A Comparative Study of Methods for Estimating Bolt Fatigue Limits

The approach to predicting fatigue limits from calculated stress concentration factors, using thread load distributions obtained from analytical theories is examined. In particular combinations of the methods of Shigley, Otaki, Heywood, Birger, Bluhm and Flanagan, Sopwith and a modification to Sopwith's theory are assessed against fatigue test data and photoelastic results for a range of bolt and nut geometries. The Snow‐Langer‐Cook, the Goodman and the Gerber methods of allowing for the effect of mean stress were also examined. It was concluded that mean stress effects are significant, that none of the theories are sufficiently sensitive to changes in geometry, and that they seriously overestimate the fatigue limit when the geometry is more severe than the standard. The use of Sopwith's theory of load distribution combined with Heywood's method and the Snow‐Langer‐Cook technique was found to be the safest for design purposes, but not the most accurate.