You are on page 1of 13

In-Situ Gel Calculations in Complex

Reservoir Systems Using a New

Chemical Flood Simulator
T. Soott, SPE, U.K. Atomic Energy Authority, Mrinfrith
L.J. Roberts, U.K. Atomic Energy Authority, Winfrith
S.R. Sharpe, U.K. Atomic Energy Authority, Winfrith
P.J. Ctifford, U.K. Atomic Energy Authority, Winfrith
K.S. Sorbic, SPE, U.K. Atomic Energy Authority, Winfrith



This paper presents results for a series of calculations on the deep emplacement of a polymer gel in a stratified
reservoir nmdel. These calculations were performed with a new chemical flocding simulator that has the facility to describe
generalized chemical reactiom between components. This code is described, and prelimimry calculations on an in-situ gel
treatment in a large model reservoir are presented. We find that to obtain significant amounts of incremental oil while avoiding
very large pressure buildup, the polymer geI system must have the correct mmnbimtion of long gelation times and good
permeability-reducing properties in the high-permeability streak (residual resistance factor FR -10 to 40 in the cases studied here).
No commercial polymerl cmsslinker systems are currently available that have the very long gel times required to obtain deep
emplacement in a large reservoir system,

During recent years several chemical systems have been suggested
for blocking off ve~-hi h-permeability channels in heterogeneous
petroleum reservoirs. 1.3 These systems involve the injection of
polymer [either polyacrylamide. (PAM) or xantban polymer] and
a crosshnking-ion or redox system to form a suitable gel, Reacrams
may be mixed jnst before i~jection imo the reservoir, or lhey may
be injected m alternating slugs of the two materials.
To date, no catcukationshave appred in the Iiteramre that atmmpt
to qatify the flow patterns .md oil recoveries expected when a
polymer gel system is injected into a heterogeneous reservoir. In
this paper, such calculations are presented, assuming a simple mcdel
of the gel kinetics and bebavior in the porous medium. These calculatiom have been carried ow witi a new chemicat tlcd simulator,
Simulator for Chemical Oil Recove~ and Polymer Injection
(SCORP1O),developed at Winfriti. SCORP1Ois a genemd-purpxe,
mukiphase, mukicomponem chemical-flood simulator fhat may be
applied to polymer, surfacta.nt, or caustic flodlng o either the field
or laboratory scale. This paper includes a description of the simw
laters underlying mathematical formulation, amplifying those features that cleat with in-situ gelation. Results from in-siru gel
calculations are discussed in some detail, particularly the effects
of gel formation rate and pore blocking on oil recovery efficiency.


is based on the metbcd of tinife differences and is desigmd
to be flexible enough to handle a wide range of petroleum engineering problems. Up to 10 chemical componems may be imluded,
and these can be distributed among up to three liquid phases (aqueous, oleic, and micellar). Flow may be treated as either compres.
sible or incompressible, making due allowance for rock
compressibility. Two new features for a chemical flood simulator
are included in SCOSPIO(1) a generalized mcdel of chemical reac.
tion that defines the rates and stoichiometry for any series of coupled chemical reactioms between components witMn a given phase
and (2) a heat-bal?mce equation that allows calculation of temperature fronts and contours within a reservoir caused by injection of
cool water into a hot reservoir. Heat flow across reservoir boundaries is treated with aquifer-type models.
The calculated temperature may feed back onto such physimt
properties as fluid viscosi~ and reaction rates. Some calculations
that use facilities of this type have been presented previously in

a study of the effects of temperature on the chemical degradation

of polymer i a stratified reservoir system. 3 This module is used
to describe the chemical reactions tiat fake place i gel fomtions.
In addition, modules are available to model the influence of mm.
pled adsorption of compocrds and the effects of frontal spreading
lhrough velocity dispersion and molemlar diftisiom The cede can
accommodate several different rock types for relative permeability. adsorption, polymer residual resistance, and capillary pressure,
in addhion to allowimg for regional variations in rock permeabili~. ~j~iOn ~d pr~OctiOn wells ~ ~ O~~t~ under ei~er PICSsure or rate constraints and may be completed in any number of
layers of the reservoir.
The compositional fornwlation used in SCORP1O
is m extemio
of that introduced by Acs et al.4 and subsequently used by
~~er~5.6 i heir de development work. This approach has b~n
a&pted to chemical flooding and has proved a good foundation on
which to establish the framework of the sinmlator.
Simulator Equations. fn thk section, the differential equations that
govern fluid flow in rhe reservoir are presented, together with a
statement of the physical significance of the various terms represeming important phenomena. The corresponding difference equations
and tie solution strategy used to solve for the primary variables
of the system are described later.
If we assume that there are N, chemical components, including
water and oil, then for each Component i, conservation of mass
yields the following differential equation i oil mser-mir







. . . . . . . . . . . . . . . . . . . . . . . ...(2)

WE ReservoirEneineerine. No.emba 1987

Mass- rather than volume-based equations are used because fhe main
relationships used when describing displacement processes arc based
on the mass conservation of each component. The quantity fi;, referred to as the mass density for Component i, is the mass of Component i per imit volume of fluid. In general, this includes the
adsorbed and mobile components, as indicated in Eq. 2.
The miss concentration of Component i in ,Phase a is denoted
by C: and is equal to the mdss of Component i per unit mass of,
Phase w In the present study, the phase will be either aqueous or
oleic. The term OR represents the density of the rock, including
its pore space, whereas pa denotes the density of Phase ct. Contributions from molecular, diffusion and velocity dispersion ire incorporated in the, term D:. Gel formation is handled through the
reaction term, R; , which accounts for the formation of Component i in Phase a; this is represented as the mass of Component
i per unit volume of Phase c-iper unit time. The well terms are represented by qi, which is the mass of Component i being injectedlproduced per unit volume per unit time.
Along with Eq. 1,which is used to evaluate the change in mass
of Component i, the model also requires the derivation of a pressure equation. This is usually obtained by summing Eq. 1 over all
i ad using constraint relations on the C;. In the present forrmdation, however, a differen: approacti is adopted.
The pressure eqwdion IS obtained by introducing the concepts
of fluid and effective PV. PV is taken to be a function of pressure
alone, whereas the timctional dependence of the fluid volume, V,
is smmizrized as V= Yf(p, T, ~), where ~ = {ml,
.P% f
Pressure is referred to by p and dependence on temperamre ~as
been included explicitly in T; mj represents the mass of Compcr
nent i. In tie applications to be described in this paper, however,
it is noi necessary to include explicit dependence on temperature
because it is reasonably assumed in chemical flood processes that.
variation in temperature takes place slowly within the reservoir.
Of course, this would not necessarily be so in the presence of a
vapor or, gas phase. Thus differentiating Vf witi, respect to
time gives
avf, afi ap

avf ami
. . ........

