Finite Element Analysis of Flaw Growth History

Finite element models of curved wide plate (CWP) samples were used to generate a family of load-deformation curves, where each curve corresponds to a flaw with a constant depth. This family of curves was then compared to the test results to find the flaw depth corresponding to each load step using two techniques. One technique is based on mapping the crack mouth opening displacement (CMOD) response obtained from finite element analysis (FEA) and tests, while the other one is based on FEA and experimental unloading compliance data. Both the CMOD mapping and unloading compliance techniques were applied to six CWP specimens and the results from the two techniques were compared. The CWP specimens included flaws either at the centreline of the girth weld or at the heat affected zone (HAZ). Nominal flaw sizes were 3 or 5 mm deep by 25, 50 or 75 mm long. For all specimens, testing continued until either maximum load was reached or specimen rupture occurred. Failure strain, defined as the remote strain at peak load, ranged from 1.1% to 4.1%. The flaw growth history curves from the CMOD mapping and unloading compliance techniques for a given specimen were generally found to be in close agreement. The prediction of flaw growth at failure for the specimens with flaw in the weld was closer to the experimental flaw growth at failure than for the specimens with flaw in the HAZ. The average FEA to test ratio of the flaw growth at failure for these two groups of specimens was 0.95 and 0.6, respectively. Additional analyses were carried out to study the effect of HAZ softening and the shape of the input stress strain curves beyond the onset of necking.