ACCELERATED TESTING

Design for high temperature applications currently calls for results from long time creep-rupture tests. Although extensive data on minimum creep rates and rupture lives extending to 10,000 hours or more now exist in generally available databanks for many widely used alloys, such data may not be available when specific needs arise. For example, for reasons other than creep strength optimization, an alloy may be modified by chemistry or process changes; design information is then needed to allow introduction of the alloy modification or to ensure that the alloy creep strength is not compromised. There are also life management decisions for major components that need to be made during routine maintenance, or special cause operating downtime, that require an assessment of the current creep strength after service exposure. Because of time constraints, both examples must rely on some form of accelerated testing.

Several options for testing are available:

If there are no major time restrictions, conventional creep rupture testing of samples over a range of temperature and stress for times up to about one year would be appropriate, with interpolated time/temperature (e.g. Larson-Miller or Manson-Haferd) parametric analysis. Note of any embrittlement in terms of ductility loss could be made if the evaluation is for service exposed material.

The next option, with some time constraint, would be accelerated testing at the design stress but higher temperatures. Extrapolation to design temperature would give the desired rupture life. This approach has been shown to be consistent with the general observation that the linear life fraction rule for failure under nonsteady conditions works quite well for temperature changes.  Some indication of embrittlement could also be obtained.

The next option is to predict life based on an interrupted creep test. If we stop the test at say 3% strain (after the minimum creep rate) we can calculate the desired rupture life several ways: using a time to 1% creep vs. time to rupture plot or a stress to 1% creep vs. stress to rupture plot (Gill-Goldhoff). We may also use the data to project failure strain using the strain rate sensitivity estimated from the creep curve (or the Omega value in the Materials Properties Council approach). Although these standard creep test approaches are helpful in some situations, the time saving compared with a long time creep to rupture test may be only between 50% and 80%. Note also, the strain rate sensitivity may be estimated from the stress relaxation test so we can get a measure of ductility (we have a universal correlation between strain rate sensitivity and ductility) from the SRT test.

The best short term (less than one week of testing) test is the stress relaxation test (SRT), with a constant displacement rate (CDR) test providing an independent measure of embrittlement. Three or four 20 hour test runs can provide a plot of creep rate vs. a time/temperature parameter. Using the Monkman-Grant relation, the rupture life at the desired stress and temperature can be estimated. Alternatively, the data may be used to compute the stress for a creep strain of 1% as a function of temperature. This may be converted to a rupture life under the desired test conditions using the Gill-Goldhoff correlation. This approach is the one generally advocated, and is described in detail on this site and in the open literature. The figure shows an example of rupture lives projected from stress relaxation data for T91 steel compared with actual long time test results.

Recently, the small punch test has been used to calibrate with standard tensile creep data. Although this has a clear advantage in terms of sample size, it does not specifically address the extrapolation issue.

A final option would be to use the correlation between room temperature hardness and rupture life in many ferritic steels to estimate remaining life. This approach has been useful to evaluate service damage due to microstructural coarsening in ferritic steels. However, it is not considered widely applicable. Hot hardness measurements can expand the scope for the application of hardness measurements, but experimental difficulties have limited their use to date.