This action might not be possible to undo. Are you sure you want to continue?
Thermal Cracking of Gas Oil in a Helical Coil Reactor
SUBMITTED TO THE FACULTY OF GRADUATE STUDIES
I PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE N
DEGREE OF MASTER OF SCIENCE M CHEMICAL ENGINEEFUNG
DEPARTMENT OF CHEMICAL AND PETROLEUM ENGINEERING
O Dahai Dong 2000
Nationai Library Acquisitions and Bibliographic Services
395 Wellington Street Ottawa ON K I A ON4
Biblioth&que nationale du Canada Acquisitions et services bibliographiques
395, rue Wellington Ottawa ON K1A ON4
Your fih Voim r e M m w 8
Our 6Ie U e r e reHrence
The author has granted a nonexclusive licence allowing the National Library of Canada to reproduce, loan, distribute or sell copies of this thesis in microform, paper or electronic formats.
L'auteu. a accorde une licence non exclusive pennettant a la Bibliotheque nationale du Canada de reproduire, przter, distribuer ou vendre des copies de cette these sous la forme de microfiche/film, de reproduction sur papier ou sur format electronique. L'auteur conserve la propriete du droit d'auteur qui protege cette these. Ni la these n des extraits substantiels i de celle-ci ne doivent Ctre imprimes ou autrement reproduits sans son autorisation.
The author retains ownership of the copyright in this thesis. Neither the thesis nor substantial extracts f?om it may be printed or othewise reproduced without the author's permission.
Hydrotreated light gas oil derived fiorn Athabasca bitumen has several properties, which
make it a less desirable transportation fuel. Its boiling point is in the range of diesel,
which is too high to be used as gasoiine. However, the gas o l also has a sigmficant i amount of fused ring compounds, which makes its cetane number too low for diesel specifications. Thermal cracking (pyrolysis) is one of the simplest and economically the most important upgrading reactions in industrial application. It was expected that the
thermal cracking method could significantly upgrade the gas oil to naphtha.
A previously designed helical coil reaction system was modified and employed to
thermally crack light gas oil to produce naphtha. Thermal cracking of hydrotreated light gas oil was studied at three different temperature levels, 500°C, 530°C and 550°C.
Extents of cracking or conversions of LGO for experiments were in the range of 15% to
For the purposes of comparing current results to the literature, a kinetic analysis
assuming a f r t order ovemll decomposition model for the light gas oil thermal cracking is
reaction was examined. The activation energy of this overall reaction was evaluated at 153 kJ/mol. The light gas oil feed used in this study has an estimated average molecular weight of 212. The activation energy found in this study was very close to that of n-
for the reaction of naphtha to gas was 2 15 kl/mol and for the reaction of gas oil to coke was 262 Wmol. According to the experimental data. the activation energy for the reaction of gas oil to naphtha was 173 kl/mol. a set of optimal reaction times at different temperatures is predicted. The activation energy for the reaction of naphtha to gas is close to 220. A more detailed reaction network involving a four-lump kinetic model was proposed. From the proposed model. 1990..hexadecane (molecular weight of 226) determined by previous researchers (Fairburn et al.8 kJlmol obtained by Kumar and K m (1985) and within the range of 212 to 228 kJ/mol found by van Damme and Froment (1981). . activation energy of 1 64 kl/mol). for the reaction of gas oil to gas was 83 kl/mol. Lumps were chosen in order to give a better understanding of the product distribution.
Acknowledgements The author wishes to express his sincere gratitude to the many individuals who have contributed to this work. . Moore for his continued interest and on-going encouragement. Leo A. D . and particularly to: Dr. Behie for his wonderful course and for teaching me about the important and practical issues of the reactor design and process modelling. patience. for his helpll discussions on fluid dynamics and heat transfer r issues. Dr. for her guidance. Gordon R. and friendship throughout the course of this work. his supervisor. Sean Cuthbert for his design of the helical coil reactor and training. B. Ayodeji A. Keng Chung for their helpful suggestions and discussions. Hyndman. Dr. Dr. Dr. for his important help on the EXCEL solver. Caroline L. Jeje. Mr. IAikhar Huq and Dr. Milne.
Paige Deitsch and Ms.Mark Misunis. the author wishes to express his deep gratitude to his wife. Ms. and Natural Sciences and Engineering Research Council of Canada are gratefully acknowledged. Mr. which occurred quite often. Appreciations are also extended to Mr. Nancy Okazawa for her help on analysis of some gas products. who gave him unlimited love and support tll during this work. Dr. Claudine Cumow for their on-going help r and guidance on the gas chromatograph analysis. . Don Mallory and Mr. Derek Verhegge for the construction of the apparatus and timely help. Financial support fiom Syncrude Canada Ltd. Ms. Jean-Marc Labontd for their valuable ideas and Wendship. Elizabeth Zalewski for her help on the simulated distillation analysis of the feed and liquid products. Mr. and his parents. Finally.Mr. Viet Bui. Mike Grigg for his help on the data acquisition system. M . Jing.
3.3.Table of Contents Approval Page Abstract Acknowledgements Table of Contents List of Tables vii xi xi11 List of Figures Nomenclature .1 Conventional Batch Reactors 2. CHAPTER 1 Introduction and Objectives I...2 Introduction Objectives CHAPTER 2 Literature Review Industrial Applications of Thermal Cracking Reactions Pilot Reactors Laboratory Reactors for Thermal Cracking Kinetic Studies 184.108.40.206 Batch Reactors 2. 1 1.2 Curie Point Py~olysis Reactor vii .3.
2 Flow Reactors 2.6 viii .3 3.2.5 The Chemistry of Thermal Cracking Reaction Kinetics for Complex Mixtures 2.2.1 Black Box Method 2.1 GC Calibration 3.1.4 Product Condenser and Collection 3.4 Lumped Kinetics CHAPTER 3 Experimental Apparatus and Methods 3.1 Helical Coil Reactor 3.4 2.2 Transfer Cylinder and Hydraulic System 3.5.2 Model Compound Method 2.1 GasAnalysis 220.127.116.11.1 Process Overview 3.3 Feed Preheaters and Gas Furnace 18.104.22.168 Process Instrumentation and Data Acquisition 22.214.171.124 Monte Carlo Simulation Method 2.5 126.96.36.199.2 Experimental Apparatus 3.4 Light G s Oil Properties a Gases Used in the Study Light Gas Oil Thermal Cracking Experiments Procedure Product Analysis 3.
4.2 Model Assumptions 4.1 Lumped Reaction Network 4.2.2 .4.3.4 Kinetic Model Deveiopment 4.3 Evaluation of Pseudo-Component Fractions 4.6 Overall Reaction Rate Product Distribution Prediction CHAPTER 5 Conclusions and Recommendation 5.5 4.2 Liquid Sample Analysis 4.3 Residence Time 4.4.1 Material Balance Product Analysis 188.8.131.52.5 Frequency Factors and Activation Energies for the Four-Lump Model 4.1 Conclusions Recommendation 5.1 TCD GC for Product Analysis 4.2 Simulated Distillation Analysis of Liquid Samples CHAPTER 4 Results and Discussion 4.4 Reaction Network Equations and Rate Constants 4.2 4.2.
References Appendix A Appendix B Appendix C Appendix D Mass Baiance Calculation for Different Runs Temperature Profile for Different Runs Pump #2 and Pump #3 Calibration Valve Switching Scheme for Gas Analysis on GC Appendix E Estimation of the Reynolds Number and Heat Transfer Coefficient Appendix F Appendix G Appendix H Estimation of the Pressure Drop along the Reactor Tube Calculations of the LGO to Nitrogen Ratio List of Failed Experiments .
4 Properties of Feed Light Gas Oil Product Specification of Gases Used in Experiments G s Chromatograph Analysis Conditions a Standard Gas Mixture Composition for GC Calibration Summary of Successful Experiments Summary of Experimental Conditions Summary of Mass Balances LGO Conversion and Pseudo-Component Distribution LGO Thermal Cracking Reaction Rate Constants Activation Energies and Frequency Factors Overall Thermal Cracking Reaction Rate Constant Optimal Reaction Times at Different Temperatures Molecular Weight Calculation of Whole Feed Molecular Weight Calculation of Feed (>196OC) Mass Balance Calculation for Run # 1 Mass Balance Calculation for Run #2 .4 Table 4. 1 Table A.1 Thermal Processing for the Purpose of Hydrocarbons Upgrading Table 3.5 Table 4.8 Table A.2 Table 3.1 Typical Properties of the Light Gas Oil Fraction of Syncrude Sweet Blend Table 2.2 Table 4.7 Table 4.6 Table 4.2 Table.3 Table 3.3 Table A.1 Table 4.1 Table 3.List of Tables Table 1.4 Table 4.3 Table 4. A.
1 xii . 10 Mass Balance Calculation for Run #3 M s Balance Calculation for Run #4 as Mass Balance Calculation for Run #5 Mass Balance Calculation for Run #6 M s Balance Calculation for Run $7 as Mass Balance Calculation for Run #8 Valve Status and Flow/Sample Status Summary of Failed Experiments Table D.7 Table A.5 Table A. 1 Table H.9 Table A.Table A.6 Table A.8 Table A.
7 Figure 3.2 Figure 4.5 Figure 3.2 Activation Energy Evaluation for n-Hexadecane Pyrolysis Kinetic Model for Fluid Catalytic Cracking Picture of LGO Thermal Cracking Experimental Set-up Simplified Schematic Diagram of LGO Thermal Cracking Apparatus Figure 3.1 Figure 3. Furnace and Hydraulic Oil Flow Meter Transfer Cylinder and Pumps (#2and #3) Picture of Condensing System Picture of Distillation Unit Chromatograph of the Standard Gas Mixture Chromatograph of Reaction Gas Product of Run #7 Chromatograph of Coke Burning Gas Product of Run #7 Simulated Distillation Curves for Feed and Products Kinetic Model for LGO Thermal Cracking Figure 3.12 Figure 4.1 Figure 4.4 .4 Helical Coil Reactor Process Flow Diagram Process Flow Diagram for the Condensing and Product Collection Systems Figure 3.1 1 Figure 3.1 Figure 2.10 Photograph of the Reactor Detailed Dimensions of the Reactor Arrangement of Reactor Heaters (Outside) Pumps (#2 and #3).List of Figures Figure 2.3 Figure 4.8 Figure 3.6 Figure 3.9 Figure 3.3 Figure 3.2 Figure 3.
Activation Energy Evaluation (LGO
+ Naphtha) Activation Energy Evaluation (Naphtha + Gas)
Activation Energy Evaluation (LGO3 Gas) Activation Energy Evaluation (LGO
Figure 4.8 Figure 4.9
Overall Activation Energy Evaluation
Component Mass Percent vs. Residence Time at 500°C Component Mass Percent vs. Residence Time at 530°C Component Mass Percent vs. Residence Time at 550°C
Figure 4.1 1 Figure 4.12
Figure B. 1 Figure B.2
Temperature Profile for Run #1
Temperature Profile for Run #2
Temperature Profile for Run #3
Temperature Profile for Run #4 Temperature Prof le for Run #5 Temperature Profile for Run #6 Temperature Profile for Run #7 Temperature Profile for Run #8 Calibration Curve for Pump #2 and Pump #3
Figure C. 1 Figure D. 1
GC Column Connection Sequence as Both Valves
in OFF Position
Valve Timing Logic
cross-section area of the reactor tube, m2
frequency factor for reaction j, f1
heat capacity of reaction fluid, J/kgK reactor tube ID, m
activation energy for reaction j, J/mol
friction factor fluid mass flux, kg/m2-s heat transfer coefficient, w / m 2 ~
thermal conductivity, W/mK
first order rate constants for reactions 1,2,3,4, s*'
first order rate constant for reaction j (j= l,2, 3,4), s-'
first order rate constant for the overall reaction, s" reaction constant in Equation (2.1) feed mass flow rate, g/s mass of pentene and heavier hydrocarbons, g mass of coke pseudo-component, g demister weight increase after experiment, which is included in
the mass of naphtha pseudo-component, g
mass of gas pseudo-component, g
mass of gas oil pseudo-component, g
mass of liquid product, g
mass of naphtha pseudo-component, g
average molecular weight of inlet feed and outlet product, g/mol molecular weight of component i in the gas bag, g/mol order of the reaction in Equation (2.1) NusseIt number reaction pressure, Pa Prandtl number gas constant, Urn01 K Reynolds number, duplp residence time, s reaction half-life, s reaction temperature, K room temperature, K fluid velocity, d s volumetric flow rate at reactor conditions, m3/s nitrogen volumetric flow rate at room temperature, rn3/s
reactor volume, m3
volume of gas bag, recorded by wet test meter, m3
initial (feed) weight percent of pseudo components 1 (naphtha)
and 2 (LGO)
1) naphtha pseudo-component mass fraction fiom SimDist gas oil pseudo-component mass hction from SimDist C02volume percent in the gas bag volume percent of component i in the gas bag Greek fluid viscosity. 3 (gas).weight percent of pseudo-component 1 (naphtha). Pas fluid density. 4 (coke) a property of the feed in Equation (2. 2 (LGO). kg/m3 Abbreviations ASTM American Society for Testing and Materials Coke Gas Gas Chromatograph Hydraulic Oil Hewlett-Packard ICFB Internally Circulating Fluidized Bed .
ID Inside Diameter L LGO Liquid Light G s Oil a Liquid Hourly Space Velocity LHSV MS Mass Spectrophometer Molecular Weight Plug Flow Reactor Simulated Distillation Thermal Conductivity Detector Wet Test Meter MW PFR SimDist TCD m .
bitumen.1 Introduction Oil sand is composed of sand. light and heavy gas oils that are combined to produce a light. (Syncrude Canada Ltd. 2000b) In the Syncrude Canada oil sand plant at Fort McMurray. mineral rich clays and water. bitumen is fed into either one of two coken or a hydroprocessing unit. In its raw state. It is thermally cracked into hydrocarbon gases. the naphtha and gas oils are treated and blended into sweet blend crude oil. As Canada's conventional reserves of light and medium oil continue to diminish while consumption rises.. asphalt-like oil. The upgraded bitumen product consists of naphtha.more than five times larger than Saudi Arabia's hydrocarbon reserves. While the hydrocarbon gases are treated for use as refinery &el. The oil sands deposits in Alberta are estimated to contain over 1. bitumen is a black.CHAPTER 1 Introduction and Objectives 1. The oil is then . sweet crude oil.7 trillion barrels . the need for such an alternative oil source is apparent. naphtha and gas oil. It requires upgrading to make it transportable by pipeline and usable by conventional refineries.
