American Oil and Gas Reporter - November 2015 - 49

FIGURE 3
Well Clustering Algorithms Enable Intuitive Field Segmentation

decrease the training time needed by algorithms to learn from the images.
If a training label is available, a classifier is trained to identify patches of
interest. In the unsupervised case, a
stacked, automatic-encoder network is
used to generate a dictionary of representative patches that can be used to locate areas of interest in new images. In
addition to extracting important hidden
patterns in seismic profiles, the same
technique can be applied to other abstract
sensor solutions, such as acoustic fluid
properties.
For example, by marrying traditional
image and attribute interpretation with a
suite of deep learning workflows, we
discern meaningful patterns to efficiently
identify direct hydrocarbon indicators
across 3-D seismic datasets.
High-Performance Analytics
While predictive analytics, forecasting
and visual analytics all work with traditional computing methods, the volume
and velocity of data streaming from oil
fields requires a more agile approach. As
sensors are added to more equipment
and sites, it's important that data be digested quickly and turned into useable
information in hours or days, not weeks
or months. Two innovative technologies
increase the speed of insights on massive
datasets.
The first is in-memory analytics, which
enables big data workflows to solve complex upstream problems in an unfettered
manner. The speed and flexibility of inmemory analysis allows analyses that

previously were out of scope because of
computing constraints.
In-memory analytics scales to a business's needs, providing concurrent, inmemory and multiuse access to data, no
matter how big or small. The software is
optimized for distributed, multithreaded
architectures and scalable processing, so
requests to run new scenarios or complex
analytical computations are handled blazingly fast.
This technology is particularly valuable
to upstream geoscientists for performing
data exploration, visualization, and descriptive statistics and modeling with advanced algorithms.
Alternative technologies such as SQL
(search and query language)-based solutions have a number of limitations, including column limits, storage constraints,
and limited data type support. Variable
selection, dimension reduction, visualization, complex analytic data transformations, and model training require multiple concurrent passes through the data-
a process not handled well by SQL and
relational technologies.
In-database analytics executes an analytic algorithm directly within a database
engine, using a native database. This allows scientists to avoid moving data,
shortens run times, and allows organizations to use existing data warehouses.
In-database processing is a flexible,
efficient way to use increasing amounts
of data by integrating select upstream
technology into databases or data warehouses. It utilizes the massively parallel
processing architecture of the database

or data warehouse for scalability and
better performance.
Moving relevant data management,
analytics and reporting tasks to where
the data reside is faster, reduces unnecessary data movement, and promotes
better data governance. For upstream decision makers, this means faster access
to analytical results, and more agile and
accurate decisions.
Oil companies operate in a highly competitive, changeable global economy. But
every problem has an opportunity attached.
By wisely harnessing data and modern
analytical techniques, companies can exploit
unconventional hydrocarbon deposits more
efficiently and profitably.
r

KEITH
HOLDAWAY

Keith Holdaway is an oil and gas
domain expert in the SAS Global Energy Practice. He joined SAS in 1997
after nearly 20 years at Shell, BP,
and two geophysical service companies. His work has included spending
a decade in the Middle East. He graduated with degrees in geology and
geophysics from Durham University
in the United Kingdom.

MORAY
LAING

Moray Laing is an oil and gas domain expert in the SAS U.S. Energy
and Manufacturing Practice. He joined
SAS in 2013 after nearly 25 years at
Baker Hughes, where he worked in
various capacities from engineering
to new technology development. He
graduated with degrees in computer
science and electronics engineering
from Robert Gordon University, and
holds an executive M.B.A. from Texas
A&M University.
NOVEMBER 2015 49

American Oil and Gas Reporter - November 2015

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