As mentioned in our last technical tip the heat input associated with welding and cooling rates of the weld and adjacent parent material causes changes to the material microstructure, which is likely to have normally deleterious effects on local properties and physical performance – particularly as far as fatigue is concerned. These microstructural changes (and the magnitude of the residual stresses) are dependent on the heat input, cooling rates, nature and number of weld passes, inter-pass temperatures, preheat, chemical composition of the material and its thermal diffusivity (density, thermal conductivity and specific heat of the material). In addition, the microstructural changes can also affect the material’s corrosion resistance (which is particularly important in stainless steels where sensitization caused by local depletion in chromium and the formation of surface oxides can occur).
In the parent material, a narrow band of material adjacent to the weld, the so called the heat affected zone (or HAZ), is where these changes are most evident. The size of the HAZ is dependent on heat input, thermal diffusivity and cooling rates and is characterized by a number of regions or zones. In the case of rolled steel plate, i) the material loses hardness and its typical rolled microstructure characterized by elongated grains in the tempered and partially transformed zones furthest from the weld, ii) becomes finer in the recrystallized zone, and then iii) coarser in the grain growth zone adjacent to the high temperature solid/liquid boundary between the HAZ and the weld (which was obviously molten during the welding process and has its own distinct microstructure).
Depending of the chemical composition of the weld and parent plate, the lack of pre-heat prior to welding can lead to rapid local quenching of the weld material and HAZ. If the rate of cooling in steels is of sufficient magnitude it can result in the formation of martensite which when untempered, is hard and brittle. So called ‘hard zone cracking’ may occur as the result of such martensite formation. The absorption of the molecular hydrogen, from moisture in the environment or welding consumables, into the weld material can compound the situation and cause cracking at micro fissures and grain boundaries, which exacerbates brittle cracking behavior. Reducing the cooling rates by proper pre-heating, using dry/baked welding consumables and the use of adequate/appropriate shielding gas can help mitigate these problems.
An additional feature of welding is the development of residual stresses which are often not well considered from a service life perspective. For example if a butt weld between two plates is considered, the residual stress in the center of the weld is highly tensile, reducing to slightly compressive in the HAZ region. Similarly in the longitudinal (welding) direction large, tensile residual stresses are developed in the center of the weld with regions of compression at the start and end of the weld. If not heat treated, or otherwise alleviated, these residual stresses can be of significant magnitude, approaching yield stress values. Indeed, the assumption in many of the assessment codes (such as fracture mechanics based code BS 7910) is that if a weld is not stress relieved the residual stress in this weld approaches the yield stress, and these stresses MUST be taken into account when evaluating the potential for fracture or fatigue. Even when the cyclic stresses are small and fatigue is unlikely, residual stresses can have serious ramifications, resulting in failure by mechanisms such as stress corrosion cracking (SCC).
An additional misunderstood aspect of welding affecting fatigue performance is illustrated by fatigue testing/conventional Stress/Life plots (SN curves). If the results of tensile fatigue tests on longitudinal flat plates (which result in optimum performance from a fatigue perspective) are compared to those in which orthogonal pieces have been welded (such as cruciform members) there is a significant reduction in fatigue life. So-called stiffeners welded onto such plates can also result in significant fatigue life reduction, even when the overall thickness seems higher. These reductions in fatigue performance are due to the stress concentration effects, which override any ostensibly beneficial affect due to stiffeners, or apparent increase in thickness. Even though this reduction in fatigue performance is not directly caused by the welding it is a consequence of welding, because of SCF effects and the integral joint/change in geometry formed by the weld. Such cases must be treated with care and circumspection and it is recommended that appropriate texts be studied to clarify this widely misunderstood phenomenon.
With proper understanding of the fundamentals these issues can largely be mitigated but the importance of careful design and material choice, attention to detail, suitably skilled/qualified welding personnel, the use of suitable welding techniques for the job and the use of appropriate welding procedures cannot be overemphasized if failure is to be prevented.