SPE 94012

Integrating Rigorous Completions Optimization into Full Field Geocellular and

Simulation Modeling
D.W
D.W.. Sobernheim;
Sobernheim, F.O.
F.O. Iwere;
Iwere, B.A.
B.A. Luneau;
Luneau, O.
O. Apaydin;
Apaydin, A.M. Aly, Schlumberger
and A.M. DataData
Aly, Schlumberger and Consulting Services
and Consulting Services

Copyright 2005, Society of Petroleum Engineers Inc.

A streamline-based flow model was utilized to upscale
This paper was prepared for presentation at the 2005 SPE W estern Regional Meeting held in geological features. The model that best captured both the
Irvine, CA, U.S.A., 30 March – 1 April 2005.
volumetric and the flow characteristics was chosen for further
This paper was selected for presentation by an SPE Program Committee following review of
information contained in a proposal submitted by the author(s). Contents of the paper, as
simulation modeling. Single-well and sector models were used
presented, have not been reviewed by the Society of Petroleum Engineers and are subj ect to to approximate the fracture properties that would be used in
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at the final simulation model. The models were calibrated to the
SPE meetings are subj ect to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
behavior of the wells and then used for forecasting
for commercial purposes without the written consent of the Society of Petroleum Engineers is performance after stimulation. A software program that took
prohibited. Permission to reproduce in print is restricted to a proposal of not more than 300
words; illustrations may not be copied. The proposal must contain conspicuous into account production and economic information was used
acknowledgment of where and by whom the paper was presented. W rite Librarian, SPE, P.O. for well and field development planning.
Box833836, Richardson, TX 75083-3836, U.S.A., fax01-972-952-9435.

