Comparison of ductile-to-brittle transition curve fitting approaches

Ductile-to-brittle transition (DBT) curve fitting approaches are compared over the transition temperature range for reactor pressure vessel steels with different kinds of data, including Charpy-V notch impact energy data and fracture toughness data. Three DBT curve fitting methods have been frequently used in the past, including the Burr S-Weibull and tanh distributions. In general there is greater scatter associated with test data obtained within the transition region. Therefore these methods give results with different accuracies, especially when fitting to small quantities of data. The comparison shows that the Burr distribution and tanh distribution can almost equally fit well distributed and large data sets extending across the test temperature range to include the upper and lower shelves. The S-Weibull distribution fit is poor for the lower shelf of the DBT curve. Overall for both large and small quantities of measured data the Burr distribution provides the best description.

► Burr distribution offers a better fit than that of a S-Weibull and tanh fit.
► Burr and tanh methods show similar fitting ability for a large data set.
► Burr method can fit sparse data well distributed across the test temperature.
► S-Weibull method cannot fit the lower shelf well and show poor fitting quality.