Вы находитесь на странице: 1из 14

SPE 167118

Best Practice of Using Empirical Methods for Production Forecast and EUR
Estimation in Tight/Shale Gas Reservoirs
Shaoyong YU, ConocoPhillips Canada

Copyright 2013, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Unconventional Resources Conference-Canada held in Calgary, Alberta, Canada, 5–7 November 2013.

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
Since 2008, a number of new empirical methods have been introduced to the petroleum industry, specifically for gas wells in
tight and shale reservoirs. Among them, Valko’s Stretch Exponential Production Decline (SEPD) and Duong’s Rate Decline
for Fractured Dominated Reservoir are two most mentioned methods by industry experts. These two methods are claimed to
be accurate in both tight and shale gas wells. However, since the reservoir rock permeability in those reservoirs ranges from
0.1 to 0.0001 mD in tight gas to less than 0.0001 mD in shale gas, are these two empirical methods still applicable in all ranges
of permeability?

This paper presents the results of an evaluation study by applying these two empirical methods to tight and shale gas wells
under a wide range of permeability generated based on both actual and synthetic production data. The key findings and
recommended improvements resulted from the study are:

1. Both SEPD and Duong’s Rate-Decline Methods are not applicable to tight gas reservoirs with a permeability ranging
from 0.1 to 0.001mD. Duong’s Rate Decline Method will significantly over-estimate EUR whereas SEPD will most
likely mismatch the production history as well as yielding a lower EUR while using its Recovery Potential Curve to
find parameters.

2. A Modified SEPD Method (YM-SEPD) developed by the author is a much easier and versatile method to use, and
most importantly it will yield a more reliable production forecast and EUR estimation.

3 With respect to reservoirs with rock permeability less than 0.01mD, a more rigorous step-by-step work flow using
Duong’s Method has been proposed. Moreover, for tight gas reservoirs Duong’s Method can only be used for
production forecast during early years, prior to pseudo-steady state (PSS) flow.

4 Hundreds of horizontal wells including both oil and gas wells, from various formations (Cadomin, Montney,
Notikewin, Cardium, etc. in Canada) and under different hydraulic fracturing conditions, have been analyzed using
these three different methods. Results indicate that the new YM-SEPD Method will yield a more reliable EUR as
well as production forecast in comparison with other two methods, especially when there is only less than 2~3 years
of production history.

5 For wells having less than 2 years of production history, the YM-SEPD Method will also be able to yield a reasonable
prediction by coupling with Duong’s Decline Method

Introduction
Within the Western Canadian Sedimentary Basin (WCSB), the development of petroleum reservoirs with low to extremely
low permeability has been substantially increased since the recent development of horizontal well drilling and multi-stage
fracturing completion technology. These types of reservoir are in fact very different from the real shale gas reservoirs in terms
of their permeability. Technologies currently available to predict a well’s production profile and its expected ultimate
recovery (EUR) are mostly analytical (Bello, and Wattenbager, 2008, 2009, 2010; Anderson, et al, 2010), numerical (Cipolla,
2 SPE 167118

et al, 2009), or empirical (Arps, 1945; Ilk, et al, 2009). Analytical methods are simple to use but very often they are just a
rough approximation due to the uncertainty in reservoir properties and the requirement of reliable pressure data. Numerical
methods, which are based primarily on obtaining a good history match, often yield non-unique solutions and, therefore, their
results can only provide us with a certain level of confidence. Besides, they also need reliable pressure data for the analysis.
However, empirical methods (production rate decline analysis methods), due to its simplicity, remain the most widely used
method in the oil & gas industry, especially for booking reserves. Therefore, several empirical methods (Valko, 2009; Duong,
2010) have been developed since 2008 and are recommended by industry experts for analysis of wells in tight/shale reservoirs
(Lee, 2012a, Lee, 2012b).

Due to the fact that there is a wide range of permeability difference within tight/shale reservoirs, questions are therefore being
frequently asked:

1. Are the empirical methods applicable to all ranges of the tight/shale reservoirs? If not,
2. What range of permeability values of the reservoirs they are directly applicable to?
3. What is the performance of Duong’s specialized plot in a fractured reservoir under boundary dominated flow (BDF)?
4. What percentage of errors of the prediction could be made by applying them to the wrong reservoirs?

This paper presents the results of a study trying to answer the above four questions. Moreover, the paper will also present how
and when to use these empirical methods. The study involved analyzing a great number of horizontal wells that are producing
from WCSB formations and were completed with multi-stage fracturing. Finally, a recommended methodology (best practice)
is presented in this paper.

Using Duong’s Rate Decline Method


As being well known in the hydraulic fractured reservoirs (Lee and Wattenbarger, 1996), the production rate history behaves
at beginning:
q = q1t − n (1)
where n = 0.5 for linear flow of the infinite conductivity fracture or n = 0.25 for bi-linear flow in the finite conductivity
fracture. From Eq.1, it can yield:
q 1− n
= (2)
Gp t
where Gp is the cumulative production of the gas; q is the rate and q1 is defined as the initial rate; t is the time of production.

