You are on page 1of 9

Reaction Kinetics of In-Situ

Combustion: Part 1-0bservations


Mohammad Reza Fassihi, * SPE, Stanford U.
William E. Brigham, SPE, Stanford U.
Henry J. Ramey Jr., SPE, Stanford U.

Abstract
Continuous analysis of produced gases from a small the overall rate will be controlled by that step. Also, the
packed bed reactor, at both isothermal and linearly in- rate of each series of steps must be equal in the steady-
creasing temperatures, has shown that combustion of state condition. However, there are no useful correla-
crude oil in porous media follows several consecutive tions for computing absorption and desorption of oxygen
reactions. Molar carbon dioxide/carbon monoxide in a porous medium.
(C0 2 /CO) and apparent hydrogen/carbon (H/C) ratios Consequently, consideration of these phenomena
were used to observe the transition between these reac- becomes extremely difficult for even simple reactions.
tions at different temperature levels. Theoretical expressions for postulated mechanisms often
A new kinetic model for oxidation of crude oil in contain 10 or more arbitrary constants. Because of the
porous media is presented in Part 2 of this paper (Page large number of arbitrary constants, several expressions
408). developed for widely different mechanisms often will
match experimental data equally well.
Introduction In general, the combustion rate, Re> of crude oil in a
The quantity of fuel consumed and the reaction rate porous medium can be described as 1-3
within the burning zone have been studied intensively for dC m
two reasons. First, the maximum oil recovery is the dif- Rc=---=kpo
dt 2
mc~, .................. (1)
ference of the original oil in place (OOIP) at the start of
the operation and the oil consumed as fuel. Second, one
of the most important factors in the economic evaluation where
of any in-situ combustion project is the cost of air com- Cm = instantaneous concentration of fuel,
pression. Excessive fuel deposition causes a slow rate of k = rate constant,
advance of the burning front and large air compression P02 = partial pressure of oxygen, and
costs. However, if the fuel concentration is too low, the m, n = reaction orders.
heat of combustion will not be sufficient to raise the
temperature of the rock and the contained fluids to a The reaction constant, k, is often a function of
level of self-sustained combustion. This may lead to temperature, T, as expressed by 4-5
combustion failure. Thus, it is necessary to understand
the reactions occurring at different temperatures as the k=w exp( -E/RT) , ......................... (2)
combustion front moves in the porous medium.
The most crucial and yet least understood zone of in- where E is the activation energy, w is the Arrhenius con-
situ combustion oil recovery is the burning front, where stant, and R is the universal gas constant.
temperature reaches a maximum value. The velocity of For heterogeneous reactions, the constant w is a func-
the burning front is controlled by the chemical reactions tion of the surface area of the rock. 4
involved. However, since crude oil is a mixture of Early studies of crude oil oxidation in a porous
hydrocarbons, it is necessary to consider a global medium were mostly qualitative. Differential thermal
description of the reaction mechanism. ana~sis (DT A) was performed on samples of crude
oil, -7 and the resulting thermograms represented the
Reaction Mechanism thermal history of each sample as it was heated at a
The reaction between fuel and oxygen in a forward com- uniform rate (usually 18°F/min [lO°C/minD in a con-
bustion process is a heterogeneous flow reaction. In- stant air flow (usually 277 mLimin [277 cm 3 /minD.
jected oxidant gas must pass through the burning zone to These thermograms indicated the presence of a number
make the burning front move. Within the burning zone, of exothermic reactions.
four known transport processes occur: (1) oxygen dif- Another method of analysis is thermogravimetric
fuses from the bulk gas stream to the fuel interface; then, analysis (TGA). Here, a sample of crude oil is weighed
perhaps, (2) the oxygen absorbs and reacts with the fuel; continuously as it is heated at a constant rate. The
(3) then combustion products desorb; and (4) products resulting curve of weight change vs. time or temperature
finally transfer into the bulk gas stream. If any of these indicates the occurrence of at least two reactions at dif-
steps is inherently much slower than the remaining ones, ferent temperatures. 8

'Now with ARCO Oil and Gas Co.

0197·7520/84/0008·8907$00.25
Copyright 1984 Society of Petroleum Engineers of AIME

