61 
54 
Notch brittleness 
163. Minute imperfections or oracks in the steel oan give rise to high 
local stresses and, in particular, there is reason to believe that triaxial 
tensions may be developed locally at the end of a sharp crack with a 
conse quent tendency to brittle fracture. This has been investigated 
extensively by testing specimens with notches cut in them and it is found 
that brittle fractures can then ocour when similar unnotched specimens 
exhibit a ductile fracture. In view of this, the term notoh brittleness 
has becon® a general term to describe brittle fracture where it is suspected 
to be due to stress concentrations arising from cracks or local inhomogeneity 
of the material. There seems little doubt that such notch brittleness is 
one of the most important factors which influence rupture. In particular, 
it seems clearly established that notch brittleness increases with decreasing 
temperature, so that for example a notch specimen may fail with a relatively 
ductile fracture at a room temperature of 60°F whereas a similar specimen 
tested in an ice-bath may exhibit a brittle fracture. With regard to 
chemical composition there is some evidence that manganese is beneficial 
in tending to inhibit brittle fracture and that a minimm manganese content 
of 0.50% is desirable, especially for ships’ plates over 4 inch thick. 
Shape and size of steel specimen or target 
164. When a normal tensile specimen is tested and exhibits a ductile 
fracture it is well know that up to @ certain load, the specimen tends to 
atretch uniformly but then develops a concentrated local neck where rupture 
finally occurs. The energy absorbed in the specimen can be considered 
broadly as composed of two parts, 
(1) the energy used in uniform extension of the specimen 
(2) energy absorbed in local necking. 
For spécimsns of given diameter but varying length made from the sam steel, 
the former energy will tend to increase in proportion to the length whereas 
the local necking energy absorption will remain relatively constant provided 
all specimens are reasonably long compared with diamster. The shape of the 
Specimen as defined by the length/diamter ratio thus influences the average 
amount of energy absorbed per unit volums of the steel even though the 
applied load is nominally uniform throughout the specimen, 
165. <A second effeot is the size effect which occurs for Specimens of 
geometrically similar shape differing only in overall dimensions, each 
Specimen being e smaller or larger soale replica of the other specimens. 
Thus, if similar notched specimens are submitted to similar statio bending 
conditions up to fracture it is found that the larger the specimen the 
Smaller is the relative deformation at fracture as measured, for example, 
by the ratio of central deflection to span if the Specimens are support.d 
at each end and subjected to a central load Correspondingly, the energy 
absorbed per unit volume up to fracture decreases as the size of Specimen 
inoreases. The precise msohanism for this size effect has yet to be 
established but it is undoubtedly related to the fact that although ths 
specimens are true soale replicas on a macroscopio scale their microscopic 
structure has not similarly been scaled, 
166. This size effect has also been observed in underwater explosion 
effects using different sizesof drum model targets constructed as scaled 
replicas and subjected to scaled charge conditions. Here it was found, 
that whereas deformations prior to rupture soaled reusonably well, large 
targets ruptured when dished (relatively) to a less extent than small 
targets at rupture. It has in faot long been knowm from empirical 
observation that damage by underwater explosions to full-scale targets 
tended to be greater than in corresponding small-scale trials, This 
Sise effect on rupture is thus especially relevant to quantitative 
interpretations of model tests. 
