RECOVERY OF OIL FROM UTAH'S TAR SANDS Final Report for Contract #ET77-S-03-1762 for the Period July 1, 1979 - November 30, 1979 by Alex G. Oblad, Principal Investigator James W. Bunger Francis V. Hanson Jan D. Miller J. D. Seader and D. K. K. M. R. V. J. Cogswell Hanks Jayakar Misra Smith Venkatesan Weeks The Department of Mining and Fuels Engineering and the Department of Metallurgical Engineering, College of Mines and Minerals Industries and The Department of Chemical Engineering College of Engineering University of Utah Salt Lake City, Utah 84112 i
Transcript
RECOVERY OF OIL FROM UTAH'S TAR SANDS
Final Report for Contract #ET77-S-03-1762
for the Period July 1, 1979 - November 30, 1979
by
Alex G. Oblad, Principal Investigator James W. Bunger Francis V. Hanson Jan D. Miller J. D. Seader
and
D. K. K. M. R. V. J.
Cogswell Hanks Jayakar Misra Smith Venkatesan Weeks
The Department of Mining and Fuels Engineering and the Department of Metallurgical Engineering, College of Mines and Minerals Industries
and
The Department of Chemical Engineering College of Engineering University of Utah
Salt Lake City, Utah 84112
i
RECOVERY OF OIL FROM UTAH'S TAR SANDS
ABSTRACT
This project is designed to develop necessary engineering data and
technology for recovery of oil from Utah's tar sands. Progress reports
for four major aspects of this project, namely Hot Water Recovery, Energy
Recovery in Thermal Processing, Effect of Variables in Thermal Processing
and Bitumen Processing and Utilization are covered. Efforts have progressed
to the point where collaboration with engineering companies for pilot plant
development in preparation for commercialization has commenced.
Hot water recovery technology has been shown to be technically feasible
for application to high and medium grade Utah tar sands. Utah tar sands are
generally believed to be oil-wet and the conditions for efficient separation
differ appreciably from those practiced in commercial operation with
Athabasca, Canada tar sands. The occurrence of high silica, low clay
content tar sands in Utah may dramatically reduce water requirements and
may eliminate the need for tailings ponds as required for Athabasca tar
sands. Further work is required to prove this point. In recent work, a
factorial design study of the major operating variables in the flotation
step of the two step hot water process using Asphalt Ridge material has been
carried out. Preliminary results are presented in this report.
Considerations of energy balance and recovery in thermal processing
show that there is sufficient energy available from combustion of coked
external energy must be input, preferably through the introduction of coal
in the combustion zone.
An energy efficient process concept, using heat pipes for energy
transfer has been tested and shown to be attractive from a conservation
standpoint. Further efforts must be made to prove out economic viability
ii
and operability relative to other thermal process configurations.
A fluidized bed recovery process employing an arrangement of steps
similar to that used widely in catalytic cracking has been studied in this
laboratory. The principal variables effecting recovery and product quality
are temperature, solids retention time, particle size, and particle size
distribution. For a specified solids retention time, an optimum temperature
for production of liquid products exists, below which insufficient production
occurs, and above which raw crude oil is cracked to form more gases. Yields
of greater than 80% raw crude oil are anticipated at residence times of less
than 20 minutes.
Detailed studies of energy recovery methods in thermal processing have
been initiated for a two stage fluidized bed system just described.
Characterization studies on extracted bitumen and the synthetic crude
liquids obtained during pyrolysis were initiated to determine molecular
composition of these tar sand oils and to develop concepts of reactions
occurring during the pyrolysis. Some results of this work are presented.
Virgin bitumen can be converted to raw crude oil by a variety of
primary upgrading processes including visbreaking, coking, catalytic
cracking and hydropyrolysis. Compared to coking, direct catalytic cracking
provides higher quality products in greater yields; however, optimum
conditions for catalytic cracking have not been identified and comparative
economic analyses have not been made. Bitumen can be converted in virtually
100% yields to hydrocarbon gases and liquids by hydropyrolysis with the
addition of 1 to 3 wt. percent hydrogen. Hydropyrolysis products show good
promise as a catalytic cracking or a steam pyrolysis feedstock.
Steam pyrolysis of bitumen to produce chemical intermediates is now
underway.
i i i
TABLE OF CONTENTS
Title Page *
Abstract ii
Table of Contents iv
List of Tables vi
List of Figures -viii
Introduction , 1
Hot Water Recovery 3
Low Temperature Separation Technology 6
Hot Water Processing of Utah Tar Sand 7
Experimental Procedure 9
Hot Water Separation Test 9
Analytical Technique 10
Bitumen Viscosity Measurements 10
Molecular Weight Determination 12
Particle Size Analysis 12
Results and Discussion 13
Tar Sand Properties 13
Bitumen Viscosity 13
Molecular Weight 18
Sand Analysis 18
Hot Water Process 20
Effect of Soda Ash 22
Effect of Diluent 26
Product Characterization . . . 26
Particle Size Analysis .26
iv
Summary and Conclusions 35
References 39
Energy Recovery in Thermal Processing 41
Energy-Efficient Thermal Processing Concept 46
Experimental Apparatus 56
Experimental Results 62
Conclusions and Recommendations 79
References 83
Effect of Variables on Thermal Processing 85
Experimental 86
Results and Discussion 89
Effect of Temperature on Product Yield and Distribution . . 89
Effect of Temperature on Product Quality 92
Effect of Solids Retention Time on Yield 94
Effect of Particle Size and Particle Size Distribution on 99
Yield
Conclusions 103
References 104
Bitumen Processing and Utilization 105
Visbreaking 105
Coking 108
Catalytic Cracking 117
Hydropyrolysis 128
Conclusions 136
References 137
Bibliography 139
v
LIST OF TABLES
World Reserves of In-Place Bitumen 4
Extent of Utah Tar Sand Deposits and Their Average
Bitumen Content 4
Existing and Proposed Commercial Operations 4
Average Molecular Weight of Utah Bitumen 19
Hot-Water Separation of Sunnyside Tar Sand 19
Thermal Recovery Processes 43
Processing of Tar Sand Triangle Material 72-74
Effect of Temperature on Yield and Product Distribution,
Sunnyside Feed 90 Properties of Synthetic Crude from Sunnyside Bituminous Sand 93
Effect of Solids Retention Time on the Yield and Product Distribution, Sunnyside Feed 95
Effects of Feed Particle Size on Yield and Product Distribution, Sunnyside Feed 100
Gradient Elution Chromatographic Analysis of Extracted Bitumen and Synthetic Liquid, Sunnyside Feed 1C2
Coking Product Yields from Various Bitumens 109
Analysis of C,-C,- Gas from Coking of Various Bitumens 109
Liquid Condensate Properties from Various Bitumens Ill
Simulated Distillation Yields of Pyrolysis Condensates
from Various Bitumens 113
Analysis of Coke from Various Bitumens 113
Order of Reaction (Power Function) for Coking of Asphalt
Ridge Bitumen at Various Temperatures 115
Group-Type Analysis of P.R. Spring Saturated Hydrocarbons 119
Results of Catalytic Cracking of Asphalt Ridge Bitumen 122
C. to C, Gas Analysis from Catalytic and Thermal Cracking
of Asphalt Ridge Bitumen 127 Analysis of Gasoline from Run Bt(4) 127
vi
Dll. Yields and Process Conditions for Hydropyrolysis of Asphalt Ridge Bitumen 130
D12. Liquid Product Characteristics from Hydropyrolysis of Asphalt Ridge Bitumen 132
D13. Comparison of Yield and Conversion Results for Primary Processing of Asphalt Ridge Bitumen 134
vii
LIST OF FIGURES
University of Utah Tar Sands Research and Development Program 2
Major Tar Sand Deposits in the State of Utah 5
Modified Hot Water Process for the Separation of Bitumen from
Low Grade Utah Tar Sands 11
Flow Curves for Sunnyside Bitumen, at Various Temperatures 14
Arrhenius-type Plot Illustrating the Effect of Temperature on Viscosity for Bitumen from Various Utah Tar Sand Deposits 15 Arrhenius-type Plot Illustrating the Effect of the Bitumen Preparation Technique on the Measured Viscosity of Asphalt Ridge Bitumen 17
Particle Size Distribution of Sand from Four Different Utah Tar Sand Deposits, Sunnyside, P.R. Spring, Tar Sand Triangle and Asphalt Ridge 21
Quality of Separation as a Function of Sodium Carbonate Concentration for the Sunnyside Sample at a Diluent to Bitumen Volume Ratio of 0.2 23
Quality of Separation and Recovery of Middling as a Function of Digestion Time at 0.2 M Sodium Carbonate and a Diluent to Bitumen Volume Ratio of 0.2 25
Quality of Separation for the Sunnyside Sample as a Function of Diluent to Bitumen Volume Ratio for Different Types of Diluent at 0.3 M Sodium Carbonate 27
Particle Size Distribution of the Sand in the Feed and Products from a Typical Hot Water Separation of the Asphalt Ridge Sample . . . . 29
Particle Size Distribution of the Sand in the Feed and Products from a Typical Hot Water Separation of the Sunnyside Sample 30
Arrehenius-type Plot Illustrating the Effect of Temperature on Bitumen Viscosity for Products from Hot Water Separation for the Sunnyside Sample 32
Influence of Flotation Temperature on the Coefficient of Separation at 0.05 M Na CO and 1000 rpm 34
The Effect of Flotation Temperature on the Size Distribution of Sand Entrapped in the Bitumen Concentrate 36
University of Utah Process 48
Conceptual Scheme for Commercial Plant 52
viii
Sample Material and Energy Balance for Pyrolysis 53
Sample Material and Energy Balance for Combustion 54
Energy-Balanced Operation 57
Laboratory System 58
Instrumentation Diagram 63
Apparatus for Fluidizing Studies 64
Fluidized-Bed Pressure Drop at 600°C 66
Minimum Fluidization Velocity at Various Temperatures 67
Material Yields with Tar Sand Triangle Feed 75
Simulated Distillation of Tar Sand Triangle Bitumen and Oil,
(Run No. 57) 80
Fluid Bed Coker for Bituminous Sands 87
Effect of Reactor Temperature on Product Yield and Distribution
for Sunnyside Feed 91 Effect of Retention Time of Solids, 6 avg, on the Yield Pattern for Sunnyside Feed 96
Effect of Retention Time of Solids, 6avg, on the Yield Pattern for Sunnyside Feed 97
Effect of Retention Time on the Optimum Temperature for Maximum
Yield of Synthetic Crude 98
Vicosity of Visbroken Bitumen 107
Arrhenius Plot for Coking of Asphalt Ridge Bitumen 116
Yields from Isothermal Pyrolysis of Asphalt Ridge Bitumen 123
ix
INTRODUCTION
The importance of developing a viable synthetic fuel industry in the
United States to provide for our economic and political health can no
longer be seriously questioned. Domestic tar sands are expected to become an
important source of this needed crude oil. Ultimately, major quantities of
synthetic crude oil must come from coal and oil shale as well as tar sands.
However, synthetic crude oil from tar sands more closely resembles con
ventional petroleum refinery feedstocks than do syncrude from coal or oil
shale and tar sand syncrude is already being produced commercially in
Canada. Thus, tar sands are a prime candidate for early development of a
synthetic crude oil industry.
The University of Utah has been engaged in recovery, processing,
characterization, and utilization research applied to Utah tar sands since
1974. The structure of this program is given in Figure-1. The program
efforts have been directed toward an above ground mining and recovery
technology while other laboratories, namely the Laramie Energy Technology
Center, have emphasized in-situ recovery technology. Both hot water and
thermal methods have been studied at the University of Utah for recovery
of the bitumen from the sand. The work has progressed to the point where
collaboration with engineering firms for design and construction of pilot
plant development has been initiated.
The following is a progress report on the four major areas of recent
research emphasis:
Section A: Hot Water Recovery.
Section B: Energy Recovery in Thermal Processing.
Section C: Effect of Variables in Thermal Recovery.
Section D: Bitumen Processing and Utilization.
1
UNIVERSITY OF UTAH TAR SANDS RESEARCH AND DEVELOPMENT PROGRAM
RECOVERY PROCESSING AND UTILIZATION
MINING STUDIES
FEED PREPARATION
FEED PREPARATION
HOT WATER continuous extraction with water recycle
THERMAL continuous fluid bed retort with combustion of coke
temperature controller, heating system, SCR speed controller and tachometer.
Inside the reactor the feed material was contacted with a mixture of the hot
aqueous solution of soda ash and diluent and stirred at constant temperature
(95°C) and speed (750 rpm) for a specified digestion time. Diluents were
added at a specified volume ratio based on the feed bitumen content.
At the end of digestion, ideally, the bitumen has been displaced from
the sand and can be separated from the sand by a modified flotation technique.
Constant air flow rate and moderate stirring speed were maintained during
flotation. Neither frother nor collector was added to the flotation cell.
During flotation relatively large lumps of nonflotable sand-bitumen aggregates
(middlings) were found with the tailing. This middling was recovered from
the tailing by screening at 20 mesh. The grade of the middling was suffi
ciently high to be recycled. A schematic representation of the processing
strategy is presented in Figure A2.
Analytical Technique
Representative samples of feed, concentrate, middling and tailing
obtained during experimentation were analysed to establish their composition
with respect to bitumen, sand and water, using a set of Dean and Stark
tube assemblies. The assemblies were set up in accordance with the procedure
described by the U.S. Bureau of Mines (A19). The specially designed flask
and operational details have been described by previous investigators (A3).
Bitumen Viscosity Measurements
The bitumen viscosity of the tar sand samples is of great importance from
a processing standpoint. Samples of pure bitumen were prepared from feed
10
Tar Sands
1 SIZE
REDUCTION DIGESTION
Diluent
Na 2 C0 3
Solution
Water
Air
0 r
\ FLOTATION /
*
-*• Bitumen Cone.
Middlings — (recycled)
SCREENING
Water
Tailings
Figure A2. Modified hot water process for the separation of bitumen from low grade Utah tar sands.
11
sources using a carefully controlled procedure, because the method of
preparation has a significant effect on viscosity measurements. In this
regard, bitumen from Asphalt Ridge, P.R. Spring, Sunnyside and Tar Sand
Triangle samples were prepared under identical conditions. Preparation
of bitumen for viscosity experiments was effected by Soxhlet extraction
using benzene as solvent and a glass extraction thimble of porosity "A"
to hold the tar sand sample. After complete extraction, the benzene
extract was filtered slowly using a 4.0-5.4ym fritted glass filter.
Removal of the solvent from the bitumen was effected by flash distillation
using a Rotary evaporator at 90 C and 4mm Hg pressure.
The viscosities of the four different bitumens were determined with
a rotational viscometer, Rotovisco. The sample was introduced in the gap
between a stationary cylindrical cup and a coaxial rotating cylinder which
was turned at a specified angular velocity. The instrument is equipped
with an electrical torsion dynamometer to measure the torque required to
maintain the specified angular velocity. From these measurements the "flow
curve" for the fluid can be established. The viscosity is then calculated
from the slope of the flow curve at a given rate of shear.
Molecular Weight Determination
Average molecular weights of the same bitumen sample that had been
used for viscosity measurements were determined by Vapor Pressure Osmometry
in benzene using the Model 117 Molecular Weight Apparatus manufactured by
Corona Electric Co., Japan.
Particle Size Analysis
From a processing standpoint, particle size distribution is an important
property of the tar sand feed. The sand size distributions for Utah tar sand
12
samples were determined by conventional wet/dry sieving techniques in the
size range of 590ym to 38ym.
Results and Discussion
Tar Sand Properties
Some of the more important tar sand properties, with respect to the
development and characterization of hot water processing technology, are
the bitumen viscosity, bitumen molecular weight and the particle size
distribution of the sand.
Bitumen Viscosity. The rheological properties of a fluid are well
described by the relationship between the shear stress applied to a fluid
element and the rate at which the fluid element is deformed under the applied
stress. The above relationship referred to as a flow curve, or rheology
diagram, is a distinct characteristic of a fluid and is useful to describe
the behavior of a fluid. This is particularly true in the case of bitumens
because their complex structure may result in unpredictable rheological
properties.
Flow curves for the Sunnyside bitumen sample are presented in Figure A3
for various temperatures. The linear response in each case indicates that
the Sunnyside bitumen sample exhibits Newtonian behavior over the range of
applied shear rates. Bitumens from Asphalt Ridge and P.R. Spring also have
been shown to exhibit Newtonian behavior (A3), results which were confirmed
in this present study. From a practical standpoint, it is important to
note that the viscosity of the Sunnyside bitumen is approximately one order
of magnitude greater than the viscosities of the P.R. Spring and Asphalt
Ridge bitumens in the temperature range studied, as shown by the data pre
sented in Figure A4. Indeed this difference in viscosity undoubtedly accounts
for the poor separation that had been reported previously (A3).
13
E o
(A
c >> •o
n i O
(f) (/) UJ
«
w
< UJ X w
Sunnyside Bitumen
10 20
RATE OF SHEAR , sec"1
30
Figure A3. Flow curves for Sunnyside bitumen, at various temperatures.
o & D
Sunnyside Asphal t Ridge P.R. Spring
J_ 2.8 3.0
>K - i
3.2
( 1 / T ) X 1 0 3
Figure A4. Arrhenius-type plot illustrating the effect of temperature on viscosity for bitumen from various Utah tar sand deDosits.
15
Furthermore, analysis of bitumen from a Tar Sand Triangle outcrop sample
which appeared to be extensively weathered indicated that the viscosity of
this bitumen was even higher than the Sunnyside sample, so much so that
measurements could not be made in the temperature range studied using
the Roto-Visco apparatus. It was estimated that the viscosity of the bitumen
from the Tar Sand Triangle outcrop sample was significantly greater than
the viscosity of the Sunnyside sample. This observation may explain why no
separation was achieved in the processing of the Tar Sand Triangle sample.
Even though there is a significant difference in viscosity, all bitumen
samples seem to obey the Arrhenius-type relationship with an apparent act
ivation energy ranging from 23.6 Kcal/mole in the case of Asphalt Ridge and
P.R. Spring to 25.7 Kcal/mole in the case of Sunnyside. These apparent
activation energies are indicative of the fact that momentum transfer is
accompanied by significant structural transformations.
Moreover, the method of preparation of the bitumen sample has a
tremendous effect on its apparent viscosity. The viscosity of Asphalt
Ridge bitumen previously reported (A3) is contrasted with the viscosity
of the bitumen from the Asphalt Ridge sample as determined during this
investigation in Figure A5. The significant difference between these
viscosity measurements is due to the difference in the procedure used to
prepare the bitumen sample. The previous Asphalt bitumen sample was not
prepared under the controlled conditions that were used for bitumen sample
preparation in this study. In fact, the high viscosity of the Asphalt
Ridge sample reported previously may reflect the more severe thermal conditions
and atmospheric exposure experienced during preparation.
Although these viscosities may not be the true viscosities of the
virgin bitumen because of possible entrainment of benzene and/or loss of
light ends, they are at the least representative of the bitumen character
16
10
10*
o 10 o tn
10 _
l. I I I I
Asphalt Ridge Bitumen
« /
/
/
/ /
/
/ s / O Sepulveda and
l/ Miller e/ A Present
/ Investigation
i i i i i
-
"
-
2.8
(1/T)X103
3.0
,°K-1 3.2
Figure A5. Arrhenius-type plot illustrating the effect of the bitumen preparation technique on the measured viscosity of Asphalt Ridge bitumen.
17
on a relative scale and are of significant value for the analysis of the
results from hot water separation experiments.
Molecular Weight. With the exception of the P.R. Spring bitumen, the
bitumen viscosities presented in Figure A4 can be correlated with the number
average molecular weights of those samples presented in Table A4. The
highly viscous Tar Sand Triangle sample exhibits the highest molecular
weight. The difficulty in achieving effective hot water separations for
the Tar Sand Triangle and Sunnyside samples together with the higher molecular
weights and higher viscosities of the respective bitumens suggest a
different state of molecular aggregation for these samples with stronger
intermolecular bonds than would be found for the Asphalt Ridge sample which
can be easily separated and whose bitumen has a viscosity which is sign
ificantly less than viscosities of the Tar Sand Triangle and Sunnyside
bitumens.
It is premature at this point to speculate whether bitumens from
different deposits exhibit differences with respect to a specific adsorption
potential at the sand surface. Bonding characteristics at the sand-bitumen
interface are currently under investigation for the Sunnyside and Asphalt
Ridge systems. It is anticipated that the results from this phase of the
research will produce a better description of the nature of the bond and
interfacial activity of the respective bitumens. Nevertheless, other bulk
phase properties such as tar sand composition, bitumen viscosity, bitumen
molecular weight and SEM photographs (A3) of the tar sand samples already
demonstrate significant differences between the various Utah tar sand deposits.
