A set of creep rupture data, for which two independent investigators have previously published models, is re-examined using the recently proposed parametric numerical isothermal datum (P-NID) approach to modelling. It is shown that neither of the existing models can be presumed reliable in extrapolation. In contrast, P-NID models are found to represent the data with much greater accuracy and to satisfy appropriate criteria for reliability in extrapolation. According to the principles of this approach, the close correspondence between the P-NID models and the data justifies extrapolation in rupture time at mid-range and lower test temperatures in excess of 15-fold. This expectation is confirmed by excluding from the analysis all data at more than 7% of the longest known rupture life, then comparing the resultant extrapolation to the excluded data. The success of the procedure suggests a potential for greatly reducing the extent of rupture testing necessary to produce reliable extrapolation to times of practical interest. This is demonstrated for a set of data including simulated results at intermediate temperatures. The study shows a potential reduction to less than one-tenth of the aggregate test time conventionally considered necessary. The analysis of further sets of rupture data shows how a P-NID model may reliably be extrapolated even when the material behaviour necessitates a model partitioned according to temperature range, or exhibits instability. Limited investigation indicates that, for a given material, creep strain data are consistent with parametric models for rupture data and may be extrapolated over similar intervals.
