Design optimization of interconnected porous structures using extended triply periodic minimal surfaces

https://doi.org/10.1016/j.jcp.2020.109909Get rights and content

Abstract

We propose an approach to design an optimized heterogeneous interconnected porous structure over a fixed macro-design mesh for performance optimization, which has seldom been addressed despite its extensive potential industrial applications. We achieve this by introducing the concept of the extended triply periodic minimal surface (ETPMS), defined by an implicit function with additional control parameters that can prescribe the element anisotropy and heterogeneity consistently. The control parameters are expressed as continuous explicit functions of spatial coordinates, and ultimately generate a porous structure with perfect geometric connections and fully interconnected pore networks. The modeling advantages of the ETPMS come at the cost of challenges in its efficient simulation and optimization owing to its large number of varied ETPMS elements and their geometric validity requirements. These issues are further resolved using a strategy of offline pre-computation; in particular, parametric homogenization and several carefully designed optimization techniques. The performance of the approach is demonstrated using a suite of three-dimensional benchmark examples.

Introduction

Porous structures appear extensively in natural objects such as bones, wood, cork, and bird beaks [1]. These structures manifest exceptional properties at relatively low densities, such as shock resistance [2], damping enhancement [3], and defect tolerance [4]. The supreme properties of porous structures arise from their complex geometries and topologies, particularly the structural heterogeneity and anisotropy, which have been tailored through thousand-year evolution given the competing constraints in nature [5]. These porous structures usually have fully interconnected pore networks within their interiors so as to enhance nutrition transfer or attachment, or to provide sufficient space for regeneration in tissue engineering [6]. The interconnection property is also mandatory when fabricating such porous structures by means of additive manufacturing, so that the unnecessary support structures during fabrication can be cleaned during postprocessing [7]. However, approaches for the efficient design of such interconnected heterogeneous porous structures have seldom been presented.

We study the problem of designing a heterogeneous interconnected porous structure while simultaneously maintaining a high degree of mechanical and structural rigidity. The macro-design mesh is assumed fixed in the study so that we have restricted the design space and thereby formulated a well-posed design optimization problem. The challenges of this problem lie in its proper geometric modeling, efficient performance simulation, and reliable design optimization.

In terms of the geometric modeling aspect, most conventional methods have mainly assumed a homogeneous element distribution, which only allows for rough shape tuning using parameters relating to the porosity and pore size. However, to utilize the design potentiality of porous structures maximally, it is highly necessary to model and control both the directional anisotropy and distributional heterogeneity. To achieve this goal, the concept of the extended triply periodic minimal surface (ETPMS) is introduced, whereby the structural volume is prescribed using an implicitly defined surface under shape control parameters. In contrast to its counterpart the triply periodic minimal surface (TPMS), the ETPMS exhibits additional design freedom along three axis directions, further enhancing the element anisotropy. Moreover, instead of using discrete constant design variables in each cell, the ETPMS controls the element distribution using certain continuous controlling parameter fields, thereby enabling smooth transition between adjacent elements. Consequently, the consistent implicit definition of the ETPMS is expected to offer significant advantages in terms of continuous connectivity, structural integrity, and the distinctive properties of porous structures; compared to conventional lattice- or strut-based architectures, the ETPMS does not have joints and struts along the body.

The modeling advantages of the ETPMS come at the cost of challenges in terms of its efficient simulation. Conventional approaches [8], [9], [10], [11], [12], [13], [14] generally simulated the performance of such biscale structures via numerical homogenization, which computes an effective elasticity tensor for each microstructure via finite element (FE) analysis and substantially reduces the overall simulation costs. However, the costs remain high as the homogenization process must be conducted for each microstructure during each optimization iteration. We resolve the issue using the strategy of offline pre-computed parametric homogenization, which explicitly expresses the material property as a function of the ETPMS design parameters and can thus instantly simulate the property of an ETPMS element at any specific parameter. It ultimately not only avoids the complex remeshing process, but also facilitates the process by building on a previous study on proper generalized decomposition (PGD) [15].

A further disadvantage lies in the effective and efficient optimization owing to the implicitly defined shape description. A set of validity constraints must be prescribed for each ETPMS element for the ETPMS-based porous structure to be solid and/or core connected. This induces a significant number of constraints for the design optimization problem and makes it difficult to derive a converged solution efficiently. The numerical techniques of constraint aggregation and adjoint-based sensitivity computation are introduced to resolve the issue. Ultimately, the proposed approach can produce smooth and heterogeneous porous structures with scaled anisotropy and fully interconnected pore networks for performance optimization, as demonstrated via a suite of three-dimensional (3D) benchmark examples.