Sand Analysis. Particle size distribution of the sand from various tar
sand samples is an important characteristic in the analysis of the hot water
separation process. In general, it would be expected that coarser sand
18
Table A4. Average Molecular Weight of Utah Bitumen
Bitumen Source Average Molecular Weight
Asphalt Ridge 763.2
Sunnyside 891.3
P.R. Spring 938.9
T.S. Triangle 1222.1
Table A5. Hot-Water Separation of Sunnyside Tar Sand
Experimental Conditions:
Digestion: Feed source: Sunnyside Wetting agent: Na^CO-Temperature: 95 C Diluent, (Toulene): Diluent/Bitumen volume ratio of .2 Percent solids: 73.5% by weight tar sands Digestion time: 30 minutes Na2C0~ concentration: 0.2 Mole/liter Feed size: Minus 4 mesh Agitation: 750 rpm
Flotation: Cell design: Cylindrical Percent solids: 20 weight percent tar sands Agitation: 900 rpm Temperature: 15 C Air Flowrate: 3000 (cc/min)
Calculated Mass Balance:
Concentrate
Tail
Middling
Feed
Weight Percent
28.49
66.08
5.43
100.00
Grade, Bitumen
30.41
.45
10.44
9.52
percent Sand
69.58
99.55
86.56
90.48
Recovery, Bitumen
90.94
3.11
5.95
100.00
percent Sand
21.91
72.71
5.38
100.00
COEFFICIENT OF SEPARATION = .69
19
distribution in the feed would be most desirable. Indeed, size classification
has been shown to occur in the processing of the Asphalt Ridge sample (A3).
The particle size distribution of four Utah tar sands are presented in
Figure A6. As can be seen, the Sunnyside sample contains more fine sand
than any of the other tar sand samples. Assuming a critical size of 100]im
it can be seen 7.7 percent of the sand in the Asphalt Ridge is finer than
lOOum. Whereas, in the case of the Sunnyside sample 25 percent of the sand
is finer than lOOum. The greater amount of fines in the sample feed stock
would be expected to have a detrimental influence on the bitumen-sand
separation process.
The sand in the Sunnyside sample contains a-quartz along with calcite,
dolomite and clay as determined by XRD analysis. The amount of clay present
has not yet been established quantitatively. The sand in the Asphalt Ridge
sample, however, contains only a-quartz and no clay or carbonate mineral
could be detected by XRD. The presence of the clay in the Sunnyside
sample undoubtedly contributes to the finer size distribution of the sand.
Hot Water Process
Because of the significant difference in the physical and chemical
properties of the low grade Sunnyside sample, as compared to the high grade
Asphalt Ridge sample, effective separations of the Sunnyside sample were not
realized (A3). Whereas a coefficient of separation of 0.9 could be achieved
for the Asphalt Ridge sample, a coefficient of separation of only 0.55 could
be achieved with the lower grade Sunnyside sample. This poor separation was
attributed to discontinuity of the bitumen and subsequent failure within the
bitumen phase rather than at the bitumen-sand interface (A3).
20
\z
OJ GO 3 C Q . 3
3 ><< CO to -a -•• 3" o. Q> a> _ j ^
r+ ^3 • - * • 73 a. • fO GO • -a
- 5
1X3 C - s fD
>
o a> c -s -5 n-Q. O — i . — i
- h fB -*> fO CO - s - '• (D N 3 ft)
Volume Fraction Finer Than Indicated Size, F3(d)
X- S»
a>
CO _ N O (B O
C-i. CL r + CO ai rt e =--s 3
W 3 ( t O " -—CO 0» C
- -$ r+
—I CO o CD Q> 3 -S 3
Q. O 00 - h cu a. 3 n> w Q.T3 0)
O 3 —I CO Q. -J - J ' -•• r+ - h cu co -s 3 • O to 3
O O til J»
- I 1—
o § 5 i i i i
O O CD
o CD Q O O
1 1—I I I
J 1 1 I I I I I I I I
In this regard an attempt was made to enrich the bitumen content of the
feed to greater than 10 weight percent by recycle of the bitumen concentrate
to the digestion stage thus promoting bitumen phase continuity. However, the
quality of the separation was not improved under these circumstances. Sim
ilarly feed enrichment accompanied by heat treatment of the low grade feed
with recycled concentrate in a furnace at 196 C prior to digestion did not
improve the separation.
Results from this study have shown that the viscosity of the Sunnyside
bitumen is one order of magnitude greater than the viscosity of the Asphalt
Ridge bitumen and suggest that the poor separation of the Sunnyside sample
may be related simply to the viscosity of the bitumen rather than to bitumen
phase continuity as previously suggested (A3). If the high viscosity of the
Sunnyside bitumen is responsible for the inferior quality of separation then
it should be possible to improve the separation by reducing the bitumen
viscosity with the addition of a diluent. In this regard, a controlled
solvent-assisted hot water process was developed. Presumably, the diluent
dissolves into the bitumen and reduces it's viscosity, such that with the
aid of a high shear, stirred tank reactor phase disengagement should be
improved.
Effect of Soda Ash. It was established by Sepulveda and Miller (A3)
that wetting agent addition is a critical factor for the success of the hot
water process. Therefore, the coefficient of separation for the Sunnyside
sample was determined as a function of sodium carbonate concentration with
and without diluent addition as shown in Figure A7. The coefficient of
separation provides a unique "one parameter" description of the efficiency,
or quality, of the separation process. It is defined as the fraction of the
22
l b s / t o n of Tar Sand
10 20 30 T
Diluent Addition :
O None
e Toluene : 0.2 cc/cc Bit.
Sunnyside Sample
0.2 0.4
Figure A7.
Na 2C0 3 ADDITION , M
Quality of separation as a function of Sodium Carbonate Concentration for the Sunnyside sample at a diluent to bitumen volume ratio of 0.2.
23
feed material which undergoes a perfect separation (A20). The coefficient
of separation can be expressed in terms of recovery and for a binary system
can be shown to be equal to the difference between the recovery of the
bitumen in the concentrate and the recovery of sand in the same concentrate.
It can be realized from Figure A7 that diluent addition has increased the
coefficient of separation significantly and that the best separation can be
achieved at about 0.2 M sodium carbonate addition for a diluent to bitumen
volume ratio of 0.2. Low wetting agent additions <0.1 M, produce a sticky
concentrate and much of the bitumen is lost in the tailing product. At
higher additions of sodium carbonate in the presence of diluent, the effective
ness of the separation is impaired apparently due to a tendency of the
system towards emulsification.
In these experiments with the addition of a diluent, about 95 percent
of the bitumen in the feed material was recovered in the concentrate and
72 percent of sand was rejected in the tailings product contained less than
0.5 percent bitumen. However, from 5 percent to 10 percent of the feed was
recovered as aggregates in a middling product which contained 10.5 percent
bitumen. Typical results for hot water processing of a Sunnyside sample are
presented in Table A5. Coefficients of separation are calculated assuming
that the middling can be separated ideally upon recycle.
In all of the above experiments 30 minutes digestion time was used. In
view of the middling stream produced, it was thought that a longer digestion
time might minimize the production of this intermediate product. In order
to test this hypothesis, a series of experiments at a diluent/bitumen volume
ratio 0.2 were performed for longer digestion times. Experimental results
presented in Figure A8, indicate that increased digestion beyond 30 minutes
causes the quality of separation (coefficient of separation) to increase
24
o 0.7
i-<
< a. w 0.6
9. 0.5
0.4
I o
Na 2 CO s : 0.2 M
Toluene : 0.2 c c / c c Bit.
O t£>
Sunnyside Sample
20
15
> •
HI > O
10 o LU
a.
o 5 o
120
and recovery of middling as a function of digestion time at 0.2 M Sodium Carbonate and a diluent to bitumen volume ratio of 0.2.
0 30
DIGESTION
Figure A8. Qual i ty
60
TIME
90
min
of separation
25
slightly. In addition, the recovery of the middling stream shows a
corresponding decrease. It appears that a 30 minute digestion time is
reasonable. This middling stream which was not produced in the hot water
processing of the Asphalt Ridge sample, will have to be redigested in order
to effect bitumen recovery.
Effect of Diluent. Preliminary results with controlled addition of
diluents, in which separation coefficients greater than 0.69 were obtained,
have encouraged further investigation in this area. In this regard, the
effect of different diluent types on the separation coefficient was de
termined as a function of diluent to bitumen volume ratio. As can be
observed from Figure A9 toluene and No. 1 fuel oil are most effective when
compared to the response obtained using kerosene. Tetraline, a hydrogen
donor solvent, is not nearly as effective as the other diluents.
The diluent to bitumen volume ratio is critical from the processing
standpoint. Even a low diluent to bitumen volume ratio of 0.01 has a sign
ificant effect on the separation coefficient. However, a better separation
is achieved at a higher diluent to bitumen volume ratio such as 0.2. Ex
cessive diluent additions (volume ratios above 0.40) have a detrimental
effect on the separation due to a tendency of the system to emulsify. Further
research in this area of diluent addition is contemplated as a part of the
future research program.
Product Characterization
Analysis of the products from the separation experiments provides
further insight regarding the nature of the phase disengagement-displacement
process.
Particle Size Analysis. The products in the hot water separation of the
26
0.8
gal lons/ ton of Tar Sand
1 2 3 4 5 T
Sunnyside Sample
N a 2 C 0 3 : 0.3 M
o ui 0.5 u. LU O o
0.4
Type of Di luent
o Toluene
D Fuel Oi l N°1
A Kerosene
a Tetra l ine
o None
0.1 0.2
DILUENT/BITUMEN RATIO , by volume
Figure A9. Quality of separation for the Sunnyside sample as a function of diluent to bitumen volume ratio for different types of diluent at 0.3 M Sodium Carbonate.
27
Sunnyside sample differs significantly from those that had been obtained in
the hot water processing of the higher grade feed stocks. In the processing
of the high grade Asphalt Ridge and P.R. Spring samples, essentially no
middling product was obtained and the sand content of both the concentrate
and the tailing consisted of free non-aggregated particles. Further, it was
shown that sand was partitioned between the concentrate and tailing on the
basis of size, i.e. size classification occurred during the process sequence
with the fine sand being preferentially recovered in the concentrate and
the coarse sand being rejected in the tailing as shown in Figure A10.
The character of these products should be contrasted to the products
obtained in the hot water processing of the lower grade, Sunnyside sample
at 0.3 M sodium carbonate addition and for a diluent to bitumen volume ratio
of 0.2. To begin with, a significant middling stream (y 10 percent of the
feed) is produced consisting largely of aggregates which remain undigested
during the processing sequence. The interparticle strength is such that the
aggregates persist even after extraction of bitumen during Dean-Stark analysis.
In addition, the aggregates can be identified in the feed, a phenomenon that
was rarely observed in the Asphalt Ridge and P.R. Spring feed stocks. These
aggregates have been separated from the feed and XRD analysis clearly shows
the presence of calcite and dolomite which enhance the aggregate strength by
acting as an interparticle cementing agent.
These results account for the lack of size classification during hot
water processing of the Sunnyside sample. As can be seen in Figure All,
the sand from the middling product has the coarsest size distribution,
followed by the sand from the concentrate, and finally the sand from the
tailing which has the finest size distribution and consists of essentially
free, non-aggregated particles, whereas the sand from the middling product
28
n i i i i i — , r o
_ o J Q i i
•
m
«_*
s 1 * * • *
TJ • a
<u _N
V)
_fl>
O ~ l_
o o_
T3 O
<D C >+- O
•r-<D 4 J
-c re +-> S-re
C Q_ •r- <L
to T3 C i -ro CD CO +J
re OJ 2
XZ +-> -M
O 4 - J = — o ro
r— ^w**
c re o u •
•i— •>— CD 4-> C r -=5 > , Q .
•r- re s- re co
+J w E n
•r- o en • D S - T 3
4 - -r -
N CO •r- + J +J CO O 1—
3 re <V TJ _C i— O Q .
u s . w +J
re E _C Q. re +J
a; 3
(P) d '3ZJS pa^oipui uDq i j e u y UO»ODJJ a uun|0A
29
1.0
o
UJ
z u_
2 U h-O < a a.
c/) < 2
N W
Q UJ
< a a z
z x ™ h-
1 1 1—
Sunnyside Sample
Na2C03 : 0.3 M Toluene : 0.2 cc/cc Bit.
Tailings Concentrate Feed Middlings
j _
10 100 1000
Particle Size pm Figure All Particle size distribution of the sand in
the feed and products from a typical hot water separation of the Sunnyside sample.
consists almost exclusively of aggregates with very few free particles. As
might be expected, no carbonate cementing compounds could be detected in the
tailing product by XRD analysis.
These apparently anomalous results, for which no size classification is
observed, indicate that sand is not transported to the concentrate by
mechanical entrapment as appears to be the case for the Asphalt Ridge and P.R.
Spring samples, but rather undigested aggregates, of a relatively coarse
size, are transported to the concentrate due to their partial hydrophobicity.
This hydrophobic character arises from the residual bitumen which together
with cementing agents bond the sand particles together and account for the
aggregate's strength.
It is clear from the results of hot water processing of the Sunnyside
sample that diluent addition is necessary in order to achieve a better
separation. The addition of diluent seems to be an appropriate strategy
because the viscosity of the Sunnyside bitumen was found to be significantly
higher than the viscosities of the Asphalt Ridge and P.R. Spring samples as
shown in Figure A4. Bitumen viscosity measurements of the concentrate and
middling streams from the solvent-assisted hot water separation of the
Sunnyside sample support the previous discussion regarding the presence and
behavior of aggregates in the Sunnyside feed stock. Note in Figure A12 that
the bitumen associated with the middling product has the same viscosity as
the bitumen from the feed, indicating that the diluent has had no effect on
the viscosity of the bitumen in the middling stream, a result which is con
sistent with the observation that the middling consists of undigested,
(apparently impervious) tightly bonded aggregates. On the other hand, the
bitumen associated with the concentrate product apparently has dissolved
the diluent and as a result is an order of magnitude less viscous than
31
Sunnyside Bitumen
. /
/ . /
. /
. / s y
y m
O Feed • Middlings • Concentrate
2.8 3.0
( 1 / T ) x 1 0 3 , °K~1
3.2
Figure Al2. Arrehenius-type plot illustrating the effect of temperature on bitumen viscosity for products from hot water separation for the Sunnyside sample.
32
either the middling or the feed bitumen. Interestingly, the viscosity of the
bitumen from the Sunnyside concentrate shown in Figure A12 has a viscosity
profile essentially the same as the viscosity profile for the Asphalt Ridge
and P.R. Spring feedstocks, both of which can be processed without diluent
addition.
These results demonstrate that for high viscosity feedstocks diluent
additions are necessary to allow for effective digestion. Further, at least
in the case of the high viscosity Sunnyside sample, the tenacious inter-
particle bonding results in the stabilization of nonreactive aggregates
which report to the concentrate and middling product accounting for the
less efficient separation relative to that which had been achieved with the
Asphalt Ridge and P.R. Spring samples.
Recent experimental results have identified some important and interesting
facets of the flotation separation of bitumen from digested tar sand.
Although contact angle measurements of pure Asphalt Ridge bitumen
indicate moderate hydrophobicity, air bubble attachment to the bitumen
concentrate taken from a 37 liter flotation cell is not possible. This
surprising result suggests that the flotation separation is dependent on air
bubble entrapment rather than on attachment due to surface hydrophobicity.
The occlusion of air bubbles in the bitumen is apparent from visual
examination of the concentrate, especially at lower temperatures. It seems
that a bitumen-bubble agglomerate forms in the impeller region of the
flotation cell, the effective density of which is such to allow the
agglomerate to float to the surface.
A factorial design of the major operating variables In the flotation
separation indicated that the quality of separation of Asphalt Ridge tar
sand was significantly dependent on the flotation temperature (see Figure A13)
33
1 I
A S P H A L T R I D G E
—
-^e^*1* 5 .-»"
5"
1 1
/^.
1 1
* « * -<•
yfir^Z*-'
9 experimental _ mode l
1 1 0 20 40 60 80
F l o t a t i o n Tempera tu re , °C
gure A13. Influence of flotation temperature on the coefficient of separation at 0.05 M Na.CO and 1000 rpm.
34
and to a lesser extent on the degree of agitation. For a flotation
temperature of 77°C, recovery of 96.7 percent was realized at a grade of
61.0 percent bitumen. The improved separation at higher flotation
temperatures was found to be due to the decrease in bitumen viscosity
resulting in more effective rejection of coarse sand from the concentrate.
The size distributions of the sand in the bitumen concentrate for concentrates
at different temperatures is shown in Figure A14.
Summary and Conclusions
Consideration of previous results and hypotheses (A3) regarding the
phenomenological description of the phase disengagement-displacement process
together with the results of this current study indicate that the criteria
for an effective separation remain the same independent of the nature of the
tar sand deposit. These criteria are:
- high shear force field
- appropriate level of wetting agent addition
- high temperature digestion
However, it appears the previous hypothesis (A3) that bitumen phase
continuity (> 10 weight percent bitumen) is required for effective hot water
separation must be dismissed. In view of the results reported in this
investigation it would seem that the correlation between bitumen content of
the feed and the effectiveness of the separation is due to two major factors:
- viscosity of the bitumen
- interfacial bonding between sand particles
For low grade tar sands such as the Sunnyside sample, the bitumen viscosity
is at least an order of magnitude greater than the bitumen viscosity for
high grade tar sands such as Asphalt Ridge. In addition, the low grade tar
sands exhibit strong interparticle bonding accounting for the presence of
Figure 14A. The effect of flotation temperature-on the size distribution of sand entrapped in the bitumen concentrate.
aggregates in the feed which maintain their integrity during digestion and
subsequent processing. These factors explain the inferior separations
obtained with the Sunnyside sample while much better separations are obtained
with high grade samples. At this point it would appear that the bitumen
viscosity and the strength of the interparticle bonding may be related to
feed grade via the genesis of the deposit. The insufficient bitumen content
of the low grade material results in the formation of strong interparticle
bonds during the rock forming process, whereas high grade material has
sufficient bitumen between the sand particles to prevent this strong inter
particle bonding and as a result facilitates the phase disengagement-dis
placement process in the hot water separation of high grade samples.
The conclusion regarding the importance of bitumen viscosity is based on
the necessity of adding diluents to the Sunnyside feed for effective separa
tions and the fact that the bitumen viscosities of the separation products
(concentrate and middling) reflect a marked difference between that bitumen
which had been separated into the concentrate successfully and that bitumen
which was not separated successfully and was recovered in the middling.
These results are contrasted to the nature of the bitumen of the higher
grade tar sands (Asphalt Ridge and P.R. Spring) the viscosity of which is
equivalent to the viscosity of the diluted Sunnyside bitumen recovered in
the concentrate.
The importance of interfacial bonding between sand particles has been
demonstrated in the case of the low grade Sunnyside sample by the identifi
cation of relatively strong aggregates which report to the middling product
and persist both in the feed and in the middling even after bitumen removal
during analysis by the Dean-Stark technique. This phenomenon is revealed
in the particle size analysis of the products of the hot water separation
37
experiments. Further, it appears that some of the bonding strength of the
aggregates arises due to cementation of the sand particles by calcareous
compounds such as calcite and dolomite which were identified by XRD.
Aggregates such as these were only rarely encountered in the processing of
the Asphalt Ridge sample.
Even with this understanding and making the necessary provisions to
improve the process, the quality of separation for the Sunnyside sample is
still inferior under present conditions to that which is achieved for the
Asphalt Ridge sample.
Asphalt Ridge
Sunnyside
Weight % Bitumen in Concentrate
64.82
27.20
Recovery Bitumen in Concentrate
96.39
96.43
Coefficient of
Separation
0.88
0.69
38
References
Oblad, A.G., J.D. Seader, J.D. Miller and J.W. Bunger, "Recovery of Bitumen from Oil-Impregnated Sandstone Deposits of Utah", Oil Shale and Tar Sands, AIChE Symposium Series, Vol. 72, No. 155, p. 69 (1976).
Sepulveda, J.E., J.D. Miller and A.G. Oblad, "Hot Water Extraction of Bitumen from Utah Tar Sands", Fuels Division, ACS (1977), Symposium on Oil Shale, Tar Sands, and related Materials-Production and Utilization of Synfuels, Division of Fuel Chemistry, ACS, 21, No. 6, p. 110 (1976).
Sepulveda, J.E., and J.D. Miller, "Extraction of Bitumen from Utah Tar Sands by a Hot Water Digestion - Flotation Technique", Mining Engineering, Vol. 30, No. 9, p. 1311, (1978).
Walters, E.J., "Reviews of the World's Major Oil Sands Deposits," Oil Sands; Fuels of the Future, L.V. Hills, Editor, Canadian Society of Petroleum Geologists, Calgary, Alberta, Canada (1974).
Ritzma, H.R., "Oil Impregnated Rock Deposits of Utah", Utah Geological and Mineral Survey, map 33, (1973).
Cottrel, J.H., "Development of an Anhydrous Process for Oil-Sand Extraction" p. 193-201, the K.A. Clark volume, A collection of papers on Athbasca Oil Sands, presented to K.A. Clark on the 75th anniversary of his birthday. Edited by M.A. Carrigy (1974).
Lowe, R.M., "Status of Tar Sand, Exploitation in the United States", paper presented at the 68th Annual Meeting of the AIChE, Los Angeles, California, November 16-20, (1975).
