Effect of thermal modeling assumptions on prediction accuracy and computational efficiency in laser-arc hybrid welding simulation

Laser-arc hybrid welding (LAHW) imposes steep, localized thermal gradients demanding computationally expensive thermo-elastic-plastic (TEP) finite element analysis. To quantify the trade-off between physical fidelity and computational efficiency, three TEP models of LAHW on a 300 mm × 500 mm × 15 mm SBHS400 butt joint are compared using identical mesh discretization. Model 1 (high-fidelity baseline) incorporates a combined cylindrical–double-ellipsoid heat source via a FORTRAN DFLUX subroutine, a 3-mm intersource offset, latent heat, and surface radiation. Model 2 retains the heat source and offset but omits latent heat and radiation. Model 3 adopts a morphology-mapped equivalent volumetric heat source (MME-VHS), replacing the analytical formulation with a GUI-based uniform heat flux mapped from the experimental fusion zone macrograph, requiring no user subroutines. Models are validated against thermocouple histories, out-of-plane distortion, and X-ray diffraction (XRD) residual stress measurements. Deactivating latent heat and radiation (M1 → M2) reduces wall clock time by 64% (57.3 h → 20.5 h), confirming these thermal nonlinearities as the primary computational cost driver. The MME-VHS achieves a further reduction to 17.5 h (69% relative to model 1) while maintaining distortion accuracy within 1.0% of the experimental mean and reproducing macroscopic residual stress distributions with good fidelity. A stress smearing effect is identified: Uniform heat flux redistributes plastic strain over a broader cross section, reducing the compressive trough at the weld centerline while marginally elevating the tensile peak at the weld toe, confined to the fusion zone vicinity. Quantitative model selection guidance is provided for efficient TEP simulation of LAHW in medium-thick structural steel.