This paper presents results from a program of residual stress measurements and modelling carried out for a pipe girth weld of 369 mm outer diameter and 40 mm thickness. The component consisted of two 316 stainless steel pipe sections joined together using a “single-V” nickel base alloy (alloy 82) weld. The residual stresses were measured using the Deep-Hole Drilling (DHD) technique and modelled using ABAQUS. Biaxial, through-thickness residual stresses were measured through the weld centreline at a total of 6 different locations around the component. At three of the measurement locations the DHD process was carried out from the outer surface of the component with the remaining three, one of which coinciding with the weld start/stop position, carried out from the inner surface of the component. The differences in DHD process application (i.e. outer-to-inner or inner-to-outer) was carried out as a sub-objective to investigate the sequence of residual stress relaxation and its influence on the measured results. Good measurement repeatability was found between all locations. The hoop residual stresses were tensile at the outer surface, increasing to a maximum of 350 MPa at 10 mm depth, then decreasing to a minimum of −325 MPa at a depth of 34 mm, before increasing again towards the inner surface. The axial residual stresses were found to be similar in profile to the hoop residual stresses albeit lower in absolute magnitude by roughly 100 MPa. For this component it was found that the hoop residual stresses showed an influence of process direction, whereas for the axial residual stresses no influence was found. The modelling of the residual stresses generated was undertaken using a 2D axisymmetric finite element analysis containing 25 discrete weld beads. Each of the 25 weld beads were analysed sequentially using the following stages: heat source modelling, thermal analysis, elastic-plastic mechanical analysis. The sensitivity of the residual stresses generated with respect to the material hardening model used was investigated (i.e. kinematic, isotropic and mixed mode – kinematic/isotropic). Generally, the isotropic hardening model produces the highest predictions, the kinematic hardening model produces the lowest predictions with the mixed mode model lying in-between. Good agreement was found between the measured and modelled residual stresses. The main discrepancy existed in the hoop direction with the modelled residual stresses being the most tensile by roughly 200 MPa at depths within 15 mm of the outer surface of the pipe.
Ultrasonic inspection of a welded stainless steel pipe to evaluate residual stresses through thickness
