The Effects of Overloads in Service Load Histories on Crack Closure and Fatigue Damage 2001-01-4079

Fatigue life predictions for notched components are typically based on constant amplitude fully reversed strain-life data derived from smooth specimens. A mean stress parameter is applied to account for the effects of mean stress in the crack initiation location. Fatigue life predictions using this approach are usually accurate for constant amplitude fatigue but are almost always unconservative for variable amplitude service loading. A considerable amount of work has now related the unconservative predictions to reductions in crack closure during the large cycles in the variable amplitude load history which result in lower crack opening stresses for the small cycles than those in the constant amplitude reference tests used in design. This increases the effective strain range and the damage done by the small cycles and results in shorter than predicted fatigue lives.

This paper reviews work dealing with the build-up of crack closure in small cracks and relates steady state crack closure levels in constant amplitude loading to mean stress effects in fatigue. The nature of crack closure reductions by near yield stress tensile overloads and compressive underloads is examined and the build-up of crack closure during small cycles between overloads is described. Effective strain-life curves describing damage during closure-free stress-strain cycles are described together with test procedures for their acquisition in uniaxial and multiaxial fatigue. A test procedure using constant amplitude and periodic overload smooth specimen tests that gives an effective strain-life curve, calibrates a crack-closure model and allows the calculation of the constants for a crack growth rate versus effective stress intensity factor curve is described.

Simplified design procedures using effective strain-life fatigue data and effective crack growth rate data together with conservative crack-closure estimates are described. The use of these procedures for variable amplitude service load and strain histories is illustrated. Fatigue life predictions given by the strain-life analysis are good for smooth specimens and blunt notches but are conservative for small notches and flaws at notch roots. For the latter, a strain-based short crack fracture mechanics analysis using effective crack growth data give good life predictions.