Statistical Methods for Creep, Fatigue and Fracture Data Analysis

Statistical methods can play an important role in the proper collection and analysis of creep, fatigue and fracture data. This article deals with what statistics has to offer in this area. The following subjects are considered: The role of statistics in problem definition; the use of stress versus time as the dependent variable in the data fitting; treatment of variability between heats; analysis of data with run-outs; exclusion of extreme high stress-low failure time data from the analysis; experimental design concepts for obtaining the most useful data. The article is based upon a paper prepared at Bob Goldhoff’s suggestion and with his substantial assistance. This paper was originally presented to a 1977 NASA/EPRI/MPC Workshop.