Summary
In this paper, we will present workflows that incorporate two
traditionally distinct disciplines within the petroleum
engineering and geological domains: rigorous single-well
completion optimization and field geocellular and simulation
modeling. W e explore the details of this workflow and discuss
the benefits of tightly linking the two disciplines. This paper
describes an example of the workflow application to a
commingled stacked tight gas play in the Rocky M ountains of
the United States.
In commingled stacked tight gas pay, gross intervals of
thousands of feet are completed to enable maximum
productivity. Typically seen are interbedded sequences of
fluvial sandstones intermixed with shales, coals, and silty-sand
deposits. Because of the heterogeneity seen on various scales,
field modeling provides value in deciding optimum well
drilling density, determining well location geometry based on
hydraulic fracture azimuth and drainage pattern, and modeling
various economic scenarios of development.
M odels were built by integrating the seismic, petrophysical,
geological, and hydraulic fracture data generated by different
disciplines. The vertical and areal resolutions of the model
were dictated by sand body sizes, reservoir gross thickness
and heterogeneity, and fracture stages. The size of the
hydraulic fracture jobs and well spacing were critical
considerations in the selection of simulation cell sizes. M ajor
challenges included upscaling both the geological features and
the hydraulic fracture properties to simulation model scale in a
reasonably accurate manner.
Introduction
Traditionally, single-well completions optimization and full-
field development simulation have occupied distinct and
unique domains within the oil and gas industry. Different
teams of personnel with different job skills would attack these
two areas separately, often having very little interaction
between one another. Because of the very large dollar impact
in each domain, i.e. in single-well completion costs and in
multi-well development on “optimized”acreage patterns, and
because of the obvious hook-up between hydraulic fracture
job size and its impact on drainage area, and thus on drilling
density needs, it seemed obvious that significant value could
be realized through effectively linking up these domains so
that the range of potential outputs of one domain, along with
its associated costs, could drive the simulation process in order
to optimize well drilling density, patterns, and cash flow.
New technology in characterizing the hydraulic fracturing
process is instrumental in allowing this new workflow to be
possible. Through newer technologies including microseismic
hydraulic fracture mapping1, rigorous rate-transient production
decline analysis2,3, and better data and modeling routines for
hydraulic fracturing simulation4, a much clearer picture is
available for both the true dimensionality (height, length,
azimuth)and the capacity (conductivity, proppant distribution)
of that hydraulic fracture. This in turn lends itself to
simulation, where the hydraulic fracture is then modeled
numerically on an upscaled geocellular grid.
In the recent past, hydraulic fracturing was much more
conjecture than science, and engineers could not reliably give
2 D.W. SOBERNHEIM; F.O. IWERE; B.A. LUNEAU; O. APAYDIN; A.M. ALY
to a simulation team reliable estimates of fracturing
parameters. This made the point moot, in that simulation
engineers were better off lumping gross fracture
characteristics across field development in order to
“accurately” model the individual wells. In our work we
focused on improved modeling using newer simulators of low-
permeability, stacked, lenticular sand pay with hydraulic
fractures.
By a more accurate representation of the hydraulic fracture,
and with some variations in the modeling technique, discussed
below, the ability to run scenarios using true pictures of the
stimulation attributes within the multi-well simulation is now
possible. This in turn allows a linking-up, or a synergy,
during field development, from the logging of individual wells
and their reservoir characterization, through stimulation
design, and that wells impact on continued field development
(well placement, patterns, stimulation sizing).
Linking Geocellular Models to Simulation
In Figure 1 a workflow is presented describing the traditional
path from building a geocellular model to upscaling to
simulation. In this workflow, transient pressure test analysis is
used to establish average hydraulic fracture properties and
these properties are spread out over all the wells in the
reservoir simulation. As previously mentioned, this
averaging, while standard practice, does not allow for better
resolution in modeling the effect of stimulation job type and
size variation. Figure 2 presents the new workflow, which
integrates the rigorous single well completion optimization
into the simulation workflow.
Of interest in this workflow is the opportunity for the
geocellular model to provide input to both the reservoir
simulation model and the fracturing simulator. This takes the
form of estimates, for a given proposed well location, of net
pay, pressure, permeability, and desired drainage area. With
this information a preliminary stimulation job design is made
which optimizes the job size and volumes based upon these
parameters. This result then is scaled to the reservoir
simulator and is used for the field modeling runs.
Once the subject well is actually logged, the reservoir
characterized, and a rigorous stimulation design undertaken,
the single-well design is analyzed for overall economic
benefit. For various job sizes, the job cost is factored against
the predicted well productivity in order to arrive at an
optimum NPV for that well. This finalized design is then fed
into the reservoir model. For this to work, a proper hydraulic
fracture model calibration (HFMC) was undertaken.
Hydraulic Fracture Model Calibration
A key to success in this new workflow relies upon obtaining a
reliable model-based predictive method for hydraulic fracture
stimulation placement. Only through being able to simulate
SPE 94012
multiple scenarios via varying job types (fluid volumes, fluid
types, proppant volumes and types, pump rates) and realis-
tically affecting the results of such treatments, can the
stimulation design be used effectively in a field simulation. In
order to create this reliable model, our work found four critical
issues necessary for success.
Reservoir Characterization. An accurate petrophysical
characterization of the reservoir and surrounding rock layers
was critical in being able to accurately delineate rock types
and rock mechanical properties. Porosity, pressure, modulus,
and in-situ stresses are some of the primary properties
determined and served to feed both the fracturing simulator
