You are on page 1of 19

SPE 77569 Analysis and Interpretation of Water-Oil-Ratio Performance

V.V. Bondar, ChevronTexaco Corp., T.A. Blasingame, Texas A&M U.

4. Application/interpretation of the following extrapolaCopyright 2002, Society of Petroleum Engineers Inc. This paper was prepared for presentation at the SPE Annual Technical Conference and Exhibition held in San Antonio, Texas, 29 September2 October 2002. 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 01-972-952-9435.

tion methods for the water-oil-ratio (WOR) and the water cut (fw) functions:
qo versus Np, log(fw) versus Np, 1/fw versus Np, fo versus Np, and log(WOR) versus Np.

Abstract This work presents the development and validation of a multivariate relation for the behavior of the water-oil-ratio (WOR) and/or water cut (fw) functions. This new model incorporates the reservoir and fluid properties for both phases (oil and water) and is based on the assumption of pseudosteady-state flow conditions. This work is an extension of traditional (i.e., steady-state) methods for the case of pseudosteady-state flow for both the oil and water phases. In this work, our pseudosteady-state model reproduces observed field performance substantially better than any of the steady-state models. We propose that this approach can be applied to any reservoir system undergoing waterflood. The specific tasks achieved in this work include:
1. Development of a rigorous model for the simultaneous

While our new formulation of the two-phase (oil-water), pseudosteady-state flow relation does not provide for a simple extrapolation formula for the estimation of recoverable oil, we do prove the utility of this relation as an interpretation mechanism. We also provide insight into the existing and proposed techniques for estimating the ultimate oil recovery. Orientation Natural water drive and/or water injection are two of the most common drive mechanisms in oil production. A detailed analysis of past performance should be conducted in order to predict the future performance of the well/reservoir system as well as to estimate the volume of in-place and recoverable fluids. The logarithm of the water-oil-ratio (WOR) and/or water cut (fw) functions plotted versus the cumulative oil production are commonly used tools for the evaluation and prediction of waterflood performance. This presumed (actually empirical) log-linear relationship of WOR (or fw) and oil recovery allows for the extrapolation of the observed straight-line to any desired water cut as a mechanism for determining the corresponding oil recovery. Such straight-line extrapolation methods are essentially empirical, and when theory is used to validate such techniques, we must make prohibitive simplistic assumptions. Two such assumptions that have been documented in the literature include the assumption that the mobility ratio is equal to unity and that a plot of log (krw/kro) versus So is linear. Our goal in this work is to develop and validate a multivariate relation to represent the behavior of the water-oil-ratio (WOR) and/or water cut (fw) functions the only significant assumption that we make in this work is that pseudosteady-state flow conditions must exist in the entire reservoir system.

flow of oil and water during pseudosteady-state flow conditions. This model has been validated using several field cases and gives an excellent representation of the WOR (or fw) data trend(s).
2. Development of a "reciprocal rate plot" for the estima-

tion of both the original oil-in-place (N), as well as the Estimated Ultimate Recovery (EUR) at current producing conditions.
3. Development of a diagnostic technique for assessing

(qualitatively) the efficiency/effectiveness of a waterflood. This technique involves the use of the following log-log format plots: WOR-derivative, the WOR-integral, and the WOR-integral-derivative functions.

V.V. Bondar and T.A. Blasingame

SPE 77569

The objectives pursued in this work include:


1. Derivation and validation of a pseudosteady-state mo-

del for the simultaneous flow of oil and water. This model is validated against 28 different field cases and in all cases, the new model gives an excellent representation of the data.
2. Development and validation of the "reciprocal rate

place and recoverable oil volumes are required for evaluation and reservoir management purposes. A number of essentially empirical methods have been proposed over time for the evaluation of waterflood performance, and these methods are typically assumed to give acceptable results. A plot of the logarithm of the water-oil ratio (WOR) (or water cut function (fw)) versus cumulative production (Np) is the most widely used technique for the evaluation and prediction of waterflood performance.1 This simple (and, we will note, empirical) method is applicable for the analysis of "late time" production behavior and the technique allows us to estimate the recoverable oil volumes by extrapolating the straight-line trend of the fw function to an arbitrary value of water cut (often fw=1, or some other "high" value, such as fw=0.95). This approach is only used when a straight line can approximate the function of interest (WOR or fw). Unfortunately, in most cases, this method is not applicable for the early stages of a waterflood (e.g., a rule of thumb is that the log(WOR or fw) versus Np plots can not be used for values of water cut function (fw) less then 0.5). Misuse of these empirical techniques can yield substantial errors in the extrapolation of recoverable reserves. We recognize that the use of a model based on the assumption of steady-state flow behavior is an approximation at best our motivation for this work is the development of a model that can be used to represent the pseudosteady-state behavior of a water-oil reservoir flow system. We also recognize that the pseudosteady-state model is an approximation as well and that mobility components can (and do) change substantially with time (which is a condition that we do not explicitly consider). However, our primary goal remains the development of a WOR (or fw) relation for pseudosteady-state flow conditions. Aside from the estimation of reservoir volumetric properties (N, Np,max, etc), WOR data can be plotted versus time (or the "material balance time" functions) on a log-log plot and used as a diagnostic tool to identify the dominant reservoir performance mechanism (uniform displacement, water coning, or water channeling).2 Our intention is to develop a methodology that combines the classic techniques for well test and production data analysis (pressure derivative, pressure integral, and pressure integralderivative functions) with our proposed model for the analysis of oil and water production data (in this case the WOR, WORderivative, WOR-integral, and WOR-integral-derivative functions). We believe that the qualitative analysis of water-oil ratio (WOR) performance data will significantly improve our evaluation and assessment of well problems, assist in injection/production balancing, and aid in the identification of the dominant reservoir drive mechanism. Development of the Pseudosteady-State WOR Model In this section we provide the derivation of the pseudosteadystate WOR equation. We begin with the rigorous, single-phase pseudosteady-state flow equation used to describe the behavior of oil and water production. This derivation makes

plot" for the estimation of original oil-in-place (N) and estimated ultimate recovery (EUR). This method is validated against field data and yields an appropriate trend in virtually all cases
3. Development and application of a series of log-log

diagnostic plots for waterflood evaluation these include:


WOR functions versus production time, WOR functions versus Np/qo, and WOR functions versus (Np+Wp)/(qo+qw).