However. because the process has the inherent flexibility to handle even the heaviest of residues. 1998). middle distillate fuel for automotive diesel engines is 40 (ASTM D975. to utilize the hydrotreated gas oil derived from Athabasca bitumen is difficult.transported via pipeline to refineries in Edmonton. the synthetic crude diesel h c t i o n has a cetane number of 34 (Yui. eastern Canada and the mid-western United States. Table 1. however.1 shows properties of the light gas oil fraction of Syncrude Sweet Blend (Yui. The low cetane number of the bitumen derived gas oil fraction is caused by a significant amount of f h e d ring compounds. The process provides essentially complete rejection of m t l while providing partial conversion to naphtha and diesel cuts. 1999). . The minimum cetane number required for a general purpose. Naphtha and gas oil streams are sent to catalytic hydrotreating units to turn nitrogen into ammonia and s u h to hydrogen sulfide and to stabilize or saturate the unsaturates. which is too low for diesel specifications. the by-products are fuel gases and coke. nitrogen and unsaturates. eas The valuable products of the coker are naphtha and gas oil. The coker naphtha and gas oil are unstable because they contain relatively high quantities of sulfur. Coking processes have been selected by many refiners as their preferred choice for bottom of the barrel upgrading. 1999). Its boiling point is in the range of diesel.
w ? t! Hydrogen.023 47 Aniline Point.3 .Table 1. w ! t ? 0.2 Chromatography). wt% Cetane Number 34.7 12. t 99 Sulfur.0 Kinematic Viscosity at 20°C.878 87.1 Typical Properties of the Light Gas Oil Fraction of Syncrude Sweet Blend (Yui. cSt 5. O C Aromatics (Supercritical Fluid 42. g h l Carbon. WYO Nitrogen. 1999) Property Density at 1 S°C. wppm Light Gas Oil 0.
1. Each process has its own characteristics concerning reaction conditions and product compositions. Among them. will be measured.2 Objectives The objective of this work is to study thermal cracking reactions of hydrotreated light gas oil for production of liquid products over the temperature of 500 tr. especially naphtha. yield of products. ranging from ethane to vacuum gas oil. Thermal cracking has also been applied to the upgrading of bitumen or heavy oil. Lumped reaction kinetics will be . At present.. a daily production of 830.000 barrels of upgraded oil has been achieved at the northern Alberta oil sand plants. 2000). The experiments were also designed to obtain measurable quantities of all products. It is expected that after on-going expansion projects.Hydrocarbon cracking is the process by which higher molecular weight hydrocarbons are converted to lower molecular weight hydrocarbons through carbon-carbon bond scission. A wide spectrum of feedstock.000 barrels may be reached in 2008 (Syncrude Canada Ltd. For a range of operating conditions. 1991). 2000a and Suncor Energy Inc. a total daily production of about 330. hydrocarbon thermal cracking (pyrolysis) is the simplest and economically the most important industrial reaction. There are three general types of hydrocarbon cracking: thermal cracking. T i hs temperature range was chosen as it is within the range of reactor temperatures for fluid coking (Kunii and Levenspiel. and hydrocracking. 550°C. catalytic cracking. is pyrolyzed to produce olefins vital to the petrochemical industry.
determined and using the experimentally determined kinetic parameters. the effect of reaction temperature and residence t h e on the optimal production of naphtha will be investigated. .
butadiene and aromatics. have been increasingly important. The by-products? such as coke and gas. Nowadays a vast industry throughout the world is based on olefins and aromatics produced by thermal reactions of hydrocarbons.CHAPTER 2 Literature Review Thermal cracking reactions of hydrocarbons have been of great interest to chemists even before the petroleum industry began. From this point.1 Industrial Applications of Thermal Cracking Reactions The thermal processing technologies in the petroleum refining industry refers to the upgrading method that utilizes heat to initiate reactions of feedstocks. are available as commercial items and all are produced by thermal means. The industry of producing chemicals from petroleum started when chemists in the 1920's produced alcohols from ethylene and propylene via thermal cracking (Stolfa. a good understanding of the industrial and the theoretical aspects of hydrocarbon thermal cracking is essential. thermal processes for the production of basic chemical feedstocks: ethene. 2. coal and lignite. Thermal reactions are still playing an important role in the production of hydrocarbons from shale. oil sands. Before a kinetic study of light gas oil (LGO) thermal cracking can be initiated. propylene. 1980). Thermal cracking . butenes.
The following processes are used by many refineries (Cheng. etc. In contemporary petroleum refining industry. the thermal processes tend to have larger capacity. One of the purposes of heavy oil thermal processing is to convert the heavy end of the crude. high pour point vacuum residue into fuel oil. . Although facing competition from other processes.1 shows some of the thermal operations used in indmtry (Stolfa. heavy metals. 1994): (1) low severity visbreaking to produce fixel oil f o atmospheric or vacuum rm residue. oxygen. thermal processes remain important. 1980). diesel.). to transform low value heavy oil into valuable light oils. gas oil and coke fiom vacuum residue. (2) thermal cracking to produce gasoline. Thermal processes can deal with residues as well as distillate oil. Table 2. diesel and fuel oil from gas oil. nitrogen. including vacuum distillate and residue. Thermal processes have good flexibility for different feeds because the reactions depend only on heat effects without the need for catalyst or solvent. the more important aspect is upgrading. into useable products. asphaltenes. (3) coking to produce gasoline. and lower utility costs. These are the essential points that help the modem thermal processes survive and continue developing.is distinct from other processes such a s catalytic and solvent processes. However. to turn high viscosity. simpler operation. For example. and can treat "cleaner" feed or feed with large amounts of impurities (sulfur.
1 Thermal Processing for the Purpose of Hydrocarbons Upgrading (adapted f?om Stolfa. gasoline. coker gas oil. middle distillate. gasoline. Thermal cracking Feed Gas oil. middle distillate. middle distillate. fuel oil I 3. vacuum gas oil. fuel oil 1 Atmospheric bottoms Gas. 1980) Operation 1. ~ or vacuum gas oil Atmospheric bottoms Gas. gasoline. gasoline. decant oil. vacuum bottoms Vacuum bottoms Gas. Visbreaking Atmospheric bottoms Atmospheric bottoms Gas. heavy coker gas oil. middle distillate. gasoline. middle distillate. thermal tar furfuraI extract. middle distillate. Coking (partial recycle) Atmospheric bottoms Gas.Table 2. gasoline. Visbreaking & gas oil cracking Atmospheric bottoms Gas. Coking (low pressure) Atmospheric bottoms or vacuum bottoms Gas. gasoline. fuel oil 2. gasoline. coke 5. he1 oil Gas. middle distillate. coke . gasoline. vacuum bottoms 4. Products Gas.
(1 988) used this reactor to study the pyrolysis of Cold Lake heavy oil. Vogiatzis et 01. A fast pyrolysis reactor called the thermovortactor has been developed at the University of Western Ontario for heavy hydrocarbon thermal cracking. The pyrolysis products exiting the reaction tube are quenched wt liquid nitrogen in a second cyclone-like device called a cryovortactor. (1992.2 Pilot Reactors Pilot reactors are scale down u i s of a commercial or potentially commercial process nt designed to test a particular process. In this reactor.2. Two examples of pilot scale thermal cracking reactors are given below. 1999) developed an internally circulating fluidized bed (ICFB) reactor. ih The system is capable of achieving temperature in the range of 550-580°C and residence time of 100-500 ms. With a pilot scale facility. high temperature nitrogen and preheated heavy oil feed are rapidly mixed in a cyclone-like device prior to entering a reaction tube. heavy oil pyrolysis experiments were carried out and it is shown that the ICFB reactor is capable of achieving the high temperature (800°C) and short reaction time (450 ms) required to . The main advantage of this design is that the ICFB successMly and simultaneously overcomes gas bypassing and high temperature instability of gas spouted bed. Milne et al. The ICFB is a modification to the gas spouted bed. They are designed based on kinetics obtained fkom laboratory or bench scale reactors similar to those obtained in this study.
Residence time and temperature must be more tightly controlled than is possible with a scale down unit.3 Laboratory Reactors for Thermal Cracking Kinetic Studies Chemical kinetic study is a very important approach to analyze the rates of chemical transformations and predict rates of reactions under new conditions. demonstration and commercial. Kinetics fi-om studies such as this one are u s e l l for reactor modelling at all scales of operation: pilot. experimental data is required.minimize coke by-product formation in the thermal conversion of heavy oils to olefins and light liquids (Milne et al. A proper kinetic model combined with suitable reactor model can predict performance for industrial scale production. When intrinsic rate equations are not available. However. The reason for this is with a laboratory reactor. It . 1994). 2. which is designed to show that a process can operate at near to commercial scale. the laboratory reactors used for kinetic studies do not necessarily resemble the commercial unit and for good reason. With pilot scale results. the next step is a demonstration unit. rather than fiom a pilot plant or commercial-scale reactor. Proper design of the reactor is crucial to a kinetic study.. It is preferred to obtain these data from a bench-scale laboratory reactor. the design as well as operating as conditions can be chosen to reduce or eliminate the effects of m s and heat transfer.
The following are different types of reactors that have been used to study kinetics of hydrocarbon thermal cracking. It is common that a batch reactor has a few hundred milliliters capacity. 23.1 Batch Reactors Before 1970's. However. 2 . the reactant is placed in a closed vessel and pyrolyzed under constant volume conditions. most laboratory gas-phase pyrolyses of hydrocarbons were carried out in batch reactors. . The reactor is usually made of high temperature glasses such as quartz or metals such as stainless steel. Batch reactors are generally quite large to hold sizeable sample and to minimize surface effects. The advantages and restrictions of each type of reactor are reviewed. Conventional Batch Reactors 311 In a conventional batch reactor. they are seldom employed on a commercial scale as productivities are low. . Temperature is maintained constant by an electric furnace or liquid metal bath.should also be noted that these bench scale reactors suitable for kinetic studies are not scaledown versions of units in commercial operation but often designed especially to obtain kinetic data.
Temperature is not uniform throughout the reactor and there can be dead volumes within the reactor. Activation energies obtained in all the models wele in the range 188 to 293 kJ/mol. the reaction is fast and the heat-up time and cool-down time become nonnegligible.. the reaction rate constant will be 7.7 x lo4 s? For the high temperatures used in the current study. the activation energy for the reaction of middle distillates to light oils was found to be 240 W o l (Model A-3 in the thesis). Over the temperature range 360 to 422OC. The most difficult problem that the conventional batch reactor faces is that it can only used to study slow reactions whose half-lives (tin) are between 15 minutes and 100 is hours (Fabuss et al. A batch reactor with 8 mm OD and 118 mrn in length was used and the reaction temperatures were in the range of 300°C to 450°C. Hayashitani (1978) carried out an extensive study of the thermal cracking of Athabasca bitumen using boiling point range criteria and solvent solubility to define pseudocomponents. However. the batch reactor does have a definite merit over other reactors: the reaction time is accurately known. if the t ~ n 15 minutes. batch reactors do not meet the ideal laboratory criteria.It is not surprising that given their large size. . 1964). For a first order reaction. M n cracking models were proposed that predicted the general ay experimental trends.
an electronically activated hammer mechanism is used to break open the glass microreactor. termed the Curie point temperature. The pyrolysis products are rapidly quenched by m x n with a stream of iig cold helium and the product vapors are flushed into a G C M S system for complete component identification and quantification. Therefore. Different ferromagnetic alloys have different Curie point temperatures. This technique was In Curie point pyrolysis. The ferromagnetic wire is composed of magnetic nickel-iron-cobalt alloy. Consequently.23. At this time.13 Curie Point Pyrolysis Reactor Curie point reactor has a radically different heating and quenching concepts from conventional batch reactors. the wire temperature stabilizes at the Curie point and this temperature is then maintained. and hence the sample. At a temperature which depends only on the alloy composition.. the high fiequency energy intake by the wire drops dramatically.it provides shorter time for feed heat up and product quenching than other types of reactors after a precisely controlled pyrolysis time (Fairbum introduced by Simon and Giacobbo (1965). Even at very high temperature (600 to 800°C). a thin ferromagnetic wire is coated with a microgram quantity of sample and then placed inside a low volume glass tube. The glass tube is iriserted into a high fiequency induction coil. When the high fiequency field is energized. et al. the ferromagnetic wire is almost instantaneous heated. After a precisely timed interval. a transition from ferromagnetism to paramagnetism occurs. the pyrolysis reaction temperature can be . 1990).
1 depicts the results used to determine the activation energy of n-hexadecane pyrolysis. Fairbum et al. Wires are available with Curie point temperatures fiom 300 to 1000°C. Peak ethylene production (28 wt%) was exhibited at 842°C and 500 ms. flow reacton can be used. are large-scale m l i t b reactors. ih Heat is supplied by the media around the reactor and the reactor wall temperature is . Figure 2. specifically at 576 to 842°C and 100 to 3200 ms. The desired product of the reaction was a ethylene. The added benefit of using this type of reactor is that many commercial pyrolysis u i s nt. ut-ue As wt conventional batch reactors. 23.2 Flow Reactors For reaction times faster than those that can be accurately studied by conventional batch reactors. but olefins up to C15 were obtained in appreciable concentrations under lower conversion (60 to 75%) conditions. Thus. especially since the 1970's.controlled by using different alloy compositions. tubular flow reactors have predominantly been employed as laboratory reactors for pyrolysis studies. which w s determined to be 164 klimol. tubular reactors are built with glass or metal tubes. (1990) studied the kinetics of n-hexadecane pyrolysis using the Curie point pyrolyser. such as those used for ethane cracking. n-Hexadecane has been pyrolysed over conditions which overlap the conventional and ultrapyrolysis regimes.
