Trajectory Planning of Non-Gravity Aligned (NGA) segments in Arbitrary Directions for Large Scale Additive Manufacturing of Metals (LSAMM)

Traditionally, slicing and path planning are done along the gravity-aligned direction of a part, causing more complex geometrical shapes to have unsupported overhangs. Wire Arc Additive Manufacturing (WAAM) has typically handled overhangs with a robotic part positioner; but, to extend the current capabilities of LSAMM, a new framework for slicing and building parts out of gravity alignment has been developed. The proposed framework focuses on segmenting more complex geometrical parts into gravity-aligned (GA), non-gravity aligned (NGA), and transition zones to support tool-path generation. GA and NGA segments can be planned with traditional slicing techniques, but the NGA tool-paths must be modified using open loop heuristics to ensure acceptable near net shape. The transition segment contains both NGA and GA features, making the planning more difficult as the geometrical complexity of the part increases. The goal of this framework is to interface different segments while maintaining consistent thermal conditions, all while ensuring build consistency and success.

A welding cell at the University of Tennessee-Knoxville was used to demonstrate builds for this work. Further builds were completed at the Manufacturing Demonstration Facility (MDF) at The Oak Ridge National Laboratory (ORNL). These builds were then scanned and used to characterize the effectiveness of torch heuristics for use in path planning techniques. The software used to generate path plans includes methods for planning, segmenting, and integration of a lumped capacitance thermal model which is used to predict whether overheating will occur within the build. The overall framework proves successful for building up to 105° of overhangs and has been used to manufacture complex geometrical features that were not previously achievable without using these methods. Future works needs to be done on improving the speed and execution of the thermal capacitance model, but it shows promise as a simplified method for predicting heat input into thin wall structures.

Fourth submission