The WOR functions include: WOR-derivative, WORintegral, and WOR-integral-derivative functions.


4. Application/interpretation of several different extrapo-

lation approaches for the water-oil-ratio (WOR) and the water cut (fw) functions. The new pseudosteady-state WOR model is given in terms of both Np and Wp, and, as such, is not amenable to extrapolation. As this issue could not be resolved (i.e., the use of the pseudosteady-state model as an extrapolation method), we chose to focus on several extrapolation techniques all of which use the cumulative oil production (Np) as the x-axis plotting function. These extrapolation plots include:
qo versus Np (constant pressure (liquid) case) log(fw) versus Np (steady-state approach) 1/fw versus Np (new approach) fo versus Np (field approach not documented) log(WOR) versus Np (steady-state approach)

As noted above, the formulation of the two-phase (oil-water), pseudosteady-state flow relation does not provide for a simple extrapolation formula for the estimation of recoverable fluids we believe that this is an area for further investigation. Introduction Historically, it has been difficult to analyze and predict oil production behavior in water-drive or waterflood reservoir systems (here we distinguish "waterdrive" as a natural condition of water influx and waterflood as a manufactured condition of water injection). Difficulty arises in how to characterize two-phase (oil-water) flow performance using analytical solutions based on single-phase flow theory, or by using simplified, steady-state solutions to represent two-phase (oilwater) flow. Neither approach is correct and yet both are used regularly for the analysis of data acquired from waterdrive/waterflood reservoir systems. As a matter of practice, we must be able to evaluate and predict waterflood performance in petroleum reservoirs in-

SPE 77569

Analysis and Interpretation of Water-Oil-Ratio Performance

several (potentially) limiting assumptions but the pseudosteady-state relation does not require the same assumptions as the steady-state case (which is much more restrictive). Specifically, he primary assumptions used in the conventional WOR (steady-state) analysis are: The pressure and the flowrate throughout the system remain constant. The mobility ratio is assumed to be equal to unity. The logarithm of relative permeability ratio versus water saturation relationship is linear. The rule of thumb for the existing WOR models is that these models can be applied only after the WOR (or fw) function develops a straight-line trend in most of the cases this occurs when the value of the fw function approaches 0.5-0.7 (or higher). Our goal is to extend the conventional WOR analyses to include the case of pseudosteady-state flow behavior and to develop a relation that can represent the entire spectrum of oil and water production performance. We do not expect such a relation to represent the entire production history for a particular well but we do anticipate significantly improved behavior when we incorporate pseudosteady-state flow characteristics into the WOR model. We begin by using the relation presented by Blasingame and Lee3 for single-phase variable-rate, pseudosteady-state flow in a bounded reservoir. This result is given as: p B 4A B = 70.6 + 0.2339 ln t mb ............ (1) C r 2 hc t A q kh e
A w

Using the general form of the pseudosteady-state flow equation (Eq. 2), we can write the following relations for singlephase oil and water flow: qo = qw = p m o t o + b psso p mwtw + b pssw (oil form)............................. (6)

(water form) ........................ (7)

where the "oil" and "water" variables are defined as:

t o = N p qo
mo = 0.2339 Bo hct A

(oil material balance time) .. (8) (oil pss slope term).............. (9)

B 4A (oil pss intercept term) (10) b psso = 70.6 o o ln 2 ko h e C A rw

tw = W p qw
m w = 0.2339 Bw hc t A

(water material balance time)... (11) (water pss slope term) ............. (12)

B 4A b pssw = 70.6 w w ln C r2 k wh e A w (water pss intercept term)......... (13) Recalling the definition of the water-oil ratio function, WOR, we have: q WOR = w qo (water-oil ratio)................. (14)

Eq. 1 is subject to the following assumptions: Pseudosteady-state (i.e., boundary-dominated) flow conditions must exist in the entire reservoir system. Homogeneous and isotropic reservoir. Constant porosity and permeability. Small and constant fluid compressibility. Constant fluid viscosity. Small pressure gradients. Negligible gravity forces. For simplicity, Eq. 1 can be written in the following, more compact form: q= p mt mb + b pss (general form) ..................... (2)

Substituting Eqs.6 and 7 into Eq. 13, we obtain an expression for the WOR function: WOR = mo t o + b psso m w t w + b pssw .................................................. (15)

where:

Eq. 15 is specifically valid for pseudosteady-state flow behavior and we implicitly assume that the entire reservoir is at pseudosteady-state flow conditions (i.e., both the oil and water phases). To extend this concept, we can re-write Eq. 15 in terms of the fractional flow of oil and water functions (fo, fw (respectively). Re-writing Eq. 15, we obtain: fw = 1+ 1 m w t w + b pssw mo t o + b psso 1 1+ mo t o + b psso m w t w + b pssw ................................................ (17) ............................................... (16)

t mb =

1 q

q dt

(material balance time)........ (3) (pss slope term) ................... (4)

