SPE 150694

In-Situ Upgrading of Bitumen and Shale Oil by RF Electrical Heating

Richard H. Snow, PyroPhase Inc.

Copyright 2011, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Heavy Oil Conference and Exhibition held in Kuwait City, Kuwait, 1214 December 2011.

This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been
reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its
officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers is prohibited. Permission to
reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract
The objectives of upgrading oil sands bitumen are:
1) Reduce viscosity so that product can be pipelined to a refinery without building a costly above-ground upgrading plant,
which requires water and auxiliary fuel such as natural gas.
2) Reduce the cost and amount of upgrading required in the refinery.
3) Reduce the amount of sulfur, nitrogen, and heavy metals which cause problems with refining catalysts, since these
elements tend to concentrate in the heaviest molecules, which end up in the coke.
4) Sequester the coke in the ground. In-situ upgrading is also desirable for oil shale.

There have previously been reports that upgrading occurs during pyrolysis, but it has not been clear what conditions lead to
upgrading. Low temperature (300-350 oC) is economically desirable to minimize cost for RF electric heating of deposits, yet
there have been suggestions that low temperature favors coke formation. Both physical and chemical variables affect
upgrading, and we attempt to explain these complications.

In the following, pyrolysis experimental methods and data are reviewed. Reaction kinetic models to represent the data are
reviewed, and used to calculate low temperature product distributions. These results are replotted as product distillation
curves and interpreted to show how upgraded product can be produced. Models indicate that coke is formed after a sequence
of reaction steps, so its appearance is slow. Therefore light product can be harvested early, before material reacts further
producing excessive coke. Data on Western US oil shale are also reviewed and similar conclusions are drawn. Further
reservoir engineering studies are needed to optimize production of light oil.

Contents
Abstract 1
PyrolysisExperimentMethods 2
BitumenPyrolysisOverallKinetics 2
MeasurementsofPyrolysisRateandComposition 3
LowTemperatureMeasurements 3
HighTemperatureMeasurements 5
ReactionKineticModels 5
ModelCalculationsofProductDistributionsatLowTemperatures 7
ResultsofCalculationsat325C 7
ResultsofCalculationsat300C 8
SignificanceforInSituUpgrading 9
EffectofTimeonYield 10
EffectofTemperatureAloneonYield 10
EffectofUpgradingonViscosity 11
2 SPE 150694

Pyrolysis/UpgradingofWesternOilShale 12
TGAExperiments 12
PipeReactorExperiments 12
Conclusions 14
References 14

PyrolysisExperimentMethods
It is difficult to draw conclusions from published experimental work because the effects of chemical and physical variables
are confounded. The following discussion will clarify this situation.

Pyrolysis Lab Experiment Methods

Bomb reactor
(totally contained, gives best kinetic data)
Destructive distillation
(Product vapor outflow -Fisher Assay, TGA)
Pipe reactor
(product liquid and gas outflow)

Previous laboratory pyrolysis experiments have been conducted in at least three ways. In the bomb method the material is
placed in a sealed bomb and heated in an oven at a predetermined time and temperature. The products remain in the bomb
and are subject to further reaction, so that light products further degrade to heavier ones plus gas.

The second method may be called destructive distillation. The sample is placed in a flask fitted with a condenser and
heated to the desired temperature. The material that volatilizes is considered the product, and is collected as it evolves. Thus
the lighter products are selectively removed from further reaction by volatilization. Fisher Assay is an example with oil
shale, and TGA is an extreme example where the light products are swept away by flowing gas.
It is difficult to deduce kinetic models from destructive distillation experiments because the composition of the reacting
material is not well known, since the effectiveness of removal depends on the experiment arrangement. This is not a problem
with the bomb method, because all products remain in the reacting vessel. So the bomb method is best for determining
chemical kinetic data, even though this method is not best for upgrading.

A third method, the pipe reactor, permits harvesting of liquid product, and is intended to simulate production in the field.
When oil sand is heated in-situ by the RF method, bitumen begins to flow around 100 oC when its viscosity is reduced. If
heating is continued further, significant pyrolysis begins around 300 to 325 oC. With oil shale the product does not flow until
solid kerogen begins to pyrolyze, also around 300 oC.

