Directed Energy Deposition (DED) Additive Manufacturing process for metallic parts are becoming increasingly popular and widely accepted due to their potential of fabricating parts of large dimensions. The complex thermal cycles obtained due to the process physics results in accumulation of residual stress and distortion. However, to accurately model metal deposition heat transfer for large parts, numerical model leads to impractical computation time. In this work, a 3D transient finite element model with Quiet/Active element activation is developed for modeling metal deposition heat transfer analysis of DED process. To accurately model moving heat source, Goldak’s double ellipsoid model is implemented with small enough simulation time increment such that laser moves a distance of its radius over the course of each increment. Considering thin build-wall of Stainless Steel 316L fabricated with different process parameters, numerical results obtained with COMSOL 5.6 Multi-Physics software are successfully validated with experiment temperature data recorded at the substrate during the fabrication of 20 layers. To reduce the computation time, elongated ellipsoid heat input model that averages the heat source over its entire path is implemented. It has been found that by taking such large time increments, numerical model gives inaccurate results. Therefore, the track is divided into several sub-tracks, each of which is applied in one simulation increment. In this work, an investigation is done to find out the correct simulation time increment or sub-track size that leads to reduction in computation time (5–10 times) but still yields sufficiently accurate results (below 10% of relative error on temperature). Also, a Correction factor is introduced that further reduces computation error of elongated heat source. Finally, a new correlation is also established in finding out the correct time increment size and correction factor value to reduce the computation time yielding accurate results.

Thermally-induced distortion and residual stresses in parts fabricated by the additive manufacturing (AM) process can lead to part rejection and failure. Still, the understanding of thermo–mechanical behavior induced due to the process physics in AM process is a complex task that depends upon process and material parameters. In this work, a 3D thermo-elasto-plastic model is proposed to predict the thermo–mechanical behavior (thermal and distortion field) in the laser-directed energy deposition (LDED) process using the finite element method (FEM). The predicted thermo–mechanical responses are compared to stainless steel 316L (SS 316L) deposition, with single and double bead 42-layer wall samples subject to different inter-layer dwell times, which govern the thermal response of deposited parts in LDED. In this work, the inter-layer dwell times used in experiments vary from 0 to 10 s. Based on past research into the LDED process, it is assumed that fusion and thermal cycle-induced annealing leads to stress relaxation in the material, and is accounted for in the model by instantaneously removing stresses beyond an inversely calibrated relaxation temperature. The model predicts that, for SS 316L, an increase in dwell time leads to a decrease in in situ and post-process distortion values. Moreover, increasing the number of beads leads to an increase in in situ and post-process distortion values. The calibrated numerical model’s predictions are accurate when compared with in situ and post-process experimental measurements. Finally, an elongated ellipsoid heat source model is proposed to speed up the simulation.

The possibility of large-scale part fabrication is the biggest novelty factor associated with Directed Energy Deposition (DED) Additive Manufacturing (AM) technology. However, issues like deformation and residual stresses in the fabricated part originated from DED process physics are still hindering the possibility of large-scale part fabrication. To overcome these bottlenecks, a DED process simulation that predicts the thermo-mechanical response of the material/workpiece can be a useful tool. There are some conventional simulation techniques that are employed commonly for other technologies like welding or Powder Bed Fusion (PBF). But using the same simulation methodologies for the DED process will lead to impractical computation time or inaccurate results. Hence, in the present work, an efficient simulation methodology dedicated to DED is proposed. The proposed model reduces the computation time drastically and also keeps the desired computation accuracy levels. An equivalent heat source is employed that efficiently models the material deposition along with the programmed deposition strategy. The inclusion of deposition strategy in the efficient model is very important for model accuracy, as deposition strategy plays a critical role in the thermo-mechanical response of the deposited material. The proposed model is developed and implemented in COMSOL Multiphysics. With a cantilever tooling, multiple Stainless Steel 316L (SS 316L) thin wall builds of 50- and 100-layers high is fabricated. Numerical results predicted with the efficient model are successfully compared with experimental data such as thermocouple’s in-situ temperature recordings and Laser Displacement Sensor’s in-situ distortion recordings at the substrate during the fabrication of 50- and 100-layers wall. The efficient model successfully captures the thermo-mechanical response of the sample. It also correctly predicts the effect of the number of layers on the accumulation of distortion during and after the material deposition.

Modélisation thermomécanique performante de grandes pièces fabriquées par procédés de fabrication additive laser par dépôt de poudre.: Les procédés de fabrication additive laser par dépôt de poudre offrent une opportunité unique pour la fabrication de grandes pièces à géométrie complexe. Cependant, les déformations mécaniques induites par ces procédés entrainent des défauts pouvant conduire à des pièces rebutées. Au cours de cette thèse, différents modèles ont donc été développés pour mieux comprendre l’apparition de ces déformations en fonction des paramètres opératoires. Un premier modèle thermomécanique prédit le comportement élastoplastique lors de la construction d’un mur en acier inoxydable 316L. L’apport de chaleur est modélisé par une source double ellipsoïdale mobile et la construction des couches se fait à l’aide d’une méthode hybride « Quiet/Active élément ». Un écrouissage isotrope non linéaire est considéré, avec prise en compte de la restauration d’écrouissage à hautes températures. Afin de réduire drastiquement les temps de calcul, une nouvelle source de chaleur est proposée utilisant une source ellipsoïdale allongée qui moyenne l’énergie sur un intervalle d’espace et de temps. Cependant, un intervalle d’espace trop grand diminue la précision du modèle. De nouveaux paramètres sont alors introduits afin d’identifier le meilleur compromis entre temps de calcul et précision. L’ensemble des modèles proposés est confronté avec succès avec des données expérimentales en termes de température et déplacement et ce pour différents paramètres opératoires. Enfin, des modèles multi-échelles basés l’activation par couche ou les méthodes de déformations inhérentes sont étudiés en vue de réduire les temps de calcul.

A novel approach to modeling and simulating multiple reflections of a laser beam in keyhole mode welding is developed based on a revised version of the conventional ray-tracing method. The ray velocity is artificially reduced to ensure compatibility with commercial finite element solvers used for thermohydrodynamic simulations. A dedicated coupling strategy is then implemented between ray-based geometrical optics and the level-set method. This integration allows both ray-tracing and level-set methods to operate within the same software environment and the same temporal solver, eliminating the need for external programming and facilitating the simulation of beam trapping phenomena in a Eulerian framework. The model dynamically solves heat transfer and fluid flow equations in solid, liquid, and gas phases, enabling a comprehensive description of both keyhole formation and melt pool behavior. Applications to laser welding of aluminum, in both static and fusion line scenarios, demonstrate promising agreement with theoretical predictions and experimental observations in terms of melt pool dimensions for different laser powers and welding speeds. The numerical results of the 3D model show its capacity to predict porosity formation.