In a laboratory. This type of reactor is called a plug flow reactor (PFR). a temperature profile and a velocity profile across the cross section of the tube exist. For flow reactors.usually deemed constant. it is difficult to realize plug flow because the Reynolds numbt . 1983). Strictly speaking. which means that no single residence time but rather a distribution of residence times exist. it is assumed that there is no mixing in the direction of flow and complete mixing in the radial direction. The calculation of the equivalent reactor volume requires an accurate knowledge of the temperature profile along the tube.. the key point that requires attention is m i t i i g the ideality of the anann PFR operation. An equivalent reactor volume concept was introduced by Hougen and Watson (1 947) to account for slight temperature differences in flow reactors. If the gas flow is turbulent. yields the same conversion as the actual reactor with its temperature and pressure profile. The reference pressure is always one bar and the reference temperature is the average of the process gas temperatures in the last 40% of the reactor. at a chosen reference temperature and reference pressure. the PFR . Therefore.is generally less than 100 (CBme. reactants may travel a long distance before reaching the reactor wall temperature. because of the poor heat transfer.. To improve the operating conditions. 1975). At the entrance and exit region. et al. The equivalent reactor volume is that volume which. the concept is only valid for a single reaction with a single activation energy or for a set of parallel reactions having identical values of activation energies (van Darnme.
the diameter is more likely 8. Depeyre et ui. (1987) studied thermal cracking kinetics of several straight-chain alkanes and their mixtures. C 12. The model made use of a simplified composition .tube diameter should be made small to reduce the temperature gradient. the pressure drop along the reactor may become non-negligible and an equation of momentum conservation is required. including C9. the activation energy was determined to be 164 kllmol. 18 species and 24 radicals were proposed. using a helical coil tube reactor ih i at atmospheric pressure. A mechanistic radical and molecular model consisted of 138 reactions. C The reactor w s 305 mm i length with an inside diameter of 8 1 mm (a diameter of 81 a n mm was quoted in the article. Gas flow through a tubular reactor experiences a pressure drop along its length. For dodecane (C12) thermal cracking.1 mm to give a reasonable length to diameter ratio). The reactor lengthldiameter ratio should be kept as large as possible so that the effects of thermal entrance and exit regions are negligible. The reactor used was 4 m in length and the inside diameter w s a 4 mm. Zhou et al. and C22. (1989) studied atmospheric gas oil thermal steam cracking kinetics over a range of 625 to 800°C w t the a m of producing olefins. When the reactor tube is long and gas flow rate is high to maintain turbulent conditions. The reactor was a straight tubular reactor under atmospheric pressure and temperature range was 350 to 620 O . C16. however. C13.
the activation energies were determined to be in the range of 212 to 228 kJ/rnol. Froment and co-workers conducted a wide range of studies on the thermal cracking of Light hydrocarbons and naphtha. iso-hexadecane represented iso- p a d f h . For typical naphtha (boiling point range: 40-195OC). naphtha cracks as if it were a single component and its disappearance can be described by a conversion. They found that the rate of cracking of single hydrocarbons closely followed a fint order law (Froment et a[. A model including 21 molecular reactions and an overall reaction of naphtha to gases w s a reported. 1981). In other words.. . n-hexadecane represented parafZns. Kumar and K u m (1985) studied naphtha pyrolysis over the temperahue of 720 to 800°C. The activation energy of the overall reaction for naphtha to gases was found to be 220. The length of the reactor was 60 cm. van Damme et al. C13H26 represented naphthenes and C I SH24 represented aromatics.8 kllmoi..of gas oil. i. 1976.. 1975). 1977. An annular tubular reactor was used. the first-order rate coefficient was shown to be independent of the conversion. Further.e. fiom a kinetic point of view. Pure compounds were w d to represent each family of hydrocarbons in the gas oil. This is true for naphtha cmcking also (van Damme and Froment.
One. is an exothermic reaction that converts the smaller molecules to larger molecules. The free-radical mechanism asserts that when a certain molecule undergoes a thermal reaction. Smaller fiee radicals such as H e. scission. isomerization reactions or olefin polymerization reactions rarely occur. larger &ee radicals are very active and are easily broken to form olefins and smaller fiee radicals.2. the other. the ultimate . When products leave the reaction system. However. The generally accepted model for paraffin cracking was first presented 60 years ago and is based on a &ee-radical mechanism (Kossiakoff and Rice. C-C and C-Hbond energy data indicate that the reactions most likely take place at C-C bonds of chain paraffins. as well as at side chains of naphthenes and aromatics. it breaks apart hrst at the position with weakest chemical bond and forms two free radicals.4 The Chemistry of Thermal Cracking In thermal processes. Without catalyst. called condensation. Therefore. free radicals will combine to form a paraffin and reactions terminate. Thus chain reactions occur. CH3 and C2Hs may exist for a short time and can consequently collide with other molecules to form new free radicals. 1943). there are two types of reactions taking place. is an endothermic reaction that makes larger molecules break to smaller molecules.
CH3 ' . + C 6 3 -* C 1H4 b + Cla33 . +CISH~ 1 . larger h e radicals release H and fonn olefins. (2) Propagation I this stage. ) C15H30 + H' . Many fke radical reactions are possible for a given reactant. several examples are shown below. 1994).products of scission will be pamfEns and olefins whose molecular weight is lower than the feed. CH3. a series of reactions occur that converts reactants to products while n leaving the concentration of radicals unchanged. C15H31 . reactions are included: The following types of reactions of radicals with other molecules. (1) Initiation Free radicals are formed from the large molecule: C16H34. This mechanism is illustrated using n-hexadecane thermal cracking as an example (Cheng.
More recently. Monte Carlo simulations have been used to construct . C53 1H1 ' C7H14 + C8H17 C8H17 * C a 8 + C4H9 C4Hp-+C3&+C&. this approach does not incorporate any chemical insight. Even the lighter hctions such as LGO and naphtha contain a continuous spectrum of hydrocarbons and make the reaction feed and product characterization very difficult. A number of approaches have been applied to deal with complex mixtures.C63 1H3 C 6 3 +H 1H2 and larger free radicals crack to smaller ones. C4H9 3 C2H4+ C2H5 (3) Termination Free radicals combine with wch other. CH3*+Ha-+Cb C2H5+ CH3 + C3H8 2.5 Reaction Kinetics for Complex Mixtures Petroleum and its upgraded products are very complex hydrocarbon mixtures. The simplest approach is to treat the reactions as a black box. Others try to understand the reaction chemistries using model compounds.
1 Black Box Method This empirical method treats the reaction process as a black box. 1997): (1) the initial composition is not represented in detail. (2) the subsequent reactions of the mixture are not represented in detail. Lumped pseudo components are also applied to represent important classes of compounds with simplified reactions. it is not suitable to predict performance outside the initial data set.ensembles of molecules to represent complex mixtures. This method has some limitations (Gray. the kinetic constants are unique to a specific feedstock. 1997) 2.5. and n is the order of the reaction. equation is used to determine the extent of reaction in a reactor: The following dx d (LHSV) = k(T)Xn where X can be a property of the feed. (Gray. LHSV is the liquid hourly space velocity in the reactor. k(T) is the reaction constant. (3) this model is reactor-dependent so that it can only be scaled to reactors of similar geometry. and therefore. and then calculate their reaction behaviour. . and therefore.
side-chain length.. The m i problem that the model compound method faces is that the key interactions an between components in the feedstock may be omitted. .53 Monte Carlo Simulation Method This method first uses statistical methods to characterize the structural attributes of a complex hydrocarbon mixture. Then the selected model compound is reacted at the desired conditions. 2.Khorasheh and Gray. 1993).52 Model Compound Method In this method. 1991.. ~Hexadecaneis a widely chosen model compound (Voge and Good. 1964. number of side chains. Fabuss. At least six attributes have been used: pa& length. number of naphthenic rings. 1994. The kinetic study of properly selected model compound can provide insight into the behaviour of more complex or larger molecules. 1949. An important example of using the model compound method is thermal cracking of n- hexadecane. Another problem is the hypothetical model compounds are sometimes very difficult to obtain. number of aromatic rings and number of naphthenic rings per aromatic molecule (Petti et aI. Depeyre and Flicoteaux. et al.2. the composition of a complex mixture is first studied in order to select structures of interest and then a model compound is established to represent the feedstock.
Then a Monte Carlo method is used to sample these distributions randomly and construct molecules. Finally. the new lumped-system rate matrix is constructed fiom the response data (Coxson and Bischoff. A detailed analysis iia of lumped-system reaction dynamics was given by Wei and Kuo (1969). as introduced by Wei and Kuo (1969) and Kuo and Wei (1969). industry is often only interested in some sort of lumped concentration. An ensemble of these molecules represents the heavy hydrocarbon mixhue. this method is extremely computationally intensive and with the difficulty of modelling different feeds. 1994). In fact.4 Lumped Kinetics In a mixture with many components. These pseudo-components are created by grouping chemical species having s m l r behaviour..Trauth et al.5. linear differential equations involving lumped pseudo-components. . The lumping method. However. 1997). the method has not yet been used to predict the observable kinetic or product properties (Gray. 1987a and 1987b). and almost invariably useless to have a complete description of all components. each molecule is reacted according to the known behaviour of model compounds and the molecuies are collected into appropriate grouphgs to define the product yield. Once an appropriate lumping scheme is established. 2. seeks to represent large monomolecular reaction systems by low-order. it is often impossible.
all of ytm the rate constants could change (Weekman. The greatest challenge for applying lumped kinetic methods for a catalytic process is that whenever the underlying catalyst s s e changes.2. the rate constants associated with lumped models are frequently functions of feedstock properties and conversion. Another problem with lumped kinetics is that the structures present in the lumps evolve as the reactions proceed. and therefore. Fortunately. is lost.The kinetic model for fluid catalytic cracking by Jacob et a[. kinetic studies of a thermal cracking process avoid this disadvantage. As a result. md thus details of the product distribution. be related to known model compound reactions. either fiorn changes in composition or aging. (1976) is shown in Figure 2. thus changing the reactivity of the lumps. One of the limitations of lumped kinetics is that information about individual components. Kinetic constants were determined by experiments on sub hctions with controlled composition. The reaction network was then validated by experiments using actual feeds. The subhctions were resolved in enough detail to have similar reactivity. . 1979). The model can predict performance even with changes in the feed composition and the operating conditions in the reactor.
2 Kinetic Model for Fluid Catalytic Cracking (Jacob et a[. 1976) ..Naphthenes 342'0- komatic substituents 342"C+ rings CIto C4 + Coke Light Components (220-342OC): PL=paraffins AL = aromatic substituents NL= naphthenes Ca = aromatic rings Figure 2.
CHAPTER 3 Experimental Apparatus and Methods This section outlines the experimental set-up and the method. A portion of the reactor outlet fluid was trapped after each condenser in a flask submerged in an ice bath.2 is a simplified schematic diagram of the apparatus. feed gas oil w s rapidly heated to reactor temperature.4 is a detailed diagram of the condensing and product collection systems. Figure 3. Nitrogen was used to dilute the reactant oil and its flow rate was adjusted to achieve different gas oil residence times in the reactor. Figure 3.1 Process Oveniew A helical coil reactor w s used for the light gas oil thermal cracking kinetic study. . products were quenched in a series of three water-cooled condensers (Figure 3. M e r the feed oil was preheated to about 350°C.4). Reynolds number was a found to be about 2500 and the temperature rose at a rate of approximately 800 Wm (see Appendix E for detailed calculations). Once inside the reactor. a Pumps pushed the hydraulic oil (HO) and therefore moved the piston of a transfer cylinder to force the hydrotreated light gas oil (LGO)feed into the preheaters. Solid black valves are always closed in the current study and the main process flow is shown by a thicker line. Figure 3.3. while a detailed diagram is shown in Figure 3. it was combined with 350°C nitrogen coming fiom the furnace. 3.1 is a photograph of the experimental apparatus. At the exit of the reactor.
Figure 3.1 Picture of LGO Thermal Cracking Experimental Set-up .
Gas Bag 4
Figure 3 2 Simplified Schematic Diagram of LGO Thermal Cracking Apparatus .
A demister r l l of rayon fibre (fibre made by Kendall Canada Inc.) absorbed the uncondensed aerosol before gas products entered a wet test meter (WRVI).
The volume of gaseous products was measured by a wet test meter, and gases were
collected in a 200-litre gas bag. After the feed was stopped, nitrogen continued to flow through reactor to sweep the remaining hydrocarbons to the reaction gas bag. Coke
make was evaluated using a bum procedure once the reactor had been well purged.
Another gas bag was employed for the gas product from the coke bum procedure.
The liquid product was collected f o the flasks. Flasks, condensers and pipes were rm
cleaned completely using dichloromethane (Anachemia Canada Inc., > 99.5% purity). Product oil was then separated from the dichloromethane by a small laboratory distillation unit inside a fume hood.
Helical Coil Reactor
The reactor coil is a total of 18 m in length. It is comprised of two pieces of stainless
steel tubes welded together. The inside diameter (ID) of the first 9 m was 3.34 mm
(3/16" tubing), while the second half of the reactor had a larger ID of 6.16 mm (5116"
tubing) to accommodate the increased volume of cracked gases. The two parts were
20 m.2.7 shows the arrangement of reactor heaters. The bar stock was heated both from the centre and the exterior. an overflow t n .welded together. As shown in Figure 3. 3. Inc. the hydraulic system comprised a hydraulic oil reservoir. the helical coil was wound into a groove around a 1Ocm diameter stainless steel bar stock. Six others were installed at the centre of the reactor assembly. flowmeters and three pumps. This heater accounts for convection effects.hydraulic oil filter.4 mm in diameter.132~ wlm2 10' power density at 230 volts.2 Transfer Cylinder and Hydraulic System As shown in Figure 3. T i assembly was contained within a larger cylindrical section of hs bar stock (see the lower section of Figure 3. They had a heated length of 3. Two heaters were coiled around the larger cylindrical section of bar stock holding the reactor assembly. ak .5).6. 7200 W power and 1.3.6. Reactor length is long enough (lengtkddiameter > 1000) to eliminate the effects of entrance and exit region. Detailed dimensions of the reactor are shown in Figure 3. Heaters are made by ARI Industries. Figure 3.5. The large thermal mass provided energy for the endothermic reactions. A third smaller heater made by A N was used to heat the bottom section only. The long reactor tube was coiled to allow it to fit in a smaller bar stock.
Figure 3.5 Photograph of the reactor .
Figure 3.6 Detailed Dimensions of the Reactor .
14 cm .Heater #I 64.
The calibration c w e for the LGO flow rate is given in Appendix C. pressure safety valves were used to protect the rest of the system fiom overpressure. Hydraulic oil was stored in a reservoir. By opening valves HI.hydraulic oil was able to circulate within 5 1. a As shown i Figure 3. A photograph of the two pumps is shown in Figure 3. These t o w pumps were set i parallel and the calibration was done with both pumps on and at the n same setting on the scale. Pumps pushed the hydraulic oil to move the piston in the transfer cylinder.3) used in the current study were made by Milton Roy Company (product number 9-2015003).H 0 H11. The volume of the transfer cylinder was 2. They were set at 0. A transfer cylinder bridged the hydraulic system and the LGO feed. As shown in Figure 3.H9. A photograph of the transfer cylinder can be found in Figure 3-9. n H . H12 and closing valve F4T.3 gls of LGO.5 MPa The pump had a knob with scale ranging from 0 to 100. a closed circuit. Once valve H12 was turned off and valve F4T was turned on.3. .8.3.3 litres and the piston diameter w s 10 crn. A hydraulic oil filter was installed before the pumps to remove the possible particles or debris. pumps could push hydraulic oil to move the piston in the transfer cylinder. They are displacement type and can raise pressure up to 40 MPa Both pumps operating together can generate a maximum flow rate of 0.The two pumps (pump #2 and #3 in Figure 3.
