Overcoming Limitations of the Conventional Strain-Life Fatigue Damage Model

Abstract

The strain-based approach to fatigue life prediction usually relies on the conventional strain-life equation which correlates the elastic and plastic strain to the life. The correlation is based on separate log-linear curve fits of the elastic and plastic components of the strain data versus the life. It is well known, however, that these linear relationships may be valid only within a specific interval of stress or strain. When material behavior approaches elastic-perfectly plastic, for instance, it is not uncommon for the test data to deviate from linearity at both very high and very low strains. For such materials a separate fit of each curve is likely to give material constants significantly inconsistent with the fit of the cyclic stress-strain curve, especially if a good local fit over a restricted interval is obtained. In this work, some of the errors that arise as a result of this inconsistency are described, and recommended methods are developed for treating these errors. Numerical concerns are also addressed, and sample results are included.