Vous êtes sur la page 1sur 8


Sp(!? 446

011Displacement Using EkrtiallyMiscible


Gas-Solvent Systems



Solvent /ioods using s~gs of solvent have been found to show continuity in behavior from tb vapor pressure of the solvent to tk crltfcal pressure for the two40mponent driving gas-solvent system. in the PIVSSUW don between the solvent vapor pressure and the critieaf pressure for the gas-solvent system, the gas and solvent are only partkdly miscible. Although compkte ?ni@bgify ~nnot be ~btained at these presstqes, complete Qil recovery is possible in principle. In two-phase solvent floods the solvent is propagated through the reservoir, primarily, in the vapor phase. The carrier gas requirements constitute a significant factor in the economicsoj the prams. A qualitative theory is PrOposed for estimating the amount of dry gas required to move. the solvent through the reservoir. The theory shows that for two-phare solvent j?iwds the total gas needed is a minimum at the vapor pressure of the solvent and at the critical pressure for the gas-solvent system, and is a maximum at somd intermediate presswe. The predictions of the theory are supprted, by expedmental studks using methane, butane and decane or methane, propane and decane in a naturalsandstone core.

theoretically, raaover all of the oll at soy pressure greater than the vapor pressure of the solvent,,.But the amount of sss reqtdred to move the solvent through the reservoir depends very much on the pressore and temperature. In the present study a maximum in the gas requirements was bcth predicted theoretically and observed experimentaUy. Ttds reatdt has not bean reported previously, and wotid not have been predictedfrom the Welge and Johnson model. The gas requirements are a minimum at the pressures eorrespending to the vdpor pressure of the solvent and for the gm-mlvent systemt agala at, the critical Prwand are a maximum at some intermediate pressure. AN APPROMIifATE THEORY FOR TWO-PHASE SOLVENT FLOODING The differences and similarities betw*n conventional solvent floods and two-phase solvent Iiecdi are best under- . stead by referring to eoncepta developi?dfor miscibledisplacement in which mkdbility is generttted in the reservoir. h Fig, l(A), a ternary CWTMM is ShOWn for a hm thetieal gas-solvent4 system. To be rigorous the three enmpmtents should each consist of a single molecuhw s~ies. The preefmre for Fig. 1(A) is greater than the critical pressure for the bb~ gas-solvent system at the specitkd temperature. Diagrams of this type are the ones mest frequently referred to in ,@cussions of enriched-gas drive and mkeible displacement. A limiting tie line is shown tangent to the gvo-phase envelope and intersecting the gas+wlventtine at Point A. To obtain generated ndwib~ty with Wi tYPe sYatemt others have shown that for an oil of Composition D, a mixture of sss and sclvent must be injeoted which is richer in solvent than that composition indicated by A. AO oil repeatedly contacted with a gas phase richer than A changes toward a eompesition which would beat equilibri~ with the fnjected mkt~, that is, a eompcsidon lying on a tie line which pthrough the injeoted-gas eompqsidou Since mo such tie line exists, the oil is enriched to the point at which it becomes directly tnkcible viith the injected mixture. At @essureslower then the critical pressure for the gasSOhtt wstem, other tYPCS of pha= @PMXM ~ Observed.The ones of interest in this paper are for pressures greater than the vapcr pressure of the solvent, but less than the critical pressure of the gas-solvent system Such a ternary diagramis shown Ori Fig. 1(B). In thk Q@% two-phase behavior is observed not only for gaadl mixtures, but also for mrtain compositions in the gesdvent

