The fracture toughness of low carbon steels : the effects of grain size and temperature
Thesis DisciplineMechanical Engineering
Degree GrantorUniversity of Canterbury
Degree NameDoctor of Philosophy
For materials that exhibit a fracture mode transition as temperature is lowered one of the important criteria for material performance is the material's fracture mode transition temperature and not necessarily the specific fracture toughness at any temperature. Therefore, it is important to establish whether operating conditions place a structure below the selected material's transition temperature or that a material is selected with a transition below that of the structure's operating conditions. Quantitative design processes, based on linear elastic (KIC) and elastic-plastic (CTOD) fracture mechanics using experimental fracture toughness data, allow the design of safer structures. In recent years standard procedures have been adopted for KIC and CTOD testing. Using the traditional Charpy V-notch impact test, detailed information on the effects of composition and grain size on the fracture mode transition temperature are known. The fracture mode transition temperature is not as equally well understood in CTOD or KIC testing, especially for low carbon steels. The CTOD and Charpy impact tests have been used to determine the grain size dependence of the fracture mode transition temperature for two low carbon (structural) steels, one of low active nitrogen content and one of high active nitrogen content. Both the CTOD and Charpy tests show a fracture mode transition over a narrow temperature range. It was established from theoretical derivation and experimental observation that there is a linear dependence of the transition temperature TC on the reciprocal square root of grain size (d-½) for both the CTOD and Charpy tests i.e. Tc = B₀ + B₁d-½ where B₀ and B₁ are constants. When the results of the CTOD and Charpy tests are compared the magnitude of B₁ is significantly different for each test. It was concluded that the difference in B₁ between the two tests is due to the different strain rates of the tests and that the strain rate significantly affects the local yield stress around the crack tip or notch. Micromechanical modelling of fracture toughness predicted a variation in transition temperature with variation of grain size but this did not show a linear dependence on d-½. The predicted transition temperature was a lower bound of the range in transition temperature. The observed decrease in transition temperature with grain refinement when using the CTOD test is explained by the increase in crack initiation and crack propagation energy necessary to overcome grain boundary resistance to fracture. For example, at the fracture mode transition temperature for the low nitrogen steel, the proportion of energy required to overcome grain boundary resistance to fracture increased from 39% at d-½ = 4.218 mm-½ to 55% at d-½ = 9.939 mm-½ of the total critical energy released. Also, it is thought that grain refinement means a lower critical crack-tip strain is needed for transition. Correlations between Charpy Impact Energy (Cv) and CTOD (δc) or KIC suggested a suitable relationship was δc (or KIC) = D(Cv)n. The constants D and n were independent of grain size but were composition dependent. The temperature shift showed a grain size dependence, given as ∆T = ∆B₀ + ∆B₁d-½ for the data available. The CTOD measured from Clip Gauge Displacement was determined to be grain size and composition dependent. From a technique using silicone-rubber replicas of the crack tip the CTOD was found to be a function clip gauge displacement (Vg) and grain size (d). namely, δt = 0.121113 Vg + 0.034222 Vg²d ½, for the compact tension specimens tested. For toughness calculations and determining the fracture mode transition the temperature and grain size dependence of the steels' yield stress (at constant strain rate) was determined. Using the Hall-Petch equation (σys = σi + kyd-½), a suitable model was found to be σys =A₁ + A₂T + A₃T² + A₄T³ + kyd-½ where the constants A₁, A₂, A₃, A₄ and ky were determined by multiple-linear regression analysis from experimental data over the temperature range -196 to +65°C.