AOSTRA

PUBLICATION SERIES

/dlberra
OILSANDS TECHNOLOGY
AND RESEARCH AUTHORITY
 SYNOPSIS REPORT NO. 50

AQUATHERMOLYSIS .

A SYNOPSIS OF WORK ON THE CHEMICAL REACTION

BETWEEN WATER (STEAM) AND HEAYY OIL SANDS
DURING SIMULATED STEAM STIMULATION

(AQSTRA CONTRACTS NO. 11, 103, 103B/C)

J.B. HYNE
PRINCIPAL RESEARCHER

DEPARTMENT OF CHEMISTRY
THE UNIVERSITY OF CALGARY
CALGARY, ALBERTA

APRIL 1986
 SYNOPSIS REPORT NO. 50

AIJUATHERMOLYSIS -

A SYNOPSIS OF WORK ON THE CHEMICAL REACTION

BETWEEN WATER (STEAM) AND HEAVY OIL SANDS
DURING SIMULATED STEAM STIMULATION

(AQSTRA CONTRACTS NO. 11, 103, 103B/C)

J.B. HYNE
PRINCIPAL RESEARCHER

DEPARTMENT OF CHEMISTRY
THE UNIVERSITY OF CALGARY
CALGARY, ALBERTA

APRIL 1986
 TABLE OF CONTENTS

Page

PREFACE 1

ACKNOWLEDGMENTS 3

REPORTS REFERENCED 4

MAJOR PUBLISHED WORKS REFERENCED 5

*
LIST OF FIGURES 7

CHAPTER 1 - AN OVERVIEW OF AQUATHERMOLYSIS 9

Introduction 9
The Aquathermo lysis Window 9
The Initiation of Aquathermo lysis 10
Acid Polymerization 11
Reaction With Water 11
The Water Gas Shift Reaction 12
C0 2 Release from Porous Medium 13
Hydrodesulphurization 13
H 2 S Production and Scavenging 13

CHAPTER 2 - EVIDENCE FOR AQUATHERMOLYSIS AS A DISTINCT PROCESS 15

CHAPTER 3 - VISCOSITY AND AQUATHERMOLYSIS 17

CHAPTER 4 - SOURCES OF C0 2 DURING AQUATHERMOLYSIS 20

C0 2 Production as a Function of Temperature 20
C0 2 from Sand Phase 21
C0 2 From Carboxylic Acids 22
C0 2 Sources - An Attempt at Mass Balance 23

CHAPTER 5 - THE GENERATION AND CONTROL OF H 2 S DURING STEAM

STIMULATION (AQUATHERMOLYSIS) 25

CHAPTER 6 - CARBON MONOXIDE IN THE AQUATHERMOLYSIS PROCESS 29

C0 2 Production from Added CO 30
Hydrogen Production through Added CO 31
Hydrogen Sulphide Production through Added CO 33
Effect on Viscosity of Added CO 34
The C0 2 Patent 35
CHAPTER 7 - THE ROLE OF METAL IONS IN AQUATHERMOLYSIS 36
Effect on Hydrogen Production - The Water
Gas Shift Reaction (WGSR) 36
Effect on H 2 S Production 37
Effect on Viscosity 38
Combined Effects of C0 2 and Metal Ions Patent 41
Patent 43

CHAPTER 8 - AQUATHERMOLYSIS AND RECOVERED OIL COMPOSITION 44

Sensitivity of Viscosity to Chemical
Composition of Oil 47

CHAPTER 9 - SOME COMMENTS ON SAMPLE INHOMOGENEITY AND

EXPERIMENTAL ARTIFACTS 49

CHAPTER 10- CONCLUSIONS AND FUTURE DIRECTIONS 52

 ! I

-1-

PREFACE

In attempting to present an easily readable yet i nformati ve and

scientifically sound review of some seven years of work and over 1,000

pages of progress reports we have encountered all of the usual problems.

Do we present the results and then analyze them? Do we present an

introductory overview and then the supporting evidence? How do we select
illustrative data from the thousands of experimental results that have been
generated? Do we depend more on graphical presentation or tabular material?

The professional reviewer is well accustomed to these problems of

selection but it is the work of others that is mainly involved in such

reviewing processes: when the reviewer is dealing with his own work
everything seems equally important!

Our task is not made easier by the fact that the topic we are reviewing

is one that is not widely regarded as being of substantial importance in

the heavy oil and oil sands field. The still widely accepted view is that
the primary if not sole consequences of injecting high temperature water
into heavy oil and oil sand formations are physical - i.e. temperature
rise, viscosity reduction, pressure drive, emulsification etc. Most
consider that the chemi ca consequences of in-situ steam stimulation are
little more than nuisances of varying degree and few believe that there may
be substantial benefit to be gained by optimizing the chemical effects.

We, of course, believe differently and have spent the last decade or so
trying to prove these beliefs. We have coined the term "aquathermo lys s" i

to describe the overall set of chemical processes that can result from
contacting the heavy oil or oil sands with high temperature water, whatever
the phase. We chose this term rather than hydrothermolysis because the
latter has already been applied to thermal breakdown using hydrogen as the
active reagent. Although the "aquathermolysi s" term mixes our etymological
roots it clearly indicates the involvement of "aqua" in the thermal "lysis"
of the substrate heavy oil material. We trust the term is not too much of
a mouthful
 Digitized by the Internet Archive
in 2016

https://archive.org/details/aquathermolysissOOhyne
 I

-2-

Our task in this review is then clear; to try to demonstrate on the

basis of experimental evidence that the aquathermo lyt c chemical processes

i

we describe do indeed play a significant role in in-situ steam stimulation

and that the results of such chemical processes can be profoundly important
in managing the development and production of heavy oils and oil sands
using steam stimulation in one of the many protocols that have been
developed, e.g. steam drive, huff and puff etc. We have attempted to do
this by first presenting an overview of what we believe to be the main
components of the aquathermo lysis system of chemical reactions and, in

particular, to illustrate how these all fit together. We then offer a

series of chapters each of which deals with a specific feature of the

aquathermo lysis phenomenon and presents supporting evidence for the general
picture offered in the overview. We also attempt, where appropriate, to
show how these features of aquathermo lysis can effect such important
practical matters as production of H2 S gas, viscosity of produced oil,
maturation of the regions of the reservoir around the production zone, etc.

A quick scan of the overview chapter will provide the reader with a

general appreciation of the work we have carried out on aquathermolysis.

We hope, however, that this will encourage rather than deter further
exploration of the more detailed material in the main body of the document.
If this synopsis serves to persuade even an additional few of the

importance of chemi ca reactions between oi I sand constituents and water

during steam stimulation, it will have served a useful purpose.
>3>Q
 .

-3-

ACKNOWLEDGMENTS

The work reviewed here was carried out over a period of years with the
financial support of AOSTRA, Petro-Canada and INTEVEP (PDVSA) . In addition

to their essential financial support, however, these organizations provided

most useful technical input through membership of their staff on a

Technical Steering Committee which met to review and critique the program
at six monthly intervals throughout the major part of the program.

In particular we wish to thank Pedro de Silva, Pierre Lichaa, Benito

Luongo, Carlos Rizek and Dino Verona of INTEVEP and PDVSA; Loren Hepler,
Ernest Wiggins, Ted Cyr and Chu Hsi of AOSTRA and Mike Goss, Mike Danyluk

and Peter Bulkowski of Petro-Canada.

The list of colleagues and research assistants who have participated

directly in the aquathermo lys s program is
i a long one. In the early days

Digby Macdonald and J.W. (Pirn) Greidanus were the only other team members.
Gary Derda I I and David Tyrer were later participants along with Peter
Clark, Mark Smith and Norman Dowling. Jeff Koo, Gerald Tsang, and Kaoru

Saito have also contributed to the project in recent years, together with
Kevin Lesage and Bob Clarke, who was of special assistance in preparing

this synopsis.

Throughout the project Dianne Jorgenson has managed the business

affairs of the research group and Bonnie King has been primarily
responsible for the production of the final format of the written reports.
Without the team work of all involved this review would not have been
poss b e
i I

J.B. Hyne
Calgary 1986
 -4-

REPORTS REFERENCED

This review covers the work presented in the following detailed reports
submitted to AOSTRA. These reports should be accessed for all details of

experimental data upon which this review is based.

Agreement #11 - 1st Progress Report, October 1, 1976 - March 31, 1977
2nd Progress Report, April 1, 1977 - September 30, 1977
3rd Progress Report, October 1, 1977 - April 30, 1978
4th Progress Report, May 1, 1978 - October 31, 1978
Final Progress Report

Agreement #103 - 1st Progress Report, period ending January 1980

2nd Progress Report, period ending June 1980
3rd Progress Report, July - December 1980

4th Progress Report, January - June 1981

Supplement to 4th - Special Report July 1981,
"Generation of H2 S during In-si tu Steam Treatment of
Heavy 0 i Is"

Final Report

Agreement #103C - 1st Progress Report, period ending January 1982

2nd Progress Report, period ending July 1982

Experimental Procedures Booklet

3rd Progress Report, period ending December 1982
4th Progress Report, January - June 1983
5th Progress Report, July 1983 - March 1984

Agreement #103C Extended - Final Report - May 1985

'
 3

-5-

MAJOR PUBLISHED WORKS REFERENCED

J.W. Greidanus, D.D. Macdonald and J.B. Hyne, Proceedings of the

Canada/Venezuela Oil Sands Symposium, Vol .17, 1977, 162-167.

US Patent, #4,157,462, J.C. Allen, issued May 29, 1979.

/

D.S. Montgomery, AOSTRA Conference on "Advances in Petroleum Recovery and

Upgrading Technology, Calgary, June 1980. -

D.W. Bennion, L.D.L. Vorndran, A.J. Serres, J.K. Donnelly, R.G. Moore,
Canadian Journal of Chem. Engineering, Vol .58, October 1980, 580-587.

J.B. Hyne, P.D. Clark, R.A. Clarke, J. Koo, J.W. Greidanus, J.D. Tyrer and

D. Verona, Proceedings 2nd International Conference on Heavy Crude and

Tar Sands, Vol. I, Chapter IX, 1982.

M.H. Akstinat, J. Petroleum Geology, 5, 4, 363-388, 1983.

M.E. Hernandez, M.T. Vives, J. Pasquali, Organic Geochemistry, Vol .4, Nos.

& 4, 1983.

P.D. Clark, J.B. Hyne and J.D. Tyrer, Fuel, 1983, Vol. 62, August, 959-962.

Canadian Patent, #1,170,444, US Patent, #4,487,264, 1984, J.B. Hyne and

J.D. Tyrer.

P.D. Clark and J.B. Hyne, AOSTRA Journal of Research, Vol.l, No.l, 1984,
15-20.

P.D. Clark, J.B. Hyne and J.D. Tyrer, Fuel, 1984, Vol .63, January, 125-128.

P.D. Clark and J.B. Hyne, Fuel, 1984, Vol .63, December, 1649-1654.
 -6-

Canadian Patent #1,189,444 and US Patent, #4,506,733, J.B. Hyne and P.D.
Clark, 1985.

R. A. Clarke, N.I. Dowling and J.B. Hyne, ASRL Quarterly Bulletin, Vo I. XXII,

No .
1 , 1985, 12-24.

A.K. Mehrotra and W.Y. Svrcek, AOSTRA Journal of Research, Vol .2, No. 2,

December 1985, 83-94.

