Favorable Attributesof Alkaline-Surfactant-Polymer FloodingShunhua Liu, SPE, Rice University; Danhua Leslie Zhang, SPE, Intertek Westport; Wei Yan, SPE, Bechtel; and

Maura Puerto, George J. Hirasaki, SPE, and Clarence A. Miller, SPE, Rice University

SummaryA laboratory study of the alkaline-surfactant-polymer (ASP) pro-cess was conducted. It was found from phase-behavior studies thatfor a given synthetic surfactant and crude oil containing naph-thenic acids, optimal salinity depends only on the ratio of themoles of soap formed from the acids to the moles of syntheticsurfactant present. Adsorption of anionic surfactants on carbonatesurfaces is reduced substantially by sodium carbonate, but not bysodium hydroxide. The magnitude of the reduction with sodiumcarbonate decreases with increasing salinity.

Particular attention was given to a surfactant blend of a pro-poxylated sulfate having a slightly branched C16–17 hydrocarbonchain and an internal olefin sulfonate. In contrast to alkyl/arylsulfonates previously considered for EOR, alkaline solutions ofthis blend containing neither alcohol nor oil were single-phasemicellar solutions at all salinities up to approximately optimalsalinity with representative oils. Phase behavior with a west Texascrude oil at ambient temperature in the absence of alcohol wasunusual in that colloidal material, perhaps another microemulsionhaving a higher soap content, was dispersed in the lower-phasemicroemulsion. Low interfacial tensions existed with the excess oilphase only when this material was present in sufficient amount inthe spinning-drop device. Some birefringence was observed nearand above optimal conditions. While this phase behavior is some-what different from the conventional Winsor phase sequence,overall solubilization of oil and brine for this system was high,leading to low interfacial tensions over a wide salinity range and toexcellent oil recovery in both dolomite and silica sandpacks. Thesandpack experiments were performed with surfactant concentra-tions as low as 0.2 wt% and at a salinity well below optimal for theinjected surfactant. It was necessary that sufficient polymer bepresent to provide adequate mobility control, and that salinity bebelow the value at which phase separation occurred in the poly-mer/surfactant solution.

A 1D simulator was developed to model the process. By cal-culating transport of soap formed from the crude oil and injectedsurfactant separately, it showed that injection below optimal sa-linity was successful because a gradient in local soap-to-surfactantratio developed during the process. This gradient increases robust-ness of the process in a manner similar to that of a salinity gradientin a conventional surfactant process. Predictions of the simulatorwere in excellent agreement with the sandpack results.

BackgroundAlthough both injection of surfactants and injection of alkalinesolutions to convert naturally occurring naphthenic acids in crudeoils to soaps have long been suggested as methods to increase oilrecovery, key concepts such as the need to achieve ultralow inter-facial tensions and the means for doing so using microemulsionswere not clarified until a period of intensive research betweenapproximately 1960 and 1985 (Reed and Healy 1977; Miller andQutubuddin 1987; Lake 1989). Most of the work during that periodwas directed toward developing micellar-polymer processes to re-cover residual oil from sandstone formations using anionic surfac-

tants. However, Nelson et al. (1984) recognized that in most casesthe soaps formed by injecting alkali would not be at the “optimal”conditions needed to achieve low tensions. They proposed that arelatively small amount of a suitable surfactant be injected with thealkali so that the surfactant/soap mixture would be optimal atreservoir conditions. With polymer added for mobility control, theprocess would be an alkaline-surfactant-polymer (ASP) flood. Theuse of alkali also reduces adsorption of anionic surfactants onsandstones because the high pH reverses the charge of the posi-tively charged clay sites where adsorption occurs.

The initial portion of a Shell field test, which did not usepolymer, demonstated that residual oil could be displaced by analkaline-surfactant process (Falls et al. 1994). Several ASP fieldprojects have been conducted with some success in recent years inthe US (Vargo et al. 2000; Wyatt et al. 2002). Pilot ASP tests inChina have recovered more than 20% OOIP in some cases, but theprocess has not yet been applied there on a large scale (Changet al. 2006).

Nevertheless, important aspects of the ASP process remainincompletely understood. One, which we address in this paper, ishow to make surfactant and soap propagate in such a way as tomaintain ultralow tension. Another is that application of surfactantprocesses, including ASP, to carbonate reservoirs received littleattention until the last few years, probably owing to concern thatthe main surfactants being considered (e.g., alkyl/aryl sulfonates)would form calcium and magnesium sulfonates that would eitherprecipitate or partition into the oil phase. An exception was thework of Adams and Schievelbein (1987), who found from labo-ratory experiments and two well pair tests that oil could be dis-placed in a carbonate reservoir using a mixture of petroleum sul-fonates and alkyl ether sulfates or alkyl/aryl ether sulfates.

Carbonate reservoirs are typically mixed-wet and often frac-tured as well. As a result, considerable effort has been directedtoward promoting spontaneous imbibition of aqueous solutions byadding surfactants to alter wettability. Standnes and Austad (2003)suggested using cationic surfactants to promote desorption of acidsfrom carbonate rock surfaces, making the rock more water-wet.Others have investigated the effectiveness of various surfactants inaltering wettability (Spinler et al. 2000; Chen et al. 2001; Xie et al.2005). Hirasaki and Zhang (2004) showed that if sodium carbonatewas used as the alkali and injected with a suitable anionic surfac-tant, the usual positive charge of carbonate rocks could be re-versed, with the result that surfactant adsorption was greatly de-creased and oil-wet surfaces were modified to intermediate wet.They found that a significant amount of oil could be recoveredfrom an initially mixed-wet carbonate core placed in an imbibitioncell containing a suitable aqueous solution of surfactant, sodiumcarbonate, and sodium chloride. The wettability modification al-lowed the surfactant solution to enter, and the resulting low inter-facial tension reduced capillary forces to the point that oil rose tothe top of the core, where it was released. No oil was recoveredwhen a sodium chloride solution was used instead. They proposedthat such alkaline-surfactant solutions could be injected into frac-tured carbonate formations to increase recovery. Mohanty and co-workers (Seethepali et al. 2004; Adibhatla et al. 2005) pursued asimilar approach experimentally and performed simulations ofhow such a process would perform for field conditions.

Scope of PaperIn this paper, we present laboratory results on phase behavior,interfacial tensions, adsorption, and oil displacement in unconsoli-

Copyright © 2008 Society of Petroleum Engineers

This paper (SPE 99744) was accepted for presentation at the 2006 SPE/DOE Symposiumon Improved Oil Recovery, Tulsa, 22–26 April, and revised for publication. Original manu-script received for review 20 February 2006. Revised manuscript received 2 May 2007.Paper peer approved 16 May 2007.

5March 2008 SPE Journal

dated sandpacks for an ASP process using a novel surfactant, apropoxylated sulfate having a slightly branched C16–17 hydrocar-bon chain. The propylene oxide provides increased tolerance tosalt and divalent ions. The branched hydrocarbon chain, recentlydeveloped by Shell Chemical with Procter and Gamble (Scheibel2004), mitigates the formation of liquid crystalline phases in theabsence of oil, thereby permitting surfactant injection as a single-phase micellar solution at ambient temperature with no added al-cohol or oil below or at optimal salinity for many oils. In contrast,alkyl/aryl sulfonates considered previously for EOR exhibited liq-uid crystal formation well below and at optimal salinity in theabsence of substantial amounts of alcohol or oil. Moreover, thelong hydrocarbon chain and the attached PO chain allow this newsurfactant to have high oil solubilization and hence low interfacialtensions over a wide range of conditions. It is studied in combi-nation with an internal olefin sulfonate (IOS-15-18), which is morehydrophilic and can be used to adjust optimal salinity of the mix-ture, and with sodium carbonate as the alkali and partially hydro-lyzed polyacrylamide as the polymer. Recovery of 98% of residualcrude oil having a viscosity of 19 cp was observed in a verticaldolomite sandpack at ambient temperature using a surfactant slugof 0.5 PV containing 0.2 wt% surfactant. Similar excellent recov-ery was also seen in a silica sandpack. A 1D finite-differencesimulator, which tracked transport of soap and injected surfactantseparately, gave predictions in good agreement with the experi-mental results in both cases. A gradient in soap-to-surfactant ratioin the direction of flow develops with ultralow tensions beingachieved near the value of this ratio corresponding to optimalconditions at formation salinity. This gradient acts to make theprocess more robust in a manner similar to a salinity gradient in aconventional surfactant EOR process. The simulations indicatethat another important factor leading to the high recovery observedin the sandpack experiments is the unusually wide range of con-ditions for which low interfacial tensions were found in the labo-ratory experiments.

