PSI - Issue 66

Umberto De Maio et al. / Procedia Structural Integrity 66 (2024) 502–510 Author name / Structural Integrity Procedia 00 (2025) 000–000

503

2

(Wijerathne et al., 2022), thermal (Du et al., 2023), and acoustic systems (Krushynska et al., 2023; Li et al., 2018), as well as in the modulation of band gaps and wave propagation (Guarín-Zapata et al., 2019; Li and Bi, 2023). For instance, structures inspired by natural designs at the microstructural level have shown significant potential in energy absorption (Ingrole et al., 2021) and thermal insulation (Siddique et al., 2022), in addition, they showed superior control over mechanical and acoustic wave propagation, offering promising applications in acoustic metamaterials (Pranno et al., 2022) and phononic crystals (De Maio et al., 2023). The adoption of this kind of materials is becoming very popular across various industries, including aircraft, civil infrastructure, and medical equipment, where their lightweight design and capabilities in terms of energy-dissipative qualities are crucial for the performance of advanced material applications (De Maio et al., 2024c). To better understand and predict their behavior under extreme loading conditions inducing mechanisms of fracture and failure (Pascuzzo et al., 2020; Slesarenko et al., 2017), these materials are often simulated numerically employing moving mesh techniques (Greco et al., 2021), cohesive fracture models (De Maio et al., 2024b, 2024a; Gaetano et al., 2022) and phase field methods (Khaderi et al., 2014). The deep-sea glass sponge, Euplectella aspergillum, has gained significant interest in engineering and material sciences (Vangelatos et al., 2023; Zhang et al., 2024) due to its extraordinary mechanical robustness combined with a hierarchical architecture across multiple length scales (Fleck et al., 2010; Robson Brown et al., 2019). Its skeleton consists of finely organized silica spicules that form a square lattice, further reinforced by diagonal struts, providing excellent mechanical properties (Aizenberg et al., 2005). Inspired by the open-cell structure of Euplectella aspergillum, several engineering approaches have been developed for applications ranging from energy absorption to modulation of acoustic and thermal wave propagation, owing to their lightweight and energy-dissipative characteristics (Liu et al., 2024). The mechanical behavior of lattice systems is determined mainly by their node connectivity and structural element thickness, especially in terms of load distribution and deformation under stress. In general, diagonal members enhance mechanical stability by creating additional load paths, which reduces the bending moments in individual elements and converts bending forces into axial forces. For instance, in (Fernandes et al., 2021) the authors investigated a glass-sponge microstructure optimizing the geometrical parameters to switch from bending-dominated to stretching-dominated behavior, greatly enhancing the mechanical performance and stability. In 3D-printed microstructured materials, due to their standard capability to handle large deformations, the onset of microscopic instability phenomena plays a crucial role in determining their mechanical performance (Li and Rudykh, 2019). Since from the literature, it is already known that the onset of both microscopic and macroscopic instabilities can result in significant changes to the material's mechanical properties, such as stiffness, strength, and energy absorption capabilities, understanding and controlling these instabilities is essential for designing performing metamaterials (Greco et al., 2024; Li et al., 2021). In this context, the present study employs a computational model to design and optimize lattice geometries capable of mimicking the mechanical response of the glass sponge. The proposed model was implemented in COMSOL Multiphysics using a LiveLink with MATLAB, ensuring that the structural optimization procedure, performed via a genetic algorithm, is further driven by artificial neural networks. The key geometric parameters are systematically varied, maintaining a constant volume fraction for the representative volume element (RVE). The genetic algorithm optimization procedure (AL-Tabtabai and Alex, 1999; Sahli Abderahmane et al., 2020) is designed to maximize the critical load factor under uniaxial compression along the vertical direction. In addition, to determine if the onset of instability is characterized by a short wavelength (microscopic) or a long wavelength (macroscopic), an artificial neural network based on convolutional layers was employed (Vedaldi and Lenc, 2015). This predictive capability allows the system to penalize designs that exhibit global instability, thus favoring the local one. Definitively, the overall structural performance is enhanced, ensuring a more stable and efficient lattice design. The results show a significant improvement in mechanical performance compared to previous bioinspired lattice designs. Specifically, an optimized lattice achieved a 70% increase in ultimate buckling deformation and a 40% improvement in the buckling load. The results highlighted the potential of coupling computational design techniques with bioinspired materials to design lightweight, mechanically efficient structures with significant applications in civil infrastructure, aircraft, and medical implants.