Вы находитесь на странице: 1из 19

SOCIETY (N?

PETROLEUM ENGINEERS OF AIME

;?!%SPE 5808
Micellar F!ooding Heterogeneities

The

Sensitivity to Reservoir

of

By ,
. ., ..

C..A..Kossack and H..L. Biltiartz; Wi, . . . . . ..,. .

Atlantic Richfield Co.

@Copyright 1976 Ameriean Iristitiitg gf,Miniag, Metallurgical,and Petroleum Engineers,Inc. . . .. ..i%IS PI@ER IS SUBJECT TO CORRECTION This paper was prepared for We Itiproved 0i2. Recovery Symposium of the Society of Petroleum. Erig@e~r,~.of AIME, to be held in Tulsa, Okla., March 22-24, 1976. IlluPermission to copyis restricted to an abstract of not more than 300 words. The abstract should contain conspicuous acknowledgstrations may not be copied. Publication elsewhere after ment of.where and by whom the paper is presented. publication in the JOURNAL OF PETROLEUM TECHNOLOGY or the SOCIETY OF PETROLEUM ENGINEERS JOURNAL is usually granted upon request to the Editor of the appropriate journal provided agreement to give proper credit is made. Three copies of any discussion should be Discussion of this paper is invited. sent to the Society of Petroleum Engineers offiae. Such discussion may be presented at the above meeting and with the paper, may be considered for publication in one of the two SPE magazines.

ABSTRACT .. . This paper provides a comparative evaluation of the effect of reservoir heterogeneityon amiscible -tertiary micellar flood. The sensitivityof slug size, preflushvolum-e, surfactant loss, and miscibility saturation of this complex displacement process was evaluated for a homogeneous, randomly heterogeneous, non-communicating layered and partially communicating layered model The five-component, finitereservoir. difference reservoir simulator used for this study is described in detail. The results show that reservoir heterogeneityis a dominant factor, adversely affecting the performance ofmicellar flooding to suchan extent that the process may notbe feasible in many non-homogeneous reservoirs. Inth.e partiall ycommunicatin glayered model reservoir, large volumes of preflush were required; and even after adequate preflushing~ 50 to70 percent of the post waterflood oil in place was unrecoverable. The results indicate References and illustrations at end of paper. thatfor a heterogeneous reservoir the most econornical slug size maybe a very small surfactant slug that displaces only the easily mobilized oil. This work also shows thata reduction in the surfactants adsorption praperties anda micellar fluid that mobilizes oil at low surfactant concentrations will, for a micellar flood with a given slug size, both substantially improve recovery. INTRODUCTION The o~jective ofthis work wasto stud) the effectof reservoir heterogeneityon the performance ofmicellar flooding using a numerical simulator. The four reservoir descriptions used in this study were homogeneous, randomly heterogeneous! noncommunicating layered, and partially comMicellar floods municating layered models. were simulatedto compare the effect of slug size, surfactant loss, and critical slug saturation upon tertiary oil recovery in the four

*
ifi ..

THE SENSITIVITY 03? MIGELLAR FLOODING TO RESERVOIR HETEROGENEITIES 5 Z i=l

sP17 5808 -...-

model reservoirs. These sensitivity studies indicate where and under what conditions the mice llar flooding concept might be applicable. The performance prediction of a micellar flood in a complex reservoir requires a numerical simulator that represents the reservoir features, chemical properties, and displacement mechanisms that affect the. flood To simulate this oil recovery performance. process the model described .below accounts f dr the flow of five separate components - -oil, water, surfactant, polymer, and pref lush. In addition, the simulator includes the adsorption of surfactant and polymer, permeability re duction, generation of miscibility, and the mixing of miscible fluids. THE MODEL Previous finite -difference siinulator development, omitting stream tube models and incompressible models, in the field of miscible flooding has been very limited due to numerical dispersion dissipating the small surfactant slug. The basic models in this field are not suited for micellar flooding but were designed for solvent and polymer floods. Todd and Longstaffl and Bondor, Hirasaki and Thom2 have developed four-component, miscible flood mode 1s for the simulation of solvent and ptilymer displacement, - re spec tively. These two papers describe methods to simulate many of the important features in a miscible flood and a mobility controlled flood, such as the flow of miscible fluids, the mixing of miscible fluids, adsorption, residual resistance (permeability reduction), arid non Newtonian effects. A numerical model approximating a system of five nonlinear partial differential equations, each representing conservation of fluids, is assumed to describe two-phase, multi -component compressible flow in rese r voirs unclergoing water external micella r flooding. These equations, obtained by combining Da rc ys law with continuity, are represented explicitly for the ith component in Eq.1. Bi k kri pi

It is also required

that

S = 1 and the i

usual capillary pressure relationship be expressed as a linear function of the aqueous saturation.

?2 -pl=p

.c2-~

sac/

(2)

The development and loss of miscibilityy, permeability reduction, and surfactant 10Ss are handled by placing auxiliary constraints on the system of equations and are satisfied explicitly to the pressure solution. The .imulator solves the pressure equation implicitly using either Gaussian elimination or the strongly implicit procedure. The saturations are then calculated explicitly following the standard IMPES (implicit pres p sure, explicit sat~. dtiOIi) rocedure. The model contains an explicit calculation in which the surfactant and polymer are removed from the flow to simulate loss of chemicals. The rat~s of reaction for the polymer adsorption and surfactant loss are assumed to be infinite (an instantaneous reaction) until the predetermined quantity of the component has been either lost or adsorbed; at that time the rate becomes 2eroi The surfactant loss rnechahism can include true chemical adsorption, and pre cipitation when in contact with multivalent ions or the 10Ss of effectiveness when mixed with a high salii~iity brine. The 10Ss is assumed known and inputted to the simulator in terms of pounds of s urfactant 10Ss per acre -foot of reservoir. Once a grid block has lost its specified amount of surfactant, no additional 10Ss is calculated for that cell, To maintain a material balance, lost surfac tant and adsorbed polymer saturations become preflush brine saturation. The amount of surfactant 10Ss is modeled as a function of preflush eff icienc yj see Fig. 1, where preflush efficiency

=
[
[

saturation of preflush brine

V*
[.

wei

Vli 1

+q j=:(q

1/

BiSi) 1 (1)

i=l,2,3,4,5

saturation of preflush brine

saturation + of reser voir brine

1..

(3)

,,

SPE 5808

C, A.

Kossack

and H. L.

Bilhartz,

Jr.

IL7 -..