. .,...........(3)

The partial volume WITI 8V/am; represents the increase in the

fluid volume caused by the addition of a unit mass of Component
i. As Dodge23 showed, these partial volumes are related to the total
volume by


~f= ~ a,ni;.


Use of the product mle for differentiation shows that


Resewoir Properties
Resewoir dimensions
length, x, ft
width, y, ft
thickness, z, ft
Mesh definition
Ax, fl
Ay, ft
for four low-permeability layers
for two low-permeability layers
Horizontal permeability, k~ (100:1 permeability ratio)
Low-permeability regiox k. =ky = 100 md
High-permeability region: k. =ky = 10,000 md
Vertical permeability, kv
Ratio (kv/kH) is varied in each region
between 0.1 and i O4.
Porosity, .$
$=0.25 (high permeability)
c$=0,20 (low permeability)
Injection rate
34,500 ftslll =6,140 BID.
Corresponds to 1 PV in 5 years and is maintained for all
cases (see text). Wells are located at each end of the
reservoir (NX= 1 and NX= 20) and are completed in all
layers except during gel placement.
Rock density, pR
Required to calculate adsohtion tevels
p.= 129,8 lbm/fts =2.08 g/cm$
Fluid Properties
ViscosiPj data, cp
Water viscosity, p ~
Oil viscosity, p.
Fluid densities, p, IbmIfts
Water (polymer solution or gel) density, Q.
Oil density, p.
Relative permeabilities, km, km
(See Fig. 1 for these curves)
The end points are the same in both region%
SW.= 0.25 SW=0.22
Compressibility data
All fluids are incompressible in these calculations
OiI in place, bbl
Total OOIP
=4.725x107 fts=8.415x106
Target (movable~ OOIP
=3.339XI07 ft =5.947 X1O3


which, when substituted into the previous Sic, leads to the tired
form of the pressure equation used in SCORPIO
Substimting this into Eq. 3 and adopting the relation in Eq. 4 and
the definition of mass density, fit =nzJVfi shows that
avf ap
ap at

+ [

av, a(+mi)


a+ ap
. . . . . . . . . . . . ...(5)





( )1

.+ avf ap
v, ap at


Again the product rule for differentiation can be applied to give

SPE Resemoir Enginemig, November 1987

Physically, Eq. 5 can be interpreted by saying that the rate of change

in volume o? fluid Eresent caused by pressure changes Plus the volumetric rate of fluid influx equals the rate of change in PV.
The derivative .3(.#@/dt is eliminated according to Eq. 1. The
governing equations of the system are therefore Eqs. 1 and 5, which
must be solved for the N,+ 1 primary variables of pressure and
overall component mass densities. Art additional equation is also
needed to calculate the temperature as a primary variable. SCORP1Oincludes such an equation, based on heat balance, but because
temperamre dependence is not included in our current study, we
omit further discussion of temperature and its calculation. Some
details on this point are included in other work. 3<7

if the fliids and rock are taken to be incompressible, then only

the summation term in Eq, 5 is nonzero.
Auxiliary relations that are required to solve Eqs. 1 and 5 in;
elude the usual saturation constraint:

Xse=l, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... . ...(6)


,wherc the saturation, Se, is defined in terms of the volume, Ye,

occupied by Phase a and the total fluid volume, Vfi such that
sa=ve/vp .......................... . . . . . . . . . . ...(7)
An additional equation is provided through the capillary relation,

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

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

where Pc denotes capillary pressure between the phase. pressure

pa and the pressure obtained by solving Eq. 5.
The soltio of Eqs. 1 and 5 by use of the constmits given Liy
Eqs. 6 through 8 is described in the following sections.
Difference Equations. The first step is to apply block itegmtion
to Eq, 1 to reduce the system of component differential equations
to discrete form, This implies the costmction of gridblcck volumes
around fixed gridblock centers with the subsequent ilerivation of
a tiite-difference equation for each block, Application of thk approach to obtain the difference equations is particularly significant
in reservoir simulation because the interface and bmmdmy conditions may be handled more readily.
Thus, suppose a typical gridblock face is dmtotcd by At, where
in dmee dimensions f= 1, . ..6. ad that Pdenotes tic center point
of this block, which has a vohime V?,. The distance between this
block center and the neighboring block that shares the face At is
specified m
AX# =(Axp+Azp,)/2,
with Art and AxV at the edge lengths i that direction of the neighboring block and the block under mnsideratio, respectively. The
final form oFEq. 1 for the gridblock with center f, ad mmpoent
mass tnl,p = Vrcptfzi is then taken as

Quantities with a bar, such as phase density, denote evsfuation

at block interfaces using weighted arithmetic averages such that
~t+ ,h=(Ax@tc +Ax@P)/(Axg+A
Themass concentmuio, C~, a?dmobility quotient, k,=lw=, inEq.
1I are calculated at block interfaces by use of either one-or twopointupstteamin~. Rock permeability iscalculated by harmonic
averaging as follows:
k8, =ktktr(A.q+Axt,


Note that the source term Qi,f, is indep&dent of the neighboring

For the pressure equation, the corresponding discredzation is obtained inaslightly different manner. As already stated, the fluid
and PVs in a block are assumed to be equaf at the end of the .
dmestep, At. AcseraL4showed thstasmall discrepancy between
fluid and PVs measured at the beginning of the time intervat will
arise because the pressures cslctdatcd at the previous time level are
not exactly correct, which leads to tie fluid in that block not filling
the effective PV completely.
Therefore, starting with theassumptiontbat ingener-al

T+ AT, ~+b+)=VP(P+Ap)

~t time fI+l wimp, T, md ,7? the primary variables at time t,

linearization leads. to the required pressure equation:


a v Ap
avf Ami
i ami At
ap At
.... .... .

ap At
. . . . . . (12)

As explained previously, temperahye dependence hss been assumed

negligible in the derivation of this equation. Such an assumption
would not hold for thermal rccove~ processes. The term iirparentheses is the volumetric error term it arises omfyio the case of compressible flow. Indeed, for incompressible flow. only tie term
involving a summation over i remains. For either type of flow,
Ami/Af is replaced for computational purposes by the expression
; l..
R. 9.
The partial derivatives in Eq. 12 are treated in various ways., For
nonzero but tixed compressibility, we use. the relations


for PV and

. ~fvf

+AtQi,tZ. . . . . . . . . . . . . . . . . . . . . .

... . ..

. (9)

for fluid volume, where the totaf fluid compressibility, c , is taken

to be a weighted sum by saturation of the individual p 1ase compressibilities,

Thcsou~e term Qifor Blockt, icldes wcllad chemical raction terms,


. . . . . . . . . . . . . . . . . . . . . . . . . . .


