ICF10-0384PR ALTERNATIVE METHODS FOR DERIVING AND FITTING J-R CURVES: HOW THEY AFFECT STRUCTURAL INTEGRITY ASSESSMENT

In this paper, results from Reese & Schwalbe’s linear normalization (LN) methodology for deriving J-R curves are compared, related to J-∆a (J-integral-ductile crack growth) data, to those obtained from traditional unloading elastic compliance (UEC) technique. Research results regarding to a nuclear grade steel exhibiting a wide range of elastic-plastic fracture resistance, agree quite well for both techniques until a certain level of toughness of the material. Below this critical level, linear normalization produces too conservative and inconsistent results for sub-sized compact testpieces. Power-law, linear and logarithmic fits were applied to the J-∆a data points within well-known limits of validity of deformation-J (JD). The results were assessed in terms of two typical J-integral criteria of the nuclear industry, namely, the crack initiation J (Ji) and the so-called Paris & Johnson’s J50 for ductile instability of cracks. It was concluded that the logarithmic fit produces conservative values for both Ji and J50 criteria, when compared to power-law, whereas the linear fitting method provides the most non-conservative failure predictions