Journal of Petroleum Science and Engineering, 2 (1989) 103-117 103

Elsevier Science Publishers B.V., Amsterdam - - Printed in The Netherlands

STATUS REPORT ON MISCIBLE/IMMISCIBLE GAS FLOODING

P E T E R G.H. BATH, O.B.E.

Shell U.K. Exploration and Production, Strand, London WC2R ODX (U.K.)

(Received October 27, 1987; revised and accepted July 14, 1988)

Abstract

Bath, P.G.H., 1989. Status report on miscible/immisciblegas flooding. J. Pet. Sci. Eng., 2: 103-117.

The gases used for flooding a reservoir - hydrocarbons, carbon dioxide and nitrogen - are injected in conditions
varying from immiscibility to full miscibility. The high mobility of gas is always a problem, and difficulties are
exacerbated by miscibility: the maximum stable rate is reduced below the immiscible critical rate, and the integrity
of slugs can be difficult to maintain, especially in stratified systems. Water and gas are often injected in alternating
slugs, and useful work has been done recently in clarifying the influence of the parameters controlling the process.
Rock wettability appears to be important in flooding by carbon dioxide: there may be cases where single-slug C02 will
give better results than WAG. Research has identified some tracers for tracking carbon dioxide fronts, as well as
classes of chemicals useful as foam additives.
Carbon dioxide has useful properties, but disadvantages too. Nitrogen is worth considering as an alternative gas
for secondary and tertiary floods in light-oil reservoirs of good permeability. Full miscibility would occur only at high
reservoir pressure, but gravity drainage combined with vaporising enrichment of the gas stream may lead to useful
recovery at lower pressures. Three-phase relative permeability is a controlling parameter in tertiary flooding, and
recent experiments suggest that it may be more favourable in some cases than is predicted by existing methods.

Summary low oil-saturation in those parts of the reser-

voir which the gas contacts; something less than
A variety of gases can be used for flooding oil miscibility may, however, be acceptable, if a
reservoirs, each having its advantages and dis- useful a m o u n t of oil can be recovered at a rea-
advantages in particular circumstances. Hy- sonable cost. The fist part of this paper deals
drocarbon gas has the benefit of occurring nat- with the basic ideas of miscibility, dynamic
urally in some proportion in reservoir oil, and miscibility and near miscibility.
hence of not contaminating the reservoir when The chapter on "Phase behaviour of carbon
it is injected. Hydrocarbon gases range from dioxide and nitrogen" outlines briefly some of
pure m e t h a n e through lean gas (separator gas ) the peculiarities of the phase behaviour of car-
and enriched gas to liquified gases such as pro- bon dioxide and nitrogen, two gases which are
pane or LPG. These can be miscible, or will de- also of considerable interest in the flooding of
velop miscibility, at corresponding m i n i m u m reservoirs. Carbon dioxide is a very versatile gas,
pressures ranging from some 6000 to 1200 psig; but is corrosive and is not always readily avail-
the behaviour depending also on the reservoir able (e.g. in the North Sea); its density exceeds
oil composition and the temperature. Miscibil- that of the oil in some reservoirs. Nitrogen be-
ity is attractive, because it leaves only a very comes miscible only at high pressures, often in

0920-4105/89/$03.50 © 1989 Elsevier Science Publishers B.V.

104

excess of 6000 psig; on the other hand, its den- grams (Figs. 1 and 2). Such a diagram is
sity is lower than that of reservoir oil, and it is incapable of showing all the mixtures which
universally available in the air, at a separation may occur in a producing reservoir, neither can
cost of less than $1 per mcf. The last chapter it represent the compositions possible in a rich
summarises its possible use in immiscible sec- injection gas. It is useful for displaying the re-
ondary and tertiary flooding. Three-phase rel- sults of experiments or detailed calculations,
ative permeability to oil is a key parameter in but is not a substitute for them.
the latter studies: recent research suggests that Experiment shows that certain mixtures,
existing interpolation methods of estimation such as composition C in Fig. 1, are unstable
may give values that are too low. and split spontaneously into a gas phase, A, and
The stability of a segregated flood front de- an oil phase, B. An interfacial tension exists be-
pends on the fluid mobility ratio and on any tween these phases, causing oil to be left behind
gravity effects. Miscibility increases the mobil- when gas A displaces oil B. The two-phase re-
ity ratio, and adds greatly to the complexity of gion - i.e. the set of all compositions which be-
the process. Mixtures occur at the flood front have like C - is bounded by a curve such as that
in all proportions, and the stable rate is reduced
to below the ordinary critical rate (if M > 1). ~'
There are also increased problems with strati- ~ k
,MM, SC,BLE
GAS
fled systems. The alternating injection of water ~OMPOS,T,ONS
and gas (WAG) is often used to reduce the dis- ~)
placement mobility, and an interesting model . l/k-~
DEVELOPED-
study has investigated the effects of varying the I/ \ ~. MISCIB,LITY
..~c, ~G2~.(CONDENSING
parameters in CO2 injection (WAG ratio, cycle F ~\GAS DR,VEI
size and slug size). Some laboratory studies
have suggested that WAG may recover less oil ..... ~-.--~-~FIRST-CONTACT
than single-slug CO2 injection, particularly in RESO,L ~,SC,B,L,r~
water-wet reservoirs. Rock wettability should / /
evidently be reported in field studies dealing °7. \)c2-6
with CO2 injection. Research into foam addi-
tives for mobility control is also active, and cer- Fig. 1. Pseudo-ternary phase diagram, showing condensing
gas drive.
tain classes of chemicals have been shown to
have promise. The movement of carbon dioxide cL
in a reservoir can be followed by halogen tracers. / A\\
k ~ 0EVELOPED-
\ M~SC~B,UT¥
Carbon dioxide should not be thought of in / \\ \IVAPOR,S,NG
terms of miscibility alone. Its high solubility and / ~ V~E ~
beneficial effects make it attractive in heavy oil, ~
where WAG is also possible. It is effective in
encouraging gravity drainage in secondary gas
caps, and is useful for enhanced oil recovery in / E,RST-CONTACT
single-well reservoirs.

