Validation of a numerical model used to predict phase distribution and residual stress in ferritic steel weldments

• We validate a numerical model to simulate the welding of ferritic steel components.
• We use a solid-state phase transformation (SSPT) algorithm to predict SSPT kinetics.
• Post-weld phase predictions are validated using micro-hardness measurements.
• Post-weld residual stress predictions are validated using neutron diffraction data.
• Residual stress predictions are found to be dependent on transformation plasticity.

Numerical finite element analyses were combined with experimental observation of a single-pass autogenous beam weld in SA508 Gr.3 Cl.1 ferritic steel. Two weldment sets were prepared using different weld heat inputs, resulting in different post-weld residual stress and ferritic phase distributions. Neutron diffraction was employed to measure the residual stress distribution while microhardness measurements were used to measure the post-weld phase distribution in each weldment. In both cases, the numerical model accurately predicts the ferritic phase distribution and residual stress field. Model predictions illustrate how the higher cooling rates associated with a faster torch speed result in an increased martensite volume fraction within the weldment. Consideration of both the transformation kinetics and transformation plasticity are proven to significantly improve model accuracy when comparing measured and predicted residual stress profiles.