Defects that develop in welds during the fabrication process are frequently manifested as embedded flaws from lack of fusion or lack of penetration. Fracture analyses of welded structures must be able to assess the effect of such defects on the structural integrity of weldments; however, the transferability of R-curves measured in laboratory specimens to defective structural welds has not been fully examined.

In the current study, the fracture behavior of an overmatched butt weld containing a simulated buried, lack-of-penetration defect is studied. A specimen designed to simulate pressure vessel butt welds is considered; namely, a center crack panel specimen, of 1.25 inch by 1.25 inch cross section, loaded in tension. The stress-relieved double-V weld has a yield strength 50% higher than that of the plate material, and displays upper shelf fracture behavior at room temperature. Specimens are precracked or notched, loaded monotonically while load-CMOD measurements are made, then stopped and heat tinted to mark the extent of ductile crack growth.

These measurements are compared to predictions made using finite element analysis of the specimens using the fracture mechanics code Warp3D, which models void growth using the Gurson-Tvergaard dilitant plasticity formulation within fixed sized computational cells ahead of the crack front. Calibration data for the finite element analyses, namely cell size and initial material porosities, are obtained by matching computational predictions to experimental results from tests of welded compact tension specimens. The calibrated methodology is then used to predict ductile crack growth in defective welds in a structural crack geometry. The R-curves measured in compact tension specimens are compared to those obtained in multi-specimen weld tests, and conclusions as to the transferability of R-curves are discussed.