You are on page 1of 57

SPE 132352 Continuous Estimation of Ultimate Recovery

S.M. Currie, Texas A&M University, D. Ilk, Texas A&M University, and T.A. Blasingame, Texas A&M University
Copyright 2010, Society of Petroleum Engineers This paper was prepared for presentation at the SPE Unconventional Gas Conference held in Pittsburgh, Pennsylvania, USA, 2325 February 2010. 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 have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper 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 SPE copyright.

Abstract Gas in place/reserves estimation in unconventional (low/ultra-low permeability) reservoirs has become a topic of increased interest as more of these resources are being developed domestically and internationally. Production data from these unconventional gas reservoirs exhibit extended periods of transient/transition flow behavior that often lead to the overestimation of gas in place/reserves with the use of simple rate-time extrapolation techniques. In this work we show that the analysis of production data (particularly rate-time data) before the onset of boundary dominated flow for these unconventional gas reservoirs leads to significant overestimation of gas in place/reserves. Consequently, we introduce the concept of "continuous estimation of ultimate recovery" (or "continuous EUR") where estimation of reserves from a single gas well is performed in a dynamic fashion in other words we continuously estimate the reserves for selected time intervals throughout the producing life of the well. We have applied the "continuous EUR" method by using simple rate-time relations such as the Arps' "hyperbolic" rate decline relation and the power-law exponential rate decline relation along with the straight line extrapolation technique, which provides a lower limit for EUR. We present the application of the "continuous EUR" method with a simulated tight gas data set and four (tight gas) field data sets. Our analyses show that the distinction in the flow behavior is more evident and the uncertainty in reserves estimation decreases significantly when reserves are evaluated continuously. Introduction Estimation of petroleum reserves is one of the most important petroleum engineering areas of research since 1910's. The use of only production data (i.e., rate-time data) to estimate the reserves has been an industry practice since the introduction of the Manual for the Oil and Gas Industry under the Revenue Act of 1918 by the United States Internal Revenue Service [1919]. Lewis and Beal [1918] provides initial guidance on the extrapolation of future production of oil wells. Cutler [1924] presents a detailed review of the oil reserves estimation procedures in 1920's. Johnson and Bollens [1928] introduce the loss ratio and the loss ratio derivative definitions which lay the basis for the exponential and the hyperbolic rate decline relations. Arps [1945] presents the empirical exponential and hyperbolic rate decline relations, which are still the widely used relations for production extrapolations of oil and gas wells, and can be assumed as valid for a variety of producing conditions for practical purposes. Arps' relations are only applicable for the boundary-dominated flow regime and the improper use of Arps' equations can yield inconsistent results and erroneous reserve estimates. Rushing et al [2007] presents a study which was designed to assess the validity of estimating reserves using the hyperbolic rate decline relation. Rushing et al's approach shows that the errors in reserve estimates based on the hyperbolic rate decline relation could be substantially higher (over 100 percent error) when the hyperbolic rate decline relation is applied during transient/transition flow regimes. The power-law exponential rate decline relation has recently been introduced in the publication by Ilk et al [2008]. The definitions of the "loss-ratio" and the "loss-ratio derivative" describe the basis of the "power-law exponential" rate equation. On the other hand, the works by Fetkovich (Fetkovich [1980]) and Fetkovich et al (Fetkovich et al [1987]) discuss the use of the Arps' hyperbolic relations along with a semi-analytical result for gas flow behavior which is only valid for constant flowing bottomhole pressure condition. Carter [1981] presents a new set of decline type curves for gas reservoir systems to address the changes in fluid properties with reservoir pressure. But, this method also suffers from the requirement of a constant flowing bottomhole pressure. These methodologies lay the groundwork for theoretically based modern production analysis methodologies which require the use of time-pressure-rate data.

S.M. Currie, D. Ilk, and T.A. Blasingame

SPE 132352

The interested reader is referred to the works by Anderson et al [2006] and Ilk et al [2007] for a detailed description and literature survey for modern production data analysis. We note that we only use the rate-time relations for our purposes in this work. To obtain a theoretically based rate-time relation (without the use of bottomhole pressures), Ansah et al [2000] use a simple linearization scheme, and develop a semi-analytical direct solution for determining average reservoir pressure, rate and cumulative production for gas wells produced at a constant bottomhole pressure during reservoir depletion. Blasingame and Rushing [2005] extend the applicability of Ansah et al relations by developing new plotting functions for direct estimation of contacted gas-in-place using only production data. It should be noted that these relations are applicable for the boundarydominated flow regime. Again we note that Ansah et al relations are not utilized in this work. The most important objectives of this work are to reduce the uncertainty in reserves estimation, and to define a consistent and systematic methodology for production extrapolation in unconventional reservoirs using simple rate-time equations. The estimation of reserves in unconventional reservoir systems is problematic due to the extremely complex geology, and because of the very low permeability of the reservoirs, longer transient flow periods are being exhibited throughout the producing life of the well. Under these circumstances the uncertainty in reserves estimation is high and usually the reserves are overestimated with the use of conventional rate-time relations. From a practical standpoint, Arps' rate-time relations are used in order to estimate reserves in unconventional reservoir systems. These relations (i.e., Arps' rate-time relations) are very convenient to use, but they usually over-predict the reserves as they are only applicable for the boundary-dominated flow regime whereas unconventional reservoirs exhibit very long transient flow periods (years or even decades). In this work we propose the use of the simple rate-time relations and a straight line extrapolation technique in a dynamic (continuous) manner to reduce the uncertainty in reserve estimates. We specify a time interval amongst the production sequence, and for each interval we perform extrapolation of the simple rate-time relations to obtain the maximum production (i.e., the estimation of ultimate recovery). For example, for 500 days of production data (i.e., rate-time data) we may select a 100-day time interval and the maximum production would be estimated using 100 days, 200 days, 300 days, 400 days, and 500 days of production data (in principle, the interval is arbitrary it could even be daily, but such a small interval would require an automated procedure, which is not addressed in this paper). In addition we use the straight line extrapolation technique to provide a conservative estimate for reserves which we use as a "lower" limit for EUR. We choose to call this procedure the continuous estimation of reserves or continuous EUR approach. For our purposes we only use the Arps' hyperbolic rate-time and the power-law exponential rate-time relations in addition to the straight line extrapolation technique for the continuous estimation of reserves. In the following subsections we provide short descriptions on the rate-time decline models that are being used in this work. Hyperbolic Rate Decline Function: The most common method of well performance (i.e. rate-time data) analysis involves the empirical rate-time relations presented by Arps [1945]. As mentioned before these equations were developed to analyze data exhibiting boundarydominated flow behavior. Arps' equations are widely used to forecast production behavior and estimate reserves. The decline behavior is described by the loss ratio, 1/D, and the derivative of the loss ratio, b, given by:
qg 1 = ...................................................................................................................................................................... (1) D dq g / dt

b=

d 1 d qg = ................................................................................................................................................ (2) dt D dt dq g / dt

When the derivative of the loss ratio, b is constant then the hyperbolic rate decline relation is obtained and this relation is given as:
q g (t ) = q gi (1 + bDi t )1 / b ............................................................................................................................................................ (3)

From the definition of the loss ratio (Eq. 1), the D-parameter for the hyperbolic rate decline relation can be formulated as: D(t ) = Di .................................................................................................................................................................. (4) (1 + bDi t )

SPE 132352

Continuous Estimation of Ultimate Recovery

The decline parameter, b, is defined as the derivative of the "loss-ratio" with respect to time (i.e. d/dt[1/D]). The b-parameter is constant for the hyperbolic rate-decline relation and should lie between 0 and 1, but we often observe values greater than 1 especially during the transient and transition flow periods for unconventional gas reservoirs. Extrapolation of the rate-time model prior to the onset of boundary-dominated flow results in the overestimation of reserves (see Rushing et al [2007] for details). We integrate Eq. 3 to obtain a cumulative production relation and the result is given as:

G p (t ) =

q gi (1 b) Di

[1 (1 + bDi t )1 (1 / b) ] ............................................................................................................................. (5)

For the calculation of maximum production (Gp,max), we take the limit of Eq. 5 to infinite time. q gi G p, max = Lim [1 (1 + bDi t )1 (1 / b ) ] .................................................................................................................. (6) ( 1 b ) D t i When the b-parameter is bounded between 0 and 1 (i.e., 0<b<1), the maximum production is (Gp,max) equal to:

G p, max =

q gi (1 b) Di

(for 0<b<1)..................................................................................................................................... (7)

