S. C. Jones, SPE-AIME, Marathon Oil Co. W. 0. Roszelle; SPE-AIME, Marathon Oil Co. Introduction To find oil and water relative permeabilities by the displacement or unsteady-state method, a small linear core usually is saturated with water, then oilflooded to irreducible water saturation. Subsequently, the core is waterflooded, and during the process, pressure drop (either constant or variable) across the entire core and water injection rate (constant or variable) are determined. Effluent fractions are collected and the amount of water and oil in each is measured. Augmented by the absolute permeability and pore volume of the core and by oil and water viscosities, these data are sufficient to develop relative permeability curves. . The average saturation in the core at any time in the flood can be found from· an over-all material balance. However, to calculate relative permeability, the saturation history at some point in the core must be determined, not the average saturation history. The Welge1 equation yields saturations at the effluent end of the core when the_ average saturation history is known. Similarly, to compute relative permeability, the point . pressure gradient per unit injection rate is needed, not the average. The equation developed by Johnson et al. 2 converts average relative injectivity to a point value, accomplishing the required task. While the equations of Welge and Johnson et al. have been used successfully for years, they require tedious computation and are subject to error because of the evaluation of derivatives. The graphical techniques presented 0149-2136/78/0005-6045$00.25 © 1978 Society of Petroleum Engineers of AI ME in this study are equivalent to these equations, but are easier to use and can give a more accurate evaluation of relative permeability. Lefebvre du Prey 3 - 5 has presented graphical constructions based on curves of volume of oil produced vs time and pressure drop vs time to develop the required point functions. These constructions are limited to constant rate displacements. The constructions presented here are general and apply to constant rate, constant pressure, or variable rate-pressure displacements. Constant-rate and constant-pressure examples are given to help clarify the methods. The graphical techniques make it easy to see that double or triple saturation values, so extensively discussed in the past6 - 9 simply do not result from the fractional flow curve generated by a single displacement, such as a waterflood or an.oilflood. Theory Ignoring gravity effects and capillary pressure, water and oil relative permeabilities (expressed as functions of saturation) are · and kro = f.LJo 2 fA 2 - 1 . (2) · · · · · · · · ............... To use these equations, the fractional flow of water or oil and effective viscosity, A. - 1 , must be determined as functions of saturation. These must be point values, not aver- This paper presents graphical constructions that simplify the calculation of relative permeability from displacement data. These constructions convert raw data to relative penneability in a less tedious, more accurate manner than the usual computations. Fractional-flow saturation curves·derivedfrom waterflood displacements are always concave downward and never yield multiple-value saturations. MAY; 1978 807 Downloaded from http://onepetro.org/JPT/article-pdf/30/05/807/2227021/spe-6045-pa.pdf by Surcolombiana University user on 02 May 2023 Graphical Techniques for Determining Relative Permeability From Displacement Experiments age ones. The most convenient location is 'the outlet end of the core because fractional flow there is the same as the produced oil or water cut. Determining Saturation Data from a constant-rate waterflood are presented in Table 1. Injected water volumes are converted into pore volumes, Qi. Given the starting water saturation, average water saturation as a function of Qi is calculated from Sw =Swi + Np/VP . ........... : .......... (3) Sw 2 =Sw- Qi dSw · .................... (4) dQi Careful inspection of the construction reveals that all terms in Eq. 4 are represented. Several more tangents may be drawn to generate a complete Sw 2 vs Qi curve (Fig. 2). Note that all tangents drawn to the average saturation curve (Fig. 1) before water breakthrough intersect the ordinate axis at the exact, beginning water saturation, fw2 = 1 - fo2· · · · · · · · · · · · · · · · · · · · · · · · · · · (6) Fig. 1 could be extended to include all data in Table 1 (to 10 PV injection) and the remaining point-saturation and fractional-flow values could be obtained as above. However, a plot of average saturation vs the reciprocal of pore volume injected has two advantages (Fig. 3). First, long tangent extensions are ayoided; long lever arms tend to amplify errors. Second, the reciprocal plot provides a convenient means to extrapolate to (1 - S0 r) at infinite throughput (1/Qi = 0). As before, tangents to the curve are extended to the vertical axis. Intercept values (d{moted S w+) are no longer equal to the outflow-end saturations. Instead, the latter are given by Sw2 = 2Sw - Sw +, . . . . . . . . . . · · . · · . · · · · · · · (7) which is derived in Appendix A. Fig. 3 shows both the average saturation and the outflow-face saturation curves, computed from Eq. 7 using the tangent intercepts, and the extrapolation to infinite throughput. Here, the average and point saturations become equal . 