Strength through defects: A novel Bayesian approach for the optimization of architected materials

Area of Science:

• Materials Science
• Computational Science
• Mechanical Engineering

Background:

• Architected microlattices offer tunable mechanical properties.
• Optimizing defect structures is crucial for enhancing material performance.
• Traditional optimization methods struggle with complex, large design spaces.

Purpose of the Study:

• To develop a novel Bayesian optimization framework for designing architected microlattices.
• To maximize the strain energy density of microlattices before failure.
• To explore the role of defect arrangement in enhancing mechanical properties.

Main Methods:

• Utilized a Bayesian optimization framework: evolutionary Monte Carlo sampling.
• Systematically designed defect arrangements in 4 × 4 × 5 3D lattices.
• Searched a design space of billions of possible lattice configurations.

Main Results:

• Identified a global optimum design with only 250 cost function evaluations.
• Achieved a normalized strain energy density 12,464 times greater than defect-free counterparts.
• Demonstrated the framework's efficiency in handling discrete, computationally expensive design spaces.

Conclusions:

• Evolutionary Monte Carlo sampling is effective for optimizing architected materials.
• Defect engineering can dramatically improve the mechanical performance of microlattices.
• The framework is applicable to diverse scientific optimization challenges.