Comparison of Parameter Methods for Extrapolating High-Temperature Data

Abstract

The use of the controversial and contradictory parameter techniques in correlation of high-temperature stress-rupture data is discussed. They are treated with a view toward their application in extrapolating short-time data to predict working stresses in serviceable heat-resistant alloys. Three parameter methods are compared on a statistical basis for their ability to reproduce the isothermal data on which they are based and in particular the longest time data currently available and usable for this purpose. Comparisons are further made with long-time extrapolations based on direct plotting of test data. The materials chosen for evaluation are alloys S-590, A-286, Nimonic 80A, and 1Cr-1Mo-1/4V steel. It is shown that prediction of long-time working stresses using parameter techniques will generally give better results than can be obtained from long extrapolations on double logarithmic plots. Among the parameters themselves, the Manson-Haferd linear parameter method gave the most reliable extrapolations. However, the results obtained using the Larson-Miller method may be improved considerably if the proper constant is selected for each set of data rather than using C = 20 for all data. For reliable extrapolations all methods require data from tests up to 1000 hr and covering adequate ranges of stress and temperature.