STATE OF ILLINOIS WILLIAM G. STRATTON, Governor DEPARTMENT OF REGISTRATION AND EDUCATION VERA M. BINKS, Director ELECTROKINETICS Development of Equations for Two-Phase Flow Through Single Capillaries Norman Street DIVISION OF THE ILLINOIS STATE GEOLOGICAL SURVEY JOHN C. FRYE, Chief URBANA CIRCULAR 276 1959 Digitized by the Internet Archive in 2012 with funding from University of Illinois Urbana-Champaign http://archive.org/details/electrokineticsi276stre ELECTROKINETICS I.— Development of Equations for Two-Phase Flow Through Single Capillaries Norman Street ABSTRACT This report is the second part of a study on the electro- kinetics of the flow of oil and water through petroleum reservoirs. The development of electric charges at the oil-water inter- face is discussed and equations are presented for the streaming potential, electro-osmotic pressure, and electroviscous effect dis- played by a single capillary filled with anannulus of wetting phase (water) and having a central core of immiscible nonwetting phase (oil). Although it is unlikely that oil and water would flow through a porous medium following the theoretical pattern set forth here, it seems certain that some fraction of the flow is of this type and would modify considerably the measured electrokinetic effects. INTRODUCTION Although the effect of electrokinetics in retarding the flow of a single phase through capillaries and porous plugs has been treated theoretically and observed experimentally, no comparable work has yet been done on the simultaneous flow of two immiscible fluids. There seems little doubt that during such flow both fluids, under suitable conditions, undergo an electroviscous retardation; but it is also possible that, under other conditions, one of the phases may be accelerated so that the measured viscosity is less rather than greater than the bulk viscosity. Such decreased viscosity actually has been observed by Templeton (1954) and Templeton and Rushing (1956) during capillary flow of water and oil through micro- scopic capillaries. Before it is possible to derive an equation for the electroviscous effect re- sulting from two-phase flow through a porous medium, it is necessary to develop a model for that flow. Generally such a model is built up from some combination of capillary tubes. Thus the simplest theoretical approach is to equate the porous medium to a single capillary of suitable length and radius for the purpose of describing its hydrodynamic behavior. The logical extension is to a combination of tubes of varying radii, arranged in parallel and series. Such a series model was used by Scott and Rose (1953) to modify the Yuster model and thus help explain the nondependence of relative permeability on viscosity ratio. [1] 2 ILLINOIS STATE GEOLOGICAL SURVEY Network models (interconnecting networks of parallel and series elements) have been analyzed by Fatt (1956) who found them useful in predicting flow behav- ior; this work has been extended by Rose (1957). The use of network models for analyzing electrokinetics in porous media has been treated by Overbeek and Wijga (1946), Mazur and Overbeek (1951),and by Lorenz (1953a). Wyllie and Gardner (1958) suggested that one might start with a bundle of capillary tubes, cut them into thin sections, and rearrange these sections at ran- dom laterally to produce a "pseudo-model. " As pointed out by Tempi eton, the first concern is to gain knowledge of the mechanism of the flow through individual capillaries. Using a "bundle of capillary tubes" model, Brownell and Katz (1947) sug- gested that both phases flow through all the tubes, whereas Wyllie and Gardner suggested that each phase follows a separate path. It is possible to combine these views (as I plan to do in another article) and to divide the flow into three regions: a) tubes flowing both wetting and non- wetting phase; b) tubes flowing only wetting phase; and c) the remainder of the tubes, wetted by an immobile film of wetting fluid and flowing the nonwetting phase. The percentage of the total flow in each region can be determined by analy- sis of relative permeability curves carried out when the electrokinetic effect is negligible. From the contribution of each flow type to the total streaming potential, or volume flowed during electro-osmosis, one can then calculate the total streaming potential or electro-osmotic pressure to be expected, and thus also predict the electroviscous retardation to flow. CHARGED OIL-WATER INTERFACE There is much experimental evidence that charged interfaces develop be- tween oil and