Types of gas flooding ,L

The different types of miscible and immis- c ~_~

cible flooding by hydrocarbon gas are com- Fig. 2. Pseudo-ternaryphase diagram, showing vaporising
monly illustrated on pseudo-ternary phase dia- gas drive.
 105

shown schematically (and exaggerated in size ) able, so the slug is diluted from both ends.
in Fig. 1. The tangent to this curve which passes Developed miscibility (vaporising gas drive).
through the composition of the reservoir oil, and If the reservoir oil is rich in intermediates (C2-
the tangent at the plait point (the limiting tie- C6), its composition may lie to the right of the
line ), define three ranges of gas richness. These limiting tie-line (Fig. 2 ). Oil gravity must be
lead to the well-known types of gas flooding, about 40 ° API or higher for this to occur at an
with which this paper is concerned: acceptable reservoir pressure. If methane (or
Immiscible gas flooding. Composition G1 lies lean gas) is injected into the reservoir, it will
on tie line ON extrapolated. If fresh gas of this displace some of the oil, and vaporise interme-
composition contacts reservoir oil repeatedly, diates from the remaining oil; the latter will be-
as would be the case around an injection well, come more viscous and relatively immobile,
it will enrich the oil to composition O, but no while the enriched gas stream will contact fresh
further. The minimum i.f.t, between gas and oil oil and repeat the process. Miscibility will soon
phases will be that of the tie-line ON. The closer be achieved, within a few feet at most. The ad-
the tie-line is to the plait point, the less the oil vantage of this process is that there is no slug,
that will be left behind a flood front. Even in no distinction between miscible fluid and driver
unfavourable cases, gravity drainage may act to gas: lean gas does both jobs. Given sufficient
increase recovery. (Figs. 11, 12). dip to stabilise the front, the method is useful
Developed miscibility (condensing gas in reservoirs developed on wide spacing.
drive ). If the gas composition is on the side of P h a s e b e h a v i o u r of c a r b o n d i o x i d e a n d
the limiting tie-line opposite to that of the res- nitrogen
ervoir oil (G2, for example), enrichment of the
oil proceeds by repeated contact with fresh in- These two gases are often used for injection.
jection gas until the line joining the oil com- Their phase behaviour in hydrocarbon mix-
position to G2 lies wholly outside the two-phase tures has certain peculiarities, which we sum-
region. Mixtures of this oil with the injection marise in this chapter.
gas in all proportions remain as a stable, single
phase. This is the definition of miscibility, and Carbon dioxide
the gas will displace the oil relatively com- Figure 3 is a phase diagram representative of
pletely. Rich gas is usually injected as a slug of the behaviour of C02 at temperatures below
the order of 15% h-c p.v., followed by cheaper about 120°F. It shows what happens when CO2
lean gas as a driver. Various processes in the comes into equilibrium with reservoir oil, in
reservoir, to be discussed, act against the integ- various proportions; it is often known as a P -
rity of the slug and may cause loss of miscibility. X diagram.
First-contact miscibility. A gas of composi- The first thing to be noted is that two phases
tion such as G~ is, according to the definition are present for CO2 concentrations above about
above, directly miscible with the reservoir oil. 65 mol%, at least for pressures up to 4000 psia.
A first-contact miscible slug is usually LPG, CO2 is not in fact miscible on first contact with
propane or butane. It must be miscible with ordinary reservoir oils at any pressure nor-
both the reservoir oil and the dry-gas driver: mally encountered in reservoirs. (The P - X
the latter condition, usually the more severe, diagram as shown describes first-contact be-
requires a minimum reservoir pressure of only haviour only. )
about 1200 psig. This makes LPG-type floods It will also be noted that the two phases pres-
attractive in shallow reservoirs. They are ex- ent at higher pressures are not vapour and liq-
pensive, however, and the slug seldom exceeds uid, but two liquids of different density. If we
10% h-c p.v. The mobility ratios are unfavour- allow the pressure to decrease, at a CO2 concen-
106

4ooo- Nitrogen
5600- SINGLE LIQUID
PHASE n Figure 4 is a partial P - X diagram showing
3200- ..~
REGION ov
UPPER LIQUID (UL}
the first-contact behaviour of nitrogen with a
AND reservoir oil of 38°API. It is evident that ex-
LOWER LIQUID (LL)
2800- oo tremely high pressures would be needed for
~ first-contact miscibility. The pseudo-ternary
2400- ~/ diagrams of Fig. 5 (for a different oil), show,
2ooo- however, that miscibility may be developed at
pressures in excess of 6000 psia. This still means
1600-
~ ~ ~ O O / o v
[...REGION OF
that full miscibility is not likely to develop in
)-TUL, LL, AND GAS many reservoirs; there will nevertheless be some
J2oo- 1-oo/° UL benfit to be had from the enrichment that re-
90 80 70 60 50 40 30 20 I0 ~'REGION OF suits from vaporisation of intermediates into
800- LL AND GAS
VOLUME PERCENT the flowing gas stream at pressures near the
LOWER LIQUID PHASE M.M.P. It may be noted that the vaporisation
400 ~:o ,,:o ~o 8'o .~o process does not normally lead to the complete
MOLE PERCENT CO2 A D D E D replacement of nitrogen. If it did, the miscibil-