0.7 TANGENT TO CURVE c. > ......... c. z J l.l fo2 = (Sw- Sw2)/Qi. · · · · · · · · · · · · · · · · · · · · · (5) Also, 0.4 o.J o~--::o':-.1---='"o.2=----:Lo.J=---o~.4~----:o:-'-.5--o-'-.6--o"'-.7----'o.a PV WATER INJECTED=W;/Vp=O; Fig. 1-Construction for determining point saturation from average saturation. Effective Viscosity Eqs. 1 and 2 require point values of effective viscosity as a function of saturation for the calculation of water and oil relative permeabilities. First, compute average effective viscosity or reciprocal relative mobility of fluids in the core. This is found by comparing ilp/q measured during the waterflood with the single- ph~e flow value, ilpb/qb, measured to find the absolute permeability of the core: 0.7 0.6 , EXTRAPOLATION TO { 1- 5 0 ,) AT INFINITE THRUPUT. 0.5 .68 0.4 .66 Sw2 0.3 .64 0.2 .62 0.1 .60 L------'------L..--....L....--..L..----,..-JL------1,--....J 0 0.2 0.6 0.8 1.0 1.2 1.4 0 0 ~ ~ ~ ~ M M Q;, P.V. Fig. 2-Saturation at outlet end of core. 808 ~ 0.8 1/0i = Vp/Wi Fig. 3-Construction for determining point saturation at large throughputs, showing extrapolation to residual saturation. JOURNAL OF PETROLEUM TECHNOLOGY Downloaded from http://onepetro.org/JPT/article-pdf/30/05/807/2227021/spe-6045-pa.pdf by Surcolombiana University user on 02 May 2023 Average saturation is plotted in Fig. 1 for the early portion of the example waterflood. Note that a line tangent to the curve at 0. 30 PV extends to intersect the ordinate axis at a saturation of 0. 534. This intercept value is the saturation at the outlet end of the core when 0. 30 PV water has been injected. The tangent construction is equivalent to the Welge equation, d_erived in Appendix A as S wi; that is, the saturation at the outlet end of the core is unchanged during a waterflood until water breakthrough. Also, the water saturation at this location is always less than the average saturation until all oil production terminates, at which time the slope of the average saturation curve becomes zero and the intercept has the same value as the average saturation. Because fractional flow of oil is equal to the slope of the tangent line in Fig. 1 at any given injection, it is calculated conveniently from the average and intercept saturations: TA~LE 1-DATA FOR CALCUl..ATION OF OIL/WATER RELATIVE PERMEABILITIES FROM A CONSTANT-RATE WATERFLOOD DISPLACEMENT wi Np (ml) (psi) Qi =W/Vp 0.00 3.11 7.00 7.84 8.43 8.93 9.30 9.65 9.96 10.11 10.30 138.6 120.4 97.5 91.9 87.9 83.7 78.5 74.2 70.0 68.1 65.4 0.000 0.100 0.225 0.360 0.523 0.780 1.260 2.001 3.500 5.000 10.000 (ml) 0.00 3.11 i.OO* 11.20 16.28 24.27 39.2 62.3 108.9 155.6 311.3 -ap A.-1 =apf.Lbqb!(apbq) Sw . =Sw 1+Np/Vp (cp) 0.350 0.450 0.575 0.602 0.621 0.637 0.649 0.660 0.670 0.675 0.681 13.50 11.73 9.50 8.95 8.56 8.15 7.65 7.23 6.82 6.63 6.37 /LI, = 1'-w = 0.970 cp /1-o = 10.45 cp su•i = 0.350 fj,p 1,/q1, = 0,1245 psi/(ml/hr) ... atS,. = 1.000 k = 35.4 md <P = 0.215 q = 80.0 ml/hr. A_, = IL• fl: j 1e:.·j . ............ ~ (8) where /J-b is the viscosity of the fluid used to find absolute permeability. Average effective viscosities for the example calculation are listed in Table 1 and plotted in Fig. 4. The graphical construction for finding point effective viscosity at the outlet end of a core is identical to that for point saturation. A tangent to the average viscosity curve is extended to the axis; the intercept is the point value. Fig. 4 shows tangents drawn at cumulative injections of 0.225, 0.300, and 0.600 PV. The equation that corresponds to the tangent construction is A2 -1 = A-1 - Qi dA. -1 dQi . . .............•... (9) Eq. 9 is derived in Appendix B. Outflow-end effective viscosities (intercept values) are plotted in Fig. 5. Comparison of Fig. 2 with Fig. 5 shows that effective viscosity at the outflow end of the core changes at the same throughput that saturation changes - that is, ef- fective viscosity does not change at this location until water breakthrough. To achieve an unchanging effective viscosity before water breakthrough (considering the tangent-intercept construction), the average effective viscosity curve (Fig. 4) must be linear from the start of the flood until water breakthrough. This requirement is useful in evaluating .experimental data - a plot of A.plq vs volume injected must be a straight line until breakthrough. Nonlinearity can result from at least two conditions: (1) the initial saturation throughout the core is nonuniform ~ insufficient fluid was injected in the pre~ vious flood to achieve uniform s~Huration and (2) the core is heterogeneous. In our experience, the first condition is the most frequent cause of nonlinearity and can be che~ked easily. If oil injection in a completely water-saturated core results in a linear Ap/q before oil breakthrough, but a subsequent waterflood does not yield a linear plot before water breakthrough, then the oil saturation just before the waterflood probably was not uniform. However, if the initial oil ·injection results in a nonlinear curve, then Condition 2 must be suspected. Injection of non-Newtonian polymer 14,---.