A process using mixtures of the same two surfactants and thesame polymer but with added alcohol and no alkali has been de-veloped in parallel by Levitt et al. (2006) for recovery of anothercrude oil from dolomite cores.

ExperimentalCrude oil samples MY3 and MY4 from a west Texas field wereused. Both came from the same well, and have very similar prop-erties. API gravity is 28, and acid number is reported to be 0.20 mgKOH/g. Water was deionized with a conductivity of 4–7 mS/cm.Anhydrous sodium carbonate powder (99.8% purity with 0.005%calcium), NaCl (certified for biological work), and CaCl2�2H2O(ACS certified) were all obtained from Fisher Scientific. A par-tially (25–30%) hydrolyzed polyacrylamide Flopaam 3330S with amolecular weight of approximately 8 million was obtained fromSNF Floerger. The two surfactants studied were ammonium C16–17

7PO sulfate (hereafter shortened to N67) supplied by Stepan andsodium C15–18 internal olefin sulfonate (hereafter IOS) supplied byShell. Table 1 lists key abbreviations used in this paper.

Surfactant adsorption was measured on calcite and dolomite.Calcite powder (SOCAL31) was acquired from Solvay Perfor-mance Chemicals. Its surface area determined by BET adsorption

was 17.9 m2/g. Dolomite powder (Carl Pool Products) was alsoused for some experiments. Surface area was 1.7 m2/g. The initialsurfactant concentration in aqueous solution was fixed, and calciteor dolomite was added in increments with equilibrium surfactantconcentration determined after each addition. Potentiometric titra-tion (Metrhom Titrino Model 716, from Brinkman Instruments)with hyamine 1622 (benzethonium chloride, from Gallord-Schlesinger Industries) was used for this purpose. The adsorptiondensity was calculated by mass balance.

Surfactant solutions and oils were mixed at a specific water/oilratio (WOR) in glass vials or pipettes. They were first shaken wellby hand for 1 minute, then mixed on a rotating shaker for 24 hours.Afterward, they were put in an upright position and allowed to settle.

Interfacial tensions were determined using a University ofTexas Model 300 Spinning Drop Tensiometer equipped with avideo camera (JEM6222, Javelin) and a video micrometer (VIA-100, Boecheler).

Dolomite sand (Unimin Corporation) and silica sand (US SilicaOttawa Foundry) were used for the sandpack experiments. Fluidswere injected with an ISCO injection pump Model 2350. A pres-sure transducer was placed at the inlet to determine the pres-sure drop across the sandpack, the outlet of which was at atmo-spheric pressure.

Phase Behavior of Oil-Free Surfactant SolutionsEffect of Salt. Fig. 1 shows the phase behavior at ambient tem-perature of aqueous solutions containing 3 wt% (active) of mix-tures of N67 and IOS as a function of added NaCl for a fixed 1wt% concentration of Na2CO3, but no alcohol at ambient tempera-ture. IOS exhibits precipitation above 4 wt% NaCl. In contrast, noprecipitation occurs and a surfactant-rich liquid phase less densethan brine forms as NaCl content increases for solutions containingN67 and its mixtures having up to 75% IOS. The line is drawn forconditions where separation of a second bulk phase was seen. Thesolutions appeared cloudy at salinities slightly below the line, in-dicating that some small droplets of the new phase were present.The phase transition occurs at higher salinities as IOS is initiallyadded to N67, reaching a maximum near 8 wt% NaCl for a 1/1blend. For the 4/1 mixture of N67/IOS, which we shall call the NIblend, it occurs at approximately 6 wt% NaCl. For N67 alone, butnot for the mixtures, some cloudiness was observed in initiallyclear solutions after several months even at low salinities (datapoints with additional +). The transition at high IOS contents be-tween formation of the second liquid phase and precipitation(dashed line in Fig. 1) was not studied, as these compositions areof little interest for purposes of the present paper.

The significance of Fig. 1 is that N67/IOS mixtures can beinjected as a single-phase micellar solution of relatively low vis-cosity at ambient temperature at salinities approaching or in somecases even exceeding optimal salinity for many oils without addi-tion of alcohol or oil, a situation rarely found for petroleum sul-fonate and synthetic sulfonate systems considered previously forEOR. Injection of a single-phase solution is important because

Fig. 1—Effect of added NaCl on phase behavior of 3 wt% solu-tions of N67/IOS mixtures containing 1 wt% Na2CO3. (See Table1 for the definition of N67 and IOS as well as for other abbre-viations in later figures.)

6 March 2008 SPE Journal

formation of precipitate, liquid crystal, or a second liquid phasecan lead to nonuniform distribution of injected material and non-uniform transport owing to phase trapping or different mobilitiesof coexisting phases.

For example, optimal conditions for the NI blend with 1 wt%Na2CO3 and MY4 crude oil occur at salinities up to approximately5 wt% NaCl (see Fig. 2, which is discussed later). Thus, the entireoptimal curve for NI blend is below the phase separation point ofFig. 1 for the oil-free solution. For comparison, optimal salinity inHealy and Reed’s classic study with a synthetic C12 orthoxylenesulfonate, tertiary amyl alcohol, and a mixture of refined oils oc-curred at 1.5 wt% NaCl (Reed and Healy 1977). However, theoil-free surfactant/alcohol/brine mixtures were dispersions of thelamellar liquid crystalline phase or a single lamellar phase forsolutions containing 0–4 wt% NaCl (Miller et al. 1986). The lamel-lar phase consists of multiple stacked surfactant bilayers. Thebranched nature of the N67 and IOS hydrocarbon chains is animportant factor in inhibiting liquid crystal formation. Straighterchains can pack more easily into compact surfactant aggregates,such as bilayers.

Separation of a surfactant-rich liquid on increasing salinity inthis anionic surfactant system is similar to that seen in nonionicsurfactant solutions as temperature is raised above the cloud point.In both cases, phase separation, which is thought to involve ag-gregation of or network formation by micelles, takes place as thesurfactant is made less hydrophilic.

Effects of Propylene Oxide, Calcium, and Polymer. In additionto the effect of hydrocarbon chain branching discussed previously,the presence of an ethylene oxide and/or propoxylene oxide chainin such anionic surfactant systems seems to promote formation ofa surfactant-rich liquid instead of lamellar phase. For instance, astraight-chain C12–13 ethoxylated sulfate having three ethylene ox-ide groups exhibited formation of a second liquid phase at 30°Cupon adding 12.5 wt% NaCl (Mori et al. 1990). Similar behaviorwas seen for a 1:1 blend of a similar surfactant and a propoxylatedsulfate having an iso-C13 hydrophobe (Zhang 2006) (TC blend, seeTable 1). Phase separation in this latter case occurred at 5% NaCl(with 1% Na2CO3 present). Optimal salinities for these surfactantswith representative oils are higher than the respective phase sepa-ration salinities. For instance, Fig. 2 shows an optimal salinity ofapproximately 12 wt% NaCl for the TC blend itself with MY4crude (i.e., at low soap-to-surfactant ratios). The TC blend could,in principle, be used in alkaline/surfactant processes for this crudeoil if injected at salinities where it remains a micellar solution, asshown by Hirasaki and Zhang (2004). However, the NI blend ispreferable in this case as its optimal salinity is not as far above thatof the naphthenic soaps but is still above field salinity. As a result,optimal conditions and hence low interfacial tensions are less sen-sitive to changes in the soap-to-surfactant ratio for the NI blend, asshown by the lower slope of its curve in Fig. 5.

