Data uncertainties are present in any engineering calucation. Their impact may be particularly important if they concenr the safety of insutrial installations. Procedures are presentded to deal with such uncertainties. Their propagation through calculations by means of a Monte-Carlo approach is shown. Methods for interpretaing the resulting distributions are presented. Among them figure the use of loss functions and entropy based relative information. Examples from safety studies for both nuclear and chemical plants illustrate practical applications. It is demonstrated that the inclusion of uncertainties provides a more reliable base for decisions and hendce helps to avoid errors.
展开▼