----.--~~--~--~--~----~--~ 13 12 "'1-t. :l.. 11 12 #<J ~,,. ).,-;, cp 11 10 8~--~~~----~--~--~--~~--~--~ 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 Q; Fig. 4-Construction for determining outlet-end effective viscosity from average values. MAY, 1978 0.8 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Q;, P.V. Fig. 5-Effective viscosity at outlet end of core. 809 Downloaded from http://onepetro.org/JPT/article-pdf/30/05/807/2227021/spe-6045-pa.pdf by Surcolombiana University user on 02 May 2023 *Point of water breakthrough. V" = 31.13 ml; L =5.002 in.; 0 = 1.500 in. TABLE 2-CALCULATED VALUES FOR OIL/WATER RELATIVE PERMEABILITIES Pore Volume Injected 0; 0.000 0.100 0.225 0.300 0.600 1.000 2.500 5.000 iO.OQO 00 Av~age __1_ . 0; 10.000 4.444 3.333 1.667 1.000 0.400 0.200 0.100 0.000 Sw 0.350 0.450 0.575 0.5~3 0.627 0.643 0.664 0.675 0.681 0.687 Intercept Sw2 0.350 0.350 0.511 0.534 0.580 0.617 0.646 0.664 0.676 0.687 Average A.-1 (cp) 13.50 11.73 9.50 9.16 8.42 7.88 . 7.05 6.63 6.37 6.07 Intercept A.2 - 1 (cp) 13.50 13.50 10.66 10.30 9.45 8.95 7.79 7.11 6.69 6.07 A2 - 1 =2A- 1 - A-1+ ............... : ..... (10) Average and point values as well as the extrapolation to infinite throughput are shown in Fig. 6. Fig. 6 is totally analogous to the saturation treatment in Fig. 3: Calculating Relative Permeability Table 2 lists average and intercept saturations and effective viscosities (taken from Figs. 1 through 3 and 4 through 6, respectively) and oil and water fractional flow values (calculated from ~qs. 5 and 6). Each line represents a particular cumulative water injection, Qi. Water and oil relative permeabilities then can be calculated from Eqs. 1 and 2, respectively, using the water and oil vis. cosities given in Table 1. The first line of Table 2 corresponds to the start of the waterflood and the second line to 0.100 PV injection before water breakthrough. Note that effluent-end saturations, effective viscosities, fractional flows, and relative 1.000 1.000 d.291 0.197 0.078 0.026 0.007 0.002 o~oo1 0.000 k,,.. '"'2 =1-fo2 0.000 0.000 0.709 0.803 0.922 0.974 0.993 0.998 0.999 1.000 k,. =fwJLwiA.2 -1 =foJL.fA2 - 1 0.000 0.000 0.065 0.076 0.095 0.106 0.124 0.136 0.145 0.160 0.774 0.774 0.285 0.200 0.087 0.030 0.010 0.003 0.002 0.000 permeabilities are identical in these two lines~ In fact, these values would be identical for all cumulative injections until water breakthrough; For this reason, it is impossible to obtain relative permeabilities for saturations between the initial and breakthrough values using the unsteady-state method. . Fig. 7 shows oil and water relative permeabilities plotted as functions of water saturation (Sw 2 in Table 2). Because all the variables ate functions of cumulative irijectioil; relative permeability is also a function of cumulative injection. Thus, the curves are directional, starting with the lowest water saturation (0. 35) and proceeding to the right. The next higher water-saturation point shown, 0.511, corresponds to water breakthrough at 0.225 PV injection. The unobtainable relativepermeability curves between this saturation and the initial, or irreducible, water saturations are drawn as dashed iines. Moving to the right, the points represent increasing throughput to infinite injection at the last pair of extrapolated points (Sw = 0.687). · Constant Pressure Example Data for a constant pressure-drop. (- D..p = 100 psi) displacement are shown in Table 3. Core and fluid properties equal those in Table 1. Cumulative water injection, assumed equal to the sum of cumulative water and oil production, and cumulative oil production are plotted vs time (Fig. 8). .\ \ kro \ X) cp \ \ \ \ \ \ \ 0.6 0.8 1.4 1/Qi = Vp/Wi Fig. &-Construction for determining point effective viscosity at large throughputs. 810 o~~--~--~~~--~--~~~~~~--~ 0 0.1 0.2 0.5 1.0 Sw Fig. 7-Relative permeability curves from example calculation. JOURNAL OF PETROLEUM TECHNOLOGY Downloaded from http://onepetro.org/JPT/article-pdf/30/05/807/2227021/spe-6045-pa.pdf by Surcolombiana University user on 02 May 2023 solutions also can cause nonlinearity. Prebreakthrough linearity is necessary for the valid application of the Welge and Johnson et al. eqmitions or the corresponding graphical constructions. Large-throughput, average effective-viscosity data are treated similarly to large-throughput, average saturation data; A- l is plotted vs 1/Qi as in Fig. 6. Tangents are drawn and outlet-end values are calculated from the intercepts, A-1+, by '·· =(Sw-Sw~IO; TABLE 3-DATA FROM CONSTANT-PRESSURE WATERFLOOD DISPLACEMENT ~~.