As indicated previously,, such behavior is different from that ofpetroleum and other alkyl/aryl sulfonates, where addition of NaCltypically produced separation of the lamellar liquid crystalline

phase at salinities well below optimal conditions for representativeoils. Similar behavior with dispersions of the lamellar phase ap-pearing well below optimal salinity in the absence of alcohol andoil was also observed for an iso-C13 propoxylated, ethoxylatedsulfate (Ghosh 1985). This surfactant was studied by Exxon for usein high-salinity reservoirs. Apparently, the branched hydrophobesof N67 and IOS produced more resistance to formation of surfac-tant bilayers than the iso-C13 hydrophobe.

Fig. 3 shows behavior when CaCl2 is added to 0.5 wt% mix-tures of the two surfactants with 2 wt% NaCl present, but noNa2CO3. A solid precipitate is observed for IOS alone and for themixture containing 80% IOS for CaCl2 concentrations exceedingapproximately 0.1 and 0.25 wt%, respectively. Except when IOScontent falls below 10%, tolerance to calcium ions is approxi-mately 1 wt% CaCl2 for mixtures having higher contents of N67.For these mixtures, separation of a viscous, surfactant-rich phasedenser than brine occurs at high concentrations of CaCl2.

As in petroleum sulfonate systems, addition of a high molecularweight polymer can cause phase separation. Fig. 4 shows thatincreasing NaCl content from 2 to 4 wt% produces phase separa-tion in a solution containing 0.5 wt% NI blend, 1 wt% Na2CO3 and0.5 wt% Flopaam 3330S, a partially hydrolyzed polyacrylamide.Larger micelles are expected at the higher salinity, which promotesseparation into surfactant-rich and polymer-rich phases. Formationof molecular complexes between the anionic surfactant and an-ionic polymer are unlikely owing to electrical repulsion.

Phase Behavior With OilImportance of Soap-to-Surfactant Ratio. Fig. 5 shows variationof phase behavior with NaCl concentration for alcohol-free solu-tions containing 0.2 wt% (active) NI blend and 1 wt% Na2CO3

Fig. 2—Optimal salinity as a function of soap-surfactant ratiofor NI and TC surfactant blends with MY4 crude oil.

Fig. 3—Effect of added CaCl2 on phase behavior of 0.5 wt% solu-tions of N67/IOS mixtures containing 2 wt% NaCl but no Na2CO3.

Fig. 4—Phase separation caused by increasing NaCl content foraqueous solution of 0.5 wt% NI blend, 1 wt% Na2CO3, and 0.5wt% polymer.

7March 2008 SPE Journal

mixed with MY4 crude oil at a water-to-oil (WOR) ratio of 3 andstored at ambient temperature for 40 days. From the reported acidnumber of this crude oil (0.2 mg KOH/g), soap-to-surfactant ratiowas calculated to be 0.35, assuming that pH is high enough tosaponify virtually all the acid. Its appearance seems to be that of aconventional Winsor I, III, II microemulsion sequence (Reed andHealy 1977) with some remaining emulsion that has not yet coa-lesced. We argue in the following that the situation is in actualitymore complex. We will use the term optimal salinity to designatethat point at which the solubilization ratios (Vo/Vs) and (Vw/Vs) ofoil and water to surfactant are equal.

Results for several such salinity scans (Zhang 2006) showedthat, if the effect of the soap was ignored, optimal salinity was afunction of both surfactant concentration of the solution and WOR.Each salinity scan had a fixed soap-to-surfactant ratio. Fig. 2shows that results for the various scans can be collapsed onto asingle curve depicting optimal salinity as a function of soap-to-surfactant ratio for both the NI and TC blends. At large and smallvalues of this ratio, optimal salinity approaches those of the soapalone (approximately 0.5% NaCl) and surfactant alone (5% and12% NaCl for NI and TC blends respectively), both in addition to1 wt% Na2CO3. For the TC blend, data points at different valuesof WOR are explicitly shown, although both WOR and surfactantconcentration were also varied for the NI blend. The optimal sa-linity of the soap formed from a crude oil is usually below fieldsalinity. In this case, the surfactant chosen for an ASP flood shouldhave its optimal salinity above the field value. Such a choice willfacilitate development of a gradient in soap-to-surfactant ratiopassing through the low-tension region, as discussed next.

Special Features of Phase Behavior. Observation over time ofthe samples below optimal salinity in Fig. 5 revealed that drops orparticles of material less dense than the lower-phase microemul-sion rose to the oil-microemulsion interface over time. As a result,a thin layer of a colloidal dispersion formed, as shown in moredetail in Fig. 6 for a similar sample with 2 wt% NaCl, but with 23days settling. As demonstrated in the following section, the mate-rial in this colloidal dispersion is essential for achieving interfacialtensions (IFT) that are very low in the Winsor I region. It is notsimply a collection of drops having the same composition as thecrude oil that have not yet coalesced with the excess oil phase,because such drops would not affect IFT. The low density of thedispersed material suggests a higher ratio of oil to brine than in thelower phase. The volume of this dispersion increases with increas-ing salinity below optimal salinity for the scan at WOR�3 shown

in Fig. 5. Moreover, its volume was significantly greater for thesame surfactant concentration at WOR�1, which contained moresoap and less surfactant. This latter result suggests that the dis-persed material has a higher soap-to-surfactant ratio than the lowerphase and hence is more lipophilic than the lower phase, with thecapability of solubilizing more oil but less brine.

The nature of the colloidal dispersion discussed previously forunderoptimum samples remains under investigation. Its dispersedmaterial may be a second microemulsion, which, because it ismore lipophilic than the lower-phase microemulsion, would havea lower IFT than the lower-phase microemulsion with excess oil.The existence of two microemulsions in equilibrium is possible formixtures of surfactants very different in structure and in hydro-philic/lipophilic properties, as is the case here. Low IFT betweenthe microemulsion phases would contribute to ease of dispersionof one in the other. Such behavior represents a deviation from theclassical Winsor I behavior of a single microemulsion plus excessoil for underoptimum conditions.

As shown in the following, optimal salinity for the salt scan ofFig. 5, as defined by conditions for equal solubilization of oil andbrine, is approximately 3.5% NaCl. Fig. 7, the same scan as seenwhen the samples are placed between perpendicularly oriented

Fig. 7—Salinity scan of Fig. 5 viewed between perpendicularlyoriented polarizers.

Fig. 5—Salinity scan for 0.2% NI blend, 1% Na2CO3 with MY4crude oil for WOR=3 after settling time of 28 days at 25°C.x=wt.% NaCl. Fig. 6—View of dispersion region near interface for sample from

salinity scan of Fig. 5 except after 23 days settling.

8 March 2008 SPE Journal

sheets of polarizing material, shows that slightly below optimalsalinity at 3.2% NaCl the lower phase becomes birefringent. Nev-ertheless, it behaves as a Newtonian fluid with a low viscosity ofapproximately 1.1 mPa�s (1.1 cp), according to our measurementsusing a cone and plate viscometer. Its interface with oil is mobilewhen the pipette is tilted and IFT is ultralow, as discussed next.For overoptimum conditions, a thin birefringent layer exists abovethe excess brine. It is separated from the uppermost phase, whichis presumably an oil-continuous microemulsion, by a dispersion ofthe birefringent and microemulsion phases.