B m = 0.2339 hct A

4A B (pss intercept term).............. (5) ln b pss = 70.6 C r2 kh e


A w

fo =

It is important to note that we have made no assumptions regarding a relationship between the relative permeability functions and water saturation although we do note that in

V.V. Bondar and T.A. Blasingame

SPE 77569

Eqs. 15-17 we presume that the mobility ratio is constant (but is not necessarily unity). dominant reservoir drive mechanism. Application of the Pseudosteady-State WOR Model In this section we use the new pseudosteady-state WOR model to analyze and interpret two field case examples. The first example is the case of a vertical well in a thick, low permeability dolomite sequence in West Texas, and the second example is the case of a vertical well in a high permeability, moderately consolidated reservoir in South Louisiana. In both cases the fields are undergoing waterfloods, but in the case of the wells from South Louisiana, water influx is also suspected. In these examples we use a standard suite of plots to evaluate our regression-based analysis of the WOR function using the proposed pseudosteady-state WOR model. The following suite of plots is used: Late-Time Extrapolation for Recoverable Oil log(fw) versus Np 1/fw versus Np fo versus Np log(WOR) versus Np Regression Analysis Global Summary Plot log(WORcal) versus log(WORmeas) Example 1: NRU Well 3106 Well 3106 from the North Robertson Unit (NRU Well 3106) in Gaines County Texas is the first case we consider, and the analysis/interpretation plots for this case are provided in Figs. 1-5. NRU Well 3106 was completed on July 9 1989. The total depth of the well is 7,350 ft with two perforated intervals 6,667-7,185 ft and 5,964-6,538 ft. The well was acidized and hydraulically fractured in two stages on 15 July 1989 and again on 17 July 1989. In Figs. 1-4 we find that the pseudosteady-state WOR model represents the production data functions extremely well especially in the late-time region, but surprisingly, the WOR model also matches all but the very earliest production data. For comparison we have plotted the "conventional" (i.e., steady-state) straight-line extrapolation models on each analysis plot and we note very good correlation of these straightline models with the late-time production data. Based on our observations in Figs. 1-4, we can conclude that the proposed pseudosteady-state WOR model best represents the performance of the WOR functions for NRU Well 3106. We can also conclude that the conventional straight-line extrapolation models also work well for this case. As a consistency check, we also provide a direct comparison of the calculated and measured WOR functions for NRU Well 3106 in Fig. 5. We note that with the exception of the very early time data (i.e., low WOR), the correlation of the calculated and measured WOR functions is excellent.

Example 2: WWL Well B41 In this case we consider the production data from Well B41 in the West White Lake Field (WWL Well B41) in South Louisiana. It is relevant to note that West White Lake Field is a Miocene sandstone reservoir sequence with good porosity and permeability characteristics. This is relevant because it may help us to understand the water and oil production performance more clearly and this case is in significant contrast to the case from the North Robertson Unit (NRU Well 3106) where the reservoir in that case is a low permeability dolstone with very heterogeneous reservoir properties. WWL Well B41 was completed on 2 July 1987. The total depth of the well is 7,300 ft and is perforated from 7,038 to 7,064 ft. The analysis plots for this case are presented in Figs. 6-9, and we immediately note good agreement between the pseudosteady-state WOR model and the measured production data. We also note good agreement for the conventional straightline (i.e., steady-state) extrapolation models. In this case, we do observe that the proposed pseudosteady-state WOR model best represents the "late time" performance. Our emphasis on the late-time data is warranted this region is where both the steady-state and pseudosteady-state models would be best expected to work. While we do not expect the pseudosteady state WOR relation to work at early times, it is worth noting that the early time data are dominated by oil flow (see Fig. 8). The pseudosteady-state relation presumes a constant mobility ratio, and this may not be the case where such a dramatic change in WOR performance occurs. Regardless, we are satisfied that the proposed pseudosteady-state relation does accurately represent the data for this case. As in the previous case (NRU Well 3106), we provide a direct comparison of the calculated and measured WOR functions for WWL Well B41 in Fig. 10, and we note also note in this case that the correlation of the calculated and measured WOR functions is excellent. For reference, the early time WOR data are not shown on Fig. 10, we have only plotted the WOR data that were actually used in the regression process. Summary Our goal in this section was to present and validate the proposed pseudosteady-state WOR relation. Recalling this result, we have: WOR = mo t o + b psso .................................................. (15) m w t w + b pssw

Substituting the definitions for material balance time for each phase, Eq. 15 becomes: mo mw Np qo Wp qw + b psso ..................................... (18) + b pssw

q WOR = w = qo

SPE 77569

Analysis and Interpretation of Water-Oil-Ratio Performance

In reviewing Eq.18, we immediately recognize that the rate and cumulative production functions can not be uncoupled, we can rearrange the variables to perhaps yield a more useful form, but the rate and cumulative functions remain. Multiplying through Eq. 18 by the (qo/qw) ratio gives: 1= or, m o N p + b psso q o m wW p + b pssw q w

recovery (EUR)) using the simple exponential rate decline relation. In particular, we utilize the following plotting techniques that are derived from the exponential rate decline relation: log(qo) versus t qo versus Np The log(qo) versus t plot is the most common production analysis mechanism unfortunately this plot is only rigorously valid for the case of liquid flow at a constant bottomhole flowing pressure. The qo versus Np plot is derived from the exponential rate decline relation, and also requires the assumption of a constant bottomhole pressure. The governing relation for the exponential production decline case is given by: (flowrate identity)

m wW p + b pssw q w = mo N p + b psso q o ........................ (19)