Djingheuzian, L.E., "Cold Water Method of Separation of Bitumen from Alberta Bituminous Sand". Proceeding of the First Athabasca Oil Sands Conference, King's printer, Edmonton, Alberta, Canada, p. 185-199, September (1951).
Bichard, J.A., C.W. Bowman, Butler, R.M. and Tiedje, J.L., "Separation of Oil from the Athabasca Oil Sands by Sand Reduction". p. 171-191 of ref. 6.
Kruyer, Jan, Oleophillic Society of Alberta, Published in Edmonton Journal, Wednesday, March 22, 1978.
Clark, K.A., Research Council of Alberta Report 1922, Edmonton, Alberta, Canada, 1923, p. 42-58.
Clark, K.A. and Pasternack, D.S., "Hot Water Separation of Bitumen from Alberta Bituminous Sands", Ind. and Engg. Chem., Vol. 24, No. 12, p. 1410 (1932).
Clark, K.A., "Hot Water Separation of Alberta Bituminous Sand," Canadian Institute of Mining and Metallurgy, Trans., Vol. 47, p. 257, (1944).
Innes, E.D. and Fear, J.V.D., "Canada's First Commercial Tar Sand Development", proc. of the Seventh World Petroleum Congress , Vol. 3, Elsevier Publishing Company, (1967).
39
A15. Camp, F.W., "The Tar Sands of Alberta, Canada", second edition, Cameron Engineering, Inc., Denver, Colorado (1974).
A16. McConville, L.B., "The Athabasca Tar Sands, The Outlook for the Future", Mining Engineering, Vol. 27, No. 1, p. 37, (1975).
A17. Porteous, K.C., "Oil Mining - The Syncrude Project", Syncrude, Canada, Ltd. Edmonton, Alberta, Canada. For presentation at "The Alternate Resources and Technologies for Fuel Symposium", University of Pittsburgh, Pennsylvania, July 31, 1978.
A18. "Separation of Bitumen From Dry Tar Sands", United States Patent No. 4,120,766, Oct. 17 (1978).
A19. U.S. Bureau of Mines, Report of Investigation, No. 4004.
The concept of recovering liquid and/or gaseous hydrocarbons from
solid-like hydrocarbon-bearing materials by thermal treatment has been
known for several centuries (Bl, B2). Thermal treatment essentially
entails processing at high temperature and is variously described as
distillation, coking, carbonization, gasification, etc., in part to
emphasize a particular aspect of the process, products or chemical
reactions involved. All the above processes are carried out by
heating the feed material in an inert or non-oxidizing atmosphere.
The mode of heating and the operating temperature largely determine
the type of changes occurring to the feed. In general, thermal
treatment can result in the following:
1. Volatilization of the low-molecular-weight components in
the feed.
2. Generation of light hydrocarbons from the original heavier
components in the feed by cracking reactions, and their subsequent
volatilization. The severity of cracking depends to a large extent
on the temperature.
3. Conversion of part of the hydrocarbons into condensed com
pounds, generally referred to as coke, by reactions such as polymerization.
In the case of feed materials such as tar sand, which contains a
significant amount of inert organic matter that remains substantially
unchanged through the thermal treatment, coke is obtained as a deposit
on the inorganic matter.
41
It Is obvious that thermal processing can require a substantial
input of energy to provide the necessary sensible, latent, and reaction
heats. However, coke, when produced as above and subsequently combusted,
can generally provide much or all of this energy requirement. Combustion,
referred to by some authors as decoking or burning, is therefore an
important aspect of thermal-recovery methods. This task was directed
toward efficient recovery of energy from combustion of the coke.
Moore, et al. (B3), classify thermal processes into two general
groups, direct heated and indirect heated, depending on whether
pyrolysis and combustion steps are carried out in one or two reaction
vessels. The processes further differ from each other with respect
to fluidized or non-fluidized state of solids in each of the two
steps. Table Bl shows a general classification scheme, which
fits most known thermal processes.
The terms pyrolysis and combustion are used to describe the two
processing steps. The vessels or zones where pyrolysis and combustion
are carried out will usually be referred to here as reactor and burner,
respectively.
In all thermal recovery processes, tar sand is subjected to high
processing temperatures, about 450-550 C (840-1025 F) for pyrolysis,
and the residual coked sand is further heated to about 550-650 C
(1025-1200 F) during the coke-combustion step. At these conditions,
an acceptable thermal efficiency can only be obtained if a significant
portion of the sensible heat in the spent sand is recovered and introduced
back into the process. Almost all the processes listed in Table Bl
provide for heat recovery from spent sand before it is discarded.
42
TABLE Bl
THERMAL RECOVERY PROCESSES
Direct Heated Indirect Heated
Non-fluidized pyrolysis, non-fluidized combustion
Fluidized pyrolysis, f lu id ized combustion
Fluidized pyrolysis, non-fluidized combustion
Non-fluidized pyrolysis, f lu id ized combustion
USBM (58) Saunders (59)
Gifford (51) Peck, et a l . (12)
University of Calgary (64)
Bennett (65) Berg (66)
Gishler and Peterson (71)
Nathan, et a l . (72) Roetheli (73) Murphree (74) Alleman (75)
No examples known
No examples known Lurgi-Ruhrgas (76)
43
In comparison to the hot-water process, thermal methods have not
received much attention. One of the major drawbacks of thermal methods
is the heat load associated with the large mineral content of tar sand.
A recent study by Flynn, et al. (B4) , directed to explore this point, is
important in this connection. They conclude that net hydrocarbon efficien
cies are virtually identical for the present technology (i.e., the hot-
water method with coking-hydrotreatment) and thermal processes even with
out heat recovery from the hot, spent sand. Further improvement in the
efficiency of thermal methods is indicated if some of the heat for heating
incoming streams can be recovered.
Incentives to develop an alternative thermal recovery technology are
real and some are noted here:
1. A direct thermal approach could eliminate huge tailings ponds and
dams usually associated with hot-water technology. This would result in
some savings in capital investment and also working capital since diking
is a continuing operation throughout project life. Disposal of dry,
thermally processed sand should be relatively easier. The volume of spent
sand from a thermal recovery unit would be about equal to the original
tar sand and hence all of it could be accomodated in the mined-out areas.
2. Since direct coking of tar sands eliminates several processing
steps in the current technology, such as extraction, froth clean-up in
volving diluent centrifuging and diluent recovery, and possibly primary
processing, some saving in capital investment is likely. The high cost
of mining and utility units could, however, offset these savings.
3. Thermal recovery of bitumen would require a minimal amount of
process water. This is particularly important to the development of Utah
44
tar sands. Ritzma (B5) and a federal study (B6) point out that scarcity
of water in Utah, compounded by projected development of an oil-shale
industry, would hamper a tar-sands industry based on hot-water technology.
4. Thermal recovery processes can process tar sands that have low
bitumen content without too much reduction of product yields. The hot-
water process, on the other hand, can experience difficulty achieving
good separation with leaner Utah tar sands (B7).
5. Higher concentration of fines in tar sand can be troublesome and
result in lower bitumen recovery by the hot-water process. Fines do not
present serious problems in thermal recovery processes.
6. Thermal recovery methods can process tar sands containing
particulate mineral matter and also those containing consolidated sand
stone matrix if suitably sized. Several Utah deposits contain consolidated
mineral.
7. As a result of factors 4, 5, and 6 above, thermal processes can
be quite flexible in handling the variations and non-uniformities in a
given deposit. This could be an important factor in their selection for
future projects which will probably have to process lower-grade, i.e.,
leaner and non-uniform, deposits. As seen before, current projects are
processing the choicest rich deposits.
8. Since thermal recovery methods produce partially upgraded oil,
the need to handle highly viscous bitumen is eliminated. It should be
noted here that most Utah tar-sand bitumens are an order of magnitude more
viscous than Athabasca bitumen. While thermally recovered product would
still need upgrading prior to its use as a crude oil substitute, it is
easily transportable as such in pipelines and an upgrading plant need not
be an integral part of a tar-sand processing project.
45
Energy-Efficient Thermal Processing Concept
General desirable features of a thermal recovery process are:
1. Pyrolysis: tar sand is subjected to a temperature of between
450 and 550 C (840 and 1025 F) to crack and volatilize the contained
bitumen; the vapors are subsequently condensed and recovered.
2. Combustion: coke produced during pyrolysis and additional fuel,
if required, are combusted at 550 to 650 C to generate heat.
3. Heat transfer: heat generated during combustion is efficiently
utilized to provide for the endothermic pyrolysis step.
4. Heat recovery: heat contained in the streams leaving the process
is recovered and utilized to improve the thermal efficiency of the process;
in particular, heat in the exiting spent sand should be recovered.
In research conducted during a prior study, a new thermal processing
concept, as shown in Fig. B-l, was conceived. Freshly mined and sized
tar sand is dropped into the upper bed of a multi-staged fluidized-bed
column. The upper bed is a pyrolysis reactor, which is maintained at a
temperature of generally between 400 and 550 C (750 and 1025 F). Here,
bitumen in the feed is cracked and/or volatilized, leaving a coke deposit
on the sand particles. The oil vapors and light hydrocarbon gases produced
are carried off by the inert fluidizing gas to fines-separation and product-
recovery sections, while coked sand flows down by gravity through a control
valve to the lower combustion bed or to the burner section of the column
where the coke is burned to generate heat. The burner is maintained at a
temperature of generally between 550 and 650°C (1025 and 1200°F). Preheated
air is used to fluidize the solids in the combustion bed and to provide
oxygen for combustion. Gaseous products of combustion, mostly nitrogen
and carbon dioxide, then flow upwards to fluidize solids in the upper bed
as noted above.
46
A number of heat pipes, as required by the heat-transfer load, are
placed vertically in the fluidized bed column such that they extend into
the pyrolysis and combustion beds as depicted in Figure B-l. What is
important is that there is sufficient area for heat transfer between the
fluidized beds and the heat pipes under operating conditions. Heat pipes
transfer excess heat generated in the burner to the pyrolysis reactor,
thus maintaining the reactor and burner at proper temperatures.
Structurally, a heat pipe consists of a closed metallic tune, the
inner walls of which are lined with a wick that can be in the form of
layers of wire screen, longitudingal grooves or channels in the wall
itself, channels covered with screen, etc. Prior to sealing the tube,
it is exhausted of all fluid contents and filled with a suitable amount
of working fluid and, in some cases (particularly with liquid metals), a
small amount of inert gas such as nitrogen or helium to aid in start up.
The amount of working fluid used is generally a slight excess over the
amount required to wet the entire wick. Potassium was selected as being
suitable for the temperature involved in tar-sand processing.
In operation, heat is transferred in the combustion bed to the lower
end of the heat pipe, causing the working fluid to vaporize. The vapor
flows to the upper, cooler end in the pyrolysis bed due to the pressure
gradient set up inside the central vapor core of the heat pipe. There,
the vapor condenses on the tube wall and inside the wick, transferring
heat to the pyrolysis bed. The condensate then returns to the warmer end,
flowing along the wick and thus completing the cyclic flow of the fluid.
The operation is continuous so long as heat is supplied to the hotter end,
known as the evaporator section, and removed at the cooler end, known as
the condenser section. Heat can be transferred to and from the heat pipe
47
Tar Sand •*• Products to Recovery
Heat Pipe-
/
\
Pyrolysis
J
Air
Combustion
'-/Tv
rn^p
Heat Recovery
Spent Sand
FIGURE Bl University of Utah process.
48
by conduction, convection and radiation. If external conditions require,
an adiabatic section can be included between the evaporator and condenser
sections.
Heat is transferred by the heat pipe as the latent heat of vaporization
of the working fluid and, hence, a large amount of heat can be transferred
if the working fluid has a large latent heat of vaporization. In operation,
the working fluid is at its saturation conditions at all points inside
the heat pipe. Since vapor pressure of a fluid changes rapidly with
temperature, the termperature gradient along the length of the heat pipe
and in the vapor core is generally very small. Hence, the heat pipe
operates nearly isothermally. Because a large amount of heat can be
transferred by a heat pipe, its so-called effective thermal conductivity
is very high.
Hot spent sand leaving the burner flows down through a control valve
to a heat recovery section, which is shown as an indirect heat exchanger
in the figure. Process air used to recover heat from the spent sand is,
in turn, suitably preheated. Sand may then be further cooled before dis
posal and the heat used to produce steam or for any other purpose. A more
detailed description of the process has been published elsewhere (B8).
The basic process described above shares several characteristics with
processes listed in Table Bl. Of those, only three—the Peterson and
Gishler process (PG), the Lurgi-Ruhrgas process (LR), and the University of
Calgary process (UC)—are referred to for comparative analysis; firstly
because only for these three processes is sufficient information available
in the literature, most others being described in patents, and secondly,
because these three processes have most of the features of the other
processes. In general, an attempt is made to secure the advantages of
49
earlier processes and at the same time avoid some of their drawbacks.
The new process retains most of the simplicity of the UC process and
other-direct heated processes. Solids move only downwards by gravity,
the equipment is essentially a single vessel, and there is no recycle of
solids. Unlike other direct-heated processes, separation of pyrolysis
and combustion beds permits better control of each process step. Thus
one can monitor and control the composition of gas entering the pyrolysis
reactor, particularly oxygen content. Hence, the contact of combustion
gases and hydrocarbon products is not considered to be a serious problem.
Most importantly, the heat-transfer features used—heat pipes, heat
recovery from spent sand to preheat process air, transfer of some heat
by combustion gases, and some radiative heat transfer from burner to
reactor—permit efficient management of the energy that is within tar
sand itself and help achieve high energy efficiency. The heat pipes
effectively link the pyrolysis reactor and the burner thermally without
necessarily imposing any other constraints on the process such as flow
patterns, reactor configuration, or dimensions of the column (except for
the volume of heat pipes, which is a small fraction of bed volumes). Heat
pipes allow separation of pyrolysis and combustion sections without
recourse to solids recycle. Heat-pipe technology has developed relatively
recently and its applications at a scale comparable to that envisaged for
commercial application in this study will require pilot-plant development.
The basic process as outlined above is very flexible and modifications
and variations can be easily incorporated into it to further improve the
overall efficiency and/or to make it more suitable for specific types of
feeds. Thus, external fuel, or recycle gas, or liquid fuels can be easily
introduced into the burner in the case of lean tar sands. By providing
50
for a purge gas stream off the top of the combustion bed, one can adjust
the flow rate of fluidizing gas to the pyrolysis bed as desired. This is
very important for lean tar sands which would otherwise have very low
product concentration in the combined exit gas stream, making product
recovery difficult. Figure B2 depicts the concept of a commercial tar-sand
processing unit with these two features and also includes facilities for
recovery and treatment of hydrogen sulfide and sulfur dioxide in the gas
streams.
The equipment and the process described here may be suitable for
recovery of hydrocarbon values from other solid hydrocarbon-bearing
materials such as oil shale, coal, etc.
As an example of the operation of the process in Figure Bl, assume
that:
1. The pyrolysis and combustion steps are carried out at temperatures
of 475°C (887°F) and 575°C (1067°F), respectively. Operating pressure is
assumed to be about 1 atmosphere.
2. All inorganic matter is completely inert under these conditions;
further, it is silica in the form of quartz.
3. Pyrolysis of tar-sand bitumen results in the conversion of bitumen
to coke, oil vapors, and light hydrocarbon gases yielding 15 percent coke,
81 percent oil vapors, and 4 percent light gases. These yields are in
dependent of the bitumen content in the tar sand.
In Figures B3 and B4 are presented the material and energy balances
for pyrolysis and combustion steps to thermally process 1 ton per hour of
tar sand containing 12 percent by weight bitumen. For simplicity, coke is
treated as just carbon.
51
Tar sand ( *
Reactor
Burner External—tx—| '•;MM-
fuel
Preheater
•••-C
Heat recovery
unit
Air cooler
Demister Pads tesHS
n ^8H
Recycle o i l
Heat exchanger
:-s->»*
Heavy o i l
Air
I (3 Spent
sand
To CO boiler
Gas holder TiQ
H2S absorption and S recovery plant
Fuel gas
FIGURE B2 Conceptual scheme for commercial plant.
« f < I
Sand
Bit
Total
lb/hr
1760.0
240.0
2000.0
Btu/hr
0
0
0
Tar Sand at 77°F
FIGURE B3
HC Gases
Oil Vap.
Total
lb/hr
9.6
194.4
204.0
Products at
Btu/hr
5,040
111,780
116,820
887°F
REACTOR
887°F (475°C)
AHR = 2160 Btu/hr c HEAT IN
473,360 Btu/hr
Coked Sand at 887°F
Sand
Coke
Total
lb/hr
1760.0
36.0
1796.0
Btu/hr
322,140
12,240
354,380
Sample material and energy balance for pyrolysis.
Coked
Sand
Coke
Total
Sand at
Ib/hr
1760.0
36.0
1796.0
887° F
Btu/hr
342144
12240
354380
N2
02
Total
lb/hr
321.0
96.0
417.0
Btu/hr
0
0
0
Air at 77°F
N2
C02
Total
lb/hr
321.0
132.0
453.0
Btu/hr
79450
33980
113430
Combustion Gases at 1067°F
BURNER
1067°F (575°C) AHC = -534,960 Btu/hr
Spent Sand at 1067°F
Sand
Total
lb/hr
1760.0
1760.0
Btu/hr
418,170
418,170
£> HEAT OUT
357,740 Btu/hr
Figure B4 Sample material and energy balance for combustion.
It is seen from Figure B3 that 1 ton of tar sand containing 12 percent
bitumen, upon pyrolysis at 887 F, produces 9.6 lb light hydrocarbon gases,
194.4 lb cracked oil vapor, and the balance of 36.0 lb of bitumen converts
to coke deposit on the sand particles. It is also seen that this step is
highly endothermic and requires a net heat input of 473,360 Btu, about 68
percent of which provides sensible heat to the sand. From Figure B4, it
is seen that combustion of the coke deposit with the stoichiometric amount
of air yields a net heat output of 357,740 Btu, which is about 75 percent
of the heat requirement of pyrolysis.
The resultant shortfall in energy of about 115,620 Btu for the above
process can be provided in one or a combination of the following ways:
1. Introducing an external source of energy such as coal, coke, fuel
oil, natural gas, etc., into the process.
2. Recycling part or all, as required, of the light hydrocarbon gases
to the burner for combustion.
3. Recycling the heaviest fraction of product oil to the burner for
combustion.
4. Recovering heat from the hot outgoing streams, viz., the hydro
carbon product stream, the combustion gases, and the spent sand, and re
introducing the energy into the process.
5. Any combination of the above.
It should be emphasized that merely balancing the process for energy
is not enough to justify use of the thermal recovery process. Recovery of
heat from hot exit streams is important in itself to improve the overall
thermal efficiency of the process. In general, the addition of external
fuel does not seem like an attractive alternative, due to the additional
cost incurred, and the overall appeal of the process would depend on how
55
efficiently it utilizes the energy available in the tar sand itself.
If the energy shortfall is provided by recovering energy from the
spent sand, an energy-balanced operation, shown in Figure B5 can be
achieved. Here, all energy requirements for the steps shown are derived
from the tar sand.
Experimental Apparatus
The experimental apparatus that was used to demonstrate the process
outlined in the previous section is shown in Figure B6. There are several
differences between the laboratory system and the conceptual commercial
plant depicted in Figure B2. For example, different fines separation and
product recovery systems are used. Further, the laboratory system does
not provide for carrying out air preheating and heat recovery from spent
sand as illustrated in Figure Bl. No provisions are made to recycle gaseous
products or to purge any combustion gases.
The laboratory apparatus shown in Figure B6 consisted of a two-staged,
fluidized-bed column constituting the primary processing unit, a mechanism
for feeding tar sand at a flow rate up to 10 pounds/hour, a system for
separation of fines from the products, and a product-recovery section.
The primary processing unit consisted, for top to bottom, of 4-inch-diameter
upper disengaging section, a 2-inch-diameter by 4-foot-long pyrolysis
section or reactor, a 4-inch-diameter central disengaging section, a 2-inch-
diameter by 40.5-inch-long combustion section or burner, and a gas inlet
section. Flanged connections were provided between the sections, which
were all insulated with Kaowool blankets. Gas distributors for the two
fluidized beds were positioned appropriately between the reactor and central
disengaging section and between the burner and gas inlet section for
pyrolysis and combustion beds, respectively. Solids flow lines and solids
flow-control valves were provided as shown to regulate the flow of solids
56
Tar Sand at 77°F 0 Btu/hr
Products at 887°F • 209,620 Btu/hr
REACTOR 887° F (475°C)
Coked Sand at 887°F 354,380 Btu/hr
Heat Transfer by Heat Pipe
452,740 Btu/hr
A
AHR = 2160 Btu/hr
Combustion Gases at 1067°F, 113,430 Btu/hr
BURNER 1067°F (575°C)
Spent Sand at 1067°F 418,170 Btu/hr ,,
AHC = -534,960 Btu/hr
Air at 997°F 95,000 Btu/hr
HEAT EXCHANGER
Spent Sand at 842°F 323,170 Btu/hr
Air at 77°F 0 Btu/hr
FIGURE B5 Energy-balanced operation.
