Artificial Intelligence could unlock efficiency and safety gains for oil and gas sector

Nevertheless, as cases and regulations around AI evolve, energy companies are beginning to find more ways to use AI. Companies are collaborating with AI startups and Big Tech companies to run pilot projects evaluating the practicality of using AI in their day-to-day business as well as exploring changes to their risk management frameworks to address technological and regulatory risks arising from the adoption of AI.

We believe that AI, which collectively includes machine learning (ML), robotics, and natural language processing (NLP), has the potential to add value to the oil and gas sector in three broad ways. Firstly, it could significantly reduce the time for various processes by automating key workflows. Secondly, it could improve the accuracy of key forecasts by using the latest tools for predictive analytics. And thirdly, it could aid, in real time, the monitoring of assets, allowing companies to pre-empt potential system failures.

Efficiency of key processes, such as seismic interpretation, reservoir modelling, and asset maintenance, can be increased as ML models are trained using a company's historical data. Predictive analytics, process automation, and reservoir engineering have been the most explored AI use cases by US oil and gas companies in 2024.

Potential challenges to AI adoption include (1) the increased threat of cyberattacks, which are likely to intensify given growing geopolitical risks; (2) competition for talent with the requisite skills as oil and gas companies compete against Big Tech companies and technology start-ups; and (3) the investment required to strengthen IT infrastructure and implement effective skill development programs.

AI has the potential to improve the efficiency and speed of seismic interpretation, a process used to create subsurface images to determine optimal drilling locations and improve the odds of finding oil and gas reserves. Quality control of traditional data collection methods requires frequent manual interventions and is time consuming. ML and AI algorithms can automate the workflows and facilitate real-time tracking of errors (called noise) based on large data sets of previous surveys to increase speed and accuracy. Furthermore, large language models can create more reliable 3D seismic subsurface images, which can be interpreted effectively by AI algorithms.

Seismic data, collected faster and more accurately, are then integrated with company's historical drilling data to identify drilling locations with more precision, increasing operating efficiency, and reducing exploration costs. For example, Shell (Morningstar DBRS Issuer rating of AA (low)) is collaborating with Big Data analytics firms for seismic interpretation. The company expects that the technology has the potential to reduce the exploration duration from nine months to less than nine days.

Similar to seismic interpretation, AI and ML algorithms can add value for upstream companies by increasing the speed and accuracy of reservoir modelling processes. Reservoir modelling is a key step in understanding fluid flow characteristics from reservoir rocks and is crucial for economic feasibility assessment and production forecasts. The models also provide inputs for enhanced oil recovery projects and reservoir management.

To predict reservoir behaviour under different conditions, simulation models are usually based on various static and dynamic parameters such as porosity, rock type, pressure, temperature, and fluid properties. Traditional simulation methods rely on multiple field assumptions and incomplete data, leading to inaccuracies in the simulation models. Moreover, integrating geological, geophysical, and historical production data to develop forecasts with existing software is often time consuming. However, using sensors to capture real-time data coupled with advanced cloud computing capabilities can provide more accurate and higher-quality inputs for simulation. Training ML models on this data under the supervision of field engineers can considerably accelerate the time period for the data collection, filtering, and processing phase. AI can perform predictive analytics on real-time parameters of online reservoirs to recommend a course of action to optimise production.

BP (Morningstar DBRS Issuer Rating of "A") has invested in an AI startup that has built a cloud-based geoscience platform wherein BP engineers are able to feed geological, geophysical, and historical production data. The platform then uses neural networks to perform rapid simulations for reservoir modelling and monitoring asset health. BP is aiming for a 90% reduction in time for data collection, interpretation, and simulation processes.

Field machinery, such as drilling equipment, compressors (for gas processing), and heat exchangers, is maintained on a predetermined schedule to monitor asset health and to prevent faults. Traditional maintenance methods are often reactive and can lead to unplanned outages and increased operational costs. Predictive maintenance, however, is the practice of using an asset's historical operational data to anticipate problems in advance and to reduce instances of equipment failure and overall downtime. The advent of technologies such as the Internet of Things (IoT), Big Data, AI, and cloud computing have made these techniques more feasible and scalable by automating the process and enhancing predictive accuracy. IoT sensors are able to collect real-time data points, such as temperature, pressure, and corrosion rates, which are then stored in a centralised database called a data lake. Subsequently, ML models are trained on each equipment's data base, and predictive analytics is used to forecast the probability of failure and estimate the useful life of assets. Advancements in natural language processing can further aid field engineers to interact with AI models and monitor asset integrity in real time.

For example, optical fibres are installed in pipelines, which aid in the real-time collection of operational and emissions data. Drones are used to collect visual data (photos, videos) from pipelines. Subsequently, AI systems are able to integrate different forms of unstructured data, including weather patterns, to identify potential leakage points. For predictive insights, Equinor (not rated by Morningstar DBRS) has developed a solution called Omnia Prevent that has over 520 ML models analyzing data from more than 18 000 sensors from over 30 oil and gas installations.

Oil and gas companies often face safety risks due to operational hazards, such as leaks of inflammable material, which could result in explosions, the risk of fatalities, and considerable damage to assets. Untrained employees, ineffectively maintained equipment (leading to corrosion), and poor communication protocols during an emergency are major causes for incidents. In addition to real-time monitoring and predictive maintenance, AI can be used to train employees to respond to dangerous scenarios. AI, in combination with virtual and augmented reality, can be used to create simulations to train employees based on root causes of historical incidents. Furthermore, with advancements in Robotics Process Automation, human involvement can be reduced in the handling of hazardous chemicals during transportation and in emergency response situations.

For example, Chevron (Morningstar DBRS Issuer Rating of AA) is deploying intelligent machines to reduce the exposure of human tank inspectors in hazardous environments. Robots, equipped with advanced sensors and HD live cameras, are being deployed across all US facilities of Chevron to increase the precision of tank bottom inspections, reducing downtime and reducing safety risks.

Although AI is in the early stages of rapid adoption, oil and gas companies are identifying risks related to its use. Shell, in its 2023 annual report, identified AI as an area of emerging risk. As the regulatory landscape around the ethical use of AI evolves, it can be expected that regulatory compliance costs will increase as AI adoption gathers pace.

Increased interconnection of systems has increased cyber security risks for all industries, particularly for the oil and gas sector because of the geopolitical uncertainty. According to Bloomberg, hackers targeted 21 US LNG producers in mid-February 2022 in the run up to the conflict in Ukraine. Oil and gas companies will have to invest in advanced cybersecurity technologies and continue to develop training programs for employees to safeguard against emerging threats.

As witnessed during the Colonial pipeline ransomware attack in 2021, cyberattacks can lead to system shutdowns lasting for several days. We believe that such risks potentially have a material impact on a company's cash flow generation and overall financial condition.

A major expected challenge will be attracting skilled talent to work in the oil and gas sector. Prospective employees trained in data science, programming, and statistics have shown a preference for working for technology start-ups and other Big Tech companies. Moreover, employment preferences of the younger generation have shifted to companies with a lower carbon footprint. These factors could slow the adoption of AI in the oil and gas sector and put upward pressure on labour expenses.

Read the article online at: https://www.worldpipelines.com/special-reports/23092024/artificial-intelligence-could-unlock-efficiency-and-safety-gains-for-oil-and-gas-sector/