Equal oil and brine solubilization corresponds to conditionswhen surfactant films have no tendency to curve toward either anoil-continuous or brine-continuous configuration and thus is fa-vorable for formation of the lamellar liquid crystalline phase,which consists of many parallel surfactant bilayers and exhibitsbirefringence. In view of its low overall surfactant concentrationand viscosity, the birefringent region seen at 3.2% NaCl in Fig. 7is likely a dispersion of the lamellar phase in a water-continuousmicroemulsion. That the dispersion has not separated after 40 daysindicates that the two phases have nearly equal densities. At highersalinities, the thin birefringent region is probably a lamellar phase.Interfaces in these overoptimum samples remained mobile after40 days.

Coexistence of a lamellar phase and a microemulsion has pre-viously been observed at relatively low surfactant concentrationswhen alcohol concentration or temperature in certain microemul-sions containing anionic surfactant fell below values needed toobtain the usual Winsor phase behavior with no liquid crystal(Hackett and Miller 1988). The Winsor sequence can be obtainedin alkaline systems with N67 and MY4 by adding alcohol (Zhang2006), but interfacial tensions are expected to rise with increasingalcohol content.

Solubilization Ratios. A solubilization ratio (Vo/Vs) can be cal-culated as the ratio of the volume of solubilized oil to the volumeof surfactant present (excluding soap) for the underoptimumsamples. The volume of solubilized oil is the difference betweenthe volume of the initial oil in the sample and that of the excess oilphase after equilibration. It represents a composite value for thecombined lower phase microemulsion and surfactant-containingphase in the dispersion. As Fig. 8 shows, values of (Vo/Vs) increasefrom approximately 7 at 2% NaCl to approximately 20 at 3.4%NaCl, just below optimal salinity. Similar values were found for ascan with 0.5 wt% NI blend, except that the salinities were higherwith optimal salinity of approximately 4.5% NaCl because thesoap-to-surfactant ratio was smaller. Above optimal salinity, asimilar procedure can be used to obtain an overall (Vw/Vs), asshown in Fig. 8. At salinities well below or above optimal, thecombination of low solubilization and some emulsion precludeddetermination of solubilization ratios from phase volumes.

Huh (1979) proposed for classical Winsor III behavior thatsolubilization ratios for the microemulsion phase and interfacialtensions of the microemulsion with excess oil and brine phaseswere related as follows:

�mo = c��Vo �Vs�2, �mw = c��Vw �Vs�

2, . . . . . . . . . . . . . . . . (1)

where c is a constant for each system. He found that results avail-able for EOR systems at that time were consistent with theseexpressions for values of c near 0.3 mN/m. Using the first of theseequations with c�0.3 mN/m and with (Vo/Vs) from Fig. 8 toestimate IFT below optimal salinity and similarly for (Vw /Vs)above optimal salinity, we obtain the predicted behavior of IFTshown in Fig. 9. Near optimal salinity where the two curves of Fig.8 cross, an estimate of both solubilization ratios is 50, which wouldlead to the lowest tensions, as indicated by the dip in the curve inFig. 9 representing the correlation predictions. The experimentalinterfacial tensions shown are discussed in the next section.

We note that some surfactant-oil-brine systems form highlyviscous phases or emulsions, which would be unfavorable forEOR. As indicated both by the rather mobile nature of the inter-faces in scans such as that of Fig. 5 and by the results of thesandpack experiments described next, formation of highly viscousmaterial does not appear to be a problem in the present system withproper process design.

Interfacial TensionA key requirement of alkaline/surfactant processes is achievingultralow interfacial tensions. If such tensions correlate with opti-mal phase behavior and high solubilization as in previous EORwork, they must depend on the soap-to-surfactant ratio, accordingto Fig. 2. Because the amount of soap present for a given volumeof crude oil is dictated by its acid number, relatively low surfactantconcentrations are required if quantities of soap and surfactant areto be comparable for experiments such as those of the precedingsection. As a result, volumes of some surfactant-containing phasesare small and, as indicated previously, they sometimes occur asdispersions. These factors make it challenging to obtain reproduc-ible measurements of interfacial tension.

For the scan of Fig. 5 with 0.2 wt% NI blend, it was found thatbelow optimal salinity, measured tensions between the lower phaseand excess oil depended on the settling time between the end of themixing process and the sampling of the lower phase (well belowthe interface) and excess oil. Fig. 10 shows that for the samplecontaining 2 wt% NaCl, low tensions (i.e., those below 0.01 mN/m) were achieved only for settling times of approximately 4 hoursor less. These results suggest that low tensions were not obtainedonce most of the colloidal material dispersed in the lower phasehad risen to the vicinity of the interface and thus was not sampled.A protocol given in Appendix A was developed to assure thatenough of the dispersed material was initially present in the lower-phase sample to achieve low tensions but not so much as to ob-scure the oil drop during the spinning-drop measurement, when thedispersed material collects in the form of a cloud of small dropletsor particles near the larger drop of excess oil. An example of thelatter effect is shown in Fig. 11, where the oil drop at the far leftis almost invisible in the cloud. Some aspects of the protocol such

Fig. 8—Solubilization ratios for salinity scan of Fig. 5. Fig. 9—Interfacial tensions for salinity scan of Fig. 5 as pre-dicted from solubilization ratios of Fig. 7 and as measured ex-perimentally.

9March 2008 SPE Journal

as the rotation, settling, and pre-equilibration times of Steps 2, 3,and 6 are specific for this system, but the basic procedure shouldbe useful in other systems with similar behavior.

With this protocol, tensions below 0.01 mN/m were obtainedover a wide salinity range both above and below optimal for thescan containing 0.2 wt% NI blend, as shown in Fig. 12. Agreementwith the solubilization ratio correlation is good except at 3.6%–3.8% NaCl (Fig. 9), where measured tensions are higher but stilllow enough (near 0.001 mM/m) to mobilize oil and prevent trap-ping. If soap were included with the surfactant in calculating thesolubilization ratios, the latter would be somewhat smaller and IFTpredictions of the correlation would be somewhat larger and closerto the data in this salinity range.

Fig. 12 further emphasizes that low tensions are not seen if thedispersed material is absent, owing either to long settling times orto its complete removal after being concentrated near the axis ofthe tube in the spinning-drop device before adding an oil drop. Thetwo lower curves show that Step 6 of the protocol, pre-equilibration of oil drop and aqueous phase in the spinning droptube before spinning, may not be required to achieve the equilib-rium IFT value. However, the time of spinning required to reachthe same constant IFT can be much longer—as much as 8 hours—without Step 6 at the lower salinities (see Fig. 13).

For a salt scan identical to that of Fig. 5 but with no sodiumcarbonate present (and hence minimal soap), less colloidal disper-sion was seen. Moreover, IFT values were below 0.01 mN/m overa much narrower range of NaCl concentrations, as the comparisonof Fig. 14 shows. This result indicates that the wide range of lowtensions with alkali present is a consequence of formation of naph-thenic soaps.

AdsorptionFigs. 15 and 16 show adsorption isotherms of N67 and IOS oncalcite powder for solutions containing no NaCl and either 0% or

1% Na2CO3. Adsorption is approximately 0.0035 and 0.007 mmol/m2 in the plateau region for the respective surfactants in the ab-sence of Na2CO3, corresponding to 0.47 and 0.23 nm2/molecule,assuming a uniform monolayer. The latter number (for IOS) isclose to the area of a single straight hydrocarbon chain. Becausethis surfactant consists of a mixture of molecules with the sul-fonate group located at various places along the hydrocarbonchain, the adsorbed molecules are twin-tailed. As a result, theirarea per molecule should be approximately twice that of a singlechain, and it seems likely that an adsorbed bilayer exists in theplateau region. Bilayers often form for adsorption of surfac-tant ions on surfaces of opposite charge at concentrations nearand above the critical micelle concentration (CMC) because theyexpose a polar surface, which has a low free energy with theaqueous phase. The arrangement of adsorbed molecules for N67 isnot clear.

