SPE

Society of Petroleum Engineers of AIME

SPE 13661

Hydrogen Generation During In-Situ Combustion

by L.E. Hajdo, * R.J. Hallam,* and L.D.L. Vorndran, BP Resources Canada Ltd.
* SPE Members

Copyright 1985, Society of Petroleum Engineers

This paper was presented at the SPE 1985 California Regional Meeting, held in Bakersfield, California, March 27-29, 1985. The material is subject to
correction by the author. Permission to copy is restricted to an abstract of not more than 300 words. Write SPE, P.O. Box 833836, Richardson, Texas
75083-3836. Telex: 730989 SPE DAL.

ABSTRACT INTRODUCTION
BP Resources Canada Ltd. is operating an oxygen BP Resources Canada Ltd. is the operator of an
in-situ oil recovery pilot at Marguerite Lake, in-situ combustion pilot at Marguerite Lake in
part of the Cold Lake heavy-oil deposit in east central Alberta. The partners in the
eastern Alberta. The pilot consists of two project include the Alberta Oil Sands Technology
principal areas: a three-well, wet air and Research Authority, PanCanadian Petroleum
combustion test and an infill-drilled, four Limited and Dome Petroleum Limited. The lease
five-spot, wet oxygen combustion project on is in the Cold Lake heavy-oil region, and
2 hectare spacing (5 acres). All the wells were production is from the Clearwater formation at a
initially steam fractured and steam operated depth of 450 m (1 476 ft). The reservoir is an
through several cycles. Thereafter, selected unconsolidated sand with a gross pay thickness
wells were converted to air or enriched-air of 34m (112ft), a porosity of 30%, a
injectors. permeability of approximately 1-3 ~m2 (1-3 Darcys)
and bitumen saturations of 64%. At the original
Eight adiabatic combustion-tube tests were reservoir temperature of 15°C the bitumen is
conducted at the University of Calgary on immobile (density of 1 000 kg/m3 [10°API]).
reservoir materials under various conditions.
During several, but not all, of the tests small The pilot consists of three independent areas
amounts of hydrogen gas were produced in a (Figure 1). Single-well tests were conducted at
discontinuous fashion. In contrast to this, well EX Tl. Wells EX T2 and EX T3 were drilled
most wells in the field have produced hydrogen, along the principal NE-SW fracture trend, steam
some consistently as high as 20 mole percent. fractured and operated through several steam
cycles. Later, EX T4 was drilled off-trend,
The present study was undertaken in order to steam fractured and cyclic steam operated
better understand the flow and reaction briefly before being converted to an air
processes that generate hydrogen in situ and injector. These three wells form the Combustion
thereby to suggest ways of influencing its Test area. The main pilot is comprised of four
production. The results are interpreted in five-spots, later in-filled with five additional
terms of five reaction processes involving wells. All the wells were initially steam
hydrogen: aquathermolysis, thermolysis, coke fractured and cyclic steam produced, typically
gasification, the water-gas shift reaction and through six cycles. The data from these
methanation. operations have led to the commercial
development of the Wolf Lake Cyclic Steam
It is concluded that the hydrogen is generated Project nearby on the lease.
by the coke gasification reaction followed by
the water-gas shift reaction. Methanation Since 1983, four wells in the main pattern have
processes also play an important role. Thermal been converted to enriched-air injectors. EX 4
cracking is a minor contributor to the gas was the first injector, and it and its
production and aquathermolysis is negligible. neighbouring producers were operated for several
months without interference from other
injectors.

References and illustration~ at end of paper.

675
2 HYDROGEN GENERATION DURING IN-SITU COMBUSTION SPE13661

Throughout the history of the project extensive yielding methane, ethane, propane, etc., as
data acquisition has been ongoing. Each well as hydrogen, carbon monoxide, carbon
producer has downhole and wellhead thermocouples dioxide and hydrogen sulphide [2].
and a bottomhole hubbletube for pressure
measurement. The observatiori wells are not 3. Water-Gas ~hift Reaction
perforated and contain a single thermocouple.
All wells are frequently surveyed over the The water-gas shift reaction is [3]
affected intervals. Oil, water and gas
production measurements and gas analyses are 0 0 0 ( 1)
typically conducted twice weekly. Oil and water
analyses are conducted regularly.
4. Coke Gasification
The production of hydrogen gas has been a
feature of all the operating schemes: cyclic Coke gasification is the term used to
steaming, air fireflooding and oxygen describe either of the reactions
combustion. The interest in the phenomenon is
twofold. Firstly, since hydrogen is generally C+H20(steam)=CO+H~ -31.4 kcal/mole ••• (2)
not a significant component of laboratory C+C02=2CO- 41.2 kcal/mole ••• (3)
combustion-tube tests (the ideal fireflood),
11 11