Doung’s Rate Decline Method


Based on the assumption that fractured reservoirs with
extremely low matrix permeability would behave linear
flow for most part of its production life, Duong (Duong,
2010) presented his Rate Decline Method using an analogy
relationship as in Eq.2:

q
= at − m (3)
Gp
Eq.3 indicates that a straight line exists between q/Gp and
time t on a log-log plot, as shown in Figure 1. In practice,
parameters a and m can therefore be easily found from the
straight line and future production and its cumulative
production can be predicted using the following two
relationships: Fig.1 – Duong’s Specialized Plot for a and m

a
(t 1− m
−1 )
q = q1t − m e1− m (4)

q (t 1−m −1)
a
G p = 1 e 1− m (5)
a
Unfortunately, Duong did not indicate what would be the appropriate range of his “low/extreme-low” permeability in which
his method can apply to. If the producing well has reached BDF at a later time, what could be the relationship between q/Gp
SPE 167118 3

and time t?

Application of Duong’s Method - Problems


To answer the above questions, the author has employed both synthetic and actual production data to investigate the true
performance of Duong’s specialized plot under various conditions.

Synthetic production data has been generated by running a commercial simulator under various values of permeability as listed
in Table 1. Other parameters of both reservoir and producing well are also listed in Table 1. Hydraulic fractures are assumed
to reach the boundary of the reservoir in order to simulate a naturally fractured reservoir, as shown in Figure 2.

Table 1 – Reservoir and Well Parameters for Generating the Synthetic Data
Reservoir Parameters: Wellbore Parameters:
Reservoir Pressure (p i )= 2200 psia Horizontal Wellbore (L ) = 3280.0 ft
o
Reservoir Temperature (T r ) = 180 F No. of Satge = 10
Fracture Conductivity (F CD ) = 2
Permeability (k ) = 0.1, 0.05, 0.01, mD
0.001, 0.0001 Fracture Half-Length (xf ) = 485 ft
Porosity (Phi ) = 6% Flowing Pressure (p wf ) = 600 psia
Water Saturation (S wi ) = 30%
Net Pay (h ) = 39.37 ft Fluid Parameters:
Drainage Area (A ) = 100 acres Sweet Dry Gas Gravity (γ g ) = 0.65
Xe/Ye (Rxy ) = 4.5 CO2 (%) = 1.3
OGIP = 991 MMscf N2 (%) = 0.5

Fig 2 – Model for Generating the Synthetic Data Fig 3 – Duong’s Specialized Plots of Synthetic Data

Duong’s specialized plots for various permeability, shown in Figure 3, indicate that inflection points have occurred at some
specific time – the greater the reservoir permeability, the earlier the inflection occurs. For reservoirs with permeability being
around 0.1mD to 0.01mD, inflection point can start less than 200 days. However, for reservoirs with less than 0.0001mD
permeability, inflection doesn’t occur until a much later time. It has also been noticed that when a well’s production has
reached the BDF, Duong’s specialized plot would inflect downwards. As a result, applying Duong’s method to early
production data would over predict both the future production and the final EURs as shown in Figure 4.

Duong’s Specialized Plots of two different reservoir permeability are presented in Figure 4. The plots are drawn based on the
first two years of production. It can be seen that the Duong’s method will over predict production for the well with a higher
permeability than the well with a lower permeability. This is simply because the higher permeability well has much earlier
inflection point, as shown in Fig4(a), compared with the lower permeability well.

A tight gas well in a WCSB formation - Cadomin has been selected to further validate the aforementioned production
behavior. This 1230-m long horizontal well has been producing gas for more than 6 years. It was initially completed based on
the slick-water fracturing technique. The initial reservoir pressure was about 3150 psia and its reservoir in-situ gas
permeability is believed to be less than 0.1mD. The Duong’s Specialized Plot for the example well is shown in Figure 5. It
can be observed that an inflection has occurred after about 1000 days. During late time, more production data points are shown
deviated from the straight line, indicating that this well had started BDF. Based on production data prior to the inflection point
to generate the related parameters (a & m) from both Eq4 and Eq.5, Duong’s method will significantly over-predict the future
4 SPE 167118

gas production and thereby, the final EUR as shown in Figure 6.

