Abstract :
[en] The combination of topology optimization and additive manufacturing has brought a recent break-through in engineering.
Topology optimization aims at generating innovative concepts with high performance to weight ratio, but the optimized
designs are often difficult to fabricate as-it-is using classical fabrication processes. In the AERO+ research project [1], we
focus on metal additive manufacturing processes, and particularly on Electron Beam Melting (EBM) and Selective Layer
Manufacturing (SLM) processes. Among others, the maximum size of structural elements has been reported as a
manufacturing limitation for these processes mainly due to overheating problems. The maximum size constraint in
topology optimization is based on restricting the amount of material within the neighborhood of each point in the design
domain [2]. Its role is to split the bulky material during the topology optimization process. The constraint introduces extra
structural members in the design, such as bars or sheets, which tend to remain very close to each other. This greatly
increases the complexity of the design and therefore of the manufacturing process, as pointed in [3]. This work aims to
improve the manufacturability of such designs. To this end, a new constraint is proposed capable of separating the
structural members according to the user’s needs.
The proposed constraint, as well as the maximum size, restricts the quantity of material within the neighborhood of each
point in the design domain. This is achieved by asking for a specific amount of voids within the test region. However,
there is a small difference in the involved parameters which allows to separate the structural members instead of
constraining their size. For this purpose, the amount of voids to be included within the test region is a function of the
distance between bars, as shown in Figure 1. The local constraints are aggregated using the p-mean function in order to
avoid the computational overburden on the optimizer. The method is validated for compliance minimization on 2D and
3D design domains (Figure 2 and 3), showing that the proposed constraint is capable of reducing the geometric
complexity of designs with maximum size constraints.