its appearance in the field ~ay be an indicator It is important to distinguish between the
of undesirable processes (e.g. low-temperature generation and production of hydrogen and other
oxidation or channelling). Hydrogen, therefore, gases during in-situ combustion. A priori, there
becomes a tool in the monitoring of in-situ is no reason to suppose that the products of the
recovery performance. A secondary concern is above-listed reactions are not subject to
that hydrogen increases potential hazards in further reactions; oxidation or any one of the
production wells and treating facilities. On following methanation reactions is possible
the other hand, if the generation of hydrogen is [3]:
not significantly detrimental to oil recovery,
its production should be enhanced for its C+2H2=CH4 +17.9 kcal/mole 0 0 0 ( 4)
com~ercial value in heavy-oil upgrading or as a C0+3H2=CH4+H10 +49.3 kcal/mole 0 0 0 ( 5)
fuel. 2C0+2H2=CH4+CO~ +59.0 kcal/mole 0 0 0 ( 6)
The present study was undertaken in an effort to Methanation reactions are a major reason for the
better understand the flow and reaction difficulty in analyzing the other processes;
processes that generate hydrogen in situ. The hydrogen and carbon monoxide are consumed to
available data include pyrolysis work conducted yield methane, carbon dioxide and water.
with the reservoir bitumen, eight adiabatic Because of the many possible sources of methane
combustion-tube tests and the early histories of (connate or produced by thermal cracking,
the air combustion test and the main pilot aquathermolysis, or methanation), it is only of
oxygen combustion project. minor use in analyzing combustion reactions.
This is also true of carbon dioxide which is
In this article the various processes and generated by high- and low-temperature
reactions which may be responsible for the oxidations, decomposition of reservoir
production of hydrogen will first be reviewed. carbonates, methanation, thermal cracking and
Then, the laboratory and field data will be aquathermolysis.
examined and some conclusions drawn as to the
source of the hydrogen production. Another complicating factor in the
interpretation of data is the solubility of the
gases. Since the solubilities and their
FLOW AND REACTION PROCESSES pressure dependencies all vary, direct
quantification of reaction rates based on
Four generic reactions have been reported which production data is not possible. This is
could generate hydrogen gas during in-situ particularly true of carbon dioxide; its
combustion. comparatively large solubility and its
domination over the minor gas components render
1. Aguathermolysis it almost useless in the investigation of
reactions other than high-temperature oxidation.
J.B. Hyne et al. [1] have shown that Lal et al. [4] have investigated the solubility
aquathermolysi s of heavy oi 1 (steam/oi 1 of C02, CO and H2 in Athabasca bitumen. It was
reactions) over the temperature range 200 found that the carbon dioxide solubility is
- 300°C produces C02, CH4, H2 and H2S. The approximately one order-of-magnitude higher, on
total quantities of gases produced by a molar basis, than that of the other two gases.
aquathermolysis were not large, being less Since the solubility of C02 in Marguerite Lake
than 1 wt% for their oil. bitumen [5] is not substantially different than
in Athabasca bitumen, it is expected that
2. Thermal Cracking (Thermolysis) similar trends apply. The solubilities of H2
and CO are sufficiently similar [4] to preclude
The heating of oil in the absence of oxygen any differences between their production rates
thermally cracks the larger hydrocarbons being attributed to dissolution or evolution.

676
SPE13661 L.E. HAJDO, R.J. HALLAM AND L.D.L. VORNDRAN 3

In the present investigation no attempt was made One can investigate the kinetics of the
to correlate gas compositions with any of the water-gas shift and coke gasification reactions
analyses of the produced liquids. The field in an attempt to estimate their relative rates.
experience has been that gases from the Guntermann et al. [7], investigating underground
combustion front are p~oduced in a matter of coal gasification, assu~ed that the following
days while the liquids lag by several weeks. rate equation describes the water-gas shift
This, in itself, causes a difficulty in that reaction.
longer transit times for the liquids smear
profiles and, therefore, clear fronts are not d[CO]=-kl [CO] ••• (7)
easily identifiable with specific gaseous crt
products. Furthermore, such properties as the
pH of water and oil viscosity were not strongly Their modelling work then yielded:
affected by the combustion processes during
initial production. kl=645 exp(-1.49x105jRT) ••• ( 8)

Before examining the five reaction processes Equation (8) is an Arrhenius expression for the
listed earlier in flowing situations, a reaction constant where T is temperature (°K)
discussion of some static considerations is and R is the ideal-gas constant
worthwhile. (8.31 J/mole·°K).
During the first month of aquathermolysis, the For the coke gasification reaction (Equation 2)
process produces approximately 1 300 ml of they similarly assumed
hydrogen (STP) per kilogram of oil (1.3 m3jm3)
[1]. Thereafter, even smaller quantities of d[C]=-k2 [C] ••• ( 9)
hydrogen are evolved, aquathermolytic reactions Cit
have been completed. Even rough estimates of
the mass of virgin bitumen which is heated to and gave
200 - 300°C each month in the pilot project have
indicated that aquathermolysis could account for k2=245 exp(-9.20x104/RT) ••• ( 10)
less than 1% of the produced hydrogen. The very
small amounts of hydrogen and hydrogen sulphide as the temperature behaviour of the
which were produced during early cyclic steam reaction-rate coefficient.
operations, prior to any air or oxygen being
injected at the site, are attributed to Guntermann [7] also assumed that the methanation
aquathermolysis. However, due to its minimal reaction (Equation 4) behaves as
contribution, this process will not be
considered further in the present study. ••• ( 11)

The contribution of thermal cracking to gas

production during combustion is not negligible with
in general. Pyrolysis experiments were
conducted for the present project on reservoir k3=0.183 exp(-4.14x104/RT) ••• (12)
bitumen in the presence of formation solids.
Table 1 gives the product gas compositions for The rates k1, k2 and k3 are evaluated at
the pyrolysis of steam-produced bitumen mixed temperatures which favour the products of the
with core material. At 420°C, after eight reactions. For the water-gas shift reaction at
hours, the bitumen will yield approximately 500°K (227°C) one finds
10m3 (H 2 )jm3 (oil). (The gas production curves
appear to level off at about 8 h, increasing ••• ( 13)
perhaps by another 50% of the value attained to
this point for long times.) The contribution of For the coke gasification process at 1000°K
thermal cracking to gas production in laboratory (727°C) one calculates
and field processes will be investigated. It
will be useful to recall that gas production due ••• (14)
to pyrolysis is characterized by an
ethane/methane ratio of 0.3-0.5 and that At 15°C and 100°C the methanation reaction rates
hydrogen sulphide is an associated product. are:
The equilibrium concentrations of the various k3=5.62x1o-\- 1 ••. (15)
products and reactants in Equations (1-6) can be k3=2.90x10- 7 s- 1
calculated from the data in Reference [6]. The
results are plotted in Figure 2 as functions of Relations (13-15) suggest that the coke
temperature. The water-gas shift reaction gasification reaction proceeds much more rapidly
favours the products up to 810°C whereas the than the other two. Relations (13) and (15)
coke gasification reaction does not favour will be discussed in the next section. There
hydrogen until about 675°C is reached. The are two important cautionary notes. Firstly,
methanation reactions consume hydrogen below the catalytic effects of the reservoir sands,
temperatures in the range of 540 - 630°C. clays and mi~erals have been ignored. There is no