8 Pumps (#2 and #3). Furnace and Hydraulic Oil Flow Meter .Figure3.
9 Transfer Cylinder and Pumps (#2 and #3) .Figure 3.
5 crn I x 79 cm length. 3. Before nitrogen was mixed with the preheated feed LGO. there was a flask to collect the condensed products.2. after each condenser. The dimensions of the furnace were 9. Each flask was placed in an ice bath.4. A demister 111 of rayon was installed after the . The maximum operating temperature was 10IO°C and the power was 1.23 Feed Preheaters and Gas Furnace Feed LGO was heated from room temperature to about 350°C by two preheaters. Cold water flowed through the shell side of each condenser.332 W. The fiunace was a Series 31 10 tube type fiunace made by Applied Test Systems Inc. The fimace was also employed to preheat air to 350°C before it entered the reactor during coke burning. The first condenser had a 40 cm long copper water jacket outside a stainless steel tube and the other two condensers were regular glassware condensers 30 cm in length.8. As shown in Figure 3.7i m. and then fixher cooled in a series of water-cooled condensers.3.4 Product Condenser and Collection The reaction effluent was quenched. it was heated to 350°C by a furnace. A photograph of the fiunace can be found in D Figure 3. It had single-zone construction and the heated length was 0.
. Coke burning product was also measured by the WTM and collected using another bag with the same capacity. flasks. drum capacity 2. The inlet thermocouple was 1.6 mm (1116 inch) i diameter and was placed about 30 cm into the n reactor tube. The gas bag had a capacity of 200 a litres. demister and pipes connecting them had to be completely cleaned using dichloromethane (normal boiling point is 4 ° ) 0C. furnace and reactor. A photograph of this unit is shown in Figure 3. Dichloromethane and oil product w s separated in a simple batch laboratory distillation a unit.5. A photograph of the condensing system and demister is shown in Figure 3. G. Zeal Ltd.1.5 a litres). 33. The inlet thennocouple may actually be touching the reactor tube wall. The gas w s then collected in a gas bag.5 Process Instrumentation and Dab Acquisition All the thermocouples used in the current study were calibrated K type made by Alltemp Sensors. As shown in Figure 3.2. They were installed at the preheaters. A photograph of the WTM and gas bag can be seen in Figure 3. .1 0.condensers to coalesce any aerosol before gas products entered a wet test meter. After each run. H.1 1. the volume of the gaseous reaction product combined with nitrogen w s measured by a WTM (DM3C-2. the condensers. Temperatures at the reactor inlet and outlet were measured by thermocouples in the fluid.
Figure 3.1 0 Picture of Condensing System .
Figure 3.1 1 Picture of Distillation Unit .
1) with a pentiurn computer was used as an on-line temperature data acquisition device. preheaters and fimace. Temperature control was realized by UT350 general-purpose controllers for the reactor. The controllers were made by Yokogawa Electric Corporation with an input sampling cycle of 250 ms. The T-connection was covered by a 10 cm thick insulation to minimize heat losses. it was then switched to Auto control mode. The reactor inlet and outlet temperature were measured and stored in a file every two seconds. The outlet thermocouple was 1.3% the span. the outlet thermocouple was measuring the fluid temperature. LabView (Version 4. 33 Light Gas Oil Properties Hydrotreated light gas oil.6 mm (1116 inch) in diameter and w s inserted into a T-connection a about 10 cm away fkom the reactor bar stock. the wall and fluid temperature equalizes within approximately 20 cm.1. because of the rapid heat transfer (see Appendix E for detailed calculations). used as feed in the current study was provided by Syncrude Canada Ltd.1% 111 scale for of process variables and an output accuracy of H. The controllers can be of operated in "Auto" or "Manual" mode. a certain percentage of power was set and once the temperature was close to the set point.However. Prior to an experiment during the heat up stage. an input accuracy of H. . Because of the T-connection. Some properties of the hydrotreated LGO are given in Table 3.
Table 3. g/rnol 212 (estimated fiom SimDist) ' (Yui. O C 370 400 Mean MW. g/ml 0.1 Properties of Feed Light Gas Oil hydrotreated light gas oil Density at l!i°C. 1) + . "C 210 90% off. "C 95% off. 1999) (Table A. "C 180 10% off.878' SimDist ' 5% off.
WTM water level was checked and adjusted as instructed in the WTM m n a . A detailed process flow diagram of the apparatus is shown in Figure 3. Therefore. Prior to an experiment.4 Gases Used i the Study n Nitrogen used in LGO thermal cracking experiments.3.5 Light Gas Oil Thermal Cracking Experiments Procedure The light gas oil thermal cracking experiments were conducted using a procedure developed in this study. Table 3.4) and the rayon for the demister were weighed. demister and process piping. It was very important that the demister should be assembled upside down so that the .3. refilled as necessary and then reinstalled. condensers were disassembled and meticulously cleaned including the product collecting flasks. and helium used in GC analysis were all produced by Praxair. The transfer cylinder was detached and checked to find out how much feed was aul left. coke analysis will not be affected. Air used in the coke burning procedure was tested for C02 and CO content on the GC. air used in the coke burning procedure. Solid black valves are always closed in the current study. The steps outlined below were strictly followed in the given order for all experiments in this study. only trace quantities were found.2 gives the product grades and quality specifications of the gases used in this study. The last one of three liquid product-collecting flasks (refer to Figure 3. Rayon was evenly put into the dernister. 3.
23. .1 ppmv ' 1 ppmv < 1 ppmv < 1 ppmv co2 CO * Neon fiee basis.2 Product Specification of Gases Used i Experiments n Nitrogen Air Helium Product Grade Oxygen Free < 0.999" < 2 ppmv < 0. % - 99.998% Hydrocarbon Free 19. < 1 ppmv < 5 ppmv 0 2 N 2 < 3 ppmv Helium.1 ppmv HZ < 0.1 ppmv < 0.Table 3.5 ppmv H20 THC < 0.5% 5 0 UHP .5 .5 ppmv 99.
could maintain the reactor temperature constant for at least 2 hours. condensers and flasks were reassembled. Gas bags for reaction and coke burning were vacuumed using the building vacuum line. Finally. there was oscillation within a range of 10°C and a period of approximately 40 minutes.rayon could be packed more tightly at the entrance and exit. the demister. Vacuum grease was put onto the joints of all glass condensers and flasks. SS-30 copper anti-seize compound (Jet-Lube of Canada Ltd. a The reactor required four to five hours to reach its final operating temperature. Manual power settings were therefore used for experiments. LabView software was then loaded and a data file was created to monitor and record the temperatures. refilled with nitrogen and then vacuumed again. The cooling water flow to the condensers was started at this time and the exhaust gas w s allowed to flow to the wet test meter. The reactor heaters power were set to 20% and h a c e power to 25% using the temperature controllers on the control board. after the reactor reached the set point. Preheaten were . Experience showed that at 46% power. The gas bags were marked and always used for the same purpose. a The valves and piping were checked and then nitrogen was introai~ced a small flow at rate of about eight ml/s before raising the reactor temperature. The reaction gas bag w s then attached to the apparatus.) was used to thoroughly coat the fitting between the first condenser and the reactor. Although the temperature controllers could be used to adjust power in the "Auto" mode to stabilize at the desired set point. This should effectively get rid of the poteatially remaining hydrocarbons or C02 and CO.
then turned on with 20% power, it took t e about 2 hours to reach 350°C. Once the hm preheaten reached the set points, they were set to "Auto" mode.
Half an hour before the reactor reached its desired temperature, the oil feed pumps #2
and #3 were started in order to circulate hydraulic oil within a closed loop and stabilize pump operation. Special attention should be paid to the low flow rate operation. A large flow rate should be used, and then reduced to desired flow rate. T i ensured there hs was flow going through the pumps. These pumps needed 20 minutes to be stabilized.
Ice was added to containers surrounding the flasks. Current atmospheric pressure was
recorded fiom a barometer. Nitrogen flow rate was increased to the desired value by
measuring flow volume going through the WTM within a known time period. Valve P6
in Figure 3.3 was opened to stop the flow going through the WTM so that it could be zeroed.
Valve F6 w s opened to prepare to bleed the feed oil to a waste oil tank. Immediately a
after valve F4T was opened, valve H w s closed to stop hydraulic oil circulating and 5 a
start to feed preheaters. Feed oil thus went through both preheaters and exited fiom
valve F6 to the waste oil tank (a 4 litre metal can). After about two minutes, the feed flow coming from F6 becomes stable. While waiting for a stable feed flow, nitrogen flow was directed to the gas bag (close P6,P7 and open P8).
The experiment started when valve F6 was closed and F7 was simultaneously opened to
feed the reactor with Light gas oil. An assistant operator started the timer to record the feed time.
After about 15 to 30 minutes of run time, valve F7 was closed and the bleed valve F6
w s simultaneously opened. The timer was stopped and feed time recorded by the a
assistant operator. Valve F6 was closed after pumps were turned off. Nitrogen flow through the reactor was continued until condensers became clear. During nitrogen
purging, the gas chromatograph (GC) w s prepared for analysis of the gas samples. To a
stop the flow of nitrogen to the gas bag and switch it to the vent, valves P6 and P7 were opened and P8 w s closed simultaneously. One should take care not to overpressure the a
WTM during this operation. Reaction gas volume was read from WTM. Data
acquisition with LabView was stopped and data saved.
SAMPLELOCK syringe was used to obtain a 10 rnl gas sample from the
reaction gas bag, which was analyzed by GC.
Before burning, the reactor temperature w s raised to 600°C and coke-burning gas bag a
was attached to the system. LabView w s started for monitoring purposes only. Ice was a
added into bucket with cooling coil for coke burning. Valve P4 was then opened. The
M condenser (shown in Figure 3.4) was detached fiom the system and the rest of the
condensers were isolated. Before the cooler for coke burning was attached to the
reactor, anti-seize compound should be thoroughly coated to its fitting.
AAer the WTM was zeroed again, valve P8 was opened and valves P6 and P7 were closed to direct nitrogen into the gas bag. Air was then introduced by closing valve F10 and opening valves F 11 and F 12. Coke burning continued for about one hour. Then air
flow was directed to vent. Total gas bag volume was recorded by the WTM. The
Hamilton SAMPLELOCK syringe w s also used to obtain this gas sample for GC a analysis.
After the run, the condensers were again disassembled and completely cleaned with
dichloromethane. Liquid product was first mixed and collected in the flask immediately
before the demister, and then the flask was weighed. After weighmg, the liquid product
w s collected in sample bottles for fuhw analysis. a
The mixture of oil and
dichloromethane was then distilled to remove dichloromethane and residue was weighed. Rayon in the demister was also weighed.
Product analysis included GC analysis of gaseous products and simulated distillation
analysis of the reaction liquid product. Because the coke make was very small, no
attempt was made to analyze the coke properties.
methane and carbon monoxide. However. dual injection ports and a thermal conductivity detector (TCD). data acquisition and evaluation.Plot Molecular Sieve (MS) 5A. Gas Analysis In the current study. Thus. including integration.361 .53 mm x H . 30 P m x 0.Plot Q. 15 m 50.3 summarizes the operating conditions used in gas sample analysis.0 pxn film thickness. these columns should not be exposed to temperatures greater than 250°C. quantification and reporting. The Plot Q column An analytical method was set up to analyze the reaction gas bag and coke burning bag. The GC has a large column oven. T i method was a collection of set points required to run a single gas sample on the hs GC.0 pm film thickness and x 0. These fbsed silica capillary columns provide a ih quicker analysis time wt better resolution than the traditional packed columns. Two columns were employed: KP . ethane and heavier hydrocarbons.53 mrn x 40. a Hewlett-Packard (HP) 6890 gas chromatograph (GC) and HP ChemStation software have been employed. Table 3. The H ChemStation P software contains all parameters for instnunent control. (Divinylbenzene/styrene Porous Polymer). The MS column was used to separate inerts (9 and N2). the maximum GC oven temperature was 250°C. separated carbon dioxide. . Details on the GC method can be found in Appendix D..
01 min 620 kPa gauge (90psig) 480 kPa gauge (70psig) . ramp to 250°C at 30°C/min Total Analysis Time 9 7 min . 60°C 2-4 min.Table 3. 22 d m i n Initial temperature 60°C 0-2min.025 mV Ml scale D t Range aa Minimum Peak Width Cylinder Pressures Helium Air 20 Hz 0.3 Gas Chromatograph Analysis Conditions Reference Flow Temperature Program Helium. System Temperature: Inlet 140°C TCD 260°C TCD Detector: Range 0.7min. ramp to 80°C at 1 O°C/min 4-9.
ASTM method D2887 was used for the current study. is a sample containing a lcnown amount of the compound to quantify. Simulated Distillation Analysis of Liquid Samples Simulated distillation methods are chromatographic techniques which provide data that emulates glassware distillation through gas chromatography. The column (Series Number 2892 15) is 10 m x 0. The response of the detector is assumed linear over the working range of concentrations for the samples of interest.53 mm ID. Single-level calibration method was used for hydrocarbons and C02 was calibrated using a multi-level method.4. The calibration table is also used for peak identification. 362 .1.1 GC Calibration Calibration is the process of determining response factors used to calculate absolute component concentrations by injecting specially prepared calibration samples.3.6. A calibration sample.. Figure 3. The external standard is ASTM D2887 reference gas .12 shows a chromatograph of the standard gas mixhue and its composition is shown in Table 3. also referred to as a calibration standard or a standard mixture. The response factor for a given component peak is given by the inverse of the slope of the calibration curve line through the point and the origin.9 micron film thickness and 0.
. . . - L .Oxygen 3 d .Methane i 1 6.N-BUTANE - 7. I . . . .- +* e e g c n M 31 . . .I .740 .241 I-- 1 3 2.626 7. .073 .521 ISO-BUTANE 6.2. . r . I .784 - - ISO-PENTANE N-PENTANE . . .763 . . d I ld I 13.
25 50.5 10 Component i-C4Hla mol % 0.4 Standard Gas Mixture Composition for GC Calibration Component moi % 1.25 0.5 0 2 CO n-caio cb CzH6 c2b 5 5 i-CSHI2 n-CsH12 i-c6H 14 n°C6H 14 0.5 0.25 2 0.25 co2 C3H8 20 2 2 N 2 cH 36 .Table 3.5 0.
and the retention time standard is SD-SS3E-05 (carbon number ranges fiom C4 to C66). The column and the standards are made by Separation Systems Inc. .oil.
such as using different condensers.1. a number of runs with and without LGO feed were performed in order to identify and debug various operating problems with the reactor. A summary of the operating conditions used in each of the successll experiments is shown in Table 4. The results of these studies prompted many design additions and modifications. Unsuccessful experiments are listed in Appendix H. A simple four-lump kinetic model for the LGO thermal cracking reaction was developed and kinetic parameters are calculated and discussed. 1).CHAPTER 4 Results and Discussion In keeping with the objectives of this study. Reactor temperature profiles are given in Appendix B. .1 is the interval between the feed starting and feed stopping (see Figure B. Besides those tests. Because of the complexity of the reaction system. as well as using the two pumps providing the smallest flow rate. The feed time shown in Table 4. hydrotreated Light Gas Oil (LGO) has been pyrolysed at temperatures varying fiom 500 to 550°C. demister. pumps. a demister and a gas bag. instrumentation and other equipment. A summary of all successful experiments appears in Table 4. eight successful runs were conducted where light gas oil w s fed to the a reactor. condenser. The simple model allows us to determine the amount and rate at which the LGO pseudo-component is converted to the desired product (naphtha) and to by-products (gas and coke).2.
