SPE 120737

A Rapid and Efficient Production Analysis Method for Unconventional &

Conventional Gas Reservoirs
C.L. Jordan, SPE, BOE Solutions; C.R. Smith, SEMCams; and R. Jackson, SPE, BOE Solutions

Copyright 2009, Society of Petroleum Engineers

This paper was prepared for presentation at the 2009 SPE Asia Pacific Oil and Gas Conference and Exhibition held in Jakarta, Indonesia, 46 August 2009.

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
Currently, there is an industry assortment of production analysis methods ranging from traditional decline and type curve
matching to rate-pressure normalization techniques and detailed production history matching. Yet despite its many
limitations, conventional decline analysis is still commonly used in gas (and oil) production analysis due to its minimal data
requirements and ease of application (regardless that without modification the method cannot be used to estimate reservoir
properties or formation damage). As a result, this paper presents an automated computer method for estimating original gas-
in-place and other reservoir flow parameter for conventional and unconventional gas reservoirs. This procedure overcomes
the limitations of convention decline in which the analyst must properly select the type of decline to avoid erroneous
forecasts, as well as more exotic methods requiring iterative computational procedures.

Onshore and offshore case studies and examples presented in this paper will demonstrate that a modified decline and
production analysis procedure will allow for proper identification of flow regimes, reliable evaluation of drainage area and
OGIP, and the prediction of future deliverability and depletion. Case studies will also show that up-scaled and aggregate
reservoir properties can provide a real measure of gas well deliverability and therefore a simpler, time-efficient model
analysis can be used. Data uncertainty, PVT error, stimulation appraisal, gas storage system (free or complex mechanisms
such as absorbed gas) and other factors will be discussed in the context of the case studies, and general reservoir
management.

When used appropriately, the approach is generally accurate and robust enough that it can provide valuable information in
circumstances of poor data quality. Finally, the procedure is extremely simple and can be implemented in desktop
applications or spreadsheets with minimal computational effort.

Introduction
Decline curve analysis has been used worldwide (particularly for single phase fluid flow) for forecasting future production
from oil and gas fields since Arps [1945] formularized the technique during the 1940s. Allegedly, the procedure can be
applied to production data for any reservoir drive or mechanism, and has been shown to be suitable for both vertical and
horizontal wells [Dou et al. 2009]. Fetkovich [1968] actually established guidelines about how decline curvature can indicate
different reservoir systems. However, despite its simplicity (and limitations of constant flowing pressure and fluid
properties), analysts typically want more information without having to incur more time consuming and costly processes; for
example, they are interested in reservoir permeability, formation damage (wellbore skin), and original fluid-in-place (OFIP)
in addition to expected ultimate recovery (EUR) generated from conventional decline. Furthermore, there is a desire to use
the aforementioned information to generate production forecasts, and evaluate varying operating conditions such as
compression.

As result, Fetkovich [1968] later proposed a substantial improvement in decline curve analysis by matching production data
onto specialized type curves for reservoir characterization: the procedure (which is a powerful diagnostic tool), used the Arps
depletion stems to analyze boundary dominated flow, and the van Everdingen and Hurst [1949] constant pressure type curves
for transient production. Al-Hussiany et al. [1966] later improved type curve analysis by addressing the effects of pressure
dependant gas properties viscosity and supercompressibility through the use of pseudo-pressure, while Agarwal [1979]
presented a pseudo-time function for incorporating pressure dependant properties gas viscosity and compressibility.

Modern methods such as those of Blasingame et al. [1991] and Argawal-Gardner [ 1998] are similar to Fetkovich in that they
use type curves for production data analysis (PDA). However, they are independent of production constraints and use
flowing pressure data, combined with analytical solutions, to evaluate hydrocarbons-in-place. Another highly popular
2 SPE 120737

method is the flowing material balance (FMB): a procedure in which variable rate-pressure data can be normalized for a
linear extrapolation to fluids-in-place [Mattar and Anderson, 2003]. Other procedures use tangent methods for evaluating
original gas-in-place (OGIP) based on the superposition of pseudo-steady-state (PSS) time functions. In general, the modern
methods (as summarized by Matter and Anderson [ 2003]) improve upon traditional techniques by normalizing variable rate-
pressure data, and handling non-linear fluid properties. These methods have become very popular and can even compete
with traditional well tests which provide many of the reservoir flow parameters (assuming suitable data is available), and
have been demonstrated to work for both conventional and unconventional gas systems including tight gas and shale gas
[Khasanov, Kransnov and Guk, 2008; Mattar, Gault, et al., 2008].

On the other hand, despite the power of modern production data analysis methods, they generally deviate from the general
work flow process and lose the simplicity that keeps the Arps decline popular: in short, economic constraints and timing of
projects prevent analysts from performing more advanced analysis. For example, a number of factors, including time to
reach pseudo-radial flow and the desire to maximize daily production, have made operators tend to decrease the number of
well tests performed [Khasanov, Kransnov and Guk, 2008]. Therefore, after acknowledging that the analyst should always
use all of the data at their disposal to develop an understanding of the production scenario [Blasingame, 2003], there is a need
for functionality comparable to the aforementioned advanced decline methods, but in a process similar to Arps decline.

