Materials-Genome-Based Multiscale Modeling of Ceramics and Laser-Assisted Machining

Xiangyang Dong, Purdue University

Abstract

This study is concerned with developing multiscale models to predict material properties and simulate laser-assisted machining of ceramics. The microstructure and composition of ceramics and computational costs are taken into consideration in the development of multiscale models. These models can predict both mechanical and thermal properties, which can further assist in selecting machining parameters for laser-assisted machining of ceramics. Quantum level, atomistic level, micro and macro scales are bridged in the multiscale modeling of ceramics. Interatomic potentials are derived by ab initio methods to achieve more accurate calculations in molecular dynamics (MD) simulations. MD calculations are carried out to predict the interfacial thermal properties and parameterize the traction-separation laws for interfacial glass phase within ceramics under thermal and mechanical loadings. The mechanical behavior of interfacial glass phase in ceramics is subsequently characterized by cohesive zone models (CZM) with the calculated traction-separation laws. The interfacial thermal conductivities and the parameterized CZMs are then input into the finite element model (FEM) to model the properties of thin interfacial glass phase surrounding the grains in the ceramics. To reduce computational costs, a materials-genome-based multiscale model is proposed to predict the material properties of ceramics by coupling multiscale model with a variational asymptotic method for unit cell homogenization (VAMUCH). A variational form for homogenization is formulated in combination with a cohesive zone model to predict crack formation within ceramics in a computationally efficient manner. The extended finite element method (XFEM) is further embedded in the formulation of the materials-genome-based multiscale modeling. The implementation of both materials genome model and XFEM enables multiscale modeling in predicting crack propagation while providing accurate predictions by considering heterogeneous microstructure. The developed multiscale model is capable of predicting the effects of microstructure, composition and temperature on materials properties and machining processes with greatly reduced computational costs. In predicting the mechanical and thermal properties of alumina ceramics, SiC ceramics and SiC/SiC composites, the favorable agreement between simulation results and experimental measurements confirms the validity of the multiscale models. A coupled thermal-mechanical multiscale model is developed to predict the thermally induced fractures within alumina ceramics under laser heating. Different compositions of alumina ceramics and crack propagation within SiC ceramics during laser-assisted machining are also studied experimentally and numerically using the developed materials-genome-based multiscale model.

Degree

Ph.D.

Advisors

Shin, Purdue University.

Subject Area

Genetics|Mechanical engineering

COinS