Anda di halaman 1dari 12

EM152066 DOI: 10.

2118/152066-PA Date: 1-February-13 Stage: Page: 5 Total Pages: 12

Resource Evaluation for Shale


Gas Reservoirs
Z. Dong, SPE, S.A. Holditch, SPE, and D.A. McVay, SPE, Texas A&M University

Summary the data sets, methodology, and tools to determine values of


Many shale gas reservoirs have been previously thought of as OGIP, TRR, RF, and economic viability in highly uncertain and
source rocks, but the industry now finds these source rocks still risky shale gas reservoirs.
contain large volumes of natural gas and liquids that can be pro-
duced by use of horizontal drilling and hydraulic fracturing. How- Petroleum Resources Management System (PRMS). The
ever, one of the most uncertain aspects of shale gas development terms resources and reserves have previously been used, and
is our ability to accurately forecast gas resources and shale gas de- continue to be used to represent various categories of mineral
velopment economics. The uncertainty of the problem begs for a and/or hydrocarbon deposits. In March 2007, SPE, the American
probabilistic solution. Association of Petroleum Geologists, the World Petroleum Coun-
The objective of our work was to develop the data sets, meth- cil, and the Society of Petroleum Evaluation Engineers jointly
odology, and tools to determine values of original gas in place published the PRMS to provide an international standard for clas-
(OGIP), technically recoverable resources (TRR), recovery factor sification of oil and gas reserves and resources (Fig. 1a). It is im-
(RF), and economic viability in highly uncertain and risky shale portant to remember that the broadest categories are also the least
gas reservoirs. Existing approaches for determining values of precise. As one moves toward the categories at the bottom of the
TRR, such as the use of decline curves or even volumetric analy- chart, the associated estimates of the amount of natural gas in
ses, may not be reliable early on because there may not be enough those categories become more and more uncertain. However,
production history for decline curves to work well or the uncer- TRR and ERR are not formally classified in the system.
tainty in the reservoir properties may be too large for volumetric
analyses to be useful.
To achieve our research objective, we developed a computer Energy Information Administration (EIA) Classification
program, Unconventional Gas Resource Assessment System System. According to the EIA, TRR are the subset of the total
(UGRAS). In the program, we integrated Monte Carlo technique resource base that is recoverable with existing technology. The
with an analytical reservoir simulator to estimate the original vol- term resources represents the total quantity of hydrocarbons
ume in place, predict production performance, and estimate the that are estimated, at a particular time, to be contained in known
fraction of TRR that are economically recoverable resources accumulations and accumulations that have yet to be discovered
(ERR) for a variety of economic situations. We applied UGRAS (prospective resources). ERR are those resources for which there
to dry gas wells in the Barnett shale and the Eagle Ford shale to are economic incentives for production. It is important to note
determine the probabilistic distribution of their resource potential that, at some time in the future, economically unrecoverable
and economic viability. On the basis of our assumptions, the resources may become recoverable, as soon as the technology to
Eagle Ford shale in the dry-gas portion of the play has more tech- produce them becomes less expensive or the characteristics of the
nically recoverable resources than the Barnett shale. However, the market are such that companies can ensure a fair return on their
Eagle Ford shale is currently not as profitable as the Barnett shale investment by extracting the resources. For our purposes, we con-
because of the higher drilling costs in the Eagle Ford dry-gas sidered TTR to be the resources that can be produced within a 25-
window. year time period.
We anticipate that the tools and methodologies developed in We rearranged categories of PRMS and present an overview
this work will be applicable to any shale gas reservoir that has suf- of how the estimates of TRR and ERR are broken down (Fig. 1b).
ficient data available. These tools should ultimately be able to Those commercial resources, including cumulative production
allow determination of technically and economically recoverable and reserves, are ERR. TRR are the subset of the total resource
resources from shale gas reservoirs globally. base that includes commercial resources, contingent resources,
and prospective resources. Estimated ultimate recovery is not a
resources category, but a term that refers to the quantities of pe-
Introduction troleum that are estimated to be potentially recoverable from an
Many gas shale plays are currently under development in the US accumulation, including those quantities that have already been
oil and gas industry. The use of horizontal drilling in conjunction produced. For the remainder of this paper, we consider only natural-
with hydraulic fracturing has greatly expanded the ability of pro- gas resources.
ducers to profitably produce natural gas from low-permeability
geologic formations, particularly shale formations. We have pre-
viously analyzed 15 basins in North America where shale gas Monte Carlo Probabilistic Approach. Shale gas plays are gen-
resources have been evaluated, and the results have been pub- erally characterized by low geologic risk and high commercial
lished (Dong et al. 2011). The total volume of original shale gas risk. Uncertainty exists in geologic and engineering data and, con-
in place for those 15 North American basins was estimated at sequently, in the results of calculations made with these data.
4,774 to 7,341 Tcf. It is clear that there are abundant volumes of Probabilistic approaches are required to provide an assessment of
natural gas in North America. The question we now need to an- uncertainty in resource estimates.
swer is what portion of the gas resource is technically and eco- Decline-curve analysis (DCA) is commonly used for future-
nomically recoverable. The objective of our work was to develop performance prediction and resource estimation when production
data are available. However, several studies (e.g., Ilk et al. 2008)
have pointed out that the Arps method provides optimistic solu-
Copyright V
C 2013 Society of Petroleum Engineers
tions for unconventional reservoirs because boundary-dominated
This paper (SPE 152066) was accepted for presentation at the SPE Hydraulic Fracturing flow (assumed in the Arps DCA model) is not reached within rea-
Technology Conference, The Woodlands, Texas, USA, 68 February 2012, and revised for
publication. Original manuscript received for review 9 February 2012. Revised manuscript
sonable times in these reservoirs. Some authors have applied
received for review 30 October 2012. Paper peer approved 3 December 2012. probabilistic approaches to DCA to quantify the uncertainty

January 2013 SPE Economics & Management 5

ID: jaganm Time: 15:58 I Path: S:/3B2/EM##/Vol00000/130004/APPFile/SA-EM##130004


EM152066 DOI: 10.2118/152066-PA Date: 1-February-13 Stage: Page: 6 Total Pages: 12

COMMERCIAL
PRODUCTION PRODUCTION

Sub-COMMERCIAL COMMERCIAL
ERR
RESERVES RESERVES

Technically Recoverable

Increasing Chance of Commerciality


DISCOVERED PIIP
Total Oil & Gas Resource Base (PIIP)

Increasing Chance of Commerciality


1P 2P 3P 1P 2P 3P

Total Oil & Gas Resource Base


PROVED PROBABLE POSSIBLE
PROVED PROBABLE POSSIBLE

Economical
CONTINGENT
CONTINGENT

Sub-
RESOURCES
RESOURCES
1C 2C 3C
1C 2C 3C

Undiscovered
PROSPECTIVE
UNRECOVERABLE RESOURCES
Low Best High
UNDISCOVERED

PROSPECTIVE
RESOURCES TRR

Unrecoverable
PIIP

Technically
Low Best High DISCOVERED

UNRECOVERABLE OGIP UNDISCOVERED


Range of Uncertainty Range of Uncertainty
(a) Resource Classification of PRMS (b) EIA definitions mapped to PRMS categories

Fig. 1Flow chart and generalized division of resource and reserves categories. (a) Resource classification of PRMS. (b) EIA defi-
nitions mapped to PRMS categories.

