Hybrid TPMS-based architectured materials (HTAM) for enhanced specific stiffness using data-driven design

Abstract

This study introduces hybrid TPMS-based architectured materials (HTAM), achieved by superimposing several triply periodic bicontinuous structures (TPBSs). This approach allows for the creation of structures that were previously unattainable using conventional single TPBS concept. We investigate the optimization of these architectured materials to enhance mechanical stiffness while reducing weight. To explore this expanded design space and identify optimal designs, we employed multi-objective Bayesian optimization (MBO) integrated with Gaussian process regression (GPR). By utilizing both the probability of hypervolume improvement (PHVI) and expected hypervolume improvement (EHVI) acquisition functions in parallel during the optimization process, we improved the efficiency of time and data usage. This facilitated the development of HTAM that form a Pareto front, approaching closer to the upper bound in the relative density and relative stiffness space. The optimized HTAM exhibited markedly higher specific Young’s modulus across various relative densities compared to conventional structures. Following optimization and manufacturability considerations, optimized HTAM designs selected from Pareto front were fabricated using selective laser sintering (SLS) at the macro scale and two-photon polymerization (2PP) at the micro scale. Compression tests confirmed the superior stiffness and exceptional yield strength of the HTAM, validating their potential for advanced engineering applications.

Keywords

• Hybrid TPMS-based architectured materials (HTAM)
• Architectured materials
• Data-driven optimization
• Multi-objective Bayesian optimization (MBO)
• Additive manufacturing