We note that although Eq. 19 could be considered a "more simple" form of Eq. 15, we can not reduce Eq. 19 further into a direct "analysis" relation. The point of this discussion is that the proposed pseudosteadystate WOR model (Eq. 15) is valid (at least for pseudosteadystate flow, presuming a constant mobility ratio). However, we can not reduce this approach into a direct analysis methodology we must use Eq. 15 as an analysis model, and fit that model to the field performance data. We are quite pleased with the performance of the pseudosteady-state WOR model (and its auxiliary models), and we believe that this approach is both robust and appropriate. While we did not obtain a direct solution or even an extrapolation formula from Eq. 15, we strongly recommend its application. At this point our recommendation is to use regression analysis it is our hope that future efforts yield a direct analysis method, in the form of a single-variable relation, or (perhaps) in the form of a "type curve" or some other type of comparative solution technique. ANALYSIS OF OIL AND WATER PRODUCTION DATA In this chapter we focus our efforts on a discussion of the "conventional" analysis methods for water and oil production data. In particular, we present methods that are commonly used in petroleum industry to estimate the recoverable oil volume (Np,max). The primary advantage of the methods we discuss is that these methods are straightforward we only require production data in order to estimate the recoverable oil volume and to make a production forecast. No tedious calculations are required, and most of the methods discussed are well accepted in the industry. The disadvantage is that most of these methods are not rigorous, and can fail (sometimes in a spectacular fashion) however, we believe that the practical value of these simple approaches makes an important contribution to the area of production data analysis. Analysis Methods This subsection addresses the inventory of analysis techniques that are used to quantitatively and qualitatively assess oil and water production data. Exponential Decline Curve Methods We begin our discussion with a focus on the estimation of the recoverable oil volume, Np,max, (or the estimated ultimate

qo = qoi e

Dit

..............................................................(20)

Integrating Eq. 20 to yield the cumulative production, then rearranging, we obtain the following relation:

qo = qoi Di Np .........................................................(21)
Conventional WOR Extrapolation Methods We will again discuss the various straight-line models proposed for the water-oil ratio (WOR) functions: WOR, the fractional flow of oil (fo), and the fractional flow of water (fw). These functions are plotted in various formats versus the cumulative oil production (Np) as indicated in the list of plotting functions given below. log(WOR) versus Np log(fw) versus Np fo versus Np The application of these functions was discussed previously however, our present goal is to apply and compare these functions with other analysis techniques. In order to utilize the fractional flow relations (fo and fw) we require these definitions. As such, the definitions of fo and fw are: fo = fw = qo ...............................................................(22) qo + qw qw ..............................................................(23) qo + qw

For the case of extrapolation using WOR versus Np and fw versus Np, we have the following implicit models for these presumed behaviors: WOR = ae f w = ae
bN p

.............................................................(24)

bN p

................................................................(25)

Another extrapolation technique that has been used extensively (but not investigated) is the case of a plot of fo versus Np, which should extrapolate to the recoverable reserves at fo =0. The presumed relationship for this case is

V.V. Bondar and T.A. Blasingame

SPE 77569

f o = a bNp ............................................................... (26)


New Methods for the Analysis of Production Data In addition to the methods previously identified for the analysis of water and oil production data, we also present two new techniques that can be used to estimate the recoverable oil, Np,max, or the oil-in-place, N. These new techniques use the following plotting functions: 1/fw versus Np 1/qo versus to (yields an estimate of Np,max) (yields an estimate of N)

Total Production Function Plots: log(WOR) versus total production (Np+Wp) log(fo) versus log(tt) Ershaghi X-Plot Technique Our final effort is to demonstrate the Ershaghi "X-plot" technique (refs. 4-5), and to compare the results of this method with the "conventional" (steady-state) straight-line extrapolation techniques.6 In all of the cases we considered, the X-plot technique gave the least consistent results compared to the other methods used. Based on our results, we believe that "conventional" straight-line extrapolation methods will produce more accurate (and more consistent) estimates of the recoverable oil than the X-plot technique. Analysis Examples In this subsection we demonstrate the analysis of production data using the techniques described above. The data for these examples are taken from corporate archives and deemed to be representative of the waterflood/water influx process. Given the large numbers of figures in this section, discussion will be limited to summary results for each "family" of analysis methodologies with exceptions as warranted. NRU 3106 Exponential Decline Curve Methods log(qo) versus t qo versus Np log(WOR) versus Np log(fw) versus Np fo versus Np 1/fw versus Np 1/qo versus to Log-Log Diagnostic Plots: WOR, WORd vs. t WORI, WORid vs t WOR, WORd vs to WORI, WORid vs to WOR, WORd vs tt WORI, WORid vs tt log(WORc) versus Np log(fwc) versus Np log(WOR) vs (Np+Wp) log(fo) versus log(tt) Ershaghi X-Plot Technique Np versus X (refs. 4-5) Fig. 45 Fig. 46 Fig. 25 Fig. 26 Fig. 27 Fig. 28 Fig. 29 Fig. 30 Fig. 37 Fig. 38 Fig. 41 Fig. 42 Fig. 31 Fig. 32 Fig. 33 Fig. 34 Fig. 35 Fig. 36 Fig. 39 Fig. 40 Fig. 43 Fig. 44 Fig. 11 Fig. 13 Fig. 15 Fig. 17 Fig. 19 Fig. 21 Fig. 23 Fig. 12 Fig. 14 Fig. 16 Fig. 18 Fig. 20 Fig. 22 Fig. 24 WWL B41

The 1/fw versus Np plotting function yields an apparent linear trend that can be extrapolated to provide an estimate of the recoverable oil, Np,max. In contrast, the 1/qo versus to plotting function yields a linear trend (predicted by pseudosteady-state theory), where the slope of the trend is proportional to 1/N. For a plot of 1/fw versus Np we implicitly assume the following relationship:

1/ f w = a bNp ........................................................... (27)


For a plot of 1/qo versus to = Np/qo we begin with Eq. 6, and upon rearranging, we have:

1 qo = a + b( Np qo ) ................................................... (28)
At the condition qo = 0, we can rearrange Eq. 28 to yield the recoverable reserves, Np,max. This result is given by:

Np ,max = 1/b ............................................................... (29)


Put simply, Eq. 29 specifies that we construct a plot of 1/qo versus Np/qo and obtain the slope (b) in order to estimate the recoverable reserves (Np,max). Qualitative Methods for the Analysis of Production Data We also provide a qualitative analysis of the behavior of the WOR, WOR derivative, integral, and integral-derivative functions plotted versus production time (t), oil material balance time (to), and total material balance time (tt). We provide a systematic presentation of examples for these plots later in this section. In particular, we review typical trends and characteristics of these plotting functions. The specific plotting functions considered in this discussion are as follows: Log-Log Diagnostic Plots: log(WOR) and log(WORd) versus log(t) log(WORi) and log(WORid) versus log(t) log(WOR) and log(WORd) versus log(to) log(WORi) and log(WORid) versus log(to) log(WOR) and log(WORd) versus log(tt) log(WORi) and log(WORid) versus log(tt) Cumulative WOR and Cumulative fw Plots: log(WORc) versus Np log(fwc) versus Np

Conventional WOR Extrapolation Methods

New Methods for the Analysis of Production Data

Qualitative Methods for the Analysis of Production Data

Cumulative WOR and Cumulative fw Plots:

Total Production Function Plots:

SPE 77569

Analysis and Interpretation of Water-Oil-Ratio Performance

Exponential Decline Curve Methods In Figs. 11 and 12 we confirm the log(qo) versus time technique for these data cases note that both cases have wellestablished semilog trends. Similarly, the rate-cumulative plots (Figs. 13 and 14) confirm the validity of the exponential de-cline model for these cases. Conventional WOR Extrapolation Methods The WOR versus Np plots for these cases are shown in Figs. 15 and 16 we note that both cases (NRU Well 3106 and WWL Well B41) illustrate a strong semilog trend. Similarly, the log(fw) versus Np plots (Figs. 17 and 18) also show very strong "late time" semilog straight line trends. Finally we consider the fo versus Np (Cartesian) plots (Figs. 19 and 20). Here we again find strong linear trends that yield consistent estimates of recoverable reserves (Np,max). New Methods for the Analysis of Production Data In this effort we "test" our proposed extrapolation/interpretation methods that can be used to estimate the recoverable reserves (Np,max). The first technique involves a plot of 1/fw versus Np (see Figs. 21 and 22) in both cases we again find strong linear trends in the data. In the second approach we use 1/qo versus Np/qo as the plotting function (Figs. 23 and 24). We find that this technique provides a remarkable "straightening" of the data for all cases considered in this work. The recoverable reserves are estimated using the slope of this trend, which can be problematic with erratic data (though this has generally not been the case in our experience). Qualitative Methods for the Analysis of Production Data In this effort we plotted the data functions for each case, then compared the variety of plots for each case (this is a "qualitative" comparison so our only effort is "interpretation," not analysis). In the first suite we consider Well NRU 3106 (Figs. 25-30) in terms of the "qualitative" log-log plots. This case is generally well behaved the "time" format plots are slightly erratic, though consistent (and changing) trends do evolve for the WORi and WORid functions. The WOR functions behave somewhat better for the "oil" and "total" material balance time functions, but the "shift" from one unit-slope trend to another is prevalent in each case. In Figs. 31-36 we consider the qualitative performance of WWL Well B41 where we immediately note that these data are not as well-behaved as the data for NRU Well 3106. Perhaps the most relevant observation is that a single unit-slope trend evolved for all of the WOR functions in the case of the "oil" material balance time function. This is the strongest trend observed for this case. Interestingly, the "cumulative" WOR and fw functions (WORc and fwc) provide apparent straight-line trends for both cases but other than this linearity, there is no clear interpretation of this behavior. This behavior is shown in Figs. 37-40. Our effort now shifts to comparisons of WOR and total production (Np+Wp) (see Figs. 41 and 43), as well as fo versus (Np+Wp)/(qo+qw) (see Figs. 42 and 44). The most relevant

comment for the log(WOR) versus (Np+Wp) trend is that an apparent semilog straight-line evolves however, the interpretation (and use) of this trend is unclear. It could be argued that the log-log plot of fo versus (Np+Wp)/(qo+qw) is simply an analog for a "rate-material balance time" plot since the trends are very similar to what we find when attempting to match production data to decline type curves. There may be an eventual application in that regard. Ershaghi X-Plot Technique As we note in both Figs. 45 and 46 the X-plot trends are very well established. The issue of analysis is relevant, at what value of fw do we extrapolate the plot? To demonstrate a failure of the X-plot method we provide the production history for WWL Well A17 in Fig. 47 note that not only is there no linear trend, the data appear to trending in the opposite direction of the proposed methodology suggests. This is hardly an indictment of the X-plot method, but we do believe that this case is "typical." Summary and Conclusions In this work we present empirical and semi-analytical models for the analysis and interpretation of oil and water production data. In particular, we provide the development, verification, and application of a new water-oil performance relation for pseudosteady-state flow conditions. This model was found to be superior to all other models considered for the representation of field production data (WOR, fw, and the water and oil flowrate functions). We provide a broad spectrum of demonstrative analyses for oil and water production data from "conventional" extrapolation plots to a several new analysis and diagnosis plots, including the new pseudosteady-state flow relation for the simultaneous flow of oil and water. The oil and water production data used in this study were obtained from the North Robertson Unit (NRU) located in Gaines County (West Texas) and the West White Lake Field located in Southwest Louisiana. Pseudosteady-State WOR Model We presented the development and validation of a pseudosteady-state water-oil ratio model. This model does not require the limiting assumptions of the conventional (steady state) WOR model (the mobility ratio is not required to be unity, nor are any assumptions made regarding the relative permeability functions). We utilized the new pseudosteady-state oil-water flow model in the following analysis/interpretation plots: Fractional flow of oil (fo) versus cumulative oil production (Np) Logarithm of fractional flow of water (fw) versus cumulative oil production (Np) Reciprocal of fractional flow of water (1/fw) versus cumulative oil production, (Np) Water-oil ratio (WOR) versus cumulative oil production (Np)