To continue, given the problems generally associated with production data including poor resolution of rate measurements, or
even lack of backpressure measurements, a rigorous analysis method is not always possible. And as noted by Anderson et
al. [2006], production data often does not contain the quality nor the frequency sufficient to produce estimates of reservoir
flow parameters such as unique combinations of permeability and skin. For reservoirs with a moderate to high permeability,
Anderson et al. [2006] stated that one should not expect to be able to estimate permeability etc. with confidence based on
monthly production rates and pressures (although monthly production data may be sufficient for wells in low permeability
reservoirs where transient flow exists for months). Although the basic theory is the same for pressure transient analysis
(PTA) and rate transient analysis (RTA), each methodology is appropriate only is its own domain: RTA tends to provide less
quality and/or accuracy than PTA, but can be utilized to provide a full life cycle analysis [Mattar et al., 2008].

Recently, Muhammad Buba [2003], following the work of Knowles [1990], presented a summary of semi-analytic identities
and plotting functions which can be used to extrapolate or estimate OGIP using only production data (qg and Gp) without a
prior knowledge of formation and/or fluid compressibility, or even average reservoir pressure. However, the work
presented by Muhammad Buba [2003] and Knowles [1990] required atypical plotting functions and plots, and fell into the
category of advanced production analysis. Yet, an evaluation of the relationship (referred to as the BK model for simplicity)
shows that it can be re-arranged for a rapid evaluation of OGIP without deviating significantly from Arps decline method.
Specifically, the analyst performs the traditional decline process, and obtains the initial decline rate (qgi) of the Arps decline,
which is used by the BK model to provide OGIP as a function of time. Once OGIP and qgi are obtained, the procedure can be
modified (and automated) to produce either actual (or effective) reservoir parameters for rapid rate forecasting, with
applications to both conventional and unconventional gas. The integration of an automated reservoir model into the decline
process would ideally help remove some of the uncertainty associated with other modern PDA methods, as it is generally
accepted that if a good history match is achieved, it can be used as a proxy (or analog) for estimating future performance. In
the majority of this work, a tank model has been coupled to the BK model for history matching as it does not require any
complex coding of functions such as pseudo-time [Hager, Brown and Jones, 2001]. Generally, the objective is to balance
speed and accuracy of the analysis method.

As implied above, the modeling process in this work is capturing an analog reservoir which can be used for forecasting. As a
result, deviations in actual versus effective permeability will manifest themselves as errors in transient periods, but long-term
production forecasting will be reliable. Nonetheless, it has been shown that a simpler model reservoir model (a single well
completed in the centre of a circular reservoir) can be used to represent a far more complex reservoir system and still provide
representative reservoir characterization and accurate production forecasting. Jordan et al. [ 2006] empirically showed (using
synthetically generated data) that radial composite reservoirs, dual porosity reservoirs, and other complex scenarios could be
effectively reduced to an equivalent radial homogeneous (ERH) system with accurate reserves, using an equivalent effective
permeability (EEP). Most of the aforementioned work by Jordan et al. [2006] was based on introducing permeability
calculations into the popular a) rate-cumulative production type curves introduced by Argawal and Gardner [1998], as well
as b) normalized rate-time plots introduced by Ibrahim and others (Ibrahim 2004; Ibrahim, Wattenbarger and Helmy, 2003;
Ibrahim et al. 2003] whose objective was to linearize variable rate-pressure data to the equivalent single rate or constant
pressures cases and then evaluate OGIP. This procedure can also be justified in that despite SCADA systems, there is many
times still a lack of proper pressure recording/measurements required for the proper adaptation of modern PDA methods
[Mattar et al. 2008].

In short, the procedure will result in an effective permeability for many cases as the rate model assumes a vertical well fully
penetrating a producing zone within a circular bounded reservoir. Given the averaging effect of production data analysis,
SPE 120737 3

non-unique matches, erratic data sets and data smoothing, it is anticipated that the less rigorous approach used in this paper
will be justified. This procedure can be compared to the well recognized and established methods which generate
productivity indices as a measure flowing capacity based on production and rate data analysis [Yang, Gao and Yang, 1995].
The modification to these processes produced an equivalent permeability as suggested by Toh [1997] who using numerical
simulation showed that random permeability fields could be represented by an average effective permeability during pseudo-
steady state which was generally equal to the geometric mean permeability). Ultimately, the examples in this work illustrates
the ability to extract an effective productivity index (or permeability, drainage area etc.), as well as OGIP without having to
deviate significantly away from traditional Arps decline

Theory & Definitions

The BK model (outlined in Appendix A) is simply a quadratic equation relating OGIP to production rate and pressure. This
implies that the OGIP can be determined directly from standard flowing data recognizing that 1) cumulative production,
flowing pressure, production rate, and finally material balance is not simply a quadratic relationship, and 2) one should not
attempt to rely solely on the identity for calculation of OGIP from one individual data point [Buba, 2003;Knowles, 1990]. As
a result, to improve the successful use of the model, it is recommended that the analyst evaluates the OGIP trend established
from a series of data measurements and that the results be used in conjunction with a simultaneous Arps decline and
production model. Basically, as the Arps decline is performed, an OGIP curve is automatically calculated (drainage radius,
and other relevant parameters such as permeability are automatically extracted), while a history match is generated based on a
tank (or other) model. In short, the analyst solely adjusts the initial decline rate (qgi) until a linear zero slope of both OGIP
and permeability trend is achieved alongside a reasonable production match.

