Metamaterials are carefully architected solids that exhibit material properties oftentimes not found in nature. This student project focuses on the computational design of a new class of cellular metamaterials (fabricable by additive manufacturing) whose carefully engineered microstructure results in controllable geometrical features (e.g., curvature, porosity, surface area, periodicity, etc.) and mechanical properties (e.g., anisotropic stiffness, strength, wave dispersion, etc.). By modeling the process of phase separation between the void and solid phases in a given periodic volume element undergoing spinodal decomposition, such cellular media with quasi-random microstructures were obtained previously. However, simulating the phase separation process is computationally expensive and limits the exploration of the design space.
This project proposes the development of a computational design framework, wherein an optimal microstructure is produced for a given set of design requirements. This inverse problem is explored, e.g., to quantitatively understand appropriate architectures for target anisotropic elastic moduli. The framework will leverage (i) the statistics of the microstructural geometry to bypass the computational requirements of simulating the phase separation process and facilitate large data sets of microstructures; (ii) machine learning techniques to explore the design space and quantitatively model the relationship between design parameters and structural properties. The project is to some extent flexibly adjustable to your ideas and interests.
Interested students should have a background or interest in at least two of the following areas:
- computational mechanics and metamaterials
- probability and stochastic processes
- data analysis, machine learning, deep learning
Additionally, programming skills in python/C++/MATLAB are preferred.
Metamaterials are carefully architected solids that exhibit material properties oftentimes not found in nature. This student project focuses on the computational design of a new class of cellular metamaterials (fabricable by additive manufacturing) whose carefully engineered microstructure results in controllable geometrical features (e.g., curvature, porosity, surface area, periodicity, etc.) and mechanical properties (e.g., anisotropic stiffness, strength, wave dispersion, etc.). By modeling the process of phase separation between the void and solid phases in a given periodic volume element undergoing spinodal decomposition, such cellular media with quasi-random microstructures were obtained previously. However, simulating the phase separation process is computationally expensive and limits the exploration of the design space.
This project proposes the development of a computational design framework, wherein an optimal microstructure is produced for a given set of design requirements. This inverse problem is explored, e.g., to quantitatively understand appropriate architectures for target anisotropic elastic moduli. The framework will leverage (i) the statistics of the microstructural geometry to bypass the computational requirements of simulating the phase separation process and facilitate large data sets of microstructures; (ii) machine learning techniques to explore the design space and quantitatively model the relationship between design parameters and structural properties. The project is to some extent flexibly adjustable to your ideas and interests.
Interested students should have a background or interest in at least two of the following areas: - computational mechanics and metamaterials - probability and stochastic processes - data analysis, machine learning, deep learning
Additionally, programming skills in python/C++/MATLAB are preferred.
Not specified
Sid Kumar,
Mechanics & Materials,
Department of Mechanical and Process Engineering,
Leonhardstrasse 21, LEE N225,
email: siddhant.kumar@ethz.ch
Sid Kumar, Mechanics & Materials, Department of Mechanical and Process Engineering, Leonhardstrasse 21, LEE N225, email: siddhant.kumar@ethz.ch