V.V. Bondar and T.A. Blasingame

SPE 77569

In each of these plots the pseudosteady-state flow model is plotted as a comparison to the given data function. We note in summary that the new pseudosteady-state model is an excellent interpolation model but we (regrettably) must also note that our new pseudosteady-state model does not provide a practical approach for the estimation of reserves by extrapolation. This is an area for future research consideration. Estimation of Recoverable Oil Though often considered an empirical (and sometimes suspect approach), the estimation of recoverable oil reserves (Np,max) by extrapolation of a function versus cumulative production (or some other production-related function) is a very popular (and effective) method for estimating such reserves. It is our belief that, because of the uncertainty in the accuracy of the "conventional" extrapolation methods, as well as the lack of a completely rigorous mathematical foundation, the best approach is to apply as many of these extrapolation techniques as possible. Such an approach will provide duplication (hence validation), and though there is no single "perfect" extrapolation technique, the comparison of results obtained from different approaches does provide consistency as well as an element of validation. Our process of estimating the ultimate recovery using straightline extrapolations of production data function incorporates a sequence of simultaneous analysis for all data functions (e.g., qo, fo, fw, WOR, 1/fw, 1/qo, and other functions versus cumulative production). The goal for such an analysis is to develop estimates of recoverable oil that best represent most, if not all, of the extrapolation methods. Specifically, we use the following extrapolation techniques for the simultaneous estimation of recoverable oil (or EUR) at current producing conditions: Oil rate (qo) versus production time (t) Oil rate (qo) versus cumulative oil production (Np) Fractional flow of oil (fo) versus cumulative oil production (Np) Logarithm of fractional flow of water (fw) versus cumulative oil production (Np) Reciprocal of fractional flow of water (1/fw) versus cumulative oil production (Np) Reciprocal of oil production rate (1/qo) versus oil material balance time (to) Qualitative Analysis of Oil and Water Production Data In addition to the estimation of recoverable oil, we also can also perform a "qualitative" analysis of the water and oil production data. These techniques provide a visual analysis of the data and may provide a qualitative assessment of the performance. While our focus is on the qualitative analysis of data, some of these techniques may also provide a quantitative estimate of recoverable oil. In particular, we considered the following analysis techniques: Logarithm of cumulative fractional flow of water (fwc) versus the cumulative oil production (Np)

Logarithm of cumulative water-oil ratio (WORc) versus the cumulative oil produc-tion (Np) Reciprocal of the water rate (1/qw) versus water material balance time (tw) Logarithm of the water-oil ratio (WOR) versus total production (Np+Wp) Logarithm of the fractional flow of oil (fo) versus logarithm of total material ba-lance time (t) Logarithm of the oil production rate (qo) versus logarithm of oil material balance time (to) Cumulative oil production versus X-function (refs. 5-6) Conclusions The following conclusions are derived from this study: Pseudosteady-State WOR Model
1. The proposed pseudosteady-state WOR model is a

combination of the analytical solutions for single-phase oil and water flow at pseudosteady-state conditions, and the only significant assumption that arises in this relation is that of a constant (non-unity) mobility ratio. While this may seem a limiting assumption, we successfully matched all of the production data cases considered in this work using this new model.
2. The proposed pseudosteady-state WOR model clearly

provides the best representation for the oil and water production data cases that we investigated.
3. The major limitation to our new model is that it does

not provide a mechanism for the prediction of future production this new model is expressed in terms of oil and water production rates and cumulative production, and it is not possible to write a predictive formula (or even an extrapolation formula) using this relation. Estimation of Recoverable Oil
1. In this work we provide a compilation of the "conven-

tional" straight-line extrapolation methods used in the industry for the estimation of recoverable oil. The techniques considered are:
log(qo) versus t qo versus Np log(fw) versus Np log(fo) versus Np

These techniques should be applied simultaneously in order to obtain consistent estimates of the recoverable oil. Application of only one or two techniques will likely lead to substantial misinterpretations of the correct model behavior, which, in turn, will lead to overor underestimation of recoverable reserves.
2. We proposed two new methods for estimating recover-

able oil reserves. These techniques are:


1/fw versus Np 1/qo versus Np/qo

A plot of 1/fw versus Np should yield a straight-line trend that can be extrapolated to 1/fw=1, which yields

SPE 77569

Analysis and Interpretation of Water-Oil-Ratio Performance

the recoverable reserves, Np,max. Similarly, a plot of 1/qo versus Np/qo should yield a straight-line trend where the slope of this trend is equal to 1/Np,max. The results obtained by these new methods correspond quite well to the results obtained by the "conventional" WOR techniques (i.e., the straight-line extrapolation methods discussed in the previous point). Analysis of Oil and Water Production Data
1. We have found that plotting the cumulative water-oil

ratio (WORc) function and/or the cumulative fractional flow of water (fwc) functions versus the cumu-lative oil production (Np) does typically yield a straight-line trend (in most cases we considered). However, the extrapolation of the observed straight-line trend does not yield a consistent estimate of recoverable re-serves (when compared to other techniques). This behavior presents an opportunity for future work.
2. We have extended the diagnostic plots proposed by

