{ "background": "The reliability of railway maintenance depot systems is critical for operational continuity and safety, yet quantitative reliability assessments for such infrastructure in developing contexts are scarce. Existing methods often fail to adequately account for operational heterogeneity and data limitations prevalent in these settings.", "purpose and objectives": "This study aims to develop and apply a novel Bayesian hierarchical modelling framework to assess the reliability of key depot subsystems, specifically focusing on traction and rolling stock maintenance facilities. The objective is to provide a robust, data-informed reliability metric that accommodates sparse and variable operational data.", "methodology": "A Bayesian hierarchical model was constructed, integrating subsystem failure data from multiple depots. The core reliability parameter, the failure rate $\lambda{ij}$ for subsystem $i$ in depot $j$, was modelled as $\lambda{ij} \\sim \\text{Gamma}(\\alphai, \\betai)$, with hyperpriors on $\\alphai$ and $\\betai$ to share information across depots. Posterior distributions were estimated using Markov Chain Monte Carlo simulation.", "findings": "The model quantified substantial variability in subsystem reliability across depots. For traction maintenance systems, the posterior mean failure rate ranged from 0.08 to 0.21 failures per month, with the 95% credible interval for the poorest-performing depot being (0.17, 0.26). This represents a reliability gap of over 150% between the highest and lowest-performing facilities.", "conclusion": "The Bayesian hierarchical model successfully provided a probabilistic reliability assessment, explicitly quantifying uncertainty and variability. It demonstrates that systemic reliability is not uniform and identifies specific subsystems and depots requiring targeted intervention.", "recommendations": "Infrastructure managers should adopt a data-centric, probabilistic approach to reliability monitoring. Investment and maintenance planning should be prioritised based on the quantified reliability gaps and posterior uncertainty intervals identified by the model.", "key words": "Bayesian inference, hierarchical modelling, infrastructure reliability, maintenance engineering, railway systems", "contribution statement": "This paper