For determining the failure probability of ceramic components in a varying stress field, there are many theories which are based on the Weibull “weakest-link” ideas. However, the difference between the predictions for many stress systems is small and because of the scatter in the failure loads for ceramic materials and the inevitable experimental errors, it has been difficult to decide which is most suitable.
In this paper, a testing regime is described which spreads the predictions from the various theories to allow the most suitable theory for a given material to be chosen. This will give more confidence when designing with ceramics.
Tests are also described on a reactor grade graphite and it is shown that of the theories examined the maximum non-coplanar strain energy release rate is the most suitable criterion although the results indicate that current ‘weakest-link’ ideas may not be suitable to describe all loading situations. Reasons for this are discussed.