We will show that for optimum production with upgrading it is important to allow some pyrolysis, but remove products
before excessive reaction converts them to coke. Reservoir flow models can determine the conditions to control flow. Once
a kinetic model has been formulated based on bomb method experiments, it can be combined with reservoir flow predictions
to determine how much material is left to react. Accordingly, the next section reviews reaction kinetics.

BitumenPyrolysisOverallKinetics
According to (Henderson JH, 1965) it has been well established that at least the initial stages of petroleum cracking can be
represented by first order kinetics:
dC/dt = kC (1)
which integrates to kt = ln (C/Co)
and the temperature dependence of the rate constant is given by the Arrhenius function:
k = A exp( -E/RT) (2)
where Co = weight fraction of reactant present initially
t = time, sec or hr
k = specific rate constant, 1/sec or 1/hr
A = frequency factor, 1/sec or 1/hr
E = activation energy, cal/g-mol or kJ/g-mol
R = gas constant, 1.987 cal/g-mol oK or 8.314 kJ/g-mol - oK

Note that the dimensions of (g-mol) cancel out in the exponential, which allows this expression to be used even when
SPE 150694 3

concentrations are expressed on a weight basis. This is necessary in kinetic representation of complex reactions, where
individual molecular species cannot be identified.

The chemistry of pyrolysis is known to follow a complex free-radical mechanism involving many reactions, which will be
reviewed below. Engineers often represent complex chemistry empirically, defining and measuring products by some
empirical means, usually distillation cuts or by classes of compounds. Extensive laboratory measurements of this sort are
available from the work of (Hayashitani, 1978), and also (Phillips CR, 1985), both of whom used the bomb method. Phillips
developed kinetic models to fit their combined data and predict product compositions.

MeasurementsofPyrolysisRateandComposition
LowTemperatureMeasurements
RF in-situ pyrolysis will be implemented commercially at low temperatures and month-long contact times, so as to minimize
the cost of electric energy input. Low temperature data are needed to determine the results of low tremperature upgrading.
But most of the published experiments have been carried out at 360 oC or more, and low temperature measurements are
limited ot overall rate measurements. We will combine these two sets of data to predict the needed results.

Hayashitani referenced several earlier low-temperature studies. (McNab JG, 1952) carefully carried out experiments at 267,
309, 357 and 372 oC using the bomb method. Those at the lowest temperature were kept at constant temperature as long as 3
to 6 months, and rate constants for the production of distillable products were determined. (Henderson JH, 1965) also
conducted long-term bomb experiments at low temperatures.

McNab and Henderson did not try to determine product distributions. They defined only three product fractions, namely
liquids, gases and residue or coke. However, they did try to account for the fact that the original bitumen contains light-
boiling material that should not be considered as product of cracking reactions because it was already present in the feed.
Both used what they defined as the 20+conversion method. For this purpose the raw bitumen was analyzed by distillation,
and a temperature determined at which 20 % distilled over. In McNabs case he used bitumen from McMurray Athabasca,
which produced 20% overhead at 359 oC. The distillation was carried out quickly to prevent pyrolysis. Then in subsequent
cracking experiments the liquid product was analyzed by distilling in the same way as the feed. The amount of the product
exceeding 20% was taken to be the reaction product, and was expressed as a fraction of the original available reactant, i.e. of
the 80%. Admittedly, it might not contain the same composition as the original -20% fraction. But in any case he used the
resulting fractions only to measure an overall first-order reaction rate constant.

Furthermore Henderson limited the length of cracking experiments so that only low conversions were measured, ranging
from 3 to 24%. The purpose was to assure that only first-order reactions would occur, while more complex higher-order
reactions might occur at later times. (Hayashitani, 1978) summarized all these low-temperature data in an Arrhenius plot
reproduced in Fig 1. The slope of the middle line has activation energy of 53 kcal/mol, which is in the high end but still in
the range of values for bitumen and shale pyrolysis reported by others.
4 SPE 150694

Fig 1 shows that pyrolysis can be carried out in 10 days at 325 oC, or in a few weeks at 300 oC. It is expected that in-situ
pyrolysis can be carried out within this time range, since one or two months are required to heat the resource to reaction
temperature, and additional weeks for the product to drain to the collection sump.