The plateau region for each surfactant represents concentra-tions well above the CMC, where micelle composition is constanteven for these commercial surfactants with numerous individualspecies because virtually all the surfactant is present in the mi-celles. The plateau region extends to lower surfactant concentra-tions for the more lipophilic N67 owing to its lower CMC.

Figs. 15 and 16 show the amount adsorbed per unit area be-cause it is a fundamental property and can be converted to theamount adsorbed for a rock sample of known mass by measuringthe pore area per unit mass of the sample. With the BET method,we found that the surface areas of two dolomite cores from a westTexas formation were 0.13 and 0.21 m2/g. Based on the average ofthese values and Fig. 15, adsorption of N67 for this formationwould be approximately 0.04 and 0.45 mg/g with and withoutsodium carbonate.

The most striking feature of Figs. 15 and 16 is the large reduc-tion in adsorption produced by addition of Na2CO3—more than anorder of magnitude for N67 and several-fold for IOS. This behav-ior is similar to that reported previously for the TC blend of an-

Fig. 11—View of cloud of dispersed material nearly obscuringdrop at far left, but not at right, during spinning-drop experiment.

Fig. 12—IFT for salinity scan of Fig. 5 with different settlingtimes and procedures.

Fig. 13—More rapid achievement of low IFT is possible whenStep 6 of the protocol (Appendix A) is followed.

Fig. 10—Dependence of IFT on settling time of sample at 2 wt%NaCl from salinity scan of Fig. 5.

10 March 2008 SPE Journal

ionic surfactants mentioned previously and occurs for the samereason: Carbonate is a potential determining ion, which reversesthe charge of the calcite surface from positive to negative, leadingto repulsion of anionic surfactant ions (Hirasaki and Zhang 2004).Hydroxide is not a potential determining ion for carbonate sur-faces, and using NaOH instead of Na2CO3 as the alkali producedno significant change in adsorption of the TC blend (Fig. 17).While the results of Fig. 17 are for dolomite powder, a series ofexperiments with the TC blend showed that the adsorption per unitarea was nearly the same for calcite and dolomite powders.

Fig. 18 shows adsorption isotherms of the NI blend on calciteat different NaCl concentrations with and without 1 wt% Na2CO3.In the absence of NaCl, adsorption is comparable to that of N67alone. As NaCl concentration increases, the beneficial effect ofNa

2CO3 is reduced, presumably because the screening effect of the

additional electrolyte decreases electrostatic repulsion betweensurfactant ions in solution and the calcite surface. Nevertheless,Na

2CO3 reduces adsorption by more than a factor of three for 3

wt% NaCl. Additional experiments showed that only 0.1–0.2 wt%Na2CO3 was sufficient to produce the same decreases in adsorp-tion as those shown in Fig. 18 for 1 wt% Na2CO3.

ASP Forced Displacement ExperimentsAn ASP process was conducted at ambient temperature using up-ward flow in a vertical, 35-darcy dolomite sandpack 1 ft long and1 in. in diameter. After the sandpack was saturated with 2 wt%NaCl brine, MY4 crude oil (viscosity 19 mPa�s or 19 cp) wasinjected. Then, the sandpack was aged in an oven at 60°C for 60hours to alter wettability of the dolomite. Waterflooding at ambienttemperature with 3.2 PV of 2% NaCl brine reduced oil saturationfrom 98% to 18%.

Next, the ASP slug (0.5 PV) containing 0.2 wt% NI blend, 1wt% Na2CO3, 2 wt% NaCl, and 0.5 wt% (5,000 ppm) Flopaam

3330S was injected at an interstitial velocity of 14 ft/day, followedby 1.0 PV of a drive with no surfactant or alkali but with the sameNaCl and polymer concentrations and approximately the sameviscosity (45 mPa�s or 45 cp) as the slug. The relatively highpolymer concentration and viscosity of both slug and drive wereused owing to the high viscosity of the crude oil. It bears emphasisthat the injected NaCl concentration of 2 wt% was well below theoptimal salinity of 5 wt% for the NI blend alone with this crudeoil (Fig. 2).

The two series of photos in Fig. 19 illustrate the formation andpropagation of the oil bank and the effluent samples. Fig. 20 showscumulative recovery of the waterflood residual oil. The oil bankbroke through at approximately 0.8 PV and surfactant at 0.99 PV(based on the first observation of a colored microemulsion con-taining solubilized oil instead of transparent brine in the effluent).The process recovered 98% of the residual oil with 80% as cleanoil. The small amount of oil solubilized in the aqueous effluent wasnot measured and hence not included in these numbers.

Fig. 21 shows that the pressure drop across the sandpack in-creased steadily during injection of the first pore volume of vis-cous slug and drive fluid, then remained nearly constant at 2.8 psi(19 kPa). That is, no highly viscous phases or emulsions developed.

A similar ASP flood was performed in a vertical silica sand-pack of the same dimensions. Permeability was 40 darcies, andless water was injected, so that remaining oil after waterfloodingwas 25% PV. Otherwise, the procedure and solution compositionswere the same as previously described, except that the sandpackwas not subjected to aging and the surfactant concentration in theslug was 0.5 wt% NI blend. Figs. 22 and 23 show photographs of

Fig. 15—Adsorption of N67 on calcite powder (17.85 m2/g) withand without 1 wt% Na2CO3 and with no NaCl.

Fig. 16—Adsorption of IOS on calcite powder (17.85 m2/g) withand without 1 wt% Na2CO3 and with no NaCl.

Fig. 17—Adsorption of TC blend on dolomite powder, compar-ing NaOH and Na2CO3 as source of alkali.

Fig. 14—IFT for salt scan of Fig. 5 with and without Na2CO3.

11March 2008 SPE Journal

the sandpack during the flood and cumulative oil recovery, respec-tively. Both development and propagation of the oil bank andrecovery of 98% of the oil remaining after waterflooding at ap-proximately 1.3 PV fluid injected are similar to the results for thedolomite sandpack, as is the pressure history (not shown). Surfac-tant broke through slightly earlier at 0.9 PV.

Another ASP flood in a similar silica sandpack was made withthe same procedure and compositions, except that NaCl content ofthe brine was 4 wt% instead of 2 wt%. As Fig. 24 shows, consid-erable oil traveled behind the oil bank, which was much less dis-tinct than in Fig. 22. Moreover, pressure drop rose to 25 psi (172kPa), nearly an order of magnitude greater than before Fig. 24. Thehigh pressure drop indicates that a highly viscous phase or emul-sion was present, presumably owing to separation of the surfac-tant slug into polymer-rich and surfactant-rich phases at this sa-linity (shown in Fig. 4). As Fig. 25 shows, the pressure drop rosesharply as the concentrated polymer phase was displaced by thepolymer drive.

SimulationA 1D, two-phase, multicomponent finite-difference simulator wasdeveloped to model the ASP process and was applied for floodingwith the NI blend. Naphthenic acids in the oil are assumed to becompletely converted to soap once the alkali front arrives. Alongthe optimal salinity curve of Fig. 2, the partition coefficient be-tween oil and water phases is taken as unity for both soap andsurfactant and interfacial tension is assumed to be 0.001 mN/m.The IFT data of Fig. 12, showing low IFTs over a wide range ofsalinities at a soap-to-surfactant ratio of 0.35, were used in gener-

ating the IFT contour plot of Fig. 26 (see Appendix B for details).Values of IFT from Fig. 26 at this soap-to-surfactant ratio shownin Fig. 12 are conservative at low (<2%) and high (>4%) NaClconcentrations, where solubilization ratios could not be determinedaccurately to confirm the low values of IFT measured. Both soapand surfactant are assumed to partition strongly into the aqueousand oil phases, respectively, at salinities somewhat below andabove optimal. Fractional-flow curves for oil and aqueous phaseswere calculated as a function of IFT and viscosities of the phases,taking into account changes in residual oil saturation and shape ofthe relative permeability curves which occur as IFT is reduced. Ahorizontal displacement was assumed, and surfactant adsorptionand longitudinal dispersion were incorporated into the simulator(Hirasaki et al. 2005). Simulation parameters used to obtain theresults described in the next paragraph are shown in Table 2. Theyare basically those for the ASP floods described previously. Fur-ther details on the simulations are given in Appendix B.