and the reservoir simulator. Using some of the most recent
technology, including more accurate resistivity and saturation
tools, borehole pressure measurement devices, and more
powerful multiple array sonic devices, the best possible
reservoir characterization models were created. They allowed
for confidence in the properties, which served as the basis for
the subsequent work in the HFMC.
Far Field Fracture Growth Monitoring. Technology using
downhole microseismic listening devices in the near vicinity
of the subject wellbore to measure seismicity created during
the hydraulic fracturing process was utilized to create a three-
dimensional picture of the fracture growth with time, during
the treatment1. Rigorous geophysical routines were used to
evaluate the measured data and create the picture of the
fracture growth in height, azimuth, and length-wise dimension
(fig. 3). The technique required that our observation well be
within 1200 ft of the subject well in order to get good data;
however, this distance appears to be variable and is dependent
on the sonic properties of the rock mass between emitter (the
hydraulic fracture) and receiver (offset wellbore).
Once measurements were made and the data processed, the
fracture dimensions were compared to the dimensions
predicted by the fracturing simulator (fig. 4). Predictions for
length and height were made to match within plus or minus
20% of the measured results. Primarily, changes to the
geomechanical earth model stresses and modulii, were used to
improve HFMC simulator prediction.
Quantitative Production Decline Analysis. Besides gross
dimensionality determined via microseismicity, rigorous
production decline analysis was undertaken in order to
understand the flow capacity properties of the created fracture.
Here, too, new technical advances referenced in the
literature2,3 have allowed for significant new capability in this
arena. Applying production logging and computational
techniques created individual reservoir layer production
profiles for the productive history of the well.
In the low permeability gas sands much of our work was
performed in, typically a year or more worth of production
data and flowing pressures was required in order to see clear
formation linear fracture flow inception. Where the end of
SPE 94012 INTEGRATING RIGOROUS COMPLETIONS OPTIMIZATION INTO FULL FIELD GEOCELLULAR AND SIMULATION MODELING
formation linear fracture flow could be discerned, the
computation yielded values for reservoir permeability, keff,
effective fracture half-length, Xf, and effective fracture
conductivity, kfbf. These factors are critical results necessary
in understanding the well’s performance.
Knowing the actual, in-situ, results of the hydraulic fracturing
treatment, was extremely valuable in that it was now possible
to further calibrate the HFMC software to reliably predict
measured properties. The parameters that were varied in the
fracturing software included such items as fracturing fluid
damage factor due to incomplete clean-up, proppant retained
permeability factors, and fluid leak-off parameters.
HFMC Simulator Technology. Much of the workflow
development centered on multi-layer, low permeability gas
reservoirs5. Typically in these completions, multiple
perforation sets within individual sand lenses are used within a
given fracturing stage to provide adequate coverage of the
reservoir pay. Because traditional fracturing simulators use a
single-layer initiation routine, it was difficult to calibrate the
model because it did not appear from microseismic data that
this was in fact what was happening. What did appear to be
occurring was multiple initiation points at the differing
perforation sets, as discerned from fracture growth monitoring
and quantitative production decline analysis.
Because of this, a multiple-layer fracturing model was sourced
which allowed for multiple points of parallel initiation and
overlapping fracture dimensional growth between layers6.
Measurements taken via microseismic data and also through
production analysis clearly indicated that the results obtained
via this sort of model were much more representative of
reality. The final model simulates individual fracture growth
concurrently and also includes perforation friction pressure
effects in allocating wellbore pumped fluid volumes to
individual fracture wing growth.
The only limitation of this simulator at this time is that it does
not simulate assymetrical fracture growth, a feature seen on
the microseismic data in varying degrees. This represents an
area for future work in simulation where, depending upon the
geocellular mechanical earth model, varying degrees of
assymmetry could be modeled.
Enhanced Reservoir Model
With the key outputs from the hydraulic fracture design
process of fracture half-length, Xf, dimensionless fracture
conductivity Cfd, and propped interval h, the reservoir model
used in simulation could be enhanced. The reservoir models
were built by integrating seismic, petrophysical, geological
and the above mentioned hydraulic fracture data, generated by
different disciplines.
3
Seismic Attributes and Geological Modeling. This
workflow involves applying seismic attributes to a seismic
cube in three dimensions (3D) in order to visualize any
anomalies. Dominated by stacked fluvial and marine
sandstone and shale sequences, the tight gas reservoirs studied
were best modeled using facies modeling (fig. 5). Well logs
were scaled up on a zone-by-zone basis and stochastic fluvial
object modeling was applied using a range of trends
determined from basin studies.
Petrophysical Reservoir Characterization. The
petrophysical challenge faced in multi-layer pays was to
develop an accurate reservoir model for the classic porosity,
water saturation and lithology computation. In typical low-
permeability gas basins, such as in this work, the classical dual
water model was used for the standard evaluations.
The basic log inputs to the standard model were deep
resistivity, neutron porosity, bulk density and gamma ray.
Total porosity was computed from the neutron density cross
plot. The bound water computation was taken from a
minimum of several indicators. The gamma ray and the
neutron porosity were sufficient in most cases in these tight
gas reservoirs. The use of neutron porosity as a shale/ bound
water indicator also eliminated the false pays due to erroneous
readings from bad hole conditions. Reservoirs were broken
into layers based on predefined criteria, including porosity,
thickness and shale volume cut-offs.
Reservoir Model. The vertical and areal resolutions of the
model were dictated by sand body size, reservoir gross
thickness and heterogeneity, and fracture stages. The size of
the hydraulic fracture jobs and well spacing are critical
considerations in the selection of simulation cell sizes. Two of
