A1 Brittle Fracture and Weibull Statistics

Topic

The strength of a ceramic is not a single value but a statistical distribution: the Weibull distribution. This distribution arises because internal defects in the ceramic are random in both size and location—each piece contains internal flaws with sizes and positions that cannot be predicted beforehand—and it is this randomness that causes strength to manifest as a distribution rather than a fixed value.

These internal defects are specifically cracks; each crack independently determines the stress level at which the ceramic fails in its vicinity. The Griffith Criterion establishes the relationship between failure stress and crack size, specifically dictating that a larger crack size results in a lower stress required for the ceramic to fail at that point. This criterion is not defined here from scratch; instead, the model adopts it from the established fracture model for metals.

Because each crack determines a different failure stress based on its size, and because crack sizes vary randomly from one piece to another, the failure stress of the entire ceramic—its strength—follows the aforementioned Weibull distribution rather than a single, fixed value.

The Weibull modulus characterizes the degree of scatter in strength within this distribution. A high modulus indicates more consistent strength with less variation between pieces, whereas a low modulus indicates greater scatter.

The effective volume of a ceramic piece—the amount of material actually subjected to the relevant stress—also influences the probability of that piece failing. A larger effective volume increases the likelihood that the volume contains one of the largest cracks within the random distribution of defects, thereby raising the probability of failure at a given stress level; this is the same size-scaling effect that explains why a large piece of a given ceramic material withstands less stress, on average, than a small one.

Therefore, predicting the failure of an actual ceramic component requires more than just knowing the material's average strength—the starting point for this model. It also requires the Weibull modulus (which describes the scatter in that strength), the effective volume of the component under stress, and the Griffith criterion (which links the size of a specific defect to the stress level at which that defect causes the ceramic to fail). None of these four elements alone makes such a prediction possible.

ID:4396