cumulative fatigue 3

[mikebb]

Hi,

I am in the lengthy process of carrying-out the fatigue analysis of a transportable pressure vessel. I am using PD 5500:2003 as my guide. Have no problems so far in calculating the fatigue life associatiated with individual stress ranges. However, I dont understand the need for check (C-6), ie that sum of n/N 's are <= 1. I guess I am struggling to understand the need to combine the effects of unrelated fatigue lives to provide a cumulative value. As it stands, my figures comply, but would like to understand the need for the check. Hope someone can illuminate me !

 

[davefitz]

I am not familiar with PD5500, but the check of cumulative life is a basic requirement for any fatigue analysis.

In the older simpler fatigue desing codes, one might have over a 50 year period 100 cold startups with each consuming 1/2000 life fraction, 2500 hot restarts with each consuming 1/10,000 life fraction. The cumulative fatigue life fraction consumed from predictable transients would then be
100/2000 + 2500/10,000 = 0.29 . If the value had turned out to be greater than 1.0, then the component would not survive the nominal 50 yr life.

Newer codes use a more sophisticated means of predicting fatigue life consumption- using the "rainflow cycle counting algorithm" to combine successive strain -time histograms - this newer procedure allows one to discard or ignore many of the smaller alternating strain ranges.

 

[TERIO]

This is called Miner's rule or Palmgren-Miner linear damage hypothesis. The idea, as Dave pointed out is that each cycle uses up a certain fraction of the materials fatigue life.

Note: even with rainflow counting Miner's rule is still used to accumulate fatigue damage due to different types of cycles.