In a recent webinar on the impact of artificial intelligence (AI) on the oil and gas industry, SparkCognition global sales executive vice president Curt Richtermeyer noted that the oil and gas industry has, over the past decade, made major investments in AI and has identified a wide range of opportunities for operational improvement as a result. According to Gartner, over 39% of oil and gas technology leaders are increasing their AI investment in 2022 and projecting a total spend in excess of $2B.
“93% of companies believe AI will be a pivotal technology to drive growth and innovation in the future.” Deloitte
The AI opportunity for a mature and critical industry
For many years, energy companies have made a practice of funneling vast quantities of performance and status information into databases and other repositories, optimistically hoping that they can then identify applications that will make effective use of the data. In fact, this rarely turns out to be the case, partly due to the sheer immensity of the information and partly because the traditional tools for analyzing such data troves are extremely complex and difficult to implement. In addition, Richtermeyer noted, this data is almost exclusively backward-looking, rendering its predictive value limited at best. AI, on the other hand, through the use of capabilities like machine learning (ML), natural language processing (NLP), and visual AI, allows the user to perform forward-looking projections based on troves of historical data.
With global infrastructure asset values above $100T, and with many of these assets either underperforming or failing outright, the onus is increasingly on industry leaders to identify ways of maximizing the value of these enormous investments. An important complication cited by Richtermeyer is the aging of the existing workforce, a collection of experts who have made careers out of maintaining systems and equipment through the depth of their amassed experience. As the workforce transitions to a new generation, this state of affairs is no longer sustainable. The new generation of workers expects data to be readily available to support their jobs, leaving it to companies to identify ways of capturing the expertise of the prior generation for use going forward.
Prior to exploring numerous real-world case studies, Richtermeyer summarized several ways in which AI has been implemented across the oil and gas value chain:
- Upstream (drilling and exploration)–reducing nonproductive time (NPT), identifying and deterring drilling anomalies, and minimizing inadvertent environmental releases.
- Midstream (transport and storage)–identifying safety and operational issues with pipelines and transport shipping and routing.
- Downstream (refining and retail)–optimizing refining operations and maximizing customer loyalty.
A critical advantage Richtermeyer pointed out in AI’s favor is its ability to evaluate performance at a system level rather than focusing exclusively on individual pieces of equipment. As shown in the above figure, while the reliability figures for individual pieces of equipment are routinely in the high 90s, that of the entire system is frequently as low as the high 70s or low 80s, a result that compromises the performance of the entire system or plant. The goal of AI is to leverage technology across the entire system to drive the overall result to much higher levels while lowering the overall number of alerts so that each alert is meaningful and truly informs operators as to what is going on.
Case studies in AI-powered value creation
Richtermeyer proceeded from this overview to discuss numerous specific case studies in which SparkCognition’s AI/ML solutions have delivered significant business value.
European Oil Supermajor: With applications deployed across numerous offshore platforms and refineries, the firm has made massive investments in AI in its quest to leverage the technology across the entire business. By analyzing data from numerous sensors and systems, they have seen a 98.7% reduction in the number of alerts and are achieving 75-day advance notice of impending equipment failures, resulting in a $30M savings in plant performance.
Large European Energy Company: This firm has analyzed vast amounts of time-series data to predict the performance of large turbines, increasing their reliability and uptime. The solution captures 75% of compressor events with an average of eight days advance notice while reducing the rate of false positives. The company operates one of the largest liquid natural gas (LNG) storage sites on the planet and is currently seeing a significant increase in the number of ships arriving to store gas at their facilities. The equipment required to safely perform this processing and storage (cryopumps, etc.) must be operating correctly at all times and the company has seen a 5.1% increase in annual operating efficiency along with a $600K reduction in annual maintenance cost at its first implementation plant.
Middle Eastern Oil Supermajor: A Middle Eastern supermajor has used AI to optimize its daily maritime shipments of 4M barrels, or 50% of daily capacity. Their implementation of SparkCognition’s solution applies knowledge representation and analytics to available data to optimize the loading and routing of petroleum shipments using computational knowledge graphs, dynamic schedule optimization, simulation and scenario planning, and fleetwide management. This has enabled them to ship over 2B barrels in an optimized manner, resulting in a 98% reduction in labor and a 2021 labor savings of $64M.
Major Oleochemical Company: The company lost nearly $1M in critical assets in a recent major fire despite the presence of numerous smoke detectors on the property. Visual AI was subsequently applied to existing camera feeds to proactively identify hazards, enabling the company to leverage multiple simultaneous use cases, including identification of workers’ personal protective equipment (PPE), appropriate training to work on equipment types, proximity to equipment and to each other, and analysis of the paths being taken as workers move about the facility. This is all accomplished using the CCTV infrastructure the client already had on-site, resulting in less than two seconds’ notice after the start of a fire, preventing a repeat of the critical asset loss suffered previously.
Large Renewable Energy Provider: The utility’s renewable/wind operations suffered from significant yaw misalignment, resulting in suboptimal energy generation and revenue creation. By using AI to continuously evaluate wind turbine yaw angles, even small changes in yaw angle have had a big impact on generation efficiency and revenue. In this case, the result was 96% accuracy in identifying yaw misalignment and a 2% increase in energy production.
Large Asian Petroleum Company: The company used visual AI to monitor retail customer time and behavior at gas pumps and then matched this information with license plates and vehicle types in the company’s CRM system. This allowed the company to track customer business volume and loyalty, as well as help to optimize employee performance, safety, and cleanliness, resulting in significant increases in customer loyalty and revenue.
AI—Pathway to excellence in the oil and gas industry
In summarizing his presentation on the impact of AI in the oil and gas industry, Richtermeyer identified several key takeaways, including increased asset lifetimes, predictive anomaly detection and downtime mitigation, and maximizing a company’s ability to optimize safety, efficiency, and environmental performance. He reiterated that AI is a collection of capabilities, including machine learning, natural language processing, visual analysis, etc. The technology works well whether the data used are structured or unstructured.
The most important aspect of AI is that the system is trained so that its understanding of normality evolves as the systems themselves evolve. This can be achieved in the presence of large quantities of historical data or very sparse data sets, reducing implementation risk and speeding deployment, resulting in faster time to value and greater scalability.