Computational and experimental analyses and optimization of architectured materials

Abueidda, Diab W

Permalink

https://hdl.handle.net/2142/108210 Copy

Description

Title

Computational and experimental analyses and optimization of architectured materials

Author(s)

Abueidda, Diab W

Issue Date

2019-12-13

Director of Research (if dissertation) or Advisor (if thesis)

Jasiuk, Iwona M

Doctoral Committee Chair(s)

Jasiuk, Iwona M

Committee Member(s)

• Ostoja-Starzewski, Martin
• Kim, Seok
• Kersh, Mariana
• James, Kai

Department of Study

Mechanical Sci & Engineering

Discipline

Mechanical Engineering

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• Architected Materials
• Topology Optimization
• Data-Driven Models
• Constitutive Modeling of Materials
• Acoustic Bandgaps

Abstract

New materials with enhanced properties are of high scientific and industrial interests. Researchers have used materials that are nature-inspired or bioinspired and developed optimization frameworks to come up with such materials. This study presents novel cellular materials (CMs) based on triply periodic minimal surfaces (TPMS), where their different physical properties are studied. One of the advantages of TPMS structures is that they do not have joints that can cause imperfections during fabrication leading to potential structural failure. Also, the absence of joints reduces the stress concentration effect. The architectures of these cellular materials induce robust performance. The electrical and thermal conductivities of six TPMS cellular materials are analyzed using finite element analysis. The conductivities vary linearly with relative density. The conductivities of the Primitive-, IWP-, Neovius-, Gyroid-, and S-foams are very close to each other. The uniaxial, shear and bulk moduli, Poisson ratio, and elastic anisotropy of these TPMS-foams are also computed and compared. To further investigate the mechanical properties, the study focuses on three TPMS cellular materials, namely, Schwarz Primitive, Schoen IWP, and Neovius. 3D printing is used to fabricate these polymeric cellular materials and their base material. Their properties are tested to provide inputs and serve as validation for finite element modeling. Two finite-deformation elastic/hyperelastic-viscoplastic constitutive models calibrated based on the mechanical response of the base material are used in the computational study of the TPMS-CMs. Furthermore, the design, fabrication, and testing of the Neovius-microlattice are discussed. The results of this study show that the Neovius-microlattice achieves outstanding mechanical properties under compressive loading. The mechanical properties (under compression) of the Neovius-microlattice are further enhanced by coating the microlattice with alumina exploiting the size effect phenomenon accompanying ultrathin ceramic layers. Also, a distinction between out-of-plane (global) and in-plane (local) buckling is made. At the local level, the Neovius-microlattice experiences in-plane buckling and plastic yielding, where the in-plane buckling occurs partially (at the openings of the Neovius-lattice). In other words, the cell does not lose its load-bearing capability even when local buckling takes place. When the slenderness ratio is increased, out-of-plane buckling starts to be the governing failure mechanism. Additionally, the acoustic band structure and sound attenuation of three cellular materials based on Schwarz Primitive, Schoen IWP, and Neovius surfaces are studied using a finite element method. The acoustic properties of TPMS structures have many potential engineering applications, where one can tune the width of band gaps based on the porosity of the TPMS structure. The band gaps reported here are wider than for similar structures reported in the literature. TPMS structures have robustness against other structures available in the literature in terms of achieving wide band gaps at higher porosity while the mechanical integrity of the structure is still not comprised. The sound attenuation results are in good agreement with the band gap analysis. Also, the first complete band gaps of our TPMS structures lie entirely in the audible range of frequencies. Another technique to obtain material architectures leading to enhanced properties is to use topology optimization frameworks. Topology optimization offers a systematic platform to achieve optimal material architectures. In this study, a framework for the design of elastoplastic materials with enhanced performance through the maximization of either toughness or end compliance has also been proposed. The objective of the optimization is to enhance the performance of an elastoplastic material obeying the von Mises plasticity, subject to mass and elastic compliance constraints. A path-dependent adjoint method to accurately and efficiently compute the sensitivities of the objective functions and constraints has been derived. Also, the effect of different design spaces has been studied, and it has been shown that, in some cases, a subset of a larger design space could yield better performance due to the nonconvexity of the optimization problem. Recent advances in the field of machine learning open the door to use data-driven models to not only predict the response of materials but also to optimize the performance of architected materials. The use of data-driven models to predict and optimize architected materials is demonstrated on a composite system. Composites are popular due to their multifunctional properties compared to those of the constituents. A convolutional neural network model that is capable of quantitatively predicting the mechanical properties (modulus, strength, and toughness) of 2D checkerboard composites has been developed. The model is trained using finite element results (ground-truth data). Then, it is tested on another dataset, which is not seen by the model throughout the training process to ensure the validity of the model. The model shows very promising capabilities; it illustrates the potential of utilizing convolutional neural network (CNN) models in structural and materials analysis. The developed CNN model is integrated with a genetic algorithm optimizer to obtain the composite configurations (material distribution and volume fraction), leading to materials with improved performance. CNN models have the potential of accelerating the current optimization techniques, and they might revolutionize the field of structural and materials design.

Graduation Semester

2020-05

Type of Resource

Thesis

Permalink

http://hdl.handle.net/2142/108210

Copyright and License Information

Copyright 2020 Diab Abueidda

Owning Collections