SPE 150694 5

HighTemperatureMeasurements
Phillips and Hayashitani (Hayashitani, 1978) (Phillips CR, 1985) used longer experiments in order to more completely
determine the development of product distributions, which they separated into several fractions. The oil fractions were first
extracted from asphaltenes using n-pentane, and the undissolved material was filtered out. Asphaltenes were then extracted
from this solid using methylene chloride. The residue on the sand was then defined as coke. The pentane-soluble oil fraction
was then distilled into separate cuts. Asphaltenes were presumed to be the final precursor of coke and gas formation because
they are large molecules. Therefore it was felt important to measure asphaltenes separately in both feed and product.

This procedure defined the following fractions:

1. Coke
2. Asphaltenes
3. Heavyoils(boiling>420oC)
4. Middleoils(180420oC)
5. Lightoilsboilingbelow180oC,and
6. Gases
Phillips conducted pyrolysis primarily at temperatures of 360, 400 and 420 oC. Most of Hayashitani data are in this range.
Reaction times ranged from minutes at the highest temperatures to 116 hr at the lowest temperature. A sample of their results
is given in Fig 2. Note that light oil first increases, then later declines as coke formation takes over.

Fig 2. Bitumen Pyrolysis at 420 C (Phillips 1985)

ReactionKineticModels
Both Hayashitani and Phillips developed empirical kinetic models. Additional empirical models and experimental data were
published by (Barbour RV, 1976) (Strausz OP, 1977) (Speight, 1970) (Millour JP, 1985) (Kapadia, PR, 2010). Some of these
models have difficulty fitting the data for production of coke, which may be important in affecting the permeability of the
deposit and the yield of light components. (Phillips CR, 1985) Model B treats coke as the end product of a sequence of
reactions, and his model appears to be adequate for this purpose. It is represented in Fig 3.

Researchers at the University of Nancy (Lannuzel et al, 2011) have investigated the free radical mechanism of bitumen
pyrolysis, with sponsorship from Total. Starting with pyrolysis of simple molecules they have built up to a mechanism of
6 SPE 150694

5200 elementary reactions, to predict the course of pyrolysis over a wide range of conditions. They find that wide changes in
temperature and pressure do profoundly affect the products, but they usually affect the overall rate too. Therefore since the
empirical mechanism of Phillips agrees with the overall rate measured at 300 to 325 oC, we feel justified in using it to predict
the empirical product distribution at these modestly lower temperatures. Of course it would be desirable to check these
results using a free radical model.

Fig 3. Phillips Pyrolysis Model B

Phillips represented this reaction model as the following rate equations: (4)

The Phillips Model B represented in Fig 3 can be expressed as matrix equation:

dC/dt = M C MT C
where C is a vector of product fractions and M is a sparse matrix of reaction rate constants and MT the transposed matrix:
SPE 150694 7

Table 1. Matrix of Rate Constants at Two Temperatures

Source-------- ------------
6 Gas 5 Lt oil 4 Mid oil 3 Heavy oil 2 Asphalt 1 Coke Product
k26 6 Gas
2.55e-4 @300 oC
14.1e-4 @325 oC

k45 5 Lt oil
40.8e-4
307e-4
k54 k34 4 Mid oil
12.5e-4
67.6e-4
k43 k23 3 Heavy oil
7.58e-4
68.2e-4
k32 2 Asphalt
10.8e-4
76.0e-4
k21 1 Coke
14.7e-4
67.8e-4

Phillips took k54 = k43 = k32 because these are similar bimolecular reactions.

ModelCalculationsofProductDistributionsatLowTemperatures
We expressed the reactions (4) of the Phillips model in the Comsol Multiphysics system and solved at 300 and 325 C . The
initial bitumen composition was that reported by (Hayashitani, 1978) for Athabaska bitumen. To calibrate the model for
lower temperatures we calculated the incremental production of light and heavy oil fractions and estimated the overall rate
constant. We found that these values at low temperatures agree with the overall rate in Fig 1. In view of this, it seems
reasonable to extrapolate the complete model to lower temperatures to determine the relative proportion of coke, gas, and
light oil. This is what we have done in the following figures at 325 C and at 300 C.