(a) (b)
Fig 4 – Duong’s Specialized Plots (a) and Their Respective Gas Production Prediction (b)
Case 1: k = 0.01mD, A=160 acres, pi=3500psia, Tr=250F, h=50ft, Stages =5, xf = 300ft. Fcd = 300, Open hole, pwf=800 psia
Case 2: k = 0.0001mD, A=160 acres, pi=3500psia, Tr=250F, h=50ft, Stages =8, xf = 300ft. Fcd = 3000, Open hole, pwf=800 psia

Fig 5 –Specialized Plot of a Well Example Fig 6 – Duong’s Prediction of a Well Example

Doung’s Method vs. Arps Hyperbolic Decline


As mentioned above, the author of this paper has found that the occurrence of infection point on Duong’s Specialized Plot is
because the well starts flowing at BDF. It is also well understood that when a gas well is flowing in Pseudo-Steady State
(PSS), its production can be easily predicted with the hyperbolic decline analysis. Therefore, there must be a certain
relationship between those two parameters (a & m) from the Duong’s Specialized Plot after the inflection point and the
hyperbolic decline parameters (b & Di). Such kind of relationship would enable people to use Duong’s Specialized Plot not
only to identify the PSS of the well production but also to calculate the correct b & Di for the hyperbolic analysis.

As illustrated in the Appendix, following relationship has been found between (a & m) and (b & Di):

1
b= (6)
m

 1   am1− m 
LogDi ≈   Log   (7)
1− m   m −1 
To validate the above relationship, synthetic data generated from commercial software had been employed and are plotted in
Figure 7. Both reservoir and well parameters are also listed in Table 2. An inflection has been observed after about 3000
days. The above relationship were also tested based on data after the well has been on production for 9 years to ensure that the
well was producing at PSS. Using an economic gas rate of 50 Mscf/D, the EUR is found to be about 2300 MMscf. From year
SPE 167118 5

9 to year 13, using a & m values obtained from the Duong’s Specialized Plot for each year to calculate the corresponding b &
Di values. Future production is then forecasted based on Duong’s Method and the Hyperbolic Decline Method to derive the
well’s EUR. Results and the EUR percentage of error for each year are summarized in Table 3. It can be seen that the

Table 2 – Reservoir and Well Parameters for Generating the Synthetic Data
Reservoir Parameters: Wellbore Parameters:
Reservoir Pressure (p i )= 2200 psia Horizontal Wellbore (L ) = 6562.0 ft
o
Reservoir Temperature (T r ) = 180 F No. of Satge = 10
Permeability (k ) = 0.01 mD Fracture Conductivity (F CD ) = 300
Porosity (Phi ) = 6% Fracture Half-Length (xf ) = 300 ft
Water Saturation (S wi ) = 30% Flowing Pressure (p wf ) = 600 psia
Net Pay (h ) = 39.37 ft Fluid Parameters:
Drainage Area (A ) = 320 acres Sweet Dry Gas Gravity (γ g ) = 0.65
Xe/Ye (Rxy ) = 4.5 CO2 (%) = 1.3
OGIP = 3368 MMscf N2 (%) = 0.5

Table 3 – % of Errors of EUR Prediction by


Hyperbolic Decline Using b & Di Values
Calculated from a & m Parameters from
Duong’s Plot for the Synthetic Well Example

Duong's Parameters Hyperbolic Parameters EUR


Year a m b Di %Error
9 19.781 1.507 0.664 0.0010954 5.65%
Inflection point
10 32.906 1.569 0.637 0.0012551 4.48%
11 51.835 1.625 0.615 0.0013831 3.59%
12 82.847 1.681 0.595 0.0014578 3.06%
Fig 7 –Specialized Plot of a Synthetic Well Example 13 123.87 1.729 0.578 0.0015083 1.90%

percentage of error are getting smaller at later years and is less that 5% after 9 years.

Best Practice of Using Duong’s Rate Decline Method


The following steps are recommended as the “Best Practice” for using the Duong’s Rate Decline Method. They are based on
the author’s experience gained after examining a good number of producing wells from various tight WCSB formation wells:

1. Knowing your reservoir prior to your analysis – which means to collect/understand your reservoir parameters
including k/μ, h, pi, Tr , and reservoir fluid properties of gas gravity (γg), the percentage of other elements (H2S, CO2,
N2, etc.). Reservoir parameters could be obtained from adjacent vertical wells through pressure transient analysis
(PTA), rate transient analysis (RTA) and static pressure gradient test as well as core analysis etc.. Fluid properties can
be collected from laboratory analysis reports;
2. Knowing your well – this should include your proposed
project well spacing (drainage area), proposed
horizontal wellbore length, proposed type of completion
(open-hole or casing-hole multi-stage fracturing);
3. Knowing your surface production condition – this is
required to determine the flowing bottom pressure
(Pwf);
4. Setting-up your mechanistic simulation model based on
the above understood parameters to create your
synthetic production profile (type-curve), thus can be
used to generate your Duong’s specialized plot.
5. Finding the inflection point (as shown in Figure 7) to
decide when you should use Duong’s Rate Decline Fig 8 – Duong’s Prediction of a Synthetic Well
Method by comparing your economic production rate to Example
6 SPE 167118

the production at the inflection point:


 Yes to use - if no inflection occurs prior to your e-limit production rate;
 Not to use directly - if inflection occurs prior to your e-limit production rate.