O C Feed Time.Table 4. Successful run at 530°C. Longest pyrolysis experiment performed in study. Over 200g of LGO processed. Successfbl repeated run. s Comments 1 2 3 99-08-31 99-11-09 99-11-17 550 530 500 1256 1212 1596 First successfbl run at 550°C. Successful run with higher feed f o rate. 500 1386 836 6 00-02-02 00-03-27 00-04-03 550 550 530 7 8 779 912 . Successfbl repeated run.1 Summary of Successfbl Experiments Run Date Reactor Temperature. lw Successful run with higher N2flow rate. 4 5 99-12-01 99-12-15 550 1584 Successfbl repeated run.
7 1.0 530 550 550 550 550 1.1314 0.058 0.5 2.58 1.35 1.1314 0.058 LGOMz mass ratio 2.0 2.1314 0.58 I 4 6 7 2.1838 6) 1.1314 0.Table 4.1314 0.0 2.41 1 .O 1.058 0.73 1.64 2 8 2 .058 0. O C Feed Flow NzFlow Rate (L/s) 0.058 0. O Residence Time Rate (g/s) 3 5 500 500 530 0.0 2.73 1.1314 0.1314 0.058 0.058 0.075 0.2 Summary of Experimental Conditions Run T.
the reaction gas products were analysed by an HP6890 TCD gas chromatography. liquid product recovered f o remaining oil in the condensers and pipes.2. Although a GC equipped with a flame ionization detector would be more accurate. More details on the material balance can be found i as n Appendix A.2 Product Analysis As outlined in Chapter 3. 4.1 and 4.1 Material Balance Products fiom each run included the gas in reaction gas bag. The liquid products were analysed by an HP6890 simulated distillation system. Example chromatograms of the reaction gas product and the coke burning gas product are shown in Figures 4. liquid product directly collected from the flasks after each condenser. and liquid rm product in the form of mist trapped by cotton in the demister. the potential error in overall mass balance was small . Table 4.2.1 TCD GC for Gas Product Analysis An HP6890 TCD GC was employed to analyse gas in the gas bags after each run. coke in the form of COz in the bum gas bag.3 gives a summary of the m s balances. 4.4.
3 0.4 3 5 2 500 500 530 530 550 550 550 550 209.8 86.2 147.6 87.2 75.40 0.8 25.3 40.05 0.0 29.1 6 7 143.9 72.8 208.4 109.30 96.9 75.0 118.4 0.6 96.4 0.05 95.1 93.6 .2 130.7 10.4 26.0 95.8 96.10 159.3 119.6 94.7 14.20 0.8 24.35 0.0 95.7 188.3 182.9 165.4 5.5 52.9 20.0 164.4 f 2.30 73.6 8 1 4 97.6 0.Table 4.8 96.3 Summary of Mass Balances Run T Feed L (g) G (g) C (g) Recovery L Wt% G of W o of t Y ("C) (g) of feed product 93.1 27.1 I product 6.1 12.9 13.2 103.0 34.2 74.
1 Chromatograph of Reaction Gas Product o f Run #7 .Figure 4.
However.since the mass of the heavier components (C5+) for which the TCD would be less accurate formed only a small percentage of the total gaseous product. simulated distillation (SimDist) is the best alternative for the liquid analysis. the currently available G C N S device could not resolve the light gas oil feed and its thermal cracking products because of the large number of isomers. after extensive preparative separation of the samples. PONA (Paraffins. 4. Therefore. Olefins. with currently available devices. However. Naphthenes and Aromatics) analysis was considered in the study.. Ternan et a[. NMR analysis is u s e l l for characterizing the average molecular characteristics. Consequently. Mass spectrometry can provide information on homologous series up to a molar mass of 1500. the mixture of components in the feed and the products defy any attempt at complete component-by-component analysis. a satisfactory PONA analysis was not possible because the column was not long enough (40 m) and could not separate the peaks effectively. Beyond this size. . but it cannot give boiling distributions of species (Gray et a[. 1994).. 1991.23 Liquid Sample Analysis Resolution of mixtures by gas chromatography is limited to hydrocarbons up to 12 carbons (molar mass of 170). the numerous isomers are too similar in their properties to be separated by a chromatographic column.
3. Gases in the reactor were at low pressure (slightly above atmospheric pressure) and high temperature (500550°C). The molecular weight of gas .As explained previously. 43 Residence Time The residence times were calculated by assuming the molecular weight of the fluid in the reactor is the average value of the inlet feed and outlet stream.From the results of SimDist. the liquid products collected were analyzed by simulated distillation gas chromatography (SimDist) according to ASTM method D2887-89. it is found that the boiling point range and end boiling point have been reduced sigmficantly after the feed light gas oil was thermal cracked. SimDist results for all eight experiments are shown in Appendix A. the lighter the product. However. The molecular weights of feed and liquid products were estimated by the results of the simulated distillation (SimDist) analysis. No attempt w s made to mix the light condensate collected in the demister with the a product oil prior to the liquids analysis. so the compressibility factor is e q d to one and the gases can be assumed to behave ideally. the oil trapped by the demister was considered part of the naphtha component because after the second condenser the s r a tem temperature had already dropped to room temperature and the demister was located after the third condenser. The higher the reaction temperature. Simulated distillation curves for feed and products at three different reaction temperatures are shown in Figure 4.
Product at 550°C Reaction-Run 6 I 20 40 60 Weight Percent Off..3 Simulated Distillation Curves for Feed and Products . .--- j Feed Product at 500°C Reaction-Run 4 -Product at 530°C Reaction-Run 2 . % Figure 4....
K room temperature. m3/s nitrogen volumetric flow rate at room temperature. Jlmol K residence time. rn3/s reactor volume. 296 K volumetric flow rate at reactor conditions. g/s average molecular weight of inlet feed and outlet product. (See Appendix A for details a of the calcdations. s reaction temperature. g/rnol reaction pressure. m3 . Pa gas constant.product w s calculated &om the gas chromatograph results.) where: feed mass flow rate.
4.4 shows the four-lump reaction network. light gas oil ytm (boiling point greater than 196OC). Wei and Kuo.4. The four lumps in the reaction s s e include naphtha @oiling point less than 196OC). Jacob et al.. C5 and C6 in the reaction gas bag and the mist trapped by the dernister are included in the naphtha boiling range. There was a large volumetric expansion in the helical coil reactor due to the cracking reactions. A set of four lumps divided by boiling point range is suggested. as . According to these categories. 1976.4 Kinetic Model Development 4. Weekman. Figure 4. In order to be consistent and to be realistic. 1979). 1969.1 Lumped Reaction Network Lumped pseudo components were applied to represent classes of compounds with simplified reactions (for example. gas (n-butane and lighter components) and coke. A four-lump cracking reaction scheme was chosen to represent the cracking reactions taking place in the reactor because of its simplicity and its ability to predict the product distribution. in this study all reaction rate calculations were performed on a m s basis.
4 Kinetic Model for LGO Thermal Cracking (numbers i brackets are pseudo-components) n .n Naphtha (1) n Coke (4) Figure 4.
43 Model Assumptions According to this scheme. Therefore. gas and condenses to coke while naphtha may crack further to form gases. may have an indirect effect on the reactions by reacting with coke via the following reaction. C (Coke)+ H 0 (Steam) -3 z - CO + H2 (4.3) Steam also contributes the formation of metal oxides on the reactor surface. 1993). It is believed that the activation energy of gas producing coke is much higher. Fe304and hydrogen. In thermal cracking studies. 1989..4. Steam. (Tsai and Albright. the coke formation reaction rate is very low and it is reasonable to assume that gas will not condense to form coke at the temperature range of interest. however. The steam treatment of temperature-resistant Incoloy 800 at 800°C for 24 hours resulted in significant formation of NiO. nitrogen or steam has been employed as a diluent in tubularflow reactors in order to control hydrocarbon partial pressure and residence time (Depeyre et al. It is assumed that the coke formed by condensation reactions f?om the naphtha firaction is negligible compared to coke formed from gas oil cracking. These inert gases can reasonably be assumed to have no direct effect on the homogeneous thermal cracking reactions if the reactions are approximately first order and if the reaction rate constants are independent of feed concentration. Khorasheh and Gray. gas oil cracks to form naphtha. 1976) .
they all contribute directly or indirectly to coke formation.-• (metal oxides).de + H2 Because steam was not employed in the study. At reactor temperatures of 500 to 550°C. it has been found that the coke formation reaction rate is very small compared to that of other cracking reactions. If there were significant coke formation in our study. Therefore. At least seven surface reactions are thought to occur during most industrial pyrolysis. the entrance regions are neglected because of their minor effect in the long reactor tube used in this project (length/diameter >1000). For the hydrotreated light gas oil feed at the reaction temperature range of this study. the effect of including the hydraulic and thermal entrance regions on the tube temperature profile is small for lengthldiameter ratios greater than 50.see Figure 4. reactor wall surface effect can be assumed to be negligible. These are i an acceptable range of turbulent flow to ensure plug flow n . Reynolds numbers for the experiments ranged from 2500 to 3000.3). Coke formation is undesirable in general because it decreases yields of desired products and increases heat transfer resistance. light gas oil is vaporized essentially instantaneously (99% vaporized at 470°C. it is unlikely that surface effects played a major role. In a complex while not completely understood manner. As indicated by Kays and Crawford (1993). it could also increase reactor tube pressure drop. Thus.(metal)surfaet+ H 0 2 .
gas and coke. The mass of gas is calculated as follows. 4. using C02 concentration obtained from GC analysis results of the coke burning gas bag and the gas bag volume fiom WTM.operation and rapid heat transfer for LGO heating up and quenching (see Appendix E for detailed calculations of heat up rate and Reynolds number). Assuming the C/Hmole ratio to be 1:1. the four pseudo-components are naphtha. The mass of gas pseudo-component was calculated using methane to n-butane concentrations fiom the GC analysis results of the reaction gas bag and the gas bag volume &om the WTM. gas oil. Detailed calculations for pressure drop are included in Appendix F. the m s of coke pseudo-component was calculated as follows: as .43 Evaluation of Pseudo-Component Fractions As explained previously.
concentration of C5+ from GC analysis results of the reaction gas bag and the weight increase of the demister.4. By considering a small volume of a reactants flowing through the reactor tube. the weight percent of each pseudo-component in the total experimental products can be obtained and is shown in Table 4. The gas oil pseudo-component is simply the heavier portion (SimDist > 196OC) of the liquid product (exclusiw demister weight increase). the kinetic equations of the plug flow reactor . 4. the mass of naphtha pseudo-component can be calculated as following: rnleid is the mass of total liquid product trapped in the demister and the three condensers.4. Therefore. Conversions of LGO pseudocomponent are also listed in Table 4.4 Reaction Network Equations and Rate Constants The reactor w s assumed a plug flow reactor.4.With the result of liquid product SimDist analysis.
31 23.wtO/o 500 500 530 530 550 550 6 15.22 46.05 0.44 48.16 94 0 6. w t % x3.61 64.8 78.84 5.5 46.67 12.09 16.75 11.83 22.13 13.1 50.4 LGO Conversion and Pseudo-Component Distribution Run T.% "C Feed 3 5 x ~ .69 80.02 0.wtYo ~q.42 24.6 0.78 6 550 550 2.52 2 8 1 4 29.29 7 50.12 24.3 48.0 61 26.69 22. Naphtha xl.3 14.12 0 0.0 .19 52.54 24.99 22.2 31.97 24.Table 4.03 0.13 66.22 0.20 44.wt0/0 LGO Gas Coke LGO Mass Conversion.13 0.76 0.
However. ut The integrated form of equations (4. which m s be integrated.9) to (4. For thermal cracking reactions. become: .12). volume expansion is significant (about 3 times under the conditions in this study).and those of the batch reactor are essentially the same as long as the volume is constant. The reaction network results in a system of four ordinary differential equations. first order reaction kinetic equations for the four reactions are outlined below.4. the batch reaction equations may still be used on a mass basis. According to the reaction network shown in Figure 4.
where: ki.2.4. kq t Xl0. s" residence time. and the residence time is calculated assuming an average molecular weight and ideal gas flow. 3 (gas).3. x20 first order rate constants for reactions 1. x2. 2 (LGO). 4 (coke) xlo and x2o are known f o the SimDist analysis of the feed. . x3 and Q are determined fiom the experimental material balances. k 2 . s initial (feed) weight percent of pseudo components 1 (naphtha) and 2 ( L O ) weight percent of pseudo-component 1 (naphtha). 4 . rm XI.
5 Frequency Factor8 and Activation Energies for the Four-Lump Model Once the reaction rate constants at different reaction temperatures have been determined. feed and product component weight percent and residence time can be found in Appendix A. Negative results were rejected due to physical constraints and the order of magnitude of reaction rates was compared to literature on thermal cracking.5. Although more than one result could be obtained. were calculated using the Microsoft EXCEL 2000 solver function. 4. a non-linear regression method was used to determine the reaction rate constants at different reaction temperatures. 14.13) to (4. The initial estimate supplied to the solver function could be changed by 30 percent without affecting the converged results. only one set was reasonable for a particular run. s" .4. activation energies and fiequency factors can be calculated by the Arrhenius relationship: where: *j frequency factor for reaction j. A set of detailed tables for the calculation of average molecular weight of feed and products. The results are shown in Table 4.Based on Equations (4.17). The reaction rate c o ~ t s kl to .
Table 4 5 LGO Thermal Cracking Reaction Rate Constants . O C . Run Temperature.
Jlmol first order rate constant for reaction j. K The linear relationship of Equation (4. is shown in Figures 4. A summary of the Arrhenius parameters is given in Table 4.J/mol.) Products (Naphtha. an .5 Overall Reaction Rate Although the activation energy for the naphtha lump reacting to form gas compares favourably with the literature. 4. Gas and Coke) The initial weight percent of LGO was 94 and naphtha was six percent.18). which is close to 220.. Therefore.activation energy for reaction j. The activation energy for the reaction of naphtha to gas was 215 k. s-' gas constant.3 14 Jlmol K reaction temperature. calculations have been performed based on a simplified overall reaction for which comparison to the literature is possible. 8. Repeating similar calculations and regressions on the Microsoft EXCEL 2000 solver function.8. there is no data in the literature with which the results of the four-lump reaction scheme can be directly compared.8 kJ/mol obtained by Kumar and Kunzru (1985) and within the range of 2 12 to 228 kl/mol found by van Damme and Froment (1 98 1). for each of the four reactions. This frt order overall reaction is assumed to be: is ko Gas Oil .5 to 4.6.
