Data-Driven Prioritization of Activities in Preventive Maintenance Planning

Maintenance planning aims to ensure a high operating performance of industrial assets by the effective use of financial and human resources, and spare parts. Although, methods for industrial maintenance management have advanced in the last decades, the technical complexity of involved human activities and the maintenance cost still pose challenges. Maintenance resources in a manufacturing system are often limited, and consequently, the engineer has to prioritize maintenance activities. The Failure Mode and Effective Analysis (FMEA) is a qualitative risk assessment tool often used for maintenance prioritization. Its main shortcoming is that it is subjective. That motivated our research in data-driven preventive maintenance planning based on clustering. The historical failure records with similar effects on availability and reliability of the machine, and the cost incurred by the performed maintenance activities are grouped together. The created clusters are ranked using a multicriteria decision-making method TOPSIS. The engineer prioritizes maintenance activities and constructs a preventive maintenance plan starting with the cluster with the highest rank. We applied and validated the proposed methodology using a case study from a food company in Brazil, which provided us with failure records collected over a 5-year period. The main benefit of the proposed maintenance prioritization is that it enables an informed trade-off between the availability and reliability of the machine and the cost of maintenance. Additionally, different type of failures in a cluster may be prevented with the same maintenance activity, which leads to reduction of the preventive maintenance cost