PSI - Issue 13

S. Lindqvist et al. / Procedia Structural Integrity 13 (2018) 1195–1200 S. Lindqvist/ Structural Integrity Procedia 00 (2018) 000–000

1198

4

During testing, the width and thickness of the specimen reduces, Fig. 4, whereas the length of the specimen increases. For the DMWs, the length of the austenitic side of the specimen increases approximately with 5 % and the ferritic side with 0.5 %. Fig. 3 b) shows that the actual final area of the remaining ligament is smaller than the final area predicted based on the initial thickness, width and final crack length. The difference between these two areas grows with increasing final crack length.The load applied during testing relative to the load of the remaining ligament at flow stress, true tensile and fracture strength was determined as a function of crack length. The results show that the remaining ligament is subjected to a load equal to the true tensile strength, assuming that the load is distributed equally within the remaining ligament which is a rough approximation, Fig. 5 (Graba 2013).

Fig. 4. The fracture surface and the degree of deformation of the cross-section after testing. The red square shows the outer bounds of initial cross-section. 3.2. The effect of constraint on tearing resistance Fig. 6 a) shows that the J 1mm obtained with SE(T) specimens is higher than obtained with SE(B) specimens of the same DMW with a 0 /W ~ 0.55 (Lindqvist et al. 2018). The exponent m for the J-R curve fit appears to be insensitive to the constraint effect and sensitive to crack location, Fig. 6 b). The m-exponents determined for the J-R curves with crack growth less than 1.5 mm are marked with hollow data points (incomplete data). In Fig. 6, the SE(T) results obtained for the 18MND5 were predicted from the SE(B) results based on the constraint correction methodology presented in (Wallin 2011). The constraint (T-stress) difference for SE(T) and SE(B) specimens can be predicted from Fig. 7 a). For positive T-stress values, the J-integral dominates the crack tip field and the differences are not significant. The J 1mm prediction of the SE(T) specimens with a 0 /W = 0.38 and based on the constraint difference of the investigated SE(B) and SE(T) specimens is calculated with the equation presented in Fig. 7 b). The prediction, Fig. 6 a), fits the data obtained with SE(T) specimens. 4. Conclusions 1) For the DMW and the HSLA-steel, the CANMET crack length prediction fits the experimental data within the validity limits of the prediction. After the crack length exceeds a/W = 0.5, the measured compliance starts to deviate from the prediction. 2) The width and the thickness of the SE(T) specimens reduce during the measurement, which was not taken into account in the analysis. 3) The J 1mm prediction accounting for the constraint difference between SE(B) and SE(T) specimens fits the experimental data obtained with SE(T) specimens. The error is of the same order as that for homogeneous materials. Acknowledgements The results presented in this study originate from a European collaboration project called MULTIMETAL funded by the European Commission (EC) within its 7th Framework Program and concluded in January 2015. The writing of this paper was funded by project LOST (long term operational aspects of structural integrity) part of SAFIR2018 (The Finnish Research Programme on Nuclear Power Plant Safety 2015 - 2018).