6. Under the situation where the production data has shown an inflection point before reaching its economic rate, using
production data points after the inflection point is recommended to generate Duong’s a&m parameters to predict
future production history and its associated EUR.

This concept can be further clarified by comparing the percentage error of the EUR predictions from the analysis of the above
synthetic well example generated from simulation model of Table 2. From the plot in Figure 7, Duong’s decline parameters
are a = 0.7743 and m = 1.078 respectively, prior to the inflection point. This plot generates an EUR greater than 4,000MMscf,
which as shown in Figure 8 indicates an error of greater than 70%. However, if the data points were used after the inflection
point, Duong’s decline parameters would yield a = 27.661 and m = 1.548, respectively, which would result in only about a 2%
error in the EUR prediction as shown in Figure 8.

Moreover, it has been noticed that when an inflection point occurs in Duong’s specialized plot, it usually occurs during the
very late stage of a well’s production life. Therefore, the percentage of the remaining reserve (or EUR) might be very small at
this time with respective to its total EUR. This actually tells us that there is no obvious advantage of using Duong’s Method to
analyze data during this period. Under such a circumstance, it might be much easier to just use the simple Hyperbolic Decline
Method.

The earlier tight gas example well was used again to validate this practice. As shown in Figures 9 and 10, early data points,
prior to inflection, would have yielded a smaller a and m values (a = 0.6732, m = 1.064), which would result in a large
positive error of both production and EUR prediction. However, by using the late stage data (after the inflection point) would
yield a much larger a and m values (a = 25.62, m = 1.609), which would forecast a much more reasonable future production
and its associated EUR, as shown in Figure 10.
.

Fig 9 –Specialized Plot of a Well Example Fig 10 – Duong’s Prediction of a Well Example

Further, by using Eq.6 and 7, the hyperbolic decline parameters are calculated as b = 0.622 and Di = 0.003467 (1/Day). Its
production forecast thus overlaps with the Duong’s forecast if the late stage data points, after the inflection point, are used as
shown in Figure 10.

Using Stretch Exponential Production Decline (SEPD)


Stretch Exponential Production Decline (SEPD) was initially developed by Valko in 2009 (Valko, 2009) to analyze the
production decline of Barnett shale wells. Later in year 2010, based on the observation that group production for a large
number of wells will follow the SEPD model, Valko and Lee (Valko et al, 2010) had used this method to provide a controlled
production forecast for an individual tight/shale gas well based on the parameters processing from a group of wells. At the
mean time, Kabir et al (Kabir et al, 2010) also used the same method to analyze oil wells in tight reservoirs and found that the
SEPD method could yield a more conservative yet more accurate solution compared to the Arps’ Hyperbolic Decline Method.

SEPD and Its Pitfalls


Having assumed that gas and oil production follows stretched exponential decline in tight reservoirs, Valko had introduced a
parameter called “Recovery Potential (γp)” defined as:
SPE 167118 7

−1
Gp  1 1  q 
γ p = 1− =  Γ    × Γ  ,− Ln  (8)
EUR   n   n  q0 

q 0τ  1 
Where EUR = Γ
n  n 
, which is defined as the cumulative production when production rate is approaching 0. Gp is the

cumulative fluid production to date, q0 is the initial maximum production; n and τ are the two critical parameters that control
the decline equation and can be easily found by forcing the intercept be 1.0 of the linear plot of γp vs. Gp.
Later, in year 2012, Yu (Yu, 2012) conducted a study by applying this SEPD Method to a number of WCSB tight gas reservoir
wells, especially for wells with permeability ranging from 0.1mD to 0.001mD. It has been found that applying the “Recovery
Potential” concept to find the two parameters, n and τ, will bring not only a very conservative EUR forecast, but also a
potential wrong production profile forecast. Figure 11 shows the SEPD forecast for the above practical example well. It
applies the “Recovery Potential” concept to the first 2-year production history, in which Figure 11a shows that it is not a
perfect straight line if all early production data points are included to reach 1.0 at the y-intercept. As a result, its predicted
production profile will deviate from the real production data and have a very conservative final EUR, as shown in Figure 11b.

(a) (b)
Fig 11 – SEPD Analysis for a Practical Well Example: (a) Recovery Potential; (b) Production Forecast

The Modified Stretch Exponential Production Decline (YM-SEPD)


In order to resolve the issues that SEPD method has encountered, Yu et al (Yu et al, 2013a) has developed a modified SEPD
method (YM-SEPD) by introducing the following equation to find the two critical parameters n and τ:

q 
Ln  0  = τ − n t n (9)
 q(t )
q 
Based on Eq. (9), one can plot Ln  0  ~ t on a log~log scale, which would result in a straight line. The slope of the
 q(t )
straight line would be the value of “n” and “τ” can be easily calculated from the intercept of the straight line.

