No abstract

A combined experimental-numerical method was used to investigate the role of microstructure on the fracture of advanced ceramics. In particular, the effect of grain size and matrix content were examined. Two dimensional representative finite volume (FV) microstructures were created using Voronoi tessellation to synthetically represent the microstructure of a two phase ceramic composite. It is shown, by comparing with real micrographs, that the method captures the features of real microstructures in terms of grain size distribution, grain aspect ratio and the distribution of second phase agglomerations. Simulation results indicate the computed elastic parameters are within the Hashin-Shtrikman bounds and also agree well with the Eshelby-Mori-Tanaka method. It is found that the underlying microstructure significantly affects the local stress and strain distributions in these advanced ceramics.

No abstract

Micromechanical models of two-phase ceramic composites are created using a modified Voronoi tessellation approach. These representative Finite Volume (FV) microstructures are used to investigate the role of microstructure on fracture of advanced ceramics. An arbitrary crack propagation model using a cell-centred finite volume based method is implemented. In particular the effect of matrix content is examined. It is shown that the underlying microstructure significantly affects the local stress and strain distributions for a two-phase ceramic containing hard particles in a softer matrix. Simulation results indicate that an increase in the volume fraction of these hard grains leads to an increase in strength of the composite material. Furthermore, it is found that the homogeneity of the microstructure affects the overall strength.

Flexural Strength data of a number of grades of polycrystalline diamond (PCD), and polycrystalline cubic boron nitride (PCBN) were analysed using Weibull, normal and lognormal distributions. The role of microstructure in the failure mechanism of such material was analysed in terms of the chosen strength distributions. The best-fit distributions were determined using the maximum log-likelihood criteria and a comparison between the best and worst fit was conducted using the Akaike Information Criteria (AIC). Both large and small specimens were tested to investigate possible volume scaling effects for these materials. The different microstructures between the two materials was shown to have an effect on the statistical strength distributions. It was found that for PCD, in general, a lognormal distribution provided a better fit than the other distributions and no specimen size effect was observed. For PCBN a significant specimen size effect was observed and * Corresponding Author