2 3 1o4 ~ Naphtha +Gas E2. H/mol Ap.s" LGO + Gas E3. Id/mol 262 8.S-I LGO ++ Coke Eq. 2 2 10" ~ 215 2.s-' . kJ/mol AZ.6 Activation Energies and Frequency Factors r LGO +Naphtha El.Table 4.81~10'~ h.4 1 x loL3 83 1 .S" 173 3 .kJ/mol A*.
O C 1000xb. the overall reaction frequency factor and activation energy were 0 determined to be 1-9 x 1' s. (1990) used a Curie point pyrolyser to study ultrapyrolysis of n-hexadecane.7 Ovcrall Thermal Cracking Reaction Rate Constant r Runs Temperature. These simplitications lead to some errors in the model predictions.9). She found that the activation energy of n-hexadecane pyrolysis was 164 W o l .overall reaction rate constant can be obtained and results are given in Table 4. ' Table 4. Fairburn et ul. as results of overall reaction kinetic parameters are very close to those of previous .7.and 153 Wmol. However. Based on these results. respectively (see Figure 4. The overall activation energy of this study (at 153 Wmol) is quite close to this value. which is very close to n-hexadecane (MW = 226). The lumping scheme represents an important tool to simplify the model. s-' 3 500 103 5 2 8 1 550 4 6 7 550 500 530 233 530 260 550 393 550 420 89 373 437 > The molecular weight of the gas oil fraction (>196OC) in the feed gas oil is approximately 221. It should be noted that LGO thermal cracking is a very complex process and it is almost impossible to identify all the chemical species present in the feedstock and study their chemical reactions.
4. 4. the desired product. According to the proposed model.10 to 4.8 M x m m Naphtha Yield. Figures 4.researchers using similar model feed. s 500 10 29 530 3 -6 550 1. a reaction rate constant at any temperature within the range tested experimentally can be obtained. A summary of the optimal reaction time at different temperatures is shown in Table 4. Table 4. Therefore. "C Residence time. wt% of product aiu 28 26 . there should exist optimal conditions for which maximum production of naphtha is possible.8 Optimal Reaction Times at Different Temperatures Temperature. naphtha.12 are the results of model predictions at several typical operating temperatures. is an intermediate product &om gas oil to gas. shown in Figure 4.6 Product Distribution Prediction Using calculated activation energies and frequency factors of four hypothetical reactions.8. it can therefore be inferred that the proposed fourlump model is reasonable and can be used to predict the product distribution.
Coke LGO LGO Data CokeData --***.10 Component Mass Percent vs. j Figure 4.Residence Time. Gas Gas Data . s / Naphtha Naphtha Data A / --. Residence Time at 500°C .
-. Gas 0 Data CokeData GasData 1 1 Figure 4. s -.Naphtha --...1 1 Component Mass Percent vs.Coke i Naphtha Data -LGO LGO A Residence Time. Residence Time at 530°C .
I.0 1 2 3 4 5 Residence Time. s Naphtha NaphthaData A ---. LGO Data Coke Data CI Gas Gas Data Figure 4-12 Component Mass Percent vs. Residence Time at 550°C .
Gas production increases continuously with residence time. .10 to 4.From Figures 4.12. it is also found that coke formation is very low. Gas is the other ultimate by-product aside fiom coke. This is expected because the feed has been hydrotreated.
Product distributions for different reaction temperatures were predicted using the proposed kinetic model. for the reaction of gas oil to gas was 83 kJ/mol and for the reaction of gas oil to coke was 262 kJ/mol. A four-lump kinetic model w s developed and evaluated to give a a better understanding of the product distribution.1 Conclusions A helical coil reaction system was capable of studying the reaction kinetics of the hydrotreated light gas oil thermal cracking. (1 990) for n-hexadecane pyrolysis in a Curie point pyrolyser.8 kl/mol obtained by Kumar and Kunrm (1985) and within the range of 212 to 228 kl/mol found by van Damme and Froment (1981). the optimal residence time was 1 0 seconds with a maximum naphtha production of 29 Wto?. at 530°C. The hydrotreated light gas oil thermal cracking reaction was studied at three different temperature levels. the activation energy for the reaction of gas oil to naphtha was 173 kJ/mol. for the reaction of naphtha to gas was 215 kJ/mol. the optimal . A kinetic analysis has been performed by assuming a fust order overall decomposition model for the light gas oil thermal cracking reaction. The activation energy for the reaction of naphtha to gas is close to 220.CHAPTER 5 Conclusions and Recommendation 5. The result is very close to the 164 kJ/mol obtained by Fairbum et a[. According to the experimental data. 500°C. It was found that at 500°C. The activation energy of this reaction was evaluated to be 153 idlmol. 530°C and 550°C.
it may be possible to understand the roles that different types of carbon play in thermal cracking reactions. naphthenes. olefins. ot . Among them.2 Recommendation Although boiling point range is of primary concern to industry.8 seconds with a maximum naphtha production of 26 wt%. aromatics) analysis on a GC and NMR spectroscopy may be the m s usell tools. 5. PONA (paraffins. With a more detailed analysis of feed and products. a model based on reactivity of different hydrocarbon structures may result in a more predictive model.residence time was 3.6 seconds with a maximum naphtha production of 28 wt% arid at 550°C. the optimal residence time was 1.
.L. 26. Academic Press. Beijing. Coulson. Cheng. 4' ed. Toronto (1983). 2.and K.and K. Ind.. P. Chem.M. 1239.F. "Lumping Strategy. 360-372 (1994). Albright.H.. D.967-976 (1989). "Modeling of thermal steam cracking of an atmospheric gas oil". Fluid Flow. Zabaniotou. Crynes and W.2151-21 57 (1 987b). Z. Depeyre.. Ind. Eng.Corcoran (eds. "Heavy Oil Processing Technology". New Y r (1990)..F. "Laboratory Reactors for Pyrolysis Reactions" in Pyrolysis: Theory and Industrial Practice. Flicoteaux. CBme. . Chem..G. China Petrochemical Press.). and J. Pergamon Press. Res. Ind. 26. L. Eng. "Lumping Strategy. C. F. Chem. Bischoff. Introductory Techniques and Applications of Cluster Analysis". Arbabzadeh and A. "Chemical Engineering volume 1.G. 1. J.B. G. A System Theoretic Approach". as ok Coxson.1248 (1987a).References ASTM D975-98% "Standard Specification for Diesel Fuel Oils" (1998). B.. Eng. Coxson. Bischoff. 28.B. Res. Res. Richardson. P. Heat Transfer and M s Transfer".
10' ed. "Thermal Cracking of Pure Saturated Hydrocarbons" in Advances in Petroleum Chemistry and Refining. Narayanan and A. Behie and W... P. Eng.J. G. Goossens. McKetta (ed).. Chem.. Saoford and O. 1537.E.J.O. B. Ind. Ind. and C.930106(1977). Sy. Achia.O.. van de Steene. P. van de Steene.K.495-504 (1976). S. 23. Nazarewycz. J.G.1545 (1990). Fuel. P.G. Satterfield. Dev. 15. B. J. Vanden Berghe and A. D. aa Gray. New Y ork: Interscience (1 964). .Y. Svrcek.. Van Damme.. Fhcoteawc. Frornent. "Ultrapyrolysis of n-Hexadecane in a Novel Micro-Reactor".C. 69. B.M. E. Froment.F. Fairburn. S. Eng.Y. H Yeniova.S. "Thermal Cracking of Light Hydrocarbons and Their Mixtures" AIChE J.. J. G. 1116-1 130 (1991). (1 994).F.N.A. M.Depeyre.O. "The Relationship Between Chemical Structure . A. "Thermal Cracking of Ethane and Ethane-Propane Mixhues". Fabuss. Gas Processors Suppliers Association. "Modeling of Thermal Steam Cracking of n- Hexadecane". Smith and C. Wanke. Process Des. Chem. "Engineering D t Book". Process D s Dev. Goossens. Knywicki. Jokuty.R. L.A. e. L. 30. U.
Department of Chemical and Petroleum Engineering. E.D. New Y o 4 884-886 (1947). McGraw-Hill Book Co. Gross.M.'LM~deling Naphtha Pyrolysis". 833-843 (199 1).. Eng. and M. J. Kumar.Kinetics and Catalysis". J. Gray. Chem.Y..Waston. and K. and D. N.. M. a Can. New York. Can. (1993). Hougen. Chem.. Eng. S. "A Lumping and Reaction B. KUIIZN.774-782 (1985). M. Jacob.. Dev. 24. Voltz and V. Alberta. University of Calgary.. Part Three . Darkly: Kinetics and Reactors for Complex Mixtures".. Thesis. Eng.. Chem. Crawford. "Therrnd Cracking of Athabasca Bitumen". M.and Reactivity of Alberta Bitumens and Heavy Oils".481-493 (1997).M.W. 22. Process of Des. 75. O. 'CThrough Glass.R. Calgary. . AIChE J. Ph.70 1-703 (1976). S. 69.Weekman.A. Canada (1 978).E. "Convective Heat and Mass Transfer". Kays. Hayashitani. John Wiley. 3rd ed. Ind. Scheme for Catalytic Cracking".. P. "Chemical Process Principles. W.
W.. Kunii. De Bmijn.39 (199 1.. Toronto. F.. Milne. F. J. Am. Bemti. F. Chem.Khorasheh.124-1 33 (1 969).Rice. J.J. p. Eng. "A Lumping Analysis in Monomolecular Reaction Systems". Berruti and L. ) Kuo. Chem. 590-595 (1 943). fluidized bed (ICFB):a novel solution to gas bypassing in spouted beds".W. Kossiakoff. 65. "High-Pressure Thermal Cracking of n-Hexadecane". and M. Eng. Levenspiel. Behie and T. Behie. Fundam. Can. 70. "Internally circulating . D.O.and J. A... 1853-1863(1993). "Fluidization Engineering".A.789 (1 994). Ind. Ind. Resonance Stabilization and Isomerization of Free Radicals" J. Wei. Milne. Butterworth- Heinemann.A. Chem.. . B. "Thermal Decomposition of Hydrocarbons.S. 8. Soc. 2nd ed. and 0.J. Eng. 32. U. L.R Gray. B J .C. b'Ultrapyrolytic Heavy Oil Upgrading in an Internally Circdating Fluidized Bed Reactor". Res. Chem..370. and F.910-9 15 (1992).Patent 5.
157-166 (1 999). "Recycling of Waste Plastics by Ultrapyrolysis . website: htt~://www.H. International Edition..T.html. Neurock.codnewsrelease/Fin~ Financeandstock-Quarted8-317. T. Green. Yasar and M.F. 570-575 (1 9 4 . D. last updated on April 19. Suncor Energy Inc. "Perry's Chemical Engineer's Handbook".Milne.. Giacobbo. and D. Stark..M. B J . Toronto (1 997). 10 1 . S. McGmw-Hill.A.M. May. Reaction. Using an Internally Circulating Fluidized Bed Reactor". 938943 (1965). Energy & Fuels.suncor. Angewandie Chemie. . Hydrocarbon Processing.2000 (2000). M.. 7'h ed. Trauth. F. "CPU Issues in the Representation of the Molecular Structure of Petroleum Resid through Characterization. '4Themal Fragmentation and the Determination of the Structure of Organic Compounds". R. Berruti. M. W. Stolfa. 9) Simon. 4. Perry. and H.W. 8. "New Roles for Thermal Cracking". Petti. and Monte Carlo Modeling". Joumal of Analytical and Applied Pyrolysis 51.109 (1980). Klein. Behie and F.L..
M..2000" (2000a). Syncrude Canada Ltd. Albright and B.D.M.518-530 (1994). 274295 (1976).syncrude.Syncrude Canada Ltd. M. 8. L. "Surface Reactions Occuning During Pyrolysis of Light Paraffh". Tsai. Rahimi..C. "The +525 Residue Before and After Hydrocracking with Bimodal Catalysts of Varying Macropore Volumes". T.F. shown in May 2000 (2000b). Petti.M. C. "Representation of the Molecular Structure of Petroieum Resid through Characterization and Monte Carlo Modeling".syncrude. in Industrial and Laboratory Pyrolysis ACS Symposium Series 32. P. Crynes (eds.. Neurock and M. "Library-News Releas- January 26.T.com/. American Chemical Society.F.D. M. . Clugston and H. Teman.. 576-580 (1994). H.L. D. and L. Washington. S. Albright.com/.). Trauth.F.EUein. D. Stark. 8. website: http://www. Energy & Fuels.. Dettman. website: htt~://www.M. Energy & Fuels. "Who We Are-Securing Canada's Energy Future---A Canadian Success Story-Syncrude in the 21st Century".
Ind.. Chem.L. SOC. Basu and I. P. 20.. AIChE. AIChE Monograph Series no 1l.H. and Kinetics in Practice".593-597 (1949). Large (eds).Models.W. "The Role of catalysts in Syncrude's Operation".. 51. Bergougnou. Chem. Propane-Propy lene Mixtures: Pilot Plant versus Industrial Data". V.. P S . Wei J. A. Chem.. Afara. Chem. 10651073 (1975).. and G. Ind.A. Eng. and J. 8.L. Briens and M.. Weekman. Voge. Good. 25-27 (1999). .Kuo.H. S. JulyIAugust.Narayanan and G.Fundam. Process Des. J.van Damme. C. Am. Eng. "Thermal Cracking of Higher P a r e " .New York (1979).71. "Scaling Up of Naphtha Cracking Coilsy'. "Lumps. and G F Froment.F. van Damme. 21.366-376 (1981). Circulating Fluidized Bed Technology II.. Vogiatzis. Yui. S. S. Can. .. News.114-123 (1969).F. P. W. "Thermal Cracking of Propane and . AIChE J. Dev.Froment. Toronto (1988). "Ultrapyrolysis of Cold Lake Bitumen". "A Lumping Analysis in Monomolecular reaction Systems".75.C. Peragarnon Press. H.S.
846-852 (1987). Ind. Eng.L.. Crynes. . Res.L. 26.. P. O.Zhou. Hollis and B. Chem. "Thermolysis of Higher Molecular Weight Straight-Chain Alkanes (C9-C22)".
