{ "background": "System reliability assessment in manufacturing is critical for operational efficiency and safety. Current reliability models often fail to adequately account for the heterogeneity and data sparsity common in industrial settings in developing economies, leading to imprecise estimates and suboptimal maintenance decisions. ", "purpose and objectives": "This study presents and evaluates a novel Bayesian hierarchical modelling framework for estimating system reliability in manufacturing plants. The primary objective is to assess the methodological performance of this approach in handling complex, multi-level failure data from heterogeneous machinery. ", "methodology": "The proposed model, \ (\, \), \ \ (\\, \²\), \ \ (\\, \²_\), was applied to failure data from electro-mechanical systems across multiple plants. Model performance was evaluated using posterior predictive checks and comparisons of posterior credible intervals with estimates from non-hierarchical models. ", "findings": "The hierarchical model produced more precise and robust reliability estimates, with 95% credible intervals for mean time between failures (MTBF) being on average 34% narrower than those from standard models. It successfully identified a dominant failure mode in rotating assemblies, accounting for approximately 62% of system downtime. ", "conclusion": "The Bayesian hierarchical framework offers a statistically rigorous method for system reliability analysis, effectively pooling information across similar units to improve inference, particularly where data are limited or uneven. ", "recommendations": "Adoption of this modelling approach is recommended for plant engineers and reliability analysts to enhance maintenance planning. Future work should integrate covariate information to model the influence of operational factors on failure rates. ", "key words": "reliability engineering, Bayesian inference, hierarchical modelling, manufacturing systems, maintenance, probabilistic methods", "contribution statement": "This paper provides a novel, validated methodological framework for reliability assessment that explicitly addresses data
Otieno et al. (Tue,) studied this question.