Data Model to Digitization of Criticality Analysis in Railway Systems

This article introduces an innovative perspective for criticality analysis in railway systems by incorporating digitization as an enabler for automated and on-demand assessment of the risk value for each asset. In contrast to previous static approaches, our method leverages digitization to enable dynamic analysis of the criticality of railway components, enhancing decision-making in maintenance management. The proposed approach follows a detailed data Extract, Transform, Load (ETL) process, starting with comprehensive data collection, including maintenance records, asset attributes, and classification. This facilitates the integration of diverse non-standardized data sources, allowing for a digital representation of assets. By characterizing assets through attributes, a quantitative evaluation of criticality is achieved, improving efficiency through an automated process. This characterization not only streamlines the assessment but also enables the application of the model in scenarios of high complexity, volume, and operational variability, such as railway infrastructures with millions of assets. The precise assessment of criticality adapts to changing railway conditions, enabling the organization to identify patterns and generate risk alerts. Criticality scores are integrated with the enterprise asset management system, triggering specific actions in response to significant risk variations. The results demonstrate significant improvements in maintenance resource allocation, including a reduction in interventions for low-criticality assets. Additionally, it enables root cause analysis for high-risk assets, resulting in cost reduction, downtime reduction, and improvement in safety. The data normalization approach through submodels allows adaptation to various railway contexts, providing a comprehensive solution for criticality analysis and other analyses. The applied case study focuses on a section of a metropolitan commuter rail network with over 700,000 assets, illustrating the effectiveness of the proposed approach in practice. Simultaneously, these results have been utilized in applications and research to forecast risks in railway infrastructures, solidifying the value and applicability of the proposed approach.