AI in Oil and Gas Supply Chains

The already complex, highly technological, and innovative oil and gas industry faces several enduring challenges: around costs, logistics and geography, public scrutiny, and environmental concerns. In recent years, advanced information technologies are providing potential solutions to many of the industry’s most stubborn efficiency-reducers. According to a report by Accenture, powerful new computing capabilities  – foremost among them AI – could help the industry save up to $1.6 trillion by 2025 (Accenture, 2018). This article discusses how AI can enhance oil industry supply chain operations, and outlines the assumptions underlying these improvements.

Predictive Analytics

This capability constitutes a major affordance of AI. Predictive analytics concerns the application of machine learning (algorithms) to the analysis of data for the prediction and modelling of future events. In the oil industry, predictive analytics can facilitate various essential activities, such as production optimization, downtime minimization, demand prediction, safety improvement, and the monitoring and pre-emptive maintenance of equipment (Alelyani et al., 2021). All offer profound cost and safety advantages. For example, diagnostics predict depletion timings so that replenishment materials can be ordered and prepared, which smooths production and improves machine efficiencies. Similarly, AI, through sensing and detection, can predict dangers ahead of occurrence, allowing crews or control systems to address potential problems before they cause downtime. AI can also maximize oil and gas production flows by, for example, predicting the optimal time to shut down a well for maintenance, or the optimal time to start a new well (Kalas et al., 2021).

Automation

This involves the use of AI in combination with robotics to execute operations that would otherwise be done by humans. Efficiencies arise and cost and human error risk are both reduced; but a far greater advantage is attained: automation of hazardous processes obviates the need for hands-on human activity, reducing the likelihood of physical danger to operatives. AI-driven drones can be used to carry out at-safe-distance visual inspections of pipelines and other features of critical infrastructure, for example. Automation can also be used to optimize logistics: AI algorithms can optimize shipping routes, lower fuel consumption, and minimize downtimes.

Numerous studies have demonstrated the benefits of automation in the oil industry. For example, Jin et al. (2018) showed  the potential of autonomous vehicles in lowering transportation costs and improving the overall efficiency of oil and gas supply chains. Another study, by Afshar et al. (2019), presented the potential of robotics in performing inspections and maintenance tasks, with positive effects on safety and costs.

Sustainability

AI can also increase the general sustainability of the oil industry’s extensive supply chains. For example, by identifying opportunities for fuel or energy reduction or more efficient usages of energy, AI can optimize system-wide energy consumption. (AI can of course also elevate the safety and reliability of renewable energy sources, such as wind and solar, by predicting weather patterns and gearing energy production accordingly.).

Various recent research papers reveal the contributions that AI can improve sustainability across the energy industry. For instance, a study by Zhang et al. (2021) demonstrated that AI can determine the optimum location and operation patterns of wind turbines, leading to increased energy efficiency and reduced costs. Another study by Chen et al. (2019) demonstrated the potential of AI in predicting solar energy production, improving reliability and reducing costs. In similar ways, it is likely that AI can contribute to the simplification and optimization of oil and gas infrastructure design, location, management, controls, procedures, capacity and flow management, and operations.

The successful implementation of AI in the oil industry supply chain will require concerted investment, experimentation, iteration, and new patterns in human-machine collaboration. The results will however rapidly offset the costs.