Master Curve analysis of high-strength and ultra-high-strength steels using subsize specimens: Implications for fracture toughness estimation and CVN−T0 correlations

This work presents a critical assessment of commonly used CVN-T0 correlations for estimating fracture toughness in high-strength steels, using Master Curve (MC) analyses based on subsize specimens as a reference framework. The study examines the influence of specimen size and material characteristics on the estimation of the reference temperature T0 and evaluates the consistency of lower-bound toughness values obtained from established correlations relative to those derived directly from the MC methodology. The analyses are conducted on an API X80 pipeline steel and two ultra-high-strength (UHS) armor steels, covering distinct strength levels and microstructural characteristics. The results indicate that subsize specimens provide consistent and physically meaningful estimates of T0, supporting their use when full-size testing is not feasible. However, the predictive performance of CVN-T0 correlations is shown to be strongly material dependent and, in the case of UHS steels, potentially inconsistent even among steels of similar class. In particular, while reasonable agreement is observed for the X80 steel, the ultra-high-strength steels exhibit markedly different behaviors. Although one material shows partial agreement between Charpy-based estimates and experimental Master Curve results, the other exhibits a complete lack of correlation between predicted and experimentally determined T0-values. A comparative assessment using the Minimum of Three Equivalent (MOTE) procedure further highlights the variability and uncertainty associated with indirect toughness estimation from limited datasets. Overall, the present results show that while CVN-based approaches provide practical engineering estimates, their applicability requires careful validation against direct MC-based measurements, especially for high-strength steels whose microstructural and fracture characteristics fall outside the typical calibration range of existing correlations.