ABSTRACT
A general framework for the risk analysis of production units of a power station is proposed using statistical methods and Bayesian networks. Input data consist of failure rates specified on the basis of inspections, measurements and expert judgements. Considered economic consequences include outages of key technological devices; social consequences cover potential injuries or fatalities. Risks are predicted using regression analyses of time series of selected parameters describing the actual states of key devices. Proposed procedures are applied to a selected production unit of a fossil power station. The proposed framework provides valuable information about individual devices and their components with regard to economic and social consequences. Methodology, intentionally simplified for operational applications, includes important factors affecting risks of production units. It appears that Bayesian networks represent a transparent method for probabilistic risk analysis of complex technological systems. Results of analyses can be updated when additional information is available.