AUGUST 1984 399


Finally, some investigators attempted to measure the factors affecting fuel availability. 15 Fuel availability
kinetic parameters by oxidizing crude oil in a packed bed showed roughly direct linear dependence on the Conrad-
and analyzing the produced gases. 2,9-11 Although the son carbon residue and an inverse dependence on API
use of an Arrhenius-type equation for the oxidation of gravity. Similar results were obtained by Showalter. 16
crude oil has been questioned by some investigators, 8 However, both the laboratory and field data show that
this approach has been accepted by others. rock lithology could be more significant than oil gravity
Tadema 6 described the existence of two main reac- in determining how much fuel is deposited. 2,12
tions, one at a low temperature and one at a higher Some investigators 15 have indicated that high oil
temperature, which are quite common for many crude saturation and low burning front temperature yield
oils. About the same time, Gates and Ramey 12 described greater fuel deposition. Others 1 have concluded that fuel
a third measurabJe oxidation reaction at the initial forma- deposition is independent of initial oil saturation but
tion temperature during air injection. Later, work on depends on the residual oil saturation at the steam
kinetics of crude oil oxidation in porous media con- plateau. In that region, steam distillation was found to be
firmed the occurrence of three major reactions in in-situ the main mechanism for fuel deposition. The amount of
combustion 2,9-11: (l) fuel deposition, (2) fuel combus- oil saturation was found to be nearly constant in that
tion, and (3) low-temperature oxidation. region. 17 In a similar study carried out in the USSR, oil
density and air flux were found to be important factors in
Fuel Deposition fuel deposition. 18
The terms "coke formation," "fuellaydown," "crack- TGA thermo grams of 15 different crude oils in a wide
ing," and "fuel deposition" all describe the process of gravity range (6 to 38°API [1.03 to 0.83 g/cm 3]) in the
leaving fuel on the solid matrix as the crude oil moves presence of both -nitrogen and air showed that (1) the ox-
forward. In this paper, the term "fuel deposition" will idation of crude oil started at higher temperatures and (2)
be used. less heat was released as the pressure was lowered. 8 For
The mechanism of fuel deposition has been discussed most of the samples studied, at 50 psig [345 kPa] and up
by many investigators. Some believe that the fuel to 500°F [260°C], the weight loss in the presence of
responsible for combustion is carbon, which is deposited nitrogen or air was about 60 %. It was deduced that
on the rock after the following pyrolysis reaction. distillation was the dominant mechanism for fuel deposi-
tion. At higher pressures, less distillation would occur
y and more fuel would be available for reaction. Only a
(CxHy)~xC~+-H2' ..................... (3)
2 few crude oils were sufficiently reactive at low
temperatures to generate the heat necessary to sustain a .
Others assume that the dominant mechanism is a low-temperature combustion front. 8
cracking phenomenon resulting from the breaking down Poettman et al. 19 discovered that the surface area has
of molecules because of heat and leaving a heavy residue a direct effect on fuel deposition characteristics of high-
called "coke" on the rock: gravity, paraffin-base crude oil. Their results showed
that high oil saturation (around 70%), relatively low
temperature (approximately 243°F [lIrC]), and air
flux around 12 scf/hr-sq ft [0.1 cm 3/cm 2 ·s] maximize
Early measurements of the deposited fuel on the reser- the amount of fuel deposition (3.65 %, carbon
voir rock were made in the late 1950's. Ignition studies weight/rock weight). Also, the minimum fuel required to
of coke-bearing samples· taken from South Belridge initiate a combustion wave was found to be about 1 wt%.
field 12 indicated fuel deposits ranging from 1 to 5.4 Given these conditions, sufficient time was necessary to
Ibm/cu ft [16 to 86.5 kg/m3] of bulk volume. This was promote the deposition of carbon on sand. Table 1 sum-
around 15 wt% of the initial oil in place. Extremely high marizes results from studies on the fuel deposition rate.
fuel deposits of 8.9 to 14.2 Ibm/cu ft [142.5 to 227.5
kg/m 3] were found in several reservoir rock samples that Fuel Combustion
appeared to be composed mainly of claystones or fine Most investigators have used the data of combustion tube
silt. The atomic H/C ratio of the coke was determined to experiments to define the combustion front and its
be 1.67 and its specific gravity averaged 1.38 (compared characteristics. 1,13,16.22-24 Their goal was to obtain cor-
with the gravity of the initial oil, which was 12.9° API relations relating the combustion front velocity and front
[0.98 g/cm 3]). temperature to experimental variables such as pressure
In another field experiment, a close agreement be- and air flux.
tween fuel deposition in the field (1.9 Ibm/cu ft [30.4 Although little quantitative work has been done on the
kg/m3] sand) and in the laboratory (1.8 to 2.21bm/cu ft reaction kinetics involved in forward combustion, there
[28.8 to 35.2 kg/m3] of sand) was reported. 13 This led are many papers on the oxidation of carbon indicating
to extensive laboratory work to identify the quantity and first -order reaction dependency on both carbon concen-
quality of fuel. However, rather than directly measuring tration and oxygen partial pressure. 25-26
the amount of fuel deposited, most investigators Similar results were obtained for combustion of
estimated the quantity of deposited fuel through residues of two different crude oils. 2 These results may
stoichiometric equations developed by Nelson and not be applicable to in-situ combustion because the ex-
McNiel. 14 . perimental conditions for fuel deposition probably did
Original and residual oil saturation, API gravity, not simulate the sequence of hydrodynamic and thermal
viscosity, atomic H/C ratio, and Conradson carbon effects experienced by a reservoir volume element as a
residue (ASTM) were found to be the most important combustion front approaches.
400 SOCIETY OF PETROLEUM ENGINEERS JOURNAL
TABLE 1-FUEL DEPOSITION RATE COMPARISON"
Rc =W exp( -EIRT)po/'C'::

E p
Crude Type Reactor Bed w(1/s) (JIg mol) m n (kPa)
Ref. 20 hydrocarbon vapor synthetic catalyst NA" 70,300 NA NA NA
hydrocarbon vapor natural clay NA 63,300 NA NA NA
Ref. 21 12.4°APloil silica sand NA 69,300 NA NA NA
(100/200 mesh)
Ref. 9 NA sand 46,000 NA NA 4x 10 4
Ref. 11 27°API sand quarry and 3.88 x 10 6 t 65,100 0 1tt 1 x 10 4
27°API kaolinite clay 3.66x10 6t 65,100 0 1tt 2.7x 10 4
27°API 8.33x10 6t 69,100 4.8x104

• R c ~ g fuel deposited on 100 g sand/s .