Therefore, in a given cell, at any time, the maximum quantity of surfactant 10Ss allowed is dependent on the pref lush efficiency. No desorption of either component is currently allowed in the model. The miscibility mechanism currently used in this model is a. first order approxi mation to many of the complex transition izone phase relationships that are currently being investigated. 3$4 The switch f rorn immiscible ? two-phase flow to a single phase is achieved through fir st contact miscibility, * which is keyed to the amount of the surfactant slug present at any point. This critical satura tion** of the surfactant is a laboratory measured parameter which varies for different micellar fluids) crude oils, and salinity. Its value represents the lowest concentration that the micellar slug can be diluted to and still be miscible with the oil. Therefore, as long as the micellar saturation at a given point is above the critical saturation, all five components are considered as miscible components of the mixed phase and the capillary pressure Miscibility y, for is set equal to zero. P= 2-1 this study, is not a function of pref lush effi ciency except through adsorption which de creases the iurfactant concentration. If the .surfactant saturation is, less than this critical +aiueDthen miscibility between the aqueous phase (reservoir brine solution, polymer solution, surfactant solution, pref lush s olu tion) and the hydrocarbon phase is assumed lost; i, e. , the flow is immiscible, and p C2-1 is non-zero. Even under this condition all four aqueous components remain miscible Each miscible within the wetting phase. component assumes a fraction of the phase relative permeability equal to its volume . *Iu this paper the word miscible implies a numerical state where all the components in a cell are mobile. No inference is intended concerning the number of phases or occurrence of fluid interfaces. **The critical saturation (expressed as a function of the cells pore volumes Sc is equivalent to the minimum surfactant concentration necessary to achieve misci bility. . . ... ..-. .. . ,

fraction in that phase, since the movement of one component through a porous medium is not impeded by the presence of other components in the same phase except for a reduction in the area available for flow. Thus, in the two-phase region, S4 < Sc

r,

Si rkr aq
i=l,3,4,5

(Saq)
aq (4a) (4b)

k 2 For miscible k r.

= k 2 flow,

(Saq) .

S4 ~ Sc (5)

= Si for all i.

A mixing parameter model is used to account for the creation of dispersed or mixing zones between miscible components A description of the within a single phase. one -fourth power fluidity mixing rule and the definitions of the effective viscosity are .Yhe mechanism of outlhed in Appendix A. permeability reduction due to polymer and surfactant contacting reservoir rock can be handled by the model, but it will not be discussed here because it was neglected for this work. To complete the mathematical formulation of the model, in::ial and boundary conditions are needed. Initial condition: @j = Ij (x, y, ; ;(o,t) surface normal to u

Z, O) =0

(6) (7)

Boundary Where:

condition: u = boundary

H, n = unit vector

Eqs. 1 through 7 describe completely the flow of fluids under the assumed mechanisms i The se equations with unknowns of pressure and saturation constitute a set of nonlinear differential equations which must be solved numerically. The solution of this mathemat ical model is accomplished by reducing the se difference equations to a single oil phase matrix problem through the use of the IMPES solution technique.

THE SENSITIVITY

OF MICELLAR

FLOODING

Numerical

Dispersion

The discretization error which occurs when the first spatial derivative of a quantity, c, is difference (first order accurate with upstream weighting) has the form5 #6 Ax Z-z ,.,J * ~z c (8) ()

This match of dispersion applies in the direction of flow when the frontal velocity is a constant throughout the reservoir; i. e. , it is valid for 1-D and 2-D vertical, crosssectional, homogeneous reservoirs. The analysis can be extended to 2 -D vertical, cross -sectional, heterogeneous reservoirs by making the following assumption: If two-dimensional (cross -sect<onal) reservoir A with constant velocity ii (specified injection and production rates), length L, time step At, and block size Ax yields a diffusivity y K1 (corre spending to numerical ii L), then the addition dispersion of K* of certain heterogeneity (not requiring re gridding) into the description of A does not increase the r.~me.rical dispersion, Thus , the additional diffusivity encountered by the fluid results from cross flowing and channeling which may be called mac~oscopic dispersion and not be due to additional truncat&nerror in the mathematical formulation.
q q

This error introduces added dispersion to the partial differential equation; thus its For example, name, numerical dispersion. in a porous medium, a flow with a frontal velocity of one foot/day, a grid block length of 20 feet, a time step of one day, and single point upstream weighting of relative pe rme ability experiences a diffusivity of 9, 5 f t2/ day. To reduce th@ to an average reservoir diffusivity of one ft~ /day would require grid blocks of 3 feet in length (this would require increasing the number of grid blocks by a factor of seven). The difference equations- used to solve for the flow of miscible fluids in this model contain fir st order accurate difference formulas; thus, using the results of Lantz, 6 ~he dimensionless numerical cliff usivit y, R*$ created by the solution of Eqs. 1 through 7 in a miscible region is given by Eq. 9. E*=T
AS-AT

Implicit in this statement is that the average longitudinal velocity in homogeneous reservoir A is the same as a weighted vertical average of the veloci~ies of heterogeneous reservoir A, and that the longitudinal compo nent of velocity is much greater than the transverse component. Also, no attempt was made -to scale the transverse. dispersion. Thus, it was assumed for this work that micella r floods could be simulated in cross-sectional (x-z strip) reservoirs with This a fixed amount of numerical dispersion. numerical dispersion represents the intra grid block microscopic dispersion (the dis persion created by heterogeneities that are less than Ax in scale). When studying micellar flood performance in a reservoir with vertical heterogeneities, the total simulator dispersion would be the sum of numerical dispersion fixed by Ax, At, and u and the additional macroscopic dispersion created by the inhornogeneous fluid flew between grid blocks where thm heterogeneity can be represented by a varying.pe rmeabilit y from grid block to grid block or fror.. layer to layer. This macrosco~ic dispersion is an approximation of the dispersion effect of real phenomena of cross flow, channeling, ctc. VW inherent assumption is that the numerical dispersion matches the int rablock micr~copic .dis persion and the simulators r&-croscopic

K G

(9)

Since it is infeasible to attempt to reduce R* to such a level that it is no longer a significant factor in the flow, the only re course is to use the numerical dispersion to simulate the real dispersion present in the formation. The numerical diffusivity, R*, can be made to match the physical dispersion, K, with the proper choices of Ax and At, as surning a cor.stant f rental velocity, u. Thus, for a 1-D reservoir displacement simulation at unit mobility, .10-acre well spacing, frontal velocit of one f t /day, and a cliff usivit y of one ft 1 /day, the numerical dispersion will match the physical dispersion with any of the following (Ax, At) pairs: (3 feet, one day), (4, 2), The use of two point upstream (5, 3), etc. weighting will replace the 2 in the denomina tor of Eq. 9 with an 8; thus the (k, At) pairs become (9, 1), (10, 2), (11, 3), etc.

