High Dimensional Multiscale Concurrent Optimization Towards Additively Manufactured Structures

This work establishes a robust multiscale framework for the design of highly optimized and additively-manufacturable structures. A spatially-varying lattice-like structure is optimized by gradually varying a microscale description, represented by a series of unit cells. Each unit cell is derived through a permutation of the microscale parameterization, with each possessing unique material properties. Unit cell properties are derived by homogenizing the results from a periodically-constrained finite-element analysis. A concurrent multiscale approach is used to minimize the number of microscale simulations needed to perform macroscale optimization. This allows the dimensionality of the microscale parameterization to be increased without a significant computational penalty. To increase efficiency, the dimensionality of the microscale parameterization can be altered locally during macroscale optimisation. Preliminary results indicate that novel morphing structures could be achieved using this framework.