Nomenclature a = intercept of log(kro/krw) versus Sw plot, fraction A = reservoir drainage area, ft2 [m2] ac = intercept of conventional log(WOR) versus Np plot, fraction aer = intercept of Ershaghi4,5 Np versus X-function plot, STB [std m3] B = formation volume factor, RB/STB [res m3/std m3] b = slope of of log(kro/krw) versus Sw plot, 1/STB [1/std m3] bc = slope of conventional log(WOR) versus Np plot, 1/STB [1/std m3] ber = slope of Ershaghi4,5 Np versus X-function plot, STB [std m3] B 4A , intercept of p/q versus t ln bpss = 70.6 2 kh e C A rw plot for general variable-rate case, psi/STB/D [kPa/std m3/d] reservoir shape factor, dimensionless total system compressibility, psia-1 [kPa-1] decline oil rate, STB/D [std m3/d] Np/N, overall reservoir recovery, fraction fractional flow of oil, fraction cumulative fractional flow of oil, fraction fractional flow of water, fraction cumulative fractional flow of water, fraction total formation thickness, ft [m] effective formation permeability, md relative formation permeability, md B 0.2339 , slope of p/q versus t plot for hct A general variable-rate case, psi/STB/D/hr [kPa/std m3/d/hr] cumulative oil production, STB [std m3] recoverable oil, STB [std m3] original oil-in-place, STB [std m3] pi- pwf, pressure drop, psi [kPa] pressure, psia [kPa] initial reservoir pressure, psia [kPa] pseudosteady-state flowing bottomhole pressure, psia [kPa] production flowrate, STB/D [std m3/d] initial production flowrate, STB/D [std m3/d] reservoir drainage radius, ft [m] wellbore radius, ft [m] water saturation, fraction connate water saturation, fraction irreducible water saturation, fraction time, days material balance time, days Np/qo, oil material balance time, days (Np+Wp)/(qo+qw), total material balance time, days Wp/qw, water material balance time, days cumulative water injection, STB [std m3] water-to-oil ratio, fraction Wp/Np, cumulative WOR function, fraction

Chan2 (i.e., the WOR and WOR derivative, integral, and integral-derivative functions plotted versus production time, oil material balance time, and total material balance time). The following observations are noted: WOR and WOR integral, and integral-derivative functions typically exhibit a clearly defined unit slope trend when plotted versus time or material balance time. The WOR derivative function is typically very erratic, and can not be used for routine analysis due to poor overall behavior. Log-log plots of the WOR functions versus t (or to) tend to be reasonably well behaved with the noted exception of the WOR-derivative function, which is typically very erratic. It is our strong recommendation that these plots only be used for diagnostic purposes although we would encourage a future study focusing on a "type curve" analysis approach for WOR data.
3. We believe that the X-plot method (refs. 4-5) provides
The

CA ct Di Er fo foc fw fwc h k kr mpss

= = = = = = = = = = = =

no substantive advantage over the "conventional" extrapolation techniques discussed previously. In fact, the X-function plot typically does not develop a clear straight-line trend. Of most concern is the observation that the estimates of recoverable oil obtained using the X-plot do not typically correspond to the estimates obtained by the conventional and proposed extrapolation techniques. Finally, the extrapolation of X-function tends to significantly overestimate the value of recoverable oil.

Np = Np,max = OOIP = p = p = pi = pss = pwf = q = qi = re = rw = Sw = Swc = Swirr = t = tm = to = tt = tw = Wi = WOR = WORc =

10

V.V. Bondar and T.A. Blasingame

SPE 77569

WORd = WORi =

d (WOR) ( x = t , to , tt , etc.) , fraction dx

1 x WOR dx ( x = t , to , tt , etc.) , fraction x 0 d (WORi ) ( x = t , to , tt , etc.) , fraction WORid = x dx Wp = cumulative water production, STB [std m3] 1 1 4,5 X = ln f 1 f , Ershaghi X-function w w

Greek: = 0.577216, Eulers constant = reservoir porosity, fraction = fluid viscosity, cp [Pa*s] Subscripts: i = initial o = oil w = water t = total References 1. Robert S. Thompson, John D. Wright: Oil Property Evaluation, Thomson-Wright Associates, Golden, CO, 1985. 2. Chan, K.S.: "Water Control Diagnostic Plots," paper SPE 30775 presented at the 1995 Annual Technical Conference and Exhibition, Dallas, TX, 22-25 October.. 3. Blasingame, T.A. and Lee, W.J.: "Variable-Rate Reservoir Limits Testing," paper SPE 15028 presented at the SPE Permian Basin Oil and Gas Recovery Conference, Midland, TX, 13-14 March 1986. 4. Ershaghi, I. and Omoregie, O.: "A Method for Extrapolation of Cut vs. Recovery curves," JPT (February 1978) 203-204. 5. Ershaghi, I. and Abdassah, D.: "A Prediction Technique for Immiscible Processes Using Field Performance Data," JPT (April 1984) 664-670. 6. Startzman, R.A and Wu, C.H: "Discussion of Empirical Prediction Technique for Immiscible Processes," JPT (December 1984) 2192-2194. Acknowledgements The authors would like to acknowledge the in-kind contibutions of Fina Oil and Chemical Co. (now TotalFinaElf) and Unocal (now Spirit Energy) for the field data used in this work.

Figure 1 Fractional Flow of Water Versus Cumulative Oil Production, NRU Well 3106.