INTRODUCTION Previously, solvent slug procmses have been found effective for oil recovery in two pressure ranges. Fin% conventional miscible displacements are possible at pressuqs greater than the eritieal pressure for the gas-solvent system. Second, Yenka,et al,x ,have shown tha~ at pres SUMSslightly in exeess of the vapor pressure of the solvent, a solvent slug can be propagated tbmugh a r6servoir by a gas @sentiaUyinsoluble in the liquid solvent. Tl&sovent bfaik displaoea the oil ahettd of [t, Beth of .oeessas, at least ideally, am capable of -v*g. ~ of the oil in the swept regions. Slug processesfor which the gas and solvent are partially misciile have not been eensidere4 that is, those @stems for which the solvent phases in which and ddvins gas fOSOltWO eqadlibrium the vapor phase contains a sigtdikant amount of aclvent and the liquid phase an appreciable amount of the driving gas. Welge and Joh&m have shown that the. gas needed to move a solvent slug @rough the reservoir increasea with increasing pressure atie the vaper pressute of the aclvent. It wii be shown that $olvent s@g proeey$es oan,


mrmaceseiwllatcndti rover. ..:: -. .- -: --:


... .. ....... IWDUUAUY.X966-


:,, . .

c4@actewith tho injection gas. Once again the oil enriehea to a composition which lies on a tie line passing throssgh the composition of the in]edion gas. l%e M o@mWition of the @srichedoil must correspond to the Cornpositlon B because that is the Iispdd eomposStion on the tie tine throssgh A llse fntereating thing to note, however, is that the fksal wsriched-ofl phase, B, eontahss none of eonskmdon of tie the originsd oil. Thsw, by successive solvent, the original oil has been wmpletely dhplaced A bank of liquid of Com@tion B is developed, displaoirsg the oil ahead of it. AlthosAgh the oil can be completely displaced in theory by this process, complete ntiwibility in the region in which solvent is being transferred Is not achieved and ctmnot be achieved for the system. In practice, the solvent can be injected directly sw a liquid dug followed by dry.gaa injection. A twskphse region will develop within the formation containing a vapor phase of Composition A and a liquid phase of Corn. position B, One of the principaldifferences between twophase solvent dlsplacesuentsand miscible displacements hi in the volume of gas mqsdred to move the solvent through the reservoir. Ideally, miscible displacementsrecover msentissllyall the oil with 1 PV of injected fluid. Even ideally, this is not true for two-phase solvent floods. The amount of gas required depends among other things, on the @m. positions of the injection M A, aud the equilibrium liquid Jphase, B. These are functionsof pressure and temperature.

A detailed calcsd@ionof the displacement performance for two-phase solvent tloods is possible by the method of Welge, et al, for enriched-gas drives? Although those authors diszussedonly systems showing the phase behavior illustrated in Fig, 1(A), the method is applicable to systems of the type shownin Fig. 1(.B). However, theeffect of pressure and ~emperature on displacement behavior tith a given gas-solvent binary system can be undtwstii

more readily using the simple displacement model shown on Fig 2, TM figure shows the poetsdeted fluid distributions for a methane-n.butaedwme rystern, This system was seleoted for experimental work befwae the requisite PVT data have been tabsdated by f$ageand bscey$ In Fig. 2 it is assumed that a slug of Iiqufd butane has been injeeted into a linear porous medium originally 100 per cent saturated with decasse.The butane slug is followed by methane injection. The flsdd distributions shown are those expeded sometime after the beginning of gas injection. Oil has been completely displaced from Region A and only methane: gas is flowing. At the bosspdary between Regions A and B, the liquid bank is evaporating. Through Rt@m ~ the gas and liquid phasea are in eqsdlibrium, The vapor @ese is a &w-point mixture of snethahe and butane, and the liquid phase [s the corresponding bubble-point miishsrk.,At the bosmdsirybetween Regions B and D successivetmrichment of the decane occurs, the butane condensing thjm the vapor phase and displacing decane, According to the arguments g$ven @ier the decane digplacensentshould be complete, altho,ughin prm tice the boundary region will not be as sharp is shown in this idealized model. In Region D the methane is at equilibrium with the decane. Afthough in the foregoing example butane was assumed to be injected directly, a similar liquid bank of butane and methane could be visualized as developing if the injection mixture had, been the equilibrium m.ethane-butanedew-point vapor. TIm three principal assumptions in the model are: (1) equilibrium is obtained at any cross-section between the gas and liquid phases; (2) the boundaries between Regions A susd B and Regions B and D are s-, and (3) the transport of solvent in the ffquid plsaw fs negligibly small compared to the transport in the vapor phase. The Sirst of these assumptions appears from experiments to be valid at rates of advance of the solvent bank less than 2 to 3 ft/D. However, the total volume of gas requiredto move the solvent through the core appears to be an important factor in whether or not equilibrium is attained. The second and Udrd assumptionsrestrict the theory to the extent that solvent slug sizes and the corresponding oil recoveries cannot be calcukted from this simple model. The gas requirements can M estimated from the model, and this is animportant economio factor for a partially miscible gas-solvent displacement process.

