-0-53 Blank 3u,m 2.0335 0.3011 0.45um 0.0134 0.0021 4.71 60.79 0.2um 0.0059 0.0023 Total TT(g) 2.3575 Average values, standard deviations and 95% confidence intervals: Average Standard Deviations 95% Confidence Interval %TI 4.81 0.11 0.21 % Wt Loss 59.43 1.18 2.17 175 Appendix G Table G3 Toluene insolubles and weight loss results at 100 ml/min Nitrogen purge with 2wt% Native solid added 390° C, 4.5 hour coking, purge wi th 100 ml/min nitrogen, 50 g sample Filters T I ( g ) % T I % W t loss 2 % Native l iquid product L iqu id washing Run-0-57 clay added < 0.45um 1.7177 0.2051 3.12 44.18 45|xm 0.2p.m 0.631 0.0063 Total T I (g) 2.5601 Filters T I ( g ) % T I % W t loss 2 % Native L iqu id product L iqu id washing Run-0-58 clay 3p.ni 0.1148 added < 0.45u.m 2.1213 0.1864 2.97 43.62 45pjn 0.2pm 0.06 0.0042 Total T I (g) 2.4867 Filters T I ( g ) % I I % W t loss 2 % Native L iquid product L iqu id washing Run-0-59 clay added < 0.45p.m 2.2738 0.2436 3.26 47.49 45(xm 0.2p,m 0.0987 0.0108 Total TT(g) 2.6269 Average values, standard deviations and 95% confidence intervals: Average Standard Deviations 95% Confidence Interval %TI 3.11 0.15 0.27 % Wt Loss 45.1 2.09 3.84 Comparing the results with the three blank experiments (see Appendix B for procedures and sample calculation). t test statistic with 95% confidence interval Toluene Insolubles x> (Degree of freedom) = 10.025 = t = 2.776 -4.1672 176 Appendix G Since the calculated t value of t falls in the rejection region, the statistical analysis indicates that the mean response is different from that of the blank case at the 0.05 level of significance. Weight Loss u (Degree of freedom) = 4 10.025— 2.776 10.025 = 0.3746 Therefore t is not in the rejection region and the average value is not statistically different from that obtained in the blank case at the 0.05 level of significance. Table G4 Toluene insolubles and weight loss results at 400 ml/min Nitrogen purge with 2 wt% native solid added 390 °C, 4.5 hour coking, purge with 400 ml/min nitrogen, 50 g sample Filters TI(g) % TT % Wt loss 2% Native liquid product Liquid washing Run-0-60 clay added < 0.45um 3.1533 0.2739 4.47 54.03 45|im 0.2um 0.0149 0.0095 Total TI (g) 3.2367 Filters TI(g) % TI % Wt loss 2% Native Liquid product Liquid washing Run-0-61 clay added < 0.45um 2.6074 0.3202 3.86 52.95 45(im 0.2um 0.327 0.0068 Total TI (g) 2.9344 Filters TI(g) % n % Wt loss 2% Native Liquid product Liquid washing Run-0-62 clay 3(1 m 2.3997 0.3405 added < 0.45um 0.3417 0.01 4.22 58.09 45|im 0.2(im 0.013 0.0023 Total TI(g) 3.1072 177 Appendix G Average values, standard deviations and 95% confidence intervals: Average Standard Deviations 95% Confidence Interval % TI 4.18 0.31 0.56 % Wt Loss 55.02 2.71 4.98 Comparing the results with the three blank experiments (see Appendix B for procedures and sample calculation). t test statistic with 95% confidence interval Toluene Insolubles D (Degree of freedom) = t 0.025 • t = 2.776 -3.3411 Since the calculated t value of t falls in the rejection region, the statistical analysis indicates that the mean response is different from that of the blank case at the 0.05 level of significance. Weight Loss a) (Degree of freedom) = 4 10.025 - 2.776 t = -2.5833 Therefore t is not in the rejection region and the average value is not statistically different from that obtained in the blank case at the 0.05 level of significance. 