The fluid volume, Vfi is evahtated as the snm of the individual

fluid-phase volumes, Va, such that

and the coet%cient .@ is a compact notation for


. . . . . . . . . . . . . . . . . . . . . . . . . . . . ..(11)

The time increment, At, is given as the difference hctween the beginning, tfl, andthcend, tn+l, of the timestep (Ar=#+l-tn),
with a similar meaning for AIrq.

v,= xv. ,
where Vu= (mass in Phase rY)/Pe.
In the present work, we assume tiat atl components apart from
oil partition wholly into the aqncous phase. Oil partitions totally
SPE Reservoir E@min& November 1987

into the olcic uhase. Thus the Dartkd volume term for a typical
block ii com@d is




where mi is wholly partitioned into Phase a.

As discussed by Watts, 6 the discrete form of the pressure equation, Eq. 12, presents an attempt to find the pressure that causes
the fluid volume in a block to fill the PV of that block exactly. For
instance,, if there is too much fluid in the block, the equation should
force the pressure to increase, thereby compressing the fluid in addition to forcing some of it to leave the block. If just the right pressure is found, the fluid remaining in the block at the end of the
timestep should precisely fill the PV of the gridblock.
fn summary, the key equations are therefore Eqs. 9 and 12. The
solution stratefg for these equations is dkcussed next.
Method of Solution. The methcd used to solve &e fundamental
system of equatiom for the primary variables is based on the familhr
implicit-pressure, explicit-saturation (3MPES) approach, 8 except
that here the component masses are evaluated explicitly. The cd:
culation of the phase saturations is then performed with the expticifty
determined component masses.
Thus we solve Eq. 12 implicitly for the oil-phase pressure. Other
phase pressures are obtsined through the capillary relation, Eq. 8,
where the basic fMPES assure tion that capilkwy pressure is constsnt over a timestep is used. ~ For each Component i, Eq. 9 is
solved to obtain the change in mass, Ami, explicitly for each gridblock. Given the mass of Component i in each gridblock for all
i, the mass in each phase is evaluated, somedmcs referred to as
the flash calculation. If the fluids are compressible, tie new phase
density is calculated (it is currently assumed that density is a function of pressure alone). Oti&iwise rhe overall calculation goes
straight to the evaluation of the volume, V., occupied by Phase
a. It is then a simple matter to obtain the phase saturations at time
~n+I from Eq,~.To adv~ce to the next timestep, the sam~ti0ris3
component masses, and phase pressures arc used to evaluate the
various coefficients appearing in Eqs. 9 and 12. The timestep is
updated and the process of solution repeated, starting with Eq. 12.
In our model; wells are treated explicitl~ fhey maybe completed in ay number of gridblock or layers of the reservoir. The controls can be either pressure or.rate consrmjnts and may be switched
from one to the other under prescribed conditions. Rates can be
defined in terms of mass or volume.
The discrete-pressure equation (Eq. 12) gives rise to 2 symmetric banded coefficient matrix that could possess up to seven bands,
imldig the diagonal. SCORPIOincorporates direct, and iterative
algorithms to solve the system of linear equations. The direcfsolvers include the multifrontal method of Duff and Reid, 9 which is
particukly suited to solving sparse mattii systems whose sparsity
pattern is unchanged from one timestep to the next. Tbe iterative
options include a preconditioned conjugate gradient method that
stems from the work of Meijerink and van der Vorst0 and Ker.
~kw, t i We have found that thk latter solution technique is extremety fart and efticient for selected premnditionings and is suifable
for relatively low numbers of gtidblocks rather than restricted to
the cbamcteristically highir number of gridblock USWJIYassociated
with iterative solvers.
The SCORPIOcode is based on a modular structure so that the
majority of modules essentially supply the various terms needed
to construct the coefflci&rs in the difference equations. Once tbcse
coefficients have been provided, the main solver module will evaluate the primay variables, i.e., pressure and the N, component
masses. .Other modules will then determine from these primary variables the seconda~ variables, such as component concentrations
in each phase present (the flash c~culation), phase densities (for
compressible flow), viscosities, phase vol~es and saturations, adsorption and reaction terms, relative permeabilities and capillary
pressures as functions of satmadon, diffusion coe.fflcients, etc. The
program has been written to take as much advantage of the vectorp~cessing capabilities of tie Cray computer as possible. Wifh some
minor modifications, however, it easily could be run on another
SPE Reservoir Engineering, November 1987

Reaction and Adsorption Modules in SCORPIO

To perform calculations in polymer gel SYStetIIS,it is necessary to
have some generalized representation of the reaction terms RL in
Eq. 1. These tetis may be very simple and describe a single-step
A +B+ C reaction, or they may be more complex and describe a
mukistep process in gel formation. It is also necessary to describe
the main physical chzacteristics of the reactants (polymer and crosslinker) and gel prcducts. The main properties that must be modeled
are (1) the effects of the polymer and gel on the mobile aqueousphase viscosi~; (2) the adsorptiqtl retention of polYmer and PIOduced gel; and (3) any corresponding reduction in permeabiliv that
results from the adso tionlretcntmn of material, i.e., residual
,e,i,mnce fader,, ~R~
The first of these relations concerning the effective viscosity of
the polymer gel systems is specified by the user tiom appropriate
experiniental data.
Reaction Module. The SCORPIOreaction package calculates the
production (or degradation) rate of Component i in any giVCnPhase
& This rate is allowed to depend on the concentration of every component in that phase, including itself, and on temperamre. Both mobile and adsorbed components (except water and oil) may be
produced or degraded by chemical reaction: The total mass change
of Component i in the gridblock is then calculated.
The reaction package has two main options for calculating rates
of change of Component i in Phase a caused by reactiotidegi-adation, i.e., (a C!#Jt)rxn. The first of these uses tabular data entered
by the user to calculate this quantity, and the second uses an empirical formula.
The total mass rate of change of Component i caused by reaction, including both mobile and adsorbed material, is given by

where the adsorbid concentration per unit rock mass in confact with
Phase a, ri,a,is assumed to be.identical for each Phase a, but with
a different reaction rate for adsorbed material in contact with each
phase. The quantity fa denotes the fractioq of rock surfacein contact with Phase e and is a fimcticm of salutation and nettability.
Note that f= is not necessarily equal to the safurafion, .%. For example, in a water-wet rock, the fraction of rock surface in contact
with the aqueous phase may be close to unity.
The empirical equation that may be used to catcutate the reactionmte term is constmcted on the assumption that the rate of production of a component from a given reaction in Phase a is proportions
toaprcduct of powers of the concentrations of all componen~
associated wifii that reatiiow i.e.,