It is worth mentioning that if the b-parameter is greater than one (i.e., b>1), Eq. 6 converges to infinity. As a result estimation of reserves is not theoretically possible unless a limiting time or abandonment rate is specified. Power-Law Exponential Rate Decline Function: Ilk et al [2008] proposed the power-law exponential rate decline relation particularly for the estimation of reserves in unconventional reservoir systems. The D- and b-parameters are defined in functional forms, which are suggested by the character of the data when D- and b-parameters are computed continuously from data. The functional forms of the D- and bparameters are given as: nt n 1 ........................................................................................................................................................... (8) D(t ) = D + D i b(t ) = (n 1)nt n D i .......................................................................................................................................................... (9) nt n ) 2 ( D t + D i

In contrast with the "hyperbolic" behavior of the "loss-ratio" function (Eq. 4), the power-law exponential "loss-ratio" function (Eq. 8) exhibits power-law behavior at early times and becomes constant at large times. The power-law exponential rate decline relation is obtained by solving Eq. 8 and is given as:
t n ] ........................................................................................................................................... (10) gi exp[ D t D q g (t ) = q i Ilk et al verify that the power-law exponential model is flexible enough to model transient, transition and boundarydominated flow by applying the model to numerically simulated and field data sets. The D term controls the late time behavior often exhibiting boundary-dominated flow behavior. For data sets that do not exhibit boundary-dominated flow, the D term can be set to 0. Johnson et al [2009] present further application of the model using type curve solutions for the power law exponential rate-time relation. Since direct integration of Eq. 10 is not possible, cumulative production must be calculated using numerical methods. Straight-Line Extrapolation using Rate-Cumulative Production: The purpose of straight line extrapolation in our methodology is to provide a "lower-limit" for the estimate of ultimate recovery. Arps' exponential rate decline relation is given as:
q g (t ) = q gi exp[ Di t ] ......................................................................................................................................................... (11)

Eq. 11 can easily be derived by assuming that the loss ratio (i.e., Eq. 1) is constant and solving the differential equation (Eq. 1). The "rate-cumulative production" relation of the Arps' exponential rate decline is also given as:
q g (t ) = q gi Di G p (t ) ........................................................................................................................................................ (12)

S.M. Currie, D. Ilk, and T.A. Blasingame

SPE 132352

Eq. 12 indicates that maximum gas production is obtained when the flowrate approaches to zero (i.e., qg(t) 0, Gp(t) Gp,max). In fact Eq. 12 can rigorously be derived for the black oil (slightly compressible liquid) case by combining the black oil material balance and black oil pseudosteady-state flow relation with constant flowing bottomhole pressure assumption (see Blasingame and Rushing [2005] and Reese et al [2007] for details on the derivation). We note that Eq. 12 should not be considered applicable for the gas flow case as nothing but a "very late transient" or "onset of boundary-dominated flow" approximation. Therefore, we may use Eq. 12 for the purpose of finding a lower-limit for maximum gas production (or estimated ultimate recovery). It is also important to state that neither EUR nor Gp,max is the fluids-in-place, but rather the quantity of fluids that will be produced at infinite time.

Development of the Methodology


We develop the continuous EUR procedure to identify the error associated with the long transient flow periods when reserves are being estimated in unconventional gas reservoirs at early times. From this approach we are able to quantify the change in EUR estimates as a function of time. In this section we describe the details about our workflow on the continuous estimation of ultimate recovery.

Fig. 1 shows the first two steps of the workflow for the "Continuous EUR" procedure. We start by obtaining rate-time data for an individual well. Prior to performing analysis, the rate-time data is edited and any points off trend from the dominant production profile are removed. The data editing process is critical since the noise in the production data is significantly amplified in the D-parameter and the b-parameter calculations (recall that D-parameter and b-parameter calculations require numerical differentiation of the data set and the Bourdet algorithm [1989] is used for numerical differentiation purposes in this work.).
The next step in the workflow is to create subsets of the production data (i.e. flowrate versus time). Each subset starts from the beginning of the production history (i.e. t = 0 days) and is analyzed individually with a rate-time relation (e.g., "hyperbolic" and power-law exponential).

Fig. 1 First step of the continuous EUR procedure: Data is reviewed and edited, and intervals are selected.

SPE 132352

Continuous Estimation of Ultimate Recovery

Fig. 2 presents the next step, which is the rate-time analysis procedure in our workflow. Each subset of the data is matched with the "hyperbolic" and power-law exponential relations and an EUR value is estimated by extrapolating the rate-time model to a specified time limit or abandonment rate. We note that for all the cases in this work, we have specified a time limit of 30 years when calculating EUR. In this work we have analyzed the flowrate data with the "hyperbolic" and powerlaw exponential rate decline relations, but this concept can be implemented with other rate-time relations as well.
In addition to the "hyperbolic" and power-law exponential relations, we use the straight line extrapolation technique to estimate a "lower" limit for EUR. We plot gas flowrate versus cumulative gas production for an individual well on a Cartesian scale. And for each previously specified interval, we fit a straight line through the part of the data exhibiting linear trend. The x-axis intercept of the line yields an estimate for Gp,max. As mentioned earlier this method can be used to estimate the maximum oil production for oil wells under the boundary-dominated flow regime. It is important to note that by applying this technique a conservative estimate for Gp,max can be obtained which yields a lower limit estimate for EUR. When short period intervals (say 30 days, 60 days etc.) are selected then Gp,max estimates are observed to increase with time particularly for the case of unconventional gas reservoirs. On the other hand, Gp,max estimate can also be obtained by limiting the interval to only two points. The derivative of flowrate (i.e., Eq. 12) is taken with respect to cumulative production. This provides the Di term in Eq. 12 in other words slope of the straight line passing through the two points is found. Next the qgi term in Eq. 12 is solved for and finally we extrapolate the straight line to the x-axis (i.e., qg 0) for the Gp,max value.

Fig. 2 Second step of the continuous EUR procedure: all of the intervals are evaluated and an EUR estimate is obtained using the rate-time relations and straight line extrapolation.

S.M. Currie, D. Ilk, and T.A. Blasingame

SPE 132352

The final step in the procedure is shown by Fig. 3. The EUR values obtained from the extrapolation of the rate-time relations and the straight line extrapolation technique are plotted together as a function of time and this EUR versus time plot provides a range for where the true EUR value should lie. We suggest that the EUR estimate from the power law exponential model can be used as an upper limit and the EUR estimate from the straight line extrapolation can be used as a lower limit for reserves. We also note that with increasing time the difference between upper and lower limits would reduce.

Fig. 3 Final step of the continuous EUR procedure: EUR and Gp,max values are plotted versus time and upper and lower limits for EUR are established.

Validation of New Analysis Methodology Numerical Simulation Study


In this section we verify our methodology by applying the continuous estimation concept to a numerically simulated case. The simulated data is obtained from the work by Ilk et al [2008], which is given as the simulated East Tx gas well example. This case is for a vertical well with a single vertical hydraulic fracture of finite conductivity which produces from a tight gas reservoir. The model parameters including reservoir and fluid properties were obtained from the work by Pratikno et al [2003] and are provided in Table 1. The synthetic gas flowrate data is generated using a numerical simulator by Ilk et al [2008]. We present the flowrate data and the cumulative production data in Fig. 4 (the entire simulation run, approximately 6 years of data, are presented). Our first task is to match the subsets of the flowrate data with the "hyperbolic" rate decline relation. In Fig. 5 we present the "hyperbolic" model matches imposed on the flowrate data, and the D- and b-parameter trends. For this case, we note that the model parameter qgi is held constant when matching the rate-time data with the "hyperbolic" relation we choose to fit the equation to the late part of data for each interval rather than fitting the equation to the entire interval as the "hyperbolic" relation is only valid for the boundary-dominated flow regime. We observe that at early times the data is matched with a "hyperbolic" b-parameter greater than 1 (b values greater than 1 may indicate the transient flow period). In Fig. 6 we observe the change in the b-parameter values over time; the b-parameter value decreases significantly at early times and begins to stabilize after about 1,000 days of production due to the onset of boundary dominated flow, and it seems to be constant after the complete boundary-dominated flow regime is established.

SPE 132352

Continuous Estimation of Ultimate Recovery

Table 1 Reservoir and fluid properties for numerical simulation case (single layer tight gas well with hydraulic fracture).