57
Tar sand Filter
1 7/7.7//*' Screw vtn.tt feeder
\
Reactor
/
Heat piDe.
/
\
Burner
Propane *
— Condenser
Fines
f\
(
Oil I
Condenser
CW
Oil X -&Bed level control valve
to
"vent
Oil
Electrostatic precipitator
w
I Oil
Bedo-level
control valve
Air
Spent sand receiver
FIGURE B6 Laboratory system.
58
from reactor to burner and from burner to a spent-sand receiver. A 0.75-
inch diameter by 7-foot-long heat pipe carrying potassium was supported
between the reactor and central disengaging section such that appropriate
portions of the heat pipe extended into pyrolysis and combustion beds.
The feeding system for tar sand consisted of a screw feeder, a feed
hopper, and a variable speed drive. The screw-feeder outlet was attached
to the upper disengaging section by means of a reducing adapter. Fines
are separated from the product stream by a cyclone and a sintered-metal
filter arranged in series and maintained at between 350 and 400 C (660-
750 F) by heat ing electrically with heating tape and nichrome heating wire,
respectively. The temperatures were controlled manually with household
dimmer switches. Filter and cyclone were supported rigidly on a slotted
angle framework and the cyclone inlet was connected to the product outlet
on the upper disengaging section by 6-inch long, stainless steel, flexible
tubing. This arrangement allowed expansion of the column upon heating to
operating temperatures without hindrance and stress.
Liquid product was recovered from gas in a system consisting of a
water-cooled condenser, a cyclone, a second condenser, and an electro
static precipitator. The uncondensed gases were vented to an exhaust
system.
In operation, suitably prepared tar sand is charged to the feed hopper.
Tar sand leaves the feeder exit and drops freely through the upper dis
engaging section into the fluidized pyrolysis bed. Some preheating of the
feed will occur before it enters the bed as it contacts the upwardly
flowing hot gases in the disengaging section. In the reactor, which is
maintained at about 450 to 550 C (840 to 1025 F), pyrolysis of bitumen takes
place and the resulting hydrocarbon product vapors leave the processing
unit along with the fluidizing gas through the upper disengaging section.
59
Coked sand produced in the reactor flows down into the burner through
the control valve, which is pneumatically actuated by a signal from a
pneumatic controller that controls the height of the pyrolysis bed. Air
is introduced into the burner in controlled amounts to fluidize the com
bustion bed and burn the coke deposit on the sand. A side port near the
bottom of the burner permits introduction of propane into the burner.
Propane is used during start-up of the equipment and very often during
the run to ensure that all, or substantially all, of the oxygen in the
air is consumed in the burner. Combustion gases are sampled for measuring
oxygen content periodically, through a sampling port provided on the
central disengaging section.
In operation, the height of the pyrolysis bed should ensure a residence
time sufficient to achieve a satisfactory degree of completion of pyrolysis.
Thermogravimetric studies indicated that about 5 minutes are required to
volatilize most of the bitumen in tar sands. For design purposes, 9 to
10 minutes was considered a maximum residence time required for complete
pyrolysis. To provide for a solids residence time of 10 minutes, the bed
height required is about 8 inches, assuming a bed void fraction of 0.7.
A much higher bed height was provided for in the design.
Several problems were encountered in transferring solids from the
pyrolysis bed to the combustion bed with the originally installed weir and
dip leg; for example, gas tended to flow up through the dip leg. There
fore, it was decided to abandon this system in favor of a simple solids
downcomer with a specially designed solids flow-control valve at its lower
end. The downcomer was a 3/8-inch diameter, 20-guage, stainless-steel tube
extending about 1 inch into the upper bed and long enough to reach the
solids flow control valve, which was located between the central
60
disengaging section and combustion section. Essentially, the valve con
sisted of a stainless-steel body and a cylindrical stem 3/8 inch in diameter.
The valve orifice had a diameter of 5/16 inch. A brass bushing and packing
nut were used to prevent seizing of the stem. High-temperature packing,
consisting of asbestos supported over molybdenum wire and lubricated with
graphite, was used to seal the valve stem.
The valve was a recurrent source of operating difficulty as it tended
to get stuck after a few runs and had to be dismantled and cleaned every
two to four runs. Flow of solids from the combustion bed was controlled
by a similar valve, which presented no operating problems.
As discussed earlier, the heat pipe transfers virtually all of the
heat required for pyrolysis. Thus, the design heat-transfer capacity of
the heat pipe can be obtained as the total heat input required for pyrolysis
plus any heat loss from the reactor. In actual practice, some heat is
transferred by the hot fluidizing gas and some by radiation from the burner.
In designing the laboratory equipment, heat losses were not accounted for
since the equipment was originally intended to be insulated to approach
adiabatic operation.
The heat pipe used was obtained from the Dynatherm Corporation and was
constructed from 3/4-inch outside diameter, 316 SS tubing with a 0.065-inch
thick wall. Two layers of 30-mesh stainless steel screen were used for
a wick on the entire 7-foot length of the heat pipe. The diameter of the
vapor core was 0.568 inch. The design heat-transfer capacity of the heat
pipe was about 3695 Btu/hr at 500 C, which is about three times the re
quired heat-transfer load mentioned above. At 600 C, the heat-transfer
capacity of the heat pipe is about 13,725 Btu/hr or about 4,000 watts.
The heat pipe was tested for 2,000 watts at 600°C at the factory.
61
Figure B7 shows the location of thermocouples, pressure taps, and
sampling points for the experimental apparatus. Type K thermocouples of
316 SS sheathed, and magnesium-oxide insulated were used for measuring
temperatures at all the locations. In addition, three thermocouples were
clamped onto the heat pipe to measure heat pipe surface temperature at
three locations, two in each of the two beds and one in the central dis
engaging section. These thermocouples were introduced into the column
through specially made adapters held between the reactor and the central
disengaging section.
Pressure signals from the three pressure taps were read by three
indicating gauges and were fed to two differential-pressure cells, which
transmitted signals proportional to the pressure drops across each bed to
Foxboro Model 40 proportional-integral pneumatic controllers, the outputs
of which were used to control the position of solids-flow valves so that
a constant pressure drop across each bed was maintained.
Gas entering the reactor was sampled at regular intervals or whenever
necessary to measure its oxygen content with a cell made by Bacharach
Instruments, Inc., that was read on a milliammeter calibrated for directly
reading the oxygen partial pressure.
Temperatures of the fluidized beds, the disengaging sections, and
fines-separation system were controlled manually using variacs and dimmer
switches.
Experimental Results
Fluidization studies were made with a single bed (combustion section)
as shown in Figure B8, using air with spent sand from Tar Sand Triangle.
A provisional pressure tap at a height of 10 inches from the distributor
was used to obtain data for estimating bed height. Runs were made at
62
Tar sand. 'B7
\tftfttttfM
'B3
To product recovery section
\ /
'B2
'B5
T B-'
Coked sand s ampling '
'A3" Y
sir Propane in,,
TA>
Propane
'Be
dp cell
I 0 cell
Air in
fl D-
r — - Q-
'Ae
Jppe^bpd level
j
dp r cell
Air in
'Av
Loyej^b^ci level
-ilj Air
FIGURE B7 Instrumentation diagram.
63
'A2
Pi
\-S
P-
w
-ex—1 Flow
control valve
Rotamete)
X Compressed air
Pressure regulator
FIGURE B8 Apparatus for fluidization studies.
64
ambient pressure (645.3 mmlb) and four temperatures—ambient temperature
(23.5°C), 200°C, 400°C, and 600°C.
Each run consisted of charging the bed with a predetermined amount
of solids, heating the bed to the desired temperature, and recording
pressures P-, and P~ from liquid manometers filled with manometer fluid
of specific gravity 2.95 at various air flow rates. Sufficient time was
allowed between readings for the temperature to reach equilibrium.
Typical data are shown in Figure B9, which shows the variation of
pressure drop across the bed of solids with air flow rate at 600 C, for
three values of bed solids hold-up. Values of minimum fluidization
velocities obtained from Figure B9 and similar plots for other temperatures
are plotted versus temperature in Figure BIO. Values calculated from the
correlation of Leva, et al. (B9) are included for comparison. It is seen
that the Leva correlation predicts a minimum fluidization velocity at
room temperature that is 11.6 percent lower than the experimental value.
The deviation increases with increasing temperature. Percent deviation
of predicted values is 14.9 percent at 200 C, 27.8 percent at 400 C, and
35.6 percent at 600°C.
Plots of AP/L, pressure drop per unit bed height, showed that AP/L is
independent of solids hold-up and is not a strong function of gas flow
rate and temperature. The void fraction of the bed is a function of
temperature and gas flow rate, but is essentially independent of bed
solids hold-up. The data obtained were used to estimate bed height for
given values of temperature, gas flow rate, and bed pressure drop.
For processing tar sands, a few runs were made initially with only one
bed to study the behavior of tar sand as it was dropped into a preheated,
fluidized bed of sand particles. In particular, it was necessary to check
for any agglomeration of tar sand after it was introduced into the bed.
65
ON
CD cr 70
CO
O L — i .
N ro Q .
1
cr n> CL
-a -s <T> ( / i in c -s ro Q . -5 O
"O
Ol r+
<y\ o o
=1 DJ t/1 i/>
< ft)
—• o o — i .
r»
*< u
CD
^ O '
-) -̂~̂ :x -s i
- h r+ '•
<-)
o o
! I M i l l -I :•! !
Bed pressure drop, Ap, inches water o o
j : : .
i I
. 1 . . . i
I
l>: :0:
\t
\11A:\:I
i a.
UiL.
1:1!
3 o
JP» o> O l
: it3
f-fr
: C> tu
CX
U
. br Cl-
<1>
;.::.Q_ i t
T,°K Bed temperature
FIGURE BIO Minimum fluidization velocity at various temperatures.
67
These runs helped confirm the notion that tar sand would pyrolyze rapidly
upon entering the preheated bed and that plugging of the bed would not
occur unless the feed rate of tar sand was too high or the bed temperature
was too low.
The effectiveness of the cyclone and electrostatic precipitator for
recovering oil was established by these runs, during which it was discovered
that significant entrainment of fines by the gases and vapors leaving the
reactor occurred. Because these fines could plug the condenser in a short
time, a cyclone for separation of fines was added. A sintered-metal filter
was included later to back up the cyclone.
Following these initial tests with a single bed, the complete equipment
was assembled, except for the heat pipe. The two-stage fluidized-bed column
was tested for performance of solids flow-control valves and the ability
to maintain steady bed heights over a period of time with changing solids
feed rate and gas flow rates. These runs were made with spent sand and
air at room temperature and at high temperatures. Runs were later made
with Tar Sand Triangle feed to further test the apparatus. It was found
during these runs that if the combustion bed was maintained at about
1025 F, the coke deposit would be substantially burned. However, com
bustion was found to be incomplete at temperatures below 930 F.
The heat pipe was installed after these initial tests and processing
runs were made with tar sand from Tar Sand Triangle. Feed preparation was
relatively simple. The ore rocks were crushed in Denver Laboratory crusher
HW 38481 and then ground in a Braun pulverizer, Type VA-53. Material re
tained on a Tyler 48-mesh screen was recycled to the pulverizer. Ground
material in size ranges -48,+65 and -65,+100 was used for the runs.
68
Each run required a lengthy start-up procedure. The beds were charged
with clean spent sand from previous runs and each section of the apparatus
was heated to predetermined temperatures, which were manually controlled
by variacs. Spent sand and air were introduced continuously during start
up, which generally took about 3 to 4 hours before reaching steady temperatures.
The flow of propane was turned on after steady temperatures were reached.
Oxygen content in the gas leaving the burner was monitored to make sure
that most of the oxygen in the air was consumed in the burner. Stoichio
metric amounts of propane and air were used for this reason. During most
of the runs, 1 to 2 pounds of coked sand from earlier runs was added to
the reactor just prior to feeding tar sand. About 4 to 5 pounds of tar sand
was added over a one-hour period during most runs.
Data recorded at intervals of 10 to 15 minutes during a run included
temperatures, pressures, oxygen cell reading, and flow rates of air and
propane. Two to three samples of coked sand and two to four samples of
gas leaving the electrostatic precipitator were taken during the second
half hour of each run. Runs were made to study the effect of varying the
temperature of the pyrolysis bed, the tar sand feed rate, and the nature
of gas in the reactor. Nitrogen was used as a fluidizing gas for a few
runs to see if the presence of carbon dioxide and carbon monoxide in
combustion gases affected the reactor performance.
Several runs were interrupted by malfunctioning of the upper solids
flow-control valve or the plugging of the solids downcomer between the two
beds. During the later part of this work, the valve was routinely dis
mantled and cleaned after every 3 to 5 runs as a precaution against seizing
of the valve stem in the packing nut. Plugging of the solids in the down-
comer was ascribed to local agglomeration of particles or bridging. Such
plugs were usually dislodged by injecting air through a blow gun. A port
69
for injecting air for this purpose was provided on the valve.
Products from each run were collected from the receivers a day or
two after the run. This allowed time for the oil to drain completely
into the receivers. This oil typically contained about 0.1 to 0.5 percent
fines, which were removed by dissolving the oil in benzene, filtering the
solution, and then recovering the benzene. Density, viscosity, and re
fractive index were determined for each sample with a Mettler-Paar DMA 40
Digital Density Meter, a Wells-Brookfield Micro-Viscometer Model LVT, and
a Bausch and Lomb Refractometer. In addition, for most product samples,
elemental analysis was performed to determine carbon, hydrogen, nitrogen,
and sulfur contents. Simulated distillation of the oil samples was per
formed with a Hewlett-Packard Model 5738 gas chromatograph. For the most
part, the above analyses were performed using facilities in the Fuels
Engineering Department.
The gas product was analyzed for oxygen, nitrogen, carbon dioxide,
and hydrocarbons using a Perkin-Elmer Model 154 Vapor Fractometer and
Perkin-Elmer Model 810 gas chromatograph in the Chemical Engineering
Department. Coked sand was analyzed for coke content by combusting a
sample in a Temco Model F-1635 furnace at 500 C for about 12 to 14 hours
and determining the resulting loss in weight.
A total of 64 runs were made, with material from Tar Sand Triangle,
some of which yielded less data than others. Earlier runs were often
accompanied by malfunctioning of some section of the apparatus, inability
to take required samples due to operating difficulties, lack of sampling
and analyzing facilities, etc. Results are presented here only for runs
that produced useful data.
Operating conditions and yield data for these selected runs are
70
summarized in Table-B-II. It is seen that a complete accounting of all the
bitumen in the feed was not achieved in any run. This could have been due
to any or all of the following reasons:
1. Oil condensed as a mist was not completely recovered.
2. The coke measured was less than the coke produced either
due to some combustion of coke in the reactor itself or during sampling.
3. The gas analysis was not accurate.
4. Runs were of short duration; this could lead to sizable end effects
and the data obtained may not be representative of steady-state operation.
Some runs had to be short out of necessity. Improvements in equipment will
have to be made for longer runs.
5. Some product condensed in the cooler portions of the top disengaging
section. This loss could not be accounted for.
6. A small error was probably caused by leaks in the electrostatic
precipitator.
7. Other unknown reasons.
For more meaningful information, operation with larger equipment, for
longer times, and/or with improved product recovery methods is necessary.
Nevertheless, some trends in yield patterns can still be discerned from
the data in Table B2.
Yield data for runs 53, 55, 56, 57, and 58, for which total product
yield ranged from 83.4 to 99.8 percent, are plotted in Figure Bll. For these
runs the feed rate of tar sand was 3.85 lb/hr and air was used to fluidize
the combustion bed. A maximum yield of oil is observed at a pyrolysis bed
temperature of 936 F. Similar results were reported by Gishler (BIO) for
Canadian tar sands with the optimum temperature lying in about the same
region. Coke yield dropped slightly and gas yield fluctuated as temperature
71
TABLE B2
PROCESSING OF TAR SAND TRIANGLE MATERIAL
Run number 38 40 44 48 53
Wt % bitumen in feed
Tar sand feed r a t e , lb /hr
Total quantity of feed, lb
Fluidizing gas to combustion bed
Entering f lu id iz ing gas flow rate, scfm1
Average temperature of pyrolysis bed, °F
3.75
5.0
4.81
a i r
3.75
4.15
4.88
a i r
0.14 I 0.20
3.75
5.0
ai r
904. G • 931.0 ! 946
4.70
5.0
a i r
0.20 i 0.14
860.0
4.70
3.85
4.45 ! 4.23 2.73
ai r
0.14
868.0
Average temperati of combustion °F
Oil yield, wt %
Gas yield, wt c,'.
Coke yield, wt c,,
% accounting of bitumen
j re bed,
1136.0
55.6
_
17.0
72.6
: 1159.0
i 53.0
-
i 20.2
i 73.2
591.0 -1032.0 '1112.C
47.5 • 50.2 | 44.5
I 16.5 30.7 I
16.6 •. 17.3 17.1
i 64.1 •• 84.0 92 c.. o
Standard conditions refer to 1 atm and 70CF.
72
TABLE B2 - (Continued)
Run number
Wt % bitumen in feed
Tar sand feed rate, Ib/hr
Total quantity of feed, lb
Fluidizing gas to combustion bed
Entering fluidizing gas flow rate, scfm1
Average temperature of pyrolysis bed, °F
Average temperature of combustion bed , °F
Oil yield, wt %
Gas yield, wt %
Coke yield, wt c.
% accounting of bitumen
55
4.7
3.85
3.87
air
0.14
1020.0
1170.0
42.6
23.0
17.8
83.4
56
4.70
3.85
4.40
air
0.14
979.0
1154.0
48.4
32.3
18.4
99.8
57
4.70
3.85
4.40
air
0.14
935.0
1093.0
52.6
25.2
21.0
98.8
58
4.70
3.85
4.40
air
0.14
887.0
1117.0
49.5
20.6
22.0
92.1
59
4.36
4.40
4.40
air
0.14
930.0
1217.0
39.1
13.8
16.4
69.3
•'Standard conditions refer to 1 atm and 70°F.
73
TABLE B2 - (Continued)
Run number
Wt % bitumen in feed
Tar sand feed rate, Ib/hr
Total quantity of feed, lb
Fluidizing gas to combustion bed
Entering fluidizing gas flow rate, scfm1
Average temperature of pyrolysis bed , °F
Average temperature of combustion bed , °F
Oil yield, wt %
Gas yield, wt %
Coke yield, wt %
% accounting of bitumen
60
4.36
3.60
4.40
air
0.14
932.0
1177.0
40.1
18.0
19.1
77.2
61
4.36
4.80
4.40
air
0.14
932.0
1137.0
41.5
16.2
17.3
75.0
62
4.36
5.30
4.40
air
0.14
932.0
1163.0
40.6
19.1
17.1
76.8
63
4.36
3.60
4.40
N2
0.14
933.0
1187.0
31.5
16.0
15.1
62.6
64
4.36
5.30
4.40
N2
0.14
937.0
1231.0
39.4
7.6
16.4
63.4
Standard conditions refer to 1 atm and 70°F.
74
800 900 1000 Temperature of pyrolysis bed, °F
FIGURE BIT Material yields with Tar Sand Triangle feed.
75
of pyrolysis was increased from 868 to 1020 F.
The change in oil yield at low temperatures is in agreement with the
data of Barbour, et al. (Bll) who observe an increase in oil yield as
temperature of pyrolysis is increased from 500 to 1000 F. Reversal of this
trend, observed here at higher temperatures, probably results from enhanced
cracking of bitumen to produce more light hydrocarbon gases at the expense
of oil yield. Other studies (B12, B13, B14, B15) generally indicate re
duction in oil yield as the temperature of pyrolysis is increased in this
temperature range.
In runs 59 to 62, for which total products yield was only 69.3 to 77.2
percent, tar sand feed rate was varied and pyrolysis bed temperature was
held essentially constant at about the optimum value shown in Figure Bll.
Little variation in yields of oil, gas and coke are observed. Rammler (Bl4)
reports an increase in oil yield as feed rate is increased. However, his
data were obtained at much higher feed rates compared to data here and a
different type of reactor was used. It is quite possible that the range of
feed rates studies here was too narrow to discern any differences. In
runs 63 and 64, the fluidizing gas was changed to nitrogen. The yield of
oil at the lower tar sand feed rate was notably low.
Before analyzing the data further, it is of interest to put the entire
set of data in perspective. Three different feed samples were used con
taining 3.75, 4.70, and 4.36 weight percent bitumen. Oil yields with tar
sand containing 4.36 percent bitumen are consistently lower than those
obtained with the other two tar sand feeds. This is thought to be due to
an error in the analysis for bitumen content of the third feed. In the
case of the tar sands containing 3.75 and 4.70 percent bitumen, the bitumen
content was determined by extraction with benzene using a Sohxlet type
76
extractor. Bitumen content thus determined is thought to be quite accurate.