UO(CL), .(cJJj



where the quantities a.. .aN+, are constants for each Component
i, and N denotes the total number of mobile and adsorbed concentrations contributing to the reaction. In some simple reactions, the
exponents a,
aN are integers, but in general they are nonnegative real numbers. The tem~erature dependence of the reaction me
is assumed to follow the Arrhenius law through the final term in
the equation. The quantity aN+ I.represents EJR, where Ea is the
activation energy of the reaction and R is fbe gas constant:
The quantities of each component produced by reaction over a
timestep are recorded and incorporated into the overall materialbaknce calculation in the simulator.
Simplified Gef Kinetics. Several smdies have appeared in the literature on the kinetics of gcl formation. 1318This may be a multistcp
kinetic process if complex mixtures of polymer and re@x systems
are used. For example, we previously studied modeling the
PAM/Cr207~- Ithiourea system in experimental floods. 19


a general function of the aqueous-phase concentrations of all components and tempaamrc. This is perfectly adequate to describe pcdymer and gel adsorption onto the rock nmtrix.
The mass density of adsorbed Component i, ?iq, is given as (see

wherepw, SW,and Ci aretheaqueous-phase densiW,saWmtiOn,
andconcerdration of Componenti, respectively, and C,q istbe entire concentwion vector in the aqueous phase. The adsorption
problem in. this case involves solv@the fOllOw&&,setOf nOnIinears imultaneouse quations for C~a!d. hence fYG. T) fOra
given mass densi~ vector, r%


Fig. lRelative-permeability
in the model reservoir.

cutves for the two rock types

A simplified reaction scheme for gel formation maybe modeled

by assuming that the reaction is of the form polymer (C I)+crosslinker (C2)+gel (C3), where the Cis refer to themass concentrations in the aqueous phase. It is assumed that the gel formation
is first order in each of the reactions, i.e., second order overall.
Defining C! o and C2 as the initial concentrations of polymer and
crosslinker, respectively, the reaction rates may be specified through
a single rate constant, K, where


= KC1C2,
-(Cl dt )



(Cl +C2)




This simple system may be integrated analytically to give the gel

concentration as a fynction of time r

q =

l+ KC, C2t

Adsorption Modufe. The central problem in a generalized adsotptionlphase package inamukiphase, multicomponent simulators
as follows: given thetotal mass densify fii inagridblock forall
components (i= 1, 2:. .N,), howmethese componenf.s distributed
between the Phases a(cz=aqueous, oleic, micellar) andthe rock
surface? A more restricted module is currently used witlin SCOR!+o, butthii is being generalized, fnthepresent work, theadsorption level of Component ion the rock surface maybe specified as

When component adsorption depends on only its own cqcentmtion (in the aqueous phase), it is possible to use a fast interpolation
mmhodtos cdvet hisproblem. Iftheadscyptioni sothermdepcnds
on two or more components, however, an iterative Newton-.
Rapbfmn20scheme is used. If the maximum number of componenb
involved in coupled adsorption is N., then asmal syst?m Of N~
nordinear equations must besolvcdto find adsorbed and mobile
levels of components This is a fairly straight fonvard process requiring the solution of an No xNa, matrix equation at ~ch iteration.
Many materials that adsorb onto reservoir rock do so !r.rcversibly; i.e., they desorb very slowly compared with other time scales
inthesystcm. Irreversible adsorption istreated intbemodnleby
tracking the amounts of each component adsorbed in each gridblock.
It is assumed that after some maximum irreversible level of adsorption is reached, further adsorption iseitber reversible or does
not occur.
Residual Resistance Factors. Permeability reduction (pore blocking) caused by adsorbed polymer or gel is treated through residwd
*&tmce faflors, FR.12 These defmethedrOp inthe mObility Of
Phase a, h=(usually tbeaqueotis phise), caustiby adso~tionas
X. =k,,&aFf@j]
These resistance factors are specified,through tabular input and may
be functions of all the zdsorbed species.
Computational Results
.TheRcsemoir Model. A simple two-dimensional layered reservoirsystcm was used in the calculations p:esentedin this paper.
This model is adequate here because it shows all the main features
of themechanism.of gel emplacement ina stratified systcrn and
itallows ustoinvestigate several sensitivities, including thee ffect.s
of vertical crossflow (kV/kH), gel reaction rate, and injection strategy, inastr~ght fOrward way.
Table 1 gives the reservoir and fluid characteristics used in the
cdcukuions.rhere semoiris 3,000ft[9l5mllong~d
III] thick and bas two distinct permeability regionsa lowpermeability region of 80-ft [24-m] total Wlckness and a highpermeability streak of 20-ft [6-m] totat thickness. In most calculations, che reservoir is represented by 20 gridblock in the x dlrection and by sixv.mtical gridblock: four inthe low-permeability
rcgionand twointie high-permeablli~ region. Calculations ona
finer mesh show that all the mtin conclusions maybe reached with
this grid. The horizontal permeability contrast is taken to be 100:1,
with the higher-petmwability streal at thebottom to reduce the helpful effects of gmvity in the oil recovery mechanism. This high permeability conhast is adopted so that very early water breakthmugb
will be observed, and in addition, a gel treatment would be expected to pefform substantially better than conventional polymer flooding. The relative permeabilbies for the two reservoir regions are
shown in Fig. l. Alinear saturation dependence isusedin tiehighpermeabili~ streak. This and the fact tludtthe mobility ratio in thk
system is very favorable (0.67) produce piston-like oildisplacement. Thus polymer will not contribute to the oil recovery mechaSPE Resenmir Engineering,November 1987



Synthetic Polymer
(Model PAM)

(Model Xanthan)

Injection concentration

1,000 ppm


4 cp at

High permea~lity.
LOW permeability



injection conceritcation

(No adsorption)


1,500 ppm

Concentrations by weight

2 CP at injection concentration

For viscositylconcentiation
curves see Fig. 2

0.28 x 104
0.44x 104

(160 lbm/acre-ft)
(250 lbm/acre-ft)

Given in fractional
concentration (Ibm polymerl

Ibm rock, assuming

PR = 121.7 lbml~ts in the
high-permeability streak ad
129.8 lbmlft3 in the 10werpermea~lity streak), a
stripping irreversible
isotherm is assumed.

