Sie sind auf Seite 1von 8

SPE 68065 Two-Phase Steady-State and Unsteady-State Relative Permeability Prediction Models

M.N. Mohamad Ibrahim, Universiti Sains Malaysia and L.F. Koederitz, SPE, University of Missouri-Rolla
Copyright 2001, Society of Petroleum Engineers Inc. This paper was prepared for presentation at the 2001 SPE Middle East Oil Show held in Bahrain, 1720 March 2001. This paper was selected for presentation by an SPE Program Committee following review of information contained in an abstract submitted by the author(s). Contents of the paper, as presented, have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material, as presented, does not necessarily reflect any position of the Society of Petroleum Engineers, its officers, or members. Papers presented at SPE meetings are subject to publication review by Editorial Committees of the Society of Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper for commercial purposes without the written consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O. Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972952-9435.

Abstract Laboratory measurement of relative permeability using either steady-state or unsteady-state methods can be expensive and time consuming. Each method has certain advantages, resulting in preferences of one particular method over the other. The primary advantage of the unsteady-state method is that it is faster than the steady-state method. This paper presents a Linear Regression Model approach for developing prediction equations for water-oil and gas-oil relative permeability from steady-state and unsteady-state experimental data. These equations offer an alternative technique in determining relative permeability values and provide a better statistical representation of relative permeability values for a reservoir as a whole since ample amounts of data were used in developing them. The data for this study were obtained from published literature of the Society of Petroleum Engineers and various industry sources. Equations were developed for water-oil and gas-oil systems based on formation type and wettability in addition to the methods of measurement. Only equations for sandstone formations are presented in this paper due to the lack of data available for other formations. The adjusted coefficient of multiple determination, R2adj is used as a criterion of a good prediction model in this study in addition to a residual plot. The equations developed in this study are easy to use: nine out of twelve developed equations require only one independent variable in order to generate a relative permeability curve, and consist of no more than seven terms in each equation. The prediction equations are compared with previously published correlations where possible and the plot of the prediction equations indicates that the unsteady-state method results in lower oil relative permeability values and higher water relative permeability values at a given saturation than the steady-state one.

Introduction In general, laboratory methods for measuring relative permeability can be categorized into two major groups which consist of steady-state and unsteady-state methods. For intermediate (mixed) wettability rocks, steady-state methods are prefered to unsteady-state methods by some reserchers1. Unsteady-state methods however, almost always yield faster results compared to the steady-state methods due to the nature of processes involved in each method. In the steady-state method, two fluids are injected simultaneously at a constant rate through a core until the produced fluid ratio achieves an equilibrium condition with the injected fluid ratio. At this saturation, the effective permeability to each phase is obtained by using Darcys law. The ratio of the effective permeability to the base permeability will yield the relative permeability at that saturation value. Relative permeability values for different saturations are obtained by changing the fluid ratio of the injection rate2. Some examples of the steady-state method are the Penn-State method, Hassler method, Single-Sample Dynamic method and Hafford method. In the unsteady-state method however, in-situ fluid is displaced by a driving fluid at a constant rate. The saturation within the core changes with time. The changes in pressure and fluid produced are measured with respect to time. Relative permeability values of the in-situ and the driving fluid are then calculated using an equation originally developed by Buckley and Leverett3. Examples of the unsteady-state methods are Johnson-Bossler-Naumann, Welge and Jones-Roszelle. Laboratory measurement of relative permeability using either steady-state or unsteady-state methods can be expensive and time consuming. Laboratory measurement is considered a micro process because a single measurement is insufficient to represent the entire reservoir. Therefore several core samples from representative facies in the reservoir must be taken and tested. Since results of the relative permeability tests performed on several samples often vary, it is necessary to average the data before a scaling up from core to reservoir scale is performed. An accurate numerical procedure for determining relative permeability values provides an alternative technique, and at the same time it can overcome the previous shortcomings. In contrast to laboratory measurement, this is a macro process which provides a better statistical representation of relative permeability values for the reservoir as a whole.