In order to validate the Method, a good number of synthetic data were generated by a commercial simulator with permeability
ranging from 0.1mD to 0.001mD. It has been found that this new methodology is capable of predicting production profile and
EUR, with reasonable accuracy, for a well having only the first 2~3 years of production history. In his paper, Yu et al explains
in great details about the reasons why this MY-SEPD Method works. In addition, the YM-SEPD Method has also provided us
with a simple way to identify a well’s BDF and the equations for calculating the corresponding hyperbolic decline parameters
b & Di.
Subsequently, Yu et al (Yu et al, 2013b) have also applied the YM-SEPD Method to various types of reservoir including wet,
retrograde gas, volatile and black oil. Yu employed both synthetic and practical well data to illustrate the application of this
YM-SEPD Method. In Yu’s study, hundreds of practical wells from various WCSB formations were analyzed and the wells
were completed with different types of multi-stage of fracturing. Yu has finally concluded that the YM-SEPD Method is
8 SPE 167118

applicable for production forecasts and EUR estimation in dry gas, wet gas, retrograde gas, volatile and black oil wells as long
as the wells’ production profiles having the typical “Hockey-Stick” profile – a characteristic that is prominent in most
horizontal wells with multi-stages fracturing in extremely tight reservoirs. In addition, Yu concluded that production data
during early months of production are not suitable to be used in the SEPD model and therefore, they should not be included in
parameter determination. If they are included, the forecasts would be more conservative.
Figure 12 presents the same practical gas well example that is now analyzed with YM-SEPD approach. A straight line has
been observed, as shown in Figure 12a, from the specialized plot which shows n = 0.3059. Using its intercept (0.3985),
another parameter “τ” can be easily calculated to be 20.24. The well’s production forecast is shown in Figure 12b. It is very
obvious that the prediction with YM-SEPD is much more reasonable than the original SEPD method.

(a) (b)
Fig 12 – YM-SEPD Analysis for a Practical Well Example: (a) YM-SEPD Specialized Plot; (b) YM-SEPD
Production Forecast

Best Practice of Using Empirical Production Decline Methods


As we all know, the utmost important criterion for any empirical analysis is to have a good knowledge of your target reservoir,
its production history and reservoir conditions. These usually include reservoir permeability range, net pay, porosity and initial
water saturation, etc. For well conditions, it will include the well type (vertical vs. horizontal), completion and production
details, well spacing (drainage area) and all surface pipeline pressure data, etc. All these essential knowledge would,
hopefully, improve the accuracy of our analysis and help the analysis to be more reasonable and realistic.

After examining more than one thousand horizontal wells that are producing from different WCSB formations such as
Cadomin, Montney, Notikewin, Cardium, Falher, Glauconite etc., which were completed under different types of multi-stage
fracturing, the following conclusions can thus be drawn as “best practice” for applying empirical methods:

For tight/shale reservoirs with permeability ranging from 0.1mD to 0.001mD

The YM-SEPD method is able to accurately predict production and EUR for both oil and gas wells having only 2~3 years of
production history, regardless the well’s flow regime and the size of its remaining reserve. When reservoir permeability
ranging from 0.01mD to 0.001mD, up to 3~4 years production history is required to be included in YM-SEPD analysis in
order to yield a reasonable forecasting.

If the well has less than 2 years of production history, a comprehensive method (Yu, et al, 2013) of combining Duong’s
method with YM-SEPD is recommended to use. When reservoir permeability is ranging from 0.01mD to 0.001mD, coupling
up to 4 years of “Pseudo” rate (actual available history + generated “Pseudo” history) is recommended to YM-SEPD analysis,
while only up to 3 years of “Pseudo” rate required for reservoirs with permeability being greater than 0.01mD. When applying
the Duong’s Method to generate “Pseudo” rates, one should always leave out the first data point of his specialized plot in order
to obtain a more reasonable Duong’s decline rate.

For tight/shale reservoirs with permeability less than 0.001mD

Under such a permeability condition, the “best practice” using the Duong’s decline method can still be applied. However, if
the YM-SEPD approach is used, a longer production history of more than 3~4 years is needed to generate a correct set of n &
SPE 167118 9

τ parameter for any forecast.

For all reservoirs

If only the first year production history is available, the YM-SEPD approach can provide us with a conservative prediction for
both production volume and the EUR. For being conservative in proved developed production (PDP) reserve booking, it is
recommended to include, at least, a few months of first year production data.

In case that BDF has been observed from the history data, Arps’ hyperbolic decline can just be applied for production and
EUR prediction due to its simplicity and versatility in many commercial types of software. Decline parameters (b & Di) can be
simply calculated from both YM-SEPD and Duong’s specialized plots.

If pressure data being available, the flowing-material-balance (FMB) method is recommended as an additional method to
analyze well’s production history, even though it most likely yields a lower limit of the EUR.

Finally, it is important to emphasis that it is always a good practice of using multiple methods to analyze a well’s production
data for future production and EUR forecasting, which should not exclude reservoir simulations based on either an actual
model or a mechanistic model.

