In-Situ Combustion Model

Keith H. Coats,

Intercomp Resource Development and Engineering Inc.

Abstract
This paper describes a numerical model for
simulating wet or dry, forward or reverse combustion
in one, two, or three dimensions. The formulation is
considerably more general than any reported to date.
The model allows any number and identities of
components. Any component may be distributed in
any or all of the four phases (water, oil, gas, and
solid or coke).
The formulation allows any number of chemical
reactions. Any reaction may have any number of
reactants, products, and stoichiometry, identified
through input data. The energy balance accounts for
heat loss and conduction, conversion, and radiation
within the reservoir.
The model uses no assumptions regarding degree
of oxygen consumption. The oxygen concentration is
calculated throughout the reservoir in accordance
with the calculated fluid flow pattern and reaction
kinetics. The model, therefore, simulates the effects
of oxygen bypassing caused by kinetic-limited
combustion or conformance factors.
We believe the implicit model formulation results
in maximum efficiency (lowest computing cost), and
required computing times are reported in the paper.
The paper includes comparisons of model results
with reported laboratory adiabatic-tube test results.
In addition, the paper includes example field-scale,
cases, with a sensitivity study showing effects on oil
recovery of uncertainties in rock/fluid properties.

Introduction
Recent papers by Ali,1 Crookston et al., 2 and
Youngren 3 provide a comprehensive review of earlier
work in numerical modeling of the in-situ combustion process.
0197-7520/8010012-8394$00.25
Copyright 1980 Society of Petroleum Engineers of AIME

DECEMBER 1980

The trend in this modeling has been toward more

rigorous treatment of the fluid flow and interphase
mass transfer; inclusion of more components, more
comprehensive reaction kinetics, and stoichiometry;
and more implicit treatment of the finite difference
model equations.
The purpose of this work was to extend the
generality of previous models while preserving or
reducing the associated computing-time requirement.
The most comprehensive or sophisticated combustion models described to date appear to be those
of Crookston et al. and Youngren. Therefore, we
compare our model formulation and results here with
those models.
A common objective of different investigators'
efforts in modeling in-situ combustion is development of more efficient formulations and methods of
solution. This is especially important in the combustion case because of the large number of components and equations involved. For a given number
of components and reactions, computing time per
grid block per time step will increase rapidly as the
formulation is rendered more implicit. However,
increasing implicitness tends to allow larger time
steps, which in turn reduces overall computing expense. To pursue the above objective, then, authors
should present as completely as possible the details of
their formulations and the associated computingtime requirements.
The thermal model described here simulates wet or
dry, forward or reverse combustion in one, two, or
three dimensions. The formulation allows any
number and identities of components and any
number of chemical reactions, with reactants,
products, and stoichiometry specified through input
data. The transient, three-phase flow is represented
by Darcy's law, incorporating temperaturedependent relative permeabilities plus gravity,
533

viscous, and capillary forces. Vaporization/condensation phenomena are handled with allowance for
any component to be distributed in any or all of the
four phases. The fourth (solid or coke) phase is
immobile and in part provides capability for treating
in-situ coal gasification problems. Oxygen and fuel
consumption are calculated in accordance with the
specified reaction kinetics and calculated fluid flow
patterns.
The following sections describe the model formulation both in general terms and in greater detail,
with equations given. Several laboratory and fieldscale applications are presented together with a set of
sensitivity runs indicating the relative importance of
different types of input data. The laboratory-scale
model results are compared with the associated
observed or experimental behavior.
While the model formulation was developed to
simulate the in-situ combustion case, it applies also
to hot-water or steamflooding and geothermal
problems. Although certain features were introduced
in the formulation with in-situ coal gasification in
mind, no applications to that type of problem have
been attempted to date.

General Comparative Model Description

Crookston et al. 2 describe a combustion model using
the components oxygen, inert (noncondensable) gas,
light hydrocarbon, heavy hydrocarbon, water, and
coke. Their model accounts for the four chemical
reactions representing oxidation of heavy oil, light
oil, and coke and cracking of heavy oil. They treat
three-phase fluid flow in one or two dimensions
within the reservoir and use a two-dimensional heat
conduction calculation for heat flow in the overburden. The additional variable introduced by the
coke mass balance in their formulation is nc Ibm mol
coke/ cu ft of formation, and this mass is assumed to
occupy negligible volume - i.e., Sw + So + Sg =
1.0 in their formulation. Crookston et al. use the
mole-fraction constraints for oil and gas phases as
primary equations and, thus, solve simultaneously a
total of nine primary equations (six mass balances,
one energy balance, and two mole-fraction constraints) using Gaussian elimination. They use a
mixture of implicit and explicit treatment for
coefficients in interblock transmissibilities and
reaction terms.
Youngren 3 describes a three-dimensional combustion model that treats the five components water, oxygen, nonvolatile oil, and two arbitrary
volatile components - which can distribute between
the oil and gas phases. He neglects coke formation
and oxidation and treats the single chemical reaction
of heavy oil oxidation. Oxygen is assumed to be
consumed instantly on contact with fuel; therefore,
combustion rate equals oxygen flux. The model
simulates three-phase flow, and a one-dimensional,
overburden heat conduction calculation is used for
heat loss. Like Crookston et al., Youngren uses a
mixture of implicit and explicit treatment but, unlike
them, he solves only four primary equations
simultaneously, using Gaussian elimination. He
reduces his six primary equations (five mass balances
534

and one energy balance) to four equations by expressing the transmissibilities of Components 4 and 5
explicitly in mole fractions. All transmissibilities are
explicit in temperature. Both models simulate
convective and conductive heat flow within the
reservoir.
The thermal model formulation described here
allows any number and identities of components and
any number of chemical reactions. Each reaction
may have any number of reactants and products,
identified through input data. Reaction rates are
functions of temperature and reactant concentrations
raised to powers specified through input data.
Without recourse to different "versions," the
formulation applies to single-component (water)
geothermal problems plus two-, three-, and fourcomponent (e.g., including inert gas and/or
distillable hydrocarbons) steam flooding or hot-water
flooding problems and N c-component in-situ
combustion problems.
If a solid (coke) is one of the components, the
model accounts for four phases, with S4 representing
the solid-phase saturation. Thus, our saturation
constraint is Sw + So + Sg + S4 = 1.0. Our experience indicates that coke (fuel) deposition can
yield S4 values in the 0.04 to 0.08 range. This may
reduce formation permeability, and the formulation
allows permeability to vary with the S4 value.
The model uses no assumptions regarding degree
of oxygen consumption. The oxygen concentration is
calculated throughout the reservoir in accordance
with the calculated fluid flow pattern and reaction
kinetics. Youngren's assumption that combustion
rate equals oxygen flux precludes application of his
formulation to reverse combustion problems and
problems where an oxidation reaction is kinetic
rather than reactant limited. Also, as Youngren
points out, his assumption prevents prediction of
spontaneous ignition and his neglect of cracking or
coke prevents prediction of fuellaydown.
In defense of Youngren's assumption regarding
oxygen consumption, we must agree that most
combustion problems probably are reactant limited.
More important, however, we feel the assumption is
simply unnecessary. That is, the assumption limits
the applicability of a combustion model but yields
little, if any, benefit in reduction of required
arithmetic or computing time.
Unlike both models just discussed, the model
presented here is completely implicit. Our experience
in steam flood modeling 4-6 has led us to the belief
that total computing expense for thermal-model runs
decreases with increasing degrees of implicitness.
The model described here is a set of mass- and
energy-balance equations and constraint equations
written for each of a number of grid blocks
representing a reservoir. The equations are written in
finite difference form and include an energy balance
and one mass balance for each of N C components.
The constraint equations express the requirements
that fluid saturations must sum to unity and the mole
fractions of components in each phase must sum to
unity. Components are pure or pseudocompounds
such as oxygen, nitrogen, carbon dioxide, water,
SOCIETY OF PETROLEUM ENGINEERS JOURNAL

"heavy oil component," "light oil component," and

solid component (coal or coke).
For maximum efficiency, these equations must be
divided into groups: (1) N c + 1 "primary" or
"flow" equations (the energy balance and N ccomponent mass balances) and (2) N p + 1 constraints equations (one saturation constraint and N p
mole-fraction constraints). N p is the number of
phases, three or four. Constraint equations are used
to eliminate Np + 1 of the total Nc + Np + 2
unknowns present in the N c + 1 primary equations.
The result is a set of N c + 1 primary equations in
N c + 1 unknowns, which is solved by the reduced
bandwidth direct solution technique described by
Price and Coats. 7
A major fraction of total model computing time
may be expended in this direct solution of the
primary equations. The arithmetic required in this
direct solution is proportional to the cube of the
number of primary equations solved. 6 Thus, it is
important to minimize this number by using all
constraint equations to eliminate unknowns.
Crookston et al. retain two mole-fraction constraints
in their set of primary equations and, thus, use direct
solution (Gaussian elimination) to solve a set of nine
primary equations for problems involving N c = six
components. The formulation described here actually results in direct solution of a set of only six
primary equations for the case of six components if
one of the components is coke. This set of N c rather
than N c + 1 primary equations is attained through
use of the coke mass balance as a constraint
equation, which in turn is possible because the coke is
immobile. This is discussed in detail in the
mathematical model description.
This discussion suggests that Youngren's approach
of solving only four primary equations, regardless of
the number of components, may result in less
computing expense than the implicit formulation
described here. His formulation certainly will, or
should, require less arithmetic on a grid-block/timestep basis. However, we must consider the penalty
associated with the method by which he attains a
primary set of only four equations. The method
involves use of explicit mole fractions x4' xs, ... , xN.
in the transmissibilities of the Component 4, 5 .. ~
N c mass balances. A stability analysis of
multi component , single-phase flow shows that this
explicit treatment give a time-step limitation,
I:.t

< Sg !:.x/v gx '

where !:.x is the grid block dimension in feet and Vgx

is gas-phase interstitial velocity in feet per day. We
can substitutel:.y/v gy or !:.z/v z for !:.x/v gx In
steam flood or combustion probfems, Vg can reach
locally large values (e.g., 10 to 20 ft/D or more), even
with low Sg values. We feel this stability limit on
time-step size will cause in many cases a sufficiently
increased number of required time steps to more than
offset the reduced arithmetic per time step. However,
for some sufficiently large number of components,
the implicit formulation requiring direct solution of
N c + 1 primary equations will require more total
computing time than Youngren's formulation.
DECEMBER 1980

The model described here simulates wet or dry

forward combustion and reverse combustion in one,
two, or three dimensions. The model equations
describe mass transport by Darcy flow, incorporating
gravity, viscous, and capillary forces. The heat
transport includes mechanisms of convection,
radiation, and conduction within the reservoir and
conductive heat loss to the overburden and underlying strata.
The model allows any component to be distributed
in any or all of four phases (water, oil, gas, and solid)
as dictated by read-in pressure- and temperaturedependent component K-values. Each phase density
is a function of its composition, pressure, and
temperature. Gas-phase supercompressibility factor s
is obtained from the Redlich-Kwong equation of
state. 8
The generality in respect to component
distribution among phases allows study of the effect
of CO 2 solution in oil with attendant reduction in oil
viscosity. Also, CO 2 or any component may be
assigned some solubility in the liquid water phase. An
interesting potential of this generality is examination
of emulsion effects. Specifying solubility of water in
the oil phase or of oil components in the water phase
and adjusting input-component viscosity data can
yield emulsion-type viscosities for the oil or water
phase.
The formulation achieves this generality of
component distribution by assigning a user-specified
"master" phase to each component with a resulting
single set of master mole fractions (Xi J. Userspecified pressure- and temperature-dependent Kvalues then give the mole fraction of each component
i in phase j as x Ii = K vijXi . This treatment simplifies
model descriptIOn, coding logic, and variable substitution logic since it avoids the need to deal with
four sets of phase mole fractions.
The inclusion of a fourth (solid) phase in the
formulation represents a step toward the capability
to simulate in-situ coal gasification. In this case, the
initialized Phase 4 saturation would be large (e.g.,
0.8), representing the coal itself. The formulation
treats relative permeabilities as functions of normalized saturations (e.g., Sw = Sw!1- S4) to
represent more reasonably the case where S4 is large
and highly variable. The model reaction kinetics
require alteration in form for the gasification
problem since coal gasification reactions generally
reach an equilibrium between reactants and
products.