Reservoir Properties: Wellbore Radius, rw Estimated net pay thickness, h Average porosity, Average irreducible water saturation, Swirr Permeability, k Fluid Properties: Gas formation volume factor at pi, Bgi Gas viscosity at pi, gi Gas compressibility at pi, cgi Production Parameters: Initial reservoir pressure, pi Original gas-in-place, OGIP

= = = = = = = = = =

0.333 ft 170 ft 0.088 (fraction) 0.131 (fraction) 0.005 md 0.5498 RB/MSCF 0.0361 cp 5.1032x10-5 psi-1 9,330 psia 2.65 BSCF

Our next task is to match the subsets of the flowrate data with the power-law exponential rate decline relation. In Fig. 7 we present the power law exponential model matches imposed on the flowrate data, and D- and b-parameter trends. We observe that the model match stabilizes and becomes identical for each interval after 1,000 days of production. We attribute this result to the boundary-dominated flow regime effects in fact, we expect that once complete boundary-dominated flow regime is established, rate-time model match would be identical for each additional interval. Also we observe that the powerlaw exponential rate decline relation matches the transient, transition, and boundary-dominated flow behavior very well. We use the straight line extrapolation technique to estimate a lower limit for EUR. In Fig. 8 we show the results of the straight line extrapolation for this case. The x-axis intercept of the lines (i.e., continuous estimation of the Gp,max) increase with time resulting in an increasing estimate of Gp,max. Again it is worth to mention that during the boundary-dominated flow regime Gp,max estimates do not seem to change significantly over time. The last step in our workflow is to calculate the EUR based on the matches obtained with the "hyperbolic" and the power-law exponential rate decline relations. In Fig. 9 we present the calculated EUR values versus production time. We observe that the EUR values obtained from the "hyperbolic" model matches decrease significantly at early times and then converge at late times to a value of 2.55 BSCF. The EUR values obtained from the power-law exponential model matches also decrease at early times but become constant at a value of 2.41 BSCF after 750 days of production. From the reservoir simulation model, the gas produced at 30 years is 2.42 BSCF, which is consistent with the EUR estimated using the power-law exponential model. The EUR values estimated from the power-law exponential model matches are more conservative than the value estimated from the hyperbolic model matches. We also observe that the EUR values obtained from both models converge or become constant once boundary-dominated flow has been established. The Gp,max values obtained from the straight line extrapolation technique are shown to increase with time and never reach the actual cumulative production (Gp) value (2.42 BSCF) during the 2,000 days of production. This result confirms that the straight line technique may not be applicable for gas flow however; it can be used to provide a "lower" limit for EUR for practical purposes. Finally we present all of the model parameters and the EUR and Gp,max values for each interval in Table 2.

Application of the Continuous Estimation of Ultimate Recovery Methodology to Field Data


In this section we present four field examples on the application of the proposed continuous EUR methodology. The field data include rate-time data of hydraulically fractured wells completed in a low permeability (tight) gas reservoir (k<0.01 md). Field Case 1: We apply our proposed methodology to a field data set acquired from a hydraulically fractured well completed in a tight gas reservoir. We present the flowrate data and the cumulative production data which spans almost 5.5 years in Fig. 10. Prior to performing analysis, the data is edited and any points off trend from the dominant production profile are removed. As a side note, the off trend rate data might be caused by liquid-loading effects and/or operational changes. Our first task is to match the subsets of the flowrate data with the "hyperbolic" rate decline relation. Fig. 11 presents the "hyperbolic" model matches imposed on the flowrate data along with the D- and b-parameter trends. We use the same matching technique as previously described in the simulated example, and hold qgi constant while matching the end portion of the data for each selected interval. We note that early portion of the data is not matched since the "hyperbolic" relation is only applicable for the boundary-dominated flow regime.

S.M. Currie, D. Ilk, and T.A. Blasingame

SPE 132352

In Fig. 12 we observe that the value of the b-parameter continues to decrease over the entire production history and does not appear to stabilize. We observe every subset (or interval) which is matched with a "hyperbolic" b-parameter greater than 1 indicates that the boundary-dominated flow regime has not been established yet. Specifically the b-parameter value decreases from 2.20 to 1.57 during the production history this result may indicate that transient flow effects are still dominant in the production behavior. However, the effects of transition to boundary-dominated flow regime start to be seen. Our next task is to match the subsets of the flowrate data with the power-law exponential rate decline relation. In Fig. 13 the power law exponential model matches imposed on the flowrate data and D- and b-parameter trends are shown. We note that for all matches a D value is not used indicating that boundary-dominated flow character is not observed (the behavior of the calculated D-parameter data trend may serve as a validation for not using the D parameter in the power law exponential relation as it exhibits only power law behavior for this case). We observe that the model matches stabilize and become relatively stable after about 1,500 days of production for each additional interval. We use the straight line extrapolation technique to estimate Gp, max. In Fig. 14 we show the lines fit through the end portion of the data for each interval. The x-axis intercept of the lines increase with time resulting in an increasing estimate of Gp,max or "lower" limit estimate of EUR. All of the model parameters for this example are presented in Table 3. The last step in our workflow is to plot the EUR based on the matches obtained with the "hyperbolic" and the power-law exponential rate decline relations versus time. In Fig. 15 we present the calculated EUR values versus production time. We observe that the EUR estimates obtained from the "hyperbolic" relation significantly overestimate reserves at early times and converge at late times to a value of 3.10 BSCF. The estimates provided by the power law exponential model appear to be constant at 3.05 BSCF after 1,500 days. We expect lower EUR values to be estimated by the rate-time relations when complete boundary-dominated flow regime is established later in the life of this well. The Gp,max values obtained from the straight line extrapolation seem to increase with time are always less than the estimates provided by the rate-time relations. The EUR of this well should be in between 2.58 BSCF (the "lower" limit given by the straight line extrapolation technique at 2,000 days) and 3.05 BSCF (the "upper" limit given by the power law exponential estimate at 2,000 days). Field Case 2: We apply our proposed methodology to a second field data set acquired from a hydraulically fractured well completed in a tight gas reservoir. We present the flowrate data and the cumulative production data which spans almost 5.5 years in Fig. 16. In this case we observe the effects of liquid loading and operational changes (specifically at 1,000 days of production) on the flowrate profile as well. The subsets of the flowrate data are matched with the "hyperbolic" rate decline relation initially. In Fig. 17 we present the "hyperbolic" model matches imposed on the flowrate data, and D- and b-parameter trends. We use the same matching technique as previously described and hold qgi constant while matching the end portion of the data where boundarydominated flow regime assumption may be made. We observe every subset of the data is matched with a "hyperbolic" bparameter greater than 1 indicating that complete boundary-dominated flow regime effects are not established yet. In Fig. 18 we observe that the b-parameter is relatively stable after about 250 days of production with a slight inconsistency occurring after about 1,000 days of production. We believe that this issue can be attributed to the changing production profile noted earlier. The b-parameter value seems to decrease from 2.37 to 1.90 during the entire production history. Next we match the selected subsets of the flowrate data with the power-law exponential rate decline relation. In Fig. 19 we present the power law exponential model matches imposed on the flowrate data, and D- and b-parameter trends. We note that for all matches D value is not used in the model as dictated by the data character (i.e., power-law D-parameter trend). We observe that the model matches are for the most part identical during the entire production history. In particular, we note that the character of the data specifically the computed D-parameter trend for the smallest interval (or subset) is almost identical to the character of the data for the largest interval resulting in almost identical power-law exponential rate model matches. Therefore, EUR estimates from the power law exponential model would be relatively constant for this case. In Fig. 20 we show the results of the straight line extrapolation technique for this case. The x-axis intercept of the extrapolated lines increase with time resulting in an increasing estimate of Gp,max. The last step in our workflow is to calculate the EUR based on the matches obtained with the "hyperbolic" and the power-law exponential rate decline relations. In Fig. 21 we present the calculated EUR values versus production time. We observe that the EUR estimates obtained from the "hyperbolic" relation converge at late times to a value of 4.11 BSCF. The estimates provided by the power law exponential relation are constant at 4.06 BSCF after 1,000 days (and close before 1,000 days). We expect the EUR values estimated by the rate-time relations to decrease when complete boundary-dominated flow regime is established later on in the life of this well.