-1 q Elapsed Time (minutes) W; Np (ml) (ml) 0.00 3.00 6.20* 9.00 12.00 15.00 20.00 26.00 60.00 100.00 150.00 0.00 3.09 7.00 10.90 15.28 19.89 27.9 37.8 99.5 176.8 276;9 0.00 3.09 7.00 '7.80 . 8.33 8.70 9.01 9.32 9.90 10.09 10.31 (from slope) (ml/hr) 57.7** 66.4 8.2.0 86.4 90.2 93.4 97.5 98.6 113.0 118.7 121.5 0; =W;IVP 0.000 0.099 0.225 0.350 0.491 0.639' 0.896 1.214 3.196 5.679 8.895 Sw =Sw;+NP/VP 0.350 0.449 0.575 0.601 0.618 0.629 0.639 0.649 0.668 0.674 0.681 = t.p p.bqbl (t.pbqJ (cp) 13.50 11.73 9.50 9.02 8.64 8.34 7.99 7.90 6.89 6.56 6.41 Core and fluid properties are the same as those shown in Table 1. S,.; = 0.350 -!lp = 100.0 psi q · = variable. Instantaneous injection rates must be determined to calculate average effective viscosities. These are found by drawing lines tangent to the cumulative injection vs time curve (Fig. 8). The slope of the curve at any point is equal to the instantaneous rate. This rate often changes most rapidly between the start of water injection and water breakthrough. Because of the short time span involved, accurate measurements are difficult to obtain, especially at the point of water breakthrough. The requirement of a linear - D..p/q vs Wi relationship until water breakthrough permits an alternative calculation of rate at breakthrough, qBT = l/(2fltBrfWiBT - 1f%), ............. (11) hand side of Eq. 14 to Eq. 9. However, it is important to recognize. a subtle difference. Water and oil perineabilities presented here are relative to the core's absolute permeability to water. Johnson's permeabilities are relative to the prewaterflood condition - that· is, oil flowing at irreducible water saturation. Thus; to convert our relative permeabilities to Johnson's basis, we must divide these by kroCSwi) or 0. 774. We prefer the absoiute basis and recommend its use, ·but the choice is arbitrary. Fractional Flow Curves Fig. 9 shows the satunltion/fractiomil-flow relationship for the example waterflood. The portion of the curve where D..tBr and WiBr are cumulative time and injected volume, respectively, from the start of water injection until water breakthrough. The initial rate, q 0 , is the same as the steady-state rate (at the same D..p) obtained at the end of oil saturation and just before water injection. Once the instantaneous rates have been determined, values of A- 1 can be calculated fromEq. 8. The values of Qi, Sw, and A- 1 are plotted exactly as in the constant rate example, and all subsequent constructions and calculations are identical to this example. Equivalence of Graphical Technique and Johnson Equation Eq. 2 shows that {:: = ~: 1 .......................... (12) But, Johnson et al. 2 have shown that d (_1___) fo2 = -~Q+-l_,__r~ kro d (Ji) , , , , , , , , , , , , , , .... ,· .. (13) where/r is defined as a relative injectivity (see Appendix B). Therefore, by comparing Eq. 12 with Eq. 13, ................. (14) TIME, minutes Appendix B demonstrates the equivalence of the rightMAY, 1978 Fig. a-Cumulative water injected and cumulative oil produced vs time for constant pressure example. 811 Downloaded from http://onepetro.org/JPT/article-pdf/30/05/807/2227021/spe-6045-pa.pdf by Surcolombiana University user on 02 May 2023 *Water breakthrough. **This rate is calculated from oil injection data at the end of oil saturation. Sw2 AT 0.5 PV INJECTED OL-~----~----~~~----~----~----~~----~----~ 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0,9 Sw Fig. g___;Fractional flow curve for example waterflood. 812 1.0 increases in saturation with continued injection. The initial straight-line portion before breakthrough is followed by a curve of positive, but continually decreasing slope with continued injection (because the slope is equal to 1/Qi). In contrast, most fractional flow curves presented in the waterflooding literature 10 have an "S" shape, or even are concave upward. Fractional flow curves that are not concave downward can be computed from kro and krw data. If relative permeability curves generated from an unsteady-state displacement (Fig. 7) are completed by interpolation between the connate water and breakthrough saturations (the dashed portions of Fig. 7); then a family of fractional flow curves can be calculated from where the viscosity ratio, Mwl p.,0 , is a parameter. The shape of a computed curve depends on the value of the viscosity ratio compared with the ratio actually observed in the experiment to generate the kro and krw data. If the viscosity ratio used iri Eq. 15 is the same as the experimental ratio and if the relative permeability curves are interpolated correctly, then the calculated fractional flow curve will have the shape of the one in Fig. 9 - a straight-line, prebreakthrough portion followed by a curve with a posit~ve, but gradually decreasing slope. This curve is corlcave downward; however, higherthan-experimental viscosity ratios will produce curves that have an ''S '' shape or that are everywhere concave upward. On the other hand, lower-than-experimental .ratios yield fractional flow curves that are concave downward, brlt that have no prebreakthrough straight-· line portion. Steady-state relative permeability experiments also can yield fractional flow curves that are not concave downward. In addition to computing fractional flow curves fromEq. 