Figure 2 Reciprocal of Fractional Flow of Water Versus Cumulative Oil Production, NRU Well 3106.

Figure 3 Fractional Flow of Oil Versus Cumulative Oil Production, NRU Well 3106.

SPE 77569

Analysis and Interpretation of Water-Oil-Ratio Performance

11

Figure 6 Fractional Flow of Water Versus Cumulative Oil Production, WWL Well B41.

Figure 4 Water-Oil Ratio Versus Cumulative Oil Production, NRU Well 3106.

Figure 7 Reciprocal of Fractional Flow of Water Versus Cumulative Oil Production, WWL Well B41.

Figure 5 Comparison of the Calculated WOR Performance (Pseudosteady-State Model) with the Measured WOR Performance, NRU Well 3106. Figure 8 Fractional Flow of Oil Versus Cumulative Oil Production, WWL Well B41.

12

V.V. Bondar and T.A. Blasingame

SPE 77569

Figure 11 Oil and Water Production Rate History, NRU Well 3106.

Figure 9 Water-Oil Ratio Versus Cumulative Oil Production, WWL Well B41.

Figure 12 Oil and Water Production Rate History, WWL Well B41.

Figure 10 Comparison of the Calculated WOR Performance (Pseudosteady-State Model) with the Measured WOR Performance, WWL Well B41.

Figure 13 Oil Production Rate Versus Cumulative Oil Production, NRU Well 3106.

SPE 77569

Analysis and Interpretation of Water-Oil-Ratio Performance

13

Figure 14 Oil Production Rate Versus Cumulative Oil Production, WWL Well B41.

Figure 17 Fractional Flow of Water Versus Cumulative Oil Production, NRU Well 3106.

Figure 18 Fractional Flow of Water Versus Cumulative Oil Production, WWL Well B41.

Figure 15 Water-Oil Ratio Versus Cumulative Oil Production, NRU Well 3106.

Figure 19 Fractional Flow of Oil Versus Cumulative Oil Production, NRU Well 3106.

Figure 16 Water-Oil Ratio Versus Cumulative Oil Production, WWL Well B41.

14

V.V. Bondar and T.A. Blasingame

SPE 77569

Figure 20 Fractional Flow of Oil Versus Cumulative Oil Production, WWL Well B41.

Figure 23 Reciprocal of Oil Rate Versus Oil Material Balance Time, NRU Well 3106.

Figure 21 Reciprocal of Fractional Flow of Water Versus Cumulative Oil Production, NRU Well 3106.

Figure 24 Reciprocal of Oil Rate Versus Oil Material Balance Time, WWL Well B41.

Figure 22 Reciprocal of Fractional Flow of Water Versus Cumulative Oil Production, WWL Well B41.

Figure 25 Water-Oil Ratio and Water-Oil Ratio Derivative Versus Production Time, NRU Well 3106.

SPE 77569

Analysis and Interpretation of Water-Oil-Ratio Performance

15

Figure 26 Water-Oil Ratio Integral and Integral-Derivative Versus Production Time, NRU Well 3106.

Figure 28 Water-Oil Ratio Integral and Integral-Derivative Versus Oil Material Balance Time, NRU Well 3106.

Figure 27 Water-Oil Ratio and Water-Oil Ratio Derivative Versus Oil Material Balance Time, NRU Well 3106.

Figure 29 Water-Oil Ratio and Water-Oil Ratio Derivative Versus Total Material Balance Time, NRU Well 3106.

16

V.V. Bondar and T.A. Blasingame

SPE 77569

Figure 30 Water-Oil Ratio Integral and Integral-Derivative Versus Total Material Balance Time, NRU Well 3106.

Figure 32 Water-Oil Ratio Integral and Integral-Derivative Versus Production Time, WWL Well B41.

Figure 31 Water-Oil Ratio and Water-Oil Ratio Derivative Versus Production Time, WWL Well B41.

Figure 33 Water-Oil Ratio and Water-Oil Ratio Derivative Versus Oil Material Balance Time, WWL Well B41.

SPE 77569

Analysis and Interpretation of Water-Oil-Ratio Performance

17

Figure 34 Water-Oil Ratio Integral and Integral-Derivative Versus Oil Material Balance Time, WWL Well B41.

Figure 36 Water-Oil Ratio Integral and Integral-Derivative Versus Total Material Balance Time, WWL Well B41.

Figure 35 Water-Oil Ratio and Water-Oil Ratio Derivative Versus Total Material Balance Time, WWL Well B41.

Figure 37 Cumulative Water-Oil Ratio Versus Cumulative Oil Production, NRU Well 3106.

Figure 38 Cumulative Fractional Flow of Water Versus Cumulative Oil Production, NRU Well 3106.

18

V.V. Bondar and T.A. Blasingame

SPE 77569

Figure 39 Cumulative Water-Oil Ratio Versus Cumulative Oil Production, WWL Well B41.

Figure 42 Fractional Flow of Oil Versus Total Material Balance Time, NRU Well 3106.

Figure 40 Cumulative Fractional Flow of Water Versus Cumulative Oil Production, WWL Well B41.

Figure 43 Water-Oil Ratio Versus Total Production, WWL Well B41.

Figure 41 Water-Oil Ratio Versus Total Production, NRU Well 3106.

SPE 77569

Analysis and Interpretation of Water-Oil-Ratio Performance

19

Figure 47 X - Plot, WWL Well A17 (example failure of the XPlot Method).

Figure 44 Fractional Flow of Oil Versus Total Material Balance Time, WWL Well B41.

Figure 45 X-Plot, NRU Well 3106.

Figure 46 X-Plot, WWL Well B41.

You might also like