An important feature of the simulator is its ability to followtransport of injected surfactant and soap generated from the oilseparately, so that the ratio of their concentrations and hence IFT

Fig. 18—Adsorption of NI blend on calcite as a function of NaClcontent with and without 1 wt% Na2CO3.

Fig. 20—Measured and predicted cumulative oil recovery forASP flood in dolomite sandpack.

Fig. 21—Time-dependence of pressure drop during ASP floodin dolomite sandpack.

Fig. 19—Photos showing behavior during ASP flood of dolomitesandpack (flow upward).

12 March 2008 SPE Journal

can be calculated as a function of position and time. For instance,the three plots on the left side of Fig. 27 show soap and surfactantconcentrations, their ratio, IFT, and oil saturation as a function ofposition after injection of the 0.5 PV slug during the dolomitesandpack simulation. A gradient in the soap-to-surfactant ratioexists, with large values corresponding to overoptimum conditionsahead of the low-tension region and small values corresponding tounderoptimum conditions behind it. When ultralow tensions existover a broad range of conditions, as in Fig. 26, the region of suchtensions in Fig. 27 has a substantial length in the direction of flow.This region propagates along the sandpack, leaving very little re-sidual oil. Because any trapping behind the oil bank occurs forunderoptimum conditions, the particularly wide region of low IFTfor such conditions shown in Fig. 26 is important for achievinghigh oil recoveries.

Another simulation was performed with a hypothetical nar-rower low-IFT region. The same optimal contour as in Fig. 26 was

assumed with the same IFT of 0.001 mN/m. However, the contoursfor 0.01 mN/m for a soap-to-surfactant ratio of 0.35 were taken atNaCl concentrations of 3.4 and 3.6%, instead of 2.0 and 4.0% asin Fig. 26. As shown on the right side of Fig. 26, total oil recoveryis reduced from 95 to 62% of the waterflood residual oil eventhough the minimum interfacial tension is the same. The reason isthat once the soap-to-surfactant ratio drops below that correspond-ing to the minimum IFT, the soap begins to transfer from the oilinto the aqueous phase. If the latter travels faster than theoil, soap-to-surfactant ratio of the solution in contact with thetrailing oil falls rapidly, leading to a rise in IFT and oil trapping.If the region of ultralow tension is of considerable length, trappingis reduced.

Good mobility control is also essential for process success.Even with the broad region of ultralow tension of Fig. 26, recoveryfalls from 95 to 86% if viscosity of the aqueous phase decreasesfrom 40 to 24 mPa s (40 to 24 cp) for this viscous oil (19 mPa sor 19 cp).

Figs. 20 and 23 show that the simulator was able to match wellthe cumulative oil recovery curves of the sandpack experimentsdescribed above for both dolomite and silica sand. Parameterswere adjusted slightly from those of Table 1 as required to beconsistent with the conditions of the experiments (e.g., measuredviscosities of the fluids, surfactant concentration, and residual oilsaturation after waterflooding).

Conclusions1. Use of Na2CO3 as the alkali in ASP processes can substantially

reduce adsorption of anionic surfactants on carbonate surfaces,

Fig. 23—Measured and predicted cumulative oil recovery forASP flood in silica sandpack.

Fig. 24—Photos showing behavior during unsuccessful ASPflood of silica sandpack where phase separation caused bypolymer has occurred (flow upward).

Fig. 25—Time-dependence of pressure drop during unsuccess-ful ASP flood of silica sandpack where phase separationcaused by polymer has occurred.

Fig. 22—Photos showing behavior during ASP flood of silicasand pack (flow upward).

13March 2008 SPE Journal

especially at low salinities. No significant reduction was seenwhen NaOH was used instead.

2. For a given synthetic surfactant and crude oil containing naph-thenic acids, optimal salinity for alkaline conditions was foundto depend only on the soap-to-surfactant ratio for the range ofconditions studied.

3. Experiments at ambient temperature with a blend of a propoxy-lated sulfate with a slightly branched C16–17 hydrophobe and aninternal olefin sulfonate but with no alcohol have exhibited:a. Surfactant/alkali/brine solutions containing no oil that remain

single-phase micellar solutions with increasing salinity untilsalinity reaches or exceeds optimal for representative oils, incontrast to most surfactants such as alkyl/aryl sulfonates in-vestigated previously for EOR.

b. Phase behavior with a crude oil for alkaline conditions andlow-surfactant concentrations that has the appearance of theusual Winsor I, III, II sequence with some emulsion that hasnot yet coalesced. However, colloidal material, perhaps an-other microemulsion having a high soap content, is dispersedin the lower-phase microemulsion for underoptimum condi-tions. Solubilization parameters are high over a wide range ofsalinities. No highly viscous emulsions or gels were ob-served, although some birefringence was seen near and aboveoptimal salinity.

c. Interfacial tensions below 0.01 mN/m over a wide range ofsalinities both below and above optimal provided that enoughof the dispersed colloidal material is present in the sampleused for the measurement. The experimental range of IFTbelow 0.01 mN/m is in reasonable agreement with that pre-dicted by Huh’s correlation based on the measured solubili-zation ratios.

d. 98% recovery of residual oil in ASP sandpack experimentsfor both dolomite and silica sands with the polymer being apartially hydrolyzed polyacrylamide and with surfactant con-centrations in the slug as low as 0.2 wt%. Injection salinity

was well below optimal for the synthetic surfactant alone,thereby avoiding phase separation in the presence of polymer.

4. A 1D numerical simulator was developed to model the ASPprocess. It predicted oil recovery curves in agreement with thoseof the sandpack experiments. By calculating transport of in-jected surfactant and soap formed from the crude oil separately,the simulations indicate that a gradient in soap-to-surfactantratio develops with conditions shifting from overoptimum nearthe rear of the oil bank to underoptimum behind the main dis-placement front. The wide range of low tensions found for thesurfactant blend used was found to be an important factor con-tributing to the high recovery predictions.

Nomenclaturec � constant in Huh’s correlation

Ej � saturation exponent of phase j in relative permeabilitycalculation

K � partition coefficientko

rj � relative permeability of phase jko

ro � endpoint permeability of oilko

rw � endpoint permeability of waterSj � saturation of phase j

SjR � educed saturation of phase jS1r � residual saturation of aqueous phaseS2r � residual saturation of oleic phaseVo � volume of excess oleic phaseVs � volume of surfactant phaseVw � volume of excess water phase

�mo � solubilization ratio between microemulsion and excess oil�mw � solubilization ratio between microemulsion and excess water

AcknowledgmentsThis research was supported by DOE grant DE-FC26-03NT15406and by the Consortium on Processes in Porous Media at RiceUniversity. We gratefully acknowledge Shell Chemical and Stepanfor supplying surfactants, Gary Pope and members of his group atthe University of Texas for useful interaction, and Brent Biseda forperforming some of the adsorption experiments.

ReferencesAdams, W.T. and Schievelbein, V.H. 1987. Surfactant Flooding Carbonate

Reservoirs. SPERE 2 (4): 619–626. SPE-12686-PA. DOI: 10.2118/12686-PA.

