Computational models for predicting transient temperature distributions, residual stresses, and residual deflections for girth-butt welds are described. Comparisons of predicted and measured temperatures for a two-pass welded pipe show agreement to within 9 percent and 17 percent of the measured values for passes one and two, respectively, the model for predicting residual stresses and residual deflections is based on a finite-element representation recognizing individual passes, temperature dependent elastic-plastic constitutive behavior, elastic unloading for material in the nonlinear stress-strain range, and changes in geometry due to the deformation of each weld pass. Load incrementation and incremental stress-strain relations are also used. Results for a two-pass girth-butt welded pipe show good correlation between residual stresses and residual deflections obtained from the computational model and data obtained from a welded 304 stainless steel pipe.

An elastic-plastic fracture mechanics methodology for treating two-dimensional stable crack growth and instability problems is described. The paper draws on “generation-phase” analyses in which the experimentally observed applied-load (or displacement) stable crack growth behavior is reproduced in a finite-element model. In these calculations a number of candidate stable crack growth parameters are calculated for the material tested. The quality of the predictions that can be made with these parameters is tested with “application-phase” analyses. Here, the finite-element model is used to predict stable crack growth and instability for a different geometry, with a previously evaluated parameter serving as the criterion for stable growth. These analyses are applied to and compared with measurements of crack growth and instability in center-cracked panels and compact tension specimens of the 2219-T87 aluminum alloy and the A533-B grade of steel. The work shows that the crack growth parameters (COA)c, Jc, dJc/da, and the linear elastic fracture mechanics (LEFM)-R, which sample large portions of the elastic-plastic strain field, vary monotonically with stable crack extension. However, the parameters (CTOA)c, R, Go, and Fc, which reflect the state of the crack tip process zone, are essentially independent of the amount of stable growth when the mode of fracture does not change. Useful, stable growth criteria can therefore be evaluated from the crack tip state at the onset of crack extension and do not have to be continuously measured during stable crack growth. The possibility of making accurate predictions for the extent of stable crack growth and the load level at instability is demonstrated using only the value of J c at the onset of crack extension.

In this paper, a computational model is developed to calculate the magnitude and distribution of residual stresses for multipass girth-butt welded pipes. The model consists of two parts, a temperature analysis model and a finite element stress analysis model. Elastic-plastic temperature dependent mechanical properties and unloading due to stress reversals are included in the model. While computational models for predicting residual stress distributions due to one- and two-pass welds have been developed, there are certain difficulties associated with predicting the behavior of multipass welds in an economical way. An approach for handling these types of problems is described. Residual stresses obtained with the model for a seven-pass weld and a thirty-pass weld are compared with residual stresses obtained from laboratory measurements. Good agreement is found between the computed and measured values of residual stresses for for both welds.

Residual stresses in a heat treated weld clad plate and test specimens obtained from the plate are determined using a combination of experimental residual stress analysis and a finite element computational model. The plate is 102 mm thick and made of A 533-B Class 2 steel with 308 stainless steel cladding. The plate is heated to 538 C and allowed to cool uniformly. Upon cooling, residual stresses are set up in the clad plate because of the difference between the coefficients of thermal expansion of the plate and the cladding. Residual stress in the clad plate is determined using both a previously verified experimental residual stress analysis technique and a computational model. Removing test specimens from the clad plate can relax the stresses in the cladding. Thus, residual stress distributions were also determined for two types of clad test specimens that were removed from the plate. These test specimens were designed to examine the effect of cladding thickness on residual stresses. Good agreement was found between the experimentally obtained residual stress values and the residual stresses calculated from the computational model. Because of the interest in tests conducted at elevated temperatures and the inherent difficulty in doing experimental residual stress analysis at elevated temperatures, the computational model was applied to examine the effect of elevated temperature on the residual stresses in the test specimens. Peak stresses in the heat treated clad plate were found to approach the yield stress of the cladding material. It was also found that removing a 32 mm clad specimen with cladding on one side reduced the residual stresses in the cladding. However, the residual stresses in the cladding were found to increase when one-half of the cladding thickness was machined away to form the second test specimen geometry. Residual stresses parallel and perpendicular to the weld direction were very similar in magnitude for all cases considered. The effect that heating the test specimens to 204 C has on residual stress distributions was to reduce the residual stress in the cladding and the plate.

No abstract