Reliable Fatigue Design, by Rigid Rules, by Magic or by Enlightened Engineering

This paper discusses some basic differences between engineering approaches to fatigue reliability assessment. The comparison is based on the fact that mechanical fatigue is a highly empirical science with large uncertainties in prior knowledge, both regarding observed data and model accuracy. It is claimed that the amount of available prior physical knowledge should decide the level of complexity in reliability tools, not computer resources nor mathematical theory. As a result the second moment statistics is put forward as a useful tool in fatigue reliability assessment, either in the regulated form like in Eurocode or in a free engineering form like through the Variation Mode and Effect Analysis. The latter method is outlined in certain detail and its practical usefulness demonstrated for some engineering reliability problems.