Xin Wang

Accurately resolving transient temperature fields in laser powder bed fusion (LPBF) is hindered by stiff, multiscale dynamics arising from localized moving heat sources, temperature-dependent thermophysical properties, and phase-change effects—conditions that impede standard Physics-Informed Neural Network (PINN) convergence. We propose a Progressive Time-Window PINN framework that advances solutions through overlapping temporal windows aligned with the scan trajectory. A causal moving-center collocation strategy concentrates training points around the laser position and its thermal wake, ensuring stable state propagation while mitigating gradient conflicts. Auxiliary neural networks represent temperature-dependent conductivity, specific heat, and density with smooth differentiability; fusion and vaporization are modeled via Gaussian-smoothed effective heat capacity, and convection–radiation Robin conditions enforce surface heat losses. Validation against a FEniCS finite element baseline across six parametric cases spanning varied scan speeds and laser powers demonstrates robust convergence and a highly accurate approximation of the finite element baseline for both point-wise thermal histories and full-field distributions. Post-training inference enables millisecond-scale evaluation, enabling rapid exploration of LPBF process parameters.

Accurate prediction of transient melt pool thermal fields in Laser Powder Bed Fusion (LPBF) is essential for understanding melt pool geometry and defect formation mechanisms, yet conventional finite element methods (FEM) impose prohibitive computational costs for parametric process exploration. A variational physics-informed neural network (VPINN) framework is presented for 3D transient thermal modeling of a GH3536 single-track LPBF scan. The framework incorporates a continuously differentiable Goldak double-ellipsoid moving heat source, temperature-dependent thermophysical property surrogates, and an effective heat-capacity treatment of latent heat associated with solid–liquid phase change and vaporization. These components are embedded in a weak-form residual-minimization scheme with octree-adaptive domain decomposition, hierarchical Legendre test functions, and sequential sliding-window time marching. Effective absorptivity is inferred jointly with the network parameters, using sparse experimental melt pool profiles as supervision. Within a parametric study covering laser powers from 100 to 140 W and scan speeds from 1000 to 1500 mm/s, the predicted melt pool width, depth, and aspect ratio agree closely with FEM benchmarks and cross-sectional optical micrograph measurements across both supervised and held-out interpolation conditions, with total relative L₂ nodal temperature errors ranging from 3.23% to 6.75%. Following a one-time offline training investment of 15,323 s that simultaneously resolves the full parametric space, surrogate inference reduces per-condition query time from 3000–4000 s (FEM) to merely 4–5 s, delivering a speedup of two to three orders of magnitude and making the framework increasingly cost-effective for high-throughput parametric studies and digital-twin integration as the number of queried conditions grows.

With the aim to improve cladding coating quality and prevent cracking, this paper established an extreme high-speed laser cladding thermo-mechanical coupling simulation model to study the evolution of the temperature field and the residual stress distribution. Process parameters that impacted the macroscopic morphology of single-pass coatings were investigated. Numerical calculations and temperature field simulations were performed based on the process parameter data to validate the effects of the temperature gradient and cooling rate on the coating structure and the residual stress distribution. The results showed that a good coating quality could be achieved using a laser power of 2400 W, a cladding speed of 20 m/min, and a powder feeding rate of 20.32 g/min. The coatings’ cross-sectional morphology corresponded well with the temperature distribution predicted by the numerical modeling of the melt pool. The microstructure of the molten coatings was affected by the temperature gradient and the cooling rate, which varied greatly from the bottom to the middle to the top. Maximum residual stress appeared between the bonding region of the coatings and the substrate, and the coatings themselves had significant residual stress in the form of tensile strains, that were mostly distributed in the direction of the laser cladding.

With the aim to improve cladding coating quality and prevent cracking, this paper established an extreme high-speed laser cladding thermo-mechanical coupling simulation model to study the evolution of the temperature field and residual stress distribution. Process parameters that impacted the macroscopic morphology of single-pass coatings were investigated. Numerical calculations and temperature field simulations were performed based on the process parameter data to validate the effect of the temperature gradient and cooling rate on the coating structure and the residual stress distribution. The results showed that a good coating quality could be achieved using a laser power of 2400 W, a cladding rate of 20 m/min, and a powder feeding rate of 20.32 g/min. The coatings’ cross-sectional morphology corresponded well with the temperature distribution predicted by numerical modeling of the melt pool. The microstructure of the molten coatings is affected by the temperature gradient and cooling rate, which vary greatly from the bottom to the middle to the top. Maximum residual stress appears between the bonding region of the coatings and the substrate, and the coatings themselves have significant residual stress in the form of tensile strains that are mostly distributed in the direction of the laser cladding speed.