,, .

=PE

5808

C. A.

Kossack

and H. L.

Bilhartz.

Jr.

lb

dispersion approximates gross fluid move ments, and that their interaction in the simulator results in a total simulator dispersion equal to the total physical dispersion in reservoirs like the ones that were modeled in this study. (Refer to Fig. 2.) THE SENSITIVITY STUDY

--

The re are limitations to our nume ri cal reservoir simulator, to our uncler standing of certain micellar f 100ding mechanisms, and to the availability y of descriptive reservoir information on specific reservoirs. However, a sensitivity analysis of the micellar. flooding concept using this simulator to evaluate the effect of various types of reser voir heterogeneity, flooding parameters, and chemical properties upon oil recovery should give us semi-quantitative insight as to the This study proapplicability y of the process. vides information on (1) the types of reser voirs where micellar flooding is more applicable, (2) the effect of surfactant slug size and critical saturation on recovery, (3) the preflush requirements needed under various reservoir conditions, and (4) the effect of surfactant 10ss on recovery. Model Reservoir . . Des criptions

(constant Ly), homogeneous reservoir model (660 feet between wells) required 74 grid blocks for the simulator dispersion to match a physical dispersion of one ft2/day. For the variable Ay configuration, miscible simula tions were run where the production concen t ration was plotted against time. Eightyeight grid blocks were required for the concentration profile to match that of a homo geneous reservoir where K equaled one ftZ/ day. The four model reservoirs were described by using various levels of permea bility to represent heterogeneity. Reservoir No. 1 was homogeneous; No. 2 was randomly heterogeneous; No. 3 was non-communicating layered; No. 4 was a partially communicating layered reservoir with 10 percent of the cells having vertical communication. A des c riptiom of the assignment of permeabilities is given in Table 2, and the procedure for the drawing of permeabilities from a log-normal distribution is given in Appendix B. The re suits from f Iooding the homo geneous reservoir were used as a comparison standard for the heterogeneous reservoir floods . The randomly heterogeneous reser voir was designed to show the effect of .lddi tional dispersion created between grid blocks with -randomly assigned permeabilities rangin~ from 90 to 3200 md (from a log-nomrial distri bution, mean 500 md, V = O. 6). Reservoir No. 3 had horizontal layer permeabilities varying from 200 md to 1220 md. Reservoir No. 4 was identical to No. 3 except a limited amount of vertical communication was allowec The four model reservoirs were water flooded in the simulator at a rate of 300 B/D This point until a WOR of 30 was reached. was considered the economic limit of secondary recovery, and the remaining oil in the reservoir was classified as %ertiary oil. 1 The recovery and rough saturation profiles following the waterflood are given in Table 3. Reservoir heterogeneity had a very slight effect on the recovery of oil because of the lack of areal sweep effects, the low viscosity of the oil, and the effect of gravity on the vertical sweep in the homogeneous and randomly heterogeneous reservoir. This result minimized the effect of waterflood performance on the results of the micellar flood.

Four reservoir deicriptio-ns W6re constructed with average properties repre senting a five spot with ten acre well spacing. A table of all the properties common to all four reservoir models is given in Table 1. The grid was constructed with a variable Ay grid in an attempt to simulate a quprter five spot with the two -dimensional, vertical cross section; see Fig. 3. This alteration of a 2 -D strip gave a better approximation of the three dimensional case with respect to the velocity distribution in the x direction and the timing of the volumetric sweep, especially when miscibility was lost before surfactant breakthrough. Since the frontal velocity varied with x, the application of the assumptions for the match of numerical dispersion was more complex than previously discussed. The total physical dispersion for the homogeneous reservoir was chosen to be one ft2/day because (1) it is typical of many sandstone reservoirs and (2) it is low enough not to be the dominant characteristic in the micellar flood. A vertical cross sectior.

50 1?eflush Studies

THE SENSITIVITY OF MICELLAR FLOODING TO RESERVOIR HETEROGENEITIES crosshatched overlap sketched in Fig. 4.

sPl? 5801 region in the case

Niwt micellar flood applications require preflushing with a particular brine. The purpose of this preflush is to act as a buffer between the reservoir brine in place at the start of preflush and the micellar slug. The presence of high total dissolved salt con..centrations and/or divalent ions (Ca+ t ) in the reservoir brine can cause a loss of oil displacing ability and an increase in adsorption upon mixing with the micella r slug. Thus, a study of the quantity of preflush required in the four model reservoirs was a necessary part of this work. The question posed was how large a slug of preflush is needed to separate the reservoir brine from the surfactant slug so that only a specific fraction of the injected micellar slug (overlap volume) is rendered ineffective by mixing with the reservoir brine ? The overlap volume is defined as the volume of the micellar slug that is contacted (mixed) by the reservoir brine above a cutoff I brine saturation and above a cutoff surfactant saturation. This is best explained by using an example along with Figure 4. With a slug made up of 15,000 mg/1 brine, O mg/1 Ca++, let us assume that the process could tolerate a maximum salinity increase of. 10 percent . (to 16,500 mg/1) and a maximu Ca+ + &oncentration of 200 ppm before the activity of the micellar slug decreases to zero. Therefore, i~~., mixture of more than 5 percent rese; vtiiv brine would result in an ineffective SIUZ f:f wo used a reservoir brine salinity of 50,900 mg/1 NaCl and 4,000 mg/1 Ca++ . This gave the definition of the cutoff brine saturation of O. 05 in Fig. 4.

A series of twenty complete micellar floods was simulated using varying amounts of preflush to determine the preflush require ments for the four reservoirs. The previ ously waterflooded reservoirs were treated ,with an 8.8 percent pore volume micellar slug with a critical saturation for loss of miscibility of O. 05 and a baseline surfactant adsorption of a constant 1000 lb/acre -foot. The volume of surfactant overlapped was calculated every four days throughout each micellar flood. The maximum overlap was then determined for each micellar displace me nt. These maximum overlaps have been expressed as fractions of total injected surfactant and plotted in Fig. 5 against the volume of preflush injected. This figure shows the results of all four reservoirs using the O. 05 cutoff! and of reservoirs 2 and 4 using the O. 01 cutoff. It is inportant, in view of the assumptions, to study these curves in a comparative sense. The O. 01 cutoff curves demonstrate the sensitivity y of the pref lush process to increasing reservoir brine salinity; and for the O. 05 cutoff! cases, approximately ten times as much preflush was required for the partially communicating layered reservoir ,. . than for the other reservoirs when compared at a low overlap value. To minimize the impact of preflush efficiency on the following micellar flooding sensitivity y study, the four model reservoirs were treated with the preflush volume that would yield a O. 10 maximum surfactant over lap. (See Figure 5.) This means that, in Fig. 4, the crosshatched region was allowed to be a maximum of 10 percent of the total injected slug in each of the four reservoirs. This resulted in a 2.85, 3.7, 2.2, and 32 percent of a pore volume preflush slug for reservoirs No. 1 to, No. 4, respectively, when the O. 05 cutoff was used. Micellar Displacements

