Failure of the ASTM E 1921 master curve to characterize the fracture toughness temperature dependence of ferritic steel and successful application of the stress distribution T-scaling method

Abstract

The fracture toughness KJc of a material in the ductile-to-brittle transition temperature (DBTT) region shows a large temperature dependence, large scatter, and specimen thickness dependence. Thus, a large amount of data is necessary to satisfy the design requirements of equipment that experiences temperature changes. The ASTM E 1921 master curve (MC) method provides an engineering approach to solve these three issues and is intended to be applicable to an arbitrary ferritic material with a yield stress of 275–825 MPa. This study tested the CrMo steel JIS SCM440, which satisfies these conditions, but the ASTM E 1921 MC failed to characterize its fracture toughness temperature dependence. The material was tested at four temperatures: −55, 20, 60 and 100 °C. The obtained reference temperature T0 (i.e., the temperature at which a 25 mm thickness shows a fracture toughness of 100 MPa·m1/2) from these tests differed in the range of −9 to 89 °C. T0 from all of the test data was 16.6 °C. Thus, the MC using any of these T0 values failed to characterize the material’s fracture toughness temperature dependence. This material showed a T0 value higher than room temperature, which is a rare case for materials to which the MC has been successfully applied. The ASTM E 1921 MC is based on the Zerilli equation, which showed a poor fitting performance of the yield stress above room temperature for this material. When this yield stress temperature dependence was corrected (i.e., a temperature shift), the ASTM E 1921 MC better fitted the material’s fracture toughness test data. However, some discrepancy still remained. Because the Zerilli equation showed a small yield stress change and may differ from the experimental results above room temperature, the ASTM E 1921 MC may fail to characterize the fracture toughness temperature dependence close to or above room temperature. In contrast, the stress distribution T-scaling method was demonstrated to successfully predict the KJc temperature dependence of SCM440.