Fig. 3. Phase diagram for wasson crude at 105°F: pressure

vs. concentration of CO2 (from Gardner et al. [20] ). 15000 -

14000

13000
tration of say 80 mol%, a vapour phase joins the
two liquid phases when the pressure reaches ,2ooo
about 1200 psia. At pressures below about 1000 .ooo
psig, only two phases are found: a vapour and a ,000o
liquid. If the temperature exceeds 120°F, a va-
pour and a single liquid are found throughout _ 9000
u
the multi-phase region. C O 2 c a n also cause as- ~ 8000
phaltenes to precipitate as a solid phase: this
may lead to loss of permeability in a reservoir. ~ ~ooo
C O 2 behaves as both a vaporising and a con- ~ 6000
densing gas. It can vaporise hydrocarbons up
to about C3o: the range which is important in 5000
the generation of miscibility is C 5 - C 1 2 , and 4000
miscibility can be developed in oils of gravity
down to about 26°API. M i n i m u m miscibility 3000
pressure (M.M.P.) ranges from some 1200 psig 200o
for light oils at low temperatures to over 4500
psig for heavy crudes at high temperatures. The 1ooo
presence of methane or nitrogen in carbon o
,o ~ io ;o 5'0 4 7'o ~'o 9'0 ,~o
dioxide may increase the M.M.P. substantially. .2 MOL % IN MIXTURE
Carbon dioxide is also soluble in reservoir oils, Fig. 4. Measured saturation pressure for nitrogen and
where it causes oil swelling and viscosity statfjord reservoir fluid (provided by Shell Group Lab.,
reduction. Rijswijk).
 107
N2
a A _ methane, for a given pressure. It is also noted
90 "~ that the presence of methane in the reservoir
oil improves miscibility with nitrogen: meth-
ane behaves somewhat like an intermediate in
its effect on the miscibility of nitrogen.
The behaviour of nitrogen resembles that of
40 carbon dioxide in that neither is (ordinarily)
~o first-contact miscible, but both can develop
~ / ~ miscibility at suitable pressures. Nitrogen re-
,o/ _. quires high pressures, however. It has low sol-
/ "~;; ubility in oil, and so has little effect on its prop-
c2. ~ ~ ~ ~ ~ ~ -~ ~o ~ c, erties. It has low solubility in water too, which
may have some bearing on its use in tertiary
b N2 flooding.

~i:i/.~~ A@
90/f/~ ~ ~
8o///\ ~-
W
I CRITICALPOINT
El RESERVOIRFLUID Estimating t h e m i n i m u m p r e s s u r e f o r
dynamic miscibility
Gravity-stabledisplacement

An injection gas at reservoir pressure may

have a viscosity of the order of one-twentieth
that of reservoir oil. The mobility ratio will be
2 o9 unfavourable, and gas will finger its way
7 ~ES.~LU,O~'~~-__-=_~ ~ through the oil;a problem to which we shall
c2+ ~ ~ ~ ~ ~ ~ ~ ~ ~ c, return in the next chapter. An attempt to esti-
mate M.M.P. by displacing oil out of a core is
Fig. 5. Pseudo-ternary phase diagram with nitrogen, show- likely to be confused by the effects of viscous
ing effect of pressure (provided by Shell Group Lab., fingering. The problem may be mitigated by ar-
Rijswijk);a: nitrogen/hydrocarbonphasebehaviourat6 0 0 0 ranging the core vertically, so that the density
psia; b: nitrogen/hydrocarbon phase behaviour at 4550 psia. difference between gas and oil stabilises the
front. The condition for stability (see next
ity question would be reduced to one of hydro- Chapter) may, however, require inconve-
carbon miscibility. The equilibrium composi- niently low rates of displacement. The slim-
tion of the flowing gas stream does, however, tube has been devised as a way of getting round
contain some nitrogen, the problem.
A ternary phase diagram does not give a very
complete picture of the behaviour of mixtures The slim-tube method
of nitrogen and hydrocarbons. Stalkup [ 3 ] uses
a pseudo-quaternary diagram (tetrahedron) to The slim tube is a means of forcing the flow
illustrate conceptually the separate parts played to be effectively one-dimensional. Viscous fin-
by methane and the intermediate hydrocar- gers will be suppressed by transverse disper-
bons. The two-phase region on the Nz/Cz_6/C7+ sion if the porous medium is small enough in
face is larger than the two-phase region on the diameter and the displacement rate is low
Cz/C2_6/C;+ face. That is, an oil must be sub- enough. Once again, this rate may be too low to
stantially richer in intermediates in order to be be acceptable. The slim tube is therefore made
miscible with nitrogen, than is the case with long, so that any fingering is confined to a small
108

proportion of the tube length. Typical dimen- effect. Figure 7 shows the recovery/pressure
siGns are a diameter of 1/4 inch and a length of points at 0.85 pore volumes injected; the lines
40 ft; the tube is usually coiled into a helix, and intersect at a M.M.P. of about 4800 psia.
is packed with sand. A rapid method of measuring the M.M.P.,
The experiment consists of saturating the based on the disappearance of a gas bubble ris-
tube with reservoir oil, and then displacing it ing through an oil column, has been described
at a trial pressure by injecting the appropriate by Marathon Oil Co. [24].
gas; oil recovery is recorded at various amounts
of gas injected (expressed in pore volumes ). The Windowed cells: M.M.P. and tie lines
procedure is then repeated at different pres-
sures, to generate a set of recovery curves (Fig. Phase changes can be analysed directly in a
6 ). The highest pressure should give a recovery windowed cell (e.g. PVT cell). The procedure
of about 95% at 1.2 pore volumes injected, is to mix some reservoir oil with injection gas,
The M.M.P. can now be estimated by plot- and let it separate into equilibrium oil and gas
ting recovery fraction against pressure, at a phases. Samples can then be extracted, for
fixed number of pore volumes injected. This measurement of the composition, density, vis-
fixed number is not indicated unequivocally, cosity and interfacial tension corresponding to
One could choose 1.2 pore volumes, since this this tie line. The vapour phase can then be
gives roughly the ultim&te recovery at pres- mixed with fresh reservoir oil (for a vaporising
sures above the M.M.P. One could argue, alter- gas drive), or the oil phase mixed with more
natively, that the number of pore volumes injection gas (for a condensing gas drive), to
should be defined by where gas breaks through give new equilibrium oil and gas phases and a
on the curves above M.M.P.: beyond this point, new tie line. The limiting tie line can be ap-
factors other than miscibility begin to have an proached by a series of such steps.

