American Oil and Gas Reporter - January 2015 - 160

Production Technology
The standard and conventional routine
of reservoir modeling is to develop a numerical reservoir model using available
static and dynamic properties. This flow
model is based on the geological model.
Considering the sequential order of the
workflow, it can be asserted that conventional reservoir modeling is a "bottom-up"
process. First, the geocellular model is built
using geological and geophysical data, followed by performing the upscaling
process. Then the flow model is developed
using well-known engineering fluid flow
principles before moving to the final
step: the history matching process.
The principal assumption that should
be made in traditional reservoir simulation
is that all the complexities of the reservoir
are known and can be modeled in terms
of the available mathematical equations.
However, every engineer knows this is not
the case in reality. It never can be claimed
that all the aspects and complications of
the geological layers are known, no matter how thorough the data acquisition
process. Neither can it be claimed that all
the phenomena happening in the reservoir
in terms of fluid flow and the interactions
of rock and fluid are well represented in
terms of mathematical models and formulas available.
This problem is even more challenging
while dealing with unconventional reservoirs. In fact, when it comes to modeling
the physics of shale reservoirs, many engineers can attest that traditional modeling results likely will be far from the reality of the actual physics within shale formations. Viable alternatives clearly are
needed to improve the accuracy of reservoir modeling in unconventional sourcerock plays.
Full-Field Simulation
As opposed to bottom-up modeling
processes, TDM takes a completely different tactic for full-field reservoir simulation.
In this state-of-the-art technique, reservoir
engineering, statistical analysis, advanced
data-driven analytics and machine learning are integrated to build a reservoir model as an alternative (or complement) to traditional reservoir modeling approaches.
Although reservoir engineering concepts are strongly adhered to, the formulation is not imposed to the reservoir model in the TDM process. Instead of imposing the current understanding of the
physics to the model, the technique lets the

model learn the behavior from the diverse
and multiscale data measured throughout
the life of the asset. However, one of the
most significant differences between
TDM and conventional numerical reservoir modeling is in the amount of data
needed.
The data acquisition requirements for
numerical reservoir modeling are extensive, time consuming and costly, and
therefore, may not be economical in some
cases, especially for smaller companies.
In addition, in many mature fields, the histories of the wells go back to a time when
not much attention was paid to recording
and maintaining data. In most cases, the
only available data are production-related, which will not meet the basic requirements for building a numerical reservoir
model.
TDM is a better alternative in these situations. It is able to provide the necessary
reservoir management tool and uses datadriven modeling techniques that can take
the most advantage of existing data, which
may not be sufficient for numerical modeling. Generally, the major differences between numerical reservoir modeling and
TDM go back to the small footprints of
TDM in development, the running and history matching process, the ability to provide
full-field models that can be used in making the management decisions using the
available data, and tackling the cost issue
for building geological and flow models.
TDM has been used in multiple projects around the world and has proven to
be a reliable alternative in cases where traditional reservoir models cannot be developed because of data limitations, the
complex physics of the reservoir, or the
fact that the model development process
is simply too costly or time consuming.
This study analyzed 145 wells in the
Wattenberg Field using various data-driven techniques to model the Niobrara
reservoir. All data were publicly available
and obtained through the Colorado Oil &
Gas Conservation Commission's website. The workflow used in the study is part
of Intelligent Solutions Inc.'s IMagine™
software application, which provided the
required tools and techniques for developing the TDM approach.
Fuzzy pattern recognition was implemented to provide the best producing locations as well as identify underperforming wells. Also, an artificial intelligencebased, history-matched model was devel-

160 THE AMERICAN OIL & GAS REPORTER

oped and validated for predictive practices,
sensitivity analysis, infill drilling locations,
and other reservoir management purposes.
Discovered in 1970, the Wattenberg
Field extends in a north-to-south direction
across 50 townships, covering an area approximately 50 x 70 miles. Hydrocarbon
production is from multiple pay zones, including the Dakota-Lakota, J Sand, D
Sand, Codell, Niobrara (A, B and C),
Shannon and Sussex. The Niobrara consists mainly of interbedded, organic-rich
shale, calcareous shale, and marl. It is
found at depths between 6,200 and 7,800
feet, and its thickness ranges from 300 to
400 feet. The overpressured Niobrara
reservoir has very low permeability (0.010.1 millidarcy) with a total organic content of 0.85 to 2.75 weight percent.
More than 10,000 wells have been
drilled and completed in the Codell and
Niobrara formations since 1981, but the remaining recoverable reserves in the Niobrara are estimated to be in the range of
2 billion-4 billion barrels of oil equivalent.
The middle branches of the Niobrara (B)
are targeted mainly by horizontal drilling
using various completion designs, including slickwater fracture treatment and
multistage fracturing with 12 to 35 stages.
The initial well spacing was approved on
320 acres, but ongoing Codell and Niobrara development activity has resulted in
spacing reductions to 80 and even 40 acres
in some cases.
Data Analysis
The area of review in this study consisted of six sections (21, 22, 27, 28, 33 and
34) located in Range 66W and Township
3N (Figure 1A). There were 222 wells that
had production from Codell, Niobrara and
J Sand formations in this area. Excluding
the wells that produced from only the
Codell or J Sand resulted in 145 wells in
six sections with production from the Niobrara formation only, or comingled with
production from the Codell. In order to allocate the approximate production to
each formation, a production ratio was
used based on the porosities, formation
thicknesses and water saturations obtained from the logs for each well.
Well logs for more than 60 wells were
analyzed to obtain the petrophysical data
over the area of review. Formation thickness, porosity and water saturation were
the most important parameters deter-

American Oil and Gas Reporter - January 2015

Table of Contents for the Digital Edition of American Oil and Gas Reporter - January 2015