resulting from use of these imperfect methods (Jochen and Spivey are hurdles commonly required by investors in the oil and gas
1996; Cheng et al. 2010). industry. The author suggested that if you could not achieve pay-
Reservoir simulation coupled with stochastic methods (e.g., out in 5 years or less, the well would not be a wise investment.
Monte Carlo) has provided an excellent means to predict produc- We developed UGRAS to help us determine the values of
tion profiles for a wide variety of reservoir characteristics and pro- ERR. In the program, we integrated Monte Carlo simulation with
ducing conditions. The uncertainty is assessed by generating a an analytical reservoir simulator, PMTx 2.0, to estimate the origi-
large number of simulations, sampling from distributions of uncer- nal volume in place, predict production performance, and estimate
tain geologic, engineering, and other important parameters. This the fraction of TRR that are ERR for a variety of economic situa-
topic has been an object of study for some time in conventional tions. In the following sections, we present the workflow of
reservoirs (MacMillan et al. 1999; Sawyer et al. 1999; Nakayama UGRAS and apply UGRAS to assess the resource potential and
2000). However, few applications to unconventional reservoirs evaluate economic viability of gas in the Barnett shale and in the
can be found in the literature. Oudinot et al. (2005) coupled Monte dry-gas window in the Eagle Ford shale.
Carlo simulation with a fractured reservoir simulator, COMET3,
to assess the EUR in coalbed-methane reservoirs. Schepers et al.
(2009) successfully applied this Monte Carlo/COMET3 procedure Methodology
to forecast EUR for the Utica shale. Shale gas reservoirs are highly heterogeneous, and well productiv-
ity depends on reservoir properties as well as completion and
stimulation parameters. Even if finite-difference reservoir simula-
Analytical Model. Given the complex nature of hydraulic-frac-
tors are available, it can be time consuming to perform a large res-
ture growth, the extremely low permeability of the matrix rock in
ervoir-simulation study. In our study, we applied UGRAS to
many shale gas reservoirs, and the predominance of horizontal
generate gas-production profiles for a variety of reservoir, well,
completions, reservoir simulation is commonly the preferred
and hydraulic-fracture scenarios. Thousands of simulations can be
method to predict and evaluate well performance. Analytical solu-
automatically run to explore combinations of unknown reservoir
tions for fluid flow in naturally fractured reservoirs were published
and well parameters across their ranges of uncertainty. We use the
by Warren and Root (1963) and Kazemi (1969). Semianalytical
investment hurdles of IRR more than 20% and payout time less
solutions for hydraulically fractured horizontal wells in fractured
than 5 years, applied on an individual-well basis, to determine the
reservoirs have been published (Medeiros et al. 2008). PMTx
fraction of TRR that is ERR for a variety of economic situations.
(PMTx version 2.0 2010), with a number of modeling options,
In future analyses, we plan to conduct sensitivity studies on these
such as the transient dual-porosity reservoir model (Kazemi 1969),
values of IRR and payout. However, for this paper, we have made
is an analytical unconventional-gas-reservoir simulator designed
the assumption that if a well does not pay out in 5 years or less, it
to quickly and easily model single-well, single-phase gas produc-
is probably not worth drilling at this time. Other places to drill
tion on the basis of near-wellbore reservoir performance under
and spend capital should be more profitable.
specified well-completion scenarios. One of the important applica-
The workflow of our probabilistic reservoir model UGRAS is
tions of PMTx 2.0 is to estimate ultimate gas recovery for horizon-
outlined in Fig. 2. First, an input file is created and uncertain
tal wells with transverse fractures in a rectangular shale gas
parameters are assigned probability distributions. There is no li-
reservoir.
mitation to the number of parameters that can be varied. The dis-
tributions are typically normal, uniform, triangular, exponential,
Economics Determination. Almadani (2010) presented a meth- or log-normal. These distributions are sampled for volumetric
odology to determine the percentage of TRR that is economically analysis and flow simulation to determine OGIP, TRR, and RF.
recoverable from the Barnett shale as a function of gas price and Then, these steps are repeated many times to generate frequency
finding and development costs (F&DC). For ERR, he applied eco- and cumulative density plots for OGIP, TRR, and RF. Finally,
nomic criteria of minimum 20% internal rate of return (IRR) and economic analysis was run to calculate the production from wells
maximum 5-year payout to recover the initial investment, which that meet economic criteria (IRR more than 20% before federal

6 January 2013 SPE Economics & Management

ID: jaganm Time: 15:58 I Path: S:/3B2/EM##/Vol00000/130004/APPFile/SA-EM##130004


EM152066 DOI: 10.2118/152066-PA Date: 1-February-13 Stage: Page: 7 Total Pages: 12

Define simulation input file TABLE 1RESERVOIR MODEL FOR SHALE GAS PLAYS

Define probability density Specifications Descriptions


function for uncertain
reservoir properties Porosity Transient dual porosity, slab blocks
Fracture conductivity Infinite
Monte Carlo Sampling for Inner boundary Horizontal TransFracs
reservoir properties Outer boundary Rectangle
Lithology Shale
Pressure step Constant
Run reservoir simulation Permeability Isotropic
Volumetric Analysis
(PMTx2.0) Well location Centered

Probability Probability Probability tude of the contribution from the matrix to the fractures. A large
distribution of distribution distribution
value indicates that fluids flow easily between the two porous
OGIP of TRR of RF
media, whereas a small value indicates that flow between the
media is restricted. We could not find any literature reporting the
Economic viability analysis values of k and x for gas shales. The storativity ratio is usually in
(ERR/ERR vs gas price etc.) the range of 0.01 to 0.1; we assumed a value of 0.01 to be repre-
sentative of shales because of small PV of the fractures. The inter-
porosity flow coefficient for dual-porosity reservoirs is usually in
Fig. 2Flow chart of UGRAS. the range of 104 to 108 (Fekete Associates Inc. 2012). The range
between 106 and 108 is assumed to be representative of shales
because of the large contrast between the permeabilities of the
income tax, payout less than 5 years) over production from all fractures and the matrix. The outer boundary is defined as a closed
wells according to different F&DC. rectangle, and the well is centered in the drainage area. Table 1
summarizes the reservoir model used for shale-gas-reservoir
simulation.
Shale-Gas-Reservoir Model
Today, shale gas reservoirs are typically developed with horizon- /ct f
tal wells that are hydraulically fractured with multiple stages. As x . . . . . . . . . . . . . . . . . . . . . . . 1
/ct f /ct m
more knowledge is gained through microseismic monitoring of
these fracture treatments, it appears that they are more likely cre- rw2 km
ating a network of fractures. Thus, shale reservoirs may behave as k 4nn 2 For slab blocks; n 1: . . . . . . 2
L2 kf
a transient dual-porosity system, with the secondary porosity sys-
tem (matrix) contributing to the primary porosity system (system)
consisting of the created fracture network and possible existing Resource Distribution of Barnett Shale
natural fractures. Transient dual-porosity systems have been used
to model naturally fractured reservoirs (Kazemi 1969; Swaan The Barnett Shale produces primarily dry gas. In this work, we
1976). This model can also be used for modeling shale gas reser- have considered only gas production and have not included any
voirs where multistage-fracture completions have created the frac- wells that may be in the oil window. Vertical wells were first
ture network. In the transient dual-porosity model, there are two drilled in the Barnett shale in the early 1980s, but development of
transientsone moving through the fracture system and the other the Barnett shale play was not seriously considered until almost
moving through the matrix toward the interior of the matrix 2 decades later with the advent of horizontal drilling in 2003
blocks. (Fig. 3). As of December 2011, the Barnett shale play had more
The transient dual-porosity model is characterized by the stor- than 12,561 wells, including 9,449 horizontal wells and 3,112 ver-
ativity ratio and the interporosity flow coefficient. The storativity tical wells. More than 8,270 Bcf of gas has been produced, of
ratio, x, is the fraction of pore volume (PV) in the fractures com- which 75% is from horizontal wells.
pared with the total PV (Eq. 1). The interporosity flow coefficient,
k, is proportional to the ratio of permeabilities between the matrix Completion History of Horizontal Wells. Horizontal wellbores
and the fractures (Eq. 2), and it determines the timing and magni- are typically oriented northwest/southeast to take advantage of

2000 10000
Vertical Well Count
Horizontal Well Count
1500 7500
Vertical Well Production
Producing Wells
Production, Bcf

Horizontal Well Production


1000 5000

500 2500

0 0
1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010
Year

Fig. 3Production has rapidly increased in the Barnett shale by horizontal wells (HPDI 2011).

