Local Approach of Fracture in the Ductile Regime and Application to VVER Materials

Abstract

Within the TACIS R2.06/96 project: “Surveillance Program for VVER 1000 Reactors”, sponsored by the European Commission, the local approach of fracture has been applied in the ductile regime. Two different models were applied and compared, namely Tvergaard-Needleman-Gurson versus Prometey model. The main tasks are: • perform special Local Approach experiments on smooth and notched cylindrical specimens; • predict JR-curve on the basis of the ductile fracture models; • compare two models of ductile fracture, namely, the Tvergaard-Needleman-Gurson model and the Prometey model. In this paper, the Tvergaard-Needleman-Gurson and Prometey models are briefly described. The parameters of both models were calibrated by using experimental data obtained on tensile specimens. While only smooth tensile specimens are used to calibrate the Tvergaard-Needleman-Gurson model, notched tensile in addition to smooth tensile specimens are used to calibrate the Prometey model. In the latter, standard smooth tensile specimens are used to determine the mechanical properties (the yield stress σy, the ultimate stress σu, the ultimate elongation δu, the area reduction Z) and notched cylindrical specimens to determine the strain at rupture. The numerical analysis comprises essentially two steps: • Step 1: finite element simulation of the smooth tensile specimen (determination of true stress-strain curve and critical void volume fraction for the Tvergaard-Needleman-Gurson model) and simulation of the notched cylindrical specimen up to rupture (determination of stress triaxiality for the Prometey model); • Step 2: finite element simulation of the 2T CT specimen and determination of the crack resistance behaviour in the ductile regime (J-Δa curve). It is found that both models were able to correctly predict the crack resistance behaviour of the investigated materials. The numerical and the experimental results were in very good agreement. The main difference between the two models is that the required number of calibrated parameters in the Prometey model is less than in the Tvergaard-Needleman-Gurson model but additional tests on notched specimens are required for the Prometey model.