AI spells opportunity and manageable risk for the oil and gas industry
The use of Artificial intelligence (AI) in the oil and gas industry is growing as more operators digitalize for cost efficiency and lower carbon emissions during production, transmission, and distribution.
The global market for AI applications in oil and gas value chains will be nearly USD 3 billion in 2024 and USD 5.2 billion in 2029, according to Mordor Intelligence.1 Nearly half (47%) the oil and gas industry professionals surveyed for DNV’s Transforming Through Uncertainty report say their organizations will use AI in operations in 2024 (Table 1).2
Table 1: What oil and gas professionals said about digitalization in early 2024 (Source: DNV 2024)
| Statement | Agree (net*) |
| Digitalization is central to my organization’s strategy | 62% |
| Digital technologies are enabling the energy transition for my organization | 45% |
| Digitalization has improved our business performance | 64% |
| In the year ahead, my organization will use AI-driven applications in our operations | 47% |
| Regulatory changes are making it more difficult for my organization to benefit from AI | 20% |
| * % who agree minus % who disagree |
While AI gains traction, only 15% of those respondents to DNV’s research say their organizations are using AI in live, day-to-day operations (12%). Only 3% report highly integrated and/or advanced use of AI. Nearly half (47%) say their AI use is in planning or piloting stages.
“Trust is the key to move beyond this low base to unlock the full opportunities and reduce the risk that AI will not meet stakeholder objectives or comply with increasing and stricter regulation,” said Kjell Einar Eriksson, Vice President – Digital Partnering, DNV. “The industry needs assurance that AI can deliver value effectively, efficiently, securely and safely. It needs to be sure it is investing in trustworthy AI, developed in-house or purchased, that can meet stakeholder expectations in a verifiable way. Hence, external developers and suppliers of AI components and AI-enabled systems need to prove through validation and verification that their products can be trusted.”
Opportunities from AI in oil and gas
In today’s oil and gas industry, AI computer programs are most commonly found as a subset of AI called machine learning (ML). ISO/IEC 22989 defines ML as a ‘process of optimizing model parameters through computational techniques, such that the model's behaviour reflects the data or experience’.3
Using ML to analyse and learn from the huge datasets collected by the industry is bringing benefits across the value chain and asset lifecycles.4,5,6
It is boosting the efficiency of oil exploration and enhancing sustainability in managing hugely complex oil and gas development projects. ML is also facilitating optimized predictive maintenance of operating assets such as oil platforms, pipelines, and refineries to avoid downtime and enhance safety. In short, AI can boost the effectiveness of integrity management to prolong the safe, economic life of assets.
Using data from sensors and from on-site, aerial, and satellite cameras, AI/ML can identify and get better at recognizing and quantifying oil spills and methane releases to air, enabling more timely and appropriate responses.
On the business side, AI can improve demand forecasting and manage price fluctuations to maximize revenue, assist with regulatory compliance processes, and drive supply-chain efficiency.
“Across all these applications, having access to the results of AI analysing huge datasets collected in real time empowers operators to act before business performance, safety, and environmental barriers starts to degrade,” said Eriksson.
Machine learning in non-destructive testing
As an example of its role in optimizing predictive maintenance, ML is being piloted in the industry to enhance the efficiency and precision of non-destructive testing (NDT).
Such testing is vital for diagnosing potential faults that could lead to equipment and system failures. Testing is the foundation of safety and maintenance planning and regimes that underpin asset risk management.
“ML shows promise in this application. But operators need proof that ML algorithms are fit for purpose, perform reliably, and that the underlying ML models are being updated to reflect the changing operational conditions common in oil and gas operations,” noted Eriksson.
Machine learning for mooring line reliability
To take a more specific example, mooring line failure can have potentially catastrophic consequences, but physical tension sensors can be difficult and costly to maintain.
ML is a more accurate and less costly method for anomaly detection, structural integrity assessment, and virtual sensors.
DNV has shown that in more than 99% of simulated test cases, the company’s ML algorithm could identify accurately the condition of offshore oil installation mooring lines.
AI in digital twins
The use of digital twins – simulation-based models of assets such as an oil platform or gas pipeline network – is now familiar within the industry.
Inputting data from databases into such models, and displaying the results on dashboards, is increasing as asset owners use these ‘twins’ to solve day-to-day challenges. The twins combine decision support, information on the status of assets, and computational models.
‘Digital twin’ means different things to different people and organizations. However, efforts are underway in the UK to achieve common definitions, standards, and governance of twins to support the vision of creating a national ecosystem of connected twins giving a 360° view of energy systems in real or near-real time. This vision is most advanced in the collaborative work being caried out in the UK around hydrogen transport by pipeline.