The main contributions of this study are as follows: An approach is proposed using implicit surfaces to design heterogeneous and anisotropic porous structures with fully interconnected pore networks, as well as a systematic approach for the implementation. The concept of ETPMS is presented for modeling the interconnected heterogeneous porous structures with consistent anisotropy and heterogeneity control, and a sufficient condition on its geometric validity is provided. Offline parametric homogenization is conducted based on PGD to predict the properties of a range of microstructures with high accuracy. A specially tailored biscale optimization approach for the ETPMS, in combination with the techniques of offline pre-computation and sensitivity computation, is demonstrated via various 3D numerical examples.

The remainder of this paper is organized as follows: pertinent studies are briefly reviewed in Section 2. The concept of the ETPMS is proposed, and its analytical and geometrical properties are described in Section 3. Thereafter, the problem of designing a heterogeneous porous structure for physical performance optimization is formulated in Section 4. The approach for computing the parametric homogenized material properties using PGD is presented in Section 5. The overall numerical optimization approach is outlined in Section 6. The numerical results from benchmark examples are discussed in Section 7. Finally, we conclude the paper in Section 8.

Section snippets

Review of pertinent studies

The current study is closely related to porous structure modeling, the TPMS, biscale topology optimization, and model reduction. Existing work on these topics is briefly reviewed in this section.

Extended triply periodic minimal surface (ETPMS)

We first introduce the following notations used throughout the paper. Let Ω:={Ωe}NR3 be a fixed 3D macro-design domain consisting of cubic elements Ωe, e=1,,N, and ΓD and ΓN are the fixed boundary and loading boundary, respectively. Let x:=(x,y,z)Ω be the spatial Cartesian coordinate of a point in Ω, and s:={se=(ae,be,ce)I,e=1,,N} is the vector of control parameters defined on each cell Ωe; I is the range of the parameters so that a connected ETPMS is generated. Accordingly, a set of

Problem statement and approach overview

We aim to optimize an interconnected porous structure using the c-ETPMS for performance optimization. The problem is first described within the context of the well-studied compliance minimization problem of linear elasticity analysis, which is followed by an overview of the approach.

Computing parametric homogenized material properties using PGD

The homogenized property DeH of a microstructure ϕ(se) inside a cell Ωe is determined by its configuration; specifically, the shape parameter se. We construct the property DH(se) explicitly in terms of se by building on previous work on PGD [15].

Numerics behind optimization

The modeling advantages of the ETPMS come at the cost of challenges in terms of its efficient optimization of Problem PD in Eq. (15) owing to its implicitly defined and varied computational domain and the large number of validity constraints on the ETPMS cells. A numerical approach for resolving this issue is explained in this section.

Numerical examples

The approach was implemented in MATLAB 2018b, and run on a PC with a 3.6 GHz Intel Core i7 CPU and 16 GB RAM. Its performance was tested in various cases, as outlined in this section. For all cases, the base material was assumed to have a Young's modulus of 1 Pa and Poisson's ratio of 0.33.

Conclusions and future work

In this study, the concept of ETPMS was introduced to model heterogeneous and anisotropic porous structures via an implicit function of additional shape control parameters. The concept results in a perfectly smooth and fully connected porous network, but also comes at the cost of challenges in the effective simulation and optimization. We further resolved these issues using a combination of offline and online computations, particularly by using a PGD-based offline parametric homogenization

CRediT authorship contribution statement

Ming Li: Conceptualization, Methodology, Software, Visualization. Liangchao Zhu: Data curation, Software, Visualization. Jingzhi Li: Funding acquisition, Investigation, Methodology, Supervision, Writing - review & editing. Kai Zhang: Software, Validation, Writing - original draft, Writing - review & editing.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

The valuable comments from the anonymous reviewers are greatly appreciated. The work of M. Li is supported by the NSF of China (No. 61872320). The work of Li was partially supported by the NSF of China No. 11971221 and 11731006, the Shenzhen Sci-Tech Fund No. JCYJ20190809150413261 and JCYJ20170818153840322, and Guangdong Provincial Key Laboratory of Computational Science and Material Design No. 2019B030301001. The work of K. Zhang is supported by the NSF of China under the grant No. 11871245,

References (38)

Cited by (14)

  • Efficient Representation and Optimization of TPMS-Based Porous Structures for 3D Heat Dissipation

    2022, CAD Computer Aided Design
    Citation Excerpt :

    However, they focused on the modeling and characterization of the TPMS-based porous structures but not the optimization of the porous structures. Ming et al. [43] proposed a porous solid structure constructed by extended triply periodic minimal surface for mechanical problems. But their method suffered from the lack of topology (porous period) optimization and a limited porosity of porous structures due to the design of their structures.

  • A review of topology optimization design methods for multi-scale structures

    2023, Hangkong Xuebao/Acta Aeronautica et Astronautica Sinica
View all citing articles on Scopus
View full text