Tool selection is a crucial stage in process planning that involves multiple machining and cost factors and necessitates the integration of CAD and CAM software. Most of the time, this activity requires human intervention through process planning engineers to search adequate cutters in tool catalogues and make decisions based on their experience. With increasing part complexity, more sophisticated CNC machines are implemented and the manual tool selection doesn’t generally lead to optimal choices.
This thesis presents an approach for tool sequence optimization in the case of 5-axis machining. Most of the reported work suggests tooling optimization methods involving parametric surfaces and CPU-enabled algorithms. In the current work, a novel voxel-based approach is presented. The main advantage of this 3D-representation is the ability to parallelize different operations executed on single voxels and run them on parallel platforms such as GPU cores. This work is realized through Sculptrprint, a voxelized GPGPU-enabled CAM software, and introduces 3 different algorithms to optimize the tool sequence selection. Each of the formulated strategies is based on the optimization of one or two machining objectives and has a GPU-only implementation.
Applications of Cloud manufacturing are also explored via Amazon Web Services and a school-hosted virtual machine by running the developed algorithms on different local and virtualized platforms. The performance of several GPUs is benchmarked and shows an efficiency optimum when using the most powerful hardware. The effects of machining and rendering parameters are investigated. The results show that generated tool sequences are strongly dependent on the chosen step over, voxel size and optimization criterion.