Professional Documents
Culture Documents
This paper was prepared for presentation at the 2009 SPE Production and Operations Symposium held in Oklahoma City, Oklahoma, USA, 4–8 April 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
Developing a predictive reservoir model involves determination or estimation of key reservoir components which
can vary through the rock volume. Sophisticated 3D grid models usually require significant input data and are
built for conventional reservoirs producing in Darcy flow.
Production from Barnett shale is not conventional. Shale rock gas flow involves a complex mixture of free and
adsorbed storage and production mechanisms. Free gas can be stored in the micro-porosity, natural fractures, or
thin lamination; existing or created during hydraulic fracturing. Adsorbed gas is contained in the organic material
randomly distributed in the bulk rock.
Horizontal multi stage fractured wellbores adds another level of complexity. Massive hydraulic fracturing of
horizontal shale has shown complex fracture networks are created along the wellbore. The mapped data suggest
multiple fracture planes are created during injection. These fracture planes can be irregular in length and are not
always symmetrical. Conventional reservoir models cannot handle this level of complexity.
A new 3-D, 4-Phase non-isothermal, multi-well Black Oil and “Pseudo-compositional” simulator has been
developed which allows placement of multiple transverse fractures along the horizontal. Its ability to model
horizontal multi wing transverse fractures and account for all three reservoir phases including injected fluid makes
this model more predictive of production.
In this paper we use mapped fracture dimensions of horizontal N. TX. Barnett wells to build the reservoir model.
Comparisons of model production to real production are made to demonstrate model predictive ability.
Introduction
Horizontal drilling and completion of the N. TX. Barnett began in 1991 and continues to the present day with over
6000 horizontal wellbores on production. Numerous well construction types and completion strategies have been
investigated including;
cemented and uncemented liners
cemented production casing
production casing with mechanical/swell packers and frac ports
cemented and uncemented casing using jet tool perforating and fracturing1
The most common completion is the cased, cemented production string utilizing the pump down perf and plug
method of multi stage completion.2
Experimentation with the above design considerations has been ongoing since 1991 to the present day. Table #1
lists the ranges and averages of these design options.
Range Average
Table #1. N. TX. Barnett Horizontal range and average completion trends.
The production outcome has been as varied as the completions. Figure #1 shows the first 12 month gas
production for 6330 horizontal Barnett completions. The 12 month cumulative gas production ranges from 0.1
BCF to 1.8 BCF with the median of 0.25 BCF.
0.9
0.8
0.7
0.6
% of wells
0.5
0.4
0.3
0.2
0.1
0
0 200000 400000 600000 800000 1000000 1200000 1400000 1600000 1800000
12 month gas mcf
Figure #1 Cumulative frequency plot of first year gas production from 6330 horizontal Barnett shale wells.
It is important to realize this spread in production outcome is the result of not only completion design or
procedure, but also involves many known and un-known factors such as the reservoir size and quality. Reservoir
parameters such as permeability and porosity distribution, reservoir pressure, saturations and relative
permeability effects vary over the +5000 square miles comprising the N. TX. Barnett. The question then
becomes, even if you had a reasonable estimate of reservoir size and quality, along with known completion
methods and mapped fracture dimensions; would you be able to model and predict a production outcome? This
is the purpose of this paper. It is not a technical discussion of reservoir modeling in fractured shale. Rather it is
an engineering approach to reservoir modeling. Understanding that detailed reservoir properties for each
SPE 120271 3
wellbore is not available, published or accepted field average values had to be used. In addition the impact on
production from multiple staged hydraulic fracture stimulation is viewed as an important factor. A model capable
of handling complex fracture geometries intersecting a horizontal wellbore is required. The model must be easy
and intuitive to use and provide reasonable production prediction from the wellbore/fracture geometry in an
unconventional reservoir environment.
The reservoir model chosen was selected because it fulfilled the following requirements.
Horizontal wellbores with multiple transverse fractures.
Accounts for three-phase, four component system (gas, oil, water, and injected fracturing fluid).
Intermittent injection and production flow periods.
Dual porosity and both free and adsorbed gas capable.
Asymmetric fracture wings with adjustable length, width, height, and conductivity.
The model was initially built to model the effect of conformance fluid injection into producing formations.3
The reservoir simulator is linked to a commercially available numerical wellbore simulator. The simulator is used
to calculate wellbore temperature and pressure profile during the injection of fluids, thus accounting for the cool
down of the formation during a sustained injection of fluid.
