The currently codified fracture toughness testing procedures, e.g., ASTM E 813, E 1152, E 1290, and BSI 7448:Part 1 use a set of inference equations to obtain J and CTOD from the measurements of global displacement and load. These inference equations were originally developed by assuming homogeneous and perfectly plastic material properties. Inaccurate results may be obtained when these inference equations are applied to non-homogeneous specimens and materials with strain hardening.

A systematic study of the relationship between crack driving force (J and CTOD)and the remote load and displacement is conducted using finite element method. The relationship derived from the study is compared with the inference equations used in the current standards. The analyzed geometry is a single-edge notched bend specimen, or SE(B), with a parallel-sided weld at the center. By varying the strength ratio between the weld and the base metal, a range of non-homogeneity is achieved. Other variables studied include crack depth, weld width, and strain hardening level.

The accuracy of the currently codified J and CTOD inference equations and those proposed in recent years (i.e., new equations) is examined. Compared with the codified inference equations, the new equations can be applied to a wide range of crack depth (0.1 ⩽ a/W ⩽ 0.5). The new CTOD inference equations provide much more accurate CTOD values for high strain hardening material than is possible using the current standards. The accuracy of the codified inference equations and the new equations is expressed in terms of weld width, strain hardening rate, and mismatch levels.

The CTOD inference equations are tested through corroboration with experimentally measured values of CTOD in an un welded HSLA structural steel. Interrupted fracture tests were performed at several temperatures. Subsequently the specimens were sectioned in the plane of the mid-thickness and quarter-thickness for measurement of the residual plastic component of CTOD. These experimentally measured values of CTOD were compared with those derived from the codified and the new inference equations. The new equations provide better agreement with the experimental measurements than the codified equations.