Vol.26, No.1, 2026, pp. 17–22
https://doi.org/10.69644/ivk-2026-01-0017

The assessment of fossil-fuel power plant reliability is of critical importance not only for operation and maintenance, but also for long-term planning and development of energy systems. The thermal power system examined in this study comprises three subsystems, with the reliability evaluation carried out using a thirteen-year failure database covering the normal service life. The reliability analysis employs several mathematical reliability models, including two-parameter Weibull, normal, lognormal, and simple exponential distributions. Parameters of each distribution are estimated using three independent statistical techniques: the graphical method (based on probability papers), method of moments (MoM), and maximum likelihood estimation (MLE) method. While the two-parameter Weibull distribution estimated by interval-censored MLE represents the most appropriate unified model for reliability analysis in terms of interpretability and decision relevance, regression analyses in this study demonstrates that the Normal distribution estimated by MoM achieves the best overall fit to the experimental data. The obtained results confirm that the application of rigorous probabilistic methods and models can provide a reliable foundation for maintenance optimisation, enhanced safety and strategic management of thermal power systems.

Keywords: • thermal power system • reliability • mathematical distributions • statistical methods