The simulator is built in with a comprehensive PVT module for multiphase simulation, especially in near-critical
fluid environment to handle black-oil reservoirs, volatile-oil reservoirs, lean- and rich-gas condensate reservoirs,
and dry gas reservoirs. A relative-permeability module provides the capability for effective multiphase simulation,
including correlation and capability to input capillary pressure data. The program provides visualization 2-D areal
and cross-sectional fluid front, 3-D full of sectional views, and can handle vertical, horizontal, and multilateral
wells.4
Like most engineering projects the input data required to be 100% technically correct is never truly available. Our
data set is a small portion of early Barnett horizontal development obtained from various customers and service
vendors. It includes actual fracture initiation measured depths, volume, rate and pounds of prop pumped. Net
zone height and TVD were given. Microseism summary reports for each case listed individual fracture half
lengths. In all cases, daily production rate of oil, water, gas, and tubing flowing pressure was provided for at least
one year. Production after that was obtained from public domain. Open or cased hole logs were not available and
reservoir information had to be assumed. Lease names, well location and operator information is not included to
maintain the confidentially of the specific information.
Individual reservoir data for each case was not available. We used published average values for the Barnett in
the Newark East.5,6,7 Basic reservoir parameters are listed in table #2.
In the three case studies the initial input values are listed in table#2. Yellow highlighted inputs were adjusted to
improve the production match. The range of adjustments is provided. The fracture half lengths (Xf) were given
as the average extent of a network spread of microseism events.
Because of the limited amount of reservoir input available, most input screen selections were selected as default
or basic correlations. For example relative permeability data was not available, so default exponent values for
gas and water (gas=2, water=2) were used. Injection and production constraints were selected to reflect the
actual conditions during fracture injection and flow back stages.
The option to use dual porosity and desorption flow was not active. Adjusting a diffusion coefficient, storativity
(), or transmissivity ratio () was beyond the scope of this study. However it may explain the percent error in the
early time production match.
Case Descriptions
The following briefly describes the three cases used in this simulation test. In all cases the casing was cemented.
Slick water was used as the fracturing fluid in all cases.
Case #1 was a four stage frac with a single perforated interval for each stage. The stages were separated by
drillable composite plugs. For this case a single fracture was assumed at each perf site. There were no flow
periods between stages.
Case #2 was a two stage treatment with three perforated intervals in the first stage and two perforated intervals in
the second. Individual fractures were assumed at each perforation site. No flow period between stages. Stages
were separated by a drillable composite plug.
Case #3 was a twenty stage treatment using a hydra-jet tool to create the perforations and initiate each fracture
described in SPE91435.8 No plugs or packers were used between stages.
Basic summary of the three cases listed in table #3.
The map views of the three cases are shown with the reported fracture half lengths at positions along the lateral
length in figures 2,3,4.
Case#1 figure shows four individual fracture wings (FW) starting from the toe to the heel FW1, FW2, FW3, FW4.
FW3 is the longest at about 1900 ft. while the rest are between 1200-1300 ft. Spacing between perforations was
600-700 ft.
Figures 5-10 show model production verses actual production as well as input values adjusted to achieve match.
5000
4500
4000
3500
3000
mcf/d
gas Act
2500
gas model
2000
1500
1000
500
0
0 200 400 600 800 1000 1200 1400 1600 1800 2000
days
Case #1 actual verses model production plot of gas in mcf/d shows reasonable early time match up to 600 days.
Then the model tends to over predict production. The sudden decline at 1100 days was due to a line pressure
increase.
The parameters changed to obtain the match were permeability and fracture conductivity.
md-ft VS Xf Case #1
35
30
Adjusted Values
25
Kg=0.0004 md
H=400 ft.
(Kg)(h) = 0.16 md-ft
20
(w)(kf) = 33 to 2 md-ft
md-ft
10
0
0 145 290 435 580 725 870 1015 1160 1305 1450 1595 1740 1900
Xf
5000
4500
4000
3500
3000
mcf/d
gas act
2500
gas model
2000
1500
1000
500
0
0 500 1000 1500 2000 2500
days
Figure #7 Production plot #2 Case #2 Actual verses model gas production rate.
In case #2 the production matches reasonably well through the entire producing time. The parameters changed
to obtain the match were permeability and fracture conductivity.
8 SPE 120271
Case #2 md-ft VS Xf
30
25
20
Adjusted Values
Kg=0.0001 md
md-ft
15
H=400 ft.
(Kg)(h) = 0.04 md-ft
10
(w)(kf) = 25 to 1 md-ft
Run time = 8 min.
0
0 145 290 435 580 725 870 1015 1160 1305 1450 1595 1740
Xf
5000
4500
4000
3500
3000
mcf/d
gas actula
2500
gas model
2000
1500
1000
500
0
0 200 400 600 800 1000 1200 1400 1600 1800 2000
days
Figure #9 Case#3 20 Stage Hydra-Jet frac. Actual verses Model Gas Production
Case #3 20 stage production match has highest error in the first 100 days of production (+50%). The model
prediction of gas rates above 4.0 mmcf/d was never observed. The error in production after 100 days falls to
below 20%. After 365 days the model and actual production match reasonably well.