15 and using steady-state-derivedkro and ktw curves and higher-than-experimental viscosity ratios, another possibility exists. A fraCtional flq~ curve derived from steady-state data does not represetH a single· displacement. Instead, it is the locus of end points from several displacements, in each of which a different water.,.oil ratio is injected. Indeed, the actual fractional flow curve for each displacement is concave downward. Fig. 10 illustrates a hypothetical, four-point, steady-state fractional flow curve that really is comprised of four separate displacements. Thefwi values listed are injected water fractions. · The important point is that any fractional flow vs saturation curve that represents a single experimental displacement always will be concave downward when the displacing phase is plotted- fw 2 vs Sw 2 for a waterflood orj02 vs S02 for an oilflood. This is true regardless of the wettability of the core or the water/oil viscosity ratio used in the displacement. Other shapes of fractional flow curves are obtained only from composite experiments or by. calculation, using relative permeability data and a water/oil viscosity ratio that is higher than the experimental ratio .. A saturation profile can be constructed frorri the derivative of the fractional flow curve. The derivative of 'the curve shown in Fig. 9, dfwldSw, is.plotted vs Sw in Fig. 11. A horizontal line corresponding to the initial JOURNAL OF PETROLEUM TECHNOLOGY Downloaded from http://onepetro.org/JPT/article-pdf/30/05/807/2227021/spe-6045-pa.pdf by Surcolombiana University user on 02 May 2023 between the initial water saturation (0.35) and water breakthrough is a straight-line segment with a slope equal to the reciprocal of pore volumes injected at water breakthrough. The straight line is a consequence -of rotating successive tangent lines (Fig. 1) around the discontinuity at water breakthrough from the initial slope ( = 1. 0) before water breakthrough to tangency just after breakthrough. Because the intercept of each tangentline is Sw 2 and its slope is ( 1 .;_ f w 2 ), saturation and fractional flow of water are related linearly when these liries originate from a single point. If satuation changes were continuous at water breakthrough, caused by smearing of the waterbank front by capillary presstire forces, then lines that are truly tangent to the average saturation curve at water breakthrough could be drawn. Points generated from these tangents also would form a virtually straight-line segment of the fractional flow curve. The main point is that a fractional flow curve that describes a waterflood or oilflood is essentially a straight line from the initial saturation and fractional flow to the point that describes saturation and fractional flow of the displacing phase immediately after breakthrough. As Welge1 demonstrated, the slope of the fractional flow curve, dfwldS w, at any point is equal to the reciprocal of pore volumes of water injected. Furthermore, a line segment that is drawn tangent to the curve and extended tofw = l.Owill temiinate at the average water saturation. TQ illustrate, a line (slope = 2.0) is drawn tangent to the fractional flow curve (Fig. 9) with a slope indicating a cumulative injection of 1/2.0 = 0. 5 PV. The extension of this line to fw = 1.0 shows that the average water saturation in the coreat 0.5 PV injection is 0.616. The saturation and fractional flow of water at the outflow end of the core (found from the point of tangency) are 0.573 and 0.910, respectively. Thus, Qi, fw 2 , Sw2• and Sw can be determined from the tangent construction. Some confusion regarding the shape of fractional flow curves exists. We have implied that any fractional flow curve that describes an actual displacement is .concave downward when the plotted phase (for example, water) water saturation, Swi' has been added to the right side of the plot. Fig. 11 is a generalized saturation profile 11 from which one can compute the saturation profile in the core at any stage in the waterflood. For example, suppose we wish to find the profile at 0. 2 PV water injected. The outflow end of the core (x = 1.0) corresponds to an abscissa value of 5. 0 .(x/qi = 1. 0/0. 2); the midpoint corresponds to 2.5, and so forth. Note that the water bank has not reached the outlet· end of the core at 0.2 PV injection. At 0.5 PV injection, the outlet-end saturation occurs at 1.0/0.5 = 2.0. The midpoirtt of the core corresponds to an abscissa value of 1. 0, and so forth. Construction of a saturation profile from the derivative of an S-shaped, fractional flow curve can lead to errors. A notable example is the triple value of saturation front that has been discussed for decades. The multiple-value saturation problem has been circumvented by use of the well known secant-tangent construction. 