January 2013 SPE Economics & Management 7

ID: jaganm Time: 15:58 I Path: S:/3B2/EM##/Vol00000/130004/APPFile/SA-EM##130004


EM152066 DOI: 10.2118/152066-PA Date: 1-February-13 Stage: Page: 8 Total Pages: 12

(a) Cemented-liner, multi-stage fracturing method (b) Open-hole, multi-stage system (OHMS) completion
(Initial Barnett shale well completion) (Latest Barnett shale well completion)

Fig. 4Lower-damage, more-intensively-stimulated horizontal-well completions (Kuuskraa 2009). (a) Cemented-liner, multistage-
fracturing method; initial Barnett shale well completion. (b) OHMS completion; latest Barnett shale well completion.

natural productivity in the Barnett shale (Hale 2010). Lateral variously been reported to be 0.00007 to 0.005 md (Grieser et al.
lengths increased from approximately 3,000 ft in 2003 to as long 2008). Productive, organic-rich portions of the Barnett shale aver-
as 8,965 ft in 2009, with an average value of 4,000 ft (Powell age 25 to 43% water saturation (Bruner and Smosna 2011). The
2010). Typical fracture half-length of the Barnett shale is between organic matter in the shale was first reported to contain 60 scf/ton,
300 and 400 ft, with seven to nine fracture stages (Kennedy but this could be as high as 125 scf/ton (Montgomery et al. 2005).
2010). It is reported that typical well spacing in the Barnett shale is 60 to
Initially, cemented liners and multistage fracturing techniques 160 acres (Hayden and Pursell 2005). Table 2 summarizes the
were used in the Barnett shale (Fig. 4a). This type of completion reservoir parameters reported from literature for the Barnett shale.
involves cementing casing in the horizontal wellbore and by use In our modeling, we used a reservoir size of 4,800  1,000 ft (111
of plug-and-perf stimulation. The inherent costs of multiple acres/well), with a lateral length of 4,000 ft, fracture half-length
interventions with coiled tubing, perforating guns, and deploy- of 350 ft, and 10 stages. Table 3 shows the fixed reservoir param-
ment of fracturing equipment needed for each stage are extremely eters used for the Barnett shale single-well reservoir simulations.
high, not to mention inefficient consumption. Production by use
of this method can also be limited, because cementing the well-
bore closes many of the natural fractures and fissures that would Model Verification. We used the HPDI (2011) database as our
otherwise contribute to overall production (Lohoefer et al. 2010). source for production data. Since 2004, 1,492 horizontal wells in
Between 2004 and 2006, a new openhole, multistage system the Barnett shale have been completed and produced more than
(OHMS) completion technology (Fig. 4b) was run in Denton 60 months. Initially, uniform distributions were assigned for six
County, Texas (Lohoefer et al. 2006). This type of completion, on key uncertain parametersnet pay, initial pressure, permeability,
average, performed better than the cemented completion method porosity, water saturation, and gas content (Table 4)honoring
(Lohoefer et al. 2010). The major advantage of OHMS is that all their reported ranges from the literature documented in Table 2.
the fracture treatments can be performed in a single, continuous We did not consider possible correlation among these parameters.
pumping operation without the need for a drilling rig, saving The distributions for three of these uncertain parametersnet
costs. pay, permeability, and gas contentwere refined until a reasona-
ble match between simulated and actual 5-year cumulative pro-
duction was obtained (Fig. 5). We did not have to make large
Reservoir Parameters. Both thickness and reservoir pressure changes in these distributions to achieve the match. Table 4
increase in the southwest/northeast direction, implying a signifi- also summarizes the final distributions for the six uncertain
cant increase in potential production rate from southwest to north- parameters.
east. The thickness of the Barnett shale ranges from 100 to 600 ft The red curve in Fig. 5 shows the distribution of 5-year cumu-
(Hayden and Pursell 2005; Grieser et al. 2008). Pressure gradient lative production from the 1,492 horizontal wells. The blue curve
is in the overpressured category, typically 0.53 psi/ft (Lafollette is the distribution from 1,000 random realizations of 5-year cumu-
et al. 2012). The most common reservoir pressure used for Barnett lative production simulated by UGRAS with the parameters in
shale reservoir simulation is 3,000 to 5,000 psi (Chong et al. Table 3 and the final distributions in Table 4.
2010). The average reservoir temperature is 200 F (Transform As a further check of the model, we compared simulated and
Software & Services 2011). The average porosity in productive actual production-decline trends. We calculated the average well
portions of the Barnett shale ranges from 4 to 5% (Hayden and monthly gas production from the 1,492 wells over the first 60
Pursell 2005). Published values for average permeability have months (Fig. 6). From the probabilistic model described in Tables
3 and 4, we plotted well production curves corresponding to the

TABLE 2SUMMARY OF RESERVOIR PARAMETERS


REPORTED FROM LITERATURE FOR THE BARNETT SHALE TABLE 3FIXED PARAMETERS FOR THE BARNETT
SHALE SIMULATION
Parameter Range
Layer Data Value
Net pay (ft) 100600
Porosity (%) 45 Reservoir temperature ( F) 200
System permeability (md) 0.000070.005 Flowing bottomhole pressure (psia) 500
Water saturation (%) 2543 Reservoir length (ft) 4,800
Gas content (scf/ton) 60125 Reservoir width (ft) 1,000
Depth (ft) 6,5008,500 Fracture half-length (ft) 350
Reservoir pressure (psi) 3,0005,000 Horizontal-well length (ft) 4,000
Well spacing (acres) 60160 Number of fracture stages 10
Vitrinite reflectance (%) 0.61.6 k (dimensionless) 7  107
TOC* (%) 2.45.1 x (dimensionless) 0.01
*
TOC total organic carbon. Bulk density (g/cm3) 2.5

8 January 2013 SPE Economics & Management

ID: jaganm Time: 15:58 I Path: S:/3B2/EM##/Vol00000/130004/APPFile/SA-EM##130004


EM152066 DOI: 10.2118/152066-PA Date: 1-February-13 Stage: Page: 9 Total Pages: 12

TABLE 4INITIAL AND FINAL DISTRIBUTIONS FOR UNCERTAIN BARNETT SHALE PARAMETERS

Initial Uniform Distributions Final Distributions

Parameters Minimum Maximum Distribution Type l r Minimum Median Maximum

Net pay (ft) 100 600 Log-normal 200 50


Initial pressure (psi) 3,000 5,000 Uniform 3,000 5,000
System permeability (md) 0.00007 0.005 Log-normal 0.0005 0.0005
Water saturation (fraction) 0.25 0.43 Uniform 0.25 0.43
Porosity (fraction) 0.04 0.05 Uniform 0.04 0.05
Gas content (scf/ton) 60 125 Triangular 60 100 125
Loglogistic (c, b, a) log-logistic distribution with location parameter c, scale parameter b, and shape parameter a.
Lognorm (l, r) log-normal distribution with specified mean and standard deviation.

