Bayesian Statistical Inference in Reliability Analysis of Power Generation Equipment at Nigerian Power Plants

This study applies Bayesian statistical inference methods to reliability analysis of power generation equipment at thermal power plants in Nigeria, addressing the critical problem of unplanned equipment failures that have contributed to persistently low electricity generation capacity utilisation in the country's power sector. Nigerian grid-connected power plants have operated well below their installed generation capacity over the past decade, with a significant portion of the generation shortfall attributable to equipment breakdowns that Bayesian reliability modelling can help predict and prevent. Equipment failure records covering turbines, generators, transformers, and auxiliary systems at three major thermal power stations, including plants in Delta, Kogi, and Kaduna states, are obtained under research access agreements covering maintenance records for the period 2016 to 2023. Weibull failure time models are estimated within a Bayesian framework using Markov chain Monte Carlo sampling implemented in R via the Stan probabilistic programming language, with prior distributions specified from equipment manufacturer reliability specifications and international industry benchmarks. Posterior predictive distributions of component failure times are derived, and Bayesian posterior predictive checks are used to validate model adequacy. Reliability functions, hazard rate functions, and mean time between failures are computed with associated posterior credible intervals. Results identify three turbine components with significantly elevated failure rates, providing a quantitative basis for targeted preventive maintenance scheduling. Keywords: Bayesian inference, reliability analysis, power generation, Weibull distribution, Markov chain Monte Carlo