American Oil and Gas Reporter - July 2021 - 21

AI Data Crunching Tools
Pick Up Notable Speed
In Advising Optimization
Recently at an event presented by the University of Kansas
Tertiary Oil Recovery Program and the Petroleum Technology
Transfer Council a Silicon Valley company addressed waterflooding
optimization to a group of Mid-Continent operators.
Now before you say, " SILICON VALLEY? " in a tone reminiscent
of the cowboys saying " New York City? " in the old salsa commercial,
stay with me. These folks' technology roots may be in
California, but they are oil field all day long.
The company, Tachyus, is named after the tachyon-the
theoretical particle in physics that travels faster than light
speed-because it is all about high-speed data processing.
Tachyus's invitation to present demonstrates KU's ongoing
commitment to the regional and national oil and gas industry.
Headed by Reza Barati, KU TORP and PTTC continue a
tradition that has endured more than 40 years. Similar to the
days when we educated operators about the personal computer's
applicability to oil and gas production, the time is right to
explore the industry's next big step changes: machine learning
and artificial intelligence.
Tachyus creates technology to optimize production. The
Houston-based tech startup focuses on AI in the oil and gas industry.
Founded in 2014 by former Microsoft developer Paul
Orland, the company moved to Houston and maintains a presence
in Silicon Valley while serving customers in the United States,
United Kingdom, Canada, Argentina, Norway and Thailand. At
their core, Tachyus' software products allow geoscientists and
reservoir engineers to interpret the deluge of data collected
from site sensors and monitors in a fraction of the usual time.
One of the company's key technologies is an application
called Aqueon, which has been designed to quickly analyze
waterflood data. Tachyus has enjoyed enough success with
the software as to sign a partnership agreement with Pan
American in Argentina to help optimize its waterfloods. So
powerful is the technology, KU TORP sought to have its
system run some regional data and develop a case study to
present to the region's operators.
The course taught participants about production optimization
across some highly challenging production environments,
including secondary and tertiary oil recovery in complex
reservoirs. Using Kansas data, the presenters employed their
software platform to integrate all relevant information sources
in real time. The AI process quickly explored millions of
scenarios to recommend an optimal operational and development
scenario. Typically, these results increase net present values
more than 10%.
The machine learning process is similar: You acquire and
upload data, apply some parameters to a small portion of that
information, train the system using all the data, receive some
system-generated options, then you select the best one and give
it a shot. The software demonstration used this same workflow.
Data acquisition generally is not a problem in places such as
Kansas, where wells have been producing for years. The next
step is to upload this production and well information into the
" Once the system is trained, it is ready to
begin running scenarios. For example, it
becomes possible to look into the
consequences of closing a certain well,
converting a producer to an injector or
changing injection rates.
"
system. The system's digital image of the data makes it easy to
spot mistakes and quality check before the model creation step.
That step puts the system to work at using small amounts of
data to begin creating the model as the operator strives to
identify the desired parameters.
During this process, the system makes correlations between
the parameters and data to measure the model's accuracy. Once
users reach a high degree of accuracy, they are ready to train the
system with complete data sets. In this case, more data is better.
Once the system is trained, it is ready to begin running scenarios.
For example, it becomes possible to look into the consequences
of closing a certain well, converting a producer to an injector or
changing injection rates. The system crunches the data to answer
these questions in minutes. Finally, after it is fully trained, the
system can begin developing optimization recommendations
that can help enable quick field decisions. As you continue to
use the system, new data also feeds back into the program,
where it continually helps refine the analysis.
At the end of the day, the presentation showed how the
Tachyus system offered another way to handle what we used to
leave to experience, instincts and luck. Unfortunately, a few
decades' experience at well optimization comes at a premium,
and so technology such as Aqueon can help us get there faster
with more accuracy.
❒
JEREMY VISCOMI is a strategic
marketing expert for the oil and
gas industry. Among his many
roles, he serves as the MidContinent
regional lead for the
Petroleum Technology Transfer
Council, providing independent
operators with access to emerging
technology aimed at reducing
risk and growing profits.
JULY 2021 21
American Oil and Gas Reporter - July 2021

Table of Contents for the Digital Edition of American Oil and Gas Reporter - July 2021
