Engineering Significance of Statistical and Temperature-Induced Fracture Mechanics Toughness Variations on Fracture-Safe Assurance

Abstract

An appraisal is made of linear elastic fracture mechanics (LEFM) as a method of fracture-safe assurance for carbon and low-alloy steels. The theoretical advantage of an exact flaw size-stress level relationship offered by LEFM is contrasted with the limitations posed in actual application. These limitations relate to statistical variations in KIc and KId data. The variations considered here are (a) data scatter at a given temperature, and (b) toughness variations between different heats of a given alloy. In an engineering context, LEFM is considered applicable only in the temperature region representing the initial development of the brittle-ductile transition that characterizes low-alloy steels. In this region statistical variations in the data suggest that critical flaw sizes could be significantly smaller than the values calculated on the basis of limited experimental data. The prime objective in determining fracture toughness is for use in evolving a fracture control plan that assures structural integrity under a variety of environmental and loading conditions. Often the exact flaw size is unknown, particularly if the structure has not yet been built. Since the toughness increases sharply in the transition region, a practical solution is to take advantage of this behavior and choose a minimum operation temperature that assures a high fracture toughness such that postulated flaws cannot propagate in an unstable manner. The objective of being able to define the temperature range and statistical distribution of KId curves is met equally by the use of Dynamic Tear (DT) and KId tests. The DT test, as contrasted with LEFM methods, is shown to be an effective engineering tool with which to determine the Fracture Transition Elastic (FTE) temperature; above this temperature, plane strain constraint is lost for the given thickness, and flaws cannot propagate at stress levels less than yield. The determination of a minimum structural operating temperature based on dynamic LEFM values, when modified by conservatisms necessitated by statistical variations in the data and inaccuracies in temperature measurement, is shown to be essentially equivalent to the FTE temperature.