U. A. Korde, M. A. Langerman, V. D. Kalanovic, South Dakota School of Mines and Technology, Rapid City, SD, Algeria

Summary:

Laser additive manufacturing has significant advantages over conventional manufacturing processes that could prove important for reusable, reconfigurable aerospace structures. These advantages, however, are offset to an extent by the need for operator oversight during the process and by the build up of internal thermal stresses in both the part and the substrate. There is therefore a need for model-based feedforward process planning algorithms that would enable easier implementation and make the overall process more amenable to automation.

In this paper, we examine a feedforward planning method for near-uniform heat distribution over a part during build up. This is expected in turn to reduce residual stress/deformations during cooling. The current feedforward plan is carried out layer by layer, and the process being developed will apply to any geometry amenable to layer by layer construction. However, this work focuses on thin-walled builds. The part geometry comprising a layer is thought to be divided into a number of geometric “primitives”, which are individual geometric units that can be arranged together in a sequence to build a layer. Through most of this research, the goal has been to ensure uniform temperature distribution within a primitive at primitive completion. While the laser traverses continuously within each primitive, laser travel may be discontinuous from primitive to primitive. In other words, the system may deposit a primitive at a particular location on a layer, turn off laser and powder, lift the head, and move it to a different location to deposit another primitive there. While LPD operators prefer continuous trajectories, a strong case could be made for discontinuous trajectories that lead to improved material/mechanical properties of the build.

The overall feedforward specification strategy includes two components: (i) part-level strategy, and (ii) primitivelevel strategy. The part-level strategy is concerned with determining the optimal primitive deposition sequence (order and location of successive primitives) over a layer. The primitive-level strategy is concerned with determining the optimal laser parameter (e.g. power, speed, powder feed, etc.) variation within a primitive. The current primitive-level strategy is based on a straightforward inversion of a discretized lumped-capacitance thermal model. Recent simulation results based on this work are discussed in this paper. The part level strategy provides an optimal sequence of primitives over a layer. In this paper, we outline an initial part-level strategy based on a genetic algorithm. Implications of the current results for complex part geometries are discussed. Additional initial experimental results with an Nd:YAG laser deposition system are also outlined. 1