1.0~

~ ~ . -= =__
G ........ ~ P = 5000 PSIA
® .............. ® P = 4500 PSIA .. . . . . . . G--.

/-
• • P = 4100 PSIA
0-8- # ~ P = 4100 PSIA (ESTIMATE) "~' .....
E" 0 .''''~"

b []
~
.~ 0.6- ~/°~
z
w
o
er

~ 0-4-

8w
n,-

0-2-

0.0 i
0"0 0'2 014 016 0"8 I"0 112 I J4 11"6
PORE VOLUMES INJECTED

Fig. 6. Oil production curves measured in slim-tube apparatus (provided by Shell Group Lab., Rijswijk).
 109

• 0"65MPV INJECTED M o b i l i t y a n d g r a v i t y e f f e c t s on
~.o - displacement

• . Mobility
0.8- ,'
_ /

/" Darcy's Law for the horizontal, linear flow of

" an incompressible fluid is:
o6- ='
/ kdp
= ,udx
0.4-

The term k / p , which plays a part in this for-

mula analogous to conductance in Ohm's Law,
w
= o2- is called mobility. The manner in which one
fluid displaces another depends on the ratio of
their mobilities, M:
O0 i i i I
3000.0 40000 5000-0 6000.0 70000
PRESSURE (psio) M - kg/flg
Fig. 7. Minimum miscibility pressure (provided by Shell k°/~t°
group Lab., Rijswijk). where suffix g denotes the displacing gas, and
suffix o denotes the reservoir oil. End-point
Finite steps of this kind may not correspond mobilities are used for segregated flow; gas mo-
to the result of continuous interaction in the bility is replaced by the sum of gas and oil mo-
reservoir; but the results can be used to cali- bilities evaluated at the Buckley-Leverett
brate an equation-of-state model, which will shock-front saturations, if the flow is diffuse.
predict the reservoir processes. This procedure If M < 1, the oil is capable of moving faster
provides the input needed by a compositional than the gas, for a given pressure differential;
reservoir simulator, and so may be more useful fingering will be suppressed, and a sharp flood
than the simple estimation of M.M.P. front is possible. This gives a favourable dis-
placement. If M > 1, fingers will develop and
Continuous multiple-contact measurements propagate, leading to early gas breakthrough.

The time required to measure phase compo- Effects of miscibility~immiscibility

sitions can be reduced substantially by per-
forming a continuous experiment, rather than If the fluids are immiscible, a favourable mo-
sampling static equilibrium mixtures. Appara- bility ratio will actually lead to a sharp flood
tus designed for this purpose is described in front. If the fluids are miscible there is no i.f.t.
[18]. Liquid production is metered from the to maintain their identity, and the front will
bottom of the cell, at a rate less than the CO2 become blurred even for M < 1. In practice,
injection rate; fluid is produced from the top of however, the mobility ratio is usually unfa-
the cell under the control of a back-pressure vourable when gas displaces oil; viscous fingers
regulator, whose function is to maintain the will develop, exacerbated by the effects of mis-
system at constant pressure. It would seem to cibility. This acts to destroy the integrity of the
be possible to mount high-pressure densitom- slug in a rich-gas or LPG type of flood, and the
eters ahead of the controllers. The apparatus effect is compounded by a similar process be-
as described does not, however, provide infor- tween lean driver gas and slug.
mation on phase volumes. There is, as we have seen, no sharp division
110

in mass transfer effects between miscibility and tion is towards a neighbouring mesh point, but
immiscibility: components will usually be ex- P fails to reach it, a Procrustean solution is em-
changed across the flood front whether misci- ployed. The result is similar to diffusion. If the
bility can be attained or not. An immiscible direction is not towards a mesh point, a lateral
flood leaves an oil saturation behind; this in- distortion is applied, and we have something
terferes with the flow of gas, so reducing the like dispersion. The method of characteristics
effective permeability and the mobility ratio is well adapted to handling saturation fronts,
somewhat. A miscible flood leaves an oil satu- but appears to be too complex for practical use.
ration of only a few percent, and so has an in- A novel scheme, employing moving point
herently less favourable mobility ratio (viz. ap- methods within a fixed finite element mesh, is
proximately the ratio of the viscosities), described in [23 ].

Influence of dispersion on miscible Gravity stabilisation

displacement
If gas is injected at the top of a dipping for-
Microscopic mixing occurs at the interface mation (Fig. 8), the tendency of gravity to
between fluids, through both molecular diffu- maintain the gas and oil segregated will help to
sion and convective dispersion. The latter is counteract the effect of an unfavourable mo-
peculiar to flow through porous media, being bility ratio. Gas fingers will be suppressed pro-
apparently caused by the convergence of dis- vided the injection rate is less than a certain
tinct stream-tubes on to a single pore space, critical rate. (The orientation of Fig. 8 assumes
Dispersion has components both in the direc- that gas is less dense at reservoir pressure than
tion of flow and at right angles to it. Transverse the reservoir oil. Carbon dioxide can be denser
dispersion is of more importance in unstable than reservoir oil, however, and it would be
displacements, where viscous fingers expose a necessary to inject it at the bottom of the
large surface area to the process; longitudinal reservoir.)
dispersion is more important in gravity-stable The ratio of gravity to viscous force is known
flooding, where fingering is suppressed, as the dimensionless gravity number, G:
Dispersion eliminates frontal perturbations Apg sin 0
below a certain critical wave length but allows G = kg (1)
larger ones to grow into fingers: this may sta- U/lg
bilise flooding on the scale of a laboratory core, The inclination of the oil/water interface is:
but cannot prevent fingering on the reservoir ~ G-M+I
scale. On the other hand, the dispersive mixing tan fl= -~ tan 0 (2)
of fluids and the coalescence of small fingers
into large ones reduces the viscosity contrast
and the effective mobility ratio. Modified black-
oil simulators represent this by employing a
mixing parameter to calculate the mobility ra-
tio in the transition zone.
Analogues of diffusion and dispersion are
created by numerical models which discretise
what is effectively a continuum. A point P,
which coincides with one of the mesh points at
time zero, is moved vectorially by the equations
of motion during a small time step. If the direc- Fig. 8. Gravity-stable displacement.
 111