“AI can be a great enabler of this vision, but connectivity between digital twins within and beyond an organization’s own boundaries naturally raises questions about security, including cybersecurity, safety, privacy, commercial confidentiality, and so on,” said Eriksson.
AI comes with new risks
These questions underline how AI/ML comes with risks that could have financial, legal, and reputational consequences.
AI/ML fed with incomplete, inaccurate, old or limited data can generate erroneous insights leading to flawed decisions that potentially compromise safety, productivity, and profitability. It may misidentify or fail to detect rare anomalies that it has not been trained to look for.
There is also cybersecurity risk as connecting AI/ML to critical control systems and data sources creates interfaces that are potential attack surfaces for malicious hackers.
Regulatory risk is a factor when the design or use of AI could contravene laws on privacy, transparency, cybersecurity, safety, availability of critical infrastructure, and other issues legislated for by lawmakers. Additionally, if regulation is not updated to reflect evolution in AI/ML, innovation can be stifled.
Managing AI risk
It follows that data quality management is essential for managing AI risk. Validation and verification of AI/ML products and services are also vital in the rapidly growing and diversifying market targeted by international suppliers and providers. Cybersecurity monitoring, testing, and updating is required.
Regulation of AI, including cybersecurity aspects, is in place in the European Union through the 2024 EU AI Act, and is emerging in the UK, US, China and other jurisdictions. Financial and other penalties are included in current and pending regulations for AI/ML. Managing regulatory risk involves taking the steps needed to ensure compliance within specified deadlines.
There is a training issue to be managed too, said Eriksson: “Designing, implementing, and maintaining AI/ML requires professionals who understand not only the AI/ML technology but the detailed functioning of the oil and gas systems within which it is being used and is integrated. There is a global shortage of this combination of skills. Also, people making decisions based on AI/ML output need to know the limits of what the technology can accurately and reliably tell them, rather than seeing it as a magical black box that can always be relied on to provide trustworthy output.”
Creating trust in AI
The current absence of formal best practice examples for AI use in the oil and gas industry need not hold back innovation. DNV recommended practices (RPs) provide guidance for organizations and industries including oil and gas throughout the digitalization journey.
The RPs cover procurement, development, and operation of AI and other digital solutions. They include the recently published DNV-RP-0671 AI-enabled systems assurance, the world’s first RP on how to ensure the trustworthiness of AI-enabled industrial systems.
The company’s roles and services related to digital trust include, among others, risk assessments; training and self-assessment; verification and validation; and testing.
Developing an AI strategy and robust solutions
As the capabilities and use of AI continue to grow and evolve, front-runner companies in energy industries are integrating ‘AI Strategy’ into their top-level management frameworks alongside Digitalization Strategy and Data Strategy.
The top 20 global oil and gas producers all have AI strategies across their upstream, downstream and, where relevant, midstream businesses.7
DNV can assist customers to confidently start using AI, by assuring the quality of the various digital building blocks in the tools organizations select. Examples include, among others, sensors, data quality, models, and the algorithms and training data used in AI tools. DNV can also help customers to build quality controls into AI applications to ensure compliance with the latest regulatory requirements as these evolve.
Learn more about DNV's AI services
References
1 ‘AI in Oil and Gas Market Size & Share Analysis - Growth Trends & Forecasts (2024 - 2029)’, Mordor Intelligence, mordorintelligence.com
2 ‘Transforming through uncertainty: Energy Industry Insights 2024’, www.dnv.com, April 2024
3 ‘ISO/IEC 22989:2022 Information technology Artificial intelligence Artificial intelligence concepts and terminology’
4 KUANG, Lichun & Liu, he & REN, Yili & LUO, Kai & SHI, Mingyu & SU, Jian & LI, Xin. (2021). Application and development trend of artificial intelligence in petroleum exploration and development. Petroleum Exploration and Development. 48. 1-14. 10.1016/S1876-3804(21)60001-0
5 Waqar, Ahsan & Othman, Idris & Shafiq, Nasir & Mansoor, Muhammad. (2023). Applications of AI in oil and gas projects towards sustainable development: a systematic literature review. Artificial Intelligence Review. 56. 10.1007/s10462-023-10467-7
6 Hussain, Muhammad & Zhang, Tieling & Nasser, Minnat. (2023). Adoption of big data analytics for energy pipeline condition assessment. International Journal of Pressure Vessels and Piping. 10.1016/j.ijpvp.2023.105061
7 ‘How Multibillion Dollar Investments in AI are Driving Oil and Gas Sector Innovation’, Gaurav Sharm, forbes.com, 14 August 2023
10/1/2024 6:00:00 AM