An Actual Well Example – Comparison of Different Empirical Methods


A horizontal well in WCSB tight Cadomin formation was chosen for the analysis under various methods mentioned above.
The Cadomin well is located in Alberta, which has slight different reservoir properties as the well in the earlier practical
example discussion. The PTA data from vertical wells in the area suggests that the in-situ reservoir permeability is about
0.05mD, initial reservoir pressure and temperature are about 2430 psia and 183 oF, respectively; the average formation pay is
about 12m; the average porosity and the corresponding initial water saturation are about 8% and 30%, respectively. The
horizontal wellbore length is about 3565ft and it was completed with 9 stages of fracturing in year 2004. The well has
therefore been on production for more than 9 years without long-term interruption, as shown in Figure 13. The bottom-hole
pressure (BHP) history as shown in Figure 13 as well was calculated from its wellhead pressures.

Figure 14 has shown the diagnostic plot of flow regimes for the example well. It can be observed from both the normalized
rate and the derivative curves that the well has reached BDF at the late stage of production. Therefore, from its FMB analysis,
the well’s controlled OGIP can be obtained to be about 2600 MMscf, as shown in Figure 15. Simply applying a volumetric
recovery factor of 85% at this area for 1000m horizontal well, the final EUR could be predicted as 2210 MMscf at an
economic rate of 25Mscf/D.

Fig 13 –Production History of a Practical Well Fig 14 –Diagnostic Plot of Flow Regime of a
Example Practical Well Example

Using a commercial PDA software package, production rate history match, as shown in Figure 16, under the actual BHP has
been obtained based on an analytical model. By that, reservoir permeability is calculated to be 0.0165mD and the total fracture
half length to be about 1215ft. From the matched model, the final EUR was estimated to be around 2300 MMscf at the same
economic rate.
10 SPE 167118

10000

Gas Rate, Mscf/D


1000

100

History Gas Rate


Analytical Model Gas Rate
10
0 500 1000 1500 2000 2500 3000 3500
Time, Days

Fig 15 –FMB Analysis of a Practical Well Example Fig.16 – Rate History Match for the Practical

The four empirical methods (SEPD, YM-SEPD, Duong’s Rate Decline and Hyperbolic Decline) have been applied to this well
in order to compare the results obtained under each of the four methods. Figure 17 shows both the YM-SEPD and the
Duong’s specialized plots, while Figure 18 shows the production forecasts from the four methods. Table 4 summarizes all the

YM-SEPD & Duong's Specialized Plot 10000


10 1.E+00
History Gas Rate
y = 0.2284x0.3172 Analytical Model Gas Rate
YM-SEPD Model
R² = 0.9685 1.E-01 Duong's Rate Decline
Hyperbolic Decline
Gas Rate, Mscf/D

1000
SEPD
q/Gp, 1/Days
Ln(q0/q)

1.E-02
1
1.E-03
100

y = 3.1942x-1.257 1.E-04
R² = 0.9786
0.1 1.E-05 10
10 100 1000 10000 0 2000 4000 6000 8000 10000 12000
Days Time, Days

Fig 17 –Specialized Plots of a Practical Well Fig 18 –Production Forecast of a Practical Well
Example Example from Different Methods

Table 4 – Results from Different Methods for The


Practical Example
EUR To Produce Parameters
Method % Error
MMscf Years
km 0.0165
Analytical 2300 4.07% 42.0 1215
T otal x f
n 0.3172
YM-SEPD 2163 -2.13% 34.4 τ 105.1393
b 0.7439948
Hyperbolic* 2272 2.81% 40.1 0.0002913
Di
a 3.1942
Duong's 2725 23.30% 61.7 m 1.2567
n 0.3912
SEPD 1888 -14.6% 21.2 τ 191.50632
* Hyperbolic Decline Parameters are calculated from YM-SEPD (Yu, et al, 2013a)

EURs and their respective percentage errors compared with the calculated value based on recovery factor. Hyperbolic decline
parameters are calculated from Eq.9 and 10 in paper (Yu et al, 2013a).

From Table 4, one can notice that YM-SEPD, Hyperbolic Decline and the Analytical Model give a very close EUR, while
SEPD, as expected, has predicted a lower value and Duong’s decline has a higher EUR prediction even from its late part slope
of the specialized plot (Fig.17)
SPE 167118 11

It is very important to remember that the above analysis is for a well that has been producing for a long period of time and has
already reached BDF. What would the forecast of both production and EUR look like if the well had only 3 years of
production?

Figure 19, 20 and Table 5 show the results from the same four methods with the only first 3-years production data being
analyzed. One can see that the YM-SEPD is superior over the other methods - the YM-SEPD has a much more reasonable
prediction. The hyperbolic decline analysis with parameters (b & Di), calculated from the YM-SEPD approach, has a different
prediction from YM-SEPD, indicating that the well had not reached BDF after 3 years of production (YU, et al, 2013). If
additional one year of “Pseudo” rate is added to the analysis, which can be generated by the Duong’s decline analysis, the YM-
SEPD can predict an EUR that is much closer to the actual value.