Data & Results: Simulated Examples

Internal Boundary
Based on the work of Horne and Sageev [1983; Horne and Sageev, 1986], a forecast was generated using an
internal boundary model assuming a constant sandface pressure (100 psia) using pseudo-pressure and pseudo-time [Al-
Hussiany and Ramey, 1966]. The internal boundary model assumes a 640 Acre reservoir with a 72 Acre hole (i.e. zero
permeability and porosity in the hole). Volumetrically, the reservoir has an OGIP of 5.4 Bcf with a reservoir permeability
of 20 md. Figure 1 shows a reservoir schematic of the internal boundary system. For analysis, the BK model is tied to both
Arps decline, as well as the tank model for gas production modeling.

Figure 2 provides the Arps Decline, and OGIP/permeability results while Figure 3 shows the history matching results
(including the NPI type curve match [Blasingame et al., 1989]), respectively. The BK model provided an OGIP of 5.4 Bcf,
and an effective permeability estimate of 10 to 16 md (depending on whether P2/z2 or pseudo-pressure is used) which
provided a suitable history match despite although the calculated permeability is lower than the true reservoir permeability of
20 md. An initial rate (qgi) of 14.5 MMscf/d was used to achieve the matches. For comparison, the popular FMB provided a
permeability estimate of 16.5 md and an OGIP of 5.4 Bcf as shown in Figure 4. In the next example, the procedure is
applied to random rock properties (i.e. permeability and net pay) in a manner similar to that suggested by Toh [1997].

Random Heterogeneity
Using an evenly distributed random number generator, values for permeability and net pay were varied from 0.05 to
21 md, and 0.3 and 59 ft (0.1 to 18 m) respectively with average values of about 5.3 md and 10.2 m (refer to Figure 5). The
reservoir was set to a square with sides of about 10,000 x 10,000 ft. Initial pressure was set to 5,000 psia, reservoir
temperature at 212 oF, and gas gravity at 0.7. Using FMB, the OGIP was estimated to be approximately 85 to 86 Bcf as
shown in Figure 6.

Using the BK model, the OGIP and effective permeability was evaluated as shown in Figures 7, with the calculated average
(effective) values to be 85 Bcf and 8.3 md. Using the effective values, a suitable production history match was achieved as
shown in Figure 8. An initial rate (qgi) of 30.1 MMscf/d was used to achieve the matches. The assumed net pay was 9.1 m.

Incidentally, similar work by Toh [1997] who evaluated the depletion performance of heterogeneous reservoirs (based on
production analysis of geostatistical models) indicated that an equivalent effective permeability (EEP) could generally be
observed, and that it remained generally constant throughout time, and that it was generally equal to the geometric mean
permeability. Toh [1997] made a general statement that the depletion performance of all reservoirs with randomly distributed
heterogeneity (and 88% of reservoirs) with sectional permeability fields can be predicted with an equivalent homogeneous
reservoir. And since most production and pressure transient methods see the reservoir as a volume average set of
properties [Blasingame, 2003], there is not necessarily a need for more detailed modeling and that a bounded radial
4 SPE 120737

homogeneous model may be suitable even if heterogeneity exists in a number of forms. Tohs [1997] results also suggested
that the EEP does may not always perform appropriately for highly heterogeneous reservoirs with the well completed in a
high permeability zone, the work presented by Jordan et al. [2006] does suggest the method is suitable to sectionally
homogenous reservoirs (i.e. triple zone composite reservoirs). Similarly, work by Yang et al. [1995] states that equivalent or
effective values determined from productivity equations in heterogeneous systems can act as a soft input for numerical
simulation. A random permeability example that incorporates structure is shown in the next section.

Random Heterogeneity Incorporating Structure

Similar to the previous example, an evenly distributed random number was used to generate values for net pay and
permeability varying from 30 to 105 ft, and 13 to 33 md, respectively (with average values both at 19). Additionally, the
formation top was set to increase approximately 422 over a distance of 14,000 ft in the north east direction. Initial pressure
was set to 1160 psia, reservoir temperature at 212 oF, and gas gravity at 0.7. Figure 9 shows the net pay and permeability
maps for this system, as well as a cross-section of the pool showing the variation in formation top and thickness. Using
FMB, the OGIP was estimated to be 23.6 Bcf as shown in Figure 10. The flowing pressure was 362.3 psia.

Figure 11 shows the Arps decline analysis, while Figure 12 shows the calculated OGIP and effective permeability.
As suggested by the analysis, the average OGIP is about 25 Bcf, while the effective permeability for this system is only
marginally below 20 md, indicating a value of 19 md (the FMB assumes an average net pay of 19 ft). Based on these results,
Figure 13 shows the rate history match using the results from the BK Model, while Figure 14 shows the corresponding
Blasingame and NPI type curves.