1 Here, the secant is drawn from initial conditions ifw = 0 andSwi) to the point of tangency on the composite fractional flow curve and become.s the straight.;. line portion of the fractional flow curve observed for a single displacement. Hysteresis DISPLACEMENT END POINT DISPLACEMENT 3 fwi = 0.75 I I I I I DISPLACEMENT 2 fwi = 0.50 I I I I The imbibition relative-permeability curves obtained from a waterflood (Fig. 7) are reproduced in Fig. 12. Two more curves have been added that result from an oilflood following the waterflood. These are drainage curves for a water-wet core. Relative permeability to water is considerably reduced in the drainage cycle, but permeability to oil, is relatively unchanged from imbibition values. This behavior is typical of our observations of water-wet cores. Like capillary pressure, relative permeability exhibits hysteresis and depends on history. The wide divergence among krw curves makes it mandatory to choose the correct one for predicting flooding behavior, particularly for tertiary recovery calculations. The waterflood and oilflood fractional flow curves (Fig. 13) also form a hysteresis loop. The oilflood part of I I DISPLACEMENT 1 fwi = 0.25 0~--~--~~----~--~----~~--~--~ 0.2 0.3 0.4 0.6 0.5 0.7 0.8 0.9 Sw Fig.10-Composite steady-state fractional flow curve. OL_~~~--~-=~====~~~~--L_ 0 0.3 0.4 0.6 0.5 0.7 _____J 0.8 Sw Fig. 11-Generalized saturation profile for waterflood example. MAY, 1978 Fig. 12-Hysteresis in waterflood vs oilflood relative permeability curves. 813 Downloaded from http://onepetro.org/JPT/article-pdf/30/05/807/2227021/spe-6045-pa.pdf by Surcolombiana University user on 02 May 2023 DISPLACEMENT 4 fwi = 1.00 Conclusions 1. Unsteady-state, displacement relative permeability data can be calculated easily and accurately using graphical constructions that are equivalent to the standard equations. Permeabilities calculated from the Johnson et al. equation are relative to the effective permeability at prewaterflood saturation. Permeabilities calculated by the techniques presented in this study are relative to absolute water permeability. 2. True end-point saturation, Sor• and effective viscosity can be estimated by graphical techniques presented here. 3. A fractional flow curve for a single displacement is always concave downward when the displacing phase is plotted. 4. Relative permeability and fractional flow curves exhibit distinct hysteresis for an oilflood following a waterflood. 5. Multiple- value saturation profiles never result from a fractional flow curve developed from a single displacement. 6. Because krw and kro are history dependent, water- flood performance should be predicted more accurately by unsteady-state, waterflood-derived data than by steady-state data, provided that capillary end effects are negligible. Nomenclature A = area normal to the axis of a linear core f = fractional flow of water or oil Ir =relative injectivity k = absolute permeability kr = relative permeability .t = length of a linear core from the injection face to an arbitrary point within the core L = total length of a linear core NP = volume of oil produced p =pressure q = volumetric injection rate Qi =pore volumes injected relative to the entire core S = saturation VP =pore volume of an entire core wi =volume of water injected x = fraction of the total length of a core z =pore volumes injected with respect to an interior location in a core d = denotes a difference between two quantities A- l =effective viscosity or reciprocal relative mobility J.L = viscosity Subscripts b =refers to single-phase flow or fluid in a core on which the absolute permeability is based i = initial (prewaterflood) or irreducible ~ =refers to an interior measurement in a linear core L =refers to an over-all measurement in a linear core o =oil r = refers to residual saturation w =water x =refers to an interior location in a core 2 =refers to outlet end location Superscript + = intercept value Acknowledgment The authors thank H. J. Ramey, Jr., and W. E. Brigham for their encouragement to publish this paper. References ru ~ 0~ OA 0~ Q6 OJ ~ Sw Fig. 13-Hysteresis in fractional flow curves. 814 M ~ 1. Welge, Henry J.: ''A Simplified Method for Computing Oil Recovery by Gas or Water Drive, "Trans., AIME(1952) 195,91-98. 2. Johnson, E. F., Bossler, D.P., and Naumann, V. 0.: "Calculation of Relative Permeability from Displacement Experiments," Trans., AIME (1959) 216, 370-372. 3. Buckley, S. E. and Leverett, M. C.: "Mechanism of Fluid Displacement in Sands," Trans., AIME (1942) 146, 107-116. 4. Lefebvre du Prey, E.: ''Deplacement de I 'huile par I 'eau dans un milieu consolide," Report Ref. 11.456 lnstitut Franc;ais du Petrole, Paris (Jan. 1965). 5. Lefebvre du Prey, E.: ''Mesure des Permeabilities relatives par Ia methode de W~lge~" ~eport Ref. 15.120 Institut Franc;ais du Petrole, Paris (Oct. 1967). 6. ''Mesure des Permeabilities relatives par Ia methode de Welge (enquete)," Groupe Petrophysique de Ia sous~Commission Laboratoires d'Exploitation du Comire des Techniciens, Chambre Syndicale de Ia Recherche et de Ia Production de Petrole et du Gaz Nature!, Paris (Sept.