S. G. Shore, J.C. Bricker, C.C. Nagel, A. A. Bhattacharyya, JACS, 107, 1985,

377-384.
 s s

-7-

LIST OF FIGURES

Fig. 1 The Aquathermo lysis Window

Fig. 2 The Initiation of Aquathermo ysi I

Fig. 3 Other Aquathermo lysis Reactions

Fig. 4 Chemical Reactions Occurring in Water/Heavy Oil Systems

Fig. 5 Main Gaseous Products from Aquathermo lysis at 300°C

Fig. 6 Viscosity Behaviour of Recovered Bitumens as Function of

Temperature and Duration of Aquathermo lysis

Fig. 7 Reservoir Temperature Profiles and Sequential Steam Stimulated

Production

Fig. 8 Aquathermo lysis of Whole Core Samples C0 2 Production vs Time and

Temperature

Fig. 9 C0 2 Production

Fig. 10 Rate of Production of H 2 S from Aquathermo lysis of Whole Core

Sample of Venezuelan Oil Sand as a Function of Time and
Temperature

Fig. 11 Rate of Production of H 2 S from Aquathermo lys s of Whole Core i

Sample of Canadian Oil Sand as a Function of Time and Temperature

Fig. 12 Relationship Between Produced H 2 S and Average Reaction Temperature

- Field and Lab Simulation

Fig. 13 CO Production and Depletion during Aquathermo lys i

Fig. 14 Effect of Adding Overpressures of CO to Aquathermo lys is Vessel on

C0 2 Production

Fig. 15 Hydrogen Production Data for Aquathermo lys is of Venezuelan and

Canadian Oil Sands at 240°C

Fig. 16 Rate of Hydrogen Sulphide Production Resulting from Added CO

Aquathermo lys is at 240°C

Fig. 17 Effect of Added CO on Viscosity of Recovered Oils from 240°C

Aquathermo lys is

Fig. 18 Trends in the Production of Hydrogen from the Aquathermo lys is of a

Venezuelan Oil Sand with Added Metal Salts over 28 Days at 240°C
-
 -8-

Fig. 19 Trends in the Production of Hydrogen Sulphide from the

Aquathermo lysis of a Venezuelan Oil Sand with Metal Salts Over 28
Days at 240°C

Fig. 20 Effect on Viscosity due to Aquathermo lysis of Oi I Sands with Added

Salts at 240°C

Fig. 21 Trends in Viscosity of Oil from the Aquathermo ys s of Canadian

I
i

Oil Sand with Combinations of Metal Salts and Carbon Monoxide Over
14 Days at 240°C

Fig. 22 SARA Properties of Recovered Oils

Fig. 23 Relationship Between Viscosity and Saturate/Asphaltene Ratio

Fig. 24 The Effect of Containment Vessel Material on Aquathermo lysis at

240° C
 -9-

CHAPIEP

AN OVERVIEW OF AQUATHERMOLYSIS

Introduction

Aquathermo lysis (aqua (l)=water; thermos (g)=hot; lysis (g)=looseni ng,

dissolution) of oil sands and/or heavy oils means the breaking down of the
components of these deposits through chemical reactions brought about by
contact with hot water either in the vapour or liquid phase or a

combination of both.

The chemical reactions that make up the overall aquathermo lytic process
are numerous and are interrelated in a complex manner. They also occur over
a wide range of temperatures depending upon the strength of the chemical
bonds that are being loosened and disrupted. This temperature range
includes a number of other chemical reactions (e.g. thermal cracking) that
can occur when oils and oil sands are subject to thermal stress, but there
seems to be a relatively narrow temperature range where the primary
chemical reaction appears to be that with water, i .e. aquathermolysis. We
have labelled this the aquathermolysis window.

The Aquathermolysis Window

In Fig. 1 we have attempted to illustrate the aquathermolysis "window"

concept. The idea of a window or opening in the overall temperature range

arises from the fact that aquathermolysis appears to be the predominant

form of chemical reaction between the thermal cracking (with or without
water) that is occurring above 350°C and the very slow chemical reactions
that characterize slow thermal maturation of the reservoir and its contents
below 200°C. We shall present data later which shows this marked change in

response to temperature and the presence of water.

 - 10 -

Fig. 1 also shows that between 200°C and 300°-350°C the chemical
reactions characteristic of aquathermo lysis rise to reach a maximum just as
the important thermal cracking reactions begin. Some of these thermal
cracking reactions may actually be suppressed in the 275°-325°C range by

the presence of water thus favouring the aquathermo lysis reactions. The
200°C to 325°C range is particularly important because it includes many of

the reservoir temperatures associated with steam stimulation (e.g. Shell,

Peace River; Esso, Cold Lake).

The relative bond energy scale is also of interest in this correlation

because it suggests that the organosu phur I
compounds in the heavy oils and

bitumens may be the first to react in the aquathermo lysis region producing
organosu phur fragments and
I
H2 S while amines and phenols resulting from
cleavage of nitrogen and oxygen containing organics are characteristic
products of higher temperature regimes.

The Initiation of Aquathermo lysis

The heart of the overview is contained in the reaction schematics

presented in Figs. 2 and 3. In these two Figures we have attempted to
identify what we believe to be all of the principal chemical reactions that
together make up the overall aquathermo ys s process.
I
i More importantly we
have tried to show how these component reactions are interrelated and how
the product of one may become the feedstock for another.

Our evidence suggests that the reactive species produced upon

aquathermo lys i
s, probably of organosu phur compounds, can either polymerize
I

(step 2 in Fig. 2) or react with water (step 1) to yield smaller fragments

that participate in a subsequent series of reactions that we consider to be
an essential part of the overall aquathermo lys is process and will be
described below. The polymerization leads to undesirable viscosity
increases; reaction with water leads to reduction in the viscosity of the
oils.
 - 11 -

Acid Polymerization (Step 2, Fig. 2)

This is believed to be the reason for observed initial i ncrease in

viscosity of some oils when steamed (see Chapter 3) and is generally

undesirable for this reason. It is highly dependent on the specific

molecular structure of the heavy oil and further HPLC and gel permeation
work on changes in the chemical nature of the oil may assist in identifying
those oils that are more sensitive. In both Canadian and Venezuelan
samples we have studied we have had one sample that was sensitive to acid
polymerization and one that was not. Generation of copious amounts of C0 2
(say from carbonates in mineral phase) can enhance acid polymerization.

This is likely a factor in some Venezuelan samples where larger C0 2

production enhances acidity. Certain added salts (e.g. vanadyl) can also
hydrolyze and produce acidic conditions. Oils that are sensitive to such
acid polymerization should obviously not be subject to acid fracturing
conditions. Caustic floods may be useful. The undesirable viscosity
effects can eventually be overcome by other beneficial reactions (see
Chapter 3) causing cleavage of the polymerized species and reversing the
increase in viscosity.

Reaction with Water (Step 1, Fig. 2)

This is, of course, the initial reaction between the high temperature
water and the organosu phur I
compounds which we believe to be the most
sensitive components of the heavy oil as far as aquathermo ys I
i s is

concerned. While details of this aquathermo ys s I

i step are only poorly
understood, there is little doubt that it takes the form of hydrolysis of a

carbon-sulphur bond and may or may not involve release of H 2 S. There are
two important features of this step.

The first is that the bond scission counters the negative effects on
viscosity of the polymerization step either by reducing molecular weight or
by changing the shape of the molecule which can also markedly affect
 - 12 -

viscosity. This is probably a contributing factor to the viscosity

maximizing effect seen in some samples we have studied (see Chapter 3).

The second important feature of this step is that it produces

intermediates (aldehydes) which can decarbony late to produce carbon

monoxide. The transient appearance of CO as a component of the gas phase

(see Chapter 6) immediately suggested that it is being reconsumed in the

overall process. This has led us to the identification of the involvement

of the Water Gas Shift Reaction (WGSR) and- subsequently to our filing a

patent for CO as a steam additive.

The aquathermo lysis reaction leading to cleavage is subject to metal

ion catalysis likely through the complexing of the metal ion with the

sulphur centre thus activating this moiety to attack by water. The metal
ion may well be hydrated thus also assisting in carrying the water molecule
into the reaction site. The importance of these steps is discussed in

Chapter 7.

The Water Gas Shift Reaction (Step 3, Fig. 3)

The WGSR has only recently been shown to occur at temperatures below

400°C. The key is the presence of suitable catalysts in order to speed the
reaction which is thermodynamically perfectly viable down to temperatures
as low as 20°C! We have demonstrated that the WGSR does indeed occur under
our reaction conditions at temperatures in the 200°-300°C range (see
Chapter 6). In the absence of added metal salts as catalysts there is

evidence to suggest that the mineral content of the sand contains

sufficient appropriate (e.g. vanadium) catalysts to promote the WGSR at
these lower temperatures. The fact that some of the metal content of the
oil sands is known to exist in complex forms with the organic matrix
suggests an even better form of WGSR catalyst may pre-exist in the deposit.
Our work with added CO leaves little doubt that abundant hydrogen can be
generated in-situ by the WGSR and, again, added metal salt catalysts can
enhance this important source of oil upgrading reagent - hydrogen.
 I

-13-

C0 2 Release from Porous Medium (see Step 4 , Fig. 3)

This reaction relates to aquathermo lys s i of the mi nera component of

the oil sand as opposed to the oil component. It can be a dominant source

of C0 2 in certain cases when reactive mineral carbonates (e.g. si derite)

are present. Copious amounts of C0 2 can be produced which provide
additional i n-s tu pressure
i drive but also tend to acidify the aqueous

system thus encouraging acid polymerization (Step 2, Fig. 1). Large

amounts of C0 2 from mineral carbonates .can also limit the WGSR by
displacing the equilibrium back to the CO and H2 0 side thus reducing
hydrogen production.

Hydrodesulphurization (see Step 5, Fig. 3)

The hydrogen produced by WGSR is of little value if it cannot be made

to react further with the heavy oil and thus upgrade the fluid. It is here
that the role of added metal ion catalysts can be critical as we have
demonstrated in results presented elsewhere in Chapter 7. This is
classical catalyzed hydrocracking at low temperatures, in aqueous medium,
in-situ. It reduces viscosity, adds hydrogen to the oil, removes sulphur
(as H 2 S) ,
but, perhaps best of all from an economic standpoint, occurs in

the cheapest high pressure reactor of all, the formation.

H 2 S Production and Scavenging (see Step 5/6, Fig. 3)

Production of H2 S as a component of the off gases during steam

stimulated recovery of heavy oils is well established. We have prepared
and circulated (1981) a special report on H 2 S production during steaming.
There is little doubt that aquathermo lys is of the organosu phur components I

of heavy oils is a principal source of H 2 S. Aquathermo lys is of pure

thiolane and pure thiophene produces mixtures of H 2 ,
C0 2 ,
CH 4 and H 2 S, the
same mixture as obtained from heavy oils (Chapter 2). We have noted,
 -14-

however, that use of certain metal salts as catalysts for the various
component reactions of aqua thermo ys s I i results in a significant reduction
in released H 2 S (Chapter 7). This we suspect is due to the scavenging of
the H 2 S by the metal to form metal sulphides. While these will be subject
to hydrolysis, releasing H2S again, a substantial amount of the H 2 S
available will still be tied up as metal sulphide in the hydrolysis
equilibrium. It is also worth noting that metal sulphides are known as
good WGSR catalysts!
 -15-

CHAPIER-2

EVIDENCE FOR AQUATHERMOLYSIS AS A DISTINCT PROCESS

Evidence for the distinct difference in the types of chemical reactions

occurring in the water/heavy oil systems as temperature increases beyond

300°C is presented in Fig. 4(a). In what we have labelled the

aquathermo ys I
i s "window" there is little or no conversion of the liquid
phase to a solid deposit (coking), and gas phase production, although
important and indicative of the ongoing aquathermolysis reactions, is

minimal compared with behaviour above 350°C. In this range thermal

cracking of the liquid phase to yield increasing amounts of gas and solid,
even in the presence of water, indicates the predominant role of thermal
cracking at such higher temperatures . In the aquathermolysis "window"
there is little gross conversion of the liquid phase. The more subtle
aquathermo lyti c reactions bring about molecular changes within the liquid
phase, however, and these are reflected in substantial viscosity variations
but little depletion of the liquid phase.

In Fig. 4(b) the logarithm of the percentage overall conversion of the

original heavy oil or bitumen to gas or solid phase products is plotted
against reciprocal temperature - a crude form of Vant Hoff plot, the slope
of the plot giving an indication of the energy involved in the chemical
conversions that are occurring. The substantially greater slope of the
curve beyond 300°C again demonstrates the change to chemical reactions with
much higher energy requirements than the more subtle, less destructive

processes that characterize the chemistry of aquathermolysis.

We have noted previously the relatively easy scission of carbon sulphur

bonds compared with other chemical linkages in the heavy oils and bitumens
and our proposed chemical mechanism (see Figs. 2 and 3) involves initial
catalyzed attack by water on such carbon-sulphur bonds. Such reactions are
therefore in keeping with the lower temperature, less destructive type of

process that characterizes aquathermolysis. Comparison of the molecular

'
 -16-

species that comprise the gas phase produced on aquathermo lysis of heavy
oil sands with that from model organosu phur I
compounds supports the

hypothesis that the organosulphur species in the heavy oils are the

principal initiators of the aquathermo yt c sequence. I

i

In Fig. 5(a) the typical principal gaseous products from

aquathermo ys s of an Athabasca oil sand

I
i at 300°C are shown as a function

of duration of aquathermolysis. They are seen to be precisely the same

chemical species as those produced from -identical aquathermolysis of

thiolane and thiophene (Figs. 5(b) and 5(c)) representing aliphatic and

aromatic type organosulphur species. The relative amounts are, of course,

different since only a small fraction of the oil sand fluid will be
organosulphur in character compared with the pure model organosulphur

species. This difference also accounts for the fact that production of the
gases approaches a limit in the oil sand cases while it appears that in the

pure model compounds the aquathermolysis is still accelerating with time

after 30 days.