M.N. MOHAMAD IBRAHIM, L.F. KOEDERITZ

SPE 68065

Objectives
With the availability of current sophisticated statistical software and the higher degree of competence in the petroleum industry, it is time for the petroleum engineers to start thinking of moving from a wet laboratory to a virtual laboratory in measuring relative permeability data. With a large amount of relative permeability data that can be obtained from the Society of Petroleum Engineers' literature (published from 1950 through 1998) as well as unpublished data from various oil and gas companies and individuals, better predicting models from the statistical view point can be achieved. In order to create predicting models that represent producing reservoirs, certain criteria in selecting data (relative permeability curves and other pertinent information) were imposed. The data selection criteria used in this study were: 1. The relative permeability curves are generated from either steady-state or unsteady-state experiments. In other words, relative permeability curves obtained from correlations or data obtained from hypothetical simulation studies are excluded in this study; 2. The core used in the experiment must be a naturally formed rock sample. Data obtained from synthetic or man-made cores such as Alundum cores is not considered; 3. Only imbibition data are used for oil-water and gas-water systems whereas for gas-oil and gas-condensate systems, only drainage data are used in the analysis; and 4. Only the primary data is selected when multiple imbibition or drainage processes are presented. 386 sets of relative permeability data that fulfill the above criteria were collected in this study. These data were classified based on system (water-oil or gas-oil), types of formation, wettability (for water-oil system) and the methods of measurement of relative permeability. Due to an insufficient number of relative permeability curves available for other formations, only predicting models for sandstone are presented in this paper. The wettability classification was made based on a modified Craigs rule4. Further divisions according to each wettability type listed in the modified Craigs rule are not feasible due to an insufficient amount of data available. It is not justified to present a predicting model based on only small amount of data and claim it as a predicting model. In this study, for oil-water systems, only water-wet and intermediate wettability models are presented. Data Manipulations The relative permeability curves used in this study did not originally have the same format, i.e., some of the curves were presented in the classical form while the rest were in normalized form. It was necessary to convert these curves into the same format (either the classical or the normalized) before the regression analysis is performed in order to be consistent. Since less than half of the data collected were in the classical form, the normalized form was chosen to be the standard form throughout this study. Moreover, it is easier to convert the classical data into the normalized form rather than converting the normalized data into the classical form. This is due to the difficulties in locating the classical relative

permeability end-point absolute permeability values (which most authors did not supply) in the articles reviewed. The classical data form was usually found in much older data, which is another justification to convert all of the data into the more current normalized form. Since the collected curves did not have the same range of saturation values (as far as the abscissa is concerned) due to the fact that some of the curves were longer than others owing to differences in the critical wetting and non-wetting phase saturations, this inconsistency would contribute to high variation in the response (ordinate). Thus, there is a need to find a way to plot each curve in its class on the same horizontal scale in order to reduce this variation so that a better prediction model can be achieved. This can be accomplished by normalizing either the wetting phase saturation or the nonwetting phase saturation which results in the horizontal axis always ranging from zero to one. For oil-water systems, the normalized water saturation is defined as5: S w - S wi * Sw = (1) 1 - S wi - S orw Unlike the oil-water system, the gas-oil system has the relative permeability to oil with respect to gas (krog) curves which are almost always longer than the relative permeability to gas (krg) curves due to the presence of the critical gas saturations (Sgc). Therefore, separate saturation normalization equations must be used for each curve as follows4:
*

Sg =

S g - S gc 1 - (S gc + S lc )

(2)

Sg S l* = 1 - 1- S lc

(3)

Regression Analysis In this study, a forward stepwise multiple linear regression technique was used in developing relative permeability prediction equations. Data for regression analysis may be extracted from experimental or nonexperimental studies. The experimental data is usually easier to deal with in regression analysis since the experimenter has control over the experiment itself. By modifying certain things in the experiment, one can obtained a desired result. Unlike the experimental data, the non-experimental or observational data provide no control over the independent variable(s). This is because the data were obtained after the experiments were completed. Data used in this study are considered as observational data even though the data originally was obtained from experimental studies. The success of a model in explaining the variation of dependent variable was measured by the value of the coefficient of multiple determination, R2. The R2 is interpreted as the proportion of observed values of Yi that can be explained by the regression model and is used to measure how well the model fits the data. The higher the