Unconventional Gas Systems

It is important to note that in order to adapt the proposed procedure to unconventional gas, the appropriate material
balance must be employed for the system under consideration. For example, to include absorded and free gas, which is
common in shale and CBM gas systems, the material balance method outlined by Clarkson et al. [Clarkson and McGovern
n.d.], or King [1993] should be used.

In this example, the permeability calculation is obtained using Eqn (10) of Appendix 1. After a number of simulations runs,
it was found that the difference in the definition of pressure between Eqn (1) and Eqn (9) can sometimes manifest itself into
serious permeability calculation errors. Specifically, Eqn (9) uses the concept of pseudo-pressure whereas the Eqn (1) uses
an approximate form of pseudo-pressure defined as P/z. Therefore, for CBM problems Eqn (10) was used to calculate
permeability.

Once the procedure is modified for absorbed gas, the BK model can then be applied to unconventional gas systems,.
In this example, a low pressure gas system (Pi = 600 psia), and with a low reservoir permeability of 0.5 md was generated
using a numerical simulator. Cleat porosity, Langmuir pressure, Langmuir Volume were set to 0.3%, 661 psia, and 250
scf/ton, respectively. Finally, gas gravity abd reservoir temperature was set to 0.554 and 120oF, respectively. As shown in
Figure 15, a 50 yr rate forecast was generated using a bottomhole pressure of 100 psia. Again, FMB (which was also
modified to account for adsorpted gas [Clarkson et al., 2007]) confirmed the OGIP to be 1.5 Bcf in Figure 16. The results
from the BK model are shown in Figure 17 which indicates OGIP and effective permeability to be about 1.5 BCF and 0.5
md, respectively. The rate history match (and NPI type curve) is shown in Figure 18.

Data & Results: Field Examples

Ballycotton Well:
Gas production data was taken from the Ballycotton Field, a major gas accumulation in the Celtic Sea, Ireland (Chu,
Fleming and Caroll 2000). Figure 19 shows a plot of the measured gas and flowing bottom hole pressures. Initial reservoir
pressure was set at 1,200 psia, net pay at 76 ft, porosity at 22.3%, formation temperature at 120 oF, and gas gravity at 0.554.
Using the BK model and an initial rate (qgi) of 53.2 MMscf/d, average OGIP was determined to be 38 Bcf (Figure 20), and
permeability was estimated to be about 109 - 100 md, respectively (not shown). The corresponding history match is shown
in Figure 21.

It is interesting to note that the measured gas rates and BHP are not entirely representative of a constant pressure condition
the BK model is robust enough to provide reasonable estimates of OGIP and proxy forecasting parameters. In the following
examples, unconventional gas field problems are addressed.
SPE 120737 5

Tight Gas Well J7

Wattenbarger et al. [1998], as well as others presented an analysis of tight gas wells from a field in South Texas, one
of which was J7. The well was hydraulically fractured and had been producing for nearly 23 years. Monthly production
rates and fluid/rock properties are the only data available. After a number of simulation runs and regressing on model
parameters, the two best fits reported by the authors were that of the linear homogeneous closed reservoir and radial transient
dual porosity closed reservoir providing an OGIP ranging from 6.9 7.1 Bcf (unique combinations of permeability and skin
could not be resolved in the original work). Average net pay, porosity, water saturation, gas gravity, and reservoir temp were
92 ft, 15 %, 47%, 0.65, and 290 oF respectively. Initial pressure was 8,800 psia.

The BK model provided an initial OGIP of approximately 4.1 Bcf (which is slightly lower than the reported results), with an
effective permeability of about 0.010 md, and an average drainage area of 40 to 48 Acres (refer to Figure 22). A review of
the production history match indicated that the results were generally suitable (Figure 23). It is obvious that the early
transient time is not matched as the well is stimulated and the parameter has not been incorporated into the analysis.
Nonetheless, a suitable production history match was obtained automatically, again by linking the tank model to the Arps
decline and the BK model.

Barnett Shale Wells

Stella Young 4, an early Barnett Shale gas well, was drilled and completed in the latter part of 1985 [GRI, 1991].
Upon completion, it was hydraulically fractured with approximately 470,000 gal of gel and 875,000 lb and 20/40 mesh sand.
Rates were initially high at 2 MMscf/d, but declined to about 0.2 MMscf/d over a period of 5 years Figure 24 shows the
measured gas rates and the associated decline line.

Average net pay, porosity, water saturation, gas gravity, and reservoir temp were 184 ft, 3.9 %, 41.2%, 0.747, and 192 oF
respectively. Initial reservoir pressure was estimated to be 4,000 psia while the flowing bottomhole pressure has been
relatively constant at 200 psia throughout the life of the well. Langmuir isotherm values PL and VL were estimated to be
400 psia and 40 scf/ton. Shale density was estimated to be 2.1 g/cm3. Performing production data analysis using
PROMATTM, the original study estimated permeability to be approximately 0.005 md and OGIP to be 1.1 Bcf with a
drainage area of 24.5 acres. The results also indicated fracture half-length to be approximately 130 150 ft (which
equivalent to a skin about -6.1).

