No abstract

The discussion about the reliability in NDE -which is just being. to become hot in Europe, -is focused around the uestion how to guarantee the right functioning of a non iestructive testing system which is composed of several devices and personnel and of a hysical method. The deterministic ap roach" relies on tie proper definition of parameters whic! ensure the defect detection from a physical oint of view. The "probabilistic a proach" pro oses $e determination of the robability of getection an% the probabilit of false c&s via the statistical evaluation of a numier of blind trials. The authors of this paper suggest not to stick on one of these a proaches, but to combine them in a reasonable manner in 8 performin a physical method by computer modelling, ii) experience about the probability of detection by evaf%tEi ex eriments and iii) raising the reliability by combining diferent methods and international expert networking ES ecially the usefulness of the statistical assessment of N8E reliability will be investigated in the paper -showing that it is only of value for a valid conclusion in case the statistical basis is provided and all the circumstances are under control. Examples from ultrasonic hand testing and digitized radiographic weld testing will be presented.forecast and optimization of the performance of a t e Introduction Similar as in medical diagnosis the correctness and the reliability of the "diagnosis result" in industrial nondestructive testing ,(NDT) is of high importance. In case of faults in nuclear power plants or air crafts this is true not only for the health of one person but is a problem of public safety.