Stable displacement requires fl> 0 and be stable if the injection velocity can be simul-
therefore: taneously greater t h a n the critical velocity for
G> M-1 the water/slug interface and less t h a n the sta-
ble velocity for the slug/reservoir oil interface.
The critical velocity is the value of u which Even then, it will be necessary to check that
makes: temporary shut-downs of injection will not al-
G= M- 1 low significant redistribution of the water and
slug.
Thus: Methods of mobility control in the absence
1 k~ of useful gravity stabilisation are discussed in
Ucrit -- M - 1 -~g-~Pgsin 0 (3) the next chapter.

(N.B. If Darcy units are employed - cm sec -1,

Darcy, cp, gm c c - 1, atm - the denominator re- Effects of mobility in stratified systems under
quires a factor 1.0133 E + 06. ) The interaction miscibility
of M and G in segregated, unstable flow
(G < M - 1 ) is illustrated [ 2 ] by two points on
the recovery curve: breakthrough occurs after Mobility between miscible fluids reduces ef-
the injection of [1~(M-G)]MOV; the mova- fectively to the viscosity ratio (c.f. previous
ble oil is recovered after the injection of [M/ section "Effects of miscibility/immiscibil-
(G + 1 ) ]MOV (MOV= movable-oil-volumes ). ity" ). Miscibility means, also, that there are no
Equation 3 uses the distinct properties of gas capillary effects between layers: exchange of
and oil and so applies strictly to segregated dis- fluids is by density difference, dispersion, and
placement. Stable displacement by a miscible (if M S 1 ) by pressure difference.
fluid requires the critical condition to be met If M= 1, frontal advance in each layer is gov-
for mixtures in all proportions. Stalkup [3 ] uses erned by its conductance ( k / 0 ) . A favourable
blending rules for viscosity and density to show mobility ratio tends to bring the fronts to-
that the stable rate for miscible flooding, ust, is gether; M > 1 acts to separate t h e m even fur-
related to the immiscible critical rate by: ther and to cause viscous fingering (Fig. 9; [4] ).
1 Cross flow (for M ¢ 1 ) will further modify the
1--- fronts. These effects are particularly serious in
Ust M
Ucrit-4 (M 1/4-1 ) (4) slug-type projects. So long as the slugs in two
layers overlap, they protect one another. As
This expressison tends to 1 as M-~I, and has, soon as they separate, they are liable to attack
for example, a value of 0.4 for M - - 5 . The cor- from all sides (Fig. 10; [4] ). It is therefore im-
rection may not be important in vaporising gas portant to try to reduce an unfavourable mo-
drives, but should be taken into account in slug- bility ratio. Methods are discussed in the next
type processes, chapter; it is unfortunate, however, t h a t w a t e r /
It may be noted t h a t water (possibly visco- gas mixtures enter stratified systems in differ-
sifted with polymer) can be injected at the top ent proportions.
of a dipping reservoir, to displace a miscible
slug. Equation l ( w i t h d a t a f o r w a t e r / s l u g ) t h e n M >t °l <t k/~
yields a negative value for G, but the stability d ~ I i Low
criterion is unchanged: V > M - 1. This can be rI i I ~ H,G,
satisfied if M < 1; the critical velocity given by M <, :~ >
Eq. 3 is now a minimum. The whole system will Fig. 9. Effect of mobility ration on frontal advance.
112

d ~////~ I representative of recent C O 2 field projects. A

"[ ~J/////~JJ/~ [ model comparison of two Amoco CO2 projects,
one with a 60 ° dip and the other with low-
permeability layers restricting vertical perme-
J ~Jf~ ] ability, shows that a 1:2 WAG ratio would give
"1 ~/~////JJ//~ I the highest production peaks [6]. A 1:1 ratio
Fig. 10. Slug exposed to degradation, would give greater ultimate recovery, and this
was selected for both reservoirs. WAG ratios in
Mobility ratio when mobile water is present hydrocarbon miscible floods are generally sim-
ilar: a figure as low as 1:4 is given for the Lev-
Water may be injected together with gas (see elland Unit [3], while 5:1 was chosen in Pru-
next chapter); mobile water and oil will be dhoe Bay [7].
present together in tertiary gas flooding. The W h e n water and gas are injected into a strat-
mobility ratio between an oil bank and the dis- ified system, water (responding to capillary
placing fluids may then be taken as: pressure) tends to flow preferentially into sands
of low permeability, leaving a low WAG ratio in
M - kg//tg + kw/p~ the high permeability sands. Depending on the
ko//to + kw //tw relative importance of recovery from the differ-
the mobilities being evaluated at the appropri- ent layers, it may be advisable to increase the
ate average saturations [3]. ratio of water injected to gas.

Mobility improvement Cycle size

The theory used for calculating the WAG ra-
Alternating injection of water and gas (WAG) tio assumes that the gas and water are flowing
together. T h e y are nevertheless injected alter-
This technique uses the interference be- nately; partly for operational convenience and
tween gas and water to decrease the relative partly because a mixture of fluids near the well-
permeability of reservoir rock to gas, and thus bore would reduce injectivity. Gas is, of course,
to reduce the mobility of the gas. There are sev- unstable when displacing water horizontally; so
eral factors to be considered in the design of a the alternate batches of gas and water will soon
WAG flood, lose their separate identity in the reservoir.
Smaller, more frequent gas/water cycles
WAG ratio control the mobility of CO2 better, according to
This governs the relative speeds at which Amoco, but at the cost of an operating night-
water and gas flow in the reservoir. If water mare [6]. Their model showed that 1.05% h-c
flows faster it may trap residual oil which the p.v. cycles would recover only 0.4% more orig-
gas may not be able to release; if the miscible inal oil t h a n 4.2% cycles. Chevron and Home
gas flows faster, the front will finger into the oil Oil have miscible hydrocarbon projects
and be diluted, which may lead to loss of integ- (WAG = 1 ) in which injection is switched at in-
rity in a slug process. The two fluids should tervals of two months [8,9].
therefore be injected in such a ratio as will en- The cycle size does not seem to be very im-
able them to travel at (roughly) equal speeds, portant to recovery, but Unocal Corp. report
Stalkup gives a method for estimating this ra- laboratory studies indicating that CO2 WAG
tio [3]; and calculates as an example a reduc- recovers less oil t h a n single-slug CO2 injection
tion in displacement mobility by a factor of 50. in some reservoirs [10]. WAG may cause
Robl et al. [5 ] quotes a WAG ratio of 1.5:1 as greater trapping of oil if the rock is water-wet
 113

