Evaluation of human factors and man-machine problems in the safety of nuclear power plants

This paper has the aim of reviewing and discussing some of the key issues in the domain of Man-Machine Interaction (MMI), with particular reference to the techniques currently adopted for the analysis and evaluation of human factors in the safety of Nuclear Power Plants (NPP). Therefore, the most known methodologies used in mechanistic PRA analysis for modelling human errors are firstly described. Then, the need of a more complete and deterministic representation of the operator's reasoning and motor behaviour is debated, showing how this represents a fundamental complement to the merely probabilistic quantification of performances for approaching correctly the overall problem of MMI. Finally, a cognitive model, based on a hierarchical goal-oriented approach and driven by fuzzy logic methodology, currently being developed within the JRC Ispra, is briefly presented and discussed as the counterpart to the "classical" methodologies for studying human errors, identifying limits and boundaries of applications of both types of approaches.