American Oil and Gas Reporter - August 2017 - 72

SpecialReport: Horizontal Wellbore Construction

Data Analytics Optimize Shale Wells
Editor's Note: The proceeding article
is the first in a two-part feature on applying
big data analytics to optimize the development of unconventional resource plays.
Part II will appear in AOGR's September
issue, and focus on the results of a Marcellus Shale application to improve completion effectiveness.

geoscientists and engineers rely on decline
curve analysis, rate transient analysis, type
curve matching, numerical simulation and
other conventional approaches to come
up with answers to challenging reservoir
development issues. However, these methods are not completely successful because
of the fact that fluid flow in unconventional
reservoirs does not adhere to the physical
principles that support conventional analytical and numerical methods.
A shale-specific big data analytical
approach offers a new and comprehensive
series of techniques to enable in-depth
evaluations of historical data, developing
predictive models based on collected
data, and optimizing well spacing and
completion practices. The approach uses
state-of-the-art solutions such as artificial
intelligence, machine learning and data
mining to address reservoir development
questions based on actual field measurements. It represents the fusion of deep
domain expertise (geology, geophysics,
drilling and reservoir engineering knowledge) with the power of data-driven analytics.
To date, the shale analytics approach
has been applied to more than 3,000 wells

By Shahab D. Mohaghegh,
Razi Gaskari
and Mohammad Maysami

MORGANTOWN, W.V.-Geologists,
geophysicists, drillers, and reservoir and
completion engineers are faced with a
series of important questions when developing unconventional assets. Many of
the biggest questions are related to completion and stimulation design, ranging
from determining the optimal proppant
concentration and number of clusters in
each stage to figuring out how to modify
a completion strategy based on changes
in geology and shale quality from one
well to another.
Above and beyond the experience that
has been gathered through the years from
observing the results of previous decisions,

FIGURE 1
30-Day Cumulative Production Cross-Plotted with Number of Stages,
Proppant/Foot, Net Thickness and Lateral Length
45,000

Number of Stages

Proppant per Foot

40,000

30-day Cum. Production (boe)

35,000
30,000
25,000
20,000
15,000
10,000
5,000
0
45,000

4

5

6

7

8

9

10
11
# of Stages

12

13

14

15

16

17

200

400

600

Net Thickness

800

1,000 1,200 1,400
Propp. (lbs-ft)

1,600

1,800

2,000

2,200

Lateral Length

40,000

30-day Cum. Production (boe)

35,000
30,000
25,000
20,000
15,000
10,000
5,000
0

100

110

120

130
Net Thickness (ft)

140

150

160

72 THE AMERICAN OIL & GAS REPORTER

2,000

2,500

3,000

3,500
4,000
Lateral Length (ft)

4,500

5,000

5,500

in the Marcellus, Utica, Niobrara and Eagle
Ford plays to understand the impact of
different reservoir and completion parameters on production, and assess the quality
of the predictions made by AI-based processes using production data from "blind"
wells. Furthermore, in the Marcellus study,
the technology identified guidelines for
best completion practices as well as for
well refracturing candidate selection.
The approach can be divided into three
phases: pre-modeling analysis (exploratory
analytics), predictive modeling (predictive
analytics) and post-modeling analysis
(prescriptive analytics). In the Marcellus
work, the pre-modeling analysis used two
data mining algorithms-well quality analysis (WQA) and key performance indicators
(KPIs)-to shed light on unclear and confusing trends in the dataset, and discover
hidden patterns in data collected during
well drilling, logging, completion/stimulation and production.
The predictive modeling integrated
all of these data from different disciplines
and operations in order to predict well
productivity. The post-modeling analysis
then used the predictive model to generate
type curves for an entire asset or specific
zone/location, perform a "look-back"
analysis to learn the best design practices
from historical data and optimize new
well completions.

Based On 'Hard Data'
Shale analytics is defined as the application of big data analytics, and encompasses any and all data-driven techniques, workflows and solutions to increase
shale play production efficiencies and
recoveries. Unlike conventional techniques
that are heavily dependent on soft data
such as fracture half-length, height, width
and conductivity, the new approach concentrates on using hard data from actual
field and well measurements.
The technology can be applied to examine historical completion practices implemented on wells that already are producing, find trends and patterns in the
seemingly chaotic behavior of the parameters that have been measured or used
for completion design, identify the importance of each reservoir and design
parameter to find the main drivers controlling production, and classify and rank
areas in a field that may respond similarly
to certain completion designs based on
American Oil and Gas Reporter - August 2017

Table of Contents for the Digital Edition of American Oil and Gas Reporter - August 2017