677
4 HYDROGEN GENERATION DURING IN-SITU COMBUSTION SPE13661

reason to think that catalysis will not dominate the complete methanation reactions consume at
the reaction rates. Secondly, the k 1 s depend least as much hydrogen as carbon monoxide.) It
strongly on the values given for the activation would appear possible, therefore, that coke
energies. A small error in the modelling or gasification is a normal component reaction
small changes in the system under consideration during wet combustion, but the generated
could substantially alter the reaction rates. hydrogen is often not produced. Therefore, one
should expect that CH4, CO and trace amounts of
H2 would be produced during wet combustion-tube
ONE-DIMENSIONAL FLOW tests but comparatively smaller amounts of
methane would he generated during a dry run.
Thermal cracking is a homogeneous reaction, not
requiring the removal of products or addition of The reaction rates (13) and (15) also explain
reactants for its sustenance. Therefore, flow the disparity between field and laboratory
processes are of secondary interest for results. In the field, the wider steam plateau
pyrolysis except insofar as they govern the size and lower fluxes allow the water-gas shift
and temperature of the steam and hot water reaction to go nearer to completion before
zones. methanation consumes the remaining carbon
monoxide and hydrogen. It might also be noted
For the coke gasification and water-gas shift 11
that in adiabatic laboratory tests, such as
11

reactions, however, flow effects are of central the ones conducted for the present study, the
importance. The water-gas shift reaction entire tube is often heated to ~ 100°C in order
requires steam and carbon monoxide as the to establish the initial gas communication.
reactants in equal molar volumes. There are This would substantially facilitate the
three possible sources of CO. Carbon monoxide methanation reactions in the laboratory as
is generated at the combustion front. Another compared to the field where the reservoir
possible source is through thermal cracking of temperatures are typically lower.
the oil. The third source of carbon monoxide is
coke gasification. From Table 1 it can be seen One may now examine the data of the idealized
that the CO~~ ratio is about 0.1. Therefore, one-dimensional fireflood: the combustion-tube
the amount of hydrogen which could be generated tests.
by pyrolysis followed by water-gas shift
reaction is no more than 10% of the direct
generation of H2 by pyrolysis. The coke COMBUSTION-TUBE TESTS
gasification reaction, Equation (2) (discussed
more extensively below), yields carbon monoxide Eight combustion-tube tests were used to
as well as hydrogen. The CO could then undergo investigate various injection schemes. The
the water-gas shift reaction to produce more first four tests were conducted years (1976-77)
hydrogen. before the second group (1982). A very major
change in the experimental methods was that
For the water-gas shift reaction to proceed, the Tests 1-4 utilized unhomogenized, repacked core
carbon monoxide must flow into a steam-laden material whereas the second set used extracted,
region at moderate temperatures (<750°C).
11 11
homogenized, repacked core material, resaturated
When wet or dry combustion is utilized, these with reservoir fluids. (Details of the
conditions prevail over a region downstream of apparatus and experimental procedure may be
the firefront. (Wet combustion widens the steam found in Reference [8]). The injection fluids
plateau, but even in a dry fireflood the connate were as follows:
water generates a steam zone.) The water-gas
shift reaction could also occur downstream of Test Number 1 Dry air combustion
the coke gasification process as long as the Test Number 2 Wet air combustion
zone is wet. ( 1. 8 kg H20/m3 of air)
Test Number 3 Wet air combustion
In one-dimensional flow situations coke (1.3 kg H20/m3 of air)
gasification as described by Equation (2) is not Test Number 4 Wet air combustion
likely to occur during dry combustion. Coke is (2. 9 kg H20/n 3 of air)
present only in a hot narrow band just ahead of Te st Number 5 Dry air combustion
the firefront with no opportunity for contact Test Number 6 Dry combustion (95% oxygen)
with steam or water. Carbon monoxide is Test Number 7 Wet combustion (95% oxygen)
produced, however, by the high-temperature (6.6 kg H20/m3 of 02)
oxidation. It is necessary to resolve, Test Number 8 Wet combustion (95% oxygen)
therefore, why good wet combustion tube burns (3.3 kg H20/m3 of 02 )
generally do not produce hydrogen. The answer
is probably found in (13) and (15). At the Test Nos. 1, 2, 4 and 5 produced some hydrogen,
comparatively high now velocities used in but never steadily throughout the run. Test
combustion tubes, with the narrow steam zones No. 2 is representative of the gas analyses that
and at the much slower reaction rates of the were obtained (Figure 3). Hydrogen and hydrogen
water-gas shift reaction, what little hydrogen sulphide were correlated in that during Test
is generated is consumed by subsequent Nos. 1, 2 and 4 these gases appeared and
methanation. (Equation (2) specifies that equal disappeared relatively concurrently.
molar volumes of CO and H2 are produced while

