Projection-based Implicit Modeling Method (PIMM) for Functionally Graded Lattice Optimization

TL;DR

This paper introduces a projection-based implicit modeling method (PIMM) for functionally graded lattice optimization that avoids homogenization, using a parametric projection and RBF interpolation, demonstrated with TPMS lattices and GPU acceleration.

Contribution

The paper presents a novel PIMM approach that links implicit lattice functions with finite element meshes without homogenization, improving stress-constrained lattice design.

Findings

Effectively resolves stress-constrained lattice design issues.

Removes sharp corners from initial designs after optimization.

Demonstrates flexibility for irregular porous scaffold and non-periodic lattice designs.

Abstract

This paper proposes a projection-based implicit modeling method (PIMM) for functionally graded lattice optimization, which does not require any homogenization techniques. In this method, a parametric projection function is proposed to link the implicit function of functionally graded lattice with the finite element background mesh. To reduce the number of design variables, the radial basis function (RBF) is utilized to interpolate the implicit design field. The triply periodic minimal surface (TPMS) lattice is employed to demonstrate the proposed method. Compared with conventional homogenization-based topology optimization, the proposed method can effectively resolve the stress-constrained lattice design; for example, sharp corners are removed from the initial design after optimization. Several two- and three-dimensional lattice design examples are presented to solve the compliance and stress-constrained problems, for which GPU-based computing is adopted to accelerate the finite element analysis. The proposed PIMM method is flexible and can potentially be extended to design graded irregular porous scaffold and non-periodic lattice infill designs.