Figure 25 shows the results from the BK model which provides an average OGIP of approximately 1.1 Bcf, and an extracted
permeability of 0.02 md. Although the calculated permeability is high compared to the results from the original analysis, it is
an effective permeability for the matrix and wellbore fracture combined and provides a reasonable history match as shown in
Figure 26 (Appendix B outlines the relationship between real and equivalent effective permeability). Incidentally, given the
low amount of absorbed gas (i.e. the low gas content) in the Barnett Shale example, this data set could be treat as purely a
tight gas example as opposed to an absorbed gas example.

For a more recent example, 32 months Barnett shale gas production data (refer to Figure 27) was obtained from an unnamed
well previously evaluated by Lewis et al. using advanced PDA methods, including type-curve analysis with pseudo-time.
The original work indicated an initial pressure, net pay, porosity, and reservoir temperature of 4,000 psia, 433 ft, 5%, and 180
o
F, respectively. Langmuir parameters VL and PL were 89 scf/ton and 635 psia, respectively (shale density was not provided
and assumed to be 1.3 g/cc). Drainage radius was estimated to be approximately 200-300 ft (28.2 Acres) and roughly equal
to an OGIP of about 1.4 Bcf. Permeability was estimated to be 0.003 to 0.009 md.

Based on these methods, the data indicated a drainage radius of 300 ft which equated to 1.4. Following the BK methodology
(Figure 28), the results indicate an OGIP of 1.8 1.9 Bcf (a drainage radius of about 345 ft which is slightly higher than that
presented by Lewis [2007]) and an effective permeability of about 0.006 md. FMB analysis (Figure 29) suggested a value
more comparable to Lewiss work. Although the original work suggested the ability to extract dual porosity and/or layered
results from the type curve analysis (Figure 30), both the results from type curve analysis and normalized decline (i.e.
flowing material balance), do not show significant transient data suitable to extracting detailed near wellbore flow
characteristics, and that the equivalent radial model is suitable.

Horseshoe Canyon (HSC) Coal Gas

In this example, data is taken from a multiple layer problem in the Horseshoe Canyon coals of the Western
Canadian Sedimentary Basin [Clarkson, Bustin and Seidle, 2008]. The individual coals are low rank (sub-bituminous) seams
and were N2 stimulated, and produced commingled. Average net pay, cleat porosity, gas saturation, gas gravity, and
reservoir temp are 49.9 ft, 0.1 %, 100%, 0.55, and 67 oF respectively. Initial reservoir pressure was low at 86 psia, while
flowing pressure was about 20 psia. Finally, coal density was 1.33 g/cm3. Flowing type curve and buildup pressure
analysis, performed by the original analysts, indicated an effective permeability of about 7.3 to 9.2 md, a wellbore skin (-1.4)
to (+0.2), and finally an OGIP of 152 MMscfd.
6 SPE 120737

Figure 31 shows the raw data, and the Arps decline match. A review of the BK model in Figure 32 provided an OGIP of 150-
170 MMscf, and an average effective permeability of 10-15 md (assuming zero skin). The associated history match is
shown in Figure 33. As indicated in the theory section of Appendix A, this analysis was performed using the modified BK
model.

Multi Well Examples

In order to evaluate production data from multiple wells simultaneously, and obtain OGIP for a pool (as opposed to the
individual wells stabilized drainage area), the BK model is adapted such that rates and cumulative production is based on
total pool rates, while pressures are evaluated at the well level. Essentially this procedure results in j OGIP and
permeability curves for a pool with j wells. Once the OGIP for the pool is evaluated, then permeability for each well is
calculated according to Eqn (9) or (10) based on well rates and flowing pressures.

In this example, a four well pool (refer to Figure 34) with an OGIP of 159.5 Bcf (i.e. a drainage area of 640 Acres) was
created using an analytical simulator [Jordan, Fenniak and Smith, 2006]. Net pay, porosity, water saturation, gas gravity,
and reservoir temp were 100 ft, 0%, 0.65, and 120oF, respectively. Initial reservoir pressure was 5,000 psia. Reservoir
permeability was 10 md. The reservoir was kept circular (refer to Figure 30) with the four wells spaced asymmetrically
throughout the reservoir. Wells A, B, C, and D had constant BHP of 1000, 1025, 980, and 1000 psia respectively. The BK
Analysis, as well as the total rate history, is shown in Figure 35.

Conclusions
The BK model, which linked to production modeling, is fairly robust can provide a reasonable estimate of OGIP and an
analog model for effective equivalent permeability (EEP) and/or effective drainage radius. Assuming PSS, the production
analog can be based on tank type models as heterogeneity and similar anomalies are assimilated through the EEP
although, there is no reason why the procedure cannot be automated using a coupled transient-PSS model incorporating
pseudo-pressure and pseudo-time. The procedure can be applied to both conventional and unconventional gas systems, as
well as abnormally pressure systems. Errors in reservoir parameters such as initial pressure and formation temperature
appear to have minimal impact on the calculated OGIP. Also, given the simplicity of the approach, coding and
implementation in spreadsheets or other desktop tools is easily accomplished.

It is understood that these linearization methods presented were developed under the assumption that PSS flow exists
(permeability, wellbore skin, net pay, etc. are not so interchangeable during transient flow periods). However, long-term
deliverability and productivity is generally of more concern than transient or flush production. The process is also assumes
that the data does exhibit traditional constant pressure decline (the method is highly comparable to conventional decline with
respect to the limitation of constant BHP). Nonetheless, the method appears to be robust and able to tolerate a small
deviation from a non-constant BHP condition.

