Key points are not available for this paper at this time.
Presented on Thursday 23 May: Session 25 Energy assets demand periodic regular maintenance to ensure safe, reliable and efficient operations. The dynamic nature of operating conditions and evolving best practices necessitate frequent optimisation of maintenance programs. Traditionally, these programs were labour-intensive and costly to deploy, often yielding unclear outcomes due to subjective decision-making. The advent of artificial intelligence (AI) presents an opportunity to challenge traditional approaches. This paper presents a case study where asset knowledge, data, and AI capabilities are leveraged to streamline maintenance optimisation using our ‘maintAI’ approach. The program addresses maintenance strategy, backlog, spares, and predictive maintenance optimisation with a focus on value creation, data-driven decisions, and consistent recommendations. A systematic methodology employs AI to sift through and eliminate non-value-adding tasks, enabling prioritisation of work and enhancing reliability and productivity throughout the production facility lifecycle. AI, including Natural Language Processing and Generative AI algorithms, enhances the speed and accuracy of failure mode classification from operational maintenance data. Reliability modelling techniques provide insights into equipment reliability. Recommendations undergo expert review before integration into a Computerised Maintenance Management System. Implementation of this data-driven approach demonstrates rapid deployment and sustained efficiency, yielding substantial gains in production uptime, cost reduction, and safety. The user-centric design ensures agility and ease of configuration. A recent project, which took only 6 weeks to deliver, led to a ~28% reduction in maintenance backlog, freeing capacity for critical focus areas. The maintAI approach proves a meaningful change for energy producers, offering a new solution in maintenance optimisation for enhanced reliability and productivity. To access the Oral Presentation click the link on the right. To read the full paper click here
Gordon Buchan (Fri,) studied this question.