In the case of the third feed, the above method was not successful due to a
higher concentration of fines. Bitumen content reported here was therefore
determined by heating a sample at 500 C for 14 hours in air and measuring
the loss in weight that was ascribed to combustion of bitumen content. To
gain some idea of the size of error involved, we may note that tar sand
samples containing 9.65 percent bitumen, as determined by extraction with
benzene, showed a loss in weight of 10.92 percent when heated to 500 C for
14 hours in air.
The higher yield of oil in run 64 compared to that in run 63 is probably
due to better product recovery. At the higher tar sand feed rate, the
concentration of oil vapor in the vapor-noncondensable gas mixture is higher
and this is thought to be favorable to the recovery of oil as it condenses.
With higher concentration of vapor in the non-condensable gas, more vapor
will condense, forming mist droplets of larger size, and making oil recovery
easier in condenser and cyclone. Not surprisingly, in run 64, 32 percent
of the oil product was recovered there. In each case, the remaining product
was collected in the electrostatic precipitator.
The lack of any trend in the values of oil yield for runs 59 to 62
can be explained by the fact that since air and propane were used in these
runs, the fluidizing gas for the pyrolysis bed contained a significant
amount of water vapor, which added to the total concentration of con-
densables in the product stream. Therefore, the relative change in con
centration of condensables due to changes in feed rates was relatively
small. The oil yield in these runs is, not surprisingly, about as high
as in run 64 for which a high tar-sand feed rate was used.
Products of combustion entering the pyrolysis reactor would contain
77
Figure B12 shows the results of simulated distillation of both bitumen
and oil from run 57. It is evident that significant cracking of bitumen
occurs during thermal recovery.
Gases leaving the reactor were found to contain hydrogen, methane,
ethane, ethylene, propane, propylene, butanes, and butylenes along with
nitrogen, carbon dioxide, and some carbon monoxide. Sulfur compounds
could not be identified because we did not have suitable equipment for
such determination.
Conclusions and Recommendations
In spite of the many difficulties that arose during the conduct of
this research, the basic processing concept was demonstrated and much use
ful information was gained on the energy-efficient thermal recovery of
oil from tar sands. Most of the difficulties could be attributed to the
result of using a relatively small experimental apparatus. It is felt
that a somewhat larger apparatus would have been relatively trouble-free
in operation and would have permitted longer run durations, at the expense,
however, of much larger quantities of tar sand feed material.
The following general conclusions can be drawn from this work:
1. The basic concept of the process is workable; equipment configuration
and the material and energy flow schemes adopted for this process permit
thermal recovery of oil from tar sands in a simple process scheme without
adversely affecting the yield of oil.
2. It appears from considerations of energy efficiency that tar sands
containing as low as 8 percent bitumen can be thermally processed without
external energy input to get satisfactory yields of oil. Tar sand with
even lower bitumen content can be processed with good oil yield if a cheaper
external energy source, such as coal, can be added to the burner to provide
energy.
79
about 10 to 11 percent by volume of water vapor when stoichiometric propane
and air were used in the combustion bed, as was the case when feed addition
was commenced for all the runs. Due to the short duration of runs, it was
not possible to turn off the propane as the runs progressed. In a few runs
propane flow was reduced as the run progressed, but was not completely
stopped as would be done in the actual process.
We can conclude from the above data that the presence of carbon dioxid
or carbon monoxide was not an important factor in affecting the yield of
products. Hence, the simplicity in configuration afforded by using two-
stage fluidized bed equipment seems justified.
No clear trends in the properties of oil, such as specific gravity
and viscosity were observed with the temperature of pyrolysis. Viscosity
of oil was in the range of 100 to 400 cps and gravity in the range of 11
to 13°API.
Measurement of the refractive index was not easy as the oil was quite
dark, but the value was about 1.54 and did not show variation with products
from different runs.
Elemental analysis of bitumen and recovered oil from run 57 was as
follows:
% by Wei
C
H
N
S
C/H
.ght Bitumen
78.9
8.8
0.4
9.6
0.747
Oil
85.4
10.7
0.2
5.5
0.665
It is seen that the C/H ratio of the product is much lower than that of
original bitumen. There is some reduction in nitrogen and sulfur contents
as well.
78
< < < « « « < < < < I
00
o
200 300 400 500 550 Temperature, °C
FIGURE B12 Simulated distillation of Tar Sand Triangle bitumen and oil, (Run No. 57).
3. The basic process is very flexible in handling variations in feed
with respect to bitumen content, fines content, and water content.
4. The process is capable of treating tar sands containing both
consolidated and unconsolidated sandstone.
5. Modifications of the basic process, such as introducing recycle
of gas and oil, allowing for purge of some combustion gas, etc., can im
prove the energy efficiency of the process and the yields of oil and gas.
6. Modifications such as purging of some combustion gas would help
in the recovery of oil product.
7. Because of incomplete recovery of products, yield of oil obtained
from the laboratory unit was lower than would be obtained in large-scale
processing of tar sands. It is thought that lower yields obtained here are
related to the limitations imposed by the scale of operation, inability to
make longer runs, inefficiencies of the recovery system, etc. Oil yields
of 75 to 80 percent by weight could be realized in a well-designed commercial
unit.
8. An optimum temperature of pyrolysis exists at which the yield of
oil would be about maximum. It is not concluded in this study if operation
at other temperatures to improve product quality, even at the cost of yield,
would be justifiable. Insufficient analysis of the products, possible errors
in analysis, and the relatively small number of runs made preclude any
judgment about influence of temperature of pyrolysis on product quality.
9. For a given flow rate of fluidizing gas in the pyrolysis step,
increase in feed rate of tar sand seems to help in the recovery of oil.
10. Presence of steam in the fluidizing gas for the pyrolysis reactor
appears to have a beneficial effect on recovery of oil.
11. Presence of carbon dioxide and carbon monoxide in the pyrolysis
reactor do not seem to affect oil yield significantly.
81
12. The solids flow-control valve between the reactor and the burner
and the solids downcomer presents some operating difficulty at higher
temperatures. In particular, it was noted that plugging of the solids
flow system occurred with singular consistency when the sand temperature
was likely to be around 575 C. Any connection of this observation to the
high-low transition of silica, which also occurs at 575 C, needs to be studied.
It is felt that the process developed during the course of this work is
simple, direct, and efficient. It is capable of wide application to pro
cessing of tar sands in Utah and Canada, and perhaps other deposits. Further
work is now in progress with samples from the Asphalt Ridge and Sunnyside
deposits in Utah.
It is recommended that the oil recovered from tar sands be studied
for use as petrochemical feedstock as well as a source of energy. Such
research is being conducted elsewhere. The hot sand leaving the burner,
or at least a part of it, may be used before it is cooled as feed for an
auxiliary process, such as for making foamed glass.
A mathematical model for the process would be helpful in design and
operation of a larger unit. Pilot-plant data can provide a basis for such
a model. Finally, the processing concepts developed should be studied for
application to process other hydrocarbon-bearing materials such as oil
shale and coal.
82
REFERENCES
Bl. Kirk-Othmer Encyclopoedia of Chemical Technology, 2nd ed. S.v. "Shale Oil," by R. E. Gustafson, 18, 1969, pp. 1-20.
B2. Kirk-Othmer Encyclopeodia of Chemical Technology, 2nd ed. S.v. "Oils, Essential," by Max Stoll, 14, 1967, pp. 178-216.
B3. Moore, R. G.; Bennion, D. W.; and Donnelly, J. K. "Anhydrous Extraction of Hydrocarbons from Tar Sands." Paper presented at local ISA Meeting, Calgary Section, April 1975.
B4. Flynn, P. C.; Porteous, D. C.; and Sulzle, R. K. "Heat and Mass Balance Implications for Direct Coking of Athabasca Tar Sands." Energy Processing, (October 1976):42-48.
B5. Ritzma, H. R. "Utah's Tar Sand Resource: Geology, Politics and Economics." AlChe Symposium Series, 72 (1976):47-54.
B6. "Energy from U.S. and Canadian Tar Sands: Technical, Environmental, Legislative and Policy Aspects." Report to the Committee on Science and Astronautics, U.S. House of Representatives. Washington, D.C.: U.S. Government Printing Office, 1974.
B7. Sepulveda, J. E. "Hot-Water Separation of Bitumen from Utah Tar Sands." M.S. Thesis, Department of Mining, Metallurgical and Fuels Engineering, University of Utah, 1977.
B8. Seader, J.D., and Jayakar, K.M. Process and Apparatus to Produce Synthetic Crude Oil from Tar Sands. Application No. 851,226 for U.S. Patent, filed 15 December 1977.
B9. Leva, M. Fluidization. McGraw-Hill Book Co., 1959.
BIO. Gishler, P.E. "The Fluidization Technique Applied to Direct Distillation of Oil from Bituminous Sand." Canadian Journal of Research, 27 (March 1949):104-111.
Bll. Barbour, R.V.; Dorrence, S.M.; Vollmer, T.L.; and Harris, J.D. "Pyrolysis of Utah Tar Sands—Products and Kinetics." Paper presented to the 172nd National Meeting, ACS, Division of Fuel Chemistry, San Francisco, September 1976.
B12. Peterson, W.S., and Gishler, P.E. "Oil from Alberta Bituminous Sand." Petroleum Engineer, 23 (April 1951):553-561.
B13. Peterson, W.S., and Gishler, P.E. "The Fluidized Solids Technique Applied to Alberta Oil Sands Problem." Proceedings of Alberta Oil Sands Conference. Edmonton (September 1951).
83
Rammler, R.W. "The Production of Synthetic Crude Oil from Oil Sand by Application of the Lurgi-Ruhrgas Process." Canadian Journal of Chemical Engineering, 48 (October 1970):552-560.
Filby, J.E.; Flynn, P.C.; and Porteous, K.C. Paper presented at the 27th Canadian Chemical Engineering Conference, Calgary, Alberta, 1977.
84
EFFECT OF VARIABLES ON THERMAL PROCESSING
The initial phases of our program relied on the extensive literature
that reported the passage of the Canadian oil sand development programs
from the laboratory to the commercial plant, that is, the GCOS (CI) and the
Syncrude, Ltd. (C2) processes. There are significant differences in the
physical and chemical natures of the Canadian and Utah bituminous sands.
The Utah deposits contain an average of 0.5% by weight connate water com
pared to an average of 4.5% by weight connate water for the Canadian
deposits. In addition, the sulfur contents of the Utah bitumens are low
compared to the Canadian bitumens (0.6% by weight for the Sunnyside deposit
versus 3.5-4.0% by weight for the Canadian deposits). The connate water
content may affect the choice of the optimum processing technique for the
recovery of the synthetic crude whereas the sulfur content will affect the
choice of the optimum processing sequence for the upgrading of the recovered
synthetic crude. These physical and chemical differences, combined with the
difference in the geographical and climatic conditions of Utah and Alberta,
may make it necessary to recover the bitumen or a synthetic crude from the
bituminous sands of Utah by a uniquely different technique than that used
to recover the bitumen from the Canadian sands.
The processing concepts currently under investigation for the recovery
of a bitumen or a synthetic crude from bituminous sands include hot water
extraction of the bitumen, thermal coking of the sand to produce a synthetic
crude and solvent extraction of the bitumen. The hot water and the thermal
coking techniques can be utilized in either an in-situ [steam injection is
considered as an in-situ hot water technique (C3)] or an aboveground mode of
operation.
85
The aboveground thermal recovery of a synthetic crude from bituminous
sands involves the mining of the deposit, the transportation of the mined
sand to the processing site and the feed sand preparation (i.e. crushing,
etc.). The thermal recovery techniques pose significant heat transfer
problems due to the high temperatures (675-925 K) required for the release
of the synthetic crude from the bituminous sand. This constraint can be
relaxed by utilizing a fluidized bed and by recycling the hot sand from
which the coke has been combusted. An integrated process in which the heat
generated in the coke combustion unit is recycled to the fluid-bed coking
unit should be energy self-sufficient provided the coked sand contains one
to two percent by weight carbonaceous residue.
The absence of a water film between the bitumen and the sand particles
in the bituminous sands of Utah and the occurrence of most deposits in the
form of consolidated sandstones led to the speculation that the fluidization
characteristics of the Utah sands as well as the synthetic crude yield and
quality may be considerably different than might be expected when processing
the Canadian sands in a fluidized bed coking unit. There have been no
reports in the literature on the thermal recovery of synthetic crudes by a
fluidized bed technique for the bituminous sands of Utah. Therefore, a
bench-scale pilot unit was constructed at the University of Utah to obtain
preliminary process variable data on the feasibility of an aboveground,
fluidized bed coker for the production of synthetic crudes from the bitumi
nous sands of Utah. The data obtained in these preliminary experiments are
presented in this communication.
Experimental
A schematic of the experimental apparatus is presented in Figure CI.
At the start of each experiment the reactor (D) was charged with coked sand
86
0 0
-O- REGULATOR VALVE -**-• CONTROL VALVE O PRESSURE GAGE -o THERMOCOUPLE — CONTROL SIGNAL
•—-"'P4—
Figure CI. Fluid Bed Coker for Bituminous Sands
produced in the preceding experiment. This precoked sand was fluidized
and the reactor was brought to the desired coking temperature and hydro-
dynamic and thermal stability were established in the bed. Nitrogen from
the gas manifold (A) was used as the fluidizing medium and its flow rate
was monitored with a calibrated rotameter (B). The nitrogen entered at the
bottom of the reactor assembly (C) and passed through a calming section
where it was preheated to the coking temperature. Pre-sized bituminous
sand was fed to the reactor from the storage hopper (H) under free fall
conditions by means of a screw feeder (F). Thermal energy was supplied to
the systems by electrical resistance heaters. The nitrogen-synthetic crude
vapor mixture passed from the reactor into the expansion chamber (E) where
the vapor and sand particles disengaged. Entrained sand fines were removed
by two cyclone separators (K.. and K„). The nitrogen-vapor mixture was
passed through a cylindrical, fine-mesh filter (M) prior to entering the
produce recovery train. The cyclones and the filter were maintained at
693 K and 653 K, respectively, to prevent condensation of the vapor. The
product recovery train consisted of a water-cooled condenser (N), a cyclone
(U) and a series of fiber mist absorbers (W) maintained at the ambient tem
perature. The synthetic crude absorbed by the cellulose fibers was stripped
from the fibers by a suitable solvent (benzene, toluene, etc.). The non-
condensable, non-absorbable light hydrocarbon gases were chromatographically
analyzed, metered and vented.
The dynamic, fluidized bed depth was maintained constant by a solids
flow valve (V-10) controlled by a differential pressure cell (£) that con
tinuously monitored the pressure drop across the bed.
The uncorrected material balances for each run exceeded 92% by weight
of the bitumen fed to the reactor. The percent recovery ranged from 92 to
88
99% by weight. The liquid yields were corrected to account for (i) the
liquid that was lost with the solvent during the vacuum stripping of the
solvent-synthetic crude mixtures obtained from the fiber absorbers and (ii)
the liquid that condensed on the screwfeeder outlet above the expansion
chamber.
The chromatographic analysis of the light gases was performed on a
Hewlett-Packard Model 5830A gas chromatograph using a Chromosorb 102
column (6.1 meters in length). A simulated distillation of the extracted
bitumen and the liquid products produced in this investigation was done on
a Hewlett-Packard Model 5734A gas chromatograph using a column packed with
three percent Dexil 300 on Anachrome Q (46 centimeters in length). Addi
tional details on the experimental apparatus and procedures can be obtained
from reference (C4).
Results and Discussion
Effect of Temperature on Product Yield and Distribution
The yields of light gas (C,-C,), naphtha (C^ -478 K), middle distillate
(478-617 K), heavy gas oil (617 K +), total synthetic crude (C*), and coke
are presented in Figure C2 and Table CI as a function of the coking bed
temperature. All yields are reported as weight percent based on the bitumen
fed. The base operating conditions for the investigation were atmospheric
pressure, a solids retention time of 27.2 minutes, a sand feed particle size
of 358.5 microns and a coking bed temperature of 773 K.
The yield of the C^ liquid passed through a maximum with temperature
(61.2 wt % at 723 K), however, at the lower temperature (698 K) a solvent
extractable liquid ("soft" coke) remained on the sand particles with the
coked bitumen (non-extractable "hard" coke). If this liquid is considered
as unliberated synthetic crude then the C- liquid yield generally decreased
with increasing temperature (dashed line, Figure C2). Although the liquid
89
TABLE CI
Effect of Temperature on Yield and Product Distribution
Sunnyside Feed
o
Experiment Number
Coking Reactor Temperature, K
Retention Time of Solids, min.
Feed Sand Particle Size, y
Gas Make, LPH at STP
Synthetic Crude Yield, gm/hr
Mass Balance (Weight Percent)
co2
Cl ' C3
C4
C 5+ Liquid
Coke
56 52 54 53 55
690
27.2
358.5
7.3
19.8
1.2
12.5
4.1
51.0
31.2
(15.8)*
723
27.2
358.5
7.3
23.9
1.3
10.8
4.0
61.2
22.6
748
27.2
358.5
10.6
18.4
1.8
18.1
5.9
55.1
19.2
773
27.2
358.5
14.5
22.2
1.5
21.0
6.5
47.5
23.5
798
27.2
358.5
16.0
22.9
2.7
22.9
6.7
45.0
22.8
Weight percent of soft coke determined by solvent extraction.
Q
>-
100
9 0
8 0
7 0
V SYNTHETIC CRUDE
O c, ° 4 GASES O COKE O NAPHTHA a MIDDLE DISTILLATES A HEAVY ENDS
693
REACTOR TEMPERATURE, K
Figure C2. Effect of Reactor Temperature on Product Yield and Distr ibution for Sunnyside Feed. Retention Time of Solids, 9avq = 27.2 rrnns. Feed Sand Particle Size, dp '- 358.5 microns
91
CM If) m •
r O N O O I N CM i n I N c o
en co
CM
+ |co
i— I N
oo
- o c «
co
to 3 O
E 3
• P •r— CO
0) •o •r-(/) >> c c
C O
CM E
o o _ J CD CJ
is 1 C J
o •r-+J O) . c +•> c >> t o
to
•r—
i -O) C L O S -
CM i n
c o I N c o si- CM i n tN r o
C M
en
CM
+ |oo
cr> i n CO
CD -a 3 i-
o
cu - E - P C
0O
CM I D
• • co I N CO CM CM i n I N co
i n s f
s f
+ |CM
CO CO CM
CO I N I N
1 3 3 CJ
a*i
a> > •M <o i —
r_ IN
sf
IN CM
CO
o o CO r—
CO Sf
CO r—
in CO
sf CM
CO sf
O co
_̂ CTl
ID CO
en sf
co sf
Sf sr o in
CM CM
IN in
in in sr co
in CM
, IN
in IN
in r-
o o o o t—
CO s f I N
co CM I N
•r— » E
<D S~ « 3 </> +J - o
i~ I— aj o ex co E 0J <4-
I— o
i - OJ O E
4-> - r -O H-<a GJ c
cc o • r -
CT>+->
c c J* 4-> o a> o a:
cu N
• r -CO
<D r— O
• r -+-> 5-ea a.
(O 4-> C O S -
<u - o Q j <D O <u i -u- a.
co en CM
i - M • • -rc 4->
c 1-4 CJ o_ o <c
o
+•> in • i - o > o 5- •!-
^ 5
•tJ
c S^ o
X I CO s-CTi ra CM CJ
•)-> C (0 O
10
wi fa o s-O - C
O CJ
e Q) E 3 •)-» •r -
ca
- a 0J
4-> O fO s-+-> X
U-i
C O i n co
re O
<— C D C O I N i n I N
c o c M c o c o c o i n o o c o s f C T i i — s f m C T i t N C o
o o in N in N
s t c o s j - m i n v o i o c o c o c o CO s f s f CM
1 3 E CJ
*«
<u > •4-> f0
i —
in sl-
co
CM 00
CO
oi CO
O r—
_ co co r^
CM CO
CM CM
_ CD
CD CM
CO CO
r»-CO
o CO
< st-
00 00
CM in
st-00
o CD
CD sf
cr> CD
CO CD
CD r-~
r-~ CO
,_ CO
o o o o r—
«0
o
LO r-» r .̂ s f co o
C M i — CTl C M !-•» <J If) N IN < •
co vo CM CM en O CT> C D CM i —
co co s f i n co co oo CO
C O
r^ i n co
i 3 <U E > 3 • ! -
CJ +-> fO
^ ? 4-> O
i n CM
I N o
CTl i — m i n
s f en
oo co en r-»
co CD en sf o I N C M C M
m co
co o
s f s f
IN O
i n s f
IN CM
C D IN
CO C M i n co
CO CM 0 0 I N CO 0 0
i n i— CM 0 0
CM CM i n CTI
s f s f 0 0 r N s f 0 0 CM s f O I N CM O CO 0 0
CO CM CO s f CD 0 0 CTl CTl
s f r—
O CTl
i n co
o o
o o
s f CM
CO r—
I 3 <D E > 3 - i -
O -t-> <o
fO
O
r— CM i— O O CTi r— CD s f I N
CO CD
s f I N CD 0 0 CTl CO s f s f CD i— I N s f I N s f
O O
r — C M C o i n c O i — i n C T i L n C T i C M i— .— r- CM CM CO
o o
CTi
o
CTi sf CM
C3 O
CTi O IN CO
C O r — C O C T i C M r — C T i C D C O c n c o c M s f c o i N C M C D i n
O i — i — C M C O C O s f i n s f C M IN C D
s_ 3
-M ca s -cg O -
E
4-> C 3 «r- C J
CJ o o Q .
o cn O CM CM CM
O un CM
i n tN C M
i n C D co
i n ro co
i n CO co
i n CTi co
i n i n CM i n s f s f
in in co oo >— co s f i n i n
c o
• r -+•> CJ <0 s-
U-
| N I 0 0 CTi CM co s f i n co 0 0
0) 3 -a
CTl i -f— in
0) a:
93
yields are low compared to the data of Peterson and Gishler (C5) the trend
is similar.
