An inverse identification method for automatic estimation of heat source model parameters for laser directed energy deposition

To properly simulate the laser directed energy deposition (L-DED) process, the heat source model is crucial as it represents the absorbed heat energy, which controls the thermal phenomena associated with L-DED. Goldak’s double-ellipsoidal heat source model has been broadly used to represent energy distribution in a wide range of laser-based simulation processes. It is therefore important to accurately determine Goldak’s model parameters for correct L-DED simulation. In this work, an original inverse identification procedure was proposed to determine Goldak’s heat source parameters by coupling a 3D finite element thermal model with a genetic algorithm. The multi-objective Non-dominated Sorting Genetic Algorithm II (NSGA-II) was used by considering two objective functions simultaneously. The temperatures at specific points in the melt pool were extracted from the 3D finite element simulations and then used to construct objective functions for the optimization procedure. By minimizing the differences between the simulated and reference temperatures, Goldak’s heat source parameters were identified based on experimental L-DED fabricated single-track deposits for different process conditions. The model accurately estimated the penetration depth and slightly overestimated the heat-affected zone (HAZ). The relative errors are reliable enough to reasonably predict Goldak’s heat source model parameters. The inverse identification procedure proposed in this study can help reduce time and experimental costs to estimate heat source parameters for the simulation of the L-DED process.