Of course, similar to all analytical methods, the methodology presented in this paper does make certain simplifying
assumptions about production data analysis. For example, it does assume single phase volumetric reservoir behavior.
However, some non-volumetric effects, such as water-drive and interference among multiple wells can be handled effectively
using influence functions (e.g. Blasingame type curves have a multi well feature that can accommodate and account for
interference effects). The assumption of single-phase production in the reservoir is, in most cases, also considered valid
especially for gas wells (as gas compressibility dominates the material balance). The primary impact of multiphase
production in gas wells is in the wellbore, where special care must be taken to ensure that the pressure loss from surface to
bottomhole conditions is estimated correctly.

References
Agarwal, R. G. "'Real Gas Pseudo-Time' - A New Function for Pressure Buildup Analysis of MHF Gas Wells." SPE Annual
Technical Conference and Exhibition. Las Vegas, Nevada, 23-26 September 1979. SPE 8279.

Agarwal, R. G., D. C. Gardner, S. W. Kleinsteiber, and D. D. Fussell. "Analyzing Well Production Data Using Combined
Type Curve and Decline Curve Concepts." SPE Annual Technical Conference and Exhibition. New Orleans, 27-30
September 1998. SPE 57916.

Al-Hussiany, R., and H. J., Jr. Ramey. "Application of Real Gas Flow Theory to Well Testing and Deliverability
Forecasting." JPT, 1966: 624-636.

Anderson, D. M., G. W.J. Stotts, L. Mattar, D. IIk, and T. A. Blasingame. "Production Data Analysis - Challenges, Pitfalls,
Diagnostics." SPE Annual Technical Conference and Exhibition. San Antonio, 24-27 September 2006. SPE 102048.
SPE 120737 7

Arps, J. J. "Analysis of Decline Curves." Trans. AIME, 1945: 228-247.

Blasingame, T. A. "Analysis of Reservoir Peformance: Introduction." Course Notes: PE 663. Texas A&M University,
November 13, 2003.

Blasingame, T. A., J. L. Johnston, and W. J. Lee. "Type-Curve Analysis Using the Pressure Integral Method." California
Regional Meeting . Bakersfield, California, 1989. SPE 18799.

Blasingame, T. A., T. L. McCray, and W. J. Lee. "Decline Curve Analysis for Variable Pressure Drop/Variable Flow Rate
Systems." SPE Gas Technology Symposium. 23-24 January 1991. SPE 21513.

Buba, M. I. Direct Estimation of Gas Reserves Using Production Data. M.Sc., College Station: Texas A&M Thesis, 2003.
Chu, W., C. H. Fleming, and K. M. Caroll. "Determination of Original Gas-In-Place in Bally Cotton, Offshore Ireland."
SPE/CERI Gas Technology Symposium. Calgary, Alberta, 2000. SPE 59791.

Clarkson, C. R., and J. M. McGovern. A New Tool for Unconventional Reservoir Exploration and Development Applications.
Internal Document, Burlington Resources.

Clarkson, C. R., C. L. Jordan, R. R. Gierhart, and J. P. Seidle. "Production Data Analysis of CBM Wells." Rocky Mountain
Oil & Gas Technology Symposium. Denver, Colordao, 2007. SPE 107705.

Clarkson, C., R. M. Bustin, and J. P. Seidle. "Production Data Analysis of Single Phase (Gas) CBM Wells." Gas Technology
Symposium. Calgary, Alberta, 15-17, May 2008. SPE 100313.

Dou, H., et al. "Analysis & Comparison of Decline Models: A Field Case Study for the Intercampo Oil Field, Venezula."
SPE Reservoir Evaluation and ENgineering, 2009: 68-78.

Fetkovitch, M. J. "Decline Curve Analysis Using Type Curves." JPT, March 1968: 1065-1077.

GRI. Analysis and Production of Well Test Data from Barnett Shale Wells Operated by Mitchell Energy Corporation. GRI
Topical Report: Contract No. 5086-213-1446, GRI, August 1991.

Hager, C. J., T. Brown, and J. R. Jones. "Analyzing Flowing Production Data with Standard Pressure Transient Methods."
Rocky Mountain Petroleum Technology Conference . Keystone, Colorado, 21-23 May 2001. SPE 71033.

Horne, R. N., and A. Sageev. "Interference Testing: Detecting an Impermeable or Compressible Region." Annual Technical
Conference and Exhibition. New Orleans, 1986. SPE 15585.

Ibrahim, M. History Matching Pressure Response Functions From Production Data. Ph.D Thesis, College Station: Texas
A&M Thesis, 2004.

Ibrahim, M., R. A. Wattenbarger, and W. Helmy. "Determination of OGIP for Tight Gas Wells - New Methods." CIPC 2003.
Calgary, Alberta, 2003. CIPC 2003-12.

Ibrahim, M., R. A. Wattenbarger, and W. Helmy.. "Determination of OGIP for Wells in Pseudo-Steady State: Old
Techniques, New Approaches." Annual Technical Conference and Exhibition. Denver, Colordao, 2003. SPE 84286.