Adibhatla, B., Sun, X., and Mohanty, K.K. 2005. Numerical Studies of OilProduction From Initially Oil-Wet Fracture Blocks by Surfactant BrineImbibition. Paper SPE 97687 presented at the SPE International Im-proved Oil Recovery Conference in Asia Pacific, Kuala Lumpur, 5–6December. DOI: 10.2118/97687-MS.

Chang, H.L. et al. 2006. Advances in Polymer Flooding and Alkaline/Surfactant/Polymer Processes as Developed and Applied in the Peo-ple’s Republic of China. JPT 58 (2): 84–89. DOI: 10.2118/ 89175-MS.

Chen, H.L., Lucas, L.R., Nogaret, L.A.D., Yang, H.D., and Kenyon, D.E.2001. Laboratory Monitoring of Surfactant Imbibition With Comput-erized Tomography. SPEREE 4 (1): 16–25. SPE-69197-PA. DOI:10.2118/ 69197-PA.

Fig. 26—IFT contour plot used in simulation based on measuredIFTs (mN/m) for NI blend.

14 March 2008 SPE Journal

Falls, A.H. et al. 1994. Field Test of Cosurfactant-Enhanced AlkalineFlooding. SPERE 9 (3): 217–223. SPE-24117-PA. DOI: 10.2118/24117-PA.

Ghosh, O. 1985. Liquid crystal to microemulsion transitions in anionicsurfactant-oil-brine systems. PhD dissertation, Houston: Rice Univer-sity.

Hackett, J.L. and Miller, C.A. 1988. Microemulsion to Liquid-CrystalTransition in Two Anionic Surfactant Systems. SPERE 3 (3): 791–800;Trans., AIME, 285. SPE-15653-PA. DOI: 10.2118/15653-PA.

Hirasaki, G.J. and Zhang, D.L. 2004. Surface Chemistry of Oil RecoveryFrom Fractured, Oil-Wet, Carbonate Formations. SPEJ 9 (2): 151–162.SPE-88365-PA. DOI: 10.2118/ 88365-PA.

Hirasaki, G.J. et al. Surfactant based enhanced oil recovery and foammobility control. Technical reports #4 (July 2005) and #5 (January2006) to DOE for Grant DE-FC26-03NT15406 to Rice University.

Huh, C. 1979. Interfacial tensions and solubilizing ability of a microemul-sion phase that coexists with oil and brine. J. Colloid Interface Sci. 71(2): 408–426. DOI: 10.1016/0021-9797(79)90249-2.

Lake, L.W. 1989. Enhanced Oil Recovery. Upper Saddle River, New Jer-sey: Prentice-Hall.

Levitt, D.B. et al. 2006. Identification and Evaluation of High-PerformanceEOR Surfactants. Paper SPE 100089 presented at the SPE/DOE Sym-posium on Improved Oil Recovery, Tulsa, 22–26 April. DOI: 10.2118/100089-MS.

Miller, C.A. and Qutubuddin, S. 1987. Enhanced oil recovery with micro-emulsions. In Interfacial Phenomena in Apolar Media, ed. H.F. Eickeand G.D. Parfitt, 117–185. New York City: Marcel Dekker.

Miller, C.A., Ghosh, O., and Benton, W.J. 1986. Behavior of dilute lamel-lar liquid-crystalline phases. Colloids Surf. 19 (2&3): 197–223. DOI:10.1016/0166-6622(86)80336-5.

Mori, F., Lim, J.C., and Miller, C.A. 1990. Equilibrium and dynamicbehavior of a system containing a mixture of anionic and nonionicsurfactants. Progr. Colloid Polym. Sci. 82: 114–121.

Nelson, R.C., Lawson, J.B., Thigpen, D.R., and Stegemeier, G.L. 1984.Cosurfactant-Enhanced Alkaline Flooding. Paper SPE 12672 presentedat the SPE Enhanced Oil Recovery Symposium, Tulsa, 15–18 April.DOI: 10.2118/12672-MS.

Pope, G.A. and Nelson, R.C. 1978. A Chemical Flooding CompositionalSimulator. SPEJ 18 (5): 339–354. SPE-6725-PA. DOI: 10.2118/ 6725-PA.

Reed, R.L. and Healy, R.N. 1977. Some physicochemical aspects of mi-croemulsion flooding: a review. In Improved Oil Recovery by Surfac-tant and Polymer Flooding, ed. D.O. Shah and R.S. Schechter, 383–437. New York City: Academic Press.

Scheibel, J.J. 2004. The evolution of anionic surfactant technology to meetthe requirements of the laundry detergent industry. J. Surfactants De-tergents 7 (4): 319–328. DOI: 10.1007/s11743-004-0317-7.

Seethepalli, A., Adibhatla, B., and Mohanty, K.K. 2004. PhysicochemicalInteractions During Surfactant Flooding of Fractured Carbonate Res-ervoirs. SPEJ 9 (4): 411–418. SPE 89423-PA. DOI: 10.2118/89423-PA.

Spinler, E.A., Zornes, D.R., Tobola, D.P., and Moradi-Araghi, A. 2000.Enhancement of Oil Recovery Using a Low Concentration of Surfac-tant To Improve Spontaneous and Forced Imbibition in Chalk. PaperSPE 59290 presented at the SPE/DOE Improved Oil Recovery Sym-posium, Tulsa, 3–5 April. DOI: 10.2118/59290-MS.

Standnes, D.C. and Austad, T. 2003. Wettability alteration in carbonates:interaction between cationic surfactant and carboxylates as a key factorin wettability alteration from oil-wet to water-wet conditions. Colloids,Surf. A 216 (1–3): 243–259. DOI: 10.1016/S0927-7757(02)00580-0.

Vargo, J. et al. 2000. Alkaline-Surfactant-Polymer Flooding of the Cam-bridge Minnelusa Field. SPEREE 3 (6): 552–558. SPE-68285-PA.DOI: 10.2118/68285-PA.

Wyatt, K., Pitts, M.J., and Surkalo, H. 2002. Mature Waterfloods RenewOil Production by Alkaline-Surfactant-Polymer Flooding. Paper SPE78711 presented at the SPE Eastern Regional Meeting, Lexington,Kentucky, 23–26 October. DOI: 10.2118/78711-MS.

Xie, X., Weiss, W.W., Tong, Z., and Morrow, N.R. 2005. Improved Re-covery From Carbonate Reservoirs by Chemical Stimulation. SPEJ 10(3): 276–285. SPE-89424-PA. DOI: 10.2118/ 89424-PA.

Zhang, D.L. 2006. Surfactant-enhanced oil recovery process for a frac-tured, oil-wet carbonate reservoir. PhD dissertation. Houston: RiceUniversity.

Appendix A—Protocol for IFT MeasurementsWith Dispersion Present

1. Mix the crude oil with the alkaline surfactant solution.2. Rotate the mixture for 24 hours.3. After settling the mixture for 4 hours, withdraw samples of

oleic and aqueous phases into different syringes.4. Because the samples in the syringes may continue to settle,

shake them before the spinning-drop measurement, so that theycan be considered as the same samples that were obtained after 4hours’ settling.

5. Before the spinning-drop measurement, centrifuge the aque-ous phase in the spinning tube. Then remove some of the dispersedmaterial by syringe, because the sample will be too dark if toomuch of it remains. The remaining dispersion should have a vol-ume slightly less than that of the excess oil drop that is subse-quently added to the spinning-drop tube.

6. Let the oil drop settle in the vertically-oriented tube for sometime (∼12 hours), so that the dispersed material can equilibratewith the oil and the lower-phase microemulsion.

