Session: DAC-06-1: Design for Additive Manufacturing

Paper Number: 147880

147880 - Improving Convergence of Topology and Orientation Optimization Through Explicit Consideration of Orthogonal Material Properties 

In material extrusion additive manufacturing of composites, extruded roads are highly anisotropic due to shear alignment of the fibers. This offers an opportunity to optimize printing toolpaths such that this anisotropy is preferentially oriented to improve performance of the printed parts. There has been significant research towards exploiting these capabilities through generative design algorithms that consider simultaneous optimization of part topology and local material orientations. However, these orientation design spaces are highly non-convex, often resulting in convergence to local minima that detract from nominal design performance and toolpath quality. Here, the authors present a novel parameterization of the orientation design space that improves convergence and final design quality. This is accomplished by leveraging the fact that the local minima often lie at a 90 degree rotation from the intuitive optimal direction, which correspond to the fiber traveling orthogonally to the anticipated load paths. By allowing the algorithm to explicitly consider the orthogonal set of material properties during optimization, stable convergence (i.e., across a variety of initial conditions) is achieved with material orientations consistently aligning to the intuitive optimal direction. This formulation is extended to also include a filtering mechanism to ensure smooth fiber orientations for deposition of continuous fiber materials. Exercising this novel problem statement on a variety of benchmark load cases has demonstrated a 10% improvement in final design quality.

Presenting Author: Joseph Kubalak Virginia Tech

Presenting Author Biography: Joseph Kubalak is a research scientist in the Design, Research, and Education for Additive Manufacturing Systems (DREAMS) lab at Virginia Tech. He received both his B.S. and Ph.D. in Mechanical Engineering from Virginia Tech in 2014 and 2020, respectively. His research focuses on the intersection between multi-axis (robotically-enabled) additive manufacturing and design optimization techniques. He has presented at the SXSW Conference (2016) and the ACCelerate Festival (2017) and won the America Makes Innovation Sprint for Smart Structures (2016).

Authors:

Joseph Kubalak Virginia Tech
Christopher Williams Virginia Tech