Mathematical Description of Model

The model consists of N equations where N = N c +
+ 2 and N p denotes the number of phases
present. N p = 4 if solid coke is present in addition to
the water, oil, and gas phases. N p = 3 if no solid
phase exists or appears. The model equations are

Np

Nc

1
1
Np
Nc

+ Np +

Component mass balance equations

Energy balance equation
Saturation constraint
Mole fraction constraints
2 Total equations.
535

The saturation constraint expresses the fact that the

Np phase saturation must sum to unity. The Np

mole-fraction constraints express the requirement

that the sum of all components' mole fractions in
each phase must sum to unity.
We refer to phases by the index or subscript J, with
J = 1 for liquid water, J = 2 for oil, J = 3 for gas,
and J = 4 for the solid phase. Components are
denoted by index or subscript I, with
Component
2
Water
2
Heavy-oil component
3
Gaseous component
4
Coke if a solid component is present
4
Gaseous or oil component if solid component is not present
5 - N c Gaseous or oil components.
I

through 4 are oSI' oS2' ... , oSN , oX I , oX2 , ... ,

oXN ' op, oT. All terms in Eqs. (through 4 can be
exprgssed in terms of these unknowns.
Treatment of Accumulation (Left-Side) Terms
The left side of the component mass balance can be
expressed as
V
Np
pjSjKvuXI )
tJ.t
j= I

- t( E
V

= tJ.t [(

Np

- ( jE=

pjSjKvuXI ) ]
n

Np

.( E

The N model equations for each grid block follow.

Component Mass Balances I = 1, 2, ... , N c

Np
f
j'"f pjSjKvuXI )

tJ.t

E [oP-m
m =I

pjSjKvuXI)]oPm' ............ (6)

j=1

where superscript f denotes the latest iterate value,

subscript n denotes the time level n value, and N is
the number of total unknowns. The N unknowns
( oPj J are changes over the iteration

NR

.(tJ.p + Mej - 'Y jtJ.Z) +

+-

Np

oP I
oP2

q Ir - ql ...... (1)

r=1

= oSI

= oS2

oPNp
OSNp
oPNp + I = oX I
oP Np +2
oX2

r=1
-QHL . ........................... (2)

OPNp +NC
OXNc
OPNp+Nc + I = op

and

OPNp+NC+2 = oT . ............... (7)

Saturation Constraint
Np

j=1

5S j =

o ........................ (3)

Mole-Fraction Constraints J

= 1, 2, ... , N p

Nc

Xu = 1.0. . ..................... (4)

1=1

We have suppressed the grid-block subscript indices

i, j, and k on all terms in these equations.
The mole fraction of component I in phase J is
calculated as
Xu

= KVUXI

, ........................ (5)

where Kvu are equilibrium K-values and XI is the

mole fraction of component I in component Fs
primary or "master" phase. Thus, if phase f is phase
I's primary phase, then Kvlf
1.0. If component lis
not present ("soluble") in phase J, then K vU = o.
The N unknowns corresponding to the Eqs. 1

536

Differentiation of a product of terms is performed by

the normal chain rule,
o
-

OC

(abc)f = (ab)f -

+ (ac)f

ob
-

+ (bc)f

oa
- .

The derivatives in the last term of Eq. 6 are evaluated

at the latest iterate (f) values of all unknowns. The
energy-balance accumulation term is expanded in the
same manner as Eq. 6.
Interblock Fluid-Flow (Right-Side) Terms
The interblock flow terms are treated implicitly. The
flow term for phase J in the component mass balance
is expanded as
[TPjKvUXI k

rj

J.l.j

(tJ.p + Mej - 'Y jtJ.Z)]

== [TPjKvUXI k

rj

J.l.j

(tJ.p+Mej-'YjtJ.Z)Y

SOCIETY OF PETROLEUM ENGINEERS JOURNAL

Component 6. The reaction stoichiometry is

slr oil 2

m=1

- 'Y J.:l

z )YbP m.

. . . . . . . . . . . . . . . . . . . . . (8)

If Eq. 8 denotes interblock flow between two

adjacent grid blocks, denoted as Block 1 and Block 2,
then the weighting of terms is upstream for (p/J-t)J,
k r]> xIJ, and
'Y J

= w'YfJI + (1 -

The stoichiometric coefficient of the first reactant,

S I " is 1.0 by definition. The reaction rate in moles of
first reactant per day for the grid block is
moles Component 2
Rr
D
1.8Er

= VAr e

w) 'Yf 12 , ..... (9)

Interblock heat flow between Grid Blocks 1 and 2

includes transport by convection (enthalpy flow
associated with mass flow), conduction, and
radiation. The flow due to conduction and radiation
is

Hr

where A is the cross-sectional area normal to flow, L

is the distance between grid-block centers, and <:3 is
read as input data. We have generally used <:3 = o.
In accordance with the implicit representations
throughout the model, this heat flow is calculated as
N
(J
M
q = qf+
-[-(TI-T2)

AA

+ -<:3Li - (4
TI -

4)]fbPm

T2

(15)

(11)

= RrHrlBtu/D

. .................. (16)

q2r = - Rr moUD
q6r = -s2r R r

................................ (10)

(JP m

The rates of appearance of energy and Components

2, 6, 3 (C0 2 ), and 1 (H 2 0), which are required in
Eqs. 1 and 2 are

AA
( T -T ) +<:3
- Li - (4
TI -T24) Btu/D,
I
2

E
m=1

1r

The heat of reaction is

Interblock Heat-Flow Terms

= LM

RT (S2P2 X22f

. (S3 P3X63f2r

where W is (Vh /[( Vh + (V>z] .

+s2r 0 2- s 3r C0 2 +s4r H 2 0 ........ (14)

q3r

s3r R r

qlr = s4r R r

and
qNc+l,r

= Hr

, ..................... (17)

where "component" Nc + 1 is energy.

The reaction rate is computed at the start of each
iteration as

The interblock weighting of enthalpy in the energybalance interblock flow terms is

H J = ()HJup +(1-()HJdown ' ........... (12)

where 0.5 :5 () :5 1.0 and "up" and "down" denote

upstream and downstream blocks for flow of phase
J. We use () = 0.5 as a default value, with provision
for the user to change the value through input data.
Treatment of Chemical Reaction Terms

The rate of the chemical reaction number r is

1.8Er

RT

with units of moles per day for the first reactant. For
gaseous reactants, the user may specify a partial
pressure in place of concentration. For the cracking
reaction, an additional multiplicative factor of 1 [S4/ (S4)max]n is used in Eq. 13. This factor was
proposed by Crookston et af. as a means of limiting
the coke (fuel) formation. If (S4)max is read as zero
in input data, then this additional factor is not used.
For illustration we consider the specific reaction (r)
of oxidation of a Heavy Oil Component 2. Oxygen is
DECEMBER 1980

.[e

RT (S2 P2X22)n

. (S3 P3X63

YfOP
2r

1r

m , . . . . . . . . . . . . . (18)

where the grid-block volume Vand the reaction-rate

constant Ar are constants. The user may specify
partial pressure rather than concentration for a
gaseous reactant so that pX63 may replace S3P3x63
in the preceding equations.
Temperature in the Arrhenius exponent is
evaluated as 0.5 (Tn + 1 + Tn) and oil saturation is
evaluated as 0.5 (S2n+ I + S2n) if all iterate values
sf. are positive and as S2n + I if any iterate value sf. is
negative.
Treatment of Well Terms

We discuss the case of a production well on

deliverability completed in three layers, denoted by k
= 1, 2, and 3. Each layer is assigned a timeindependent productivity index Pl k . The phase
production rates from each layer are
537

bottomhole well bore pressure opposite the center of

grid block k. A flowing well bore gradient is
calculated for each layer as

I~

I.J (krI
-PI'I)
I.J (krI)
p,
1=1
III
k 1=1
III k
.................................. (20)

I'wbk

which is a volumetric average gradient. The l' wbk are

calculated at the start of each time step and are used
unchanged for succeeding iterations. The flowing
well bore pressure then is calculated as

f
f
f

KuU

qIk = PI k ( -krI) (Pk-Pwbk) , .......... (19)

III k
where Pk is grid-block pressure and Pwbk is flowing

H,U

(p, T)
(p,T)

(p)

Here, X denotes dependence on {Xi} and S denotes

dependence on {S I}. The Appendix describes the
PVT treatment of rock and fluid properties.
Treatment of Constraint Equations
The saturation constraint (Eq. 3) is expressed simply
at each iteration as
Np

aSI

O ....................... (24)

1=1

In the mole-fraction constraints (Eq. 4), we have

Pwbk

= Pwbk-I + (I'wbk-l<Xk-1
+I'wbk<Xk)(Zk -Zk-I), ......... (21)

where Zk is depth to the center of grid block k and

Pwbl is a specified limiting bottomhole flowing
pressure. The weight factors account for differing
layer thicknesses, <Xk = 0.5b.Z k l(Zk - Zk-I) and
<Xk-I = 1 - <Xk' where b.Z k is the (vertically
measured) thickness of layer k.
The production rate of component I from layer k is
calculated as
qjK

Pl k [itl ( : kr )IKuUXj (P-Pwb)

+Pl k

N -a
E
m=1 ap

3
(PE
11
1=1

J:

kr KuUXj
I

e
. (p-PWb)Lap m , ............... (22)

where Pwbk is held constant in the differentiation.