The "hard" coke yield increased with increasing temperature up to 723 K
and remained approximately constant at 19-23% by weight based on bitumen fed
above 723 K. A similar trend was observed by Filby _et. al. (C6) and despite
the chemical differences in the natures of the Canadian and Utah sands and
in the coking bed temperatures the weight percent bitumen converted to coke
was about the same in both investigations.
The light gas production increased with increasing temperature at the
expense of the 617 K + heavy gas oil, however, the increase in naphtha
reported by Filby e_t. a_l. (C6) was not observed with the Utah sands. This
may be due in part to the difference in the chemical nature of the Canadian
and Utah sands, or to the differences in operating conditions. The carbon
dioxide in the light gas is believed to have been produced by the decompo
sition of carbonates in the sand matrix.
Effect of Temperature on Product Quality
Selected physical properties of the extracted bitumen and the effect
of the coking reactor temperature on the physical properties of the
synthetic crudes are presented in Table C2. The API gravity of the liquid
decreased with increasing coking reactor temperature concomitant with an
increase in the Conradson carbon residue. A marked decrease in the syn
thetic crude viscosity was observed as the coking reactor temperature
increased. The simulated distillation data are discussed in terms of an
813 K (540°C) cut point. The amount of liquid boiling below 813 K (540°C)
is greater at the lower coking reactor temperature and decreases with in
creasing temperature. The liquid boiling below 638 K (365 C) increases
92
with increasing coking reactor temperature, an indication that the hydro
carbon species boiling above 698 K (425 C) are undergoing thermal cracking
at the higher reactor temperatures. A similar observation was reported by
Peterson and Gishler (C7).
Effect of Solids Retention Time on Yield
The solids retention time (8, minutes) was defined as
6 = 60 W/F
where W is the weight of solids in the bed, kg, and F is the sand feed rate,
kg h . In this investigation, the retention time was varied by increasing
or decreasing the sand feed rate while keeping the bed height and mass con
stant. The effect of solid retention time on the synthetic crude yield
and coke make is presented in Table C3 and Figures C3 through C5. The
yield of synthetic crude decreased with increasing retention time and
the yield of light gas increased with increasing retention time at each
temperature studied (Fig. C3 and C4). The amount of coke produced was
relatively insensitive to changes in the solids retention time (VL9-23% by
weight of bitumen fed). The increased solids retention time would appear
to increase the residence time of the liberated hydrocarbon vapor in the
coking zone thus leading to more extensive thermal cracking of the vapor.
Decreasing the solids retention time shifted the temperature at which
the maximum liquid yield was obtained and increased the yield of liquid at
the maximum temperature (Fig. C5). At a retention time of 20.4 minutes the
maximum liquid yield was 67.4% by weight at 773 K whereas at a retention
time of 27.2 minutes the maximum yield was 61.2% at 723 K.
Retention times below 20 minutes were not investigated due to a limita
tion in the reactor throughput capacity. If we can reasonably extrapolate
94
« « « I € t I
TABLE C3
Effect of Solids Retention Time on the Yield and Product Distribution
Sunnyside Feed
Experiment Number Coking Bed
Temperature, K Retention Time of
Solids, min.
Feed Sand Particle Size, v
Gas Make, LPH at STP Synthetic Crude
Yield, gm/hr
Mass Balance (Weight Percent)
co2
c1 - c3
C4
C 5+ Liquid
Coke
68
723
31.4
358.5
9.3
25.1
2.3
15.2
4.3
55.8
22.5
52
723
27.2
358.5 7.3
23.9
1.3
10.8
4.0
61.2
22.6
61
723
20,4
358.5 7.7
32.6
0.9
8.3
2.6
50.4
37.9
(20.9)*
66
773
31.4
358.5
15.9
23.1
2.8
26.8
8.0
42.8
19.7
53
773
27.2
358.5 14.5
22.2
1.5
21.0
6.5
47.5
23.5
64
773
20.4
358.5
9.0
44.0
1.3
8.7
2.6
67.4
20.0
67
798
31.4
358.5
14.3
22.8
5.2
23.3
6.8
42.2
22.5
55
798
27.2
358.5
16.0
22.9
2.7
22.9
6.7
45.0
22.8
65
798
20.4
358.5 12.2
41.1
1.6
11.8
4.1
61.8
20.6
* Weight percent of soft coke determined by solvent extraction.
1001—
90
80
70
60
Q 50
UJ
>•
40
30
20
10
• SYNTHETIC CRUDE OC,-C4 GASES m COKE • GAS MAKE
32 IS 20 24 23
RETENTION TIME OF SOLIDS, 0A V G > , MINUTES
Figure C3. Effect of Retention Time of Solids, 0avg, on the Yield Pattern for Sunnyside Feed. Reactor Temperature, T = 773K Feed Sand Particle Size, dp = 358.5 microns
96
• SYNTHETIC CRUDE
O C,-C4 GASES
m COKE
IS 20 24 28
RETENTION TIME OF SOLIDS, 0A VG.
32
MINUTES Figure C4. Effect of Retention Time of Solids, Oavg, on the Yield Pattern for Sunnyside Feed. Reactor Temperature, T = 798 K Feed Sand Particle Size, dp = 358.5 microns
97
100
9 0
80
<2 H r-». t>
*• Q _ !
>-
LU O ZD ££ O
O
X F-:£ >~ co
70
60
50
40
30
20
10
RETENTION
OF SOLIDS,
• 20.4 MIN. A 27.2 MIN.
TIME
&AVG.
698 723 748 773 793 823
REACTOR TEMPERATURE, K
Figure C5. Effect of Retention Time on the Optimum Temperature for Maximum Yield of Synthetic Crude.
Feed Sand Particle Size, dp 358.5 microns
98
the data obtained in this investigation a liquid yield of 80% by weight
of bitumen fed would be obtained at a coking reactor temperature of 773 K
with a solids retention time of 16 minutes. The solids retention time-
liquid yield data in the literature lead to conflicting interpretations,
that is, Matchen and Gishler (C8), Safonov et. al. (C9) and Filby et. al.
(C6) reported no effect of sand retention time on liquid yield. Rammler
(CIO) observed that the plant capacity (directly related to feed rate)
has a definite influence on the liquid yield in the Lurgi-Ruhrgas direct
coking process when processing a Canadian sand.
Effect of Particle Size and Particle Size Distribution on Yield
The effects of particle size and particle size distribution of the
feed sand on the yield and product distribution are presented in Table C4.
The particle size data were acquired at a coking bed temperature of 773 K
and a solids retention time of 20.4 minutes. A reduction in sand particle
size from 358.5 microns to 253.5 microns had little or no effect on the
liquid yield and on the product distribution. However, a significant
shift in product distribution was observed when the sand particle size was
increased from 358.5 microns to 507.5 microns. The light gas yield in
creased from 11.3 to 26.7% by weight of bitumen fed while the C_+ liquid
yield decreased from 67.4 to 51.8% by weight. Thus a substantial portion
of the C,+ hydrocarbon was thermally cracked to lighter species, in parti
cular, C.-C- gases. We speculate that a fraction of liberated hydrocarbon
vapor was "trapped" in the pore structure of the larger particles. The
diffusion time for these species to transfer from the internal region of
the sand particles to the bulk fluid phase was therefore increased. This
increased residence time within a microscopic thermal cracker (i.e., the
99
TABLE C4
Effects of Feed Particle Size
on Yield and Product Distribution
Sunnyside Feed
Experiment Number 71 64 69 59
Coking Bed Temperature, K 773 773 773 773
Retention Time of Solids, 6avg>
m l n
Feed Sand Particle Size
Rate of Bitumen Feed to Reactor, gm/hr
Gas Make, LPH at ST?
Synthetic Crude Yield,
!» V
gm/hr
Mass Balance (Weight Percent)
co2
C 1 " C 3
C4
C5+ Liquid
20.4
253.5
75.5
9.2
40.8
2.3
9.7
2.9
65.1
20.4
358.5
82.8
9.0
44.0
1.3
8.7
2.6
67.4
20.4
507.5
91.0
21.9
40.6
2.3
20.6
6.1
51.8
25.5
162.0
76.5
11.9
32.0
1.3
13.6
4.3
63.5
Coke 20.0 20.0 19.2 17.4
100
pore structure of the sand) led to conversion of the higher molecular
weight species to C,-C, gases.
A single experiment was made with a wide cut feed sand (Tyler Sieve:
20-150 mesh fraction) to determine the effect of the sand size distribution
on the liquid yield. The yield was similar to that obtained with the
smaller feed sand particles, that is, 63.5% by weight liquid and 17.9% by
weight C -C, gases. A size distribution analysis on the sand indicated that
65% of the feed sand was finer than 358.5y and it would be expected to
exhibit yields more nearly like the smaller feed sand particles (>. 358.5u)
than like the large particles (>_ 507.5u).
Characterization studies on the extracted bitumen and the synthetic
liquid were initiated to determine the molecular compound types present and
to gain an insight into the possible reaction pathways that occur during the
pyro-distillation of bitumen from the sand. The atomic ratio of hydrogen to
carbon computed from elemental analysis showed that the liquid product
became more aromatic at higher reactor temperatures. Gradient Elution
Chromatographic procedures used in the U.S. petroleum industry to analyze
heavy fractions have been applied to Utah's tar sand bitumens and products
(C11.C12).
These analyses (see Table C5) show that the liquid product contains
35-45 weight percent in MNA-DNA oil fraction compared to 15 percent for
extracted bitumen. However, the GEC analysis of the bitumen may be compli
cated by the high asphaltene content of the bitumen. Beyond the optimum
temperature of 723 K, it was observed during pyro-distillation of Sunnyside
bituminous sand that the yield of light gases increased considerably with
moderate increase in the yield of coke at the expense of liquid yield.
This might arise from the additional cracking of the oil fraction, in
101
Table C5
GRADIENT ELUTION CHROMATOGRAPHIC ANALYSIS OF EXTRACTED BITUMEN AND SYNTHETIC
LIQUID SUNNYSIDE FEED
Reactor Temperature, K
GEC Fractions
Saturates
MNA-DNA Oil
PNA - Oil
PNA - Soft Resins
Hard Resins
Polar Resins
Asphaltenes
Non-eluted
Asphaltenes
353a
14.7
14.7
3.3
3.5
5.4
2.0
43.0
13.4b
698
19.2
34.0
2.7
11.8
2.1
2.6
24.2
3.4
723c
10.5
45.3
1.2
8.1
5.7
2.7
20.3
6.2
773
9.4
36.6
2.9
13.1
3.4
3.7
27.4
3.5
Notes: a - Soxhlet Extraction of bituminous sand using bezene as solvent
b - Benzene soluble asphaltenes
c - Optimum reactor temperature at a solids retention time of 27 minutes
102
particular the saturates and the MNA-DNA oil fractions. The compound type
separation technique will be continued to provide process data on the chemical
nature of the various synthetic liquids derived from Utah bituminous sands.
Conclusions
1. An aboveground, fluidized bed thermal process for the recovery of
a synthetic crude from bituminous sands of the Sunnyside (Utah) deposit could
be a feasible alternative to a modified hot water process.
2. Synthetic crude (C<-+ liquid) yields in excess of 80% by weight of
bitumen fed might be attainable at solid retention times below 20 minutes.
3. The consistency of the coke yields indicates a commercial or
large scale pilot plant could be maintained in thermal balance regardless
of the coking bed temperature, the feed sand retention time or the feed sand
particle size.
4. The process variable data obtained in this exploratory investigation
indicate the size of a commercial thermal coking unit for the recovery of
a synthetic crude from bituminous sands would be consistent with engineering
technology currently employed in the petroleum industry.
5. Gradient chromatographic analysis of tar sand bitumens and products
show that terminal recovered products are considerably higher in mononuclear-
dinuclear aromatic materials than the feed while asphaltenic materials are
lower.
103
References
Bachman, W. A. and Stormont, D. H., Oil Gas J. 69, Oct. 23 (1967).
Nulty, P., Fortune, 72, May 22 (1978).
Thurber, J.L., and Welbourne, M.E., Petrol. Eng., 31, November (1977).
Venkatesan, V.N., Ph.D. Dissertation, University of Utah (1979).
Gishler, P.E. and Peterson, W.S., Canad. Oil Gas Ind. 3^, 26 (1949).
Peterson, W.S., and Gishler, P.E., Canad. J. Res. 28, 66 (1951).
Matchen, B. and Gishler, P.E., C51-51S, National Research Council of Canada, Ottawa, Canada (1951).
Safonov, V.A., Indyukov, N.M., Loginova, S.M., and Shevtsov, I.S., Sb. Tr. Inst. Nabtekhim Protssov, Adad. Nauk Azerb. SSR #4, 272 (1959).
Rammler, R.W., Can. J. Chem. Eng. 48, 552 (1970).
Middleton, W.R., "Gradient Elution Chromatography Using Ultraviolet Monitors in the Analytical Fractionation of Heavy Petroleums," Anal. Chem., 39, 1839, (1967).
Callen, R.B., Bendoraitis, J.G., Simpson, C.A. and Voltz, S.E., "Upgrading Coal Liquids to Gas Turbine Fuels. I. Analytical Characterization of Coal Liquids," Ind. Eng. Chem., Prod. Res. Dev., 15 (4), 222 (1976).
104
BITUMEN PROCESSING AND UTILIZATION
Characteristics of the virgin bitumen have been previously determined to
be grossly similar to distillation residues from petroleum (Dl, D2). Processes
useful for upgrading petroleum residues are, therefore, potential candidates
for upgrading of tar sand bitumen. Processes which are used commercially for
upgrading resid are solvent refining (deasphalting, dewaxing, etc.)* visbreaking,
hydrocracking, hydrotreating and coking (D3). With tar sand bitumens the high
costs of recovery of the bitumen may so influence the overall economics that
the most profitable approach for upgrading resid may not be the best approach
for upgrading bitumen.
The principal chemical objective for primary upgrading of tar sand bitumen
is to reduce the molecular weight (and heteroatom content, if possible) with
out extreme losses of yield of distillates and without expensive catalyst or
hydrogen requirements. It is with this chemical (and economic) objective in
mind that studies of both conventional and unconventional primary processes
have been studied. Processes which have been studied are visbreaking, coking,
catalytic cracking and hydropyrolysis. These processes have been studied for
the effect that process variables have on product composition and structure.
This approach allows a qualitative comparison to be made of the efficiency of
the various processes for converting bitumen to valuable products.
Visbreaking
Visbreaking represents one of the simplest processes which can be applied
to heavy bitumens (D3, D4). The objective of visbreaking is to subject the
material to a mild thermal cracking to produce lower molecular weight species.
Conditions of residence time and temperature are selected to minimize coking.
In petroleum processing, visbreaking is used to increase the yield of dis
tillate from the bottoms (D5). However, with tar sand bitumen, another
105
objective might be to reduce viscosity and molecular weight without rendering
the material more difficult to process. The final visbroken product may be
fed to a distillation column to produce distillate and residue. Distillate
could be hydrorefined or catalytically cracked while residue may be used for
fuel oil or coker feedstock (D3-D5).
Results of visbreaking Asphalt Ridge, Utah bitumen are given in Figure
Dl and show viscosity of the liquids as a function of space time for three
different temperatures and two different pressures. Visbroken products were
evaluated principally by viscosity. Viscosity will fail to give, by itself,
important data on product structure; and coking, if it occurs, cannot be
detected by viscosity. However, viscosity exhibits a direct relationship to
molecular volume (D6) (which is approximately related to molecular weight)
and should provide a good index of the severity of reaction conditions. The
amount of gases produced was less than 2% for the 6 atm. runs and less than
1% for the 0.9 atm. runs. Results show that at atmospheric pressure a product
having a viscosity of 450 poise was produced at a reaction temperature of
500 C and three minutes space time. When back pressure is applied to the
system, a dramatic reduction in viscosity is seen at given temperature and
space time conditions.
Results of the visbreaking experiments show that rather severe conditions
may be required before a pumpable fluid is produced. A viscosity of less than
10 poise is normally required for efficient pumping and handling (D7) and only
in the case of 500 C, elevated pressures and three minutes space time, is this
viscosity approached. Efforts have not been made to ascertain the effects of
higher pressures (̂ 500 psig) on product viscosity but the results of coking
at higher pressures indicate that there will be an enhanced tendency to form
aromatics and coke, thus decreasing the value of the resulting product.
106
VISCOSITY OF VISBROKEN BITUMEN
p =.9 otm P = 6.6 otm
TEMPERATURE OF V1SBREAKING °C
T r n r 2 4 6 8
SPACE TIME-T(MINUTES)
1 10
Figure Dl. Viscosity of Visbroken Bitumen.
107
Conclusions which can be drawn based on the visbreaking results are:
(a) Utah bitumen is quite responsive to visbreaking, but a large molecular
weight reduction must be effected before desirable products are produced, and
(b) high severity visbreaking is required to reduce the viscosity to a pumpable
fluid but such conditions may adversely affect the subsequent processing of
the heavy ends.
Coking
Results of bitumen characterization (Dl) showed that tar sands differed
appreciably as a function of the source deposit. Significant differences were
also noted between tar sand bitumen and conventional crude oil residue. The
primary aim in this aspect of the study was to process bitumen from various
sources under a set of reproducible reaction conditions^ in order to provide a
basis for assessing the relative value of products derived.
The gravimetric results of the coking study are given in Table Dl. Gas
yields, which range from 4.8 to 7.5 percent, do not exhibit any obvious
correlation with bitumen properties. The majority of the gas was produced
above 500 C when condensate production was tapering off. The light gases
were apparently not co-produced from the cracking reactions giving rise to
higher molecular weight condensate molecules, implying that the dominant
reactions for production of each are different.
Gas chromatographic analysis of the C,-C,- gases are given in Table D2.
Results show a significant variation in gas composition between source
deposits. The Athabasca sample produced large amounts of methane compared
not only to the Uinta Basin samples, but also to the Tar Sand Triangle
sample. Saturate to olefin ratios are highly variable and results have not
been explained on the basis of bitumen structure. Certain similarities
appear to exist between P.R. Spring and Tar Sand Triangle on the one hand
108
Table Dl
Coking Product Yields from Various Bitumens
Product ATH TST AR PRS
Table D2
WIL
Gases (Cc and lighter,
by difference)
Liquid Condensate
(C6 - 535°C)
Coke
7.52
76.52
15.96
5.31
72.82
21.87
4.80
82.85
12.35
7.41
76.05
16.54
6.03
77.04
16.93
ATH (Athabasca); TST (Tar Sand T r i ang le ) ; AR (Asphalt Ridge);
PRS (P. R. Spr ing) ; and WIL (Wilmington).
Analysis of C,-C5 Gas from Coking of Various Bitumens
and the other three bitumens (ATH, AR and WIL) on the other.
Liquid condensate yields exhibit a good correlation with hydrogen to
carbon atomic ratios of the virgin bitumen. A secondary inverse correlation
appears to exist with molecular weight; for a given H/C ratio, the higher
the molecular weight, the lower the yield. An insufficient number of data
points are available to determine a reliable correlation by regression of the
data. The Asphalt Ridge sample, which is a likely candidate for early
commercial development, gives a high yield of almost 83 percent.
Elemental analysis and physical properties of condensates derived from
various bitumens are given in Table D3. Carbon/hydrogen ratios exhibit
moderate variation with values grossly following that in the original bitumen.
The Tar Sand Triangle condensate exhibited a substantial increase in hydrogen
content over the native bitumen while other samples exhibited less marked
enrichment. Generally, the bulk properties are remarkably similar for all
samples.
Sulfur and nitrogen contents reflect the relative concentrations of
these elements present in the native bitumen, although both are reduced in
the products. Oxygen has been essentially removed, presumably as water and
the oxides of carbon and sulfur. Small amounts of water were produced at
reaction temperatures of 350 to 400 C indicating that the same bonds
associated with oxygen functionalities are quite thermally labile. Molecular
weights are nearly identical for the tar sand bitumens while API gravity and
refractive index show minor variations.
The boiling point distribution of liquid products was determined from
simulated distillation by gas-liquid chromatography. Results are given
in Table D4. In this analysis significant differences are obvious. This is
rather surprising in light of the similarities in the other properties. The
110
Table D3
Liquid Condensate Properties from Various Bitumens
Carbon (wt.