the major challenges that had to be overcome are upscaling,
with reasonable accuracy, the geological features and
hydraulic fracture properties to simulation model scale7. A
streamline based flow model was used to upscale geological
features (fig. 6).
Multiple models were generated at different upscale ratios.
Boundary conditions—injector-producer pairs, injection-
production rates, and others—were imposed on each model
and the results compared with the geocellular scale model.
Since the reservoir is tight, pseudo fluid properties (low
viscosity) and a very long prediction period were used. Fluid
recovery and injected fluid breakthrough times were compared
to those of the geocellular model. The model that best captures
both the volumetric and flow characteristics was chosen for
further simulation modeling.
Single-well and sector models were used to approximate the
ultimate fracture properties that would be used in the final
simulation model. First, local grid refinement is used to
represent the fracture properties. Then a parametric study is
conducted by calculating effective pore volume, permeability
and/or skin. These properties are used to represent hydraulic
4 D.W. SOBERNHEIM; F.O. IWERE; B.A. LUNEAU; O. APAYDIN; A.M. ALY
fractures and the results—production and pressure
performance over a long period—are compared to those
obtained by local grid refinement. Effective permeability and
pore volume calculation yield the most favorable comparison
and, therefore, are used during the history-match phase.
After the models are successfully calibrated against the
behavior of the wells, they are used for forecasting well
performance under 40-, 20-, 10- and 5-acre spacing. The
production streams for the wells and economic parameters are
fed into an economic package, which gives the economic
performance of the various cases.
Integrated Models
Figure 2 shows how the well-centric, single-well completion
process is integrated with the full-field modeling work. The
fracture model inputs of reservoir net thickness, porosity and
effective gas permeability are taken from the geocellular field
model. These data, particularly permeability or flow capacity,
are conditioned with data from fall-off/leak-off tests,
production logs, production and pressure tests, and
microseismic hydraulic fracture stimulation diagnostics, if
available. The conditioned data also serve as input in the
simulation model.
Via this integrated model, dual objectives of both (1) an
optimized single-well completion design, and (2) an improved
single-well data-set for use in the field modeling simulation,
are realized. Because the underlying inputs to both processes
are the same, a cross-disciplinary synergy is realized which
links the high value objectives of optimum stimulation design
and economics with well spacing and optimum drainage
patterns.
Case History
For one low permeability, stacked multi-pay dry gas reservoir
in the Rocky Mountain region of the United States, this
process was applied. The first step was to build a detailed
reservoir characterization model of the interval to delineate
porosity, net pay, pressure, and calculate mechanical earth
properties via long-spaced sonic processing. Once this was
completed, two production logs taken over the first year’s
producing life were used to allocate production from the
various layers over time. The data requirement for this step
was detailed daily production history and flowing wellhead
pressure which, combined with the production logs, allowed
for an equivalent individual completed interval production
history to be calculated.
The individual zone rate histories were then fed into a
fractured well rate-transient diagnostics program for
evaluating reservoir and fracture properties in each flow
regime of the rate-transient behavior of the well. All the zones
analyzed showed flow regimes through the end of formation
linear flow, allowing for a rigorous calculation of effective
SPE 94012
fracture half-length, fracture proppant conductivity, and
formation permeability. Additionally, boundary effects were
examined to see if they were present, however none were
noted indicating that the hydraulic fractures had not reached
beyond the well’s drainage area. This technique of production
analysis was found to be far superior to conventional pressure
transient testing via build-ups, as the reservoir was so low-
permeability (<0.01 md) that build-up testing would have
required shut-in times on the order of several months to
several years to yield meaningful results regarding fracture
half-length.
The calculated fracture properties were then fed into an
analytical hydraulically fractured well production model to
create a history match of both the individual layer’s allocated
production history, as well as the commingled well’s total
production, prior to passing the results to both the fracturing
simulator model and the numerical production simulator. The
fracturing simulator was set to run in a multiple-layer,
independent fracture growth mode (over-lapping mode),
allowing for multiple initiation of hydraulic fractures in
individual sand layers grouped within a fracture stage. By
working with the mechanical earth model, including stress and
modulus, a correspondence between predicted behavior and
calculated actual results was performed. This step allowed for
much greater confidence in future fracture design work that
the simulator output would be representative of the results
generated in-situ.
The results were passed to the numerical reservoir simulator
where, initially, a single-well model was constructed to match
the well’s performance and thereby tune the simulator8 (fig.
7). The fracture half-length dictated the number of grid cells in
the reservoir model representing the hydraulic fractures and
the properties of these cells (fracture effective permeability
and pore volume) were calculated from the fracture
conductivity obtained previously and history matching of the
well historic performance (fig. 8). Several field development
scenarios, including well spacing, fracture treatments and
completion schemes, drilling schedule, etc were investigated
using the calibrated model.
Conclusions
1. The data and activities of multiple domains were
integrated to allow many different development scenarios
to be considered for optimal well placement and
completion practices.
2. The workflow presented allows for optimization of both
individual well stimulation treatments and field
development decisions for tight gas reservoir
development and exploitation.
3. Through linking multiple domains, a framework is in
place that allows for modeling how decisions made
regarding particular practices in one domain impact
SPE 94012 INTEGRATING RIGOROUS COMPLETIONS OPTIMIZATION INTO FULL FIELD GEOCELLULAR AND SIMULATION MODELING 5