The enthalpy production rate in Eq. 2 from grid
block k is
QHK=PI k [

(~kr)HI(p-pWb)r
11
I
k

1= I

E
1= I

(~k)
11

r I

............. (23)

Heat-Loss Treatment
The implicit calculation of heat loss is calculated as
described in detail in Ref. 4. The QHL term in Eq. 2
is replaced by a term of type <XI - <X2aT.
Functional Dependency
The fluid and rock properties are dependent on the
system variables:
P1,111'1' I

krI

"

k
Mf
538

f
f
f
f
f

(p,T,X)
(S, T)
(Sg)
(S4)

(T)

Xu

(KuUXj)c+

N
a
C
E
-(KuijXj ) aP m ,
m=1 ap

m
....................... (25)

so that the constraints (Eq. 4) become

j~1 [m~1
=

1-

a:

Nc
E
( j=1

(KuUXj) apmJ

KuUXj

)C

. . ............. (26)

Consolidation of N c + 2 Model Equations

The preceding sections describe the expansion of all
terms in Eqs. 1 through 4 in linearized expansions
consisting of coefficients times iterate changes aPm'
m = 1, 2, ... , N, where N = N c + N p + 2. Only
the first N c + 1 of these N equations possess rightside flow terms that introduce unknowns {aP m} of
neighboring grid blocks. These N c + 1 equations are
designated "primary" equations. The remaining N p
+ 1 equations are constraint equations, and they are
used to eliminate N p + 1 unknowns from the
primary equations. This elimination is possible
because the constraint equations for a given grid
block involve only unknowns {aP m} of that grid
block.
For that moment, we neglect the interblock flow
terms in Eqs. 1 through 4 and consider the accumulation and all other terms. After the expansion
described previously, the N equations can be written
as
N

E
1=1

CijaP I

bi; i

1,2, ... , N, ...... (27)

where we locally use i as the equation number and j

as the column or unknown index. The equations for i
= 1, 2, ... , N c are component mass balances. The
equation for i = N c + 1 is the energy balance. The
equation for i = N c + 2 is the saturation constraint.
The equations for i = Nc + 3, Nc + 4, ... , N are
the mole-fraction constraints. The coefficients Cij
and b i terms are evaluated at latest iterate (0 values
of the unknowns {p. }.
The last N p of t6ese equations are mole-fraction
constraints and coefficients Cij are all zero for j = 1,
SOCIETY OF PETROLEUM ENGINEERS JOURNAL

2, ... , N p , since unknowns oP I through oP Np are

oSI through OSN and the mole-fraction constraints
do not involve fl~id saturations. Coefficients Cij for
j = Np + 1, Np + 2, ... , N are generally nonzero.
We diagonalize this submatrix of N p equations by
Gaussian elimination so that the last N p (molefraction constraint) equation's matrix is reduced to
j= Np +1
i=N c +3
i=Nc+ 4

i=Nc +Np +2
j =

N p +2

2Np +1

0
0

... 2Np

v'

i=Nc +4

v'

- f
oP)

+ oP) , ................. . (30)

where a damping factor may be applied to oP) in Eq.

30 if one or more of the iterate changes OP) are
excessive.
C, T, and R in Eq. 29 are re-evaluated using the
latest iterate values of {p) J, and Eq. 29 is solved
again. These iterations are terminated when all the
following are satisfied.
max
j= 1,Np

loS)

:::;

TOLS (0.01) ,

max
i=1,Nc

10Xi

:::;

TOLX (0.01) ,

max

I op I

:::; TOLP (1.0),

max

loT

:::; TOLT (1.0) ,

and

... N

i=Nc +3

- f+ I
OP)

OPNp + 1
OP Np + 2

v'

..... (31)

where "max" denotes maximum over all grid blocks

and default closure tolerance values are given in
parentheses.

v'

Variable Substitution
i=NC +Np +2
v'
oPN
v' ,
................................ (28)

where the check marks indicate nonzero coefficients.

These diagonalized N p mole-fraction constraints are
used to eliminate Columns (unknowns) j = N p + 1,
N p + 2, ... , 2Np in the first N C + 1 equations of
the equation set (Eq. 27).
The saturation constraint (Eq. 24) then is used to
eliminate Column N p from the first N C + 1
equations of Eq. 27. This now leaves Nc + 1
equations (which involve right-side interblock flow
terms) and Nc + 1 unknowns. We started with Nc
+ N p + 2 unknowns, but we eliminated N p
unknowns (mole fractions) using the N p molefraction constraints and eliminated one unknown
(saturation) using the saturation constraint.
.
The coefficients in the reduced mole-fractlOnconstraint matrix (Eq. 28) are saved or stored for two
reasons: (1) they are needed to eliminate the mentioned columns from the expansions on the interblock flow terms and (2) they are needed after
solution of the N C + 1 equations to compute the
eliminated unknowns. Similarly, oSN ' which was
eliminated, is calculated after solution bf the N C + 1
equations.
The N C + 1 primary equations can be written in
matrix form as

Cr: =

Ll(T~) , ...................

(29)

where C and Tare (Nc + 1) x (Nc + 1) matrices

and Rand Pare N C + 1 column vectors. These
equations are solved by the reduced bandwidth di~ect
solution technique. 7 Eqs. 28 and the saturatIOn
constraint then are used to calculate the eliminated
unknowns. The cumulative changes over the time
step then are augmented as
DECEMBER 1980

The N p + 1 constraint equati~ns. are used to

eliminate N p + 1 unknowns, resultmg m a set of N c
+ 1 (mass- and energy-balance) primary equations in
Nc + 1 remaining unknowns. In the general case,
where all four saturations are nonzero, the
eliminated unknowns are oX I , oX2 , oX3 , oX4 , and
OS4' leaving oSI' oS2' oS3' oX5 , , OXNc' op, and
Tas the N C + 1 "primary" unknown~.
.
If any saturation S) is :ero f?r some Iteratl?n, then
no mole-fraction constramt eXists for phase J and we
ostensibly have on: less unknown (o~) and one less
equation. In actuality, however, we still have N C + 1
primary equations in N C + 1 unknowns, with one
less constraint equation counterbalanced by one less
eliminated unknown. That is, oX) is not eliminated
and oX takes the place of oS) in the N c + 1
unknow'ns. The oS) is known a priori as - S),,!, An
example of this substitution of oX) for oS) IS the
onset of the superheated water state, where SI
becomes zero but oX I still requires calculation since
the gas phase will contain a water mole fraction of
K u13 X I
. .
..
One important exceptIOn to thiS deSCrIptIOn of N c
+ 1 primary equations in N C + 1 primary variables
is introduced by the immobility of coke. If coke is
one of the N C components, we can reduce the system
of N equations to only N C rather than N c + 1
primary equations. This is important because the
computing time required for simultaneous, direct
solution of the primary equations (1) is proportional
to the cube of the number of primary equations and
(2) can constitute 80llJo or more of total computing
time. 6
If coke is one component, then the primary
equations (involving right-side flow terms) are the
energy equation and the N C - 1 mass -balances for
the N C - 1 mobile components. The coke massbalance equation becomes a constraint equation. The

539

mole-fraction constraint for Solid Phase 4 is used to

eliminate OX4 [mole fraction of Component 4 (coke)
in the solid phase] from all N equations and then the
resulting coke balance is used to eliminate OS4 from
the primary equations. The saturation constraint is
used to eliminate oSjrnax from the primary equations,
where jmax is less than 4 and is the largest value of j
for which
is positive.

SJ

Miscellaneous
Model initialization requires specification of initialphase mole fractions, temperature, gas/oil and/or
water / oil contacts, and pressure at a specified subsea
datum. The model initializes pressure either by
capillary-gravitational equilibrium or by setting
pressure to a constant, at the user's option. If a gas
phase is initially present, the model automatically
will recalculate the mole fraction of a specified
"swing" component to attain a sum to unity of gasphase mole fractions. This swing component
generally is specified as N2 or a N2 + COx
pseudocomponent. The reason for this internal
resetting of a gas-phase mole fraction is the difficulty
in calculating by hand the initial mole fractions of
distributed components so that the gas-phase molefraction sum will exactly equal unity.
The material balance at each time step for each
component is calculated as

1 + mass in place - (initial mass in place

+ cumulative injection - cumulative production

+ net production by reactions)

+ max (cumulative injection, cumulative
production)

+ max (reaction production, reaction consumption) .

The energy balance is calculated in the same
manner with "energy" substituted for "mass."
Values of 1.0000 for these balances simply indicate
that the finite difference equations are solved accurately.
Automatic time-step control is used with time-step
size controlled by the previous step size and the
previous step changes (maximum over grid) in
saturation pressure, temperature, and mole fraction.

Applications
In all the model runs described below (S4)rnax and
heats of vaporization for volatile oil components are
zero. Our experience indicates a negligible effect of
heat of vaporization. No temperature dependence of
relative permeability or capillary pressure is present
in any of the problems. Units missing from problem
data listed in tables are given in the Nomenclature.
Partial pressure is used in lieu of concentration for
oxygen in all oxidation reactions.
While the ASTM form for oil-component viscosity
(Eq. A-20) currently is programmed in the model, the
model runs described here used the form
II
rJJ
--

540

ae b / T

for liquid phases J = 1 or 2, which is the

representation used by Crookston et al. 2
All model runs were performed using midpoint
convective weighting (8 = 0.5 in Eq. 12).
Smith and Perkins Experiment

Smith and Perkins 9 conducted wet, forward combustion experiments in a vertical, adiabatic combustion tube. They injected a constant-ratio
air/water stream into a 700-md, 120F vertical tube
5.75 ft long after preheating the top to 600F. They
reported (1) measured temperature profiles at 2, 4,
and 6 hours, (2) percent unused oxygen in the
produced gas phase vs. cumulative air injection, and
(3) amount of oil burned. They also described a
numerical model and compared their calculated with
their experimental results.
Smith and Perkins reported two experiments - a
low-pressure test using Crude A and a high-pressure
test using Crude B. Results discussed here are for the
8I5-psia low-pressure run. Crude A is very light with
API gravity and viscosity of 36 and 2.8 cp,
respectively. They presented a temperature profile at
time zero corresponding to the heating at the tube top
price prior to air injection. They further stated that
the air/water injection-stream temperature varied
from an initially high value (due to preheating of the
upper tube apparatus) to a later value of
(presumably) 120F. We simply initialized the first
20070 of tube length to 500F and the remaining 80070
to 120F and injected at a constant 120F.
Smith and Perkins reported an initial period of air
injection with no water. This air injection was
continued until they noticed vigorous combustion
and no oxygen in the effluent gas. They did not
report the time period of this dry air injection. We
used 28.8 minutes. The subsequent water/air ratio
was constant at 0.18 Ibm/ scf.
Smith and Perkins stated their insulation was
unusually permeable and heat convected outside the
tube from the burning front upward to the trailing
edge. They accounted for this external convection in
their numerical model and reported better trailingedge agreement than we obtained.
We changed Smith and Perkins' six-component oil
analysis to three components and treated the sixcomponent system given in Table 1. We calculated Kvalues for each of the original six oil components
from In P v = A - B/ T, with A and B determined
from boiling-point and critical data given in Ref. 10.
Viscosity coefficients were obtained for the six
components from the same reference. K-value and
viscosity parameters for the three pseudo oil components then were calculated by simple mole-fraction
weighted averages.
Smith and Perkins represented the oil oxidation by
d0 2
dT

= A OP0 2 e

--

-cJ
, . . . . . . . . . . . . . . . (32)

where Po2 is partial of oxygen; A o ' B, and C were

determine d from experimental data; and f is the
fraction of oil oxidized. We used a reaction
proportional to
SOCIETY OF PETROLEUM ENGINEERS JOURNAL

EXPERIMENTAL DATA

RUN \,10 BLOCKS

- -

RUN 2,20 BLOCKS

6.

6. RUN 3,30 BLOCKS

EXPERIMENTAL DATA

--6

RUN 1,10 BLOCK:',

RUN2,20BLOCKS
RUN3,30BLOCKS

600

0,

[j.[j

400

400

T, OF

T ,oF

200

200

0'------''------'--'---'---'---'---'---'---'-----'
o

1.0

x/L

x/L

Fig. 1 - Calculated and observed 2hour temperature

profile - SmithPerkins experiment.

Fig. 2 - Calculated and observed 4hour temperature

profile - SmithPerkins experiment.