\ ~


OAS c, + cm

A ~









Rc..i-!kxk+rmc. DIActi


. ..!- ,

ttEQUIREhtENTfi The total pore mhumeeof drivinggas requiredto move the solventbankthroughthe reservoircan be derivedfrom a materialbalanceat the leadingedge of the solventbank. U the leading edge of the bank advancesa distanw dl in time tit as shown in Fig, 2, a material balance can be written for the volume #Adl. Sep@e brdance$are requiredfor the gae and the eohwnt.The derivationis #ven in the Appendk The principal equiationfrom that derivation is the following,


where pv~v.d= volume of ,carrier gas produced at tie time of breakthrough of leading edge of solvent bank, = molar volqme of carriergas at the con. Vnll ditions for which the volume of prci ducedgas is measured, = molar vohune of Iiquid phase in solvent VML bank, & = liquid saturation in the solvent bank, and K. = equilibrium constant for the binary gassolvent system at reservoir pressure and temperature. For the simple model assumed in Fig. 2, the volume of gas produced at solvent-front arrival depends only on the PVT properties of the two.component system of gas and solvent, and not on those of the displaced OK For a two-mmponent system the effect of either pressure or temperature can be included on the phase diagram, A pressure-composition phase diagram is shown for the methane-n-butane system on Fig. 3. The compositions of the equilibrium liquid and vapqr phases are shown as functions of pressure at 73 F. These data were calculated from data of Sage and Lacey at other temperatures. fhe diagram shows that the compositions of the equilibrium phases are equal at 34 and 1,940 psia. Since K, equals the ratio of the molefraction of solvent in tlM vapor phase to that in the liquid, K, is 1.0 when these mole fractions are equal. When K, equals 1.0,

=++ .() . .. . . .


the number of pore volumes of gas produced equalI3zero. from 13q.L Therefore,the pore volumesof gas produced for the methan-n-butane system at 73 F windd be pr~ dieted to be zero at 34 PA (the vapor pressure of n-Mane) and at 1,940 PA (the critical pressurefor the methane-n-butane system). The number of pore volumm of methane requiredto move the leading edge of a butane bank through a linear rmervoir is shown as a function of pressure on Fig. 5. More important than the specific volumes shown is the trend. f& equations predict a minimum volume at the vapor prepmre of the solvent, a mmhnum at about 210 psia, and decreasing volumes up to the critical presauke. Experiments to be described subxquently confirmed this qualitative trend in methane requirements.







l! Ew


The objectives of the experimental work were to measure the oil resoveryby a slug processin two-phasesolvent flooding and to check the gas requirements as a function of pressure predicted from the equation developed previously.The fluids used were, primariiy, methane, n-butane and decane because PVT data were available for this system, Some experiments were made with propane instead of butane. A schematic diagram of the apparatus is shown on Fig. 4. The equipment was designed to operate at pressures up to 2,000 psia. The core was mounted in an epoxy plastic end inserted in a thick-walled steel tube. An anntdus pressure in excess of the maximum pressure within the core could W maintained on the plastic-mounted core. Pressure within the core was maintained by a Grove backpressure regulator at the etlluent end. Experiments were conducted at room temperature, about 73 F. The core was Berea sandstone, 5-ft long and 2-in. in diameter. Its permeability and porosity were 560 md and 21.7 per cent, respectively. The total pore volume was 656 cc. The core was 100 per cent saturated with oil at the beginning of be solvent flocids.No interstitial water was present iE tiny of the experiments, In some of the experiments with deoane, the decane was saturated with methane at the core pressure at the beginning of the run, For most of the runs, however, the decane contained no methane, Normslly, the butane or propane was injected in a dewpoint vapor, generated by bubbling the methane through the liquid solvent in the samplti bomb at the upstream end of the COm.The liquid volume of solvent used was usually 0.1 PV. The slug volume of enriched gas was, of course, much larger than tlds; Following ipjaction of the solvent slug, methane was injected until no more oil was recovered. TIM e.tiiueptfluids passed through a heater to separate the methane and solvent from the oiI phase, The oil was cdkted in a burette and measured. The separated gas phase was measured through a wet-test me~r. Samples of gas were collected periodically and analyzed for butane or propane using a gas chromatography: Rates were selected to advance the solvent slug from 1 to 2 ft/D. Rates greater than 2 to 3 ft/D apparently did not givechemical equilibrium between the gas and liquid phases. Oil recoveries were generally lower at the higher rat&s.