Jordan, C. L., C. R. Smith, and M. F. Fenniak. "Case Studies: A Practical Approach to Gas-Production Analysis and
Forecasting." Gas Technology Symposium . Calgary, Alberta, 15-17, May 2006. SPE 99351.

Jordan, C. L., M. J. Fenniak, and C. R. Smith. "An Efficient Model for Gas Rate Forecasting in Complex Reservoirs." Gas
Technology Symposium. Calgary, Alberta, 2006. SPE 99352.

Khasanov, M., V. Kransnov, and V. Guk. "Reservoir Parameters Evalaution Based on Production Data Analysis." SPE
Russian Oil & Gas Technical Conference in Moscow. Moscow, Russia, 28-30 October 2008. SPE 117406.

King, G. R. "Material-Balance Techniques for Coal-Seam and Devonian Shale Gas Reservoirs with Limited Water Influx."
SPE Reservoir Engineering, February 1993: 67-72.

Knowles, R. S. Development and Verification of New Semi-Analytical Methods for the Analysis and Prediction of Gas Well
8 SPE 120737

Peformance. M.Sc. Thesis, College Station: Texas A&M University, 1990.

Lewis, A. M. Production Data Analysis of Shale Gas Reservoirs. MSc Thesis, Louisiana State University , 2007.
Mattar, L., and D. M. Anderson. "A Systematic and Comprehensive Methodology for Advanced Analysis of Production
Data." SPE Annual Technical Conference and Exhibition. Denver, 5-8 October 2003. SPE 84472.

Mattar, L., et al. "Production Analysis and Forecasting of Shale Gas Reservoirs: Case History-Based Approach." SPE Shale
Gas Production Conference. Fort Worth, Texas, 16-18 November 2008. SPE 119897.

Sageev, A. Pressure Transient Analysis of Reservoirs with Linear or Internal Circular Boundaries (SGP-TR-65). Geothermal
Program Interdisciplinary Research Report, Stanford University, 1983.

Toh, S. The Depletion Performance of Hetergensous Reservoirs. PhD Thesis, College Station: Texas A&M University ,
1997.

van Everdingen, A. F., and W. Hurst. "Application of the Laplace Transformation to Flow Problems in Reservoirs." Trans.,
AIME, 1949: 306-324.

Wattenbarger, R. A., A. H. El-Banbi, and M. E. Villegas. "Production Analysis of Linear Flow into Fractured Tight Gas
Wells." Rocky Mountain Regional Low Permeability Reservoirs Symposium. Denver, Colordao, 5-8, April 1998. SPE 33931.

Yang, H., Y. Gao, and A Yang. "Application of Well Production Data to Reservoir Characterization." Annual Technical
Conference and Exhibition. Dallas, USA, 1995. SPE 30708.

Nomenclature
Greek Variables

a = Quadratic equation variable

b = Arps decline exponent, dimensionless
b = Quadratic equation variable
C1 = Constant for PSS equation
c = Quadratic equation variable
Di = Arps constant (initial decline rate)
Gp = Gas producced, MMscf
k = Reservoir permeability, md
keff = Reservoir permeability, md
OGIP = Initial Gas-In-Place, BCF
Pi = Initial Reservoir Pressure, psia
Pwf = Flowing BHP, psia
qgi = Arps Initial gas rate,
qg = Gas Rate, MMscfd
rw = wellbore radius, ft
re = reservoir radius, ft
s = Skin, dimensionless
tf = formation/reservoir temperature, oF/oR
z = z-factor, dimensionless

Greek Variables
wf = Wellbore flowing pseudo-pressure, psia/cp2
i = Initial Reservoir Pseudo-pressure, psia/cp2
SPE 120737 9

APPENDIX A: BACKGROUND THEORY

Muhammad Buba [2003] and Knowles [1990], presented the semi-analytic quadratic rate-cumulative
production relation as shown in (A-1).

2qgi qgi
qg = qgi Gp + G p2 (A-1)
pwf
2
pwf
2

zwf z wf
1 OGIP 1 OGIP2
i p i p
zi zi

A general review of (A-1) will show that it is a quadratic equation which can be simplified to (A-2) where a, b,
and c are defined by equations (A-3), (A-4), and (A-5) shown below.

q g = c b(G p / OGIP) + a (G p / OGIP) 2 (A-2)

qgiG p2
a=
pwf
2

zwf
1 (A-3)
pi
zi

2qgiG p
b=
pwf
2

zwf (A-4)
1
pi
zi

c = q gi (A-5)

Since (A-2) is a quadratic equation, solving for OGIP can now be easily accomplished as shown in (A-6), where
is defined as shown in (A-7), and qgi is one of Arps variables as shown in (A-8).

1 (2 2 ( + qg qgi ) )G p
2 0.5
(A-6)
OGIP =
2 (qgi qg )

1 qgi
=
2 pwf
2
(A-7)

1 zwf
pi
zi


10 SPE 120737

qgi
qg = (A-8)
(1 + bDit )1/ b

If one performs a traditional decline, an evaluation of OGIP can be automatically evaluated by substituting qgi
from (A-8) into (A-6) given knowledge of flowing pressure and initial pressure. Assuming that the drainage
area of the well is constant during PSS, then a relatively linear plot of OGIP should also be produced as function
of time.