ResultsofCalculationsat325C
0.5
C1(Coke)
C2(Asphaltene)
0.4
Concentration,wt%

0.3

0.2

0.1

0
0 100 200 300 400 500
Time,hr

Figure 4. Concentrations of Components 1 (Coke) and 2 (Asphaltene) at 325 C. Initial asphaltene fraction on the abscissa is
the raw bitumen asphaltene content.
8 SPE 150694

0.8

C3(HeavyOils)
0.7
C5(LightOils)
0.6
Concentration,wt%

0.5

0.4

0.3

0.2

0.1

0
0 100 200 300 400 500
Time,hr

Figure 5. Concentrations of Components 3 (Heavy Oils) and 5 (Light Oils) at 325C

0.12

C4(MidOils)
0.1 C6(Gases)

0.08
Concentration,wt%

0.06

0.04

0.02

0
0 100 200 300 400 500
Time,hr

Figure6. Concentrations of Components 4 (Middle Oils) and 6 (Gases) at 325C

ResultsofCalculationsat300C
0.5
C1(Coke)
C2(Asphaltene)
0.4
Concentration,wt%

0.3

0.2

0.1

0
0 500 1000 1500 2000 2500
Time,hr

Figure 7. Concentrations of Components 1 (Coke) and 2 (Asphaltene) at 300C

SPE 150694 9

0.8

C3(HeavyOils)
0.7
C5(LightOils)
0.6
Concentration,wt%

0.5

0.4

0.3

0.2

0.1

0
0 500 1000 1500 2000 2500
Time,hr

Figure 8. Concentrations of Components 3 (Heavy Oils) and 5 (Light Oils) at 300C

0.14

C4(MidOils)
0.12
C6(Gases)

0.1
Concentration,wt%

0.08

0.06

0.04

0.02

0
0 500 1000 1500 2000 2500
Time,hr

Figure 9. Concentrations of Components 4 (Middle Oils) and 6 (Gases) at 300C

Results for In-Situ Upgrading

SignificanceforInSituUpgrading
The amounts of each boiling cut in Figures 4 to 6 are replotted in the form of distillation curves of the products of pyrolysis at
contact times of 100, 150 and 250 hrs in Fig. 10. The coke fraction has been left out, i.e. the curves represent the normalized
composition of the oil and gas fractions. For comparison the raw bitumen curve is shown, as well as a distillation curve for
32 API light crude. Pyrolysis products have more light fractions than even the light crude, but they also contain larger
amounts of the heaviest fraction. (These curves are only approximate, because of the wide boiling temperature range of the
bitumen fractions.)
10 SPE 150694

100

90

80

70

60 150Hrs

50 API32Crude
Bitumen
40
100Hrs
30
250hrs
20

10
o
C
0
0 200 400 600 800 1000

Fig 10. Distillation curves for feed and products resulting from pyrolysis at 325 C.

EffectofTimeonYield
In addition to lightness of product, yield is also important. The proportion of light products increases with pyrolysis time, but
the formation of coke (not shown in graph) also increases, decreasing the yield. These results were computed from Figs 4-6:

Time,hrs %Upgraded OilYield,% GasYield,% CokeYield,%

100 25 79 4 17
150 31 72 5 23
250 35 69 7 31

Pyrolysis at 300 C gives similar results, except that longer time (weeks) at this temperature is required.

EffectofTemperatureAloneonYield
Results in the first two rows of Table 5 show that there may be a small temperature effect even for the same degree of
conversion. But much bigger production of coke occurs when reaction is continued for long time, as in the third row. This
corresponds to the curve at 420 C in Fig 2, which shows that light oil yield reaches a maximum at 2 hrs, after which it drops
while the coke yield continues to rise, reaching 42 per cent at 5 hrs. Our conclusion is that the ratio of light oil to coke
depends largely on how far the reaction is carried on.

Table 5. Results of Bitumen Pyrolysis

Light Oil, % Coke,%

Figs 4, 5 @ 90 hrs, 325 C 19 8

Fig 2 @ 0.3 hrs, 420 C 19 7

Fig 2 @ 2.0 hrs, 420 C 30 25

SPE 150694 11

These results show that for an in-situ process, harvesting product at a partial stage of pyrolysis appears to be the key to
producing an upgraded product. Determining how to harvest product at the right time is a problem in reservoir engineering
combined with chemical kinetics, which is a task remaining.