678
SPE13661 L.E. HAJDO, R.J. HALLAM AND L.D.L. VORNDRAN 5

High-temperature burning was necessary, but not temperatures with the production of hydrogen and
sufficient, for the production of hydrogen. hydrogen sulphide suggests that thermal cracking
(Peak temperatures of active firefronts ranged contributes to the generation of these gases~
between 500-860°C with steam banks of
200-260°C.) A slight negative correlation
between CO and Hz can be read into Test Nos. 1, MULTI-DIMENSIONAL FLOW
2 and 4 whereas Test No. 5 suggests a positive
correlation. There was no apparent correlation The idealization of a one-dimensional fireflood
between hydrogen and methane. is not obtained in the field and only
approximately true in a combustion tube. In the
The earlier speculation regarding methanation reservoir, both macroscopic and microscopic
11 11

reactions in wet and dry tests was supported by multi-dimensional processes occur, e.g.,
the five runs conducted with air injection. The fingering, over-riding and by-passing around
dry tests produced 0.01-0.1% CH4 while the wet clay lenses and inhomogeneities.
tests produced 0.1%-0.5% methane.
Unfortunately, these numbers do not prove the The most important effect of channelling and
hypothesis because wet combustion generally by-passing is expected to be on the coke
produces wider steam plateaus than dry tests gasification process. Once two- and
and, therefore, more thermal cracking could three-dimensional considerations enter the
occur. For the current wet and dry tests the conceptual model, a distinctively different
temperature contours of the leading edges did process than the one described earlier for the
not differ sufficiently to account for the coke gasification reaction can be envisaged. It
order-of-magnitude variation in CH4 production is quite likely that in a reservoir pockets of
due to thermolysis alone. However, other hydrocarbons are left behind the principal
experimental variations (air flux rates, combustion front, or - seen another way - the
ignition temperatures, initial temperatures) air/oxygen flow streams alter, leaving
were sufficiently important as to preclude a incompletely burned coke exposed. If this coke
definitive conclusion here. is then washed by steam due to water injection
or migration of connate water, coke gasification
Considering that thermal cracking, the water-gas will occur. The heterogeneities and channelling
shift, coke gasification and methanation which account for the remnant coke also help
reactions involve CO, H2 and CH4 (carbon dioxide explain why the hydrogen is not consumed before
and water concentrations being not discernible) produ~tion. The hydrogen, either by itself, or
the lack of clear patterns in their productions aided 'by the steam/water, channels
argues that two or more of the reactions took through/around the front to a producer. If this
place in an uncorrelated fashion. For example, process were not feasible, one would expect that
temperature fluctuations, controlling reaction all of the hydrogen should be consumed at the
rates, alone could account for the incoherent firefront during simultaneous air/water
gas concentrations. Inhomogeneities in the core injection. The idea that substantial pockets of
(even in the homogeneous tests packing
11 11
coke are left behind the firefront could also
densities are known to vary) and consequent explain periods of sustained hydrogen generation
by-passing, as well as fluctuations in the water and production in the absence of continued air
saturations add to the variations in the gas injection.
productions.
The implications of multi-dimensional effects on
The pattern in the combustion-tube tests was thermal cracking are probably not major but, in
that non-ideal burning was necessary to produce general, by-passing and channelling could be
hydrogen, e.g. during start-up (Test Nos. 2 and expected to enlarge the heated zones. For
5), increasing water levels (Test Nos. 1 and 4) example, the fingering of oxygen ahead through a
or when plugging occurred (Test Nos. 2, 3 and cold oil bank would create local heating and
4). Channelling and/or incomplete burning, thermal cracking well in advance of the
leaving pockets of hydrocarbons, would be common combustion front. As had been noted, however,
to all these events. All four runs without thermal cracking appears to be limited in its
hydrogen generation left clean cores except at capability of forming gases. Therefore,
the ends. (In some instances, known channelling pyrolysis with channelling will not ultimately
(Test No. 6) or the possibility of plugging or produce more gases than would be generated in a
channelling, (Test Nos. 3 and 7) did not produce one-dimensional flow.
hydrogen - or possibly it was consumed before
production.) Heterogeneities in the reservoir and flow fields
could be expected to reduce the extent of the
The lack of a clear correlation between carbon water-gas shift reaction. The carbon monoxide
monoxide and hydrogen production, the apparent which is produced by the combustion, coke
link between hydrogen and unburnt hydrocarbons gasification and thermal cracking processes can
and the necessity for high temperatures strongly only be expected to react with H20 in the steam
suggest that coke gasification, rather than the zone ahead of the firefront. If channelling
water-gas reaction, is the source of Hz in a occurs, the gas would pass through/around the
combustion tube. It does not necessarily steam plateau to a producer.
follow, however, that the water-gas shift
reaction is a minor contributor to hydrogen
generation in situ. The concurrence of high

