Page 72 - MSAM-2-4

Materials Science in Additive Manufacturing Base shape generation for HAM

illustrated in Figure 12. This step is intended to evaluate 4.1. Case study 1
and search for the optimal base shape. Since the material The model described in this case study was derived from a
waste rate determined by volume could directly impact study on cold spraying process . Using this method, the
[48]
the AM and NAM processing time, we only focused on authors built the whole model from zero volume, causing
the volume of material in this research. To facilitate the wastage of many materials. After analyzing the CAD
searching process, an objective function concerning waste model, we learn that a quite large volume can be fabricated
material volume is defined in Equation V: by traditional method easily and only very small complex
min: F = ω (f − f ∩ f ) + ω (f − f ∩ fi) (V) structure, like the four feet, can be fabricated by iterative or
p
r
p
p
i
i
1
2
where F is the material use rate, fi is the base shape sequential AM and NAM processes. Hence, we generated
a base shape with an aim to save material and time for this
r
(Figure 13A), and f is the volume of the finished part case. The CAD model of this case is shown in Figure 14.
p
volume. Figure 13B shows the base shape after machining It is a reproduction of the example CAD model used in
from the casting shown in Figure 13A. Figure 13C Lynch et al. .
[48]
shows the subparts which need to be added through
AM processing (CS spraying AM module is used as Based on the CAD model of the bracket, the voxelization
example in this research) to the qualified base shape. ω and generation of skeleton were operated in accordance
1
and ω are the weights assigned to the cost of AM and with the above-mentioned method in MATLAB, as
2
NAM processing, respectively, in terms of subtractive illustrated in Figure 15A and B. A modified skeleton was
manufacturing processes and materials. Here, ω and adopted to replace the original one. The original subparts
1
ω are related in an equation, that is, ω + ω = 1. In this can be decided by the branches based on the joint points,
2
1
2
research, the weights were set at 0.95 and 0.05, respectively. similarly with the tree model. The subparts of the bracket
This objective function could be directly used as fitness are shown in Figure 16.
function in the PSO. In this research, we defined two main The selected branch combination after GA calculation
evaluation criteria for illustration and general application. is shown in Figure 18A. The following optimization
In industrial application, different specific criteria can be is conducted by the mentioned PSO algorithm with
defined according to HAM processing constraints and consideration of the above-mentioned constraints and
other objectives related to time, cost, and quality. material use rate. The optimization procedure is depicted
in Figure 17, which shows that the optimal solution is
4. Case studies found after 32 iterations.
In this section, we use two examples manufactured by a To simplify the whole process, circle was used as cross-
cold spraying-based HAM platform in different complexity section in this case study, and ball shapes were used as
levels to demonstrate the proposed method. joints to connect the adjacent branches, as illustrated in
Table 1. Parameters for Genetic Algorithm (GA) in Galapagos editor
Option Max. stagnant Population Initial boost Maintain (%) Inbreeding (%)
Description 50 50 2 5 75

A B C

Figure 13. (A) Base shape of the tree structure; (B) qualified base shape after machining; and (C) cold-sprayed subparts.

Volume 2 Issue 4 (2023) 9 https://doi.org/10.36922/msam.2103