or of mixed wettability. They cite a number of oxide and propylene oxide. The report says that
field studies in support: the WAG cases need experiments on gas mobility, conducted in lin-
checking to see whether the water/gas ratio was ear sand-pack models, show only a general cor-
correct for the fluids to travel to equal speeds, relation with the static foam test. This is per-
It is evidently important that field studies haps not unexpected, since some doubt exists
should always include information on rock as to whether foams are capable of existing in
wettability, so that its effect on performance porous media unless the permeability is very
can be assessed, high.
The direct thickening of carbon dioxide
Slug size would have some advantages over methods re-
The amount of oil recovered will usually in- quiring water injection: corrosion would be re-
crease with the size of the slug, but the return duced, and oil trapping by mobile water avoided.
becomes progressively less. A CO2 slug repre- Research into the synthesis and properties of
sentative of recent field projects is about 40% polymers of alpha-olefins in the C~-C12 range
h-cp.v. [5]: 12-14% is q u o t e d f o r t h e SACROC is reported in [12]. An alternative technique
unit (Kelly-Snyder Field; [3] ), while Amoco under investigation is to use monomers misci-
have picked a slug size of 60% [6]. This is a ble with carbon dioxide, and polymerise them
parameter which is to be optimised economi- in situ while the carbon dioxide is supercritical
cally for each set of circumstances, but the trend [ 21 ].
seems to be towards larger slugs.
Miscible hydrocarbon slugs are, by contrast,
somewhat smaller. Stalkup gives figures mostly Chemical tracers for carbon dioxide
in the range 2-12% h-c p.v. for first-contact
miscible projects (secondary and tertiary), and The movement of carbon dioxide in a misci-
2-15% for condensing gas drive projects [3]. ble flood can be monitored by certain halogen
tracers [13]: halocarbons 11,12 and sulphur
Foam additives; direct thickening hexafluoride. Tracer adsorption on rock sur-
faces appears to be minimal. Detection limits
Sweep efficiency is controlled not only by the
are as low as 0.5 ppb.
mobility ratio, but also by the variation in
permeability. Assuming no communication be-
tween layers, the latter can be measured by the G e n e r a l u s e s of CO2
Dykstra-Parsons co-efficient Ik, which varies
from 0 for a homogeneous reservoir to 1.0 for
infinite heterogeneity. The 1984 National Pe- Carbon dioxide is a versatile injection gas,
troleum Council Paper [5] considers that mis- with a number of useful properties not depen-
cible CO2 flooding requires foam technology if dent on the achievement of miscibility. Screen-
Ik > 0.7. Dense CO2 and surfactant solution are ing guides for miscible CO2 injection recom-
injected simultaneously to create a foam. mend that reservoir oil should have a gravity of
A report on CO2 flooding by the U.S. DOE at least 26°API. Immiscible CO2 is useful for
[ 11 ] recommends three basic chemical struc- causing oil swelling (of the order of 10% ) and
tures as showing the best promise for control of viscosity reduction in heavy oils. Mungan
gas mobility: ethoxylated Cs-C14 linear alco- quotes an example of crude oil equilibrated with
hols, sulphate esters of ethoxylated C9-C16 lin- CO2 at 2000 psia [ 17 ]: the viscosity of a 5 cp oil
ear alcohols (possibly limited to 130 ° F), and is reduced 10-fold, while that of a 1000 cp oil is
low molecular weight copolymers of ethylene reduced 50-fold.
114

