Lattice Structure Design Advisor for Additive Manufacturing Using Gaussian Process

Abstract

Additive manufacturing (AM) exemplifies the potential of lattice structures to revolutionize structural design. It enables light weight lattice structures to be produced while maintaining the desirable structural performance. Lattice design can vary in different shapes and dimensions. Obtaining the structural performance of a particular lattice structure design is not a straight-forward process. Significant effort is required to perform mechanical testing experiments or to perform finite element analysis (FEA) to characterize the lattice design. In view of that, a guidance system to determine lattice design parameters based on desired functional performance for a specific lattice type is developed, which can be used in interactive design processes and workflows. Homogenization using FEA experiments is applied to characterize the macroscopic lattice structural properties. Mechanical properties of orthotropic cubic lattice f2ccz are estimated. It follows with a design of experiment study to characterize the effective structural properties of 39 lattices with respect to lattice design parameters (unit cell length and strut diameter). A Gaussian process is applied to develop models relating the lattice design parameter to macroscopic structural properties (forward model), and vice versa (inverse model). Both the forward and inverse models are examined and shown to be capable of modeling the FEA experimental dataset of 39 lattices. To illustrate the potential application of the lattice design advisor framework, a structural design use case including lattice part is presented. In the use case, the lattice structure design advisor is proven to be able to estimate an accurate homogenized material property of arbitrary lattice design parameter. This lattice structure design advisor can simplify and streamline the design, modeling and simulation process of lattice-filled structural designs.