178 Appendix H APPENDIX H - Derivation of V and TI as Functions of Time from Wiehe's Series Kinetic Model 1. Parallel Reaction Model This section provides the equations for Wiehe's (J993) parallel reaction kinetic model given in section 4.9.1. In the following equations, A© is the initial asphaltene concentration (wt%) and Ho represents the initial concentrations of the heptane solubles (wt%). The concentration of reactant asphaltenes, A+, reactant heptane solubles, and volatiles, V, are unaffected by whether coke is forming or not. Thus, fort £0, V = (l-a)HQ(l - e x p ( - ^ r ) ) + (l-m-n)AQ(l - exp(-KAt)) (H.3) A+ = AQ e x p C - i ^ O H+=H0exp(-KHt) (ELI) (H.2) During the coke induction period A* = oH0{\ - e x t f - i ^ O ) + mA0(l- e x p ( - i ^ 0 ) A = A++A* (H.4) (H.5) H* =nAo(l-exv(-KAt)) H = H++H* (H.6) (H.7) 77 = 0 (H.8) The coke induction period ends when 179 Appendix H A*=SL(H++H*) ^ Thereafter, A , is given by the above equation and H* = + 714,(1 - exp(KAt)) (H.10) = 1- V E [aH0 Aa + SjH0^-Kut)Am-nSL)A^(l-ex^-KAt))] (H.12) Using equation for the Cold Lake vacuum residuum at 400 °C, the following constants were used by Wiehe: KH = 0.013 min- 1 KA = 0.026 min*1 4,=25wt% ^ 0 = 7 5 w t % S_=0.49 2. Series Reaction Model This section includes the derivation of expressions for the %toluene insolubles and %volatiles (% weight loss) as a function of time according to Wiehe's series reaction kinetic model. We first start with the with the basic reaction equations of the model: H+ K h > bA+ + (1 - b)Vx (H.13) A+ — ^ U CA* + dH*+(l-c- d)V2 (H-14) 180 Appendix H A m . =  S°L ( # + +  H ' ) 0 p e " r e a C t O T (H.15) 4 n a x = ^ ~ - 4 c (H.16) For heptane solubles (H.17) Therefore, [ H + ] = if 0exp(-^r) From reaction H. 13 i dvx 4 * 1 . _ • H + 1-6 <ff </r ff' ^ ( l - f t ^ . i T =(1-A)^. i f 0 expC-^O (H.18) From reaction H. 14 1 dV2 _ dA+ Tc~^d~dT~~ dt CH-19) From reactions H. 13 and H. 14 - ——— = KAA+ — bKHH+ (EL20) dt A H From equations H. 19 and H.20 181 Appendix H 1 dV, l-c-d dt *- = KAA+ -bKffH+ (H.21) Then dV* dt = (l-c-d)(KAA+-bKffH+) (H.22) From reaction H. 14 dt dA* dA+ dt = -c^ = c{KAA*-bKHH+) (H.23) (H.24) The total volatiles V=Vi+V2. Thus, then from equations H. 18 and H. 22 dV dV, dV2 dt dt dt =(l-b)KHHQcxy(-KHt)+(l-c-d)(KAA+-bKHH+) = (l-b-(l-c- d)b)KH. Ho expQ-Kat) + {l-c-d)KAA+ = (l-2b + bc + bd)KH. H0 exp(-KH t) + (l-c-d)KAA+ (H.25) For an open reactor system dA' dt = 5 rdH+ dH*^ dt + dt (H.26) 182 Appendix H dC dA* dA max dt dt dt (H.27) _______ dt dt (H.28) From equations H.26-H.28 dTI dA dAmar dA max _ dt dt dt dt dH_ dH_ dt + dt (H.29) Insert equations H. 14 and H.24 into equation H.29 dTI dt = c(KAA+ -bKHH+)-S°L(-KHH+ +dKAA+ -dbKHH+) = (c-S°Ld)KAA+ +(S°L +dbSl —bc\KHH+ ={c-Sld)KAA+ +(S°L +dbSt°L —bcJKffHo exp(rKHt) (H.30) From equation H.20 dA+ dt Therefore, KAA+ - bKHH+ = KAA+ -bH0KH exp(-KHt) dA+ dt = -KAA+ + bH0Kff eM-Kfft) Take>'=y4+, x = t, <Z = KA, fi = bHoKH and y=-KH 183 Appendix H Then dy dt f ay- fSQxpirx) y = exp(-J cedx) x (J/?exp(rc).exp(J + c) = exp(-%) x (J /3exp(rx) exp(ax)dx + c) \ cc + r J (H.31) Therefore, A+ =exp(-KAt)x Whent=0, A+=Ao bKHH0 (bKHH0exp(KA-KH)t c = A0 K A ~ K H (H.32) (H.33) Inserting equation H.33 into equation H.32 yields A+ =exp(-KAt)x hf"H" exp((KH-KA)t) + AQ-~KA bKHH0 KH~KA (H.34) Inserting equation H.34 into equation H.25 then gives dV dt bKA (l-2b+bc + bd) + (l-c-d)—-^ KA~KH J + (l-c-d)KA bKHHQ ° KA-Ka\ expC-i^O (H.35) 184 Appendix H Integrating the above equation results in: V = |(26 - be - bd -1) + (c+d-1) ^ _ ^ J # o e x p ( - ^ r ) + (c + d-l) 4> cxp(-KAt) + K2 Also inserting equation H.34 into equation H.30 results in dTI dt =J(c-SL°d)KA K b H ° K +SL°H0(l + db)-bcH0jxKHexp(-*„0 (H.36) x ^ x cxp(-KAt) (H.37) Integrate the above relation 77 = SL°HQ(l + db)+bcH0 x exp(-.K i /f) ( c - 5 _ V ) ^ - ^ ^ - j xeXp(-^r) + ^ (H.38) Ki and K 2 are the integration constants. They were determined with the corresponding TI and V values at the end of induction time (Initial values). For example: At t = 0 (end of induction time) TIo = OATherefore, inserting t = 0 into equation H.38 results in: 185 Appendix H TI0=\(SL°d-c)KA H SL°H0(l + db)+bcH0 -(c-SL°d)\AQ Hj 1 = 0.4 As all other parameters in the above equation are known (from curve fitting data to the equation H.38) Ki is determined by solving the above equation. Same procedures were followed for calculating K2 with equation H.36 and corresponding weight loss value at the end of the induction time (t=0). 3. Rate constants for Cold Lake Bitumen This section summarizes the derivation of the rate constants for the Cold Lake bitumen, assuming that the corresponding values for the Cold Lake residue at 400 °C is also valid for the Cold Lake bitumen. At 400 °C for the Cold Lake residuum (Wiehe 1993): K A = 0.013 min1 and EA=147695.2 J/mol (35300 cal/g-mol) K H =0.026 min-1 and EH= 185351.2 J/mol (44300 cal/g-mol) According to the Arrehnius equation, if the rate constant (KTI) at the temperature Ti and activation energy (E,) for a reaction are given, then the rate constant can be calculated for a second temperature T2. That is, since 186 Appendix H KTi — AQ exp KT2 = AQ exp •E. \ 2J K. T2 (H.39) (H.41) (H.40) Therefore from the above values of K H and K A at 4 0 0 ° C for the cold lake residuum, then-values could be calculated for 3 8 0 ° C and 3 9 0 ° C . They are listed in Table HI. Table ELI Rate constants for Cold Lake residuum for different temperatures Temperature(0Q K^mm 1 ) KnCmin 1) 380 0.012 0.005 390 0.018 0.008 400 0.026 0.014 4. Curve Fitting to Determine Missing Parameters This section includes the procedures and the 'Mathcad T calculations used for curve fitting equations H.36 and H.37to the measured % weight loss and % toluene insolubles data, in order to obtain the stoichiometric coefficients b, c, d and the solubility limit ( S L ° ) . When equations H.36 and H.38 are examined, it can be seen that each contains two identical exponential terms, and that the coefficients of these terms contain the unknowns, b, c, d and SL° . The coefficients of the exponential terms were found directly by curve fitting the toluene insoluble and volatiles data separately, to a function which is a linear combination of the two exponential terms and a constant value (Corresponding to the integration constant in each equation). 