Dissimilar metal welds between different grades of ferritic steels or between ferritic steel and austenitic nickel alloys are used extensively in power plants. When such weldments are exposed to high temperature conditions, as might be found in service in a thermal power plant, local microstructural evolution will occur. This is due to diffusion, driven by chemical potential gradients, of solute atoms. Such diffusion can cause major changes in hardness and mechanical properties of joints and can lead to the formation of embrittling phases and/or softened zones. This can potentially lead to premature component failure by, for example, high temperature creep.
Whilst finite element modelling of mechanical behavior and damage evolution is well established this is not the case for chemical diffusion and microstructural evolution at weld interfaces. In the present study, the general purpose linked thermodynamic and kinetic software packages Thermo-Calc and DICTRA have been applied to simulate chemical diffusion and precipitation/dissolution (i.e. phase fraction evolution) in dissimilar weld joints using commercially available thermodynamic databases TCFE7 and TTNI6. Two approaches for modelling multiphase, multicomponent systems using this software will be presented and discussed and their implementation will be illustrated.
The paper will present results on modelling a range of dissimilar metal interfaces of both the ferritic-ferritic type and the ferritic-austenitic type (for example, grade 22 to grade 91 steel and grade 22 to Inconel 625). Ferritic-ferritic case studies will compare model predictions with a number of previously published experimental studies and it will be shown that the current approach can give good quantitative agreement in terms of carbon composition profiles and carbide depleted/carbide enriched zones. The results obtained from modelling a grade 22 steel-Inconel 625 system where the crystal structure of the matrix is different on either side of the weld will be compared with experimental observations on a weld overlaid tube component. The experimental results will include scanning and transmission electron microscopy studies of the weld interface regions and it will be shown that the predictions of diffusion and precipitate formation compare well with observations made experimentally following exposure at 650 °C. Also discussed are the options for further refining the computational model based on empirically observed phenomena, such as the unmixed zone of a weld.