The most important feature of the comparison, however, is the identity

of the chemical species released to the gas phase. It would be fortuitous

indeed if this identity were not a reflection of the similarity of reacting
species in the two cases, i.e. the organosulphur centres in the molecular
structures. The production of C0 2 in the aquathermolysis of the model
organosulphur compounds shows that the aqueous phase must be chemically
involved in the cleavage reaction as there is no other source of oxygen

available in the system. All model reaction systems were carefully

degassed with pure nitrogen before aquathermolysis. The production of
small amounts of hydrogen in the model compound reactions also suggests
that even in these less complex systems the Water Gas Shift Reaction (WGSR)
is occurring, utilizing carbon monoxide produced in the aquathermolysis
cleavage described in Fig. 2. It is very unlikely that there would be any
significant thermal cracking to yield free hydrogen at 300°C. The much
higher hydrogen production in the case of aquathermolysis of the oil sand
suggests that the presence of a wide range of potential catalysts in the
accompanying mineral phase may have enhanced the water gas shift reaction
in this case.
"<& >-9
 .

-17-

CHARHR 3

VISCOSITY AND AQUATHERMOLYSIS

One of the most important properties of heavy oil or bitumen is its

viscosity. Viscosity determines the ease with which the fluid moves in the
reservoir and the manner in which it can be handled and transported on the
surface. The effect of aquathermo lysis on viscosity is therefore of
critical importance. Indeed throughout this- study we have used viscosity

as a benchmark property in examining the effect of variables and additives

on aquathermo lys s i

The viscosity of heavy oils or bitumens that have been subjected to

aquathermo lys is (in contact with host porous media) is dependent on
temperature and time. This is well illustrated in Fig. 6 where viscosity

of recovered fluid is plotted against time of aquathermo lys is at three

different temperatures. The sample was an Athabasca oil sand.

Aquathermo lys is at 200°C results in a steady i ncrease in viscosity. At
300°C, however, the opposite effect was observed and a substantial decrease
in viscosity over a 30 day aquathermo ys s period resulted.
I
i This reversal
of behaviour, however, is not solely dependent on temperature as seen in
the viscosity data for aquathermo ys s I
i at the intermediate 240°C
temperature. Under these conditions viscosity first increases, as at
200°C, but after prolonged aquathermo ys s I
i (60 and 90 days), as would be
characteristic of actual field operations, the viscosity of the recovered
fluid decreases again.

We suggest that this behaviour is a reflection of the interplay between

polymerization and molecular cleavage (lysis) reactions as discussed in
Chapter 1 and illustrated in Fig. 2. The polymerization reactions,
catalyzed by acid conditions caused by initial fast release of C0 2 (see
Chapter 4), are dominant at lower temperatures probably because of the
higher activation energies of the chemical reactions that form the
subsequent steps of the overall aquathermo lys is process (see Fig. 3). At
 -18-

higher temperatures or longer aquathermo lysi s times at intermediate

temperatures, the molecular cleavage reactions become dominant and
viscosities decrease.

We sought supporting evidence for this hypothesis in molecular weight

measurements on the whole fluid. Such average molecular weights varied
between 580 and 630 but did not show any relationship to viscosity. This
does not necessarily invalidate the hypothesis since substantial changes in

molecular structure could be occurring in a small fraction of the total

fluid (e.g. the organosu phur I

species) that had little effect on the
overall average molecular weight but which could significantly affect
viscosity. A good illustration of this is the effect that small residual
amounts of extracting solvent can have on the measured viscosity of heavy
oils and bitumens (Bennion et al).

The viscosity changes may also be more the result of changes in

molecular shape or conformation than of molecular weight. The opening and

reclosing of sulphur linkages in high molecular weight species may well
result in considerable reorganization of the spacial arrangement of

asphaltenic and other large molecular structures. Very recent work on

asphaltenic fractions of aquathermo yzed I
oils using gel permeation
chromatography has shown that there is indeed a substantial change in the
ability of this fraction to permeate the gel structure after
aquathermolysis, despite little change in overall molecular weight,
supporting the hypothesis that change in molecular shape may be as
important a contribution to viscosity variation as change in molecular
weight.

The temperature and time dependence of the effects of aquathermolysis

on viscosity could be of considerable importance in heavy oil and oil sand

reservoir management.

Consider the diagrammatic presentation in Fig. 7. This depicts an

initial steam soak and drive in-situ stimulation scenario, with steam
 -19-

injection at 300°C and bottom hole production well conditions around 260°C.
The temperature range in the active sweep zone will therefore be 260°-

230°C. Most of the fluid in this zone will therefore be exposed, over

time, to conditions that lead to an overall reduction of viscosity as seen

in Fig. 6. However, peripheral to the active sweep zone there will be a

set of temperature contours defining a thermal gradient all the way from
the temperature of the active zone to natural reservoir conditions.
Temperatures at the upper end of this gradient will be in the range where
polymerization and viscosity increase was observed in Fig. 6.

Following completion of the sweep of the original active zone

communication may be established between the injection well and a new
producer along a second radius. In attempting to establish such
communication, however, the fluid in place that is encountered along the
pathway will not be virgin reservoir fluid as was the case in the first
active zone, but fluid that has been exposed to temperatures in the 140°-
200°C range for some period of time. The natural water content, if not
some water breakthrough from the active sweep zone, will permit low
temperature aquathermo ys I
i s to occur which our results indicate may result
in slow polymerization and viscosity increase.

The slow, low temperature aquathermo ys I

i s of the reservoir volume
around the initial active sweep zone may thus generate a higher viscosity
fluid in place and result in a substantially different degree of difficulty
in establishing communication with second and subsequent producing wells.
In subsequent sections of this synopsis we discuss possible methods for
minimizing the viscosity increase observed at lower aquathermo ys I
i s temper-
atures .
 - 20 -

C HAPTER -4

SOURCES OF C0 2 DURING ARUATHERMOLYSIS

Carbon dioxide is a major component of the gas phase produced during

aquathermo lysis. The amount and the rate at which it is produced is

important not only as a measure of the extent of chemical reaction that

produces it but also because it has been shown (Svrcek et al.) that when
dissolved in the heavy oil it can cause significant reduction in viscosity.

In the introductory Chapter of this synopsis we indicated several sources

of C0 2 . Aquathermo lysis of the organosu phur components of the heavy oil
I

of both the thioether (thiolane) and thiophenic type can yield C0 2 ;

mineral

carbonates in the porous (sand) medium can release C0 2 ;

and, as will be

demonstrated in this Chapter, polar carboxylic acids adsorbed on the sand

granules can also be a source of C0 2 as a result of decarboxylation. The
quantitative aspects of these phenomena will be examined below.

C0 2 Production as a Function of Temperature

In some of our earliest work performed on Canadian and Venezuelan oil

sand samples we examined the rate of production of C0 2 as a function of

temperature during aquathermo lys i s of the samples. Illustrative data is

presented in Fig. 8. The two samples used in this work produced

significantly less C0 2 than some of the later samples we have examined but
the work done on C0 2 production on these samples was more extensive both in

terms of time and temperature. They are thus most appropriate as an

introduction to the topic.

The Venezuelan sample produced more C0 2 than the Canadian core material
at all temperatures, but the general pattern of the rate of production is
the same in all cases. There is an initial rapid release of C0 2 to the gas
phase, which we have labelled below the plots as zone I, followed by a

slower and declining rate of production over a much longer time period
 - 21 -

(zone II). We suggest, and will support later in this chapter, that the
zone I behaviour is primarily due to release of C0 2 from mineral carbonates
in the sands probably due to hydrolysis. X-ray diffraction indicated
substantially more carbonate material in the Venezuelan sample although
different carbonates will decarboxy late at different rates and it requires
relatively little reactive carbonate to produce the kind of C0 2

concentrations measured in this work. The amount of C0 2 produced in zone I

(full line) is relatively constant for each sample irrespective of

temperature suggesting that the C0 2 production from the mineral carbonate

occurs quickly and more or less completely at all three temperatures. It

should be noted, however, that at 200°C the zone I behaviour occurs over 7
days, at 240°C over approximately 3 days and at 300°C in less than 2 days.
The decarboxylation of the mineral carbonates, rapid though it may be in

all cases, is still faster at 300°C than at 200°C, as it should be.

After mineral decarboxylation is essentially complete a much slower C0 2

production ensues (dotted line) which we attribute to aquathermo lysis of
the organic constituents of the oil sand, in particular the organosu phur I

compounds and the carboxylic acids. Both the amounts and rates of C0 2
production are highly temperature dependent as would be expected from
kinetic theory. The shape of the dotted curves is in general accord with
first order kinetics where the concentration and reactivity of the organic
substrate controls the rate and the aqueous reactant is present in excess.
The limited amount of reactive organic substrate in the heavy oils
(probably organosu phurs) I
is indicated by the trend toward a maximum in C0 2

production at the highest temperature.

C0 2 From Sand Phase

We have postulated above that the initial rapid production of C0 2 is

due to decarboxylation of the mineral carbonates in the sand. If this is

so then the sand alone should produce an equivalent amount of C0 2 when

aquathermo yzedI
. In order to check this postulate we chose a Venezuelan
 1 .

- 22 -

sample known to produce substantial quantities of C0 2 . Aquathermo lysis of

the whole core sample produced C0 2 as indicated in Fig. 9 again showing the
rapid initial production of C0 2 ,
followed by a slower release of C0 2 .

A sample of the same homogenized core was then extracted with methylene

chloride and alcoholic sodium hydroxide to generate an organic free sand.

An amount of this clean sand equivalent to that present with 1,000 g of
bitumen in the whole core was then aquathermo y zed under precisely the same I

conditions as the whole core. The amount of C0 2 produced was 21,720 mis as
indicated in Fig. 9. The initial rapid production of C0 2 is thus likely
associated with hydrolysis of the mineral carbonates in the sand as

postulated

C0 2 From Carboxylic Acids

It has been argued (Montgomery et al) that decarboxylation of

carboxylic acids is a significant source of C0 2 in steam treatment of oil

sands. We have examined this possibility by extracting a core sample with
CH 2 C 2 ~ a relatively non-polar solvent in which most of the organic phase
1
is soluble.

Polar carboxylic acids, strongly adsorbed onto the sand phase, however,
may not be removed. Thus such sand extracted with CH 2 CI 2 may indeed still
contain significant quantities of carboxylic acids. When subjected to
aquathermo ysi s at 240°C for 14 days
I
it yielded 350 mis of C0 2 per kilogram
of sand. Another sample was then washed with methylene chloride and 1 m
alcoholic NaOH which would remove any adsorbed carboxylic acids. Infrared
analysis of the recovered extract showed evidence of carboxylate groups

1. Methylene chloride has been claimed to be more polar and thus remove
more polar material from the sand than benzene or toluene. While this
may be true in relative terms methylene chloride is not significantly
different from the aromatics when compared to alcoholic NaOH as a
sol vent.
 s

-23-

-1
(1710, 1440, 1360 and 1020 cm ). Sand so extracted was again subjected to
aquathermo lysis and the C0 2 production per kilogram of this sand was 160

mis compared with 350 mis without caustic ethanol wash.

The C0 2 production from CH 2 CI 2 extracted sand thus drops by more than

50% upon further extraction with NaOH. This suggests that the carboxylic
acids present in sand are retained on the surface of the CH 2 CI 2 extracted

sand and are a significant source of C0 2 as a result of decarboxylation

under aquathermo lysis conditions.

C0 2 Sources - An Attempt at Mass Balance

If C0 2 is generated from both the mineral and organic components of the

whole oil sand it should be possible to perform three aquathermo ys I
i

treatments on - a) the whole core, b) the separated bitumen, and c) the

separated sand, and obtain a mass balance with respect to C0 2 produced in

the three aquathermo lyses . Our results from this experiment are shown in

Table 1. Two separated oil sand samples were examined with substantially
different C0 2 production, especially from the mineral phase.

