Leveraging Artificial Intelligence for Prioritizing and Streamlining Maintenance Backlogs

Abstract Objectives/Scope The safety of industrial facilities relies heavily on maintenance, yet planned tasks may face delays due to factors like surging work demands or insufficient resources. This results in a backlog of overdue activities, posing operational, environmental, and safety risks to assets. Pandemic-induced challenges, shortages of skilled staff, and inflation have exacerbated this situation, compelling us to seek innovative solutions. This suggested methodology prioritizes backlog tasks that improve reliability and eliminates non-value-adding items by combining artificial intelligence and technical expertise. This analysis considers multiple factors including resourcing capacity, aging infrastructure, marginal economics, and competing priorities. Methods, Procedures, Process The suggested methodology in this work combines artificial intelligence, automated data processing, and operations domain knowledge to empower asset owners. By eliminating non-value-adding backlog and providing targeted reliability improvement guidance, we enhance operational efficiency. This innovative methodology integrates natural language processing algorithms and generative artificial intelligence models, with advanced reliability assessment techniques. This methodology transforms raw data into actionable insights at significantly improved pace, accuracy, and consistency. Reliability simulations are utilized to understand the next time a failure may occur. Reliability engineers then use this data to prioritize working on equipment with poor reliability, while work on equipment with good reliability is risk- assessed for cancellation. Done at scale and with pace, our backlog solution considers a range of factors within the assessment of overdue work, including criticality, reliability, consequences of failure, and remaining life of the assets. Results, Observations, Conclusions This methodology enables us to prioritize the most important work by understanding how backlog items affect asset safety and reliability. The strategic focus on value-adding work leads to improved safety, higher production, and lower costs. A recent project utilized this process and examined over 200,000 hours of maintenance backlog tasks. The project eliminated 67,000 hours of maintenance backlog that did not add value, which reduced the operation cost by more than $5M. Utilising artificial intelligence allowed us to finish the project in 8 weeks, from the start of the project to the update of maintenance plans in CMMS. Novel/Additive Information Traditionally, maintenance engineers relied on manual review to prioritize backlog items. However, when faced with hundreds or thousands of activities in the backlog, this task became exceptionally challenging. Our innovative approach, leveraging artificial intelligence and automation, not only streamlines the process but also enhances accuracy and consistency, revolutionizing the landscape of backlog optimization.