679
6 HYDROGEN GENERATION DURING IN-SITU COMBUSTION SPE13661

In summary, multi-dimensional considerations late June. This period provided an excellent

offer a mechanisms whereby: 1. coke test case for the mechanis~s of hydrogen
gasification can occur behind the combustion generation. With the resumption of air
front; 2. the generated hydrogen can channel injection, hydrogen production climbed to 10-15%
through/around an active front to a producer; 3. and tapered off to zero by early April. It
sustained hydrogen generation and production can appears, therefore, that despite the
occur during a period of water interruption of air injection for three weeks,
injection/invasion without the continued hydrogen production was relatively stable (with
injection of air. some decline). However, at the beginning of
April when water injection was terminated,
In light of the discussions above, the pilot hydrogen production ceased. The next appearance
production data can now be examined. of H2 , in early June, followed three weeks of
low-level (1.5 m3/d) water injection, still
maintaining air at 10 000-15 000 m3/d. It might
COMBUSTION-AREA BEHAVIOUR be possible to interpret the small decline in
hydrogen concentration (and large decline in H2
During the second half of November 1979, air and production volumes) in late March, and cessation
water injection was started into EX T4 (Figure 4). by April 3rd, as being a delayed response to the
Gas production response was rapid at EX T2, termination of air injection on March 10 rather
coming through in a couple of days. Hydrogen than the cessation of water injection on March
production (Figure 5) averaged 15% initially, 26. There are two arguments countering this
declining to zero in early January, a month view, however. Firstly, gas responses to
after the termination of air injection. (All injection changes were generally on the order of
gas concentrations are reported on a days rather than three weeks. But secondly, and
nitrogen-free basis.) Carbon monoxide rose ~ore significantly, the resumption of air
(<1%) during the air injection period, returning injection on April 1 did not produce hydrogen.
to zero quickly at its termination. Carbon The evidence seems to be that hydrogen
dioxide rose from 20% to 60% while H2S climbed generation requires water but not oxygen. The
to -3% in late December before resuming its coke gasification process fulfils this
former levels. The carbon dioxide and carbon criterion, whereas the water-gas shift reaction
monoxide are products of complete and incomplete does not. (It is assumed, with a good deal of
combustion, respectively. The carbon dioxide confidence, that carbon monoxide is not present
was also generated by several other reactions: in the reservoir for long after its generation;
thermal cracking, methanation and decomposition it is either consumed or produced. Therefore,
of carbonates. The CO production quickly if H2 is generated without air injection, coke
disappeared after the cessation of air gasification must be occurring.) Carbon
injection, due both to its low solubility in oil monoxide production appeared concurrently with
(compared to C02) and its higher reactivity, the start-up of air injection at the beginning
e.g., methanation with the abundant hydrogen. of February and April, i.e. some incomplete
The hydrogen sulphide and ethane production is combustion accompanied the re-ignition process.
due to thermolysis of the oil. With the As earlier, ethane and hydrogen sulphide
doubling of the air and water injection rates at production jumped sharply, but briefly, after a
the end of November an expanding steam plateau large slug of water was injected in late March.
would have developed through December. In this
region pyrolysis generated H2S and C2H6 • This High-rate (45 000 m3/d) air injection in early
process depleted the reactants rapidly at July was followed by high-rate (200m3/d) water
300-400°C (in a matter of hours or days) so that injection which was decreased to -10 m3/d beyond
the decline in the production of hydrogen August 2. In response, hydrogen production
sulphide and ethane is understandable. The became more consistent and, as large volu~es of
continued injection of cold water would also gas came through, almost simultaneously with the
have helped quench the reactions. The methane air slug, the hydrogen concentration was
had several sources: connate methane, maintained. Hydrogen production only tapered
thermolysis and methanation. The thermal off to zero in late September, almost three
cracking experiments showed that thermolysis months after the termination of air injection.
generates approximately 40% as much ethane as Again this indicates that the water-gas shift
methane. Since the ethane fraction peaked reaction alone was not responsible but that coke
briefly at 8%, perhaps a 20% CH4 concentration gasification and/or thermal cracking was
would be expected due to thermal cracking. involved. The cessation of H2S production by
Since the actual value was 40%, about half of early August suggests, however, that thermal
the methane can be attributed to the other cracking accounts only for part of the hydrogen
sources. generation.
A second slug of air at 40 000 m3/d in It is interesting to note that the sharp
February/March was followed by a brief period of increase of air injection rates from 15 000 m3/d
high-rate water injection (up to 100m3/d). to 45 000 m3/d caused a major jump in the CO
Thereafter, air alone was injected at production from 0.0% to 2%. Since combustion
20 000 m3/d until the end of March when low-rate was in progress, the evidence of incomplete
water co-injection was restarted at -2.5 m3/d. combustion suggests that new channels or regions
Air and water. injection was maintained until of oxidation were developed when the injection