100
90
80
Percentile, %

70
60
50
40
30
20
10
0
0 1 2 3 4 5
5-year Cumulative Production, Bcf/well
Field Data Simulation Result

Fig. 5Probability distribution of cumulative-gas-production (5-year) match result for the Barnett shale.

mean, P10, P50, and P90 5-year cumulative production values 0:02329 ZT
(Fig. 6). The simulated production curves were run out to 25 years where Bgi
to observe the long-term production trends. Both the actual and p
simulated production profiles show high initial production rate In addition to free gas, shale can hold significant quantities of
with steep production decline in the first year and significant gas adsorbed on the surface of the organics in the shale formation.
decrease in the decline rate in subsequent years. Adsorbed gas can be the dominant in-place resource for shallow,
organic-rich shales. A Langmuir isotherm is established for the
prospective area of the basin by use of available data on TOC and
OGIP, TRR, and RF. Detailed geologic and reservoir data were thermal maturity to establish the Langmuir volume (VL) and pres-
assembled to establish the free gas and adsorbed gas in place for sure (PL).
the Barnett shale. Free gas becomes the dominant in-place Adsorbed gas in place (GIP) is then calculated by use of Eq. 4:
resource for deeper, higher-clastic-content shales. Eq. 3 is used to
calculate the free gas in place for shale gas reservoirs: Adsorbed GIP 43; 560Ahqc Gc 1  / . . . . . . . . . . . 4
43; 560Ah/1  Sw VL  p
Free GIP . . . . . . . . . . . . . . . . 3 where Gc .
Bgi PL p

1.E+05
Field Data From Barnett Shale
Simulated by UGRAS(P90)
Simulated by UGRAS(Mean)
Production, Mcf/Month

Simulated by UGRAS(P50)
Simulated by UGRAS(P10)

1.E+04

1.E+03
0 50 100 150 200 250 300
Month

Fig. 6Average Barnett shale gas production of 1,492 wells overlaid by the simulated TRR distribution.

January 2013 SPE Economics & Management 9

ID: jaganm Time: 15:58 I Path: S:/3B2/EM##/Vol00000/130004/APPFile/SA-EM##130004


EM152066 DOI: 10.2118/152066-PA Date: 1-February-13 Stage: Page: 10 Total Pages: 12

100% 100%
90% 90%
80% 80%
70% 70%
Percentile

60%

Percentile
60%
50% 50%
40%
40%
30%
30%
20%
OGIP Simulated by UGRAS 20%
10% TRR Simulated by UGRAS
Loglogistic (1.5,10.6,5.3) 10%
0% Lognorm(2.5,1.5,Shift(0.05))
5 50 0%
OGIP, Bcf 0.4 4
TRR, Bcf
Fig. 7Probabilistic distribution of OGIP/111 acres for the Bar-
nett shale. Fig. 8Probabilistic distribution of TRR/111 acres with a 25-
year life for the Barnett shale.
100%
90%
80% TABLE 5ECONOMIC PARAMETERS FOR BARNETT SHALE
70%
Operating Cost USD/Mcf
Percentile

60%
50% Working interest 100%
40% Royalty burden 25%
30% Severance taxes 7%
20% Gas shrinkage 6%
Simulated by UGRAS
10%
Lognorm(20,10,Shift(0.3))
0%
0 20 40 60 80 100 The simulation results yielded a TRR distribution with a P10
Recovery Factor, % value of 1.1 Bcf/111 acres, a P50 of 2.2 Bcf/111 acres, and a P90
of 4.5 Bcf/111 acres (Fig. 8). The distribution is quite wide, indi-
Fig. 9Probabilistic distribution of RF with a 25-year life for the cating significant uncertainty in forecasting Barnett shale gas pro-
Barnett shale. duction. RF of Barnett shale also follows a log-normal
The probability distributions for OGIP, TRR, and RF are gen- distribution (Fig. 9), with P10, P50, and P90 values of 10, 18, and
erated by reservoir parameters in Table 2 and finalized density 35%, respectively.
functions in Table 3. The P10, P50, and P90 values of OGIP are
8.4, 12.2, and 17.8 Bcf/111 acres, respectively (Fig. 7). Economic Evaluation. We next examined the economic impact
We chose a 25-year production history rather than 30 or 40 of different gas prices and F&DC on ERR in the Barnett shale. To
years. In the current economic environment, no one really cares do this, for each realization we determine the gas price required to
how much gas will be on the books 40 years from now. However, just meet the economic hurdles for particular F&DC. The eco-
in future work, we plan to vary the well life to determine whether nomic analysis is performed at the assumptions listed in Table 5.
it really makes a difference in terms of both TRR and ERR. Gas shrinkage results from the usage of a percentage of produced
gas for mechanical compression along the pipeline.
Ranking the realizations, we can determine the fraction of
1
TRR that is economically recoverable for a particular combina-
tion of gas price and F&DC. We then repeat this to determine the
ratio of ERR/TRR over a range of gas prices and F&DC (Fig. 10).
0.8 With typical F&DC of USD 3 million for Barnett shale wells and
a gas price of USD 4/Mcf, 20% of the Barnett shale gas TRR is
ERR/TRR, Bcf/Bcf

economically recoverable. If we increase the gas price, we will


0.6 increase the fraction of technically recoverable resource that is
economically recoverable.
With an estimated acreage of 3.2 million acres and assumed
0.4 well spacing of 111 acres, 29,000 wells could be drilled in the
Barnett shale. Thus, the resource potential for the entire Barnett
shale is estimated at 352 Tcf of OGIP (P50), 63 Tcf of TRR
0.2 (P50), and 12 Tcf of ERR (20% of TRR) at USD 4.0/Mcf gas
price and USD 3 million/well F&DC (Table 6).

0
1 10
Gas Price, $/Mcf TABLE 6RESOURCE POTENTIAL FOR THE BARNETT
F&DC = 1, MMS F&DC = 5, MMS SHALE PLAY
F&DC = 2, MMS F&DC = 6, MMS
F&DC = 3, MMS F&DC = 7, MMS Category P10 P50 P90
F&DC = 4, MMS OGIP (Tcf) 242 352 513
TRR (Tcf) 32 63 130
Fig. 10Ratio of ERR/TRR as a function of gas price and F&DC
ERR (Tcf) 6 12 26
for the Barnett shale.

10 January 2013 SPE Economics & Management

ID: jaganm Time: 15:58 I Path: S:/3B2/EM##/Vol00000/130004/APPFile/SA-EM##130004


EM152066 DOI: 10.2118/152066-PA Date: 1-February-13 Stage: Page: 11 Total Pages: 12

Fig. 11Eagle Ford extends across south Texas with updip oil, middip condensate, and downdip gas windows.