It is seen that in both cases the total C0 2 produced from the separated
sand plus that from the bitumen exceeds that from the whole core by 33% in

sample 1 and 19% in sample 2. We suggest that this apparent lack of mass
balance is due to the fact that in the whole core the mineral carbonates in

the sand are "protected" from attack by high temperature water to some
extent by the surface layer of bitumen and the absorbed carboxylic acids.
Once these are stripped off by solvent extraction and the clean sand is

aquathermo yzed I
the mineral carbonates are subject to more direct
aquathermo lysis and produce more C0 2 .

The CO data in Table 1 provide a good illustration of the important

catalytic role played by the minerals in the sand in the WGSR conversion of
CO to C0 2 . When the separated bitumen is aquathermo yzed substantial
I
 n

-24-

quantities of CO are detected. This suggests that while the minerals i

the sand may not be critical catalysts for the initial cleavage of the
organosu phur species which eventually
I
results in CO production, they are
essential in the subsequent conversion of CO to C0 2 . The very much lower
CO levels detected in the gas phase produced on aquathermo lyz ng the whole i

core (with sand present) demonstrates the effective WGSR catalytic role of
the minerals in the sand.

The diagramatic presentation below Table, 1 attempts to summarize the

sources of C0 2 as we have attempted to justify above. The mineral sand
particles are believed to be surrounded by a thin layer of absorbed water
in which we suggest the polar - COOH end of the high molecular weight

carboxylic acids (humics, fulvics etc.) may be absorbed and thus linked to
the polar sand minerals. When this sand, stripped of its oil/bitumen
cover, is aquathermo yzed
I
the carboxylic acids (RCOOH) and mineral
carbonates (MC0 3 ) will be the main source of C0 2 through decarboxylation.

When the oil sand is "whole" and the oil/bitumen is in place, the low
polarity hydrocarbon ^tails" of the adsorbed carboxylic acids (and other
polar organics) can act as a transition medium between the oil and the sand
particle by interacting with the low polarity oil phase. The adsorbed
carboxylic acids thus form from the bridge between the polar mineral phase
and the non-polar oil acting as a sort of a "reverse soap" enabling the

polar sand to "dissolve" in the non-polar oil. In such a situation

aquathermo lysis acts on both the oil and the sand producing C0 2 from both
sources but with reduced efficiency from the sand because of the adsorbed
oil layer. The presence of the sand minerals in close proximity to the
aquathermo lyz ng oil enables the minerals
i to act cata lyti ca I
ly in the WGSR
conversion of the CO produced by aquathermo lysis of the oil to yield C0 2
and hydrogen.
 .

-25-

CHAPTER 5

THE GENERATION AND CONTROL OF H 2 S

DURING STEAM STIMULATION (AQUATHERMOLYSIS)

The generation of H 2 S during steam flood/drive operations in heavy oil

deposits appears to be the consequence of chemical reactions of the

organosu Iphur components of the heavy oil. While metal sulphides in the
sand phase may, on occasion, be a source ef H 2 S such components of the

porous medium are likely to be more important as catalysts in H 2 S

generation. Thus, it does not appear to be possible to avoid the

generation of H 2 S at temperatures that are of any value in steam treatment
(i.e. injection >200°C) when bitumen with substantial organosu phur content I

is involved.

There can be little doubt that, apart from the nature of the
organosu phur constituents of the
I
heavy oils, the parameters of time and
temperature are the most important variables determining the generation of
H2 S in steam treatment of oil sands. We deal with them together because
they are so intimately connected both in the laboratory simulation and in

the field operations.

We have shown by separate aquathermolysis of extracted sand that the

contribution of H 2 S from the porous medium is not a major factor in these

cases. No H 2 S whatever was generated from aquathermolysis at 240°C of the
sand from the Venezuelan sample illustrated in Fig. 10 and a Canadian
sample sand generated some 2% of the total H 2 S generated when the whole

core sample was aquathermo yzed I

. We believe this was due to small amounts
of iron sulphide in the sand. Interestingly, when the separated bitumen
from both samples was aquathermo lyzed in the absence of sand and any added
catalyst no H 2 S was generated. This suggests that the minerals in the sand
have an important catalytic role in hydrodesulphurization when the whole
core material is aquathermo lyzed
'
 . .

-26-

The relevant data for H 2 S generation are plotted in Figs. 10 and 11 for

Venezuelan and Canadian samples respectively. The vertical axis of these

plots is a logarithmic scale of the H2 S production expressed in mis
(STP)/1000 g bitumen and the horizontal axis is aquathermo lysis time in

days. The logarithmic scale was chosen for convenience of data

presentation in a manner that would permit comparison, but there may well
be theoretical grounds for using such a log (concentration) format for
expressing the rate of production of H 2 S.

It is immediately clear that the generation of H 2 S at 240°C is rapid

and that it reaches what is effectively an equilibrium value within a few

days. Comparison of the full line (Haste I I

oy vessel) 240°C plots with the
corresponding dotted line (quartz tube) demonstrates that although both
reach the same equi I ibr ium value the rate in Haste I I
oy vessel is very

significantly greater. Thus, we have clear evidence of a catalytic role in

hydrodesulphurization for the sulphided interior walls of the Hastelloy

vessels. This should be compared with the effect on C0 2 production

discussed in Chapter 9. The relatively rapid rise of H 2 S content in the
gas phase has, however, been observed under field conditions in Venezuela
where initial values in the range of 20 ppm had risen to some 6,000 to

10,000 ppm within a few months and remained there for several years.
Although the time scale is different, as are the volumes involved, the
relatively rapid rise to an equilibrium value and maintenance of that value
for a long period of time does appear to be characteristic of field
cond t ons
i i

It is seen that the same pattern of rate behaviour occurs at 300°C and
200°C as occurred at 240°C in the Hastelloy vessel. The difference is the
equilibrium H 2 S production reached at each temperature. Our best estimate
of these values is shown at the extreme right hand end of each curve in
Figs. 10 and 11

Two points should be noted when comparing the Canadian (Fig. 11) and
Venezuelan (Fig. 10) curves. The behaviour at 200°C and 240°C is very
 '

,
 -27-

similar as far as H2 S production from equal amounts of bitumen are

concerned. This is not the case when C0 2 production is compared (see
Fig. 24, Chapter 9). However, at 300°C the amount of H 2 S produced from

1000 g of Canadian bitumen exceeds that from 1000 g of Venezuelan by a

factor of two. The rate of achieving equilibrium concentration in the

Canadian case is also considerably slower than in Venezuelan especially at

200°C. What do these two observations mean?

We have noted previously (Chapter 1) that at 300°C thermolysis as well

as aquathermolysi s is occurring. This means that there will be less

selectivity as to which organosu phur compounds in the heavy oil or bitumen

I

release H 2 S and the rate of H2 S production from Canadian and Venezuelan

samples will more nearly reflect the ratio of available organosu phur in I

the heavy oil. The Canadian sample contains 5.055 sulphur and the
Venezuelan sample 2.355: a ratio of 2.17. The equilibrium amount of H 2 S
from aquathermolyzing the Canadian sample at 300°C is 1400 ml s/1000 g
bitumen and from the Venezuelan sample 620 ml s/1000 g bitumen: a ratio of

2.26! Bearing in mind the approximations and the averages involved in such
an analysis we will say nothing more than that at 300°C the H 2 S production
appears to reflect the total sulphur content of the bitumen!

The somewhat slower rate at which the Canadian sample reaches an H 2 S

equilibrium value at 200°C compared with the Venezuelan case is supportive
evidence for the claim that the organosu phur I
compounds in the Canadian
sample are more resistant to aquathermo lys s i than are the organosu phur I

compounds in the Venezuelan sample. We have commented elsewhere that the

proportion of the more readily aquathermo lyzab e thioether type compounds
I

was greater in the Venezuelan than in the Canadian sample.

Comparison of laboratory generated data with field data is always

fraught with uncertainty because of the number of uncontrolled variables
that may be operating in the field situation. We have nonetheless been
able to compare our laboratory data on H 2 S generation with field data for

the Venezuelan sample. The data is plotted in Fig. 12. The field H 2 S
 -28-

production is plotted against bottom hole (formation) temperature together

with the higher temperature laboratory simulation data from this work. The
lab data fits almost too well on the upper end of the field data thus

lending credence to the applicability of our laboratory model to field

conditions at least in the case of this sample and within the limits of
accuracy claimed for the data. The curvilinear nature of the relationship
in Fig. 12 is expected since there is obviously an upper limit to H 2 S
production from the heavy oils irrespective of temperature.

Temperature and time can obviously be used to control the amount of H 2 S

produced during steaming although the use of time is likely to be severely
circumscribed by other operating factors. If satisfactory production can

be achieved at lower temperatures, however, it might be economically

preferable to do so rather than seek higher production at higher
temperature at the cost of having to handle more produced H 2 S if the
reservoir fluid was one which more readily produced H 2 S. Control can also
be affected to some extent by use of added metal salts which form sulphides
and trap H 2 $. This feature is discussed in greater detail in Chapter 7.
 -29-

CHAPTER 6

CARBON MONOXIDE IN THE AQUATHERMOLYSIS PROCESS

Carbon monoxide (CO) has been proposed (see Figs. 1 & 2) as an

important intermediate in the aquathermo lysis process. In this Chapter we

present evidence to support this proposal and describe how the

identification of the involvement of CO in aquathermo lysis led to a patent
for enhancing the beneficial effects of aquathermo lysis by addition of

carbon monoxide during in-situ steam stimulation.

The transient nature of the carbon monoxide component of the gas phase
produced during aquathermo lysis is clearly seen in Fig. 13. The behaviour
of two quite different oil sand samples is illustrated; that in Fig. 13(a)

being a Canadian Athabasca sample and in Fig. 13(b) a Lagunillas (M-6)

Venezuelan sample. The initial rapid rise in the amount of CO in the

produced gas phase is in accord with the generation of CO by

decarbony at on of the
I i intermediate aldehydes resulting from the initial
steps in the aquathermo lysis process (see Fig. 1). The produced CO,
however, is the feed for the water gas shift reaction (WGSR) and thus
subsequent depletion of CO in the gas phase is observed. Although we do
not wish to over nterpret
i the quantitative aspects of this data it is
noteworthy that the amount of CO generated is greater at the higher
temperature as might be expected from the enhanced aquathermolysis
occurring at 300° C compared with 200° C.

The WGSR has recently been shown by Shore et al. to occur at relatively
low temperatures if the appropriate catalysts are present .

Thermodynamically the WGSR equ i I i br i urn predicts significant quantities of

hydrogen and carbon dioxide will be present in the 200°-300°C
aquathermolysis window temperature range.

cata lyst
s
co + h 2 o co 2 + h2

consumed in hydrocracking etc.

 -30-

This equilibrium will continue to be displaced to the right if the hydrogen

is consumed in other "upgrading" reactions. As this occurs more C0 2 will

be produced and thus the ratio of C0 2 /H 2 in the gas phase resulting from
the WGSR will not necessarily be 1:1. In the reverse direction, of course,

any other source of C0 2 such as decarboxylation of mineral carbonates will

tend to suppress the WGSR and minimize the production of hydrogen in-situ
and reduce the beneficial upgrading reactions resulting from hydrogen
uptake. These factors thus suggest that the nature of the porous medium
and in particular its mineral carbonates may have a significant effect on
the role played by aquathermo lysis during in-situ steam stimulation.

C0 2 Production from Added CO

The effect of adding various amounts of CO to the aquathermolyz ng oil i

sand system is shown in Figs. 14 to 17. In these graphical presentations

we examine the effect on production of gas phase components and interpret

these effects in terms of the proposed chemical mechanisms of

aquathermolysis. In Fig. 17 we examine the effect on the all important

property of viscosity of the recovered liquid phase.

In Fig. 14 the effect of adding various overpressures (psi) of CO to

the aquathermolysis vessel on C0 2 production is shown. CO additions are

also expressed in terms of weight percent of the total system. In the

cases of both the Venezuelan and Canadian oil sand samples C0 2 production
rose markedly as increasing amounts of CO were added, reaching or exceeding
100,000 mis per kilogram of oil /bitumen content in the presence of some 1

wt % added CO. This observation leaves little doubt that the added CO is
being converted to C0 2 via the WGSR in the aquathermolysis process.

It is interesting to note that although the added CO is essentially the

same in both systems the extent to which it is converted to C0 2 in the
presence of the two oil sand samples is quite different. Although, as
described elsewhere, we have carried out separate experiments to check the
'
 .