68.0
SPE13661 L.E. HAJDO, R.J. HALLAM AND L.D.L. VORNDRAN 7

pressure was raised. The effect persisted only gases. Hydrogen was still being produced in mid
for 10 days. October at 5% concentration and in substantial
quantities. The occasional anomalous data point
The ethane production response was somewhat between the end of August and the resumption of
different from its previous history. Rather enriched air injection is either due to
than a pronounced, short-lived surge, a gradual difficulties in sampling low gas production
rise in the C2H6 concentration was recorded, volumes or analysis errors.
which peaked at 3% in August and declined
monotonically to 1% by November. It is not In October air/oxygen injection was resumed. A
clear why the H2S and C2H6 responses differed substantial burst of carbon monoxide (-3%) was
this time but not in the earlier instances. One produced. Again, this is attributed to
may speculate that in some regions the incomplete combustion and gas channelling at the
temperature was warm enough to cause thermal front which had been quenched for 3.5 months.
cracking and distillation of the light ends but It is significant that hydrogen production
not hot enough to create H2S (the sulphur being disappeared immediately, and reappeared when the
primarily in the larger molecules). CO concentration returned to zero. This is
considered to be clear evidence of the
importance of methanation reactions at the
MAIN-PATTERN ENRICHED AIR INJECTION Marguerite Lake site. (Although oxidation may
have contributed to the consumption of hydrogen,
Air injection was started into EX 4 in March the simultaneity of the reappearance of H2,
1983, soon blended up to high levels of oxygen during oxygen injection, with the disappearance
enrichment. For both operational and of CO is not well explained by oxidation.) The
reservoir-related reasons, injection rates renewed heating of bitumen caused a drop in the
varied until late May when pure nitrogen was methane/ethane ratio and the onset of H2S
injected for two weeks and then enriched air was production due to thermal cracking.
resumed (Figure 6). Water injection followed at
a high rate in July, increased further in August
and was then cut back to very low rates until DISCUSSION
the resumption of enriched air injection in
October. Gas production at EX 5 followed the The most important difference between the
EX 4 injection by four days with 10% hydrogen laboratory and the field data is the much
and 1% CO (Figure 7). The C02 fraction rose to greater production of hydrogen at the pilot.
70%, showing the rapid initiation of combustion Based on the thermal cracking data, it has been
at the temperatures left by the cyclic steaming. estimated that thermolysis could account for no
The methane/ethane ratio dropped substantially, more than 10% of the H2 produced at the site.
signalling the onset of ethane generation by Another important variance is the sustained
thermal cracking. The ratio, however, generally production of carbon monoxide in combustion
remained sufficiently high (-20) to indicate tubes but not in the field. As was found in
that thermolysis was a minor source of methane. this study, Ejiogu et al. [8], Alexander et al.
The H2 concentration climbed to 15%, but only [9], Moss and Cady [10], Miller and Jones [11],
dropped slightly during the nitrogen injection. and Fassish et al. [12] have all reported
H2S rose from 0% to -0.1% as the heated oil laboratory CO/C02 ratios ranging over 0.1 - 0.5.
zones underwent thermal cracking. The nitrogen Typically, with air injection these figures
injection reduced the C02 production, but imply 1-3% CO in the produced gases, while for
because of the substantial storage of C02 in the oxygen 20% CO has been reported [10]. Generally,
oil, this drop was not as sharp as it might have gas analyses have not been published for
been. In contrast, the CO concentration, which combustion projects, but the Badeau fireflood
had hovered around 0.5%, went completely to 0%. [13] produced only small quantities of carbon
Hydrogen production declined belatedly in the monoxide (0.3-0.6%), consistent with the
latter half of June to 8%. The methane/ethane Marguerite Lake experience. Since the appearance
ratio jumped as the interruption of the of hydrogen appears to be intimately related to
combustion process reduced the thermolysis. The the existence of CO, the anomaly needs
resumption of enriched air injection raised the resolution.
hydrogen concentration to 20%. An accompanying
small burst of CO indicated the onset of Fulford [14] conducted a broad study of the
exothermic reactions. The CH4/C 2H6 ratio fluid changes during the Badeau fireflood.
resumed its former levels. Hydrogen gas analyses were not reported, but
based on other evidence Fulford surmized that
Water injection in July caused no immediate incomplete oxidation of the oil produced carbon
change in hydrogen concentrations. Carbon monoxide which then underwent the water-gas
monoxide disappeared. Due to thermal cracking shift reaction (Equation 1) and then H2 and CO
of oil in regions newly swept by steam, hydrogen went to methane (Equation 6). If the
sulphide levels climbed to their highest values. methanation reaction were slower, this process
The methane/ethane ratio was higher in early would produce an abundance of H2 as compared to
July than in late June, due to the renewed carbon monoxide. If the water-gas shift
oxygen injection, but then dropped by late reaction is rate limiting, as would appear from
August to its lowest-ever value (~3.0) as the earlier analysis, then excess carbon
thermolysis became the principal source of both monoxide should be produced. Generally, in the

681
8 HYDROGEN GENERATION DURING IN-SITU COMBUSTION SPE13661

laboratory, very little H2 is yielded, and the 3. The generation of carbon monoxide by
latter view is borne out. In the field, on the in-situ combustion appears to be
other hand, without any CO production, copious substantially less than observed in
quantities of H2 have been found. So either the laboratory combustion-tube tests, and the
water-gas shift reaction is very complete difference is attributed to the much higher
(leaving no CO for methanation) or very little field temperatures. Incomplete combustion,
CO was generated by the combustion front in the yielding carbon monoxide, occurs
first place. The latter view is preferred here. predominantly upon the start-up of air
Recently, at the pilot the approach of a injection or when an increase of air
combustion front to a producer raised the CO injection rates creates new zones of
production only to 3%, with negligible H2. It oxidation.
would appear, therefore, that an active in-situ
combustion front produces less CO than 4. Hydrogen generation at the Marguerite Lake
anticipated and little H2, if any. The pilot is due principally to the creation of
explanation ~ay perhaps be found in the data H2 and CO by the coke gasification reaction
reported by Fassihi [12] both on their own work (steam and coke) followed by the water-gas
and that of Rardon and Gadelle [15] which show shift reaction.
that above 420°C there is a radical increase in
the C02/CO ratio of the gas production. The 5. Methanation reactions are a significant
present field pilot has found temperatures in part of the combustion process at Marguerite
excess of 900°C. It may be possible, therefore, Lake.
that heat losses, and the consequent lower
temperatures, in laboratory tests have prevented
the observation that CO production is lower at ACKNOWLEDGEMENTS
typical field combustion temperatures.
The work of the operating staff in data
The absence of carbon monoxide production due to acquisition and fluid sampling, and the gas
coke gasification requires explanation. analyses by C. Murray and his group are
Methanation alone does not provide an answer acknowledged. Encouragement to conduct this
because excess CO rather than H2 would be study and technical contributions by J. K.
obtained. It seems that in the regions upstream Donnelly were appreciated. The pyrolysis and
of the combustion front, where coke gasification combustion-tube tests were conducted by members
occurs, the water-gas shift reaction must of the Chemical Engineering Department of the
rapidly follows the coke gasification process. University of Calgary.
This view requires that the water-gas shift
reaction be fairly complete in situ and favoured Permission by the Albert a Oi 1 Sands Technology
over methanation, at least upstream of the and Research Authority and BP Resources Canada
combustion front. A recent patent [16], based Limited to publish these results, and reviews by
on laboratory tests, seems to support this Dome Petroleum Ltd. and PanCanadian Petroleum
contention; it was found that below 260°C, steam Ltd. are gratefully acknowledged.
and carbon monoxide undergo the water-gas shift
reaction in oil sands. In turn, in order to
explain the minimal H2 and CO production REFERENCES
downstream of an active combustion front, the
model requires one of three possibilities: 1) 1. Hyne, J.B., et al.: 11 Aquathermolysis of
the front generates little CO; 2) methanation is Heavy Oi 1s 11 , The Future of Heavy Crude and
favoured over the water-gas shift reaction ahead Tar Sands, Second International Unitar
of the front; 3) carbon monoxide is consumed by Conference, Venezuela (1977), 404-411.
organic reactions near the front. The authors•
opinion is that, during steam bathing of coke, 2. Hayashitani, M., Bennion, D.W., Donnelly,
hydrogen is produced by coke gasification J.K., and Moore, R.G.: 11 Thermal Cracking of
followed rapidly by the water-gas shift reaction Athabasca Bitur~en , The Future of Heavy
11