TABLE 1 - DATA FOR SMITHPERKINS LABORATORY TEST

Vertical tube length, ft
Tube crosssectional area, sq ft
Permeability, md
Porosity
Rock heat capacity, Btu/cu ft rockoF
Thermal conductivity, Btu/ftDoF
Initial temperature, of in first 20% of tube length
'F in last 80% of tube length
Po,psia
Initial water, oil, gas saturations
Initial mole fractions, Xl = 1.0, X 2 = 0.2063, X3 = 0.71,
X 4 = 0.084, X5 = 0.7097, X6 = 0.29
Number grid blocks (three onedimensional runs)
Number of components
Number of reactions
Component /
1. H2
2. C20
3. N2 -COx
4. C32 + C 47
5. C6 +C8 +C 12
6. O2

Molecular
18
282.6
32
508.9
131.4
32

V~2

= 0.01767 cu ft/lbm

V~2

= 0.01635 cu ftllbm
= 0.02228 cu ftllbm

V%2

~
3,206.2
162
750
50.7
336.6
730

Tc('R)

1,165.4
1,381
350
1,598
1,067
277.9

5.75
0.0834
700
0.25

34 [1 + 0.00055 ( T - T; ) I
3.0
500
120
815
0.3, 0.4, 0.3
10, 20, and 30
6
3

Viscosity Data
3
/122 = 0.02772 e ,363fT
/123 = 0.08
0.0002127 J'l.702
/133
0.01 e 4 ,908fT
/142
2
/152 = 0.0358 e ,135fT
0.03
/153
0.0002196 J'l.72l
/163

=
=

=
=

Reaction Data
C; denotes Component i

(3;

CP1 (Btu/Ibm)

10- 5

0.00038

10- 5
10- 5

0.00038
0.00038

0.6
0.226
0.6
0.6
0.24

Stocktank densities for Components 2, 4, and 5: 56.6,61.15, and 44.9 Ibm/cu ft

Reaction r

Stoichiometry

2
3

C2 + 25.08 C6 - 16.5 Cl + 20.8 C3

C4 + 45.17 C6 - 29.71 C l + 37.46 C3
C3 + 11.66 C6 - 7.67 Cl + 9.67 C3

2
3

Capillary pressure = 0
Sw; = Sw;r

Rate Constant
A r x10- 10

Activation
Energy Er

Heat of Reaction
Hrl (Btullbm oil)

0.28164
3.4054
0.4035

18,214
18,214
18,214

16,000
16,000
16,000

KValue Data
K v23 = (3.11 x 10 6 /p) e- 13 ,620fT
KV43
KV53

Relative Permeability and Capillary Pressure Data

= 0
= (7.668 x 10 5 /p) e- 8 ,33lfT

DECEMBER 1980

= Sorw = 0.25
Sorg = 0.9
k rwro = 0.5
kro;w = 1.0
k rgro = 1.0

nw

= now = nog = ng = 2.0

Injection/Production Data
108.1
Air injection mode, scf/D
Water/air ratio for first 28.5 minutes
o
Ibm/scf air after 28.8 minutes
0.18
Production at tube outlet at constant pressure, psia
815
541

EXPERIMENTAL DATA

RUN 1 ,10 BLOCKS

10 BLOCKS

RUN 2,

20 BlOCI(S

RUN 3,

3QBLOCKS

II

lOIP

009~4

SHI

6.

RUN 3,30 BLOCKS

.

l)

RUN'

- - - RUN2,20BLOCKS

l:l

-- -

_ _ _ _ ""- _ _ _ _ _ _

o~

6/
0/

:t

/
0

/0

'Y
/

400

/0

/~

T,OF

?
,6

200
0

/"

TIME,HOURS

Fig. 5 - Calculated cumulative oil recovery vs. time SmithPerkins experiment.

x/l

Fig. 3 - Calculated and observed 6hour temperature

profile - SmithPerkins experiment.

SMITH AND PERKINS NUMERICAL MODEL RESULTS

EXPERIMENTAL DATA

6. 6

RUN 3,30 BLOCKS

Cl

RUN 3,

30 BLOCKS

RUN

10 BLOCKS

1,

,,"

"

- - - RUN 2,20 BLOCKS

"if> ;-

z",

"'"'
"0
Xw

60

"',

w<t

u
... =>

So

"I

40

Uo.

a:

WZ

0.-

'"

"I

ZO

w~

"0

RUN 1,10 BLOCKS

100

80

"

"

20

" j,

""

0
0

" I

"

20

40

60

80

AIR INJECTED,SCF

Fig. 4 - Calculated and observed oxygen breakout times

- SmithPerkins experiment.

xI l

Fig. 6 - Calculated oil saturation profile at 6 hours SmithPerkins experiment.

, ................ (33)

where Xi is mole fraction of oil component i in the

oil phase. This second-order dependence on oil
concentration partially compensates for our
neglecting Cj. The -value of A ~ was chosen so that
the rates of Eqs. 32 and 33, expressed in consistent
units, are equal at initial oil-component concentrations.
Reaction stoichiometry and other data listed in
Table 1 were obtained from Ref. 9. Since that
reference gives no capillary-pressure or relativepermeability data, we used zero Pc and analytical
relative permeability curves with exponents of 2.0.
Our one-dimensional model Runs 1, 2, and 3 were
made using the data of Table 1 and using 10, 20, and
30 grid blocks, respectively. Figs. 1, 2, and 3 show
the agreement between calculated and observed
542

temperature profiles at 2, 4, and 6 hours, respectively. We attribute the discrepancy along the trailing
edge to the external heat convection. These figures
indicate (1) reasonably good agreement between
observed and calculated (20 grid blocks) profiles at
all three times, (2) virtually identical calculated
results for 20 and 30 grid blocks at each time, and (3)
significant but not extreme numerical dispersion in
the results calculated using 10 blocks.
Fig. 4 shows calculated and observed effluent
oxygen content vs. time. Calculated results using 20
and 30 grid blocks agree quite well with the observed
oxygen breakout time, while the results using 10
blocks show effects of numerical dispersion.
Fig. 5 shows a moderate effect of numerical
dispersion on calculated oil recovery vs. time for the
case of 10 grid blocks. Smith and Perkins reported an
estimated figure of 8.8 wtOJo of the original oil in
SOCIETY OF PETROLEUM ENGINEERS JOURNAL

place burned. The calculated figure for the 30-block

case was 10.38 wt07Q.
Fig. 6 shows the calculated oil saturation profile at
6 hours for the 10- and 30-,block runs. The sharp
jump in oil saturation occurs at the leading edge of
the sharp temperature profile. A sharp jump from
essentially zero to 0.80 to 0.90 in mole fraction of
light oil also occurs at this temperature and oil
saturation" front."
A favorable aspect of the combustion process
frequently emphasized in the literature is the in-situ
upgrading of a heavy oil by cracking. Cracking was
presumed negligible in this experiment due to the
relatively low temperature and light oil. However, an
"upgrading" type of effect is still present through
distillation or preferential vaporization and recovery
of lighter ends. For example, the calculated total oil
recovery on a stock-tank-barrel basis was 88.69070.
However, the calculated percentages recoveries for
C 6 + C 8 + C 12 , C 20 , and C 32 + C 47 were 99.8, 79,
and 69070, respectively.
The reaction-rate e - Cj factor used by Smith and
Perkins resulted in their numerical model's
calculated reaction rates declining very rapidly with
increasing fraction of oil burned in the range off = 0
to 0.1. Our second-order reaction gives a much less
severe decline in reaction rate with increasing
fraction of oil burned. They determined their large
rate decrease (large C) from combustion tests in an
isothermal tube with no water and an initial oil
saturation of 0.15. We feel that that type of test
might give a misleading e- Cf factor due to the
wetting, low-value oil saturation occupying the tight
or small, relatively inaccessible portion of the pore
space. This might lead to strongly diffusioncontrolled reaction rates. In the actual combustion
tube tests, water saturation of 20 to 50070 might force
the oil into much more accessible (to air) portions of
the pore space, resulting in a decrease of diffusion
control.
All model runs were performed with midpoint
convective weighting (8 = 0.5 in Eq. 12). Use of
upstream convective weighting resulted in a
significantly greater discrepancy between calculated
results for 10, 20, and 30 grid blocks.
Computations for all runs were performed for 18
hours or 0.667 days. Model computing efficiency for
the three runs can be summarized as follows.
Number Number Number
of
of Grid of Time
Steps Iterations
Run Blocks
160
42
10
1
250
20
53
2
62
330
30
3

Computing
Time,
CDC 6600
(seconds)
83
251
500

Parrish and Craig Data

Parrish and Craig II reported results from wet,

forward combustion tests in a vertical adiabatic tube
with length and diameter equal to 12 and 1 ft,
respectively. Original oil viscosity for their 11 tests
reported varies from 2.0 to 2,900 cp at 100F.
Pressure level was 500 to 600 psia.
DECEMBER 1980

We feel the data presented by the authors are not

sufficiently complete to warrant a serious matching
effort. However, we have made a number of model
runs seeking qualitative agreement with the experimental temperature profiles reported.
We selected Parrish and Craig's 2,900-cp Test 2 for
matching. Five components were represented water, heavy (original) oil, N2 + COx' light
(cracked) oil, and oxygen. The two chemical reactions were oxidation and cracking of heavy oi!. Coke
formation and oxidation were neglected.
Their reported Test 2 experimental conditions are
given in the first section of Table 2. Injection rates,
time period of initial dry air injection, relative
permeability, and kinetic and stoichiometric data
were not reported. Table 2 gives the values used for
these data. The authors reported experimental
temperature profiles at three times for a "typical"
COFCAW test.
We present computed results here for two runs.
The runs differed in respect to the following data.

Run
Number
4
5

Heat of
Reaction,
Oxidation
Reaction
(Btu/Ibm oil)
16,000
20,000

Rate
Constraint,
Cracking
Reaction
106
0.25 X 10 6

We heated the first 10070 of tube length to 500F at

time zero, then injected 20 scflhr of air at 130F for
28 hours. Maximum injection pressure was 900 psia,
and producing back pressure was 500 psia. The injection well went on injectivity during the period of
about 17 to 58 hours due to banking of cold (2,900cp) oil ahead of the combustion front. Starting at 28
hours and continuing to 96 hours, injection rates
were 0.19713 STn [cold-water equivalent (CWE) per
day] of water and 480 sefiD of air - an air/water
ratio of 2,435 scflbb!. The authors stated their tests
required 30 to 100 hours. Our 4-day run is 96 hours.
Figs. 7 and 8 indicate grid sensitivity through
comparison of Run 4 with 10 and 20 grid blocks. The
20-block run calculated a lower injectivity due to
sharper oil-bank definition; therefore, the curve of
cumulative air injected vs. time for the 1O-block run
lies somewhat above that for the 20-block run.
Fig. 7 shows peak temperature in each grid block
vs. grid block position. The 10- and 20-block runs
agree well, with the exception of the blocks in the 0.1
to 0.3 X I L interval, where the 20-block-run peak
temperature is less due to lower injection rate during
the corresponding time period. The decline in peak
temperature with distance traveled reflects a gradual
"extinction" of the dry combustion zone upstream
of the steam plateau.
Fig. 8 shows reasonably good agreement between
oil recovery curves computed with 10 and 20 grid
blocks.
The effect of a 25070 increase in heat of reaction
from Run 4 to Run 5 was one of increased peak
temperature and sustaining of the dry combustion
zone throughout the duration of the run. That is,
543

20 BLOCKS

900

RUN 4

= (6,000

BTU/LB Oil
_---:,----,,.-----,,~~-

IOIP : 341 STB

RUN 4

800

H= 16,000BTU/LB 01
o

o
-208l0CKS

TpEAK,"F 700

I08LOCKS

OIL

RECOVERY,
ST.

