[en] Metal Additive Manufacturing (MAM) offers many advantages such as fast product manufacturing,
nearly zero material waste, prototyping of complex large parts and the automatization of the manufacturing process in the aerospace, automotive and other sectors. In the MAM, several parameters influence the product creation steps, making the MAM challenging. In this thesis, we modelize and control the deposition process for a type of MAM where a laser beam is used to melt a metallic wire to create the metal parts called the Laser Wire Additive Manufacturing Process (LWAM). In the dissertation, first, a novel parametric modeling approach is created. The goal of this approach is to use parametric product design features to simulate and print 3D metallic objects for the LWAM. The proposed method includes a pattern and the robot toolpath creation while considering several process requirements of LWAM, such as the deposition sequences and the robot system. This technique aims to develop adaptive robot toolpaths for a precise deposition process with nearly zero error in the product creation process. Second, a layer geometry (width and height) prediction model to improve deposition accuracy is proposed. A machine learning regression algorithm is applied to several experimental data to predict the bead geometry across layers. Furthermore, a neural network-based approach was used to study the influence of different deposition parameters, namely laser power, wire-feed rate and travel speed on bead geometry. The experimental results shows that the model has an error rate of (i.e., 2∼4%). Third, a physics-based model of the bead geometry including known process parameters and material properties was created. The model developed for the first time includes critical process parameters, the material properties and the thermal history to describe the relationship between the layer height with different process inputs (i.e., the power, the standoff distance, the temperature, the wire-feed rate and the travel speed). The numerical results show a match of the model with the experimental measurements. Finally, a Model Predictive Controller (MPC) was designed to keep the layer height trajectory constant, considering the constraints and the operating range of the parameters of the process inputs. The model simulation result shows an acceptable tracking of the reference height.
Interreg V-A Grande Région
Fabrication Additive par Dépôt de Fil (FAFil) project (Ref. 3477)