Gravity drainage ervoir oil: this might leave an asphaltene resi-

due, which could plug the reservoir. The ex-
CO2 can be injected immiscibly into a sec- traction process seems to be happening, since
ondary gas cap to encourage gravity drainage, oil is being produced at 24-25 ° API. There have
The CO2 causes the residual oil to swell, and been no injection problems, however.
also reduces its viscosity. As an example, Mar- The second problem is the presence of a free
athon Oil Co. are injecting CO2 into a second- gas saturation; this could provide a channel,
ary gas cap in the east side of Yates Field (Pe- enabling the CO2 to bypass the oil. This ap-
cos County) [8]. Crude gravity is 30°API and pears not to be happening, since reservoir per-
reservoir pressure about 750 psig. formance is in line with predictions. A recovery
of 35% of original oil is expected, using a CO2
Small [auk-block (single-welD reservoirs slug of 15% h-c p.v.
This particular field might not respond so
EOR processes often call for separate injec- well to the WAG technique, since it contains
tion and production wells, but small reservoirs hydrophilic clays. Nevertheless, a paper [ 16 ]
may not be economically capable of supporting on the recovery of heavy oil by CO2 injection
multiple wells. Cyclic COe injection (huff 'n confirms that immiscible CO2 can be effective
puff) is a possible technique in such cases, even in the WAG process: it improves the waterflood
for light oils under strong water drive, efficiency by increasing the mobility of the
Amoco report substantial success in 25 field heavy oil. WAG injection is better than the old
projects, both on and offshore Louisiana, and carbonated water process, because slugs of CO2
backed by laboratory rechearch [14]. Reser- make direct contact with the oil.
voir depths ranged from 1200 to 13,140 ft, and The first test of immiscible CO2 tertiary oil
oil gravity from 23 to 38°API. An average of 9 recovery in a late-life waterflood reservoir has
mmcf CO2 was injected per cycle in 24-48 h; been going on in Wilmington field since 1981
wells were then soaked for from two to eight [22]. Oil in the pilot area is 14°API; the pro-
weeks. First-cycle production lasted for a year ducing wells had been near the economic limit
on average, starting at about four times the pre- before the test, but have subsequently shown a
vious rate. Oil swelling and viscosity reduction seven-fold increase in rate.
are mentioned as likely mechanisms, together I m m i s c i b l e gas f l o o d i n g
with hydrocarbon extraction and relative-
permeability effects. Attic-oil recovery would Miscible flooding can lead to high recoveries,
presumably be possible also, in cases where and will commonly be chosen provided the mo-
wells were not located optimally, bility problem can be handled. A dipping res-
ervoir, for example, may stabilise the front at a
Heavy-oil C02 flood reasonable rate. There are often cases, how-
ever, where a miscible gas is unavailable or
Westgrowth Petroleum Ltd. are injecting CO2 would be too expensive. The purpose of this
successfully in Retlaw Field (Alberta), where chapter is to consider the alternative of gravity
oil gravity is 23 ° API [15]. Minimum miscibil- drainage.
ity pressure is well above fracture pressure, so
Choice of injection gas
the flood is immiscible. Interest attaches to this
flood because of two possible problems: The properties of hydrocarbon gas make it
CO2, although immiscible, should extract particularly attractive for injection: it does not
some of the lighter hydrocarbons from the res- contaminate the reservoir, is non-corrosive, and
 115

will usually benefit from significant mass ex- Reservoir behaviour in secondary flooding
change with the reservoir oil. The problem is
that there is a ready market for the sale of hy- Figure 11 illustrates the process of gravity
drocarbon gas; it seems unlikely, therefore, that drainage in secondary immiscible flooding. The
it would be economic to develop a gas reservoir gas/oil contact moves down-dip as oil is pro-
for the express purpose of injectingthe gas back duced, leaving some oil behind in the gas-in-
underground and leaving it there for several vaded zone. This oil drains downwards slowly,
years. It might, however, be an economic prop- until it reaches an impermeable layer: it then
osition to buy hydrocarbon gas from a source flows down-dip in a segregated layer to join the
developed primarily for the general market. If descending oil column. In the gas-invaded re-
hydrocarbon gas is not available in the quan- gion the flow of oil is dominated by gravity: gas
tity required, for whatever reason, a useful al- is injected to replace the voidage created by oil
ternative is nitrogen, and it is this gas that we production down-dip.
shall be especially considering in this chapter. Gravity drainage under gas injection has few
The phase behaviour of nitrogen is summarised constraints. The rock is assumed to be water-
in the chapter on "Phase behaviour of carbon wet and connate-water bearing. The amount of
dioxide and nitrogen", and illustrated in Fig. 5. oil left behind the gas/oil contact depends pri-
The process requires high reservoir permeabil- marily on the relative permeability to oil, and
ity and light oil. the vertical permeability; the drainage rate of
oil depends on its viscosity also. Thus for good

~ X~ ~jy

Um
mOX X : k h

mo ,y "
&pg sin a

cos Q

ma,y

owe
WHERE
Umax is maximum Dorcy gravity
drainage rate
J ']GAS-INVADED Uo is actual Dorcy
displacement rote
~OIL - SATURATED
VGOC,x is velocity of GOC
along dip

Fig. 11. Schematic cross section of draining reservoir with relevant flow characteristics (from Bath [ 1 ] ).
116

recovery in a reasonable time we need high oil ~ OCAAPS A

mobility and high vertical permeability. One /'///'~/sEco,~O~T~E
further factor to be considered is the gas/oil U,RODU0EDD,L L A Y E R ~ ~ : : J "
capillary pressure. The thickness of an individ- ° °°°°.° °°
a © o o o GAS INJECTION
WATER AND o o°o°o°o °

tion-zone height by a factor of ten at least. ° ....

A heterogeneous reservoir will require nu- °°=°° TER~,~R~A~E
merical simulation, and care is then needed with ~ °,° o o /
the vertical grid spacing. The segregated stream
of oil flowing down-dip (which may occur on ~'~TE~R~N~'
intermediate shales of limited extent, as well as ~'::~
at the base of the sand) may be quite thin. It 7 -
should ideally be represented by a grid layer of
comparable thickness. If the grid layer is much Fig. 12. Schematicrepresentation of tertiary gas displace-
ment (from Bath [1 ] ).
thicker, the segregated oil will be converted into
a low oil saturation spread over the thickness placement rate must be limited to allow gravity
of the layer: this must then move down-dip with to stabilise the interface. A potentially serious
a low oil relative permeability, and recovery will problem is that nitrogen may, in some circum-
be under-estimated. If there are a number of stances, replace oil as the spreading phase.
intermediate shales at various heights, it will
not be feasible to grid correctly. Simple homo-
Three-phase relative permeability
geneous models (which may be analytical) can
then be used to generate pseudo-relative
permeability and pseudo-capillary pressure Experiment has shown that three-phase oil
curves. It may be noted in passing that the relative permeability may not be so poor as aug-
presence of some shales may actually help the gested by standard interpolation methods. Fig-
drainage process, since the oil will then spend ure 13 shows some results [ 19 ] obtained by the
more time flowing in segregated form. centrifuge technique. The direction taken by the
isoperms measures the extent to which mobile
Tertiary gravity drainage k 3vo~
=