Also, we assumed that the micellar slug was ineffective at saturations below O. 051 this defined the surfactant cutoff 1 saturation of 0.05 in Fig. 4. The total volume of potentially active surfactant in contact with re se r voir brine between the O. 05 cutoffs is crosshatched in Fig. 4; i. e. , this overlap region contains surfactant made ineffective by contact with high salinity brine and Ca+ + ions. If a very high salinity reservoir brine . were encountered, say 200, 000 mg/1 NaCl and 20, 000 mg/1, Cat+, then the brine ~cutoff saturation would be lowered to O. 01 and .- the volume labeled (a) would be added to the
,..

The standard micellar flood was designed with typic~i values of the process variables to establish a base performance for the four reservoirs. The process was conducted on the four reservoirs in an

. . . .

SPE

5808

C. A.

Kossack

and H. L.

E+ilhartz,
.

Jr.

151

ve rtical communication between layers identical fashion so any difference in recov The partial communicating usually exists. ery can be attributed to the cliff erences in The standard ,,. ,~,case left an additional 20 percent of the reservoir heterog,eneity. tertiary oil in the reservoir compared to micellar flood included all of the data of the non -communicating case. Table 1 and began after the preflmh injec The surfactant slug vol.urne was 6.13 tion. A sensitivity y study of the process percent of the total pore volume of the reser variables was made to further evaluate voir which provided sufficient surfactant micella r f 100ding in the four model re ser (!hemical to satisfy a 1000 lb/ac re.-foot voira. The study consisted of 28 micellar surfactaht 1os3 for all of the rock in the entire floods that were identical to the correspondquarter of a five spot. Thus, in a homogen ing standard micellar flood except that one eous reservoir, in the absence of gravity$ The variable in each simulation was altered. with piston -like displacement, and at a 100 sensitivity y study was developed around the percent efficient preflush, no surfactant perturbation of the following variables: (1) The critical saturation would be produced. the slug size, (2) the amount of surf actant of the micellar slug for the loss of misci loss, and (3) the critical saturation of the bility was 0.05 (miscibility means 100 perslug for 10Ss of miscibility, The surfactant cent displacement of the oil was allowed). slug sizes injected were 4 percent and 12 perThe standard surfactant loss curve was cen,t of the pore volume. The high and low assumed to be linear with pref lush efficiency surfactant loss curves are plotted on Fig. 1. with end points of 1000 lbs/acre, -foot at a The critical saturations used were O. 10, preflush efficiency of 1.0 and 3000 lbs/acre 0.025, and 0.001. The tertiary oil recovery foot at 0,0 efficiency, as shown in Figure 1. efficiency for all the simulations in reservoir The micellar slug was followed by a 40 perNo. 1 are plotted against pore volumes of cent pore volume slug of polymer solution, fluid injected after preflush in Figs. 7a, 7b, Drive water was then injected into the for and 7c. The same efficiencies are plotted mation until the water/oil ratio in the profor reservoirs No., 2, 3, and 4 on Figs. 8a, ducer reached 30. 8b, 8c, Figs. 9a, 9b, 9c, and Figs. 10a, 10b, 10c, respectively. We calculated the tertiary oil recovery and oil/ surfactant ratios for the standard The sensitivities of the tertiary oil flood. Those re suits are labeled ! standard recovery and oil/ surfactant ratio are listed ~. micellar flood in Tables 4 and 5. The in Tables 4 and 5. The upper number in normalized tertiary oil recovery is plotted Table 4 is the fraction of the oil in place against the pore volumes of f hid injected recovered by the tertiary process, whereas, after preflush in Fig. 6. A tertiary oil the upper number in Table 5 is the oil recovery efficiency of 1.0 would mean that recovered- surfactant volume ratio. The the process recovered all the tertiary oil lower number is the sensitivity y (percent (total oil in the reservoir at the end of the change) of the parameter variation when waterflood). The results show that reser co~pa red to the standard micellar flood in voir heterogeneity y has a pronounced effect that reservoir. No special simulations to on the tertiary oil recovery efficiency. The optimize any of the variables and no s imula additional dispersion caused by the randomly tions with a combination of altered variables heterogeneous reservoir resulted i,~ 18 perwere run.. cent less oil recovery than that for the homo geneous reservoir. The layering of reservoir DISCUSSION No. 3 (with no vertical communication) caused the production of 31 percent less than the An overview of the sensitivity study homogeneous case. When some vertical communication was added to the layered case, reveals that reservoir heterogeneity and su.rfactant loss are major factors in the one -half of the recoverable tertiary oil was recovery of tertiary oil. A severely hetero left behind. The comparative results begeneous reservoir or a reservoir with a high tween reservoirs No. 3 and No, 4:are signifisurfactant loss will leave as much as threecant because many micellar floodfield evalufourths of the tertiary oil in the formation.. ations are estimated from a non ~communi Specific- observations which. can be drawn ., eating layered model, wlien in reality some .

D-52

THE SENSITIVITY OF MICELLAR FLOODING TO RESERVOIR H. rEROGENEITIES from Tables 4 and 5 concerning the effects of heterogeneity and process variables follow.

SPE 580$

~ ~
=

contact a large portion of the reservoir was negated by ,the 10Ss of surfactant to the rock. Thus, the high surfactant 10Ss systems looked equally bad in all types of reservoirs. (5) The effect of low surfactant loss is surprisingly constant in reservoirs No. 1, No. 2, and No. 3, where an additional 25 percent recovery was gained.. Reservoir No. 4 did not increase as much as the others, again because of the dominance of heterogeneity on the process. (6) The low value (O. 10) of critical saturation of the slug for 10Ss of miscibility had its largest effect for reservoirs No. 1 and No. 2 because the formation was contacted by the slug more evenly and miscibility was lost in all layers at some distance from the reservoir entrance. The location in each layer when miscibility was lost is given, along with standard micellar flood recoveries, in Table 3. Reservoirs No. 3 and No. 4 showed little effect of this low dilution as most of the surfactant entered only the most permeable layers where dilution was not severe. (7) The change of the critical satura tion from O. 05 to O. 025 (generally considered the most probable range) showed little effect for any of the cases. The largest effect was fetid for the No. 2 rese#voir case because dispersion of the slug dominated the behavior. (8) The asymptote for the critical saturation of surfactant (O. 001) gives the results of a completely miscible displacement; i. e. , if any surfactant is present, then the system is miscible. These recoveries are the most optimistic for the reservoirs with the standard surf actant slug and sur factant loss curve. The recovery results show that the more homogeneous reservoirs (No. 1 and No. 3) had nearly. reached their asymptote by the O. 025 simulations, yet the more heterogeneous reservoirs (No. 2 and No. 4) showed significant increases in recovery at O. 001. Thus, the development of micellar fluids effective at lower concentrations of surf actant ne~ded to produce miscibility could have a favorable effect on the micellar flooding of heterogeneous re ser .voirs.