Resource Distribution in the Gas Window of the highest. In future work, we plan to evaluate the gas-condensate
Eagle Ford window and the oil window in the Eagle Ford. For now, all results
The Eagle Ford shale in south Texas is in its infancy in terms of in this paper were computed by use of gas-production data only
development compared with other shale plays in the US. The from the dry-gas portion of the Eagle Ford.
Eagle Ford shale is directly beneath the Austin chalk. Wells in the In the Eagle Ford shale, there were seven producing gas wells
Eagle Ford shale produce from depths between 4,000 and 14,000 in 2008 and more than 509 wells producing in 2011 (Fig. 12). As
ft. The Eagle Ford dips toward the Gulf of Mexico. It is consid- of December 2011, more than 434 Bcf of dry gas has been pro-
ered to be the source rock for the Austin chalk. The Eagle Ford is duced. Fig. 12 shows dry-gas production from the Eagle Ford shale
best known for producing variable amounts of dry gas, wet gas, is increasing annually. But because of the low gas price, the aver-
natural-gas liquids, condensate, and oil. In late 2008, the first few age daily dry-gas production reached only 854 MMcf/D in 2011.
exploration wells in the Eagle Ford were drilled in LaSalle The average lateral length of gas wells in the Eagle Ford was
County in the gas window of the play (Fig. 11). In this paper, we 5,600 ft in 2011 (Fig. 13). Unlike many other shale plays, the
have evaluated only the dry-gas portion of the Eagle Ford, which
is also the deepest portion of the play where drilling costs are the 8000

400 600
Well Count 6000
Laternal Length, ft

Gas
300 450
Production, Bcf

4000
Well Count

200 300
2000
100 150

0
0 0 2008 2009 2010 2011
2008 2009 2010 2011 Year
Year
Fig. 13The trend of average lateral length of Eagle Ford hori-
Fig. 12Annual dry-gas production in the Eagle Ford shale zontal wells over time (data provided by Unconventional
(HPDI 2011). Resources LLC, personal communication with G. Vonieff).

January 2013 SPE Economics & Management 11

ID: jaganm Time: 15:59 I Path: S:/3B2/EM##/Vol00000/130004/APPFile/SA-EM##130004


EM152066 DOI: 10.2118/152066-PA Date: 1-February-13 Stage: Page: 12 Total Pages: 12

TABLE 7RESERVOIR PARAMETERS OF THE EAGLE FORD SHALE*

Upper Eagle Ford Lower Eagle Ford

Range Mean Range Mean

Depth (ft) 5,50014,300 11,700 5,80014,400 11,800


Net pay (ft) 3236 100 8326 163
System permeability (md) 0.00010.0005 0.0003 0.00010.0007 0.0004
Water saturation (%) 1244 23 944 18
Porosity (%) 39 6 312 8
Gas content (scf/ton) 796 41 18118 82
Bulk density (g/cm3) 2.442.65 2.55 2.362.63 2.46
TOC (%) 0.34.0 1.9 0.75.4 3.6
*Personal communication. 2011. Houston: W.D. Von Gonten & Company.

Ford shale have been completed and produced for more than 12
TABLE 8THE FIXED PARAMETERS AND RESERVOIR months. We initially assigned uniform density functions for net
MODEL FOR EAGLE FORD SHALE SIMULATION pay, initial pressure, water saturation, porosity, permeability, and
gas content within their parameter ranges (Table 9). We did not
Layer Data Value consider possible correlation among these parameters. These six
Reservoir temperature ( F) 
247 density functions were refined until a reasonable match between
simulated and actual 1-year cumulative production was obtained.
Flowing bottomhole pressure (psia) 500
Table 9 also summarizes the final distributions for the six uncer-
Reservoir length (ft) 6,400
tain parameters.
Reservoir width (ft) 1,000 The red curve in Fig. 14 shows the cumulative probability dis-
Fracture half-length (ft) 350 tribution of 1-year cumulative production from the 152 horizontal
Horizontal-well length (ft) 5,600 wells. The blue curve is the cumulative probability distribution
Fracture stage 18 from 1,000 random realizations of 1-year cumulative production
k (dimensionless) 106 simulated by UGRAS with the parameters in Table 8 and final
x (dimensionless) 0.01 distributions in Table 9.
Bulk density (g/cm3) 2.51 As a further check of the model, we compared simulated and
actual production-decline trends. We calculated the average well
monthly gas production from the 152 wells over the first 12
Eagle Ford shale does not exhibit natural fracturing within the for- months (Fig. 15). From the probabilistic model described in
mation. The carbonate content of the Eagle Ford can be as high as Tables 7 and 8, we plotted well production curves corresponding
70%. The high carbonate content and consequently lower clay to the mean, P10, P50, and P90 1-year cumulative production val-
content make the Eagle Ford more brittle and easier to stimulate ues (Fig. 15). The simulated production curves were run out to 25
through hydraulic fracturing than other shales with less carbonate. years to observe the long-term production trends.
Typical fracture half-length of the Eagle Ford shale is 350 ft, with
8 to 10 fracture stages (Kennedy 2010).
Resource Assessment. Figs. 16 through 18 show the probability
distributions for OGIP, TRR, and RF generated by reservoir pa-
Reservoir Parameters. We obtained the ranges of reservoir pa- rameters in Table 7 and finalized density functions in Table 8.
rameters from 121 horizontal gas wells in the Eagle Ford shale The values of OGIP range from 7.5 (P10) to 25.3 (P90) Bcf/147
(Table 7). Table 8 lists the fixed parameters used for the Eagle acres. TRR for a 25-year recovery period range from 2.3 (P10) to
Ford shale simulation. The case was modeled on a well with 11 8.5 (P90) Bcf/147 acres. Eagle Ford RF ranges from 25 (P10) to
multistage hydraulic transverse fractures, with fracture half-length 40% (P90). Again, we chose 25 years to determine TRR as a rea-
of 350 ft and a total wellbore length of approximately 5,600 ft, sonable well life that is of interest to us now. In the future, we
producing natural gas for a period of 25 years. The assumed well will run sensitivity studies to determine whether the answers
spacing is 147 acres/well. change significantly for other values, such as 30 or 40 years.

Model Verification. We used the HPDI database as our source Economic Evaluation. The economic analysis for the Eagle
for production data. Since 2010, 152 horizontal wells in the Eagle Ford shale was performed at the assumptions in Table 10. The

TABLE 9INITIAL AND FINAL DISTRIBUTIONS FOR UNCERTAIN EAGLE FORD SHALE PARAMETERS

Initial Uniform Distribution Final Distribution

Parameters Minimum Maximum Distribution Type l r a b Shift

Net pay (ft) 3 326 Log-normal 130 50


Initial pressure (psi) 4,300 10,900 Log-normal 7,200 1,100
Water saturation (fraction) 0.09 0.44 Gamma 0.17 0.06 3.8 0.03 0.06
Porosity (fraction) 0.03 0.12 Inverse Gaussian 0.1 6.8 0.04
System permeability (md) 0.0001 0.0007 Log-normal 0.0004 0.0001
Gas content (scf/ton) 7 120 Gamma 49 19 7 7
Invgauss (l, k) inverse Gaussian distribution with mean l and shape parameter k.
Pearson 5 (a, b) Pearson-type V (or inverse gamma) distribution with shape parameter a and scale parameter b.

12 January 2013 SPE Economics & Management

ID: jaganm Time: 15:59 I Path: S:/3B2/EM##/Vol00000/130004/APPFile/SA-EM##130004


EM152066 DOI: 10.2118/152066-PA Date: 1-February-13 Stage: Page: 13 Total Pages: 12

100
90
80

Percentile, %
70
60
50
40
30
20
10
0
0 0.3 0.6 0.9 1.2 1.5 1.8
1-year Cumulative Production, Bcf/well
Field Data Simulation Result

Fig. 14Distribution of cumulative production (1-year) match result.

