It's been a slow road for AI in oil and gas industries
Analytics and AI can help oil and gas companies better predict supply and demand, involuntary flaring and where to drill. But the companies struggle to get talent.
While the oil and gas industries have been around for more than a century, they have been slow to adapt to new technologies.
It wasn't until the last few years that AI in the oil and gas industries took off, and still, the industries lag in AI compared to healthcare, finance and other major vertical markets.
AI in oil and gas
Oil and gas, highly sensitive to economic cycles, saw a renaissance in digital technology after the shale bust in 2015-2016, said AJ Abdallat, CEO and founder of AI vendor Beyond Limits. The collapse forced many oil and gas companies to file for bankruptcy and prompted technological changes.
Launched in 2014 as a startup spun out of the California Institute of Technology, Beyond Limits provides a cognitive system that specializes in codifying knowledge. British oil and gas giant BP is a major investor in Beyond Limits.
Beyond Limits works on a number of projects with BP, some of which revolve around economic and environmental optimization. Their partnership began, in part, due to BP's infamous Deepwater Horizon oil spill.
The disaster, which started in 2010 and is considered the largest marine oil leak ever, cost BP more than $60 billion and forced it to dramatically reduce its workforce. BP began looking into AI to help avoid further leaks, which is how it found Beyond Limits shortly after the startup launched in 2014.
In one project, BP and Beyond Limits worked on an AI-powered system for sand control. Wells in soft formations tend to produce sand, which, over time, can erode equipment and plug wells.
It may be difficult for BP's experts on sand control to put their expertise into words or numerical data, as they've learned certain fundamentals by simply doing their job over many years. Beyond Limits' cognitive system can help quantify and capture their expertise by examining their sand control recommendations over a period of time, according to Abdallat.
Most AI applications in the oil and gas business deal with simple optimization, such as temperature, well placement and oil flow optimizations.
While AI-based machine optimization technology has been available for years, it is still relatively new to them, said Aamir Aftab, senior technical lead at oil and gas exploration company Chevron.
Using AI sensors
Speaking in a session during the Ai4 2020 virtual conference, Aftab highlighted permanent downhole gauges, pressure or temperature gauges installed permanently in an oil or gas well. Sensors such as these, augmented with predictive analytics, can help monitor well integrity, pressure, temperature and other vital information.
Still, Aftab said, despite their usefulness, many wells don't have these types of Preventive maintenance capabilities available to them due to their high cost.
In an interview, George Hackford, SparkBeyond's lead impact strategist for energy, power and resources, also noted the slow pace of adoption of AI in the oil and gas industries.
If the crew needs to identify where specifically some of these lines are running, it's just amazing. It's right in front of you.
Deepthi ChandrasekaranLead business partner of advanced data and analytics, National Grid
Between their aging technologies, difficulty attracting new talent and slipping financials, oil and gas industries "have a lot of catching up to do" with analytics and AI, he said.
Founded in 2013, SparkBeyond sells an AI-powered research engine meant to augment pattern recognition and problem solving using machine learning. The company does a lot of work within the energy industry, including helping oil and gas companies optimize inventory and production and improve safety.
Using the SparkBeyond platform, oil and gas companies can aggregate economic data to predict, for example, how an oversupply of gas in Germany could impact gas prices around the world, or how different weather conditions affect gas sales.
With the platform, companies can "optimize [their] site of assets to replicate perfect conditions," or replicate poor conditions to plan ahead, Hackford noted.
Using SparkBeyond, oil and gas companies could also help reduce flaring, the controlled burn of gas during oil and gas production. Flaring can be voluntary, such as when companies make small burns to deplete some gas and make room for the more expensive oil. It can also be involuntary, when a large amount of gas is burned quickly, as with a machine malfunction or a fire.
National Grid uses AI and AR to enable their crews to identify underground gas, water, and sewer pipes
By inputting data collected by hundreds of sensors into SparkBeyond's machine learning platform, the platform predicts when a piece of a system will break, enabling crews to repair it before it breaks and causes involuntary flaring.
AR strengthens AI for oil and gas exploration
Multination utility National Grid also uses AI combined with augmented reality to help crews locate underground assets, including water, sewer and gas pipelines, and ensure they receive proper maintenance.
In a separate session during Ai4 2020, Deepthi Chandrasekaran, lead business partner of advanced data and analytics at National Grid, showcased how crews can use a smartphone application to identify where the pipelines are under the ground.
Crews can click on individual pipelines, clearly visible through the phone's camera display thanks to GPS location, computer vision and an AR overlay, to get information on the asset, including what type of maintenance was done on it, who did it, and when.
"If the crew needs to identify where specifically some of these lines are running, it's just amazing. It's right in front of you," Chandrasekaran said.
Crews can manually adjust any information they see in case it is inaccurate, she said.