-31-

possible catalytic effects of the walls of the Haste I I

oy vessels used in
this work (see Chapter 9) the differences in C0 2 production seen in Fig. 14
can only be ascribed to the difference in oil sand present. The vessels
and all other conditions in the experiments were identical. Thus the fact
that the WGSR appears to be less efficient in the Canadian oil sand case
suggests that the mineral components (or perhaps organometa I I ic species)

present in the Canadian oil sand sample are less effective WGSR catalysts
than those in the Venezuelan sample.

The field operating significance of this observation is that the extent

to which desirable hydrogen may be produced in-situ as a result of WGSR

conversion of added or aquathermolytica I

ly generated CO will depend to some
extent on the catalytic activity of the minerals present in the porous
medium of the reservoir or the metals incorporated in the organometa I I ic

structures of the heavy oils and bitumens. It was, in part, for this

reason that we subsequently examined the effect of adding potentially

catalytic salts to the aqueous phase as is described in Chapter 7 of this

synops s i

Hydrogen Production through Added CO

The C0 2 production from the WGSR must be accompanied by hydrogen

production. The hydrogen production data for aquathermolysis of both
Venezuelan and Canadian oil sands at 240°C in the presence of various
amounts of added CO is plotted in Fig. 15(a) and 15(b). An immediate
difference compared with C0 2 production is seen. More hydrogen appears in

the gas phase in the Canadian case than in the Venezuelan sample and,
probably more important, the hydrogen concentration in the Venezuelan case
passes through a maximum clearly suggesting that over longer reaction times
it is being consumed in the system faster than it is being produced. Since
less C0 2 was produced in the Canadian case than in the Venezuelan case when
similar amounts of CO were added it would have been expected that less
hydrogen would also have been produced and appear in the gas phase. The
 -32-

fact that this is not so must mean that, once again, the mineral phase in

the Canadian sample is cata lytica I

ly less active than the Venezuelan sand
but this time in reactions that consume hydrogen, i .e. the all important

upgrading reactions. Thus less of the produced hydrogen is consumed in the

Canadian case and the amount remaining in the gas phase is greater than in
the Venezuelan case.

The relative extent of consumption of hydrogen produced by the WGSR can

be demonstrated by comparing the volumes of -C0 2 and H 2 in the gas phases
after 30 days aquathermo lysis at 240°C with 155 added CO. The Canadian
sample has a gas phase containing approximately 93,000 mis of C0 2 per 1000

g of bitumen in the oil sand which, when corrected for the 7,000 mis of C0 2
that is produced in the absence of added CO yields 86,000 mis of C0 2
generated by WGSR. A similar calculation for hydrogen shows that of the
original 86,000 mis of H 2 that must have been produced in the WGSR together
with the 86,000 mis of C0 2 approximately 75,000 mis of H 2 remain in the gas
phase. Thus some 9,000 mis of hydrogen appear to have been consumed.

In the Venezuelan sample, however, 92,000 mis (124,000-32,000) of C0 2

and thus the same amount of H 2 appear to have been produced by WGSR of the
added CO but only 5,000 mis of H2 remain in the gas phase after 30 days
aquathermolysis at 240°C. Thus it appears some 47,000 mis of hydrogen were
consumed in other reactions, some five times as much as in the Canadian oil
sand case. These numbers are at least strongly supportive of both the role
of the WGSR in aquathermolysis and the important function of appropriate
catalysts in the mineral phase in utilizing the hydrogen produced by the
WGSR.

This internal consumption of the hydrogen produced as a result of

addition of CO and its participation in the WGSR is an extremely important
phenomenon. If it can be optimized by appropriate adjustment of reaction
conditions, catalysts etc., then i n-si tu upgrading of the heavy oil by
in-situ consumption of in-situ generated hydrogen will be accomplished.
This could result in easier recovery of a higher quality oil. Do we have
 -

.
 -33-

any additional evidence for this apparently enhanced internal consumption

of hydrogen especially in the Venezuelan case?

In both the Canadian and Venezuelan cases the relative amount of light
hydrocarbons (C x ->
C4 ) produced increases significantly as the amount of CO
added increases. This suggests enhanced hydrocracking of the bitumen due
to the availability of WGSR generated hydrogen. Hydrogen sulphide
production is also indicative of enhanced internal consumption of hydrogen.

Hydrogen Sulphide Production through Added CO

More dramatic perhaps is the effect of added CO on the rate of hydrogen

sulphide production as seen in Fig. 16. This observation is particularly
important because production of H 2 S, although something of a problem as far
as handling the produced gas is concerned, is a direct measure of the
amount of in-situ desulphurization that is occurring as a result of
aquathermo lys i s . Thus, whether the interest is in knowing how much H 2 S
will have to be handled in the produced gas or in achieving maximum in-situ
desulphurization, the results in Fig. 16 and the reasons for them are of
considerable practical importance.

The greater rate of H2 S production in the Venezuelan sample is in

accord with the higher hydrogen consumption and the greater

hydrodesu phur
I
i z i
ng catalytic activity of the minerals in the Venezuelan
sample. A further important factor affecting the ease of desulphurization
and H 2 S formation, of course, is the relative reactivity toward hydrogen of
the organosu phur I
components of the two oil sand samples. There is

evidence to suggest that the Venezuelan sample which we used had a much
higher thioether or thiolane type organosu phur content than the Canadian
I

sample which was rich in thiophenic sulphur. Reference to Fig. 5 in

Chapter 2 will show that the thioether/thiolane type organosu phur species I

are much more aquathermo lytica I

ly active than thiophenic species.
i
 -34-

A though the production

I of hydrogen rich gases is evidence of the
consumption of hydrogen produced in aquathermo lysis the most desirable type
of consumption would be the incorporation of the hydrogen into the liquid

phase. We will discuss the molecular properties of the liquid phase

following aquathermo lysis in another chapter, but we turn now to an

examination of the viscosity of the liquid phase following aquathermo lysis

and the effect of the presence of added CO.

Effect on Viscosity of Added CO

It is first necessary to examine the effect of aquathermo ysi s on the I

two samples without addition of CO.

In the Venezuelan case we have an example of an oil sand that is more

susceptible to polymerization and the viscosity rises initially from 400

cent poise (at 85°C) to over
i
600 cps before dropping again as beneficial
cleavage reactions of aquathermo lysis begin to dominate. The interplay of
these two factors was discussed in Chapter 3. Although the viscosity data
is somewhat scattered (see Chapter 9 on sample reproducibility etc.) there

can be little doubt that the presence of CO reduces the height of the
viscosity "hump". We suggest that this is due to the fact that the added
CO enhances the cleavage reactions characteristic of aquathermolysis by

increasing the amount of hydrogen available to participate in these

cleavage reactions and competing with the opposite effect of the more

favoured polymerization. The net effect is to reduce but not eliminate the
viscosity "hump". At aquathermolysis temperatures above 240°C we would
predict that the "hump" would disappear.

The presence of CO in injected steam might thus help to minimize the

undesirable thermal maturation effects in the low temperature regions
around the main production zone as discussed in Chapter 3 and Fig. 7.
 s

-35-

The generally greater reactivity of the Venezuelan sample as reflected

in catalytic activity of the mineral phase and ease of desulphurization is

in keeping with the apparent dominance of initial polymerization as seen in

the viscosity behaviour in Fig. 17(a). Correspondingly, the lesser

reactivity of the Canadian sample used in this particular work (different

from that used in Fig. 6) is reflected in the absence of a significant
viscosity hump in Fig. 17(b), even for oil recovered from aquathermolys i

without CO addition at 240°C. The addition of 0.5 or 1.0 wt % of CO

reduces the viscosity of the recovered liquid- oil phase by some 20% to 400
centipoise or lower from an initial value of over 500 centipoise. While
this is not a dramatic reduction it is in the desired direction and is
achieved with a sample that appears to have a relatively low catalytic
activity as far as the desirable aquathermo lytic cleavage reactions are
concerned. Indeed the delay in the effect of added CO on the viscosity may
be yet a further illustration of the limited overall catalytic activity of
the minerals in the Canadian sample.

The CO Patent

The work on the effect of added CO has led to the issuing of patent
coverage in the US, Canada and Venezuela, under the title "The Use of
Hydrogen-Free Carbon Monoxide with Steam in Recovery of Heavy Oil at Low

Temperatures" (Canadian Patent #1,170,044, US Patent #4,487,264). The

claims cover injection of CO either separately or simultaneously with steam
and the possibility of using partial combustion of high sulphur coke (as

from synthetic crude plants) as a source of carbon monoxide. The presence

of small amounts of sulphur compounds in such a CO feed would not adversely
affect downhole use and these undesirable sulphur components might well be
scrubbed out in the reservoir and not reappear in the produced gases. The
patent recognizes the 1979 US Patent #4,157,462 issued to J.C. Allen and
Texaco and demonstrates the essential differences between it and the patent
based on our work.
.
 -36-

CHAPTER -?

THE ROLE OF METAL IONS IN AQUATHERMOLYSIS

The important catalytic role played by metal ions (M*) in the various
proposed component reactions of the aquathermo lysi s process has been

clearly indicated in Figs. 2 and 3 of Chapter 1. The ions are believed to

be involved in the initial cleavage of the organosu phur bonds,
I
in the
water gas shift reaction and in the hydrocracking or hydrodesulphurization
steps, among others. While the porous medium or sand component of heavy
oil sand core samples that we have aquathermo yzed undoubtedly contain
I

cata lytica I
ly active quantities of many metal ions, the question obviously

arises as to the possible effects of added metal ions on the

aquathermo lytic reaction products and, more importantly, the viscosity of
the recovered heavy oil. These aspects of aquathermolysis are examined in

this Chapter.

Effect on Hydrogen Production - The Water Gas Shift Reaction (WGSR)

The effect of added metal ions on the WGSR should be most clearly
indicated by the corresponding effect on the hydrogen production since this
is one of the two main products of the WGSR. Hydrogen is a preferable
measure of any such effect over carbon dioxide, the second main product of
the WGSR, because of the complications that arise due to other major
sources of C0 2 production in the system (see Chapter 4). Using hydrogen as
a measure of the effect of metal ions on the WGSR is not without its own
complications since the degree to which hydrogen may be reconsumed in other
upgrading reactions will also affect the amount detected in the product gas
phase. Nevertheless, we believe that the general trends in hydrogen levels
measured in the gas phase following aquathermolysis are valid indicators of
the effects of added metal ions.
 i

-37-

In Fig. 18 we present such data for a variety of added metal ions and

compare the hydrogen production with that for aquathermo lysis using water
alone. The concentration of the added metal ion salt was held constant
1
throughout (1CT Molar) and aquathermo lysis conditions were the same in all

cases. The same Venezuelan oil sand sample was used throughout. It is

seen that the amount of hydrogen observed in the product gas phase does
indeed vary considerably as the nature of the added metal ion varies and in

most instances is substantially greater than that observed for

aquathermolysis with water alone. This observation leaves little doubt

that the metal ions are indeed involved in the aquathermolysis process and
more particularly in the WGSR. The effect of the concentrations of salts
added further supports the suggestion that the primary role is catalytic.
Similar although less dramatic results have been obtained using Athabasca
oi I sand samples.

Effect on H 2 S Production

If metal ions have a catalytic effect on the WGSR it is very likely

that they will affect hydrodesulphurization and hence H 2 S production. The
data in the bar graph in Fig. 19 illustrate this effect. Like hydrogen
production, however, the effect is not a simple one and requires
explanation. In every case except aluminum and titanium the amount of H 2 S
in the produced gas phase is lower than that observed for aquathermolysis

with water alone. If the metal ions were having a positive catalytic
effect on hydrodesulphurization and hence H2 S production it would be
expected that H 2 S production would be h gher especially in cases where the
metals also increased the amount of hydrogen available by catalyzing the
WGSR. The only cases where this is observed is with aluminum and titanium
and this is because these two metals are characterized by having unstable
metal sulphides. All of the other metals have relatively stable metal
sulphides which form by reaction with produced H2 S and reduce the free
gaseous H 2 S levels in the produced gas mixture (see Fig. 3, Chapter 1). It
is also interesting to note that the sulphides of these metals are known to
 -38-

be particularly good catalytic species in many of the types of reactions we

consider to be important in overall aquathermo lysis.

