Over the past 18 months, a large data set has been developed on an A533B plate material removed from the decommissioned Shoreham nuclear power pressure vessel. The specimens in this data set were tested in the ductile-to-brittle transition regime using the procedures of the new ASTM Standard, Standard Test Method for Determination of Reference Temperature, T0 for Ferritic Steels in the Transition Range (E 1921). Results presented earlier  showed that crack length ratio has a clear effect on the reference transition temperature, and also that SE(B) and C(T) specimens define clearly different reference temperatures for this material. The observation that the extent of plasticity had little effect on the observed specimen geometry shift led to an investigation of the elastic T-stress as a potential measure of constraint for these results. It was found that T0 did correlate well with the median T-stress  obtained from standard equations  for the sample geometries used.
The effect of T-stress on specimen crack-tip constraint on T0 has been investigated recently by Gao and Dodds , who noticed the strong effect that the T-stress had on plane strain finite element models under small-scale yielding conditions. Using finite element analysis and a Weibull stress-based failure model, Gao and Dodds calculated and developed polynomial fitting relationships for their T-stress scaling function . Application of this method of constraint adjustment seemed promising  to predict the shift of the T0 reference temperature between different specimen geometries.
This work applies the T-stress based constraint correction functions of Gao and Dodds to the new A533B data set. The analysis shows that a Weibull modulus can be chosen that resolves the observed constraint differences between the shallow and deep crack SE(B) data sets, while a slightly different Weibull modulus is required when deep crack C(T) data sets are substituted for the deep crack SE(B) data sets. However, the constraint correction procedure does not account for the difference observed in this data set between the deep crack SE(B) and C(T) data sets.