600

500

.2

.3

.5

.6

.7

.8

.9

10
CUMULATIVE AIR INJECTION, MCF

X/L

Fig. 7 - Calculated peak temperature vs. distance along

tube - Parrish and Craig COFCAW experiment.

Fig. 8 - Calculated oil recovery vs. cumulative air injection

- Parrish and Craig COFCAW experiment.

TABLE 2 - DATA FOR PARRISH AND CRAIG COFCAW TEST 2

Data From Ref. 11

Viscosity Data
1'-22 = 0.0002 e 9 ,2371T
1'-33 = 0.0002127 .,-07030
1'-42 = 0,01044 e 2 ,6181T
1'-43 = 0.01
1'-53 = 0.0002196.,-0721

13.5
Oil gravity, API
2,900
Oil viscosity at 100F, cp
0, 0.712, 0.228
Initial Sw, So, Sg
130
Initial cell temperature, of
2,435
Air/water ratio, scf/bbl
30
Porosity, percent
Permeability, md
1,000
12
Tube length, ft
12
Tube diameter, in.
Initial pressure (outlet backpressure), psig
500

K-Value Data
K v23
KV43

Additional Data

1.
2.
3.
4.
5.

H 2O
Heavy oil
N2 +CO x
Light oil
O2

= (1.157

10 6 /p)

e- 9 ,933/T

Reaction Data

Rock heat capacity, Btu/cu ft rock- F

Thermal conductivity, ft-D- F
I nitial mole fractions X1
1.0, X2
1.0, X3
0.9919, X4
Number of grid blocks
Number of components
Number of reactions
First 10% of tube length raised to 520 F to start ignition
Component /

= 0

Molecular
Weight

Pc

Tc(OR)

18
300
32
170.3
32

3,206
50.7
750
264.6
730

1,165.4
1,598
350
1,184.9
277.9

35

Cj denotes component i

= X5 = 0

10 and 20

5
2

Cpl
Cj

(Btullbm)

(31

10- 5

0.38 x 10- 3

10- 5

0,38x 10- 3

= 0.01642 cu ttllbm
= 0.01874 cu ttllbm
Stock-tank oil densities (Ibm/cu tt) Components 2 and 4

0.5
0.226
0,5
0.24

V~2

v~2

Reaction r
1.
2.

1.
2.

= 61.15 and 53.35

Stoichiometry
C2 + 32.143 C5 - 21.429 C 1 + 21.429 C3
C2 -1.761 C 4
Rate Constant
A,xlO- 6

Activation Energy

42.25
0.25

18,214
16,000

E,

Relative Permeability and Capillary Pressure Data

Same as in Table 1
Injection Production Data
Injection rate: 480 scflD air at 130F
Production on deliverability at 500 psig

Heat of oxidation, Reaction 1 = 16,000 Btullbm oil for Run 4

= 20,000 Btullbm oil for Run 5
Heat of cracking reaction
= 20,000 Btullbm mol oil

544

SOCIETY OF PETROLEUM ENGINEERS JOURNAL

peak temperature did not decline with distance

traveled in Run 5. Also, at the higher heat of reaction, reduction of the cracking. reaction-rate constant
from 10 6 to 0.25 X 106 caused a moderate increase
in calculated peak temperature.
The dashed curves in Fig. 9 are plots of the
"typical" COFCA W test presented by Parrish and
Craig. The three times of the experimental, dashed
profiles are "start COFCAW," "steam breakthrough," and "combustion zone breakthrough."
The times corresponding to the three Run 5 calculated profiles are 28,57, and 89.5 hours, respectively.
For Run 5, the calculated oil recovery was 93.6%
compared with Parrish and Craig's reported 87.9070.
Calculated oil burned was 7.15070 compared with
Parrish and Craig's reported 7.88070 of initial oil in
place. The calculated figures are a little low in that
combustion was continuing (So = 0.15) in the 10th
block at the end of the run. Reported and calculated
air/recovered-oil ratios are 6,020 and 5,166 scflbbl,
respectively.
On a volumetric stock-tank-barrel basis, 24070 of
the oil recovery calculated in Run 5 was Light
Component 4. One hundred percent of the light oil
formed by cracking was produced, since final oil
saturation was zero and we included no oxidation
reaction for the light oil.
Run 5 used 10 grid blocks and required 108 CDC
6600 seconds, 61 time steps, and 302 total iterations
for the 4 days of real time.

Crookston, Culham, and Chen Problem

Crookston et al. Z simulated oxygen injection into a
one-dimensional, 200F reservoir 164 ft long, 115 ft
wide, and 21 ft thick. Reaction and fluid-property
data were estimated from values given in the
literature. Their four reactions represented oxidation
of light and heavy oil components, cracking of heavy
oil, and oxidation of coke.
Table 3 lists our model input data for this problem.
Our data are identical to those of Crookston et al.
with the following exceptions. Our oil relative
permeability curve is only similar to theirs. They llsed
a multiplicative factor of 1 - (nc/ncmaJ5 in their
cracking reaction-rate expression, with ncmax =
0.07692 Ibm mol/cu ft of formation. We used no
maximum equivalent (S4)max' since our calculated
coke laydown did not exceed their n cmax '
As given in Table 3, the problem includes the six
components water, heavy oil, Nz-CO x ' coke,
oxygen, and light oil. Thus, our simulation required
simultaneous solution of six primary equations one energy balance plus five component mass
balances - with the immobile coke mass balance
treated as a constraint equation.
Crookston et al. used propane and dodecane
properties for light and heavy oils, respectively, with
an increased viscosity for dodecane. At 200F, the
liquid-component viscosities of light and heavy oil
are 0.089 and 139 cp, respectively.
Crookston et al. described their calculated results
in some detail both verbally and graphically. Since
our results agree very closely with theirs in all
DECEMBER 1980

- - - - TYPICAL COFCAVI' TEST (PARRISH AND CRAIG)

CALCULATED, RUN 5 6H '20,00Q BTU/LS OIL

1100
COMaUSTION FRONT

1000

900
800

700
TI

OF

600

500
400
300

200
0
0

X/L

Fig. 9 - Calculated and experimental

perature profiles.

COFCAW

tem-

respects, we will not repeat the extended discussion

and figures.
Table 4 summarizes the agreement between the two
sets of calculated results. Oil recovery is volumetric
oil recovery expressed as a percentage of initial oil in
place. Stock-tank oil densities are 45.313 and 23.633
Ibm/cu ft for heavy oil and light oil, respectively, at
72 OF and 14.7 psia. Initial water and oil in place are
5,168 and 12,590 STB, respectively. The light oil
recovery is expressed as volumetric percentage of
total oil recovery. Heat loss (to overburden) is expressed as a percentage of total heat of reactions.
Combustion-front velocity is calculated as
0.7(164)/(19 tz), where ti is time in days of the
temperature peak in grid block i and 164 ft is the
system length. Tpavg is average peak temperature of
Cells 2 through 9, inclusive, peak temperatures.
Water recovery is expressed as a percentage of initial
water in place. Heat of reaction is total heat of all
reactions. Computing time is in CDC 6600 CPU
seconds.
Table 4 shows fairly good agreement between the
Crookston et al. results and our Run 6 results. Both
we and they calculated a steam zone or plateau
throughout the system by the time the first cell
peaked in temperature. The reason for this very rapid
steam-zone development is the passage of unreacted
oxygen throughout the system in the first 10 to 20
days prior to first-block spontaneous ignition. This
oxygen reacts sufficiently throughout the system to
raise temperature throughout to about 300F.
Energy balance and component material balances
for Run 6 were excellent, ranging only from 0.9999 to
1.0001.
Crookston et al. reported a computing time for
this problem of 1,071 seconds on an IBM 370/168
computer. Run 6 required 242 seconds on a CDC
6600 computer. CDC 6600 computer times are about
2.2 times larger than IBM 370/168 times for our
steam flood model. 6 Much of the difference between
the computer time reported by Crookston et al. and
that reported here reflects the fact that their for545

TABLE 3 - DATA FOR CROOKSTON ET AL. ONEDIMENSIONAL FIELDSCALE PROBLEM

Reservoir length, ft
Reservoir width, ft
Reservoir thickness, ft
Permeability, md
Porosity
Rock heat capacity, Btu/cu ft rock F
Thermal conductivity, Btu/ftD F
Overburden thermal conductivity, Btu/ftD F
Overburden heat capacity, Btu/cu ft overburdenoF
Ti , of
Pi, psia

164
115
21
4,000
0.38
35
1.6
1.6
35
200
65

K-Value
K
v23
Kv63

= (184,900/p)e- S ,7391(T-1S7.1)
= (130,700/p)e- 3 ,370/(T-45.29)

Reaction Data
Ci denotes component i
V~2

V~2
~4

Initial water, oil, gas saturation

0.2,0.5,0.3
Initial mole fraction X1 = 1.0, X2 = 1.0, X3 = 0.806, X4 = X5 = Xs = 0
Number of grid blocks (onedimensional)
Number of components
Number of reactions
Component /

Molecular
Weight

1. H 2 O
2. Heavy oil
3. CO 2
4. Coke
5. O 2
6. Light oil
Reaction
1

2
3
4

18
170
44
13
32
44

J1.22

Ci

(31

3,206.2
264.6
1,073

1,165.4
1,184.9
547.7

10- 5

0.000382

730
615.9

CP1 (Btullbm)

2
3
4

=
=

ro
ro

0.34 + 0.0003547 T
0.25
0.3
0.24
0.000769 -1.1843 + 0.003452 T

277.9
665.6 2.2 x 10- 4

Relative Permeability and Capillary Pressure Data

Stoichiometry

Swi = Swir = 0.2

Sorw = 0.3

C2 +18C 5 -12C 3 + 13C 1

Cs + 5 C5 - 3 C3 + 4 C 1
C2 - 2Cs + 4.67C 4 + 1.33C3
C 4 + 1.25C5 - C3 + 0.5C 1

Sorg = 0.09
Sgc
k rwro

k roiw
k rgro

Heat of Reaction

= 0.0003624 e 8 ,4851T

= 0.3926

10- 5 T1. 102

J1.33
0.0002127 702
J1.53
0.0002196 721
J1.S2 = 0.02083 e959.SIT
J1.S3 = 0.2166 X 10- 4
J1.23

Tc(OR)

= 0.0125 cu ftllbm

Viscosity Data

10
6
4

Pc

= 0.0235 cu ftllbm

= 0.05 cu ftllbm

Rate Constant
Ar x 10- 6

Activation
Energy Er

(Btu/mol first reactant)

1.0
1.0
0.3
1.0

18,500
18,500
16,000
13,000

3.49 x 106
0.948 X 10s
20,000
0.225 X 106

Hr1

= Sgr = 0.05
= 0.25
= 1.0
= 0.7
= now = nog = 3

nw
ng = 1
P cwo
P cgo

= 0.75
= 5.45

- 1.5S w
- 8.45S L

Injection/Production Data
Oxygen injection rate: 300 Ibm mollO, 200F
Production on deliverability at 60 psia
Productivity index: 300 RBcplDpsi

TABLE 4 - COMPARISON OF CROOKSTON ET AL. WITH

FORWARD COMBUSTION PROBLEM

RUN 6 RESULTS FOR EXAMPLE ONEDIMENSIONAL

Crookston
Oil recovery, %
Light oil, %
Heat loss, %
Velocity, ftlD
Water recovery, %
Tpavg, of
Heat of reaction, Btu x 10- 10
Computing time, seconds
Number of time steps
Number of iterations
Average M, days

eta/.

