PSI - Issue 13

Mihaela Iordachescu et al. / Procedia Structural Integrity 13 (2018) 554–559 M. Iordachescu, A. Valiente, E. Scutelnicu/ Structural Integrity Procedia 00 (2018) 000–000

557

4

a number of solidified layers of filler metal, the heterogeneities corresponding to the interlayers, and small areas of HAZs. Despite this, the curves indicate that NW and RW have the strength of the filler metal, which exceed that of BM, the difference being significantly higher in the repaired joint. In contrast, the welding effect on the joints ductility is exactly the opposite, as resulting from the maximum uniform elongation comparison (Fig. 2c). Then, fracture toughness tests were performed according to ASTM E-1820 (2015) on fatigue-precracked sets of two CTs containing NW, RW and BM. The specimens were instrumented with a crack-opening displacement (COD) extensometer to determine, by successive partial un-loadings, the compliance changes that accompany the crack size variation. The load-COD data presented in Fig. 2d indicate that ductile fracture occurred during the critical crack propagation through the HAZs of NW and RW. In all cases, the corresponding crack size estimated from the un loading slopes widely exceeded the limits given in the ASTM E-1820 (2015) regarding the specimen capacity to ensure a fracture process governed by the stress field at the crack tip. As this condition is hardly compatible with the crack growth through a fully yielded resistant ligament, a plastic collapse model adapted from Kachanov (2004) for the CT specimen was used as a quantitative basis to assess the effectiveness of the repairing procedure with respect to the overall performance of the NW. 3. Plastic collapse model and fracture analysis The highest strength and lowest ductility obtained by tensile testing cylindrical specimens containing NW and RW are hardening indicators but the ductile crack growth occurred during fracture testing of CTs discards the brittle behavior of the respective HAZs. However, such experimental evidences are not enough to ascertain if the repairing procedure alters the weld capacity to behave as a structural plastic hinge since the fracture tests do not provide a direct quantitative basis to compare the HAZs influence in such behavior. Regarding this issue it is worth noting that a cracked CT allows its two symmetrical halves rotate together by plastic deformation even when these are connecting a peculiar zone, as it is the weld HAZ. The fact that the fracture of the tested CTs takes place along a fully plastic resistant ligament suggests the possibility of quantifying the plastic rotation capacity of the specimen from the fracture data on the basis of a plastic collapse model applied to the resistant ligament. The decreasing size of the resistant ligament due to the ductile propagation of the crack can be taken into account by deriving the crack size from the slope of each elastic unloading made in the fracture test. The plastic collapse load of CT specimens results from different limit load analyses based on the assumptions of plane-strain and ideally rigid-plastic materials (Hu and Albrecht, 1991). All of them postulate the formation of yielded regions symmetric to the resistant ligament whose stress field is known as a function of yield strength. The un-yielded regions of the CT specimen are two equal rigid halves that symmetrically rotate around a point of the resistant ligament. The load and the position of the rotation center are obtained from the balance between the external forces and the stress field at the ligament. For engineering applications, the yield strength of the ideal rigid plastic material is replaced by an intermediate value between the conventional yield strength and the tensile strength of the real material. Most of the fracture testing standards use the mean value between the yield and tensile strength to obtain the plastic collapse load of the CT specimen as a reference value to limit the fatigue precracking loads (ASTM E-1820, 2015). The plastic collapse model here adopted for the CT specimens made of BM, or containing NW or RW, assumes plane stress rather than plane strain for being a hypothesis more adjusted to the real deformation of the specimen (Fig. 6). Fig. 3a shows the plane stress collapse model adapted from that of Kachanov (2004) for a notched flexural bar obeying the von Mises criterion. The plastic yielded region consists of a band of displacement discontinuity that extends from the crack tip to the rotation center along the ligament and a triangular prism that extends from the band up to the CT backside. The stress state is simple compression in the prism and biaxial tensile in the discontinuity band, a principal stress being double than the other. The load F and the depth of the rotation center   x (relative to the W) are obtained from the equilibrium condition of a half specimen. The resulting values, given in terms of yield strength σ Y , relative crack size   a (with respect to W) and specimen thickness B, are:   x  2 3  3  2 3  4    a 2   a (1)