STOCHASTIC CRACK-PROPAGATION IN OFFSHORE STRUCTURES - THE SENSITIVITYOF COMPONENT LIFETIME TO WAVE DISTRIBUTION MODELS

Citation
A. Ahmadshariff et al., STOCHASTIC CRACK-PROPAGATION IN OFFSHORE STRUCTURES - THE SENSITIVITYOF COMPONENT LIFETIME TO WAVE DISTRIBUTION MODELS, Journal of offshore mechanics and Arctic engineering, 120(1), 1998, pp. 43-49
Citations number
13
Categorie Soggetti
Engineering, Marine","Engineering, Mechanical
ISSN journal
08927219
Volume
120
Issue
1
Year of publication
1998
Pages
43 - 49
Database
ISI
SICI code
0892-7219(1998)120:1<43:SCIOS->2.0.ZU;2-K
Abstract
Stress variations induced by wave loading can lead to fatigue crack gr owth in structural components of offshore structures. This paper is co ncerned with the influence of the form of the statistical distribution s for wave height on the damage accumulation and lifetime of a structu ral component. Damage accumulation is modeled by a stochastic Paris-Er dogan equation in which the increase in crack size is proportional to a power (m) of the range of the stress intensity factor Analytic expre ssions for the mean and variance of damage, and approximate mean lifet ime, of a component are derived for the case in which m equals 2. It i s seen that these depend on both the mean and variance of the stress d istribution. The results are compared with those obtained by simulatio n, and the adequacy of the approximation is demonstrated. Simulation r esults using Rayleigh and Weibull distributions for wave heights are a lso given for the case in which m equals 3. It is shown that the Weibu ll distribution gives a better fit to empirical wave height distributi ons than does the Rayleigh distribution. Furthermore, when m equals 3, there is a substantial difference between results obtained by fitting Rayleigh and Weibull distributions to wave height data. The former le ads to considerable overestimation of lifetimes. Ir is argued that Wei bull distributions are more appropriate in determining lifetimes since the two parameters in the distribution allow more accurate representa tion of the mean and variance of the stress distribution.