Thermo-mechanical-fluid coupled modeling for residual stress prediction in thick plate welding

Welding thick plates induces more complex residual stress distributions than thin plates, with traditional thermo-mechanical model struggling to accurately predict the impact of molten pool flow and weld bead morphology. This study proposes a novel thermal fluid mechanical model (CFD-FEM) by integrating the computational fluid dynamics (CFD) model and the finite element method (FEM) model, considering molten pool flow. The gas-metal interface is captured employing the Volume of Fluid (VOF) method, enabling accurate extraction of the solidified weld bead morphology for each pass in thick multi-pass welding. This morphology is then utilized in the FEM model to reconstruct weld geometry for residual stress prediction. For double-V thick-plate butt welds, the CFD-FEM model, validated by blind-hole tests, reduces Root Mean Square Error (RMSE) by 47.77 MPa and Mean Absolute Percentage Error (MAPE) by 27.22 % compared with traditional model. The results demonstrate that convex weld bead formation is primarily driven by fluid momentum, surface tension gradients, and the Marangoni effect. The alternating welding strategy effectively reduces surface residual stress. By incorporating the cladding effect and providing more realistic weld bead morphology, the CFD-FEM model predicts residual stress of 478.43 MPa at adjacent weld junctions, compared with 243.26 MPa for the traditional model, better capturing local stress concentrations at the weld interface.