Run6

97
30
27
1.2
119
849
0.7925
1,071'
764
3,056
0.1965

97.18
27.7
26
1.35
118.2
836
0.7183
242"
122
530
0.122

0.1
0.2
90
30

0.4
None
20
80

Time Step Controls

Saturation
Mole fraction
Pressure, psi
Temperature, of
546

Cell
1
2
3
4
5
6
7
8
9
10

Peak Temperature F (days)]

600(33.3)
534(32.3
775(44.9)
728(42.1)
855(57.5)
861(52.7)
850(71.5)
878(65.9)
868(85.9)
924(79.9)
820(99)
816(92)
875(113.1)
824(103)
870(126.6)
836(116)
880(140.5)
822(127)
780(150.2)
653(137)

IBM 3701168.
"CDC 6600.

SOCIETY OF PETROLEUM ENGINEERS JOURNAL

TABLE 5 - SENSITIVITY RUN RESULTS - ONEDIMENSIONAL FORWARD COMBUSTION

Oil
Water
Heat of
Recovery Light Oil Heat Loss Velocity Recovery
Reaction
Time
Tpavg
(OF)
(%)
(%)
'(%)
(ft/O)
(Btu x 10- 1) (days)
(%)
93.5
19.9
34.9
1.39
123.8
622
0.399091.5
Base case
93.7
20.3
35.8
1.36
123.4
633
0.3965
88.1
k = 0.5x
74
13.4
30.1
2.23
129.2
475
0.2737
87.4
<I>
0.63x
93.9
19.7
29.2
1.43
122.7
640
0.3842
88.7
Aoa
0.5x
94.5
18.9
34.9
1.47
120.9
628
0.3918
80.6
(H')1
1.2x
93.9
19.1
35.0
1.51
122.4
662
0.4136
87.4
(H 4 ),
1.2x
92.9
14.0
40.6
1.31
122.7
647
0.3615
65.3
E1
0.8x
99.2
86.1
52.9
0.44
137.9
1,244
0.869
160
E3
0.8x
89.4
20.4
43.8
0.52
132.4
646
0.5039
146
E4
1.77x
93.5
17.6
35.7
1.31
122.6
606
0.3754
75.9
A1
2x
95.2
30.7
37.1
1.19
124
705
0.4484
96
A3
2x
93.7
20.0
34.4
1.55
123.2
608
0.3920
89.6
A4
2x
93.4
18.7
34.3
1.58
123.4
605
0.3894
90
now = nog
2
93.5
20.6
34.8
1.35
123.9
631
0.4040
91.4
now = nog
4
93.4
19.6
35.3
1.44
123.8
621
0.3985
91.9
Pcwo = Pc~o
0
94.6
55.6
46.6
0.71
130.1
1,021
0.6979
143
(K V4 2
8,000
93.4
18.9
34.4
1.57
123.5
604
0.3913
89.7
(K V4 )6
5,000

Run
7
8
9'
10
11
12
13
14
15"
16
17
18
19
20
21
22
23

=
=
=
=
=
=
=
=
=
=
=
=
=
=
=

First block did not "ignite." Spontaneous ignition occurred in Block 2.

"Coke was not completely burned out of Cells 1 and 2.

mulation required about 764 time steps compared

with 122 for Run 6. Their larger computing time per
block step reflects, at least in part, their direct
solution of nine primary equations compared with
our six primary equations.
We feel the considerably smaller number of time
steps in our Run 6 is primarily due to the completely
implicit formulation, which allows larger steps. We
have noted very little time truncation error in forcing
our runs to use smaller time steps.
Sensitivity Runs
A number of one-dimensional sensitivity runs were
made using the Crookston et at. data. These runs
were made to indicate which data are "important" in
the sense of having a significant influence on
calculated results. The results serve as a guide for
history-matching efforts in that they indicate the
types of response to changes in various input-data
variables. The utility of the sensitivity-run results is
limited because they are only applicable to the
Crookston et at. data and do not reflect two- or
three-dimensional flow effects.
The base case (Run 7) is a reservoir 82 ft long, 115
ft wide, and 21 ft thick. The 82-ft length is
represented by five 16.4-ft grid blocks. A well
productivity index of 1,328 was used together with
the data listed in Table 3.
Base Run 7 results are tabulated in Table 5. Tpav
is the average of peak temperatures in Cells 2 throug6
4. Velocity is 2 X 16.41(t4 - t 2 ), where ti is time of
peak temperature in Cell 1. Time tabulated in Table 5
is the time at which the last cell (Cell 5) peaked in
temperature; oil recovery was complete at some time
prior to that time in all cases. The notation in the last
column of Table 5 (e.g., E1 = 0.8) means that activation energy of Reaction 1 is 0.8 times the basecase value. K V4 is defined for heavy and light oil
Components 2 and 6 by
KV4i

Ki

(K V2i/P) e

DECEMBER 1980

T-Kv5i

where Ki is the equilibrium K-value (Yi = Kixi,xi =

oil-phase mole fraction, and Yi = gas-phase mole
fraction) of component i. Thus, increasing the K V4
value decreases the component volatility. Base-case
values of K V4 for Components 2 and 6 are 6,739 and
3,370, respectively.
A number of runs made are unreported in Table 5
because of insensitivity of results to (1) 20070 increases
in component or rock specific heats, (2) 50070
reduction in oil-component thermal expansion
coefficient, and (3) change from 44 to 32 of COx
gaseous Component 3 molecular weight.
Run 8 indicates little sensitivity to permeability in
the range of 2,000 to 4,000 md. Actually, reduced
permeability may have a pronounced effect if it is
reduced to a value low enough to increase reservoir
pressure significantly and, hence, reaction rate.
Run 9 indicates a pronounced effect of lowered
porosity. Reaction rate decreases due to smaller
reactant concentrations and oil recovery falls from
93.5 to 74070 as porosity is decreased from 0.38 to
0.24. Also, of course, more rock is associated with a
given pore-space volume as porosity is decreased.
Reduction of overburden thermal conductivity by
a factor of two reduces the heat loss moderately but
has little effect on oil recovery and peak temperature
(Run 10).
Runs 11 and 12 show that a 20070 increase in cokeoxidation heat of rection has a greater effect than a
20070 increase in the direct-oxidation heat of reaction.
Run 13 shows increased heat loss and peak temperature caused by decreased activation energy of the
direct-oxidation Reaction 1. Run 14 shows that a
decreased cracking-reaction activation energy
doubles the peak temperature to 1,244F and gives a
much lower combustion-zone velocity and higher
(volumetric) oil recovery, with light-oil volumetric
fraction in recovered oil rising from 19.9 to 86.1.
Increasing the coke-oxidation activation energy by
54070 (Run 15) moderately increases peak temperature
and heat loss. Runs 16 through 18 indicate a greater
547

o INJECTION WELL

PRODUCTION WELL

GRI D BLOCK CENTER

1=2

C>X =6y =

i =4

k= 1

1146.8
(2140)

k=2

608.5
(712)

431.6
(1190)

9945
(285)

TEMP,OF

(TIME,DAYS)

396.5
(1440)

=3

k=4

27.33 FT.

Fig. 10 - One-eighth of a five-spot grid.

sensitivity to cracking reaction-rate constant than to

direct-oxidation or coke-oxidation reaction-rate
constants. The sensitivities noted here should be
viewed with the knowledge that only about 3070 of the
oil is consumed by the direct-oxidation reaction for
the base-case data.
Oil relative permeability was increased considerably in Run 19 and decreased in Run 20. Heat
loss, light oil production, velocity, and peak temperature increase slightly with decreased oil relative
permeability_ Run 21 shows a negligible effect of
capillary pressure.
Run 22 shows that a reduced volatility of heavy oil
Component 2 significantly increases peak temperature and decreases velocity_Percentage of light
oil in the recovered oil rises from 19.8 to 55.6 and
total heat of reaction increases by 50070 _ Run 23
shows a much smaller effect of reduced volatility of
light oil Component 6. Component 6 is so volatile in
the base-case data that virtually none of it is oxidized
by Reaction 2. That is, light oil is vaporized nearly as
rapidly as it is formed and is carried in the gas phase
downstream from the combustion front.
Three-Dimensional Run
Runs 24 and 25 simulated one-eighth of a 1.24-acre
five-spot with an injection/production well distance
of 164 ft and formation thickness of 80 ft. Initial
reservoir pressure and temperature were 500 psia and
200F. Air injection was 336.8 McflD at 500F for
10 days with the production well produced on
deliverability at 500 psia. Starting at 10 days, 93.18
STB/D of water were injected along with 336.8
McflD of air at 200F for an air/water ratio of 2,615
scflbbl.
Four 20-ft layers were used with an upper gas-filled
layer and lower water-saturated layer. Initial
saturations were as follows:
548

=3

869.9
(1440)

5X3GRID

640.3
(480)

Fig. 11 - Peak temperature of a 5 x 4 wedge - Run 25.

Layer

Swi

Soi

1
2
3
4

0.2
0.2
0.2
1.0

0.09
0.75
0.75
0

0.71
0.05
0.05
0

The injection completed in Layer 3 only while the

producer was completed in Layers 2 and 3, with a
total well productivity index of 1,123 split evenly
between the two layers. Initial gas composition was
77 .44070 N 2 and 20070 O 2 , with the balance water
vapor. Initial water and oil in place were 14,050 and
13,590 STB, respectively. The remaining data were
identical to those given in Table 3.
Three-dimensional Run 24 utilizes a 5 x 3 x 4
grid, with Ax = ~y = 27.33 ft and Liz = 20 ft. Fig.
10 is an areal view of this grid illustrating the parallel
orientation 14 with the x axis coincident with the line
joining the wells. The cross-hatched small "tip"
blocks are eliminated, with their pore volumes included in their neighbor blocks. 4-6
Run 25 is a 5 x 1 x 4 two-dimensional, wedgeshaped cross section with variable width reflecting
the triangular one-eighth five-spot shape. Ax is 32.9
ft, and except for the two-dimensional variable-width
aspect, all data are identical to those for the threedimensional Run 24.
The runs were terminated at 300 days. For the
three-dimensional Run 24 at that time, oil production
rate had declined to less than 5 STB/D and instantaneous WOR and GOR were 15.5 STB/STB and
82.5 McflSTB, respectively.
The initial gas-filled layer along the top of the
reservoir resulted in a strong "over burn, " as indicated by the peak temperatures and time noted in
Fig. 11. Peak temperatures less than 500F reflect the
presence of liquid water and corresponding constraint on temperature.
SOCIETY OF PETROLEUM ENGINEERS JOURNAL

In both runs the upper layer burned dry, with

maximum coke saturations reaching the 5070 level.
Oil was pushed into the upper layer, reaching
saturations as high as 60070 compared with the
original 9070. Oil also was pushed downward into the
aquifer layer, reaching saturations as high as 22070.
Calculated oil recovery at 300 days for the threedimensional Run 24 was 52.3070. On a volumetric
basis, light oil formed by cracking was 28.5070 of this
total recovery. Calculated oil production rate peaked
at 61 STB/D at 41 days and averaged 32 STB/D over
the first 200 days and 7.6 STB/D during the period
200 to 300 days.
Fig. 12 compares calculated heat loss and oil
recovery Runs 24 and 25. The two-dimensional run
calculates somewhat higher oil recovery but approximates the three-dimensional run moderately
well.
Table 6 summarizes the model computing efficiency for Runs 24 and 25. The three-dimensional
run using 52 active blocks required 55 minutes of
CDC 6600 computer time. The automatic time-step
selection in the runs used maximum changes per step
in saturation, pressure, and temperature of 0.2, 100
psia, and 80F, respectively.