EffectofUpgradingonViscosity
(Henderson JH, 1965) showed that upgrading has a striking effect on the viscosity of the product, caused by the

Figure 11. Viscosity from (Henderson JH, 1965)

change in composition. He was concerned with lowering viscosity so the product could be pipelined to a refinery. He
measured the viscosity of various samples of Athabaska bitumen after cracking, with the results shown in Fig 11.
A viscosity of 100 to 200 cp should be readily pipelined, and corresponds to 15 to 20% upgrading in Fig 11. As described
above, Henderson defined per cent upgrading as the percent of fraction boiling below 359 C, after subtracting the 20 per
cent in his raw bitumen. Fig 10 shows that the raw bitumen in this study based on data of (Phillips CR, 1985) contains only 5
per cent below 359 C. After only 100 hrs the curve in Fig 10 shows 29 per cent material below 359 C, which corresponds
to 25 per cent upgrading for this bitumen. It should be suitable feed for a refinery equipped to carry out some further
processing. Hendersons untreated bitumen had a viscosity of 5242 cp, which he measured at 150 F.

In the RF in-situ process the product will be hot as it is being extracted, so its viscosity will be even lower, and production
out of the deposit by gravity flow should be effective.
12 SPE 150694

Fig 12. Effect of temperature on viscosity of Asphalt Ridge bitumen (IITResearch institute, unpublished)

Pyrolysis/UpgradingofWesternOilShale
Production of Western US oil shale inherently involves pyrolysis and upgrading, since kerogen must be broken down to oil
before it can flow out of the deposit. There have apparently been no experimental efforts as described above with bitumen to
measure several pseudo-components of oil as the reaction evolves, other than oil, gas and char. There do not appear to be
any bomb pyrolysis experiments reported for oil shale. Instead, pyrolysis laboratory experiments have been done under
conditions where liquid or vapor outflow could occur.

Oil shale initially has a very low permeability until kerogen breaks down, which opens up pores for flow. It differs from oil
sands, which have high permeability. It also has fine pores resulting in high capillary forces, which hold the oil in place until
autogenous gas pressure or gravity pushes it out. Therefore it is questionable how much the production is limited by
physical processes, and how much by chemical kinetics.

TGAExperiments
(Sresty GC, 1982) found that the overall rate of product formation under thermo-gravimetric analysis (TGA) conditions was
50 times greater than when liquid was produced from a pipe reactor. This shows that in TGA experiments the physical
processes of product removal may be less limiting than the chemical kinetics rate, in contrast to pipe reactors where the flow
conditions governing escape of product are very important. Fig 13 shows the overall rate constants from Srestys TGA tests
on western oil shales, which gave activation energy of 45.4 kcal/mol - K.

PipeReactorExperiments
In Fig 14 we also present results of (Sresty GC, 1982)s low-temperature experiments to compare with bitumen pyrolysis.
These experiments were done on cores in a heated tubular reactor. The objective was to show that upgraded liquid product
could be partially cracked to lighter materials and removed before it degraded further to form excessive amounts of coke.
Therefore liquid product was allowed to exit. It is believed that capillary forces tended to retain some liquids, while
autogenous gas generation tended to push liquids out of the pores. The cores were too small for gravity to have a drive
effect. However a later field test showed that gravity is effective to produce oil if the height is larger, in this case 19 ft.
SPE 150694 13

Sresty reported liquid products obtained from the lab experiments as a percent of the Fisher assay of the raw shale. At the
lowest temperatures of 280 and 300 C the experiment continued for 10,000 hrs (1.2 years.) The results are plotted in Fig 14.
The slope gives the activation energy of 39.6 kcal/mol -K, lower than for the TGA experiments. Since physical processes
usually have lower activation energy than chemical reactions, this suggests that physical processes are affecting the rate of
production of liquid.

Experiments showed that longer residence time gave higher coke yields and lower product yields. The coking loss when
pyrolysis was carried to completion was
28% of FA at 280 C
18% at 300 C
6.9% at 360 C
These differences were believed due to the physical effect of lower autogenous pressure pushing liquid out at lower
temperatures, rather than due to kinetics.

Fig 13. Arrhenius plot for oil shale TGA pyrolysis (Sresty, 1982)

The Phillips model represents coke as the end product of a series of complex reaction steps, which take some time to occur.
Oil shale starts with kerogen molecules, which are larger than bitumen and may proceed more directly to coke. But still,
analogy with bitumen and limited data suggest that physical effects which determine effectiveness of harvesting early
products may be more important than chemistry for upgrading shale oil. In other words the success in producing a good
yield of light product depends on harvesting the product before it reacts too far. By analogy with bitumen pyrolysis, any
effect of lower temperatures on coke formation can likely be overcome by early product harvesting.
14 SPE 150694

Fig. 14. Arrhenius plot of Rate Constants for Pyrolysis of Oil Shale in a Tubular Reactor (Sresty, 1982).

