Page 74 - MSAM-2-4

Materials Science in Additive Manufacturing Base shape generation for HAM

validity and necessity of the proposed method. The model and a coarse skeleton was obtained (Figure 20A). After
has many features and no obvious bases as well as many simplification, the approximate skeleton was generated
radial subvolumes in different directions (Figure 19). (Figure 20B).
Similarly, the CAD model in STL file was voxelized In this case, all the branches are important since they
have similar volumes. Hence, the subparts of the original
A B CAD model, as shown in Figure 21, were generated using
the joint points of branches of the simplified skeleton
(only smoothing process applied). The optimization
calculation of base shape considers the constraints
mentioned above and the material use rate as the
objective function is calculated using the PSO algorithm
in Silvereye, one plugin of Grasshopper. The result of each
Figure 20. (A) Voxelization and skeleton in MATLAB; (B) modified iteration was obtained, as shown in Figure 22. It can be
skeleton.
found that the convergence occurred at the 7 iteration
th
(Figure 22). The optimization results of case study 2 are
presented in Figure 23.
5. Conclusion

In this research, we analyzed a key process planning
problem, called base shape generation and optimization,
for the CAPP of HAM. This problem is important for
the industrial application, although it has rarely been
investigated. For sequential HAM processes, where
there is no iterative AM and NAM processing, the
determination of an optimal base shape directly affects
the final manufacturing complexity, time, cost, and
Figure 21. The subparts of case study 2.
quality. To address the problem at the generic level, this
paper introduces an optimization method and presents
two examples for demonstration purposes. The case
studies show that the proposed optimization method
can well resolve the base shape determination problem
for complex geometries, for example, the tree model and
bracket model in the paper. The expected advantages are
obvious, especially in terms of material saving in the CS
HAM process. It is can also be adopted for other similar
sequential HAM processes, such as wire arc AM and
cladding-based HAM process. However, the method still
has room for improvement. For instance, the skeleton
generation and CAD model decomposition algorithms can
be improved with more consideration of the constraints of
manufacturing the base shape, which are not discussed in
this paper. We intend to address this gap by developing a
Figure 22. The optimization process of base shape with Particle Swarm CS HAM platform to further verify the proposed method
Optimization. and its application value.
A B C

Figure 23. (A) Optimized skeleton by genetic algorithm; (B) optimized base shape by Particle Swarm Optimization; and (C) remaining subparts for CS
AM module.

Volume 2 Issue 4 (2023) 11 https://doi.org/10.36922/msam.2103