.0 0 ,2




W FWXION METHANE. ., Fgi..3-COMPQSXTION OFEQttUSRRM PrlAsmAs PorwrioriOF -Ptwamis tm Ma%aqI+BusxN8SY5@r M-73 F.

The primary variable in the experiments WApressure. For a-given gas-solvent systeti; ~e. qiuditative efk+ctof- ----. ..













~ II




pressure on gas requirements could be predicted from theory. Other variables were iovestigrited to a somewhat lesser extent. The measured pore volumes of methane required to move propane or n-butane through the core and recover the oil are shown on Fig, 5. The volume of methane is shown at core pressure at 73 F, The oil phase was decane. The effeet of pressure on the total amount of methane required is shown on Fig. 6 in which the volume of methaae is given at 1 atm, To obtain the total cumulative vohune of methane, the logarithm of the cumulative produced gas was plotted vs cumtdative oil production, as illustrated on Fig. 7. l%ese curves give a stisight-line portion which was extended to the final oil reriovery.The cumulative gas on the extended line at the total oil reeovery was taken as the total methane requirement. TMs procedure was found to be more reproducible than attempting to aeleet a methane volume from the final rounded portion of the curve, The results &ow satisfactory agreement with the computed volume of methane for the methane-butane system when the liquid saturation Sk is assumed to be 0.4. For the methane-propane, somewhat. better agreement would have been obtah.wdif SLwere assumedlarger. In general, however, the volume of methane required to move the solvent through the core exe@ed that oaleulated when the eideulated requirements were less than 4 to 5 PV. This was true for both the methane-butane and metbaneprQpanesystems.Po@ble reasons for this will M.diseu$sed later. Nevertheless, the effects of pressure and of solvent were in qualitative agreement with the predictions of the simple theory. PRO~uC~C wS.OIL @TXOS .-The -produ-hiog gti=o~lia&-a8 a&&a-hf &..: .. _:_ _. ..-..::. ,,. -:,; . . .. . , .oil heti


G C,+


lb iii


i E 0

s =

l!! 5! @ E 8



o 0



.,- ..:. ... . .. ,- ..-. -..: . 1 .,?: ,,,, ,. . .-: ... ..,. .. -,.. .- ...* .-:-.

------- ..:, ,..

IOKWM*-OF ~~RO=U=TEcENoL@Gy-.>. -.-..,


- --:,-.

furnidw acritical testofthemedelumd intheoalculatiom. Qas9ilratfm for the Metbe@Wtane sy@em are shown for core pressuresof 4S, 81S and 1,815 psia on Fig. 8. Gas vohunes were measured at atmospheric pressure. FrOOItbe model one would pmdiet a period of 033 production for which the producing ges.oil ratio would






(T=731? ANVS4-0.4).

