ESTIMATION OF EXTREME BENDING MOMENTS ON SHIPS FROM LIFETIME FATIGUE LOADS

Abstract

The U. S. Coast Guard uses a U. S. Navy methodology and a computer program called SPECTRA to probabilistically characterize wave-induced bending moments on surface vessels. SPECTRA is primarily used for fatigue design based on defined cells of vessel operation with specified heading, sea condition and speed in order to calculate bending response using the probability a ship is within each cell in a specified time period. In this study, the SPECTRA output for a hypothetical ship was obtained to examine its appropriateness to be used as a basis to characterize lifetime extreme design bending moments on ship hulls. The objective was to develop a method to utilize the SPECTRA fatigue load output to estimate the parameters of an extreme value distribution, such as the Weibull probability distribution, for the largest bending moment of k years. The study examined how to appropriately interpret and use the mean and variance of the bending moments obtained from SPECTRA for this purpose. A four step method is proposed in this thesis involving first getting the statistical moments of the data from the SPECTRA histograms, estimating the parameters of the Weibull using these moments, finding the moments of the largest in k years from the generated distribution, and finally estimating the parameters of the Weibull for the largest in k years from these moments. The study also includes the development of an efficient and robust method of estimating the parameters and moments that is called the adaptive technique, involving exact calculation and numerical integration. The method is illustrated using a hypothetical case and verified using extreme value computations. It is also observed that the SPECTRA output based on specifying two or more years produces only minor enhancements in the estimated moments for one year and does not produce the statistical moments of extreme loading.