SPE 132352

Continuous Estimation of Ultimate Recovery

The Gp,max values obtained from straight line extrapolation are shown to increase with time, and they are always less than the estimates provided by the rate-time relations. Also we observe that Gp,max values obtained from straight line extrapolation still appear to increase at late times Gp,max values appear not to stabilize. Consequently, the EUR of this well should be in between 3.57 BSCF (the "lower" limit given by the straight line extrapolation technique at 1,930 days) and 4.06 BSCF (the "upper" limit given by the power law exponential estimate at 1,930 days). We summarize all the model parameters in Table 4 for this well. Field Case 3: The third field data set also includes the daily rate-time data a hydraulically fractured well completed in a tight gas reservoir. We present the flowrate data and the cumulative production data which spans almost 5.5 years in Fig. 22. Our first task is to match the subsets of the flowrate data with the "hyperbolic" rate decline relation. In Fig. 23 we present the "hyperbolic" model matches imposed on the flowrate data, and D- and b-parameter trends. We observe that every subset of the data is matched with a "hyperbolic" b-parameter greater than 1 indicating that complete boundary-dominated flow has not been established yet. In Fig. 24 we observe that the b-parameter value changes over time and is relatively stable after about 1,000 days of production in particular, b-parameter value decreases from 4.79 to 1.66 during the production history. This change corresponds to the final EUR estimate, which is three times less than the initial EUR estimate as predicted by the "hyperbolic" equation using the initial value of the b-parameter in the first subset of the data. Next we employ the power-law exponential rate decline relation for rate-time extrapolations for each interval (or subset). In Fig. 25 we present the power law exponential model matches imposed on the flowrate data and D- and b-parameter trends as well. Again for this case we note that for all matches, a D, value is not used indicating that boundary-dominated flow character is not observed. We observe that the model matches stabilize and become relatively constant after about 1,000 days of production for each additional interval. In Fig. 26 we show the results of the straight line extrapolation technique for this case. The x-axis intercept of the extrapolated lines increases with time resulting in an increasing estimate of Gp,max over time. Table 5 presents the model parameters along with the EUR estimates for each interval using all the rate-time models and straight line extrapolation. The last step in our workflow is to calculate the EUR based on the matches obtained with the "hyperbolic" and the power-law exponential rate decline relations. In Fig. 27 we present the calculated EUR values versus production time. We observe that the EUR values obtained from the "hyperbolic" and power-law exponential model matches decrease at early times but appear to stabilize after about 1,000 days of production. The EUR values estimated from the "hyperbolic" model matches converge at late times to a value of 2.39 BSCF, and the power-law exponential model matches also converge at late times to a value of 2.28 BSCF. The EUR of this well should be in between 2.04 BSCF (the "lower" limit given by the straight line extrapolation technique at 2,000 days) and 2.28 BSCF (the "upper" limit given by the power law exponential estimate at 2,000 days). Field Case 4: We apply our proposed methodology to the production data acquired from a hydraulically fractured well completed in a tight gas reservoir. We present the flowrate data and the cumulative production data which spans over 6 years in Fig. 28. Our first task is to match the subsets of the flowrate data with the "hyperbolic" rate decline relation. In Fig. 29 we present the "hyperbolic" model matches imposed on the flowrate data, and D- and b-parameter trends. In all of the model matches using the "hyperbolic" relation, we obtain b-parameter values greater than 1 again indicating that transient flow regime effects are still dominant. In Fig. 30 we observe that the b-parameter value decreases significantly over the production history and does not appear to stabilize at late times (for longer intervals). Specifically, the b-parameter value decreases from 3.66 to 1.70 during the production history (from the shortest subset of 50 days to entire production history). Following our procedure, we next use the power-law exponential model to match the subsets of the flowrate and estimate the EUR values. In Fig. 31 we present the power-law exponential model matches imposed on the flowrate data, and D- and bparameter trends. Once again D value is not used in the power-law exponential model which suggests that boundarydominated flow regime effects have not started to evolve. We observe that the model stabilizes and becomes relatively constant after about 750 days of production. Contrary to the second field example (i.e., Field Case 2), the rate model matches are not identical despite the power-law trend exhibited by the computed D-parameter trend. We note that the data quality for this example is not as good as the second field example resulting in slightly different matches for the computed D-parameter data trend and consequently for the flowrate data. In Fig. 32 we show the results of the straight line extrapolation technique for this case. The x-axis intercept of the extrapolated lines increases with time resulting in an increasing estimate of Gp,max. Finally we plot the EUR based on the matches obtained with the "hyperbolic" and the power-law exponential rate decline relations. In Fig. 33 we present the calculated EUR values versus production time. We observe that the EUR values obtained from the "hyperbolic" model matches decrease significantly at early times (almost twice).

10

S.M. Currie, D. Ilk, and T.A. Blasingame

SPE 132352

The EUR values seem to somewhat stabilize but still decrease at late times at a slower rate. The EUR values obtained from the power-law exponential model matches decrease significantly at early times and stabilize after about 750 days of production. As mentioned earlier the EUR values estimated from the power-law exponential model matches are more conservative the than value estimated from the hyperbolic model matches. After 2,384 days of production, both models converge to an EUR value of 3.78 BSCF. The EUR of this well should be in between 3.39 BSCF (the "lower" limit given by the straight line extrapolation technique at 2,000 days) and 3.78 BSCF (the "upper" limit given by the power law exponential estimate at 2,000 days). For reference the model parameter values and the EUR values for each interval are presented in Table 6.

Summary and Conclusions


Summary: In this work we propose the continuous EUR method to reduce the uncertainty in reserves estimation particularly for unconventional gas reservoirs. We utilize rate time functions and a simple extrapolation technique in a dynamic manner to offer a profile of EUR versus time. The proposed method is found to be especially useful to provide "upper" and "lower" limits for EUR prior to the onset of boundary-dominated flow. Conclusions: We state the following conclusions based on this work: 1. The uncertainty in reserves estimation for unconventional reservoirs is relatively high. A systematic methodology is needed to reduce the uncertainty associated with reserves estimation for unconventional gas reservoirs. The proposed continuous EUR method offers an effective way to reduce the associated uncertainty with reserves estimation by performing dynamic rate-time analysis/extrapolation throughout the producing life of the well. 2. The proposed "continuous EUR method evaluates EUR as a function of time and shows that EUR estimates stabilize when boundary dominated flow is established. 3. In this work we use the "hyperbolic", power law exponential with the proposed continuous EUR method. Other rate decline relations can also be implemented with this approach. 4. The straight line extrapolation technique offers a "lower" limit for EUR and can be used to establish a range for the reserves as well as to prevent the overestimation of reserves.

Nomenclature
Field Variables b D Di D D
i

EUR EURLL EURUL G Gp Gp,max k n qg qgi


gi q

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

Arps' decline exponent, dimensionless Loss ratio, D-1 Initial decline constant for hyperbolic rate relation, D-1 Decline constant at "infinite time" for power-law exponential relation [i.e., D(t=)], D-1 Decline constant for power-law exponential relation, D-1 Estimate of ultimate recovery, BSCF Lower limit for estimate of ultimate recovery, BSCF Upper limit for estimate of ultimate recovery, BSCF Original (contacted) gas-in-place, MSCF Cumulative gas production, MSCF Maximum gas production (t), MSCF Formation permeability, md Time exponent for power-law exponential relation, dimensionless Gas production rate, MSCF/D Gas initial production rate for hyperbolic rate relation, MSCF/D Rate "intercept" for power-law exponential relation [i.e., qg(t=0)], MSCF/D
Production time, days Time limit for reserves extrapolation, days

t tlim

SPE 132352

Continuous Estimation of Ultimate Recovery

11

References
Anderson, D.M., Stotts, G.W.J., Mattar, L., Ilk, D., and Blasingame, T.A. 2006. Production Data Analysis Challenges, Pitfalls, Diagnostics. SPE paper 102048 presented at the SPE Annual Technical Conference and Exhibition, San Antonio, Texas, 24-27 September. Ansah, J., Knowles, R.S., and Blasingame, T.A. 2000. A Semi-Analytic (p/z) Rate-Time Relation for the Analysis and Prediction of Gas Well Performance. SPEREE. 3 (6): 525-533. Arps J.J. 1945. Analysis of Decline Curves. Trans. AIME: 160, 228-247. Blasingame, T.A. and Rushing, J.A. 2005. A Production-Based Method for Direct Estimation of Gas-in-place and Reserves. SPE paper 98042 presented at the SPE Eastern Regional Meeting, Morgantown, West Virginia. 14-16 September. Bourdet, D., Ayoub, J.A., and Pirard, Y.M. 1989. Use of Pressure Derivative in Well-Test Interpretation. SPEFE 4 (2): 228-293-302. Carter, R.D. 1985. Type Curves for Finite Radial and Linear Gas Flow Systems: Constant-Terminal Pressure Case. SPEJ. 25 (5): 719-728. Cutler, W.W., Jr. 1924. Estimation of Underground Oil Reserves by Oil-Well Production Curves, Department of the Interior, U.S. Bureau Of Mines Bulletin 228. Fetkovich, M.J. 1980. Decline Curve Analysis Using Type Curves. JPT. 32 (6): 1065-1077. Fetkovich, M.J., Vienot, M.E., Bradley, M.D. and Kiesow, U.G. 1987. Decline Curve Analysis Using Type Curves: Case Histories. SPEFE. 2 (4): 637-656. Ilk, D., Mattar, L., and Blasingame, T.A. 2007. Production Data Analysis Future Practices for Analysis and Interpretation. Paper CIM 2007-174 presented at the 58th Annual Technical Meeting of the Petroleum Society, Calgary, Alberta, Canada, 12-14 June. Ilk, D., Perego, A.D., Rushing, J.A., and Blasingame, T.A. 2008. Exponential vs. Hyperbolic Decline in Tight Gas Sands Understanding the Origin and Implications for Reserve Estimates Using Arps' Decline Curves. Paper SPE 116731 presented at the SPE Annual Technical Conference and Exhibition, Denver, Colorado, 21-24 September. Johnson, R.H. and Bollens, A.L. 1928. The Loss Ratio Method of Extrapolating Oil Well Decline Curves. Trans. AIME 77: 771. Johnson, N.L., Currie, S.M., Ilk, D., and Blasingame, T.A. 2009. A Simple Methodology for Direct Estimation of Gas-in-Place and Reserves Using Rate-Time Data. Paper SPE 123298 presented at the SPE Rocky Mountain Petroleum Technology Conference, Denver, Colorado, 14-16 April. Lewis, J.O., and Beal, C.H.. 1918. Some New Methods for Estimating the Future Production of Oil Wells. Trans. AIME 59, 492-525. Manual for The Oil and Gas Industry Under The Revenue Act of 1918, Treasury Department United States Internal Revenue Service. 1919. Pratikno, H., Rushing, J.A., and Blasingame, T.A. 2003. Decline Curve Analysis Using Type Curves Fractured Wells. Paper SPE 84287 presented at the SPE Annual Technical Conference and Exhibition, Denver, Colorado, 5-8 October. Reese, P.D., Ilk, D., and Blasingame, T.A. 2007. Estimation of Reserves Using the Reciprocal Rate Method. Paper SPE 107981 presented at the 2007 SPE Rocky Mountain Oil & Gas Technology Symposium, Denver, Colorado, 16-18 April. Rushing, J.A., Perego, A.D., Sullivan, R.B., and Blasingame, T.A. 2007. Estimating Reserves in Tight Gas Sands at HP/HT Reservoir Conditions: Use and Misuse of an Arps Decline Curve Methodology. Paper SPE 109625 presented at the 2007 Annual SPE Technical Conference and Exhibition, Anaheim, California, 11-14 November.

