Prediction of creep strain and creep strength of ferritic steels for power plant applications

Of the conventional time-temperature parameters (TTP's) for creep rupture, one of the most common is the Larson-Miller (LM) expression, which unfortunately predicts relatively inaccurate and optimistic values of extrapolated life. Some improvement can be expected by using better models such as the Manson-Haferd (MH) expression, which also can apply a low number of fitting constants, or other more complex standard models if there is sufficient data. However, similarly accepted models have not been standardised for creep strain data. In an example of constant stress creep data of 2¼Cr-1Mo (10CrMo9-10) steel, acceptable fitting of the time to given strain was obtained using a simple and robust modified MH model. The model involves four free fitting parameters, and provides a relatively good fit and an easier fitting procedure than a the more complex strain hardening model tested on the same data. Comparison to the standard creep strength of the 2¼Cr-1Mo steel suggests that the predicted long term creep strength from short term data of the example material tends towards increasingly optimistic values with increasing time and temperature. This difference appears more pronounced for time to rupture than for time to 1% creep strain, and is probably at least partly because the standard values originate from constant load testing with a systematically shorter time to rupture than under constant stress. The example involves a single data set, but demonstrates the need to avoid excessive extrapolation beyond the range of data. The conventional rule of maximum extrapolation up to three times the longest testing time is usually accepted for creep rupture. A somewhat more extensive extrapolation could be tolerated for the time to a modest strain up to about 1%