YM-SEPD & Duong's Specialized Plots


10 1.E+00
y = 0.1688x0.3564
R² = 0.9712
1.E-01

q/Gp, 1/Days
Ln(q0/q)

1 1.E-02

y = 1.35x-1.134 1.E-03

R² = 0.9941
0.1 1.E-04
10 100 1000 10000
Days

Fig 19 –Specialized Plots of a Practical Well Fig 20 –Production Forecast of a Practical Well
Example (3 years data) Example from Different Methods (3 years data)

Table 5 – Results from Different Methods for The


Practical Example (3 years data)
EUR To Produce Parameters
Method % Error
MMscf Years
km 0.0255
Analytical 2391 8.19% 42.4 801
T otal x f
n 0.3564
YM-SEPD 2003 -9.37% 25.4 τ 147.18594
n 0.3414
YM-SEPD+Duong's ** 2180 -1.36% 30.7 τ 140.1324
b 0.882912
Hyperbolic* 2377 7.56% 45.6 0.0006651
Di
a 1.35
Duong's 4187 89.46% 133.6 m 1.1343
n 0.4765
SEPD 1329 -39.9% 6.5 τ 200.454

* Hyperbolic Decline Parameters are calculated from YM-SEPD (Yu, et al, 2013a)
** Added additional one year "Pseudo" rate to the YM-SEPD analysis generated by
Doung's decline

Conclusions
Based on the evaluation study, which has applied empirical methods to various tight and shale gas wells under a wide range of
permeability, the following “best practices” to use empirical methods can be concluded as:

1. Both SEPD and Duong’s Method are not applicable to tight/shale reservoirs with permeability ranging from 0.001md
~ 0.1mD. If they are applied, the SEPD approach will give an inaccurate production profile and a very conservative
EUR while the Duong’s decline approach will over-predict both production profile and EUR.

2. YM-SEPD gives more reasonable results over other empirical methods for production and EUR forecast especially
12 SPE 167118

when a well has very limited production history.

3. For reservoir permeability less than 0.001mD, step by step procedures have been proposed to employ Doung’s
decline method. If the YM-SEPD is used, up to 3~4 years of production history is required for a reasonable
production forecast result.

4. When a well has only limited production history, coupling the YM-SEPD Method with the Duong’s decline approach
will generate a few more years of “Pseudo” rate into analysis, which would significantly improve the production
forecast accuracy.

5. If pressure data is available, FMB and other analytical model analyses should not be excluded in the analysis. Such
analyses would unquestionably increase the confidence level of results obtained through empirical methods.

Acknowledgments
The author of this paper would like to thank the management of ConocoPhillips for their time to review and permission to
publish this paper. Many thanks are also due to Mr. Kien Wong of ConocoPhillips Canada for his time to review this paper.

Nomenclature
a = Duong’s Rate Decline Parameter, 1/Day
b = Hyperbolic Decline Exponent, dimensionless
Di = Hyperbolic Decline Rate, 1/Day
EUR = Expected Ultimate Recovery, MMscf or MSTB
Gp = Cumulative Gas Production, MMscf
m = Duong’s Rate Decline Parameter
n = Parameter in SEPD equation, dimensionless
q0 = Initial Gas or Oil Rate, Mscf/D or STB/D
qg = Gas Rate, Mscf/D
qo = Oil Rate, STB/D
q(t) = Gas or Oil Rate, Mscf/D or STB/D
t = Time, days
γp = Recovery Potential, defined by SEPD method
τ = Parameter in SEPD equation, dimensionless
Γ = Gamma Function

