Development of the T-0 reference temperature from precracked Charpy specimens

The master curve approach specified within the new ASTM E1921 Test Standard is a significant advance in defining an indexing temperature, T-0, the med ian fracture toughness, and associated failure probability bounds for ferri tic steels in the ductile-to-brittle transition regime. An objective in dev eloping: this standard test procedure has been that it should, if possible, allow the use of precracked Charpy-size specimens to measure T-0 and the a ssociated fracture performance. However, the supporting technical basis doc ument for E1921 [NUREG/CR-5504, US Nuclear Regulatory Commission, Washingto n, DC, November 1998] presents no experimental evidence which demonstrates the accuracy of T-0 estimates from precracked Charpy specimens. There was also little independent experimental verification presented [NURE G/GR-5504, US Nuclear Regulatory Commission, Washington, DC, November 1998] to support the limiting deformation criterion (M = 30) contained in the st andard. Computational and analytical support [NUREG/CR-5504, US Nuclear Reg ulatory Commission, Washington, DC, November 1998] of the deformation crite rion for precracked Charpy specimens is also lacking. The most applicable r ecent computational work by Ruggieri and Dodds [Engng. Frac. Mech. 60 (1) ( 1998) 19] supports a deformation criterion that is almost twice as stringen t as the E1921 standard to ensure that constraint loss does not occur. Hist orically, constraint loss has been evaluated far individual specimens, whil e E1921 implicitly only requires that constraint loss does not significantl y alter the median toughness. This work experimentally evaluates the use of precracked Charpy specimens t o estimate T-0 by comparing results obtained on both precracked. Charpy and 1T (or larger) fracture toughness measurements. The data demonstrate that the Charpy-size specimens tend to have a non-conservative bias, even when t he data fall within the allowable E1921 deformation criterion. This effect may be a function of the degree of crack tip constraint loss, which strongl y depends on the material flow properties. An attempt is made to correct fo r any constraint loss in the Charpy-size specimen and remove any bias in th e measured T-0 using a simple constant stressed volume cleavage failure cri terion [Int. J. Frac. 74 (1995) 131]. While this method does decrease the d ifference between T-0 values measured using Charpy and conventional 1T spec imens, more detailed cleavage initiation models which can be independently calibrated will be required to rigorously account for any constraint differ ences between Charpy and larger specimens. (C) 2001 Published by Elsevier S cience Ltd.