Mixed-Integer Approach to the Optimization of Trajectories in Laser Powder Bed Fusion Additive Manufacturing
Please login to view abstract download link
Due to the uneven heat distribution created during Laser Powder Bed Fusion Additive Manufacturing (LPBF), thermal stress during the production can lead to significant warping after cooling. As the material type is usually fixed for a given part and the substrate height is fixed by the machine, a good way to reduce thermal stress and thus warpage is to optimize the scanning strategy. In fact, it has been shown that a simple heuristic on the scanning strategy can lead to a reduction of thermal stress of up to one third compared to a standard strategy. We use a mathematical model combining the continuous heat-equation with discrete decisions on the path of the laser in each layer. Here, the continuous heat equation is approximated with a finite differences scheme and combined with the combinatorial constraints to form one joint mixed-integer linear programming problem. As warping is difficult to model and predict, we use several approaches to approximate this effect indirectly, such as computing heat gradients and the temperature in the printing layer. These approximations are used to generate different objective functions for the mixed-integer linear program. Furthermore, we analyze the performance of different finite difference schemes and compare the results. A distinct advantage to such an approach is the flexibility of the model: It is relatively simple to add further combinatorial constraints if this is necessitated by the shape of the product. Furthermore, on can easily adapt to changes in the manufacturing process, e.g., multiple lasers and restrictions resulting from the chamber gas flow. We present an approach which aims to compute mathematically optimal solutions for the printing order problem using a mixed-integer formulation with an integrated temperature distribution model. We discuss different methods to compute the temperature distribution and present different objective functions for optimization. The computed printing orders are used in a refined FEM simulation, which gives a good picture of the temperature distribution during printing and an evaluation of the thermal stress and warpage. Furthermore, the computed laser trajectories are compared with with standard-practice trajectories.