Although the formation of the stable metal sulphides complicates the

interpretation of H 2 S production as a measure of the catalytic role of the

added metal ions, it provides important evidence of a very beneficial

secondary effect of adding these metal salts - the reduction of the

concentration of H 2 S in the gases produced due to aquathermo lysis! As is
demonstrated in Chapter 5, H 2 S concentrati ons-of up to 20,000 ppm or higher
can be generated in the gas phase produced by aquathermolysis. This is

well beyond lethal limits (less than 1,000 ppm) and certainly high enough
to require at least flaring and possibly sweetening. Thus the use of added
metal ions not only has a beneficial catalytic role but also serves as a

partial in-situ H2 S scrubbing mechanism. Enhanced desulphurization to

produce the H 2 S corresponds, of course, to in-situ upgrading.

Concern has been expressed about the possibility of metal sulphide

deposition in the formation resulting, eventually, in loss of porosity

between the sand particles or cementation of the inorganic sand matrix

especially around bottom hole. We have not observed this in our small

laboratory samples. Metal sulphides can be detected on the sand grains and
occasionally in the separated bitumen but the amounts have always been low
enough to give reasonable grounds for believing that, at the added salt
levels that would be economically feasible, there is not much likelihood of
formation plugging.

Effect on Viscosity

Like the effect of added CO, the effect of added metal ions on

viscosity of the recovered bitumen or heavy oil is also a beneficial one.

The relevant results are shown in bar graph form in Fig. 20 and are quite
dramatic in the case of the Canadian sample. The fact that it is the
Canadian sample that shows big effects of added metal ion on viscosity,
 -39-

while it was the Venezuelan sample that showed the largest effects on
hydrogen production, is a good illustration of the complexity of these
various features of aquathermo lysis and particularly the complex manner of
their interaction.

In Fig. 20(a) the substantial reduction in viscosity below both the

value obtained after aquathermo lysis with water alone and the viscosity of

the heavy oil in-situ before steaming (unreacted) is clearly seen. In the
2+ =
case of Fe and Mo0 4 the reduction is well over 6055 and indeed the oils

recovered from these catalyzed aquathermo lyses flow readily under gravity
compared with the tarry, viscous flow behaviour of both the original
unsteamed bitumen and that aquathermo yzed I
without added salts. The
beneficial effect of added salts in the steam stimulation of this

particular Canadian heavy oil sand sample is thus substantial and indeed
might produce, with some metals, a pipelineable well head product.

The dependence of effect of added metal ion on sample type, however, is

equally dramatically demonstrated in Fig. 20(b) where the corresponding

viscosity values for a Venezuelan sample are shown. This bar graph is a

very different one. Indeed in most cases the viscosity of the bitumen
recovered after aquathermo lys s with added i metal ion is greater than that
of the unsteamed (unreacted) value and in the case of vanadyl the viscosity
is actually higher than that for the oil recovered after steaming with
water alone.

The reason for this substantial difference in behaviour is, we believe,

related to the greater reactivity of the Venezuelan sample and, in

particular, to its greater sensitivity to acid catalyzed polymerization

(see Step (2) in Fig. 2, Chapter 1). This sensitivity was illustrated and
discussed in Fig. 17 in Chapter 6. In the absence of any additives the
viscosity of the Venezuelan sample first increases with time of

aquathermolysis due to polymerization compared with the Canadian sample

which showed little tendency toward polymerization and hence increase in
viscosity when aquathermo yzed at 240°C. I
'
 -40-

The sensitivity of the Venezuelan sample to acid catalyzed

polymerization is even more clearly seen in the effect that added vanadium
ion has on viscosity. The acidity of the solution of the vanadyl ion is

approximately pH=4 compared, for example with pH 9-10 for the manganous and
molybdate ion solutions. The catalytic effect of the ac d tv of the
i i

vanadyl solution on the polymerization outweighs any other effects that

this ion has on other steps in the aquathermolyti c process and an increase
in the viscosity results. In order to check this hypothesis the vanadyl

catalyzed aquathermolysis was carried out again in the presence of 1 molar

caustic soda. As is seen in Fig. 20(b) there is a complete reversal of
effect on viscosity. This we believe is due to the elimination of acidic
catalysis of the polymerization reactions to which the Venezuelan sample is
particularly sensitive and possibly a beneficial catalytic role by one of
the vanadate species on the aquathermolysis cleavage reactions which lower
the viscosity of the liquid phase. The generation of basic steam
stimulation conditions may therefore be desirable in reservoirs where oils
or bitumens are sensitive to acid polymerization during aquathermolysis.

All of the other ions shown in Fig. 20(b) have some beneficial effect

on the viscosity compared with that of the recovered bitumen sample

following aquathermolysis without added salt. In other words they help
reduce the viscosity "hump" discussed in Chapter 6. Only the manganous,
nickel and molybdate ions, however, actually reduce the viscosities below
the original "in place" value for the heavy oil and, as noted above, we

believe this is in large part due to their basic characteristics and

suppression of polymerization to which this particular sample is prone.

These results indicate that the aquathermolytic activity of each

particular heavy oil or oil sand reservoir is an important parameter in the

design of production facilities and operating procedures. Although we have
compared two geographically widely separated samples in this discussion
other more recent work on Canadian samples that are much more proximate and
indeed in some instances from different depths in the same hole show
substantial differences in activity. The different acid sensitivity of the
I

I
 .

- 41 -

Canadian and Venezuelan samples discussed here is an alert against

injecting acidic fluids into the sensitive reservoirs and raises such
questions as the sensitivity of heavy oils in the Carbonate Triangle of
North Central Alberta to aquathermo lysis if substantial amounts of acidic
C0 2 are generated during steaming. Addition of salts, on the other hand,

might help suppress the undesirable viscosity effects of slow low

temperature thermal maturation around the main sweep zone as discussed in

Chapter 3 .

Such studies of the variability of activity of oil sand samples from

different locations is the major thrust of our current ongoing work and
part of this research protocol involves the study of the beneficial effects
of added ions. We have chosen the molybdate ion as the most likely one to
optimize these effects for a variety of economic and chemical reasons.
Although ferrous ion appears to be almost as good in the case of the
Canadian sample it is very difficult to keep in the ferrous state; the
ferric ion we have shown to be less effective. The molybdate ion, while
more expensive, is stable and quite basic in aqueous solution thus
suppressing any tendency toward acid polymerization. The molybdate species
is also known to be a good hydrogenation catalyst. Combinations of
molybdate with other metal ions may also yield enhanced aquathermolysis.

Combined Effects of CO and Metal Ions

The generally beneficial effects of added CO and metal ions discussed

above and in Chapter 6 naturally poses the question as to the possible

combined effects of these two "addends" on the aquathermo lytic process. We
have attempted to summarize the combined effects in the bar graph presented
in Fig. 21

Because of the substantially greater effects of the metal ions in the

Canadian sample this was used throughout the comparisons shown in Fig. 21.

All other aquathermolytic conditions besides the presence or absence of

 .

'
 .

-42-

addends were held constant throughout. The parameter used to monitor the
effect of the various changes in aquathermo lysi s conditions was the

viscosity of the recovered heavy oil following aquathermo lysis.

The effect of adding CO was to reduce the viscosity from a value of 590
centipoise after aquathermo lysis without addend to 370 centipoise after 14

days aquathermolysis at 240°C. The same aquathermo lysis in the presence of

molybdate or ferrous iron and in the absence of added CO, resulted in a

viscosity of 170 to 190 centipoise - a substantially larger decrease in

v i scos ty
i

It might have been hoped that a combination of both CO and added ions
would have resulted in an additive combination of their beneficial effects.
Such, however, is not the case as is seen from the 210 to 230 centipoise
viscosity values obtained when CO and molybdate or ferrous iron are

combined. We do not attach any great significance to the slightly higher

viscosity values obtained with the combined "addends" although there is
%
much significance in the non-additive nature of the combination. This
indicates that there is a limit to the beneficial effects that can be
generated through aquathermolysis and once these have been optimized by
either CO cn ionic addend no further improvement can be made. From an
economic standpoint it may be advantageous to use less of each addend in
order to achieve optimal conditions in a combination but this has not been
tested. The limit on the degree of optimization, however, is in keeping
with the hypothesis that the controlling factor in the overall process is
the amount of reactable (organosu phur) I
material present in the oil or
bitumen phase. Once this has been utilized by reactions that are affected
by either the WGSR (added CO) or by other aquathermolysis reactions that

are affected by addition of ionic catalysts, no further beneficial results

can be obtained within the aquathermolysis window.

As we have demonstrated, however, these benefits can be substantial in

the case of certain reservoir fluids and suggest that the aquathermo lytic
activity of such fluids in each potential production zone be carefully
 MM

«
examined as part of the overall investigation of suitability for in-situ
steam stimulation.

Patent

As a result of the work described in this Chapter Canadian (§1,189,444)

and US Patents (#4,506,733) on the use of Additives for Inclusion in a

Heavy Oil Reservoir Undergoing Steam Injection were issued in 1985. The
claims cover all of the metal salts noted in Fig. 20 with and without the
addition of carbon monoxide.
pfl
 I .

-44-

CHAPIEB-8

AQUATHERMOLYSIS AND RECOVERED OIL COMPOSITION

The bulk of evidence that has been presented and discussed in this
synopsis of aquathermo lysis has dealt with the chemical composition of the
produced gas phase or the physical properties of the recovered liquid
phase. We turn now in this Chapter to a consideration of the effect of
aquathermo ysi I
s on the chemi ca composition of the liquid phase.

The chemical composition analysis has been primarily one of examining

changes in the SARA components of the liquid phase. More recently our work
has been extended to examination of the actual molecular species undergoing
change but the work being reviewed here dealt primarily with a more wide
ranging exploration of the less detailed SARA classification. Both
Canadian and Venezuelan samples were subjected to aquathermo lysis under

varying conditions and the oils recovered from the aquathermo y zed (as well I

as unreacted) oil sands were subjected to SARA analysis and viscosity

measurement. The results are shown in bar graph and diagrammatic form in
Fig. 22.

At the extreme left and right of Fig. 22 the SARA analyses for the
unreacted extracted oils are shown serving as reference values for the
effects of subsequent aquathermo lysis. Between the two unreacted oil SARA
analyses we have presented the corresponding SARA’s for oils recovered
after three different aquathermo lysis treatments; the first with water
alone; the second with added CO; and the third with added CO and molybdate
ion. Conditions of concentration etc. are essentially the same
throughout. The key to the identification of the SARA components in each
group of bar graphs corresponding to each reaction condition is shown at
the bottom of the Fig. together with the sulphur content of each fraction
as recovered after water only aquathermo ys I
i s . The viscosity of each
recovered oil is also shown diagrammatica I ly
I

"

I
I

I
 -45-

The first observation of significance is that in both cases the

asphaltene content i ncreases upon aquathermo lyz ng i with water only. This
is particularly noticeable in the Canadian case. This is not surprising in

view of the substantially higher total sulphur content of the Canadian oil
and that the sulphur content of the asphaltene is significantly higher than
any other fraction. If indeed the organosu phur components of the oils are
I

the most reactive species in the aquathermo lysis window it might be

expected that a substantial change would occur in that fraction containing

most sulphur.

The relative amount of asphaltene in the Venezuelan sample does not

appear to vary much with aquathermo lysis conditions remaining close to 20?S.

However, the presence of addends CO or molybdate significantly reduces the

asphaltene content of the aquathermo y zed Canadian sample. I

This we believe
is due to the suppression of polymerization type reactions due to the
production of hydrogen and the catalysis of both the WGSR and hydrocracking
and hydrodesulphurization reactions.

The behaviour of the resin fraction in the Canadian sample is

noteworthy. It would appear that in the simple water aquathermo lysis it is

primarily the resins that are being converted into asphaltenes. The
suppression of that conversion in the presence of addends is reflected in

the higher resin content (and lower asphaltene content) in oils recovered
from addend aquathermo ys s I
i . There are less dramatic changes in resin
content in the Venezuelan sample but this fraction appears to be the source
of material for conversion to saturates when CO and molybdate are present.
This is very likely a reflection of hydrogenati on/hydrodesu phur zation by I
i

WGSR hydrogen, catalyzed by molybdate.

It is interesting that the aromatic SARA fraction appears to be the

least affected by aquathermolysis. There are changes but these are
relatively minor and suggest that the aromatic structures are less reactive
at the temperatures characteristic of the aquathermolysis window. This is

in keeping with our observations on the aquathermo yti c activities of

I

thiolane and thiophene discussed in Chapter 2.