Compositions of pseudo-components are calculated using Equations (4. The sum of Yi for the gas analysis includes only the hydrocarbons. the KOH concentration in the solution was very low (about 0. Tables A.10 are mass balance calculations for the Runs #1 to #8.3 to A.03 M) and the gas flow rate was high so that not much C02 was absorbed. An estimation of its average MW is given based on the MW of normal paraffins.4) to absorb any H2Swhich could potentially be produced. Runs #1 showed similar CO2 production to replicate runs.2 is the average MW of LGO pseudosomponent (>196"C) in the feed and a comparison to n-hexadecane. Table A. the balance is nitrogen. A Erlenmeyer flask containing 0.8). A KOH solution had been used for Runs #1 and #2.5 M Zn(Ach solution w s installed immediately before a the WTM (see Figure 3.2). A switch was made to t!e Z ~ ( A C ) ~ solution. The residence time was calculated based on Equations (4.1) and (4.Appendix A Mass Balance Calculation for Different Runs Table A. Zn(Ac)l . There was only 2 cm contacting length (the KOH solution was shallow).1 is the SimDist analysis of the whole feed light gas oil.5) to (4. including SimDist results of reaction liquid products and GC analysis results for the reaction gas bags. The weight percents of <9') naphtha hction ( 1 6 C and gas oil hction (>196"C) are also shown. because C02 could be absorbed by KOH solution.
.solution absorbs H2S but not C02. this absorption step was probably not necessary. For hydrocarbon feed such as the one used in this study.
0008463 0.5431 3 .1 0 I9751 8 2 6 0.0846267 0 0 0 0 1 0 3 0.57 0.0451 15 0.0533387 7.81 0.0328952 1 .0776626 18.6 .0078698 0 0 66499 5.2 0 0 0.0004 0.186246 .01 0. 2 0.4 4 8 0 0 55484 0.0367083 0.1057834 .6 .32841 0.6 1 212.4485384 .3963407 .0005778 0.639032 1 . 5 3.1084488 27.19041 1 .9 0 0 0 9.5 .0022081 0.0023563 0.000738 0.0522727 01 105917 21 8 7 6 .2961935 .0781 17 12.036923I 0. 2 .04 71 17E-05 0.wt% X30.0323962 4.0002929 0.7 .0057895 0 0122486 f .87 0.1 1 6 0.4 0 0 0.363E-05 0. .0001389 0.0512598 0.Table A.6276761 .04894 0.279369 .0846267 0 6 0.0001981 0.0 0 2 0.101989 18.0001779 0.1035433 17.0 SUM 2 2 24 26 28 30 32 34 36 38 40 42 SUM .472664 0.6 0 1 0. 1 .5500736 .wt% <196OC >196OC Gas Coke 9 1 0 11 1 2 1 3 1 4 1 5 1 6 1 7 18 3. wt% X40.24694 4 9 71 0.0955044 20.04 M W XlO.32 0.6 8.0482065 0.545993 30 8 6 0.5740904 1 3 5 7 0.91 1 10.58 0. 1 Molecular Weight Calculation of Whole Feed Feed C# wt% mol mol% ave MW 4 5 6 7 8 0 0 0 0 0. 3 8.0006197 0.0002938 0 0 1 1567 .0001506 0.0908988 20.9943638 .0252 1 0.87 0.8406309 0 4 0 0 1 1 137 0.0044536 0.004671 1.602355 12 8.0003763 0.0429646 0.% 0 01 3125 0. wt% X20.0094222 3.0306383 0.3 7 0.3 0. 9.35 0.0648205 18.1467087 .0003488 0. 0.765973 .06347 .0012224 0.
4931 4 22.091836735 0.8221 93 20.0005124 0 1133037 .09 0.565E-07 0.1046788 0.1204801 7 8 6 0.02827381 8.0005202 0 1150304 .847874 24.0001653 0.0045224 1 ave M W 13. 8.32 0.81 0.76 0.08 1 0.138392857 0.66 0.14 1 E-06 0.05 1232993 0.891 E-06 0.1228608 9.307286 1 1.00053 1463 9.328833 6.0940152 0 221.70685 1 30. 01 1 7 1 .0004251 7. 10.0862785 13.04 0.61 10989 0. 3 .831 1 2 2.14 0.103210034 0.5381 1 9 19.0006884 .004995748 1.601952 20.0051899 0.0104675 .0004798 0.0026177 0.0005556 0.734E-05 0.0002201 0.05 0.000956633 1.82 0.57 0.Table A.3290532 0 21 15342 .07201 1 .06122449 0.87 0.25201 1 7 3.347E-05 0.0003902 0.184E-05 0.555165 20. 5 9 0.87 0.0867833 8.088435374 0.4 4.02 0.326437 22.000418 1 0.0005449 0.1060997 9 71 0. 8.001488095 3 1 13E-06 0.35 0.47 0.1009832 .12378 .101721939 0.952E-07 0.365E-05 0 0184971 .26 0.009247449 2.1 0.0003925 0.0004567 0.002763605 6.0003257 0. 0. 1 6 0 0 732568 4.0365446 2.094281463 0.0448309 1.093643707 0.110012755 0.0001673 7 0 0 0 0 94.2 Molecular Weight Calculation of Feed (>196"C) Feed (>196OC) C# 11 1 2 MW 13 1 4 156 170 184 198 212 226 240 254 282 310 338 366 394 422 450 478 506 534 562 590 wt % 15 1 6 17 1 8 20 22 24 26 28 30 32 34 36 38 40 42 SUM normalized mol mol% 5.001358 0.
033 1858 0.16 0.35 159.61 1.5 5.574 83.3 Mass Balance Calculation for Run #1 outlet Iiq C# outlet gas W V 4 5 6 wt% mol 7 8 9 10 11 12 13 14 I5 16 17 18 20 22 24 26 28 30 32 34 36 38 40 42 58 72 86 100 114 128 142 156 170 184 198 212 226 240 254 282 3 10 338 366 394 422 450 478 506 0 0.02 3.02 9.5261838 0 0 0 58 54 72 70 86 4.0096766 0.87 0 0 0 0 0 0 0.0002778 0.32 125.52 4.0055 161 0.5%946 outlet fluid G% L% MW(G) MW(L) Ave M W 25.1088525 0.0748291 0.0414151 0.0224567 0.61 32.81 0.38 0.596946 175. g recovery 0.57627 1 32. s C 0.06205 13 0.22 0.42 0.68 7.0276378 0.5213028 0.191323 18.48 2.226 38.0530808 0.0594565 0.78 7.2 120.35 trapped by rayon.6 172.45 5.770134 19.g 0.314725 6.002574 0 0 0 0 0 534 562 590 SUM T 550 0 0 0 0 99.8 <196OC.84 10.0726518 0.1043007 0.314725 6.2411525 0.0140747 0.156757 9.0441228 0.51 8.02 1.71 0.0345 0.06%471 0.0931 162 0.0021923 I w 0 4 .402177 13.09012 139 Material Balance fed.96 1.2 4.0002193 212 1. g I 0 0 0 0 0 0 1 175.02 0.5 >l%"C.Table A.968 >C4 in gas bag.00451 53 0 0 0 0 0 0 0 0 0 ave MW MW CI C2= Y i 16 28 30 42 56 0 0.8237984 9.36 4.437002 15.09012 12.1 2 g <196°C > 196OC X I0 6 X20 94 G V(Reactor) v.69 3.g 40.94 10.3 704.5 0.0494366 0.0004873 0.% 0.46 7.1221773 0.0774017 0.88 27.0582157 0. g 165 SimDist for Liquid product g 118.62 C2 C3= C4= C4 C4H6 C5 Cyc-CS C6 SUM Ave MW (Gas) = 16.78 G.9997406 1.0426563 0.5781 191 12.22 0.0080233 0.44 78.06 MiYi 104.05 21.68 11.980988 20. % c.0 128014 0.0867234 0.2353523 0.0003471 0.3327857 2.04 15169 0.0350847 1.9640507 12.0484832 0.03 5.5700493 0 0 0 0 mol YO 0 0. mA31s MW(feed) Rest.0236667 0.2 104216 6.65 X1 X2 X3 X4 wt?! 0.423729 74.1 2.5829728 SUM 37.16 65.0521082 8.06052 1 1 0.
1187063 20.00 15574 0.6926769 .0060134 2.0033 3 0.6751 3 11 6 Material Balance feed.0009387 0.425926 86. 5 % 1 6.80651 .2 16 0.717664 .0772983 16.8 .0002118 212 1.0028263 1. g G. 2.18 1 0 9 <196OC.1301028 6.0558333 0.82 rccovery <196OC >196OC 534 562 590 SUM 0 0 0 0 0 0 1 0 11 0.03441 4 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 181.951 >C4 in gas bag.0246 0.58 0.39 0.0425943 0.035 0.6387516 SUM 34.0578261 0.55 0.48 0.0440141 0.Table A.0654118 0. >196"C.308994 06 1 .% 03 0.339667 181.g 2 . 2 0. 93 0.1049402 19.0635164 8.2 SimDist for Liquid product 0.886646 8.3 81 1 0. 4 0.09862 19.26 0. 2.1013238 15. X1 X2 X3 X4 0.0337795 0.0613016 15.0025 0.67513 SUM I Ave M W (Gas) = 1.57 0.9599355 .25 0. g 159.33%67 370.0132302 4.4642975 . 6. 6 0.56 outlet fluid G% L% MW(G) MW(L) Ave MW 13.526765 07 9 0 0.6412682 0.0072903 0.46 0.9 .3 26 29.0651224 14.0325257 .0308251 8.1274769 0.387239 .0272917 0.044643 4.0798748 1 1.3266559 1 0 0 0125581 0.0227899 1.1520198 3. s C 0.812 97.5510385 0.8 0.0495277 1 1.1 7 1 . rnA3/s MW(feed) Rest.2 114 5.1013465 1 1 0.0003471 0.0045369 0.0297368 0 0 3 5 4.570599 4 7 0 0169858 0.180077 11 1 .0005172 0.03 0.2299339 0. L.0544427 0I .0543434 0. % 3.34221 9 8 7 0. g g tWo 1 G T 530 X10 6 V(Reactor) v.035885 0. X20 94 0.574074 29.00132 1 I .4 M s Balance Calculation for Run #2 as outlet Iiq C# oullet gas MW 4 5 6 58 72 86 100 114 f 28 142 156 1 70 14 8 18 9 212 226 240 254 282 31 0 338 366 394 422 450 478 wt% 7 8 9 1 0 II 1 2 1 3 1 4 I5 1 6 1 7 1 8 20 22 24 26 28 30 32 34 38 40 42 mol mol% ave M W 0.088 1.
76587 2 1.14 8.5 W! XI 1 194.48 0.01 0.0654657 0. mA3/s MW(fecd) 0.7869122 0.0421503 0.1358515 0.0337324 0.0002488 1 .3952446 9.1205608 0.Table A.71399 184 28 394 0 0 0 30 32 34 36 38 40 42 422 450 478 506 534 562 0 0 0 0 0 590 SUM 0 0 0 100.580979 23.338908 c2 c3= C4= C4 WH6 C5 Cyc-C5 C6 SUM Avc MW (Gas) = outlet fluid MW(G) MW(L) Ave MW G% L% 6.7307183 SUM 30.0870755 0.371329 93.628671 32.0009722 0.1 <196OC.0073 0.0604348 0.3 0.0448673 0.0629412 0.958 >C4 in gas bag.95 X2 X3 X4 0.5 Mass Balmce Calculation for Run #3 outlet liq C# MW 4 5 6 7 8 9 10 11 12 13 14 wt% mol 58 72 86 100 114 128 142 !56 170 184 198 212 226 240 254 IS 16 17 18 0.79 7.4 167369 2.12 12.1 LI g 188.03504 17 0.1221522 0.0003346 0.0621212 0.09405 1 0. % CI 6 0.67196 20.4 0.0 123684 0.0512264 0.05 trapped by rayon.1516298 0.0002006 Res 4 s C 1.86 10.321585 20.3 10.688616 194.78 4.5822206 1.1 172879 0.04 12598 mol % 0.47 158.7 1 1.0141674 0. % 0.2961229 14.33 0.0034884 0. g recovery 0.13 12.56 10.05 200.0240038 0.8 >196"C.04846 15 0.076387 19.71399 212 <1%OC >196OC G T 51WI XI0 6 V(Reactor) v.679058 16.07 0.0018868 0.0001724 0.0003471 0.871046 21.00677 0. g g 12.736437 5.0680066 0.0800744 sve MW 0.0612093 0.73 1.4 1 2. g 0.71~ i r n ~ ifor Liquid product st fd.0994169 0.5152686 X20 94 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Material Balance 209.9 G i I ! 12.4 1 10.3 0.0194074 0.0217188 0.
0005172 mol% avc MW 0.000138 9.078235 73.2 13MS 5.2 3 w 0 .0243299 0.0698323 0.960 X 4 in gas bag g 0.0020034 0.1396646 0.38 X2 0.8610229 3.42 0.0005584 0.13 0.76 a= C2 c3= C4= C4 C4H6 C5 Cyc-CS C6 15 16 17 18 20 22 24 26 28 30 32 34 36 38 40 42 212 226 240 254 282 3 10 338 366 394 422 450 478 506 534 562 590 8.4 X4 0.03 0. 13 208.808E-05 6.24414 SUM Ave MW (Gas) = .0003471 0.0251667 0.0139362 0.5728009 0.06 SUM T 550 X 10 6 Material Balance feed.0410377 0.88 0.78 G.905E-05 5.0596355 0.918E-05 1 15.0002191 0.4 recovery 0.62 3.2483684 0.% 0.04 W F by rayon.03 0.0551629 1.0523742 0.559E-05 3.5043844 X3 0.2 SirnDist for Liquid product L.0523742 M W Cl 16 Y i 6.0009748 0.0439362 0.92 1765 32.g 52 0.0523742 0.002002 199.000903 0.0001255 7.303052 6.72 6.22 >196"C.0002194 212 1.0349161 0.071644 0. mA3/s M W ( f d ) Res 4 s C 0.0349161 175.226955 0.624 XI 0.1047484 0.93 1.6 M s Balance Calculation for Run #4 as outlet liq C# outlet gas M W 4 58 wt% mol 0.0003 104 0.776 100.0045453 0. L: 147.0003081 0.08 0. 32.5821185 S U M 0.02 >196OC 100.39B05 0.0098554 0.0056452 0.0005378 0.3840776 0.544676 13.2452452 G 49 X20 V(Reactor) v.4 C 196OC.7 7.7332391 0.5363105 0.0001778 0.477633 10.24414 18.8 1 94 0. .188524 13.75 0.02 W? & <196OC 49.22 0. g ~ (A 0.04 7.924579 outlet fluid (3% L% MW(G) MW(L) Ave MW 138 26.924579 175.03 0.Table A.0026036 0.034 1593 0.0011475 0.618E-05 3.61 MiYi 105. % 0.