7. Begin the spinning-drop measurement and measure IFT.

Appendix B—Additional Informationon SimulatorFractional Flow. The fractional-flow curve is a function of re-sidual saturations and relative permeabilities, which are deter-mined by IFT. For residual saturations, we calculate them from Eq.B-1, which is similar to the approach used by Pope and Nelson(1978):

S1r = 0, S2r = 0 when IFT � 0.005 mN�m

S1r = 0.3 * �1 + �log10�IFT��2.3��

S2r = 0.3 * �1 + �log10�IFT��2.3��

when 0.005 dyne�cm � IFT � 1 mN�m, . . . . . . . . . . . . . . . . (B-1)

Fig. 27—Spatial distribution of soap and surfactant concentra-tions; their ratio, IFT, and oil saturation after injection of 0.5 PVASP slug for wide and narrow regions of low IFT.

15March 2008 SPE Journal

S1r = 0.3, S2r = 0.3 when IFT � 1 mN�m,

where S1r is the residual saturation of aqueous phase, and S2r is theresidual saturation of oleic phase.

For the relative permeability, we calculate as follows:

SjR =�Sj − Sjr�

�1 − S1r − S2r�j = 1, 2 . . . . . . . . . . . . . . . . . . . . . . . (B-2)

krj = krjo SjR

Ej j = 1, 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (B-3)

kr1o = krw

o + �1 − krwo ��0.3 − S2r��0.3

krwo is endpoint permeability of water . . . . . . . . . . . . . . . . . . (B-4)

kr2o = kro

o + �1 − kroo ��0.3 − S1r��0.3

kroo is endpoint permeability of oil . . . . . . . . . . . . . . . . . . . . . (B-5)

Ej = 1.0 when IFT � 0.005 mN�m

Ej = 1.5 + 1�6 * log10�IFT�

when 0.005 mN�m � IFT � 1 mN�m . . . . . . . . . . . . . . . . . . . (B-6)

Ej = 1.5 when IFT � 1 mN�m

At high IFT, these parameters are for a water-wet system.

Interfacial Tension and Partitioning. Fig. 25 was constructed asfollows: Based on the lowest measured IFT in Fig. 12, IFT wastaken to be 0.001 mN/m along the entire curve for the NI blendgiven in Fig. 2, showing dependence of optimal salinity on soap/surfactant ratio. For a soap/surfactant ratio of 0.35, where optimalsalinity is approximately 3.5% NaCl, the point in the underopti-mum region on the IFT contour for 0.01 mN/m was estimated tobe at 2.0% NaCl, using the data of Fig. 12. This choice is conser-vative because IFT values near 0.01 mN/m were seen at evenlower salinities, according to Fig. 12. However, emulsions presentat these low salinities prevented confirmation of the low tensionsby solubilization ratio measurements. In the overoptimum region,IFT was taken to be 0.01 mN/m at 4.0% NaCl for the same soap/surfactant ratio. Here too, the choice is conservative because mea-sured IFT values were near 0.01 mN/m at higher salinities, butemulsions precluded estimates of solubilization parameter. For thesame soap/surfactant ratio, IFT was taken to be 0.1 mN/m at 0.5%and 5.0% NaCl. In view of Fig. 12 these choices are also conser-vative. Simulation results were insensitive to the exact location ofthe 0.1 mN/m IFT contours. Points on the IFT contours for othersoap/surfactant ratios were located by assuming that a particularIFT (e.g., 0.01 mN/m) was the same percentage below (or above)the optimal value as when the ratio was 0.35.

The partition coefficient K between oil and aqueous phases forthe soap/surfactant mixture was calculated by the following em-pirical method.

Above Optimal Salinity. K � 0.01 * 100R

where R =Salinity

Optimal Salinity

Below Optimal Salinity. K � 1�(0.01 * 100R)

where R =Optimal Salinity

Salinity.

Clearly, K�1 at optimal salinity as required. Recovery pre-dicted by the simulator was found to be relatively insensitive to thedetailed expression for calculating K except that numerical insta-bilities were encountered if K varied too rapidly near optimalconditions.

Numerical Details. Table 2 indicates that a value of dt/dx of 0.05was used. This value is the largest where stability could be main-tained during the finite-difference calculations with minimal nu-merical dispersion. For dt/dx�0.1, sometimes there was an insta-bility problem. Similarly, the choice of 100 gridblocks reflects thesmallest number for which the solution does not change signifi-cantly with an increasing number of gridblocks.

The surfactant breakthrough curve was determined for 1D non-adsorbing tracer experiments in dolomite sandpacks with a differ-ent but similar anionic surfactant mixture. When the simulator wasused to match the results, a Peclet number of 500 was obtained. Asimilar value was found for experiments where surfactant waspresent in addition to the tracer.

SI Metric Conversion Factorscp × 1.0* E–03 � Pa�spsi × 6.894 757 E+00 � kPa

*Conversion factor is exact.

Shunhua Liu is a PhD candidate in chemical engineering atRice University. His research interests include surfactant sci-ence, enhanced oil recovery, and numerical simulations. Heholds BS and MS degrees from Tsinghua University, China, bothin chemical engineering. Danhua Leslie Zhang joined IntertekWestport Technology Center in January 2007 after a year withHalliburton. Her research interests include formation evalua-tion, enhanced oil recovery, fracturing fluid formulation, andrheology. She holds a BS degree in chemical engineering fromTianjin University, China, and a PhD degree in chemical engi-neering from Rice University. Wei Yan held industrial positionsfor 7 years in China before coming to the US, and is now aprocess engineer with Bechtel (IPSI) in Houston. He holds a BSdegree in chemical engineering from Tianjin University, China,and a PhD degree in chemical engineering from Rice Univer-sity. Maura C. Puerto is a Complimentary Visiting Scholar atRice University. She retired from ExxonMobil as a Research As-sociate after more than 25 years of service in the ReservoirDivision. She is known as an expert in application of surfactantsto processes involving flow through porous media and is skilledin the area of designing laboratory coreflood testing and his-tory matching the results by simulation. Before joining Exxon in1974, she was employed in quality control and product devel-opment for Almay Hypoallergenic Cosmetics and was techni-cal director of the cosmetic branch at the Cuban Institute ofSoap and Cosmetics. In 1996, she received, based on her workon CO2 solubility in LVP solvents for aerosols, the Quality Awardfrom Exxon Chemicals for turning environmental regulationsinto opportunities. Since her retirement from Exxon, she hasparticipated on a part-time basis in research on enhanced oilrecovery and aquifer remediation at Rice. She holds a chemi-cal engineering degree from Oriente University in Cuba. Shewas the recipient in 1980 of the SPE Cedric K. Ferguson Award,George J. Hirasaki had a 26-year career with Shell Develop-ment and Shell Oil Companies before joining the ChemicalEngineering faculty at Rice University in 1993. At Shell, his re-search areas were reservoir simulation, enhanced oil recovery,and formation evaluation. At Rice, his research interests are inNMR well logging, reservoir wettability, enhanced oil recovery,gas hydrate recovery, asphaltene deposition, emulsion coa-lescence, and surfactant/foam aquifer remediation. He wasthe 1999 recipient of the Society of Core Analysts TechnicalAchievement Award, and is a member of the National Acad-emy of Engineers. He holds a BS degree in chemical engineer-ing from Lamar University and a PhD degree in chemical en-gineering from Rice University. He received the SPE Lester UrenAward in 1989, and was named an Improved Oil RecoveryPioneer at the 1998 SPE/DOR IOR Symposium. Clarence A.Miller is Louis Calder Professor of Chemical and BiomolecularEngineering at Rice University and a former chairman of thedepartment. Before coming to Rice University, he taught atCarnegie-Mellon University. He has been a Visiting Scholar atCambridge University, the University of Bayreuth (Germany),and the Delft University of Technology (Netherlands). His re-search interests center on emulsions, microemulsions, andfoams and their applications in detergency enhanced oil re-covery and aquifer remediation. He is coauthor of the bookInterfacial Phenomena, the second edition of which was re-cently published by CRC Press. He holds BS and PhD degreesfrom Rice University and the University of Minnesota, both inchemical engineering.

16 March 2008 SPE Journal