12

S.M. Currie, D. Ilk, and T.A. Blasingame

SPE 132352

Fig. 4 (Semi-log Plot): Production history plot flowrate (qg) and cumulative production (Gp) versus production time. Tight gas well with a vertical hydraulic fracture (Ilk et al [2008]).

Fig. 5 (Log-log Plot): qDb Plot flowrate (qg), D- and b-parameters versus production time and "hyperbolic" model matches.

SPE 132352

Continuous Estimation of Ultimate Recovery

13

Fig. 6 (Cartesian Plot): Hyperbolic b-parameter values obtained from model matches with production data.

Fig. 7 (Log-log Plot): qDb Plot flowrate (qg), D- and b-parameters versus production time and power-law exponential model matches.

14

S.M. Currie, D. Ilk, and T.A. Blasingame

SPE 132352

Fig. 8 (Cartesian Plot): Rate Cumulative Plot flowrate (qg) versus cumulative production (Gp) and the linear trends fit through the data.

Fig. 9 (Cartesian Plot): EUR and Gp,max estimates from hyperbolic, power-law exponential model matches and straight line extrapolation.

SPE 132352

Continuous Estimation of Ultimate Recovery

15

Fig. 10 (Semi-log Plot): Production history plot flowrate (qg) and cumulative production (Gp) versus production time. Hydraulically fractured tight gas well.

Fig. 11 (Log-log Plot): qDb Plot flowrate (qg), D- and b-parameters versus production time and "hyperbolic" model matches.

16

S.M. Currie, D. Ilk, and T.A. Blasingame

SPE 132352

Fig. 12 (Cartesian Plot): Hyperbolic b-parameter values obtained from model matches with production data.

Fig. 13 (Log-log Plot): qDb Plot flowrate (qg), D- and b-parameters versus production time and power-law exponential model matches.

SPE 132352

Continuous Estimation of Ultimate Recovery

17

Fig. 14 (Cartesian Plot): Rate Cumulative Plot flowrate (qg) versus cumulative production (Gp) and the linear trends fit through the data.

Fig. 15 (Cartesian Plot): EUR and Gp,max estimates from hyperbolic, power-law exponential model matches and straight line extrapolation.

18

S.M. Currie, D. Ilk, and T.A. Blasingame

SPE 132352

Fig. 16 (Semi-log Plot): Production history plot flowrate (qg) and cumulative production (Gp) versus production time. Hydraulically fractured tight gas well.

Fig. 17 (Log-log Plot): qDb Plot flowrate (qg), D- and b-parameters versus production time and "hyperbolic" model matches.

SPE 132352

Continuous Estimation of Ultimate Recovery

19

Fig. 18 (Cartesian Plot): Hyperbolic b-parameter values obtained from model matches with production data.

Fig. 19 (Log-log Plot): qDb Plot flowrate (qg), D- and b-parameters versus production time and power-law exponential model matches.

20

S.M. Currie, D. Ilk, and T.A. Blasingame

SPE 132352

Fig. 20 (Cartesian Plot): Rate Cumulative Plot flowrate (qg) versus cumulative production (Gp) and the linear trends fit through the data.

Fig. 21 (Cartesian Plot): EUR and Gp,max estimates from hyperbolic, power-law exponential model matches and straight line extrapolation.

SPE 132352

Continuous Estimation of Ultimate Recovery

21

Fig. 22 (Semi-log Plot): Production history plot flowrate (qg) and cumulative production (Gp) versus production time. Hydraulically fractured tight gas well.

Fig. 23 (Log-log Plot): qDb Plot flowrate (qg), D- and b-parameters versus production time and "hyperbolic" model matches.

22

S.M. Currie, D. Ilk, and T.A. Blasingame

SPE 132352

Fig. 24 (Cartesian Plot): Hyperbolic b-parameter values obtained from model matches with production data.

Fig. 25 (Log-log Plot): qDb Plot flowrate (qg), D- and b-parameters versus production time and power-law exponential model matches.

SPE 132352

Continuous Estimation of Ultimate Recovery

23

Fig. 26 (Cartesian Plot): Rate Cumulative Plot flowrate (qg) versus cumulative production (Gp) and the linear trends fit through the data.

Fig. 27 (Cartesian Plot): EUR and Gp,max estimates from hyperbolic, power-law exponential model matches and straight line extrapolation.

24

S.M. Currie, D. Ilk, and T.A. Blasingame

SPE 132352

Fig. 28 (Semi-log Plot): Production history plot flowrate (qg) and cumulative production (Gp) versus production time. Hydraulically fractured tight gas well.

Fig. 29 (Log-log Plot): qDb Plot flowrate (qg), D- and b-parameters versus production time and "hyperbolic" model matches.

SPE 132352

Continuous Estimation of Ultimate Recovery

25

Fig. 30 (Cartesian Plot): Hyperbolic b-parameter values obtained from model matches with production data.

Fig. 31 (Log-log Plot): qDb Plot flowrate (qg), D- and b-parameters versus production time and power-law exponential model matches.

26

S.M. Currie, D. Ilk, and T.A. Blasingame

SPE 132352

Fig. 32 (Cartesian Plot): Rate Cumulative Plot flowrate (qg) versus cumulative production (Gp) and the linear trends fit through the data.

Fig. 33 (Cartesian Plot): EUR and Gp,max estimates from hyperbolic, power-law exponential model matches and straight line extrapolation.

SPE 132352

Continuous Estimation of Ultimate Recovery

27

Table 2a Analysis results for numerical simulation case (fractured tight gas well) "hyperbolic" model parameters.

Time Interval, days 50 100 250 500 750 1,000 1,250 1,500 1,750 2,000

qgi (MSCFD) 4,400 4,400 4,400 4,400 4,400 4,400 4,400 4,400 4,400 4,400

Di (D-1) 0.01000 0.01380 0.01600 0.01000 0.00713 0.00600 0.00540 0.00500 0.00497 0.00497

b (dimensionless) 2.80 2.60 2.15 1.50 1.10 0.90 0.85 0.78 0.75 0.75

EURhyp (BSCF) 9.46 7.67 5.80 3.90 3.09 2.82 2.72 2.60 2.55 2.55

Table 2b Analysis results for numerical simulation case (fractured tight gas well) power-law exponential model parameters.

Time Interval, days 50 100 250 500 750 1,000 1,250 1,500 1,750 2,000

gi q

(MSCFD) 6,292,146 7,325,965 1,132,541 812,662 812,662 812,662 812,662 812,662 812,662 812,662

(D-1) 6.216 6.361 4.440 4.158 4.158 4.158 4.158 4.158 4.158 4.158

D i

n (dimensionless) 0.05 0.05 0.07 0.07 0.07 0.07 0.07 0.07 0.07 0.07

D (D-1) 0 0 0 0.000421 0.000440 0.000460 0.000460 0.000460 0.000460 0.000460

EURPLE (BSCF) 6.74 6.37 4.99 4.02 2.41 2.41 2.41 2.41 2.41 2.41

Table 2c Analysis results for numerical simulation case (fractured tight gas well) straight line extrapolation.