-Oct. 1973) 28, No.5, 695. JOURNAL OF PETROLEUM TECHNOLOGY Downloaded from http://onepetro.org/JPT/article-pdf/30/05/807/2227021/spe-6045-pa.pdf by Surcolombiana University user on 02 May 2023 the loop is concave upward because the displaced phase, not the displacing phase, is plotted. History dependence raises the question of uniqueness. If one chooses to generate steady-state relative permeabilities using eight steps instead of four, would the same krolkrw curves be generated? Because two sets of relative permeability curves are formed when the direction of saturation change is reversed, might not a family of curves be generated that depends on the saturation history of the core? These questions can be answered only with careful, accurate experimental measurements. Because relative permeability and fractional flow curves depend on history, we believe that waterflood performance is described more accurately by unsteadystate, waterflood-derived, relative permeability curves than by the composite curves obtained from simultaneous injection of water and oil. 7. Cardwell, W. T., Jr.: ''The Meaning of the Triple Value in NonCapillary Buckley-Leverett Theory," Trans., AIME (195.9) 216, 271-276. 8. Fayers, F. J. and Sheldon, J. W.: "The Effect of Capillary Pressure and Gravity on Two-Phase Fluid Flow in a Porous Medium," Trans., AIME (1959) 216, 147-155. 9. Hovanessian, S. A. and Fayers, F. J.: "Linear Water Flood With Gravity and Capillary Effects," Soc. Pet. Eng. J. (March 1961) 32~36; Trans., AIME, 222. 10. Craig, F. F., Jr.: The Reservoir Engineering Aspects of Waterflooding, Monograph Series, Society of Petroleum Engineers of AIME, Dallas (1971). 11. Parsons, R. W. and Jones, S. C.: "Linear Scaling in Slug-Type Processes- Application to Micellar Flooding," Soc. Pet. Eng. J. (Feb. 1977) 11-26. Therefore, from the definition of a derivative, So[Qi,x] =So[Qi,x] +xaSa~~i,x] . .............................. (A-3) If z is defined as Q1/x, then from the chain rule we get aS::[Qi,X] - a"S:"[Qi,X] . ~ , ........... (A-4) ax. az ax but, for single-fluid injection: a"S:"[Qi,x J= d5;;[Qi,x] , ................. (A-5) · az d(Qi/x) and, ~ = -Q/x2 • Calculating Point Saturations in a Linear Core From Production Data ax is odx = lim ~~o~ x+.U lim ~~o (A-6) Therefore, For a given porous medium and fluid system, a displacement resulting from the injection of a single fluid is said to be linearly scalable if saturation at any point is a function of only the number of pore volumes of fluid injected with respect to that point. 11 The pore volumes of fluid injected in a core segment can be increased in two equivalent ways. First, more fluid can be injected. Second, the volume of rock on which the injected _PV of fluid - is calculated can be decreased. Regardless of basis size, however, the average saturation (of oil, for example) remaining in the rock and the shape of the saturation profile depend on only the pore volumes of fluid, Q/x, injected in the rock, or portion of the rock, under consideration. For a linearly scalable displacement, the saturation at any point in a linear core can be obtained from a material balance around a small segment of the core. All that is required is a plot of average saturation as a function of the pore volumes of fluid injected. Although data obtained for this plot /are the average saturation over the entire core and the pore volumes injected based on the entire core, these may be plotted as if applied to only an arbitrary segment of the core (from 0 to x), based on the above arguments (see Fig. 14). The shaded area in the ~x increment is the difference in shaded areas from 0 to x + ~ and from 0 to x. These areas also can be expressed as an average saturation times the appropriate distance. As ~approaches zero, the average saturation over the increment approaches the point saturation at x. These considerations lead to the following equations. x+.U • ••••••••••••••••••• So[Q i•x J = -so[Q·v x J - Qi Ji-SJQi,x I J . . . . . (A-7) ----,;c... x . .. d(Qi x)· At the outflow face, x = 1, and Eq :-A-7 becomes d~r Qi T (A-8) - -so [Qi J - nSo2 [Q . i Jlrt.i --dQi.- · · · · · · · · · Because Sw = 1 - S0 , Eq. A-8 can be written as Sw2[Qi] =Sw[Qi] -Qid~~fi],_ ........ (A-9) which is Eq. 4 in the text. _ Large throughput data, plotted as Sw vs x!Qi, are handled slightly differently. Starting with Eq. A-3, this time written in terms of water instead of oil saturation: x x+Ax x Lsodx - [ Sodx ~ · {from 0 to x) ~ lim (x+~)Sa(Qi,x+~)-xSa(Qi,x) dx~o ~x . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A-1) AVERAGE SATURATION FROM 0 to x +Ax, AT A VOLUME, W;, OF WATER INJECTED. AVERAGE OIL SATURATION FROM 0 to x, AT THE SAME W; . and S0 (Qi,x) = lim ~~o [s(Q· +~) + o vx xe5:"[Qi,x+~]~so[Qi,x])] Llx PV INJECTED = Wi /( Vpx) · .............................. (A-2). MAY, 1978 Fig.14-Relationship between the saturation profile at a given instant in the injection history and the average saturation vs injection history. 