Also, recognizing the PSS relationship in given in (A-9), an estimate of gas permeability (kg) can be calculated if
reservoir pressure (material balance) calculations are automated using the average OGIP estimated from (A-6) and
the measured field data. However, the best results for permeability extraction are obtained if (A-9) is modified
to (A-10) to be consistent with the definition of normalized pressure in (A-1)

( R wf )k g h
qg = (A-9)
r 3
C1T f ln e + s
w
r 4

P 2 P
2
(A-10)
R wf k h

z z g
qg = R wf
re 3
C1T f ln + s
rw 4

For adaptation to coal-gas, systems where there is absorbed and fee gas, (A-6) can be used. However, testing has
shown the best results are obtained when a constant of 10/12 is included in the solution as shown in (A-11). This
constant accounts for the fact that empirical testing showed that the BK results were typically 60% of the actual
OGIP.

10 1 (2 2 ( + q g qgi ) )G p 10 (2 2 ( + qg qgi ) )G p
2 0.5 2 0.5

OGIP = =
6 2 (qgi qg ) 12 (qgi qg ) (A-11)

For adaptation to pool evaluations, Eqns (A-6) and (A-11) can be adjusted such that gas rate and cumulative gas
produced is based on total pool production rates, but flowing pressure (found in ) is based on the individual
well. These comments are summarized by Eqns (A-12), (A-13), and (A-14).

1 (2 2 ( + qg , pool qgi, pool ) )Gp, pool

2 0.5

OGIP =
2 (qgi, pool qg , pool ) (A-12)

10 (2 2 ( + qg , pool qgi, pool ) )Gp, pool

2 0.5

OGIP = (A-13)
12 (qgi, pool qg, pool )
SPE 120737 11

1 q gi, pool
=
2 p
2
wf , welli
z wf ,welli (A-14)
1 pi



zi

APPENDIX B: EQUIVALENT PERMEABILITY ESTIMATES

The procedure outlined in Appendix A describes how to extract an equivalent effective permeability from
production data. This procedure assumes that subject well is a vertical well, completed in the center of
homogeneous reservoir. To estimate the equivalent effective permeability, one takes Eqn (B-1) and equates it to
Eqn (B-2), resulting in (B-3). Eqn (B-3) provides the effective or equivalent effective permeability.

( R wf )kh
qg = (B-1)
r 3
C1T f ln e + s
rw 4

( R wf ) keff h
qg = (B-2)
r 3
C1T f ln e
rw 4

k re 3
keef = ln (B-3)
re 3 rw 4
ln + s
rw 4
12 SPE 120737

Figure 1: Internal Boundary Schematic (NTS)

Figure 2: Decline and BK Analysis for Internal Boundary Example

SPE 120737 13

Figure 3: Production History Match for Internal Boundary Example

14 SPE 120737

Figure 4: FMB Analysis for Internal Boundary

10,000 ft

10,000 ft

Figure 5: Permeability (md) and Net Pay (m) Distributions

Figure 6: FMB Analysis for Random Heterogeneity Example

SPE 120737 15

Figure 7: Decline and BK Analysis for Random Permeability Example

16 SPE 120737

Figure 8: Decline Analysis and BK Analysis for Internal Boundary Example

SPE 120737 17

Figure 9: Permeability, Net Pay, and Cross-Section for Structure Example

Figure 10: Flowing Material Balance for Structure Example

18 SPE 120737

Figure 11: Arps Decline for Structure Example

Figure 12: BK Model Results for Structure Example

SPE 120737 19

Figure 13: Rate History Match for Structure Example

Figure 14: BK Model Results for Structure Example

20 SPE 120737

Figure 15: Simulated Raw Data for CBM Example

Figure 16: FMB for Simulated CBM Example

SPE 120737 21

Figure 17: OGIP and Permeaibility Results for Simulated CBM Example

Figure 18: Rate History Match and NPI Type Curve for Simulated CBM Example
22 SPE 120737

Figure 19: Raw Data for Offshore Example

Figure 20: OGIP Analysis for Offshore Example

Figure 21: Rate History Match for Offshore Example

SPE 120737 23

Figure 22: Raw Data, OGIP Analysis, and History Match for J7

Figure 23: Rate History Match & Type Curve for J7

24 SPE 120737

Figure 24: Raw Data and Decline Match for Stella Young 4

Figure 25: OGIP and Permeability Analysis for Stella Young 4

Figure 26: History Match for Stella Young 4

SPE 120737 25

Figure 27: Raw Data for Lewis Barnett Shale Example

Figure 28: OGIP and Permeability Analysis for Lewis Barnett Shale Example
26 SPE 120737

Figure 29: FMB Analysis for Lewis Barnett Shale Example

Figure 30: History Match & NPI Typecurve for Lewis Barnett Shale Example
SPE 120737 27

Figure 31: Raw Data And Decline Match for HSC Example

Figure 32: OGIP and Permeability Analysis for HSC Example

Figure 33: History Match for HSC Example

28 SPE 120737

re = 2978.9 ft

Figure 34: Reservoir Schematic for Multi-Well Example

Figure 35: OGIP and Permeability Analysis for Multi-Well Example