Ashley Woodbridge, Field CTO Infrastructure Solutions Group – Lenovo META, says that for the oil and gas sector AI is not about technology for technology’s sake.
From computer vision systems that keep employees safe at wellheads, to algorithms which prevent costly plant shutdowns, AI is reshaping the way the oil and gas sector works.
Computer vision in particular is emerging as an essential tool for operators in the industry to improve safety and private LLMs are helping to analyse huge amounts of data and democratise access to information within organisations. Other predictive AI techniques are helping operators to avoid costly repairs and plant shutdowns.
The companies that are using AI successfully are those who start by looking at a business problem rather than organisations who deploy AI for its own sake – with forward-thinking organisations in the sector already seeing measurable business value from AI in everything from employee safety to exploration.
For many companies, computer vision is already seen as transformative to operations and new use cases keep emerging.
Analytic models have long been widely used in predictive maintenance in the oil and gas sector, but computer vision is transforming how this is delivered, offering a 360-degree view of what is happening at a site.
For example, a company can opt to use computer vision to monitor whether people are wearing protective equipment at various operational stages in the oilfield and then act rapidly if needed. It’s mature technology – an extension of what many companies were doing during COVID with mask detection. You get immediate results, but once you have that infrastructure there, it’s a net-zero investment to add more use cases and use the cameras to detect other issues. You can detect more things at a much lower incremental cost, because you are not having to add sensors.
In the oil and gas sector, hazardous environments such as well pads, drilling rigs and other high-risk sites often pose challenges for deploying IoT sensors due to strict safety requirements and high costs. To overcome these limitations, computer vision enabled cameras placed outside these zones can provide predictive monitoring from a safe distance. Other important uses of computer vision in the sector include corrosion identification, leak detection, gauge surveillance and brownfield site safety monitoring.
Machine learning and AI innovator nybl, part of Lenovo’s AI Innovator program, is already using AI to great effect in the oil and gas sector, with an AI capability built for oil wells which utilizes real-time operational data like vibration, temperature, pressure and other metrics. By integrating this data, nybl’s AI helps predict potential failure, reduce downtime by up to 97%, increasing productivity by 15-20% and extending the life of wellhead assets by up to 30%.
Among other tools like n.lift, designed for Electric Submersible Pumps (ESPs), nybl is also building models capable of deep computer vision analytics around human behaviours and are researching around how computer vision can be used in upstream oil field services in the future.
In the exploration stage, AI can also have a powerful impact. When identifying oil and gas prospects, organisations have to deal with enormous amounts of seismic and geological data. High-performance computing and AI algorithms can make sense of this data, identifying oil prospects rapidly and accurately.
When it comes to operating plants, AI also has multiple uses, from event prediction to production optimisation, using techniques such as pattern recognition to predict outcomes. Combining multiple AI techniques such as predictive simulations and closed-loop optimisation can help operators to boost efficiency, by adjusting equipment parameters in real time and optimising pumping rates.
AI can also help prevent plant shutdowns, which can cost enormous amounts of money. Global energy giant Woodside Energy is using AI algorithms combined with thousands of sensors to detect and prevent foaming incidents at its Pluto Liquid Natural Gas plant in Western Australia. Foaming incidents require the plant to be shut down. One incident reportedly cost Woodside $300 million in lost revenues, so the company added an AI system that can detect the early signs of foaming up to four days in advance. A cloud IoT platform ingests data from 10,000 sensors within and around the plant’s acid gas removal units, looking for the early signs of foaming. The system offers clear warnings of a foaming event long before it happens, meaning that the plant can adjust operations or perform planned maintenance, rather than losing revenue. Woodside now plans to expand the system to five other onshore and offshore facilities and vessels.
LLMs are also beginning to find uses in the sector.
Previously a workflow might be that an oil producer would want to understand what might happen if they increased output: they would go to a business analyst who would give it to a data scientist and get a one-time report with the results. Now that the models can write and run code, you can ask a question using conversational language – it writes the code, runs it and gives you your report immediately.
LLMs can also provide an easy way to access information such as repair manuals, offering a way to put information into technicians’ hands at drilling sites.
Companies in the oil and gas space are routinely dealing with highly private data, so they opt for private LLMs tailored to the industry, which can help streamline workflows across the organisations, allowing workers to automate reporting and democratise access to insights between business units. Customers retrain models to be very specific, to only get data from reputable sources, so they get accuracy and ultimate value.
Organisations that are successful with AI in the oil and gas sector are those that are focused on the tangible.
For the oil and gas sector, AI is not about technology for technology’s sake: it is about improving efficiency, improving safety and delivering long-term value.
