An elastic properties-based topology optimization algorithm for linear orthotropic, functionally graded materials

TL;DR

This paper introduces a topology optimization algorithm for linear orthotropic materials that directly optimizes six elastic properties, resulting in more versatile and efficient strain-homogenized structures without volume constraints.

Contribution

It extends existing TO algorithms to orthotropic materials, increasing design space and eliminating the need for volume constraints, with two developed approaches for optimization.

Findings

Outperforms isotropic algorithms in load case designs.

Effective for shear-dominated problems like torsion.

Comparable computational cost to existing methods.

Abstract

Topology optimization (TO) has experienced a dramatic development over the last decades aided by the arising of metamaterials and additive manufacturing (AM) techniques, and it is intended to achieve the current and future challenges. In this paper we propose an extension for linear orthotropic materials of a three-dimensional TO algorithm which directly operates on the six elastic properties -- three longitudinal and shear moduli, having fixed three Poisson ratios -- of the finite element (FE) discretization of certain analysis domain. By performing a gradient-descent-alike optimization on these properties, the standard deviation of a strain-energy measurement is minimized, thus coming up with optimized, strain-homogenized structures with variable longitudinal and shear stiffness in their different material directions. To this end, an orthotropic formulation with two approaches -- direct or strain-based and complementary or stress-based -- has been developed for this optimization problem, being the stress-based more efficient as previous works on this topic have shown. The key advantages that we propose are: (1) the use of orthotropic ahead of isotropic materials, which enables a more versatile optimization process since the design space is increased by six times, and (2) no constraint needs to be imposed (such as maximum volume) in contrast to other methods widely used in this field such as Solid Isotropic Material with Penalization (SIMP), all of this by setting one unique hyper-parameter. Results of four designed load cases show that this orthotropic-TO algorithm outperforms the isotropic case, both for the similar algorithm from which this is an extension and for a SIMP run in a FE commercial software, presenting a comparable computational cost. We remark that it works particularly effectively on pure shear or shear-governed problems such as torsion loading.