Sg : I00%

Figure 12 illustrates the principle of tertiary 9o/~.o

gas injection in a watered-out reservoir. Gas is 2 k kX)
injected to drive the oil rim down-dip and re- 8V~ kro
connect the oil ganglia trapped by waterflood- ~of/'~//~ \ ~ ~° o ~,,o-4
ing. Gas, oil and water all flow in the gas-in- 6o/a////~ ~° .,o-~
ended zone, so three-phase relative permeabil- ~off , ~ " / / / \ \~° D ,0-~
ity to oil governs the drainage efficiency in this / %// \ \~o a o
region; nitrogen strips the light hydrocarbons ~o/~ /~/ ~
out of the residual oil, and so may inhibit after- //
drainage. On the other hand, the enriched ni-
trogen is a more efficient agent for displacing ~)/" ~ ~
oil. These complicated interactions are dis- /~ ~ ~ ~ ~ ~
cussed in [19]. s~=,oo-/. -b ~o -& ~ 'b ~ ~ ~\~, so=~oo°/o
The mobility ratio will be unfavourable where s:,
gas displaces the oil/water bank, so the dis- Fig. 13. Three-phase oil relativepermeabilities.
 117

water interferes with the flow of oil. The exper- 5 Robl, F.W., Emanuel, A.S. and Van Meter, Jr., O.E.,
imental technique is fully described in [ 25 ]. 1986. The 1984 National Petroleum Councilestimate
of potential E.O.R. for miscible processes. J. Pet.
Technol., Aug.
Cost in offshore fields 6 Enhanced Recovery Week, 1986, 26 May.
7 Williamson, A.S., Gondowia, M., Pavlas, E.J., et al.,
1986. The planning of a large-scale miscible flood at
Tertiary nitrogen floods will be expensive PrudhoeBay. J. Pet. Technol. Oct.
offshore, because a platform is needed to carry 8 EnhancedRecovery Week, 1986, 2 June.
9 Enhanced Recovery Week, 1986, 25 Aug.
the nitrogen plant. Costs were estimated at $35/ 10 Enhanced Recovery Week, 1987, 9 Feb.
incremental barrel in two of Shell/Esso's 11 Patton, J., 1986. Enhanced oil recovery by carbon
North-Sea fields; nitrogen consumption was dioxide flooding. DOE/MC/16551-19. Reviewed in
about 14 m c f / i n c r e m e n t a l barrel, at a cost of Enhanced Energy Recovery News, Dec. 1986.
$0.40/mcf. 12 Enhanced Energy Recovery News, 1986, Oct.
13 Craig, F.F., 1985. Field use of halogen compounds to
trace injected carbon dioxide. Soc. Pet. Eng., 14309;
Symbols and S.I. u n i t s 60th Annu. Tech. Conf. Exhib. Soc. Pet. Eng., Sept.
14 Enhanced Recovery Week, 1986, 15 Dec.
15 Enhanced Recovery Week, 1986, 10 March.
k =permeability, m 2 16 Sankur, V., Emanuel, A.S. and Di Julio, S.S., 1985.
p =pressure, N m -2 Recovery of heavy oil by C02 injection. Houston
u = rate/area, m s - 1 AIChE 1985 PetroExpo. Reviewed in Enhanced En-
ergy Recovery News, April 1985.
g = acceleration of gravity, m s - 2 17 Mungan, N., 1981. Enhanced oil recovery using water
tt = viscosity, kg m - 1 s - 1 as a driving fluid. World Oil, Aug.
p = density, kg m -3 18 Orr, Jr., F.M., Silva, M.K., Lien, C.L. and Pelletier,
a,fl, Oare angles as shown M.T., 1982. Laboratory experiments to evaluate field
prospects for C02 flooding. J. Pet. Technol., April.
19 Niko, H., Schulte, A.M., Drohm, J.K. and Cottrell,
Acknowledgements c.w., 1987. The feasibility of Tertiary nitrogen injec-
tion in waterflooded volatile oil reservoirs in the North
Sea. Presented at the 4th European Symposium on
The author expresses his t h a n k s to the Man- E.O.R., Hamburg, Oct. J. Pet. Sci. Eng., 2: 119-128.
agement of Shell U.K., Shell Internationale 2o Gardner, J.W., Orr, F.M. and Patel, P.D., 1981. The
effect of phase behaviour on C02-flood displacement
Petroleum Maatschappij and Esso U.K. for efficiency. J. Pet. Technol., Nov.
permission to publish this paper. 21 Terry, R., Zaid, A., Angelos, C. and Whitman, D., 1987.
Polymerization in supercritical carbon dioxide to im-
prove carbon dioxide/oil mobility ratios. AIChE
References PetroExpo. Reviewed in Enhanced Energy Recovery
News, April 1987.
1 Bath, P.G.H., 1985. Enhanced Oil Recovery in the 22 Saner, W.B. and Patton, J.T., 1986. C02 recovery of
North Sea. Trans. Int. Seminar on North Sea Oil and heavy oil: Wilmington field test. J. Pet. Technol., July.
Gas Reservoirs. Norwegian Inst. Technol., Dec., Gra- 23 Pietlicki, R. and Archer, J.S., 1987. Novel scheme for
ham and Trotman Publ. control of numerical diffusion in reservoir simulation.
2 Dake, L.P., 1978. Fundamentals of Reservoir Engi- Can. J. Chem. Eng., 65, Jan.
neering. Developments in Petroleum Science, 8. El- 24 Christiansen, R.L. and Kim, H., 1984. Rapid mea-
sevier, Amsterdam. surement of minimum miscibility pressure using the
3 Stalkup, Jr., F.I., 1983. Miscible displacement. Soc. rising bubble apparatus. Soc. Pet. Eng., 13114; 59th
Pet. Eng. Monogr. Ser., Vol. 8. Annu. Techn. Conf.
4 Lambeth, N. and Dawe, R.A., 1987. Viscous effects for 25 Van Spronsen, E., 1982. Three-phase relative perme-
miscible displacements in regular heterogeneous po- ability measurements using the centrifuge method,
rous media. Chemical Engineering Research and De- Soc. Pet. Eng./DOE 10688; Third joint Symp. on
sign, Trans. Inst. Chem. Eng., 65 (1). E.O.R.