• • NA ~ not available.
t If a first·order reaction with respect to oxygen partial pressure is assumed, then w would be around 0.6 to 1.8.
tte m =q;poSo·

In another study, 10 the combustion reaction was found stream. In other words, more oxygen reacts with the
to be first-order with respect to oxygen partial pressure hydrocarbons than can be found in the produced gases.
but second-order with respect to carbon concentration. It This reaction occurs with the oxygen dissolved in the
also was found that this reaction is controlled by diffu- whole volume of the dispersed oil phase. Oxygen diffu-
sion although diffusion was discounted in a previous sion to the hydrocarbon molecules is faster than the ox-
study. 2 idation process. 10
Thomas et al. II showed that operating pressure had no Some of the products of oil LTD are alcohols,
effect on the activation energy but that it did affect the aldehydes, ketones, acids, and peracids. It also has been
Arrhenius constant. However, their data were not ac- reported that aldehydes promote the reaction. 30 It ap-
curate enough to detect low-temperature oxidation pears that the reaction rate is prolJortional to the specific
(LTD) reactions. Table 2 summarizes measured surface area of the matrix raised to a power of between 0
combustion-rate parameters from several studies. and L 10,31 In addition, certain soils and metallic
derivatives have catalytic effects on this reaction. 2,30,32
Low-Temperature Oxidation By subjecting a sample of oil sand to a heating
LTD takes place upon air injection either before or after schedule, Alexander et aI. ls found that LTD reactions,
ignition, when oxygen is available downstream from the over an extended period of time, promote t~e deposition
combustion front. 13,27-29 This may result from (1) in- of coke-like residues on the sand grains. For a 2L8°API
complete oxygen consumption in the high temperature [O.92-g/cm 3] crude, the amount of this material was
combustion zone, (2) air channeling around the front, or maximized at an LTD reaction temperature of 425°F
(3) a tilted combustion front surface. 10 [218°C] and then it decreased sharply to zero at 700°F
LTD reactions are characterized by either no carbon [371 °cj. They also found that if the heating rate
oxides or low levels of carbon oxides in the effluent schedule is decreased, more fuel is deposited as a result

TABLE 2-COMBUSTION REACTION RATES COMPARISON"


Rc = w exp( -EIRT)po,mc,{'

E p
Crude Type Reactor Bed w(1/Pa's) (JIg mol) m n (kPa)
Ref. 25 metallurgical coke fluidized 7.4x10-2 121,400 1 x 10 2
hardwood fluidized 2.0x10- 3 66,500 1 X 10 2
charcoal
graphite fluidized 1.3x 10 3 205,800 1 X 10 2
Ref. 26 coke c sand pack 6.67x10 5d 157,300 1 X 10 2
Ref. 9 NAb sand pack NA 125,500 4x 10 3
Ref. 2 13.9°API berea sand 2.37x10- 3 61,900 3x 10 2
(170/230 mesh)
13.9°API 80% berea, 2.43x10- 4 48,400 5x 10 2
20% clay
22.1°API berea sand 9.14x10-3 59,800 2x 10 2
Ref. 10 19.9°API berea sand 1.38x10-3 58,900 2x 10 2
precoked 60 mesh
Ref. 11 27° API sand quarry, 3.Ox10 1 58,800 1 X 10 4
kaolinite (5%) 1.8x 10 1 58,800 2.8x 10 4
4.84x101 58,800 4.8x 10 4

aRc ~ g carbon burnedll00 g sand-so


b NA ~ not available.
c Obtained from chromatographic decolorization of heavy oil (percolation); it was purged at 430°C with nitrogen to remove residual
liquid.
d This is obtained at high oxygen partial pressure. If a first-order dependency on oxygen partial pressure is assumed, the w would be
31.3.
e Here m is the exponent of (<!>PfS f)' where (<!>PfSf)= fuel laid down minus luel burned off.

AUGUST 1984 401


TABLE 3-LTO REACTION RATES COMPARISON'
Roo =W exp( -E/RT)P0 2m

W E P
Crude Type Reactor Bed (g mol/Pam ·cm 3 's) (J/gmol) m (kPa)
Ref. 9 NA" sand pack NA 44,200 NA NA
Ref. 28 14.4°API sandpack 9.2x102 73,700 0.46 NA
12.8°API sand pack 4.0 x 10 2 72,200 0.45 NA
NA sand pack 7.78x101 71,900 0.57 NA
NA sand pack 3.31 73,900 0.79 NA
NA sand pack 2.01 x 10 4 85,500 0.48 NA
NA sand pack 1.24x 10 4 78,900 0.31 NA
Ref. 2 13.9°API pressure bomb 5.1x10-2t 53,300 1 1.38x 10 3
Ref. 10 19.9°API sand pack 3.51 tt 70,500 0.5 1 x 10 2
A'=0.91 m2/g
27.1°API sand pack 1.6 tt 73,300 0.75 1 x 10 2
A'=0.91 m2/g
27.1°API sand pack 1.25 tt 73,500 0.75 1 x 10 2
A'=0.52 m2/g
Ref. 33 27.0 o API sand pack 9.4x 10- 2 70,700 0.7 9.1x103

-R O , = g mole oxygen/s·cm"'
--NA= not available.
tReported in terms of (A,'X 0,) ratherthanp02 ,whereA,'=8.9x 105 minute -, andX = fraction of oxygen concentration.
02
Thus, w=0.051 P0 •
2
tt(A,'C ) used rather than Po" where A,'=0.33xl0 6 minute-' (Ug mol)o.s and C =g mol/L. Therefore,
02 O2
w=3.51 x P 0$
02

of increased LTO reactions. They emphasized that LTO Experimental Equipment


reactions increase the oil viscosity and alter the distilla- Fig. 1 shows a flow diagram of the apparatus used in this
tion characteristics of crude oil. Crude oils generally study. Its main components were a combustion cell
.gain weight in an air atmosphere at low temperatures. 8 placed in a furnace, three continuous gas analyzers, and
'Light crudes are more susceptible to partial oxidation temperature and flow rate measuring equipment.
than are heavy oils. 10 The combustion cell consisted of a stainless steel tube,
Several authors have indicated that the ignition delay 1 Va-in. [2.85-cm] OD with O.02-in. [O.OS-cm] wall
corresponds to the initial part of the oxidation process thickness and 5-in. [12.7-cm] length (Fig. 2). A coil
during which a very small fraction of crude oil is altered. made up of a Va-in. [O.3-cm] stainless steel tubing of
They used the heat released by the oxidation reaction to O.Ol-in. [O.025-cm] wall thickness was wrapped around
estimate the spontaneous ignition time for crude the tube to preheat the air. The wrapping also created
oils. 28 ,29 LTO reactions thus affect the movement of oil two air cushions around the tube (Fig. 2). This promoted
and water banks and, ultimately, the performance of an more uniform temperature in the reactor.
in-situ combustion process. 15 Some of the reported LTO To prevent contamination at the bottom of the cell, the
rates are shown in Table 3. bottom flange had several grooves and a ledge. A