FB is quoted for the synthetic

Residual resistance factors,

High permeabifky
Low permeability




Polymer degradation



Injection strategy

Q=24,500 ft31D (6,140 BID)throughout. Polymer injected for

150 days after 450 days of waterlood (i.e., an 6.2% PV slug
of polymer)

W8S of polymer injected

Amount of oil recovered, per
unit mass of polymer

32.3 x 104 Ibm

(147 tonnes)
306 Ibm oilllbm polymer (1 .0S

polymer at the mwimum

adsorption level (see above]
it is.taken to be a hear
function of adsorbed
No polynier degradation is
assumed in these
calculations (see Ref. 3)
Because of the low visco&ieS
and FRs for these
polymers, large pressure
buildup was not found for
these injection rates

(220 tonnes)

48.4x 104

212 Ibm oill!bm polymer (0.75


Results at 1,500 days

nism by significantly impmving microscopic displacement efficiency

in this region. Hence, any additional oil will arise through fluid
diversion and crossflow, which are the important mechanisms in
these systems. 3,21
Onc of the qmin objectives is to investigate plugging by a suitzble polymer gel in-depth. That our reservoir model has a large
volume and well spacing is ve~ important because the requirement
of long gel times is implied. We intend to define some of the key
pttrametcrs that are required for a successful application of this
chemical system. We stress that we are not studying small nearwell gel treatments,
Water amf ?olymer Ffooding Resufts. In both the water and polymer floods, water is injected into the whole formation at a rate of
34,500 fc3/D (6,140 B/D [980 m3/d]). This corresponds to the injection of 1 reservoir PV in 5 years. It is desirable to maintain this
rate, although high pressure buildup may arise in some cases. In
such cases, we define maximum acceptable differences between the
pressures of the injectorlpmducer well pair. .
.The properties of the two polymer typs used in the calculations
arc pm.scnrcdin Table 2. We assume that .fheinjection water is quite
saline and hence that PAM is a poorer viscosilier than xanthan.
However, this is compensated for by the pore-blocking properties
of PAM. Fig. 2 shows the viscosilylconccntration curves assumed
for the two polymers.
Polymer flooding calculations were perfommd for both xamhan
and PAM and compared with the corresponding waterflood calculations over a range of vertical-to-horizonti permeability ratios from
10-1 to 104. fn all cases, a polymer slug was injected over 150
days after 450 days of previous watertloochg, when the water cut
had risen to 95 %, At tie injection rate used, this led to an 8.2%
PV slug of polymer. The cumulative oil recovery profiles with time
are shown in Fig. 3 for kvlkH =0.01 for a waterflood, a xanfhan
flood, and a PAM flood. Corresponding water-cut profiles are
shown in Fig. 4. The effect of kvlkH on oil remve~ at 1,5CQdays
is summarized in Fig. 5. The following is noted.
8PE Reservoir Engineerig3November 1987

Fig. 2Concentration dependence of viscosity for the model

xanthan and polyacrylamlde




.,,, 5,,



















i9. 3Cumulative oil recoveiy

]ast%case polymer floods. ~.

profiles for .waterfl.aod and

1., For the polymer properties assumed in this work (fable 2,

Fig. 2), both xznthan and PAM give very simifar amounts of oil,
although they operate through slightly different mechanisms.
2. Both the waterflood and polymer tlocds recover more of the
target oil. for the higher values of kvlk~.
3. Polymer flocding gives a larger incrementzi oil recove~ contpared with the waterflood at higher values of kv/kH because of increased oil crossflow from the low- into the higher-permeability
zone during polymer flooding, as well as fluid diversion. 3.21
4. At the highest crossflow level investigated (~v/kH =0. 1), the
polymer produces anadditiotxd 9% of the,target oil, i.e., a 30%
improvement over waterflood.
Analysis of these results has led us to choose the kv/kH =0;01
case as our base case. This gives an improvement of about19%
over the corresponding waterflood (i.e., a further 6% of the brget
oil), which, given that the simulation model tends to be rather optimistic, in these systems, is sufficiently poor tojustify a gel
G&I Properties. In the simple kinetic model described above, it
is assumed that the polymer and cmsslinker are in stoichiometric

ratios, whicli in this case are taken such that l,OWl ppm by weight
of polymer reacts with 20 ppm of crosslinker to give 1,020 ppm
of gel. Therefore, the time taken for conversion of half of the
1,000-ppm-polymer/20-ppm-cmsslinker mixture to gel is given by

KC, C20

Thus, a 10-&y half-conversion time corresponds. to K=5 x.106

day - 1; in our applications, werequire t,A to be about 100 days
and hence KG 5 X 10s day 1. The time profile of gel conversion
for a range of K values is shown in Fig, 6. Here, we are primarily
interested in the longer gel times, which allow deep penetration
of the polymer into the large reservoir model.
The physical propetiies of the resulting gel in tbe porous medium must be de,tined to,describe bow the gel actually behaves in
the reservoir system. In practice, polymerl crossligkef fluids appear to operate by first aggregating into small (pregel) clusters of
polymer molecules and, in some cases,. subseqtkntly forming full
tl@e-dimensional structures. Even in the pregel stage, the aggregates will be retained and will reduce the effective permeability
of the aqueous phase. It is this pore-blocking mechanism that is
inco~orated into the model gel considered in.thk work., Neither
the polymer nor the crosslinker is adsorbed onto the reck, and oidy
the gel is assumed to be adsorkdretained.
It is known that polymers and some crosslinkers, such as A13+ and Cr3 +, ~e adsorbed
on reservoir reck. Ordy prelimikuy results ate ptesented here, tiowever, and further work is in progress to study the effects of adsorption in more detail.
The properties of the model gel are given in Table 3. Two cases
are defined showing higher and lower levels of residud resistance
factor, FR. fn each of these cases, the FR in the low-permeabiiigJ
region is taken m be four times that in the high-penneabitity w.@.
Whife this factbr is somewhat arbkmy, it reflectt the fact that blocking is moresevere if geI enters or is formed in the low-permeability
strata and follows earlier reported practice. 7.22 However, the absolute value of the FR in the high-permeability region has been
found m be a much more important parameter. As expected, the





, !,


., -


,., , !, @A,,.

,.,. , !,

8 ,

r. ,s . ,,.

,, f





s !,, ,,,,,


i9. 5Effect Of vertical-to-horizontal

permeability ratio on
111recovety (after 1,500 days) for v@terfloods and polymer


SPE Reservoir Engineering, November 1987








Polymer, Co= 1,000 ppm

.Cmssfinker, Co= 20 ppm

Injection concentrations

t,fi (day)

K(d~y -l)

Rate constants

These are fixed for all calculations

A range of rate constants is Studfee
te is the time for conversion of half
the original material to gel (C and
Co as above)










(C)=0.5 +1.75X103
+1.75 XI06 c;

Polymer viscosity


This is a convenient analytical form at

C, = 1,000 ppm, the polymer
viscosity is 4 cp
1. Gel is assumed to adsorb
irreversibly with a stripping

Polymer crosslinker does not adsorb


Case 1
Case 2

5.65S x 105
8.844x 105
5.658 x 105
8.644 x10-5

2, Given in tractional concentration.