SPE 68065

TWO-PHASE STEADY-STATE AND UNSTEADY-STATE RELATIVE PERMEABILITY PREDICTION MODELS

value of R2, the more successful the model is in explaining the variation of Y. The R2 value will keep increasing if more independent variables are introduced into the model. To balance the use of more parameters against the gain in R2, many statisticians use the adjusted R2 value (R2adj)6 . R2adj is defined as follows:
2 R adj =

(n - 1)R 2 p n - (p + 1)

(4)

where n is the number of independent observations or data points and p is the number of independent variables or terms in the model. Very simply, the R2adj value approaching R2 indicates that excessive terms were not included in the model. Discussion The prediction equations developed in this study are listed in Appendix A and defined in Table 1 which summarizes the characteristics of the equations developed in this study. Notice that in Table 1, the number of data points for gas-oil systems used in the oil and gas equations do not match with each other although both equations were placed in the same methods row. For example, Eq. A9 has a larger number of data points compared to Eq. A10. This is due to the fact that the krog curve is longer than the krg curve due to the presence of the critical gas saturations (Sgc). Tables 2 and 3 list the ranges of rock properties and fluid saturations used in developing the prediction equations for oil-water and gas-oil systems, respectively. Each equation developed in this study has the following criteria: 1. The equation has a relatively high R2adj value; i.e. at least 75%. For observational data, an R2 value slightly higher than 60% was considered highly satisfactory6; 2. The R2adj values are equal to the R2 values thus indicating a reasonable fit without excessive terms; 3. The hypothesis test was conducted at 0.05 significant level. That means the probability of committing a type I error is less than or equal to 5%; and 4. A plot of residuals has the general pattern as shown in Figure 1 which suggests that the model has a constant variance 2. Figure 2 graphically illustrates predictive steadystate and unsteady-state oil-water relative permeability values for a water-wet sandstone formation. The relative permeability of oil with respect to water (krow) curves for the steady-state method are consistently higher than the unsteady-state values. While for the relative permeability of water (krw) curves, the unsteady-state values are almost always higher than the steady-state ones. The plots for intermediate (mixed) wettability systems depict the same trend as the water-wet plots. These plots agree with the definitions given by the modified Craigs rule as well as the original Craigs rule7 in defining wettability types. Figure 3 shows predictive steadystate and unsteady-state gas-oil relative permeability values for sandstone formations. In this figure, the relative permeability of oil with respect to gas (krog) curve for unsteady state is higher than for the steady-state one

while the reverse occurs for the relative permeability of gas (krg) curve. These phenomena should be expected since the unsteady-state method is a faster process compared to the steady-state method. In Figure 2 for example, the krow values for unsteady-state were more rapidly approaching zero as the driving fluid (water) pushes off the in-situ fluid (oil). In other words, as we move from left to right of the curves, the krow values for unsteady-state approached zero much faster for any given saturation than the steady-state method. Figures 4 and 5 show the plots of relative permeability values for various correlations for oil-water and gas-oil systems, respectively. In Figure 4, the steady-state and the unsteady-state plots are compared with the correlations given by Ibrahim-Koederitz4 and Honarpour et. al.8. Notice that these correlations did not specify the method of measurements. The plot of Honarpours correlations was not shown in Figure 5 because the correlation requires assuming the endpoint value for krg. For Figures 2 and 4, the authors used the following values to generate the curves: = 15% , ka = 100 md, Swc = 20% and Sorw = 20% whereas in Figure 3 and 5, the following values were used: = 15% , ka = 100 md, Swc = 15% and Sorg = 15%. Conclusions The prime objective of this paper is simply to provide readers with sets of steady-state and unsteady-state relative permeability prediction equations for oil-water and gas-oil systems. Six pairs of steady-state and unsteady-state two phase relative permeability prediction equations have been developed through extensive trial and error model building processes using linear regression analysis. The plot of the prediction equations for oil-water systems indicates that the unsteady-state method results in lower oil relative permeability values and higher water relative permeability values at a given saturation than the steady-state one. The plot of the prediction equations for gas-oil systems however, indicates that the steady-state method results in lower oil relative permeability values and higher gas relative permeability values at a given saturation than the unsteadystate one Comparison with other correlations indicate that they are in close agreement with each other. Acknowledgements The authors would like to express their appreciation to the Universiti Sains Malaysia and the University of MissouriRolla for the financial support of this project. Nomenclature Capital Letters 2 R = coefficient of multiple determination 2 R adj = adjusted coefficient of multiple determination = gas saturation, fraction Sg = critical gas saturation Sgc Sl = liquid saturation, fraction Slc = total of critical liquid saturations present in the system, fraction