60

030
~

80

25

120

1040

160

160

200

220

2-40

260

280

300

Fig. 12 - Comparison of oil recovery with heat loss for

Runs 24 and 25.

the relatively expensive direct solution by Gaussian

elimination.
Model results agree moderately well with experimental results from two independent, published,
laboratory adiabatic-tube studies. However, the
second study did not present sufficient measured
data to attach quantitative significance to the
model's match.
The model required less than 1 hour of CDC 6600
computer time to solve a six-component, 5 x 3 x 4,
three-dimensional, wet forward combustion example
problem representing one-eighth of a five-spot.
The model results agreed very well with those
calculated by Crookston et al. for a one-dimensional,
field-scale, forward combustion problem involving
six components. We attribute our significantly lower
computing time for this problem to our implicit
formulation and minimization of primary equations
requiring direct solution.
A number of sensitivity runs using data for the
above-mentioned problem indicate that calculated
combustion behavior is strongly dependent on
relatively few of the many model input-data values.

Summary

24

100

TIME ,DAYS

The formulation described here for numerical

simulation of in-situ combustion processes is more
general than previously described models. The
number and identities of components, number of
reactions and corresponding reactants, products, and
stoichiometry are not limited or "hard cored" but
rather are specified through input data.
Vaporization/condensation phenomena are treated
with maximum flexibility by allowing any component
to be distributed among all phases.
One aspect of the formulation's generality is the
applicability to thermal problems ranging from
single-component geothermal problems to
multicomponent steam flood and in-situ combustion
problems, without recourse to different model
versions.
The model uses an improved variable substitution
method using single set of "master-phase" mole
fractions. The formulation is implicit, reflecting our
belief that computing expense or arithmetic for most
practical problems will decrease as the formulation is
rendered more implicit. The formulation takes
maximum advantage of constraint equations to
reduce the number of primary equations requiring

Run

3D5x3X4RUNZ4

o 020 '5X4 WEDGE CROSS-SECTION RUN 2'5

Nomenclature
A = cross-sectional area normal to flow,

sq ft (m 2 )

TABLES - COMPUTING TIMES AND TlMESTEP SIZES

FOR TWODIMENSIONAL 5 x 4 AND THREEDIMENSIONAL 5 x 3 x 4 RUNS
Cumulative
Average
Cumulative
Cumulative
Number of
Time Step
Cumulative
Time per
Number of
Number of
Active
Time
(days)"
Iterations
Time'
Block Step
(days)
Time Steps
Blocks
1.87
2,944
200
107
464
52
9.1
517
3,280
0.535
300
118
20

200
300

88
107

385
477

525
651

0.304

2.27
5.26

'Cumulative computer time, CDC 6600 CPU seconds.

"Average time step, days, in 0- to 200-day or 200- to 300-day intervals.

DECEMBER 1980

549

A r = reaction-rate constant for reaction r

cI =

C IJ

C p1

Er
HIJ =

Hf
HJ
Hr
HrI

H SL

H SV =
t:.H vb

/=
=

J1 =

k =

k ro
k rw
k rgro

k roiw

k rwro

KV/J
K VI

K V5
f =
L =

Kv

Mj =
550

(Eq. 15)
compressibility of component / (Eq.
A-5), vol/vol-psi (l/kPa)
coefficients in matrix of linearized
model equations
ideal gas-state heat capacity of
component /, Btu/Ibm mol- of
(kJ/kmolK)
activation energy for reaction r,
cal/g mol (kJ/kmol)
partial enthalpy of component /
in phase J, Btu/Ibm mol (kJ/kmol)
ideal gas-state enthalpy of component
/, Btu/Ibm mol (kJ/kmol)
enthalpy of phase J, Btu/Ibm mol
(kJ/kmol)
heat of reaction r for grid block
Btu/D (kJ/d)
heat of reaction r of first reactant,
Btu/Ibm mol (kJ/kmol)
enthalpy of saturated liquid, Btu/Ibm
mol (kJ/kmol)
enthalpy of saturated vapor, Btu/Ibm
mol (kJ/kmol)
heat of vaporization at boiling point,
Btu/Ibm mol (kJ/kmol)
component index
phase index
primary or master phase for component/
absolute permeability, md x 0.00633
value of k read as input data
relative permeability to gas
relative permeability to phase J,
fraction
oil relative permeability, fraction
water relative permeability, fraction
relative permeability to gas at
residual liquid saturation
relative permeability to gas in a
gas/oil system with connate water
relative permeability to oil at
irreducible water saturation
relative permeability to oil in an
oil/water system
relative permeability to water at
residual oil saturation
equilibrium K-value for component /
in phase J
coefficients for calculation of KVIJ
iteration number In to base e
distance between adjacent grid block
centers, ft (m)
heat of vaporization, Btu/Ibm mol
(kJ/kmol)
reservoir-rock heat capacity, Btu/cu
ft rock- of (kJ/m 3 rock-K)

MI = molecular weight of component /

M J = molecular weight of phase J
ng = exponent on gas saturation for krg
nj,r = order of reaction r in thejth reactant
nag = exponent on oil saturation for krog
now = exponent on oil saturation for krow
nw = exponent on water saturation for k rw
N=N p +Nc +2
Nc = number of components
N p = number of phases
N R = number of chemical reactions
N R,r = number of reactants in reaction r
P = gas-phase pressure, psi a (kPa)
Pc = critical pressure, psi a (kPa)
Pel = critical pressure of component /, psia
(kPa)
P g = gas-phase pressure
Pi
original reservoir pressure
Pj
phase J pressure, P + P eJ
Pk
pressure in grid block, layer k, at
well, psia (kPa)
Pa
oil-phase pressure, P - P ego
Pw
water-phase pressure,
P - P ego - P ewo
Pwb
wellbore flowing pressure, psia (kPa)
Pwbk
wellbore flowing pressure opposite
center of layer k, psi a (kPa)
P Cgo = gas/oil capillary pressure, P g - Po
= phase J capillary pressure, P J - P

= water/oil capillary pressure, Po

= mth or jth unknown (Eq. 7)

- Pw

= time-dependent productivity index of

layer k, res cu ft-cp/D-psi (res m 3 )

q Ik

QHL
R

Rr
Sir

production rate of component / from

grid block, mol/D
= rate of production of component /
from layer k, mol/D
rate of creation of component / in
grid block due to reaction r,
mol/D
rate of production of phase J from
layer k, res cu ft/D (res m 3 /d)
production rate of enthalpy from grid
block association with fluid
production, Btu/D (kJ/d)
rate of production of enthalpy from
layer k, Btu/D (kJ/d)
= heat-loss rate to overburden from
grid block, Btu/D (kJ/d)
= gas-law constant, 1.987 Btu/Ibm molOR (kJ/kmol.K)
= rate of reaction r in grid block, Ibm
mol (kmol) of first reactant/D
stoichiometric coefficient of ith
participant in reaction r
critical gas saturation

SOCIETY OF PETROLEUM ENGINEERS JOURNAL

Sgr = residual gas saturation

(Sg) max = maximum historical gas saturation in

SJ
SL

= SJI(l - S4)
= liquid saturation, 1 -

SL

=
=

Ai
Aob
/-tIJ

Sg
/-tJ

SLI(l - S4)

residual oil saturation to gas

Sorg
residual oil saturation to water
Sorw
connate water saturation, fraction
Swi
irreducible water saturation, fraction
Swir
solid
phase saturation
S4
=
maximum
allowed coke saturation
(S4)max
(Eq. 13)
T = temperature, oR (K)
Tb = boiling point, oR (K)
T br = TblTc
Tc = critical temperature, oR (K)
Ti = original reservoir temperature
Tr = reduced temperature, TITc
TI or T2 = temperature of Grid Block 1 or 2
V J = grid block volume, Ax~yLiz,
cu ft (m 3 )
vIJ = partial volume of component I in
phase J, cu ft/mol (m 3 Ikmol)
V~J = see Eq. A-5
v J = specific volume of phase J, cu ft/mol

PJ
7

7c
7R

cp

x IJ =
Ax

XI

~y

z=
Liz =

(m)

Z = subsea depth measured positively

vertically downward, ft (m)
Zk = subsea depth to center of layer k at
well
Z lor Z2
subsea depth to center of Grid Block
1or2
~Z

= ZI

- Z2

(3 I = thermal expansion coefficient for

component I (Eq. A-5), vollvol- of
(IlK)
'Y J = specific weight of phase J, psilft

(kPa/m)
'Y Jk = value of 'Y J in layer k
'Ywb = well bore
fluid gradient, psi/ft
(kPa/m)
() = weight factor for interblock enthalpy
DECEMBER 1980

(Eq.

A-I),

value of Aread as input data

= overburden thermal conductivity
partial viscosity of component I in
phase J, cp (Pa s)
= viscosity of phase J, cp (Pa s)
density of phase J, moll cu ft
(kmollm 3 )
= fluid-flow transmissibility, kAI L,
res cu ft-cp/D-psi (res m 3 )
heat-conduction transmissibility,
MIL
= radiation transmissibility, Acx3 A I L
= porosity, fraction

f = iteration number

Subscripts

i = x-direction grid block index

ijr = component number of jth reactant in
reaction r
I = component number
j = y-direction grid block index
J = phase number
k = z-direction grid block index
mjr = phase of jth reactant in reaction r
n = time-step level
r = reaction number

= grid-block volume, Ax~yLiz,

cuft(m 3 )
mole fraction of component I in
phase J
grid-block dimension in x direction,
ft (m)
mole fraction of component I in
component Fs primary base
grid-block dimension in y direction,
ft (m)
gas-phase supercompressibility factor
grid-block dimension in z direction, ft

conductivity

Superscript

(m 3 /kmol)

thermal

Btu/ft-D- of (WIm.K)

agrid block
saturation of phase J

SJ

Difference Notation
X is any quantity or arithmetic expression.
bX

= X n+ I

oX

== Xn+1 - XC

oX

Xn '

or
X c+ I - XC ,

where the (f+ 1)th iterate,

toXn + l
~

xe+ I , is an approximation

= XI - X 2 ,

where Subscripts
Blocks 1 and 2.
~(7~) -

and 2 refer to adjacent Grid

~x(7x~xX)

+
~x(7x~xX) -

+ ~y(7y~yX)

~z (7Z~ZX) .

7xi+y,(Xi + 1 -Xi)
- 7x i-Y,(Xi

-Xi - l ) ,

where 7 xi + y, is x-direction transmissibility for flow

between Grid Blocks i and i + 1.