the surfactant slug (1) Increasing size from 4 to 12 percent pore volume increased the oil recovery substantially as expected. Yet the oil/ surfactant volume ratio, a measure of the worth of the process, decreases (with the exception of reservoir No, 1). An economic analysis to determine the optimum slug sizes should include a weighted combination of recovery (Table 4) and oil/ surf actant ratio (Table 5). These results suggest that for expensive surfactants the injection of a small slug into a hete ro geneous reservoir, displacing only the easily mobilized oil, may be more profitable. (2) Decreasing the slug size had the least effect on recovery for the reservoir No. 4 case because a reduction in slug volume means less slug is cycled through the reservoir and/or lost to tight layers. Decreasing the slug size had the largest effect on recovery for the homogeneous reservoir No. 1 case since it directly affected the volume swept by active sur factant. (3) Increasing the slug size had the smallest effect on recovery for the reser ., voir No. 3 case because th-esurfactant cycled through the more permeable layers. Increasing the slug size had the largest effect for the reservoir No. 2 case because dispersion led to an earlier inactivation of a small surf a ctant slug. (4) The high surfactant loss parameter has the smallest effect for the reservoir No. 4 case because ths heterogeneities reduced the surfactant contact with the reservoir to such an extent that high surfac tant 10Ss had little effect on the recovery. The largest effect of high surfactant 10SCS occurred for the reservoir No. 1 case, because the contact was the largest. A plot of the tertiary oil recovery efficiency vs. pore volume of fluid injected after preflush for the four model reservoirs with high sur factant loss (see Fig. 11) shows that the four curves are grouped very tightly. This indi cates that a high surfactant. loss in a reser - . voi.r greatly reduces the effects of hetero gene it y; i. e., the advantage of the slug to
..I

SPE

5808 PROGRAM LIMITATIONS

C. A. Kossackam

H. L.

Bilhartz

15:

h evaluating the model and the results p~esented, it is important to recognize the limitations introduced by the assumptions and the simplifications that were needed to construct a tractable solution. The model does not rigorously represent the thermod~amic. and transport phenomena which are neces sary to describe the mechanisms and details of the various processes. The model attempt to describe the pertinent features, as a first order approximation, so it can be used for the evaluation of micella r floods in various The sensitivity study model reservoir types. should be viewed in a comparative sense, not When viewed as the absolute performance. in the proper sense, the effects of heterogeneity and chemical properties between, simulations give a semi-quantitative re suit for the effects in a real reservoir. Certain assumptions and simplifications listed below lead to higher oil recoveries than could be expected from real reservoirs. (1) (2) Transverse Areal dispersion. 7

quantities in excess of 50 percent of a pore volume will likely be required to be effective in stratified reservoirs with communication between the strata. (3) The sensitivity analysis of the surf actant slug size showed that as the slug , size increases the oil recovery, as expected, increases; but the oil produced/ surfactant injected ratio (as a measure of process efficiency) decreases in the heterogeneous reser voirs. Therefore, the optimum slug size for a heterogeneous reservoir is likely to be less than that for a homogeneous reservoir. (4! For a 6.13 percent pore volume surfactant slug, a reduction in the s urfac tantfs adsorption and other loss characteristics caused an increased oil recovery from the more homogeneous reservoirs but led to only a slight increase in the partially communicating la ye red reservoir. (5) A micellar fluid that mobilizes oil at much lower surfactant concentrations than can be achieved with e urf actants currently available will not substantially increase Yet, recovery in homogeneous reservoirs. in heterogeneous reservoirs a micellar fluid that is miscible at lower concentrations will greatly increase the tertiary oil recovery efficiency . ... ; . . ... ,. .. NOMENCLATURE

sweep effects. effects . .

(3) Stratification and gravity within a grid block. ,,. ;. ,.. _. . . ., CONCLUSIONS

(1) The numerical simulator study of tertiary micellar flooding showed that reservoir heterogeneity has a pronounced detrimental effect on the oil recovery efficiency. For example, in the standard micellar flood the oil recovery of the pa rtiall communicating layered reservoir was less than one -half that of the homogeneous re ser voir. The oil recovered-micellar fluid volume ratio for this heterogeneous reser voir was only 1.7 compared to 3; 3 for the homogeneous reservoir. (Z) The partially communicating layered reservoir required ten times the qgantity of preflush needed for the more homogeneous reservoirs to provide adequate separation _between the reservoir brine. and the micellar fluid. Despite the optimistic fluid flow mechanism built into the simulator, the result of this study showed that pref lush

Bi ~a+ + K * K k kr L

= formation . calcium

volume ions

factor,

STB/RB

= physical

dispersion,

(length)2 /time diffusivity md fraction to

= dimensionless = absolute = relative

numerical

permeability, permeability,

= total system length, injector producer distance, length = pressure, oil-water


= .= strength

P P=
2-1 qj

psi capillary pressure -

of source or sink repre senting wells, ST B/time

..

54 RK3

THE SENSITIVITY OF MICELLAR FLOODING TO RESERVOIR HETEROGENEITIES = polymer component to permeability reduction = polymer RK4 residual resistance factor r RK4, = micellar permeability reduc tion factor, input value saturation, fraction = relative M m RR3 = median = mixing

SPE 5808

REFERENCES 1. Todd, M. R. and Longstaff, W. J,: The Development, Testing, and Application of a Numerical Simulator for Predicting Miscible Flood Per formance, J. Pet. Tech. (July, 1972) - 874. Bondor, P. L. , Hirasaki, G. J. , and Thorn, J. J. : Mathematical Simulation of Polymer Flooding in Complex Reser voirs, r Paper SPE 3524, presented at SPE -AIME 46th Annual Fall Meeting, New Orleans (Oct. 3-6, 1971). Healy, R. N. and Reed, R. L.: Physiochemical Aspects of Microemulsion Flooding, Sot. Pet. ~. (October, 1974) 491.

s.1 s aq
t

total aqueous phase saturation time, frontal days velocity, V/q 2. length/time

u
v

darcy velocity,

v
v v u i b{ # ~

permeability y variation gradient dive rgence effective viscosity, cp Ax/L 4. Hubberts potential = 3.

