Manual walking inspections are a federally mandated and operationally crucial component of railway track safety system. Despite generating extensive longitudinal records of observed defects, the analytical use of walking inspection data for risk-informed scheduling decisions has been limited across both transit and freight rails. Our current work, conducted under prior CoE funding, developed a data-driven framework that transforms multi-year, segment-level walking inspection records into interpretable cumulative risk indicators and uses the slope of cumulative effective risk per mile to drive dynamic inspection scheduling. The framework has been validated across two structurally distinct case studies: Metro Rail Observation Data from an urban transit network and Freight rail’s FRA Defects Data. Both applications confirm that the slope-based scheduling logic produces inspection schedules consistent with observed risk accumulation behavior, ensures optimal resource allocation: reduces surveillance effort in low-activity periods, and concentrates inspection resources during documented periods of elevated defect activity. The framework helps mitigate risk in track with reduced inspection efforts.
The proposed Year 3 research extends this framework in two directions identified during the current study as the most consequential limitations and opportunities. First, the framework currently operates as a retrospective characterization tool: it identifies periods of elevated accumulation after the fact but does not probabilistically project how risk will evolve in subsequent months. Markov Chain Monte Carlo (MCMC) simulation will be used to model segment-level risk state transitions, enabling the framework to generate forward-looking inspection frequency recommendations grounded in both historical behavior and probabilistic deterioration forecasts. Second, the current framework relies exclusively on walking inspection records. Railway track safety assurance in practice involves multiple inspection modalities operating in parallel, including automated geometry measurement and ultrasonic testing. Integrating walking inspection risk indicators with automated inspection outputs will allow the framework to produce joint resource allocation recommendations that account for what each inspection modality observes and what it misses, moving toward a unified inspection planning model. Together, these extensions advance the framework from a descriptive retrospective tool toward a prospective, operationally deployable decision-support system with direct applicability across freight and transit contexts.