59346 03 9.8958214 94.041771 0.0597826 0 1184919 21.3 0. g 0.0029%1 0.0003471 0.74 0.DO C1 C2= C2 C3= 0% C4 C4H6 C5 C~C-C5 C6 1 6 28 30 42 56 58 5 4 72 70 86 SUM Ave MW (Gas) = 17 .033662 0. 7 1 .1 1.09 09 .9 00 .0018613 0.710482 27. 04 . 9 0.9 2.868 9.0166909 1.0212411 0.012 0 6.0566524 0.7 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 100.5747934 O % 0. 1 .0490385 0.0792818 0 0.0002861 1 C 1 C-r + .71995 7 189 Material Balance feed. 0.13 0.77 0.52 outlet fluid G% L% MW(G) MW(L) Ave MW 5.29 0.9027644 . <1 %OC > 196°C T 500 X 10 6 94 G V(Reactor) v.67 0.084807 19.04 0.8 .0009391 0.1369499 0.84 48.84 46.3 16 .1041 9 32.05 trapped by rayon.02 0.Table A.84 0.75 X1 X2 X3 X4 wtO/o 0.6686521 .56 0.1 10 . 1 0.0633529 0.7 Mass Balance Calculation for Run #5 outlet lq i C# outlet gas MW 58 72 86 100 14 1 128 12 4 156 170 1 84 18 9 212 226 wt% mol mol% ave MW MW Yi MiYi 4 5 6 7 8 9 1 0 11 1 2 1 3 1 4 1 5 1 6 1 7 1 8 20 22 24 26 28 30 32 34 36 38 40 42 240 254 282 310 338 366 394 422 450 478 506 534 562 590 SUM 0 0 0 0 0.872457 . >196"C.097139 20.0421009 1 1.0092258 0 0 8286 5.1 10 .4667142 0 3 0.37 0.710482 198.0490094 0.03375 0.0740383 4.802508 .% 6.7333571 .932 140.36 28.960>C4 i gas b g g n a I3 96 . 1 1 9 0.744913 .001511 0.5045291 1 198.0668941 16.0396409 5.3 42.% 03 0.48 1 1.095 0.78 0.0971965 15.162653 10.19 0 0.166387 81 0.0040074 1.1255685 21.5 174.84 0 212. mA3/s MW(feed) Res f s C 0.008421 1 0.0827933 21.06671% 9.8061116 0.2 08 . 2. t3 Sg mg e (2 g recovery 182 SimDist for Liquid product 1 4 4 <I 96OC.71995 .7292161 SUM 23.0040533 0.497 32.0005556 0.257666 6 0.02 0 6.0002007 212 1.4741808 7.0080337 2 7154033 .0029 0.0020219 0.0057479 0. 0.0011 1 1 0.26 0.02951 06 5 9 0.054574 .1 1 .0605051 01 199238 23.0427876 0.65 0.86 0.346638 1 1 0.
02 17083 0.87 3.75 29 >196"C.61 3.2609838 0.75 0 366 394 422 450 478 506 534 562 590 0. N .84 0 483.2553174 0 0 0 0 0 0 0 0 outlet fluid G% L% MW(G) MW(L) Ave M W 27. g g 7.0022189 0 0 0 0.9 SimDist for Liquid product 75.325 51.4878223 0.0413167 0. C-.0109929 0.7 I c .735385 93.0002219 212 1.735385 167.1 1.003714 SUM T 550 XI0 6 0 0 0 0 0 0 0 0 0 100.1886454 2.23 0 0.953 >C4 in gas bag.2483286 0.25 0.37 0.0363345 0.024685 0. 6 0 0 0 0 0 0 0 0 0 1 167.27 3.36 0.0028653 104.2 17 18 20 22 24 26 28 30 32 34 36 38 40 42 240 254 282 3 10 338 5.0003471 0.48 0 13.8 Mass Balance Calculation for Run #6 outlet liq outlet gas MW wt% C# mol mol?40 ave MW MW Yi MiYi C1 a= Q C3= C3 C4= C A C4H6 C5 C6 16 28 30 42 44 56 58 54 72 86 SUM Ave MW (Gas) = 5. g 0 L&? G.777778 72.0073969 0.5974584 XO 2 94 0 0 0 0 0 8.075 26 0.82 1.3 150.9782 1 130 Material Balance feed. g 0.36 18.2930465 1. mA3/s M W ( f d ) Rcs t.7202717 10.72 0 16.2 1 6.0044194 0.222222 29.58 0.3 trapped by rayon.Table A.3 G V(Reactor) v.25 29.34 0 51.4 < 196°C % 0.494454 5.9~~1 6s recovery 109.96 48.92 106. % 0.42 0 0.0183995 0. s C 0.5645645 SUM XI X2 X3 X4 wt?? 0.3 3 <196°C >196°C 27.
353979 (3% L% MW(G) MW(L) Ave MW Material Balance 38 40 42 534 562 590 SUM T XI0 0 0 0 0 0 0 100.29 4.5075922 0.36 0 18.0021692 .3 XI X2 X3 X4 0.64 133.773 155 C6 SUM Ave M W (Gas) = outlet fluid 22 3 10 1.39 0.12 0 601.56 0 69.5 0. mA3/s MW(feed) 37.9661>~4 gas bag.9 Mass Balance Calculation for Run #7 outlet liq C # outlet gas M W wt% mol mol YO avc M W M W C1 Y i 16 28 30 42 44 56 58 54 72 86 7.1 4.96 0 20.Table A.2 30.5904896 X20 0 0 0 0 0 0 1 169.44 0 MiY i 116.12 0.2697035 0.6 19.0044839 0.32 0.89293 recovery <196°C > 196°C G Res f s C 0.78 2. g in 6 wt% 3-91 V(Reactor) v.3 0.2205351 0.3 70.19 29.0075935 2.84 C2= C2 C3= C3 C4= C4 C4H6 C5 63 180.32 0 0.
1.l shows the temperature of the inlet and outlet fluid. The average of the reactor inlet and outlet temperature (dashed lines) in the figures was used as reaction temperature.1 shows the different stages during an experiment and how the reaction temperature was evaluated. Nitrogen continued to purge after the feed oil was stopped. . Figure B. The temperature data were obtained by the LabView on-line data acquisition system once every two seconds from thermocouples #19 (inlet) and #20 (outlet). Figure B.Appendix B Temperature Profde for Different Runs Figures B l to 8. After feed oil entered the reactor. it affected the reactor fluid flow rate and caused the dynamic stage. #8 respectively.8 show the temperature profile along with time during the Runs #I to . The reaction temperahue was determined using the steady state interval shown in the figure B. Nitrogen had been purging before feed oil started.
The pump had a knob with scale ranging from 0 to 100. These two pumps were set i parallel and the n calibration w s done with both pumps on and at the same setting on the scale. .Appendix C Pump #2 and Pump #3 Calibration Figure C.l shows the calibration results for the pumps #2 and #3. The a pumps were stabilized with hydraulic oil circulating in a recycle loop for one half hour and then the flow w s opened to the transfer cylinder for 10 minutes before the a calibration for each point.
When the valve is switched clockwise. with the shaft rotated counter-clockwise. the columns' relative position in the sequence changes. corresponding hydrocarbon position and emerging sequence in the columns at different times. This strategy was used to realize an effective separation of CO and CO2. Once valve #1 is turned on. Figure D 2 shows the valve timing .Appendix D Valve Switching Scheme for Gas Analysis on GC Two valves (#I and #2) were installed in a heated valve box on top of the HP 6980 GC oven. 1 shows column connection sequence with both valves in OFF position. Normally the valves were in the OFF position. Valve #I has ten ports and valve #2 has six ports. Table D. The MS column and its contents are shown on the left.s and 6. it is the ON mode. Therefore.. 6 and 7. Figure D. . 10 and 1 are connected. port 1 and 2.1. These pneumatic valves were automatically actuated. illustrates I the operating valve switching sequence.3 and 4. logic. for valve #1 in Figure D. 4 and 5. 8 and 9. when it is in OFF mode. while the PLOT Q column and its contents are shown on the right of Figure D 2 .7 and 8. For example. port 2 and 3.9 and 10 are connected.
lw Sample trapped in 25 pl sample loop.0 min OFF ON Flow reversed on MS. Plot Q isolated. Light gases to CO through Plot Q to MS.35 rnin OFF OFF OFF OFF OFF Helium f o through MS to Plot Q. C02 and heavier hydrocarbons go to TCD.0 min 1. containing C 0 2 and heavier hydrocarbons.1 min OFF OFF Flow through MS then Plot Q. Light gases to CO ON ON ON are analyzed by TCD. 4.Table D l Valve Status and FlowISmpie Status . COz and heavier hydrocarbons still trapped on Plot Q. Time Valve #I Valve #2 Flow/Sample Status Initid Injection 0. 4. .
35 min LON.2 Valve Timing Logic . Valve 2-OFF MS PLOT Q TCD [ I CO Cl N2O2 Adjustable Restrictor TCD 1. 2-ON I TCD --I b TCD - b TCD Figure D.Valve 2 -OFF.
m fluid velocity. where: Re d u P Ir Reynolds number reactor tube ID. m2 For Run #1. Pas fluid mass flw.2 were used to estimate the Reynolds number.13 14 g/s and the LGO/N2 mass ratio was 2.34 mm was used to obtain a reactor fluid .Appendix E Estimation of the Reynolds Number and Heat Transfer Coefficient Equation E. kg/rn3 fluid viscosity. the LCO mass f o rate w s 0. 1 and E. To give an estimate of the Reynolds number at the inlet end. the smaller reactor tube ID of 3. m/s fluid density. kg/s as G k A cross-section area of the reactor tube.0 lw a (see Appendix G for detailed calculations). kg/m2~s fluid m s flow rate.
W/mK Pr h k . Nitrogen viscosity at 500°C was found to be 3 . p. The viscosity of the hydrocarbon gases in the reactor was estimated to be 1.23-39). 343). 6 lo-' Paas w s used to calculate the Reynolds number.2-321).5 kg/m2-s. 1997. 1990. w/rn2~ thermal conductivity. p. 1994. An average reactor fluid viscosity of 2 . p.6x10-~ Paas based on molecular weight (200g/mol) and temperature (GPSA data book. it can be assumed that the reactor fluid is in the turbulent regime.mass flux of 22. By ~ a considering the enhancement of helical geometry.. It was found to be 2900. The Colbum equation was used to calculate the heat transfer coefficient (Coulson and Richardson. where: Nu Re Nusselt number Reynolds number Prandtl number heat transfer coefficient. 6 ~ Pa*s (Perry and Green.
substitution of Pr = 0.74.74 in equation E.05 WImK. 1990. the Prandtl group usually has a value of about 0.23-47). 399) and the heat transfer coefficient for the coil geometry was found to be 21 0 w/rn2~. The following equation w s used to account for this effect (Coulson and a Richardson. p. somewhat improved heat transfer is obtained for the same physical conditions. From the above equations. L .TJua mc.When gases are heated or cooled. 1994. p. 1990.053 WImK was used for the reactor fluid thermal conductivity. for With a helical coil. An average value of 0. 344): The thermal conductivities of nitrogen and hydrocarbon gases at 500°C are 0.056 WlrnK and 0. p. respectively (GPSA data book.3 gives the equation (Coulson and Richardson. h(coi2) = h(straight pipe) d(tube ID) d(he1ir) From the energy balance equation: AT -= d h ( T d . the heat transfer coefficient can be calculated as 187 w/rn2~ straight tube.
To heat the fluid entering the reactor f o 350°C to 500°C would require 0. . Boiling heating transfer rates are generally higher or on the same order of magnitude as wall to gas heat transfer rates. 1994. respectively. J/kgK AT - L fluid temperature change along the reactor tube.where: CP heat capacity of reaction fluid.7. An average value of 2075 JIkgK was used for the reactor fluid heat capacity. p.23-3). rm It should be noted that the proceeding calculations are based upon wall to gas heat transfer.185 m or approximately 20 cm. the fluid temperature rose at a rate of 8 10 W m along the reactor tube. Wm The heat capacities of nitrogen and hydrocarbon gases at 500°C are 1050 JIkgK and 3100 J/kgK (GPSA data book. From Equation E. the lower bound of wall to gas heat transfer has been used in the analysis. Therefore.
5 kg/m2*s(Equation E.73 s was used. 53) was used to calculate the friction factor. rn fiction factor fluid mass flux. rn/s For Run #I. G was calculated to be 22. p. kg/m2+s f G L u reactor length. rn fluid velocity. The Blasius equation (Coulson and Richardson.2). 1990. . 54). the average fluid velocity can be calculated using the residence time obtained in Appendix A and the reactor length of 18 m. 1990 p.Appendix F Estimation of the Pressure Drop along the Reactor Tube The pressure drop along the reactor tube can be calculated using the following equation (Coulson and Richardson. Here a residence time of 1. where: d reactor tube ID.
The actual pressure drop must be based upon the actual fluid velocity. . 27 kPa is estimated to be an upper bound on the pressure drop over the reactor. which changes along the length of the reactor. the pressure drop along the reactor ) tube was estimated as 27 kPa. These two effects counteract each other.Assuming a Reynolds number of 2900 (Appendix E . The velocity tends to increase due to cracking but decreases because of increased diameter in the last half of the reactor. This is low enough to be neglected in the current study.
Appendix G Calculations of the LGO to Nitrogen Ratio The ideal gas equation of state was used to calculate the nitrogen mass flow rate because of the high temperature and the low pressure. .058 Lls at 23OC which is equivalent to a mass flow rate of 0. If the m s flow rate of the feed oil is as 0. where: mass flow rate of nitrogen.13 14 g/s. J/mol K room temperature. m3/s gas constant. 296 K For Run #1.067 gls can be obtained.0. the nitrogen volume flow rate is 0. kg/s volume flow rate of nitrogen at room temperature. the LGO to nitrogen mass ratio will be 2.
Appendix H List of Failed Experiments .
"C Comments 1 2 3 4 99-05-26 99-06-05 - 500 500 600 620 580 550 40 Run failed due to high feed f o rate and very short feed time. s Temperature. feed ran out during bleeding. 1 Summary of Failed Experiments Run Date Reactor Feed Time. Poor mass lw balance. - . 99-06-17 99-06-23 99-06-25 99-07-16 99-07-28 62 57 Run failed due to high temperature and insufficient packing of rayon. no oil enter the reactor. Oil went into the piping after the demister.Table H. The SimDist results showed no thermal cracking occurred. Some liquid products went into piping after the demister and gas bag. Pump problem. LabView did not record temperatures properly (large fluctuations). 500 877 8 99-08-08 99-08-1 9 99-13-23 540 9 10 540 - No products exited the reactor. 5 6 7 60 Run failed because no feed entered the reactor. Feed went through reactor very fast because of elevated preheater pressure. Feed flow rate was too high. Run failed due to no bleeding of feed prior to directing it to the reactor. Thermocouples were out of order after heating up to 600°C. Forgot to refill the transfer cylinder.
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue reading from where you left off, or restart the preview.