Data Sampling Schemes for Microstructure Design with Vibrational Tuning Constraints

Abstract

Microstructures significantly impact the performance of sensitively engineered components, such as wireless impact detectors used in military vehicles or sensors used in aircrafts. These components can operate safely only within a certain range of frequencies, and frequencies outside that range can lead to instability because of resonance. This paper addresses optimization of the microstructure design to maximize the yield stress of a galfenol beam under vibration tuning constraints defined for the first torsional and bending natural frequencies by using a data-driven solution scheme. In this study, two carefully designed algorithms are used to sample the entire microstructure space. Classical optimization techniques often lead to a unique microstructural solution rather than yielding the complete space of optimal microstructures. Multiple optimal solutions are imperative for the practicality of design because conventional low-cost manufacturing processes can generate only a limited set of microstructures. The current data sampling-based methodology outperforms or is on par with other optimization techniques but also provides numerous near-optimal solutions, which is two to three orders of magnitude more than previous methods. Consequently, the proposed framework delivers a spectrum of optimal solutions in the microstructure space that can accelerate material development and reduce manufacturing costs.