PSI - Issue 34

Federico Uriati et al. / Procedia Structural Integrity 34 (2021) 184–190 Author name / Structural Integrity Procedia 00 (2021) 000–000

186

3

To run a topology optimization the part geometry is divided in the “design space” where material is present but can be removed to achieve an optimal geometry, and the “non-design space”, where material and geometry are fixed to guarantee necessary interfaces (bearing seats etc.), that is the boundary conditions. Fig. 1b shows a simple volume larger than the original part. The commercial software used to complete the preparation of the three-dimensional model of the part and execute finite-element-based topological optimization adopts the SIMP-algorithm (solid isostatic material with penalization) that attributes a varying density to each element, (Bendsøe 1989), (Inspire, Altair Engineering Inc., USA). A sufficiently large design space provides freedom for the generation of alternative solutions to be subsequently analyzed and selected. The specific material properties are for an aluminum alloy: Elastic modulus E = 75 GPa, yield stress σ y = 275 MPa and density δ = 2.7 g/cm 3 . The optimization goal is defined in terms of minimization of the mass with a realized safety factory (SF) of 1.6 and global maximization of the stiffness. This optimization allows to obtain the best ratio with a limitation on the maximum admissible local stress. The best solution is shown in Fig.2a. It respects imposed limits and is characterized by the best stiffness-to-weight ratio associated to weight saving of 57%. Upon selection of the optimal solution. The topological optimization software generates several solutions, each one characterized by a roughly defined geometry that cannot be directly manufactured, this is used as a reference by the designer when modeling the part manually with regularized shape, smooth surfaces and considering applicable DfAM rules, see fig. 2b. The commercial software used in this phase is a non-parametric CAD software using a poly-NURBS modeling approach (Inspire Studio, Altair Engineering Inc., USA) because it gives freedom to reproducing the organic shape typically obtained from topology optimization. The connection areas between optimized and non-design volumes which are typically critical from the stress distribution point of view and require ad-hoc smoothing. a) b) Figure 2. (a)Topology optimized geometry with boundary interfaces; (b) Remodeled part with smooth surfaces Validation of the reconstructed and optimized part geometry requires repeated finite element analyses (FEA) to iteratively identify possible critical points and local model refinement as schematically shown in Fig.3 considering details such as sharp fillets or thin trusses to control local stress concentrations. The best solution is shown in Fig.2a. It respects imposed limits and is characterized by the best stiffness-to-weight ratio associated to weight saving of 57%. The advances in AM technology gives design freedom from the geometrical point of view, however requirements related to DfAM rules should be considered during modeling. Overhanging surfaces angles, minimum wall thickness and minimum truss size are examples of limits to be known from the designer. The job preparation involves the placement of the parts into the build chamber, orientation and nesting, this procedure requires specifical knowledge and experience to define the correct position and successful result. The build design and the preparation phase were developed with the Materialise Magic software (Materialise, Belgium). Supporting structures are important for the L-PBF manufacturing process not only to print down skin-oriented surfaces but to control local heat dissipation. Minimization of supports for economic reasons and functional control of parts distortions.