The problem of creep-rupture life (CRL) prediction for high-temperature alloys and special steels is very substantial for designing turbojet engines, turbines, chemical, and nuclear reactors, steam boilers, and other power equipment. The problem has a long history and extensive literature. Difficulties in the resolution of this problem are caused by a colossal complexity of micro-physical mechanisms that form the long-term destruction processes in modern steels and alloys at creep.

Numerous traditional methods for CRL prediction based on the theory of thermo-activated processes do not give enough reliable results. They include the widely known F. R. Larson's and J. Miller's time-temperature parametric method, suggested in 1952, and a number (about 100) of similar time-temperature parameters. At the end of the 70s of the last century, a special group led by R. M. Goldhoff generalized and improved this experience by creating the Minimum Commitment Method recommended for use in the USA. However, a significant part of empirical knowledge in the time-temperature parameters causes the uncertainty inherent to this methodology and limits the generality of the obtained findings. The extrapolation power (the ratio of the obtained time-to-rupture prediction to the maximum test time) of traditional methods is usually no more than approximately 20.

In the traditional approach, predicting equations contain as a rule a few special empirical coefficients. Their values depend on the classes of the tested materials and test regimes. To determine the coefficients for each specific material, a significant amount of preliminary experimental and analytical work is required.

In the suggested new approach such predetermined coefficients are absent. Thus, it becomes possible to apply this approach to the development of new materials that have no analogs.

The suggested approach uses a specially developed version of the known phenomenological theory that interprets the gradual rupture caused by creep as a process of accumulation of diffused damage (micropores, microcracks) in the material over time. While known models of damage accumulation lead to the hypothesis of linear damage summation at non-stationary loads, the new approach gives a more realistic and universal nonlinear damage summation principle.

The new approach along with its algorithmic realization provides a good approximation to the real kinetics of damage. This allows one to achieve high certainty of the CRL predictions along with extrapolation power of more than two orders of magnitude relative to the maximum test time. Thus, significant savings in test time, labor, and money are provided when working out new high-temperature metallic materials and exploring CRLs of existing ones.
In practical terms, results from standard CRL tests with 1 to 4 (or up to 6) months in duration will give a reliable prediction of CRL over a period of 10 to 40 years.

The method requires some time-limited results of standard creep rupture tests at a given constant temperature. The results of the tests represented as pairs of values σ – t*, where σ – stress, t* – time to rupture, are being used as the only required initial data for the next steps in the calculations. The calculation process based on the method theory includes classical and modern statistical procedures and allows obtaining time-to-rupture predictions for the given (comparatively small) stress. Predictions can be made for both the medium probability of rupture (p≈0.5) and the corresponding low probability (e.g. p≈0.025), and if necessary, a more detailed “survival analysis” of the material can be carried out. To obtain a CRL prediction, a work in interactive user mode with a specially developed computer program is required.

Some examples of the results, obtained by using the method for diverse materials at various test regimes, are presented in Fig.1 – Fig.10:

The power of extrapolation of the method, confirmed by the available data of prolonged control tests (courtesy NIMS), in these examples was in the range from ~65 to ~170 while the predicted time-to-rupture was in the range from ~90,000 h to ~358,700 h (from ~10 years to ~41 years).

Conclusions:

  • The suggested method provides reliable CRL predictions with exceeding the maximum test time approximately by two orders of magnitude

  • The method is applicable for high-temperature metallic materials of various chemical compositions
    •
  • The method may be used for the materials running at a wide range of temperature-stress conditions
    •
  • There is a possibility to apply the method to develop new high-temperature metallic materials that have no analogs
    •
  • An objective evaluation of this new method is possible based on the results of short-term standard tests on CRL (at least three pairs of values ​​σ – t* without specifying the class of material and test temperature are required)
    •