| Abstract Scope |
Codes and standards approved weld fatigue S-N curves are represented as a series of parallel lines in a log-log plot depending on the loading mode, joint type, plate thickness, etc. Because very high tensile residual stresses are assumed in as-welded weldments, the cyclic load levels are not normally considered for fatigue design and evaluation.
However, the level of residual stress in weldments is not always high in tension. The residual stress level in the weld can vary depending on joint complexity, post-weld treatment (e.g., application of stress relieving or weld improvement techniques), etc. The residual stress level in post-welded treated joints is not high in tension and is usually observed as a low level of tensile or compressive stress.
For joints with weld improvement techniques applied, the residual stresses in welds become compressive and the associated S-N curves are subdivided according to the material yield strength level (e.g. IIW recommendation for high-frequency mechanical impact (HFMI)), and the negative inverse slopes of the S-N curves become much shallower than those for as-welded cases.
However, the S-N curve relationship between as-welded and post-weld treated joints is not clearly defined in codes and standards. Consequently, it is necessary to have a fatigue design and evaluation procedure that can be applied from a high tensile residual stress level to a compressive residual stress level.
This presentation proposes a pragmatically unified design and evaluation procedure incorporating the effects of residual stresses in welded steel joints.
By employing the equivalent structural stress parameter adopted by ASME and API, the fatigue S-N data of welded joints can be consolidated into a small scatter band regardless of the loading mode, joint type, and plate thickness. Therefore, using the equivalent structural stress parameters, a single master S-N curve for welded joints can be constructed for as-welded joints where high tensile stresses are expected.
A modified equivalent structural stress parameter was formulated to incorporate different residual stress levels for fatigue design and evaluation of welded joints. Because the fatigue strength is affected by both the cyclic load level and the residual stress level, an effective stress ratio combined with the cyclic load level and the residual stress level was introduced. The procedures of the common weld improvement techniques become standardized in codes and standards and their residual stress levels can be quantified based on the investigation of well-documented test results. Therefore, the proposed unified procedure is applicable from as-welded joints to post-weld treated joints.
The proposed unified procedure was validated using various weld toe fatigue test results such as as-welded, stress relieved, preloaded, and HFMI treated.
With the modified Equivalent Structural Stress parameter, the post-weld treated fatigue data are mingled well with the as-welded fatigue data, and the post-weld treated weld data falls within the scatter band of the as-welded joints, regardless of the material yield strength and residual stress level as well as loading mode, joint type, and plate thickness. Therefore, the unified procedure generates a weld fatigue single master S-N curve, and its master S-N curve can be used for the design and evaluation of welded joints with various residual stress levels. |