In 1939, Weibull [1] proposed a model for the failure of materials that could be used to predict the increased strength observed in bending specimens when compared with tensile specimens. The model was based only on the statistical scatter in experimental results and the assumption that the weakest point in the structure governs the failure of the entire structure. The weakest-link assumption implies that there is a greater probability for failure in larger structures than in equally stressed smaller structures. This follows from the fact that there is a greater probability for a critical flaw in the volume of the larger structure. Of course, weakest-link theory will result in conservative predictions for the failure probabilities, since stress can be redistributed when the weakest point fails. Hence, the conservatism may significantly overpredict the failure probability in metallic structures, but it should be accurate in ceramic structures. Composite material structures should lie intermediate between metallic and ceramic structures. Many aspects of the effects of size on strength have been examined. Harter [2] has produced a review of the early literature on size effects in material structures. Weakest link has been extended to the prediction of failure locations [3], time dependent failure [4], multi-axial stress states [5], and composite material structures [6].This chapter will first present the basic theory, and then this will be extended to the prediction of failure locations. Time-dependent failure will then be discussed, and the results will be extended to composite materials.
Basic Theory for Static StructuresThe development of weakest-link theory for static structures presented here will follow the discussion provided in the literature [6]. Consider an infinitesimal volume element, d V , at a point x in a total volume V , and let the stress be s ij at point x . Assume that the infinitesimal probability for failure, d f , in volume element d V is proportional to the volume element and is a function of the stress state. ThenFinally, weakest-link failure can be readily applied to the design of composite material structures. Although the accuracy for composite structures is not as good as for ceramic structures, it still provides considerable accuracy and can be readily adapted to represent the large number of competing failure mechanisms present. An excellent guide for developing these models is the use of existing damage models.