PSI - Issue 13

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

1196

2

2. Materials and methods The J-R curves of the two materials were characterised with SE(T) specimens. The width and the thickness of the specimens were 20 mm, and the daylight-length was 200 mm. The target initial crack length was 7.5 mm (a 0 /W = 0.38), and pre-crack length was 2.5 mm. The specimens were side-grooved to a depth of 5 %. The test temperature was 26 °C. The first material was a narrow-gap DMW plate mock-up consisting of ferritic low-alloy steel 18MND5, austenitic stainless steel AISI 316L and Ni-base weld metal Alloy 52. The cracks were located in the HAZ adjacent to the fusion boundary between SA 508 and Alloy 52. The same DMW was characterised in (Lindqvist et al. 2018) with high constraint SE(B) specimens. The same procedure, for extraction of the specimens from the mock-up and determination of the crack location relative to the fusion boundary, was applied in this study. The other material was a heavy plate LASER 420 MC plus HSLA-steel. The cracks were oriented in the longitudinal-transverse (L-T) direction. The specimens were extracted from the plate with an electro-discharge wire cutter. The material has an elastic modulus, E, of 205 GPa and Poisson’s ratio of 0.3. The specimens were fatigue pre-cracked using a resonant testing machine with load ratio R = 0.1 and maximum stress intensity factor was less than 21 MPa√m. For the DMW specimens, the CMOD was measured 2.05 mm from the front face, which does not have a significant effect on the CMOD. The CMOD for the HSLA-steel was measured on the front face. The unloading compliance method was applied. The measurement was displacement controlled, and load controlled during the compliance. The specimens were loaded in a MTS universal servo-hydraulic testing machine equipped with a 250 kN load cell and CMOD was measured with an Epsilon displacement gauge. The DMW specimens were clamped at the front and back surface, and the HSLA-steel specimens were clamped on the sides. The dwell time before unloading was 15 s, and no hysteresis was observed. The compliance was determined from all points belonging to the linear load decrease and increase. The J-R curves were calculated with equations given in the CANMET recommended practice. For the J-R curve analyses, the principles described in ASTM E1820 were followed. For initial and final crack length measurements, the 9-point measuring technique from ASTM E1820 was employed. The crack front curvature for the initial and the Fig. 1 shows the J-R curves for the two materials. The J-R curves are rising and the specimens fail by the ductile fracture mechanism. In Fig. 1 b), the HAZ cracks of the DMW are identified by the distance of the fatigue pre-crack tip to the fusion boundary. The tearing resistance is lowest for cracks with the fatigue pre-crack tip adjacent to the fusion boundary. Some of the DMW SE(T) specimens were subjected to crack back-up/”negative crack growth”. The location of these J-R curves was shifted by determining the blunting line to the linear part after the crack back-up, and moving the curve so that the blunting line crosses the origin. Fig. 2 shows the comparison between the measured and the calculated initial crack length, a 0 and a oq , and crack growth. The difference between the measured, Δa, and calculated, Δa predicted , crack growth is large for the specimen with over 8 mm of crack growth. Otherwise, the differences are small, demonstrating the quality of the applied crack length estimation procedure. The crack length estimate was corrected for rotation and necking. The correction in CANMET was developed for 0.25 < a/W < 0.5 (W is the width), which can explain the difference between prediction and the measurement for longer cracks. For the correction, the flow stress and instantaneous crack length was applied. Fig. 3 a) shows that the prediction of the compliance, C, multiplied with the correction factor, ρ (BCE) , developed by Huang and Zhou (2015) fits the experimental data up to a p /W of 0.55, where a p is the final crack length. The maximum error of the compliance prediction within 0 < a/W < 0.9 is ± 4.2 %. For the DMW data point at a final /W = 0.67, the crack has grown both in the weld metal and the HAZ, Fig. 4 c), which moves the data point to the right, resulting in a good fit. The initial dimensions of the specimen were applied to determine the experimental data points to Fig. 3 a) but the specimen dimensions change as the specimen is loaded. By applying the instantaneous dimensions, the data points in Fig. 3 a) would move to the right and downwards. final crack was under 20 %. 3. Results and discussion 3.1. The J-R curves, the quality of the data and deformation of the remaining ligament