A thermal-physics-informed machine-learning framework for transferable melt-pool prediction in laser powder bed fusion, trained on single-track scans of 23 pure elemental metals.
| Laser power P | W | |
| Scanning speed V | mm / s | |
| Normalized enthalpy ΔH / hs | — |
| Density ρ | g / cm³ | |
| Melting point Tm | °C | |
| Boiling point Tb | °C | |
| Thermal conductivity λ | W / (m·K) | |
| Specific heat capacity Cp | J / (g·K) | |
| Linear expansion coefficient α | 1 / K | |
| Heat of fusion ΔHf | J / g | |
| Heat of vaporization ΔHv | J / g | |
| Reflection coefficient R | — |
Initializing model…
Schematic cross-section (scaled to predicted geometry):
Diagram is a schematic representation of the predicted melt-pool width and depth; it is not a physical simulation of the melt-pool shape. Predictions for multi-element compositions are extrapolative — the model was trained on pure elemental substrates and should be used as ranking guidance for alloy systems.