potential results achieved in another, in both technical and Fracturing Design,” paper SPE 68175 presented at the 2001 SPE
economic terms. Middle East Oil Show held in Bahrain, 17-20 March 2001
8. Iwere, F.O., et al.: “Numerical Simulation of Thick, Tight Fluvial
4. A common geocellular and petrophysical model serves as Sands”’ paper SPE 90630 presented at the 2004 SPE
International Petroleum Conference in Mexico held in Puebla,
the basis for both single-well completion optimization and Mexico, 8-9 November 2004.
reservoir simulation.

5. A calibrated hydraulic fracture simulation model is

necessary to achieve true predictability in the integrated
process.

6. Improved reservoir simulation techniques for

hydraulically fractured, tight, multi-layer, gas wells and
field modeling have been tested and validated.

Nomenclature

CfD = dimensionless fracture conductivity

keff =formation effective permeability, L2, mD
kfbf = fracture conductivity, L3, mD-ft
Xf = effective fracture half-length, L, ft
Mcf/D =gas production rate, L3/T, thousand cubic
ft/day
MMcf = cumulative gas production, L3, million
cubic ft

Acknowledgment

The authors wish to thank Schlumberger for giving permission

to publish this technology paper.

References

1. Maxwell, S.C., et al.: “Microseismic Imaging of Hydraulic

Fracture Complexity in the Barnett Shale,” paper SPE 77440
presented at the 2002 SPE Annual Technical Conference and
Exhibition held in San Antonio, Texas, 29 Sep. –2 Oct. 2002.
2. Poe, B.D. Jr., et al.: “Advanced Fractured Well Diagnostics for
Production Data Analysis,” paper SPE 56750 presented at the
1999 Annual Technical Conference and Exhibition, Houston,
TX, Oct. 3-6.
3. Poe, B.D. Jr.: “Production Performance Evaluation of
Hydraulically Fractured Wells,” paper SPE 59758 presented at
the 2000 SPE/CERI Gas Technology Symposium held in
Calgary, Alberta Canada, 3–5 April 2000.
4. Mack, M.G., and Warpinski, N.R.: “Mechanics of Hydraulic
Fracturing,” Chapter 6 in “Reservoir Stimulation,” ed. M.J.
Economides and K.G. Nolte, John Wiley & Sons, 2000.
5. Sobernheim, D.W., et al.: “An Integrated Technique to Optimize
the Completions in Low Permeability Gas Reservoirs While
Improving Efficiency and Productivity,” paper SPE 84171
presented at the 2003 SPE Annual Technical Conference and
Exhibition held in Denver, Colorado, USA 5-8 Oct. 2003.
6. Gu, H., Desroches, J., and Elbel, J.L.: “Computer Simulation of
Multilayer Hydraulic Fractures,” paper SPE 64789 presented at
the SPE International Oil and Gas Conference and Exhibit, held
in Beijing, China, 7-10 Nov. 2000.
7. Aly, A.M., et al.: “Optimization of Gas Condensate Reservoir
Development by Coupling Reservoir Modeling and Hydraulic
6 D.W. SOBERNHEIM; F.O. IWERE; B.A. LUNEAU; O. APAYDIN; A.M. ALY SPE 94012

Petrel/Frontsim
GeoCellular Model

Upscaled
Model

Full Field
PTA / DCA
Model

Fig. 1 Basic Simulation Workflow Fig. 3 – Microseismic Imaging of Multiple Stage Tight Gas
Fracturing

FMI Natural
Petrel/Frontsim Mechanical
Fracture GeoCellular Model Earth Model

Upscaled DSI / Microseismic

Model

Production Log
Sector FracCADE
Model/NWM
Microseismic
Tracer Survey
Full Field Well Prod.
Model PTA / DCA
Model
PSPLITR

Economics ProFIT
Risk/NPV
Integrated Completion Optimization
Field Development Plan

Fig. 2 – Enhanced Simulation Workflow Including Completion Fig. 4 – Calibrated Model Detail of Hydraulic Fracturing Simulation
Optimization Showing Microseismic Events
SPE 94012 INTEGRATING RIGOROUS COMPLETIONS OPTIMIZATION INTO FULL FIELD GEOCELLULAR AND SIMULATION MODELING 7

Fig. 5 – Facies Model for Multi-layer Stacked Pay Showing

Porosity Distribution

Fig. 7 – Single Well Model Including Hydraulic Fractures Using

Local Grid Refinement

Fig. 6 – Streamline based Flow Model Used to Upscale Geological Fig. 8 – Production History Match of the Well’s Historic
Features Performance Using Local Grid Refinement