TWI Industrial Member Report Summary 114/1980
By S J Garwood
Background
Criteria governing the use of laboratory bend specimens for the prediction of conservative estimates of structural crack growth resistance curves in ductile steels are discussed.
For the characterisation of through-thickness cracks it is sufficient to test full plate thickness bend specimens. However, for the description of surface defects a specimen thickness criterion must be satisfied.
It is suggested that the criterion B > 25JR/sigmaY (which is equivalent to B > 50delta) is used for the entire crack propagation region. Hence, as the resistance curve increases so the thickness criterion becomes more severe.
The validity of this criterion is investigated using oversquare bend specimens, 250mm thick, to compare with:
i. surface notch tension wide plate tests with an effective thickness of 508mm,
and
ii. 55 and 110mm thick bend specimens in A533B Class I tested in a previous research programme.
The oversquare specimens have a lower resistance curve than the thinner bend specimens and this curve also falls below the surface notch tension scatterband. However, the oversquare specimen data lie within the semi-elliptical surface notch tension scatterband, but this may be due to the conservative estimates employed in the evaluation of the results from this latter geometry.
Deeply side-grooved specimens are also employed in the surface notch orientation to induce high constraint and eliminate the formation of shear lips. These specimens give very conservative estimates of both the tension geometries and of the bend configurations.
It is not clear at present whether:
side-grooving induces a greater degree of constraint than thickness alone can provide and that the oversquare specimens are in fact giving 'plane strain' resistance curves,
or
if the thickness criterion presently suggested is not restrictive enough and the side-grooving actually allows thin specimens to achieve 'plane strain' type conditions which are not attained in the oversquare specimens.
The use of the maximum load toughness values from laboratory bend tests for the prediction of structural ductile instability toughness is discussed. It is concluded that, provided the conditions relating to the attainment of lower bound resistance curves are adhered to, then the use of maximum load values in design curve procedures should provide conservative estimates of critical flaw sizes in ductile structures. The level of over-conservatism inherent from the use of initiation of tearing values for the prediction of ductile instability in tension is also demonstrated.
It is recommended that, to provide a consistent definition, the crack opening displacement measurement in the presence of a ductile tear should refer to the original crack tip position.
Finally, various methods of allowing for the effect of the extending crack on the J resistance curve are investigated. It is concluded that these methods have very little effect on the resistance curves in A533B Class I whilst within the presently suggested limitations governing J controlled growth (i. e. crack extension of <6% of the remaining ligament). Hence, if evaluations of J are to be restricted to this regime, crack growth corrections to the standard experimental evaluation method for bend specimens are unnecessary. Crack growth corrections do, however, have significant effects for larger amounts of crack extension and must be applied to ensure the prediction of conservative estimates of structural resistance curves from laboratory tests.