I
I
, I
I
1. Pressure gauge
I I 15. Pressure regulator
I I 2. Drierite 16. Needle valve "
R
~, I
0- 10
3. Three-way valve
4. 60-~ filter,
17.
18.
Fischer gas partitioner
Bubble flowmeter
2 5. Mass flow controller 19. 2-~ filter
I!J 11 6. Furnace 20. CO 2 analyzer
28 7. Temperature controller 21. CO analyzer
4 8. Digital temperature reader 22. Oxygen analyzer
2 -+
9. Temperature reader 23. Rotameter
10. Separator 24. 20-psia backpressure regulator
11. Ice bath 25. Wet test meter
12. Pressure transducer 26. Four-way valve
13. Pressure indicator 27. l-psi relief valve
14. Recorder indicator 28. Five-way valve
2
.. :
II, - FLOW LINES

[3] - - - ELECTRICAL LINES


- . - CALIBRATION LINES
CO CO
He
CO 2 CO2

Fig. 1-Schematic diagram of apparatus for kinetics studies of in-situ combustion.

402 SOCIETY OF PETROLEUM ENGINEERS JOURNAL


200-mesh Screen was inserted between the ledge and the
O-ring. Flanges were sealed with brass gaskets.
A thermowell inserted at the center of the reactor was
used to check the isothermal behavior of the reactor and
to read the temperature. Another thermocouple was
placed around the air preheater, outside of the reactor, to
indicate the radial temperature differences. The gas was
supplied by high-pressure gas cylinders and the gas flow
rate was controlled by an electronic mass flow
controller.
A 0- to 500-psi [0- to 3.4-MPa] adjustable back-
pressure regulator was used to obtain the required
pressure in the system. Another absolute backpressure
regulator was used to adjust the pressure in the gas
analyzers.
Two infrared analyzers were used to measure CO 2 and
CO levels. Oxygen was analyzed using a paramagnetic
oxygen analyzer. Also, a Fischer gas partitioner was
used for nitrogen and methane analysis. Detailed de-
scription of the equipment can be found in Ref. 34.

Procedure
Four different kinds of crude oil were used in these series
of runs. The viscosity of these oils ranged from 0.0001
to 0.015 cp [0.1 to 15 Pa' s] and oil gravity ranged from
10 to 18° API [1 to 1.05 g/cm 3 ]. The details of the oil FLOW PATTERN
properties are given in Table 4. Sandpack consisted of
Ottawa sand with an average sieve size of 28. For some
runs, fire clay was added to the sandpack. By using x-
ray diffraction, it was found that the clay consisted main-
ly of kaolinite.
The combustion cell was tamped with the oil sand
(around 40 g). Then dry sand was added to fill the tube.
No water was added to the oil sand because published
work has shown little effect of initial water saturation on
the fuel availability. 15 After leak testing, the tube was Fig. 2-Combustion cell.
placed in the furnace. Two sets of temperature histories
were applied to this system to determine reaction rate
data: (1) constant-temperature steps and (2) variable-
temperature, increasing at a constant rate. under high temperature for a short period of time.
Therefore, variable-temperature runs were conducted.
Constant-Temperature Runs. The bed was brought to
the desired temperature level while nitrogen was flowing Variable-Temperature Runs. If the bed is subjected to.
through the system. This commonly would last 1 to 1.5 a linear rise in temperature (such as in DTA) while air is
hours. Then air was admitted to the bed. This was con- flowing through the cell, it is possible to get useful infor-
tinued until no change could be seen in the exit gas con- mation about the reaction kinetics. However, this heat
centrations. The process was repeated at a higher rate should be slow enough to give a minimum
temperature at the same pressure and flow rate, and temperature differential across the reactor. This process
again at a higher temperature until all fuel was burned. was continued until there were no carbon oxides gases in
Similar runs were made at different pressures and flow the effluent line: This type of run is very similar to the
rates. behavior of a reservoir element ahead of the combustion
Although this method is straightforward, it is time front. A heating rate of 131 °F/hr [55°c/h] was chosen
consuming (40 to 50 hours) and does not simulate true to improve heat transfer and to provide an experiment of
frontal behavior where each element of the reservoir is reasonable duration. The temperature history described

TABLE 4-CRUDE OIL PROPERTIES OF FOUR FIELDS

Gravity C H Aiomic N S Molecular


Zone (OAPI) (%) (%) H/C Ratio. (%) (%) Weight
San Ardo Lombardi 11.2 84.94 10.86 1.53 0.96 4.74 593
Venezuela Jobo Field No.2 9.5 83.24 10.55 1.52 0.60 4.06 681
Huntington Beach lower main 18.5 84.50 11.53 1.64 0.86 0.00 386
Lynch Canyon' NA 10 74.97 10.81 1.73 0.71 0.00 435