Time Interval, days 50 100 250 500 750 1,000 1,250 1,500 1,750 2,000

Slope, 10-6 D-1 15,250 3,648 2,478 1,724 1,294 1,176 922 818 738 738

Intercept, MSCF/D 6,891 3,912 3,352 2,814 2,381 2,245 1,894 1,733 1,601 1,601

Gp,max (BSCF) 0.45 1.07 1.35 1.63 1.84 1.91 2.05 2.12 2.17 2.17

28

S.M. Currie, D. Ilk, and T.A. Blasingame

SPE 132352

Table 3a Analysis results for field example 1 (fractured tight gas well) "hyperbolic" model parameters.

Time Interval, days 50 100 250 500 750 1,000 1,250 1,500 1,750 2,000

qgi (MSCFD) 4,100 4,100 4,100 4,100 4,100 4,100 4,100 4,100 4,100 4,100

Di (D-1) 0.0094 0.0094 0.0131 0.0146 0.0146 0.0146 0.0158 0.0158 0.0152 0.0152

b (dimensionless) 2.20 2.05 1.98 1.94 1.83 1.75 1.68 1.63 1.59 1.57

EURhyp (BSCF) 6.04 5.84 4.92 4.53 4.12 3.82 3.42 3.24 3.17 3.10

Table 3b Analysis results for field example 1 (fractured tight gas well) power-law exponential model parameters.

Time Interval, days 50 100 250 500 750 1,000 1,250 1,500 1,750 2,000

gi q

(MSCFD) 72,500 67,426 67,426 67,426 67,426 67,426 67,426 66,500 66,500 66,500

(D-1) 1.785 1.684 1.684 1.684 1.684 1.684 1.684 1.684 1.684 1.684

D i

n (dimensionless) 0.14 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15

D (D-1) 0 0 0 0 0 0 0 0 0 0

EURPLE (BSCF) 3.10 3.09 3.09 3.09 3.09 3.09 3.09 3.05 3.05 3.05

Table 3c Analysis results for field example 1 (fractured tight gas well) straight line extrapolation.

Time Interval, days 50 100 250 500 750 1,000 1,250 1,500 1,750 2,000

Slope, 10-6 D-1 22,385 21,845 3,915 1,843 819 728 605 569 454 443

Intercept, MSCF/D 8,248 9,881 4,045 2,714 1,738 1,558 1,356 1,310 1,139 1,145

Gp,max (BSCF) 0.37 0.45 1.03 1.47 2.12 2.14 2.24 2.30 2.51 2.58

SPE 132352

Continuous Estimation of Ultimate Recovery

29

Table 4a Analysis results for field example 2 (fractured tight gas well) "hyperbolic" model parameters.

Time Interval, days 50 100 250 500 750 1,000 1,250 1,500 1,750 1,930

qgi (MSCFD) 3,495 3,495 3,495 3,495 3,495 3,495 3,495 3,495 3,495 3,495

Di (D-1) 0.0160 0.0153 0.0141 0.0140 0.0140 0.0140 0.0130 0.0120 0.0121 0.0121

b (dimensionless) 2.37 2.25 2.14 2.11 2.10 2.09 1.95 1.90 1.90 1.90

EURhyp (BSCF) 5.05 4.75 4.57 4.44 4.34 4.28 4.18 4.11 4.11 4.11

Table 4b Analysis results for field example 2 (fractured tight gas well) power-law exponential model parameters.

Time Interval, days 50 100 250 500 750 1,000 1,250 1,500 1,750 1,930

gi q

(MSCFD) 26,695 28,606 30,654 32,849 35,200 40,421 40,421 40,421 40,421 40,421

(D-1) 1.58 1.61 1.65 1.69 1.73 1.93 1.93 1.93 1.93 1.93

D i

n (dimensionless) 0.12 0.12 0.12 0.12 0.12 0.11 0.11 0.11 0.11 0.11

D (D-1) 0 0 0 0 0 0 0 0 0 0

EURPLE (BSCF) 4.54 4.43 4.32 4.20 4.08 4.06 4.06 4.06 4.06 4.06

Table 4c Analysis results for field example 2 (fractured tight gas well) straight line extrapolation.

Time Interval, days 50 100 250 500 750 1,000 1,250 1,500 1,750 1,930

Slope, 10-6 D-1 12,459 3,068 1,762 1,023 625 421 354 288 268 242

Intercept, MSCF/D 3,805 2,544 2,082 1,686 1,401 1,127 1,023 930 910 863

Gp,max (BSCF) 0.31 0.83 1.18 1.65 2.24 2.68 2.89 2.23 3.39 3.57

30

S.M. Currie, D. Ilk, and T.A. Blasingame

SPE 132352

Table 5a Analysis results for field example 3 (fractured tight gas well) "hyperbolic" model parameters.

Time Interval, days 50 100 250 500 750 1,000 1,250 1,500 1,750 1,995

qgi (MSCFD) 1,900 1,900 1,900 1,900 1,900 1,900 1,900 1,900 1,900 1,900

Di (D-1) 0.01920 0.01470 0.01430 0.00862 0.00825 0.00737 0.00755 0.00700 0.00700 0.00713

b (dimensionless) 4.79 4.20 2.93 2.16 1.97 1.81 1.75 1.71 1.66 1.66

EURhyp (BSCF) 6.19 5.75 3.88 3.12 2.82 2.64 2.50 2.46 2.42 2.39

Table 5b Analysis results for field example 3 (fractured tight gas well) power-law exponential model parameters.

Time Interval, days 50 100 250 500 750 1,000 1,250 1,500 1,750 1,995

gi q

(MSCFD) 424,256 92,676 185,041 23,247 30,654 23,247 20,244 21,966 18,632 22,928

(D-1) 4.82 3.25 3.97 1.84 2.17 1.77 1.73 1.77 1.58 1.54

D i

n (dimensionless) 0.05 0.07 0.06 0.11 0.10 0.12 0.12 0.12 0.13 0.14

D (D-1) 0 0 0 0 0 0 0 0 0 0

EURPLE (BSCF) 3.34 3.27 3.16 2.86 2.56 2.43 2.35 2.30 2.29 2.28

Table 5c Analysis results for field example 3 (fractured tight gas well) straight line extrapolation.

Time Interval, days 50 100 250 500 750 1,000 1,250 1,500 1,750 1,995

Slope, 10-6 D-1 15,589 2,436 2,010 1,043 833 525 465 368 348 292

Intercept, MSCF/D 2,351 1,614 1,421 1,114 1,036 789 755 669 659 595

Gp,max (BSCF) 0.15 0.66 0.71 1.07 1.24 1.50 1.62 1.82 1.89 2.04

SPE 132352

Continuous Estimation of Ultimate Recovery

31

Table 6a Analysis results for field example 4 (fractured tight gas well) "hyperbolic" model parameters.

Time Interval, days 50 100 250 500 750 1,000 1,250 1,500 1,750 2,000 2,250 2,384

qgi (MSCFD) 8,315 6,704 5,606 5,280 4,852 4,500 4,000 3,932 3,905 3,932 3,669 2,880

Di (D-1) 0.8310 0.2800 0.1100 0.0810 0.0560 0.0390 0.0250 0.0251 0.0242 0.0230 0.0151 0.0070

b (dimensionless) 3.66 3.28 2.81 2.65 2.52 2.36 2.21 2.19 2.16 2.10 1.90 1.70

EURhyp (BSCF) 7.29 6.34 5.24 4.93 4.71 4.44 4.24 4.15 4.07 4.02 3.85 3.78

Table 6b Analysis results for field example 4 (fractured tight gas well) power-law exponential model parameters.

Time Interval, days 50 100 250 500 750 1,000 1,250 1,500 1,750 2,000 2,250 2,384

gi q

(MSCFD) 688,395 243,999 344,776 522,057 261,467 300,246 344,776 354,446 57,116 80,706 75,314 75,314

(D-1) 4.77 3.79 4.08 4.46 3.74 3.82 3.91 3.93 2.28 2.44 2.41 2.41

D i

n (dimensionless) 0.05 0.06 0.06 0.06 0.07 0.07 0.07 0.07 0.10 0.10 0.10 0.10

D (D-1) 0 0 0 0 0 0 0 0 0 0 0 0

EURPLE (BSCF) 5.88 5.55 5.08 4.40 4.02 3.98 3.92 3.90 3.82 3.81 3.78

3.78

Table 6c Analysis results for field example 4 (fractured tight gas well) straight line extrapolation.