References
1. Arps J.J. 1945. Analysis of Decline Curves. Trans. AIME: 160, 228-247
2. Anderson, D.M., Nobakht, M., Moghadam, S. and Matter, L.: “Analysis of Production Data from Fractured Shale Gas Wells, ” paper
SPE 131787 presented at the SPE Unconventional Gas Conference held in Pittsburgh, Pennsylvania, USA, 23-25 February 2010
3. Bello, R.O. and Wattenbarger, R.A.: “Rate Transient Analysis in Naturally Fractured Shale Gas Reservoirs,” paper SPE 114591
presented at the CIPC/SPE Gas Technology Symposium 2008 Joint Conference held in Calgary, Alberta, Canada, 16-19 June 2008
4. Bello, R.O. and Wattenbarger, R.A.: “Modeling and Analysis of Shale Gas Production with a Skin Effect,” paper PETSOC-2009-082
prepared for the Proceedings of the Canadian International Petroleum Conference (CIPC) 2009, Calgary, Alberta, Canada, 16‐18 June
2009
5. Bello, R.O. and Wattenbarger, R.A.: “Multi-Stage Hydraulically Fractured Shale Gas Rate Transient Analysis,” paper SPE 126754
presented at the SPE North Africa Technical Conference and Exhibition held in Cairo, Egypt, 14-17 February 2010
6. Cipolla, C.L., Lolon, E.P. and Mayerhofer, M.J.: “Reservoir Modeling and Production Evaluation in Shale-Gas Reservoirs,” paper
IPTC 13185 presented at the International Petroleum Technology Conference held in Doha, Qatar, 7-9 December 2009
7. Duong, Anh N.: “An Unconventional Rate Decline Approach for Tight and Fracture-Dominated Gas Wells,” paper SPE 137748,
presented at the Canadian Unconventional Resources & International Petroleum Conference held in Calgary, Alberta, Canada, 19–21
October 2010
8. Ilk, D., Rushing, J.A., and Blasingame, T.A.: “Decline Curve Analysis for HP/HT Gas Wells: Theory and Applications” paper SPE
125031 presented at the 2009 Annual SPE Technical Conference and Exhibition, New Orleans, Louisiana, 4-7 October 2009
9. Kabir, C.S., Rasdi, F. and Igboalisi, B.: “Analyzing Production Data from Tight Oil Wells,” paper SPE 137414 presented at the
SPE 167118 13

Canadian Unconventional Resources and International Petroleum Conference held in Calgary, 19-21 Octoboer 2010
10. Lee, J., and Wattenbarger, R.A. : Gas Reservoir Engineering, SPE Text Book Series, Vol. 5 Texas (1996)
11. Lee, W.J. 2012a:“Estimating Reserves in Unconventional Resources,” a SPE Webinar presentation on Jan. 26, 2012
12. Lee, W.J. 2012b:“Estimating Reserves in Unconventional Resources,” a SPE Webinar presentation on Feb. 23, 2012
13. Valko, P.P: “Assigning Value to Stimulation in The Barnett Shale-A simultaneous Analysis of 7000 Plus Production Hystories and
Well Completion Records,” paper SPE 119369 presented at the 2009 SPE Hydraulic Fracturing Technology Conference held in the
Woodlands, Texas, USA, 19-21 Jan., 2009
14. Valko, P.P. and Lee, W.J.:“A Better Way to Forecast Production from Unconventional Gas Wells,” paper SPE 134231, presented at the
SPE Annual Technical Conference and Exhibition held in Florence, Italy, 19-22 September 2010.
15. YU, S. 2012: “A Modified SEPD Method to Predict EUR & Production for Extremely Low Permeability Wells with Short Production
History,” prepared for peer review in ConocoPhillips Company
16. Yu, S. and Miocevic, D. J. (2013a): “An Improved Method to Obtain Reliable Production and EUR Prediction for Wells with Short
Production History in Tight/Shale Reservoirs,” paper prepared for presentation at the Unconventional Resource Technology Conference
held in Denver, Colorado, USA, 12-14 August 2013
17. Yu, S., Lee, J.W., Miocevic, D. J., et al, (2013b): “Estimating Proved Reserves in Tight/Shale Wells Using the Modified SEPD
Method, ” paper SPE 166198 presented at the SPE Annual Technical Conference and Exhibition held in New Orleans, Louisiana, USA,
30 September – 2 October 2013

Appendix – Derivation of Hyperbolic Decline Parameters (b & Di) from Duong’s Decline Parameters (a & m)
From hyperbolic decline, production rate q(t) is defined as:
qi
q (t ) = (A-1)
(1+ bDi t ) b
1

And the cumulative production will be:

qi  ( )
b −1

Gp = 1 − 1 + bD t (A-2)
(1 − b )Di  
b
i

Re-write Eq.(A-1) to be:

qi qib q(1t−)b
Gp = − (A-3)
(1 − b )Di (1 − b )Di
From Eq.(A-1) and Eq.(A-3), it can be obtained that,

q( t )
=
(1 − b )Di =
(1 − b )Di (A-4)
(1 + bDi t )b − (1 + bDi t )
b 1
Gp qi q 
− i
q(t )  q( t ) 
At later time when production time value t is getting bigger, in Eq.(A-4), there will be:

(1 + bDi t )b 〉〉 (1 + bDi t )
1
(A-5)

Therefore, Eq.(A-4) becomes

q( t )

(1− b )Di (A-6)
(bDi t )b
1
Gp

Compare Eq.(A-6) with Eq.3, it is not hard to have:

1
b= (A-7)
m
14 SPE 167118

And

a=
(1− b )Di (A-8)
(bDi )b
1

After replacing b in Eq.(A-8) with Eq.(A-7), it will become:

1
am −m Dim−1 = 1 − (A-9)
m
Then, it is not difficult to have:

 1   am1− m 
LogDi ≈   Log   (A-10)
1− m   m −1 

Вам также может понравиться