ABSTRACT
Brittle materials such as cast-iron are weak in tension. Tensile strength is limited by the presence of flaws, the character and distribution of which varies between nominally identical pieces of the same material. It follows that a range of failure loads will be obtained from tests on a batch of matched specimens and that strength prediction methods with a statistical basis may be needed. The probability that a piece of cast-iron will contain significant flaws that will cause fracture increases with the size of the piece, hence the term 'size effect'. This is a rather imprecise term as the type of loading and the shape of the piece are also important. For a prismatic tensile test specimen, the whole of the waisted section between the grips is subject to uniform tensile stress and is therefore at risk. For the same specimen tested in bending, the quantity of material subject to tension is smaller. Beam bending stresses are lower towards simple supports and towards the neutral axis, presenting a less serious condition within the volume at risk when compared to that for the tensile test.