• The properties are those of the displaced crude in Combustion Run 79-7. 34

AUGUST 1984 403


TABLE 5-INITIAL CONDITIONS FOR KINETICS RUNS

Flow Average Injected Burned Fuel/Sand Bed


Rate Pressure Oxygen Fuel Ratio Thickness
Run Oil (scc/s) (kPa) (mol%) Condition (g) (wt%) (cm)
101 San Ardo 16.7 690 21 isothermal 2.402 2.69 9
102 San Ardo 16.7 690 21 non isothermal 0.401 1.72 2
104 Huntington Beach 8.3 690 21 non isothermal 0.866 3.64 2
107 San Ardo 16.7 550 21 isothermal 0.604 2.81 2.1
108 Venezuela 8.3 550 21 nonisothermal 0.514 4.48 1.2
110 Huntington Beach 7.5 550 8.3 nonisotherma! 1.777 3.99 4.3
111 Venezuela 8.3 415 8.3 isothermal 1.894 3.56 5.5
112 Venezuela 8.3 415 8.3 nonisothermal 1.234 3.75 3.5
114 San Ardo 8.3 550 21 nonisothermal 0.775 3.45 2.5
(clay added)
115 San Ardo 8.5 275 8.3 non isothermal 0.677 3.39 2
(clay added)
116 Huntington Beach 10.0 550 21 isothermal 1.88 3.80 4.5
117 Huntington Beach 10.0 1035 21 non isothermal 1.260 3.16 4
118 Huntington Beach 10.0 138 21 non isothermal 0.730 3.41 2.2
119 Huntington Beach 10.0 552 21 non isothermal 1.035 3.52 3
120 Huntington Beach 10.0 1104 21 nonisothermal 0.984 3.63 2.5
121 Huntington Beach 10.0 690 21 non isothermal 0.757 3.41 2.2
122 Huntington Beach 10.0 138 8.3 non isothermal 1.103 3.44 3
123 Huntington Beach 10.0 17 21 nonisothermal 1.177 3.75 3
124 Huntington Beach 10.0 235 21 nonisothermal 1.441 3.32 4
125 Venezuela 10.0 828 21 non isothermal 0.784 3.22 2.5
126 Venezuela 10 690 21 non isothermal 10.242' 4
(original core)
127 San Ardo 10 690 21 nonisothermal 0.916 2.54 3
128 Huntington Beach 10 41 21 non isothermal 0.931 3.2 2.7
129 San Ardo 10 138 21 non isothermal 0.814 2.55 3.1
130 Lynch Canyon 10 690 21 non isothermal 3.3
(original core)
131 Lynch Canyon 10 690 21 non isothermal 3.321' 1.6
(original core)

'Total weight reduction of oil sand including water loss.

is similar to that observed in a long tube run with a com- Table 5 presents the initial pack conditions for dif-
bustion front velocity of approximately 0.4 to 1.2 ferent kinetic runs. A variety of parameters were tested
ili.!min [1 to 3 cm/min.]. Practically, this type of run is in different runs to observe their effects on the reaction
similar to a summation of constant-temperature runs at rates.
different temperature levels.
Analysis of the Results of Kinetics Runs
Operation of the Reaction Kinetic Apparatus
As mentioned earlier, two sets of experiments were car-
The operating pressure and flow rate were held constant ried out in kinetics runs. In this section, the results of
throughout each run. The backpressure regulator never both constant- and variable-temperature runs are dis-
digressed more than ± 1 psi [6.9 kPa] from the set cussed. Also, some parameters such as molar CO 2 /CO
pressure. The maximum pressure drop was 2 psi [13.8 ratio and apparent H/C ratio are graphed vs. reaction
kPa]. The mass flow controller maintained a constant in-
temperature to identify the reactions occurring. Finally,
jection rate. However, because of drift in the electric
a model is proposed to analyze these reactions and the
flow controller, the actual flow rate was about 1 % lower
results are discussed.
than the set flow rate at later periods.
The oil saturation in the sandpack (below 15 %) and
the bed thickness were chosen so that the oxygen con- Isothermal Runs. The data for a typical isothermal run
sumption and therefore the heat generation in the sand- are presented in Fig. 3. Each portion of the diagram
pack could be minimized. The oxygen consumption graphs the data obtained at a specific temperature level
never exceeded 15 % of the inlet concentration. Because vs. the run time. The changes in produced gas concentra-
of the low level of combustion, in most runs the most tions at each temperature level are plotted against the left
sensitive span range of the gas analyzers was used. ordinate and the sand temperatures are plotted against the
The heat generation in the pack never caused the bed ordinate on the right. Each abscissa represents the oxida-
temperature to exceed the average temperature by more tion time during which the fuel was in contact with air at
than about 5°F [3°C]. However, the existence of higher a given temperature level.
oil saturation, more reactive oil, or a thick bed caused The isothermal data in Fig. 3 for Run 107 exhibit
formation of hot spots in the pack. If the bed had not unusual kinetic reaction results. When the temperature is
been subjected togas drive, these hot spots could have raised, the reaction rate increases (more carbon oxides
led to spontaneous ignition. are produced) as expected. However, assuming usual
404 SOCIETY OF PETROLEUM ENGINEERS JOURNAL
0.5 3.0 800
420 RUN NO. 127
RUN NO. 107 410 2.5 AVERAGE PRESSURE ~ 690 kPa
FLOW RATE ~ 10 sec I s
<f- D CO 2 PRODUCED 400 0 TEMP
700
<ll
(5
6CO PRODUCED
o TEMPERATURE
390 -
d'
2.0
° CO 2 PRODUCED
x CO. PRODUCED
t
w
.s 380
370
'"
0
1.5 ° <lC
Oi
02 CONSUMED
600 ~
a:
=>
UJ S a:
(!)
z 0 360 <J)
« w
"-
<I:
I
6" '" 1.0 ::;
W

O:~E
f-
() 500
L
Z I-- 0.5
0
i= ui
<I: a:
a: :::> 0.0 400
I-- I-- 0 6
Z <I: TIME (hrs)
UJ a:
() UJ
z a.
0 ~ Fig. 4-Gas composition and temperature VS. time for San
UJ

r:~
Ardo oil.