be constant, but the ex@mentat rosuits show only a period of elowly inereasmg gas-oil ratios. 8M a reeidt would be entioipatedbeoause the boLlndary bEtweerl the solvent end$beoif wilfnotbeasehsr pasindieatedby the medeL The relative pmdtions of the GOR curves for the three pressuresare eons&tentwith the theory, l%e GOR twrve b lowcatfor the 45=pdarim, andM@est for tho 81S@a run. At 1,815 IX@,the results am between .Mse for the other pressures. An equilibrium-gas-drive curve is shown at 815 f.wda for comparison, one other factor was bn@ent for the GOR behavior. Most of ths experiments started with the core saturated with deoane, free of dhsolved gas, During the runs the decane became saturated with methane at the existing core pmrsure. Some dewne was produoed from swelling. Tb+ @oted the observed GOR curves. A @mparison ia shown on Fig. 9. of runs at 1,81S pda with equilibrium dwane and unsaturated decane. The methane-saturated deoane shows a more definite GOR plateau, ne total fraodon of original in-phwe decsne recovered was about the same whether or not the decane originally contained dissolved methane. AIao, the total volume of methane required to move the butane through the system was not changed aigoMoantly by the dissolved ,metbane in the dec%ne.Consequently, most of the runs were made, for . convenience, with dead deeane originally %-place. , SOLVENT RECOVERY The model of the displacementprocess shown on Fig,, 1 leads to tlw prediction that all of the oil would be pfoduced @fore the solvent. Figs, 10 and 11 show that this prediction is approximately correct. Fig. 10 shows oil production and butane eontemtof the produced gas as functions of the total gas produced at reservoirpressure. ) I I I I L -1 Kwx-

Moot ~ 6olfu 3WM : z * = s ~ EJ .~ * [OK ; 34$ SK,/, .2 ,, .6 ,, :





OIL REMvEITA - F,RACTIOS ORIGINAL STO FIG, 7-lLLUsTjUATION_OP .METSiOD l?OfiSTiXWNIN C~ltTXAliE EtSQUIiIED (P==@5 NO .I979)+ h. 8-EPmcT OF hssom ON 12AS.OIL RATIO AND OIL RseoYsaY sortMEZWS-BUTASW . .:- . . L rkueawarmi <T-73P). .. ---- . . . ... . :, ...-.--: .,. ...~.-, -... ,. .:-.


., .,..- :



: .. _-, . ...-,. , -.

, .,, ..,.. . . .. . .. . ..,.,..:,..,+. . ...

. ,.. . .

,-..:, .. . .

.m9. <.. :: . .. . . ,. . ., ..: . . ,+, ,f.. . .. . :, ---- . . . . .

. .... .


. - . ..

Fig. 11 shows the butane content of the produced gas as a function of oil reoovexy. The example Aown is for a run at 1,81S pda with a 0.2-W slug of butane.Total oil rewvery was 0.96 PV. Over 80 per a?nt of the original oil in place was produced before any substantialbutane productionwas obtained, and most of the butanewas recovered after 90 per cent of the oti hadbeen produced.
I Kmoo t




The maximum bntane cont&ntof the produm?d gas at atmosphericpressurewas 11.9 snol percent. IMeobservation is not isssonsistent with the pro@sed modaI. The butane oontent of a dew.pdnt gs8 ~t 1,81S @a is 13 mol per c+. 17Mproduced vapor phase would be expeoted to be at equilibriumwith the liquid phase at the end of the core. Therefore, the liquid phase must hsive had a eompmitionnot too ditlerentfrom that of a bubble-point mixture of methane and butane as prediotedfrom the model, although some of the butane could have been produced in a threeompmwnt liquid phase. For experiments with 0.1 PV of butane injested, the butane slugs were more deteriorated when finally produqzd. The maximum butane concentration in the produced gas was only about half that of a dew-point gas at core pressure. Df8CU8810N


