The level and distribution of residual stresses in welds arises from the complex thermo-mechanical history of heat flow and thermal expansion at very high temperatures. It is not possible to make assessments of these with the methods that are used to determine service stresses. Simulation techniques have been developed over many years making it increasingly possible to predict residual stresses. These models need accurate materials data including, where applicable, the effect of phase transformations. In nuclear reactor pressure vessel welds, it is necessary to consider welding as a metallurgical problem as well as a thermo-mechanical one and FE simulations of these require a wide range of material data in order to create suitable input parameters.
It is crucial that models of ferritic steel welds simulate the effects of phase transformations because the different phases have different thermal expansion coefficients. Partly due to differences in thermal expansion coefficient attributed to the different phases, but more significantly because of the associated transformation strain and transformation plasticity. Further to this, predicting the distribution of the phase fractions enables structural simulations to account for the distribution of mechanical properties throughout a weld. In this work, a simplified approach to producing an empirical model to simulate phase transformations in SA-508 Gr3 pressure vessel steel is presented. A commercial finite element package is used to implement the model which calculates the volume fraction of bainite, martensite and austenite and the thermal strains that evolve over the thermal excursions. The results of these FE simulations are compared to experimental data.