Workloads: Analytic Database, Data Science, Data Engineering
Key Components: Apache Impala, Apache Kudu, Cloudera Data Science Workbench
Data Science Tools:
IoT-enabled predictive maintenance
Maintains production volumes with one-sixth the capital expense
Predicts operational shutdowns up to 72 hours in advance to reduce downtime
Extends equipment life
Big data scale
4 PB with 70 trillion sensor data points streaming in
Chesapeake Energy dramatically improved production capabilities, reduced costs, and decreased well downtime with Cloudera.
Chesapeake Energy Corporation is one of the largest unconventional producers of oil and natural gas in the United States, extracting these fuels from a variety of sources, including oil shales, oil sands, and gas to liquid processes. With operations in Louisiana, Ohio, Oklahoma, Pennsylvania, Texas and Wyoming, Chesapeake Energy produces approximately 200 million barrels of oil equivalent (the combined measure of oil and natural gas reserves) annually.
The scale of Chesapeake Energy’s operations is enormous. The company operates approximately 6,500 wells, drilling through rock formations to increase U.S.-based energy production. Over the course of its history, Chesapeake Energy has drilled 126 million feet of wells—enough to circle the globe.
Each well operates like a small factory—with numerous rigs, all dependent on a complex web of components, from drills to pumps to compressors. Two years ago, the company launched a significant transformation, merging its operations and IT teams to put data and technology at the center of its business.
“We wanted to get more value out of that data to reduce costs, increase production, and achieve zero-downtime,” said Jason Pigott, executive vice president of Operations & Technical Service at Chesapeake Energy.
“Our rich history, experience and information coupled with our growing knowledge of how to use it is making a difference for Chesapeake Energy,” said Pigott.
Chesapeake Energy uses Cloudera to optimize well operations and production. The Cloudera platform enables 24x7 real-time monitoring and uses machine learning and Internet of Things (IoT) to uncover new insights and evaluate each layer of operations in context for predictive maintenance, improved site selection and well construction, and much more.
More than four petabytes (PB) of data is analyzed, including:
IoT data streaming in from more than 70 trillion sensor data points and robotic automation systems. IoT data spans drilling operations (e.g., drill rate, rig rate, and volume and velocity of mud pumped); well operations (e.g., production rate, velocity, and pressure of wells); and machinery and robotic process automation systems (e.g., temperature, pressure, vibration of pumps, compressors, separators, and other parts).
Log data from more than 2.5 million well logs.
Financial data from the company’s SAP enterprise resource planning system.
Geomodeling data from geologist observations.
According to Pigott, implementing Cloudera instead of out-of-the-box applications for drilling wells provides Chesapeake Energy with a competitive advantage. “With out-of-the-box solutions, we would have the same analytics as our competitors and wouldn’t have any unique insights,” said Pigott. “With Cloudera, we can use our logs and data in novel ways to predict what’s ahead of the drill bit. This allows us to adjust to what’s coming next more effectively so we can drill faster than our competition.”
Historically, the company's data has been stored on prem. However, the company is currently moving some data to the cloud. “Cloudera enables us to do this work no matter what environment we operate—both on prem and in the cloud,” said Pigott.
During the past six years, Chesapeake Energy has maintained production levels while slashing capital costs by more than 80 percent. Pigott attributes much of this success to using data to increase efficiency at all levels of the organization, maximize output of its oil wells, and better select new drilling sites.
We can deliver nearly the same volumes of oil and natural gas with one-sixth the capital expenditure.
“We started with 60 rigs and now need only 15 rigs for the same production.”
Take, for example, how the company constructs new wells. “In the past, we had a cookie-cutter system, with each well completed the same way, and it was expensive to make changes,” said Pigott. “Now we have the data to see what drives different wells and which changes will make a difference to increase production. We don’t have to build a new US$30 million well to test if changing a knob will make a difference. The data can tell us.”
Additionally, predictive maintenance enables operations staff to anticipate operational shutdowns up to 72 hours in advance. This is critical in helping staff act early to prevent equipment breakdowns and achieve their goal of zero downtime.
“A one percent decrease in downtime is worth US$50 million a year for us,” said Pigott. “The ability to identify small changes early keeps our uptime higher and helps us avoid situations, such as overheating, that reduce equipment life.”