water, but the distribution of charges at this interface is open to discussion. Thus Booth (1951), in deriving equations for the calculation of zeta potentials from the electrophoretic mobility of oil droplets dispersed in an aqueous medium, considers three possible distributions of charge: a) the charge is concen- trated in a thin layer at the interface; b) the charge is distributed in uniform density throughout the droplet; c) the charge is distributed in an electric double layer de- veloped in the oil phase. The equations developed for distribution c reduce to those for distribution b when the double-layer thickness becomes greater than the radius of the droplet. One may consider that distribution c is the intermediate one and that the double layer can expand into uniform density, b, or compress into a thin interfacial layer, a. Thus if we picture the general charge distribution at the oil-water interface to be that of two diffuse double layers, we can illustrate it schematically as in figure 1 . Here, since the double diffuse double layer as a whole must be electrically neutral /°0 /° CO f yOjMdx + J p 2 (x)dx = (1) "CO ^° Aqueous phase Oil phase Fig. 1. - Schematic representation of the distribution of space charges in the oil and water phases TWO-PHASE FLOW THROUGH SINGLE CAPILLARIES which will also describe the charge condition in the event of reduction to either a or b. p i, A, = charge density in phase 1 and phase 2, respectively. The double-layer thickness in either phase is generally given by the expression where X- /^ Do 2 kT 2 (2) 2nz z e z k = Boltzmann's constant T = absolute temperature n = number of dissociated molecules per m^ z = valency of ions e = charge of an electron D = relative dielectric constant Do = absolute dielectric constant of a vacuum However, for purposes of comparison between aqueous solutions and hydro- carbons the expression given by Klinkenberg and van der Minne (1958) is much to be preferred. This expression may be arrived at from (2) by a series of substitu- tions, thus kT where and m 61777a A m - coefficient of molecular diffusion a = radius of an ion 77 = viscosity of medium where 6TT77a v = ionic velocity under the influence of an electric force, F F = zeE (E is electric field strength) It is also true that the current, I, flowing under the influence of a potential difference, E, in a liquid of conductivity, K, is I = KE = 2nvze Hence it follows that in (2) kT may be replaced by * c j o ? ? „ Eze Eze 6-m?a v , c A m 6tT77a and 2nz^e^ = K — — = K — y~ L =K6-n-r;a and so ^ = DD A m K Klinkenberg and van der Minne point out that A m may be assumed to have about the same value in hydrocarbon and water, because in the hydrocarbon the somewhat larger size of the organic ions will be compensated for by the lower viscosity of the medium. Early experiments on the mobility of oil droplets dispersed in water were reported by Ellis (1912), Mooney (1924), and Limberg (1926). Some authors have reported their results in terms of mobilities (microns per second per volt per centi- _L SOLID-WATER INTERFACE I OIL PHASE WATER PHASE Fig. 2. - The flow model showing the dis- tribution of the phases and of their associated double layers 4 ILLINOIS STATE GEOLOGICAL SURVEY meter), and others as zeta potential (millivolts). Although for large spherical solid particles dispersed in water at 25°C the zeta potential is numerically equal to the mobility multiplied by 14.3, such a simple relationship does not apply be- tween the mobility and zeta potential of oil droplets. First there is the problem of the charge distribution, which may be one of the three possible distributions listed by Booth, and, second, there is the fact that the droplets are liquid, which causes them to have a velocity different from that of an equivalent solid particle. Booth suggests that the mobility of the equivalent solid particle will be greater, but Jordan and Taylor (1952) suggest it will be less; Jordan and Taylor's hypothesis is supported to some extent by experimental evi- dence. It would appear that the zeta potentials of paraffin oil droplets may approach values of -86 mV in water and -151 in 0.0036M sodium oleate (Urbain and Jensen, 1936). Powney and Wood (1940) report mobilities of 4.35 u./sec/v/cm for Nujol droplets decreasing to about 2 at 0.005N. The mobility was considerably increased when the droplets were dispersed in sodium oleate and sodium dodecyl sulphate solutions. Newton (1930) had reported velocities of 3.9 in water with a conduc- tivity of 4xl0~6 ohm~l cm~l. More recently Douglas (1950) found that dissolved sugar reduced the mobil- ity of hydrocarbons in solutions of nonionic substances. He explained this as being due to the preferential adsorption of the sugar, which reduced the ionic adsorption and consequently the surface charge density. Many workers have studied hydrocarbons other than paraffin. Taylor and Wood (1957) found zeta potentials of -70 to -100 mV for decalin droplets in NaCl solutions, and Carruthers (1938) attempted to correlate measured mobilities with chemical composition using n-Octadecyl, n-Undecyl, n-Octyl, and A^'^n-Un- decenyl alcohols and n-Octadecane and A-^ ^n-Octadecene a n dispersed in solu- tions 0.01N with respect to sodium ion. Douglas (1943) determined the mobility of droplets of paraffin wax, dodecane, octadecane and A^-' ^octadecene, and of mixtures of hexane-dodecane, cyclohexane-dodecane, benzene-dodecane and decalin-dodecane, getting values of 1 - 1.5 u,/sec/v/cm in solutions kept 0.01N in sodium ion at various pH's. DEVELOPMENT OF EQUATIONS This paper deals with the electrokinetics to be expected when two phases are present in one capillary tube, and considers the flow through a single capillary under the conditions suggested by Brownell and Katz (1947) and by Yuster (1951), namely that the nonwetting phase (oil) should flow in a cylindrical portion of the wetted (by water) capillary and concentric with it (fig. 2). These authors assume that the velocities of the oil and aqueous phases are equal at the interface during such flow, and Yuster gives the following expression for the velocity distribution in the oil phase TWO-PHASE FLOW THROUGH SINGLE CAPILLARIES P 4L ( R 2- r 2) t ( r 2_ a 2) (4) where u Q = velocity of laminar layer of radius a R = radius of capillary r = radius of oil phase 7] w = viscosity of water phase 7] Q = viscosity of oil phase P = pressure drop across capillary L = length of capillary The average velocity of the oil phase is P 8'7 Lr2 2 ( R 2 r 2_ r 4) 7 7o + r 4 (5) /w — L / VV J and the velocity (v w ) of any laminar layer (radius a) in the aqueous phase is '» = T^r ' R2 - a2 l < 6 » Streaming Potential Assume that both the capillary and the oil surfaces are negatively charged so that mobile positive counter ions are located in the aqueous phase. (The analysis will remain valid, with due attention to signs of the zeta potentials, in the event of other charge distributions.) As only one ionic solution is in contact with both the capillary wall and the oil surface, it is apparent that the double layers associated with both surfaces will be of the same thickness (X). Then assuming, for simplicity in the derivation, that all the counter ions are located in a Helmholtz double layer, it is possible to derive an expression for the streaming potential caused by the movement of charges in the aqueous phase (Wood, 1946; Butler, 1940; Perrin, 1904). The streaming potential arises partly from the transport of the positive ions of both the double layers in the aqueous solution, namely those associated with the capillary wall charge and those associated with the oil surface, and partly from transport of the negative charges associated with the oil itself. It is of course essential that the thickness of the aqueous phase be such that the double layers do not overlap and that the distribution of charges in the oil phase be known. It is possible to suggest that the charges in the oil phase are located in a Helmholtz double layer at a distance (X ) of approximately X = DqK . x from the oil-water interface (D Q and K are the dielectric constant and the conduc- tivity of the oil, respectively). However, the double-layer thickness in the hydrocarbon phase will be great, and, as we consider only capillaries of small diameter, it is more likely that the charge is homogeneously distributed throughout this phase as suggested by Klinkenberg and van der Minne (1958). ILLINOIS STATE GEOLOGICAL SURVEY (1) Oil Phase -/: If electrokinetic measurements in the aqueous phase indicate a surface charge density of 'CO /^Mdx = -CT2 'o on the oil phase, then the surface charge per unit length is -2-rrra2; an expression that describes the charge per unit length of this phase whatever the charge dis- tribution. If the oil should move with an average velocity, v , the current of con- vection is I = -2Trrcr 2 v and substituting the value of v from equation 5 gives Io = " ?!l2l.T 9l o2r2_Wh3o^ 4 Vo -[2(R^-r4)^ + r«] (7) 1 ' w (2) Aqueous Phase Assuming that the surface-charge density in the Helmholtz double layer associated with the capillary interface is