3QOO0 -

i? -. : E? = ~ 5 & OEAD 1,000 ~ OECANE ~o~o .Illooo : METHANE SATURATED

300 -

100 o

I 2

I .4

I .6

I # 1.0




The results of &e two-phase solvent floods show that solvent flooding need not be restricted to reservoirs for which the pressures are greater than the critical pressure for the solvent-gassystem used in the solvent slug process, nor does it need be restricted to pressures ody slightly in exam of thevapor pressure of tbe solvent. Solvent slugs can recover, at least in tbcory, all of the oil from alinear system at any pressure greater than the vapor pressure tem@rature. But for pressures of the solvent at reservoir between the vapor pressure of the solvent and the critical pressure of the gas-solvent system, the gas and solvent are only partially miscible. CXmsequently,tnisoible displacement in the usual sense is not possible. A gas phase and a hydrocarbon liquid phase are both present in the region from which oil is being displaced. Nevertheless, experimental and theoretical results show that M@ oil recoveries are possible with reasonable quantities of solvent and gas. The principal &erence betw~n the two-phase displacement processes and the completely miscible methods 3s in the quantity of driving gas requiredto move the solvent through the reservoir. Ideally, the miseib!e processes recover all of the oil with 1 PV of injected solvent and gas. Even ideally, this is not true for the pa@dly misdble gas-











Fw 11hfoLsPNR C

. .
.... . . :;. :

AND SLUC VOLubfE=(i.~ PV).

Om. x


(P==MU%~Y@ F . . ... . .

: _ . .;-.,.
(. :...:~



_-, -. -:-


.,- . . . . .. . . . . ..-.

. ... . .


solvent displacpent method.The volumeof sss required has been ahown tode~on. theresmfok~ww. An -t factor in an economic analysis of the twophase solvent drive is, in additionto the Solventrequirements, the amount of gas requiredto move the solvent throughthe rwen@ir. For a mdume-butshe system at 73 F, the labomtoIY realts have shown that, except at pressuresaround 800 to Lm whk produc@ gas-oil ratiosdo~excecd tokraMeW for ~injeedon pmjeots.Wkh propaneor with butaneat normalreamvolr @mpemtur%the gas.oil ratios would not be excessive h a linear system for anYpressure in the two-phaseregion. lEo effects on calculatedgas requirements of temperature and pressure are shown for some exampk gas.solvent systems-in Table 1. fho assumed liqoid@vettt setumtion was 0.4, the same as that asumed for the ineth-butatte J wqnwiments.These ealculatioas show that with increasing , temperature Mgher-molectdar-wdghtsolventsgive behavior similar to that of lower-moleeukw-weight solvents at lower temperatures. fhus, at higher temperatures the ~ requirements for the metharw-butane system are about equal to these of the methne-propsne system at a lower temperature. Lower selubility of the gas in the solvent Mreases the pressure range for which two-phase solvent _ is possible.. For ttltrogen-butene at 100 F, the critical pm=ssureis about 4,200 psia. At 3,500 psia and 100 F, the nitrogen requirements are not exegesive for this system The data for the nitr@en-butane aysteip are astd Mcibtta! V~LWSSeatirnatd from those Of ~rta methmdmtane data. In praodcal applications, fhw gas could serve as a nitrogen source. The carbon dioxide in fluegas would not beexpectdtoafkt theprocas adversely. Table 1 is intended to show only the relative effects of pressure, temperatureand the gas-solvent.system on carriergas requirements. Experiments have shown that, for gas requirements 1*- than 4 PV at reservoirconditions, the amountof gas tends to QXCAthat predictedfrom the simple theory. . Anotherimportanteconomic factor in any 5eld prooess is the solvbnt slug size. The theory does not predictslug sizes, but the core floods gave high oil recoveries with about,10 per cent of a hydrocarbon pore volumeof liquid solvent. If one assumesthat the advent=oilmixing re@on grows in direct prppmtionto the length of travel,the slug volume determinedin gore tests should be applicableto Iifloods in the deld. A linear rate of growth for the mixing zone for two-phasedisplacementsis supportedby Welgesanalysis of enriched-gasdrive? 4 possible X%ntage of two-phasesolvent floods may be that the solvent is more uidformlydistributedthen it is for single-phase miscible displacements. Z& solvent moves mostly in the vapor phase so that liquid transpoit is not an importantnvichanitauFurthermore,the aolyent lags the drivinggas by en a~recfable ainouty and, therefore, is moving in a segion in whichthe sss is well dutributed. That the solvent dntains a coherent bank is indicatedby.the late butaoe breakthrough shown on F@.