~.

dimensionless

Jell size,

.>

[1
J o
j

*.L P.. .: 1

, height .. . .. .. .. 5*

Healy, R. N., Reed, R. L., and Stenmark, D. G,: r!M~tipha se Micro emulsion Systems, Paper SPE 5565, presented at SPE-AIME 50th Annual Fall Meeting, Dallas (Sept. 28 -Oct. 1, 1975). Todd, M. R. , ODell, P. M., and Hirasaki, G, J.: Methods for Increased Accuracy in Numerical lators, Trans. AIME Reservoir (1975) &, Simu515.

porosity, density

fraction

P A7 w

dimensionless

time uAt/L, pv 6. Lantz, R. B.: Quantitative Evolution of Numerical Diffusion (Truncation Error), II Sot. Pet. ~. ~. (September, 1971) 315xo. Koonce, K. T. and Blackwell, R. J.: Idealized Behavior of Solvent Banks in Stratified Reservoirs. Sot. Pet. ~: J. (December, 19~5) ~8 -3~0 #Dykstra, H. and Parsons, R. L.: The Prediction of Oil Recovery by Water Flo~di in: Secondary Recovery of Oil in the United States, 2nd Edition, API, New York, New York (1950).

the fluidity mixing parameter indicating-the degree of mixing

Subscripts 7. aq D e i = aqueous = , standard = effective = 1,2, 3,4,5, brine, oil, preflush component: reservoir polymer, surfactant, . deviation 8.

,,

SPE 5808 APPENDIX Fluid Mixing A

C. A.

Kossack

and H. L.

Bilhartz,

Jr.
(1 - W4)

155

U)4
v

IJ

V4

and Effective

Viscosity

1,2,

3,4,5 (A-2)

A mixing parameter model was used to account for the creation of dispersed or mixing zcnes between miscible components within a single phase. 1 If the dispersed zone completely occupied a cell in the model, then the effective viscosity. of each component was that defined by a one-fourth power fluidity mixing rule as given in Eq. A-1. n

where ~ is the one-fourth power fluid mixing rule (Eq. A-1) and @i is the mixing pa -, rameter indicating the degree of mixing. The limiting cases of no mixing and complete mixing correspond to ~i = O and ~i = 1, respectively. The model included the option of specifying separately the degree of mixing for each component (when the components share a dispersed zone). APPENDIX B

Assignment of Permeabilities from a Log-Normal Distribution The Dykstra -Parsons8 concept of scaling the rock permeability to a log-normal distribution with a permeability variation parameter measuring the degree of heterogeneity provides an unbiased means of assigning permeability to layers or grid blocks. When the cumulative distribution is plotted on a log-probability scale, the location of the (assumed) straight line is fixed by two parameters, the permeability y at the median kM and at one standard deviation (84. 1) above the mean kD. ~ The slope is commonly called the permeability variation, V; see Fig. B-1. The average value of 0,6 was used for the simulations since most sandstone reservoirs have a macroscopic permeability variation of 0.4 S V <0,8. Thus, given the mean (500md) and V = 0.6, a straight line, Fig. B-1, can To obtain five pe rmeabilities from be drawn. this log -normal distribution to be assigned to the five reservoir layers as was done in reservoirs No. 3 and No. 4, the permeabil ities at cumulative probabilities of 16.6, 33, 3 50.0, 66.7, and 83.3 are read from the graph These five * ~lues were then assigned randomly to ti. five layers. For reservoir No. 2, the J8 permeabilities at cumulative probabilities of 1.12, 2.24, 3.37, . . . . 97.8, 98.9 are read from this graph. These 88 values were then, drawn ra.ndo~y for each grid block- in a given layer. Each of the five layers was constructed from a separate random drawing in an attempt to create a truly random (tibiased) reservoir description.

(A-1) where L = set of components a dispersed zone which can share

h = total number of components set L 4

in

identified each component in set L. . . .-. -..... .,, , , On the other hand, when the size of the dispersed zone was small compared to cell size, the component viscositiess became those of the pure components. It is reasonable to expect the actual effective viscosities to fall somewhere between the mixing limits; thus, the effective viscositiess are defined as: ,, .-, (1 - u) Me i i= = Pi v 1,3,4,5 1, 3, and 5 (for water, polymer,. and preflush) U)i 8

v 2

(1 - W2) v = V2 2, (2 = oil, 4 fluid)

4 = micellar

TABLE 1 - DATA FOR ALL 2-D VERTICAL CROSS SECTION SIMULATIONS


/ System Reservoir Dimension: 10 acre spacing, quarter 5-spot, 660 x Variable YX45

Data - before

waterflood
0.70 0.30 0.25 360 M bbl 252 M bbl 0.2 500 md .000003 psi-i 2000 psia

Initial Oil Saturation Connate Water Residual Oil to Waterflood Pore Volume Original Oil in Place Porosity, p Absolute Permeability y, k Rock Compre risibility Initial Reservoir Pressure Relative Permeability KR W 0 0.01314 0.0555 0.129 0.2348 0.30 1.0 Data .. Table

Sw
. 3 .4 5 :6 .7. .75 1.0 Fluid

KRH
1.0 0.628 0.3371 0.131 0.0172 0.0 0.0 ,.. .

Oil Viscosity Water Viscosity Preflush Viscosity Micellar Fluid Viscosity Polymer Viscosity Formation Volume Factors (All Components) Polymer Content ration Maximum Polymer Adsorption Mixing Parameters (All) Aqueous Densities (at standard conditions) Oil Density (at standard conditions) Water Compressibility Oil Compressibility Surfactant Concentration Capillary Pressure Wi .

l.o.cp 1.0 Cp 1.0 Cp 4.0 Cp 4.0 Cp 1.0 RB/STB 1000 ppm 50 lb/ (acre -ft) 1.0 62.4 lb/cu. ft. 55.0 lb/cu. ft. .0000004 psi- 1 .00005. ,ps,i- 1 30,000 ppm O at all, saturations l.O i= 1,2, 3,4,5

TABLE 2 - ASSIGNMENT OF PERMEABILITIES IN THE FOUR RESERVOIRS


Reservoir No. 1 rmeabilityof 500 md. -,

Homogeneouspe kv/\ Reservoir =

0.1

No. 2 permeability distribution

Randamly heterogeneous Mean permeability

=500md

Permeability variation, V=O.6


Permeability

(see Appendix B)

from 88 permeabilities from a log-normal


Reservoir No. 3 reservofz

assigned randomly to each cell (one layer at a time) distribution with V = 0.6.

