Atanu Basu, CEO of the big data analytics firm Ayata, admits that when he and his team first approached the oil and gas industry they were speaking different languages.
But times have changed for the 11-year-old company.
“Now we don’t talk about computer efficiency, machine learning and natural language processing. We say, ‘Your fractures per stage in that well are 30%different from a well in similar geology. How are you addressing that?’” Basu said Nov. 18 during Deloitte’s 2014 Oil & Gas Conference in Houston. The company then suggests the operator try it their way for five days and then try the Ayata way for five days and see what happens. “If we can’t help you, what we do doesn’t matter anyway.”
With a combination of structured and unstructured data, prescriptive analytics software is used to predict what will happen downhole, when it will happen and why to help operators become more efficient and improve production.
“Foresight is great as long as you can understand the dynamics around things—what’s ahead of you, what’s behind you and what’s around you as it keeps changing. That technology known as prescriptive analytics,” Basu said, pointing out how the technology can be used with sensors and a downhole camera monitoring fluid flow.
The oil and gas industry is an $8 trillion business globally, he said, and there are enormous datasets containing information such as flow rates, pressure, temperature, sound and location. But when it comes to shale completions, the data reveal that 60% of the frack stages are ineffective.
Using an onshore shale operation as an example, Basu described how sometimes “bad geology” gets blamed when results are not as expected although all the sand, water and other chemicals were pumped according to plan.
“You hear the oil and gas industry [say] it’s all about the rock. Not necessarily,” Basu continued. “Yes, it’s all about the rock and it’s about what you are doing. … There are over 1,000 variables that affect completions. More than 550 of them are actionable, something you can do something about, for drilling, production.” Of these, more than 115 can be adjusted to affect completions.
If an operator doesn’t like the recommendations, changes can be made, including changes based on available funding. “You may decide not to do the best well but do the best $6 million well, do the best $5 million well.”
“We are trying to recommend settings to complete the wells before they are drilled. All you are doing is playing with the computer,” Basu said, explaining how adjustable settings can be modified based on the biggest factors that influence a particular well’s production in a certain geological setting. “We were able to prescribe a set of completion settings taking into account the geology for where the well was going to be drilled.”
He later pointed out a problem that some companies are facing—production coming from only four or five of more than 30 frack stages. “One size doesn’t fit all in different geologies by different operators, different energy service companies, different frack fluids, different sands and different proppants,” he said.
Analytical concepts also can be applied to forms of artificial lifts, including electrical submersible pumps (ESP). Using data from sensors combined with other data such as sound, video, images, text and numbers, ESP failures can be predicted more accurately enabling the operator to prescribe a planned outage or make modifications to extend the lifetime of a pump, he explained.
Crucial to prescriptive analytics are data, which would be difficult to come by without sensors.
“We’re seeing sensors everywhere on everything on every one,” said Steven Koonin, director of New York University’s Center for Urban Science and Progress. The instrumentation of everything to create data that can be transformed into knowledge was among the science and technology areas he said will be of interest during the next 20 to 30 years. “We need to learn how to fuse information or data from multiple sources.”
Having data everywhere from everything and learning how to make sense of these data will affect people in their daily lives, he added, noting, “I think you will see it more and more in the business world.”
These sensors are already bringing in new streams of data for many industries, including oil and gas.
“What is happening in oil and gas is Mother Nature meets the Internet of Things. Mother Nature is G&G—geology and geophysics, [as well as] petrophysics,” Basu said. The Internet of Things is a computing concept in which sensors are placed on machines and other objects capable of improving or tracking performance. “The Internet of Things is already here. What is microseismic, what is a downhole camera, what is a fiber-optic sensor, what is a tracer? They are being deployed in mass scale by the large independents, the IOCs [international oil companies] and the NOCs [national oil companies] in the world for all fracking operations. They are collecting data and sending it back in real time. What they are telling you is whether what you are doing to Mother Nature is working according to the plan, and most of the time it is not.”
Contact the author, Velda Addison, at firstname.lastname@example.org.