Welding processes induce residual stresses and distortion in the welded joint and the connected components. For manufacturing purpose distortion is the main issue and up to now the problem is handled by post weld corrective actions. Welding residual stress fields are not considered at the design stage in French codes and standards. However, it is well known that residual stresses are likely to increase the risks of fatigue or corrosion and may cause failure in brittle materials. Ferritic parts of large components are post-weld heat treated; allowing disregarding the influence of residuals stresses thanks to their relief. Preventive measures, including mitigation by fine polishing are undertaken in corrosion sensitive zones. The influence of residual stresses on fatigue is more complex to analyze: in low cycle fatigue, residual stresses should be relieved or redistributed after few cycles with plastic straining, and for high cycle fatigue, residual stress effects are accounted for through a mean stress offsett.
When considered, residual stress fields are often represented in a very crude manner by a membrane distribution of the most influent stress component through the thickness of the structure. In a less rough way, several codes or fitness-for-purpose guidelines (API [1], British standards [2]) propose residual stress profiles relative to several weld configurations. Nevertheless for a given case, the given profiles may differ significantly for several reasons: the degree of conservatism, the number of covered cases, the embedded margins accounting for uncertainties.
Some ill-posed benchmark problems have shown that numerical simulation of residual stresses may deliver very scattered results. AREVA has therefore developed a methodology to validate welding simulations. The scope is limited to fusion welding. The simulations are based on a Thermo-Metallurgical Mechanical model in which the welding energy is represented by an equivalent heat source.
This paper presents the actual state of development of this methodology which will be illustrated through 4 examples of residual fields in Dissimilar Metal Welds. Residual stress measurements have been performed for each of the four mock-ups by different techniques. Based on this important experimental and numerical campaign some actions of improvement of the validation methodology are finally listed.