AI still has a lot to learn about humans in the natural gas and oil industry

Artificial intelligence (AI), which tech giant Microsoft Corp. defines as “any technique allowing computers to imitate human intelligence”, is fully in line with digital transformation.

Exploration and production (E&P) and oilfield services (OFS) companies have been embracing digitalization for years, and the momentum is not expected to wane anytime soon.

In a recent study, global technology market research firm ABI Research predicted that oil and gas companies will spend $15.6 billion on “digital transformation technologies” by 2030.

“The role of technology is evolving from helping oil and gas companies monitor their large, complex and dangerous operations, to helping them optimize their facilities to manage the volatility of their operating environments,” said Michael Lerner, analyst Principal of Industry and Manufacturing at ABI.

Lerner ticked off for NGI a list of the ways E&P and OFS companies are using AI.

“To keep workers and personnel safe, prevent leaks and increase yields, oil and gas companies are investing in technologies to analyze and monitor their operations, such as their wellheads or drill rigs, deploy sensors to signal status of pipes and pipelines, and operations centers to keep up to date,” Lerner said.

Lerner added that companies are using AI to monitor local geology to understand the extent of reserves and operators to monitor local seismic activity. For example, BP plc and ExxonMobil have demonstrated their prowess in this regard. Canadian company Meg Energy Corp., which uses the technology to optimize production at its Christina Lake facility, is also applying AI.

“Operators also need to proactively monitor local conditions underground and another aspect that needs continuous monitoring is the weather so they can shut down facilities and evacuate personnel,” Lerner said.

Ted Wilson, head of energy business development at digital information technology company SoftServe, told NGI that E&Ps are also using AI to optimize production, automate control systems, support worker safety and monitor emissions and the carbon footprint of the supply chain.

[Special Edition: Download “ESG Reshapes the Natural Gas Pricing & Investment Landscape” now, and access NGI’s extensive coverage of where the market is headed in 2022 and beyond.]

When asked where he sees the greatest need for AI deployment in E&P and OFS operations, Lerner said one area uses data analytics to assess risk and execute scenario plans for wellhead condition monitoring.

“Historically, operators would move to another site once a site’s productivity began to depreciate,” he said. “As new springs are often located in harsh environments, such as much deeper under the sea, operators are investing in analytics to model their active reservoirs to understand how they can increase yield levels.”

In the pipeline industry, “operators with assets in remote locations are turning to digital technologies to perform remote monitoring,” Lerner added.

AI offers oil and gas companies additional opportunities in addition to automating and improving worker safety and reducing operating expenses, Wilson noted.

He said “there is a greater opportunity in the AI ​​market, which is for companies to understand where an AI-based system can actually increase intelligence and create areas of optimization where Humans simply cannot function as well or as quickly”.

For example, Wilson noted that machine learning (ML)-based systems help geologists and geophysicists interpret massive volumes of seismic data. Additionally, he said ML technology could help an E&P increase production from a well by training how to optimize output parameters based on variables such as equipment, production, and temperature data.

What about human factors?

Behaving in a more humane way is an ongoing pursuit of a sub-discipline of AI called cognitive AI.

Mike Krause of software company Beyond Limits told NGI that cognitive AI involves making diagnoses or predictions with AI that are based on accumulated institutional knowledge.

Krause, the company’s senior director of AI solutions, said cognitive AI adds “explainability to AI outcomes in a way that’s meaningful to people.” He added that the subspecialty refers to “the inclusion of knowledge within an AI framework.”

At a relatively basic level, an example of cognitive AI would use an image recognition algorithm to allow the AI ​​to identify a stop sign, Krause said. An example of cognitive AI in the oil and gas industry would be coding technology to recognize a well, he added.

“I give it attributes like casing, drilling…all the material in a well,” and then various other “explanatory layers” of information, he said. “Once you apply the cognitive view, you really expand the possibilities.”

Krause emphasized the importance of granular detail, especially in an industrial AI application that recommends, say, how to modify a factory workflow or how to drill a multimillion-dollar well.

“I want to know why, and I want to make sure this recommendation is based on something that’s real and makes sense,” he said.

Cognitive AI relies more on knowledge than on data, but Krause said it still depends on capturing “that information, or that knowledge, in the first place into a system that can then be exploited by a wide range of users across the organization”.

While deploying AI can help streamline the operations of an E&P or OFS firm, it also presents legal and ethical issues, intellectual property (IP) attorney Terrell Miller, a partner at the firm, told NGI. of Foley & Lardner lawyers.

“For example, job loss and workforce morale are two ethical issues that can arise when AI is deployed in a way that can replace human employees,” he said. .

Wilson, however, said the “human element” is often a key ingredient for successful AI deployments.

“For example, many of our best ML solutions have required deep subject matter expertise from well engineers, operations professionals and geoscientists to be successful,” he said. “Data by itself is sometimes powerless without being paired with the domain knowledge of human experts who know the patterns behind it.”

Data dependency, in turn, can create other problems, Miller said.

From a legal perspective, the “largely data-driven” nature of AI systems carries potential issues related to cybersecurity and data reliability, he said.

Because many AI systems are subject to IP protections, their use “in the oil and gas industry comes with very real concerns about third-party intellectual property infringement, such as infringement of patents,” Miller said. “As such, it is important to license any use of AI systems against existing registered intellectual property or third parties.”

Otherwise, “companies expose themselves to increased risk of patent litigation or other intellectual property litigation, which can come with significant liability risk,” he said.

Comments are closed.