(Ibm polymerl[bm rock assuming
PR = 121.7 Ibmift in the highpermeability streak and 129.8 lbm/ft
in the lower-permeability streak)

factor (3)

(300 lbm/acr6-ft)
(500 lbm/acre-ft)
(300 Ibmlacre-ft)
(5001 bm/acre-ft)

levels of F;for the gel case are hi~her than those for the ordinary
polymer flood (see Table2) presented previously.
Remtfts of Gel Calculations. When a gelling system is injected
into a reservoir at a constant rate, there is inevitably a pressure
buildup. To put these pressures in context, consider two limiting
cases in our model reservoir system (1) when all injected water
enters the flooded-out 10-darcy streak, at the injection rate considered here (34,500 ft3/D [980 m3/d]), then, if 100% water flow
is assumed, AP across the reservoir is -140 psi [-965 @a], and
(2) when all injected fluid is diverted at the same rate into the 100-md
Iow-psnmeability tigion as a result of the total blocking of the streak
by a ve~ efficient gel treatment; then, if 100% oil flow is assumed,
the.correspottding AP across the reservoir is -2,300 psi [- 15.9


S. Twocases with. higher and lower

levels of pore blocking by gel are
studi@FR isquoted at the
maximum adsorption level

For simplicity in our calculations, we maintain a constant injection rate at above, but we specify a maximum acceptable pressure
drop across the system of 2,300 psi [15.9 MPa]. If the pressure
rises shove this value in a given gel flood, we assume that the flood
is not feasible. This pressure drop does not include the pressure
drop between the wellbore and the adjacent formation. This depends very much on the local conditions, such as the well skin factor and well fracturing, and therefore it not included. Only relatively
low-viscosity fluids ~e being injected into tbe fimnation (maximum polymer viscosi~ is 4 cp [4 mpa. s]), however, and unless
there is considerable reduction in pe~eability close to the well,
injectivity should not be impaired vew significantly.
In all cases,, the 8el injection strategy is similar to that for the
polymer cases discussed previously, The gel systems are injected,
however, into only tie high-permeability streak for 150 days after







...- >














Fig. 7Effect of reaction rate and residual resistance factor on cumulative

ter 1,500 days) and injectivity for pol ymerlcrossfinker

8PE Rse;oir

Engineering, November 1937


oil recovety










Fig. 8Cumulative

lg. 9Waler-cut development

md Case 1 gel calculations.

oil rec6ve!y

for waterflood,

profiles for waterflood,



450 days of water flooding. This avoids excessive gel blocking iii

the low-permeability region close to the injection well. Watertlo@ithen resumed over all layers for the remainder of the flood.
[n the first series of calculations, with the data in Table 3, the
effects of both residual resistance factors, FR, generated and die
gel reaction rate, K, were investigated. The effects of FRand K
on oil recovery (after 1,500 days) and maximum pressure buildup
across the resctvoir are summarized in Fig. 7. The maximum pressure value cccurs at the end of polymerlcrosslinker injection. Fkst
consider the higher blocking situation (Case 1, Table 3), whereat
maximum gel adsorption, the FR is 40 and 80 in the higlier and
lower strata, respectively. A range of reaction rats with
2.5x103 SKS8X 104 day-l is considered for Case 1; the maximum reaction at which tie AP across the reservoir remains ~cceptablc is K=8 x 104 day 1. ~is is equivalent to t ,Aof 625 days,
which is above rhe mrredy available practical gel times by a factor
of about 20. For boih this case and that where K=2.5 x 104
day -1, it is noted that increnmmal oil recoveries are considerably
above those for either the waterflood or the polymer floods; these
two cases recover a futtber 17 and 27 % of the farget movable oil
(quoted at 1,500 days), r+spcctively, compared with the polymer
ing is


polymer flood,. and Case 1 gel calcu

lle recovery profiles and water cuts for tlese cases are also
shown in Figs. 8 and 9, where they are compared with the waterflood and polymer flood. Note that the gel applications show approximately the sanie onset time for increased oil production rate
(or wider mt drop). The better gel application lowers water cuts
in the simple kesemoir model fmm -95% to as low as 32%, however, compared with a lowest water cut of - 72% in tic polymer
flood. Because of the permeability-reducing action of the gel, the
effect oit water cut is more persistent. Final water cuts in tie two
cases shown in Fig. 9 ~e -85 and 80% compared with a worse
postflood water cut of - 97% in the case of the mo~$lV-cOntrOl
polymer. Contours of water saturation, residual resistance factors,
and pressure Sre shown i Figs. 10, 11, and 12, respectively, at
a rangeof times during the tlood for the Case 1 example with
K= 8 x 104 day-1. The resistance factors in Fig. 11 are a measure of where the gel is actually located. The coriespondmg water
saturations in Fig. 10 show the progress of the aqueous phase during and after the gel injection. Fig. IOa shows that very little water
has entered the low-permeability region after 450 days. After the
polymerlcrosslinker reactants have been injected (at 600 days), tic
water invasion of the Iow-petineability strata is much improved,
as shown in Fig. 10b through d..
It is inmuctive to examine the pressure contoup at different stages
of the flood shown in Fig., 12. Before polymerlcrosslinker injec~on, pressures aie low (s 150 psi [s 1034 kPa]) and somewhat
higher in the.streak than in the low-permeability region. The pressure reaches a maximum of.- 2,100 psi [- 14.5 MPa] after 600
days at the end of chemical injection. At this time and over the next
300 days, there is a lower-pressure region just in front of the gel
slug that causes some crossflow of oil from the low- to the highpermeabiliw zones. The effat of this can be seen in Fig. 10b and
c, where water saturations in tbe streak clearly drop a little. These
crossflow mechanisms are discussed elsewhere. 21 Long afief the.
resumption of normat waterflooding, the pressure tield is as shown
i,nFig. 12d. The pressure across the system has:dmpped to -1,400.
psi [-9653 !&a], and pressures in the high- and low-permeability
streaks have approximately equalized. tn this case, the effect of
the treatment is clearly very long-lived because gel degradation has
not been included in. our calculations. 3
If gel blocklng, FR, is reduced by a factor of four in both the
high- and low-permeability regions (Case 2), then increased oil
recoveries are obtained only at higher gel reaction ratei, as shown
in Fig. 7. For a value of K-4.5 x105 day-i (ZIL-111 days),
however, the pressure constraint across the reservoir is again just