Hydrogen
Nitrogen
Sulfur
Oxygen
C/H atomic rat io
pet . )
Average molecular weight (VPO -benzene)
ATH
84.7
11.3
.19
3.75
0-trace
.631
279
TST
85.2
11.6
.16
2.68
0-trace
.616
280
Bitumen
AR
87.1
12.0
.58
.32
0-trace
.615
282
PRS
86.5
12.1
.57
.29
0-trace
.598
280
WIL
86.5
11.7
.43
1.43
0-trace
.618
313
Specific gravity (20/20) .923 .910 .898 .895 .920
API gravity 21.9 24.0 25.8 26.5 22.3
Refractive index -n 20 1.5191 1.5130 1.5106 1.5053 1.5174
Heating value (Btu/ lb, calculated)
18,630 18,800 19,080 19,150 19,000
i l l
Wilmington condensate is noticeably heavy which is consistent with its
higher molecular weight and slightly higher degree of ring condensation.
The coke produced in this study was characterized by elemental analysis
and heating value and results are given in Table D5. All of the cokes had a
shiny appearance with infrequent pores characteristic of a relatively high
density material. Results show that Uinta Basin cokes are extremely low in
sulfur whereas coke from Athabasca and Tar Sand Triangle possesses quite high
concentrations of sulfur. This factor is one of the major problems in
utilization of Athabasca coke in the commercial operations existing in Canada
today. The low sum of the elemental composition is thought to be caused by
concentration of minerals in the coke. Although care was taken to dry the
samples prior to analysis, these chars showed a strong propensity to adsorb
water and small amounts of adsorbed water present would have an appreciable
effect on the elemental balance. Calculating heating values do not vary sign
ificantly from one source to another and are approximately 15,000 Btu/lb.
By relating the results from Utah samples to those derived for Athabasca
and a representative petroleum residue, comparison is established with
commercially processed samples. Results for the Athabasca sample compare
favorably with literature results (D8) with gas productions and compositions
being similar to delayed coking operations while liquid yields are higher
(77 vs. 70) and coke yields are lower (16 vs. 22) in the present study.
These results are explained by the longer residence time under higher hydro
carbon partial pressures during the commercial operation. Such conditions en
hance the second order condensation reactions relative to first order
cracking reactions.
Attempts were made to estimate the kinetics of coking reactions by
monitoring the gas rate of production. It was known from the destructive
112
Table D4
Simulated Distillation Yields
of Pyrolysis Condensates from Various Bitumens
ATH TST AR PRS WIL
Gasoline Cg -200°C
Kerosene 200 - 275°C
Gas oil 275 - 325°C
Heavy gas oil 325 - 450°C
Vacuum gas oil 450 - 538°C
Subtotal
Residue
7.5
12.9
13.7
48.0
17.9
100.0
0 - 4 %
(W< siqht Percent of D is t i l l ab les )
7.2
11.5
13.0
51.4
16.9
100.0
0 - 3 %
11.9 10.4
19.9 14.7
16.9 12.8
.34.0 46.6
17.3 15.6
100.0 100.0
0 - 5 % 0 - 2 %
9.1
12.2
8.5
40.5
29.7
100.0
5 - 10%
Table D5
Analysis of Coke from Various Bitumens
ATH TST, AR, PRS WIL
Carbon (wei
Hydrogen
Nitrogen
Sulfur
Heating val
ght
ue
percent)
(Btu/lb)
88.6
2.5
1.8
6.0
14,960
87.7
2.8
1.5
6.2
14,950
87.9
3.0
2.9
0.4
14,860
87.7
2.6
2.9
0.5
14,720
89.8
2.9
3.0
1.5
15,165
113
distillation studies that gas production did not correspond with condensate
production in non-isothermal runs. However, for a series of isothermal runs,
comparison of the rates of gas production at various temperatures approximately
describes the overall temperature dependence of the rate of production.
(Variation in liquid/gas ratios at different temperatures was small compared
to differences in rates of production.) In this analysis, initial heat-up
period (10 minutes for the 460 C run) was disregarded; and kinetics were based
on t0 = t(Tf. 1 - 10 C). This approximation is not thought to produce
significant error.
Assuming that the rate of gas production is representative of the overall
rate of reaction, expression of this rate can be made on the appearance of
gas. Letting m = mass of gas generated/gram charge, then -r~ = kfa^-m) .
Application of the data to this equation gave zero-order dependence for low
temperature and fractional order dependence at higher temperatures. The
results are summarized in Table D6. An Arrhenius plot (Figure D2) gave an
activation energy of 34.6 Kcal/mole and a pre-exponential factor of 2.3 x
10 g/min-g feed. For the regime covering the first 75% of the reaction, the
integration expression, (n = o),
m = [2.308 x 107 g/min-g feed) e " 3 4 6 0 8 / R T ] ( t m i n > )
was found to satisfactorily describe the observed results. The activation
energy of 34,608 cal/mole has little significance to the reaction mechanism
because of the multiplicity of molecular reactions which contribute to this
value. A similar calculation based on volume rate of production gave an
activation energy of 29,350 cal/mole. These values are consistent with
values reported for gas evolution from coal pyrolysis by Campbell (D9).
Attempts were not made to define quantitatively the kinetics of the
final portion of the reaction where concentration of unreacted bitumen becomes
114
Table D6
Order of Reaction (Power Function) for Coking
of Asphalt Ridge Bitumen at Various Temperatures.
Temperature, °C Reaction Order (n)
380
402
415
430
454
460
115
0.0
0.1
0.2
0.3
0.5
0.3
-|-«I7.5 Kcal/mole
A » 2.3 X I0 7g/min-g feed
1 1 1.3 1.4 1.5
Y X IO"*3(l/°K)
Figure D2. Arrhenius Plot for Coking of Asphalt Ridge Bitumen.
1.6
a significant factor. Diffusional resistances probably play a significant
role as solid coke is formed.
Conclusions which can be drawn from this set of experiments are that
(1) conversion is primarily a function of hydrogen content, (2) condensable
product character, although considerably uniform from one bitumen feed to
another, tends to correlate with bitumen character, (3) products derived from
tar sand bitumens compare favorably with those derived froma representative
petroleum residue, (4) coke generated from Uinta Basin, Utah samples is
extremely low in sulfur, (5) hydrogen demands in hydrotreating processes will
probably be lower for Utah samples than for Athabasca samples, and (6) compared
to the Athabasca bitumen products, the Uinta Basin products are low in sulfur,
high in hydrogen, and are produced in higher yields. The comparative features
of Uinta Basin bitumen to Athabasca bitumen strongly suggest that for coking
a more valuable raw material occurs in the Uinta Basin.
Catalytic Cracking
Catalytic cracking is the principal means of converting gas oil to high
octane gasolines and is, perhaps, the most important of all refining operations
in meeting demand for motor gasolines. In catalytic cracking, catalyst is
regenerated on a continuous basis and one of the major limitations to equipment
size and process costs is the amount of coke which must be burned (D10). High
carbon residue feedstocks such as tar sand bitumens may be expected to impose
economic penalties on direct catalytic cracking of bitumen. It is also known
that basic nitrogen which is a significant constituent of Uinta Basin bitumens (Dl),
adsorbs strongly to the surface of cracking catalysts and temporarily poisons
acid sites (Dll). However, potential improvements in primary process conversion
and/or product quality compared to coking may prove to be adequate to offset
the expected increase in costs compared to gas oil cracking.
117
The predominantly naphthenic structure of Uinta Basin bitumen suggests
that these materials may be highly responsive to direct catalytic cracking.
It has been known for many years that naphthenic gas oils give high yields
of high octane gasoline (D12). The highly naphthenic character of Uinta Basin
tar sand bitumens is exemplified by the group type analysis of the P.R. Spring
saturated hydrocarbons (D2). This analysis is given in Table D7 and shows that
only 7.1% of the saturates are paraffins or isoparaffins. Over 60 percent of
the saturated hydrocarbons consists of perhydronaphthalenes, and perhydro-
phenanthrenes and -anthracenes. The saturated hydrocarbons from this particular
sample represented 26 weight percent of the total bitumen and exhibited an
average molecular weight of 325. The other 74 percent of the bitumen contains
either aromatic or polar groups, or both, and the generally naphthenic character
of the saturated hydrocarbons is though to prevail in the saturated substituents
on the aromatics as well.
Virgin tar sand bitumen contains large quantities of material which are
not volatile at cracking conditions. This necessitates contact of the catalyst
with liquid phase bitumen. Because normal temperatures for catalytic cracking
are well into the region where thermal cracking occurs, competitive thermal
reactions are expected. To the extent that inefficient contact occurs between
bitumen and catalyst as a result of mass transport resistances, one can expect
thermal reactions to be relatively more important. In the case of tar sand
bitumen it is conceivable that the largest molecules could be completely ex
cluded from the pores of catalysts having average pore diameters of 30 to 40 A
or less. The effective exclusion of large molecules may be particularly
important when the feedstock exhibits an appreciable coke forming propensity
due to deposition of coke on the exterior of the catalyst and around the pore
mouths.
118
Table D7
Group-Type Analysis of P. R. Spring Saturated Hydrocarbons (D2)
•Number of Ri 0
1
2
3
4
5
6
Monoaromati
ngs
cs
Wt. % of Saturates 7.1
12.3
29.4
31.5
14.1
4.4
1.3
0
119
From previous experience with liquid chromatographic separations of bitumen
samples, it is felt that a catalyst possessing a pore diameter in excess of 40 A
is required to minimize exclusion from the interior of the catalyst particles.
Results of liquid chromatographic separations of 535 -675 C boiling petroleum
hydrocarbons on silica gel indicate that aromatic hydrocarbons of this boiling
range are excluded from a 22 A pore diameter gel but are readily adsorbed on a
60 A pore diameter gel (D13).
Two catalysts were selected for application, one an amorphous silica-alumina
possessing 82 A average pore diameter (D14), the other a molecular sieve catalyst
possessing a 200 A average pore diameter for the support (D15). These catalysts
are thought to be representative of commercial acid cracking catalysts capable
of handling high molecular weight feeds. Catalysts were used as equilibrium
catalysts, that is, catalysts which were recovered from commercial catalytic
cracking units. For certain experiments the catalyst was ground and sieved
to include about equal quantities of 25-35, 35-200, -200 mesh catalyst. The
purpose of using finely ground catalyst was to assess whether or not adsorp
tion or coke deposition on the external surface of the catalyst particle was
inhibiting the activity of the interior of the catalyst particle. Grinding
increases the relative importance of exterior surface area over interior
surface area and effects of intra-particle diffusion could be observed using
catalysts of varying particle size.
The virgin bitumen is partially vaporized at normal cracking temperatures
and obtaining reproducible catalyst-bitumen contact presented a significant
problem. Two modes of operation were utilized. In the semi-batch mode,
patterned after the Cat-A test (D16), the catalyst and feed were separately
preheated before contact was made in a downflow fixed bed reactor. Preheat
of the feed was held to 350 C to minimize thermal cracking reactions prior to
120
contact. This reactor had the advantage of contact at reaction temperature,
but had the disadvantage that meaningful contact and catalyst to oil ratios were
not achieved. The batch mode had the advantage of providing ultimate contact
with the bitumen but had the disadvantage of longer heatup times of approximately
10 minutes.
Gravimetric results of the various experiments are given in Table D8 along
with representative results from coking of the same feed [Bt(10) and Bt(ll)].
Temperatures of 412-415 C were chosen because this was the temperature of
maximum increase in production from thermal cracking (Figure D3) and catalytic
vs. thermal effects would be more easily discernable at this temperature. The
temperature of 460 C was chosen to correspond with the temperature studied
most for coking.
Results of Bt(2) and Bt(8) show good selectivity of the molecular sieve
catalyst toward the production of liquids. Results of these two runs illustrate
general agreement between the two reactor configurations. Results of these two
runs also indicate significant catalytic activity as illustrated with the
higher liquid production when compared to the purely thermal cracking results
of Bt(10).
Results of mild temperature catalytic cracking on the amorphous silica-
alumina catalyst are given in runs Bt(l) and Bt(9). Comparison of results of
Bt(l) and Bt(10) show enhanced cracking activity using the catalyst when the
pelleted form of the catalyst is used, but reduced activity when the ground
catalyst is used [Bt(9)]. Semiquantitative data obtained from using the
ground catalyst in the SB mode also showed extremely high coke/residue yields.
These results strongly suggested that the silica-alumina catalyst was sus
ceptible to pore mouth plugging and/or catalyst deactivation by adsorption of
virgin bitumen. This rendered the catalyst inactive and in the course laid
121
Table D8
Results of Catalytic Cracking of Asphalt Ridge Bitumen
N3
Designation
Bt(2) Bt(8) Bt(l) Bt(9) Bt(10)
Bt(6) VB(7)
Bt(3) Bt(4) Bt(12) Bt(13)
M(5) HP(14) Bt(ll)
Catalyst
MS(f) MS(f) S/A
S/A(f) T
MS MS
S/A S/A S/A S/A
S/A S/A T
T°C
412 414 412 412 415
460 460
470 460 460 460
460 426 460
Mode
SB B SB B B
B B
SB B B B
B B B
Cat/Oil
1.8 1.0 1.3 1.0 -
3.0 3.0
1.3 2.0 5.0 10.0
2.0 2.3 —
Gas
Weight
1 2 6 7 2
7 4
11 10 7 9
10 2 4
Liquid
Percent
79 76 67 50 58
80 83
76 74 73 67
78 80 81
Residue (Coke)
Yields
20 22 25 43 40
13 13
13 16 20 24
12 18 15
Stpl Liquids
29.5 29.3 27.9 31.7 26.9
27.1 28.8
25.1 30.8 35.6 41.5
32.0 33.9 27.1
Symbol Designation: Bt (v i rg in bitumen); M (pentane soluble maltenes from virgin bitumen); VB (visbroken bitumen, 425°C, 150 psig, Ml min.) ; HP (hydropyrolized bitumen, 525°C, 1500 psia H?, 18 s e c ) ; S/A (silica-alumina cata lyst ) ; MS (molecular sieve cata lys t ) ; SB (semi-batch Cat A mode); B (batch mode); f (powdered catalyst) ; T (thermal coking).
£ZT
WEIGHT PERCENT YIELD OF LIQUID AND RESIDUE
to c -s fD
CD
Q -C/>
- t l -s o
o
fD -s
-a << -s o
<< c/)
>
cu
Q .
c 3 rD 3
T 1 1 ' 1 ro .& CD
WEIGHT PERCENT YIELD OF GAS
down carbonaceous material which cracked to lighter products in the absence of
a catalyst. Thus, it is concluded from the results obtained at moderate
temperatures that an 82 A pore diameter is inadequately small for handling
virgin molecules but that 200 A appears to accept these molecules. This
exclusion may be due to either intrinsic factors or adsorption of nonvolatile
species on the exterior surface of the catalyst, or both. The higher yields
exhibited in Bt(l) compared to Bt(10) resulted partly because mild thermal
cracking reduced the molecular weight prior to the catalytic cracking and
partly because the presence of catalyst enhanced the overall cracking activity.
A series of experiments were run at 460 C. This temperature is closer to
the temperature used in commercial units operating at low severity. It was
felt that low severity testing would minimize the thermal effects and would
maximize the differences experienced by changes in variables other than temp
erature. Results of run Bt(6), Bt(3) and Bt(4) show yields similar to or better
than coking, Bt(ll). Analysis of the gases produced in Bt(6) revealed that
significant thermal cracking took place simultaneously with cat cracking which
helps explain the relatively high yield of gases. When a 500 average molecular
weight visbroken bitumen was charged to the molecular sieve catalyst under
identical conditions, VB(7), the result was a greater catalytic effect. This
effect is apparent in the selectivity toward production liquid products and no
notable degradation of liquid product quality when the API gravity of VB(7)
liquid is compared to Bt(6) liquid.
Runs Bt(4), Bt(12) and Bt(13) show the effect of cat to oil ratios. Gas
yields are variable and probably reflect the effects of competing reactions;
thermal reactions are almost certainly relatively more influential at low
catalyst to oil ratios than at high catalyst to oil ratios because of the greater
opportunity for catalyst reaction to occur at higher ratios. Total liquid and
124
gas yields decrease with increasing catalyst to oil ratio but API gravity of
liquids increases significantly with increasing ratio. These results are
consistent with published literature on the effect of catalyst to oil ratios
on gas oil cracking which show that total yields of gases and liquids decrease,
conversion to gasoline increases, and octane number of gasoline increases as
catalyst to oil ratio increases (D12).
The effect of asphaltenes on the results are seen in the comparison of
Bt(4) and M(5). The feedstock for M(5) was the n-pentane deasphaltened bitumen
(maltenes) and represented 90% of the total bitumen. Only 4% greater yields
were experienced in M(5). These results are explained by observing that the
asphaltenes from tar sand bitumen consist of the high molecular weight species
but that they are also quite rich in hydrogen (C/H = 0.83). The molecules which
comprise the asphaltenes actually contribute significantly to the liquid product,
albeit the heavy products as seen in the API gravity results. Conversely, not
all coke precursors are removed by deasphaltening as shown by the 12% coke
yield for run M(5). Significant quantities of aromatics are soluble in the
precipitating solvent n-pentane. (See, for example, a comparison of compound
type analysis for whole bitumen and a deasphaltened bitumen (D17)). These
results illustrate that a one-to-one precursor-product mechanism does not
exist between molecules comprising the asphaltene fraction and those responsible
for coke formation. Correlations between asphaltene content and the propensity
for coke formation are largely fortuitous.
Run HP(14) utilized as a feedstock a more drastically upgraded bitumen. The
feedstock in this case was produced by hydropyrolysis in a tubular reactor. The
feedstock represented 73% of the total bitumen (27% gases were made in hydro
pyrolysis) and possessed an elemental composition quite similar to the original
bitumen. However, the average molecular weight of the feed was 321, or less
125
than half that of the original bitumen. Results of HP(14) show excellent
responsiveness and selectivity at very mild severity.
The gases and liquids from selected runs were subjected to detailed comp
ositional analysis. Table D9 gives the C to C, gas analysis and compares the
results from cat cracking with thermal cracking. With run Bt(6) only moderate
catalytic activity was experienced as evidenced by the moderate increase in
C~ and C, content compared to Bt(ll). Somewhat better catalytic activity was
experienced with Bt(4) as evidenced by the substantial amounts of C.'s produced
and the high i-butane/n-butane ratio. The relative production of C, and C,
hydrocarbons compared to C. and the ratio of isoparaffins to n-paraffins are
important indicators of catalytic activity (D12). These structural indicies
result because catalytic cracking proceeds largely through carbonium ion
chemistry while thermal cracking proceeds through free radical chemistry (D18).
Results of Bt(6) are almost certainly less than the optimum that can be achieved
by this catalyst but illustrate the difficulty which may be encountered due to
competing reactions when cracking these high molecular weight bitumens.
Approximately 22% of the original bitumen was converted to 0^-200 C
gasoline in run Bt(4). This gasoline was isolated by spinning band distillation
and analyzed for its composition and octane rating. Results are summarized in
Table D10. Results show somewhat low octane ratings which are attributable to
the presence of thermal products as seen with the presence of n-paraffins and
low octane isoparaffins. Over 12% of the product was unidentified and was
assigned an octane number of zero. Results of this analysis reveal that
optimum conditions have not been achieved and that further efforts must be
made to produce a high quality gasoline. Attempts should be made to contact
the feed and catalyst at preheated conditions to more closely simulate con
ditions in a modern fluid catalytic cracker (FCC).
126
Table D9
Methane
C-| to C^ Gas Analysis from Catalytic
and Thermal Cracking of Asphalt Ridge Bitumen
Molecular Sieve^ Bt(6)*
32.3
Silica-Alumina Bt(4)
Weight Percent
18.9
Thermal Bt(ll)
41.0
Ethane
Ethyl ens
Propane
Propylene
n-butane
i-butane
Butylenes
16.1
5.0
15.4
10.2
5.4
4.0
11.6
10.3
4.2
12.2
15.4
3.3
18.0
17.7
16.5
15.1
10.0
11.6
1.3
1.4
3.1
See Table 08 for explanation of symbols used.
Table D10
Analysis of Gasoline from Run Bt(4)
Volume % Weight R.O. **
M.O. **
n-paraf f ins
i - p a r a f f i n s
Naphthenes
Olef ins
Aromatics
Unc lass i f ied
C-5 to 200°F
200 - 392°F
Calculated Blended Values
5.0
20.7
10.7
11.1
40.1
12.4
16.8
83.2
100
4.3
18.2
10.3
9.8
45.1
12.3
-4 .9
45.2
79.1
86.4
99.4
0
85
70
73
-4.5
48.5
74.4
73.5
89.6
0
74
60
62
See Table D8 for description of run.
R.O. and M.O. are research octane number and motor octane number, respectively.