1.E+06 100%
Field Data From Eagle Ford Shale 90%
Simulated by UGRAS(P90)
Simulated by UGRAS(Mean) 80%
Production, Mcf/Month

Simulated by UGRAS(P50) 70%


1.E+05 Simulated by UGRAS(P10)

Percentile
60%
50%
40%
30%
1.E+04
20% Simulated by UGRAS
10% Loglogistic(1.6,12,3)
0%
3 30
1.E+03 OGIP, Bcf
0 50 100 150 200 250 300
Month Fig. 16Probabilistic distribution of OGIP/147 acres for the
Eagle Ford shale gas window.
Fig. 15Average Eagle Ford dry-gas production of 152 wells
overlaid by the simulated TRR.

100% 100%
90% 90%
80% 80%
70% 70%
Percentile
Percentile

60% 60%
50% 50%
40% 40%
30% 30%
20% 20%
Simulated by GURAS Simulated by UGRAS
10% 10% Pearson5(43,1651,Shift(7.6))
Invgauss(5,17)
0% 0%
1 10 20 30 40 50 60 70
TRR, Bcf Recovery Factor, %

Fig. 17Probabilistic distribution of TRR/147 acres with a 25- Fig. 18Probabilistic distribution of RF with a 25-year life for
year life for the Eagle Ford shale gas window. the Eagle Ford shale gas window.

ERR/TRR ratio in the dry-gas portion of the Eagle Ford shale for shale is economically recoverable. It is clear that either higher gas
different gas prices and F&DC is shown in Fig. 19. With typical prices or better technology to increase average well recovery or
F&DC of USD 9 million and gas price of USD 4/Mcf, only a decrease well costs is required to economically produce the large
small fraction of TRR in the dry-gas portion of the Eagle Ford amount of natural gas in the dry-gas portion of the Eagle Ford.
Because of the economic environment, it can be observed that
virtually all current drilling (other than to hold acreage) is occur-
ring in the gas-condensate window or the oil window in the Eagle
TABLE 10ECONOMIC PARAMETERS FOR EAGLE
Ford (Fig. 11). The addition of liquid production to the natural-
FORD SHALE
gas production significantly improves the average product price
and the economic profitability of these Eagle Ford wells. In addi-
Operating Cost USD 1.3/Mcf
tion, the gas-condensate and oil portions of the Eagle Ford are
Working interest 100% shallower than the dry-gas portion; thus, drilling costs are lower
Royalty burden 25% in the shallower portion of the play.
In the dry-gas window, the estimated productive acreage is
Severance taxes 7%
estimated to be 3 million acres. If we assume an average well
Gas shrinkage 6%
spacing of 147 acres, 20,000 wells could be drilled in the dry-gas

January 2013 SPE Economics & Management 13

ID: jaganm Time: 15:59 I Path: S:/3B2/EM##/Vol00000/130004/APPFile/SA-EM##130004


EM152066 DOI: 10.2118/152066-PA Date: 1-February-13 Stage: Page: 14 Total Pages: 12

1
TABLE 11RESOURCE POTENTIAL FOR DRY GAS IN THE
EAGLE FORD SHALE PLAY
0.8
Category P10 P50 P90
ERR/TRR, Bcf/Bcf

OGIP (Tcf) 153 278 516


0.6 TRR (Tcf) 47 90 173

0.4
first because the reservoir pressures are greater. Second, it appears
that operators generally drill longer laterals with more stages and
use larger fluid and proppant volumes per stage in the hydraulic-
0.2 fracturing treatments pumped in Eagle Ford dry-gas wells (thus,
our assumed well spacing of 147 acres/well for the Eagle Ford
gas window vs. 111 acres/well for the Barnett shale). These two
0 reasons are more likely to have a greater impact in the TRR dif-
1 10 ference than differences in petrophysical properties between
Gas Price, $/Mcf basins (Baihly et al. 2010). Higher reservoir pressure in the Eagle
F&DC = 6, MMS F&DC = 10, MMS Ford gas window is the main reason that Eagle Ford has higher
F&DC = 7, MMS F&DC = 11, MMS
RF than Barnett shale.
Table 12 lists opportunities for increasing ERR in both the
F&DC = 8, MMS F&DC = 12, MMS Barnett shale and Eagle Ford shale gas window by increasing gas
F&DC = 9, MMS prices and/or decreasing F&DC. For instance, by decreasing
F&DC cost to USD 2 million/well, 50% of TRR could be eco-
Fig. 19Ratio of ERR/TRR as a function of gas price and F&DC nomically recoverable from the Barnett Shale at a gas price of
for the Eagle Ford shale. USD 4/Mcf. If gas price increases to USD 6/Mcf, 25% of TRR
could be recovered economically in the Eagle Ford gas window at
portion of the Eagle Ford shale. Thus, the resource potential for an F&DC of USD 9 million/well.
the entire Eagle Ford dry-gas window is 278 Tcf of OGIP (P50) The technology and tools described in this paper can be useful
and 90 Tcf of TRR (P50) (Table 11). The value of ERR will be a in assessing TRR and ERR in shale gas plays. However, it is im-
function of the average gas price in the future. Currently, the natu- portant to acknowledge the assumptions and uncertainties inher-
ral-gas price is approximately USD 4/Mcf or less, which suggests ent in the results presented in this paper. First, we assumed a 25-
that many of these wells are not economic. However, the industry year well life for calculation of TRR and economic hurdles of
is working to increase natural-gas demand, which should increase IRR of more than 20% and payout time of less than 5 years for
the natural-gas price. As such, much of the TRR will be recovered calculation of ERR. Although we believe these are reasonable cri-
whenever the natural-gas price increases to a point at which more teria and many operators and investors in shale gas plays use simi-
drilling will occur. lar values, we acknowledge that other operators may use different
criteria and, thus, may obtain different results.
Discussion Our resource assessments are high-level assessments.
Although we estimate resources for entire plays, we do not model
The portion of the Eagle Ford shale in the dry-gas window has reservoir and well properties on a well-by-well basis. Instead, we
more TRR per well on average than the wells in the Barnett shale, model each play as a whole, by use of probability distributions
that encompass the variability in reservoir properties across the
TABLE 12OPPORTUNITIES FOR INCREASING field as well as the uncertainty in these properties. For example,
THE ERR/TRR RATIO the OGIP varies from county to county because of differences in
net thickness and other properties across the field. The distribution
Barnett Shale Eagle Ford Gas Window of net thickness we used in the Barnett study covered the greater
net thickness in Tarrant County and the lower net thickness in the
ERR/ F&DC (USD Gas Price F&DC (USD Gas Price southwestern Barnett shale. Another limitation of our high-level
TRR 106/well) (USD/Mcf) 106/well) (USD/Mcf) assessments is related to vertical variability in properties. We did
not consider vertical variations in properties, such as fracturabil-
75% 1 3.1 6 6.5 ity, throughout the zones evaluated. In some areas, the net thick-
2 5.1 7 7.2 ness of the shale gas plays is so great that the entire pay zone
3 7.1 8 8.3 cannot be completed and produced. However, we used the same
4 9.0 9 9.0 distributions of net pay for the OGIP calculation and TRR predic-
5 10.5 10 10.0 tion for the Barnett and Eagle Ford shales in this study.
6 12.0 11 10.6 We also acknowledge the uncertainty in the production fore-
casts generated by the probabilistic analytical simulator. The input
50% 1 2.8 6 5.2
parameters used to generate production forecasts for both the Bar-
2 4.0 7 6.0
nett and the Eagle Ford gas window were obtained from the litera-
3 5.1 8 7.0 ture and well data. Operators and reserves evaluators reviewed
4 7.0 9 7.2 the parameter values and forecasts in these plays to verify their
5 8.0 10 8.0 reasonableness. The probabilistic forecasts for the Barnett shale
6 9.8 11 9.0 were calibrated against actual 5-year cumulative production data,
25% 1 2.1 6 4.2 although this of course does not guarantee the accuracy of 25-year
2 3.1 7 5.0 forecasts. Few performance data exist for the Eagle Ford shale.
3 4.1 8 5.5 Even though we calibrated the Eagle Ford dry-gas forecasts
4 5.0 9 6.0 against actual production data, there is uncertainty in these
forecasts.
5 6.0 10 6.2
The ERR/TRR ratio was calculated on an individual-well ba-
6 7.0 11 7.0
sis. That is, ERR/TRR is the TRR from wells that individually