and that neither CO nor H2 is produced in Crude and Tar Sands, Second International
substantial quantities at the combustion front. Unitar Conference, Venezuela (1977),
233-247.

CONCLUSIONS 3. Davis, B.E., and Jennings, J.W.:

11
State-of -the-Art Summary for Underground
1. Aquathermolysis of the bitumen is barely Coal Gasification .. , J. Pet. Tech. (January
detectable and is not a significant 1984) 15-21.
contributor to the production of any
gases. 4. La 1 , D. , Otto, F. D. , and Mat her, A. E. :
..Solubility of CO, H2 and CO-H2 Mixtures in
2. Thermal cracking of the bitumen is a minor Athabasca Bitur:1en 11 , Proceedings of the 34th
generator of gases but can clearly be Canadian Chemical Engineering Conference,
recognized in the production of ethane and Quebec City (September, 1984) 510-513.
hydrogen sulphide and probably contributes
to the hydrogen production as well. 5. Me hrot r a , A. K. , and Svr ce k , W. K. :
11
Measurement and Correlation of Viscosity

682
SPE13661 L.E. HAJDO, R.J. HALLAM AND L.D.L. VORNDRAN 9

and Gas Solubility for Marguerite Lake

Bitumen Saturated with Carbon Dioxide 11
,

AOSTRA J. Research (September, 1984)

51-62.
6. United Catalyst Inc.: Physical and 11

Thermodynamic Properties of Elements and

Compounds 11
•

7. Guntermann, K., Gudenau, H.W., and

Mohtadi, M.: Mathematical Modelling of
11

the In-Situ Coal Gasification Process 11

,

Proceedings of the Eight Underground Coal

Conversion Symposium, (August, 1982)
297-306.
8. Ejiogu, G. C., Bennion, D.W., Moore, R.G.,
and Donnelly, J.K.: Wet Combustion - A 11

Tertiary Recovery Process for the Pembina

Cardium Reservoi r Pet. Soc. of CIM, paper
11
,

No. 78-29-32 (1978).

9. Alexander, J.D., Martin, W.L., and Dew,
J. N.: Factors Affecting Fuel Availability
11

and Composition During In-Situ Combustion 11

,

J. Pet. Tech. (October, 1962) 1154-1162.

10. Moss, ,J. T., and Cady, G. V.: Laboratory 11

Investigation of the Oxygen Combustion

Process for Heavy Oil Recovery Paper SPE 11
,

10706 presented at the 1982 California

Regional Meeting, San Francisco (March,
1982).
11. Miller, J.S., and Jones, R.: Laboratory 11

Experiments Simulating Fire Flooding

Through a Fractured Reservoir 11
,

U.S. Department of Energy Report

DOE/BETC/R1-83/2 (April, 1983).
12. Fassihi, M.R., Satman, A., Williams, R.L.,
Pettit, P., Grim, J., and Ramey Jr., H.J.:
Laboratory Combustion Tube Studies •••
11

Part II, U.S. Department of Energy Report

DOE/ET/12056-10 (March, 1981).
13. Fulford, R.S.: 8odcau In-Situ Combustion
11

Project; Produced Fluid Analyses U.S. 11

Department of Energy Report DOE/ET/12057-7

(July, 1981).
14. Fulford, R.S.: Produced Fluid Changes
11

during a Fi reflood Paper SPE 9005 11

,

presented at the Fifth International

Symposium on Oilfield and Geothermal
Chemistry, Stanford, California (May,
1980).
15. Bardon, C., and Gadelle, C.: Essai de 11

Laboratoire Pour L'Etude de la Combustion

In-Situ11
Institute Francais de Petrole,
,

Paris (May, 1977).

16. Hyne, J.B., and Tyrer, J.D.: The Use of 11

Hydrogen-Free Carbon Monoxide with Steam in

Recovery of Heavy Oil at Low Temperatures 11
,

Canadian Patent No. 1,170,444 (July, 1984).

683
 LEGEND

e - PRODUCER

Q - OBSERVATION

-~
TABLE 1 () - I NFILL WELL
THERMAL CRACKING OF COLD LAKE OIL

H2
CH4
10 h/360°C
mole fract1ons

6.07
23.74
8 h/420°C
mole fract1ons

4.73
41.16
• 4• Q03"'-
Q 04
~0 14 Surf. Loc.

