VoxeLogic: A voxel-based, mesh-free model for fast, high-fidelity temperature prediction and process planning in directed energy deposition

Directed Energy Deposition (DED) is increasingly adopted for manufacturing large-scale metal components where conventional methods are impractical. While it offers material efficiency and tailored properties, it also introduces challenges associated with repeated thermal cycles. Even for a single geometry, different decomposition strategies, toolpaths, and process parameters can significantly alter the resulting thermal histories, which in turn govern microstructure and final properties. For process plan optimization, conventional finite element (FEM) models can capture these cycles, but their reliance on meshing, specialized expertise, and long runtimes restrict use in real-world cases. This research aimed to develop a framework that balances computational efficiency with predictive fidelity while remaining suitable for both academic and industrial deployment. To this end, the VoxeLogic Heat Model was developed: a new, mesh-free formulation built from first principles to simulate heat transfer directly from deposition toolpaths, providing complete temperature–time histories across the build. Rooted in physical principles rather than training datasets, VoxeLogic is broadly applicable across geometries and process conditions. Applications include single-layer deposition on flat and curved substrates with complex curvilinear toolpaths. Benchmarking against experimentally validated FEM simulations and thermocouple measurements showed that VoxeLogic reproduced temperature–time profiles with errors below 5% for peak temperatures and 10% for cooling rates. Microstructure-relevant thermal metrics were also captured with accuracy suitable for engineering analysis. This fidelity was achieved while reducing computation time by 99.8%, from hours to seconds. These results also establish VoxeLogic as a foundation for extending voxel-based thermal simulation to multilayer 3D deposition and future digital twin applications.