I--
'---
Similar behavior was' observed with other crude
oils. 34 However, the first peak may be smaller or higher
0 than the second one, depending on the nature of the

P:k;
0 .5
:;; ~:
1.5
crude oil. For example, the combustion peak for the
heavy, viscous Venezuelan oil was much higher than the
low-temperature peak. 34 This indicates the propensity of
this crude oil for fuel deposition. In contrast, for a
TIME (hrs) French oil, because of its high reactivity with oxygen at
low temperatures, the first peak is much higher than the
Fig. 3-Gas composition and temperature VS. time for second peak. 33
Isothermal Run No. 107. To investigate these reactions and the chemical nature
of the fuel burned and to see the importance of distilla-
kinetic theory, one would expect the exit gas composi- tion and pyrolysis on these reactions, the apparent HIC
tion to level off to a nearly constant value. Although this ratio and the molar C0 2 /CO ratio were calculated. The
seems to occur at 446°F [230°C], at higher temperatures results of these calculations are discussed in subsequent
the carbon oxides concentrations gradually diminish paragraphs.
toward zero. This implies that the reactant has disap-
peared. However, at even higher temperatures, there is Molar C0 2 /CO Ratio. The molar ratios of the carbon
another rapid rise in the produced gas concentrations oxides (F co lCO) produced by the oxidation reactions of
followed by another decline toward zero. crude oil at different temperatures are shown as triangles
The overall result indicated by Fig. 3 is that the system in Fig. 5. The changes in F C0 2 /CO indicate the transition
acts as though the amount of material available to react between reactions at different temperatures. 25 .
increases with temperature. Also, since the relative por- It appears that the value of F C0 2 /CO has two distinct
tions of reacting carbon and hydrogen vary with regions. Below l,112°F [600°C], it is variable, increas-
temperature level, the chemical nature of the material ing from 0.8 to 2.0. At higher temperatures, it is nearly
available to react changes with temperature. This type of cimstarit at about 2 to 3, depending on the pressure. The
behavior was observed originally by Bousaid and calculated F co ICO at very low and very high tempera-
Ramey 2 but has not been considered in previous kinetic tures were excluded because of the lower accuracy in
models. Therefore, a new kinetic model formulation is measurement of small concentration of carbon oxides
needed. produced.
F C0 2 /CO being nearly constant at high temperatures
Variable-Temperature Runs. Fig. 4 presents gas com- indicates that carbon oxides are being produced by the
position and temperature vs. time for a non-isothermal same re\lction. By the same token, the reactions at low
experiment. The consumed oxygen (shown by ..:lC o ) temperatures must be numerous and nonunique because
and the produced CO 2 and CO are plotted against tl~e below 600 K, F co 2 ICO varies with temperature. •
left ordinate while the ordinate on the right represents the It has been reported that the molar ratIO of
temperature of the sandpack (in 0q. The abscissa shows CO/(CO+C0 2 ) for combustion reactions of charcoal,
the run time (in hours) from the beginning of air graphite, and coal in a fluidized bed is around 0.25 (i.e.,
injection. F C0 2 /CO =3). 25 This number being near the value of
There are two apparent peaks in the production of car- F co ICO obtained at high temperatures indicates that the
bon oxides at different temperatures. At high fuel burned in the combustion reaction is a heavy residue
temperatures (about 700 K), the amount of consumed similar to carbon in chemical characteristics.
oxygen is comparable to the amount of produced carbon Generally, the longer heating schedule and residence
oxides (i.e., CO 2 +0.5 CO). But at low temperatures, time of isothermal runs resulted in greater CO 2 produc-
the oxygen consumed is greater than the carbon oxides tion and greater values of F C0 2 /CO compared with
produced. At even lower temperatures, some oxygen is nonisothermal runs. 34 Apparently, the larger residence
consumed but no carbon oxides are produced. times resulted in more nearly equilibrium conditions.
AUGUST 1984 405
3.0
COMBUSTION REACTION
0'" HUNTINGTON BEACH OIL
0
u
0
- N
(J • VENEZUELAN OIL

U
• • SAN ARDO OIL
LL

c:i
10' r- >=
«
a:
25
0
-
U

0
N o 228
.
In p
O2
• 1 404
U
a:
Q «
...J
o VARIABLE T RUNS
I- 0 C, CONSTANT T RUNS
« ::;
II:

2 0lLO~~--'--~-'---'--'--J...L-L1.1-L00~~--'-~-'----L.--'--.L..l~1000

PARTIAL PRESSURE OF OXYGEN p (kP al


O2
. .;(
Fig. 6-Molar C0 2 /CO ratio VS. pressure.