127
Conclusions derived from the results of catalytic cracking are: (a)
bitumen derived from the Asphalt Ridge, Utah deposit was found to be res
ponsive to catalytic cracking, (b) catalytic cracking provided higher
quality products at similar yields when compared to coking, (c) the feasability
of using an acid catalyst for more selectively cracking of virgin bitumen to
valuable products has been demonstrated, (d) the extremely high molecular
weight of virgin bitumen resulted in an inhibited rate of catalytic cracking
and allowed competing thermal reactions to adversely influence the product
quality, (e) octane numbers for the gasoline produced were lower than desirable
as a result of competing thermal reactions, (f) the amount of coke produced
was substantially higher than presently experienced commercially for gas-oil
cracking, (g) higher aromatics partially responsible for producing coke were
not quantitatively removed by prior deasphaltening and more severe deasphal-
tening conditions resulted in coprecipitation of substantial amounts of reactive
species as well, (h) mildly upgraded products with a reduced molecular weight
were found to be more reactive to catalytic cracking than the original bitumen.
Good reactivity to catalytic cracking was predicted from considerations of the
high content of alkyl and naphthenic carbon present. Further study of more
optimum reactor configurations and process conditions is indicated. Whether
the feed is a virgin bitumen or, more likely, an upgraded bitumen product,
catalytic cracking will probably play an important role in commercial development
of Uinta Basin tar sands.
Hydropyrolysis
The term "hydropyrolysis" is used to describe non-catalytic pyrolysis in
the presence of hydrogen (D19). Such reactions can be carried out at moderate
temperatures, <480 C, and long residence times, 1 hr., or at higher temperatures,
128
>650 C, and shorter residence times, <10 seconds. The mechanism of the inter
action of hydrogen in this system is not well known. It is generally agreed
that for thermal dissociation of a hydrogen molecule to occur, temperatures in
excess of 600 C are required. However, it is clear from the work at the
Canadian Centre for Mines, Energy and Technology (D20, D21) that hydrogen is
reactive at temperatures as low as about 430 C. The mechanism of hydropyrolysis
of model compounds has been studied by Ramakrishnan (D19) and his results con
firm the activity of hydrogen at moderate temperatures.
The Asphalt Ridge bitumen was subjected to hydropyrolysis in order to
ascertain yields and product qualities obtainable. An important objective was
to determine the effect of temperature, pressure, and residence time variables
on yield and product structure. Results of these runs are given in Table Dll
and show that conditions were found which prevented or severely inhibited coke
formations. Conditions of 650 C and 200 psig, HP(4), were found to be too
severe and coking occurred. Coking was also observed at conditions of 525 C
and 1200 psig in run HP(7). Apparently, the minimum pressure for operation at
525 C lies between 1200 and 1500 ps ig, according to data given in Table Dll.
Conditions stated in Table Dll were not precisely controlled and inter
pretation of the data must recognize variations possible during any given run.
The temperature control was good to approximately +6 C of the stated temperature.
Pressure control was accurate to +100 psig. Residence time, as measured by the
reactor volume divided by the volumetric flow rate at reaction conditions, was
highly variable and fluctuated as much as +5 seconds during the course of a run.
Attempts were made to hold the residence time constant for runs HP(5), HP(6) and
HP(7), but results show these attempts were unsuccessful. Values reported are
the mean values calculated at the end of the run by dividing the time of the
run by the total number of reactor volumes passed. Correction for temperature
and pressure was made assuming ideal gas behavior.
129
Table Dll
Yields and Process Conditions
for Hydropyrolysis of Asphalt Ridge Bitumen
Reaction Conditions Yields
HP(1)
HP(2)
HP(3)
HP(4)
HP(5)
HP(6)
HP(7)
Temp. °C
500
525
575
650
525
525
525
Pressure (psig)
1500
1500
1500
200
1825
1500
1200
Average Residence Time
(seconds)
18
18
18
18
10
13
15
Gas
17
27
NA
28
23
Weiqht Percent
Liquid
83
73 NA
COKE FORMED
72
77
COKE FORMED
Coke
NIL
NIL
NIL
NIL
NIL
130
Results in Table Dll show that gas make increases both as a function of
temperature and pressure. The difference in results between HP(2) and HP(6) may
be attributable to a difference in residence time. The difference may, however,
be within experimental error and more definitive work must be done to quantify
the effects shown.
Elemental analysis and physical properties of liquid products are given in
Table D12. Also shown are the properties for the feed material. The major
chemical change accomplished by hydropyrolysis is shown in the physical property
data. The API gravity has been doubled and the apparent average molecular weight
has been cut in half by the treatment. Refractive indicies are slightly higher
than those determined from coker distillates. Refractive index increases as
aromaticity increases (D22). The most aromatic molecules present in the original
bitumen are included with the hydropyrolysis products, whereas many aromatics are
removed in the form of coke during the coking operation. Thus, the low refractive
index of 1.51 to 1.52 for the hydropyrolysis products is an important indication
that substantial aromatization did not occur.
Hydrogen consumption was calculated by first determining;a material balance
on carbon to assure that all of the carbon fed to system was accounted for. This
was accomplished by elemental analysis of the liquids and by gas chromatographic
analysis of the gases. The hydrogen content of the total gaseous and liquid
products was determined and compared to the hydrogen contained in the feed
bitumen. The assumptions and method of calculations are such that the values
calculated do not underestimate the amount of hydrogen added. Results shown
in Table D12 reveal that 1.6 and 2.6 weight percent hydrogen was added for runs
HP(1) and HP(2), respectively. Comparison of results in Table D12 with the yield
of gases shown in Table Dll reveals that hydrogen consumption is directly re
lated to the non-condensable gas production. This result is consistent with the
131
Table D12
Liquid Product Characteristics
from Hydropyrolysis of Asphalt Ridge Bitumen
Feed
86.2
11.3
1.1
0.4
0.9
.640
12.7
.981
HP(1)
86.7
11.6
0.8
0.3
0.3
.627
22.1
.921
1.52
Ru
HP(2)
86.8
11.4
0.8
0.3
0.3
.639
25.2
.903
1.52
* n
HP(5)
86.8
11.4
0.75
0.35
0.4
.637
24.2
.910
1.51
HP(6)
87.0
11.5
0.74
0.36
0.31
.633
24.3
.910
1.51
Carbon (weight percent)
Hydrogen (weight percent)
Nitrogen (weight percent)
Sulfur (weight percent)
Oxygen (weight percent)
C/H Ratio
API Gravity
Specific Gravity
Refractive Index
Average Molecular Weight 713 336 321 294 289
Weight Percent H? Added
to Total Products 1.6 2.6 **
SCF H2/bbl Feed 1200 2000
**
A representative sample of the gases for HP(5) and HP(6) was not obtained which prohibited calculation of an accurate carbon and hydrogen balance for these runs.
it
See Table Dll for run conditions.
132
observation that the C/H ratio of the liquid products is virtually unchanged
over the narrow range of reaction variables studied.
The amount of n-pentane insoluble material was determined on the HP(1)
products. Results revealed that 3.5% of the liquids are insoluble in pentane
(compared to 11.8% for the original bitumen) and that the resulting asphaltenes
are quite aromatic (C/H = 1.03). The asphaltenes contain an extremely large
amount of nitrogen (5%) and moderate amounts of sulfur (0.6%). Over 20% of
the nitrogen, presumably the most refractory material, are removed from the
hydropyrolysis liquids in this fashion and this result suggests the possibility
of solvent deasphaltening prior to secondary processing of hydropyrolysis
products.
Conclusions which are based on the hydropyrolysis work are: (a) virgin
bitumen can be converted in 100% yields to gaseous and liquid products by
hydropyrolysis, (b) the major chemical effect of hydropyrolysis is to reduce the
average molecular weight; elemental compositions including heteroatom contents
are not significantly changed, compared to the feed material, (c) moderate
quantities of hydrogen are required for production of low molecular weight
liquids and significant quantities of valuable gases, and (d) a wide range of
products can be produced through proper control of the important variables.
Results of simulated distillation, combined with the gravimetric results
of yields of liquid products can be used to calculate a conversion index allowing
a standard comparison between the various processes studied. Conversion is
defined as the percentage of the original 60.2% of the bitumen which boiled
above 538 C which was converted to material distillable below 538 C. Gases
were assumed to be 100% distillable and coke/residues were assumed to be 100%
non-distillable. Results of this analysis are given in Table D13.
133
Table D13
Comparison of Yield and Conversion
Results for Primary Processing of Asphalt Ridge Bitumen
Yield
Process
Visbreaking (VB)
Coking TC(80)
Catalytic Cracking
Coking TC(0)
Hydropyrolysis (HP)
(CC)
Gases
1
7
10
4
27
Liquids
99
70
74
83
73
Total Gases + Liquids
Weight Percent
100
77
84
87
100
% Liquids Distillable
67
100
99
97
85
Conversion
46
62
72
74
82
Conversion is defined as the percentage of >538°C boi l ing material converted to <538°C bo i l ing material
Several factors are apparent from the results given in Table D13. The
highest conversion is achieved by hydropyrolysis which yields a distillable
liquid product superior in terms of mean boiling point to that obtained with
atmospheric pressure coking. Further, 13% more yields are experienced and the
18% unconverted material (only 11% of the virgin bitumen) remains a tractable
material. These improved products resulted from the addition of 2.6% hydrogen
which was not added in other processes. Coking at atmospheric pressure gave
the second highest conversion but liquid products were distributed heavily
toward the heavy gas oil region.
The selectivity induced by catalytic cracking and alluded to above can be
seen in the significantly higher conversion and yields compared to the coking
at 80 psig. Comparison is properly made with TC(80) because the product
distribution for (CC) was similar to the product distribution for TC(80).
Product quality for (CC) was superior also to coker distillate because of the
presence of high octane gasoline components.
Results for visbreaking reveal the 46% conversion was achieved without
generation of intractable material (coke). Visbreaking may serve as a primary
process as discussed before or possibly as a pretreatment for catalytic
processes which are susceptible to pore diffusion remains a possible option.
Catalytic cracking and catalytic hydrocracking (the latter process was not
examined in this study) are examples of processes which may benefit from
prior visbreaking.
Studies of the steam pyrolysis of Utah tar sand bitumens and products
have been initiated. The purpose of these studies is to determine the
potential of these oils for producing the basic chemicals: ethylene, propylene,
butadiene, benzene, toluene and xylenes. A furnace capable of high temperature
short residence time conditions together with the necessary auxiliary experiment
135
is being built and will be calibrated with conventional feedstocks. Feed
stocks for use in these studies are now being prepared. Results will be
presented in future reports as it is obtained.
Conclusions
Results of this research have shown that bitumen conversion is quite
sensitive to both process conditions and processing approach. The capital
intensive nature of tar sand recovery suggests the efficiencies of the
conversion process will have major economic importance. From the standpoint
of process efficiency, hydropyrolysis appears to be particularly attractive.
Work has progressed to a point where comparative economics based on
process development unit data should be initiated.
Further work on hydropyrolysis using a tubular flow reactor which is
more useful as a process development unit than previously used equipment
is now being carried out. Results from this program and the steam pyrolysis
of bitumens and products to produce basic chemicals will be covered in future
reports.
136
References
Dl. J.W. Bunger, K.P. Thomas and S.M. Dorrence, "Analysis of Compound Types and Properties of Utah and Athabasca Tar Sand Bitumens", Fuel, Vol. 58 (3), (1979).
D2. J.W. Bunger, "Characterization of a Utah Tar Sand Bitumen", Chapter 10 of Shale Oil, Tar Sands and Related Fuel Sources, T.F. Yen, ed., Advances in Chemistry Series 151, ACS, 121-135 (1976).
D4. Petroleum Processing Handbook, W.F. Bland and R.L. Davidson, eds., McGraw-Hill, New York, (1967).
D5. H. Beuther, R.G. Goldthwait, and W.C. Offult, "Thermal Visbreaking of Heavy Residues", The Oil and Gas Journal, 57_ (46), 151-157 (1959).
D6. J.A. Dixon and W. Webb, "Viscosity-Pressure Relationships for High-Molecular-Weight Fluids", Proceedings, Division of Refining, Amer. Pet. Inst. 42 (3), 146-151 (1962).
D7. Chemical Engineer's Handbook, 5th Edition, R.H. Perry and C.H. Chilton, eds., McGraw-Hill Book Co., New York, 9-11, (1973).
D8. F.W. Camp, "The Tar Sand of Alberta, Canada", 2nd ed., Cameron Engineers, Denver, (1974) 77 pp.
D9. John H. Campbell, "Pyrolysis of Sub-bituminous Coal in Relation to In-Situ Coal Gasification", Fuel, 57 (4), 217 (1978).
D10. A.G. Oblad, personal communication, 1976.
Dll. A.G. Oblad, T.H. Milliken, Jr., and G.A. Mills, "Chemical Characteristics and Structure of Cracking Catalysts", Advances in Catalysis, 2_, 199-247 (1951).
D12. A.G. Oblad, T.H. Milliken and G.A. Mills, "The Effects of Variables in Catalytic Cracking", Chpt. 28 in The Chemistry of Petroleum Hydrocarbons, B.T. Brooks, S.S. Kurtz, Jr., C.E. Boord and L. Schmerling, eds., Reinhold Publishing Co., New York, 2, 165-188, (1954).
D13. J.W. Bunger, unpublished results obtained while the author was employed by the Laramie Energy Research Center, U.S. Bureau of Mines, Laramie, Wyoming (1973).
D14. J.J. Rim, "Chemical Characterization of Molecular Sieve Catalysts", Ph.D Thesis, Dept. of Fuels Eng., Univ. of Utah, 113, (1974).
D15. S.C. Eastwood, C.J. Plank and P.B. Weisz, "New Developments in Catalytic Cracking", Proceedings, 8th World Petroleum Congress, 4L, 245-254, (1971).
137
D16. J. Alexander and E.G. Shrimp, "Laboratory Method for Determining the Activity of Cracking Catalysts", National Petroleum News, _36 (31), R537-R538 (1944).
D17. J.W. Bunger, "Techniques of Analysis of Tar Sand Bitumens", Preprints, Div. Petr. Chem., ACS, 22 (2), 716-726 (1977).
D18. B.S. Greensfelder, "Theory of Catalytic Cracking", Chpt. 27 in The Chemistry of Petroleum Hydrocarbons, B.T. Brooks, S.S. Kurtz, Jr., C.E. Boord and L. Schmerling, eds., Reinhold Publishing Co., New York, 2, 137-164 (1954).
D19. R. Ramakrishnan, "Hydropyrolysis of Coal Derived Liquids and Related Model Compounds", Ph.D. dissertation, Dept. of Fuels Engineering, Univ. of Utah (1978) 132 pp.
D20. W.H. Merrill, D.H. Quinsey, M.P. Pleet and J.M. Denis, "Description of a High Pressure Combined Liquid and Vapour Phase Hydrogenation Pilot Plant", Report FD67/139-PE, Fuels Research Centre, Department of Energy, Mines and Resources, Ottawa, (1967).
D21. W.H. Merrill, R.B. Logie and J.M. Denis, "A Pilot Scale Investigation of Thermal Hydrocracking of Athabasca Bitumen", Research Report F-281, Fuels Research Centre, Department of Energy, Mines and Resources, Mines Branch, Ottawa, (1973) 25 pp.
D22. K. Van Ness and H.A. Van Weston, Aspects of the Constitution of Mineral Oils, Elsevier Publishing Co., New York (1951).
138
BIBLIOGRAPHY
University of Utah Tar Sands Research and Development Program
Bunger, J. W., S. Mori and A. G. Oblad, "Processing of Tar Sand Bitumens. Part I. Thermal Cracking of Utah and Athabasca Tar Sand Bitume.is",. Preprints, Division of Fuel Chemistry, Amer. Chem. Soc, 21, 6, 147 (1976).
Oblad, A. G.» J. D. Seader, J. D. Miller and J. W. Bunger, "Recovery of Bitumen from Oil-Impregnated Sandstone Deposits of Utah", Oil Shale and Tar Sands, AIChE Symposium Series, Vol. 72, No. 155, p. 69 (1976).
Sepulveda, J. E., J. D. Miller and A. G. Oblad, "Hot Water Extraction of Bitumen from Utah Tar Sands", Fuels Division, ACS (1977). Symposium on Oil Shale, Tar Sands and Related Materials - Production and Utilization of Synfuels, Division of Fuel Chemistry, ACS, 2j_, No. 6, p. 110 (1976).
Sepulveda, J. E., "Extraction of Bitumen from Utah Tar Sands by a Hot Water Digestion-Flotation Technique", Garr Cutler Energy Award for scholarly paper which makes the most significant contribution in the general area of energy development, University of Utah, March, 1977.
Oblad, A. G., and J. W. Bunger, "Development of Uinta Basin Tar Sands", paper presented before the Uinta Basin Association of Governments, Vernal, Utah (May, 1977).
Bunger, J. W., "Development of Utah Tar Sands - A Status Report", Mines and Minerals Reporter, No. 5, 1-10 (October, 1977).
Bunger, J. W., "Techniques of Analysis of Tar Sand Bitumens", Preprints, Division of Petroleum Chemistry, Amer. Chem. Soc, 22, 2, 716-726
TT977T:
Bunger, J. W., D. E. Cogswell and A. G. Oblad, "Processing of Tar Sand Bitumens. Part II. Catalytic Cracking of Asphalt Ridge Bitumen", Preprints, Div. of Petr. Chem., Amer. Chem. Soc, 22, 3, 1008 (1977).
Sepulveda, J. E., "Hot Water Separation of Bitumen from Utah Tar Sands", M.S. Thesis in Metallurgy, University of Utah, 1977.
Weeks, J. K., Jr., "Fluidized-Bed Processing of Utah Tar Sands", M.S. Thesis in Chemical Engineering, University of Utah, 1977.
Bunger, J. W., "Tar Sand Resources and Technology", Proceedings, Alternate Resources and Technologies for Fuel Production, Symposium, Dept. Chem. Eng., Univ. of Pittsburgh, July 31, 1978.
139
Sepulveda, J.E. and J.D. Miller, "Separation of Bitumen from Utah Tar Sands by a Hot Water Digestion-Flotation Technique", Mining Engineering, 30 (9) 1311, (1978); (also published in Trans SME/AIME, Sept., 1978).
Bunger, J.W., D.E. Cogswell and A.G. Oblad, "Influence of Chemical Factors on Primary Processing of Utah Tar Sand Bitumen", Preprints, Div. of Fuel Chem., ACS 23 (4), 98-109 (1978).
Bunger, J.W., D.E. Cogswell and A.G. Oblad, "Thermal Processing of a Utah Tar Sand Bitumen", Proceedings of the Canada-Venezuela Oil Sands Symposium-77, Edmonton, Alberta, Canada, 1977; "The Oil Sands of Canada-Venezuela-1977, D.A. Redford and A.G. Winestock, editors, CIM Special Volume 17, pp. 178-182 (1978).
Bunger, J.W., D.E. Cogswell and A.G. Oblad, "Utah Tar Sands - Progress in Processing and Utilization Research", presented at the 4th Rocky Mountain Fuel Symposium, Salt Lake City, Utah, (Feb. 9-10, 1979).
Jayakar, K.M., K.C. Hanks and J.D. Seader, "Production of Synthetic Crude Oil from Utah Tar Sands by an Energy-Efficient Thermal Process", paper presented at the 4th Rocky Mountain Fuel Symposium, Salt Lake City, Utah, Feb. 9-10, 1979.
Bunger, J.W., "Processing Utah Tar Sand Bitumen", Ph.D. dissertation, Dept. of Fuels Engineering, Univ. of Utah, 217 pp., (1979).
Bunger, J.W., D.E. Cogswell and A.G. Oblad, "Catalytic Cracking of Asphalt Ridge Bitumen", Refining of Synthetic Crude Oils, ACS Advances in Chemistry Series, (1979), to be published.
Misra, M. and J.D. Miller, "The Effect of Feed Source in the Hot Water Processing of Utah Tar Sands", Trans. SME/AIME, to be published, (1979).
Misra, M., "Applicability of Hot Water Processing Technique to Low Grade Utah Tar Sands", March, 1979.
Venkatesan, V.N., F.V. Hanson and A.G. Oblad, "The Thermal Recovery of a Synthetic Crude from the Bituminous Sands of the Sunnyside (Utah) Deposit", Proceedings, First International Conference on the Future of Heavy Crude and Tar Sands, Edmonton, Alberta, Canada, June 4-12, 1979.
Seader, J.D. and Jayakar, K.M., "Process and Apparatus to Produce Synthetic Crude Oil from Tar Sands", U.S. Patent 4,160,720, July 10, 1979.
Jayakar, K.M., "The Thermal Recovery of Oil from Tar Sands", Ph.D. dissertation Dept. of Chem. Eng., Univ. of Utah, Aug. (1979), 236 pp.
Hanks, K.C, "Chemistry of Oil Production from Tar Sand", Master of Science thesis, Dept. of Chem. Eng., Univ. of Utah, Aug. (1979), 93 pp.
![TAR SANDS: WORTH THE ENERGY? AN ANALYSIS OF THE FUTURE OF UTAH'S TAR SANDS · 2015-01-15 · 2007] TAR SANDS: WORTH THE ENERGY? 325 Middle East. I2 The largest tar sand resources](https://static.documents.pub/doc/80x56/5f8587e1cd7dc20fb530642f/tar-sands-worth-the-energy-an-analysis-of-the-future-of-utahs-tar-sands-2015-01-15.jpg)
