The disparate data problem: The calibration of creep laws across test type and stress, temperature, and time scales

Abstract

Creep data exists in a diaspora. It is not always available in the regime (short, intermediate-, and long-term) or form (minimum-creep-strain-rate, creep deformation, stress-rupture, etc.) of interest. This is particularly true for new alloys where the time-to-qualification can take between 10 and 20 years. For new alloys, extrapolation is performed using short-term creep data. When incorporating the uncertainty of the creep data, the probability distribution in extrapolations expands with time; eventually leading to unrealistic predictions. Standards recommend that extrapolation should not exceed three times the duration of the longest experiment. For example, a creep deformation model calibrated using short-term creep deformation data (e.g. 103 h) is highly unlikely to generate realistic creep deformation predictions near 105 h. This study proposes integrating disparate creep data in the calibration process of continuum damage mechanics based constitutive laws that are capable of simultaneously predicting minimum-creep-strain-rate, stress-rupture, deformation, and damage. It is hypothesized that cross-calibrating CDM constitutive models to any and every form of available creep data (across test type, stress range, and temperature range) will lead to more reliable extrapolations. This is demonstrated by integrating minimum-creep-strain-rate, stress-rupture, and short-/intermediate-term creep deformation data into the calibration process of two CDM constitutive models. Successful extrapolations are performed for minimum-creep-strain-rate, stress-rupture, and creep deformation. In addition, the approach facilitates the development of creep design maps that can be used for material selection and design and to compare the interpolation and extrapolation ability of the existing creep models.