MODELING OF NONUNIFORM COMPOSITE MICROSTRUCTURES

The microstructural effects of non-uniform composite microstructures a re modeled. Relationships are observed between the degree of non-unifo rmity and fiber spatial information which may be measured via image an alysis or derived from Dirichlet cell tessellations. For artifical pat terns containing chain-like clustering which simulate composite micros tructures: (a) the nearest-neighbor distances of random and clustered patterns are smaller than those normally estimated by square or hexago nal arrays, (b) increased clustering may be associated with increased mean cell volume fraction, cell volume fraction distribution standard deviation, number of cell sides, distribution standard deviation, near est-neighbor distance distribution skewness, and decreased nearest-nei ghbor distances and (c) accounting for non-uniform fiber diameters is generally unnecessary, except possibly at low fiber volume fraction or with patterns exhibiting a high degree of chaining. For Nicalon SiC/z irconia titanate composite samples with microstructures which exhibit clustering, the maximum value of skewness, determined from sub-regions of the sample, correlates with the flexure strength of that sample.