Computational discovery of extremal microstructure families

Chen, Desai; Skouras, Mélina; Zhu, Bo; Matusik, Wojciech

doi:10.1126/sciadv.aao7005

Cited by 82 publications

(59 citation statements)

References 42 publications

(54 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The processes by which mechanical structures are designed have evolved to include a variety of computational tools that have been successful in producing structures with highly tuned properties (1)(2)(3)(4)(5)(6)(7)(8)(9)(10). However, realizing high-performance mechanical structures often involves optimizing properties that cannot be reliably and rapidly predicted using computation, namely, nonlinear mechanical properties (11)(12)(13)(14)(15)(16).…”

Section: Introductionmentioning

confidence: 99%

A Bayesian experimental autonomous researcher for mechanical design

et al. 2020

View full text Add to dashboard Cite

While additive manufacturing (AM) has facilitated the production of complex structures, it has also highlighted the immense challenge inherent in identifying the optimum AM structure for a given application. Numerical methods are important tools for optimization, but experiment remains the gold standard for studying nonlinear, but critical, mechanical properties such as toughness. To address the vastness of AM design space and the need for experiment, we develop a Bayesian experimental autonomous researcher (BEAR) that combines Bayesian optimization and high-throughput automated experimentation. In addition to rapidly performing experiments, the BEAR leverages iterative experimentation by selecting experiments based on all available results. Using the BEAR, we explore the toughness of a parametric family of structures and observe an almost 60-fold reduction in the number of experiments needed to identify high-performing structures relative to a grid-based search. These results show the value of machine learning in experimental fields where data are sparse.

show abstract

Section: Introductionmentioning

confidence: 99%

A Bayesian experimental autonomous researcher for mechanical design

et al. 2020

View full text Add to dashboard Cite

show abstract

“…Responsible Editor: Ming Zhou Chikwesiri Imediegwu ci214@imperial.ac.uk 1 Department of Aeronautics, Imperial College London, London, UK These constitute constraints on the problem that ensure the feasibility of the optimal design. Inspired by the pioneering work of Lakes (1987), researchers have developed frameworks for designing global auxetic phenomena (materials with negative Poisson's ratio) (Zhu et al 2017;Chen et al 2018;Ha et al 2016;Xia and Breitkopf 2015a) as well as fabrication techniques for the derived micro-structured materials (Lee et al 2012). Such phenomena were previously limited to small deformations but recently, simple parameterization that facilitate programmable extremal properties which remain constant over large deformations have been developed (Babaee et al 2013;Wang et al 2014;Clausen et al 2015).…”

Section: Introductionmentioning

confidence: 99%

“…This optimized tiling leveraged the precomputed library of micro-architecture families in search of best-suited neighboring cells for macro-scale structural optimization. Zhu et al also populated a "gamut" of discrete evaluations of material properties by alternating stochastic sampling with continuous optimization and in so doing discovered families of microstructures with extremal properties (Zhu et al 2017;Chen et al 2018). Xia and Breitkopf solved an inverse homogenization problem by determining the best 2-D micro-architecture for a prescribed deformation, thereby populating a strain energy density property space (Xia and Breitkopf 2015b).…”

Section: Introductionmentioning

confidence: 99%

Multiscale structural optimization towards three-dimensional printable structures

Imediegwu

Murphy

Hewson

et al. 2019

Struct Multidisc Optim

View full text Add to dashboard Cite

This paper develops a robust framework for the multiscale design of three-dimensional lattices with macroscopically tailored structural characteristics. The work exploits the high process flexibility and precision of additive manufacturing to the physical realization of complex microstructure of metamaterials by developing and implementing a multiscale approach. Structures derived from such metamaterials exhibit properties which differ from that of the constituent base material. A periodic microscale model is developed whose geometric parameterization enables smoothly changing properties and for which the connectivity of neighboring microstructures in the large-scale domain is guaranteed by slowly changing largescale descriptions of the lattice parameters. A lattice-based functional grading of material is derived using the finite element method with sensitivities derived by the adjoint method. The novelty of the work lies in the use of multiple geometry-based small-scale design parameters for optimization problems in three-dimensional real space. The approach is demonstrated by solving a classical compliance minimization problem. The results show improved optimality compared to commonly implemented structural optimization algorithms.

show abstract

“…With the emergence of micro- and nanofabrication techniques and high-content imaging, novel combinatorial and high-throughput approaches have been developed [ 171 , 172 , 173 , 174 ]. These libraries are based on miniaturised platforms, which are able to simultaneously characterise a high number of varying surface properties, such as topography [ 37 , 38 , 39 , 41 , 43 , 174 ] and chemistry [ 42 , 174 , 175 , 176 ]. Together with machine learning algorithms this offers a great tool to screen for properties that induce desired cell behaviour in vitro [ 177 ].…”

Section: Discussionmentioning

confidence: 99%

“…This approach can be applied both on a structural and chemical level. In addition, high-throughput platforms exist to screen for desirable properties in the field of material science [ 41 , 42 , 43 ]. Although high-throughput platforms offer an unbiased method for discovering optimal material properties, they have their limitations since only a restricted part of the material design space is covered.…”

Section: Introductionmentioning

confidence: 99%

Natural Architectures for Tissue Engineering and Regenerative Medicine

Honig

Vermeulen

Zadpoor

et al. 2020

JFB

View full text Add to dashboard Cite

The ability to control the interactions between functional biomaterials and biological systems is of great importance for tissue engineering and regenerative medicine. However, the underlying mechanisms defining the interplay between biomaterial properties and the human body are complex. Therefore, a key challenge is to design biomaterials that mimic the in vivo microenvironment. Over millions of years, nature has produced a wide variety of biological materials optimised for distinct functions, ranging from the extracellular matrix (ECM) for structural and biochemical support of cells to the holy lotus with special wettability for self-cleaning effects. Many of these systems found in biology possess unique surface properties recognised to regulate cell behaviour. Integration of such natural surface properties in biomaterials can bring about novel cell responses in vitro and provide greater insights into the processes occurring at the cell-biomaterial interface. Using natural surfaces as templates for bioinspired design can stimulate progress in the field of regenerative medicine, tissue engineering and biomaterials science. This literature review aims to combine the state-of-the-art knowledge in natural and nature-inspired surfaces, with an emphasis on material properties known to affect cell behaviour.

show abstract

Computational discovery of extremal microstructure families

Abstract: We report the first fully automatic method for discovering microstructure families with extremal physical properties.

Cited by 82 publications

References 42 publications

A Bayesian experimental autonomous researcher for mechanical design

A Bayesian experimental autonomous researcher for mechanical design

Multiscale structural optimization towards three-dimensional printable structures

Natural Architectures for Tissue Engineering and Regenerative Medicine

Contact Info

Product

Resources

About