 -46-

The viscosity behaviour of the various recovered oils reflects a

complex dependence on chemical composition. A possible simplified
relationship is discussed in a following section where it is suggested that
at a particular saturate/asphaltene ratio the viscosity i>ecomes very

sensitive to small changes in composition. In the Canadian sample the

failure of the substantial increase in asphaltene content on

aquathermolyzing with water alone to result in a larger increase in

viscosity is presumably due to the original higher percentage of saturates

(20%) which actually increases slightly on aquathermo lysi s . In the

Venezuelan case there is a higher initial asphaltene content and a lower

saturate content which decreases somewhat on aquathermo lysi s with water

alone and obviously cannot "solubilize" the increasing asphaltene. The
result is the substantial increase in viscosity of the Venezuelan oil on
aquathermo lysis with water alone which has been noted in Fig. 17 in

Chapter 6.

The decrease in viscosity noted in the Canadian sample as "addends" are

introduced is^almost certainly due to the drop in asphaltenes back toward
the unreacted level. The fact that the viscosity is actually less than the
original unreacted oil is a further illustration of how sensitive the
viscosity is to small increases in the amount of solubilizing saturate as
well as decrease in the resin content. This combination of saturate and
resin content is also likely the reason for the decline in viscosity noted
when addends are present in the Venezuelan aquathermo lysis.

We are already at the point where a knowledge of SARA composition and

sulphur content of the fractions permits us to make a crude prediction of
aquathermolytic (steam stimulation) sensitivity of a particular untreated
oil extracted from an oil sand core. If we can add to this a knowledge of

the principal molecular types involved and how they respond to

aquathermolysis, polymerization (acid), uncatalyzed hydrogenation, metal
catalysis etc., we will then be in an excellent position to answer the
challenge to develop a predictive capability regarding aquathermolytic
behaviour based on the chemical analysis of virgin core material.
'
 -47-

Sensitivity of Viscosity to Chemical Composition of Oil

We have argued in other sections of this synopsis that the chemical

reactions that occur during aquathermo lysi s may be limited in terms of the

quantities of reactants involved and new products generated. This

hypothesis is related to the role played by the organosu phur compounds as I

the reactive species in the aquathermo lysis window. Despite the limited
nature of the chemical changes occurring, however, we have seen some

substantial changes in the important property of viscosity of the fluid

phase. The question that then arises is how sensitive is that viscosity to
relatively small changes in composition?

Hernandez et a I have reported an interesting relationship between the

viscosity of the heavy oil phase and the ratio of the weight percent

saturated components/asphaltenes (S/As) in the heavy oil. Their

relationship is shown in Fig. 23(b). In our study we examined four
Canadian heavy oil sand samples and four Venezuelan samples each from
significantly different locations. A similar relationship to that of

Hernandez et a I is shown in Fig. 23(a). Both plots show a distinct

tendency toward a marked change in dependency of viscosity on S/As ratio at
2.3 in the Hernandez case and 1.3 in our data. The difference in these
break points may be due to a somewhat different method used to carry out
the SARA (Saturates, Aromatics, Resins, Asphaltenes) analysis. It is clear
from the plots in Fig. 23 that heavy oils with S/As ratios lower than the
break points shown, are very prone to major changes in viscosity resulting
from minor changes in chemical composition. Consequently these oils could
exhibit desirable reductions in viscosity due to relatively small amounts
of aquathermo lytic products produced in-situ. Miscible solvent flooding,
which could raise the effective saturate content, might also result in a

large viscosity decrease of the produced oil. Those heavy oils with a S/As
ratio greater than the breakpoint will show less tendency to change in

viscosity as a result of a small change in chemical composition, such as

might result from aquathermolysis.
I

I
 - 48 -

This effect may be explained in terms of the molecular aggregation of

the components which make up the oil. If the S/As ratio reaches a limiting

value above the breakpoint then there will be sufficient saturate molecules
to "solubilize" the asphaltenes and other large molecules. Otherwise the
large molecules will tend to aggregate into effectively larger molecules.
At this point the state of aggregation of the oil becomes very susceptible
to small changes in composition.

Although this model does not take into account other important factors
that could change during aquathermolysis (e.g. functional group reactions
that could change i ntermolecu lar forces) it does serve to rationalize the
observation that in some heavy oil sand cases the small changes in chemical
composition that can result from aquathermolysis can have a substantial
effect on viscosity while in others the effect is minimal.
 -49-

CHAPIER-9

SOME COMMENTS ON SAMPLE INHOMOGENEITY AND EXPERIMENTAL ARTIFACTS

Our studies have provided us with an opportunity to work with a wide

variety of heavy oil sand samples from many locations and with a range of

storage and handling histories. Although the research community is now

reasonably familiar with many of the pitfalls that the unwary may
encounter, we thought it useful to highlight some of our experiences in

this synopsis.

The gross composition of core samples from the same field and indeed
from the same hole are often significantly different. If such samples are

not homogenized, very large differences in aquathermo lytic activity can be

observed between sub-samples of what is presumed to be the same sample.
This also has important field consequences. It indicates that varying
degrees of activity such as H2 S production, viscosity change, C0 2
production etc. can be expected across a given reservoir and even with
depth in a particular hole. While these variations may be less than those
observed between major reservoirs they may be sufficient to cause
difficulties in well head handling equipment (e.g. H 2 S sweetening) if it is

not sized to handle the range of fluctuations that may be encountered.

Even after homogenization there may still be significant variations

between sub-samples. In Table 2 we show the variation in water/bitumen/
solids (sand) composition among four sub-sample bags of the same
homogenized total core. While the variations are not great they are
sufficient to result in data scatter in results such as produced gas
analysis that might not be acceptable in "pure" quarters. The simple facts
of the matter are that the substrates being worked with are not pure and
never will be and the uncertainty in data replication must simply be
accommodated. The major caution must be against over nterpretat ion of
i

small changes in any measurement.

I

I
I

I
I

I
I

J
I

I
 -50-

We have shown elsewhere in this synopsis that small changes in the

chemical composition can have large effects on viscosity. Oxidation can

have a significant effect on the number and type of polar functional groups
in resins which in turn can affect the i ntermolecular forces and viscosity
of the heavy oil. In Table 3 we show the dramatic effect that drying out

and ageing can have on a sample from the same field as far as

aquathermolytic behaviour is concerned. The new sample is clearly much

more reactive as, of course, would be the actual material in-si tu.

Since changes in viscosity are particularly important in

aquathermo lysis work it is important that experimental artifacts that could

interfere with viscosity measurement be fully understood and eliminated.
Bennion et a I have reported previously the effect of residual amounts of
extracting solvent on oil viscosity. Our data in Table 4 confirm this
showing that difficult to remove traces of benzene solvent can lower
bitumen viscosity by 100 cent poisei
compared with that obtained by

methylene chloride extraction, a solvent that is much easier to remove and

to detect in trace residue amounts.

The effect of water on the viscosity of methylene chloride extracted

bitumen is dramatically illustrated in Table 5. Each drop of water
represented some 2% by weight of the oil sample. Although it is unlikely
that 16% water contamination would go unnoticed the data do illustrate the
substantial viscosity increase that can result from a wet sample. Drying
agents are thus frequently used and residues of these materials or
suspended clays can significantly affect viscosity as seen in Table 4.

Samples not filtered (micropore) can have substantially higher viscosities

due to the presence of residual suspended solid materials.

As part of the overall reporting of this work as it progressed we

prepared an Experimental Procedures Booklet in which are collected data and
reports on many of the operating artifacts that we encountered in working
with oil sand samples.
'
 -51-

The bulk of our aquathermo lysis work was carried out in reaction

vessels constructed from Hastelloy C276 alloy. Concern regarding the

possible catalytic effect of the walls of the vessel led us to perform

parallel experiments in quartz reactors. We understand that even quartz

has been criticized as having a possible catalytic effect leading to the
use in some cases of gold lined steel reactors. Our argument has been that

if quartz is catalytic then all steam-oil interactions occurring in-situ

must be subject to such catalysis in view of the ubiquitous presence of

ample supplies of silica in the sand.

Whatever the relative catalytic effects of the reactor wall materials

the plots in Fig. 24 show that they are not significantly different as
between Hastelloy and quartz. In Fig. 24(a) there may be some evidence of

a difference in C0 2 production rates in both the Venezuelan and Canadian

samples in the initial aquathermo ys I
i
s period but the final equilibrium
value obtained is not significantly different (see also effect of vessel

wall and quartz on H 2 S production in Chapter 5). The same conclusion can
be drawn from the rates of production of other gases. The effect on
viscosity of the recovered oil is also seen in Fig. 24(b) to be very small.
The initial difference between Venezuelan and Canadian sample viscosity
behaviour is due to the greater sensitivity of the Venezuelan sample to

polymerization as discussed in more detail in Chapter 3.

These results also serve to illustrate the effects of prolonged

aquathermolysis. Data for periods of aquathermo ys s I

i in excess of six
months show that the bulk of aquathermo yti c activity has taken place over
I

a period of a few weeks and thereafter any further reaction is very slow at
240°C. The viscosity data for the two samples substantiate the earlier
argument that the initial rise in viscosity in the Venezuelan sample caused
by the greater polymerization sensitivity is eventually overcome by the
slower more energy demanding aquathermolysis processes resulting in a

substantial reduction in viscosity over time for both samples. Thus the
effect of thermal soak in the case of these two samples under the
aquathermolysis conditions noted will be a substantial reduction in

viscosity.
I
 i .

-52-

CHAPTER IP

CONCLUSIONS AND FUTURE DIRECTIONS

Significant chemical reaction has been shown to occur between oil sands
and water (steam) in the temperature range of 200°-300°C which we have
labelled the aquathermo lysis window. Thermal cracking is only significant

above 300°C and below 200°C the reactions are not fast enough to be
meaningfully measured over usual laboratory experiment time scales. Their

effect on reservoir areas surrounding the active steamed zone, however, may
be one of slow thermal maturation, raising oil/bitumen viscosity and
influencing subsequent production.

Both the oil/bitumen and the host porous sand medium undergo
aquathermo lysis. The primary sand reaction appears to involve reactive
mineral carbonates yielding varying amounts of C0 2 which can affect the
acidity of the system and the viscosity of the oil by dissolving in it.

The most reactive species in the oil/bitumen phase appear to be the

organosu phur compounds,I

the carbon-sulphur bonds of which may cleave
during aquathermo lysi s especially if suitable metal ion catalysts are
present to coordinate with the sulphur as a ligand. Hydrogen sulphide is

produced as a result of this sensitivity of the organosu phur species and I

at higher temperatures in the aquathermo ys s I

i range the amount of H 2 S
produced appears to be proportional to the total amount of sulphur in the

oil /bitumen phase. Addition of metal salts that can form stable sulphides
by reaction with H 2 S, may be a method of controlling H 2 S production but

removal of sulphur from the oil by this mechanism, in-situ, may be viewed

as a useful upgrading step lessening the load on surface desulphurization

f ac I i t es
i

Heavy oils/bitumens appear to undergo competitive chemical

transformations that affect viscosity in opposite ways during
aquathermo lysis. Once reactive species have been generated by

aquathermo lyti c cleavage of sensitive sites such as the carbon-sulphur

 .

- 53 -

bond, they can either polymerize and thus raise the viscosity or they can
initiate a series of further reactions that result in the generation of
products (e.g. saturates) which can reduce viscosity even when present in

relatively small amounts. The balance between these two opposing effects
appears to be dependent on a number of factors including type of

oil/bitumen, time of aquathermo lys s (steaming) and acidity. i

One of the most important component reactions of the aquathermo lys is

process is the Water Gas Shift Reaction. Initial cleavage of the reactive

species (organosu phurs) I

in the oil or bitumen can trigger a series of
reactions that yield carbon monoxide as a result of decarbony lation. This
carbon monoxide can react with the high temperature water to yield carbon
dioxide and hydrogen and the hydrogen, in turn, can participate in

upgrading type hydrocracking and hydrodesulphurization reactions with the

oil /bitumen phase. Thus hydrogen from water used to upgrade the oil
in-situ

Optimization of this highly desirable direct use of hydrogen in water

to upgrade heavy oils and bitumens in-situ would be economically
attractive. Our work has shown that this can be accomplished by use of
additives to enhance the aquathermo lys s reactions involved. i In particular

addition of carbon monoxide and metal salts to the aqueous phase (steam)
results in substantial reductions in viscosity with the potential for
producing a we I I head fluid that is pi pel ineable. The added CO drives the
WGSR equilibrium to the desirable hydrogen side and the metal salts act as
critical catalysts in the several component reactions of the
aquathermo lys i s sequence. Generation of basic conditions by addition of
caustic, or choice of salts which themselves are basic may be of further
assistance in enhancing these reactions that lead to viscosity reduction.