il. Solvent breakthrough at nmeb lower ofl recovery woidd be ex@cted for a miscible flood in a core of this geometry. TIM re&lts obtrdned for the methanepropane floods show that larger slugs probably would be required for propane to obtainthe same oil reeoveryobtebiedwith butane. AIso, tbc amountof metlymerequiredto ntove tbe solvent through the Core &yiatcd farther from theary for propane than for butane.However, the calculatedvolume of methane also was substantitiy below that observedfor butane when the methanethroughputwas less than 4 to 5 PV at core pressure.One probablereasonfor the deviations at low methane production ia a less-uniformdis. tribution of gas in the region through which the solvent is being trsnspmted. At any given cross-sectioncontaining solvex the gas pltase,is probably hotat equilibrium with the liquid phase in the section. E@iiibrium is one of the mom of tie theory. As a limit, at the eriticsl pr*sure of t& methane-soi+entsystem the methane produced before the leading edge of the solvent bank most -nly does not go to aero. More exact calculations which take into account flow behavior and the corresponding lack of a sharp boundary between the solvent and oil might reduce some of the. difference between calculated and observed behavior; but, if nonequiliirhunis important,the dii?erenceswould still be substantialnear the critical pressure of the gas-solventsystem. in the two.phese solvent flooding process, two alternative methods of injedng the solvent are possible, A requirement for maximum oil recovery is that a dew-point mixture of the SSS and solvent at reservoir temperature be injected into or generated within the reservoir. A mixture containing the dew-point proportions of solvent and gas at reservoir temperatures will normally form two phases at surface temperatures. If such a two-phase mixture can,% injected in the reservoir without separation of the gas and liquid phases (for e=ple, * a fit), WS method of injecting the solvent is satisfactory or, perhaps, even preferable. The alternate procedure is to inject the liquid solvent directly into the reservoir and follow with sss injection, One other requirement of the theory is that the 5olventgss system is composed of only two molecular components, i.e., methane-propane or methane-butane, A solvent containing two mster@s, such as propane and butane, witl fractionate into its two components during the chromatograpbic transport through the reservoir. Arnold, et uJ, have shown ~ia previously for enriched-gas drives?. In practice, small amoupts of other components in an essentially one-component solvent should not affect the behavior of the tlo@appreeiably. As noted previously, the main advtmtege of two-phase solvent fleg is thst the method extends the prmsure regions in which solvent flooding may be used. The process by which the Qolventis transferred through the reservoir may offer some advantage in the control of the solvent bank 7he solvent is moved, primariiy, in the vapor pk. me pmbiibii,~ Of solvent finser formation, tierefore, is km thin for singl&plu@e miseibIedisplacemer$a. However, es the pressure approaches the pressure, for miscible displecemen$ fingering becomes (not un~tedly) an increasing problgm even for two-phase displace.: ments.

Naiwm.burtm . . ~-,=



- _Pxevious s@@ei of inerLgtwfriven .@vepi fdqg pessim havebeetie-ti6nd&d fo e@#ids i Wch the ddvMB


gas is OpprCCiiibijlsoluble in the


advent. fheresl!ka SftQW aohddlity affects, considerably, tba d&placement

1, Solvent al~ processes show continuity in behavior from the vapor preswre of the solvent up *o* the criticalprewureforthe two-component gawolvent system. 2. A biting facto? for WhCltt fkfa at &esauM8 batween the vapor prestmm of the solvent and tba Cdtleai pressureof the gaa-adventEyatemis the volume of gaE required to move die advent through the reaelvoir. An elementarytheory haa been derivedfor estimatingthe gaa requirements.AIthoughmere degant rnatbematieaf tmatntentaam PesaiMe,the aiutpletheory does ifluatratethe general princlpfeeof advent transferfor these sywcms. 3. l?xory and experimentsshow that the gas requirements are minimum at the vapor pressureof the advent, and agafn at the crhical pressure for the gas-advent Syaterm 4. l%m-phaw advent tleods may offer some advantage in controllingsolvent @ering, whfeh is a serious probtem in completelymisciblesystems.