815 Downloaded from http://onepetro.org/JPT/article-pdf/30/05/807/2227021/spe-6045-pa.pdf by Surcolombiana University user on 02 May 2023 APPENDIX A Application of the chain rule results in S [x}Q.] =S[xiQ·] w w t A-1 = +~ dS:[x/Qi] Qi l d(x/Qi) kA = qb/1-t/..- , . : ...................... (B-2) -dpb where - tJ.pb is the pressure drop across the entire core which, at the outlet face, becomes [VQ-] =S[l/Q.] +-1 dS:[ +Qi] : t w t Qi d(l/Qi) w2 ............................. (A-12) during the single-phase flow of a fluid of viscosity /1-b at rate qb (for example, during an absolute permeability determination); also, since But from the tangent construction (see Fig. 3), s w (B-1) But, · · · · · ........................ (A-ll) s k:- (- ~~) .................... R=xL, ............ ·................. (B-3) then +[liQ·] =S[+Q·] __ 1 dS:[liQi] . l w l Qi d(l/Qi) Substituting Eq. A-13 into Eq. A-12 yields Sw2[11Qi] =2Sw[11Qi] -sw+[l!Qi], ..... (A-14) Similarly (referring to Fig. 15), the average effective viscosity from 0 to x is which is Eq. 7 in the text. Ax - APPENDIXB Derivation of Point Effective Viscosity From Over-All Pressure/Rate Measurements in a Linear Core (~;:) ~(:) ............. ~ ... (B-4) 1 = e:~;: y~x . .. ................ (B-5) Point effective viscosity now is found in a similar manner to point saturation, using linear scaling principles: x+Ax A [Q,,x] = lim LA -tdx By definition, the effective viscosity (reciprocal relative mobility) at any point in a linear core is -1 tJ.x tJ.x~o x+Ax lim tJ.x~o x lim LA - dx -faA - dx ax~o tJ.x 1 1 (x+fJ.x~ [Qi,x+fJ.x] -x0[Qi,x] 11x · ........................... , .. (B-6) or A-1[Qi,x J =A -1[Qi,x] +x aFtJ_~hx J . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (B-7) PV INJECTED WITH RESPECT TO ENTIRE CORE=Wi/Vp= Qi a. Because A- 1 is a function of Q/x (Fig. 15), we must apply the chain rule to evaluate the partial derivatives in Eq. B-7. Ifz = Q/x, thenforsingle-fluidinjection u a8[Qi,x] = aA - 1 ax -----az- ~,..ll :ta. X\._ , EVAL~ATED AT x '-< PV INJECTED WITH RESPECT TO SEGMENT OF CORE FROM wQ· 0 to i = Vp(t/L) = -f 0 0 -t.pL -t.p 1 or -t.p. -t.px+ll.x ............................. i 1 +t:.L I r--t.p ll.x ! i ~ (B-9) At the outflow face (x = 1), the effective viscosity is I I I ~ ................ (B-8) Substituting Eq. B-8 into Eq. B-7, the effective viscosity at any point in the core is 1 A-1[Q· J =A-1[Q· J _Qi dA- [Qi,x] . uX uX X d(Qi/x) AND THE SAME W; ~)( 1- A,-'[Q,] = A- 1 [Q,]- Q,~~;,] , . I I x x + t.x Fig. 15-lllustrating the basis for calculating point-pressure gradients from over-all pressure measurements in a linear core. 816 az = d8[Qi,x] ax dz (~~i) 1.::. ' • ............................. (B-10) which is Eq. 9. FromEq. 2, JOURNAL OF PETROLEUM TECHNOLOGY Downloaded from http://onepetro.org/JPT/article-pdf/30/05/807/2227021/spe-6045-pa.pdf by Surcolombiana University user on 02 May 2023 1 A- = ............................. (A-13) fo2 = A2 - t . . . . . . . . . . . . . . . . . . (B-11) kro JLo But Johnson et al. 2 have shown that Substituting Eq. B-16 into Eq. B-14, we obtain JLo d . d ............ (B-12) tQ~r) -_----:-.----.--,------. p.,(flp/q), d (~J (Ji) JLb(Ap/q)i d (Ji) ............................. (B-17) But it is apparent from Eqs. B-11 and B-15 that where the relative irtjectivity, Ir, is lr = (lp) I tlPji' ...... (B-l3) _k_ = kro !o2 kro [swt.] kro 'kro [Swi ] = = JLJ'o2 A2 - 1 "·o r d d To demonstrate the equivalence of the two derivations, we must show that d ( 1 ) --'--=Q-7--/,__,r:. .:.___· . .. .. .. .. . (B-14) d(Ji) e:) l (a:),' .... (~) (Ji) -kro [s ·] Wt dQi dQi d(l!Qi) ............................. (B-20) d~/Qi) = _ Q.2 d~/Qi) ·d(l/Qi) dQi l ' ........ . (B-21) - ----=1 - Qi dA-1 -A - , .......... (B-22) dQi . which is equaJ to our A2 - 1• Therefore, (B- I 5) and the two derivations are equiyalent. Therefore, Ir = J.Lb(Aplq)i . . ............ (B-16) A- 1(Ap/q)b MAY, 1978 1 But, .............................. (B-11a) = !Lb ' JLb(Ap/q)i because at irreducible water saturatioh,/02 = 1 and A2 is equal to the average value,· A- 1 . Therefore, Eq. B-17 becomes d(l!Qi) A-• = JLo(Ap/q)b ............................. (B-18) d~/Qi) - d~/Qi) By definition, 1-Lo A- 1 JPT Original manuscript received in Society of Petroleum Engineers office July ~Q, 1976. Paper accepted for publication June 18, 1977. Revised manuscript received Dec. 20, 1977. Paper (SPE 6045) was presented at the SPE-AIME 51st Annual Fall Technical Conference and Exhibition, held in New Orleans, Oct. 3-6, 1976. This paper will be included in the 1978 Transactions volume. 817 Downloaded from http://onepetro.org/JPT/article-pdf/30/05/807/2227021/spe-6045-pa.pdf by Surcolombiana University user on 02 May 2023 where (q/ Ap )i is the injectivity just before the start of waterflood when oil flows at irreducible water saturation. Because two different bases were used in Eqs. B-11 and B-12, that is, our permeabilities are relative to the absolute brine permeability and Johnson's permeabilities are relative to the prewaterflood effective permeability, the left side of the equation can be written as kro [Swi]