of produced gases for each combustion run as a function


of temperature. These calculations were based on the
assumption that all the oxygen not observed in the exit
10-' J I I I
gas had reacted to fomi water. This assumption is not
450 500 550 600 650 700 750 800
rigorous because of the oCCurrence of LTO reactions.
TEMPERATURE, T (oK)
The calculated F HIC for four runs are graphed as
Fig. 5-Molar C0 2 /CO ratio and apparent H/C ratio for circles in Fig. 5. The values of F HIC at both very high
French oil. and very low temperatures should be discarded because
of less accurate measurement of the small concentration
of gases produced.
The average values of F C0 2 /CO calculated at high
In most runs, a general, decrease ;in the apparent HIC
temperatures in different runs are plotted vs. partial·
ratio with an increase in temperature was observed and
pressure of oxygen in Fig. 6 .. It appears that F C0 2 /CO in-
was found to follow a simple semilogarithmic
creases· slightly with higher pressure. The straight line
relationship:
shown is drawn through the data obtained using Hunting-
ton Beach oil and has the following equation.
In FH/C=a+bT, . . . . . . . . . . . . . . . . . . . . . . . . . . (9)
Fcoz/co=0.228 In P0 2 +1.404, ............. (5)
where a and b are constants and can be evaluated for
where Po is partial pressure of oxygen (in kPa). This each oil. 34
pattern is in accordance with the findings of others. 10 The atomic HIC ratios of the distillation cuts of crude
The pressure dependency of F C0 2 /CO may be a result oils used were calculated to investigate whether distilla-
of the occurrenCe of the following reactions. tion in porous medium is responsible for lowering F HIC
CO+ Y2 O 2 -C0 2 ......................... (6) (see Ref. 36 on methodology). .•
The HIC ratios of distillation cuts only dropped froni
and 1. 95 to 1.5 as the temperature w~s increased. This drop
is much less than the one ob§erved in kinetic ex"
2CO-C02 +C. . ......................... (7) periments. In these runs, at temperatures below 560 K,
F HIC was higher than the atomic HIC ratio of the parent
Calculated thermodynamic equilibrium· constants 35 for oil. However, as the temperature was increased, F HIC
these reactions showed that the second reaction is more decreased more until.it reached a minimum (around 0.5).
likely to occur ·~t pressures and temp'eratures encountered This is in accordance with the results of Burger and
in the experiments. In fact, at 800 K, the following rela- Sahuquet 31 who studied the effect of pyrolysis on the
tionship exists between F C0 2 /CO arid the partial pressure atomic HIC ratio of the fuel. They found that the carbon
of CO 2 , Pco z . content of fuel was almost constant at all temperatures,
but the hydrogen content was·only constant up to arou9d
F C0 /CO
2
=0.971 (Pco )0.5.
2
. ... ; . . . . . . . . . . . . (8) 572 OF [300°C] after which it decreased as the
temperature was increased. Hence, they observed that
Similar equations can be derived for F C0 2 /CO at dif- F HIC decreased linearly with an increase in temperature.
ferent temperatures by using the published values of the
eqUilibrium constant for this reaction. However, Eq. 5 Conclusions
cannot fully explain the observed behavior shown in Fig.
6 because Eqs. 3 and 6 have different constants. 1. At lower temperatures, the crude oil undergoes ali
oxidation reaction without generating carbon oxides. For
Apparent HIC Ratio. The apparent HIC ratio of the some reactive oil, heat released during this period may
fuel consumed, F HIC, was calculated from the analysis lead to spontaneous ignition.
406 SOCIETY OF PETROLEUM ENGINEERS JOURNAL
2. As the temperature is raised, distillation coupled Combustion Project," Oil and Gas J. (1961) 39, No. 23, 58.
with pyrolysis produces hydrogen gas and some light 15. Alexander, J.D., Martin, W.L., and Dew, J.N.: "Factors Affect-
ing Fuel Availability and Composition During In-Situ Combus-
hydrocarbons in the gas phase. A major part of these tion," J. Pet. Tech. (Oct. 1962) 1154-64; Trans., AIME, i25.
hydrocarbons are produced without being oxidized. 16. Showalter, W.E.: "Combustion-Drive Tests," Soc. Pet. Eng. J.
However, oxygen reacts with the remainder of these (March 1963) 53-58; Trans., AIME, 228.
hydrocarbons and, hence, medium-temperature oxida- 17. Wu, C.H. and Fulton, P.F.: "Experimental Simulation of the
Zones Preceding the Combustion Front of an In-Situ Combustion
tion occurs. Process," Soc. Pet. Eng. J. (March 1971) 38-46; Trans., AIME,
3. At higher temperatures, this reaction is completed 251.
and a heterogeneous reaction begins. Here, the reactants 18. Nagorny, L.A.: "Investigation of Coke Formation During In-Situ
are oxygen in the gas phase and a heavy residue of oil Combustion Process in Water Saturated Reservoirs with Oil of
Average Gravity," Neft Khoz. (March 1974) No.3, 58-61.
deposited on the solid matrix. 19. Poettman, F.1., Schilson, R.E., and Surkalo, H.: "Philosophy
Nomenclature and Technology of In-Situ Combustion in Light Oil Reservoirs,"
Proc., Seventh World Pet. Cong., Mexico City (1967) 3, 487-97.
A = cross-sectional area, sq in. [cm 2 ] 20. Crawford, P.B. and Cunningham, W.A.: "Carbon Formation in
A = specific rock surface area per unit bulk
I
Cat Cracking," Pet. Refiner (Jan. 1956) 35, No. I, 169-73.