Layered heterogeneous Mean permeability Permeability Permeability

without vertical communication

= 500 md V = 0.6
layer

variation,

varies among the layers onlfi each

has a

constant permeability.
Ayes

1 (top) 2

kx = kx = k= =

500 740 205

Layer

Layer 3 .. ., Layer 4 -

kx- = -. 12z0 kx = 340

Layer 5 (bottom)

kv/\
Reservoir

= 0.
No. 4

with some vertical communication. Layered heterogeneous reservoir 8ame a~ Reservoir No. 3 except vertical communication was permitted thrmgh periodic windowc spaced every eight cells in the x direction staggered from layer to layer so vertical flow cannot trave~ more than one cell up or down without flowing ltorizontelly four cells.

.-

TABLE 3 - RECOVERY AND DISPLACEMENT RESULTS FOLLOWING WATERFLOOD AND STANDARJMICELLAR FLOOD
OilRecovery

-..

Waterflood

Reservoir

Initial Oil In Place (Mbbls)

atWORof 30 (%ofinitial oilinplace)

Unit Displacement* Effect in Layers 1-5

.-

IncrementalSurfactant (distancefrom
injector, feet) Flood Recovery atWORof 30 {% of initial oil ~ place) Loss of Miscibility** Location in Layers 1-5 (distance from injector, feet) (layer) 1 -443 2 -450 3 - 45a 4-465 5 -495 1 2 3 4 5 1 2 3 4 5 :435 -465 - swept -304 - swept -383 - swept - 195 - swept -278

No. 1

252

63.8

(layer) 1-450 2-485 3-525 4 -608 5 - swept*** 1-383 2-428 3-450 4-518 5-548 1 - swept 2 - swept 3-233 4 -swept 5-428 1-443 2 - swept 3-263 4 -swept 5-345

28.8

No.2

252

63.7

23.9

No. 3

252

63.4

20.2

No. 4

252

62.6

15.2

1 -315 2 -473 3-338 4 - swept 5 -240


l~o above residual oilto waterflood.

*lhe location (feet from the injection well) where the waterflood left mobile oil at

#*~he location (feet from the injection well) where the surfactant slug lost miscibility with the oil and two-phaae flow began. #**Swept means the layer was swept clean of mobile oil.

TABLE4 - SENSITIVITY OF TERTIARY OIL RECOVERY


(Barrels Produced/Barrels Tertiary Oil In Place)

Standard Micellar Flood

4% Pv

12% Pv

Micellar
Flood

Mcellsr
Fkcd

High
Surfactant Leas

Low Surfs.xant Loo,

0.10

0.025

0.001

Critical fiaturation

Critical Saturation

Critical Saturation

Reservoir No. 1

Reservoir

No. 2

Reservoir No. 3

Reservoir No. 4

*Lower

f$gureg are percent change when compared to standard rnicellar flood in a particulm

zeaervoir.

S TABLE5 - SENSITIVITY OF TERTIARY OIL PRODUCEDISURFACTANTINJECTED (B!)l Bbls)

Standa d
Micellar

Flood
t Reservoir No. 1

4% Pv Micellar Flood

12% Pv Micellez Flood

High Surfactant Loee

Low Surf~ctant Loss

0.10
Critical Saturation

0.025
Critical Saturation 3.45

0.001
Critical Saturation 3,56

I 328 ~,. 13X


2*7 x 2.67 / 2~3% 1.46 7 -

+5%

/ 3.25

+9% + 0%

Reservoir

No. 2

z,% :X6%
/

.93

+0%

/ 2.40

2.37

Reservoir

No. 3

2:30

I
Reeervoir No. 4
~ *70 z% X%.zo z% z%

/
1.75

+ 3%

/
2.10

+4%

+3% reservoir.

/ :...

+ 24%

*Lower

figures are percent changewhen compared to standard micella r flood in a particular

-o

~...
0.25
0.50 0.75

1.0
curves
INTERGRID SLOCK MACROSCOPIC DISPERSION (MiXINGOUETO GROSS FLUID MOVSMSNT)

.i

PREFLUSH
Fig. I vs pref
SCALE: WELL TO WELL

EFFICIENCY

- Surfactant lush efficiency.

loss

INTRAGRID BLOCK

EIMULATOR

TOTAL SIMULATOR DISPERSION

NUMERICAL DISPERSION

FORCS THIS SOUALITY

MATCH WITH PROPER AX At

APPROX_lMATSLY EQUAL

REAL RESERVOIR

TOIAt PHYSICAL DISPERSION

MICROSCOPIC DISPERSION

MACROSCOPIC OISPSRSION (MIXING OUE1O GROS5 FLUID MCtVEMSNT)

Fig. 2 - Relationships of dispersion in the simulator real reservoir.

longitudinal and in

the

Fig. 3 a quarter

Variable AY five spot.

configuration

of

1.0 r

..-

... .

--

..- --- ..-_ 1

(o) ADDITIONAL VOLUME ADDED TO INTEGRAL WHEN CUTOFF IS 0.01

SURFACTANT SLUG

I BRINE SAT. = 0.05/ -+ 0 X, DISTANCE IN DIRECTION

CUTOFF SURFACTANT . OF FLOW . _____ . .

Fig. 4preflush

Definition study.

of

overlap

region

in

o RESERVOIR o RESERVOIR v RESERVOIR & RESERVOIR

#1 #2 #3 #4

/ .

\ 1 I 2 # 4 6 VOLUME 8 PREFLUSH

-+4 f+p++10 12 INJECTED (% PV)

Fig. four
1.0
I

5 - Resul ts reservoirs.
I

of
I
I

the
I

pref
I

lush
I

study
I
I

in
I
I

the

o RESERVOIR o RESERVOIR A RESERVOIR

#1 #2 #4

o -0

~-

0.2
PORE

0.4
VOLUMES

0.6
OF FLUID

0.8

l.O,

1.2
AFTER

I*4

1.6

INJECTED

PREFLUSH

Fig.

6 - Standard

micel

Iar

floods.

m s 0.8 * ~ 0,6 : w ;

G z

1.0.
o

I
STANDARD MICELLAR (6.13% PV SLUG) 0 12% PV SLUG A 4% PV SLUG FLOOD

I
c ~
G 0.8 c ~
q

o STANDARO MICELLAR FLOOD o HIGH SURFACTANI LOSS A LOW SURFACTANT LOSS

~ 0.6 g u : 0.4 0
q

0.40 * S 0.2; o 0 0.2 0.4


slug

~ 0,2 -

(SEE FIGURE IFORsuRFACTANT

LOSS

0.6

Fig. to surfactant

PORE VOLUMES 7A - I - homogeneous

1.0 OF FLUID INJECTED AFTER PRE.

0.8

1.2

1 .4
H

1 L6

& o 0 0.2 0.4 0.6 0.S 1.0 1.2 1.4 PORE VOLUMES OF FLUID INJECTED AFTER PREFLUSH 1.6

reservoir

sens

size.