M.N. MOHAMAD IBRAHIM, L.F. KOEDERITZ

SPE 68065

Sorg Sorw Sw Swc Swi Yi

= = = = = =

residual oil saturation in oil-gas system, fraction residual oil saturation in oil-water system, fraction water saturation, fraction critical (connate) water saturation, fraction initial water saturation, fraction th i observed value where i =1,2,3,.....n

Lowercase Letters ka = absolute permeability, md krg = relative permeability of gas, fraction krog = relative permeability of oil with respect to gas, fraction krow = relative permeability of oil with respect to water, fraction = relative permeability of water, fraction krw n = number of independent observations p = number of independent variable in a model Greek Symbol = 2 = Superscript * = ^ = References
1. Gorinik, B. and Roebuck, J. F.:Formation Evaluation through Extensive Use of Core Analysi, Core Laboratories, Inc., Dallas, Texas, (1979). 2. Unalmiser, S. and Funk, J. J : Engineering Core Analysis, Journal of Petroleum Technology, April, (1998). 3. Buckley, S. E. and Leverett, M. C. :Mechanism of Fluid Displacement in Sands, Trans. AIME, vol. 146, (1942), p.107 4. Mohamad Ibrahim, M. N. and Koederitz, L. F. : Two-Phase Relative Permeability Prediction Using a Linear Regression Model, SPE 65631 (2000). 5. Koederitz, L. F., Harvey A. H. and Honarpour M. : Introduction to Petroleum Reservoir Analysis, Gulf Publishing Company, Houston, Texas, (1989). 6. Devore, J. L. : Probability and Statistics for Engineering and the Sciences, Duxbury Press., California, (1995), pp. 474 ff. 7. Craig, F. F., Jr. : The Reservoir Engineering Aspects of Waterflooding Monograph, Vol. 3, Society of Petroleum

porosity, fraction variance

normalized value predicted value

Engineers of AIME, Henry L. Doherty Series, Dallas, Texas, (1993), p. 20 8. Honarpour, M., Koederitz, L. F. and Harvey, A. H.: "Empirical
Equations for Estimating Two Phase Relative Permeability in Consolidated Rock", Trans. AIME, vol. 273, (1982), pp. 2905 ff.

SPE 68065

TWO-PHASE STEADY-STATE AND UNSTEADY-STATE RELATIVE PERMEABILITY PREDICTION MODELS

Appendix A
* k* row = 1 - 2.5486421 S w + 1.54958751 S * 1.5 w

(A1)

k* rw = 0.05886035

* 2.5 Sw

(1 )

3. 5

0.4 2 1.5 *2.5 3 * - 31.494896 2 S 0.6 + 0.09325366 S * wc S orw S w w + 0.63098161 (S w S wc )

4 + 0.00337566 S *2 w (S wc ln k a )

(A2)

* k* row = 1 - 4.3807857 S w + 4.9439179 S

* 1.5 w

- 1.562024 S *2 w

(A3)