Our work to date has been directed more toward a study in depth of a

limited number of heavy oil sand samples. While the comparison of the
behaviour of the Canadian and Venezuelan samples provided us with data from
two widely different sources it only scratched the surface of the general
 I
1

I
 I

-54-

question of large variability in the aquathermo lytic behaviour of heavy oil

sands. We believe we now know what to look for in attempting to predict

the chemica behaviour of such reservoirs when they are steam stimulated
but we are only at the start of the process of identifying the extent of
the variation. Emphasis should thus be placed on examining the

aquathermo lyti c behaviour of a wider range of heavy oil sand samples and

establishing a protocol for predicting the behaviour based on laboratory

examination of cores.

It is hoped that field tests of some of the ideas that have been

generated (and patented) in this work may be possible. Some of the

concepts could be tested to some extent on existing pilots by gathering

relevant production data over time. It is hoped that this synopsis may
encourage operators to examine their pilots in a manner that might assist
in checking the validity of some of the postulates. Ideally, we would hope
that injection of CO and/or salts could be tested to determine whether the
beneficial effects on viscosity observed in the laboratory could be
duplicated in the field.

Although work at lower temperatures (< 200°C) is very time consuming we

believe that the potential long term effect of slow thermal maturation of
polymerization sensitive oils around the active steam sweep zone is an

important one. Field sampling through step-out boreholes into reservoir

regions at temperatures between active zone and natural reservoir values
might be most helpful. A laboratory examination of the properties of these
cores as a function of time would provide at least some form of test of the
validity of the slow thermal maturation hypothesis that we have advanced.

Much remains to be done at the molecular level to characterize and

understand the chemical processes that occur during aquathermo ys s I
i . Our
work to date has identified changes in the SARA composition of oils as a

function of degree and type of aquathermo lysis. We have only begun to

improve the level of understanding by identifying changes at the molecular
level. Modern liquid and gel permeation chromatography coupled with gas
 -55-

chromatograph c/mass spectrometric examination

i is quite capable of probing
these changes at the molecular level. This is not an academic exercise
seeking to know why for the sake of knowing why. Without this information
we cannot hope (except by chance) to select and design catalytic systems
(addends) that will optimize and enhance the beneficial aspects of

aquathermolysis. That this is possible in certain instances has already

been demonstrated by reducing some heavy oil viscosities by well over 60%.
We still do not understand fully, however, why this happens in the
particular sample studied, nor how it happens with sufficient assurance to
be able to optimize the addends and the aquathermolysis conditions.

Finally, we turn to viscosity and its dependence on chemical

composition. While we have not been engaged primarily in work directed
toward this end we have, of necessity, been directed into this problem
because of the effect that aquathermolysis has on the composition of the
oils. It is surely through this effect on composition that aquathermolysis
affects viscosity. The degree of chemical change that can occur due to
aquathermolysis may well be limited by the small amount of material in the
oil that is reactive in the aquathermolysis window. However, if this can
be optimized by use of addends, and if the small amount of chemical change
produces the correct new product it is well known that such small amounts
of new chemical species in heavy oils and bitumens can have an effect on
viscosity out of a I I proportion to the amount present. Therein may lie the
future utility of aquathermolysis.
-

I
 SYNOPSIS REPORT NO. 50

AQUATHERMOLYSIS -

A SYNOPSIS OF WORK ON THE CHEMICAL REACTION

BETWEEN WATER (STEAM) AND HEAVY OIL SANDS
DURING SIMULATED STEAM STIMULATION

(AQSTRA CONTRACTS NO. 11, 103, 103B/C)

J.B. HYNE
PRINCIPAL RESEARCHER

DEPARTMENT OF CHEMISTRY
THE UNIVERSITY OF CALGARY
CALGARY, ALBERTA

APRIL 1986
 SYNOPSIS REPORT NO. 50

ARUATHERMOLYSIS -

A SYNOPSIS OF WORK ON THE CHEMICAL REACTION

BETWEEN WATER (STEAM) AND HEAVY OIL SANDS
DURING SIMULATED STEAM STIMULATION

(AQSTRA CONTRACTS NO. 11, 103, 103B/C)

J.B. HYNE
PRINCIPAL RESEARCHER

DEPARTMENT OF CHEMISTRY
THE UNIVERSITY OF CALGARY
CALGARY, ALBERTA

APRIL 1986
 may i 4 t99i

c
o

I
» I
Chemical
C-S-C < C-N-C < C-O-C < C-C Bond

Reservoir Reservoir Reservoir

Around Producing Producing Water
Production Under Steam Wet Combustion
Zone Stimulation Conditions Reservoir

Figure X The Aquathermolysis Window

 NMR sugqests intracomplex interactions.
1(b) Intermediate thiol production detected by potentiometric
titration of SH.
1(c) Enol rearrangement can yield aldehyde. Aldehydes detected by
IR and formation of 2,4 DNPH complex.
1(d) Decarbonyl ation to yield CO shown possible by aquathermolyzing
n-butyraldehyde under same conditions to yield CO.

Fig. 2 The Initiation of Aquathermolysis

 —

Gases in boxes are principal components of gas phase produced

from aquathermolysis of both heavy oil sands and thiolane and
thiophene, (see Fig. 5)

——— Metal ions from mineral phase or intentional addition act as

catalysts for both WGSR and hydrodesulphurization possibly in
sulphided form through reaction with produced H 2 S.

(3) Intermediate nature of CO explains transient appearance in

produced gas. (see Chapter 6)

Fig 3 Other Aquathermolysis Reactions

 Phase

Liquid

of

Conversion

Phase

Liquid

of

Conversion

%
log

Figure U Chemical Reactions Occurring in Water/Heavy Oil Systems

 Oil

Gas/kg

(STP)

mis

Compound

Organosulphur

(STP)/ml

Gas

Moles

10 20 30
Days (aquathermolysis)

Figure 5 Main Gaseous Products from Aquathermolysis at 300°C

(85°C)

Centipoise

Fig 6 Viscosity Behaviour of Recovered Bitumens as Function of Temperature

and Duration of Aquathermolysis
 Fig. 7 Reservoir Temperature Profiles and Sequential Steam Stimulated Production

180°

slow aquathermolysis 160°

during 1st production

140°

1st Active Zone Sweep

2nd Active Zone Sweep

Injection

Production
O*
 Sample

Core

in

Bitumen

ml/1000

Production

2
C0

Fig 8 Aquathermolysis of Whole Core Samples C0 2 Production vs Time and Temperature

 Bitumen

1
kg
gas

Production/mls

Gas

Figure 9 C0 2 Production
Figure 10 Rate of Production of H 2 S from Aquathermolysis of Whole Core Sample
of Venezuelan Oil Sand as a Function of Time and "J^mperature
Bitumen

(STP)/1000g

S2
H
mis

Figure 11 Rate of Production of from Aquathermolysis of Whole Core Sample of

Canadian Oil Sand as a Function of Time and Temperature
 gas

produced

in

H2S

ppm

Bottom injection hole or lab conditions

Fig 12 Relationship between Produced H S and Average Reaction Temperature

2
Field and Lab Simulation
 (a) Canadian

Sand

Oil

in

Bitumen

CO/IOOO

ml

(b) Venezuelan

Fig 13 CO Production and Depletion during Aquathermolysis

Bitumen

g
/1000

2
C0
mis

Days Aquathermolysis

Figure 14 Effect of Adding Overpressures of CO to Aquathermolysis

Vessel on CO 2 Production
Bitumen

H2/IOOO

mis

Days

Figure 15 Hydrogen Production Data for Aquathermolysis of

Venezuela and Canadian OH Sands at 240°C

#
Figure 16 Rate of Hydrogen Sulphide Production Resulting From Added
CO Aquathermolysis at 240°C

Wt.% CO
(cps) (a) Venezuelan

Viscosity

(cps)

Viscosity

(b) Canadian

Fig 17 Added CO on Viscosity of Recovered

Effect of Oils from
240°C Aquathermolysis
Fig 18 Trends in the Production of Hydrogen from the Aquathermolysis of
a Venezuelan Oil Sand with Added Metal Salts over 28 days at 240°C
 1050

1000 -

S
E
2 150 -
ffi

B>

s
o

o 100
«5
O
3
*D
2
Cl

%
X
50 -

H 2 0 A1 3+ T1 3+ V0 2+ Cr 3+ Mn 2+ Fe 2+ Co 2+ 2+ Cu 2+ Zn 2+ Mo0 4 2
'
Ni

Fig. 19 Trends in the Production of Hydrogen Sulphide from the

Aquathermolysis of a Venezuelan Oil Sand with Metal
Salts over 28 days at 240<>C
 (a) Canadian

unreacted value
ps a TBp" ir
500 h 2o
i ?

400
VO2*
Zn-

+
300 Ni 2
g:

+
+ Mn 2 Cu 4

3+ Cr2
Al Co-
200
MoO>

Fe-

*
VO2
h 2c )

600 (b) Venezuelan

pH4
:

::

P<'

500 :

•
i Cu 2 +
Co-

*
Cr2 Fe<
If
'

::
Zn 4

400
V Unreacted
2+ Mo0 4 ^
P H9 Ni;
Value
Il pH 10
pH > 12 Mn-

300
V0 2++ NaOH
Fig 20 Effect on Viscosity due to Aquathermolysis of Oil Sands with
Added Salts at 240°C
Fig. 21 Trends in Viscosity of Oil from the Aquathermolysis of Canadian Oil Sand
with Combinations of Metal Salts and Carbon Monoxide over 14 days
at 240°C
 Canadian Venezuelan

Unreacted h 2o H 2 0/C0 h 20/CO/MoO*J H 20 H 2 0/C0 h 2 0/C0/MoO*J Unreacted

40

30

wt %

20 SL I
i
I
I i ::

li

If s a
10 -SM M.
500
Viscosity

300
100

asphaltenes resins resins saturates

SARA
Steamed (H 2 0 only) oils; Sulphur Content
V77A asphaltene

resin
9.0% S — 5.2% S
ESI
m aromatic
5.7% S
4.3% £
—
- 5.0% S
3.1% S
f I saturates 0.5% S 0.2% S

Fig 22 Sara Properties of Recovered Oils

Fig 23 Relationship between Viscosity and Saturate/Asphaltene Ratio
 Bitumen

(mls/1000

Production

2
C0

(cps)

Viscosity

Fig 24 The Effect of Containment Vessel Material on Aquathermolysis

at 240°C
 Table 7 An Attempted CO 2 Mass Balance

Aquathermolyzed Sample

Whole Core Bitumen Sand

Sample 1 CO 2 (mis) 7980 1050 9453

CO (mis) 82 3005

Sample 2 C0 2 (mis) 5075 2076 3958

CO (mis) 29 2241

Sources of CO 2

CO.

CO.
Catalyst
 Table 2

Sub-Sample Composition of Homogenized Venezuelan Core

Bag 1 2 .3 4

Water 1.03 1.69 : 1.12 1.26

Bitumen 8.74 9.58 . 9.71 8.62
Solids 90.23 88.73 89.17 90.12

Table 3

Comparison of Gas Production from Old and New Venezuelan Reservoir

Oil Sand Sample Aquathermolyzed for 14 days at 240°C
Bitumen H 0 Ratio of 0.62
: (mis gas STP/1000 g Bitumen)
2

Gas Old Sample New Sample

H 2S 74 153
C0 2 3242 10582
CO 0.5 32
H2 351 824
ch 4 122 507
c 2+ 86 221

Total Gas (mis) 3875.5 12319

Table 4 Effect of Residual Solvent and
Suspended Clays on Bitumen Viscosity
Viscosity
Centipoise
%A

CH 2 Cl 2 Extracted, 520
Filtered + 37%
CH 2 CI 2 Extracted 712
Not Filtered

C6H 6 Extracted, 420

Filtered + 36%

C6 H 6 Extracted, 570
Not Filtered

Table 5 Effect of Water on Bitumen Viscosity

Added
Centipoise
Drops H
2
0
0 520
1 530
2 550
4 590
6 740
8 832
-a

mat
 OILSANDS TECHNOLOGY
AND RESEARCH AUTHORITY