The use of constant probability fatigue curves in predicting deaerator vessel failure

This paper gives an assessment of some of the studies which have attempted to apply statistical distributions, or probability models, to the fatigue c rack growth process in an effort to structure fatigue crack growth predicti ons in a more reliable manner. The fatigue crack growth process is defined in terms of the well known Paris-Erdogan law which successfully predicted t he fatigue process at intermediate crack growth rates, da/dt. The following expression was developed to generate constant probability fat igue crack growth curves, viz.: da/dt = A(l)(DeltaK)(m)(-ln p(i))(beta), where da/dt is the crack growth rate, DeltaK the stress intensity factor ra nge, P the probability level while m and P are constants. Crack growth in deaerator feedwater storage vessels was taken as a useful p ractical application to demonstrate the benefits of this probability approa ch to fatigue crack growth. In this instance the average fatigue crack grow th law for low alloy steels at 200 degreesC in water was utilised to assess the fatigue crack growth, which initiated from a surface corrosion pit, an d extended through the vessel wall. Realistic probability of failure, POF, levels were established for a variet y of circumstances and the effects of a delay in an inspection on the incre ased levels of POF in working deaerators were quantified. Finally, this paper illustrates how probabilistic calculations can be pract ically incorporated into any decision making process. (C) 2001 Elsevier Sci ence Ltd. All rights reserved.