Time Interval, days 50 100 250 500 750 1,000 1,250 1,500 1,750 2,000 2,250 2,384

Slope, 10-6 D-1 15,697 5,449 2,064 1,086 705 552 392 342 311 282 247 240

Intercept, MSCF/D 4,033 2,908 2,066 1,635 1,379 1,236 1,030 980 959 893 832 815

Gp,max (BSCF) 0.26 0.53 1.00 1.51 1.95 2.24 2.63 2.87 3.09 3.17 3.37 3.39

SPE 132352
Continuous Estimation of Ultimate Recovery
S.M. Currie, Texas A&M University D. Ilk, Texas A&M University T.A. Blasingame, Texas A&M University Department of Petroleum Engineering Texas A&M University College Station, TX 77843-3116 +1.979.845.4064 stephanie.currie@pe.tamu.edu
2010 SPE Unconventional Gas Conference Pittsburgh, PA 2325 February 2010 SPE 132352 Continuous Estimation of Ultimate Recovery (Currie/Ilk/Blasingame) Stephanie M. CURRIE Texas A&M University (23 February 2010)

Slide 1/26

Rationale for this Work Literature Review/Orientation Development of the Method Validation: (Numerical Simulation Case) Illustrative Examples:
East Tx Gas Well (SPE 84287) Field Example East Tx Gas Well (SPE 84287) Field Example Fractured Gas Well

Summary and Conclusions

2010 SPE Unconventional Gas Conference Pittsburgh, PA 2325 February 2010 SPE 132352 Continuous Estimation of Ultimate Recovery (Currie/Ilk/Blasingame) Stephanie M. CURRIE Texas A&M University (23 February 2010)

Slide 2/26

(Outline)

Presentation Outline

Uncertainty in Reserves Estimation for Unconventional


Gas Reservoirs:

Objectives:

Very low permeability. Arps' Decline Curves.

[longer transient flow periods] [boundary-dominated flow only]

To show that the analysis of production data before the onset of

boundary-dominated flow will lead to significant overestimation of reserves. To reduce uncertainty in reserves estimates by applying the continuous EUR approach.

2010 SPE Unconventional Gas Conference Pittsburgh, PA 2325 February 2010 SPE 132352 Continuous Estimation of Ultimate Recovery (Currie/Ilk/Blasingame) Stephanie M. CURRIE Texas A&M University (23 February 2010)

Slide 3/26

(Rationale)

Rationale For This Work

Definition of "Loss-ratio"

qg 1 D dqg /dt
"Loss-ratio" derivative

d 1 d b dt D dt

qg dqg /dt

Arps' Hyperbolic Rate-Time Relation

q (t ) =

qi (1 + bDi t ) (1 / b)

Discussion: "Hyperbolic" Rate Decline Relation


2010 SPE Unconventional Gas Conference Pittsburgh, PA 2325 February 2010 SPE 132352 Continuous Estimation of Ultimate Recovery (Currie/Ilk/Blasingame) Stephanie M. CURRIE Texas A&M University (23 February 2010)

Slide 4/26

"Hyperbolic" relation is the mostly used relation for reserves estimation. b-parameter should lie between 0 and 1 but is greater than 1 during transient or transition flow regimes. Extrapolation of "hyperbolic" model prior to the onset of BD flow will result in overestimation of reserves (Rushing et al [2007]).

(Orientation)

Orientation: "Hyperbolic" Rate Decline Relation

From: Ilk, D., Perego, A.D., Rushing, J.A., and Blasingame, T.A. 2008. Exponential vs. Hyperbolic Decline in Tight Gas Sands Understanding the Origin and Implications for Reserve Estimates Using Arps' Decline Curves. Paper SPE 116731 presented at the SPE Annual Technical Conference and Exhibition, Denver, Colorado, 21-24 September.

tn] i exp[ D t D q=q i

Discussion:

2010 SPE Unconventional Gas Conference Pittsburgh, PA 2325 February 2010 SPE 132352 Continuous Estimation of Ultimate Recovery (Currie/Ilk/Blasingame) Stephanie M. CURRIE Texas A&M University (23 February 2010)

Slide 5/26

b(t) and D(t) are evaluated continuously using numerical differentiation. D(t) trend indicates "power-law" behavior converging to a constant at late times basis for the power-law exponential relation. Very flexible model that can be used to match transient, transition, and boundary-dominated flow data.

(Orientation)

Orientation: Power-Law Exponential Rate Decline

Review and Edit Data

Specify Time Intervals

50 days

250 days

...

2000 days

2010 SPE Unconventional Gas Conference Pittsburgh, PA 2325 February 2010 SPE 132352 Continuous Estimation of Ultimate Recovery (Currie/Ilk/Blasingame) Stephanie M. CURRIE Texas A&M University (23 February 2010)

Slide 6/26

(Development)

Development: Continuous EUR Method

Analyze All Intervals Using the Hyperbolic Relation

Analyze All Intervals Using the Find Lower Limit Using qg vs. Gp Straight Line Power-Law Exponential Extrapolation Relation

50 days EUR = 6.04 BSCF EUR = 3.10 BSCF Gp,max = 0.37 BSCF

250 days EUR = 4.92 BSCF EUR = 3.09 BSCF Gp,max = 1.03 BSCF

2000 days Slide 7/26 EUR = 3.10 BSCF EUR = 3.05 BSCF Gp,max = 2.58 BSCF

2010 SPE Unconventional Gas Conference Pittsburgh, PA 2325 February 2010 SPE 132352 Continuous Estimation of Ultimate Recovery (Currie/Ilk/Blasingame) Stephanie M. CURRIE Texas A&M University (23 February 2010)

(Development)

Development: Continuous EUR Method

...

Plot Gp Data and EUR Estimates from Models vs. Time for All Intervals
EURhyp EURPLE Gp,max

Identify the Upper Limit for EUR Using the Power Law Exponential Model Identify the Lower Limit for EUR Using the Straight Line Extrapolation Technique
2010 SPE Unconventional Gas Conference Pittsburgh, PA 2325 February 2010 SPE 132352 Continuous Estimation of Ultimate Recovery (Currie/Ilk/Blasingame) Stephanie M. CURRIE Texas A&M University (23 February 2010)

Slide 8/26

(Development)

Development: Continuous EUR Method

Discussion: East Tx Gas Well (SPE 84287)

Previously obtained model parameters (see Pratikno et al (2003) were

2010 SPE Unconventional Gas Conference Pittsburgh, PA 2325 February 2010 SPE 132352 Continuous Estimation of Ultimate Recovery (Currie/Ilk/Blasingame) Stephanie M. CURRIE Texas A&M University (23 February 2010)

Slide 9/26

used to generate the synthetic flowrate response (with constant flowing wellbore pressure). Tight gas well having a vertical fracture with finite conductivity (k=0.005 md, FcD=10, and Ginp = 2.65 BSCF Gp (t=30 years) = 2.42 BSCF).

(Numerical Simulation Case)

Validation: Numerical Simulation Case

Discussion: Continuous EUR using Arps' Hyperbolic Decline

2010 SPE Unconventional Gas Conference Pittsburgh, PA 2325 February 2010 SPE 132352 Continuous Estimation of Ultimate Recovery (Currie/Ilk/Blasingame) Stephanie M. CURRIE Texas A&M University (23 February 2010)

Slide 10/26

Subsets of data are matched using the Arps' hyperbolic decline relation. EUR of each subset is estimated (qgi is fixed). b-parameter value decreases (as expected) over time.

(Numerical Simulation Case)

Validation: Numerical Simulation Case

Discussion: Continuous EUR using Power-Law Exponential Model


2010 SPE Unconventional Gas Conference Pittsburgh, PA 2325 February 2010 SPE 132352 Continuous Estimation of Ultimate Recovery (Currie/Ilk/Blasingame) Stephanie M. CURRIE Texas A&M University (23 February 2010)

Slide 11/26

EUR of each subset of data is estimated progressively. We do not fix any model parameter for the power-law exponential model. EUR stabilizes when boundary-dominated flow regime is established.

(Numerical Simulation Case)

Validation: Numerical Simulation Case

Discussion: Gp,max using Straight Line Extrapolation


2010 SPE Unconventional Gas Conference Pittsburgh, PA 2325 February 2010 SPE 132352 Continuous Estimation of Ultimate Recovery (Currie/Ilk/Blasingame) Stephanie M. CURRIE Texas A&M University (23 February 2010)

Slide 12/26

Gp,max for each interval is referred as the "lower-limit" for EUR using the straight line extrapolation (boundary-dominated flow assumption). Gp,max is expected to increase with time for unconventional reservoirs.