"
'e 8
I
0

co 0.79 0.37 () 22
C02 49.44 12.06 QOS
se Qo6
C2H4 0.50 0.27
C2H6 9.30 21.12
H2S 0.7 0.49
() 21
C3Ha 5.34 11.05
o7Q.Qos
i -C4H1 o 0.90 1.81 10
n-C4H1o 1.86 4.58
T 66-R5W4M
i-CsH12 0.38 0.7 -P~ANT-S~; l
n-CsH12 0.69 1.26 ---..., I

C6H14 0.29 0.4 I I

en I I
-a Average MW 34.07 29.05
I I
"'" Mass Fraction 3.2% 16.7%
I I MARGUERITE LAKE
PHASE-A
PROJECT

I--" I I
\..N L __ _j _C1JB '!_E.lJ.!. WELL LOCATIONS
0' en
._.
0'
0~
250 500 FT
0 lOOM

Fig. 1-Marguerite Lake in-situ combustion pilot well locations.

 I(J()(](XXJJ().

LEGEND
IOOOJOOO. EQUATION 1
EQUATION 2
@ EQU'lTION 3
EQUATION 4
EClU'lTION 5
1000000. EQUATION 6

100000.

10000.
I-
z
a:
I-
(/)
HXXJ.
z
0
u
:L 100.
::J
..... a.
gj:.=:
10 .
...J
::J
C!l
w l.

0.1

0.01

0.001

0.000\J. 200. 600. 800. 1000.

[ °Cl

Fig. 2-Temperature dependence of reaction equilibria.

lO.r-----------~~----------~~----------~~----------~~----------~~------~--~~-----------,

l.

I-
z
w
u
a::
w
0...

z
0
I- 0.1
a:
a::
I-
z
w
u
z
0
u
(/)
cr
(!)

0.01

+ H2S
X H2
A CO
0 CH4

Fig. 3-Produced-gas concentrations for combustion-tube Test 2.

SP·E
 .f
I-
X
w
Cii
s
.8
-~
(/)
Q)
~
ci ci ci ci ci ci ci ci ci ci ci ci
-:t g ci ci ci ci ci ci ci
~ ~ ~ Cl
0 0 0 0 0 0 0 t" (.0 (T)
0 0 0
0 0 0 0 0 0 0 0 0 (T) N N N
I.J) 0 I.J) 0 I.J) 0
~ g I.J)
-:t -:t (T) (T) N N
(J..~O/EW) NOilJ3rNI ~3lt1M
(J..~O/EWl NOilJ3rNI ~ItJ
SPE 1366.1
 40.
:-..:
80.-
z
8
I-
a:
a::: 20.
I-
z
w
u -N
z
a
u 10.
(\J
::c
vi
8.
:-..:
z 2.
8
I-
a:
a:::
I-
z
w
u
z
a
I.
u
a
u
100.
so .•
!I
:-..: 80.- N
1-
z 70.· X
a w
~ 60. Qi
a: $
0:::
I- 50.· .2
z en
w 40. ·u;
u ro
z .0
a 80. Q)
u
(\J
a
20. ~
Q)
0>
u 10. £;
80. :sen
c
0
~
:-..:
70.- ·~
z 60. g
a 8
~50. en
ro
a: 0>
~
z
w
40.-
l
-o
0
~ 80. a:
a
u
...
::c
u

20.-
10.
10.-

\) I
ll)
til
u:
:-..:
8.
z
a
......
I-
a: 6.
0:::
I-
z
I
w
u 4.
z
a
u
~
(£) 2.-
::c
(\J
u
20.-
18.
:-..: 16.
z 14.
a
I- 12.
a:
0:::
I- 10.
z
w 8.
u
z
a 6.-
u
(f.) 4.
(\J
::c 2.
D.
NOV OCT
SP'E 1 3 66 1
 C!l
::::l
a: .¢
X
w
Q)
s
0
-~
U)
...J Q)
::::l
"")
~
c
0
(T)
(D
"t5Q)
~ :~
Ci5
(ij
z 3:
::::l "0
"")
c
t ro
.....
C'l "(ij
z "0
_! Q)
..c
-~
w
c
>-
a: I
CD
z:
c,
u::
ci ci ci ci ci ci 8 0 0 ci
0
0
0
0
0
0
0
0
0
0
Ul
N l\1
g Ul
Ul 0
~ g Ul
N l\1
CA~O/EWJ NOilJ3rNI ~3l~M
CA~O/EWJ NOil:J3rNI ~It:f 03H:JI~N3
SP·E 1 3 661
 80.
:::: 25.
az 20.-
1-
a:
a::
1-
15.
z
w
u
z
a
10.
u
N 5.
I
4.
:-.:
8.
z
a
.....
I-
a:
g: 2.
z
w
u
z
a
u I.
a
u
100.·
90.
:-.: 80.· u"i
z X
UJ
a 70. a;
~ s
a: so.
a:: 50. .8
1-
z
w
(/)
-~
u 40.
.c
z
a 80.
Q)
~
u
20.
~
N Q)
a g
Cl
u 10.
:s
(/)
80. c::
0
~ 70. -~
:-.:
z so.
g
a 8
~50. ~
Cl
a: -6
a::
1- 40. g
z "0
w 0
~ 80. 0::
a ,....I
u 20. c,
...
I
u::
u 10.
10.·
:-.:
8.
z
CJ
.....
I-
a: s.
a::
I-
z
w
u 4.
z
a
u
(0
2.
I
N
u
2.5
:-.: 2.
z
a
~ 1.5·
a:
a::
1-
z
w
u I.
z
a
u
(J) 0.5·
N
I
0.·
MAR
_SP'E 13 661