* 3.5 *4 * 1.5 4 0.2 k* + 0.08061213 S * rw = 0.13842961 S w w S orw - 0.9545567 S w S orw - 0.0322872 S w

k6 a

3 - 202.86687 S*0.8 S3 w orw S wc

2 1.5 + 12.2147902 S *0.6 S wc S orw - 0.5239345 S *0.4 (S wc ln k a ) 2 3 w w

(A4)

* k* row = 1 - 2.36277 S w + 1.842713 S

*2 w

0.472066 S *3 w

(A5)

*2 * *3 k* rw = 0.2071174 S w 0.429379 S w + 0.6381827 S w

(A6)

*2 * *3 k* row = 1 - 3.417802 S w + 4.1266403 S w 1.721169 S w

(A7)

* k* rw = 0.2155152 S w 0.121423 S

*2 w

+ 0.2768078 S *3 w

(A8)

* k* rog = 0.3365223 S l 2.742128 S

*2 l

+ 8.6732888 S l*3 - 10.84364 S l*4 + 5.5670795 S l*5

(A9)

* *2 k* rg = 0.8553047 S g 0.037865 S g

(A10)

* k* rog = 0.3625777 S l 1.258902 S

*2 l

+ 1.8952425 S l*3

(A11)

*2 *2 *2 k* rg = 0.7920128 S g 1.839124 S wc S g + 1.3264416 S g

(A12)

M.N. MOHAMAD IBRAHIM, L.F. KOEDERITZ

SPE 68065

Table 1. Summary of characteristics of equations develop in this study System Wettability Water-Wet Unsteady-State Oil-Water Intermediate Wettability (Mixed-Wet) Gas-Oil Unsteady-State Steady-State Unsteady-State Steady-State Method Steady-State Equation A1 A2 A3 A4 A5 A6 A7 A8 A9 A10 A11 A12 Number of Data Sets 304 304 543 543 179 179 209 209 483 388 413 353 R2 91 80 96 75 93 90 96 92 94 90 98 87 R2adj 91 80 96 75 93 90 96 92 94 90 98 87

Table 2. Ranges of rock properties and fluid saturations used in developing oil-water relative permeability equations Equations A1, A2, A3 & A4 A5, A6, A7 & A8

(%) 8.4 37.1 8.0 32.6

ka (md) 0.52 8,440 3.4 10,500

Swc (%) 3.6 67.5 5.0 38.9

Sorw (%) 6.6 47.3 11.09 44.4

Table 3. Ranges of rock properties and fluid saturations used in developing relative permeability equations for gas-oil systems
Equations A9 & A11 A10 & A12

(%) 6.3 39.0 6.3 39.0

ka (md) 1.48 5,580 1.48 3,650

Sgc (%) 0.6 25.0 0.6 25.0

Swc (%) 3.28 50.0 3.28 50.0

Sorg (%) 3.5 48.0 5.0 48.0

Figure 1: Residuals plot

SPE 68065

TWO-PHASE STEADY-STATE AND UNSTEADY-STATE RELATIVE PERMEABILITY PREDICTION MODELS

1 0.9 0.8 Unsteady-State 0.7 0.6 kr* 0.5 0.4 0.3 0.2 0.1 0 0 0.2 0.4 0.6 0.8 1 Sw (fraction) Steady-State

Figure 2: Oil-water Relative permeability plots for water-wet sandstone

Steady-State 0.8 Unsteady-State

0.6 kr* 0.4 0.2 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 Sg (fraction)

Figure 3: Gas-oil relative permeability plots

M.N. MOHAMAD IBRAHIM, L.F. KOEDERITZ

SPE 68065

1 Steady-State 0.8 Unsteady-State Ibrahim-Koederitz Honarpour et al 0.6 kr* 0.4 0.2 0 0.2 0.3 0.4 0.5 Sw (fraction) 0.6 0.7 0.8

Figure 4: Oil-water relative permeability plots for various correlations

1 Steady-State 0.8 Unsteady-State Ibrahim-Koederitz

0.6 kr* 0.4 0.2 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 Sg (fraction)

Figure 5: Gas-oil relative permeability plots for three different correlations

Das könnte Ihnen auch gefallen