The fatigue behavior of aluminum-zinc-magnesium (Al-Zn-Mg) alloy fillet-welded joints was analyzed in fracture mechanics terms. Basic crack propagation data were obtained with −2 ⩽ R ⩽ + 0.5 and correlated using formulas in the literature and, more successfully, in terms of ΔKeff, based on the results of crack closure experiments. The form of the da/dN versus ΔK relationship was influenced by the specimen geometry. A fracture mechanics analysis of the fatigue life of Al-Zn-Mg alloy fillet welds based on the da/dN versus ΔKeff relationship indicated that the weld toe was less severe from the fatigue viewpoint than the same region in a steel fillet weld. This was compatible with the fact that metallurgical examination of Al-Zn-Mg alloy fillet welds has failed to reveal toe defects similar in magnitude to those which act as fatigue crack initiators at the toes of steel fillet welds. The analysis showed that the fatigue life obtained from the Al-Zn-Mg alloy weld could be predicted on the basis that defects only one tenth the size of those observed in steel were present. Fatigue failure from the weld root in a cruciform joint was also analyzed and the optimum weld design, which gives an equal chance of failure from the root and toe, was determined. The analysis was supported by fatigue test results. Comparison with results obtained for steel added confirmation to the finding that if toe defects are present in Al-Zn-Mg alloy welds, they are smaller than those in steel.

The premature fracture of a jack body during pressure testing drew attention to the need to examine the integrity of the body in the presence of defects, from the point of view of fracture and fatigue. and to reexamine the adequacy of existing inspection measures. This assessment was based on fracture mechanics. and the methods used and conclusions drawn are described in the present paper.

Fillet welds have a low fatigue resistance compared with other types of weld detail, and it is difficult to design structures without using them. Results are presented of screening tests on methods of life improvement that have been applied successfully to fillet welds in steels under fluctuating tensile loading and are applicable to heat-treatable aluminum alloys. The welding method or shielding gas used did not affect fatigue life: (a) improvement of the weld profile during or after welding resulted in slightly increased lives; (b) local peening treatments had to be sufficiently intense to produce compressive residual stresses some depth below the surface to cause a worthwhile increase in life; and (c) specimen overloading treatments applied both initially and periodically generally were effective but the results for exclusion of the atmosphere by a conventional paint system were inconsistent. Results of tests on as-welded specimens, at various stress ratios which include compression in the loading cycle, corresponded with the results of treatments which include compressive residual stresses. The number of tests of each method was small, and further investigation of the more promising methods is recommended before they can be applied generally without qualifying tests.