SPE Reservoir Engineering, November 1987


mm sANwrm6 .
a !Ca




days (start





a 3m
a 4m
a 6m
a 7m
1. cm




t =




(end of polymer/cro.s linker injection)


i9. 10W4W SatUmtiOn COntOUW during polymer gelation treatment (Case 1; K= a x 104


m=imum acceptable value. The

this case is between the Case 1 values for K=2.5x

oil recove~ for

104 and 8 x 104

We not; that the FRvalues used in these two cases for the highpermeabilhy streak (40 for Case 1 and10 for Case 2) are relatively low. It has been found that if high gel blocking is assumed, then
the penetration depth is much redu;cd and pressures become uriacceptably high, Gel adsorption levels are also quite low in this study
(see Table 3) for the same i.e., at high adsorption levels,
the gel system cannot penetrate deeply into fhe streak. Higher blocking Td adsorption levels may be acceptable for smaller systems;
this is currently under study.
.%unmsry of Results. A number of points emer~e from these
preliminary studies of gel formation in a layered system.
1. To obtain deep penemationof large stmthied systems, gels with
ve~ long setting times are required if pressure problems are to be
avoided. No systems with such long gel times are currently known.
2. Both the incremental oil recove~ and pressure buildup incr.$ase
as the pore blocking, FR, and reaction rate increase over the range
SPE Reservoir E@meting, November 1987

of reaction rates studkd. For higher-blocking gels, however, the

reaction rate must be sufflcicnOy low toavoid high pressure buildup across the reservoir.
3. It is clear from our calculations that the gel properties have
not been optimized. For example, in the reaction where
K=8.OX 104 day -1 in Case 1, Fig. 6 shows that the conversion
of reactants is less than 62% after 1,000 days of reaction. Clearly,
kjection cancenuntions and slug size of reac+ts could be oprimizgxt
in terms of oil recovery for 8iven F.q values and rate behavior.
4, If the technical problems of designing a very lonXtime-setting
polymer)croislinker system can be overcome, then this approach
may have considerable potential in large reservoir systems where
str@4ng is known to occur. The size of the reservoir studied in
this work imposes severeconstraints on the rate constant in the gel
system. If these constraints are relaxed somewhat, so that requited
penetration depths are about 2Cilrather than 1,000 II [60 rather than
300 m] and setting times, t,h, are 30 to 60 days rather than 100
m 1,000 days, then the time-setting properties of a suitable gel are
more technically feasible. Future work will examine the size sealing of systems in which gel applications may be performed.


t = 450 days (,*care of polymerlcrms linker i.jec. ion)

c = 600


z Om
& 003

days (end of polyme.1.ross linker ipjectiom)

r = 900 days

t = 1500 devs
Fig. 11-Residual
material. (Case 1; K=8x104

factors for polymer gel treatment

Concluding Remark? and Future Work

A number of simplified gel calculations have been carried out with
a new chemical flood simulator, SCORP1O.This is the first time that

been presented in the Iiteramre. The

simulator has performed very satisfactorily in these calculations.
Taken as a whole, SCORP1O
represents the cmrent state of chemical
flooding technology; it is flexible and has a wide range of versatility. Care must be taken, however, with the well models whm bighviscosity fluids or permeability reduction of the formation are involved. Further work is in progress to incorporate h more implicit
treatment of wells in tbe SCORP1O
Akhough our calculations are quite prelimina~, some interesting coriclusions on gel systems have been ~achd. The results in
this paper apply to large resefvoir systems coritaining very severe
high-permeability streaks in which we have attempted to specify
condkions for deep gel emplacement. Our most important conclusion is that, to obtain significant amounts of incremental oil while
avoiding excessive pressure buildup; the gel system must have a.
acceptable combination of long gelation times and good
calculations of this type have


showing deep penetration



properties in the high-penneabiliv


(FR -10 to 20 in the cases stadied).

Areas of work that require further study include the following.
1. More refined models of the gel kinetics that allow for the multistep nature of tie coagulation-type reaction andlor describe the
kinetics of a time-delay redox system to produce tie cmsslinker.
2. The effects of interactions, such as those betieen themck
matrix and the crosslinker-e. g., when Cr3 + isthe cr:sslinking
ion, it is known to adsorb onto the rock.
3. The effects of reservoir sizs in this work a large reservoir
system was use~ that requires very long gel times hcycmd ,tiose
of currently available polymer-gel technology, but in sma!ler sYstems, gel times on the order of 20 to 30 days may have considerable beneficial effect:
4. Beiter model descriptions of the physical bebav~or of tie gel
in the porous medkun (in this work, the gel was adsorbedlrefained
and permea.biliy reduction was described through residual resistance
factors, FR).
5: A wider range of sensitivities than has been, covered in this
papertie effects of gelation rate, initial rw.cmnt concentmtions,
SPE Reservou Engineering, November 1987


c = 450 days (start of polymerlcro~s

t = 600 days.(end of p.lpcrlcmss

linker injection)

linker injection)

t = 1500 days

i9. 12Resem~ir pressure distribution through polymer





fi2 [mz]

CP =
c =
$ =


total fluid compressibility, psi-1 [kpa -I]

pore compressibili~, psi -1 [kPa -1]
compressibility of Phase a, pi-1 &Pa-] .
adsorbed component comqitratio vector
.= mass concent;tion of Component i in Phase a
of Component
= initiallinjection
= depth below line of zero gravitational potential,
ft [m]
=, physical diffision of Component i in P@se, a,
c t12/D [mZ/d]
= activation energy, cal/g mol. [kJ/kMOll

SPE R.?semoirEngineering,November 1987



= residual resiswnce factor

g = gr2vit2ti0nal

aj = coefficients in general chemical reaction relation

4P. = area of interface between Gridblock t and!,




f. = frmtibn of Phase a in contact witi adsorbed





slug size, permeability coritrast, and blocking characteristics need

,to be examined.




rock permeability,


relative permeabili~ for Phase o!

horizonml rock permeability, md
vertical rock permeabili~,

K = reaction

rate constant

~ = vector of component
mi = mass of Component

= mass

for Phase a




for crosslinhg
masses, lbm
i, Ibm [kg]




lbm [kg]
densily of Component
i, lbm/ft3 [kglm3 ].
coefficient =(k,a/pJp@C~
N = total number of adsorbed and mobile reacting


= mass

ML =




. .

AJa = mtmkr of components involved in co~pied

including water
N, = total number of components,
pa pressure of Phase a, psi [kpa]
AP = pressure increment over tinre=pn+ 1Pn, psi [kpa]
Fe = capillary pressure, psi [kPa]

= pressure




for Component

9! = SOUrCe/SiItk tCrm
[kg/m3. d]

psi [kPa]
i, lbtn/ti3


Q = overall source fsink term for Component i idding

lbm/ft3 -D
chemical generationldcgradation,
[kg/m3 d]
R = gas constant, cal/gmol-K [MJ/kmoI K]
R~ = reaction rite of Component i in Phase & lbrnlt13-D
S= = satrxztion
f = time,

of Phase a, fraction


At = length of timestep =t +1 -f,


= half-conversion



of second-order




= temperature, F ~C]
= total volume occupied by all fluids, ft3 [m3]
= volume of Gridblock PI, R3 [m3]
= effective PV, ft3 [m3]
= block PV
Vm = volume occupied by Phase a, ft3 [m3].
Ax = incremental disfance, ft [m]
Ti = adsorption of Component i
of Component
i on rock surface in
Ii,a = adsorption
contact with Phase a
@= typical quantity to be evaluated at block interfaces
km = mobility of Phase a, l/cp [l/Pa.s]
I% = viscosity of Phase a, cp [Pa. S]
PW = density .of Phase a, lbm/t13 [kg/m3]
PR = density of rock, including pore space, Ibmlft3
+ = pOrOsity
i,j = component label

f =
rxn =
w =
.a =

gridblock connected to Gridblock ?,

current gridblock
term associated with chemical reacti.m
phase label (aqueous or oleic)

n = timestep index

Corwersion facfms ceded m achieve mmistency are implicit i

the equations.
We thzmkK.L. Rolf for her assistance with the compwing and R.I.
Hawes for his helpful comments on the manuscript. This work was
timded by the U.K. Dept. of Energy.