Needham (1976) reported an experimental study of oil shale kinetics, intended to elucidate the effects of temperature and
heating rate. It included pipe reactor tests. He pyrolyzed cores of Green River shale mounted in a pipe reactor such that the
shale expanded and sealed against the reactor walls. Sresty observed that the higher grades of shale expanded and sealed
against the pipe reactor, while the lower grades did not expand but left a passage between the core and the reactor walls.
Needham also tested other experimental arrangements. A few of his product yields and quality are reported below:

Table 3. 30 gal/ton Oil Shale Pyrolysis Data (Needham, 1976)

Crushed shale with Gas combustion Pilot Pipe Reactor, Pipe Reactor,
gas recycle, 433 C Retort Process 371 C 427 C
API gravity 19.7 20.3 32 32
Wt % FA oil 83 85 78 81
recovered

The product from the pipe reactor was mostly recovered as liquid forced out through the core by autogenous gas pressure,
and was believed to be subject to cracking as it made its way through the heated core. This resulted in upgrading, as shown
by the increase in API gravity. The yield was less from the other reactors, which had more complex operating conditions.

Conclusions
Upgraded oil can be produced from oil sands by harvesting oil from reaction zone at 10 to 20% conversion. This applies at
lower temperatures of 300-325 C, as well as at higher temperatures. Limited available experimental data indicate it is true for
oil shale as well.

References
Al-Ayed OS and Sulieman QAE, Variable Activation Energy to Model Oil Shale Pyrolysis Kinetics, 27th Oil Shale
Symposium, 2007.
SPE 150694 15

Baker-Jarvis J, and Inguva R, Mathematical Model for In Situ Oil Shale Retorting by Electromagnetic Radiation, Grant DE-
AS20-81LC10783, University of Wyoming, Laramie,1984.
Barbour RV, Dorrence SM, Vollmer TL, and Harris JD, Pyrolysis of Utah Tar Sands, Products and Kinetics, Preprints Am
Chem Soc Fuel Div 21(6) 278-289, 1976.
Hayashitani, M, Thermal Cracking Models for Athabasca Oil Sands, SPE 7549, 1978.
Henderson JH, and Weber L, Physical Upgrading of Heavy Crude Oils by the Application of Heat, J Can Petrol Tech, Oct-
Dec, 206-212, 1965.
Kapadia, PR et al, A Comprehensive Kinetic Theory to Model Thermolysis, Aquatheroloysis, Combudstion, and Oxidation
of Athabaska Bitumen, SPE 129660, 2010
Lannuzel, F, Burkle-Vitzthum, V, Bounaceur, R, Marquaire, P-M, and Michels, R, Reaction Mechanisms in Petroleum: From
Experimentation to Upgrading and Geological Conditions, published online, 2011.
McNab JG, Smith PV Jr., and Betts RL, The Evolution of Petroleum, Ind & Engr Chem 44 (11) 2556-63, 1952.
Millour JP, Moore RG, Bunion DW, Greenback MG and Gee DN, A Simple Implicit Model for
Thermal Cracking of Crude Oils, SPE 14226, 1985.
Needham RB, Oil yield and quality from simulated in-situ retorting of Green River oil shale, SPE 6069, 1976.
Phillips CR, Hider NI, and You CP, Kinetic Models for the Thermal Cracking of Athabasca Bitumen, Fuel 64 (5) 678-682,
1985.
Speight JG, Thermal Cracking of Athabasca Bitumen, Asphaltenes and Deasphalted Heavy Oil, Fuel 49 (2) 134-145, 1970.
Sresty GC, Dev H, Snow RH, and Bridges JE, Kinetics of Low Temperature Pyrolysis of Oil Shale by the IITRI RF Process,
15th Oil Shale Symposium, 1982.
Strausz OP, Jha KN, and Montgomery DS, Chemical Composition of Gases in Athabasca Bitumen and in Low-Temperature
Thermolysis of Oil Sand, Asphaltene and Maltene, Fuel 56, 114, 1977.