Predicting the Fatigue Resistance of Welds

Abstract

Because the fatigue resistance of welds is generally less than that of the members they join, it has been extensively studied through tests on full-size weldmentsan expensive and time-consuming procedure. Munse (I), Gurney (2), and Sanders (3) have reviewed the results of the fatigue testing of weldments and have identified the main variables influencing their fatigue resistance. Extensive bibliographies of the literature relating to the fatigue of weldments have been compiled (4-6). Munse and Sanders have each created computerized weldment fatigue data banks (7, 8). Others, such as the British Welding Institute, may also have similar resources. The fatigue data obtained by testing often exhibit significant scatter­ ing, two major sources of which are uncertainty as to the actual state of stress in the weldment and seemingly small variations in weld geometry. To reveal the causes of this scatter, to interrelate the parameters that influence the fatigue life of welds, and to accurately predict weldment fatigue resistance, an analytical model for estimating the total fatigue life of welds has been developed (9). It assumes that the total fatigue life of a weldment (NT) is composed of a fatigue crack initiation period (NJ) and a fatigue crack propagation period (Np) such that