14 January 2013 SPE Economics & Management

ID: jaganm Time: 15:59 I Path: S:/3B2/EM##/Vol00000/130004/APPFile/SA-EM##130004


EM152066 DOI: 10.2118/152066-PA Date: 1-February-13 Stage: Page: 15 Total Pages: 12

meet the economic hurdles at a particular gas price and particular References
F&DC divided by the TRR from all wells. This does not account Almadani, H.S. 2010. A Methodology to Determine Both the Technically
for the practice of budgeting and drilling packages of wells in Recoverable Resource and the Economically Recoverable Resource in
statistical shale gas plays. Because there is a lot of variability in an Unconventional Gas Play. MS Thesis, Texas A&M University, Col-
individual shale-gas well performance, a package of wells may lege Station, Texas (August 2010).
meet the economic hurdles overall, whereas some wells will indi- Baihly, J.D., Altman, R.M., Malpani, R., et al. 2010. Shale Gas Production
vidually meet the economic hurdles and some will not. Once Decline Trend Comparison Over Time and Basins. Paper SPE 135555-
drilled, wells will be allowed to continue producing to a net-cash- MS presented at the SPE Annual Technical Conference and Exhibi-
flow economic limit even if they do not meet the economic hur- tion, Florence, Italy, 1922 September. http://dx.doi.org/10.2118/
dles previously specified. Thus, actual ERR/TRR ratios for the 135555-MS.
Barnett and Eagle Ford gas window are potentially greater than Bruner, K. R., and Smosna, R. 2011. A Comparative Study of the Missis-
the ERR/TRR ratios presented in this paper. sippian Barnett Shale, Fort Worth Basin, and Devonian Marcellus
Finally, although a probabilistic reservoir model was used in Shale, Appalachian Basin. US Department of Energy and National
this work, the ERR/TRR ratios presented in this paper are deter- Energy Technology Laboratory, Washington, D.C. (April 2011).
ministic and do not capture the uncertainties described in the pre- Cheng, Y., Wang, Y., McVay, D.A., et al. 2010. Practical Application of a
ceding paragraphs. We plan to address these limitations in future Probabilistic Approach to Estimate Reserves Using Production
work. Decline Data. SPE Econ & Mgmt 2 (1): 1931. http://dx.doi.org/
10.2118/95974-PA.
Conclusions Chong, K.K., Grieser, W.V., Passman, A., et al. 2010. A Completions
In this study, we have evaluated resource potential and economic Guide Book to Shale-Play Development: A Review of Successful
viability in the Barnett shale and the dry-gas window in the Eagle Approaches toward Shale-Play Stimulation in the Last Two Decades.
Ford shale by use of a probabilistic, analytical reservoir model. Paper SPE 133874-MS presented at Canadian Unconventional Resour-
We conclude the following. ces and International Petroleum Conference, Calgary, Alberta, Canada,
1. The median (P50) resource potential of the Barnett shale is 1921 October. http://dx.doi.org/10.2118/133874-MS.
estimated at 352 Tcf of OGIP and 63 Tcf of TRR. The value Dong, Z., Holditch, S. A., McVay, D.A., et al. 2011. Global Unconven-
for ERR is a function of F&DC and gas prices. tional Gas Resource Assessments. Paper SPE 148365-MS presented at
2. The median (P50) resource potential of the gas window in the Canadian Unconventional Resources Conference, Alberta, Canada,
Eagle Ford shale is 278 Tcf of OGIP and 90 Tcf of TRR. The 1517 November. http://dx.doi.org/10.2118/148365-MS.
value of ERR is a function of F&DC and gas prices. Fekete Associates, Inc. 2012. Interporosity Flow Coefficient (k), http://fekete.
3. The natural-gas window in the Eagle Ford shale has more tech- com/software/welltest/media/webhelp/Interporosity_Flow_Coefficient.
nically recoverable resources than the Barnett shale. However, htm.
the natural-gas window in the Eagle Ford shale will not be Grieser, W.V., Shelley, R.F., Johnson, B.J., et al. 2008. Data Analysis of
fully developed until drilling costs are reduced and/or natural- Barnett Shale Completions. SPE J. 13 (3): 366374. http://dx.doi.org/
gas prices increase from the current USD 4/Mcf range. 10.2118/100674-PA.
Hale, B.W. 2010. Barnett Shale: A Resource Play Locally Random and
Regionally Complex. Paper SPE 138987-MS presented at SPE Eastern
Regional Meeting, Morgantown, West Virginia, 1214 October. http://
Nomenclature
dx.doi.org/10.2118/138987-MS.
A area, acres Hayden, J. and Pursell, D. 2005. The Barnett Shale: Visitors Guide to the
Bgi gas formation volumetric factor, cf/scf Hottest Gas Play in the US. Report, Pickering Energy Partners, Inc.,
ct total compressibility, 1/psi Houston, Texas (October 2005).
d fracture spacing, ft
HPDI, version 6.0.1.2. 2011. Austin, Texas: DrillingInfo, Inc.
Gc initial gas content, scf/lbm
Ilk, D., Rushing, J.A., Perego, A.D., et al. 2008. Exponential vs. Hyperbolic
H net pay, ft
Decline in Tight Gas Sands: Understanding the Origin and Implications
kf fracture permeability, md
for Reserve Estimates Using Arps Decline Curves. Paper SPE 116731-
km matrix permeability, md
MS presented at SPE Annual Technical Conference and Exhibition, Den-
L characteristic length of a matrix block
ver, Colorado, 2124 September. http://dx.doi.org/10.2118/116731-MS.
n number of flow dimensions, dimensionless
Jochen, V.A. and Spivey, J.P. 1996. Probabilistic Reserves Estimation
P10, P50, P90 values for which the probability is 10, 50, or
Using Decline Curve Analysis with the Bootstrap Method. Paper SPE
90% that the value will not be exceeded, indi-
36633-MS presented at the SPE Annual Technical Conference and Ex-
cated by the 10th, 50th, or 90th percentile on a
hibition, Denver, Colorado, 69 November. http://dx.doi/org/10.2118/
cumulative probability plot
36633-MS.
rw wellbore radius
Kazemi, H. 1969. Pressure Transient Analysis of Naturally Fractured Res-
Sw water saturation
ervoirs with Uniform Fracture Distribution. SPE J. 9 (4): 451462.
k interporosity flow coefficient, dimensionless
http://dx.doi.org/10.2118/2156-A.
l mean
qc bulk density, lbm/cf Kennedy, R.B. 2010. Shale Gas Challenges/Technologies Over the Asset
r standard deviation Life Cycle. Oral presentation given at the US-China Oil and Gas
/ formation porosity Industry Forum, Fort Worth, Texas, 1 September.
x storativity ratio, dimensionless Kuuskraa, V.A. 2009. Worldwide Gas Shales and Unconventional Gas: A
Status Report. Oral presentation given at the United Nations Climate
Change Conference, Copenhagen, Denmark, 9 December.
Lafollette, R., Holcomb, W. D., and Aragon, J. 2012. Practical Data Mining:
Acknowledgments Analysis of Barnett Shale Production Results with Emphasis on Well
We thank the Research Partnership to Secure Energy for America Completion and Fracture Stimulation. Paper SPE 152531-MS presented
and the Crisman Institute in the Department of Petroleum Engi- at SPE Hydraulic Fracturing Technology Conference, The Woodlands,
neering at Texas A&M University for supporting this research. Texas, 68 February. http://dx.doi.org/10.2118/152531-MS.
We would like to acknowledge William D. Von Gonten with Lohoefer, D.S., Seale, R.A., and Athans, J. 2006. New Barnett Shale Hori-
W.D. Von Gonten & Company for providing relevant data sets zontal Completion Lowers Cost and Improves Efficiency. Paper SPE
and valuable feedback to calibrate our research findings and anal- 103046-MS presented at the SPE Annual Technical Conference and
ysis. We thank John P. Spivey with Phoenix Reservoir Engineer- Exhibition, San Antonio, Texas, 2427 September. http://dx.doi.org/
ing for providing PMTx 2.0 for this research. 10.2118/103046-MS.