References
I. Ali, F.S.M.: "Multiphase, Multidimensional Simulation of InSitu Combustion," paper SPE 6896 presented at SPE 52nd
Annual Technical Conference and Exhibition, Denver, Oct. 912, 1977.
551

2. Crookston, H.B., Culham, W.E., and Chen, W.H.:

"Numerical Simulation Model for Thermal Recovery
Processes," Soc. Pet. Eng. J. (Feb. 1979) 37-58; Trans.,
AIME,267.
3. Youngren, G.K.: "Development and Application of an In-Situ
Combustion Reservoir Simulator," Soc. Pet. Eng. J. (Feb.
1980) 39-51.
4. Coats, K.H., George, W.O., Chu, Chieh, and Marcum, B.E.:
"Three-Dimensional Simulation of Steamflooding," Soc. Pet.
Eng. J. (Dec. 1974) 573-592; Trans., AIME (1974) 257.
5. Coats, K.H.: "Simulation of Steam flooding With Distillation
and Solution Gas," Soc. Pet. Eng. J. (Oct. 1976) 235-247.
6. Coats, K.H.: "A Highly Implicit Steamflood Model," Soc.
Pet. Eng. J. (Oct. 1978) 369-383.
7. Price, H.S. and Coats, K.H.: "Direct Methods in Reservoir
Simulation," Soc. Pet. Eng. J. (June 1974) 295-308; Trans.,
AIME,257.
8. Redlich, O. and Kwong, J.N.S.: "On the Thermodynamics of
Solutions. V. An Equation of State. Fugacities of Gaseous
Solutions," Chern. Reu. (Feb. 1949) 44, 233-244.
9. Smith, F.W. and Perkins, T.K.: "Experimental and
Numerical Simulation Studies of the Wet Combustion
Recovery Process," J. Cdn. Pet. Tech. (July/Sept. 1973) 4454.
10. Reid, R.C., Prausnitz, J.M., and Sherwood, T.K.: The
Properties of Gases and Liquids, third edition, McGraw-Hill
Book Co. Inc., New York City (1977).
II. Parrish, D.R. and Craig, F.F. Jr.: "Laboratory Study of a
Combination of Forward Combustion and Waterflooding ~
The COFCAW Process," J. Pet. Tech. (June 1969) 753-761;
disc., J. Pet. Tech. (July 1969) 801-802; Trans., AIME, 246.
12. Braden, W.B.: "A Viscosity-Temperature Correlation at
Atmospheric Pressure for Gas-Free Oils," J. Pet. Tech. (Nov.
1966) 1487-1490; Trans., AIME, 237.
13. Stone, H.L.: "Estimation of Three-Phase Relative Permeability and Residual Oil Data," J. Cdn. Pet. Tech.
(Oct.! Dec. 1973) 53-61.
14. Todd, M.R., O'Dell, P.M., and Hirasaki, G.J.: "Methods for
Increased Accuracy in Numerical Reservoir Simulators," Soc.
Pet. Eng. J. (Dec. 1972) 515-530; Trans., AIME, 253.

APPENDIX

Treatment of Fluid and Rock Properties

Thermal conductivity is treated as a function of gas
saturation as

v~J[I+!h(T-Ti)][I-CI(P-Pi)]'

VIJ

........................... (A-S)
where v~J' component thermal expansion coefficient
(3 J> and component compressibility C I' are read as
input data. Liquid-phase density PJ is then 11 vJ.
Liquid-water partial volume is calculated as a
function of temperature pressure from a read-in
table.
Gas-phase density is calculated as

P
zRT

P3 =

mol
cu ft

, . . . . . . . . . . . . . . . . . . (A-6)

where R = 10.73 (here only), and the supercompressibility factor z is calculated from the
Redlich-Kwong equation of state. 8
The specific weight of phase J is

"IJ

= PJM J II44 psilft

, . . . . . . . . . . . . . . . (A-7)

where molecular weight is

Nc

MJ

xIJ MI . . . . . . . . . . . . . . . . . . . (A-8)

1=1

The ideal gas-state heat capacity of each component is calculated as

C Opl = CPI (l)

+ CP2 (/) T + CP3 (/) T2. . (A-9)

The "partial" enthalpy of Component 1 (water) as

liquid or vapor is calculated as a function of temperature and pressure from a read-in steam table.
The ideal gas-state partial enthalpy of every other
component /(l-:;t.I) is

Hf (T)

=~

CCPIdT . ................ (A-lO)

Tj

Gas-phase enthalpy is calculated as

Nc

H3 =

xJ3Hf

+ (H-Ho)

, ........ (A-II)

1=1

where CYI should be in the range of 0 to 0.8 and Ai is a

thermal conductivity read as input data. Formation
absolute permeability is dependent on solid-phase (J
= 4) saturation as

ki(l

+ CY2S4)md

, . . . . . . . . . . . . . . . . . (A-2)

where CY2 must be zero or negative and k i is permeability read as input data. Porosity is calculated as
cP d 1 + C r (p - Pi)]'
Reservoir-rock heat capacity is calculated as

The specific volume of a liquid phase is calculated

by Amagat's law of partial volumes:
Nc

vJ

HIJ = Hf - LuIJ (1- T rI ) 0.38, ......... (A-I2)

where LuIJ can be obtained from the two equations
(Ref. 10, Page 208)*

1.093 RTc Tbr

Lub
and

............................. (A-4)
The partial volume of component / in phase J is

In Pc-I

0.93 - Tbr

1 - Tr )0.38
Lub ( - - .
1- Tbr
The enthalpy of liquid phase J is

xIJvIJ cu ftlmol of phase J .

1=1

552

where the pressure correction (H ~) IS

calculated using the Redlich-Kwong equation of
state.
The partial enthalpy of component / in liquid
phase Jis

' ..... (A-l3)

. ..... (A-I4)

Nc

HJ

xIJHIJ, ................. (A-IS)

1=1
*Critical pressure here is in atmospheres.

SOCIETY OF PETROLEUM ENGINEERS JOURNAL

with HIJ given by Eq. A-12. Internal energy is taken

as equal to enthalpy for solid and liquid phases.
Internal energy of the gas phase is

. (T) [ 1 - S wir ( T) - S org ( T) - 5g ]

rolW
1 - S . (T) - S
( T)
wlr
org

=k

k
rag

nog

........................... (A-25)

U 3 = H3 - 144p/778p3 . . . . . . . . . . . . . . (A-16)

The gas-phase viscosity is

rgro

5g

1_ S . _
wlr

Sgr

_.
Sorg Sgr

ng

. .... (A-26)

Nc

/l3

xJ3/lJ3 ' . . . . . . . . . . . . . . . . . .

(A-17)

1=1

Stone's second method,13 in modified form, is used

to calculate oil relative permeability as

where
VISO(l,3)TNVIS (l,3) , . . . . . . . . . . . (A-18)

/lJ3 =

k ro

Liquid-phase viscosity is

II
1=1

/l;Y, . . . . . . . . . . . . . . . . . . . . .

(A-19)

where /lIJ is calculated from the ASTM standard

form l2 :
In[ln(/lIJ +-yp] +alnT

= b,

............ (A-20)

where -y is about 0.6 and p is liquid density in grams

per cubic centimeter. We drop the term -yp and place
a minimum value on /lIJ. The values of a and bare
specified through input data and can be calculated
from known values of component viscosity at two
temperatures. Eqs. A-17 through A-19 follow Ref. 2.
Liquid water viscosity is obtained as a single-valued
function of temperature from a read-in table.
The mole fraction of component I in phase J is
XIJ

KvIJX I ' . . . . . . . . . . . . . . . . . . . . .

(A-21)

where KVIJ is an equilibrium K-value and XI is the

mole fraction of component I in component I's
master or. primary phase, J I' That is, K vIJI == 1.0.
ForJ~JI'
K vIJ

= [K VI (l,J) + K

V2

(l,J) / p

+K V3 (l,J)pJe

T-Kvs(/,J), . . . .

(A-22)

where K VI through K V5 are read as input data. The

form of Eq. A-22 is that used by Crookston et al. 2
For water, K V13 is calculated as Ps/p, where Ps is
water vapor pressure, a single-valued function of
temperature. If T exceeds Tc for water, then Ps is
taken equal to water critical pressure.
Relative permeability and capillary pressure data
may be calculated analytically or interpolated from
read-in tables. Analytical calculation of relative
permeabilities is as follows.
rw

rwro

(T)

[ 1-

5W -

S . (T)
]
wlr
( T) - S . (T)
wlr
orw

nw

........................... (A-23)

1 - S orw ( T) -

row

5w

] now

=k (T) [
rOlw
1- S
( T) - S . (T)
.
wlr
orw
........................... (A-24)

DECEMBER 1980

+ k rw

+krg)-krw-krgJ. .. . (A-27)

Gas relative-permeability hysteresis is handled as

follows.

(5g ) max

' . . . . . . . . . . . (A-28)
1 - S wir - S org
w~ere Sgrc is current residual gas saturation,
(Sg ) max IS historical maximum gas saturation in the
grid block from start of the run, and Swir and Sorg
are read in (nontemperature-dependent values). Sgr
is read-in residual gas saturation. Sgrc' after
calculation from Eq. 6, is changed to satisfy Sgc :S:
Sgrc :S: Sgro where Sgc is read-in critical gas
saturation and S$r ~ SgC' Current effective residual
gas saturation (~gr) for use in Eq. A-26 is calculated
as
grc

gr

Sgr = wSgrc

+ (1 -

w) Sgc ' . . . . . . . . . . . . (A-29)

where

w=
As long as Sg is increasing, always ~ (Sg) max' these
equations give Sgr = S c' and kr from Eq. A-26
will follow the "original$, krg vs. ~ curve. Then, as
Sg decreases toward zero, krg will approach zero at
SgrC'

K V4 (/,l)

krow
k roiw ( T) [( k roew ( T)

'(k krog
rOlw (T)

Nc

/ll

This hysteresis is generally not important and has

an effect (not a large one, from our experience) only
in cyclic steam stimulation runs.
Gas-phase relative permeability is treated with no
temperature dependence.
The temperature-dependent quantities Swir (T),
k rwro (T), etc., all are treated as
X

=X

+a(T- T i )

, ................

(A-3l)

where XO and the temperature derivatives (a) are

read in as data.
The normalized saturations in the above equations
are

5w
50

Sw
I-S4
So
I-S4

and
553

s~g =

~
I-S4

SI Metric Conversion Factors

Analytical expressions for capillary pressures are

Pcwo(psi) = [PCC(I) + PCC(2) x (1- 5w )]
+PCC(3) x (1-

5w ) 3]

x [1 - PCC( 4) x (T - Ti ) ] ' .. (A-32)

and
PCgo(psi) = [PCC(5) + PCC(6) x 5g
+ PCC(7) x

5lJ x [I - PCC(8)

X (T - Ti ) ] ..............

(A-33)

API
atm
bbl
Btu
cp
cu ft
of

x
x
x
x
x

ft
Ibm
Ibm mol
psi,psia
scf
sq ft

x
x
x
x
x
x

141.5/(131.5 + API)
1.013 250*
E+02
E-Ol
1.589 873
1.055 056
E+OO
1.0*
E-03
E-02
2.831 685
CF - 32)/1.8 + 273.18
3.048*
E-Ol
4.535 924
E-Ol
4,535 924
E-Ol
6.894757
E+OO
E-02
2.863640
E-02
9.290304*

g/m3
kPa
m3
kJ
Pas
m3
K
m
kg
kmol
kPa
std m 3
m2
SPEJ

PCC(I) through PCC(8) are input data. The tem-

perature dependence uses the assumption that interfacial tension is a linear function of temperature.

554

Original manuscript received in Society of Petroleum Engineers office Sept.

23. 1979. Paper accepted for publication April 21, 1980. Revised manuscript
received Aug. 18, 1980. Paper (SPE 8394) first presented at the SPE 54th Annual
Technical Conference and Exhibition, held in Las Vegas, Sept. 2326,1979.

SOCIETY OF PETROLEUM ENGINEERS JOURNAL