Fig, 7B - I - homogeneous to the amount of surfactant

reservoir loss.

sensitivity

1.0 -:
k o ;
u

1.0 q
z o

~ ~ In b ~ > ~ u u * o * z ~ w %

0.8 -

STANDARD MICELLAR FLOOO(O.05) ~ 0.10 CRITICAL SATURATION A 0.02S CRITICAL SATURATION v O.001 CRITICAL SATURATION

w G 0.8 =

STANDARD MICELLAR FLOGD (6.13% PV SLUG) ~ 12% PV SLUG A 4% PV SLUG

0.6 \ 0.4 -

*
0.2 :0.2 F= * g I [ I 0.2 0.4 0.6 0.8 1.0 1.2 1.4 PORE VOLUMES OF FLUID INJECTED AFTER PRSFLUSH 7C I 1 1.6 . Fig. -

o0

Fig.

to

critical

Homogeneous ressrvoir slug saturation.


-

sensitivity

1 I I 1 I I I 0.2 0.4 0.6 0.8 1,0 1,2 1.4 1.6 PORE VOLUMES OF FLUID INJECTED AFTER PREFLUSH BA - 2 - Randoml heterogeneous reservoir

sensitivity

to

surfac

%ant

slug

size.

1.0
*

1.0r

~ 0.8 .5 w * ~ 0.6 * ~ w * 0.4 -

s w

o STANDARO MICELLAR FLOOO o HIGH SU,RFACTANT LOSS a LOW SURFACIANT LOSS

5
w u 0.8 ~ III * ~ 0.6 * o ~ : 0.4 ://;

Q G A v

STAhU3AR0 MICSLLAR FLOOD (0.0S) O.IO CRITICAL SATURATION 0.02S CRITICAL SATURATION 0.001 CRITICAL SATURATION

* o * z 0.2 s ~
w o~ 0 1

[SEE FIGURE I FOR SURFACTANT 1 I 0.2 0.4 0.6 0.8 1.0 1.2 1.4 PORE VOLUMES OF FLUID INJECTSD AFTER PREFLUSH

LOSS 1 1.6

5 * :0.2 ~ ~ o

Fig. 8B - 2 - Randomly heterogeneous reservoir sensitivity to the amount of surfaotant loss.


1.0 5 ~ o I I I I ( , I I {

0.2 0.4 0.6 0.8 1.0 !!2 1.4 0 PORE VOLUMES OF FLUID INJECTED AFTER PREFLUSH Fig. 8C - 2 - Randomly heterogeneous reservoir sensitivity to critical slug saturdon.

1.6

~ 0.8 ~ u w ~ 0.6 E :0.4 3 P :0.2 ~ s o 0 -

STANDARD MICELLAR FLOOD a 12ZPV SLUG a 4% PV SLUG

(6.13% PV SLUG)

STANDARD MICELLAR FLOOD HIGH SURFACTANT LOSS LOW SURFACTANT LOSS

Q (SEE FIGURE 1 FOR SURFACTANT LOSS CURVES)


1 1

0.2 0.4 0.6 0.8 l+o 1.2 1.4 PORE VOLUMSS OF FLUID INJECTED AFTER PREFLUSH

1,6

01 o

Fig. 9A - 3 - non-communicating sensitivity to surfactant slug

layered size.

reservoir

0.2 0.4 ?OREVOLUMES

0.6 0,8 1.0 1.2 1.4 OF FLUID INJECTED AFTER PREFLUSH

1.6

Fig. 9B sensitivity

3 - non-communicating layered to the amount of surfactant

reservoir loss.

o u a v

STANDARD MICELLAR FLOOD (0.0S) 0,10 CRITICAL SATURATION 0.025 CRITICAL SATURATION O.001 CRITICAL SATURATION 1
J

1.0 [
o ~ a v

!i w
&

c 0.8 =

STANDARD MICELLAR FLOOD (0.05) 0.10 CRITICAL SATURATION 0.02S CRITICAL SATURATION 0.001 CRITICAL SATURATION

91

-1
I

g
w

0.2 0.4 0.6 0.8 1.0 1.2 1.4 PORE VOLUMES OF FLUID INJECISD AFTER PRSFLUSH g, 9C - 3-non-communicating layered reservoir ssitivity to critical slug saturation.

1.6

:0.4 5 * w *.0.2 = m E o 0

PORE VOLUMES OF FLUID INJECTED AFTER PREFLUSH Fig, IOA - 4 - ~artially communicating layered reservoir sensitivity to sdrfactant slug size.

Lfd
0.2 0.4 0.6 0.8 1.0
7 A

J?----

1.2

1.4

1.6

I I

o STANDARD MICELLAR FLOOD(6.137S u 12% Pv SLUG & /4~PVSLUE3

PV SLUG) I

(0~~:---5 0,8 = :

RESERVOIR RESERVOIR

~ 0.6 > t

7 (SEE FIGURE I FOR THE HIGH SURFACTANT

#3 #4

LOSS CIIRVE) I

0.2 0.4 0.6 0.8 1.0 1.2 1.4 PORE VOLUMES OF FLUID INJECTED AFTER PREFLUSH IOB - 4 - partially communicating layered -rvoir sensitivity to the amount o? surfactant
3.

1.6

0,2 0.4 0.6 1.0 1.2 1.4 1.6 0.8 PORE VOLUMES OF FLUID INJECTEO AFTER PREFLUSH Fig. IOC - 4 - partially communicating layered reservoir sensitivity to critical slug saturation.

2<: ~LA ~
3000 I I
CUM. PROB.

, ) I !
k 20s 340 500 740 1220

1-

0 STANDARD MICELLAR FLOOD o HIGH SURFACTANT LOSS A LOW SURFACTANT 10SS

16.6 33.3
50.0 66.7 83.3

-1
o~ &o

I v= /M

kD-k ko

FOR SURFACTANT LOSSCURVESI 1 1 1 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 PORE VOLUMES OF FtUID INJECTED AFIER PREFLUSH

(SEE FIOURSI

,oL&&EYYl
1
10 50 80 90 99 CUMULATIVE PROBABILITY DISTRIBUTION five permeabil ities

3.
=covery

1! -

Influence from the

of four

high surfactant reservoirs.

loss

on

Fig.
from

B-1

- The selection of a log-normal distribution.

Вам также может понравиться