4. Acs, G., Doleschall, S., and Farkas, E.; General Purpose Compmitiond Model, SPEI (Aug. 1985) 543-53.
5, Kendall, R.P, a cd.: .$Developmentof a MultipleApplicationReservoir Simulator farsUse cma Vector Cmnpww,,, paper SPE 11483
mesentcd at the 1983 SPE Middle East Oil Tectuical Conference,
Bahr.uh, March 14-17.
6. Watts, 1.W.: ..A CompositionalFormulation of the and Smmticm Equation,,, SPERE (May 1986) 243-52.
7. Sorbic, K. S., Robem, L, J., and Foulser, R. W, S,:.. Polymer Flooding Calculations for Highly Stratified Brent Sands in the North Sea,
Pmc,, Secmd European Symposium on EOR, Paris (Nov. 19g2)

8. Aziz, K. and Settari, A.: P.Iroltwm Re.?en$ofr Simudarfon, Applied
Science Publishers Ltd., London (1979) 135-38.
9. Duff, 1.S. and Reid, J.f$.: .The MukifrontalSoludon of Indefinite
Sparse Symmetric Linear Systems,, Tntns., AWL for CmnPutig
Machinery, Mafh. Software (19g3).9, 3~-25.
10. Meijerink,J.A. andvan&r Vorst,H.A.: A. IterativeSol.tim Meth&
for Limar Systemsof Which the Cmtlicimt Matrix is s. Symmetric
M-Matrix,,. Mtifh. COW.. (1.977) 31, 148-62.
11. Kershaw, D. S.: .Thc [ncomplefe Cholesky-Conjugate Gradient Method
for the Iterative Solution of Systems of Linear Equations, J. Commit.
Phys. (1978) 26,43-65.
12. Jennings, R.R., Rogers, J.H., and West, T.J.: .$Fa.!ors Iutl.emig

MobilityCmtrol by PolymerSolutions,,>JPT (March 1971)391-401;

Trans., AIME, 251.
13. Dovm, H.T. and Hutchim, R.D,: kDevelopmmtof a New Alunimm/PolymerGelSystemfor PermeabilityAdjwrnem,, SSPERE (May
19S7) 177-83.

t4. Pmdhomme,R,K. et al.: .RheologicalMonitoringof the Formation

of PolyactylamidelCr3+Gels, SPEJ (Oct. 1983)8W-08.
15. Aslam, S., Vosso.gbi, S., and Wrllhite,G.P.: Viscom.trtc Mess
meme.t of ChromiumlU-PolyacrylarnideGelsby WeissenbergRheogoniom.ter,,, paper SPE 12639 presented at tie 1984 SPE/GGE
Enhanced Oil, Recovery Symposium, Tulsa, April [5- IS
16. Pmdhomme,R.K. andUtd, J.T.: Xinetiis of Polymer/Metal-IonGelation, paper SPE 12640presentedat the 1984SPEiD.OEEnhanced
Oil RecoverySymposium,Tulsa, April 15-18.
17. Southard,M,Z,, Green, D,W., ad Willhke, G.P,: Kineticsof the
ChromiumVtlThioureaReactionin fhePresenceof Polyacrylamidc,
DaDerSPE 127[5mese.ted at the 1984SPEIDOEEnhawedOil Recov~ti Symposium,?.1s., April 15-18,
18. Huang,C., Green, D.W..,ind Willhhe, G.P.: An ExperimentalStudy
of the In-sit Gelationof Chromium(+ 3)/PolyamylamidePolymeri.
Pomw Mti,,, , SPERE (No. 1986) 583-92.
19. Sortie, K.S., Roberts, L,J., and Clifford, P.J.: ..Cdculado.$ m the
Behavi.w of Tirne-SctdngpolymerGelsi PorousMedia,,, pqer 62.

presentedat the 1985 AIChE NatL Spring Meeting, Houston, March

~cr~d, C.F.: Aj@ed Numericcd Analysis, se&mdedhion, Addison-

Wesley PublishingCo. Inc., Reading, MA (1980) 15-20.

21. Clifford, P.J. and Sortie, K. S.: .Polymer Flooding in Stratifkd Systems Recnvery Mechanisms and the Effert of Chemical Degradation,
m EOR, Tmmibeim (Oct. 1984).
22. Vela, S., Peaceman, D, W., and Sm.ik, E. I.: .GEvalwuionof Polymer

dation,,, paperSPE5102 presented at the 1974 SPE Amu.1 Meet;ng,

Hcmstm. Ott. 6-9.
23. Dodge, B.F.: ChemicalEngineeringT&nnodynamc.t M.Graw-.Hill
Book Co., New York City (1944) 104-07.

S1Metric Conversion Factors

acre-ft x 1.233489
bbl x 1.589873
Cp x 1.0*
ft x 3,048*
ft3 X 2.g31 685
lbm X 4.535924
lbm/ft3 X 1.601846
tonne X 1.0*

1, Balycky, J.P., Maii, B.B., md Milmz, G.: CAStudy of the Application of PolymmicGelsin Porom Media,,, paperSPE 10620presented

at the 1982SPEItl, SyrnpmsiwncmO,ltieldmd GeothermalChemistry, DaUaS, Jan. 25-27.

2. Navradl, M., Sovak,M,, and Mitchell, M,S.: Dive~ng Agentsfor
Sweepknprovemmts i ,FloodingOperationsLaboratoryStudies,,,
paper SPE 10621
presented at the 1982SPE 1.0. Symposiumon Oilfield and GeothermalChemimy, Dallas, Jam 25-27,
3. Clifford, P.J, md Sorbic, K.S.; $TheEffects of ChemicalDegradstio on polymer Ftooding,,3paper SPE 13586presmted at the 1985
SPE ImL Syrnposiun m Oilfieldand Gco!herm?JChemistry,Phoe-.
nix, April 9-11.


fm.r !. e=+






Ho. June 11, 1986,Revised rnanus.xlplmcelwd Sepl, ?7, 1986,PaPer(SPE 14234)first
Presentedat the 1985SW AnnualTechnic.)Conference..d Exhwtim held,. Lasvwa,,
Sept. ,985.

SPE Rese&ok

En8i.neetig, November 1987