A cost-driven process planning framework for selective laser melting

Selective Laser Melting (SLM) is an additive manufacturing (AM) method, capable of producing end-use metal parts by selectively melting layers of powder with a moving laser. The process can create complex lightweight geometries in a single build, and offers an opportunity to improve product value through greater design freedom, part consolidation, reduced tooling and reduced supply chain complexity for a range of different industries. However, SLM also poses a number of materialand process-related challenges, which prevent wider adoption of this process for enduse part production. Due to high residual stresses characteristic of this process, build failure and poor part quality are still common issues. The cost of this process is also relatively high, while the need for post-processing leads to additional, and sometimes unexpected, costs. Effective process planning can mitigate these problems, however, this is not an easy task due to the large number of interdependent process parameters and their combined effects on the process cost, time and quality. Part build orientation is one such important parameter, which affects the surface roughness, build cost and time, as well as the need for support structure. Due to the anisotropic material properties produced by SLM and the effects of part geometry on the residual stress profile, there is also a complex relationship between build orientation, part quality and the risk of build failure. Moreover, the build orientations of multiple parts determine how many SLM machines are required to build all those parts. The placement of parts inside AM machines is known as a bin packing problem, and its link to build orientation has been acknowledged throughout the literature. Existing approaches have optimised build orientation and bin packing as two separate problems, limiting the optimality of the overall solution. Thus, this thesis provides a cost-driven framework for optimising build orientation and bin packing of multiple parts in SLM, and provides two heuristics for solving these two problems simultaneously. The cost model used to drive the framework is developed based on a thorough literature review, and incorporates build cost and post-processing cost. Additionally, by considering SLM in the context of lean manufacturing two distinct production scenarios are identified, and referred to as Identical Batch Production (IBP) and Mixed Batch Production (MBP). Thus, each heuristic vi is aimed at a specific production scenario. For IBP, a Tabu search procedure is developed by solving the build orientation problem along with a cutting-stock problem, as bins are assumed to be identical. To improve the efficiency of this search heuristic an area-based approximation strategy is proposed, which can reduce the solution time by as much as 50%, as indicated by the computational results. The effect of build orientation on residual stresses is also considered, by extending the above Tabu search to address the multi-objective problem of cost and residual stresses in the context of IBP. Due to the high computational cost of this problem, two alternative approximation strategies are proposed, and coupled with the multiobjective Tabu search. The two strategies are demonstrated on a single test part of medium complexity; the results demonstrate the validity of this approach, as better, albeit less obvious trade-off solutions were found by the proposed Tabu search than by an experienced SLM operator. Finally, an Iterative Tabu Search Procedure (ITSP) is developed for the MBP scenario. Because the solution space of this problem is much larger and more complex than the IBP, the ITSP consists of six distinct stages, where each stage is aimed at exploring a different area of the solution space. The procedure is benchmarked against commercial software, indicating an average cost improvement of 14.6% for 27 test instances