Comparative assessment of project FALSIRE—results

Abstract

The analysis results of the recently completed Phase I for the Project Fracture Analysis of Large Scale International Reference Experiments (FALSIRE) are summarized in a comparative manner. Thirty-nine analyses of the pressurized thermal shock experiments NKS-3 and NKS-4 from MPA-Stuttgart (FRG), PTSE-2 from ORNL (USA) and spinning cylinder SC-I and SC-II from AEA Technology (UK) were evaluated. Discussion of the results has focused on the discrepancies of the finite element results and on comparisons with the estimation scheme analyses. A set of quantities such as crack mouth opening (CMOD), strains, stresses, J-integral and constraint have been selected and compared for the different analyses to approximate the structural behaviour of the test specimens and the fracture behaviour of the cracks. A database of the results has been established. The influence of boundary conditions, approximation of material properties and calculational methods is shown in detail. The structure mechanics behaviour of the test specimens could be approximated well for the NKS experiments but not for PTSE-2. Most differences between the various analyses could be explained. In SC tests structural mechanics results could not be compared with experimental measures. The application of J R methodology to predict crack extension was partially successful in some cases (NKS experiments) but not in others (PTSE-2). The quality of fracture assessment is closely connected with the structural mechanics simulation. In all analyses with good structural mechanics approximation, the fracture prediction was reasonable. Fracture assessments based on CT-specimens overestimate stable crack growth in the case of NKS-4 and SC-I/II, because the crack resistance in the large-scale test specimens is greater than predicted by small specimens (e.g. CT-25). SC-I/II fracture results show that crack growth can be described quite well with the J-integral and the J R -curves of the large-scale test specimen. Therefore, future work has to be concentrated on extension of the J R methodology by a parameter which controls the geometry and load dependence of the crack resistance. This can only be achieved by close connection between numerical simulation and fracture mechanics testing.