SPE 120271 9
md-ft VS Xf Case #3
4
md-ft
3
Adjusted Values
Kg=0.0001 md
2
H=340 ft.
(Kg)(h) = 0.034 md-ft
(w)(kf) = 5 to 1 md-ft
1 Run time = 16 min.
0
0 100 200 300 400 500 600
Xf
The net height for Case #3 was changed to the net height of an offset vertical completion. The conductivity
reduction was the result of the low prop volume pumped per stage which averaged 10,000# of 40/70 sand.
Results
The error in production prediction ranged from as low as 10% to as much as 50%. While this may be viewed as a
poor simulation, the match is reasonable when you consider the uncertainty that existed in the most basic
reservoir input values. We believe the results illustrate the utility of the program showing that it can adequately
predict the production outcome one might expect from a multiple stage fracture treatment in horizontal shale
wellbore. The ease of data input, the automatic gridding function, and the ability of accounting for the injected
frac fluid, makes this software user friendly even for the novice simulation modeler. The small grid size (250,000)
and rapid run time allows multiple runs in a short period of time using refined adjustments in selected input
parameters. Improvements in production matches may be obtained if;
More accurate estimate of bottom hole flowing pressure was available
Production from secondary porosity, natural or created during fracturing is used.
Further refinements can be made to other factors such as distribution of permeability measured from nuclear
magnetic resonance (NMR), lab generated relative permeability curves, and affect of offset well production.
However as stated earlier, a complete and exact data set is usually never available. Well construction and
completion decisions are required early in the development phase. These decisions require a basic
understanding of the reservoir bulk average properties and the completion practices that affect productivity. A
predictive easy to use reservoir simulator capable of modeling complex fracture geometries can help with
development drilling and completion decisions. Production sensitivity of various completion factors can be made
providing an engineering tool to use as a predictive model rather than simple reliance on “close-ology”.
1 SPE 91435 Successful Application of Hydra jet Fracturing on Horizontal Wells Completed in a Thick Shale Reservoir
Loyd E. East, Jr., William Grieser, B.W. McDaniel, and Bill Johnson, Halliburton, Randy Jackson, Devon Energy, Kevin Fisher, Pinnacle Technologies 2004 SPE Eastern
10 SPE 120271
Regional Meeting held in Charleston, West Virginia, U.S.A., 15–17 September 2004.
2 SPE 112377 Method to Pump Bridge/Frac Plugs at Reduced Fluid Rate
Don Smith and Phillip Starr, Halliburton 2008 SPE International Symposium and Exhibition on Formation Damage Control held in Lafayette, Louisiana, U.S.A., 13–15 February
2008.
3 SPE 99697 Optimization of Conformance Decisions Using a New Well-Intervention Simulator. M.Y. Soliman Halliburton; S.A. Ali, Chevron; C. Moreno, Repsol YPF; R.
Jorquera, J. Warren Halliburton. 2006 SPE/DOE Symposium on Improved Oil Recovery. Tulsa Oklahoma 22-26 April 2006
4 SPE 107629 Advanced Numerical Simulator to Predict Productivity for Conventional and Non-conventional Well Architecture
Eduardo Pacheco, Rene Castro, and M. Y. Soliman, Halliburton; Fernando Flores-Avila, PEMEX, 2007 SPE Latin American and Caribbean Petroleum Engineering Conference
held in Buenos Aires, Argentina, 15–18 April 2007.
5 Thermal Maturity of the Barnett Shale Determined from Well Log Analysis. H. Zhao, N. Givens, B. Curtis. AAPG Oct. 27 2006.
6 SPE 90051. Optimizing Horizontal Completion Techniques in the Barnett Shale Using Microseismic Fracture Mapping. M.K. Fisher, J. Heinze, C. Harris, B. Davidson, C.
7 Barnett Shale Gas Production, Fort Worth Basin, Issues and Discussion. K. Bowker. AAPG June 19 2006
8 SPE 91435 Successful Application of Hydra jet Fracturing on Horizontal Wells Completed in a Thick Shale Reservoir
Loyd E. East, Jr., William Grieser, B.W. McDaniel, and Bill Johnson, Halliburton, Randy Jackson, Devon Energy, Kevin Fisher, Pinnacle Technologies 2004 SPE Eastern
Regional Meeting held in Charleston, West Virginia, U.S.A., 15–17 September 2004.
9 SPE 115769 The Relationship Between Fracture Complexity, Reservoir Properties, and Fracture Treatment Design. C.L. Cipolla, N.R. Warpinski, M.J. Mayerhofer, E.P.
Lolon, M.C. Vincent. 2008 SPE Annual Technical Conference Denver CO 21-24 September 2008.