Accelerating implementation of condition-based maintenance with artificial intelligence

Sammanfattning

ABSTRACT Purpose – This thesis investigates how organizations can accelerate the implementation of Condition-Based Maintenance (CBM) and effectively leverage Artificial Intelligence (AI) inindustrial maintenance contexts. The study aims to identify critical enablers and barriersacross technological, organizational, and human dimensions, and to propose a maturity-basedframework to guide firms through the staged adoption of AI-assisted CBM. Method – The research is based on a qualitative, explorative case study conducted incooperation with company Alphas’ liquified natural gas (LNG) site. Nineteen semi-structured interviews were held with professionals from both the case company and externalorganizations. Data was analyzed thematically with People, Process and Technologydimensions in focus, to identify maturity patterns and implementation dynamics. Findings – The study identifies four maturity stages of CBM implementation: CBMpreparation, Organizational CBM engagement, Introducing intelligent CBM, and Scalingintelligent CBM. Each stage is characterized by specific technical requirements,organizational structures, and cultural conditions. Key findings include the importance ofdigital literacy, leadership involvement, cross-functional collaboration, and trust in AIoutputs. The study also highlights the shift from rule-based systems to conversational,generative AI tools and the resulting impact on roles, workflows, and governance. Implications – The proposed maturity framework offers practical guidance for firms seekingto implement or scale AI-assisted CBM. It enables managers to assess organizationalreadiness, align CBM efforts with strategic goals, and design interventions tailored to theircurrent maturity level. The research also contributes to the theoretical understanding of digitaltransformation in asset-heavy industries. Limitations and future research – The findings are primarily based on a single in-depth casestudy within the energy sector, which may limit generalizability. Future research couldvalidate the maturity model across different industries and explore quantitative methods tomeasure CBM readiness and performance outcomes. Additionally, the integration of largelanguage models in CBM warrants further exploration.