Structural reliability benchmark exercise for primary-circuit components life prediction methods

Results of NDT inspections, crack growth data taken from small specimens, and results of stress calculation together with a load time history are used to predict the growth to failure of defects intentionally introduced into the welds of a 1/5 scale pressure vessel. The prediction results are compared with further NDT results taken during the fatigue test. Analysing the discrepancies between the analytical and the test results, the question, how to define an equivalent crack to each detected defect, is found to be the weakest link in the procedure of crack growth and life prediction. The NDT results differ considerably from one inspection to another and contain only the positions and sizes of the defects but no information on the shape and whether a defect is crack-like or not. Therefore it is necessary to define equivalent cracks with assumed shapes e.g. elliptical, and to take into account an incubation time of crack initiation at the blunt (instead of sharp) edges of the defects. However, NDT gives no information to set up the necessary calculation parameters, especially for the decision, whether a defect of critical size will start to grow immediately when in-service, after a short (or long) incubation time, or never