21. Madgewick, G.G., Amberg, C.H., and Peterson, W.S.: "The
volume, cm - I or per unit weight of Pyrolysis of a Heavy Crude Oil in a Fluidized Bed," Cdn. J.
sample, cm 2 /g Chern. Eng. (April 1959) 65-70.
Cm fuel concentration, g fuel/1oo g sand 22. Martin, W.L., Alexander, J.D., and Dew, J.N.: "Process
Variables of In-Situ Combustion," J. Pet. Tech. (Feb. 1958)
C O2 O 2 concentration, mol % 28-35; Trans., AIME, 213.
CC0
2
CO 2 concentration, mol % 23. Benham, A.L. and Poettmann, F.H.: "The Thermal Recovery
CN 2 N 2 concentration, mol % Process-An Analysis of Laboratory Combustion Tube Data," J.
Pet. Tech. (Sept. 1958) 83-85; Trans., AIME, 213.
E = activation energy, Jig mol 24. Pehberthy, W.L. and Ramey, H.1. Jr.: "Design and Operation of
F HIC = apparent HlC ratio Laboratory Combustion Tubes," Soc. Pet. Eng. J. (June 1966)
F C0 2 /CO = molar C0 2 /CO ratio 183-98; Trans., AIME, 237.
25. Lewis, W.k., Gilliland, E.R., and Paxton, R.R.: "Low-
P0
2
= oxygen partial pressure, psi [kPa] Temperature Oxidation of Carbon," Ind. Eng. Chern. (June 1954)
PC0
2
CO 2 partial pressure, psi [kPa] 46, No.6, 1327-31.
R universal gas constant 26. Weisz, P.B. and Goodwin, R.D.: "Combustion of Carbonaceous
Deposits Within Forous Catalyst Particles: Part 2-Intrinsic Burn-
T = temperature, OF [0C] ing Rate," J. Catalysis (June 1966) No.6, 227-36.
w = Arrhenius constant, cm 3 I g mol· s 27. Ramey, H.1. Jr.: "Transient Heat Conduction During Radial
Movement of a Cylindrical Heat Source-Application to the Ther-
References mal Recovery Process," Trans., AIME (1959) 216, 115-22.
I. Wilson, L.A. et al.: "Some Effects of Pressure on FOIward and 28. Tadema, H.1. and Weijdema, J.: "Spontaneous Ignition of Oil
Reverse Combustion," Soc. Pet. Eng. J. (June 1963) 127-37; Sands," Oil and Gas J. (Dec. 14, 1970) 77-80.
Trans., AIME, 228. 29. Burger, J .G.: "Spontaneous Ignition in Oil Reservoirs, " Soc. Pet.
2. Bousaid, 1.S. and Ramey, H.1. Jr.: "Oxidation of Crude Oil in Eng. J. (April 1976) 73-80.
Porous Media," Soc. Pet. Eng. J. (June 1968) 137-48; Trans., 30. Crawford, P.B.: "Water Technology-Underground Combustion
AIME,243. of Hydrocarbons," Producers' Monthly (June 1968) 15.
3. Burger, J.G. and Sahuquet, B.C.: "Chemical Aspects of In-Situ 31. Burger, J.G. and Sahuquet, B.C.: "Les Methodes Thermiques de
Combustion-Heat of Combustion and Kinetics," Soc. Pet. Eng. Production des Hydrocarbures-Combustion In-Situ, Principes et
J. (Oct. 1972) 410-22; Trans., AIME, 253. Etudes de Laboratoire," Revue de L'Inst. Franrais du Petrole
4. Smith, J.M.: Chemical Engineering Kinetics, second edition, (March-April 1977) No.2, 141-88.
McGraw-Hill Book Co. Inc., New York City (1970). 32. Johnston, H.R. and Burwell, E.L.: "Carbon Deposition for Ther-
5. Carberry, J.1.: Chemical and Catalytic Reaction Engineering, mal Recovery of Petroleum-A Statistical Approach to
McGraw-Hill Book Co. Inc., New York City (1976) 495. Research," Producers' Monthly (July 1968) 8-15.
6. Tadema, H.J.: "Mechanism of Oil Production by Underground 33. Bardon, C. and Gadelle, C.: "Essai de Laboratoire Pour L'Etude
Combustion," Proc., Fifth World Pet. Cong., New York City de la Combustion In-Situ, " Revue de L'Inst. Franrais du Petrole,
(1959) Sec. II, paper No. 22,279-87. (May 1977).
7. Berry, H.1.: "An Experimental Investigation of Forward Com- 34. Fassihi, M.R.: "Analysis of Fuel Oxidation in In-Situ Combus-
bustion Oil Recovery," PhD dissertation, Texas A&M U., Col- tion Oil Recovery," PhD dissertation, Stanford U., Stanford, CA
lege Station (1968). (1981).
8. Bae, J .H.: "Characterization of Crude Oil for Fireflooding Using 35. Kanury, A.M.: Introduc.tion to Combustion Phenomena, Gordon
Thermal Analysis Methods," Soc. Pet. Eng. J. (June 1977) and Breach Science Publishers (1977) 366-68.
211-11. 36. Watson, K.M. and Nelson, E.F.: "Characterization of Petroleum
9. Weijdema, J.: "Determination of the Oxidation Kinetics of the In- Fractions," Ind. Eng. Chern. (Nov. 1935) 27, 1460-64.
Situ Combustion Process," KoninkIijke/Sheli E&P Laboratorium,
Rijswijk, The Netherlands (1968). SI Metric Conversion Factors
10. Dabbous, M.K. and Fulton, P.F.: "Low-Temperature-Oxidation °API 141.5/(131.5+ ° API) g/cm 3
Reaction Kinetics and Effects on the In-Situ Combustion Proc-
ess," Soc. Pet. Eng. J. (June 1974) 253-62.
cp x 1.0* E-03 Pa's
II. Thomas, G.W., Buthod, A.P., and Allag, 0.: "An Experimental
OF (OF-32)/1.8 °C
Study of the Kinetics of Dry, Forward Combustion-Final g mol x 1.0* E-03 kmol
Report," U.S. DOE Report No. BETC-1820-1 (Feb. 1979).
12. Gates, C.F. and Ramey, H.J. Jr.: "Field Results of South
in. x 2.54* E+oo = cm
micron x 1.0* E+oo p.m
Belridge Thermal Recovery Experiment," J. Pet. Tech. (Oct.
i958) 236-44; Trans., AIME, 213. psi x 6.894757 E+oo kPa
13. Moss, J.T., White, P.D., and McNiel, J.S. Jr.: "In-Situ Combus- ·Conversion factor is exact.
tion Process-Results of a Five-Well Field Experiment in
SPEJ
Southern Oklahoma," J. Pet. Tech. (April 1959) 55-64; Trans., Original manuscript received in the Society of Petroleum Engineers office April 11,
1980. Revised manuscript received Feb. 24, 1983 and accepted for publication Oct.
AIME,216. 12, 1983. Paper (SPE 8907) first presented at the 1980 California Regional Meeting
'14. Nelson, T.W. and McNiel, J.S. Jr.: "How to Engineer an In-Situ held in Los Angeles April 9-11.

AUGUST 1984 407

You might also like