187 Appendix H A system of equations is constructed by equating the coefficients of the exponential terms obtained by curve fitting to the expressions for coefficients of exponential terms in equations H.36 and H.38. Then these four unknowns were calculated by solving this system of non-linear equations. Example calculations are shown below for estimating the parameters b, c, d and SL° from the TI and weight loss results at 400 °C : Only the data obtained after the induction time were used, because equations H36 and H.38 only apply at times exceeding the induction time. T:=400°C umeTI400:= 0 30 60 90 120 150 180 210 240 TItest := 0.52 | 2.03 4.35 5.18 7.53 8.57 9.96 11.19 11.06 For the Cold Lake bitumen it is assumed (Table H. 1) kA := 0.026 kH:= 0.0134 It is also assumed that the target function is a linear combination of two exponential terms and a constant, i.e., FTI(x) := exp(-kH-x) J_exp(-kA-x) (the curve-fitting basic functions) 188 Appendix H sir := linfi<timeTI40Q TTtest, FTI) (linear fit of the identified functions ) sTI = 11.978 "24.2861 13.002 (corresponding fitted coefficients of the basic functions) Now the toluene insoluble results are predicted as follows: TI400(t) :=FTI(t)sTI Plotting the experimental and calculated results together: i:=0.. 8 r:=0,30„ 24 TTtest. l +++ TT400(r) The same procedures were followed for the exponential volatiles function with the %weight loss data: 189 Appendix H Vtest := 46.6 52.4 58.6 60.9 65.2 68.6 69.9 72.0 74.2 timeV400:= FV(x) := 1 exp(- klTx) [exp -^kA-x) sV :=linfi<timeV40Q Vtest, FV) sV = 74.719 -48.238 20.459 0 30 60 90 120 150 180 210 240 V40fl[t) :=FV(t)sV Plotting the experimental and calculated result together: i:=0„ 8 r:=0,30..24 Vtest. l +++ V400(r) 150 200 timeV400.,r 190 Appendix H The system of equations involving four equations and four unknowns is constructed as follows: The initial asphaltene and heptane soluble contents of the feed along with the rate constants are: kA = 0.026 kH:=0.013 H0:=84. A0:=15. Initial guesses of unkown parameters are: SL:=0.3 b :=0. c :=0. d:=0.3 Given kA(SLd- c)- b H ° -t-H0(bc-SL(bd + l))=sTL kA - kH (SW-c).(AD-Si5SW \ kA-kH/ 2 h-lcA 1 2 b - b c - b d - H-(d-hc-l)- H0=sV, k A - kHj 1 t , L n bkHH0\ _. (d-f-c- 1)- AO =sV0 \ k A - k H / 2 coefficient of expCKjj) in TI coefficient of exp(KA) in TI coefficient of expfKjj) in V coefficient of exp(KA) in TT 191 Appendix H Solving the system of equations results in: find(SL,b,c,d) = 0.088 0.743 0.282 0.318 Same procedure was used to determine the parameters for data obtained at 390 and 380 °C. For T= 390 °C according to Table H. 1: kH.-0.00813 kA:=0.0174 The fitted curves and experimental results are plotted as follows: i:=0~ 7 r:=0,30.. 21 192 The parameters are: find(SL,b,c,d) = 0.0751 0.64 0.284 0.844 ForT=380 °C: kA := 0.01158 kH := 0.004858 The fitted curves and experimental results are plotted as follows: 193 Appendix H i:=0„4 r :=0,30. . 15 TBctt. i TI(r) 150 Vtest. I V(r) The parameter are: find(SL,b,c,d) = 0.05 0.762 -5.80>10~ 0.774 194 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0059002/manifest

Comment

Related Items