January 2013 SPE Economics & Management 15

ID: jaganm Time: 15:59 I Path: S:/3B2/EM##/Vol00000/130004/APPFile/SA-EM##130004


EM152066 DOI: 10.2118/152066-PA Date: 1-February-13 Stage: Page: 16 Total Pages: 12

Lohoefer, D.S., Snyder, D., Seale, R.A., et al. 2010. Comparative Study of Transform Software & Services. 2011. A Comparative Study of North
Cemented Versus Uncemented Multi-Stage Fractured Wells in the Americas Largest Shale Gas Reservoirs. http://www.transformsw.
Barnett Shale. Paper SPE 135386-MS presented at the SPE Annual com/papers-and-presentations/studies.html#Barnett.
Technical Conference and Exhibition, Florence, Italy, 1922 Septem- Warren, J.E., and Root, P.J. 1963. The Behavior of Naturally Fractured
ber. http://dx.doi.org/10.2118/135386-MS. Reservoirs. SPE J. 3 (3): 245255. http://dx.doi.org/10.2118/426-
MacMillan, D.J., Pletcher, J.L., and Bourgeois, S.A. 1999. Practical Tools PA.
to Assist History Matching. Paper SPE 51888-MS presented at the
SPE Reservoir Simulation Symposium, Houston, Texas, 1417 Febru-
Zhenzhen Dong works as a reservoir engineer with Schlum-
ary. http://dx.doi.org/10.2118/51888-MS. berger Consulting Services. She holds a PhD degree in petro-
Medeiros, F., Ozkan, E., and Kazemi, H. 2008. Productivity and Drainage leum engineering from Texas A&M University; an MS degree in
Area of Fractured Horizontal Wells in Tight Gas Reservoirs. SPE Res petroleum engineering from Research Institute of Petroleum
Eval & Eng 11 (5): 902911. http://dx.doi.org/10.2118/108110-PA. Exploration and Development, PetroChina; and a BS degree
Montgomery, S. L., Jarvie, D. M., Bowker, K. A., et al. 2005. Mississip- in mathematics from Northeast Petroleum University, China.
pian Barnett Shale, Fort Worth Basin, North-Central Texas: Gas-Shale Dong has focused her research in areas involving unconven-
Play with Multi-Trillion Cubic Foot Potential. AAPG Bull. 89 (2): tional gas reservoirs, reserves and resource assessment, well
155175. http://dx.doi.org/10.1306/09170404042. testing, and performance analysis.
Nakayama, K. 2000. Estimation of Reservoir Properties by Monte Carlo Stephen A. Holditch is the Director of the Texas A&M Energy
Simulation. Paper SPE 59408-MS presented at the SPE Asia Pacific Institute, the Director of the Crisman Institute for Petroleum
Conference on Integrated Modelling for Asset Management, Yoko- Research, and a professor of petroleum engineering. He was
hama, Japan, 2526 April. http://dx.doi.org/10.2118/59408-MS. head of the Harold Vance Department of Petroleum Engineer-
Oudinot, A.Y., Koperna, G.J., and Reeves, S.R. 2005. Development of a ing from January 2004 until January 2012. Holditch was SPE
President in 2002 and SPE Vice PresidentFinance and a mem-
Probabilistic Forecasting and History Matching Model for Coalbed
ber of the Board of Directors 19982003. In addition, he served
Methane Reservoirs. Oral presentation given at the International as a trustee for the American Institute of Mining, Metallurgical,
Coalbed Methane Symposium, Tuscaloosa, Alabama, 5 May. and Petroleum Engineers (AIME) during 199798. Holditch has
PMTx, version 2.0. 2010. College Station, Texas: Phoenix Reservoir Soft- received numerous awards in recognition of his technical
ware, LLC. achievements and leadership. In 1995, he was elected to the
Powell, G. 2010. Shale Energy: Developing the Barnett: Lateral Lengths National Academy of Engineering and in 1997 to the Russian
Increasing in Barnett Shale. World Oil 231 (8). Academy of Natural Sciences. In 1998, Holditch was elected
Sawyer, W.K., Zuber, M.D., and Williamson, J.R. 1999. A Simulation- to the Petroleum Engineering Academy of Distinguished Grad-
Based Spreadsheet Program for History Matching and Forecasting uates. He was elected an SPE and AIME Honorary Member in
2006. In 2010, Holditch was honored as an Outstanding Gradu-
Shale Gas Production. Paper SPE 57439-MS presented at the SPE
ate of the College of Engineering at Texas A&M University. He
Eastern Regional Conference and Exhibition, Charleston, West Vir- currently serves as Editor-in-Chief of SPEs peer-reviewed tech-
ginia, 2122 October. http://dx.doi.org/10.2118/57439-MS. nical journals.
Schepers, K.C., Gonzalez, R.J., Koperna, G.J., et al. 2009. Reservoir Mod-
eling in Support of Shale Gas Exploration. Paper SPE 123057-MS pre- Duane McVay is the Rob L. Adams 40 Professor in the Depart-
ment of Petroleum Engineering at Texas A&M University in Col-
sented at the Latin American and Caribbean Petroleum Engineering
lege Station, Texas. His research interests include reservoir
Conference, Cartagena de Indias, Colombia, 31 May3 June. http:// simulation, uncertainty quantification, and unconventional
dx.doi.org/10.2118/123057-MS. reservoirs. Previously, McVay spent 16 years with S.A. Holditch
Swaan, A.D. 1976. Analytic Solutions for Determining Naturally Frac- & Associates, a petroleum engineering consulting firm. He is a
tured Reservoir Properties by Well Testing. SPE J. 16 (3): 117122. Distinguished Member of SPE and earned BS, MS, and PhD
http://dx.doi.org/10.2118/5346-PA. degrees in petroleum engineering from Texas A&M University.

16 January 2013 SPE Economics & Management

ID: jaganm Time: 15:59 I Path: S:/3B2/EM##/Vol00000/130004/APPFile/SA-EM##130004

Anda mungkin juga menyukai