(Numerical Simulation Case)

Validation: Numerical Simulation Case

Discussion: Continuous EUR


2010 SPE Unconventional Gas Conference Pittsburgh, PA 2325 February 2010 SPE 132352 Continuous Estimation of Ultimate Recovery (Currie/Ilk/Blasingame) Stephanie M. CURRIE Texas A&M University (23 February 2010)

Slide 13/26

Upper and lower limits of EUR are established. EUR estimate decreases over time and stabilizes at late times. EUR estimate using the power-law exponential model stabilizes earlier and is more conservative.

(Numerical Simulation Case)

Validation: Numerical Simulation Case

Discussion: East Tx Gas Well

2010 SPE Unconventional Gas Conference Pittsburgh, PA 2325 February 2010 SPE 132352 Continuous Estimation of Ultimate Recovery (Currie/Ilk/Blasingame) Stephanie M. CURRIE Texas A&M University (23 February 2010)

Slide 14/26

Hydraulically fractured gas well completed in a tight gas reservoir

(Bossier field) (see Pratikno et al (2003) for well/reservoir properties). Off-trend data are removed prior to analysis.

(Tight Gas Well 1)

Field Example: East Tx Gas Well (SPE 84287)

Discussion: Continuous EUR using Arps' Hyperbolic Decline


2010 SPE Unconventional Gas Conference Pittsburgh, PA 2325 February 2010 SPE 132352 Continuous Estimation of Ultimate Recovery (Currie/Ilk/Blasingame) Stephanie M. CURRIE Texas A&M University (23 February 2010)

Slide 15/26

b-parameter value ranges between 1.57 (all data) and 2.2 (earliest interval) for the subsets of data. No strong evidence of boundary-dominated flow is observed.

(Tight Gas Well 1)

Field Example: East Tx Gas Well (SPE 84287)

Discussion: Continuous EUR using Power-Law Exponential Model


2010 SPE Unconventional Gas Conference Pittsburgh, PA 2325 February 2010 SPE 132352 Continuous Estimation of Ultimate Recovery (Currie/Ilk/Blasingame) Stephanie M. CURRIE Texas A&M University (23 February 2010)

Slide 16/26

All of the matches for each part of the data set are almost outstanding. EUR estimates from the hyperbolic and power-law exponential relations are close. However, power-law exponential relation matches all data.

(Tight Gas Well 1)

Field Example: East Tx Gas Well (SPE 84287)

Discussion: Gp,max using Straight Line Extrapolation


2010 SPE Unconventional Gas Conference Pittsburgh, PA 2325 February 2010 SPE 132352 Continuous Estimation of Ultimate Recovery (Currie/Ilk/Blasingame) Stephanie M. CURRIE Texas A&M University (23 February 2010)

Slide 17/26

Straight line extrapolation yields the minimum value for the reserves estimate as boundary-dominated flow assumption is made. Gp,max(t) does not stabilize at late times.

(Tight Gas Well 1)

Field Example: East Tx Gas Well (SPE 84287)

Discussion: Conclusions

2010 SPE Unconventional Gas Conference Pittsburgh, PA 2325 February 2010 SPE 132352 Continuous Estimation of Ultimate Recovery (Currie/Ilk/Blasingame) Stephanie M. CURRIE Texas A&M University (23 February 2010)

Slide 18/26

Almost 50 percent decrease is observed in the EUR estimate when power-law exponential and hyperbolic relations are used. Character of flowrate data do not seem to be influenced by the boundary-dominated flow behavior.

(Tight Gas Well 1)

Field Example: East Tx Gas Well (SPE 84287)

Discussion: Fractured Gas Well


to analysis.

2010 SPE Unconventional Gas Conference Pittsburgh, PA 2325 February 2010 SPE 132352 Continuous Estimation of Ultimate Recovery (Currie/Ilk/Blasingame) Stephanie M. CURRIE Texas A&M University (23 February 2010)

Slide 19/26

Hydraulically fractured gas well completed in a tight gas reservoir. 2000 days of daily production data are available. We edit the data prior

(Tight Gas Well 2)

Field Example: Fractured Gas Well

Discussion: Continuous EUR using Arps' Hyperbolic Decline

2010 SPE Unconventional Gas Conference Pittsburgh, PA 2325 February 2010 SPE 132352 Continuous Estimation of Ultimate Recovery (Currie/Ilk/Blasingame) Stephanie M. CURRIE Texas A&M University (23 February 2010)

Slide 20/26

Subsets of data are matched using the Arps' hyperbolic decline relation. EUR of each subset is estimated (qgi is fixed)

(Tight Gas Well 2)

Field Example: Fractured Gas Well

Discussion: Continuous EUR using Power-Law Exponential Model


2010 SPE Unconventional Gas Conference Pittsburgh, PA 2325 February 2010 SPE 132352 Continuous Estimation of Ultimate Recovery (Currie/Ilk/Blasingame) Stephanie M. CURRIE Texas A&M University (23 February 2010)

Slide 21/26

EUR of each subset of data is estimated progressively. We do not fix any model parameter for the power-law exponential model. Boundary-dominated flow regime is not observed (straight line D(t) trend).

(Tight Gas Well 2)

Field Example: Fractured Gas Well

Discussion: Gp,max using Straight Line Extrapolation


2010 SPE Unconventional Gas Conference Pittsburgh, PA 2325 February 2010 SPE 132352 Continuous Estimation of Ultimate Recovery (Currie/Ilk/Blasingame) Stephanie M. CURRIE Texas A&M University (23 February 2010)

Slide 22/26

Gp,max(t) increases with time and does not appear to converge as boundary-dominated flow is not established. Straight line extrapolation provides a conservative way to identify the "lower-limit" for reserves.

(Tight Gas Well 2)

Field Example: Fractured Gas Well

Discussion: Conclusions

2010 SPE Unconventional Gas Conference Pittsburgh, PA 2325 February 2010 SPE 132352 Continuous Estimation of Ultimate Recovery (Currie/Ilk/Blasingame) Stephanie M. CURRIE Texas A&M University (23 February 2010)

Slide 23/26

EUR estimate decreases over time and stabilizes at late times as the effects of transition to boundary-dominated flow are being established. Production data exhibit transition to boundary-dominated flow behavior EUR estimate may decrease further at a lower rate of stabilization.

(Tight Gas Well 2)

Field Example: Fractured Gas Well

(Summary):

reserves particularly in unconventional reservoirs. We employ simple rate-time relations (i.e., hyperbolic and power-law exponential) in a "dynamic" manner in our procedure. Time intervals are specified and rate-time relations are extrapolated for each interval to create a "time-dependent" profile of EUR vs. time. Straight line Gp,max extrapolation is performed to obtain the "lower-limit" for reserves. This process also yields a "timedependent" Gp,max(t) vs. t profile.

This work presents a new procedure for the estimation of

2010 SPE Unconventional Gas Conference Pittsburgh, PA 2325 February 2010 SPE 132352 Continuous Estimation of Ultimate Recovery (Currie/Ilk/Blasingame) Stephanie M. CURRIE Texas A&M University (23 February 2010)

Slide 24/26

(Summary and Conclusions)

Summary and Conclusions:

reservoirs is relatively high and a systematic methodology is required to reduce the uncertainties associated with reserves estimation. Continuous EUR procedure evaluates the EUR as a function of time, and eventually EUR stabilizes when complete boundarydominated flow regime is established. In this work we use the hyperbolic and power-law exponential rate decline relations in conjunction to estimate the EUR. However, additional rate-time relations may be implemented in the procedure. "Lower-limit" of EUR (obtained by straight line extrapolation) can be used as a "safeguard" to prevent overestimation of reserves for each interval.

The uncertainty in reserves estimate for unconventional

2010 SPE Unconventional Gas Conference Pittsburgh, PA 2325 February 2010 SPE 132352 Continuous Estimation of Ultimate Recovery (Currie/Ilk/Blasingame) Stephanie M. CURRIE Texas A&M University (23 February 2010)

Slide 25/26

(Summary and Conclusions)

Summary and Conclusions:

(Conclusions):

SPE 132352
Continuous Estimation of Ultimate Recovery
End of Presentation
S.M. Currie, Texas A&M University D. Ilk, Texas A&M University T.A. Blasingame, Texas A&M University Department of Petroleum Engineering Texas A&M University College Station, TX 77843-3116 +1.979.845.4064 stephanie.currie@pe.tamu.edu
2010 SPE Unconventional Gas Conference Pittsburgh, PA 2325 February 2010 SPE 132352 Continuous Estimation of Ultimate Recovery (Currie/Ilk/Blasingame) Stephanie M. CURRIE Texas A&M University (23 February 2010)

Slide 26/26

You might also like