The change in the moles of carrier gas in the region @dl is the 3naf tnolea minus the ildtial nwles present. FM . @ff[~ 1-x,) +(1 -m (1-Y,)



.. f&tl=#dl~. .






where & = assumwtitmnobile fiquid solvent saturation in R@orI B, Vm = rm$larVOhJ&of tiqukf phi%%le in advent bank, x, = mole fraction of whrsnt fn Iiqaddphaae, and S.a = oil saturation in Region II. Therefore* (1-Y.) @ dt = 1 [ K*F (1 S&)(1 Y.) 0 SOD)
@7*+ v, ;Mn . [ I

Jenks,L EL,CsaIpbdl, . B. and Binder G. GvJr.: A Field Tact of the G&Drivm L@ d Prnpsne&thod of (M kCOV. err. Trons.. AIME (1%7) 210.$4. 2. kisonj E.-F. and W& fL j: An Analysis, of the Linear
Diaplammentof Oil by C?~9rken ~Solrent Paper P06.C
pN!OMtCd at Wad &lDll@ fau -~

. . . . . . . (6) $indlarly. a brdamxean be writtenfor the wlvent assumfng the flow rate of liqufd solvent is negligible. Sl,x.+(l-SL)Y, ~ dt = +Adl ~ . . ...(7)

vu, 1 [ One other equation ia r&@red to relate the flow rates

(O@ 6-9, 1957).

of &E fn ~sh

3, Wel , H. J, Johnson, ti. F%Ewing, S. P. Jr. G nd %fnkxoan, F. ~t Who 13near Dfsplaeement of Oil &om Pomns Media Pet. ~ed. (Aw 1%1) 7S7. by EnrSohedGas, 10W 4. Sage B. H. sad on AP Researeb Proieot 37. API {1950). 5, Reb L R sad I&K= J. J.: Va r-Liquid in then.haaeNitrogen Systam,A[C&/our., ( 1% T ).?~i~ 6. &DOl& C. W., Stwe, H, J. and I& i). L: Dis laeement of Oil by RklMhs Banks, TraorwM% (l%o) 2~9, 305.

in the varioub regfons, If it is assumed the fluids are incompressible with no exchange of components between the carrier and the originaf oil, then

the fiffwar Para&&Y%C%Zh&%%&%?%%$%X$

where q is the flow rate of gas in Region A, and q, is

the oil flow rate in R@on D,

qn from 4. 8 and 48 from q. 7 me substitute iII JAI.6. When the resulting equation is simplitled and integrated, the followingis obtained.



= ~


To derive ~~ 1 given in the texg a material balance is made for the fluids in the volume Udf of Fig. 2, Separate balances must be ~ade for thi carrierga{-aird-for *e solvent. The moles of carriergas entering#Adl duringtime dc, 1sss the moles feaving, must equaf the change in the munber of moles present in that vcdume.Ilnw, -Y.)zh moles of carrier sss fn = q. (1 . (2) ~M, ,.. and . Qdft moles of earner gas out = ~ . . .< . (3)

in which

PV,., is the total Wre volumes of carrier gas which must be injeeted to produce the leading edge of the solvent. The volume of the @quidsolvent bank is negligible. If the wlvent is injected fn the vapor phase, the number of pore volumes of gas produced before the leading edge Of.the solvent bauk is independent of the bank size and ia equal to one less than tbe @re volumes injected,




The volume calculated

where q, = gas flowrate in Region B, qo~f3S3f10Wratein Region D, y, = mole fraction of solvent fn the vapor phase, Region B, vJf,= molar volume of vapor ht Region B, . = molarvolumeof gas ht Region D. vim

from RI. io is at r&irvoir pressure and temperature. If the volume of gas ismeasured at another prwaure and temperature, the equation becomes




in which VM ia the &cdar volume of the carrier gas at the specffied measuringcondidona. K, has been aubatltuted for y,/x,.
>. ..